Split (3)

train · 24.8k rows

Search is not available for this dataset

article stringlengths 4.36k 149k	summary stringlengths 32 3.35k	section_headings listlengths 1 91	keywords listlengths 0 141	year stringclasses 13 values	title stringlengths 20 281
Allosteric regulation is a key component of cellular communication , but the way in which information is passed from one site to another within a folded protein is not often clear . While backbone motions have long been considered essential for long-range information conveyance , side-chain motions have rarely been considered . In this work , we demonstrate their potential utility using Monte Carlo sampling of side-chain torsional angles on a fixed backbone to quantify correlations amongst side-chain inter-rotameric motions . Results indicate that long-range correlations of side-chain fluctuations can arise independently from several different types of interactions: steric repulsions , implicit solvent interactions , or hydrogen bonding and salt-bridge interactions . These robust correlations persist across the entire protein ( up to 60 Å in the case of calmodulin ) and can propagate long-range changes in side-chain variability in response to single residue perturbations . Allostery is an essential feature of protein regulation and function . Allosteric regulation acts by linking distant sites of a protein together in such a way that information about one site is transmitted to and influences the behavior of another . Chemical modifications as subtle as the phosphorylation of a serine residue can cause dramatic changes in protein function [1] , and shifts in structure as small as 1 Å have even been shown to modify behavior in a domain up to 100 Å away [2] . Traditionally , allostery has been understood as a feature of symmetric , multi-subunit proteins where the binding of a ligand to one subunit facilitates the binding of similar ligands to the other subunits , resulting in cooperative binding transitions [3] . However , allosteric behavior has now been observed within a single protein domain [4] and its definition extended to include any shift in protein structure and function at one site resulting from modification at another . Moreover , it was proposed some time ago that information regarding the binding of a ligand or other modification at one protein site could be transmitted through altered protein fluctuations , even if the protein's average structure remains unaffected [5] . Two particularly clear examples of this kind of dynamic allostery have been recently observed in the binding of cAMP to the CAP dimer and in the subsequent binding of the cAMP-activated CAP dimer to DNA [6] , [7] . In the first step , the binding of cAMP to one monomer of CAP lowers the binding affinity of cAMP to the second even though no structural changes are observed , and calorimetric analysis suggests that the negative cooperativity results entirely from entropic effects [6] . The observed allosteric effect of protein fluctuations has led to the idea that allostery may be present in all proteins [8]–[10] , and that functional allostery simply exploits and refines pre-existing long-range correlations and interaction networks . In fact , such networks are to be expected given the physical constraints of the densely-folded , yet fluctuating , protein . Just as in any condensed phase , significant fluctuations in this packed environment are permitted through correlated motions . Qualitative experimental evidence for long-range correlation abounds in studies demonstrating allosteric regulation , as exemplified in [1] and [2] . However , attempts to quantify these long-range correlations using NMR techniques have proven difficult [11]–[13] , and much of our current understanding of correlated motions has come from analyses of molecular dynamics ( MD ) simulations . Traditional MD trajectories evaluated with covariance matrices and principle component analyses [14] have shed light on important features of intra-protein correlations , such as how backbone motions tend to be significantly correlated within secondary structural units [14] and how a few flexible hinge residues can cause large motions within otherwise stable folds [15] . Energy-perturbative MD simulations , such as anisotropic thermal diffusion [16] and pump-probe MD [17] , have been used to observe the rapid anisotropic diffusion of an energy perturbation within the protein . However , these MD studies are limited in their ability to characterize sluggish rearrangements and have largely neglected the contributions of correlated side-chain fluctuations . Within the folded protein , side-chains are significantly less ordered than the backbone [18] , and alternative side-chain configurations in protein crystals are more prevalent than previously thought [19] . The thermodynamic importance of this side-chain variability in calmodulin-ligand binding has been highlighted in Refs . [20]–[22] . In addition , the participation of side-chain fluctuations in long-range networks has been demonstrated through NMR mutational studies [9] , [23] . In one MD simulation designed to incorporate data from NMR experiments , correlations were even observed between side-chains whose motions appeared decoupled from those of their backbone atoms [24] . Double mutant cycles [25] have also been applied to examine the dependence of folding and binding on interactions between specific residue pairs . While such mutational studies can demonstrate the interactions of certain residue pairs , they are experimentally demanding , making it difficult to obtain a comprehensive picture of any long-range side-chain interactions present , in particular those involving residues essential for folding stability . As an alternative , an evolutionary statistical network analysis method has been developed to determine networks of correlated residues that are common to evolutionarily related proteins [26] . Although this method has had some success in identifying allosterically-related regions within proteins [27] , its robustness has been challenged in a study on artificially-generated sequences [28] . In principle , it is also limited to detecting correlated changes in residues during evolution , presumably highlighting only correlated networks with a selected function and can therefore say little about the presence or absence of other correlations . In this study , we employed an atomistically detailed model to examine the kinds of correlations that emerge among side-chain fluctuations within the natively-folded protein . The computationally inexpensive nature of our model energy function [21] , together with a variety of advanced Monte Carlo sampling techniques , allowed an unprecedentedly thorough investigation of the correlations among these fluctuations that result from different types of interactions . Keeping the backbone fixed , we find that long-range correlation of side-chain fluctuations can emerge from different types of atomic interactions , that significant correlations persist across the entire folded protein , and that these correlations alone can propagate changes in structure and mobility over scales as large as 50 Å . In order to investigate the correlated rearrangements that arise from side-chain fluctuations alone , it was necessary to isolate these motions from other sources of configurational change . For this reason , we held the backbone fixed in its folded conformation throughout the calculations described in this paper . While fluctuations resulting from bond stretching and angle bending are important and likely to give rise to a great deal of correlated motion , we focused here instead on side-chains' torsional degrees of freedom , as these rotations give rise to the changes in atomic configurations that are largest in magnitude . This study made use of a model we designed to roughly capture the essential physical determinants of side-chain behavior within the folded protein , namely , steric repulsions , van der Waals attractions , hydrogen bonding , salt-bridge interactions , and solvation [21] . While not fully realistic in every particular ( e . g . , resolving the positions only of nuclei heavier than hydrogen ) , the model properly represents the variety , strength , and anisotropy of the side-chain interactions and the physical constraints of the folded backbone on which they reside . We explored this model with Monte Carlo ( MC ) sampling ( see Methods ) . Each MC step consisted of the proposed rotation of a single randomly-chosen side-chain dihedral angle . To promote broad sampling of thermally accessible configurations , we permitted moves through sterically disallowed regions of state space . Using exact correction methods , we constructed equilibrium averages with contributions only from sterically allowed structures ( those in which each heavy atom excludes a spherical volume with radius 0 . 75 times its van der Waals radius ) . See [21] for details . Boltzmann-weighted ensembles of the side-chain configurations determined using this sampling procedure include a diverse set of rotamer states and correlate well with experimental observations of side-chain fluctuations and changes in entropy upon ligand binding [21] . We therefore applied this method to investigate correlations among the diverse set of rotamer states . Several experimental approaches that probe correlations within proteins mutate single residues and observe the resulting changes in structure , function , or dynamics [9] , [23] , [29] , [30] . We began our examination of side-chain correlations in a similar way by modeling the changes in torsional variability that occured throughout a small globular protein , barstar , as a result of perturbations to a single side-chain . Previous MD simulations of barstar suggested a relatively rigid backbone , as well as significant variability in side-chain packing [31] . Additional results from NMR experiments showed that the P27A mutation results in detectable dynamic changes even in residues more than 12 Å away from the mutation site , and suggested that the motion of barstar's side-chains gives rise to a network of correlated residues [32] . Quantifying such correlated fluctuations requires a metric that can report on the extent of local variability at the single residue scale . For this purpose , we calculated the Gibbs entropy for each residue , , associated with occupying distinct rotameric states . ( 1 ) where denotes the set of torsional variables for each of the rotatable - hybridized bonds belonging to residue , and denotes the set of ideal torsional angle values for the th torsional angle in residue . While - hybridized bonds were allowed to rotate , they were also excluded from the statistical analysis due the difficulty in determining ideal dihedral angles [33] . The probabilities of these states were calculated in simulations by constructing histograms over the course of importance sampling from the Boltzmann distribution of side-chain configurations . In doing so , we focused on the inter-rotameric rearrangements ( those between the three most likely energy basins for the torsional angle of an - hybridized bond ) , which allowed the calculation of absolute local entropies that would have been impractical at a higher level of resolution . However , intra-rotameric fluctuations ( those within a single torsional energy basin ) are sensitive to the structural perturbations we applied , and it is necessary to allow deviations from these ideal angles , , in order to fully account for the variety of possible side-chain configurations [34] , [35] . A quadratic energy is associated with these deviations ( see Methods ) . Fig . 1 shows the change in that resulted from a single-residue perturbation . Residues shown in red demonstrated a statistically-significant increase in side-chain variability , while the variability of those shown in blue decreased ( see Methods ) . The perturbations shown , a mutation of isoleucine to glycine at position 86 ( Fig . 1 ( a ) ) and a constraint of the glutamate in position 46 to its crystalline configuration ( Fig . 1 ( b ) ) , were chosen to demonstrate the types of changes possible . ( A comparison to the previously studied P27A [32] was not possible , since neither proline nor alanine residues have rotatable side-chains in our model . ) Surprisingly , removing the isoleucine side-chain at position 86 ( circled in Fig . 1 ( a ) ) not only affected the local entropy of a few neighboring residues , but also altered the side-chain variability of residues much farther from the site of mutation . Motions of even distant residues must therefore be linked to those of residue 86 . Because the interaction potentials in our model are short in range , the changes in fluxionality that resulted from this mutation must propagate through neighboring residues to those farther away . Fig . 1 ( b ) shows analogous changes that resulted when a residue , E46 ( circled ) , is merely frozen into its crystalline conformation . Such a reduction in motion of one side-chain might be expected to result in the increased variability of its nearest neighbors . However , we found that even so subtle a constraint resulted in unexpected and wide-spread changes in the side-chain fluctuations . Some residues near the frozen amino acid even became slightly more constrained while the variability of a few residues farther away increased . A similar effect was observed in NMR experiments upon ligand-binding in stromelysin 1 , where the few residues participating in strong interactions with the ligand lost mobility , but the order parameters of those farther away actually decreased upon binding , indicating an increase in their entropy [36] . It was suggested that the increased fluctuations far from the binding site may counter the loss of entropy at the binding site itself and therefore assist in modulating the thermodynamics of binding . The changes in side-chain statistics we observed as a result of these single residue perturbations are not readily intuited . Increasing or decreasing disorder at one site may result in the same or opposite effect in other regions of the folded protein , and the effects cannot be easily predicted from the spatial arrangement of the residues . Correlated fluctuations within the folded protein are commonly quantified using Pearson correlation coefficients [14] . Despite their limitations in detecting nonlinear correlations and correlations between the motions of particles moving orthogonally to one another [37] , Pearson coefficients have yielded important information regarding correlated motions . These coefficients are most appropriate for backbone motions as these motions are expected to be correlated in similar directions and to be linear in nature due to the stiffness and collective motions of various secondary structural elements [14] . However , in a study analyzing the results of molecular dynamics simulations of protein G and lysozyme , a generalized correlation measurement based on mutual information was able to detect significantly more correlation than the Pearson coefficient [37] . Side-chain motions are even more likely to fall outside the purview of the Pearson coefficient , dominated as they are by dihedral angle rotations . A parameter based on mutual information is able to provide a more robust measurement of correlated side-chain fluctuations [37] , and can be readily derived from simulation data in a similar way to the entropies calculated in the preceding section . We therefore chose to consider the mutual information associated with each pair of residues within a folded protein . Pairwise mutual information is a measure of the correlation between two random variables . In our case it reports on the degree of correlation between the rotameric state populations of two residues . The mutual information between residues and can be calculated as ( 2 ) where denotes the probability of each of the joint states of residues and , and is the number of rotatable - hybridized bonds in residue . After rearranging Eq . 2 and substituting in Eq . 1 , this becomes ( 3 ) where is the Gibbs entropy associated with the discrete rotameric states for residues and considered jointly . Thus when the fluctuations of the two residues are completely independent of one another , and . However , when the residues are correlated , their entropies are inseparable , and . One difficult feature of mutual information is that a numerically-calculated estimate of two completely uncorrelated variables only approaches zero at the limit of infinite sampling . For any finite sampling , a small amount of spurious mutual information will be observed , regardless of the actual coupling between the two variables [38] . When calculating numerically , this inherent bias in the noise must be accounted for in order to determine the mutual information's statistical significance . We used two approaches to address this bias . In the first , we subtracted out the expected spurious mutual information to estimate the true amount of correlation between the two variables . The resulting excess mutual information , , between residues and is defined as ( 4 ) is the numerically-calculated mutual information measured over a finite sampling period consisting of MC steps . is the same measurement , but this time computed within a non-interacting reference state , where no correlations are possible ( see Methods for details ) . is then an estimate of the mutual information of the infinitely-sampled ensemble . In the second approach , we focused on the robustness of the mutual information measurement , calculating its signal-to-noise ratio , . The extended structure of calmodulin ( 3cln [39] ) , as shown in Fig . 2 ( a ) , provides an exemplary test case for examining how side-chain fluctuations are correlated within the folded protein . Although in solution this chain collapses , the structure of the crystal is extended , featuring two globular regions connected by an extended -helix . Any information shared between the two lobes must pass through this extended -helix , since the pairwise interactions in our model largely decay by 7 Å . We calculated the pairwise excess mutual information , , for all residue pairs in -bound calmodulin , as well as the ratio of in order to gauge the significance of the measured correlations . Both quantities are shown in Fig . 3 as functions of the residues' position along the backbone . For reference , we present in Fig . 2 ( b ) the spatial distance between residues in the native structure as a function of the same indices . Panels ( b ) – ( e ) of Fig . 3 indicate mutual information resulting from various interaction types considered in isolation . Panel ( f ) gives results for the full model . Different types of inter-atomic interactions in our model gave rise to different patterns of correlated fluctuations . In Fig . 3 ( b ) , correlations that result solely from steric repulsions are shown . While the signature of calmodulin's -helical structure can be clearly seen along the diagonal , where residues and or and are often highly correlated , many other residue pairs appear significantly correlated as well , even those that are spatially distant . In Fig . 3 ( c ) , the correlations that result from the implicit solvent alone are shown . These correlations are more limited , restricted almost completely to residues that are nearby in space , as can be seen when comparing Fig . 3 ( c ) to Fig . 2 ( b ) . Again the -helical residues display appreciable correlation , even more than that resulting from the repulsive sterics , as might be expected from their high degree of solvent exposure . The correlations that result from considering van der Waals attractions along with the repulsive sterics is shown in Fig . 3 ( d ) . While the correlations along the -helix remain strong , many other correlations emerge as a result of these attractions . Hydrogen bonding and salt bridge interactions , taken alone , generate highly significant correlations throughout the entire structure ( see Fig . 3 ( e ) ) , which appear remarkably insensitive to spatial distance . Since only a subset of the residues participate in such interactions , the fluctuations of the remaining residues are completely uncorrelated in this restricted version of our model . The full potential , used to generate the data in Fig . 3 ( f ) , results in both the most significant signal-to-noise ratios and the largest excess mutual information values , indicating a large degree of correlation that spans the full range of inter-residue distances while retaining features of the dominant -helical structure . To further explore how different interactions give rise to long-range correlations in both a small globular protein as well as the extended calmodulin structure , we calculated the average excess mutual information per residue pair for all residue pairs in calmodulin and barstar , resolved by the spatial inter-residue distance between atoms . ( See Fig . 4 . ) In both proteins , steric repulsions alone give rise to small , but significant , correlations that persist across the entire protein structure . The same is true for the implicit solvent interactions and their combination , S+IS . However , much larger correlations emerge when van der Waals attractions are considered in addition to the steric repulsions . Hydrogen bonding and salt bridge interactions are clearly the most correlating types of interactions considered . However , the full potential , which combines all these interaction types , results in the largest overall correlation . An additional feature within these plots deserves mention; in both proteins , correlation is at a maximum around 6 Å for all subsets of interactions excepting hydrogen bonding and salt bridges . This short-distance peak indicates that residue pairs adjacent in the amino acid sequence ( whose -carbons are separated by Å ) do not interact as strongly on average as do residue pairs that are positioned slightly farther apart . In -helices , neighboring residues point in different directions and , while still likely to interact with their sequential nearest neighbor , are more likely to interact strongly with their and neighbors . In -sheets , however , residues alternately point towards different faces of the sheet , so that the side-chains on residues and are much more likely to interact with one another than do those on and . Residues influenced only by hydrogen bonding and salt bridge interactions , when artificially freed of the steric constraints that would keep them from collapsing back on themselves , still correlate most strongly with their nearest neighbors . Substantial long-range correlation is seen throughout both barstar and calmodulin . Moreover , the fact that so many subsets of the full potential independently give rise to long-range correlations suggests that correlated side-chain fluctuations should be a robust characteristic of most protein sequences and nearly any globular fold . Through correlated side-chain fluctuations , local perturbations to a protein ( e . g . , due to small ligand binding ) could in principle be transmitted over substantial distances . We scrutinized this possibility by examining the consequences of mutating a single residue in calmodulin . Such a mechanism of communication was described earlier for barstar ( see Fig . 1 ) , whose size limited our conclusions to distances of less than 30 Å . Calmodulin , in its extended structure , provides a better test of the ability of side-chain correlations to transmit information over long distances . We focused this analysis on correlated fluctuations involving one particular residue in calmodulin , 30K , which we observed to be significantly correlated with several other residues ( see Fig . 3 ( f ) ) . In Fig . 5 ( a ) , 30K is colored black , while the pairwise mutual information between this side-chain and all others is indicated in bluescale . Appreciable correlations are apparent throughout the lower globular region near residue 30 . The correlations become stronger within the spatially-constricted alpha-helical bridge and spread out again and weaken in the far lobe . To determine whether these correlations could transmit structural and dynamical information over significant distances , we mutated residue 30K to glycine . The resulting change in for each residue is shown in Fig . 5 ( b ) . A significant decrease in entropy was detected in some neighboring residues , while both increases and decreases in entropy were found for residues farther from the mutation site . Although unexpected , the reduction in entropy at residue resulting from the removal of a neighboring bulky residue can be readily explained if that nearby mutation results in the loss of a potential hydrogen bonding partner for residue . Such a loss can result in the probability associated with the hydrogen-bonding subset of residue 's configurations being greatly reduced . We found statistically significant changes in entropy even in the globular region opposite that of residue 30 . Thus we conclude that side-chain fluctuations alone can reliably propagate the effect of a single point mutation over at least 50 Å . When comparing Fig . 5 ( a ) to Fig . 5 ( b ) , it is clear that some , but not all , of the strongly correlated residues in the wild-type calmodulin experience detectable changes in their side-chain variability as a result of this particular mutation . Even some residues that are minimally correlated with residue 30K show significant shifts in their side-chain statistics . Although the mutual information can tell us a great deal about the degree of correlation between two side-chains in our model , it is not a discriminating predictor of changes in side-chain variability upon mutation . The discrepancies are likely due to the fact that our calculation of mutual information lacks contributions from correlated intra-rotameric fluctuations , which are still able to convey information in our model and will therefore influence the detected changes upon side-chain mutation . Furthermore , observing the statistically significant changes in Fig . 5 ( b ) requires a great deal of sampling – were more sampling feasible , additional changes would likely be detected . If the side-chain motions of a protein's different residues were negligibly correlated , then the total entropy associated with transitions among distinct rotameric states could be calculated as a simple sum of single-residue contributions , . The excess mutual information , summed over all residue pairs , provides a rough measure of the error in such a mean-field estimate . Correspondingly , the quantity characterizes the global thermodynamic significance of inter-residue correlations . For crystalline barstar modeled with the full potential , is calculated to be 72 kJ/ ( mol300 K ) . The higher-order correlations expected in such a dense environment [40] ( see Fig . 3 where a single residue is often significantly correlated to several others ) make this value an overestimate of the total correlation . Even so , its magnitude is noteworthy . In addition , while allowing intra-rotameric fluctuations , this calculation neglects their contribution to the total correlation , which were found to be essential in reproducing the calorimetric of calmodulin with its ligands in [21] and are likely to be substantial . The rigidity of the peptide backbone in these calculations justifies to some extent our schematic model of side chain interactions: For our purposes the potential energy function need not resolve subtle thermodynamic differences among diverse chain conformations , but instead serves to establish basic length and energy scales for rearrangements within the native state's basin of attraction . In considering the biological relevance of our results , backbone rigidity is in part justified empirically by the observation that only weak correlations exist between backbone NMR order parameters , , and their associated side-chain order parameters , [41] . This weak correlation is likely due to the fact that side-chain and backbone fluctuations largely occur on different time-scales [42] , with typical side-chain fluctuations ranging from picoseconds to nanoseconds , while typical collective backbone fluctuations range from nanoseconds to seconds and longer . However , it is important to assess how variations in backbone configuration of the folded protein might influence the side-chain correlations we have calculated . Toward this end we examined four different structural models from an NMR structure of barstar ( 1btb [43] ) . These four conformations were chosen to represent the range of models included in the NMR structure ( see Methods ) . In each case plots of per pair vs . inter-residue distance for the full potential closely resemble results for the crystal structure ( see Fig . 6 ) . Since the statistics of side-chain rotations in a fluctuating backbone environment can be rigorously decomposed into contributions from sub-ensembles in which the backbone is held fixed , the consistent nature of the observed long-range correlation from one backbone structure to another establishes their robustness to typical backbone motions . Larger backbone fluctuations , however , such as partial unfolding events or the motions of hinged regions , are certain to disrupt many of these correlations and may limit their role in conveying allosteric information . In particular those correlations arising from contact between residues that are spatially proximal , but distant within the protein's amino acid sequence , will attenuate as backbone motions carry them away from one another . However , correlations between residues linked through a path of sequential neighbors , such as those observed along the central -helix of crystalline calmodulin in Fig . 5 , may persist . As a result , some information may continue to be transmitted through side-chain fluctuations even after significant backbone rearrangements , as long as the secondary structure , which is responsible for many of the observed correlations between sequential neighbors ( see Fig . 3 ( f ) ) , remains intact . In addition to scrutinizing the effect of different types of atomic interactions , we also examined how a protein's amino acid composition might contribute to stronger or weaker correlations among its side-chain fluctuations . To do so , we took a set of twelve small globular proteins with different sequences and folds ( see Methods ) and calculated the average excess mutual information per pairwise interaction for each amino acid across the entire set of proteins . The result is plotted in Fig . 7 . In general , amino acids with the most - hybridized rotatable bonds resulted in the largest values . The amino acid arginine is clearly the most strongly correlated residue , followed closely by lysine . While both of these amino acids have four rotatable bonds , arginine is considerably bulkier than lysine , with more potential hydrogen bonding partners . In addition , arginine has been found to take on fewer alternate rotameric states in simulations of folded proteins than lysine [44] . Similarly , the bulky aromatics ( Phe , Tyr , Trp ) were more correlating than their single - hybridized rotamer would indicate , while isoleucine and leucine are both much less correlating than the other residues with two rotatable bonds: glutamine and glutamate . Glycine , alanine , and proline all have , since they possess no rotatable bonds in the model . Recent NMR measurements on eglin C provided a good opportunity to compare our results with experimental evidence of wide-spread changes in side-chain fluctuations resulting from small perturbations [9] , [30] . In this work , a series of valine residues were mutated to alanine at various positions in eglin C , a small globular protein with a relatively static backbone , and the resulting changes in the order parameters of side-chain methyl groups were measured . The changes in the NMR-measured order parameters were in many cases quite low; the majority of the statistically significant changes were less than 0 . 05 ( order parameters range from 0 . 0 for a completely disordered vector to 1 . 0 for a completely ordered one ) , with only a few residues showing changes greater than 0 . 1 [9] , [30] . The magnitude of these NMR-measured changes combined with the significant statistical errors in our calculations ( the average standard deviation was 0 . 02 ) rendered such a comparison difficult . In the cases where we could resolve the changes in our MC-calculated order parameters enough to make a meaningful comparison to the experiments we found little correspondence between our data and the NMR measurements . NMR order parameters are expected to underestimate the full range of side-chain motion , as they neglect motion slower than the tumbling time of the molecule , and recent work demonstrates that such motion can be substantial [45] , [46] . Similarly , our calculation is also expected to underestimate the full range of motion accessible to the side-chains due to the fixed backbone , which we observed previously to be particularly problematic in calculating methyl group order parameters for alanines [21] . As a result , a poor correspondence between our calculated methyl-group order parameters and those derived from NMR relaxation experiments , in particular those involving mutations to alanine , is perhaps to be expected . In past work we nevertheless demonstrated a clear correspondence to the measured NMR order parameters for wild-type eglin C using the same computational approach [21] . Even if the model is not sufficiently accurate or detailed to make quantitative predictions for altered side-chain fluctuations in the specific case of eglin C , the conclusions we draw here for long-range correlations among side-chain fluctuations should be pertinent to the biophysics of folded proteins in general . The propagation of information across long distances within the folded protein is of great importance in allosteric regulation . While backbone structural changes and fluctuations have long been studied as the bearer of this information , correlated side-chain fluctuations provide potential information conductance pathways that have often been overlooked . In this work we have examined these side-chain correlations and found that changes in fluctuations in response to even single residue perturbations , such as a point mutation or residue immobilization , are statistically significant , widely distributed , and not easily intuited from the protein's sequence or structure . The correlations emerge independently from several different sources: steric repulsions , solvation , and hydrogen bonding and salt bridges . Together , these interactions give rise to robust , statistically-significant correlations that persist across the entire spatial extent of both barstar and calmodulin . In the calculation of the mutual information values , we model the protein's backbone as a rigid structure , enabling us to investigate the degree of correlation among side-chain fluctuations alone . An understanding of the role of backbone fluctuations is necessary , however , to judge the biological relevance of these observed correlations . Interestingly , we found that significant correlations are present among the side-chains of barstar in several different backbone structures , which collectively represent the range of its typical conformational fluctuations in solution . Moreover , since the time scales characteristic of backbone and side-chain motions are likely well separated in many cases , communication via side-chain rearrangements as we have described may often occur on an effectively static backbone . However , upon larger backbone rearrangement , such as hinge motions or partial unfolding events , a significant fraction of the observed side-chain fluctuations are likely to decouple , and thus the role of side-chain correlations in allosteric regulation where large backbone rearrangements are known to occur may be limited . We also utilized a simple implicit solvent model to account for the solvent's mean field effect on the protein . While this approach allows us to focus directly on the protein's degrees of freedom , it neglects solvent fluctuations . Physically realistic solvent fluctuations are likely to influence side-chain fluctuations in the same way that side-chain fluctuations influence one another , and the reverse is also true . This potential for correlation between solvent and side-chain fluctuations suggests that fluctuations of the solvent shell may also convey information from one site on the protein to another . Indeed , we observed that even the mean field effect of solvation mediates the correlation of side-chain fluctuations , as seen in Fig . 3 ( c ) . However , more and stronger correlations were observed to arise from hydrogen-bonding and salt bridge interactions ( see Figs . 3 ( e ) & 4 ) , and it is largely through these very effects that the solvent molecules would influence the side-chains . While we are not able to explore the resulting implications in our current work , the demonstration of allosteric effects mediated through solvent fluctuations would be quite intriguing . Finally , it is important to note that the magnitude of the correlations measured here is only a fraction of the total magnitude possible among the side-chains , since correlated fluctuations within each individual rotameric well , which are not included in our correlation metric , are sure to contribute significantly , perhaps even to a greater degree , to the overall amount of correlation . Even so , correlations amongst inter-rotameric fluctuations alone reveal much about the way side-chain fluctuations give rise to long-range correlations within the folded protein . The role of these robustly correlated side-chain fluctuations in allosteric regulation should be considered further . Our model is defined by an energy function that depends on the atomic coordinates of the protein's residues . The only degrees of freedom in our model are the dihedral angles , where represents the ideal angle of each rotameric state taken from an empirical rotamer library [33] , and describes torsional variations within the potential energy well of each rotamer . Thus the energy function depends on the full set of and values ( denoted and ) for all residues , ( 5 ) The potential is piecewise quadratic in , biasing dihedral angles towards their ideal values; includes a Lennard-Jones function that governs both repulsive sterics ( with a hard-sphere cutoff at 0 . 75 times the van der Waals radius ) and attractive van der Waals interactions , as well as hydrogen-bonding and salt-bridge terms; and accounts for solvation using an approach based on the solvent-accessible surface area . A detailed description of the model is given in [21] . The analysis outlined in this paper requires a good structural model of the protein considered . For barstar , we use a crystal structure of the mutant C82A at 2 . 8 Å resolution ( 1a19 [47] ) , except for the analysis in Fig . 6 using different NMR structural models , where the structure 1btb [43] was used . In this case four models ( numbers 3 , 4 , 16 , and 29 ) were chosen to represent the structural variety within the full set of NMR models , as assessed by the RMSD values calculated between all possible pairwise combinations of the four models and compared to those of the full ensemble . For calmodulin , a crystal structure at 2 . 2 Å was used ( 3cln [39] ) . In our calculations the positions of calcium ions and the side-chains bound to them were held fixed . The results in Fig . 7 also included the following proteins: eglin c ( 1cse [48] ) , GB3 ( 1igd [49] ) , protein L ( 1hz6 [50] ) , PYP ( 1f9i [51] ) , PZD2 ( 1r6j [52] ) , SH2 ( 1d1z [53] ) , CspA ( 1mjc [54] ) , ubiquitin ( 1ubq [55] ) , and tenascin ( 1ten [56] ) . Side chain configurations were sampled from the Boltzmann distribution using Metropolis Monte Carlo techniques . In order to calculate the highly converged measurements of needed to produce the results shown in Figs . 1 & 5 ( b ) , additional adaptive umbrella sampling techniques and biased sampling procedures were utilized [21] , [57] . Averages at 300 K were finally constructed from such calculations by summing results for different energies with appropriate statistical weights .	Allosteric regulation occurs when the function of one part of a protein changes in response to a signal recognized by another part of the protein . Such intra-protein communication is essential for many biochemical processes , allowing the cell to adapt its behavior to a dynamic environment . Most studies of the information conveyance underlying allostery have to date focused on the role of backbone motions in mediating large structural changes . Here we focus instead on more subtle contributions , arising from fluctuations of side-chain torsions . Using a model for side-chain bond rotations in the tightly packed environment imposed by native backbone conformations , we observed significant sensitivity of side-chain organization to small , localized perturbations . This susceptibility arises from correlations among side-chain motions that can propagate information within a protein in complex , heterogeneous ways . Specifically , we found appreciable correlations even between side-chains distant from one another , so that the effect of a minor perturbation at one site on the protein could be observed in the altered fluctuations of side-chains throughout the protein . In conclusion , we have demonstrated that the statistical mechanics of correlated side-chain fluctuations within a model of the folded protein provides the basis for an unconventional but potentially important means of allostery .	[ "Abstract", "Introduction", "Results", "Discussion", "Methods" ]	[ "protein", "chemistry", "entropy", "chemical", "biology", "thermodynamics", "chemistry", "biology", "computational", "chemistry", "biophysics", "simulations", "biophysics", "theoretical", "chemistry", "physical", "chemistry" ]	2011	Long-Range Intra-Protein Communication Can Be Transmitted by Correlated Side-Chain Fluctuations Alone
Schistosomiasis is a debilitating disease that affects over 240 million people worldwide and is considered the most important neglected tropical disease following malaria . Free-swimming freshwater cercariae , one of the six morphologically distinct schistosome life stages , infect humans by directly penetrating through the skin . Cercariae identify and seek the host by sensing chemicals released from human skin . When they reach the host , they burrow into the skin with the help of proteases and other contents released from their acetabular glands and transform into schistosomula , the subsequent larval worm stage upon skin infection . Relative to host invasion , studies have primarily focused on the nature of the acetabular gland secretions , immune response of the host upon exposure to cercariae , and cercaria-schistosomulum transformation methods . However , the molecular signaling pathways involved from host-seeking through the decision to penetrate skin are not well understood . We recently observed that heat shock factor 1 ( Hsf1 ) is localized to the acetabular glands of infectious schistosome cercariae , prompting us to investigate a potential role for heat shock proteins ( HSPs ) in cercarial invasion . In this study , we report that cercarial invasion behavior , similar to the behavior of cercariae exposed to human skin lipid , is regulated through an Hsp70-dependent process , which we show by using chemical agents that target Hsp70 . The observation that biologically active protein activity modulators can elicit a direct and clear behavioral change in parasitic schistosome larvae is itself interesting and has not been previously observed . This finding suggests a novel role for Hsp70 to act as a switch in the cercaria-schistosomulum transformation , and it allows us to begin elucidating the pathways associated with cercarial host invasion . In addition , because the Hsp70 protein and its structure/function is highly conserved , the model that Hsp70 acts as a behavior transitional switch could be relevant to other parasites that also undergo an invasion process and can apply more broadly to other organisms during morphological transitions . Finally , it points to a new function for HSPs in parasite/host interactions . Schistosome parasites have six different morphological stages during their life cycle , which requires an intermediate molluscan and a definitive mammalian host that the parasite must correctly identify and invade . Free-swimming , freshwater cercariae ( singular: cercaria ) are released from infected molluscs and invade mammals and humans for further development into larval worms called schistosomula ( singular: schistosomulum or schistosomule ) . Schistosomula adapt to survival in the host blood environment , evade the immune system , develop a gut to begin digesting red blood cells , elongate and traverse the human circulatory system , and eventually develop into egg-laying adult worms [1] . Cercariae are highly adapted for swimming and invading their mammalian hosts . Transcriptional studies show that cercariae have elevated expression of genes associated with metabolism and motility when compared with other stages [2 , 3] . Free-swimming cercariae have a limited energy supply and a limited duration during which they can infect their host [4] . Thus , they must correctly identify and quickly respond to an appropriate host ( or source of chemoattractant ) , swim toward it , and begin the host penetration process . For the purposes of this report , we call this behavior cercarial honing or simply , honing . Swimming cercariae respond to changes in light levels , to thermal gradients , and to chemicals such as linoleic acid and L-arginine released from human skin [5–9] . After reaching the skin , the cercariae crawl along the skin surface until they identify a suitable location to penetrate . Parasite invasion through the skin involves the physical motion of swimming into the skin , in coordination with release of their acetabular gland contents , which include mucins to enhance the attachment to skin and proteases to degrade skin molecules [10–12] . While the ultrastructure of cercariae has been described before and after entry into the host [13–15] , protein regulators of cercarial honing and invasion have not been studied , with the exception of two reports [16 , 17] . In 1991 , Matsumura and others proposed that protein kinase C and calcium metabolism are involved in proteolytic enzyme release from cercariae acetabular glands [16] . Almost 25 years later , Ressurreição followed up on the work by Matsumura and recently reported that PKC , ERK , and p38 MAPK phosphorylation is involved in release of proteolytic enzymes from cercarial acetabular glands following the observation that inhibition of PKC , ERK , and p38 MAPK activities blocked linoleic acid-induced release of acetabular gland contents [17] . The current report further explores the molecular requirements for cercarial host invasion . We identify heat shock protein 70 ( Hsp70 ) as a potential molecular component involved in cercarial honing and show that inhibition of Hsp70 can bypass the requirement for linoleic acid , L-arginine , or any host-derived signal to induce cercarial host targeting behavior . Interestingly , numerous reports corroborate regulatory interplay between Hsp70 , PKC , ERK , and p38 MAPK activities [18–21] . Several studies led us to investigate the potential role for a heat shock pathway during cercarial honing and invasion . First , the heat shock response has traditionally been associated with cellular stress [22–24] , and cercariae are no exception to this , since they must transition from a cooler , low-saline and freshwater environment to the warmer , saline environment of a human host . Second , we recently observed an unexpected localization of heat shock factor 1 ( Hsf1 ) , the major transcriptional activator responsible for transcribing heat shock genes ( such as HSP70 and HSP90 ) , to the acetabular glands of cercariae [25] . This observation helps corroborate the findings of another study that showed the presence of Hsp70 in released acetabular gland contents [26] . Third , the heat shock response may play a role in other stages of schistosome infection as well . In particular , an induced heat shock response in the schistosome intermediate host Biomphalaria glabrata renders them susceptible to schistosome infection , while absence of a strong heat shock response leads to resistance [27] . Together , these studies suggest an important role for a heat shock pathway in parasitic schistosomes . Hsp70 , a member of the heat shock protein ( HSP ) superfamily , is structurally and evolutionarily conserved from prokaryotes to eukaryotes and generally functions as a chaperone protein that aids in ( re ) folding nascent and denatured proteins through interactions with its substrate domain and ATP hydrolysis ( for review , [28] ) . However , additional roles for Hsp70 outside of its well-established chaperone functions have also been described . Together with various co-chaperones , Hsp70 can also direct signaling pathways that control cell death , differentiation , homeostasis , and proliferation by modulating the function of key regulatory proteins ( client proteins ) [29] . This is observed in the regulation of tumor necrosis factor receptor 1 ( TNFR1 ) signaling [30] . Aggregation of TNFR1 leads to cell death; however , TNFR1 aggregation is inhibited when TNFR1 interacts with silencer of death domain ( SODD ) . Hsp70 is thought to bind to SODD , modifying it to induce SODD/TNFR1 interaction , thereby inhibiting TNFR1-dependent cell death [30] . Hsp70 also plays a role in modulating Smad-mediated transcription [31] . Smad proteins are essential transducers of the transforming growth factor superfamily . Smad-mediated transcription is enhanced by the activity of the melanocyte specific gene ( Msg1 ) protein , a transcriptional activator that cannot independently bind DNA but does so indirectly through interaction with p300/CBP . Hsp70 forms a complex with Msg1 , suppressing its interaction with p300/CBP , and consequently blocks Msg1 enhancement of Smad-mediated transcription [31] . As another example , in clathrin-mediated endocytosis , Hsp70 binds and holds clathrin triskelia , preventing their aggregation during the uncoating of clathrin-coated vesicles; in the other half of the clathrin cycle , Hsp70 releases the triskelia to allow the coating of new vesicles upon activation by some unknown signal ( s ) [32] . The role of Hsp70 in clathrin-mediated endocytosis resembles that which we propose here for cercarial honing , especially with respect to the sequestering of important cellular components until Hsp70 receives an activating signal to release its client protein . While identification of the mechanism for Hsp70 mediated regulation for clathrin-mediated endocytosis is a topic of much interest [33–35] , a similar mechanism and question just as interesting may apply to the cercarial honing and invasion process . In this study , we treated cercariae with modulators of Hsp70 protein that inhibit or activate Hsp70 via different mechanisms to explore whether Hsp70 functions in cercarial host invasion . Of interest , we found that 2-phenylethynesulfonamide ( PES ) , also known as pifithrin-μ , initiated the process of cercarial honing and invasion in the absence of any host-specific stimulants such as skin lipids or linoleic acid , and it did so with 100% effectivity , which is greater than that observed with either skin lipids or linoleic acid , albeit at a slower rate . PES specifically binds to Hsp70 ( Kd ~ 2 . 9 μM ) , and its derivatives do not interact with Grp75 or Grp78 , organelle-specific members of the Hsp70 family [36 , 37] . X-ray crystallographic analysis shows that PES interacts with residues L394 , P398 , L401 , G484 , N505 , and D506 in human Hsp70 . We propose a model that Hsp70 is involved in a signaling pathway that causes cercariae to begin host invasion maneuvers and that inhibition of Hsp70 bypasses the need for upstream host signals that normally initiate this process . We have recorded and observed over 200 videos of cercarial mobility in response to small molecule modulators that target Hsp70 , heat shock protein 90 ( Hsp90 ) , or apoptosis . To our knowledge , this is the first investigation of a molecular signaling pathway in cercariae that points to a role for Hsp70 as a regulatory factor for the transition between parasite development stages . In addition to providing a potential pathway to which we can direct drug development against schistosomes , these data could apply more broadly to other parasites and to other organisms during transitions or periods of rapid development [38] . Finally , we add to the current model in describing cercarial host invasion . Protein sequences of Hsp70 from various species most closely related to that of Schistosoma mansoni ( NCBI accession numbers: CCD76164 ( Smp_106930 ) and CCD76236 ( Smp_049550 ) were identified by the NCBI BLASTp function [39] and aligned using ClustalW2 using its default parameters [40] . A phylogenetic tree was generated using the output of the ClustalW2 alignment and TreeView X software . Biomphalaria glabrata snails infected with S . mansoni ( NMRI strain ) were obtained from Biomedical Research Institute ( BRI; Rockville , MD ) . Cercariae were collected from infected snails by light-induced shedding: the snails were kept in the dark overnight and then placed under bright light for 2 hours [41] . Cercariae were observed in 12-well or 24-well culture plates ( respectively about 1 , 000 or 500 cercariae per well ) using an inverted ( VanGuard 1493INi ) and upright stereo ( Olympus SZ30 ) microscope fitted with a camera ( Canon T5i ) . Videos were captured with the focus on the bottom of the wells at 40× and 10× magnification and a camera setting of 1280 by 720 at 60 fps . Images shown in figures are frames extracted from the videos . Treatments of cercariae included the addition of the following substances; the treatment concentrations were chosen based on those used in the studies indicated ( typically increased several-fold over those used in cell-based studies ) : human skin lipid ( finger swipe ) , linoleic acid ( Sigma L1012 ) [42] , Hsp70 modulators 2-phenylethynesulfonamide ( PES; Sigma P0122 ) [36] , MKT-077 ( Sigma M5449 ) [43] , 115-7c ( Stressmarq SIH-123 ) [44] , and VER-155008 ( Sigma SML0271 ) [45] , Hsp90 inhibitors geldanamycin and 17-dimethylaminoethylamino-17-demethoxygeldanamycin ( 17-DMAG; these Hsp90 inhibitors were a kind gift of Giselle Knudsen and Jonathan Choy from the Small Molecule Discovery Center at UCSF ) [46] , pan-caspase inhibitor Z-VAD-FMK ( Santa Cruz Biotechnologies sc-3067 ) [47] , anthelmintic praziquantel ( Sigma P4668 ) [48] , and adenosine phosphates ATP ( Sigma A1852 ) , AMP-PNP ( non-hydrolyzable ATP analog; Sigma A2647 ) , and ADP ( Sigma A2754 ) [49 , 50] . These substances were either vortexed with a volume of water before treatment or added directly to water containing cercariae . Cercariae were treated within 3 hours of collection , and the time points expressed in this report refer to the time elapsed after the administration of a given treatment . We obtained a 637 amino acid protein sequence for S . mansoni Hsp70 from NCBI ( CCD76164 ) and used this sequence as a query ( NCBI BLASTp ) to identify homologous Hsp70 proteins from different organisms . Using the available sequences ( incomplete sequences were omitted ) , we performed an alignment using ClustalW2 to determine the phylogenetic relationship among these proteins ( S1 Fig ) . As expected , we found that SmHsp70 proteins are highly conserved across organisms with greater than 50% identity and that they cluster into different Hsp70 classes [51] . SmHsp70 ( NCBI accession CCD76164 , 637 amino acids ( aa ) ) clustered with the human Hsp70 ( NCBI accession NP_006588 , 646 aa ) , which is constitutively expressed and recognized as the heat shock protein 70 cognate ( Hsc70 ) protein . The second SmHsp70 protein ( NCBI accession CCD76236 , 648 aa ) represents a non-constitutive heat-inducible form of Hsp70 , and it clustered with HsHsp70 ( NCBI accession AAI12964 , 655 aa ) , also called heat shock protein 70 family A ( Hsp70 ) member 5 , which is localized to the lumen of the endoplasmic recticulum ( ER ) where it is thought to mediate protein trafficking of ER-derived proteins , thereby regulating protein signaling [52] . Previously , we published the observation that SmHsf protein is localized to the acetabular glands of schistosome cercariae [25] . SmHsf is a transcriptional activator of HSPs . While we do not think that Hsf1 can directly regulate the actions of its transcriptional targets in acetabular glands , we became interested in the idea that Hsf1 or HSPs may be involved in the transition between cercariae and schistosomula , either for cercarial invasion or for newly transformed schistosomula . We began by experimentally repeating observations of cercarial responses to human skin lipid that have been well established since the 1970s [26 , 53] . Our descriptions of cercariae are based on observations from inverted and upright microscopes . However , because cercariae continuously moved vertically in our 1 mL water samples , a consistent location to image between samples was not possible . Thus , images described here focus on the bottom of the culture wells , with approximately 1 , 000 cercariae per well for a 12-well culture plate or 500 cercariae per well for a 24-well culture plate . When observing cercariae by microscopy in a culture well , the relatively large depth of the water column and the nature of standard microscopes precludes a meaningful side-view visualization . Swimming cercariae , in wait of a host , are distributed vertically in a water column with few touching the bottom surface of a culture well . Thus , most cercariae will not be seen at the bottom of a culture well from this viewpoint . In contrast , when the cercariae have settled in response to a stimulus , many more cercariae can be observed at the bottom of a culture well ( Fig 1 ) . The apparent lack of cercariae in some of the images described later is not caused by a discrepancy in the number of cercariae added , but rather by their specific distribution ( vertical and horizontal ) in the water column . Since many drugs are often diluted or dissolved in DMSO , we established a baseline for cercarial DMSO tolerance , relative to what we observed in water . We compared cercariae treated with filtered water , 0 . 5% DMSO , and 1% DMSO . Cercariae treated with water and 0 . 5% DMSO were distributed in a similar manner and exhibited a similar swim ( up ) -sink-swim behavior at both 10 minutes and 2 hours ( S1 Video ) . Given the potential connection for a heat shock response during the cercaria-schistosomulum transformation and that Hsp70 is widely conserved , we compared the effect of treating cercariae with human skin lipid , linoleic acid , and PES ( Fig 2; S2 Video ) . PES has been shown to prevent Hsp70 from interacting with several Hsp70 client proteins [36] . Experimentally , cercariae respond to a skin lipid smear on the bottom of a petri dish by settling to the bottom of the petri dish and beginning the penetration process [26] . Our observations confirmed this . However , only cercariae located in close proximity to the skin lipid smear seemed to gather at the site where skin lipid was placed; the majority of the cercariae settled to the bottom of the well without regard for the location of the lipid smear . It should be noted that when cercariae were exposed to human skin lipid or linoleic acid ( mixed into water and added to cercariae ) , the cercarial honing response occurred within minutes . We also note that not all cercariae in our 1 mL sample responded to the skin lipid stimulus , as some cercariae could be seen swimming higher in the water column , out of the focal plane ( S2 Video ) ; this may correlate with the 60–70% cercarial response previously described in response to human lipids or L-arginine [9] . We next tested the effect of PES , a selective inhibitor of Hsp70 . When cercariae were exposed to PES , they initially behaved similarly to the 0 . 5% DMSO control treatment , whereas the cercariae exposed to skin lipid responded immediately and started settling to the bottom of the well and swimming into or crawling along the surface ( Fig 2D , 2G and 2J; S2 Video ) . However , we were surprised by the result just minutes later . After 5–10 minutes , PES ( 250 μM ) -treated cercariae began to swim to the bottom of the culture plate well , eventually losing their tails to transform into schistosomula ( Fig 2J , 2K and 2L ) . We observed the same effect with a lower treatment concentration ( 50 μM ) of PES but at a later time point ( S4 Video ) . While the majority of the cercariae treated with human skin lipid or linoleic acid honed downward , we observed that 100% of the PES-treated cercariae settled to the bottom of the well and began the penetration behavior ( S2 Video ) . When we co-treated cercariae with PES and skin lipid , the cercariae responded with an effect similar to that of PES alone: all of the cercariae were present at the bottom of the well ( S3 Video ) . We observed that after exposure to skin lipid ( 9 minutes ) or PES ( 51 minutes ) , the cercariae formed clusters ( Fig 2H and 2K ) ; this effect was not seen in the 0 . 5% DMSO control treatment ( 57 minutes , Fig 2E; S2 Video ) . Cercariae under PES treatment had not yet formed these clusters at 20 minutes ( S4 Video ) . A majority of the cercariae lost their tails by 1–3 hours in the PES treatment and 1 hour in the skin lipid treatment ( Fig 2L and 2I; S2 and S4 Videos ) ; again , this effect was not seen in the 0 . 5% DMSO control treatment ( 1 hour 56 minutes , Fig 2F; S4 Video ) . Within 3 hours , both PES and skin lipid-treated cercariae transformed into schistosomula . For PES-treated cercariae , it should be noted that this honing and transformation occurred in the absence of any host signaling molecules . Transformation involves several events , notably the loss of tails and loss of water tolerance . The flat appearance of the heads of the cercariae in the skin lipid- and linoleic acid-treated sample at 2 hours indicates the loss of water tolerance and lysis , and further progression in the transformation to the schistosomulum stage , as compared with the corresponding PES-treated sample , in which the heads have a round appearance and are motile ( S2 Video ) . We should also note that the timing for all events seemed to vary somewhat , albeit consistently between cercarial sheds . For example , in one cercarial shed , honing with skin lipids may begin within a minute , in another 3 minutes . To further determine whether the effect of PES is specific to Hsp70 , we treated cercariae with several different Hsp70 modulators , including MKT-077 , 115-7c , and VER-155008 . MKT-077 functions as an allosteric inhibitor of Hsp70 , binding within the nucleotide binding domain of Hsp70 next to its ATP/ADP binding pocket and inhibiting ATP turnover rate . MKT-077 is a rhodacyanine dye originally identified as an anti-tumor agent , and it has been shown to bind mortalin , an Hsp70 family member , and disrupt its interaction with p53 [43] . However , we found no obvious change in behavior of the cercariae in our MKT-077 treatments at 100 , 250 , or 500 μM concentrations ( Fig 3D ) , with the exception of increased death at 22 hours ( S5 Video ) . Note that the mechanism of action of MKT-077 differs from that of PES , which binds to the Hsp70 substrate binding domain and competitively blocks protein-protein interactions of Hsp70 and its client proteins . While most pharmacological agents target and inhibit the function of proteins , 115-7c has the unusual property of acting as an activator of Hsp70 protein folding function , leading to an enhanced rate of substrate refolding [44] . It binds to Hsp70 and promotes complex formation between Hsp70 and Hsp40 . In our treatments of cercariae with 115-7c , we observed the induction of honing behavior by 2 hours , especially in the 400 μM treatment ( Fig 3E ) ; by 22 hours , a majority of the cercariae had lost their tails ( S6 Video ) . VER-155008 at the concentration used ( 100 μM ) is insoluble in water , and it did not change the behavior of the cercariae ( Fig 3F; S7 Video ) . While there are numerous inhibitors of Hsp70 , most utilize a similar mechanism of action . For example , all of the following Hsp70 modulators inhibit Hsp70 nucleotide binding activity or ATPase activity: apoptozole , JG-98 , methylene blue , MKT-077 , VER-155008 , YM-01 , and YM-08 ( stressmarq . com ) . Since Hsp70 can work with other HSPs as a major effector of the heat shock response pathway , we asked whether another highly conserved HSP , Hsp90 , could be involved . We treated cercariae with the Hsp90 inhibitors geldanamycin and 17-DMAG , a water-soluble derivative of geldanamycin . However , treatment with these compounds did not produce a change in cercarial behavior; the cercariae resembled those treated with 1% DMSO ( Fig 4; S8 Video ) . Although PES is a potent inhibitor of Hsp70 , it was initially described in a screen to identify molecules that block p53-dependent transcriptional activation and apoptosis [54 , 55] . PES can also block cisplatin-induced p53 interaction with mitochondrial Bak , a pro-apoptotic molecule responsible for the permeabilization of the mitochondrial membrane , and which thereby blocks p53-dependent activation of apoptosis-associated caspases 8 and 3 [56] . However , it is thought that PES inhibition of p53 acts by inhibition of Hsp70 , as PES does not directly interact with p53 , BAK , BCL-xL , Grp78 , Hsc70 , or Hsp90 [36] . The molecular targets or mechanism for p53 regulation of apoptosis is unclear . To determine whether the apoptosis pathway is involved in the honing behavior of cercariae , we blocked caspase activity by treating cercariae with a pan-caspase inhibitor , Z-VAD-FMK . When cercariae were treated with Z-VAD-FMK , we found no change in cercarial honing behavior . Co-treatment with Z-VAD-FMK and PES resulted in a honing behavior similar to that of PES treatment alone ( S9 Video ) . As an additional treatment , we included praziquantel , the long-standing drug treatment for human schistosome infection . The efficacy of praziquantel treatment depends on the parasite stage for schistosomes; notably , while it can kill cercaria and adult stage schistosomes , it cannot kill the intermediate schistosomulum stage schistosomes [48 , 57] . Our treatment of cercariae with 300 nM praziquantel resulted in settling , similar to honing behavior; however , at 24 hours , we observed that while most of the cercariae had died , very few had lost their tails , in contrast to the PES treatment , which resulted in tail loss ( in addition to death ) for nearly all of the cercariae ( S10 Video ) . Functional roles for Hsp70 in the regulation of signal transduction through the binding of client proteins have been recently described and correlate with its intrinsic ATPase activity [29] . When Hsp70 is in an ADP bound state ( Hsp70ADP ) , Hsp70 interacts with its client protein stably and the Hsp70 lid is in a “closed” state , preventing release of the client protein . When in the ATP bound state ( Hsp70ATP ) , the Hsp70 lid is opened , allowing the release of the client protein and increasing the on/off rate at the substrate interaction domain [58] . We propose in the regulation of cercarial honing that Hsp70 binds a client protein and functionally inhibits the client protein’s ability to initiate cercarial honing ( Fig 5 ) . In accordance with this , if Hsp70 is critical to honing , then we predict that increasing the ATP concentration should cause the Hsp70 lid to open , leading to the release of the client protein and consequentially result in cercarial honing ( Fig 5 ) . To test this , we treated cercariae with ATP , AMP-PNP ( a non-hydrolyzable ATP analog ) , and ADP , each at a concentration of 5 mM ( intracellular ATP concentration in mammalian cells has been suspected to occur in the millimolar range [59] ) . While ATP and ADP treatments at this concentration did not show any difference compared to the water alone control treatment , AMP-PNP induced honing behavior within 2 hours 30 minutes ( Fig 6; S11 Video ) . Understanding the requirements for schistosome infection at the parasite-host interface can expedite the identification of novel targets for prevention of infection or the elimination of newly established infections . This has been observed using a topical skin treatment with inhibitors of the schistosome proteases used by the larval cercarial form , during invasion [60 , 61] . The process by which cercariae invade a mammalian host has been well described , but molecular requirements regulating this process are unknown . We present evidence that Hsp70 is involved in the process of cercarial honing and plays a role in a signal transduction pathway to regulate cercarial invasion behavior . Cercariae are released from their molluscan host and have less than 24 hours to find a mammalian host before depletion of their glycogen stores in their tail and body prevents their ability to penetrate host skin [4] . In search of a host , cercariae are distributed in the water column with minimal up and down motion , presumably lying in wait in what might be described as a “still hunting” mode . Cercariae swim randomly in response to water turbulence , light and shadows , and it is thought that they swim toward their host through gradients of body heat and skin chemicals , including linoleic acid , human skin lipid , and L-arginine , with the latter two being the most directionally significant [8 , 12 , 62] . This chemotactic process is modeled by placing cercariae in water and exposing them to a surface streaked with skin lipid as stimulus . We observed limited chemotaxis in our treatments of cercariae with skin lipid , such that only the cercariae in close proximity to the site where the skin lipid was placed made contact with the lipid . This suggests that cercariae do not swim toward the host over long distances , but lie in wait for a host that comes into close proximity and swim more actively to increase the chance of making contact with the host . Pharmacological targeting is one way to dissect the molecular pathways that may be involved in cercarial honing . Here , we used a selection of chemical compounds to query a role for Hsp70 , Hsp90 , and apoptosis in this honing behavior . Based on our observations , we propose that a heat shock pathway is specifically involved in cercarial honing for host invasion . HSPs have been identified in cercarial gland secretions [26 , 63] and are among the highest abundance transcripts identified in newly transformed schistosomula [3] . In fact , HSPs have been correlated with cercarial transformation since the late 1980s [64] . However , the role of HSPs has traditionally been connected with the stress response ( for review , [22–24] ) , correlating with the transition from cercaria to schistosomulum , which involves a temperature change from that of ambient water to 37°C host body temperature . Recent evidence in other systems has suggested that HSPs have more diverse functions outside of stress response , including roles in oogenesis and development , lifespan extension , regulation of cancer , fertility and viability [65–71] . In the schistosome molluscan host B . glabrata , the snail heat shock response is necessary for snail susceptibility to infection , such that a reduced heat shock response in the snail results in resistance to schistosome infection [27]; this suggests an important function in host HSP level for schistosome host invasion . Our observation that cercariae treated with PES undergo a behavioral change is novel , and it allows for the initial identification of molecular components involved in cercarial honing . Honing occurs in response to skin lipid [72]; however , since cercariae treated with the Hsp70 inhibitor PES show a similar behavior to cercariae treated with skin lipid , we predicted that Hsp70 plays a regulatory role in the signaling required for the honing behavior . Honing induced by PES is concentration-dependent with time , such that lower concentrations require more time for induction to occur ( S4 Video ) . Our treatment of cercariae with two other Hsp70 modulators , MKT-077 and 115-7c , resulted in different behaviors . MKT-077 treatment resulted in a lack of honing , similar to the water control treatment , while treatment with the Hsp70 activator 115-7c resulted in honing behavior , similar to the PES treatment . We propose that while all three Hsp70 modulators tested in this study bind to Hsp70 , only PES and 115-7c actively promote the release of its client protein , which can then function to initiate cercarial honing . Treatment with ATP and its non-hydrolyzable analog , AMP-PNP , could also cause Hsp70 to release its client protein by skewing the Hsp70ATP/ADP binding state distribution toward Hsp70ATP ( the state at which Hsp70 has low affinity for client proteins ) . Specifically , we model that in the uninduced honing state ( absence of lipid stimulus ) , Hsp70 interacts with and negatively regulates or inhibits the function of an Hsp70 client protein , which we call Hsp70 Honing Factor ( HHF ) . Upon upstream signal activation by skin lipid , Hsp70 releases HHF , which allows the activation of further signaling to trigger the honing behavior ( Fig 5 ) . This model is not in disagreement with current models describing a function for Hsp70 signaling [29] . A signaling pathway required to induce cercarial honing implies that many potential signaling factors could be involved , beginning from the receptor ( s ) that senses skin lipid , potential kinases or phosphatases , through Hsp70 , HHF and its targets . We reasoned that other HSPs , such as Hsp90 , could be involved , as Hsp90 is reported to interact with client proteins in signaling as well [73] . However , in our treatment of cercariae with Hsp90 inhibitors geldanamycin and 17-DMAG , a geldanamycin derivative , we did not observe any obvious change in the behavior of the cercariae . Next , we considered the potential involvement of apoptosis in honing induction . PES can block cisplatin-induced p53 activation of apoptosis [36] . Our treatment with the apoptosis inhibitor Z-VAD-FMK also did not result in any obvious change in the behavior of the cercariae . Interestingly , praziquantel treatment led to honing behavior similar to that resulting from PES treatment; however , at 24 hours , most of the cercariae had not lost their tails , indicating that transformation did not occur . In contrast , tail loss occurred for the skin lipid , linoleic acid , PES , and 115-7c treatments by 24 hours ( S2 Video ) . This observation leads us to speculate that cercarial honing involves specific signaling to cause the loss of tails ( transformation ) in addition to a change in the swimming pattern ( settling ) . Further effort will be necessary to identify the signaling components involved in cercarial honing , including the proposed Hsp70 client protein , HHF , and to better understand the relatively recently described role of Hsp70 in signaling [29] . Under ideal circumstances , genetics approaches such as gene knock-downs and knock-outs would be appropriate to identify honing components . However , these tools have not been thoroughly developed for use in developing or mature cercariae . Our group and others are working on developing methods to overcome these technical challenges [74–79] . Analysis in cercariae is challenging , as cercariae are short lived , transient , and the necessary proteins for swimming and host invasion have already been produced prior to exit from the snail host . Genetic manipulations of early developing cercariae within sporocysts may be possible , but in the case of Hsp70 and potentially other proteins , knock-down or knock-out could result in the loss of viability or production , not because of protein targeting problems , but because of the multipurpose nature of this particular protein . HSPs are the most abundant proteins expressed in the schistosome egg and miracidium [80] . However , a reduction of the hest shock response in the intermediate snail host , B . glabrata , makes the snail resistant to schistosome infection [27] , suggesting a critical role for the heat shock pathway for intermediate host susceptibility . Consequently , it would not be a far stretch to speculate whether inhibition of miracidial HSPs could affect invasion of the snail host . In this study , we have just pierced the surface and glimpsed at molecular components that contribute to cercarial honing . We have found no similar observation where Hsp70 signaling affects a whole organism and its behavior directly , leading to stimulating questions such as: how does signaling quickly and directly regulate cercarial behavior , and are there other organisms that are similarly regulated ? Additionally , schistosomiasis affects nearly 240 million people globally . Understanding the molecular requirements for cercarial honing and invasion , as well as those for early schistosomulum survival , could identify new potential drug targets and transition schistosome control from treatment to prevention .	Parasitic schistosome worms cause morbid disease in over 240 million individuals worldwide . Acute infections with these worms can lead to Katayama fever , while chronic infections can lead to portal hypertension , enlarged abdomen , and liver damage . The infective larval stage , called cercariae , are free-swimming and can detect , seek , and penetrate human skin to enter the human host circulatory system , eventually developing into egg-laying adult worms that cause schistosomiasis . Molecular pathways associated with the initial cercarial invasion of the host , however , are largely unknown , especially with respect to the parasite-specific signals involved in host detection and subsequent decision to invade . Here , we describe a role for Hsp70 in cercarial invasion behavior . To date , only generic stimulation with skin lipid , linoleic acid or L-arginine are known to induce cercarial invasion behavior; thus , we can begin an initial investigation of molecular requirements for host invasion and environment transition for schistosomes and possibly other parasitic organisms .	[ "Abstract", "Introduction", "Methods", "Results", "Discussion" ]	[ "skin", "schistosoma", "invertebrates", "medicine", "and", "health", "sciences", "cellular", "stress", "responses", "swimming", "integumentary", "system", "pathology", "and", "laboratory", "medicine", "helminths", "cell", "processes", "parasitic", "diseases", "biomechanic...	2016	Hsp70 May Be a Molecular Regulator of Schistosome Host Invasion
Identifying key reservoirs for zoonoses is crucial for understanding variation in incidence . Plague re-emerged in Mahajanga , Madagascar in the 1990s but there has been no confirmed case since 1999 . Here we combine ecological and genetic data , from during and after the epidemics , with experimental infections to examine the role of the shrew Suncus murinus in the plague epidemiological cycle . The predominance of S . murinus captures during the epidemics , their carriage of the flea vector and their infection with Yersinia pestis suggest they played an important role in the maintenance and transmission of plague . S . murinus exhibit a high but variable resistance to experimental Y . pestis infections , providing evidence of its ability to act as a maintenance host . Genetic analyses of the strains isolated from various hosts were consistent with two partially-linked transmission cycles , with plague persisting within the S . murinus population , occasionally spilling over into the rat and human populations . The recent isolation from a rat in Mahajanga of a Y . pestis strain genetically close to shrew strains obtained during the epidemics reinforces this hypothesis and suggests circulation of plague continues . The observed decline in S . murinus and Xenopsylla cheopis since the epidemics appears to have decreased the frequency of spillover events to the more susceptible rats , which act as a source of infection for humans . Although this may explain the lack of confirmed human cases in recent years , the current circulation of plague within the city highlights the continuing health threat . Plague , like other zoonoses , can exhibit large temporal variation in incidence at the same location , sometimes re-emerging after long periods of silence or apparently disappearing . Such variation in exposure may be due to changes in the presence or prevalence of the pathogen within the reservoir community or in the contact between humans and infected reservoirs ( or vectors ) . Identifying key reservoirs of a zoonotic disease is essential for understanding disease dynamics , as well as for the design of surveillance or control strategies . However , long-term data from reservoir populations are relatively rare , especially in resource poor settings . Yersinia pestis , the causative agent of plague , is usually transmitted to humans from rodents via the bite of infected fleas . Such bubonic plague cases can develop into pneumonic plague that may be transmitted human to human . Globally , most plague cases occur in Africa , with Madagascar , the country most seriously affected , reporting an average 400 human cases annually between 2010 and 2015 [1 , 2] . Plague arrived in Madagascar in 1898 at the port of Toamasina , subsequently spreading to other ports and then the Central Highlands from 1921 . Plague became endemic in areas above 800 meters [3] but apparently disappeared from coastal areas after 1928 . However , between August 1991 and April 1992 , plague reappeared after more than 60 years of silence in Mahajanga , a city on the north-west coast of Madagascar with 202 suspected cases including 41 probable and confirmed human cases [4–6] , after recolonization by a strain originating from the Central Highlands foci [7 , 8] . From 1995 to 1998 , annual epidemics occurred with a total of 1702 suspected and 297 confirmed cases [9 , 10] . The last confirmed human case occurred in November 1999 . Since then no new human cases were detected . From a public health viewpoint it is important to understand the role of different reservoir hosts in allowing plague to re-emerge in Mahajanga in the 1990s and whether plague persists today . In rural areas of the Central Highlands in Madagascar , where most human cases occur , the epidemiology of plague is relatively well understood , primarily involving the black rat Rattus rattus and two flea vectors Xenopsylla cheopis and Synopsyllus fonquerniei [11 , 12] . Black rats in these areas have acquired resistance [13 , 14] and this may play an important role in persistence as the co-existence of resistant and susceptible hosts is often thought to play an essential role in the maintenance of plague [15] . In Mahajanga , plague epidemiology is poorly understood . The human plague season in this coastal focus is July to November while in Central Highlands foci it is September to April [16] . Preliminary studies found that the shrew Suncus murinus dominated the small mammal population with some infected by the plague bacillus and carriers of X . cheopis [6 , 17] . S . murinus is a small mammal belonging to the family Soricidae , order Soricomorpha . Originating from India , it is often commensal with man , living in and around houses , and has colonized South East Asia , the Pacific islands and the coasts of the Indian ocean [18 , 19] and was introduced to Madagascar in the 11th to 14th centuries [20 , 21] . The role of shrews as a reservoir of plague has been alluded to in Vietnam [22 , 23] but it remains hypothetical . Integrating data from different approaches can significantly strengthen understanding of complex multihost systems . Using data and samples collected during and immediately after the epidemics in 1990s and recently ( 2011–2014 ) , this study examines the role of S . murinus in maintaining the plague focus and as a source of the human epidemics by ( i ) comparing seasonal and inter-annual changes in reservoir host and flea abundance and seroprevalence of antibodies to plague amongst reservoir hosts; ( ii ) comparing sensitivity of current populations of potential reservoir hosts to two Y . pestis strains isolated from R . norvegicus and S . murinus during the epidemics; and ( iii ) conducting genetic analyses on whole genome sequences of Y . pestis isolates from Mahajanga . The study has been conducted in accordance with the Institut Pasteur guidelines ( http://www . pasteur . fr/en/file/2626/download ? token=YgOq4QW7 ) for animal husbandry and experiments . All experiments were performed at Biosafety level 2+ and verbal informed consent was obtained for sampling rats and shrews in the household . Trapping of small mammals was conducted in August 1991 and June to November 1995 in central Mahajanga ( Abattoir and Aranta ) ( Fig 1 ) , close to Marolaka market where the original re-emergence occurred in 1991 and where cases occured each epidemic season . From 1997–2001 and 2011–2014 datasets have been obtained with the same trapping regime ( 30 houses with one wire-mesh trap and one Sherman trap in each ) in two areas: central and suburbs , Tsararano ( Tsararano ambony and Tsararano ambany ) that also had plague cases in 1995–1999 ( Fig 1 ) . Traps were set for three consecutive nights either inside houses or in the immediate surroundings , baited with dried fish and onion each afternoon and checked every morning . Verbal informed consent was obtained for sampling rats in the household . As the last confirmed human case was November 1999 , we classified trapping sessions into 3 periods: epidemic ( quarterly from May 1997 to May 1999 ) ; post-epidemic ( biannually from November 1999 to November 2001 ) and recent ( biannually from November 2011 to November 2014 ) . The Abattoir site was not sampled in November 2011 . The periods of human plague outbreaks and animal trapping are schematized in Fig 2 . For each captured animal , fleas were collected . Proportion of animals infested and flea index ( number of fleas per animal ) were calculated for each host species-site-trapping session combination . Blood samples were collected on seropads ( LDA 22440 , Zoopole–Ploufragan ) . For dead animals , spleen samples were removed and stored in Cary Blair transport medium . In May and November 2012 , live sub adult and adult individuals ( weighing between 30 to 75g for shrews and 70 to 200g for rats ) were brought to the laboratory for experimental infection . Blood samples were collected from the tail tip . Other captured animals were euthanized by cervical dislocation and necropsied . Two diagnostics tests were used: ( i ) culture on selective Cefsulodin Irgasan Novobiocin ( CIN ) agar for isolation of Y . pestis from spleen followed by identification with API 20E and phage lysis test [24 , 25]; ( ii ) enzyme-linked immunosorbent assay ( ELISA ) to detect specific anti-F1 IgG antibodies against Y . pestis in blood as previously described for mice and rats ( with an optical density threshold of 0 . 05 ) [26–28] . For shrews , protein A-peroxidase diluted 1/10 , 000 was used as a secondary antibody , with a threshold of 0 . 1 . Plates were read at 492 nm using Lab system Multiscan . Shrews and rats brought to the laboratory were quarantined for at least 15 days before the experiment and all animals used for infection were negative for anti-F1 antibody . During the experiment , animals were housed individually in cages with filtered enclosures at ambient room temperature with food and water ad libitum . Two Y . pestis strains isolated during the 1998 epidemic in Mahajanga were used in the experiment: 163/98Sm , isolated from a S . murinus and 164/98Rn , isolated from a R . norvegicus . Virulence of the two strains was determined by measuring the lethality of OF1 mice infected subcutaneously with 100 colony-forming units ( CFU ) of Y . pestis . Based on previous studies on wild rats from Madagascar , two different doses of Y . pestis were used: 102 CFU ( LD50 previously reported for rats from the plague free zone in Madagascar ) and 105 CFU ( LD50 for rats from the central highland plague focus ) [13 , 14] . For each bacterial dose , 12 S . murinus , 10 R . rattus and 10 R . norvegicus were infected subcutaneously . The concentration of Y . pestis administered was estimated by measuring the optical density at 600 nm ( LKB Biochrom , Ultrospec PLUS ) and confirmed by colony enumeration of ten-fold dilutions on selective CIN agar plates . Animals were monitored twice daily for 21 days . A rapid diagnostic test ( RDT ) was carried out on the spleen of dead rats to detect the presence of Y . pestis F1 antigen [29] in order to confirm that death was due to plague . Blood was collected on seropads at 0 , 2 , 4 , 7 , 14 and 21 days post infection to allow detection of anti-F1 IgG antibodies as described above . At the end of the experiment , rats were euthanized by CO2 asphyxiation and necropsied . Whole genome sequencing of Y . pestis strains ( Table 1 ) was performed using the NEXTflex PCR-Free DNA-Seq kit for Illumina ( Bio Scientific ) , and a HiSeq2000 machine ( Illumina , San Diego , CA ) to yield paired-end reads of 100 bases . Image analysis , base calling , and error estimation were performed using Illumina Analysis Pipeline version 1 . 8 . Paired-end Fastq files were uploaded into the Yersinia database available in the Enterobase website ( http://enterobase . warwick . ac . uk/ ) . De novo assembly of the genomes and SNP calling against the CO92 reference genome ( after correction according to Morelli et al . , 2010 ) [30] were performed automatically using EnteroTools in the Enterobase website . Generalized linear models ( GLM ) were used to examine changes in the abundance of S . murinus , R . norvegicus and fleas ( poisson errors and a log link ) and the proportion of animals carrying fleas ( binomial errors and a logit link ) . For models of flea abundance , the natural log of the number of host was used as an offset ( these models analyze the flea index described above ) . Explanatory variables included month , spatial area and a variable to examine long-term changes . Two global models were considered , one considered a linear time trend , whilst the second included time period ( epidemic , post-epidemic , recent ) . Two-way interactions between site and month and site and the long-term change variable were included . Initial choice between global models was based on Akaike Information Criterion ( AIC ) [31] . Where there was evidence of overdispersion we subsequently used quasipoisson , negative binomial or quasibinomial models , with model selection was based on the F test for “quasi” models and AIC for negative binomial models , with the most parsimonious model within 2 of the model with the lowest AIC selected . Due to lower power , seroprevalence data were analyzed using chi-square tests . To remove any effects of season , tests of long-term changes only used May and November data . All analyses were performed using R software [32] . Four species of small mammals were caught during trapping: S . murinus , R . norvegicus , Mus musculus and R . rattus ( Table 2 ) . S . murinus was systematically the most abundant small mammal captured over the entire study period followed by R . norvegicus . R . rattus were rare in 1991 and 1995 and then absent from the central areas . M . musculus were rare all along the study , except a slight increase recently in suburbs . All fleas collected were X . cheopis and fleas were significantly more abundant on R . norvegicus than S . murinus ( p<0 . 001 ) ( S1 Table ) . GLM analyses of host and flea abundance indicated significant long-term , as well as seasonal , changes , whilst site differences were limited to a moderate increase in flea abundance on S . murinus in suburbs ( Table 2 , S2 Table ) . For S . murinus abundance , comparisons of the initial two global models indicated that time period ( AIC = 318 . 08 ) provided a better fit to changes in S . murinus abundance than a linear trend ( AIC = 341 . 87 ) . For all other datasets , a linear trend was the most parsimonious model . Relative abundance of S . murinus abundance was significantly lower in both the post-epidemic and recent periods , compared with the epidemic period , whilst R . norvegicus has shown a moderate increase in abundance over time ( Fig 3A , S3 Table ) . S . murinus abundance appears to peak in May , whilst R . norvegicus does not show strong seasonality ( S3 Table ) . For S . murinus , both the proportion infested with fleas and flea index showed a decline over time , whilst there was no evidence of long-term changes in flea abundance on R . norvegicus ( S2 Table ) . However , flea index of S . murinus and R . norvegicus have tended to increase in 2014 ( Fig 3B ) . In terms of seasonality , the flea index of R . norvegicus , the proportion of S . murinus carrying fleas and the S . murinus flea index peaked in August ( S3 Table ) . Sixteen strains of Y . pestis strains were isolated during human plague epidemics ( 1991 , 1995 , 1997 , 1998 and 1999 ) from small mammals: 9 from S . murinus , 6 from R . norvegicus and 1 from R . rattus ( S1 Table ) . In 1998 , fleas collected from 2 R . norvegicus and 1 S . murinus were infected with Y . pestis . In 2014 , 15 years after the last confirmed human case , one strain was isolated from R . norvegicus caught close to the market of Marolaka , the place where the plague outbreak started in 1991 . Among seropositive animals , we have found that antibodies against plague were detectable in shrews throughout the entire period of study . Seroprevalence in S . murinus was higher in May ( 12 . 1% , 55/456 ) and August ( 12 . 0% , 15/125 ) , than November ( 6 . 3% , 18/284 ) and February ( 2 . 6% , 3/114 ) ( χ23 = 12 . 76 , p<0 . 01 ) . There was no evidence that seroprevalence in S . murinus changed between time periods . In contrast , seroprevalence in R . norvegicus was higher during the epidemic period ( 16 . 1% , 13/81 ) than in the post-epidemic ( 0% , 0/75 ) or recent periods ( 2 . 3% , 4/172 ) ( Fisher’s test: p<0 . 001 ) ( S1 Table ) . Upon subcutaneous infection of OF1 mice with 100 CFU of either 163/98Sm ( isolated from S . murinus ) or 164/98Rn ( isolated from R . norvegicus ) , no significant differences in terms of lethality were observed between the two strains: 6/6 and 5/6 mice died after infection with 164/98Rn and 163/98Sm , respectively . S . murinus was by far the most resistant species since no animals died after infection with 100 CFU , and most animals ( ≥75% ) survived an infection with a high dose of 105 CFU . R . rattus was highly susceptible to Y . pestis , even at a low dose of 100 CFU , while R . norvegicus exhibited intermediate levels of resistance ( Fig 4 ) . No major difference in lethality caused by the two strains was observed ( p = 0 . 706 ) . Anti F1 antibodies were detected in both shrews and rats from day 7 following the infection . To determine whether there was a possible exchange of strains between shrews and other infected hosts and vectors ( humans , fleas and rats ) during plague epidemics in Mahajanga , we evaluated their genetic relatedness . A SNP analysis was performed on the genomes of 6 Y . pestis strains isolated from S . murinus and 12 strains isolated during the same periods from humans , fleas and rats ( Table 1 ) . The resulting minimum spanning tree shows that the 7 Y . pestis strains from the 1995 epidemic were identical to each other , although they were isolated from different hosts ( Fig 5 ) . Five of the 7 strains isolated in 1998 from rats , shrews , fleas and humans were also identical , and they differed by a single SNP from the strains of the 1995 outbreak . However , 4 strains isolated from shrews in 1997–1998 differed from the others by a much higher number of SNPs and were also more distant among themselves ( 6 to 33 SNPs ) ( Fig 5 ) . Since SNPs accumulate with bacterial divisions , this may be indicative of a very active multiplication of Y . pestis in shrews or a longer persistence in this animal population , leading to a higher bacterial diversification . The Y . pestis strain ( 89/14 ) isolated from a rat in Mahajanga in 2014 branched with the separate shrew branch and was most closely related to a shrew strain ( 103/97 ) , although this strain was isolated 17 years earlier ( Fig 5 ) . Our study investigated the seasonal abundance of reservoirs and vectors for plague in Mahajanga ( the only coastal foci in Madagascar ) , the dynamics of the antibody response in shrews and rats , the susceptibility of shrews to plague and the genetic relationship between Y . pestis strains isolated from humans , rats , shrews and fleas . Integrating data from the different approaches used in our study provide strong evidence that S . murinus , a non-rodent host , was the most important reservoir of plague in the port of Mahajanga during epidemics in the 1990s , and since that date may have functioned as the maintenance host for plague within the city , despite a lack of confirmed human cases . A reservoir can be defined as one or more epidemiologically connected populations which together achieve the critical community size and can , therefore , permanently maintain a pathogen and transmit the infection to the defined target population [33 , 34] . Thus , a species should be regarded as an important reservoir of plague if it is essential to maintenance of Y . pestis within a community of small mammals and/or if it acts as a source of infective fleas that pose a risk to humans . The predominance of S . murinus in traps during the epidemics in Mahajanga , their carriage of known flea vectors , and the isolation of Y . pestis strains from some of them suggest they played an important role in plague transmission and maintenance . Moreover , the close genetic relatedness among Y . pestis strains isolated from humans , the usual hosts and vector ( rats , fleas ) and from shrews , is indicative of an active circulation of Y . pestis strains between all these sympatric populations during outbreaks . Despite the lack of human confirmed cases , the isolation from a rat in 2014 of a Y . pestis strain closely related to a strain isolated from a shrew in 1997 and the lack of long-term change in seroprevalence in shrews strongly argue for the maintenance of the plague bacillus in the shrew population over nearly 20 years . The findings from the experimental infection experiment are consistent with the ability of S . murinus to act as a maintenance host . The co-existence of susceptible and resistant hosts is thought to be important for persistence of plague within endemic foci [15 , 35] , with susceptible hosts developing the high bacteremia needed to reliably infect fleas [36 , 37] , and resistant hosts allowing the long-term maintenance of host and flea populations . In the desert plague foci of Central Asia , the great gerbil , with its high but variable rates of resistance , is thought to be the primary host [35 , 38] . Our results demonstrate that S . murinus exhibits a similar pattern , with high resistance overall but heterogeneous responses to infection at the individual level . Alternatively , some authors have suggested that individual hosts that survive infection could become chronically infected [39] , potentially acting as a source of infection at a later date , possibly when declines in immunocompetence associated with ageing or coinfections reduce the ability of the host to control the infection . However , evidence for this is limited , and in laboratory experiments such chronic infections were caused by isolates lacking the ability to express the F1 antigen on their surface [39] , an antigen that prevents phagocytosis of infected macrophages [40] and is thought to be essential for achieving the high bacteraemia necessary for transmission to fleas [41] . Thus , we believe heterogeneous responses , evidenced by our plague challenge experiments , in S . murinus are a more likely explanation . The presence of genetically divergent strains of Y . pestis from shrews provide additional evidence that plague persists within S . murinus , but occasionally spills over into the rat and human populations through partially-linked transmission cycles . The seasonal patterns of host-vector abundance , seroprevalence and human cases are also consistent with the spillover of infection from S . murinus . Host abundance above a threshold has been shown to be critical for plague epizootics in other systems , such as within great gerbil ( Rhombomys opimus ) populations in Central Asia [42 , 43] . In Mahajanga , the increase in S . murinus abundance around May appears to lead to increased transmission ( S . murinus seroprevalence peaks in May and August ) ; whilst the increase in X . cheopis in August-November , when S . murinus have declined , suggest that this is the period that infected fleas are most likely to be seeking other hosts ( rats ) , with the higher sensitivity of rats leading to more infected fleas that pose a risk to humans . In the Central Highlands in Madagascar , the start of the plague season in humans also coincides with a peak in flea abundance and a trough in host abundance [11] . Altogether then , our data point at shrews as a plague reservoir maintenance host , and suggest that this population may be at the origin of occasional resurgence of plague in Mahajanga . Although epidemiological data from Vietnam has previously suggested they may play a role in plague transmission along with rats [22 , 23] , in Mahajanga they appear to be critical for persistence , with rats primarily acting as a “source population” for human infections . A similar epidemiological cycle of a relatively resistant reservoir allowing plague persistence and a second species facilitating transmission to humans was proposed for Java [44] . To fully understand the role of S . murinus in persistence , experiments are needed to establish the relative ability of susceptible and resistant individuals to infect fleas . Our results also indicate that changes in the host-vector community may explain the lack of seasonal outbreaks in human cases since 1999 , with a significant decrease in S . murinus abundance that appeared to primarily occur just after the outbreaks in the 1990s . The observed decline in S . murinus and X . cheopis since the epidemics appears to have decreased the frequency of spillover events to the more susceptible rats , which act as a source of infection for humans . However , based on seroprevalence data , this has not reduced plague transmission within S . murinus populations but the significant decline in seroprevalence in R . norvegicus after the outbreaks in humans . The lesser abundance of R . rattus can be explained by a lower resistance to Y . pestis but also by competition with the larger R . norvegicus ( the sewer rat ) strongly favored by urbanization of the central districts . So today , R . rattus is only present in the suburbs with an important tree and shrub cover , a habitat more suitable for this arboreal species . However , they live in very close proximity to humans . In the period post epidemics , the abundance of R . norvegicus increases while S . murinus decreases: this suggest also competition between these two species , sharing the same subterranean and moist habitat and living greatly on the same food resources . X . cheopis is the only effective vector in Mahajanga , consistent with evidence that the endemic flea S . fonquerniei is limited to the Central Highlands [16] . Although an experimental study indicated that X . cheopis live longer at higher relative humidity [45] , the abundance of fleas on shrews presents a marked increase in August , the driest month in Mahajanga . This may provide an explanation for the discrepancy observed in plague seasons between Mahajanga and the main central foci in Madagascar . The optimal relative humidity for fleas may not occur at the same time in the two areas . In conclusion , our results indicate that S . murinus acts as a maintenance host for Y . pestis in Mahajanga as well as a source of infection for other hosts . Although changes in the abundance of reservoirs since 1999 may have decreased the probability of infections in the human population , the confirmation that plague still circulates within the city and can spillover into other , more sensitive hosts highlights the continuing health threat . Given the spread of S . murinus to other countries , especially in port cities , they could also act as important reservoirs if plague is inadvertently introduced elsewhere . Our study also emphasizes the importance of collecting longitudinal data and integrating information from different approaches to elucidate the relative role of potential reservoir hosts of zoonoses .	The reemergence of plague is related to the persistence and dynamics of its reservoirs and vectors . During the human plague outbreaks that re-occurred in Mahajanga harbor , Madagascar in the 1990s , the shrew Suncus murinus was found infected with Yersinia pestis . Combining field surveys , experimental infections and genetic analysis , we examined the role of Asian shrew Suncus murinus in plague transmission and maintenance comparatively with others potential hosts in Mahajanga . The genomes of nineteen Y . pestis isolates recovered from humans , shrews , rats and fleas in this focus were sequenced and compared . We observed the predominance of S . murinus captured during the epidemics and their carriage of the flea vector . Shrews exhibited high resistance to Y . pestis experimental infections . Genetic analyses of the strains isolated from various hosts were consistent with two partially-linked transmission cycles , with plague persisting within the S . murinus population , occasionally spilling over into the rat and human populations . The isolation of a Y . pestis strain from a rat in Mahajanga in 2014 genetically close to shrew strains reinforces this hypothesis . The current circulation of plague within the city highlights the continuing health threat .	[ "Abstract", "Introduction", "Methods", "Results", "Discussion" ]	[ "invertebrates", "medicine", "and", "health", "sciences", "plagues", "immune", "physiology", "pathology", "and", "laboratory", "medicine", "shrews", "spleen", "pathogens", "geographical", "locations", "microbiology", "vertebrates", "animals", "mammals", "bacterial", "dise...	2017	The Asian house shrew Suncus murinus as a reservoir and source of human outbreaks of plague in Madagascar
Local weather influences the transmission of the dengue virus . Most studies analyzing the relationship between dengue and climate are based on relatively coarse aggregate measures such as mean temperature . Here , we include both mean temperature and daily fluctuations in temperature in modelling dengue transmission in Dhaka , the capital of Bangladesh . We used a negative binomial generalized linear model , adjusted for rainfall , anomalies in sea surface temperature ( an index for El Niño-Southern Oscillation ) , population density , the number of dengue cases in the previous month , and the long term temporal trend in dengue incidence . In addition to the significant associations of mean temperature and temperature fluctuation with dengue incidence , we found interaction of mean and temperature fluctuation significantly influences disease transmission at a lag of one month . High mean temperature with low fluctuation increases dengue incidence one month later . Besides temperature , dengue incidence was also influenced by sea surface temperature anomalies in the current and previous month , presumably as a consequence of concomitant anomalies in the annual rainfall cycle . Population density exerted a significant positive influence on dengue incidence indicating increasing risk of dengue in over-populated Dhaka . Understanding these complex relationships between climate , population , and dengue incidence will help inform outbreak prediction and control . Dengue virus ( DENV ) [1] transmission occurs in more than 100 countries; however , the burden of dengue is not evenly distributed . Approximately half of the global population at risk of acquiring dengue infection resides in the South-East Asia Region of the World Health Organization [2] , a region characterized by strong seasonal weather variation and heavy monsoon rainfall . This reflects the influence of local weather , particularly temperature and rainfall , on the transmission of DENV by Aedes mosquitoes . Higher temperature , for example , shortens mosquito development time [3] , increases the frequency of blood feeding presumably by decreasing body size [4 , 5] , and reduces the extrinsic incubation period of DENV within mosquitoes [6] . However , transmission of DENV is influenced not only by average temperature , but also by diurnal temperature range ( DTR , the difference between daily maximum and minimum temperature ) . Temperature-dependent empirical and mathematical experiments show that temperature fluctuation influences vectorial capacity of Aedes aegypti , the principal mosquito vector of DENV , via biting rate , DENV transmission probability , extrinsic incubation period , and vector mortality rate [7–10] . At high mean temperatures , vectorial capacity increases with narrow daily temperature variation [7–9] . At low mean temperatures , the relationship between DTR and vectorial capacity is reversed [7–10] . Temperatures above 30°C reduce survival of adult Ae . aegypti [11] as does either very low or very high rainfall [12] . The positive relationship between rainfall and dengue incidence has been observed in several locations [13–15] . Seemingly paradoxical is the observation that the incidence of dengue increases in the dry season in some locations [16] . Large scale climatic events , such as the Southern Oscillation , resulting from the interplay of large scale ocean and atmospheric circulation processes in the equatorial Pacific Ocean have been identified as a remote driver of inter-annual weather variability across the globe . The warm and cold phases of the Southern Oscillation , El Niño and La Niña , respectively , are known to influence local temperature and rainfall and hence year-to-year variations in dengue incidence [13 , 17 , 18] . Socio-demographic and economic factors also influence dengue incidence . While the population at risk of dengue is likely to rise with increasing population , economic development would be expected to reduce risk [19] . Bangladesh , a member country of the World Health Organization South-East Asia Region experienced its first epidemic of dengue fever in 2000 after more than three decades of sporadic dengue [20] . Dengue is highly seasonal in Bangladesh with increased incidence during the monsoon . From 2000 to 2009 , cases have been reported from 29 of the 64 Bangladeshi districts , with around 91 . 0% from the capital , Dhaka [21] . Since 2010 very few cases have been notified from districts other than Dhaka [21] presumably because of a change in reporting criteria requiring confirmatory laboratory diagnosis . Studies of dengue in Bangladesh before ours have not considered daily temperature variation [22 , 23] . We present an analysis of the influence of daily temperature variation on the transmission of dengue adjusted for rainfall and population density , using a monthly dengue case time-series over 10 years from Dhaka . We also considered anomalies in sea surface temperature ( SSTA ) , an index for El Niño-Southern Oscillation ( ENSO ) that is associated with extreme weather in Bangladesh and has not been included in other studies . Analyses such as ours are critical for understanding the associations between weather , population , and dengue incidence and will allow the development of a reliable dengue early warning system . The study was approved by The Australian National University Human Research Ethics Committee . The national surveillance data of dengue fever cases was anonymized . Dhaka district , comprising Dhaka Metropolitan area ( DMA ) and adjacent sub-districts , is a 1 , 464 km2 area near the center of Bangladesh . Of the 64 districts this is the most densely populated , currently with 8 , 229 people per square kilometer . Over the years 2001 to 2011 , there was a 41 . 0% increase in the population density of Dhaka [24] . More than 37 . 0% of the population of DMA live in slums with a population density of 220 , 246 people per square kilometer [25] . Slums have no access to piped water and temporary containers like drums and earthen jars are commonly used to store water in which Ae . aegypti lays eggs [26] . Inadequate supplies of piped water and an absence of proper waste management in most locations of Dhaka result in abundant indoor and outdoor mosquito breeding sites . Both Ae . aegypti and Aedes albopictus , the latter a secondary vector of dengue , were observed in Dhaka during the 2000 epidemic [27] . Unscreened doors and windows permit mosquito entry to dwellings . Dhaka has a hot and humid tropical climate , with an average temperature of approximately 25°C , which nearly always permits mosquito development and DENV transmission . Rainfall is highly seasonal , with the wettest period ( June to September ) occurring during the warmest months . About 80 . 0% of the annual rainfall of 2 , 000 mm falls during the monsoon . Rainfall in Bangladesh is partly influenced by the Southern Oscillation with El Niño years usually associated with less than average monsoon rainfall while the opposite has been observed in La Niña years . However , the influence of the Southern Oscillation on monsoon rainfall is not linear and is inconsistent , as observed in the moderate El Niño years causing flooding while some La Niña events during the monsoon preceded by El Niño are associated with reduced monsoon rainfall in Bangladesh [28 , 29] . Monthly dengue cases for Dhaka district , from January 2000 to December 2009 , were obtained from the Directorate General of Health Services . This time period was chosen to avoid the influence of the change in reporting practice started in 2010 . The daily maximum , minimum , and mean temperatures ( °C ) , relative humidity ( % ) , and rainfall ( mm ) data for Dhaka were collected from the Bangladesh Meteorological Department . A single missing value for maximum temperature was replaced by linear interpolation . Diurnal temperature range was derived as the difference between maximum and minimum daily temperature . Monthly means of these climatic variables were calculated from the daily records . A monthly time series of SSTA over the Niño 3 . 4 region was obtained from the United States National Oceanic and Atmospheric Administration Climate Prediction Center ( http://www . cpc . ncep . noaa . gov/data/indices/ersst3b . nino . mth . 81-10 . ascii ) . The Niño 3 . 4 index was used because of its correlation with Indian Ocean region monsoon rainfall . An increase ( decrease ) of >0 . 5°C ( <-0 . 5°C ) in three-month moving average of SSTA is referred to as an El Niño ( a La Niña ) event . Population estimates were extracted from the 1991 , 2001 , and 2011 census data ( there was no census taken between these years ) of the Bangladesh Bureau of Statistics . Linear interpolation was used to calculate the monthly population for each of the years between 2000 and 2009 . The population density ( people/km2 ) for Dhaka was estimated by dividing the district population size by the area ( km2 ) . To examine temporal patterns over the study period , monthly dengue cases and climatic averages were plotted over the 10-year period . To display seasonal patterns , monthly averages of mean temperature , DTR , and rainfall , and monthly numbers of total dengue cases over the 10 years were aggregated and plotted by month . Overall correlation between dengue cases and climatic variables ( mean monthly temperature , mean monthly DTR , mean monthly relative humidity , mean monthly rainfall , and monthly SSTA ) were examined using Spearman's rank correlation test . To avoid multicolinearity arising from correlated variables , the final set of candidate variables was restricted to those with pair-wise correlations of ≤0 . 8 . Cross-correlation functions of dengue cases with each of the climatic variables were then estimated to investigate their lagged effects on dengue incidence ( p≤0 . 05 ) . Time lags were included to account for the influence of climatic variables on the development , maturation , and survival of the vector ( Aedes mosquitoes ) as well as the extrinsic incubation period of DENV in the vector and the intrinsic incubation period in the human host . Lags of up to three months were considered for all weather variables , with SSTA also considered at a lag of four months . The counts of dengue cases were then fitted by a generalized linear model ( GLM ) with negative binomial distribution to allow for overdispersion in dengue counts . The population of Dhaka was added as an offset to the model on a logarithmic scale to adjust for population size . Population density was also included in the model to account for the potential influences of associated socio-demographic changes on dengue transmission in Dhaka . An indicator variable for outbreak months was added to prevent occasional extreme values from distorting the analyses . A month with the number of dengue cases exceeding the 10-year mean plus two standard deviations was defined as an outbreak month . To account for the long term trend in dengue incidence over time , an indicator variable for year was incorporated in the model . An autoregressive term at order 1 was also included to allow for autocorrelation in monthly numbers of dengue cases . To determine whether seasonal variation had any influence on dengue incidence , a categorical variable for winter ( December–February ) , pre-monsoon ( March–May ) , monsoon ( June–September ) , and post-monsoon ( October–November ) was also considered . The analyses were performed using STATA 13 . 1 ( StataCorp . , Texas , USA ) and figures were drawn using RStudio ( R development Core Team , 2015 ) . Inter-annual and seasonal variations for dengue and weather over the period 2000–2009 are presented ( Figs 1 and 2 ) . The number of dengue cases during winter is low and starts to increase from June ( Fig 2 ) with the advent of the monsoon with considerable annual variation ( Fig 1 ) . The peak comes one month after the initial rainfall peak in July and starts declining afterwards . Temperature reaches its peak in April and plateaus until October when it drops ( Fig 2 ) . Because of the high correlation with mean temperature and DTR , relative humidity was excluded at the initial stage of model formulation . Consideration of both temperature and rainfall was , however , expected to minimize the potential confounding effect of relative humidity on dengue incidence . The categorical variable for season was also subsequently removed because it did not improve model fit . Therefore , the model finally fitted is as follows: yt~NegBin ( μt , θ ) log ( μt ) =α+∑j=03β1jTjt+∑j=03β2jDTRjt+∑j=03β3j ( Tjt×DTRjt ) +∑j=03β4jRjt+∑j=04β5jSSTAjt+β6Popdent+outbreak+year+yt−1+log ( Population ) +εt ( 1 ) where yt is the dengue count in Dhaka in month t ( t = 1 , … , 120 ) ; μt is the corresponding mean dengue count; T , DTR , R , and SSTA are the mean monthly temperature ( °C ) , mean monthly diurnal temperature range ( °C ) , mean monthly rainfall ( mm ) , and monthly sea surface temperature anomaly respectively; ( T×DTR ) represents the interaction between mean monthly temperature and mean monthly DTR; j = 0 , … , 4 represent the time lag periods in months; outbreak is the categorical variable for outbreak months; year represents time trend; yt-1 is the dengue count of previous month; and εt is the error term . Table 1 shows estimates of the significant covariates from model ( 1 ) . Mean temperature , DTR , and the interaction between these two variables are all significant predictors of dengue incidence at a lag of one month . However , the opposing directions of main and interaction effects indicate a negative synergy between mean temperature and DTR . Therefore , dengue incidence increases with higher temperature and lower DTR or lower temperature and higher DTR in the previous month but decreases when both are either high or low . Rainfall at lag one and two months was found to be positively associated with dengue incidence , suggesting that increased incidence of dengue in a given month is associated with higher rainfall during the previous two months . The negative effect of SSTA on dengue incidence at lag zero month indicates that the incidence goes up with increasing negative values of the SSTA in the current month , while the inverse relationship was observed at lag of one month . Increasing population density , as anticipated , increases dengue incidence . To investigate how SSTA influences climatic anomalies in Dhaka , standardized anomalies of temperature , relative humidity ( S1 Fig ) , and rainfall were calculated and plotted with SSTA over the study period ( Fig 3 ) . Simple linear regression of temperature , relative humidity ( S1 Fig ) , and rainfall anomaly on SSTA at lag of zero and one month revealed a weak negative correlation between rainfall and SSTA ( Fig 4 ) even though the relationship is not temporally consistent ( Fig 3 ) presumably due to a non-linear relationship between them . Model diagnostics were performed as follows . Firstly , a model was run without the interaction terms and compared with model ( 1 ) . The likelihood ratio test confirmed that the addition of interaction terms resulted in a significantly improved fit compared to the model without interactions ( p<0 . 000 ) . The Pearson dispersion statistic ( 0 . 98 ) also provided evidence for the goodness-of-fit of the model ( 1 ) . Secondly , residual analyses were performed to ensure that the model provided an adequate fit to the data . Serial autocorrelation of the residuals was checked by examining a time plot and a partial autocorrelation plot of the residuals ( S2 and S3 Figs ) . In addition , observed vs fitted plot of dengue cases was examined ( S4 Fig ) . It is well established that temperature influences vector and virus biology and therefore dengue transmission . Monthly changes in average temperature have been reported to be positively associated with dengue transmission in Puerto Rico [30] . In addition to average temperature , temperature fluctuations also have an impact . Large fluctuation around warmer temperature reduces transmission whereas around cooler temperature this speeds up the process and vice versa [7 , 8 , 10] . However , studies of climate and dengue usually ignore diurnal temperature variation . We found that dengue incidence in Dhaka was significantly influenced by mean temperature , DTR , and their synergistic effect , after adjusting for rainfall , anomalies in sea surface temperature , population density , autoregression and the long term temporal trend in dengue incidence . Although mean temperature and DTR were positively associated with dengue incidence , the opposing direction of their interaction term suggested a negative synergy between these two variables . This indicates that although increased mean temperature and reduced DTR or reduced mean temperature and increased DTR increase dengue incidence one month later , an increase or decrease in both lessen dengue incidence . This is consistent with studies showing a positive association between DTR and dengue at low temperatures and a negative association at high temperatures [7 , 8 , 10] . Use of mean temperature alone in predicting dengue outbreaks will therefore fail to capture the full complexity of the relationship between temperature and dengue transmission . We demonstrated that increased incidence of dengue in Dhaka was associated with an increase in rainfall in the previous two months . However , an earlier study in Dhaka identified a significant positive association only at lag of two months [22] . The effect of rainfall on Ae . aegypti breeding is lessened by the species’ egg laying in artificial containers filled with water by humans . But Ae . albopictus has also been found in Dhaka [31] . Its dependence on rain-fed outdoor artificial containers as larval habitats might explain the positive association between rainfall and dengue incidence . Such a relationship has also been reported in other countries [13 , 32] . In Puerto Rico , rainfall has been proposed to have caused increases in dengue incidence by increasing Ae . aegypti density , egg laying in water storage containers and discarded tires [33] . In Thailand , monthly dengue incidence and epidemics of dengue have been associated with ENSO , which is believed to cause changes in temperature and relative humidity [34] . At time lags of one to 11 months , both epidemics and monthly cases are correlated with El Niño , which is associated with higher temperature and in some places with lower relative humidity [34] . A multivariate ENSO index , lagged at one to six months , alone explains a maximum 22% of the variations in monthly dengue cases [34] . An increase in the number of dengue cases following El Niño was also observed in Mexico , French Guiana , Indonesia , Colombia , and Surinam [13 , 18] . The role of ENSO in the inter-annual variability of monsoon rainfall in Bangladesh has been examined demonstrating that El Niño is generally associated with lower rainfall , whereas La Niña and sometimes moderate El Niño generate higher rainfall [35] . However , the relationship is not consistent over time and ENSO is only partially responsible for the rainfall anomalies in Bangladesh . Our study found a negative effect of SSTA on current dengue incidence together with a positive effect at a lag of one month . Possible explanations for the negative association with current SSTA could be that the dry weather resulting from a strong El Niño or the heavy rainfall associated with a moderate El Niño both reduce adult mosquito survival [11 , 12] and thereby reduce DENV transmission . Heavy rainfall , on the other hand , could increase transmission because people do not cover themselves in the post-rainfall humid weather resulting in increased human-mosquito contact . The positive effect of SSTA on dengue incidence at a lag of one month is biologically plausible because moderate rainfall is needed for mosquito development , and is also consistent with our findings of a positive influence of rainfall on dengue transmission at a lag of one month . However , heavy rainfall washes away mosquito larvae reducing vector numbers thereby transmission in the following month . Consideration of the non-linear influence of ENSO on rainfall may provide a richer insight into the relationship between dengue and SSTA . Socio-demographic and economic factors , as well as climate , powerfully influence dengue incidence . A study projects the population at risk of dengue in 2050 under global climate change considering gross domestic product per capita ( GDPpc ) as an indicator of socio-economic development [19] . The study reports 5 . 0% and 4 . 0% increases in the population at risk of dengue in 2050 compared to the baseline risk population in 2000 considering only the projected increase in population and the projected changes in both climate and GDPpc , respectively . Positive but non-significant effects of population growth on dengue cases have also been reported in Mexico [13] . In our study in Dhaka population density was used as a proxy for socio-demographic factors and was found to be positively associated with dengue incidence . The strength of the present study is that we considered both small and large-scale climatic influences on dengue incidence along with the interaction between mean temperature and DTR and included population density in the model as a proxy for socio-demographic changes over time . However , while we demonstrated significant associations between temperature and rainfall with dengue transmission we did not model non-linear relationships , and we excluded relative humidity from our model due to its strong correlation with mean temperature and DTR . We used months as our temporal unit of study because daily data on dengue incidence were not available . As a consequence , short-scale influences of climatic parameters on dengue incidence may not be fully captured by our model , and lag effects cannot be determined at a fine time-scale . Another limitation of the model used here is that it did not allow for under-reporting from passive surveillance data or possible changes in the rate of under-reporting . However , inclusion of a temporal trend variable in the model may indirectly capture variation in the rate of under-reporting . In conclusion , our findings indicate that the association between weather and dengue transmission is complex , which is further confounded by socio-demographic factors like population density . Models designed for forecasting should account for this complexity in order to minimize the risk of overestimation in relation to increasing mean temperature , thereby optimizing resource allocation in tropical overpopulated countries with limited resources .	The sensitivity of mosquito vector and dengue virus biology to diurnal temperature variability has been established , but this study is the first analyzing these relations with dengue occurrence . We show that Dhaka’s tropical hot monsoon climate and small variation in daily temperature enhance dengue transmission one month later . Large-scale climatic events like El Niño-Southern Oscillation and increasing population density of Dhaka also increase incidence . Our results therefore enable us to accurately estimate dengue transmission dynamics in densely populated areas that are also vulnerable to global warming by considering diurnal variability . Our approach reduces the chance of overestimating the effect of increasing temperature on dengue transmission intensity with the ultimate goal of outbreak prediction and control .	[ "Abstract", "Introduction", "Materials", "and", "Methods", "Results", "Discussion" ]	[]	2015	Interaction of Mean Temperature and Daily Fluctuation Influences Dengue Incidence in Dhaka, Bangladesh
Ivermectin ( IVM ) is a widely-used anthelmintic that works by binding to and activating glutamate-gated chloride channel receptors ( GluClRs ) in nematodes . The resulting chloride flux inhibits the pharyngeal muscle cells and motor neurons of nematodes , causing death by paralysis or starvation . IVM resistance is an emerging problem in many pest species , necessitating the development of novel drugs . However , drug optimisation requires a quantitative understanding of GluClR activation and modulation mechanisms . Here we investigated the biophysical properties of homomeric α ( avr-14b ) GluClRs from the parasitic nematode , H . contortus , in the presence of glutamate and IVM . The receptor proved to be highly responsive to low nanomolar concentrations of both compounds . Analysis of single receptor activations demonstrated that the GluClR oscillates between multiple functional states upon the binding of either ligand . The G36’A mutation in the third transmembrane domain , which was previously thought to hinder access of IVM to its binding site , was found to decrease the duration of active periods and increase receptor desensitisation . On an ensemble macropatch level the mutation gave rise to enhanced current decay and desensitisation rates . Because these responses were common to both glutamate and IVM , and were observed under conditions where agonist binding sites were likely saturated , we infer that G36’A affects the intrinsic properties of the receptor with no specific effect on IVM binding mechanisms . These unexpected results provide new insights into the activation and modulatory mechanisms of the H . contortus GluClRs and provide a mechanistic framework upon which the actions of drugs can be reliably interpreted . Pentameric ligand gated ion channels ( pLGICs ) are membrane-bound receptors that facilitate the diffusion of ions across cell membranes in response to the binding of agonists . The glutamate-gated chloride channel receptor ( GluClR ) , first identified in arthropods , such as insects and crustaceans [1–3] , is an anion-selective pLGIC found at neuronal and neuromuscular inhibitory synapses [4] . GluClRs are also present in other major metazoan phyla , including platyhelminths and nematodes [4] , but have not yet been identified in vertebrates . GluClRs can exist as homo- or hetero-pentamers [5] . Crystal structures of the homomeric GluClR from the nematode , C . elegans , have recently been determined in ligand-bound [6] and apo [7] states . The mechanism of agonist activation has been studied extensively in vertebrate pLGIC members , such as the glycine ( GlyR ) [8] , acetylcholine ( AChR ) [9–11] , serotonin ( 5-HT3R ) [12] and GABAA ( GABAAR ) [13–15] receptors , as well as ELIC , which is a bacterial pLGIC [16] . A detailed study of the biophysical properties of GluClR activation has not been undertaken , even though GluClRs constitute a major group of pLGICs , many organisms that express them are serious parasitic pests , or vectors for disease transmission and they are a major target for anthelminthic drugs . For instance , O . volvulus and W . bancrofti are nematodes that cause river blindness ( onchocerciasis ) and elephantiasis ( lymphatic filariasis ) , respectively , in humans . Another nematode , H . contortus [17] is a serious pathogen in ruminant agricultural animals such as cattle , sheep and goats . The sea lice ( arthropod ) species C . rogercresseyi [18] and L . salmonis [19] , ravage salmon and trout farming industries worldwide . The cereal cyst nematode H . avenae devastates broad acre cereal crops across temperate wheat-producing regions of the world [20 , 21] . A . gambiae is the mosquito vector that transmits malaria in over 90% of world-wide cases [22] . Finally , the flatworm blood fluke , S . mansoni , inflicts schistosomiasis ( associated with serious systemic morbidities ) on hundreds of millions of people in underdeveloped communities [23] . Macrocyclic lactones ( MLs ) such as ivermectin ( IVM ) , moxidectin , abamectin and emamectin are widely used to control all of these , as well as many other , nematode and arthropod pests [24] . IVM works by activating GluClRs in pharyngeal muscle cells and motor neurons of these organisms , thereby causing death by flaccid paralysis or starvation [25] . Unfortunately , however , IVM resistance is emerging as a serious problem in many pest species [21 , 26–29] prompting the need for new generation treatments . Functional and crystallographic studies have recently delineated the binding pocket of IVM and identified potential residues that IVM interacts with [6 , 30 , 31] . The main structure of the pocket is formed by first ( TM1 ) and third ( TM3 ) transmembrane domains of adjacent receptor subunits , at the level of the upper leaflet of the cell membrane [6] . Site-directed mutagenesis of transmembrane domains has identified critical residues that drastically affect IVM potency in the avr-14b subunit of H . contortus , such as TM3-G36’ [30] and TM1-P230 [31] , and in the α subunit of C . elegans , such as TM1-L279 and TM1-F276 [32 , 33] . The glutamate binding site and TM3 domain are also sites that harbour mutations in wild ML-resistant strains of C . elegans [27] , whereas ML resistance in wild isolates of pest species have been attributed to mutations at , TM3-30’ in P . xylostella [34] and TM3-36’ in T . urticae [35 , 36] . Of particular note , a Gly at the 36’ position is thought to be essential for exquisite IVM [30] and abamectin [34] sensitivity , and larger substitutions at this location were proposed to reduce ML sensitivity by hindering access to its binding site [31 , 34 , 37] . However , the effects of these mutations are generally evaluated using functional assays that lack the resolution needed to distinguish discrete functional states in the activation process . A detailed mechanistic understanding of how wild-type and mutated receptors respond to glutamate is a prerequisite to understanding how IVM and other modulating ligands affect the receptor . This aim is best achieved through the study of single channel currents mediated by individual receptors [38] . Without a quantitative understanding of activation and modulation mechanisms of the receptor , attempts to design drugs with higher potency and selectivity for the receptor would be intractable . Four GluClR subunits have been identified in H . contortus [α3A ( avr-14a ) , α3B ( avr-14b ) , β and α] , all of which express on motor neuron commissures [39 , 40] . However , the native stoichiometric combinations of these subunits is unknown [4] . Here we investigated homomeric receptors comprising the avr-14b subunit , which is also expressed in pharyngeal neurons [39 , 40] . We will refer to this subunit as α ( avr-14b ) . In heterologous expression systems , homomeric receptors comprising either α ( avr-14b ) [30 , 41] or α subunits form high affinity IVM binding sites , whereas the β subunit homomers do not [42] . In this study we investigated the biophysical properties of the homomeric α ( avr-14b ) GluClR from H . contortus as: 1 ) H . contortus is a major parasitic pest of domestic ruminant animals , 2 ) IVM is used widely to control H . contortus , 3 ) IVM resistance has emerged as a major problem in this species [43] , and 4 ) GluClRs comprising or containing the α ( avr-14b ) subunit are most likely the major biological IVM target in this species [4 , 25] . Here we sought to quantify the activation properties of this receptor in the presence of glutamate and IVM , and to explore the mechanism by which the TM3-G36’A mutation reduces IVM sensitivity to a level that is similar to vertebrate GlyRs and GABAARs [30 , 44–46] . Single receptor currents ( Fig 1A ) were measured and plotted as a function of voltage ( Fig 1B ) to determine the single channel conductance of the receptor . Using Eq 1 and a mean current amplitude of 1 . 80 ± 0 . 03 pA ( n = 7 , at ‒70 mV ) , the estimated single channel conductance of the homomeric GluClR was 22 . 9 ± 0 . 3 pS . The i-V was nearly linear ( Fig 1B ) . The slight inward rectification and relatively small conductance of the homomeric GluClR is very similar to that determined for ternary GABAARs containing , α , β and γ subunits [13 , 14] . A recent study has also estimated the current amplitude of the heteromeric GluClR of C . elegans at ‒ 90 mV to be ~ 1 . 9 pA [32] . Single channel activity was recorded in the presence of a broad range of L-glutamate concentrations ( 10 mM– 5 nM ) to determine the receptor’s sensitivity to glutamate , the active durations of single receptors , the total time spent in conducting configurations ( PO ) and the shut and open dwell characteristics within each active period . Continuous sweeps of single channel activity , recorded from a patch in the presence of 200 μM glutamate is shown in Fig 1C . At this and higher concentrations the activity of single receptors occurred as clearly defined periods of variable duration , termed ‘activations’ , where the receptor oscillated between conducting and non-conducting configurations . These active periods were interrupted by relatively long intervals of inactivity where the receptor adopted desensitised states . These states are distinct from ligand-bound shut states both structurally [47 , 48] and functionally [49] . With few exceptions , desensitised states are much longer-lived than shut states . Mean dwell times of the shut and open durations within activations were generated by plotting histograms and fitting these to mixtures of exponentials ( Fig 1D ) . The shut dwell data were best described by two exponential components , whereas the open dwell histogram was best fit to three exponential components . The dwell time constants were similar when the receptors were exposed to 10 mM and 1 mM glutamate , but at 200 μM the time constant of the longer shut component increased ( S1 Table ) . Reducing the glutamate concentration to 30 μM resulted in similar single channel activity ( Fig 1E ) . The number of components and the time constants of both dwell histograms ( Fig 1F ) , appeared little changed , except for a further increase in the time constant of the longer shut component ( S1 Table ) . In addition , the mean duration of the active periods appeared to become shorter as glutamate concentration decreased ( S2 Table ) . At low glutamate concentrations there appeared to be a transition from mostly tightly grouped to loosely grouped periods of activity and isolated open-shut events . For example , 2 μM glutamate elicited activity that comprised a mixture of isolated open-shut events and activations consisting of openings and shuttings in rapid succession , as with the higher glutamate concentrations . However , these latter more complex activations were more likely to occur in shorter bursts ( Fig 2A ) . The dwell histograms also exhibited two shut and three open components with similar time constants to those for the higher concentrations of glutamate , but the time constant of the longer shut component continued to increase and the fraction of the longest open time constant diminished ( Fig 2B , S1 Table ) . 30 nM glutamate elicited activations that occurred as bursts of loosely spaced openings and brief open-shut events ( Fig 2C ) . Moreover , long stretches of record corresponding to receptor desensitisation were mostly absent . The dwell histograms revealed changes to both shut and open components . Here both shut components increased and the longest open component disappeared ( Fig 2D , S1 Table ) . As 30 nM glutamate was effective at eliciting single channel activity the concentration was lowered even further , to 5 nM . Remarkably , even at this concentration GluClRs were activated . Most of the activity comprised simple shut-open events , but the occasional activation of loosely spaced openings was also apparent . In contrast , in the absence of glutamate , receptor openings were extremely rare , brief and essentially negligible ( Fig 3A ) . These data show that 1 ) the homomeric GluClRs are exquisitely sensitive to glutamate and 2 ) from 2 μM glutamate and below , the activity becomes increasingly simple and brief , likely reflecting an effect consistent with agonist dissociation from partially liganded receptors . The dwell histograms derived from 30 nM and 5 nM glutamate showed distinct differences compared to those of higher concentrations . Here both the longer and briefer shut components increased ( Fig 3B ) and the third , longest open component was absent , whereas the remaining two open components decreased ( Fig 3C , S1 Table ) . The invariant open dwell components for concentrations ≥ 2 μM glutamate are consistent with an optimal degree of ligation for receptor activation , as has been shown for the GlyR [8] . In contrast , the decrease in the remaining two open component time constants at nanomolar concentrations of glutamate is consistent with sub-optimal activation of receptors . We infer that at 30 nM and 5 nM each receptor is bound to fewer ligand molecules than at the higher concentrations , giving rise to openings with briefer lifetimes [50] . We also infer that the steadily increasing longer shut component at ≤ 200 μM glutamate is additional evidence that the receptors are able to activate without all glutamate binding sites being occupied [8] . The total time spent in open states was also compared across glutamate concentrations . This analysis demonstrates that PO increases as a function of glutamate concentration ( Fig 3D ) . Consistent with the inference that the receptors are highly sensitive to glutamate , the PO at glutamate concentrations ≥ 10 μM were all > 0 . 90 . PO showed a significant decrease at 2 μM glutamate and dropped to 0 . 21 at 30 nM and 0 . 14 at 5 nM ( Fig 3D , S2 Table ) . A Hill fit to the PO plot revealed a maximum of 0 . 99 , an EC50 of 70 nM and a Hill coefficient of 0 . 82 . The mean duration of activations was also plotted and showed that active durations declined from ~500 ms to ~330 ms between 10 mM and 30 μM glutamate . Fitting the data to a Hill equation produced an EC50 of 31 μM , a Hill coefficient of 0 . 56 , a maximum duration of 500 ms and a minimum of ~80 ms ( Fig 3E ) . The activation properties of homomeric GluClRs were also investigated at an ensemble current level using rapid solution exchange [14 , 51] of glutamate onto macropatches . As GluClRs are located at inhibitory synapses , these experiments were carried out to mimic synaptic activation conditions by determining the response of many ( ~20–100 ) receptors to rapid glutamate exposure . By avoiding the distorting effects of receptor desensitisation encountered with slower agonist application methods , rapid solution exchange techniques also establish a more accurate ligand concentration ‒ peak current relationship . Peak current was achieved by rapidly applying glutamate for either 50 ms ( 5 mM– 20 μM ) or 500 ms ( 10 μM– 0 . 5 μM , Fig 4A ) . These data were fitted to a Hill equation , yielding an EC50 for glutamate of 43 μM and a Hill slope of 0 . 8 ( Fig 4B ) . Whole-cell experiments on the same GluClR produced an EC50 for glutamate of ~15 μM and a Hill slope of ~1 . 7 [30] . The 2–3 fold difference in EC50 and Hill slope between whole-cell and macropatch data are consistent with open and desensitised states , which have a higher affinity for ligand , having made a significant contribution to the whole-cell data . Similar experiments were carried out to determine the relationship between the activation rate of the current and agonist concentration . Normalised examples of these recordings are illustrated in Fig 4C and the group data are summarised in Fig 4D . A Hill fit to this plot produced an EC50 of 0 . 95 mM and a Hill slope of 1 . 0 . The upper asymptote of the activation plot was ~9000 s−1 , representing the maximum activation rate [51 , 52] , whereas the lower level was ~10 s−1 . Homomeric GluClRs containing the G36’A mutation exhibit a markedly reduced IVM sensitivity ( EC50 ) when recorded in whole-cell configuration [30] . However , any changes to the intrinsic properties of the receptor conferred by the mutation have yet to be examined in mechanistic detail . G36’A-containing receptors were first examined on a single channel level . Applied glutamate elicited a similar current amplitude to wild-type receptors , suggesting the G36’A had no appreciable effect on channel conductance ( Fig 5A ) . A current amplitude of 1 . 81 ± 0 . 02 pA ( n = 7 ) for the mutant at ‒70 mV yielded a conductance of 23 . 0 ± 0 . 2 pF , if it is assumed that under the same recording conditions the reversal potential is similar to that for wild-type . However , moderate to high ( 30 μM– 10 mM ) concentrations of glutamate revealed two distinct types of activations in the mutant receptor ( Fig 5A ) , whereas the same concentration of glutamate elicited homogeneous activations in the wild-type receptor ( Fig 5B ) . The two activation modes mediated by the mutant GluClR were quantified on the basis of PO and duration only for 1 mM glutamate . The analysis revealed a very low PO activation mode of 0 . 14 ± 0 . 02 ( n = 6 ) and mean active periods of 1333 ± 222 ms duration and a higher , more wild-type like mode with a PO of 0 . 87 ± 0 . 05 and a mean active duration of 309 ± 62 ms ( Fig 5C ) . In contrast , 1 mM glutamate produced a single PO of 0 . 99 ( S2 Table ) in the wild-type receptor consistent with fewer shuttings within each activation ( Fig 5D ) . The two activation modes observed in the mutant receptors were pooled for further analysis for all concentrations where they were apparent so as to determine the net effect of the mutation on PO and active durations , and facilitate a more direct comparison to wild-type receptors . The dwell histograms for the mutant receptor at 1 mM glutamate required two shut and three open components ( Fig 5E ) , but the longer shut component was substantially increased compared to wild-type receptors ( Fig 5F ) and the two longest open components were reduced ( S1 Table ) . Two distinct gating modes were also observed at a moderate ( 30 μM ) glutamate concentration ( Fig 6A ) , but were difficult to distinguish at a low ( 2 μM ) concentration because the activations became too brief and simple ( Fig 6B ) . Over the concentration range tested , the mean duration of activations of the mutant receptor were considerably shorter than wild-type with a maximum mean duration of 200 ms , as was the mean PO , which peaked at 0 . 73 ( Fig 6C , S2 Table ) . Fig 6D summarises the dwell component data over the glutamate concentrations that were tested on the mutant receptors . Consistent differences to wild-type receptors include an increase in the long shut component and briefer open components ( Fig 6E ) . At 2 μM glutamate only one shut component and two open components were resolvable ( Fig 6F , S1 Table ) . The briefer active periods exhibited by the G36’A mutant receptors is indicative of accelerated ensemble current deactivation [14 , 53] and desensitisation [49] . To investigate whether receptor desensitisation was affected by the G36’A mutation , the long quiescent periods corresponding to desensitisation in single channel records were quantified , then corrected for channel number [49] . Sample recordings for wild-type and mutant receptors are shown in Fig 7A and 7B , respectively . A saturating concentration of glutamate ( 10 or 1 mM ) was first rapidly applied onto each patch , ensuring that all the receptors in the patch were activated , after which constant agonist perfusion was maintained over the patch for the remainder of the recording . Clearly defined steps corresponding to the single channel amplitude ( ~2 pA ) became apparent as all the receptors desensitised back to baseline . The number of steps was then taken as an estimate of the total number of receptors contained in each recorded patch . Only patches expressing 1–10 steps ( channels ) were accepted for analysis . The long desensitised periods were estimated by plotting shut dwell histograms for the entire record , as is illustrated in Fig 7C and 7D . The shut events could be divided into two broad components . The briefer component corresponded to shut events within active periods . This component could be subdivided into briefer components ( as in Figs 1 , 2 , 4 and 5 ) . The longer component represented the mean desensitised lifetime , and it was this component that was corrected for channel number . This method of analysis produced a mean desensitised lifetime for wild-type receptors of 91 s ( n = 5 ) , and was used to determine a re-sensitisation transition rate constant ( ω , Fig 7E ) of 0 . 011 s‒1 . Similarly , the desensitisation rate constant ( δ ) was estimated from the mean duration of the active periods at saturating glutamate concentrations ( 500 ms , Fig 3B ) to be 2 . 00 s‒1 [49] . This produced an equilibrium constant ( δ/ω ) for desensitisation of 182 for wild-type receptors . A similar analysis for the G36’A mutant receptor produced a mean desensitised lifetime of 125 s ( n = 8 ) and an ω of 0 . 008 s‒1 . However , a more significant change was estimated for the mean duration of active periods for the mutant , which saturated at 200 ms ( S2 Table ) , producing a δ value of 5 . 00 s‒1 and an equilibrium constant of 625 . Thus , mutant receptors desensitised ~3 . 4 times more rapidly than wild-type receptors . From this analysis it can be inferred that the G36’A mutation increases the likelihood of the receptors entering desensitised states . To determine if the estimates of receptor desensitisation reflected current decay and desensitisation in ensemble currents , macropatches expressing wild-type or G36’A mutant GluClRs were exposed to a saturating ( 3 mM ) concentration of glutamate for either ~1 ms or 500 ms . In response to a ~1 ms application , the deactivation phase of macropatch currents mediated by wild-type GluClRs was adequately described by two standard exponential functions with a weighted time constant of 67 ± 4 ms ( Fig 8A and 8B ) . The individual time constants ( and fractions ) are tabulated in Table 1 . To allow comparison with other , better characterised pLGICs , heteromeric α1β GlyRs and α5β3γ2L GABAARs were tested under similar conditions . A ~1 ms pulse of 3 mM glycine applied to α1β GlyRs elicited macropatch currents that also exhibited a two component decay phase with a weighted mean of time constant of 22 ± 3 ms ( Fig 8A ) . In contrast , a ~1 ms pulse of 3 mM GABA applied to α5β3γ2L GABAARs activated macropatch currents that decayed considerably more slowly than those of GluClRs . They also deactivated with two components , with a weighted time constant of 275 ± 2 ms ( Fig 8A and 8B , Table 1 ) . 3 mM glutamate was also rapidly applied for ~1 ms onto patches expressing the G36’A mutant GluClR ( Fig 8C ) . The weighted time constant from a two component fit was 11 ± 1 ms , which was 2-fold faster than those of the α1β GlyR and over 6-fold faster than the wild-type GluClR ( Fig 8B ) . The activation phase of the currents was also measured by fitting 10–100% of the rising phase of the current to Eq 2 . The measurements , summarised in Fig 8D , demonstrate that currents mediated by all receptors tested activate with similar time constants , which ranged between 0 . 1 ‒ 0 . 2 ms . In contrast to the deactivation kinetics , the G36’A mutation had no significant effect on the ability of the receptor to activate upon exposure to glutamate . Ensemble desensitisation was examined by rapidly applying 3 mM glutamate onto macropatches for a duration of 500 ms ( Fig 8E ) . Wild-type GluClRs desensitized with single time constant of 492 ± 38 ms , whereas the G36’A mutant receptor required two exponential functions to adequately describe the desensitisation phase of the current ( Table 1 ) . The weighted desensitisation time constant for the mutant receptor was 252 ± 26 ms ( Fig 8F ) . We infer that the number of components that were needed to describe single receptor and ensemble desensitisation is related to modal activation in the mutant receptor . Consistent with this inference , estimates of the mean active durations for both receptors at saturating glutamate match very closely with the time constants of ensemble desensitisation ( Tables 1 and S2 ) . Overall , these data demonstrate that the G36’A mutation abbreviates single channel active periods , which manifest as accelerated deactivation and desensitisation in ensemble currents . These alterations to the intrinsic activation properties of the receptor are likely the underlying reasons for the order of magnitude rightward shift in the whole-cell concentration-response relationship for glutamate , reducing its sensitivity ( EC50 ) from 15 μM to 154 μM . However , studies have also revealed a parallel shift in IVM sensitivity ( EC50 ) , from 40 nM to 1 . 2 μM in the G36’A-containing receptor [30 , 45] . IVM is both a direct agonist and a potentiator of glutamate responses at the GluClR . In our final set of experiments we wished to test the hypothesis that the changes to the functional properties of the receptors conferred by the G36’A substitution gives rise to the reduced sensitivity to IVM , as it does for glutamate . To test this idea , we recorded single channel currents in the presence of 5 nM IVM alone ( direct activation ) or in 5 nM IVM + 2 μM glutamate ( potentiation ) . In both experiment types the receptors opened to an amplitude of 1 . 8 pA ( e . g . , Fig 9A ) , suggesting that the presence of IVM had little effect on the permeation pathway . Wild-type receptors exhibited a substantial degree of potentiation and direct activation by IVM . However , the recordings also revealed that these experiments were not ‘steady state’ . We confined our analysis the steady-state phase of both experiments types ( direct activation and potentiation ) . When membrane patches expressing wild-type receptors were exposed to 5 nM IVM alone , no receptor activity was apparent for the first 41 ± 4 ms . After this initial silent period the activations were initially well separated , but increased in duration for 47 ± 25 s , after which the active durations reached a steady-state equilibrium of almost continuous activity ( Fig 9A ) of all the receptors present in each patch ( between 1–4 receptors ) . The mean active duration of the receptors at steady-state was 9 . 5 ± 2 . 6 s and had a PO of 0 . 65 ± 0 . 07 ( Table 2 ) . The shut intervals were best described by three components whereas the open interval histograms required four exponential components for fitting ( S3 Table ) . The presence of additional shut and open components suggests that IVM alone induces activity of greater complexity or exposes state lifetimes that are not easily resolvable when glutamate is present . Receptor desensitisation by IVM alone had a mean lifetime of 536 ± 140 ms ( ω = 1 . 87 s‒1 ) . A mean active duration of 9 . 5 ± 2 . 6 s ( δ = 0 . 105 s‒1 ) yielded an equilibrium constant of 0 . 06 . Direct activation of G36’A mutated receptors by 5 nM IVM produced a similar lag time before equilibrium was reached ( Fig 9C ) . At equilibrium the receptors were active for a mean duration of 46 ± 8 ms and a PO of 0 . 85 ± 0 . 01 ( Fig 9D , Table 2 ) . These active periods were much briefer than wild-type receptors when activated by IVM directly ( 9 . 5 s ) . Dwell histograms revealed two shut and three open components , which is less complex than wild-type ( S3 Table ) . Moreover , the mutated receptors desensitised for a mean of 2004 ± 268 ms , yielding a desensitisation equilibrium constant of 41 . 7 s‒1 ( ω = 0 . 499 s‒1 and δ = 20 . 8 s‒1 ) . The mean active durations and PO data are summarised as bar plots in Fig 9E and 9F , respectively . Wild-type receptors activated rapidly upon exposure to 5 nM IVM and 2 μM glutamate . For the first 66 ± 18 s after commencement of the recording , the active periods were well-separatedincreasing in duration over time ( Fig 10A ) , until an apparent steady-state equilibrium was reached ( Fig 10B ) . An analysis of the active durations , PO and the dwell time components at equilibrium produced a mean active duration of 15 . 9 ± 4 . 2 s and a PO of 0 . 88 ± 0 . 02 ( Table 2 ) for potentiation of wild-type currents . The dwell histograms were best described by three shut and three open components ( S3 Table ) . The time constants for the first two shut and open components were similar to those in the presence of low concentrations of glutamate ( S2 Table ) . In contrast , the longest open component was at ~ 200 ms and represented about 40% of the total open intervals . To estimate receptor desensitisation , long stretches of record consisting of single receptor activity were analysed to obtain the main shut component that separated discrete active periods ( Fig 10B ) . This shut component produced a short-lived mean , after correcting for channel number , of 223 ± 36 ms and thus an ω of 4 . 48 s‒1 . Using a mean active duration of 15 . 9 s ( δ = 0 . 063 s‒1 ) , the calculated equilibrium constant for receptor desensitisation was 0 . 01 in the presence of IVM and glutamate . Similar experiments were carried out for the G36’A-containing mutant . Here too the active periods initially increased in duration ( Fig 10C ) , before equilibrating to steady-state activity ( Fig 10D ) . However , steady-state activity was not near continuous , as was observed for the wild-type receptors . Instead , individual receptors were active for a mean duration of 113 ± 25 ms and had a PO of 0 . 60 ± 0 . 04 ( Table 2 ) . Receptor desensitisation was also unlike that of wild-type receptors . The mean shut lifetime for long stretches of record was 2381 ± 657 ms ( ω of 0 . 420 s‒1 ) . This yielded an equilibrium constant for desensitisation of 21 . 1 . As for glutamate-gated activity , IVM produced briefer active periods and induced greater desensitisation in the G36’A GluClRs than wild-type . The summary of the mean active durations and PO is provided in Fig 10E and 10F , respectively . The POs were significantly different between direct activation and potentiation for both wild-type and mutant receptors . However , because direct activation by IVM of the mutant receptors produced brief , simple activations , the PO determined for this activity was relatively high . In summary , IVM acted as an agonist and potentiated currents in the presence of glutamate at wild-type and G36’A mutated GluClRs to elicit significantly longer active periods and markedly reduce receptor desensitisation . In addition , the sparse activity at the start of the recordings , which equilibrated to steady-state activity implies that , as with glutamate , additional binding of IVM molecules to each receptor saturates receptor activation . The two broad aims of this study were firstly , to examine the activation properties of GluClRs expressed by a parasitic species in the presence of its physiological agonist and secondly , to explore the mechanism of IVM sensitivity . To achieve the first aim glutamate-gated currents were examined over a wide concentration range on single receptor and ensemble levels . The conductance of the receptor channel was determined to be ~23 pS , which is close to that of GABAARs that comprise α , β and γ subunits [13 , 54] . Upon binding to glutamate , wild-type GluClRs activated rapidly ( ~9000 s‒1 , Fig 4 ) , comparable with the rate of other pLGICs , including the G36’A mutated GluClRs ( Fig 8 ) . The experiments also revealed that wild-type GluClRs were highly responsive even at low nanomolar concentrations of glutamate and exhibited active durations and an open probability that was concentration-dependent . These parameters saturated at ~500 ms and 0 . 99 , respectively ( Fig 3 ) . Dwell interval analysis of active periods demonstrated that the receptors have multiple components , indicating that each receptor oscillates between multiple functional states during receptor activation [8 , 11 , 13] . The pattern of dwell components also indicated that at ≥ 2 μM glutamate an optimal number of bound glutamate molecules achieves efficient receptor activation . This is similar to GlyR activation , whereby fitting data to postulated kinetic schemes it was deduced that three bound glycine molecules are sufficient for optimal activation [8] . The decrease in open dwell times at nanomolar concentrations of glutamate clearly showed that at these concentrations fewer glutamate molecules were bound on average to each receptor [50] . The G36’D and G36’E mutations have been identified in the ML-resistant agricultural pest mite T . urticae [35 , 36] . These mutations occur on different subunit isoforms , suggesting that heteromeric GluClRs containing different substitutions to G36’ could work either individually or synergistically to reduce ML sensitivity [36 , 55] . The G36’E mutation is particularly effective at reducing ML sensitivity on its own and homomeric receptors expressed in oocytes demonstrate complete insensitivity to two MLs ( abamectin and milbemycine A4 ) [55] . Our data suggest that the G36’A mutation gives rise to significant functional changes , such as a reduced active duration and an increase in desensitisation of single receptors , which manifest as faster current decay and reduced sensitivity to glutamate and IVM . Whether these functional changes are also present in G36’D or E has yet to be determined . However , given that both substitutions contribute large side groups that are likely negatively charged , it is likely that these too would affect the activation properties of the receptors . The physico-chemical properties of aspartate and glutamate may also restrict access of IVM to its binding site . We chose to study the G36’A mutation because it dramatically decreases IVM sensitivity [30] and is located on the TM3 domain where crystallographic data show that it contributes one side of the IVM binding site [6] ( Fig 11 ) . Given its location , it is tempting to hypothesise that the G36’A substitution reduces IVM sensitivity simply by disrupting the binding of IVM . However , the mutation also decreases the EC50 of glutamate [30] , which binds at an extracellular domain site that is over 3 nm from the site of the mutation . Another mechanism that could reconcile the parallel decrease in glutamate and IVM sensitivities is that the actions of both ligands reveal changes to the intrinsic activation properties of the receptor conferred by the mutation . Distinguishing between these two possibilities is critical to understanding the mechanism of action of IVM . This is of particular importance given that IVM resistance in H . contortus and other pest species is an emerging concern [21 , 26 , 29] . To help distinguish between these two possibilities we analysed glutamate- and IVM-gated currents in wild-type and G36’A mutated receptors . Clear evidence that the G36’A mutation markedly compromised receptor activation was gleaned in the presence of glutamate alone . The mutation gave rise to two distinct and stable modes of activation; one that was similar to wild-type , and another with a much reduced PO ( Figs 5 and 6 ) . The wild-type like mode was briefer than the activations mediated by wild-type receptors over the glutamate concentrations tested and both modes had lower POs than wild-type . When analysed together , the net effect of these modes produced a maximum mean active duration of ~200 ms and a PO of 0 . 70 ( Fig 6 ) . These parameters underlie the reduced sensitivity to glutamate observed in the G36’A mutated receptors . Indeed , where 2 μM glutamate elicited robust activity in wild-type receptors , it produced only sparse , simple activity in the mutant . These results led to the hypothesis that the mutation impaired receptor desensitisation and ensemble current decay . This was tested at the single receptor level ( Fig 7 ) and in macropatches ( Fig 8 ) . The single receptor experiments yielded desensitisation equilibrium constants of ~180 and ~625 for wild-type and mutant receptors , respectively , representing a 3 . 4-fold greater likelihood of adopting desensitised states in the mutant . Ensemble deactivation and desensitisation rates were also much abbreviated in the mutant , producing mean time constants that corresponded well to the mean active durations of single receptors ( Fig 8 ) . It is notable that other pLGICs , such as α1β GlyRs and α1β2γ2 GABAARs have a similar sensitivity to IVM [30 , 44] and also exhibit similar rates of current decay [14 , 51 , 53] to the G36’A mutant . Moreover , pLGICs that exhibit low IVM sensitivity also contribute non-G36’-containing TM3 domains to their IVM binding sites [30 , 37 , 44] . IVM acted as a ligand on its own and synergistically with glutamate to enhance currents elicited by glutamate . It did not affect the single channel conductance even though it binds at a site within the transmembrane segments and is predicted to interact with the pore-lining TM2 domain [6] . At GABAARs it has been demonstrated that 10 nM IVM alone lengthens the durations of single receptor currents without changing single channel conductance [46] . 2 μM glutamate applied alone at wild-type GluClRs gave rise to a mean active duration of ~150 ms . When 5 nM IVM and 2 μM glutamate were applied together , current potentiation manifested as prolonged active durations with a mean of ~16 s , representing a two order of magnitude increase . The same combination of IVM and glutamate at G36’A mutated receptors produced active durations with a mean of 113 ms ( Fig 10 ) , compared to a mean of ~11 ms elicited by 2 μM glutamate alone . This also represents an order of magnitude change , but the absolute durations were much briefer than in wild-type receptors . A similar pattern was observed between wild-type and G36’A receptors in the presence of 5 nM IVM alone . The mean duration of active periods for wild-type receptors was ~9 . 5 s , whereas that for the mutant was a mere 48 ms ( Fig 9 ) . As for glutamate-gated currents , the active periods of the G36’A mutated receptor were much briefer when activated by IVM alone or in conjunction with glutamate compared to wild-type receptors . Receptor desensitisation in the presence of IVM was estimated by fitting shut histograms to long periods of record that contained successive single receptor activations ( Figs 9 and 10 ) . Receptor saturation , where all the receptors in each patch became active , was then used to count active receptors and correct for the desensitisation time constant . This analysis revealed that desensitisation was nearly abolished at wild-type receptors , especially in the presence of IVM and glutamate . The mean lifetimes of desensitised states were between ~220 ms and ~540 ms and yielded equilibrium constants of ~0 . 01 for IVM plus glutamate and ~0 . 06 for IVM alone , respectively . IVM alone induced a mean desensitisation lifetime of 2002 ms and an equilibrium constant of ~42 in the mutant receptors . This represents a significant increase in desensitisation compared to wild-type receptors under the same recording conditions . These data demonstrate that in the presence of each agonist alone and when they are co-applied , the G36’A mutated receptors exhibited briefer active periods and enhanced desensitisation compared to wild-type . Our data strongly support the inference that the loss of sensitivity reported for both agonists [30] is due to the same mechanistic process , and not fundamentally related to IVM binding interactions at the 36’ position . Although we cannot categorically rule out an IVM binding effect our data show that the wild-type and the G36’A mutant receptors are similarly affected even when receptor activation is at saturation throughout the recording . These conditions also correspond to ligand saturation where occupancy of receptors in unbound states is negligible . That this is the case for glutamate ( Figs 5 and 6 ) and IVM ( Figs 9 and 10 ) strongly suggests that both agonists are less efficacious at activating the mutant receptors . A notable difference between the actions of glutamate and IVM was that the onset and equilibration of currents in the presence of IVM were much slower than observed for glutamate . A lag time of over ~1‒1 . 5 minutes was apparent between the initiation of channel activity and the time when activations equilibrated to a constant mean duration for both mutant and wild-type receptors . Indeed , no activity was seen when IVM was applied alone for the first minute or so . Diffusion limited binding rates , calculated for ligands that encounter receptor binding sites directly from aqueous solution , including ligands of similar dimensions to IVM are in the range of ~5‒7 x 109 M‒1s‒1 [56] . For instance , the upper limit of the diffusion rate for a small aqueous molecule like glutamate is ~109 M‒1s‒1 [57] . The binding energy and correlated structural changes at binding sites can reduce these values by about two orders of magnitude ( ~106‒108 M‒1s‒1 ) [56] . These diffusion rates are far too high to account for the lag time observed in the recordings , suggesting the existence of other rate-limiting factors [58] . Structural evidence indicates that IVM binds to an inter-subunit cavity in the upper leaflet of the lipid bilayer [6] , as do other highly lipophilic ligands such as neurosteroids [59] and anaesthetics [60] . The IVM binding pocket in GluClRs is likely to be partly occupied by lipid , requiring its displacement by IVM for access to the pocket [6 , 7] . Due to its lipophilic nature , IVM is believed to partition into cell membranes [61] where it reaches a high local concentration , consistent with persistent whole-cell currents [30] . Thus , much of the ‘binding energy’ of IVM could be derived from the nonspecific free energy of membrane partitioning , giving rise to a high apparent affinity , whereas the actual ligand-channel interaction could be relatively weak [62] . Our data are consistent with IVM partitioning in the lipid membrane and diffusing to its binding pocket [63] , where its concentration would increase to produce current saturation over the course of several minutes in patches of membrane . The increase in the active durations of individual receptors over the initial phase of the recordings and the emergence of a long open time constant at saturation also suggests that multiple IVM molecules bind to each receptor over course of the experiment to produce saturation . Heteromeric α1β2γ2 GABAARs have also been shown to bind multiple IVM molecules , to produce interface-specific potentiation and direct current activation [44] . It has been suggested that the flexible ‘hinge’ function of glycine residues found within K+ channels [64 , 65] and pLGICs [66] can serve to isolate protein segments , or even entire domains , from surrounding protein conformational changes during channel activation [65] . According to their respective high resolution molecular structures , the TM3 domain backbones of the α1 GlyR ( which contains an endogenous A36’ residue ) and the GluClR are closely aligned in the shut state ( Fig 11A ) . However , upon IVM binding , the GlyR TM3 undergoes a larger displacement ( Fig 11B ) . This differential displacement is also observed when the TM3 domains corresponding to shut and IVM-bound states are overlaid from the same receptor ( Fig 11C and 11D ) . This strongly suggests that the A36’ residue confers structural rigidity to the TM3 . The structural comparisons in Fig 11 illustrate that the G36’ acts to minimise deformation of the TM3 between state transitions during the conformational activation ‘wave’ of pLGICs [67] . Because the G36’A mutation causes briefer active durations and an increased likelihood of adopting glutamate and IVM-induced desensitised states we conclude that the alanine destabilises open states via reduced backbone flexibility and a larger TM3 displacement . Functional studies have established that pLGIC activation and desensitization are mediated by structurally distinct sets of conformational changes at the both extracellular-transmembrane domain interface [48 , 49] and at the intracellular end of the pore [47] . The difference in IVM-induced TM3 displacement in the wild-type and G36’A mutant GluClRs will cause TM3 to interact differentially with one or both of these regions , and could thus explain the differential effect of the mutation on desensitization . The H . contortus α ( avr-14b ) GluClR is an important biological target for IVM , although IVM resistance is emerging as a problem in this pest species . Here we quantified the effects of glutamate and IVM on these receptors with the aim of understanding the structural and functional bases of their modulatory effects . We found the receptor to be highly responsive to low nanomolar concentrations of both ligands . Dwell interval analysis of active periods demonstrated that the receptor oscillates between multiple functional states during activation by either ligand . However , we also observed that the duration of activations increased with increasing ligation of receptors by either ligand . The G36’A mutation , which was previously thought to hinder access of IVM to its binding site on the receptor , was found to decrease the duration of active periods and increase receptor desensitisation . On an ensemble macropatch level these changes gave rise to enhanced current decay and desensitisation rates . There are two main reasons why we consider these effects are due to impaired channel gating and not impaired IVM binding . First , the impairment to gating was quantitatively similar for the two ligands which bind to structurally distinct sites , and second , the impairment was observed at saturating concentrations of either ligand , thus ruling out a contribution to gating from binding and unbinding events . We infer that G36’A affects the intrinsic properties of the receptor with no specific effect on IVM binding . These results provide new insights into the activation and modulatory mechanism of the GluClR and provide a mechanistic framework upon which the actions of new candidate anthelmintic drugs can be reliably interpreted . HEK AD293 cells ( ATCC cell lines , VA USA ) were seeded onto poly-D-lysine coated glass coverslips and transfected with cDNAs encoding the GluClR subunit avr-14B ( pcDNA 3 . 1+ ) of H . contortus using a calcium phosphate-DNA co-precipitate method . The cDNA encoding the CD4 surface antigen was also added to the transfection mixture and acted as a marker of transfected cells . Cells were used for experiments 2–3 days after transfection . The point mutation , TM3-G36’A , was incorporated into the subunit using the QuickChange site-directed mutagenesis method . Successful incorporation of mutation was confirmed by sequencing the mutated DNA . All experiments were carried out at room temperature ( 21–24°C ) . Single-channel and macropatch currents were recorded from outside-out excised patches at a clamped potential of −70 mV , unless indicated otherwise . The patches were continuously perfused via a gravity-fed double-barrelled glass tube . Out of one barrel flowed an extracellular bath solution containing ( in mM ) , 140 NaCl , 5 KCl , 1 MgCl2 , 2 CaCl2 , 10 HEPES , and 10 D-glucose and titrated to pH 7 . 4 . The adjacent barrel contained agonist dissolved in this extracellular solution . Glass electrodes were pulled from borosilicate glass ( G150F-3; Warner Instruments ) , coated with a silicone elastomer ( Sylgard-184; Dow Corning ) and heat-polished to a final tip resistance of 4‒15 MΩ when filled with an intracellular solution containing ( in mM ) 145 CsCl , 2 MgCl2 , 2 CaCl2 , 10 HEPES , and 5 EGTA , pH 7 . 4 . Stock solutions of L-glutamate were also pH-adjusted to 7 . 4 with NaOH . A 10 mM stock of IVM ( Sigma-Aldrich ) was dissolved in 100% DMSO and kept frozen at ‒20°C . Fresh working stocks of IVM at 5 nM were prepared by dissolving the appropriate quantity directly in extracellular solution . 100% DMSO when dissolved in extracellular solution alone at the same concentration as is present in working solutions containing 5 nM IVM had no effect on patches excised from cells transfected with GluClRs or from untransfected cells . Excised patches were directly perfused with extracellular solution by placing them in front of one barrel of the double-barrelled glass tube . Single channel currents were elicited by exposing the patch continuously to agonist containing solution , flowing through the adjacent barrel . 1–2 glutamate concentrations were applied to most patches for single receptor experiments . A ~1 minute wash with agonist-free extracellular solution was applied between each glutamate application . Because IVM does not readily wash out , either 2 μM glutamate + 5 nM IVM or 5 nM IVM alone were applied to a given patch . Macropatch currents were elicited by lateral translation of the tube from the agonist free to agonist containing barrel using a piezo-electric stepper ( Siskiyou ) . This achieved rapid solution exchange ( <1 ms ) . Currents were recorded using an Axopatch 200B amplifier ( Molecular Devices ) , filtered at 5 kHz and digitized at 20 kHz using Clampex ( pClamp 10 suite , Molecular Devices ) via a Digidata 1440A digitizer . The experiments that were carried out can be broadly divided into 1 ) single receptor currents at steady-state and 2 ) ensemble currents , which are phasic . The two types or experiments are complimentary and provide different data . Single channel recordings yield information on receptor conductance and functional state complexity ( eg . active durations and dwell histograms ) . The fast application ( ~1 ms ) ensemble measurements mimic synaptic currents . Single-channel current amplitudes were measured in Clampfit . In current-voltage ( i-V ) experiments , the amplitude was measured at voltages of , ±70 mV , ±35 mV , ±15 mV and 0 mV . The data were fit to a polynomial function in Sigmaplot ( Systat Software ) and the reversal potential was read directly from the plots . Single-channel conductance ( γ ) was calculated from the single-channel amplitude ( i ) using Ohm’s law: γ=iVhold−Vljp−Vrev Eq 1 Where Vhold is the holding potential ( −70mV ) , Vljp is the liquid junction potential and Vrev is the reversal potential . Vljp was calculated to be 4 . 7 mV for the solutions used in the experiments [68] . We confined our analysis to the largest , main conductance level . QuB software was used to analyse the kinetic properties of GluClR activations . Segments of single-channel activity separated by long periods of baseline were selected by eye and idealized into noise-free open and shut events using a temporal resolution of 70 μs . Idealized data were initially fit with a simple activation scheme in which open and shut states were added to a central shut state . This fit was used to determine the critical time ( tcrit ) , which was taken from the shut interval durations and used to divide the idealized segments into clusters ( or bursts at < 2 μM glycine ) of single receptor activity . Clusters and bursts will be referred to as activations . tcrit applied to single channel records of wild-type activity varied between 5–30 ms for concentrations ≥ 2 μM and 120–180 ms for 30 nM and 5 nM glutamate . Activation mode analysis for G36’A-containing receptors at 1 mM glutamate required tcrit times of 180–200 ms ( low PO ) or 15–50 ms ( high PO ) . Pooled data obtained from G36’A-containing receptors were defined using tcrit times of 20–50 ms . Finally , IVM-induced single channel currents were defined using tcrit values of 50–120 ms for both wild-type and mutant receptors . This analysis yielded mean cluster durations and intra-activation open probabilities ( PO ) . All data are presented as mean ± SEM of between 3 and 16 patches . The shut periods that correspond to receptor desensitisation were estimated by generating shut histograms for long stretches of record ( several minutes ) that exhibited single receptor activations . Receptor desensitisation was modelled as a single transition from an open conducting state ( ARo ) to a desensitised state ( ARd ) . Where A is the agonist , R is the receptor and the superscripts denote open ( o ) or desensitised ( d ) . δ denotes the desensitisation rate constant whereas the re-sensitisation ( or re-activation ) rate constant is denoted by ω . The equilibrium constant for desensitisation is δ/ω . Macropatch currents were analysed by fitting the onset phase of the current to a single exponential of the form: I ( t ) =Imax ( 1−e−kobst ) Eq 2 Where I ( t ) is the current at time t , Imax is the maximum current amplitude and kobs is the pseudo-first order rate constant for current activation . The decay phase of the macropatch currents were fit to two standard exponential functions . Data are presented as mean ± SEM . Power analysis of our data sets for IVM revealed power levels of 0 . 9‒1 . 0 . Two-tailed , unpaired t-tests were used to compare wild-type and mutant current parameters in the presence of IVM and p < 0 . 01 was taken as the significance threshold . The alignments of TM3 domains of the GluClR and GlyR were done using the Internal Coordinate Mechanics software ( ICM-Pro Molsoft LLC , San Diego , CA ) . The α-carbon atoms of the N-terminal residues from TM3-29’ to TM3-36’ were superimposed and used as a fixed reference . The displacement between α-carbon atoms at position TM3-56’ were then measured between two given TM3 domains The structures used for this analysis were , the GluClR in a non-conducting conformation ( PDB , 4TNV [7] ) , the GluClR in complex with IVM ( PDB , 3RHW [6] ) , the α1 GlyR in a non-conducting conformation in complex with strychnine ( PDB , 3JAD [69] ) , and the structure of the α1 GlyR in complex with IVM ( PDB , 3JAF [69] ) . The final representations were created using the Pymol Molecular Graphics System , Version 1 . 3 .	IVM is a gold standard anti-parasitic drug that is used extensively to control invertebrate parasites pest species . The drug targets the glutamate-gated chloride channel receptor ( GluClR ) found on neurons and muscle cells of these organisms , causing paralysis and death . However , IVM resistance is becoming a serious problem in human and animal health , as well as human food production . We provide the first comprehensive investigation of the functional properties of the GluClR of H . contortus , which is a major parasite in grazing animals , such as sheep and goats . We compared glutamate and IVM induced activity of the wild-type and a mutant GluClR , G36’A , that markedly reduces IVM sensitivity in wild populations of pests . Our data demonstrate that the mutation reduces IVM sensitivity by altering the functional properties of the GluClR rather than specifically affecting the binding of IVM , even though the mutation occurs at the IVM binding site . This study provides a mechanistic framework upon which the actions of new candidate anthelmintic drugs can be interpreted .	[ "Abstract", "Introduction", "Results", "Discussion", "Materials", "and", "methods" ]	[ "single", "channel", "recording", "neurochemistry", "chemical", "compounds", "aliphatic", "amino", "acids", "social", "sciences", "species", "delimitation", "neuroscience", "organic", "compounds", "mutation", "substitution", "mutation", "membrane", "electrophysiology", "spe...	2017	Effects of glutamate and ivermectin on single glutamate-gated chloride channels of the parasitic nematode H. contortus
Reinitiation is a gene-specific translational control mechanism characterized by the ability of some short upstream uORFs to retain post-termination 40S subunits on mRNA . Its efficiency depends on surrounding cis-acting sequences , uORF elongation rates , various initiation factors , and the intercistronic distance . To unravel effects of cis-acting sequences , we investigated previously unconsidered structural properties of one such a cis-enhancer in the mRNA leader of GCN4 using yeast genetics and biochemistry . This leader contains four uORFs but only uORF1 , flanked by two transferrable 5′ and 3′ cis-acting sequences , and allows efficient reinitiation . Recently we showed that the 5′ cis-acting sequences stimulate reinitiation by interacting with the N-terminal domain ( NTD ) of the eIF3a/TIF32 subunit of the initiation factor eIF3 to stabilize post-termination 40S subunits on uORF1 to resume scanning downstream . Here we identify four discernible reinitiation-promoting elements ( RPEs ) within the 5′ sequences making up the 5′ enhancer . Genetic epistasis experiments revealed that two of these RPEs operate in the eIF3a/TIF32-dependent manner . Likewise , two separate regions in the eIF3a/TIF32-NTD were identified that stimulate reinitiation in concert with the 5′ enhancer . Computational modeling supported by experimental data suggests that , in order to act , the 5′ enhancer must progressively fold into a specific secondary structure while the ribosome scans through it prior uORF1 translation . Finally , we demonstrate that the 5′ enhancer's stimulatory activity is strictly dependent on and thus follows the 3′ enhancer's activity . These findings allow us to propose for the first time a model of events required for efficient post-termination resumption of scanning . Strikingly , structurally similar RPE was predicted and identified also in the 5′ leader of reinitiation-permissive uORF of yeast YAP1 . The fact that it likewise operates in the eIF3a/TIF32-dependent manner strongly suggests that at least in yeasts the underlying mechanism of reinitiation on short uORFs is conserved . Translation of the majority of eukaryotic mRNAs encoding almost exclusively only a single large open reading frame ( ORF ) is initiated by the canonical mechanism involving formation of the 48S pre-initiation complex ( PIC ) at the mRNA's 5′ cap structure followed by scanning through the 5′ untranslated region ( UTR ) for usually the nearest AUG start codon ( reviewed in [1] ) . According to recent reports , however , in approximately 13% of yeast and 50% of human transcripts the main ORF is preceded by one or more short upstream ORFs ( uORFs ) [1] , [2] , consisting of the AUG start codon and at least one additional coding triplet . Presence of a short uORF in mRNA's 5′ UTR generally leads to significant reduction in expression of a main ORF [2] , the degree of which depends on the “strength” of the nucleotide context surrounding the uORF's initiating AUG ( called the Kozak consensus sequence ) [3] . Short uORFs with a relatively poor initiation context can be skipped by at least some 48S PICs via leaky scanning , which decreases their inhibitory impact . On the other hand , there is growing evidence that there are many non-AUG-initiating short uORFs that , if in a good context , may serve as very potent inhibitors [4] , [5] . Short uORFs may also down-regulate expression of a main ORF by their special ability to mediate ribosome stalling at coding or termination codons , or by influencing the mRNA stability through the Nonsense Mediated Decay ( NMD ) pathway ( reviewed in [6] ) . On the other side of the spectrum of short regulatory uORFs are those that permit the small ribosomal subunit to stay mRNA-bound post-termination and resume scanning for efficient reinitiation ( REI ) downstream . It has been shown that the ability of some uORFs to retain the 40S subunit on the same mRNA molecule after it has terminated translation at the uORF's stop codon depends on: ( i ) cis–acting mRNA features , ( ii ) the time required for the uORF translation , which is determined by the relative length of a short uORF and the translation elongation rates , and ( iii ) on various initiation factors ( for review see [6]–[8] ) . The last two requirements are united in the idea that eIFs important for promoting reinitiation remain at least transiently associated with the elongating ribosome , and that increasing the uORF length or the ribosome transit time increases the likelihood that these factors are dropped off [9] . There is now genetic evidence for this hypothesis showing that in yeast S . cerevisiae eIF3 remains 80S-bound for several rounds of elongation and critically enhances the REI capacity of post-termination 40S ribosomes [10] ( see also below ) . With respect to cis-acting features , with the exception of the uORF-mediated translational control of the budding yeast GCN4 described below , there is virtually nothing known about what other REI-promoting mRNA features are required . Finally , REI efficiency is also directly dependent on ( iv ) the distance between the uORF termination codon and a downstream initiation codon owing to the fact that the rescanning PICs require a certain time for de novo recruitment of the eIF2•GTP•Met-tRNAiMet ternary complex ( TC ) to be able to decode the next AUG start site [11] . The GCN4 mRNA encodes a transcriptional activator of mainly amino acid biosynthetic genes and its leader sequence contains four short uORFs ( Figure 1A ) . Independent of amino acid availability , most ribosomes translate the first REI-permissive uORF ( uORF1 ) and , following termination , about a half of them resumes scanning downstream . When amino acid levels are high , re-scanning ribosomes reacquire the TC relatively rapidly afterward and preferentially reinitiate at one of the last three uORFs , none of which supports efficient REI ( see our model in Figure 1A ) . When amino acid levels are low , deacylated tRNAs accumulate , activating the eIF2α kinase GCN2 . As a result , the TC levels are decreased and the re-scanning ribosomes must travel for a longer period till they have rebound the TC . This significantly increases the likelihood of bypassing all three REI-nonpermissive uORFs to reach the GCN4′s start codon . Thus , whereas the global protein synthesis is significantly down-regulated , translational expression of GCN4 is concurrently induced ( derepressed ) . A failure to derepress GCN4 expression is called the Gcn- ( general control nonderepressible ) phenotype . A similar regulatory mechanism has been also shown to govern expression of for example the mammalian functional homologue of GCN4 , the ATF4 transcription factor [12] . The pressing question of why ribosomes readily reinitiate after translation of uORF1 but not the other uORFs has baffled the translational field for many years . Mutational analyses indicated that AU-rich sequences surrounding the stop codon of uORF1 ( dubbed the 3′ enhancer herein ) might favor resumption of scanning and REI [13] ( Figure 1B ) . In addition , sequences 5′ of uORF1 were also shown to be critical for efficient REI [14] ( Figure 1B ) . In contrast to the 3′ enhancer , the molecular mechanism of which remains to be elucidated , the molecular contribution of the 5′ sequences has been recently proposed on the basis of our characterization of the N-terminal truncation of the a/TIF32 subunit of eIF3 [10] . The N-terminal domain ( NTD ) of a/TIF32 was previously shown to interact with the small ribosomal protein RPS0A in vitro [15] , and we subsequently found that the N-terminal truncation in a/tif32-Δ8 severely reduced association of eIF3 and its associated eIFs with the small ribosomal subunit in vivo [10] . ( RPS0A is positioned near the mRNA exit pore on the solvent side of the 40S subunit [16] ) . Unexpectedly , however , a/tif32-Δ8 also produced a severe Gcn- phenotype as it failed to up-regulate GCN4 expression under starvation conditions by preventing the post-termination ribosomes from resuming scanning downstream of the uORF1's stop codon . Detailed genetic analysis suggested that besides RPS0A , the a/TIF32-NTD also interacts with a yet to be identified element ( s ) within the uORF1's 5′ sequences . Together our findings led to a working model in which wild-type eIF3 remains at least transiently associated with the translating 80S ribosome , and if it does not drop off prior to termination , the a/TIF32-NTD interacts with the 5′ sequences to permit ribosomal recycling of only the large 60S subunit while aiding to preserve the small subunit on the GCN4 mRNA [10] ( Figure 1A and 1B ) . This last step serves as a critical prerequisite for subsequent resumption of scanning by the 40S subunit for REI downstream . Interestingly , we have only recently showed that the eIF3g/TIF35 subunit of yeast eIF3 also critically contributes to this process , but the mechanism of its involvement seems to differ from that of a/TIF32 [17] . Besides the uORF1 of GCN4 , there is another well described example of a REI-permissive uORF in yeast represented by uORF of the YAP1 gene , an AP1-like transcription factor [18] . The intriguing question is whether the molecular aspects of its reinitiation mechanism are similar to that of GCN4's uORF1 , which could indicate a broad mechanistic conservation of reinitiation on short uORFs . In this study we have subjected the ∼220-nt long 5′ sequences of uORF1 as well as the first 200 amino acid residues of the a/TIF32-NTD to an in-depth mutational analysis to identify specific elements/regions required for their common REI-promoting activity . Four such elements designated REI-promoting elements ( RPEs ) are described that together make up what we now call the 5′ enhancer . In addition , two distal regions within the NTD of a/TIF32 were identified and shown to promote REI in the 5′ enhancer-dependent manner . Enhanced computer modeling taking into account a progressive character of mRNA folding combined with classical enzymatic probing surprisingly revealed that the 5′ enhancer contains only two well-defined structural features in a 9-nt long stem and a double-circle hairpin representing the RPEs ii . and iv . , respectively . Strikingly , a similar structural motif working in concert with the a/TIF32-NTD was also found upstream of the REI-permissive uORF of YAP1 . These findings thus strongly suggest existence of a conserved short uORF-mediated mechanism of reinitiation , whereby the a/TIF32-NTD of the post-termination 80S-bound eIF3 must contact the specifically folded cis-acting REI-promoting elements 5′ of uORF in order to facilitate efficient resumption of scanning of the 40S ribosomal subunit . A considerable difference in efficiency of resumption of scanning following translation of uORF1 versus uORF4 in the GCN4 mRNA leader is thought to be attributable to the distinct sequences surrounding the termination codons of these two uORFs . Replacing the last codon and 10 nt downstream of the uORF1 stop codon ( Figure 1B , dubbed the 3′ enhancer ) with the corresponding nucleotides from uORF4 was sufficient to make uORF1 as inhibitory for REI on GCN4 as is uORF4 [13] . Similarly , sequences located in the leader region >20 nt upstream of the AUG start codon of uORF1 ( Figure 1B ) were also shown to be critically required for efficient REI downstream [14] . However , individual contributions of both of these stimulatory sequences to the overall REI efficiency have never been directly compared in a single experiment . To do that , we divided the two uORFs and their surrounding sequences into four segments: segment A ( 166 bp in length from position -181 to -16 relative to the AUG start codon corresponding to the 5′ REI-promoting sequences of uORF1 ) ; segment B ( 15-bp long segment ( −15 to −1 ) designated previously as linker [10] ) ; segment C ( 3 coding triplets and a termination codon ) ; and segment D ( 25 bp downstream from the uORF stop codon including the aforementioned 3′ enhancer of uORF1 ) ( Figure 1C ) . It should be noted that the A segment of uORF4 has the start codons of the preceding uORFs 2 and 3 mutated out to compare the effects of only uORFs 1 and 4 . Also , in contrast to A , C and D segments , the sequence corresponding to the B-linker region of uORF1 was previously shown to play a negligible role for efficient REI [19] . Three hybrid uORFs were constructed by the substitution of some or all of uORF1 segments with the corresponding segments derived from uORF4 in the GCN4-lacZ construct lacking all three uORFs naturally occurring downstream of uORF1 ( compare Figure 1A and 1C ) . When all four uORF1 segments were replaced by the corresponding uORF4 segments ( Figure 1D; row 2 ( construct 4-4-4-4 ) ) , the GCN4-lacZ expression dropped by ∼20-fold to the background level ( Figure 1D , row 2 [bg] versus 1 [wt] ) in accord with previous findings demonstrating the two uORFs' highly disparate capacities to promote efficient REI [19] . Selective replacements of either the 5′ sequences or the entire 3′ enhancer ( row 4 ( construct 4-4-1-1 ) versus row 3 ( 1-1-4-4 ) ) of uORF1 resulted in 6-fold or 17-fold reductions in β-galactosidase activities , respectively . These data indicate that both elements closely co-operate to promote highly effective REI downstream of uORF1 , but probably by mechanistically distinct processes . Interestingly , whereas the 3′ enhancer is sufficient to stimulate resumption of scanning to at least some degree ( by ∼13% after background subtraction ) , the 5′ sequences are not . This fact could imply that the 3′ enhancer acts first and its stimulatory activity is required for the subsequent action of the 5′ enhancing sequences . It is important to note that the transfer of both sequence elements into the sequence context of REI-nonpermissive uORF4 converts it into a REI-permissive uORF [19] . Hence the mechanism of their combined action appears to be general , not specific to uORF1 only . Whereas the molecular mechanism by which the 3′ enhancer promotes REI is unknown , our recent genetic epistasis analysis suggested that the 5′ sequences ( in segment A ) emerging from the 40S mRNA exit channel promote REI by interacting with the NTD of a/TIF32 upon termination on the uORF1 stop codon . This interaction was proposed to stabilize association of the post-termination 40S subunit with the GCN4 mRNA so that it could resume scanning for REI downstream [10] . Partial deletions of the 5′ sequences in the GCN4-lacZ construct containing solitary uORF1 had severe deleterious effects on efficiency of REI in the wt a/TIF32 background but not in the a/tif32Δ cells expressing a viable a/tif32-Δ8 allele lacking sequences encoding the extreme N-terminal 200 amino acid residues . Given that the 5′ enhancing sequences comprise a rather long stretch of ∼160 nt , however , it is fairly unlikely that such a long segment contacts eIF3 bound to the 40S as a whole . In fact , previously published data suggested that it may consist of at least two critical elements , as deletions of 40 , 80 and 120 nt from nt −21 upstream reduced the GCN4-lacZ expression by a similar fold ( from 2 . 5- to 3-fold ) , whereas the largest deletion of 160 nt resulted in ∼6-fold reduction [14] . In order to precisely map the minimal region ( s ) responsible for the REI-promoting role of the uORF1's 5′ sequences that work in concert with the a/TIF32-NTD , the 5′ sequences were progressively deleted ( beginning at a position −16 nt relative to the uORF1 AUG codon ) in a GCN4-lacZ construct containing solitary uORF1 ( Figure 2A ) . For example , constructs DEL6 and DEL36 had internal deletions of 6 nt ( from −16 to −21 ) and 36 nt ( from −16 to −51 ) , respectively . As a specific background control , the 4-4-1-1 construct devoid of the entire 5′ enhancing sequences ( defined in Figure 1C ) was routinely used ( bg* ) . All deletion constructs were expressed in both the a/TIF32 wt and a/tif32-Δ8 mutant strains and the levels of β-galactosidase activities were measured in at least three independent experiments with three individual transformants in triplicates for each construct . These values were then expressed relative to the value obtained with the wt uORF1-GCN4-lacZ construct that was set to 100% in both strains . The mean values of the resulting percentages ( with standard deviations ) from all experiments were calculated and plotted ( Figure 2B ) . We opted for this percentage expression because it enables a better comparison of the effects of the 5′ sequences deletions on relative β-galactosidase activities independently in each strain . It is important to remember , however , that the a/tif32-Δ8 mutation itself reduces expression of the GCN4-lacZ from the uORF1-GCN4-lacZ constructs by ∼70% when compared to wt a/TIF32 [10] , and the chosen way of data presentation does not reflect this dramatic difference in activities . Owing to this “scaling up” we set a cut-off line of 80% for changes that are considered significant in the a/tif32-Δ8 mutant cells . ( For comparison , the raw , not-normalized data for some of the constructs are shown in Figure S1A and S1B ) . It is also important to note that mRNAs produced from all GCN4-lacZ constructs used throughout the study are highly stable in both wt and a/tif32-Δ8 strains thanks to the fact that they all contain an intact stabilizer element ( STE ) that protects the natural GCN4 mRNA from NMD [10] , [20] ( Figure S1C ) . As shown in Figure 2B , deletions of up to 16 nt from the 3′ end of the 5′ sequences ( DEL6 and DEL16 ) did not produce any significant changes in the GCN4-lacZ expression in the wt cells . In contrast , larger deletions of 26 , 36 , and mainly of 46 nt ( DEL 26 , DEL36 , and DEL46 ) reduced β-galactosidase activities by ∼10% , ∼40% , and ∼60% , respectively . None of the largest deletions ( DEL56 through DEL109 ) decreased the levels of GCN4-lacZ expression any further ( i . e . above 60% of DEL46 ) . In striking contrast to the wt cells , DEL36 had virtually no effect in the a/tif32-Δ8 cells , whereas DEL46 led to a substantial drop in activity ( by ∼40% ) . None of the largest deletions decreased the GCN4-lacZ expression in a/tif32-Δ8 any further , just like in a/TIF32 . Taken together , these results indicate the existence of two REI-promoting elements ( RPE ) falling between nt −31 and −61 . The first element ( RPE i . ; −31 through −51 ) appears to function in the a/TIF32-NTD-dependent manner , since its removal in DEL36 shows genetic epistasis ( non-additive phenotype ) with the a/tif32-Δ8 mutation . The second REI-promoting element ( RPE ii . ; −51 through at least -61 ) , however , operates independently of the a/TIF32-NTD as its deletion together with the RPE i . in DEL46 produced a sharp decrease in β-galactosidase activities in both wt as well as mutant cells . Next we wanted to examine whether the far upstream sequence between nt in positions −143 and −181 constitutes yet another REI-promoting element of the 5′ sequences as originally proposed by Grant and co-workers [14] . Towards this end , we deleted the corresponding region from the wt leader in DELup39 ( Figure 2A ) and observed ∼25% and >30% reductions of activities in wt and a/tif32-Δ8 mutant cells , respectively ( Figure 2B ) . These results thus unambiguously reveal the presence of a third REI-promoting element ( RPE iii . ; −143 through −181 ) in the 5′ sequences of uORF1 that seems to be less potent than the other two and that enhances the efficiency of REI in the a/TIF32-NTD-independent fashion . To conclude , our deletion analysis identified three RPEs that together make up what we designate the 5′ enhancer of uORF1 thereafter . Having identified three RPEs in the 5′ enhancer of uORF1 , we wished to predict a potential secondary structure that the entire 220 nt long segment of the uORF1 5′ UTR might progressively fold into during scanning for , translation elongation of , and termination on uORF1 . Note that we excluded the most 3′ terminal 9 nt from our analysis as they are highly likely buried in the mRNA binding channel of the 80S ribosome terminating at uORF1 [10] . The computer modeling was carried out by the RNA fold software [21] . Our prediction was based on two facts: 1 ) the 5′ enhancer is not a standalone molecule with a rigid structure; its fold forms and changes dynamically as the sequence emerges from the ribosomal mRNA exit pore; and 2 ) the overall underrepresentation of Guanosines ( the nucleotide composition of the entire 5′ UTR of uORF1 is: A 40% , C 22% , G 7% , T 31% ) . Since the Gs are missing especially at the very 5′ end of the sequence , we reasoned that their absence might leave this region unstructured , after it has emerged from the mRNA exit channel , owing to the fact that no local G–C pairs can be formed . To take these assumptions into account in our model , we divided the 5′ UTR of uORF1 into three consecutive segments represented by the extreme 5′ end 66-mer ( AU-rich ) , the middle 81-mer , and the extreme 3′ end 73-mer that is also AU-rich . We first folded the extreme 5′ segment and found that , in agreement with our reasoning , the 66-mer showed no predictions of any secondary structures ( Figure 2C ) . Importantly , it is believed that the AU-rich sequences have a stronger tendency to interact with proteins than those rich in Gs [22] . Hence it is conceivable that the extreme 5′ AU-rich RNA stretch remains unstructured to engage in binding to ribosomal proteins and/or translation factors situated in the vicinity of the mRNA exit pore . Given this potential , we further stipulated that this 66-mer would not directly pair with the downstream sequences gradually leaving the exit channel during ribosomal scanning . To account for this , we added the middle 81-mer to the 66-mer and modeled the folding of the resulting 147-mer by blocking potential contacts between both individual segments . As a result , a short double-circle hairpin relatively GC-rich was predicted to form at the very 3′ end of the 147-mer ( Figure 2C ) . Interestingly , the same hairpin formed when the complete sequence of the 5′ UTR of uORF1 was analyzed by RNA fold without any restraints ( data not shown ) , and , furthermore , when homologues sequences from numerous yeast species were subjected to computer modeling ( JP and LV , unpublished observations ) . These results indicate that the double-circle hairpin is a conserved structure , at least among various yeasts , that may have a functional significance in the translational control mechanism of GCN4 ( see below ) . Finally , we added the remaining extreme 3′ end segment to the pre-folded 147-mer and sought predictions of the overall structure of the 5′ sequences . As shown in Figure 2C , the 73-mer remained mostly unfolded with the exception of a 9-nt long stem loop , situated only 6 nt downstream of the 3′ end of the double-circle hairpin , with one 3-nt topical bulge and one 1-nt bulge close to its 3′ end . Taken together with our genetic deletion analysis presented above , we propose that both the RPE i . and RPE iii . remain unstructured , whereas the RPE ii . folds into a stable stem loop with two bulges ( Figure 2C ) . To test our computer predictions experimentally , we subjected a commercially synthesized 79-mer containing both the double-circle hairpin and the RPE ii . stem to enzymatic probing . ( The 79-mer that was chosen based on RNA fold predictions starts 2 nt before the hairpin and ends 2 nt after the RPE ii . stem ( Figure 3A ) . ) The 79-mer was 5′-end labelled by T4 polynucleotide kinase with [32P]-γATP , heated at 90°C for 3 minutes , slowly cooled down to room temperature to stimulate proper re-folding , and probed by RNases T1 and V1 prior to analysis on denaturing polyacrylamide gels . As shown in Figure 3B , the data for enzymatic probing were in good agreement with the computationally predicted secondary structure of this 5′ enhancer section . Formation of all three stems , two of which occur in the double-circle hairpin [nt 3–7 base-paired with nt 45–49; and nt 22–24 base-paired with nt 31–33] , and the third forms the RPE ii [nt 56–64 base-paired with nt 68–77] , was confirmed by specific cleavages by RNase V1 ( cuts based-paired nucleotides only; lane V1 ) . As expected , V1 cuts of the RPE ii . stem are preferentially detected in the strand that is more proximal to the 5′-radiolabel . On the other hand , V1 cuts are only detected in the more distal strand of the longer stem of the double-circle hairpin owing to the fact that the other strand is too close to the 5′ end label ( nt 3–7 ) . Since all four G's that are distal to the 5′-radiolabel ( namely G23 , G31 , G48 , and G75 ) were predicted to occur in the based paired regions , no cleavages with RNAse T1 ( cleaves at 3′ end of single-strand G's ) should be detected . The fact that we did reproducibly observe cuts at all four G's ( lanes T1 ) suggests that the 79-mer is metastable , undergoing dynamic unfolding/folding cycles in our sample . This is expected , however , given that the REI process requires the ribosome to smoothly scan through this region before it translates uORF1 , stops at its stop codon and primes itself for resumption of scanning . It is understood that under given circumstances a highly stable secondary structure would actually impede swift translational remodeling of this critical region . Indeed , a critical support for the proposed structure identity was provided by the T1 enzyme under denaturing conditions ( lane T1 denatur ) that showed a substantially stronger T1 cuts compared to the folded sample ( lane T1 fold ) . Next we subjected individual RPEs to an in-depth analysis in order to provide additional support for their importance in the REI mechanism of GCN4 . The RPE i . acts in the a/TIF32-NTD-dependent manner and appears to be unstructured . Hence it is highly likely that the putative direct interaction between the a/TIF32-NTD and the RPE i . is sequence specific . To test that , we divided the RPE i . into three consecutive segments with the first two comprising 9 nt ( −31 through −39 in SUB31; and −40 through −48 in SUB40 ) , and the third one being composed of 6 nt ( −49 through −54 in SUB49 ) and ending at the base of the RPE ii stem ( Figure 4A and Figure 2C ) . We then substituted sequences of these segments with complementary nt and tested the resulting constructs for efficiency of GCN4-lacZ expression . As shown in Figure 4D , whereas neither of the substitutions significantly affected expression in the a/tif32-Δ8 cells , SUB31 produced ∼25% , and SUB40 and SUB49 even ∼40% reductions , respectively , in wt cells . Hence the results obtained especially with the latter two substitutions nicely correlate with DEL36 that removes the entire element ( Figure 2B and Figure 4F ) and suggest that mainly the nature of nt situated at the 5′ end of the RPE i . is critical for its function in REI . The RPE ii . forms a stem with two bulges and does not seem to be involved in the functional interaction of the 5′ enhancer with the a/TIF32-NTD . We designed two constructs one of which removed all stem-forming nt and the other one replaced them with a stretch of multiple CAA triplets , which minimizes formation of secondary structures [23] ( Figure 4B ) . As predicted , both constructs reduced the GCN4-lacZ expression by ∼40% in wt as well as in a/tif32-Δ8 cells clearly confirming the importance of this element for resumption of scanning after uORF1 in the a/TIF32-NTD-independent fashion . We also swapped both strands of the stem either preserving the sequences of both bulges or replacing them with complementary nt to find out whether the structure or sequence , or both is important . In either case the GCN4-lacZ expression went down by consistent ∼40% in both strains ( data not shown ) , suggesting that certainly the sequence is critical for function of this element . The question of the fold importance could not be satisfactorily answered . As shown in Figure 2B , removal of the RPEs i . and ii . in DEL46 ( −16 through −61 ) produced ∼60% drop in the β-galactosidase activity in wt cells and any of the larger deletions up to −125 nt that we tested did not make it any worse . These findings may indicate that a nucleotide sequence from the 5′ base of the RPE ii . stem ( nt −76 ) upstream ( at least up to nt −125 ) is dispensable for the 5′ enhancer function in REI . Interestingly , however , our computer modeling suggested that a nucleotide stretch spanning nt −129 through −83 folds into the conserved double-circle hairpin ( Figure 2C ) that , by definition , would be expected to be functionally important . To test that , we employed computer modeling and designed a triple nucleotide substitution ( C-129A , G-128A , G-109C ) that should completely disrupt base-pairing between nt forming both stems while preserving the length and the rest of the sequence of this rather long segment intact ( Figure 4C ) . As shown in Figure 4F , the resulting AA-C construct indeed reduced the GCN4-lacZ expression by ∼40% but only in the wt cells . In principle , it behaved the same as the RPE i . -deletion construct DEL36 indicating that the RPE i . and this hairpin may closely cooperate with each other and also with the NTD of a/TIF32 . If true , then combining DEL36 and AA-C mutations ( Figure 4C ) should be epistatic; and this was exactly observed ( Figure 4F ) . These findings thus identify a fourth REI-promoting element ( RPE iv; −129 through −83 ) within the 5′ enhancer that adopts a conserved higher-order structure and acts in synergy with the RPE i . and the a/TIF32-NTD . All experiments described so far were carried out with GCN4-lacZ constructs carrying only uORF1 of the four uORFs from the GCN4 mRNA leader and under non-starvation conditions . To perform an ultimate test of our findings , we examined effects of selected mutations on GCN4 induction in wt cells treated with 3-aminotriazole ( 3-AT; an inhibitor of histidine biosynthetic genes that mimics starvation conditions ) using a construct containing uORF1 and uORF4 that together suffice for wt regulation of GCN4 expression ( Figure S2 ) . As described in detail in the Text S1 , obtained results underpinned the functional importance of all three major 5′ enhancer's RPEs ( i . , ii . , and iv . ) in their task to ensure efficient REI on GCN4 when cells are starved for nutrients such as amino acids . The a/tif32-Δ8 mutation was shown to reduce the REI efficiency by two distinct mechanisms: ( i ) decreasing retention of eIF3 on elongating ribosomes translating uORF1 by reducing the binding affinity of eIF3 to 40S subunits and ( ii ) impairing functional interaction of a/TIF32 with the 5′ enhancer of uORF1 [10] . To identify residues in the extreme NTD of a/TIF32 that are responsible for these two roles , and to possibly separate them , we introduced Ala substitutions in consecutive blocks of 10 residues between amino acids 1 and 200 ( dubbed Boxes 1 to 20 , Figure 5A ) . None of these mutations was lethal and only Boxes 6 ( residues 51–60 ) , 8 ( 71–80 ) , and 17 ( 161–170 ) produced slow-growth ( Slg− ) phenotypes and , most importantly , significant Gcn− phenotypes ( Figure 5B ) indicating an impairment of the GCN4 induction . Indeed , our GCN4-lacZ reporter assays with the wt GCN4-leader confirmed the derepression defect ( Figure 5C , construct i ) . Interestingly , combining Boxes 6+17 and 8+17 but not 6+8 exacerbated both the Slg− and Gcn− phenotypes of the single mutants ( Figure 5B and 5C , construct i ) suggesting the presence of two functionally partially redundant regions within the a/TIF32-NTD , with the first one represented by Boxes 6 and 8 , and the other by Box17 . Importantly , in a striking analogy with the a/tif32-Δ8 mutation [10] , all three Boxes as well as their combinations decreased β-galactosidase activities measured from constructs carrying only uORF1 at three different positions relative to GCN4-lacZ by a similar number ( ∼50–80% ) ( Figure 5C , constructs ii . – vi . ) strongly indicating that the failure to induce GCN4 expression emanates from the inability of 40S subunits to resume scanning after translating uORF1 . Remarkably , in contrast to a/tif32-Δ8 , neither of the Boxes either alone or in pair wise combinations affected the overall eIF3 affinity for 40S subunits in vivo ( Figure S3A and data not shown ) . Furthermore , binding of the in vitro synthesized a/TIF32-NTD to GST-fused RPS0A was also not affected by these mutations ( Figure S3B ) . Together these findings strongly suggest that the a/tif32-Boxes impact REI specifically by impairing the a/TIF32-NTD interaction with the 5′ enhancer . To demonstrate directly that the amino acid regions represented by the latter Boxes mediate the REI-promoting interaction between the a/TIF32-NTD and the 5′ enhancer , we analyzed β-galactosidase activities of the selected GCN4-lacZ constructs described in Figure 2 and Figure 4 eliminating the key RPEs in the background of the Box6+17 and Box8+17 mutations ( Figure 5D and data not shown ) . Whereas neither DEL36 , SUB40 and SUB49 ( impairing RPE i . ) nor AA-C and DEL36+AA-C ( impairing RPE iv . either alone or together with RPE i . ) significantly exacerbated deleterious effects of the double-Box mutations on REI efficiency in the mutant cells , CAAII impairing eIF3-independent RPE ii . showed an additive effect when combined with either of the double-Box mutations . These results thus clearly corroborate identification of the two critical 5′ enhancer-dependent regions that together account for the REI-promoting activity of the a/TIF32-NTD independently of its 40S-binding activity . Next we asked whether the just described mRNA and protein features required for efficient REI on the GCN4 mRNA are unique to its uORF1 . We took advantage of two genes , YAP1 and YAP2 , both encoding stress related transcription factors , the mRNA leaders of which contain short uORF ( s ) with well described regulatory roles . Whereas the YAP1's uORF permits post-termination 40S ribosomes to efficiently resume scanning for REI on the main ORF ( similar to GCN4's uORF1 ) , the uORFs 1 and 2 of YAP2 act to block ribosomal scanning after their translation by promoting efficient termination followed by rapid mRNA decay [18] . To our knowledge the uORF of YAP1 is the only short uORF in yeast experimentally proven to promote efficient REI besides GCN4's uORF1; however , in contrast to GCN4 , the exact link between its REI-mediated translational control mechanism and its stress-protective cellular role ( s ) is still not fully understood . We first computationally predicted potential secondary structures of the 5′ sequences of YAP1's uORF ( −81 to −1 ) and of YAP2's uORF1 ( −101 to −4 ) occuring behind the trailing edge ( the mRNA exit channel ) of the post-termination 40S ribosome , using an analogous folding model as that described for the GCN4 5′ sequences above . The predicted secondary structures were compared with that occurring in the corresponding region of GCN4′s uORF1 ( −131 to −10 ) ( Figure 6A ) . The structure similarities , computed using the RNA distance program [21] , revealed a remarkable resemblance between predicted secondary structures of 5′ sequences of YAP1's uORF and RPEs of GCN4's uORF1; the similarity score reached the value of 35 ( compared to 46 for YAP2 versus GCN4; the higher the number , the lower the similarity ) , which is highly significant considering that the compared sequences are fairly short ( ∼90 nt ) . It mainly arises from ( i ) the occurrence of a double-circle hairpin and ( ii ) similar lengths of unstructured sequences indicating congruent positioning of the structured elements in the overall folds . It is worth noting that no significant sequence similarities were observed ( data not shown ) suggesting that these particular structural features might truly play an important role in the REI mechanism . To examine that , we replaced the entire 5′ leader of uORF1 of GCN4 excluding the promoter region with the corresponding sequences from both YAP genes in our GCN4-lacZ construct containing solitary uORF1 ( Figure 6B ) and measured β-galactosidase activities in wt as well as a/tif32-Δ8 cells . Whereas the 5′ leader of uORF1 of YAP2 ( in Y2-uORF1 ) showed background levels in both strains , as expected , the 5′ sequences of uORF of YAP1 ( in Y1-uORF1 ) stimulated the GCN4-lacZ expression by ∼2-fold over the background in wt cells ( Figure 6C ) . Strikingly , this activity dropped by ∼80% in a/tif32-Δ8 . The similar reduction was also obtained when we fused the YAP1 gene with its intact 5′ leader with lacZ ( in Y1-lacZ ) . In contrast , a lacZ fusion with the YAP2 gene containing its natural 5′ leader ( in Y2-lacZ ) showed no β-galactosidase activity at all in accord with previous observations implicating both uORFs of the YAP2 mRNA in promoting its rapid degradation [18] . Importantly , point mutations designed to disrupt the conserved double-circle hairpin ( in Y1-uORF1-hairpin_G-45U C-57A ) reduced the Y1-uORF1 activity by ∼30% in wt cells and showed the epistatic interaction with a/tif32-Δ8 ( Figure 6D ) , in good agreement with the data presented in Figure 4F . In contrast , mutations disrupting the predicted non-conserved bulged-stem ( in Y1-uORF1-“stem”_C-32G G-33C ) showed no reduction in either of the strains indicating that it is either not functionally important or not affected by our mutations . Taken together these results clearly demonstrate that the specifically structured 5′ enhancers of REI-permissive uORFs of GCN4 and YAP1 are at least partially functionally interchangeable and critically require the NTD of a/TIF32 for their function . Hence a possibility for a common mechanism of translational control operating on short REI-permissive uORFs seems highly likely . We first tested the individual contributions of both uORF1's enhancers on efficiency of REI by their individual replacements with the corresponding sequences of the REI-nonpermissive uORF4 . Previously , a similar cassette replacement mutagenesis was carried out [19]; however , it did not include the uORF1's 5′ sequences . In accordance with Grant et al . [14] , the uORF1's 3′ enhancer alone ( Figure 1D; construct 4411 ) was still capable to allow some REI on GCN4-lacZ ( ∼4-fold higher than the background control in uORF4; construct 4444 ) , albeit the overall REI activity was strongly reduced by ∼80% when compared to wt ( construct 1111 ) . On the contrary , the REI activity of the 5′ enhancer alone containing all four RPEs ( construct 1144 ) dropped to the background levels of uORF4 . Thus rather than making additive contributions to the uORF1's ability to support a high frequency of REI at GCN4 , as originally proposed , it seems that the 3′ enhancer acts first and its action is a prerequisite for the subsequent contribution of the 5′ enhancer . We propose the following model of the sequence of events on the uORF1 that follow termination of its translation and that , in the light of our YAP1 data , could be applicable to short uORFs with the REI-permissive character in general ( Figure 1A and 1B ) . Upon stop codon recognition , the 3′ enhancer , buried for its most part in the mRNA binding channel , interacts somehow with the ribosome and ensures that the 40S subunit remains attached to the mRNA during the first ribosomal recycling reaction that removes the large ribosomal subunit and is thought to be catalyzed by RLI1/ABCE1 [30] . This alone suffices for a certain level of elevated efficiency of REI . In the meantime , the 5′ enhancer that has gradually emerged from the mRNA exit channel progressively folds into its secondary/tertiary structure and contacts the a/TIF32-NTD , previously shown to interact with RPS0A occurring near the mRNA exit pore [10] , [15] , to further stabilize the 40S subunit on the GCN4 mRNA . This second step considerably boosts the efficiency of REI as it prevents recycling of at least 50% of small subunits [14] . Consistent with our model , mammalian eIF3a was shown to interact with mRNA in the 48S PIC in a way extending the mRNA binding channel beyond the exit site [31] . In addition to a/TIF32 , the g/TIF35 subunit of eIF3 also promotes this process , however , by an unknown mechanism that does not depend on the 5′ enhancer and awaits a detailed investigation [17] . Interestingly , plant eIF3g together with eIF3h were similarly shown to support efficient REI [32] , [33] , however , their mechanistic contributions also remain to be explored . Once the mRNA-40S complex is sufficiently stabilized , eIF3 most probably facilitates recruitment of scanning-promoting factors namely eIF1 and eIF1A . These factors were shown to trigger conformational changes of the 40S head region resulting in the open/scanning conducive conformation that is required for linear scanning from the mRNA′s 5′ cap [34] . It is very likely that similar conformational changes are also needed for the mRNA-bound post-termination 40S subunit in order to resume scanning . How the 3′ enhancer performs its initial task is currently under investigation in our laboratory . Previous work suggested that its AU-rich content ( ∼60% ) rather than a particular sequence could be critical for its function . In fact , it was proposed that the AU-rich sequence would not form strong base-pairing interactions with the 40S subunit and would allow it to promptly resume scanning [13] . However , with the exception of uORF4 ( AU-content ∼40% ) , the sequences corresponding to the 3′ enhancer of other two GCN4′s uORFs ( 2 and 3 ) have even higher AU-content ( ∼85% and ∼70% , respectively ) , yet they do not promote REI as uORF1 . Besides , our model posits that the ribosome terminating on uORF1 spends longer than usual time on the termination/recycling steps to allow the 5′ enhancer to fold and interact with the a/TIF32-NTD . Hence we think that a simple enrichment in A and U nt is unlikely to be the key to this puzzle , and it is still possible that the 3′ enhancer contains a less stringent sequential motif that contacts some component of the post-termination complex , presumably 18S rRNA ( Figure 1B ) . If true , this mechanism would bear a significant resemblance to the termination/reinitiation mechanism that is the best described for the polycistronic mRNA of feline calicivirus [35] , [36] . A specific 87-nt element ( called TURBS ) preceding the overlapping termination/initiation site of two long ORFs 2 and 3 folds into a specific secondary structure that in fact resembles our double-circle hairpin . A part of this structure interacts with a complementary segment of 18S rRNA and also with eIF3 via several subunits including eIF3a and eIF3g to prevent dissociation of the mRNA/eIF3/40S complex in order to allow efficient REI on ORF3 . Even though this system operates on long ORFs , its mechanistic likeness with the short uORF-mediated REI does not seem to be accidental from the evolutionary point of view . The original data by the Hinnebusch′s group suggested that the ∼160 nt-long 5′ sequences may contain two critical REI-promoting motifs [14] . In agreement , we identified not only two but together four individual elements denoted RPEs that together account for the stimulatory effect of the uORF1's 5′ enhancer on REI . Individual mutations of the unstructured RPE i . and the structured RPE iv . , as well as the combination of these mutations were found to be epistatic with the a/tif32-Δ8 mutant . These results clearly suggest that both elements are needed to contact the a/TIF32-NTD and thus it seems conceivable that they might fold together in a higher-order structure . The fact that the RPE iv . is structurally conserved among various yeasts ( JP and LV , unpublished observations ) may suggest that the RPE iv . provides a structural basis for the 5′ enhancer–a/TIF32 interaction , whereas the RPE i . lends a specificity to it . Whether it is a direct interaction is currently being explored in our laboratory in the living cells . In addition to that , our genetic epistasis experiments revealed that the RPEs i . and iv . interact with the a/TIF32-NTD via its two relatively distal REI-promoting regions represented by Boxes 6 and 8 , and by Box17 , respectively ( Figure 5 ) . Importantly , neither of these regions mediates a direct contact of a/TIF32-NTD with RPS0A to facilitate eIF3 binding to 40S ribosomal subunits in vivo ( Figure S3 ) clearly suggesting that they promote efficient REI solely in the 5′ enhancer-dependent manner . Interestingly , unlike in the case of RPEs i . and iv . , combination of mutations in both of these REI-promoting regions exacerbated the effect of the individual mutations . Hence it seems likely that even though each region may contact the 5′ enhancer individually , their mutual co-operation is required to establish a strong interaction . Mutations in RPEs ii . and iii . showed additive effects when combined with a/tif32-Δ8 indicating that the molecular mechanism of their involvement in REI differs from that of RPEs i . and iv . The model structure predicts that the RPE ii . forms a 9 nt-long stem loop whose sequence and less likely also the structure are crucial for its stimulatory activity . At present we can only speculate about the molecular nature of the roles of these two RPEs . They could either contact other eIF3 subunits or other eIFs , or act independently , for example by interacting directly with the ribosomal components . Even though there is an increasing number of short uORFs demonstrated to permit efficient REI after their translation [12] , [28] , [37] , [38] , perhaps none of them , besides uORF1 of GCN4 [8] , [10] , has been studied deeply enough to draw any general conclusions regarding the molecular details of the short uORF-mediated REI mechanism . Until now , this has also applied on the only other well defined REI-permissive uORF in yeast occuring in the mRNA leader of the transcription factor YAP1 [18] . Here we showed that its 5′ sequences share significant structural similarity predictions with GCN4′s uORF1 and , most importantly , stimulate REI on YAP1 in a strict dependency on the NTD of a/TIF32 ( Figure 6 ) . Hence the functional if not direct interaction between the a/TIF32-NTD and the specifically folded sequences upstream of a short REI-permissive uORF represents the first generally applicable requirement of this type of a regulatory mechanism described to date , at least in yeast . Considering the remarkable similarity with the aforementioned termination/reinitiation mechanism utilized by viruses , it is very likely that the analogous principles apply also to uORFs promoting efficient REI in higher eukaryotes . Future work exploring the mechanistic details of some of these uORFs , especially those connected with pathophysiological mechanisms , will certainly tell us more about the evolutionary conservation of this important translational control process . Lists of strains ( Table S1 ) , plasmids ( Table S2 ) , and PCR primers ( Table S3 ) used in this study and details of their construction can be found in the Text S1 . Commercially synthesized 79-mer RNA ( East Port ) was 5′-end-labeled using γ-32P-ATP and T4 Polynucleotide Kinase ( Fermentas ) . Free radioactive nucleotides were removed by NucAway Spin Columns ( Ambion ) . The RNA was then subjected to limited digestion with RNase T1 ( cleaves after single-stranded G residues ) or RNase V1 ( cleaves within double-stranded RNA ) . RNase T1 was used in RNA Sequencing buffer or in RNA Structure buffer to induce denaturing ( denatur ) or folding-promoting ( fold ) conditions , respectively . Alkaline hydrolysis of the RNA was used to generate appropriate reference landmarks . ( All enzymes , buffers and protocols were provided by Ambion ) . The digested products were then separated on 10% polyacrylamide ( 8M urea ) sequencing gel in 1xTBE buffer . β-galactosidase assays were conducted as described previously [13] . GST-pull-down experiments , preparation of whole-cell extracts , sucrose gradient separations and Western blot analysis of gradient fractions were essentially conducted as described in [39] , [40] .	Protein synthesis is a fundamental mechanism capturing the rejuvenation of DNA–encoded genetic information by its translation into molecular effectors—proteins . Its regulation can be used to change the protein content and thus to adapt a cell to changing environmental conditions . Translation requires mRNAs delivering genetic information of corresponding genes , tRNAs carrying amino-acids , ribosomes as the molecular translators , and accessory proteins/factors facilitating the entire process . There are numerous regulatory mechanisms that modulate translation at its various stages . Here we describe one such a translational control mechanism called reinitiation . Most eukaryotic mRNAs contain only a single translatable gene ( ORF ) ; however , in many of them this gene is preceded by a short coding sequence ( uORF ) that is in some cases translated first . In order to reinitiate translation on the downstream main ORF , a ribosome has to stay bound to mRNA after it has terminated short uORF translation . This requires a concerted action of specific mRNA elements surrounding the uORF and selected initiation factors . Our results delineate how these key players interact with each other and suggest a sequence of general events that ought to take place on short uORF to enable the ribosome to reach and translate the main ORF downstream .	[ "Abstract", "Introduction", "Results", "Discussion", "Materials", "and", "Methods" ]	[ "genetic", "mutation", "gene", "regulation", "gene", "function", "protein", "synthesis", "molecular", "genetics", "sequence", "analysis", "proteins", "gene", "expression", "biology", "molecular", "biology", "mutagenesis", "biochemistry", "rna", "nucleic", "acids", "prot...	2011	Translation Reinitiation Relies on the Interaction between eIF3a/TIF32 and Progressively Folded cis-Acting mRNA Elements Preceding Short uORFs
We analyze the problem of obstacle avoidance from a Bayesian decision-theoretic perspective using an experimental task in which reaches around a virtual obstacle were made toward targets on an upright monitor . Subjects received monetary rewards for touching the target and incurred losses for accidentally touching the intervening obstacle . The locations of target-obstacle pairs within the workspace were varied from trial to trial . We compared human performance to that of a Bayesian ideal movement planner ( who chooses motor strategies maximizing expected gain ) using the Dominance Test employed in Hudson et al . ( 2007 ) . The ideal movement planner suffers from the same sources of noise as the human , but selects movement plans that maximize expected gain in the presence of that noise . We find good agreement between the predictions of the model and actual performance in most but not all experimental conditions . The experimental task illustrated in Figure 1 contains many of the elements of our coffee-cup example , and is reminiscent of the kind of obstacle avoidance behavior that has been studied extensively both in terms of its neurophysiological substrates [3] , [4] , [5] and in identifying sensory/motor factors that influence the movement trajectory [6] , [7] , [8] , [9] , [10] , [11] , [12] , [13] . We will describe it in detail in the next section . To study obstacle-avoidance reaches within the framework of Bayesian decision theory , we translated the above example to one where there is an explicit reward ( ) associated with touching a target and an explicit cost ( ) associated with inadvertently intersecting an obstacle that is placed between the starting point of the hand and the target . Contact with a physical obstacle placed along the reach path might change the physical character of the reach and such an obstacle would constitute an intrinsic cost whose value we could not easily measure or manipulate . To avoid these issues , we used virtual obstacles that could not impede the reach . Although the virtual obstacle is invisible , a visual indication of its leftmost edge ( at ) is presented on the monitor prior to each reach . Figure 1A shows a front view of the experimental apparatus with the virtual obstacle shown in transparent blue . The blue line on the monitor marks its edge ( at ) . The subject incurs the cost if the fingertip passes through the virtual obstacle while reaching toward the target ( centered on , with width ) . One part of training will allow subjects to become familiar with the location of the obstacle in depth and how its edge relates to the visual marker ( the blue line ) . Across experimental conditions we varied the location of the obstacle and target and the cost incurred by passing through the obstacle as described in the next section . In all conditions there was a constant relative distance between the obstacle edge and the center of the target . Figure 1B show the same setup but from an overhead viewpoint . The left and right panels differ in the location of the obstacle-target pair . Seven naive subjects participated in the experiment . Subjects were paid for their time ( $10/hr . ) and also received a bonus based on points earned during the experiment that amounted to $ . 01 per point ( an additional $5–$10 over the hourly rate ) . All participants provided informed consent and research protocols were approved by the local Institutional Review Board . Subjects were seated in a dimly lit room 42 . 5 cm away from a fronto-parallel transparent polycarbonate screen mounted flush to the front of a 21″ computer monitor ( Sony Multiscan G500 , 1920×1440 pixels , 60 Hz ) . Reach trajectories were recorded using a Northern Digital Optotrak 3D motion capture system with two three-camera heads located above-left and above-right of the subject . Subjects wore a ring over the distal joint of the right index finger . A small ( 0 . 75×7 cm ) wing , bent 20 deg at the center , was attached to the ring . Three infrared emitting diodes ( IREDs ) were attached to each half of the wing , the 3D locations of which were tracked by the Optotrak system . Further details of the apparatus are given in a recent report [16] . The experiment was run using the Psychophysics Toolbox software [17] , [18] and the Northern Digital API ( for controlling the Optotrak ) on a Pentium III Dell Precision workstation . Subjects attempted to touch targets on a computer screen , represented visually as a vertical [6 . 5 mm×15 cm] strip , whose locations were chosen randomly and uniformly from a set of three locations [0 , 38 , 75 mm] relative to the monitor center . Rewards and penalties were specified in terms of points . Hits on the target earned subjects two points , and passing through the obstacle incurred a cost of one , two or five points . Missing the target earned no points , and too-slow reaches incurred a cost of ten points . All reaches . All trials proceeded as follows: subjects brought their right index finger to a fixed starting position at the front edge of the table ( 15 cm to the right of screen center ) , triggering the start of the trial . Next , the target ( and obstacle ) was displayed ( Figure 1A ) , followed 50 ms later by a brief tone indicating that subjects could begin their reach when ready . Movement onset was defined as the moment the fingertip crossed a frontal plane 3 mm in front of the table edge , itself located 35 cm from the screen; the fingertip was required to reach the screen within 600 ms of movement onset . Both the fingertip endpoint , the location where the fingertip passed through the plane of the obstacle ( during obstacle practice and experimental reaches ) and a running total of points ( during experimental reaches ) were displayed on-screen at reach completion . Before each experimental session , subjects ( fitted with IREDs ) touched their right index finger ( pointing finger ) to a metal calibration nub located to the right of the screen while the Optotrak recorded the locations of the six IREDs on the finger 150 times . Linear transformations converting a least-squares fit of the three vectors derived from the 3 IREDs on each wing ( left and right; each defining a coordinate frame ) into the fingertip location at the metal nub were computed . During each reach we recorded the 3D positions of all IREDs at 200 Hz and converted them into fingertip location using this transformation . The 3 IREDs on the left and right wings were used to obtain fingertip location independently , and the two estimates were averaged when all IRED locations were available for analysis . This redundancy allowed data to be obtained even if IREDs on one wing or the other were occluded during some portion of a reach . Because we cannot predict the biomechanical costs associated with reach speed and overall length of reach trajectory that might accompany the longer and faster reaches necessary to reach targets within the timeout interval for , e . g . , midline vs . right-of-midline target locations , we restrict the cost function that must be minimized by an optimal reach planner to the target and obstacle costs defined by and . Thus , the only factors entering into the optimal reach plan are fingertip positional uncertainty ( i . e . , the standard deviation of fingertip position in the relevant plane ) , average fingertip coordinates at the two critical planes , , and target and obstacle costs . To compute optimal reach plans , we model the empirical relationship between mean excursion , , and the remaining kinematic variables , the two sample standard deviations , and at the obstacle and target planes , respectively . The relationships were close to linear and we thus fit three lines relating empirical fingertip standard deviation to mean excursion separately for each of the three obstacle positions because we allowed for the possibility that fingertip standard deviations will change differently for excursions around nearby and further-away obstacles . Similarly we fit three lines relating empirical fingertip standard deviation to mean excursion . These six lines allowed us to predict and as a function of any planned excursion . While it is plausible that we could develop a single equation to predict each of the standard deviations , and by incorporating the obstacle location itself we could only do so at the cost of additional assumptions; the equations we use are sufficient for our purposes . After having obtained a function relating excursion size and fingertip uncertainty ( at both the target and obstacle planes , for all three obstacle positions ) , it is possible to predict fingertip standard deviations for theoretical excursions ( ) not observed experimentally , around any of our obstacles . This in turn allows one to compute the expected gain associated with any theoretical excursion . Maximizing the expected gain function yields the prediction of the optimal reach planner ( i . e . , the theoretical excursion maximizing expected gain , ) in each of the 9 conditions of the experiment . In the previous section we outline our method of predicting the obstacle avoidance behavior of an optimal Bayesian reach planner based on modeled changes in uncertainty , both at the obstacle plane and the target plane , of making reaches that deviate from their natural unobstructed trajectory . Because we parameterize the expected gain function in terms of obstacle-plane excursion , we can test the hypothesis that data conform to the predictions of the optimal Bayesian reach planner by comparing predicted and observed ( ) obstacle-plane excursions . Data conforming to the Bayesian ( optimal planning ) model will fall along the identity line of a plot showing observed vs . predicted excursions . Notice that we manipulated value to get the range of data needed to predict the standard deviations and given the planned excursion , and we then use these equations to predict the optimal excursion for each condition . The reader may be concerned that there is an apparent circularity in our use of the Dominance Test . The circularity is only apparent , not actual; This is because , no matter how well the empirical fits ( relating planned excursion to standard deviations and ) fit the data , there is no guarantee that the average excursion ( ) observed in a particular condition , of all possible excursions , will produce the largest possible gain; i . e . , that it happened to fall at the theoretical MEG excursion ( ) for that condition . Suppose , for example , that the subject consistently chose excursions that are 80% of the way between the edge of the obstacle and the theoretical MEG excursion ( ) . While the observer has failed to maximize expected gain in every condition , the fits relating planned excursions to standard deviations and will be little affected . We refer the reader to the second experiment of Hudson et al . [15] , which used a similar Dominance Test and demonstrated such a patterned failure . We compare performance to that predicted by the optimal planning model using standard Bayesian model comparison techniques ( see Supplemental Text S1 ) . This analysis yields a measure of evidence [19] ( given in decibels ) for the optimal planning model ( or conversely , against non-optimal planning models ) , based on the odds ratio comparing the probability of the optimal planning model given the observed data and the probability of any of the non-optimal models on the same data . For example , evidence of between 3 and 4 . 75 dB ( or odds of between 2: and 3∶1 favoring one model over the alternative[s] ) is usually considered a lower bound for statistically significant evidence [see e . g . ] , [ 15] , [16] , [19] , [20 , 21] . Several features of the data can be observed directly in the value diagrams ( Figure 3 ) . First , higher costs lead subjects to avoid the obstacle region by a greater margin: there is an increasing deviation between obstacle-plane crossing points and the obstacle edge as magnitudes increase , across all targets . However , this change in crossing-point is not accompanied by within-target changes in average target-relative endpoints: no matter how great an excursion the finger took around the obstacle , the location of the distribution of target endpoints was unchanged . This relation of endpoint error with target position alone ( i . e . , independent of excursion ) allowed us to model as the average endpoint error in each condition ( ) , regardless of excursion size . In addition , covariance ellipses consistently increase in size as magnitudes increase ( within each target location ) . These four functions , relating changes in positions and standard deviations to magnitude , are plotted in Figure 4 . One can also see a slight positive correlation ( “counterclockwise tilt” ) in value diagram covariance ellipses ( Figure 3 ) . That is , a rightward deviation from the mean in the obstacle plane tends to be paired with a rightward deviation in the target plane . This correlation implies that there is a component of the trial-to-trial variation in trajectories that affects the entire reach , and is therefore detectable at both obstacle and target planes . This tendency is quite small , however , and is ignored in our modeling . We have developed a simple empirical model of the relationship between horizontal excursion within the obstacle plane and horizontal variance . While the model allows us to predict optimal behavior , we make no claims regarding the factors affecting horizontal variance . Our study was not designed to determine the origins of positional uncertainty , a separate and intriguing question . There are very likely many factors that contribute separately to sensory and motor uncertainty and we implicitly assume that those factors ( in our task , direction of gaze , body posture , etc . ) are selected by the visuo-motor system so as to provide the best possible tradeoffs between hitting the target and avoiding the obstacles . To compute optimal reach plans based on the data available in the value diagrams , we re-organize the plots in Figure 4 to predict target- and obstacle-plane fingertip positional uncertainty as functions of the observed obstacle-relative fingertip excursion ( Figures 5A and 5B , respectively ) . Fitting straight-line functions to these data by linear regression ( i . e . , a line was fit to the data from each obstacle condition separately; R2 ranged from 0 . 8 to 0 . 99 ) , we can predict target- and obstacle-plane uncertainties at unobserved fingertip excursions . By varying the theoretical planned excursion ( ) , we compute the expected gain ( Equations 1–3 ) at the obstacle plane ( ) , the target plane ( ) and overall , predicted as a function of any possible ( i . e . , non-positive ) planned obstacle-plane excursion for each obstacle location and magnitude . An illustration of the computation is given in Figure 5C , corresponding to the middle target location and the middle obstacle cost . The maximum of the expected gain curve as a function of theoretical excursion , , corresponds to the excursion in the obstacle plane that maximizes expected gain , denoted . The mean observed excursion across subjects is plotted versus the excursion maximizing expected gain in Figure 5D . The confidence intervals are 95% confidence intervals across subjects . An optimal reach planner would produce data along the identity line of this plot . Overall , the Bayesian evidence measure we computed is 12 . 99 dB ( about 20∶1 odds ) favoring the hypothesis that data do , in fact , fall along the identity line . However , there are deviations when the predicted MEG excursion ( ) is large in magnitude ( leftmost point in Figure 5D ) where the mean observed shift is almost a factor of two smaller than the predicted shift . While human performance for smaller excursions is not far from optimal , there is a clear failure of optimality for the largest predicted excursion . Subjects passed too close to the obstacle in following their trajectory to the target . The optimal reach planning model described here assumes that the distribution on is stationary ( does not change across time ) . We considered the possibility that subjects might employ a within-block “hill-climbing” strategy designed to discover the MEG excursion by initially making too-large excursions around the obstacle and reducing their size over the following few reaches until an appropriate point was found . We verified that this was not the case in the Supplement ( Supplemental Figure S2 ) . There , we show that the distribution of excursions does not vary appreciably over the course of each block . To further investigate the possibility of similar cognitive strategies , we computed autocorrelations for each subject and block up to lag 15 . No significant autocorrelations were found , suggesting that cognitive “contamination” was not present in our results . Reaches have goals . Although particularly obvious when reaching around an obstacle , this aspect of reach planning in the presence of an intervening obstacle has previously been ignored . This has created something of a dilemma for subjects , who must choose how much ‘weight’ to assign to accidentally contacting an obstacle vs . successfully touching the target ( reminiscent of studies where one is instructed to perform a task ‘as quickly as possible without sacrificing accuracy’ ) . Subjects must resolve the conflict created by these contradictory goals by choosing a relative weighting , a weighting that cannot generally be inferred from the data alone . Here , we avoid these problems; obstacles are assigned a cost , giving a clear indication of the relative ‘importance’ of accidentally contacting an obstacle and of contacting the reach target . Not only does our value manipulation allow us to avoid the uncertainty associated with arbitrary target and obstacle weightings that change by subject ( and possibly by experimental condition ) , it is also a necessary element of an optimal model of obstacle-avoidance reach trajectories . The value component of ( 1 ) allows us to quantitatively predict the excursion magnitudes that form the basis of the comparison shown in Figure 5D . This in turn allows us to separate the optimal planning model ( data on the unity line of Figure 5D ) from other models of trajectory planning around the virtual obstacle that might make the same qualitative predictions , but are nevertheless quantitatively sub-optimal ( though not observed , such data would lie along a non-unity-line in Figure 5D ) . Such a separation between qualitative and quantitative optimal performance is demonstrated in Tassinari et al . [28] and in Hudson et al . [15] . Our data have implications for a class of popular models of obstacle avoidance and reach planning in general based on optimal linear feedback control [29] , [30] , [31] . One important prediction of these models is that 2D and 3D variance may be partitioned among the axes to produce the best task performance; for example more precision may be required along the horizontal than along the vertical dimension , as in the current experiments . Such a system is capable of partitioning more variance to the dimension requiring less precision . Here , for the first time , we are looking at a task where variance at two points along the trajectory of a reaching movement affects the outcome of the movement . We find no evidence that subjects partition their covariance in response to rewards or costs . Had they done so , there should have been increased vertical variance , not increased horizontal variance . That is , any manipulation that in fact increased horizontal variance should have been ‘referred’ to the vertical dimension , where it would not have adversely affected performance . We confined analysis to the intersection of trajectories with the obstacle and target planes . The subject's reward is determined by these two points: fingertip position at the intersection of the obstacle and target plane , nothing more . The subject should select a movement plan , , with the criteria that means and covariances in passing through these two critical planes maximize expected gain . Movement plans that satisfy these criteria clearly form a subset of all possible plans , but are they unique ? Does the choice of the movement plan that maximizes expected gain in our experiment determine the entire trajectory bundle ? Or , are there multiple planned trajectories ( , , etc . ) that match in mean and covariance at the two critical planes , but that deviate from elsewhere ? We cannot exclude this possibility nor can we exclude the possibility that a subject chooses now , now , now , as he pleases . All would count as optimal choices of movement plan . The constraints we impose in the obstacle and target planes serve to select a set of equivalent optimal movement plans but further research is needed to determine the effect of the constraints we impose on the trajectory outside of the obstacle and target plane . In particular we avoided using data from outside the obstacle and target plane precisely because measured means and covariances at points along the trajectory outside of the obstacle and target planes may not reflect any single movement plan and it would be inappropriate to analyze them as if they were determined by the constraints of our task . In our task the location of the fingertip at just two points along the trajectory determines the resulting reward or cost . We can readily generalize the task by adding additional obstacles along the path to create tasks for which the subject must consider his covariance at many points along the trajectory . This sort of generalization would allow investigation of the possible covariance structures along the reach trajectory available to the motor system . It also serves as a model task mimicking the constraints of many natural tasks where the goal is to maneuver around multiple obstacles to reach a goal , as in reaching into a computer chassis to extract one component . We found that subjects' performance was close to that of a Bayesian decision-theoretic movement planner maximizing expected gain except for the most extreme conditions where the optimal choice of trajectory required a large excursion ( “detour” ) around the virtual obstacle . One possible explanation is that such movements entail a large biological cost and that the subject includes biological costs in the computation of expected gain . In effect he “prices” biological cost and is willing to reduce his monetary gain in order to reduce biological cost as well ( see discussion in [32] ) . Although our current data cannot speak to this possibility , one might predict that separate measurements of biomechanical cost would allow these extreme conditions to be predicted as well . The costs in our task are monetary but in theory would also apply to tasks where movement constraints are the results of injury or disease to the motor system [33] , [34] . Patients might limit their motor repertoire in order to prevent undesirable outcomes such as pain or clumsiness , leading to long-term , conditioned motor deficits . This idea forms the basis of a now well-established rehabilitation approach , Constraint Induced Movement Therapy , in which the reward/cost structure of the environment is manipulated in ways that encourage the use of the previously avoided regions of motor space [35] . The conclusions we draw are based on movements confined to a narrow , clearly visible region of space immediately in front of the reviewer . Subjects presumably have considerable experience in coordinating eye and hand in this region of space before they begin the experiment . It would be interesting to investigate in future work with a full range of arm movements , including whether movement plans tend to avoid awkward or unusual movements . We examined the problem of obstacle avoidance from the standpoint of Bayesian decision theory . Our approach is different from other work in the area of obstacle avoidance . Previously , this problem has been approached from the standpoint of theories that suggest that the CNS minimizes kinematic or dynamic variables ( e . g . , total force production ) , with the constraint that the hand path not intersect an obstacle . Of course , this approach fails to take account of two major contributions to real-world movement plans: the uncertainty of visual estimates and motor outcomes ( even for the same real-world obstacle and planned trajectory ) , and variable costs associated with intersecting different kinds of obstacles ( accidentally toppling a cup of water is very different from toppling a cup of scalding coffee ) . Instead , such models always predict the smallest possible trajectory deviation that does not contact the obstacle ( with no ‘room for error’ , so to speak ) . Moreover , the approach confounds the effect on trajectory of hitting an impenetrable obstacle and the cost to the subject . To return to the example we began with , it is easy to imagine circumstances where one would smash through the coffee cup to grasp something on the other side , such as a child in danger of falling . We see that obstacle avoidance , when viewed from the standpoint of Bayesian decision theory , can explain the amount of deviation around a virtual obstacle based on the cost of accidentally intersecting it , and the visuo-motor uncertainty in predicting the location of the fingertip when it passes the obstacle and when it reaches the target .	In everyday , cluttered environments , moving to reach or grasp an object can result in unintended collisions with other objects along the path of movement . Depending on what we run into ( a priceless Ming vase , a crotchety colleague ) we can suffer serious monetary or social consequences . It makes sense to choose movement trajectories that trade off the value of reaching a goal against the consequences of unintended collisions along the way . In the research described here , subjects made speeded movements to touch targets while avoiding obstacles placed along the natural reach trajectory . There were explicit monetary rewards for hitting the target and explicit monetary costs for accidentally hitting the intervening obstacle . We varied the cost and location of the obstacle across conditions . The task was to earn as large a monetary bonus as possible , which required that reaches curve around obstacles only to the extent justified by the location and cost of the obstacle . We compared human performance in this task to that of a Bayesian movement planner who maximized expected gain on each trial . In most conditions , but not all , movement strategies were close to optimal .	[ "Abstract", "Introduction", "Materials", "and", "Methods", "Results", "Discussion" ]	[ "systems", "biology", "biology", "neuroscience", "genetics", "and", "genomics" ]	2012	Speeded Reaching Movements around Invisible Obstacles
RNA pseudoknots are a kind of minimal RNA tertiary structural motifs , and their three-dimensional ( 3D ) structures and stability play essential roles in a variety of biological functions . Therefore , to predict 3D structures and stability of RNA pseudoknots is essential for understanding their functions . In the work , we employed our previously developed coarse-grained model with implicit salt to make extensive predictions and comprehensive analyses on the 3D structures and stability for RNA pseudoknots in monovalent/divalent ion solutions . The comparisons with available experimental data show that our model can successfully predict the 3D structures of RNA pseudoknots from their sequences , and can also make reliable predictions for the stability of RNA pseudoknots with different lengths and sequences over a wide range of monovalent/divalent ion concentrations . Furthermore , we made comprehensive analyses on the unfolding pathway for various RNA pseudoknots in ion solutions . Our analyses for extensive pseudokonts and the wide range of monovalent/divalent ion concentrations verify that the unfolding pathway of RNA pseudoknots is mainly dependent on the relative stability of unfolded intermediate states , and show that the unfolding pathway of RNA pseudoknots can be significantly modulated by their sequences and solution ion conditions . RNAs can fold into complex three-dimensional ( 3D ) structures to carry out their various biological functions [1] . An RNA pseudoknot represents a very common structure motif , which is not only one of the fundamental structure elements in various classes of RNAs such as human telomerase RNA , self-splicing introns of ribozyme and S-adenosylmethionine-responsive riboswitches , but also involved in many biological functions , including regulation and catalysis [2 , 3] . For instance , an RNA pseudoknot can be present within the coding regions of an mRNA , where it stimulates programmed -1 ribosomal frameshifting to control the relative expression levels of proteins [2–4] . Generally , an RNA pseudoknot is formed when a sequence of nucleotides within a single-stranded loop region forms base pairs with a complementary sequence outside that loop [2 , 3 , 5 , 6] . Many experiments have shown that this special 3D topology is key to realize the various functions of RNA pseudoknots [2–4] . In addition , the stability of RNA pseudoknots can also play important roles in modulating their biological functions , and structure changes of RNA pseudoknots could cause diseases such as dyskeratosis [3 , 7 , 8] . Thus , to determine 3D structures and quantify stability of RNA pseudoknots is essential to unveil the mechanisms of their functions and to further aid the related drug design [5 , 9] . There have been several successful experimental methods to obtain 3D structures of RNAs , such as X-ray crystallography , nuclear magnetic resonance spectroscopy , and newly developed cryo-electron microscopy [9–12] . However , it is still very time-consuming and expensive to derive high-resolution 3D structures of RNAs and the RNA structures deposited in Protein Data Bank ( PDB ) are still limited [9 , 12] . To complement experimental measurements , some computational models have been developed to predict 3D structures for RNAs [13–22] . The knowledge-based models [23–34] such as MC-Fold/MC-Sym pipeline [24] , FARNA [25] , 3dRNA [29 , 35 , 36] , RNAComposer [30] and pk3D [31] are rather successful and efficient in constructing 3D structures for RNA pseudoknots through fragments assembly based on limited experimental structures/fragments or reliable secondary structures , while it is still a problem to exactly predict secondary structures of RNA pseudoknots [11 , 20] . Furthermore , most of the above methods cannot give reliable predictions for the thermodynamic properties of RNA pseudoknots from their sequences [9–11] . Simultaneously , some coarse-grained ( CG ) models have been developed to predict the thermodynamic stability of RNAs including pseudoknots [37–46] . The Vfold model enables predictions for the structure , stability , and the free energy landscape for RNA pseudoknots from sequences through enumerating loop conformations on a diamond lattice [37 , 38] . The model is applicable to secondary structure folding while the 3D structures need to be built through fragment assembly based on secondary structures [47] . Several other CG models such as the iFoldRNA [39] , the HiRE-RNA [40] and the oxRNA [42] have been used to predict 3D structure and stability for a few RNA pseudoknots , but the parameters of these models may need further validation for quantifying RNA thermodynamics to accord with experiments . In addition , due to the polyanionic nature of RNAs , metal ions ( e . g . , Na+ and Mg2+ ) in solutions can play an essential role in RNA folding [48–53] , and Mg2+ can play a more special role in stabilizing the compact folded structures of RNA pseudoknots [54–57] . However , the above structure prediction models seldom consider the conditions departing from the high salt ( e . g . , 1M NaCl ) . Although all-atomic molecular dynamics simulations can be used to probe ion-RNA interactions , it is still difficult to simulate RNA structure folding at present due to the huge computation cost [56 , 57] . In simplified CG models , the effect of ions ( especially Mg2+ ) is seldom properly involved due to the interplay between ion binding and structure deformation as well as the particularly efficient role of Mg2+ beyond mean-field description [51–53] . Recently , a Gö-like CG model has been introduced to reproduce the folding thermodynamics of several RNA pseudoknots in the presence of monovalent ions [46 , 58 , 59] , and another structure-based model can well capture the ion atmosphere around RNAs with an explicit treatment of divalent ions [60] . However , the two structure-based models could not be used to predict 3D structures for RNA pseudoknots solely from the sequences [11 , 20 , 46 , 60] . Therefore , it still remains an important problem to predict 3D structures and thermodynamic stability for RNA pseudoknots especially in monovalent/divalent ion solutions only from the sequences . In this study , we focused on predicting 3D structures and stability for extensive RNA pseudoknots in monovalent and divalent ion solutions from their sequences through our previously developed three-bead CG model [61 , 62] . In the following , we first revisited the key features of our CG model such as the CG representation and the implicit-solvent/salt force field for RNAs . We then employed the model to predict 3D structures for various RNA pseudoknots from their respective sequences . Afterward , we made the prediction for the stability of typical pseudoknots with different lengths and sequences over a wide range of monovalent/divalent ion concentrations . Finally , we made the comprehensive analyses on the unfolding pathway for various RNA pseudoknots in ion solutions and examined the effect of monovalent/divalent ions on the unfolding pathway of RNA pseudoknots . Throughout the article , we have made the comparisons between the predictions and the extensive experimental data as well as the comparisons with the existing models . In our model , an RNA is represented as a chain of nucleotides , where each nucleotide is reduced to three beads retaining the key structure features of an RNA chain [46 , 47 , 61 , 62] . As shown in Fig 1A , the backbone phosphate bead ( P ) and sugar bead ( C ) coincide with the phosphate and C4’ atoms of a nucleotide , and the base beads ( N ) are placed at the base atoms linked to the sugar , that is N1 atom for pyrimidine or the N9 atom for purine [61 , 62] . The P , C and N beads are treated as spheres with van der Waals radii of 1 . 9Å , 1 . 7Å and 2 . 2 Å , respectively , and each P bead has a charge of–e on its center [61 , 63] . In the CG model , the effective potential energy of an RNA conformation is given by [61 , 62] U=Ub+Ua+Ud+Uexc+Ubp+Ubs+Ucs+Uel , ( 1 ) where bond length energy Ub , bond angle energy Ua and dihedral energy Ud account for chain connectivity and angular rotation for an RNA chain , and Uexc represents for excluded volume interactions between two CG beads . Ubp and Ubs are the base-pairing and base-stacking interactions , and Ucs is the coaxial stacking interaction between two neighbor stems . The last term Uel corresponds to electrostatic interactions between phosphate groups , which are ignored by most of the existing predictive models for RNA 3D structures [11 , 20] . The detailed description of the potentials in Eq 1 and the determination of the potential parameters have been described in S1 Text and also in Refs . [61 , 62] . Briefly , two sets of parameters of the bonded potentials ( Ub , Ua and Ud ) , Paranonhelical used in RNA folding process and Parahelical used only in structure refinement for helical stems , are derived respectively from single strands/loops and stems in the PDB [12 , 61 , 64] . The sequence-dependent strength of base-staking energy is derived from the combination of the experimental thermodynamic parameters [65–67] . In most occurring pseudoknots with interhelix loop length ≤ 1nt , two helical stems can be often coaxially stacked to form a quasi-continuous double helix ( Fig 1 ) , and the strength of Ucs depends on sequences of two interfaced base pairs [65] . The coaxial stacking could stimulate high levels of -1 frameshifting [3 , 4] , and consequently , could be important for stabilizing functional structures of RNA pseudoknots . The electrostatic interaction Uel is taken into account through the combination of the Debye-Hückel approximation and the concept of counterion condensation ( CC ) [68] . Notably , based on the tightly bound ion ( TBI ) model [51 , 52 , 69] , the competition between monovalent and divalent ions was also taken into account in Uel to enable the CG model to simulate RNA pseudoknot folding in mixed monovalent/divalent ion solutions [61 , 62] . Although the present model has been described by us in Refs . 61 and 62 , the model is still not employed for 3D structure predictions of extensive RNA pseudoknots and it has never been used to predict the stability of RNA pseudoknot in ion solutions , especially in the presence of divalent ions [61 , 62] . Here , the model will be tested by extensive RNA pseudoknots on 3D structure prediction , and be further used to predict thermodynamic stability and the unfolding pathway for various RNA pseudoknots over the wide range of monovalent/divalent ion conditions . Based on the CG force field , the Monte Carlo ( MC ) simulations with simulated annealing algorithm are used to predict 3D structures of RNA pseudoknot [43 , 61 , 62] , where an initial simulation is started at a high temperature and a given solution condition from a totally random chain configuration generated from an RNA sequence . The system is then gradually cooled in steps , and the ion condition is fixed during the cooling process . At each temperature , RNA conformational changes are accomplished via the pivot moves which have been demonstrated to be rather efficient in sampling conformations of polymers [63] , and the changes are accepted or rejected according to the standard Metropolis algorithm [43 , 61] . The final structures obtained at the lowest target temperature ( e . g . , room/body temperature ) are the folded conformations of the RNA predicted by the CG model . Notably , the recorded trajectories at different temperatures during the cooling process allow us to analyze the stability of the RNAs [61 , 62] . Beyond 3D structure prediction , the present model was also employed to predict the stability of RNA pseudoknots in monovalent and divalent salt solutions . Since intermediate states of RNAs can be important to their biological functions [5 , 76 , 80–85] , unfolding pathway of RNAs including some pseudoknots has been studied through theoretical modeling and experiments [75–77 , 81–88] . To examine the unfolding pathway of RNA pseudoknots , we made comprehensive analyses for six RNA pseudoknots; see Fig 7 and S4 Fig . Based on the simulations for each pseudoknot at a given solution condition , beyond the fractions of states F and U , the fractions of different intermediate hairpin states ( named as S1 and S2 for intermediate states reserving one of Stem 1 and Stem 2 , respectively ) at different temperatures can also be calculated; see Figs 7 and 8 and S4 and S6 Figs . Furthermore , we employed the model to predict the unfolding pathway for various RNA pseudoknots in monovalent/divalent ion solutions and examined the effect of monovalent/divalent ions on the unfolding pathway of RNA pseudoknots , which was seldom covered in previous studies since the effect of divalent ions is generally difficult to be involved . It is important to predict 3D structures and stability of RNA pseudoknots in monovalent/divalent ion solutions from their sequences . In this work , we employed our previously developed model to address this problem . Beyond mainly focusing on reproducing structures , as many previous structure prediction models have done , the present model enables us to predict and analyze 3D structure stability for RNA pseudoknots in different monovalent/divalent ion solutions . The following are the major conclusions: Despite the extensive agreements between our predictions and experiments , the present model has several limitations that should be overcome in future model development . First , the present model does not treat possible noncanonical interactions such as base triple interactions between loops and stems , self-stacking in loop nucleotides and special hydrogen bonds involving phosphates and sugars , which could be important for some more complex pseudoknotted structures [7 , 17 , 38] . Beyond the common H-type pseudoknots ( ≤ 56nt ) used in this work , larger RNAs with complex structures should be incorporated in to further improve the present model [19 , 90–94] . Second , the effect of monovalent/divalent salts is implicitly accounted for in the present model by the combination of CC theory and the TBI model . Such implicit-salt treatment may be responsible for the underestimation on the stability of RNA pseudoknots at high [Mg2+] . Mg2+ can play an efficient and special role in stabilizing compact RNA structures [51–54 , 79] , and further development may need to involve Mg2+ explicitly in our model . Third , in this work , we mainly focused on the 3D structures and thermodynamic stability of RNA pseudoknots , and did not involve the stability under mechanical force . Mechanical forces can be not only considered as a useful probe for RNA stability , but also important for the functions of some RNA pseudoknots [3 , 81 , 86 , 95–97] . For example , the frameshifting efficiency may be affected by the magnitude of unfolding force for RNA pseudoknots [3 , 81 , 96] . Fortunately , the present model can be extended to study the mechanical stability of RNA pseudoknots by including external force in the energy functions of the model [67 , 86 , 95] . Finally , the 3D structure predicted by the present model is at the CG level , and it is still necessary to develop the model to reconstruct all-atomistic structures based on the CG structures for further practical applications . Nevertheless , the present model could be a reliable predictive model for predicting 3D structures and stability of RNA pseudoknots in ion solutions from their sequences and the analyses can be helpful to understand the physical mechanism for the unfolding pathway of RNA structures .	RNA pseudoknotted structures and their stability can play important roles in RNA cellular functions such as transcription , splicing and translation . Due to the polyanionic nature of RNAs , metal ions such as Na+ and Mg2+ in solutions can play an essential role in RNA folding . Although several computational models have been developed to predict 3D structures for RNA pseudoknots to further unveil the mechanisms of their functions , these structure prediction models seldom consider ion conditions departing from the high salt ( e . g . , 1M NaCl ) and temperatures from the room temperature . In this work , we employed our coarse-grained model to predict 3D structures and thermodynamic stability for various RNA pseudoknots in monovalent/divalent ion solutions from their sequences , and made comparisons with extensive experimental data and existing models . In addition , based on our comprehensive analyses for extensive pseudoknots and the wide range of monovalent/divalent ion conditions , we confirmed that the thermally unfolding pathway of RNA pseudoknots is mainly determined by the relative stability of intermediate states , which has been proposed by Thirumalai et al . Our analyses also show that the thermally unfolding pathway of RNA pseudoknots could be apparently modulated by the sequences and ion conditions .	[ "Abstract", "Introduction", "Materials", "and", "methods", "Results", "Discussion" ]	[ "rna", "structure", "prediction", "condensed", "matter", "physics", "rna", "stability", "sequence", "motif", "analysis", "thermodynamics", "research", "and", "analysis", "methods", "sequence", "analysis", "rna", "structure", "bioinformatics", "gene", "expression", "melt...	2018	Predicting 3D structure and stability of RNA pseudoknots in monovalent and divalent ion solutions
Altered metabolism is one of the hallmarks of cancers . Deregulation of ribose-5-phosphate isomerase A ( RPIA ) in the pentose phosphate pathway ( PPP ) is known to promote tumorigenesis in liver , lung , and breast tissues . Yet , the molecular mechanism of RPIA-mediated colorectal cancer ( CRC ) is unknown . Our study demonstrates a noncanonical function of RPIA in CRC . Data from the mRNAs of 80 patients’ CRC tissues and paired nontumor tissues and protein levels , as well as a CRC tissue array , indicate RPIA is significantly elevated in CRC . RPIA modulates cell proliferation and oncogenicity via activation of β-catenin in colon cancer cell lines . Unlike its role in PPP in which RPIA functions within the cytosol , RPIA enters the nucleus to form a complex with the adenomatous polyposis coli ( APC ) and β-catenin . This association protects β-catenin by preventing its phosphorylation , ubiquitination , and subsequent degradation . The C-terminus of RPIA ( amino acids 290 to 311 ) , a region distinct from its enzymatic domain , is necessary for RPIA-mediated tumorigenesis . Consistent with results in vitro , RPIA increases the expression of β-catenin and its target genes , and induces tumorigenesis in gut-specific promotor-carrying RPIA transgenic zebrafish . Together , we demonstrate a novel function of RPIA in CRC formation in which RPIA enters the nucleus and stabilizes β-catenin activity and suggests that RPIA might be a biomarker for targeted therapy and prognosis . Colorectal cancer ( CRC ) is one of the most common forms of cancers and results in more than 600 , 000 deaths annually [1–3] . Mutations in adenomatous polyposis coli ( APC ) and β-catenin , members of the Wnt signaling cascade , are among the major causes of colon tumorigenesis [4–6] . APC acts as a cytoplasmic scaffolding protein and induces the ubiquitin-mediated degradation of β-catenin [7] . In addition to its cytoplasmic activity , APC also modulates nuclear β-catenin levels as a result of its intrinsic nuclear-cytoplasmic shuttling capability [8–11] . Truncation of APC protein results in accumulation of nuclear β-catenin in CRC cells [12–14] . However , existing APC truncation mutants differentially affect the phosphorylation and ubiquitination of β-catenin , suggesting that these functions may be controlled by different APC domains [15 , 16] . In the nucleus , β-catenin acts as a coactivator with T-cell transcription factor 4 ( Tcf-4 ) /lymphocyte enhancement factor ( LEF ) to activate the transcription of downstream targets such as Cyclin D1 ( CCND1 ) and Cyclin E2 ( CCNE2 ) . Abnormal activation of the β-catenin signaling pathway can lead to increased cell proliferation and immortalization [17–19] . For example , a human CRC cell line expressing wild-type ( WT ) APC and a mutant version of β-catenin protein ( with a single amino acid deletion at residue S45 ) is sufficient to induce a cancerous phenotype [20] . However , the precise activation process of β-catenin signaling is still largely unknown . The pentose phosphate pathway ( PPP ) is critical for cancer cell survival and proliferation [21 , 22] . Ribose-5-phosphate isomerase A ( RPIA ) is an important integral member of the PPP and regulates cancer cell growth and tumorigenesis [2 , 23 , 24] . In pancreatic ductal adenocarcinoma ( PDAC ) , RPIA expression is required for maintenance of tumor cells overexpressing KRasG12D , an activated form of Ras [23] . Our previous study showed that in hepatocellular carcinoma ( HCC ) , RPIA regulates tumorigenesis via PP2A and extracellular signal-regulated kinase ( ERK ) signaling [24] . Studies performed in colon tumor tissues expressing microRNA-124 revealed that cells expressing low RPIA levels led to a reduced tumor size , while high RPIA expression was correlated with reduced survival and increased tumor growth [2] . Here , we report that in CRC tissue RPIA is significantly up-regulated , and it is expressed at multiple stages of tumorigenesis , including early stages . It directly interacts with β-catenin and APC to activate target genes downstream of β-catenin that are important for carcinogenesis . High levels of RPIA expression stabilize β-catenin levels by preventing phosphorylation and ubiquitination of β-catenin . Transgenic zebrafish overexpressing RPIA under the control of a gut-specific promoter exhibited enhanced β-catenin expression and elevated mRNA levels of the colon cancer marker gene ccne1 . Our work uncovers a new role of RPIA and provides a molecular mechanism of RPIA-mediated β-catenin stabilization and activation necessary for colon cancer formation . To assess the role of RPIA in the progression of CRC , we measured RPIA protein levels using immunohistochemistry ( IHC ) with tissue arrays from stage I through IVB and metastatic colon cancer . The RPIA immunoreactive score ( IRS ) was calculated by multiplying the staining intensity by the proportion of positive cells [25] ( S1A Fig ) . Highly elevated RPIA expression was found in all stages of colon cancer when compared to non-cancerous samples ( Fig 1A ) . A control , non-immune antibody was employed to assess background staining and was stained negative . In Fig 1A , “normal colon” shows dark staining in an epithelial region to the lower left that is of much lower intensity than that of the tumor samples . IRS quantification revealed that RPIA expression is significantly up-regulated in all stages of colon adenocarcinoma and even in metastatic carcinoma ( Fig 1B ) . To examine whether the RPIA mRNA level is also up-regulated in CRC patients , 80 paired tissues , including tumors and the adjacent normal tissues , were analyzed using real-time quantitative PCR ( qPCR ) . In 78% ( 62 of 80 ) of the CRC specimens , RPIA mRNA was more than 2-fold overexpressed from stage I to IV and in metastatic carcinoma ( Fig 1C ) . Taken together , we found that RPIA is overexpressed at both the mRNA and protein levels in all stages of colon cancer formation . To examine the effects of RPIA overexpression on cellular proliferation , two colon cancer cell lines , HCT116 and SW480 , were used . SW480 is a human colon cancer cell line with APC C-terminal truncation at 1338 , but the β-catenin binding region is retained; HCT116 is a highly metastatic cell line with WT APC and both an S45 mutation and the WT allele for β-catenin , but most of the β-catenin protein comes from the mutant allele . WST-1 assays examine metabolic activity that represent the viability of the cell . We tested three small interfering RNAs ( siRNAs ) ( S1B Fig ) , which were pooled or treated separately , and found all three siRNAs had similar effects . Therefore , we selected number 3 siRNA for the rest of the experiments . Knock down of RPIA significantly decreased cell proliferation in both cell lines ( Fig 2A and S2A Fig ) . Conversely , overexpression of RPIA increased cell proliferation in both HCT116 and SW480 cells ( Fig 2A and S2A Fig ) . In the knockdown and followed by overexpressing the RPIA for rescue , the results clearly showed that RPIA small interfering RNA ( si-RPIA ) decreased proliferation and β-catenin protein level can be rescued by overexpression RPIA in both HCT116 and SW480 cell lines ( Fig 2A and S2A Fig ) . Knockdown of RPIA also dramatically decreased the colony formation ability in both cell lines ( Fig 2B and S2B Fig ) , and overexpression of RPIA increased the colony formation ability in both HCT116 and SW480 cells ( Fig 2B and S2B Fig ) . These data suggest that the RPIA expression level is positively correlated with cellular proliferation and colony formation ability in colon cancer cells . Aberrant β-catenin accumulation is a major cause of uncontrollable proliferation in colon cancer cells [26 , 27] . β-catenin exerts its proliferation-promoting effects via translocation to the nucleus where it binds to T-cell transcription factor ( TCF ) /LEF to activate the transcription of downstream β-catenin target genes [28] . Therefore , we were interested in determining whether RPIA affects the β-catenin level in colon cancer cells using HCT116 and SW480 cells . Our results indicate that knockdown of RPIA decreased the β-catenin protein level in both HCT116 ( Fig 2C ) and SW480 cells ( S2C Fig ) without affecting β-catenin ( encoded by CTNNB1 gene ) mRNA levels . Conversely , RPIA overexpression increased β-catenin protein expression levels in both HCT116 ( Fig 2D ) and SW480 cell lines ( S2D Fig ) , while β-catenin mRNA levels were unchanged . Using TOPflash/FOPflash luciferase reporter assay , we found that β-catenin activity was significantly attenuated upon RPIA knockdown ( Fig 2E and S2E Fig ) and dramatically increased when RPIA was overexpressed ( Fig 2F and S2F Fig ) . Using qPCR to detect the expression levels of known downstream targets of β-catenin , including CCND1 , CCNE2 , and AXIN2 , we found that knockdown and overexpression of RPIA reduced ( Fig 2G and S2G Fig ) and increased ( Fig 2H and S2H Fig ) , respectively , the expression of these target genes . However , the effect of overexpression of RPIA was not as dramatic as that of knock down of RPIA , likely because CRC cell lines already overexpress RPIA . These results suggest that the RPIA expression level is positively correlated with β-catenin protein levels and its transcriptional activity . Our previous study indicates that ERK signaling participates in RPIA-mediated hepatocarcinogenesis [24] . These observations , in combination with other studies demonstrating crosstalk between β-catenin and ERK in other types of tumors [29 , 30] , led us to investigate whether ERK signaling might also play a role in RPIA-mediated tumorigenesis in colon cancer . Therefore , the effects of both RPIA overexpression and reduction on both β-catenin and ERK protein levels were examined in HCT116 and SW480 cells . Reduction of RPIA by knockdown significantly decreased nuclear β-catenin protein levels in HCT116 ( Fig 3A ) and both cytoplasmic and nuclear β-catenin protein were decreased in SW480 ( S3A Fig ) , but did not affect the levels of activated , phosphorylated ERK ( pERK ) ( Fig 3B and S3B Fig ) . Conversely , overexpression of RPIA increased β-catenin protein levels in HCT116 ( Fig 3A ) and SW480 ( S3A Fig ) cells without altering both cytoplasmic and nuclear level of pERK and ERK levels in these cell lines ( Fig 3B and S3B Fig ) . Other mechanisms important for intestinal cell proliferation such as epidermal growth factor receptor ( EGFR ) signaling were also examined . Neither EGFR protein level nor the phosphorylated EGFR ( pEGFR ) were altered upon overexpression or knockdown RPIA in both cell lines ( S3C Fig ) . Furthermore , IHC staining analyses revealed that both RPIA and β-catenin protein levels were significantly higher in the nuclei of colon cancer tissues than in the nuclei of normal tissues . In addition , a positive correlation was presented between RPIA and β-catenin protein levels in the nuclei of colon cancer tissue ( Fig 3C ) . These data suggest that promotion of β-catenin signaling , but not ERK or EGFR signaling , is involved in transducing the effects of RPIA-mediated colon cancer tumorigenesis . As only β-catenin protein levels were affected by the levels of RPIA expression , we proposed that RPIA increased β-catenin protein stability in colon cancer cells . To test this hypothesis , the protein synthesis inhibitor cycloheximide ( CHX ) was used to determine the half-life of β-catenin in cell lines with either reduced or elevated levels of RPIA . Reduction of RPIA by knockdown decreased the β-catenin protein half-life from 3 to 1 . 1 h in HCT116 cells ( Fig 3D ) and from 2 . 7 to 0 . 9 h in SW480 cells ( S3D Fig ) . Conversely , overexpression of RPIA strongly increased the half-life of β-catenin from 5 . 6 to 10 . 4 h in HCT116 cells ( Fig 3E ) and from 4 . 9 to 9 . 9 h in SW480 cells ( S3E Fig ) . Therefore , we conclude that RPIA increases β-catenin protein stability in colon cancer cells . Because it has been shown that β-catenin protein levels can be controlled by ubiquitination and subsequent proteasome degradation [31] , we tested whether RPIA could modulate β-catenin protein levels in colon cancer cell lines by changing the ubiquitination-mediated degradation process . As shown previously , RPIA knockdown resulted in a reduction in β-catenin protein levels ( Fig 3F and S3F Fig , left panel ) . Treatment with 5 μM MG132 , a proteasome inhibitor , rescued the reduction in β-catenin protein levels observed in cells expressing RPIA siRNA ( Fig 3F and S3F Fig , left panel ) . Immunoprecipitation ( IP ) revealed that more ubiquitin was coprecipitated with β-catenin in RPIA-siRNA-treated cells than in negative control siRNA ( si-NC ) -treated cells ( Fig 3F and S3F Fig , right panel ) . In addition , the phosphorylated , targeted for degradation form of β-catenin ( with phosphorylation at residues Ser33/Ser37 ) was elevated upon RPIA knockdown relative to total β-catenin ( Fig 3G and S3G Fig ) . Moreover , expression of a non-degradable β-catenin mutant ( S33Y ) rescued the reduction of proliferation upon RPIA knockdown in HCT116 and SW480 cell lines ( Fig 3H and S3H Fig ) . To demonstrate that β-catenin is indeed required downstream of RPIA , the β-catenin inhibitor ICRT14 was applied to the RPIA overexpression cells . The results showed that RPIA-promoted cellular proliferation was attenuated in a dose-dependent manner by ICRT14 ( Fig 3I and S3I Fig ) . This confirms that β-catenin is required for RPIA overexpression-mediated cell proliferation . The levels of an inactive form of GSK3β ( with phosphorylation at residue ser9; pGSK3βser9 ) , which does not have the ability to phosphorylate β-catenin , were further examined . Because phosphorylated GSK3βser9 was elevated upon overexpression of RPIA in both HCT116 and SW480 cell lines and there is no difference in the nucleus GSK3β , we suggest the RPIA modulate GSK3β only in the cytoplasm ( Fig 3J and S3J Fig ) . Moreover , treatment with GSK3β inhibitors ( lithium chloride [LiCl] or CHIR99021 ) rescued the reduction of proliferation upon RPIA knockdown , indicating the involvement of GSK3β in this process ( Fig 3K and S3K Fig ) . These data show that RPIA retains a novel function to protect β-catenin from phosphorylation-mediated ubiquitination and degradation via proteasomes . The cytoplasmic complex that targets β-catenin for degradation includes the scaffolding protein APC [32] . In addition , APC associates with β-catenin in the nucleus and directs the nucleocytoplasmic export of β-catenin [10 , 11] . We used immunostaining to detect RPIA localization . In the pcDNA3 vector only control ( pcDNA ) , RPIA was expressed in the cytoplasm exclusively . Overexpression of RPIA in both HCT116 and SW480 cells resulted in an increase in nuclear and cytoplasmic RPIA expression ( Fig 4A and S4A Fig ) with a punctate pattern of RPIA in the nucleus of the DAPI-negative nucleoplasm [33] . IP of different proteins followed by western blotting indicated that RPIA can form a complex with APC and β-catenin in the nucleus in both HCT116 and SW480 cell lines ( Fig 4B and 4C and S4B and S4C Fig ) . Interestingly , we noticed the minor difference between these cells . In HCT116 , RPIA interacted strongly with APC and β-catenin , respectively , in the cytoplasm ( Fig 4C ) . In addition , RPIA/APC and RPIA/β-catenin complex levels are induced by RPIA WT ( RPIA-WT ) ( Fig 4B and 4C ) . In HCT116 , the nuclear interaction of RPIA-β-catenin is much weaker than in cytoplasm . We suspect the interaction between RPIA and β-catenin in the nucleus might be indirectly through APC . In SW480 , the nuclear RPIA-β-catenin interaction is much stronger than in cytoplasm . However , only the β-catenin bond to RPIA and promoted from RPIA-WT in cytoplasm in SW480 ( S4B and S4C Fig ) and the nuclear β-catenin-RPIA interaction can be regulated by the RPIA amount . The differences in HCT116 and SW480 might be caused by the truncated APC in SW480 , and HCT116 has an S45 mutation from β-catenin . The RPIA protein sequence is conserved among humans ( Homo sapiens ) , mice ( Mus musculus ) , and zebrafish ( Danio rerio ) ( Fig 5A ) . To determine which protein domain ( s ) are important for RPIA-mediated tumor cell proliferation , five RPIA deletion mutants were generated . These include RPIA-ΔA ( deletion of the active domain of RPIA ) , RPIA-ΔB ( deletion of the catalytic domain of RPIA ) , RPIA-Δ ( A+B ) , RPIA-ΔC , and RPIA deletion domain D mutant ( RPIA-ΔD ) ( Fig 5B ) . The WT and five deletion mutants were transfected into HCT116 and SW480 cells . WST-1 assays were performed to examine metabolic activity , and the RNA and protein levels from the WT and five deletion mutants were verified ( S5A and S5B Fig ) . We noticed that different RNA constructs might have different regulation of RPIA stability . Interestingly , only the expression of RPIA-ΔD failed to enhance cell proliferation , while the other mutants produced no significant changes in proliferation ( Fig 4C and S4C Fig ) . These data suggest that the C-terminal domain D of RPIA ( AAs 290 to 311 ) is essential for RPIA-mediated tumor cell proliferation . Domain D also seems to be necessary for the RPIA-mediated increase in β-catenin protein stability in colon cancer cells because overexpression of RPIA-ΔD did not stabilize β-catenin protein levels like the WT RPIA . Following overexpression of RPIA-ΔD , the half-life of β-catenin was approximately 6 . 1 and 3 . 3 h in HCT116 and SW480 cells , respectively , which was similar to that of the pcDNA vector alone ( 5 . 6 and 4 . 9 h in HCT116 and SW480 cells , respectively; Fig 3E and S3E Fig ) . Moreover , RPIA-ΔD did not interact with APC and β-catenin in either the cytoplasm or nucleus ( Fig 4B and S4B Fig ) . Furthermore , RPIA-ΔD was unable to elevate TCF reporter activity in the colon cancer cells ( Fig 5D and S5D Fig ) . These results demonstrate that domain D of RPIA is essential for the RPIA-mediated increase in β-catenin protein stability , activation of β-catenin target genes , and cell proliferation advantages seen in colon cancer cells . Together , our data also indicated that the RPIA D domain exhibits a novel function in addition to the enzymatic region . Histopathologically , many of the features found in human colon adenocarcinoma are similar to those in zebrafish , an important vertebrate cancer model system [34 , 35] . To test the effects of RPIA overexpression on colon cancer formation , we generated transgenic zebrafish that overexpressed RPIA under the control of a gut-specific promoter ( ifabp ) . In particular , we analyzed the histopathology of the intestinal bulb ( IB ) , middle intestine ( MI ) , and posterior intestine ( PI ) collected from non-transgenic and transgenic ( ifabp:RPIA ) fish of different ages . Increased nuclear-to-cytoplasmic ratio , nuclear atypia , and moderately differentiated adenocarcinoma were observed in 3- and 5-month-old Tg ( ifabp:RPIA ) fish relative to the WT controls ( Fig 6A and 6B , upper panel ) . In WT zebrafish , the β-catenin protein is detected at low levels in the intestinal villi at intracellular junctions [36] . Using IHC , we also observed low-level and intracellularly located β-catenin expression in WT fish , while overexpression of RPIA in transgenic fish resulted in increased β-catenin expression and nuclear localization ( Fig 6A and 6B , lower panel ) . We next explored the β-catenin target genes , including ccne1 , ccnd1 , cdkn2a/b , myca , mycb , and lef1 . A log 10-fold change value was used to show the level of genes [37] in 3-month-old Tg ( ifabp:RPIA ) fish , and the expression levels of β-catenin target genes were found to be positively correlated with RPIA expression levels ( Fig 6C and 6E ) and more highly expressed in the PI ( Fig 6E ) than in the IB ( Fig 6C ) . In addition , as ccne1 is required for cell cycle/proliferation and tumor growth in CRC [29 , 38] , it was used as colon tumorigenesis marker . The mRNA levels of ccne1 were elevated in 3-month-old Tg ( ifabp:RPIA ) fish , especially in the PI , which had dramatically increased RPIA expression . In addition to 3-month-old fish , we also analyzed β-catenin target genes in 5-month-old fish ( S6A–S6C Fig ) . In accordance with the 3-month-old Tg ( ifabp:RPIA ) fish , the β-catenin target genes were up-regulated more significantly in the PI than in the IB ( S6C Fig ) . Then , we noticed that RPIA and the expression of most of the β-catenin target genes were slightly decreased in 5-month-old fish compared with 3-month-old fish . Whether this phenomenon was the result of mature recovery in zebrafish remains to be determined [37 , 39] . A number of physiological changes are associated with cancer in human patients [40 , 41] including decreased body weight , decreased body width , decreased body length , and reduced intestinal length . With the exception of reduced body length , all these changes were observed in transgenic 1-year-old fish ( S6D–S6G Fig ) . These results demonstrate that in vivo , in an important model system , RPIA increases β-catenin protein levels and induces colon tumorigenesis , resulting in overall weaker and smaller fish . Recent findings have revealed that the non-oxidative PPP is a critical pathway for tumor formation [21] . Aberrant activation of the canonical Wnt/β-catenin pathway has also been shown to be involved in gastrointestinal cancers [36] . In cancer cells , β-catenin protein has a dual function: at the membrane , β-catenin coordinates adherent junctions for maintenance of epithelial cell barriers , while in the nucleus , β-catenin acts as a transcriptional activator to regulate proliferation genes [42–44] . In this study , we demonstrate that RPIA exhibits a novel role in CRC through association with and activation of β-catenin . High levels of RPIA expression were detected early and throughout multiple stages of 80 paired samples in CRC human patients . These results are consistent with the Human Gene Database and the Human Protein Atlas , which indicates about the 10-fold higher expression of RPIA in CRC patients than in normal tissues . Furthermore , we found RPIA stabilizes and subsequently promotes activation of β-catenin downstream target genes . We suggest that the increased cellular proliferation and oncogenicity are induced by RPIA through β-catenin pathway . In other cancer types , RPIA promotes tumorigenesis via different mechanisms [2 , 23 , 24] . In pancreatic and hepatic cancers , RPIA expression is required for maintenance of tumor cells overexpressing KRasG12D , while in HCC , RPIA regulates tumorigenesis via PP2A and ERK signaling . Interestingly , the RPIA-mediated CRC tumorigenesis does not involve the activation of ERK and presumably Ras signaling . This observation raises an interesting question: "Is RPIA-mediated stabilization and activation of β-catenin merely in CRC ? " If valid , it may influence the decision of choosing different therapeutic targets of molecules and/or signaling pathways for treating different cancer types . In the canonical β-catenin signaling pathway , APC binds to β-catenin in the cytoplasm in normal cells ( Fig 7 ) . This recruits GSK3β phosphorylates β-catenin , resulting in the eventual proteasomal degradation of β-catenin . We propose that in CRC cells , overexpression of RPIA results in the binding of RPIA to β-catenin and protects β-catenin from phosphorylation and subsequent cytoplasmic degradation . Intriguingly , we also found that RPIA interacts with APC and β-catenin in the nucleus . According to current studies , superfluous β-catenin is shuttled from the nucleus to the cytoplasm by APC [18 , 38] . We propose that RPIA might interrupt the APC-mediated process of exporting β-catenin by forming a complex in the nucleus . Consequently , colon cells developed tumorigenesis . Moreover , we found that C-terminal 22 amino acid of RPIA D domain is required for the RPIA-mediated β-catenin activation , stabilization , and enhanced colon cancer cell proliferation . This region is distinct from its enzymatic domain and a portion of the protein not previously identified as playing a role in CRC . Thus far , nothing is known about the function of the D domain , except this report . Cross species comparison of RPIA protein sequences between human , mouse , and zebrafish reveals that D domain is highly conserved across species . Accordingly , we hypothesize that the RPIA D domain exhibits a novel function in addition to the enzymatic region . It may associate with important partners , such as APC , β-catenin , and other proteins and form multimolecular complexes in both the cytosol and the nucleus . The phenomenon raises the questions such as “Does RPIA act as transcription co-activator ? ” and “Does RPIA have different protein partners in various cancer types ? ” We are currently searching for additional proteins that interact with RPIA in cancer cells . The observation that RPIA is expressed at high levels early and throughout CRC is consistent with a role for RPIA in initiation and maintenance of carcinogenesis . Linking the clinical samples to our in vivo studies in zebrafish , misexpression of RPIA in the intestines of zebrafish is sufficient to induce spontaneous tumor formation in fish as young as 3 month and to cause additional physiological hallmarks of cancer , including reduced body weight , body width , and intestinal length in adult fish . These in vivo results are consistent with the observation that in cancer cell lines , downregulation of RPIA using microRNA reduces cell growth and colony formation ability , while overexpression of RPIA is correlated with enhanced growth and lower survival rates . In addition , the examination of apc/+ zebrafish revealed high levels of β-catenin that are disorganized and accumulate both in the cytoplasm and nucleus [45] , and similar to our study , these fish develop spontaneous , intestinal tumors . In human patients , we noticed that RPIA expression was slightly decreased at the metastasis stage ( Fig 1B and 1C ) . It was reported that invasive CRC cells exhibit low levels of proliferation markers [46] . We suggest that RPIA is necessary for primary tumorigenesis and that the RPIA level decreases at the metastasis stage so that tumor cells undergo epithelial-mesenchymal transition ( EMT ) . Taken together , our studies demonstrate that RPIA functions as an activator for β-catenin-mediated colon tumorigenesis at the initiation stage . One of the important functions of PPP is to generate ribose-5-phosphate for nucleotide synthesis . ATP provides the phosphate group via the salvage pathway [47] , and RPIA mediates this enzymatic step in the cytoplasm . It has been proposed that knockdown of RPIA hinders tumor cell proliferation by reducing nucleotide synthesis [2] . However , in colorectal cells , the RPIA enzymatic and catalytic domain deletion mutants still promoted cell proliferation and activated β-catenin downstream target genes , revealing that at least in this type of cancer this is not how RPIA modulates tumor growth . In these cells , the RPIA D domain is necessary for cell proliferation , stabilization of β-catenin , and formation of a β-catenin/APC complex . Accordingly , D domain may be a therapeutic target for inhibiting the oncogenicity ability without affecting RPIA canonical enzymatic function in PPP . We therefore suggest that combination of anti-RPIA D domain therapy with conventional chemotherapy might improve the inhibition of CRC progression . The mRNA from 80 paired tissues , including CRC and the adjacent normal tissues were obtained from Taipei Veterans General Hospital , procedures were undertaken in accordance with the Institutional Review Board of Taipei Veterans General Hospital , and the IRB number is 2015-04-010-AC . All adult participants provided written informed consent and there were no child participants . All zebrafish experiments were approved by the Institutional Animal Care and Use Committee ( IACUC ) of the NHRI and were in accordance with International Association for the Study of Pain guidelines ( protocol number: NHRI-IACUC-104157-A ) . Taiwan Zebrafish Core Facility ( TZCF ) at NHRI or TZeNH is a government-funded core facility , and since 2015 , the TZeNH has been AAALAC accredited . The CRC cell lines HCT116 and SW480 were maintained in Dulbecco’s Modified Eagle’s Medium ( DMEM ) ( Invitrogen , Carlsbad , CA ) supplemented with 10% fetal bovine serum ( FBS ) , 100 units/ml of penicillin , and 100 μg/ml of streptomycin and incubated at 37°C and 5% carbon dioxide . DNA typing of the cell lines was verified by Mission Biotech ( Taipei , Taiwan ) using a Promega GenePrint 10 System . Transient transfection of siRNA was performed using Lipofectamine RNAiMax ( Invitrogen ) according to the manufacturer’s manual . Three individual RPIA siRNAs ( HSS117931 , HSS117932 , and HSS117933 ) and si-NC were purchased from Invitrogen . pcDNA3 . 0-RPIA , RPIA-ΔA , RPIA-ΔB , RPIA-Δ ( A+B ) , RPIA-ΔC , and RPIA-ΔD were constructed by subcloning full-length or truncated RPIA cDNAs into a pcDNA 3 . 0 expression vector . Truncated RPIA products were amplified by specific primer sets: RPIA-ΔA ( nt 1–522 deletion ) ( forward ) 5ʹ ATAGAATTCATGGGCGGAGGCTG CCTGAC 3ʹ and ( reverse ) 5ʹ AGACTCGAGCTTGCAGGGTCAACAGAAAGGCT 3′; RPIA-ΔB ( nt 523–567 deletion ) ( forward ) 5ʹ ATAAGTCGCTTCATCGTGATCGCT 3ʹ and ( reverse ) 5ʹ AGAACCCTTGATGAGATTGAGATCA G 3′; RPIA-Δ ( A+B ) ( nt 1–567 deletion ) ( forward ) 5ʹ ATAGAATTCATGAGTCGCTTCATCGTGATCGCT 3ʹ and ( reverse ) 5ʹ AGACTCGAGCTTGCA GGGTCAACAGAAAGGCT 3′; RPIA-ΔC ( nt 568–867 deletion ) ( forward ) 5ʹ ATAATGGCTG AGAGAGTCTACTTTGGGATG 3ʹ and ( reverse ) 5ʹ AGAAGCATAGCCAGCCACAATCTTCT 3′; RPIA-ΔD ( nt 868–936 deletion ) ( forward ) 5ʹ ATAGAATTCACTTCAGCGGAGGCCGGAG 3ʹ and ( reverse ) 5ʹ AGACTCGAGGTTGATGAATAGGCCTGTGTCC 3ʹ . Transient transfection of the plasmid DNAs was performed using Lipofectamine 2000 ( Invitrogen ) according to the manufacturer’s manual . Colon cancer tissue staining data were obtained by using the tissue array CDA3 from SUPER BIO CHIPS and the tissue array MC5003a from US Biomax . The slides were incubated with mouse anti-RPIA ( 1:100 ) or rabbit anti-β-catenin ( 1:150 ) primary antibody at 4°C overnight after the dewax , rehydration , and antigen retrieval steps . The HE staining procedure was performed as outlined in our previous report , and tissues were examined with light microscopy [48] . Total protein was extracted from cells using whole-cell extract ( WCE ) lysis buffer . Lysates were vibrated for 30 min and centrifuged at 13 , 200 rpm for 20 min at 4°C . Western blotting was performed as outlined in our previous report [24] , and the fractionation protocol used has been described previously [49] . Primary antibodies include RPIA ( Cat# ab67080; Abcam , Cambridge , United Kingdom ) , β-actin ( Cat# GTX109639; GeneTex , Inc , Irvine , CA ) , β-catenin ( Cat# GTX61089 , GeneTex; Cat# ab22656 , Abcam ) , β-catenin ( phospho Ser33/Ser37 ) ( Cat# GTX11350 GeneTex ) , APC ( Cat# GTX61328 , GeneTex ) , Ubiquitin ( Cat# 3936; Cell Signaling Technology , Danvers , MA ) , K48-linkage specific polyubiquitin ( Cat# 4289 , Cell Signaling Tecnology ) , β-Tubulin ( Cat# ab52866 , Abcam ) , and Lamin A/C ( Cat# ab108922 , Abcam ) . For IP , 100 μg of protein lysate was incubated with primary antibody overnight and subsequently incubated with protein A/G-Sepharose beads for 1 . 5 h . The interaction results were assessed with western blotting . RNA was extracted from paired samples of patient tissues , transgenic zebrafish tissue , or cell lines homogenized in TRIzol . cDNA was reverse transcribed from RNA using a High Capacity RNA-to-cDNA Kit ( Cat# 4387406; Applied Biosystems , Foster City , CA ) . qPCR was performed using an ABI Prism 7500 Sequence Detection System ( Power SYBR Master Mix , Cat#4367659 , Applied Biosystems ) . Gene expression was amplified with the primers listed in Supporting information: S1 and S2 Tables . Cells were transfected with TOPflash ( containing a WT TCF binding site ) or FOPflash ( containing a mutated TCF binding site ) , which were purchased from Millipore , and Renilla luciferase was used as an internal control . The transfected cells were harvested 48 h post-transfection and lysed by the buffer supplied in the Dual-Glo Luciferase Assay Kit ( Cat# E2940 , Promega , Madison , WI ) , and luciferase activity in lysates was measured with a luminometer . Transfected cells grown on cover slips were fixed in 4% paraformaldehyde for 10 min at room temperature and permeabilized in 0 . 5% Triton for 10 min . After 1 h of blocking in 2% FBS at room temperature , slides were incubated with anti-RPIA or anti-APC primary antibody at 4°C overnight . Secondary antibodies conjugated with Texas Red or FITC were used , and DAPI was used to stain nuclei . The images were scanned and captured with confocal microscopy . The coding region of human RPIA ( NM_144563 . 2 ) was amplified by PCR with the attB1-F-RPIA and attB2-R-RPIA primer pair using cDNA from the HEK293 cell line as a template . PCR was performed using a KOD FX ( Toyobo , Osaka , Japan ) and a 994-bp amplicon . The following forward primer was used for Gateway cloning: attB1-F-RPIA ( Tm:59°C ) :5ʹGGGGACAAGTTTGTACAAAAAAGCAGGCTATGCAGCGCCCCGGGCC3ʹ , and the following reverse primer was used for Gateway cloning: attB2-R-RPIA ( Tm:58°C ) :5ʹGGGGACCACTTTGTACAAGAAAGCTGGGTTCAACAGAAAGGCTTCTCCCTCATG3ʹ . PCR comprised the following steps: stage I: 94°C for 5 min; stage II ( 35 cycles ) : 95°C for 30 sec , 58°C for 30 sec , and 72°C for 2 . 5 min; stage III: 72°C for 7 min; and stage IV: 4°C . Gateway cloning was performed to generate the final expression construct , namely , pTol2-ifabp: RPIA; myl7: EGFP , using a MultiSite Gateway Three-Fragment Vector Construction Kit ( Invitrogen ) . The transgenic zebrafish model was established via microinjections of the above constructs , which were performed as described elsewhere , and the transgenic fish were selected as described previously [50] . The F2 generations of Tg ( ifabp: RPIA; myl7: EGFP ) zebrafish ( n = 36 ) and AB line ( WT ) zebrafish which served as controls ( n = 12 ) were analyzed in this study . The zebrafish were maintained at the TZCF under an automated 14:10-h light:dark cycle and a constant temperature of 28°C under continuous flow . MS 222/tricaine methanesulfonate ( 160 mg/L ) were applied for anesthesia .	The pentose phosphate pathway generates NADPH , pentose , and ribose-5-phosphate by RPIA for nucleotide synthesis . Deregulation of RPIA is known to promote tumorigenesis in liver , lung , and breast tissues; however , the molecular mechanism of RPIA-mediated CRC is unknown . Here , we demonstrate a role of RPIA in CRC formation distinct from its role in these other tissues . We showed that RPIA is significantly elevated in CRC . RPIA increased cell proliferation and oncogenicity via activation of β-catenin , with RPIA entering the nucleus to form a complex with APC and β-catenin . Further investigation suggested that RPIA protects β-catenin by preventing its phosphorylation , ubiquitination , and subsequent degradation . In addition , the C-terminus of RPIA ( amino acids 290 to 311 ) , a portion of the protein not previously characterized , is necessary for RPIA-mediated tumorigenesis . Finally , we observed that transgenic expression of RPIA increases the expression of β-catenin and its target genes and induces tumorigenesis . Our findings suggest that RPIA can enter the nucleus and associate with APC/β-catenin , and suggest precise treatment of human CRC by targeting its nonenzymatic domain .	[ "Abstract", "Introduction", "Results", "Discussion", "Materials", "and", "methods" ]	[ "medicine", "and", "health", "sciences", "fish", "immune", "cells", "gene", "regulation", "antigen-presenting", "cells", "biological", "cultures", "cell", "processes", "cancers", "and", "neoplasms", "vertebrates", "immunology", "animals", "oncology", "animal", "models",...	2018	Identification of a noncanonical function for ribose-5-phosphate isomerase A promotes colorectal cancer formation by stabilizing and activating β-catenin via a novel C-terminal domain
Cornelia de Lange Syndrome ( CdLS ) is a multi-organ system birth defects disorder linked , in at least half of cases , to heterozygous mutations in the NIPBL gene . In animals and fungi , orthologs of NIPBL regulate cohesin , a complex of proteins that is essential for chromosome cohesion and is also implicated in DNA repair and transcriptional regulation . Mice heterozygous for a gene-trap mutation in Nipbl were produced and exhibited defects characteristic of CdLS , including small size , craniofacial anomalies , microbrachycephaly , heart defects , hearing abnormalities , delayed bone maturation , reduced body fat , behavioral disturbances , and high mortality ( 75–80% ) during the first weeks of life . These phenotypes arose despite a decrease in Nipbl transcript levels of only ∼30% , implying extreme sensitivity of development to small changes in Nipbl activity . Gene expression profiling demonstrated that Nipbl deficiency leads to modest but significant transcriptional dysregulation of many genes . Expression changes at the protocadherin beta ( Pcdhb ) locus , as well as at other loci , support the view that NIPBL influences long-range chromosomal regulatory interactions . In addition , evidence is presented that reduced expression of genes involved in adipogenic differentiation may underlie the low amounts of body fat observed both in Nipbl+/− mice and in individuals with CdLS . Cornelia de Lange Syndrome ( CdLS; OMIM#122470 ) is characterized by developmental abnormalities of the cardiopulmonary , gastrointestinal , skeletal , craniofacial , neurological , and genitourinary systems [1]–[3] . The clinical presentation ranges from subtle dysmorphology to conditions incompatible with postnatal life . Common structural birth defects observed in CdLS include upper limb reduction ( significant in just under half of cases ) , cardiac abnormalities ( especially atrial and ventricular septal defects ) , and craniofacial dysmorphia ( including dental and middle ear abnormalities , occasional clefting of the palate , and highly characteristic facies ) [2]–[8] . Other findings include small head size , lean body habitus , hirsutism , ophthalmologic abnormalities , pre- and postnatal growth retardation , and structural abnormalities of the gastrointestinal tract ( duodenal atresia , annular pancreas , small bowel duplications ) [2] , [3] , [9]–[11] . Physiological disturbances in CdLS include moderate to severe mental retardation [12] often accompanied by autistic behaviors [13] , and severe gastrointestinal reflux [14] . Although prevalence has been estimated at between ∼1/10 , 000 and 1/50 , 000 births [8] , [15] , wide phenotypic variability in the syndrome makes it likely that large numbers of mildly-affected individuals are not being counted . A genetic basis for CdLS was uncovered in 2004 with the demonstration that many affected individuals carry mutations in Nipped-B-like ( NIPBL ) , so named for its homology to the Drosophila gene , Nipped-B [16] , [17] . Heterozygous NIPBL mutations are found in about 50% of individuals with CdLS [18] . As many of these mutations are expected to produce absent or truncated protein , haploinsufficiency is the presumed genetic mechanism [19] . NIPBL/Nipped-B protein is found in the nuclei of all eukaryotic cells , where it interacts with cohesin , the protein complex that mediates sister chromatid cohesion [20] , [21] . The NIPBL ortholog in fungi plays a role in loading cohesin onto chromosomes , and a role in unloading has been suggested as well . The fact that a minority of cases of mild CdLS result from mutations in the SMC1L1/SMC1A ( ∼5%; OMIM 300590 ) and SMC3 ( 1 case; OMIM 610579 ) genes , which encode two of the four cohesin structural components , supports the view that CdLS is caused by abnormal cohesin function [22] , [23] . Consistent with the hypothesis that cohesin plays important roles during embryonic development , it was found that mutations in the cohesin regulatory protein ESCO2 cause Roberts'-SC phocomelia syndrome , another multi-organ systems birth defects syndrome [24] , [25] . In mice , deletion of the cohesin regulators PDS5A and PDS5B also produces a wide variety of developmental defects , some of which overlap with CdLS [26] , [27] . In addition , there has recently been a report of one family showing atypical inheritance of CdLS , in which both affected and unaffected siblings harbor a missense mutation in the PDS5B gene , raising the possibility of some genetic association between PDS5B and CdLS [27] . How alterations in cohesin function give rise to pervasive developmental abnormalities is largely unknown . Cohesin is involved in sister chromatid cohesion and DNA repair in many organisms , but observed alterations in cohesion and repair in individuals with CdLS are mild at best [28] , [29] . More recently , observations in model organisms and cultured cells have suggested that cohesin plays important roles in the control of transcription [reviewed in 18] . In Drosophila , for example , changes in levels of Nipped-B or cohesin structural components alter the expression of developmental regulator genes , such as homeodomain transcription factors [30]–[33] . Such effects on gene expression , which have been proposed to reflect the disruption of long-range promoter-enhancer communication , occur with small changes in Nipped-B or cohesin levels that do not produce cohesion defects; they can also occur in postmitotic cells , in which chromosome segregation is presumably not an issue [18] . Studies using Drosophila cell lines have demonstrated that cohesin and Nipped-B binding are concentrated near the promoters of active transcriptional units [34] . In mammalian cells , cohesin often binds , in an NIPBL-dependent manner , to sites occupied by the transcriptional insulator protein CTCF , where it plays a significant role in CTCF function [35]–[37] . Recently , NIPBL has also been shown to bind and recruit histone deacetylases to chromatin [38] . These observations suggest that cohesin and NIPBL may interact in multiple ways with the transcriptional machinery . As a first step toward understanding the molecular etiology of CdLS , we generated a mouse model of Nipbl haploinsufficiency , which replicates a remarkable number of the pathological features of CdLS . Cellular and molecular analysis of mutant cells and tissues revealed widespread , yet subtle , changes in the expression of genes , some of which are found in genomic locales in which transcription is known to be controlled through long-range chromosomal interactions . We propose that the aggregate effects of many small transcriptional changes are the cause of developmental abnormalities of CdLS , and present evidence that one set of transcriptional changes may explain the notably lean body habitus of many individuals with CdLS . Two mouse ES cell lines bearing gene-trap insertions into Nipbl were obtained and injected into C57BL/6 blastocysts to produce chimeras ( see Materials and Methods ) . Male chimeras were bred against both inbred ( C57BL/6 ) and outbred ( CD-1 ) mice . For only one cell line ( RRS564 , which contains a beta-geo insertion in intron1 , and is predicted to produce a truncated transcript with no open reading frame; Figure S1 ) was ES cell contribution to the germline obtained ( as scored by coat color; Table 1 ) . Whereas Mendelian inheritance predicts that half the germline progeny of chimeric mice should be heterozygous ( Nipbl+/− ) for the gene trap insertion , the observed frequency was much lower . When chimeras were bred against CD-1 females , 22 out of 113 germline progeny ( 19% ) carried the mutant allele ( Table 1 ) . With C57BL/6 females only one out of 18 germline progeny carried the mutation ( 5 . 5% ) , and this animal , although male , did not produce any progeny when subsequently mated . In view of these data , it was decided that further analysis of the Nipbl− allele would take place through outcrossing onto the CD-1 background . As shown in Table 1 , when the Nipbl+/− offspring of chimera by CD-1 crosses ( N0 generation ) were bred against wildtype ( CD-1 ) females , 17% of surviving adult progeny carried the mutant allele . When animals of this “N1 generation” were again outcrossed against CD-1 , 18% of surviving progeny ( N2 ) carried the mutant allele . Similar survival ratios were observed for subsequent generations of outcrossing . The data imply that 75–80% of Nipbl+/− mice die prior to genotyping ( typically done at 4 weeks of age ) , a fraction that remains stable as the mutant allele is progressively outcrossed onto the CD-1 background . To determine whether lethality occurs in utero , we examined litters for Nipbl+/− embryos just before birth ( gestational days E17 . 5 and E18 . 5 ) . With no visible marker available for the ES-cell derived progeny of chimeras , this test was carried out with progeny of the N0 generation , in which the Mendelian expectation for the mutant allele is 50% . Mutants were found to comprise 41% ( 30 out of 67 ) of progeny , a frequency not significantly different from expected for this sample size ( Table 1 ) . These data imply that most mutants die at or after birth . To evaluate the extent to which Nipbl+/− mice provide a good model for CdLS , we performed an analysis in which we examined these animals for a number of different structural phenotypes analogous to common clinical findings observed in CdLS ( summarized in Table S1 ) . Among the most common clinical features of CdLS are small body size , often evident before birth; heart defects; and upper limb abnormalities ranging from small hands to frank limb truncations [2]–[5] , [7] , [8] , [39] . As shown in Table 2 , Nipbl+/− embryos examined shortly before birth ( E17 . 5–E18 . 5 ) were 18–19% smaller than wildtype littermates ( P<0 . 001 ) , a reduction not accompanied by decreased placental size . Nipbl+/− embryos at earlier stages were also noted to be slightly smaller than littermates ( data not shown ) . Nipbl+/− embryos did not display limb or digit truncations , or obvious loss of any other bony elements . However , upon staining embryonic skeletons , we observed delays in ossification of both endochondral and membranous bones of Nipbl+/− embryos . As shown in Figure 1A–1D , delayed ossification of the skull and digits was apparent between E16 . 5 and E18 . 5 . Measurement of long bones and digits at E17 . 5 revealed , in addition to a symmetrical reduction in bone length ( consistent with smaller body size ) , a significant decrease in the relative extent of ossification ( Figure 1E ) . Otherwise , the patterning of cartilaginous elements was relatively normal , although some subtle differences in morphology were consistently observed , e . g . the shape of the olecranon process of the ulna was consistently abnormal in Nipbl+/− mouse embryos ( Figure 1F–1G ) . Interestingly , dys- and hypoplastic changes of the ulna are common findings in CdLS [40] . Among the cardiac defects that occur in CdLS , atrial and ventricular septal defects are especially common [2] , [5] , [7] . Atrial septal defects , which were typically large , were observed in about half of Nipbl+/− mouse embryos , ( Figure 1H–1K; Table S1 ) , and could be detected as early as E15 . 5 , shortly after atrial septation normally finishes . A reduction in atrial size was also seen in some mutants , but was not a consistent finding . No defects were detected in the atrioventricular valves or septum , outflow tract , or pulmonary vasculature . However , many mutant embryos displayed subtle abnormalities of the ventricular and interventricular myocardium , including abnormal lacunar structures and disorganization of the compact layer , especially near the apex ( data not shown ) . Significantly , no histological or functional cardiac abnormalities were detected among mutant mice that survived the perinatal period ( data not shown ) . This implies that the cause of perinatal mortality is either cardiac , or correlates strongly with the presence of cardiac structural defects . Histological examination of other organ systems in late embryonic mutant mice revealed no obvious anatomical abnormalities of the lungs , diaphragm , liver , stomach , spleen , kidney or bladder . Brains of neonatal Nipbl+/− mice displayed relatively normal gross anatomy , although a single mutant was observed to have a large brainstem epidermoid cyst ( not shown ) . Most Nipbl+/− mice that survived the perinatal period reached adulthood , and appeared to have a normal lifespan . However , marked decrease in the body size of mutant mice was evident at birth and throughout all ages ( Figure 2A and 2B ) . Indeed , the 18–19% weight difference between mutant and wildtype mice observed before birth ( Table 2 ) widens to 40–50% by postnatal weeks 3–4 ( Figure 2C–2E; this finding has remained consistent over 6 generations [data not shown] ) . To investigate early postnatal growth of Nipbl+/− mice in more detail , litters fathered by N1 and N2 generation animals were subjected to daily weighing from shortly after birth until sexual maturity ( 5–6 weeks of age; Figure 2F ) . Most mutant mice exhibited failure to thrive during the first weeks of life , with many undergoing several days of wasting followed by death ( Figure 2F , inset ) . By 3 weeks of age , the average weight of surviving mutants was only 40% of wildtype , but after weaning this pattern abruptly changed: mutants ( even ones that had already begun to show wasting ) underwent rapid catch-up growth ( Figure 2F ) , such that by 9 weeks of age they had reached 65–70% of wildtype weight . These observations suggest that , in addition to being intrinsically small , Nipbl+/− mice may have difficulty with suckling , or may receive inadequate nutrition from milk . Remarkably , the weights of children with CdLS also fall further behind age norms during the first year of life , but show significant catch-up growth later on [11] . The distinctive craniofacial features of CdLS , including microbrachycephaly , synophrys , upturned nose , and down-turned lips , play an important role in clinical diagnosis [3] , [6] . Micro-CT analysis was used to assess whether Nipbl+/− mice also display consistent craniofacial changes . Analysis of the skulls of 63 adult mice showed significantly smaller size ( microcephaly ) among all mutants ( N = 23 ) , as well as a variety of significant shape changes ( Figure 3 ) . The latter included foreshortening of the anterior-posterior dimensions of the skull ( i . e . brachycephaly ) and an upward deflection of the tip of the snout ( Figure 3B–3E ) . The upturned nares ( Figure 3C and 3E ) reflect reduced size of the ethmoid and sphenoid bones , which produces a sunken midface . Together , these shape changes in the basicranium and face are consistent with a greater reduction in the size of chondrocranial , as opposed to dermatocranial , elements within the skull . In addition , an 8% average decrease in bone thickness was also observed ( ANOVA , df = 47 , F = 18 . 6 , p<0 . 01 ) . Neurological abnormalities in CdLS include mental retardation , abnormal sensitivity to pain , and seizures [41] . Although Nipbl+/− mice have not been subjected to intensive long-term neurological or behavioral tests , several distinctive behaviors were observed: Repetitive circling ( Videos S1 , S2 , S3 ) was noted in 20% ( 34/173; 15 females and 19 males ) of adult Nipbl+/− mice ( >5 weeks of age ) , across all generations examined ( N0–N4 ) . Repetitive behaviors—including twirling in place [42]—are common symptoms in children with CdLS . In addition , 30% ( 4/13; all males ) of Nipbl+/− mice were noted to adopt opisthotonic postures in response to administration of a normal anesthetic dose of avertin ( see Materials and Methods ) , strongly suggesting seizure activity . Seizures are also common in individuals with CdLS [43] , [44] . We also observed that 15% of Nipbl+/− adult mice ( 24/158; 11 females and 13 males ) displayed reflexive hindlimb clasping when suspended by their tails ( Videos S4 , S5 ) , whereas only 2% ( 6/268 ) littermates showed the same behavior ( Table S1 ) . Hindlimb clasping has been observed in several mouse models of neurological disorders , including Rett's syndrome [45]–[47] , mucolipidosis type IV [48] , infantile neuroaxonal dystrophy and neurodegeneration with brain iron accumulation [49] , and Huntington's disease [50]–[53] . Histological examination of mutant brains revealed the presence of all major brain structures , grossly normal lamination of the cerebral and cerebellar cortices , but an overall reduction in brain size , consistent with a 25% reduction in endocranial volume observed with micro-CT ( Figure 4A , two-tailed T-test , df = 28 , T = 5 . 7 p<0 . 01 ) . Absence or reduction in size of the corpus callosum was occasionally observed in Nipbl+/− mice ( Figure 4B ) . Obvious patterning defects were noted only in the midline cerebellum , where lobe IX displayed specific reductions ( Figure 4C ) . Interestingly , midline cerebellar hypoplasia is one of the few consistently-reported changes in brain anatomy in CdLS [54]–[56] . Children with CdLS display a range of ophthalmological abnormalities including ptosis , microcornea , nasolacrymal duct obstruction , strabismus , blepharitis and conjunctivitis [57]–[59] . We noted that 22% of Nipbl+/− mice exhibited one or more gross ophthalmological abnormalities ( Table S1 ) . Most frequently observed was ocular opacification , observed in 14% of animals ( Figure 4D ) ; opacities were often evident as early as three weeks of age . In several cases , this condition was associated with marked periorbital inflammation , and progressed to permanent closure of the eyelids ( not shown ) . Histological analysis revealed inflammatory and fibrotic changes within the corneal epithelium and stroma ( Figure 4E ) , consistent with repeated abrasion or injury . Such injury might arise from neglect due to abnormalities in corneal sensation , from abnormal production or composition of tear fluid , or secondary to periorbital inflammation or infection ( e . g . blepharitis; cf . Table S1 ) . Some degree of hearing loss is observed in almost all individuals with CdLS , and this may play a role in the marked speech disability often seen in this syndrome [60] , [61] . To assess hearing in Nipbl+/− mice , we measured auditory brainstem evoked responses ( ABR [62] ) . Abnormalities were found in the majority of mutant mice examined ( Table S1 ) . In a few cases , markedly increased thresholds to stimulation were observed ( Figure 4F ) . More commonly , stimulus thresholds were within normal limits , but the relative intensities of the components of the ABR were altered . In particular , mutant mice displayed a characteristic reduction in the amplitude of the third peak ( at about 3 msec following stimulus ) , a latency consistent with an abnormality in the auditory nerve and/or early brainstem neural pathways ( Figure 4G ) . The Nipbl564 gene-trap mutation is expected to produce a truncated message lacking all but the first exon ( Figure S1 ) . Therefore , the level of full-length Nipbl mRNA in Nipbl+/− mice should provide an indication of the activity of the wildtype allele . To measure this level , we used an RNase protection assay based on hybridization to sequences found in exons 10 and 11 . Total RNA was analyzed from two tissues: adult liver and E17 . 5 brain , using age-matched littermate controls . As shown in Figure 5 , Nipbl levels in mutants , as a percentage of wildtype levels , were 72–82% in adult liver , and ∼70% in embryonic brain . When western blotting was used to quantify levels of NIPBL protein in Nipbl+/− embryo fibroblasts ( MEFs ) , a reduction to about 70% of wildtype levels was observed ( Figure S2 ) . The observation that Nipbl+/− mice exhibit only a 25–30% decrease in transcript and protein expression , rather than an expected decrease of 50% , is consistent with Nipbl gene being autoregulatory . An alternative explanation is that the mutant allele is “leaky” , i . e . alternative splicing around the gene trap cassette produces some wildtype message . We favor the former explanation because , in both Drosophila and man , the evidence indicates that null mutation of a single allele of Nipped-B/NIPBL produces only a 25–30% drop in transcript levels , the same decrease we observe in Nipbl+/− mice [31] , [63] , [64] . Thus , even if the Nipbl allele studied here is not null , it is probably quite close to being so . More importantly , the degree of decrease in Nipbl expression in Nipbl+/− mice is comparable to that which causes CdLS in man . Overall the data from multiple species strongly argue that pervasive developmental abnormalities result from remarkably small changes in NIPBL levels . There has been one report of precocious sister chromatid separation ( PSCS ) in cell lines derived from individuals with CdLS [28] , which was not seen in a second study [29] . We found no statistically-significant elevation of PSCS in cultured Nipbl+/− MEFs ( Figure S3 ) , Nipbl+/− embryonic stem cells ( data not shown ) , or adult B-lymphocytes ( Figure S3 ) . These results suggest that cohesion defects in the Nipbl heterozygotes , if present , are very subtle; they are also in accord with findings in Drosophila , where PSCS is seen only when both alleles of Nipped-B are mutated [31] . To investigate whether heterozygous loss of Nipbl leads to alterations in transcription , we turned to expression profiling of tissues and cells from Nipbl+/− mice . Because such mice display pervasive developmental abnormalities , transcriptome data can be expected to reflect not only the direct consequences of reduced Nipbl function , but also a potentially large number of transcriptional effects that are secondary consequences of abnormal morphology and physiology . In an effort to minimize the detection of such secondary effects , we focused on profiling samples in which frank pathology was not seen , or had yet to develop by the time of profiling . The samples chosen for analysis were embryonic day 13 . 5 ( E13 . 5 ) brain , and cultures of fibroblasts derived from E15 . 5 embryos ( mouse embryo fibroblasts; MEFs ) . Although mature brain appears to be functionally abnormal in Nipbl+/− mice ( see above ) , at E13 . 5 it at least appears anatomically normal . Cultured MEFs were chosen because they are established with similar efficiency from both mutant and wildtype embryos; exhibit similar morphology and growth characteristics in culture; and by virtue of being maintained ex vivo , are freed of the secondary influences of any systemic metabolic or circulatory derangements within Nipbl+/− embryos . Transcriptome analysis was performed using Affymetrix microarrays . MEF RNA samples were obtained from 10 mutant and 9 wildtype embryos taken from three litters ( 19 separate microarrays ) ; brain RNA was analyzed from 10 mutant and 11 wildtype embryos from two litters ( 21 separate microarrays ) . Gene expression changes were detected in both comparisons . In the brain ( Table S2 ) , 1285 probe sets , corresponding to 978 genes , displayed statistically significant differences in expression between wildtype and mutant mice ( per-probe-set false discovery rate of Q<0 . 05 ) . By and large , the effects were small: 97 . 5% of changes were within 1 . 5-fold of wildtype expression values; >99 . 6% were within 2-fold . The single largest statistically-significant change was 2 . 5-fold . Genes encoding products of virtually all structural and functional categories could be found among those affected , with no dramatic enrichment of any particular functional sets ( by Gene Set Enrichment Analysis [65]; data not shown ) . In cultured Nipbl+/− MEFs , 89 probe sets , corresponding to 81 genes ( Table S3 ) , displayed statistically-significant ( Q<0 . 05 ) differences in expression between wildtype and mutant mice . Again , effects were small: 89% of changes were within 1 . 5-fold of wildtype , and 99% were within 2-fold . The single largest statistically-significant change was 2 . 1-fold . The lower number of transcriptional changes identified in MEFs versus brain may not be biologically meaningful , as MEFs happened to display a somewhat higher average within-sample variance than E13 . 5 brain , making it more difficult for small changes to be judged significant . As with embryonic brain , transcriptional effects in MEFs involved genes that encode a wide variety of proteins . Although automated analyses failed to single out any particular functional class as being highly overrepresented , manual curation revealed significant changes in the expression of a number of genes implicated in adipogenesis ( Figure 6A ) . For example , Cebpb and Ebf1—which encode transcriptional factors central to the process of adipocyte differentiation [66]–[68]—were both down-regulated in Nipbl+/− MEFs , as were Fabp4 and Aqp7 , well-known adipocyte markers [69] , [70] . Other genes down-regulated in Nipbl+/− MEFs ( Table S3 ) could also be found , through literature searches , to exhibit expression positively correlated with adipocyte differentiation , including Adm , Lpar1 , Osmr , and Ptx3 [69] , [71] , [72] . Several additional genes ( Amacr , Avpr1a , Il4ra , Prkcdp , S100b ) down-regulated in Nipbl+/− MEFs can be inferred , from publicly-available expression data , to be enriched in pre-adipocytes and/or brown or white adipose tissue [73]–[75] . Conversely , Lmo7 , which is normally down-regulated during late adipogenic differentiation [71] , was found to be up-regulated in Nipbl+/− MEFs . Furthermore , we noted that genes such as Cebpa and Cebpd ( transcriptional activators of adipocyte differentiation [66] , [76] ) , Il6 ( a cytokine stimulator of adipocyte differentiation that controls adiposity in man [77] , [78] ) and Socs3 ( an intracellular signaling regulator induced by Il6 [79] ) , were also down-regulated in the MEF samples , but at false-discovery rates slightly too high to permit their inclusion in Table S3 ( Q = 0 . 065 , 0 . 085 , 0 . 075 , and 0 . 17 , respectively ) . Together , these data raise the possibility that Nipbl+/− mice are specifically impaired in adipogenesis . Support for this idea was obtained by weighing intrascapular fat dissected from adult mutant and wildtype littermates [80] . As shown in Figure 6B , both brown and white fat are substantially depleted in Nipbl+/− mice . To correct for the fact that mutant mice are generally smaller than their wildtype littermates , we normalized fat measurements to brain weight ( which scales with overall body size ) . As shown in Figure 6C , even by this measure , Nipbl+/− mice displayed a significant , substantial reduction in body fat . As mentioned earlier , lean body habitus is also a characteristic of CdLS . To investigate whether the reduction in body fat in Nipbl+/− mice reflects an intrinsic defect in the differentiation potential of mutant fibroblasts , we studied adipogenic differentiation in vitro . It is known that embryonic fibroblasts can be converted , in large numbers , to adipocytes by treatment with agents such as glucocorticoids , PPAR-γ agonists , isobutylmethylxanthine and insulin , which stimulate the activity of a core network of pro-adipogenic transcription factors ( C/EBPα , C/EBPβ , C/EBPδ , PPARγ; [81] , [82] ) . In response to such agents , we observed no significant difference between Nipbl+/− and wildtype MEFs in terms of the number of adipocytes or adipocyte colonies produced ( data not shown ) . However , when we omitted these pharmacological agents , and measured the ( much lower ) level of spontaneous adipogenic differentiation that occurs in MEF cultures [83] , we observed a substantially-lower level in mutant cultures ( Figure 6D–6F ) . The observation that Nipbl+/− MEFs are impaired in spontaneous , but not induced , adipogenesis implies that their primary defect does not lie downstream of the targets of pharmacological inducers . Of the 80 genes ( not counting Nipbl itself ) with significant differential expression in Nipbl+/− MEFs ( Table S3 ) , 20% ( 16/80 ) are also found among the 978 genes whose expression was altered in Nipbl+/− embryonic brain ( Table S2 ) . Using a more stringent false discovery rate cutoff of Q<0 . 02 for both samples , we find that 23% ( 9/40 ) of differentially expressed MEF genes are among the 560 that are differentially expressed in brain . These data suggest that common transcriptional targets exist in the two tissues . Further support for this idea is obtained by correlating fold-increase or -decrease of affected transcripts . In this case a less conservative approach to false discovery is justified ( the goal is to estimate overall correlation between samples , not implicate individual genes ) , so the log-fold changes for all probe sets that exhibited differential expression exceeding an arbitrary t-statistic threshold ( t>2 ) in both tissues were plotted against each other ( shown in Figure 7 ) . The data are clearly strongly correlated ( R = 0 . 77 ) , suggesting that at least some of the transcriptional effects of Nipbl deficiency are shared across tissues . Among the genes in which expression changes contributed substantially to the correlation are four members of the protocadherin β cluster ( Pcdh17 , Pcdh20 , Pcdh21 , Pcdh22; all down-regulated ) , Lpar1 ( also down-regulated; encoding the lysophosphatidic acid receptor ) , Vldlr ( down-regulated; encoding a receptor involved in both lipid metabolism and cerebral cortical development ) , and Stag1 ( up-regulated; encoding SA1 , a cohesin component ) . Interestingly , in Drosophila , inhibition of Nipped-B expression also leads to up-regulation of the ortholog of Stag1 [31] . Recently , STAG1 up-regulation has also been seen in lymphoblastoid cell lines of individuals with CdLS [64]; Table S5 . Among the most significant changes common to mutant MEF and brain samples were decreases in expression of transcripts from the 22-gene Pcdhb ( protocadherin beta ) cluster on chromosome 18 ( Table S2 and Table S3 , Figure 7 ) . As shown in Figure 8A , affected transcripts included Pcdhb7 , 16 , 17 , 19 , 20 , 21 and 22 , which lie predominantly at the 3′ end of the cluster . This observation raised the possibility that the transcriptional effects of Nipbl might be related to the physical locations of genes . However , as genes at the 5′ end of the Pcdhb cluster tend to be expressed at lower levels than those at the 3′ end , lower signal-to-noise ratios might have made small changes in expression at the 5′ end more difficult to detect . To resolve this issue , and to provide independent confirmation of microarray data , quantitative RT-PCR was used to measure transcripts levels at multiple locations throughout the Pcdhb cluster ( Figure 8B ) . For these experiments , brain mRNA was prepared at a later developmental stage ( E17 . 5 , when most Pcdhb transcripts are more highly expressed ) from 13 independent samples ( 7 mutant and 6 wildtype embryos ) . Robust RT-PCR signals were obtained for 14 of 15 transcripts tested ( Pcdhb2 , 3 , 4 , 5 , 7 , 8 , 9 , 10 , 13 , 14 , 16 , 17 , 19 , and 22; but not Pcdhb1 ) . As shown in Figure 8B , the data support the microarray results from the earlier embryonic stage , and indicate that most transcriptional changes in Nipbl+/− brain indeed occur preferentially at the 3′ end of the cluster ( Pchdb13 , 14 , 15 , 16 , 17 , 19 , 22 ) . Additionally , they suggest that at least one 5′ gene , Pcdhb2 , may also be affected . A more revealing analysis of the data can be obtained by correlating Pcdhb transcript levels in each tissue sample , regardless of genotype , against Nipbl transcript levels within that sample ( i . e . treating Nipbl expression as a quantitative trait; Figure 8C , Figure S4 ) . This approach offers greater discriminatory power because Nipbl expression in individual samples varies significantly , even within mutant and wildtype groups , and occasionally overlaps between the two groups . Indeed , the results of the analysis indicate that Pcdhb expression correlates strongly with Nipbl transcript level , lending support to the view that Pcdhb transcription is directly affected by the amount of NIPBL present in cells . In Figure 8C , the results of such correlations for all 13 tested Pcdhb transcripts are summarized by plotting the slopes of regression lines ( the sensitivity of each transcript's expression to Nipbl level ) against gene location , with error bars reflecting the strength of correlation for each gene . The results strongly suggest a continuum of sensitivity to Nipbl across the entire Pcdhb cluster , with genes at both the 5′ and 3′ ends being the most sensitive , and those in the middle being least affected . We show here that mice heterozygous for a gene-trap mutation upstream of the first coding exon of Nipbl displayed many features of human CdLS , including pre- and postnatal growth retardation , cardiac septal defects , delayed bone development , lean body habitus , microbrachycephaly with characteristic craniofacial changes , behavioral disturbances , ophthalmological abnormalities , cerebellar hypoplasia , and hearing deficits ( Figures 1–4 , Table S1 , Videos S1 , S2 , S3 , S4 , S5 ) . These phenotypes remained stable through many generations of outcrossing , and occurred in the context of modest ( 25–35% ) reductions in levels of Nipbl mRNA in every tissue measured ( embryonic brain , MEFs , adult liver ) . Similarly modest reductions have recently been reported in cell lines derived from individuals with CdLS [63] , [64] . In some cases , quantitative agreement between the mouse model and CdLS is remarkable , e . g . fall-off and catch-up in growth rates during early postnatal life , the upturned nose . Yet some common features of CdLS are not observed in the mouse model , such as reduction and fusion abnormalities of the upper limb , which is seen in up to 30–50% of children with CdLS ( depending on criteria used ) . The mutant mouse heart also displays only atrial and not ventricular septal defects , whereas both occur at similar frequency in CdLS . Mutant mice also display some pathological features , such as corneal opacities , that are atypical of CdLS ( corneal scarring has been noted , however [58] ) . Furthermore , the frequency of perinatal mortality in CdLS is estimated at about 10% [8] , not as high as in the mutant mouse ( although this may simply reflect better postnatal care ) . Despite these differences , it is clear that the Nipbl+/− mouse is an excellent animal model for many features of CdLS , and provides the first experimental verification that Nipbl mutations cause the syndrome . Interestingly , wide variation in the penetrance or severity of phenotypes , a distinctive feature of CdLS , was also observed in mutant mice . Because the mouse line was maintained on an outbred ( CD-1 ) background ( given high mortality of heterozygotes , it was not practical to maintain the line on an inbred background ) , genetic heterogeneity could have accounted for some of the variability . It is fascinating that the diverse and severe pathology observed in this study is caused by only a 25–35% decrease in the level of Nipbl transcripts ( Figure 5; also see Table S2 and Table S3 ) . A recent study of a rare familial case of CdLS involving a mutation in the 5′-untranslated region of NIPBL suggests that a mere 15% decrease in transcript levels is associated with a clinically significant phenotype [63] . Given the extraordinary sensitivity of development to the level of expression of this one gene , it would not be surprising if an unusually high proportion of disease-causing mutations in Nipbl occur in regulatory DNA , where they would be difficult to detect . This could help explain why such a large proportion of CdLS mutations ( ∼40% ) have yet to be identified [23] , [84] , [85] . Results of the present study support the view that changes in Nipbl level have significant , yet modest , effects on transcription throughout the genome . At present , it is impossible to know how many observed gene expression changes ( Table S2 and Table S3 ) are primary—due to direct transcriptional actions of NIBPL—and how many are downstream consequences of gene misregulation . Among the affected MEF transcripts that were noted to participate in adipogenic differentiation , for example , many are transcriptional targets of each other ( Figure 6A ) , raising the possibility that direct actions of NIPBL may be confined to a subset of these . Among the most likely candidates for direct NIPBL “targets” are those genes that displayed similar expression changes in both cultured MEFs and E13 . 5 brain ( Figure 7 ) . Prominent among these were genes of the protocadherin beta ( Pcdhb ) cluster . Measurements in later-stage brain confirmed that alterations in gene expression occur throughout the Pcdhb locus in Nipbl+/− mice , but in a manner that is positionally graded across the cluster ( Figure 8 ) . Such effects are consistent with a role for NIPBL in the long-range , coordinated regulation of sets of genes . Additional evidence for this hypothesis can be found in the E13 . 5 brain expression data: As shown in Table S4 , there are at least 13 other examples of small clusters ( usually 2–4 genes ) of related ( paralogous ) genes , in which Nipbl+/− mice display similar expression changes in more than one paralog . Two of these are the well-studied β- and α-globin loci [86] , in which long-range cis-regulatory elements ( locus control regions ) are known to control and coordinate expression of different transcripts ( the globin transcripts in brain RNA presumably come from fetal erythrocytes in the tissue ) . Interestingly , whereas decreases in expression were seen at all four β-globin genes in Nipbl+/− samples , the magnitudes varied greatly among the genes within each cluster ( arguing against the trivial possibility that Nipbl+/− brains simply have less blood in them ) . In fact , the single greatest gene expression change in the entire study ( ∼2 . 5 fold decrease ) involved one of the transcripts ( Hbb-bh1 ) of the β-globin locus . It is known that the transcriptional insulator protein CTCF plays an important role in establishing chromatin boundary elements at the β-globin locus [87] . The Pcdhb locus is also flanked , at least in man , by sites occupied by CTCF [88] , the functional significance of which has yet to be studied . CTCF insulation is also involved in control of the myc-locus [89] , which is highly significantly down-regulated in E13 . 5 Nipbl+/− brain ( Table S2 ) . Even the Igf2/H19 locus , at which long-range , CTCF-dependent regulation of gene silencing has been shown to occur [90] , [91] , displayed evidence of H19 down-regulation ( by ∼20% ) in the Nipbl+/− brain , albeit at lower statistical significance ( Q = 0 . 14 ) . In view of recent work showing that cohesin and CTCF binding sites extensively co-localize in the mammalian genome ( including at the β-globin , Igf2/H19 and myc loci [35]–[37] ) , and that cohesin contributes to CTCF function [35] , [37] , it is reasonable to speculate that at least some of the transcriptional effects in Nipbl+/− tissues arise from impaired CTCF function . It should be noted , however , that CTCF sites are far more common ( >13 , 000 per mammalian genome ) than Nipbl-sensitive genes , and we find no clear correlation between the locations of Nipbl-sensitive genes ( or the magnitudes of transcriptional effects in those genes ) and known or predicted CTCF-sites ( X . Xie and N . Infante , personal communication ) . It remains possible that only a subset of NIPBL transcriptional effects is related to CTCF function . Indeed , it is possible that NIPBL acts primarily by influencing other aspects of long-range cis-regulatory interaction ( e . g . histone methylation , DNA looping ) , which simply take place frequently at CTCF-regulated loci . Despite the extensive overlap between the phenotypes of Nipbl+/− mice and CdLS , it is interesting to note that the gene expression changes recently reported in lymphoblastoid cell lines of CdLS individuals [64] exhibit only limited overlap ( 6–8%; cf . Table S5 ) with those we observed in mouse embryo fibroblasts and embryonic brain ( Table S2 and Table S3 ) . So far it is unclear whether this stems from a high degree of tissue-specificity in the expression of genes that are directly affected by NIPBL; a high proportion of indirect NIPBL targets ( which might be more likely to vary from tissue to tissue ) ; or the effects of differences in genomic organization between mouse and man . Among the gene expression changes detected in Nipbl+/− MEFs and brain ( Table S2 and Table S3 ) one can find many genes that , when mutated in mice or man , produce phenotypes that overlap with CdLS . These include skeletal and craniofacial abnormalities ( Lpar1 , Pitx2 , Satb2 , Tcof1 , Trps1 ) ; heart defects ( Adm , Cited2 , Cxcl12 , Gja1 , Hey2 , Pitx2 , Mef2c ) ; reduced body size ( Ebf1 , Lpar1 , Hsd3b7 , Mef2c ) ; decreased adiposity ( Cebpb , Ebf1 , Lpar1 , Npy , Vldlr ) ; behavioral abnormalities ( Avpr1a , Ctnnd2 , Lpar1 , Vldlr ) ; seizures ( Cdk5r1 , Gabrb1 , Gabrb2 , Neto1 , Nr4a3 , Plcb1 , S100b , Sv2b ) ; and hearing deficits ( Cldn11 , Eya , Gjb2 , Gjb6 ) [66] , [92]–[124] . For most of the genes mentioned above , however , phenotypes are observed only with complete loss of gene function . In the few cases in which significant heterozygous phenotypes are seen ( e . g . Satb2 , Pitx2 , Trps1 , Tcof1 , Cited2 ) , expression changes in the same genes in Nipbl+/− samples are not even as great as would be expected for heterozygous loss . Of course , it is possible that greater expression changes occur at other stages , or in other tissues , than those sampled here . However , the alternative interpretation is that phenotypes in Nipbl+/− mice , and in individuals with CdLS , arise from the collective effects of small changes in the expression of many genes . Although we are not yet in a position to distinguish between these hypotheses , we recognize that this issue is closely related to a major unanswered question in human genetics: whether most common disease phenotypes arise from large effects at a few loci , or from very many loci of small effect . The results of the present study suggest that further study of CdLS , and related “cohesinopathies” [125] , [126] , could shed light on a fundamental question of widespread importance . Ethics Statement: All animals were handled in strict accordance with good animal practice as defined by the relevant national and/or local animal welfare bodies , and all animal work was approved by the University of California Irvine Institutional Animal Care and Use Committee ( protocol 1998-1656 ) . A search for Nipbl sequences in mouse gene-trap databases ( http://www . genetrap . org/ ) initially identified two targeted ES cell lines ( generated using the E14 parental cell line , which has a 129/Ola background ) . One of these ( RRS564 ) contains a gene-trap in the intron between exon1 and exon 2; the other ( RRJ102 ) in intron 25 ( the exon numbering of [16] for the human gene is used here ) . Gene-trap constructs are designed to terminate transcription and translation , producing a truncated or absent protein product . Both cell lines were injected into blastocysts of C57BL/6 mice ( for the RRJ102 cell line , 83 blastocysts were injected; for RRS564 , 324 blastocysts were injected ) . Multiple male chimeras were obtained and bred against outbred ( CD-1; Charles River ) females . Germ line progeny ( distinguishable by chinchilla coat color ) were obtained only from RRS564-derived chimeras , and further work on RRJ102 was suspended . The RRS564 allele is hereafter referred to as Nipbl564 and mice heterozygous for this allele as Nipbl+/− for simplicity . Nipbl+/− mice were maintained under normal laboratory conditions , and the line propagated by successive rounds of outcrossing to CD-1 mice . Offspring were genotyped using LacZ- ( Forward 5′-TGATGAAAGCTGGCTACAG-3′ and Reverse 5′-ACCACCGCACGATAGAGATT-3′ ) primers . Anatomical and histological evaluations were performed using fresh-frozen or paraformaldehyde fixed tissues . In some cases , fixation was carried out by cardiac perfusion . Alcian Blue/Alizarin Red staining was carried out as described [127] . Hematoxylin-eosin and cresyl violet staining was carried out using standard techniques . Micro-CT analysis of adult ( >90 days ) skulls was performed using a Scanco VivaCT as described [128] , [129] . Craniofacial shape was assessed using geometric morphometric techniques , and cranial vault thickness was assessed in 3D as described [130] , [131] . Scapular fat pads were dissected and measured as described [80] . Auditory brainstem response recordings were generated as described [62] . Briefly , a cohort of young adult mutant and littermate control animals were anesthetized with avertin ( 2 . 5% solution of tribromoethanol in tert-amyl alcohol; 20 µl/g body weight administered by i . p . injection ) , and subcutaneous electrodes inserted at the level of the brainstem to record neural potentials evoked by a variety of clicks and tones introduced into one ear . Embryo fibroblasts ( MEFs ) were cultured from E15 . 5 Nipbl+/− and wildtype littermate embryos as described [127] . Metaphase spreads of MEFs were prepared from cells cultured for 12 hrs in medium supplemented with 0 . 1 µg/ml colchicine . Trypsinized cells were pelleted , incubated in 75 mM KCl for 20 min at 37°C , then re-pelleted and fixed in 3∶1 methanol/acetic acid . Cells were dropped onto glass slides and stained with 4′-6-Diamidino-2-phenylindole or Giemsa . Microscopic assessment was carried out for each slide by three independent observers who were blinded to the genotypes of the sample . B cells were isolated from mouse spleens by immunomagnetic depletion with anti-CD43 beads ( Miltenyi Biotech ) , cultured in RPMI1640 with 10% fetal bovine serum , and stimulated with lipopolysaccharide ( 25 µg/ml; Sigma ) and IL4 ( 5 ng/ml; Sigma ) for 3 days . Cells were arrested at mitosis by treatment with 0 . 1 µg/ml colcemid ( Roche ) for 1 hour , and metaphase chromosome spreads prepared following standard procedures . Cells were stained with 4′-6-Diamidino-2-phenylindole and images of metaphases acquired with an Axioplan2 upright microscope ( Zeiss ) , using Metamorph software . Measurements of Nipbl levels by RNase protection were made according to standard methods . The Nipbl probe contained 39 bases of exon 10 , all 183 bases of exon 11 , and 4 bases of exon 12 . There is no expressed sequence tag evidence supporting alternative splicing of exons 10–11 , and in situ hybridization studies in mouse embryos indicated they are ubiquitously expressed , so it was felt that this probe would provide a good indication of overall levels of expression . Briefly , for each reaction , 20 µg of total RNA was hybridized with 32P-labeled probes for Nipbl ( 90 , 000 cpm ) and Gapdh ( glyceraldehyde 3-phosphate dehydrogenase; containing 116 bases of exon 4 and 15 bases of exon 5; 20 , 000 cpm ) and processed according to manufacturer's instructions ( Ambion RPA III kit ) . Samples were run on a 5% polyacrylamide/8M urea gel , dried , and bands quantified by phosphorimager . For measurement of NIPBL protein levels , MEFs were lysed in cold buffer containing 50 mM Tris ( pH 7 . 4 ) , 150 mM NaCl , 1% Triton X-100 , 1% sodium deoxycholate , 0 . 1% SDS , and protease inhibitors [1 mM phenylmethylsulfonyl fluoride ( PMSF ) , 1 mM EDTA , 2 µg/ml aprotinin , 2 µg/ml leupeptin , and 2 µg/ml pepstatin] . Total protein ( 20 µg ) from 3 wildtype MEF homogenates and 3 Nipbl+/− MEF homogenates was separated , in duplicate , on 7 . 5% SDS-PAGE gels and transferred onto Immobilon-P membranes ( Millipore , Bedford , MA ) . The membranes were blocked with 2% BSA and sequentially incubated with the anti-GAPDH ( 1∶200 , 000 , 6C5; Ambion , Austin , TX ) and anti-NIPBL-N ( 1∶60000 , anti-NIPBL antibody was produced in rabbit from a GST fusion protein containing amino acids 1–380 of human NIPBL , and was affinity purified using original antigen ) . The membranes were then incubated with the horseradish peroxidase-conjugated anti-rabbit antibody ( 1∶10000 ) and detected using chemiluminescence . Images were scanned and densitometry performed . To evaluate spontaneous adipogenic differentiation , MEFs ( passage 2 ) were seeded into 96-well plates at 7 , 500 cells/well , maintained in Dulbecco's modified Eagle's Medium with 10% fetal bovine serum , 100 U/ml penicillin and 100 µg/ml streptomycin at 37°C , cultured to confluence , and maintained for 7 additional days . Lipid accumulation was visualized by staining with Oil Red O . Briefly , cells were washed with PBS and fixed with 10% formaldehyde ( 30 minutes at room temperature ) , rinsed , and permeabilized in 60% isopropanol for 2–5 minutes . Isopropanol was removed and Oil Red O solution ( Chemicon , Temecula , CA ) was added for 30 minutes . Wells were washed several times with PBS then counterstained with Hoechst 33258 ( 0 . 01 mg/ml ) . Four wildtype and five Nipbl+/− MEF lines were analyzed . Each was plated in triplicate wells and 7 fields ( at 10× magnification ) within each well were photographed and lipid-containing cells ( stained lipids appear red under phase contrast ) and nuclei ( appear blue with fluorescent microscopy ) were counted using Image J software ( NIH ) to detect nuclei ( nucleus counter plug-in: approximately 1000–1200 nuclei were detected per field ) and lipid-containing cells ( point picker plug-in ) . Analysis of gene expression was performed on total RNA isolated using the trizol method from MEFs , adult liver , or manually dissected E13 . 5 brain . Hybridization and data collection were carried out by the Broad Institute ( Cambridge , MA ) . RNA was labeled and hybridized to Affymetrix Murine 430A 2 . 0 ( for MEFs ) or 430 2 . 0 ( for brain ) array chips using the protocol described at http://www . broad . mit . edu/mpr/publications/projects/Leukemia/protocol . html , and data were analyzed using GenePattern software ( http://www . broad . mit . edu/cancer/software/genepattern ) . Expression data sets were assembled from individual CEL files using the RMA algorithm with quantile normalization . Data were log2-transformed and transcripts with near-background expression filtered . Measures of statistical significance were obtained by permutation testing [132] , using the Comparative Marker Selection module . Significance is presented in terms of per-sample false discovery rates , or Q-values [133] . Data were also analyzed using D-chip software ( http://www . biostat . harvard . edu/complab/dchip ) , which yielded similar enrichment sets ( not shown ) . Probe sets were annotated , and gene locations obtained , according to the NCBI m37 mouse assembly , and Affymetrix annotation files ver . 28 ( March 2009 ) . For measurements of transcript abundance by quantitative PCR , RNA ( 5 µg ) from E17 . 5 mouse brains was reverse transcribed with Superscript II , oligo dT , and random hexamers according to manufacturer's instructions ( Invitrogen , Carlsbad , CA ) . Reactions were assembled using iQ SYBR Green Supermix ( BioRad , Hercules , CA ) and processed in 20 µl volumes with 1 µl of cDNA ( diluted 1∶25 ) and primers at a final concentration of 100 nM . Specificity of amplification was verified for each reaction by examination of the corresponding melt curve . Normalization was carried out using beta-2 microglobulin as a standard , and genomic amplification controlled for using samples prepared without reverse transcription . All PCR reactions were performed on an iQ5 iCycler ( BioRad ) . Cycling conditions were 95°C for 4 min and then 40 cycles of 95°C 10 sec , 61°C 30 sec and 72°C 30 sec . Primers were: Pcdhb2: agcccacctggtagatgttg and attggggatgattggtttca; Pcdhb3: cctggaaatacaccgcagaa and cctagacatggacccagcaa; Pcdhb4: cagtcagtcccaacctcca and tgaactgtggtcatcccagac; Pcdhb5: cagaggggaaatcaggaaca and gggcttaaactggcaatgaa; Pcdhb7: accccacacaggaagttgag and ctttatccccacgaaaagca; Pcdhb8: gccttggcttctgtgtcttc and caccactgacatccaccaag; Pcdhb9: atgcctggtgaacactttcc and gcagtggggactttccataa; Pcdhb10: gctgaccctcacctctcttg and accaccacgagtaccaaagc; Pcdhb13: ggcttctctcagccctacc and cagcaccacagacaagagga; Pcdhb14: cattgcacataggcaccatc and tgatggagatgagcgagttg; Pcdhb16: tggcttctctcagccctacc and aacagcagcacagacaccag; Pcdhb17: gcaagtcctggctttctttg and ggatatctctgccaggtcca; Pcdhb19: gacaaggcaagtcctgcttc and ccccaggtcctttaccaaat; Pcdhb22: tatcatcgctcaccaatcca and cagagctccatctgtcacca , beta-2-microglobulin : , atgggaagccgaacatactg and cagtctcagtgggggtgaat , and Nipbl ( exons 6–7 ) : agtccatatgccccacagag and accggcaacaataggacttg . PCR product sizes were between 107 and 182 bp .	Cornelia de Lange Syndrome ( CdLS ) is a genetic disease marked by growth retardation , cognitive and neurological problems , and structural defects in many organ systems . The majority of CdLS cases are due to mutation of one copy of the Nipped B-like ( NIPBL ) gene , the product of which regulates a complex of chromosomal proteins called cohesin . How reduction of NIPBL function gives rise to pervasive developmental defects in CdLS is not understood , so a model of CdLS was developed by generating mice that carry one null allele of Nipbl . Developmental defects in these mice show remarkable similarity to those observed in individuals with CdLS , including small stature , craniofacial abnormalities , reduced body fat , behavioral disturbances , and high perinatal mortality . Molecular analysis of tissues and cells from Nipbl mutant mice provide the first evidence that the major role of Nipbl in the etiology of CdLS is to exert modest , but significant , effects on the expression of diverse sets of genes , some of which are located in characteristic arrangements along the DNA . Among affected genes is a set involved in the development of adipocytes , the cells that make and accumulate body fat , potentially explaining reductions in body fat accumulation commonly observed in individuals with CdLS .	[ "Abstract", "Introduction", "Results", "Discussion", "Materials", "and", "Methods" ]	[ "genetics", "and", "genomics/disease", "models", "developmental", "biology/developmental", "molecular", "mechanisms", "molecular", "biology/chromatin", "structure", "genetics", "and", "genomics/gene", "expression" ]	2009	Multiple Organ System Defects and Transcriptional Dysregulation in the Nipbl+/− Mouse, a Model of Cornelia de Lange Syndrome
The complement system is key to innate immunity and its activation is necessary for the clearance of bacteria and apoptotic cells . However , insufficient or excessive complement activation will lead to immune-related diseases . It is so far unknown how the complement activity is up- or down- regulated and what the associated pathophysiological mechanisms are . To quantitatively understand the modulatory mechanisms of the complement system , we built a computational model involving the enhancement and suppression mechanisms that regulate complement activity . Our model consists of a large system of Ordinary Differential Equations ( ODEs ) accompanied by a dynamic Bayesian network as a probabilistic approximation of the ODE dynamics . Applying Bayesian inference techniques , this approximation was used to perform parameter estimation and sensitivity analysis . Our combined computational and experimental study showed that the antimicrobial response is sensitive to changes in pH and calcium levels , which determines the strength of the crosstalk between CRP and L-ficolin . Our study also revealed differential regulatory effects of C4BP . While C4BP delays but does not decrease the classical complement activation , it attenuates but does not significantly delay the lectin pathway activation . We also found that the major inhibitory role of C4BP is to facilitate the decay of C3 convertase . In summary , the present work elucidates the regulatory mechanisms of the complement system and demonstrates how the bio-pathway machinery maintains the balance between activation and inhibition . The insights we have gained could contribute to the development of therapies targeting the complement system . The complement system is pivotal to defending against invading microorganisms . The complement proteins recognize conserved pathogen-associated molecular patterns ( PAMPs ) on the surface of the invading pathogens [1] to initiate the innate immunity response . The complement activity also enhances adaptive immunity [2] , [3] and participates in the clearance of apoptotic cells [4] as well as damaged and altered self tissue . The complement proteins in the blood normally circulate as inactive zymogens . Upon stimulation , proteases in the system cleave the zymogens to release active fragments and initiate an amplifying cascade of further cleavages . There are three major complement activation routes: the classical , the lectin and the alternative pathways [5] . Regardless of how these pathways are initiated , the complement activity leads to proteolytic activation and deposition of the major complement proteins C4 and C3 , which induces phagocytosis , and the subsequent assembly of the membrane attack complex which lyses the invading microbes . However , complement is a double-edged sword; adequate complement activation is necessary for killing the bacteria and removing the apoptotic cells , while excessive complement activation can harm the host by generating inflammation and exacerbating tissue injury . Dysregulation of the balance between complement activation and inhibition can lead to rheumatoid arthritis [6] , systemic lupus erythematosus [7] , Alzheimer's disease [8] and age-related macular degeneration [9] . Since the final outcome of complement related diseases may be attributable to the imbalance between activation and inhibition [10] , manipulation of this balance using drugs represents an interesting therapeutic opportunity awaiting further investigation . In light of this potential , complement inhibitors such as factor H and C4b-binding protein ( C4BP ) are critical since they play important roles in tightly controlling the proteolytic cascade of complement and avoiding excessive activation . Therefore , a systems-level understanding of activation and inhibition , as well as the roles of inhibitors , will contribute towards the development of complement-based immunomodulation therapies . Complement is usually initiated by the interaction of several pattern-recognition receptors with the surface of pathogens . C-reactive protein ( CRP ) [11] and ficolins are two initiators of the classical and lectin pathways , which boost immune responses by recognizing phosphorylcholine ( PC ) or N-acetylglucosamine ( GlcNAc ) , respectively , displayed on the surface of invading bacteria [12] , [13] , [14] . Recently , it was discovered that under local infection-inflammation conditions as reflected by pH and calcium levels , the conformations of CRP and L-ficolin change which leads to a strong interaction between them [15] . This interaction triggers crosstalk between classical and lectin pathways and induces new amplification mechanisms , which in turn reinforces the overall antibacterial activity and bacterial clearance . On the other hand , C4BP , a major complement inhibitor is synthesized and secreted by the liver . The estimated plasma concentration of C4BP is 260 nM under normal physiological condition [16] but its plasma level can be elevated up to four-fold during inflammation [17] , [18] . Through its α-chain [19] , [20] , C4BP modulates complement pathways by controlling C4b-mediated reactions in multiple ways [21] , [22] , [23] . Further , C4BP has been proposed as a therapeutic agent for complement-related autoimmune diseases on the premise that mice models supplemented with human C4BP showed attenuation in the progression of arthritis [24] . Therefore , it is important to understand the systemic effect and the underlying inhibitory mechanism of C4BP . With this background , we constructed a detailed computational model of the complement network consisting of a system of ordinary Differential equations ( ODEs ) . The large model size and the many unknown kinetic rate parameters lead to significant computational challenges . Using the technique developed in [25] , we approximated the ODE dynamics as a dynamic Bayesian network [26] and used it to estimate the model parameters . After constructing the model , we investigated the enhancement mechanism induced by local inflammation and its interplay with the inhibition mechanism induced by C4BP . Our studies confirmed and further elucidated the previous experimental findings [15] . Specifically , using our model we established a detailed relationship between the antimicrobial response and the strength of the crosstalk between CRP and L-ficolin as determined by various combinations of the pH and calcium levels . We also found that C4BP prevents complement over-activation and restores homeostasis , but it achieves this in two distinct ways depending on whether the complement activity was initiated by PC or GlcNAc . Finally , the computational model suggested that the major inhibitory effect of C4BP is to potentiate the natural decay of C3 convertase ( C4bC2a ) . These findings regarding the role of C4BP were experimentally validated . An earlier mathematical study [27] of the complement system focused on the classical pathway . This study assumed the dynamics to be linear , which is a severe restriction . A later study by Korotaevskiy et al [28] more realistically assumed the dynamics to be non-linear . It also included the alternative pathway . The main focus was to derive quantitative conclusions regarding the lag time of the immune response as the initial concentrations of the constituent proteins were varied . Relative to [28] , our model additionally includes the lectin pathway and the recently identified amplification pathways induced by the crosstalk between CRP and L-ficolin [15] . On the other hand , given our focus on the up- and down- regulation mechanisms of the complement , we do not model the alternative pathway in detail since its role is to maintain a basal level of complement activation . Instead , this basal activity and the effects of other mechanisms such as C2 bypass [29] are implicitly captured by the kinetic parameters in our model . Given our focus on the amplification and down-regulation mechanisms of complement , we included in our model only the key proteins in the classical and lectin pathways . The basal activity maintained by the alternative pathway and other mechanisms are implicitly captured by the kinetic parameters in our model . A schematic representation of the model structure is shown in Figure 1A . The cascade of events captured by the model can be described as follows . The classical pathway is initiated by the binding of antibodies or CRP to antigens or PAMPs . In our model , in order to decouple the involvement of adaptive immune response , the classical pathway is triggered by the binding of CRP to PC , which is a ligand often displayed on the surface of the invading bacteria [30] , [31] . Deposited CRP then binds to C1-complex ( formed by C1q , two molecules of C1r , and two molecules of C1s ) that is further activated . The activated C1-complex recruits C4 leading to the cleavage of C4 to its fragments , C4b and C4a . After binding of C2 to C4b , the same protease complexes are responsible for generating fragments , C2a and C2b , by cleaving C2 . The C2a and C4b then form the C4bC2a complex , which is an active C3 convertase , cleaving C3 to C3a and C3b . The formation of C3b exposes a previously hidden thioester group that covalently binds to patches of hydroxyl and amino groups on the bacterial surface [32] . The surface-deposited C3b plays a central role in all subsequent steps of the complement cascade: ( 1 ) it acts as an opsonin that enhances the binding and leads to the elimination of bacteria by the phagocytes , ( 2 ) it induces the formation of membrane attack complex leading to the lysis of bacteria . Since the concentration of the deposited C3 reflects the antibacterial activity of complement , we terminated our model at this step to simplify the network . On the other hand , the lectin pathway is initiated by the binding of mannose-binding lectin ( MBL ) or ficolins to PAMPs on the pathogen surface . In our model , we focused on the lectin pathway initiated by L-ficolin as it can interact with CRP and induce crosstalk between classical and lectin pathways . L-ficolin recognizes various PAMPs on the bacterial surface via the acetyl group on the GlcNAc moiety [33] , [34] . Therefore , in our model the lectin pathway was triggered by binding of L-ficolin and GlcNAc onto the bacterial surface . Subsequently , a protease zymogen called MASP-2 is recruited and activated . Activated MASP-2 cleaves C4 and C2 to form C4bC2a which is C3 convertase . At this point , the classical pathway and lectin pathway merge at the cleavage step of the central complement protein , C3 , and hence constitutes the endpoint of our model . As discovered in [15] , infection-induced local inflammation conditions ( slight acidosis and hypocalcaemia ) provoke a strong crosstalk between CRP and L-ficolin [15] . This elicits two new complement-amplification pathways , which reinforce the classical and lectin pathways . Since we aimed to study the complement activation and modulation under pathophysiological conditions , we included these two amplification pathways ( Figure 1A , purple ) in our model . Infection by bacteria containing PC will induce the CRP:L-ficolin mediated amplification pathway: PC→CRP:L-ficolin→MASP2→C4→C2→C3 . On the other hand , infection by bacteria containing GlcNAc will induce the CRP:L-ficolin mediated amplification pathway: GlcNAc→CRP:L-ficolin→C1→C4→C2→C3 . The complement allows a rapid attack to intruding bacteria while at the same time protecting the host cells from over-activation . C4BP , a major inhibitor of complement activation , was reported to either accelerate the decay of the convertases or aid proteolytic inactivation of key players in the pathway into inactive forms such as factor H [32] but the systemic effect of C4BP has remained unclear . Hence , in our model , we included this major multifunctional inhibitor . Upstream of the complement cascade , C4BP competes with C1 for the immobilized CRP [23] . Downstream to this , C4BP binds to C4b and serves as a cofactor to the plasma serine protease factor I in the cleavage of C4b both in the fluid phase and when C4b is deposited on bacterial surfaces [21] . In addition , C4BP is able to prevent the assembly of the C3 convertase and accelerate the natural decay of the complex [35] . All of the above effects of C4BP are considered in our model and the relevant components are depicted as red bars in Figure 1A . The reaction network diagram of the model is shown in Figure 1B . Processes such as protein association , degradation and translocation are modeled with mass action kinetics and processes such as cleavage , activation and inhibition with Michaelis-Menten kinetics . The resulting ODE model consists of 42 species , 45 reactions and 85 kinetic parameters with 71 unknown . The details can be found in the supporting information ( Text S1 ) . Due to the large model size and many unknown kinetic parameters , tasks such as parameter estimation and sensitivity analyses became very challenging . Hence , we applied the probabilistic approximation technique developed by Liu et al [25] to derive a simpler model based on the standard probabilistic graphical formalism called Dynamic Bayesian Networks ( DBNs ) [26] . Briefly , this approximation scheme consists of the following steps: ( i ) Discretize the value space of each variable and parameter into a finite set of intervals . ( ii ) Discretize the time domain into a finite number of discrete time points . ( iii ) Sample the initial states of the system according to an assumed uniform distribution over certain intervals of values of the variables and parameters . ( iv ) Generate a trajectory for each sampled initial state and view the resulting set of trajectories as an approximation of the dynamics defined by the ODEs system . ( v ) Store the generated set of trajectories compactly as a dynamic Bayesian network and use Bayesian inference techniques to perform analysis . A more detailed description of this construction can be found in the Methods section while we explain in the Discussion section how we fixed the number of trajectories to be generated and the maximum time point upto which the trajectories are to be constructed . In the ODE model the PC-initiated and GlcNAc-initiated complement cascades are merged for convenience . By suppressing these two cascades to one at a time ( by setting the corresponding expressions in the reaction equations to zero ) , we constructed two dynamic Bayesian networks; one for the PC-initiated complement cascade and the other for GlcNAc-initiated complement cascade . The range of each variable and parameter was discretized into 6 non-equal size intervals and 5 equal size intervals , respectively . The time points of interest were set to {0 , 100 , 200 , … , 12600} ( seconds ) . Each of the resulting DBN approximations encoded trajectories generated by sampling the initial values of the variables and the parameters from the prior , which was assumed to be uniform distributions over certain intervals . The quality of the approximations relative to the original ODEs dynamics was sufficiently high and the details can be found in the supporting information ( Figure S1 ) . The values of initial concentrations and 14 kinetic parameters were obtained from literature data ( Table S1 and Table S2 ) . To estimate the remaining 71 kinetic parameters , we generated test data by incubating human blood under normal and infection-inflammation conditions with beads coated with PC or GlcNAc followed by immunodetection of the deposited CRP , C4 , C3 and C4BP in time series . For PC-beads , the concentration levels of deposited CRP , C4 , C3 and C4BP were measured at 8 time points from 0 to 3 . 5 h ( Figure 2A , B , red dots ) . For GlcNAc-beads , the concentration levels of deposited MASP-2 , C4 , C3 and C4BP were also measured at 8 time points from 0 to 3 . 5 h ( Figure 2C , D , red dots ) . To estimate unknown kinetic parameters , a two-stage DBN based method [25] was deployed . In the first stage , probabilistic inference applied to the discretized DBN approximation was used to find the combination of intervals of the unknown parameters that have the maximal likelihood , given the evidence consisting of the test data . As mentioned above , each unknown parameter's value space was divided into 5 equal intervals and the inference method called factored frontier algorithm [35] was used to infer the marginal distributions of the species at different time points in the DBN . We then computed the mean of each marginal distribution and compared it with the time course experimental data . To train the model by iteratively improving fitness to data , we modified the tool libSRES [36] and used its stochastic ranking evolutionary strategy ( SRES ) , to search in the discretized parameter space consisting of 571 combinations of interval values of the unknown parameters . The result of this first stage was a maximum likelihood estimate of a combination of intervals of parameter values . In the second stage we then searched within this combination of intervals having maximal likelihood . Consequently , the size of the search space for the second stage was just 1/571 of the original search space . We used the SRES search method and the parameter values thus estimated are shown in Table S2 . In principle , given the noisy and limited experimental data and the high dimensionality of the system , one could stop with the first stage [37] and try to work an interval of values for each parameter rather than a point value . However , in our setting we wanted to use the ODE model too for conducting in silico experiments such as varying initial concentrations including the down and over expression of C4BP . This would have been difficult to achieve by working solely with our current DBN approximation . We address this point again in the Discussion section . Figure 2A–2D shows the comparison of the experimental time course training data ( red dots ) with the model simulation profiles generated using the estimated parameters ( blue lines ) . The model predictions fit the training data well for most of the cases . In some cases , the simulations were only able to reproduce the trends of the data . This may be due to the simplifications assumed by our model and further refinement is probably necessary . We next validated the model using previously published experimental observations [15] . In particular , normalized concentration level of deposited C3 was used to predict the antibacterial activity since C3 deposition initiated the opsonization process and the lysis of bacteria . We first simulated the concentration level of deposited C3 at 1 h under different conditions . We next normalized the results so that the maximum value among them equals to 95% which is the maximum bacterial killing rate reported in the experimental observations [15] . The normalized values were then treated as predicted bacterial killing rates . The simulation results are shown in Figure 2E and 2F as black bars . Consistent with the experimental data ( Figure 2E , grey bars ) , our simulation showed that under the infection-inflammation conditions , the P . aeruginosa , a clinically challenging pathogen , can be efficiently killed ( 95% bacterial killing rate ) by complement whereas under the normal condition , only 28% of the bacteria succumbed ( Figure 2E , black bars ) . Consistent with experimental data , our simulation results show that in the patient serum , depletion of CRP or ficolin induced a significant drop in the killing rate from 95% to 33% or 25% respectively , indicating that the synergistic action of CRP and L-ficolin accounted for around 40% of the enhanced killing effect . However , in the normal serum , depletion of CRP or ficolin only resulted in a slight drop in the killing rate from 28% to 18% or 10% respectively . Furthermore , simulating a high CRP level ( such as in the case of cardiovascular disease ) under the normal healthy condition did not further increase the bacterial killing rate . As shown in Figure 2F , the simulation results matched the experimental data . Thus , our model was able to reproduce the published experimental observations shown in both Figure 2E and 2F with less than 10% error . This not only validated our model thus promoting its use for generating predictions , but also yielded positive evidence in support of the hypothesized amplification pathways induced by infection-inflammation condition . It also suggested that the antibacterial activity can be simulated efficiently by the level of deposited C3 and this was used to generate model predictions described in later sections . We performed local and global sensitivity analysis of the model to identify species and reactions that control complement activation during infection , and to evaluate the relative importance of initial concentrations and kinetic parameters for the model output . To identify critical species , we first calculated the scaled absolute local sensitivity coefficients [38] for initial concentrations of major species using the COPASI tool [39] . The model outputs were defined as the peak amplitude ( maximum activation ) and integrated response ( area under the activation curve that reflects the overall antibacterial activity ) of C3 deposition . The results are shown in Figure 3A . Both the peak amplitude and integrative response were strongly influenced by initial concentrations of C2 and C3 , and were mildly influenced by initial concentrations of C4BP , C1 and C4 . In contrast , the low sensitivities of CRP , MASP-2 and L-ficolin indicate that over-expression of these proteins is unlikely to increase the antibacterial activity . Interestingly , it was observed that the integrative response was more sensitive than the peak amplitude to the changes in the initial concentration of PC . Since the concentration of PC is correlated to the amount of invading bacteria , this result implies that the maximum complement response level may not increase as the amount of bacteria increases but the overall response ( i . e . the area under the curve obtained by integrating the response level over time ) will be enhanced to combat the increased number of bacteria . In order to identify critical reactions , we next computed global sensitivities for kinetic parameters . To reduce complexity , we used the DBN approximations . Multi-parametric sensitivity analysis ( MPSA ) [40] was performed on the DBN for PC-initiated complement cascade ( the details are presented in the Materials and Methods section ) . The results are shown in Figure 3B . Strong controls over the whole system are distributed among the parameters associated with the immobilisation of C3b with the surface , interaction between CRP and L-ficolin , cleavage of C2 and C4 , and the decay of C3 convertase ( see Figure 1B , reactions labeled in red ) . The sensitivity of reactions associated with C3 , C2 and C4 is consistent with the local sensitivity analysis , which highlighted the significant role of major complement components . The high sensitivity of interaction of CRP and L-ficolin confirms that the overall antibacterial response depends on the strength of the crosstalk between the classical and lectin pathways . In addition , since the decay of C3 convertase is one of the regulatory targets of C4BP , the sensitivity of the system to a change in the rate of decay of C3 convertase suggested that the regulatory mechanism by C4BP plays an important role in complement . Since the critical reactions identified are common in PC- and GlcNAc-initiated complement cascades , MPSA results using the other DBN will produce similar results and hence this analysis was not performed . We next focused our investigation on the enhancement mechanism by the crosstalk and the regulatory mechanism by C4BP . Under infection-inflammation conditions where PC-CRP:L-ficolin or GlcNAc-L-ficolin:CRP complex is formed , the amplification pathways are triggered . Model simulation showed that if C1 and L-ficolin or CRP and MASP-2 competed against each other , the antibacterial activity of the classical pathway or lectin pathway might be deprived of the amplification pathways ( see Figure S2 ) . Therefore , in order to achieve a stable enhancement , C1 and L-ficolin ( or CRP and MASP-2 ) must simultaneously bind to CRP ( or L-ficolin ) . Further , the abilities of CRP and L-ficolin to trigger subsequent complement cascade were not affected by the formation of this complex . This is consistent with the previous experimental observation that two amplification pathways co-exist with the classical and lectin pathways [15] . According to [15] , slight acidosis and mild hypocalcaemia ( pH 6 . 5 , 2 mM calcium ) prevailing at the vicinity of the infection-inflammation triggers a 100-fold stronger interaction between CRP and L-ficolin compared to the normal condition ( pH 7 . 4 , 2 . 5 mM calcium ) . This can be explained by the fact that the pH value and calcium level influence the conformations of CRP and L-ficolin which in turn govern their binding affinities . Therefore , the overall antibacterial response which is influenced by the binding affinity of CRP and L-ficolin will be sensitive to the pH value and calcium level . To confirm this and further investigate the effects of pH and calcium on the antibacterial response , we simulated the complement system under different pH and calcium conditions . Based on the previous biochemical analysis [15] , we first estimated functions using polynomial regression to predict the binding affinity of CRP and L-ficolin for different pH values and calcium levels ( Figure 4A , B , right panel ) . In the right panels of Figure 4A and 4B , the reported binding affinities [15] were normalized and are shown as dots . By curve fitting the dots , we estimated polynomial functions that can be used to predict the binding affinity . The curves of these functions are shown in red . We then simulated the C3 deposition dynamics using the predicted binding affinities at pH ranging from 5 . 5 to 7 . 4 in the presence of 2 mM and 2 . 5 mM calcium . The simulation time was chosen to be 3 . 5 h which is the time frame of the response peaks . The results are shown in Figure 4A and 4B . Under both 2 mM and 2 . 5 mM calcium conditions , decreasing pH induces not only the increase of the peak amplitude ( maximum activation ) but also hastens the peak time ( time of maximum activation ) . To further compare the effects of the two calcium levels , the dose-response curves were generated as shown in Figure 4C . The antibacterial response was predicted by simulating the system for 1 . 5 h . At 2 mM calcium ( blue curve ) , the antibacterial response was clearly greater than at 2 . 5 mM calcium ( pink curve ) indicating that slight hypocalcaemia enhanced the antibacterial activity in a stable manner . In addition , the pH-responses were reaching saturation levels when pH was near 5 . 5 ( Figure 4C ) , implying that the undesirable complement-enhancement by extreme low pH condition can be avoided . This also suggests that the saturation of the pH-response was influenced by the calcium level in the milieu . We next investigated the complement regulation by the major inhibitor , C4BP , under infection-inflammation conditions . We varied the initial concentration of C4BP and simulated the PC- and GlcNAc- initiated complement under infection-inflammation conditions . The simulation time was chosen to be 5 h which is slightly beyond the largest time point of our training experimental data . The predicted effects of the initial concentration of C4BP on the antibacterial response in terms of C3 deposition are shown in Figure 5A and 5B . For PC-initiated complement activation , when the starting amount of C4BP was perturbed around the normal level of 260 nM [16] , increasing C4BP level only delayed the peak time but did not decrease the peak amplitude significantly . In contrast , reducing the initial C4BP level clearly hastened the complement activation and maximized the activity . Interestingly , the GlcNAc-initiated complement activation ( Figure 5B ) behaved differently from the PC-mediated complement activation ( Figure 5A ) . Around the normal level of 260 nM , perturbing the initial C4BP changed the maximum activity but did not affect the peak time , suggesting that C4BP plays distinct roles in regulating the classical and lectin pathways . To experimentally verify the model predictions , we perturbed the initial amount of C4BP in the patient sera by ( i ) spiking with purified C4BP ( high C4BP ) and ( ii ) reducing it by immunoprecipitation ( low C4BP ) . The resulting C4BP levels in the normal and patient sera are shown in Figure S3 . The sera were then incubated with PC- or GlcNAc-beads to initiate complement . Unaltered serum served as the normal control . The time profiles of the deposited C4BP level was measured over 4 h using Western blot ( Figure 5C ) . Comparing the kinetic profiles in the C4BP deposition initiated by both PC and GlcNAc , we observed the following order of peak time: high C4BP>normal C4BP>low C4BP , indicating that the pre-existing initial level of C4BP was indeed the driving force controlling the deposition of complement components onto the simulated bacterial surface . We then measured the time profiles of deposited C3 . Figure 5D shows that with PC-beads , high C4BP sera induced an early peak and low C4BP delayed the peak of C3 deposition . The peak amplitude for all three conditions was at a similar level . These observations are consistent with the simulation results shown in Figure 5A . With GlcNAc-beads , reducing C4BP led to a slight increase in the peak height although the peak coincided with the normal condition . In contrast , spiking the sera ( high C4BP ) delayed and lowered the peak amplitude of C3 deposition . Thus the experimental results broadly agree with our model predictions presented in Figure 5B . We next investigated how C4BP mediates its inhibitory function . As shown in Figure 1A , the inhibitory effects of C4BP target different sites in complement: ( a ) binding to CRP and blocking C1 , ( b ) preventing the formation of C4bC2a by binding to C4b , ( c ) acting as a cofactor for factor I in the proteolytic inactivation of C4b , and ( d ) accelerating the natural decay of the C4bC2a complex , which prevents the formation of C4bC2a and disrupts already formed convertase . To identify the dominant mechanism , we employed in silico knockout of the reactions involved for each mechanism and performed simulations . Figure 6A–6D shows the model predictions . Among the four inhibitory mechanisms , only the knockout of reaction ( d ) significantly enhanced the complement activation suggesting that facilitating the natural decay of C4bC2a ( C3 convertase ) is the most important inhibitory function of C4BP . This is consistent with our previous observations derived from sensitivity analysis , which identified the decay of C3 convertase as a critical reaction . In addition , as the inhibitory effect of reaction ( d ) is stronger than others , knocking out reaction ( a ) and ( b ) can even reduce the complement activity , which is counter-intuitive and emphasizes the significance of the systems-level understanding . To confirm our hypothesis that the major inhibitory role of C4BP relies on accelerating the decay of C3 convertase , we measured the C4 cleavage at different time points . Figure 6E ( black triangles ) indicates the inactive C4b fragments presented from the time points of 20 , 30 and 90 min under high , normal , and low C4BP conditions , suggesting that C4BP aided cleavage and inactivation of C4b , and thereby caused the natural decay of the C4bC2a . Here , we developed an ODE-based dynamic model for the complement system accompanied by DBN-based approximations of the ODEs dynamics to understand how the complement activity is boosted under local inflammation conditions while a tight surveillance is established to attain homeostasis . Previously published models of complement system have focused on the classical and alternative pathways [27] , [28] . Our model includes the lectin pathway and more interestingly , the recently identified amplification pathways induced by local inflammation conditions [15] . It also encompasses the regulatory effects of C4BP in the presence of enhanced complement activity . The ODE model incorporated both the PC-initiated and GlcNAc-initiated complement together for convenience . By setting the corresponding expressions to zero one at a time , two DBN approximations were then derived; one for the PC-initiated complement cascade and the other for GlcNAc-initiated complement cascade . For constructing the DBN approximation from an ODE model , one needs to fix , the maximal time point upto which each trajectory is to be explored and , the number of trajectories to be generated . is set to be suitably beyond the largest time point for which experimental data is available . In the present study 3 . 5 h , is the largest time point of our training experimental data . Based on this we set to be 5 h . After constructing the model , we simulated the system upto 10 h and found no relevant dynamics after 3 . 5 h . As for the choice of , the number of trajectories , ideally one would like to specify the acceptable amount of error between the actual and the approximated dynamics and use to determine . This is however difficult to achieve due to the following: The dynamic Bayesian network we construct is a factored Markov chain . It approximates the idealized Markov chain induced by the ODEs dynamics . This idealized Markov chain is determined by the discretization of the value spaces of the variables and the parameters , the discretization of the time domain and the prior distribution of the initial states . As observed in , Liu et al [25] , given an error bound , a confidence level and the transition probabilities of the idealized Markov chain , we can estimate ( upper bound ) the required to fall within the given error bound with the required confidence level . However , our high dimensional ODE system does not admit closed form solutions and hence the transition probabilities of the idealized Markov chain will not be computable . Hence one must make a pragmatic choice of . Our approach has been to use a sampling method by which we can provide a minimum coverage of at least samples for each possible combination of interval values of the unknown parameters in the equation for each variable . This will ensure that the dynamics , governed by the set of equations ( one for each variable ) is being sufficiently sampled at least on a per equation basis . To achieve the required coverage , one will need samples , where is the maximal number of unknown parameters appearing in an equation and is the number of variables in the system . In our experience , seems to be an adequate choice . Based on this , we sampled initial points and generated the corresponding trajectories . The quality of the approximations relative to the original ODEs dynamics was sufficiently high and the details can be found in the supplementary information ( Figure S1 ) . How to determine with guaranteed error bounds is however a basic problem and we are continuing to study this issue . The study here has involved a tight integration of computational and experimental aspects . First , we used available biological information to form the biochemical network and the corresponding ODE system . We then experimentally generated test data to train the model during the process of parameter estimation . After constructing the model , one part of the computational exploration of the model was guided by the previous experimental study reported in [15] . Specifically , we computed through simulations the antibacterial response curves for varying combinations of pH values and calcium concentrations starting with the data provided in [15] . On the other hand , the second part of the study started from the computational side , namely , sensitivity analysis . Once C4BP was confirmed to be an important inhibitor through sensitivity analysis , we explored its regulatory mechanisms through simulations and generated the hypotheses concerning the differentiated influence of C4BP on PC-initiated and GlcNAc-initiated complement activity as well as the decay of the C3 convertase being the main inhibitory activity of C4BP . These hypotheses were then experimentally validated . At present , we have used the DBN approximation to mainly aid the tasks of parameter estimation and sensitivity analysis . The key idea is to use the DBN approximation and probabilistic inference to first reduce the search space and then apply conventional search techniques to this reduced search space in the second stage . For a -dimensional search space with discretized intervals for each dimension , the first stage can reduce the search space by a factor of . For analyses involving multiple initial conditions ( such as the in silico experiments involving C4BP ) , we found it more convenient to use the ODE model . This is due to the fact that the size of the DBN approximation increases significantly if it must encompass multiple initial conditions . Alternatively , one must construct a separate DBN for each choice of initial conditions . Related probabilistic formalisms such as Multi Terminal Binary Decision Diagrams ( MTBDDs ) and Probabilistic Decision Graphs ( PDGs ) are also available for analysis . It is not clear at present how they can be derived directly from the ODE model . One could however try to convert our DBNs to MTBDDs for purposes of model checking [41] or develop statistical model checking methods [42] . As compact representations of the probability distributions , PDGs are , in spirit , similar to Bayesian networks [43] and can be computationally as efficient as Bayesian networks [44] . Further , probabilistic inference can be carried out with a time complexity linear in the size of the PDGs [44] . Thus , it will be an interesting future direction for us to explore the performance of PDGs in our setting . Finally , we are aware that model construction is rarely complete . In the present setting , we included as much of the relevant and available biological information as possible in our model . Once the model was calibrated using the test data and was validated using reported bacterial killing rates , we were reasonably confident that it could be used as a platform for studying the up- and down- regulation mechanisms of the complement under local inflammation conditions . In exploring the model , we were guided by both the previous experimental study [15] and standard techniques such as sensitivity analysis . It is clear that the model will have to be refined and modified as new experimental findings become available . Indeed , we consider the systematic incremental updating of a computational model as new data become available , to be an important task [45] . Turning next to the biological insights gained from this study , we have shown that increase in PC concentration , representing the inoculum size of the invading bacteria , affected the overall classical pathway response time more than the peak amplitude . Our model analysis confirmed that the enhancement of complement activity under infection-inflammation condition was attributable to the synergistic action of CRP and L-ficolin and supported the existence of the amplification pathways . We showed that to achieve a steady enhancement , C1 and L-ficolin ( or CRP and MASP-2 ) should not compete with each other and the activities of CRP and L-ficolin should remain after forming the complex CRP:L-ficolin . Our computationally derived antibacterial response curves corresponding to varying pH values and calcium levels showed that the overall complement response was sensitive to pH and calcium levels . Through model analysis we found that with under PC-activation , perturbation of the initial C4BP level only affected the peak time but not the amplitude of the response . In contrast , in the case of GlcNAc-activation , perturbation of the initial level of C4BP only affected the peak amplitude and not the peak time . These results imply that C4BP regulates the lectin pathway more stringently than the classical pathway , which is consistent with previous experimental findings [46] . Further , for PC-initiated complement cascade , the over-expression of C4BP only delays but does not “turn off” the antibacterial response . In contrast , increased C4BP can efficiently inhibit GlcNAc-initiated complement activation . This may explain previous observations that bacteria such as Yersinia enterocolitica , Streptococcus pyogenes , Neisseria gonorrhoeae , Escherichia coli K1 , Moraxella catarrhalis , Candida albicans , Bordetella pertussis [47] , [48] , [49] , [50] , [51] , [52] , [53] can exploit C4BP to evade complement . Through in silico knockouts , we found that , of the four documented inhibitory roles , C4BP mainly aided the natural decay of C3 convertase . As the enhancement mechanism by the crosstalk between CRP and L-ficolin occurs upstream of the cascade , we envisage C4BP acts downstream to ‘quality control’ and modulate C3 convertase activity . Thus our results suggest that efficient regulation of complement can be achieved by targeting the C3 convertase , where the complement pathways merge . In summary , by integrating our computational model and experimental observations we have obtained novel insights into how the complement activation is enhanced during infection and how excessive complement activity may be avoided . This introduces a new level of understanding of the host defense against bacterial infection . It also provides a platform for the potential development of complement-based immunomodulation therapies by exploiting the sensitivities of the perturbations of the pH , calcium and C4BP levels .	The complement system , which is the frontline immune defense , constitutes proteins that flow freely in the blood . It quickly detects invading microbes and alerts the host by sending signals into immune responsive cells to eliminate the hostile substances . Inadequate or excessive complement activities harm the host and may lead to immune-related diseases . Thus , it is crucial to understand how the host boosts the complement activity to protect itself and simultaneously establishes tight surveillance to attain homeostasis . Towards this goal , we developed a detailed computational model of the human complement system . To overcome the challenges resulting from the large model size , we applied probabilistic approximation and inference techniques to train the model on experimental data and explored the key network features of the model . Our model-based study highlights the importance of infection-mediated microenvironmental perturbations , which alter the pH and calcium levels . It also reveals that the inhibitor , C4BP induces differential inhibition on the classical and lectin complement pathways and acts mainly by facilitating the decay of the C3 convertase . These predictions were validated empirically . Thus , our results help to elucidate the regulatory mechanisms of the complement system and potentially contribute to the development of complement-based immunomodulation therapies .	[ "Abstract", "Introduction", "Results", "Discussion" ]	[ "computational", "biology/systems", "biology", "immunology/innate", "immunity" ]	2011	A Computational and Experimental Study of the Regulatory Mechanisms of the Complement System
Pyrethroid insecticides are widely utilized in dengue control . However , the major vector , Aedes aegypti , is becoming increasingly resistant to these insecticides and this is impacting on the efficacy of control measures . The near complete transcriptome of two pyrethroid resistant populations from the Caribbean was examined to explore the molecular basis of this resistance . Two previously described target site mutations , 1016I and 1534C were detected in pyrethroid resistant populations from Grand Cayman and Cuba . In addition between two and five per cent of the Ae . aegypti transcriptome was differentially expressed in the resistant populations compared to a laboratory susceptible population . Approximately 20 per cent of the genes over-expressed in resistant mosquitoes were up-regulated in both Caribbean populations ( 107 genes ) . Genes with putative monooxygenase activity were significantly over represented in the up-regulated subset , including five CYP9 P450 genes . Quantitative PCR was used to confirm the higher transcript levels of multiple cytochrome P450 genes from the CYP9J family and an ATP binding cassette transporter . Over expression of two genes , CYP9J26 and ABCB4 , is due , at least in part , to gene amplification . These results , and those from other studies , strongly suggest that increases in the amount of the CYP9J cytochrome P450s are an important mechanism of pyrethroid resistance in Ae . aegypti . The genetic redundancy resulting from the expansion of this gene family makes it unlikely that a single gene or mutation responsible for pyrethroid resistance will be identified in this mosquito species . However , the results from this study do pave the way for the development of new pyrethroid synergists and improved resistance diagnostics . The role of copy number polymorphisms in detoxification and transporter genes in providing protection against insecticide exposure requires further investigation . Aedes mosquitoes have shown a remarkable ability to develop resistance to insecticides [1] . Today , resistance to DDT , organophosphates and pyrethroids is widespread in the major dengue vector , Aedes aegypti [2] and this resistance is negatively impacting on control efforts . For example , in the Caribbean , resistance to pyrethroids is reducing the efficacy of pyrethroid space spraying in La Martinique and organophosphate resistance in Cuba is reducing the duration of control obtained by larviciding [3] , [4] . Resistance to pyrethroids is of particular concern as this class of insecticides is increasingly replacing organophosphates in space spraying ( WHO , 2011 ) and curtains , impregnated with pyrethroids , have also shown initial promise in reducing dengue transmission [5] . Most studies on the molecular basis of pyrethroid resistance focus on target site mutations [6] , [7] , [8] . Amino acid substitutions in the voltage gated sodium channel cause a resistance phenotype to pyrethroid insecticides known as knockdown resistance or kdr . At least four amino acid substitutions in the sodium channel ( I1011M , V1016G , V1016I and F1534C ) have been linked to resistance in Ae . aegypti [9] , [10] , [11] , [12] . Two of these alleles , 1016I and 1534C , are widely distributed in the Caribbean [9] . The role of other resistance mechanisms is less clearly understood . Biochemical assays are frequently used to screen for metabolic resistance caused by elevated activities of cytochrome P450s , carboxylesterases and/or glutathione transferases . Although these assays lack sensitivity they have provided preliminary evidence that metabolic resistance is involved in conferring pyrethroid resistance in Caribbean populations of Ae . aegypti from Cuba , Grand Cayman , La Martinique and Trinidad . To determine the molecular basis of this metabolic resistance and , identify other pathways potentially involved in conferring the resistance phenotype , we utilized microarray and quantitative PCR to analyse the near complete transcriptome of pyrethroid resistant populations from Cuba and Grand Cayman . The results confirm that elevated cytochrome P450 activity is strongly associated with pyrethroid resistance in these populations . Comparative analysis of the data from this , and earlier studies on populations from Latin America and Southeast Asia , indicates that the CYP9J family of P450 enzymes is primarily responsible for metabolic resistance to pyrethroids in Ae . aegypti . Three strains of Ae . aegypti were used in this study . The NEW ORLEANS ( NO ) strain is a laboratory strain that is susceptible to all known insecticides and was originally colonized by the Center for Disease Control and Prevention ( CDC ) Atlanta , USA . The pyrethroid resistant CAYMAN strain was colonized from larvae collected in routine field surveillance sites in Grand Cayman in 2008 . This strain has very high levels of resistance to DDT ( >90% survival after 8 hours exposure to 4% DDT ) and pyrethroids ( resistance ratio of 109-fold to permethrin and 30-fold to deltamethrin compared with the susceptible New Orleans strain [9] ) . The CUBA-DELTA SAN 12 strain ( CUBA-DELTA ) was collected in 1997 in Santiago de Cuba . It was selected for 12 generations at the larval stage with deltamethrin at the Institute ‘Pedro Kouri’ in Havana , Cuba . CUBA-DELTA larvae were highly resistant to this insecticide ( >1000-fold ) and this resistance was also manifested at the adult stage [13] . Egg papers from the CAYMAN strain and the CUBA-DELTA strain were sent to the Liverpool School of Tropical Medicine , UK and the mosquitoes were reared under standard laboratory conditions ( 26°C , 80% RH ) and a 12∶12 hours light∶dark cycle . The prevalence of the 1016I and 1534C kdr mutations in the CAYMAN strain has been reported previously . For the CUBA-DELTA strain , 38 mosquitoes were genotyped for the 1534C mutation using the tetraplex assay described in [9] and for the 1016I mutation using the hot oligonucleotide ligation assay ( HOLA ) [11] . For each strain , total RNA was extracted from three pools of 30 , three day old , non blood-fed females using Pico Pure™ RNA Isolation Kit ( Applied biosystems , Foster city , CA , USA ) . The strains were reared in parallel to minimize variation resulting from breeding conditions . Each biological replicate consisted of mosquitoes from distinct generations to control for stochastic variations . The quality and concentration of RNA was assessed using a 2100 Bioanalyzer ( Agilent technologies , Santa Clara , CA , USA ) . Then , 100 ng of total RNA were used for RNA amplification and labeled with Cy-3 and Cy-5 fluorescent dyes using the Two Colors Low Input Quick Amp Labeling Kit ( Agilent technologies ) according to manufacturer's instructions . Labeled cRNAs were purified with the Qiagen RNeasy spin columns ( Qiagen , Hilden , Germany ) . Quantification and quality assessment of labeled cRNA were performed with the Nanodrop ND-1000 ( Thermo Scientific , DE , USA ) and the Agilent 2100 Bioanalyser ( Agilent Technologies ) . Purified labeled cRNAs were stored at −80°C until microarray hybridizations . Hybridizations were made to the ‘Liverpool Aedes aegypti Agilent 8×15K v1’ microarray ( A-MEXP-1966 ) designed by the Liverpool School of Tropical Medicine . Each array contains 60mer oligo-probes representing >14320 Aedes aegypti transcripts ( 93% of the putative gene count , 79% of putative transcripts –the lower coverage of transcripts is a consequence of the multiple putative transcripts for some genes ) . Labeled cRNA from CAYMAN and CUBA-DELTA were co-hybridized with age-matched NO samples , in direct pairwise comparisons . For two out of the three biological replicates , dye swaps were performed making a total of five hybridisations per comparison . Labeled targets were hybridized to the array for 17 h at 65°C and 10 rpm rotation and then washed according to Agilent protocol . Slides were scanned on Agilent G2565AA/G2565BA Microarray Scanner System using Agilent Feature extraction software ( Agilent technologies ) . Genespring GX 11 . 1 software ( Agilent technologies ) was used for normalization and statistical analysis . To account for multiple testing , p-values were adjusted adopting the approach of Benjamini and Honchberg [14] to control for the false positives . Transcripts showing an absolute fold change >2-fold in either direction and a t-test P-value lower than P<0 . 01 after multiple testing correction were considered as significant . Descriptions and GO-terms of transcript-IDs were extracted from VectorBase [15] using BIOMART [16] and completed with Blast2GO software ( BioBam Bioinformatics S . L . ( Valencia , Spain ) ) [17] . GO term Enrichment analysis was performed on the significantly up-regulated genes ( 72% of transcripts present on microarray have GO-terms ) using Blast2GO software with Fisher's exact test and false discovery rate ( FDR ) <0 . 05 . Selected microarray data were validated using quantitative reverse transcription PCR ( qRT-PCR ) . Primers were designed using the Oligo7 Primer Analysis Software ( Molecular Biology Insights , Cascade , CO , USA ) based on cDNA sequences retrieved from VectorBase . An aliquot of 4 µg total RNA from each of the three biological replicates , for each strain , served as a template for cDNA synthesis with Superscript III ( Invitrogen , Carlsbad , CA , USA ) using oligo-dT20 , according to the manufacturer's instructions . The resulting cDNAs were diluted 20 times in ultra-high quality water for qRT-PCR reactions using a MiniOpticon System ( Biorad , Hercules , CA , USA ) . PCR reactions of 25 µl contained Fast Start SYBR Green Master Mix ( Roche , Penzberg , Germany ) , 0 . 3 µM of each primer ( Table 1 ) and 5 µl of diluted cDNA . Melt curve analysis was performed to test the specificity of amplicons . A serial dilution of cDNA was used to generate standard curves for each gene in order to assess PCR efficiency and quantitative differences among samples . Primer sequences are provided in Table S1 . The fold-change of each target gene , normalized to the 60S ribosomal protein L8 ( AAEL000987 ) and 40S ribosomal protein S7 ( AAEL009496 ) , and relative to NO , was calculated according to the 2−ΔΔCT method incorporating PCR efficiency [18] . In most cases , two independent primer sets were used for each gene ( Table S1 ) . Genomic DNA ( gDNA ) from three batches of ten adult mosquitoes from each strain was extracted using DNAzol ( Invitrogen ) according to the manufacturer's instructions . DNA quality and quantity was assessed by Nanodrop ND-1000 spectrophotometry and by running an aliquot on a 1 . 5% agarose gel . Quantitative PCR reactions were performed as described above on the same genes chosen for transcript analysis . The PCR efficiency for target and control genes ( RPS7 and RPL8 ) was calculated from standard curves generated from a pool of gDNA for all three strains . 150–200 ng of gDNA was used as template and primer concentrations were between 100–300 nM for all genes . Primer sequences are provided in Table S1 . The relative copy number fold-change was calculated using the 2−ΔΔCt method [18] . Two target site mutations , both previously associated with pyrethroid resistance in Ae . aegypti , are present at high frequencies in the two resistant strains from the Caribbean . In the CAYMAN strain , kdr frequencies of 0 . 79 for the 1016I allele , and 0 . 68 for the 1534C allele have been reported previously [9] . In the current study , 38 individuals from the CUBA-DELTA strain were genotyped and the frequency of the resistant alleles were 0 . 51 ( 1016I ) and 0 . 88 ( 1534C ) . Differences in gene expression in whole adult female mosquitoes of the pyrethroid-resistant strains from CUBA-DELTA and CAYMAN and the NO susceptible strain were assessed using a 15K Ae . aegypti microarray platform . The data have been deposited in ArrayExpress ( accession number E-MTAB-868 ) . Using an arbitrary cut off of fold change >2-fold in either direction and a t-test P-value lower than P<0 . 01 after multiple testing correction , 981 transcripts ( 5 . 4% ) were differentially transcribed between CAYMAN and NO ( 410 up regulated and 566 down regulated ) and 414 genes ( 2 . 2% ) were differentially transcribed between the CUBA-DELTA and NO strains ( 213 up regulated and 201 down regulated ) ( Figure 1 ) . Of the 516 up-regulated genes , 107 ( 20 . 7% ) were over expressed in both resistant populations . In the down regulated subset , 99 of 668 ( 14 . 8% ) genes were under expressed in both CAYMAN and CUBA-DELTA populations relative to NO . Five genes showed opposing patterns of gene expression between the two comparisons ( Table S2 ) . The predicted functions of the genes differentially expressed in both populations were identified by BLAST2GO . More than 43% of the differentially expressed genes in the Cuba vs New Orleans comparison are annotated as ‘conserved hypothetical proteins’ in Vectorbase and 49% for Cayman vs New Orleans . These are listed in tables S2 to S4 but are not discussed further in this manuscript . For further analysis we focused primarily on the subset of genes that were differentially expressed in both populations ( Table 2 ) , although other genes of interest are also discussed . Enrichment analysis was used to identify particular GO terms that were over represented in the subset of transcripts up regulated in both resistant populations . Thirteen GO terms were significantly over represented in the up-regulated subset ( Figure 2 ) . However , after Benjamini and Hochberg multiple testing correction ( Pval<0 . 05 ) only the GO term designating moooxygenase activity was significantly differentially represented . This initial screening via GO terms was followed up with a manual examination of the putative functions of each of the 107 transcripts up-regulated and 99- down regulated in each population . This gene set , ranked by fold change in the CAYMAN population , is listed in Table S2 . Genes with putative detoxification functions are listed in Table 1 . Detoxification genes comprised 15 . 8% of the commonly up-regulated subset but were not represented at all in the down –regulated subset of the commonly expressed genes . Several additional detoxification genes were found up-regulated in either the Cayman or Cuba strains ( Table 1 ) . A total of 18 and 13 CYPs are over expressed >2-fold in the CUBA-DELTA strain and CAYMAN strain relative to the susceptible NO population respectively ( Table 1 ) . Seven of these CYPs were up-regulated in both strains: CYP6BB2 , CYP9J9 , CYP9J10 , CYP9J26 , CYP9J27 , CYP9J28 and CYP329B1 . Three P450s were down regulated in the CAYMAN strain and two in the CUBA-DELTA strain but none of these are common to both strains . Twelve of the 24 up-regulated CYPs ( and five of the seven up-regulated in both strains ) belong to the CYP9J family . Further genes with roles in oxidative metabolism of xenobiotics were amongst the subset of genes over expressed in both resistant populations . This included a dimethylanaline monooxygenase ( AAEL00834 ) , a member of the cytochrome b561 family ( AAEL012836 ) and subunit 4 of NADH dehydrogenase ( AAEL009076 ) . Other detoxification genes included the glutathione transferase , GSTe4 ( AAEL007962 ) , and two glucosyl glucornosyl transferases ( AAEL003099 and AAEL014246 ) ( Table 1 ) . Seven P450s ( plus an ABC gene described below ) were selected to validate the microarray results by qPCR . These included six out of the seven P450s over expressed in both strains and an additional P450 that is only over expressed in the CAYMAN strain , CYP9J19 . In general there is good agreement between the qPCR and microarray data ( Table S5 ) with the exception of the CYP6BB2 gene , ( AAEL014893 ) . The high level of over expression of this gene observed in the microarray could not be confirmed by qPCR using two alternative primer sets . However , the qPCR confirmation of over expression of the CYP9J genes adds further support for these enzymes playing a role in resistance to pyrethroids in these Caribbean populations . An ABC transporter gene , AAEL006717 , was expressed at approximately 5-fold and 2-fold higher levels in the pyrethroid resistant populations from Cayman and Cuba respectively , relative to the susceptible NO strain . This gene is potentially of interest because elevated ABC transporters have been linked to insecticide resistance in several species [20] , [21] , [22] , [23] although the physiological mechanism by which these transporter proteins act to reduce insecticide susceptibility is unknown . The over expression of AAEL006717 , which is an orthologue of the An . gambiae ABCB4 gene , [24] was confirmed by qPCR for the CAYMAN strain but not attempted in the Cuban population ( Figure 3 Table S5 ) . In the CAYMAN strain , 8 transcripts for odorant binding proteins ( OBPs ) were up-regulated . Furthermore , when the CAYMAN population was analysed alone , the GO term ‘odorant binding’ was the most differentially represented term in the up-regulated set of transcripts ( Figure 2 ) . OBPs facilitate the passage of semio-chemicals across the antennae , and other sensory appendages , to the olfactory neurones . To date , no specific role for OBPs in insecticide resistance has been demonstrated but this is not the first time that OBPs have been identified as being over expressed in insecticide resistant populations . A study of bendiocarb resistance in Anopheles identified an OBP gene that was overexpressed in Ghanaian resistant populations ( S Mitchell , unpublished data ) . Quantitative PCR was used to compare gene copy number between the two resistant and the susceptible Ae . aegypti strains . Using the same cut off of >2 fold change in expression , gene amplification was observed for two genes , CYP9J26 and the ABC transporter , ABCB4 ( AAEL006717 ) . The copy number of CYP9J26 , measured using two different primer pairs , was between 6 . 5 and 8 . 1-fold higher in the resistant CUBA and CAYMAN strains respectively compared with New Orleans ( Figure 3 ) . Similarly , the ABCB4 gene was amplified approximately 7-fold in the Cayman strain relative to the New Orleans ( Figure 3 ) . Pyrethroid resistance is widely distributed in Ae . aegypti throughout its range but relatively little is known about the mechanisms responsible for this resistance . Target site resistance is present in both the Cuban and Cayman populations although neither of the two mutations , 1016I or 1534C , were fixed in either population , despite several rounds of laboratory selection with deltamethrin in the Cuban strain . Interestingly the 1534C mutation has recently been shown to confer a selective advantage against type I pyrethroids but not affect the sensitivity to type II pyrethroids such as deltamethrin [25] . As far as we are aware , electrophysiological experiments have not been performed to examine the impact of the V1016I substitution . Thus , target site resistance may be partially responsible for the high levels of deltamethrin resistance in both these populations but it is likely that other mechanisms are involved . In this study , a microarray containing probes for the vast majority of annotated genes in the Ae . aegypti genome was used to compare gene expression in the two Caribbean populations with a standard lab susceptible strain . A potential limitation of this approach is the use of a single laboratory susceptible strain that originated from the United States . Ideally , a range of susceptible strains including wild populations from similar genetic regions would be included in the study . Unfortunately , such strains are becoming increasingly difficult to find . The differential gene expression observed may be partially attributed to the different genetic background of the strains although , encouragingly , earlier experiments have shown no significant difference in expression of detoxification genes between New Orleans and another well established laboratory susceptible strain , Rockefeller [19] . In light of this potential criticism , the analysis focused primarily on genes that were up-regulated in both resistant populations . Interestingly , a smaller number of genes were found differentially transcribed in the Cuban strain , which had been subject to extensive laboratory selection , than the Cayman strain , which was resistant upon colonization . By using GO term enrichment analysis , the functions or processes that were enriched in the subsets of genes up or down regulated in the pyrethroid resistant populations were identified . Only one GO term was significantly enriched in this analysis . Eleven transcripts with the GO term GO:0004497 , monooxygenase activity , were found amongst the subset over expressed in both insecticide resistant populations . This supports the well documented role of cytochrome P450s in conferring pyrethroid resistance [26] . Aedes aegypti has an extensive repertoire of between 160 and 180 CYP genes [19] . The uncertainty over the exact gene count is partly due to the fragmented nature of the Ae . aegypti genome assembly; several supercontigs containing clusters of P450s most likely represent alternative haplotypes . P450s contained within these putative duplicate clusters have been assigned independent accession numbers in VectorBase but named as allelic variants of the same P450 by the P450 nomenclature committee ( designated v1 or v2 ) . This issue is discussed further in the supplementary material of Strode et al ( 2008 ) [19] . For ease of discussion , in the current manuscript , official P450 nomenclature has been used to discuss the P450 family and the v1/v2 nomenclature omitted . However , the finding that two variants of the same gene frequently show similar fold changes in expression , adds confidence to the analysis as the probes were not designed to be able to distinguish allelic variants . Seven cytochrome P450 genes were up-regulated in both resistant populations . This included two genes in the CYP6 clade , CYP6BB2 and CYP329B1 , but neither was confirmed by qPCR . The five remaining P450 genes belonged to the CYP9J family . Four of these have been found to be up-regulated in resistant strains from other geographical localities . CYP9J9 and CYP9J10 were found over expressed in Thai and Latin American populations of Ae . aegypti . CYP9J28 is over expressed in pyrethroid resistant populations from Peru and Mexico [19] , [27] , and has also been shown to be over expressed in pyrethroid resistant Ae . aegypti from Vietnam ( Warr and Ranson , unpublished data ) . CYP9J27 is over expressed in Thailand [19] and is also one of the candidates emerging from the Vietnam study . Other CYP9J genes have also been implicated in resistance ( Figure 4 ) . In fact , in total , ten CYP9J genes have been found over expressed in at least two pyrethroid resistant populations . Four of these , CYP9J24 , 26 , 28 and 32 , have now been biochemically characterized [28] and have all been shown to metabolize pyrethroids ( a single CYP6 , CYP6CB1 was also expressed but had no activity against this insecticide class ) . Aside from the CYP9s , the only additional clade of Ae . aegypti P450s that is found repeatedly over expressed in resistant strains is the CYP6Z subfamily . Genes CYP6Z6 , Z8 and Z9 have been found over expressed in multiple populations from southeast Asia , Latin America and Caribbean [19] , [27] , [29] and CYP6Z8 and Z9 were over expressed in Cuba and Cayman populations respectively in the current study . The CYP9 family in Ae . aegypti is greatly expanded compared to other insect species with over three times as many members as found in An gambiae and nearly six times as many as in D . melanogaster [19] , [30] . The degree of genetic redundancy in the P450 family of Ae . aegypti makes it unlikely that a single gene responsible for pyrethroid resistance in all strains will be detected , particularly if resistance is emerging independently in different populations . However , the identification of a small subset of genes , consistently over expressed in resistant populations does suggest it should be possible to develop specific inhibitors of these metabolic pathways that could be used as insecticide synergists . Gene amplification was associated with the over expression of one of the P450 genes , CYP9J26 , in both strains , with approximately 7 –fold increase in copy number compared to the susceptible strain . Increased gene copy numbers have been associated with P450 mediated resistance in An . funestus , D . melanogaster and Myzus persicae [31] , [32] , [33] . The increased transcript levels of the other CYP9 genes were not associated with an increase in gene copy number although copy number polymorphisms appear to be common in the P450 family in Ae . aegypti ( Strode et al , 2008 ) . CYP9J26 , 27 and 28 are arranged sequentially in the Ae . aegypti genome within a large cluster of CYP9 genes on supercontig 1 . 1188 . It is not yet known if the CYP926 duplications are found in tandem . Multiple copies of the ABC transporter were also present in the Cayman resistant population . Gene amplification is being increasingly recognized as an important mechanism conferring metabolic resistance to insecticides with examples reported in all the major families of detoxification enzymes from several insect species [34] . Resistance to pyrethroid insecticides is now widely established in Ae aegypti populations throughout its distribution [2] . Elucidating the mechanisms responsible for this resistance will facilitate resistance monitoring and pave the way for the development of effective resistance reversal approaches . The microarray approach used in this study is not itself a field applicable screening approach . However , using this tool to analyse additional pyrethroid resistant populations will help define a subset of genes that are responsible for pyrethroid resistance . As discussed above , although it is unlikely that a single diagnostic mutation will be detected , the identification of a panel of candidate resistance associated genes is an important prerequisite for developing simple , molecular diagnostics that are urgently needed by dengue control programmes . Given the key role that pyrethroids play in controlling this disease vector , and the lack of affordable , acceptable alternative insecticides , it is imperative that efforts are made to monitor for resistance and reduce the impact that this resistance may impose on vector control interventions .	Dengue is the most rapidly spreading arboviral infection of humans and each year there are 50–100 million cases of dengue fever . There is no vaccine or drug to prevent dengue infection so control of the mosquitoes that transmit this virus is the only option to reduce transmission . Removing mosquito habitats close to human homes can be effective but in reality most dengue control programmes rely on a small number of chemical insecticides . Therefore , when the mosquito vectors develop resistance to the available insecticides , dengue control is jeopardized . In this study we examined the causes of resistance to the insecticide class most commonly used in mosquito control , the pyrethroids . We found that a group of genes , which have been implicated in detoxifying these insecticides in other populations of dengue vectors , were highly over expressed in both these Caribbean populations and we investigated the molecular basis of this increased expression . The next steps , which will be a considerable challenge , are to utilize this information to develop effective means of restoring insecticide susceptibility in dengue vectors .	[ "Abstract", "Introduction", "Materials", "and", "Methods", "Results", "Discussion" ]	[ "medicine", "infectious", "diseases", "global", "health", "genetics", "biology", "genetics", "and", "genomics", "toxicology" ]	2012	Gene Amplification, ABC Transporters and Cytochrome P450s: Unraveling the Molecular Basis of Pyrethroid Resistance in the Dengue Vector, Aedes aegypti
Unmodified or as a poly[lactide-co-glycolide] nanoparticle , tetraiodothyroacetic acid ( tetrac ) acts at the integrin αvβ3 receptor on human cancer cells to inhibit tumor cell proliferation and xenograft growth . To study in vitro the pharmacodynamics of tetrac formulations in the absence of and in conjunction with other chemotherapeutic agents , we developed a perfusion bellows cell culture system . Cells were grown on polymer flakes and exposed to various concentrations of tetrac , nano-tetrac , resveratrol , cetuximab , or a combination for up to 18 days . Cells were harvested and counted every one or two days . Both NONMEM VI and the exact Monte Carlo parametric expectation maximization algorithm in S-ADAPT were utilized for mathematical modeling . Unmodified tetrac inhibited the proliferation of cancer cells and did so with differing potency in different cell lines . The developed mechanism-based model included two effects of tetrac on different parts of the cell cycle which could be distinguished . For human breast cancer cells , modeling suggested a higher sensitivity ( lower IC50 ) to the effect on success rate of replication than the effect on rate of growth , whereas the capacity ( Imax ) was larger for the effect on growth rate . Nanoparticulate tetrac ( nano-tetrac ) , which does not enter into cells , had a higher potency and a larger anti-proliferative effect than unmodified tetrac . Fluorescence-activated cell sorting analysis of harvested cells revealed tetrac and nano-tetrac induced concentration-dependent apoptosis that was correlated with expression of pro-apoptotic proteins , such as p53 , p21 , PIG3 and BAD for nano-tetrac , while unmodified tetrac showed a different profile . Approximately additive anti-proliferative effects were found for the combinations of tetrac and resveratrol , tetrac and cetuximab ( Erbitux ) , and nano-tetrac and cetuximab . Our in vitro perfusion cancer cell system together with mathematical modeling successfully described the anti-proliferative effects over time of tetrac and nano-tetrac and may be useful for dose-finding and studying the pharmacodynamics of other chemotherapeutic agents or their combinations . Tetraiodothyroacetic acid ( tetrac ) is a deaminated thyroid hormone analogue that binds to the integrin αvβ3 receptor for thyroid hormone [1] , [2] . Tetrac inhibits binding of agonist L-thyroxine , T4 , and 3 , 5 , 3′-triiodo-L-thyronine , T3 , to the integrin on cultured cells [1] , blocking nongenomically-initiated effects of T4 and T3 on signal transduction pathways [2]–[4] . Tetrac also has actions at the receptor independent of T4 and T3 , including inhibition of cancer cell proliferation [2]–[4] and angiogenesis [5] , [6] . The integrin is largely expressed on tumor cells and dividing blood vessel cells [7] . Acting at the surface of cancer cells , tetrac alters expression of differentially-regulated cancer cell survival pathway-relevant genes . These include upregulation of expression of pro-apoptotic BcL-x short form [3] and other pro-apoptotic genes [8] , upregulation of anti-angiogenic thrombospondin 1 and downregulation of several families of anti-apoptotic genes [8] , [9] . Covalently bound to the exterior of a nanoparticle , tetrac does not gain access to the cell interior—where it may have thyromimetic activity [10]—and has biological activity at the integrin receptor similar to that of unmodified tetrac , but with desirable effects on cell survival pathway genes that differ from the parent thyroid hormone analogue [8] , [9] . To further characterize in vitro the anti-proliferative pharmacodynamics ( PD ) of tetrac and nanoparticulate tetrac ( nano-tetrac ) , with and without other chemotherapeutic agents , we developed a perfusion bellows cell culture system based on a perfusion ( ‘hollow fiber’ ) model . The hollow fiber model was modified by two co-authors ( AL , GLD ) from a previous system that explored antibiotic pharmacodynamics [11] . The hollow fiber model and perfusion bellows cell culture system allow simulation of concentration-time profiles ( pharmacokinetics ) expected in humans in an in vitro system and study of the effects over time ( PD ) of anti-infective and anti-cancer agents in vitro [12] , [13] . Such in vitro systems in combination with mathematical modeling can support translation from in vitro to animal models and human clinical trials . The developed pharmacodynamic model describes the full time course of drug effects at various concentrations simultaneously and may be used to predict the effects of other than the studied dosage regimens . We report here that tetrac and nano-tetrac inhibit cancer cell proliferation on a concentration-dependent basis that can be cell line-specific . Harvesting cancer cells from the perfusion bellows cell culture system permits fluorescence-activated cell sorting ( FACS ) analysis of cell cycle , and for apoptosis , quantitation of specific pro-apoptotic and anti-apoptotic gene expression by RT-PCR or microarray . Unmodified tetrac and nano-tetrac were tested in this model system for anti-proliferative efficacy alone or in combination with two other anticancer agents , the stilbene resveratrol [14] , and commercially-available anti-epidermal growth factor receptor ( EGFR ) monoclonal antibody ( cetuximab , Erbitux ) . Additive effects were obtained with combinations of tetrac or nano-tetrac and those other chemotherapeutic agents . We report studies in several human cancer cell lines to infer the applicability of the model and to confirm , not surprisingly , that there are dose-dependent differences in responses of specific cell lines to the chemotherapeutic agents tested . Overall our aim to develop a mechanism-based pharmacodynamic model that characterizes the action of tetrac on human cancer cells in a newly developed perfusion bellows cell culture system was well achieved as described in the present report . The pharmacodynamics of tetrac as an anti-proliferative agent against human cancer cell lines were examined in the perfusion bellows cell culture system depicted in Fig . 1 . Stability of tetrac in the culture system was determined by LC/MS/MS . Without cells , 75% of the original tetrac concentration was detected after 24 h incubation in medium with 10% FBS at both room temperature and 37°C . Tetrac decayed by 12% when incubated with cells at 37°C , indicating that tetrac is relatively stable in the perfusion bellows cell culture system . At first tetrac induced anti-proliferation of cancer cells was studied in the non-perfusion system . Human glioblastoma U87MG cells were treated with different tetrac concentrations ( 10−9–10−5 M ) for 7 d , with daily replenishment of tetrac . A model incorporating effects of tetrac on both growth rate and success of replication ( Fig . 2 ) adequately described the time course of cell counts as shown by comparison of the model fitted lines to the observed data ( Fig . 3A ) . Tetrac caused a concentration-dependent reduction in U87MG cell proliferation ( Fig . 3A ) , where 10−9 M was least effective , and 10−8 and 10−7 M caused 15% and 28% decreases in cell counts compared to untreated cells after treatment for 7 d ( Fig . 3A ) . Both effects on growth rate and probability of successful replication were required to describe inhibition of cell proliferation of U87MG cells , as shown by a statistically significant ( p<0 . 01 ) difference in NONMEM's objective function . Parameter estimates suggested U87MG cells being more sensitive to tetrac's effect on growth rate than to the effect on success of replication ( IC50k<IC50R , Table 1 ) . However , the capacity ( i . e . the largest possible effect at very high concentrations of tetrac ) was higher for the effect on success of replication than the effect on rate of growth ( ImaxR>Imaxk ) . For this model the cell count on day 0 was fixed based on the number of seeded cells . From simulation-estimation experiments ( 50 replicates , very rich sampling , additive error on log10-scale = 0 . 1 ) the median bias was −4% for Imaxk , +25% for IC50k , +0 . 4% for ImaxR , and −2% for IC50R , using the MC-PEM algorithm in S-ADAPT . When the same bootstrap datasets plus 50 additional ones were run in NONMEM , the median bias was −2% for Imaxk , +16% for IC50k , −2% for ImaxR and −7% for IC50R ( nominal results from NONMEM shown in Table 1 ) . Bootstrap results for the actual sampling times in the experiments were similar to those from the rich design ( Table 1 ) . The two effects were therefore estimable and distinguishable , both under ideal conditions and in the actual sampling schedule which was employed in our experiment . For additional model evaluation , S-ADAPT with the MC-PEM algorithm was also used to estimate the parameters from the observed data . The S-ADAPT results for ImaxR and IC50R were within 15% of the results from NONMEM , while Imaxk was 22% lower and IC50k was 70% higher than the results from NONMEM . All other parameters were within 40% of their NONMEM estimates . The satisfactory agreement of parameter estimates from two completely different algorithms suggests adequate estimability of the model parameters . In addition , estrogen receptor-α ( ERα ) -negative human breast cancer MDA-MB-231 cells ( MDA-MB ) were treated with 7 different concentrations of tetrac ( 10−8 to 10−5 M ) for 19 d and total cell counts determined every 1–2 d ( Fig . 3B ) . A model with effects on both rate of growth and success of replication ( Fig . 2 ) adequately described the data ( Fig . 3B ) . Parameter estimates from NONMEM are shown in Table 1 . The parameter estimates suggest a higher sensitivity for the effect on probability of successful replication ( IC50R<IC50k , Table 1 ) and a larger capacity of the effect on growth rate ( Imaxk>ImaxR ) . Simulation-estimation experiments ( 50 replicates , additive error on log10-scale = 0 . 1 ) showed a median bias of +3% for Imaxk , +9% for IC50k , −2% for ImaxR , and +6% for IC50R , using the MC-PEM algorithm in S-Adapt . For 100 datasets in NONMEM the median bias was +0 . 5% for Imaxk , −0 . 4% for IC50k , +0 . 5% for ImaxR and +4% for IC50R . The bootstrap results based on the actual sampling design which was also rich were similar ( Table 1 ) . As for the action of tetrac on U87MG cells , both effects were therefore estimable and distinguishable . In S-ADAPT ( MC-PEM ) , the parameter estimates based on the observed data were within 20% of those from NONMEM for 5 parameters and were within 50% for the other 3 parameters . Although tetrac had a growth-suppressive effect late in the treatment period , it may also have a proliferative effect on cancer cells ( results not shown here ) . This is thought to reflect access of the agent to the cell interior where it is a modest thyroid hormone agonist ( thyromimetic ) [9] , [10] , [15] rather than an inhibitor , as it is exclusively at the cell surface receptor . To prevent uptake of tetrac by cancer cells , it was reformulated as poly[lactide-co-glycolide] nanoparticle [8] , [9] , [16] . MDA-MB cells were treated with constant concentrations of 10−6 and 2 . 5×10−6 M tetrac or nano-tetrac for 9 d . Results indicate that the anti-proliferative effect of nano-tetrac in MDA-MB cells is greater than that of unmodified tetrac ( Fig . 4A ) . MDA-MB cells were also treated with 4 different concentrations of nano-tetrac ( 10−9 to 10−6 M ) for 9 d ( Fig . 4B ) . Based on mathematical modeling , the sensitivity of the MDA-MB cells for the nano-tetrac effect on probability of successful replication was considerably higher than for the effect on growth rate ( IC50R = 0 . 0086 µM , IC50k = 6 . 3 µM , Table 1 ) , while the capacity was similar for both effects ( Imaxk = 1 . 0 , ImaxR = 1 . 0 at time = 0 ) . Simulation-estimation experiments ( 50 replicates , additive error on log10-scale = 0 . 1 ) showed a median bias of +12% for IC50k , −0 . 8% for kiR , and +2 . 5% for IC50R , using the MC-PEM algorithm in S-ADAPT . For 100 datasets in NONMEM the median bias was +4 . 0% for IC50k , −2 . 5% for kiR , and −1 . 3% for IC50R . The bootstrap results based on the actual sampling design are shown in Table 1 . The anti-proliferative effect of nano-tetrac was also concentration-dependent in human glioblastoma U87MG cells . At a nano-tetrac concentration of 10−9 M , cell number was reduced by 36% ( control vs . 10−9 M nano-tetrac = 1 . 45×108±3 . 3×107 vs . 2 . 28×108±1 . 9×107 , average±S . D . ) after 7 treatment days ( Fig . 4C ) . Modeling suggested a higher sensitivity for the effect on rate of growth ( IC50k<IC50R , Table 1 ) and a higher capacity for the effect on replication ( Imaxk < ImaxR ) . Both IC50k and IC50R were lower for nano-tetrac than unmodified tetrac in U87MG cells indicating a higher sensitivity to nano-tetrac . For both MDA-MB and U87MG cells , the model includes a decrease in ImaxR of nano-tetrac over time in order to adequately describe the observed cell counts . Such a decrease in ImaxR might potentially be due to functional adaptation or presence of subpopulations with different sensitivities to tetrac . Simulation-estimation experiments ( 50 replicates , additive error on log10-scale = 0 . 1 ) showed a median bias of +2 . 1% for Imaxk , −2 . 8% for kiR , and +5 . 7% for IC50R , using the MC-PEM algorithm in S-Adapt . For 100 datasets in NONMEM the median bias was +1 . 5% for Imaxk , −1 . 5% for kiR , and +1 . 3% for IC50R . The bootstrap results based on the actual sampling design are shown in Table 1 . The individual measurements presented as symbols in Fig . 4B and 4C are the results from 3 repeat experiments , i . e . one data point represents one experiment at each time point . The error bars in Fig . 4A are standard deviations from 3 experiments . The plots of observed versus predicted cell counts are presented in Fig . 5 for unmodified and nano-tetrac in U87MG and MDA-MB cells and show that the time course of cell counts was adequately described . Cells were harvested from the perfusion bellows cell culture system for flow cytometry analysis after 1–3 d of treatment with 10−8 to 10−5 M tetrac . There was a 1 . 8-fold increase of apoptotic cells with 10−5 M tetrac compared to untreated cells at 1 d ( Fig . 6A ) . By days 2 and 3 , all tetrac concentrations caused apoptosis , as determined by TUNEL assay . In cells continuously exposed to tetrac for more than 10 d , only 10−5 M tetrac produced apoptosis consistently ( Fig . 6B ) , suggesting that tetrac may induce some cell proliferation , although the G1 phase was decreased by 50% after 12 d of tetrac treatment . The degree of apoptosis induced by 10−6 M nano-tetrac was 3-fold that of 10−6 M tetrac ( Fig . 6C ) . We have recently reported that tetrac and nano-tetrac induce gene expression profile changes in MDA-MB cells [8] and medullary thyroid carcinoma cells [9] . Experiments presented here examined pro-apoptotic gene expression in tetrac- and nano-tetrac-treated glioblastoma U87MG cells and MDA-MB cells in the perfusion bellows cell culture system . RNA was extracted from the harvested cells at the end of treatment for RT-PCR studies . Treatment of cells for 2 d with nano-tetrac ( 10−6 M ) increased expression of PIG3 , BAD , p21 and p53 in both U87MG and MDA-MB cells ( Fig . 7 ) . In contrast , tetrac ( 10−6 M ) did not significantly increase expression of this panel of genes in U87MG cells and , except for c-jun , gene expression in the MDA-MB cells was enhanced to a lesser extent by tetrac than by nano-tetrac . We have previously observed several differences between gene expression profiles that result from treatment with unmodified tetrac and nano-tetrac [9] . Experiments of flow cytometry and gene expression demonstrate the practicality of harvesting tumor cells from polymer flakes in the perfusion bellows cell culture system for studies of post-treatment states of the cells . We also determined whether tetrac and nano-tetrac had anti-proliferative actions on immortalized non-malignant cells . Monkey kidney epithelial CV-1 cells and human embryonic kidney 293T cells were treated daily with 10−6 M tetrac or 10−6 M nano-tetrac for 7 d . There was no significant change in cell numbers or morphology ( results not shown here ) when untreated cells were compared with those exposed to tetrac or nano-tetrac . A naturally-occurring stilbene , resveratrol [14] , induces apoptosis in human follicular thyroid cancer cells [4] , [17] . Thyroid hormone analogue T4 inhibits the apoptotic action of resveratrol [3] , [4] and tetrac has been shown to restore the pro-apoptotic effect of the stilbene in presence of T4 [3] . This effect of tetrac reflects displacement by tetrac of T4 from the iodothyronine receptor site on integrin αvβ3 . Resveratrol is capable of binding to the integrin αvβ3 [3] , [18] , at a site distinct from that for tetrac and other thyroid hormone analogues [3] , [4] . In the present studies , the anti-proliferative effect of combined resveratrol and tetrac exposure was tested . Cancer cells were treated with resveratrol ( 0 . 1 µM ) in presence or absence of 10−7 M tetrac . Both tetrac and resveratrol individually caused anti-proliferative effects in MDA-MB cells ( Fig . 8A ) , while their combination was additive , based on comparison of cell counts on day 8 and Loewe additivity . Human follicular thyroid cancer ( FTC ) cells were treated daily with resveratrol ( 0 . 1 µM ) in presence or absence of 10−7 M tetrac . Compared with breast cancer cells , FTC236 cells were less sensitive to tetrac ( Fig . 8B ) . The inhibitory effects of resveratrol and tetrac in combination were additive also in FTC cells , based on cell counts on day 10 . Cetuximab is a monoclonal antibody targeted to the extracellular domain of the EGFR intended for use in patients with metastatic colorectal carcinoma and certain other tumors [19] , [20] . Effectiveness is variable [21] , [22] . The drug has been combined clinically with various other chemotherapeutic agents in colorectal cancer patients [21] , [22] and recently has been tested adjunctively in vitro against breast cancer cells [23] . Combining cetuximab with various chemotherapeutic agents has revealed additive or potentiated growth inhibition in various cancer cell lines [21] , [22] . To determine whether tetrac potentiates cetuximab-induced anti-proliferation , human breast cancer MDA-MB cells were treated with cetuximab ( 0 . 1 µg/mL ) in presence or absence of 10−7 M tetrac . Individually , both agents suppressed proliferation of MDA-MB cells ( Fig . 9A ) . After 8 d treatment with cetuximab and tetrac the average total cell counts were decreased by 34% and 38% , compared to control . Combined tetrac and cetuximab was more effective , reducing total cell numbers on average by 63% . Application of an empirical mathematical model to all treatments and time points simultaneously also suggested an approximately additive effect of both compounds . The empirical model was a disease progression type model where the cell counts in the control treatment were described by a simple exponential function . The effect of tetrac was described as an offset , i . e . a change from baseline cell counts while tetrac is present . The effect of cetuximab was described in the same way , only including an additional lag-time of effect . When both drug effects were added the resulting profile adequately described the cell counts during combination treatment for the studied concentrations and observation period . An approximately additive effect was also found for the combination of nano-tetrac and cetuximab in human colon cancer Colo-205 cells ( Fig 9 B ) . Colo-205 cells grown in T-75 flask were treated with either nano-tetrac ( 10−8 and 10−7 M ) , cetuximab ( 4 and 40 µg/ml ) , or combination . Medium was refreshed with agents daily . Cells were harvested and counted as indicated up to 16 days . A model including effects of both drugs on the probability of successful replication and a noncompetitive interaction adequately described the observed cell counts ( Figs . 9B and 9C ) . The effect of the combination treatments was slightly larger than predicted by a competitive interaction model , where both drugs work on the same pathway , and slightly smaller than predicted by a purely noncompetitive interaction model , where the drug works on completely different pathways . Therefore a factor ψ was included ( see equation in the Materials and Methods section ) which was estimated at 5 . 6 . The ImaxR and IC50R for inhibition of the probability of successful replication were 0 . 12 and 7 . 0 nM for nano-tetrac and 0 . 13 and 3 . 3 µg/mL for cetuximab . Using a novel perfusion bellows cell culture system developed in our laboratory ( Fig . 1 ) , we have compared the pharmacodynamics in vitro of unmodified and nanoparticulate formulations of tetrac as anti-proliferative agents . The system revealed that nano-tetrac had a higher potency than tetrac as an anti-proliferative agent ( Fig . 4 ) . Neither nano-tetrac nor tetrac affected proliferation of two non-cancer cell lines even at high concentrations ( 10−6 M ) . The anti-proliferative effect of tetrac and nano-tetrac on cancer cells in the perfusion bellows cell culture system was seen starting 3 d after start of treatment ( Fig . 3 , 4 ) . The anti-cancer effects of tetrac and nano-tetrac in human tumor cell xenografts were well-established within 3 d after onset of drug administration [9] . These results in the perfusion system thus reproduce findings obtained earlier in cells grown in culture dishes and xenografts . While the tetrac effects in xenografts have been shown to involve both primary effects on tumor cell proliferation and an anti-angiogenesis effect [6] , the effect of tetrac and nano-tetrac in the perfusion bellows cell culture system of course is limited to suppression of cell proliferation . In vitro models such as described here can save animals by decreasing the number of animal studies which need to be conducted , by employing well-defined conditions which allow for investigation of individual factors impacting the PD and permitting the simulation of human pharmacokinetics ( PK ) based on data from clinical trials . Limitations of the method described here which need to be considered are that the impact of tissue penetration and the effect of the immune system are usually not directly taken into account; PK/PD models based on animal or clinical studies that include measurement of drug concentrations in tumor need to be developed . In the perfusion system cells are exposed alternately to fresh medium and air . This paradigm optimizes growth conditions for cancer cells by maximizing nutrient uptake and oxygen transfer and supported experiments of up to 3 weeks' duration ( Fig . 3B ) . Information obtained in longer studies about both the slope of the growth/proliferation phase and the plateau of the cell count with regard to time permitted mathematical modeling to identify two different effects of tetrac on cancer cells: inhibition of growth rate and inhibition of success of replication ( Fig . 2 ) . In addition to treating the cells with constant drug concentrations , reflecting in vivo continuous infusion treatment , the in vitro system described here allows to study other dosing regimens . Multiple short-term or intermittent infusions or brief injections can be studied in the perfusion system by adjusting the flow rate of the medium and the dosing schedule . Drug concentration/time profiles that are expected or have been obtained in human or animal studies can be simulated and effects on cancer cells of changing drug concentrations as anticipated in vivo may be observed in the system . Together with mathematical modeling , these in vitro paradigms can support optimization of design of subsequent animal and human studies thereby saving time and expense . Because a wider range of drug concentrations can be studied in vitro than in animal models , dose selection for in vivo studies may become more efficient . Mathematical modeling was utilized to increase the amount of information gained from the reported experiments . By considering the entire time course of cell counts in response to multiple concentrations of tetrac and control treatment simultaneously , more insight can be gained into the dose-response relationship and mechanism of action of a drug . Purely empirical growth models , e . g . , the Weibull model , often do not include meaningful parameters , but offer arbitrary coefficients . For simulating other scenarios , e . g . , cells with faster growth rates , mechanism-based models may be more adequate . While only total cell counts were available in the perfusion bellows cell culture system experiments reported here the applied model is based on mechanisms of action . Inclusion of flow cytometry results in the model will be performed for future experiments in order to enhance the mechanism-based modeling approach . For U87MG cells studied here , mathematical modeling suggested a higher maximum effect but lower sensitivity of the effect on probability of successful replication , compared to the effect on growth rate for both unmodified and nano-tetrac . For both effects the sensitivity favored nano-tetrac over unmodified tetrac . This may be explained by the ability of unmodified tetrac to penetrate into cells and thereby exert low-grade thyromimetic ( proliferative ) effects in addition to the anti-proliferative effects initiated at the cell surface integrin receptor . Therefore the net anti-proliferative effect of unmodified tetrac is decreased . The model describes the net effects of unmodified tetrac . Nano-tetrac does not gain access to the cell interior and shows a more robust anti-proliferative effect . MDA-MB cells had growth rate sensitivity to nano-tetrac that was similar to unmodified tetrac , but a higher sensitivity to nano-tetrac for the effect on success of replication . For both unmodified tetrac and nano-tetrac MDA-MB cells were more sensitive to the effect on success of replication than the effect on growth rate . The uncertainty of the parameter estimates was explored by bootstrap runs . A very rich sampling design was used to ensure the general estimability of the model by two different algorithms . In addition , the estimability was tested under the sampling designs of the actual experiments . For the models of unmodified tetrac , the 10% percentile to 90% percentile intervals ( P10–P90 ) were relatively narrow . A larger uncertainty was seen for the IC50 parameters in the nano-tetrac models , especially for the IC50k in MDA-MB cells . The latter suggests that the effect on rate of growth was not apparent in all of the randomly created bootstrap datasets . Optimal design was not applied to structuring those experiments but will be utilized in future studies . It is important to note that the studied concentrations were 10-fold different between the treatment arms and , based on that factor , the uncertainties in IC50 and the differences in the estimates between NONMEM and S-ADAPT are acceptable . Overall our mechanism-based models have adequately described the cell counts over time in our studies and the effects of a wide range of tetrac concentrations and will support the design of future experiments . In addition to the pharmacodynamic studies in vitro and in animals , also the pharmacokinetics of tetrac will be studied in vivo to more fully characterize the pharmacokinetic / pharmacodynamic relationship for tetrac in vivo . We have previously shown that resveratrol induces apoptosis in human cancer cells , an effect which requires the nuclear translocation of COX-2 and activated ERK1/2 for support of p53-dependent apoptosis [3] , [17] . Resveratrol and tetrac both bind to plasma membrane integrin αvβ3 [1] , [18] , but at discrete sites that apparently do not interfere with one another [3] , [7] . In the present studies , the combination of resveratrol and tetrac was additive in the in vitro perfusion bellows cell culture system in terms of suppression of cell proliferation in two human cancer cell lines . The ability to detect such additivity—or potentiation , if present—is obviously a requirement of the perfusion system . Therapeutic epidermal growth factor receptor ( EGFR ) targeting with cetuximab , either as single agent or in combination with chemotherapy , has demonstrated variable clinical activity [19] and may benefit only select patients [20] . In the perfusion bellows cell culture system , concurrent treatment with tetrac and cetuximab resulted in highly effective inhibition of proliferation of MDA-MB cells by day 8 ( Fig . 9A ) . The model system thus offers the prospect of efficiently exploring a variety of drug combinations . An empirical disease progression model was employed for the combination treatment of MDA-MB cells with tetrac and cetuximab , and revealed an approximately additive effect for the combination . While such an empirical model has limitations it is not feasible to develop a receptor occupancy model for a drug combination without data at multiple drug concentrations . Two concentrations each of nano-tetrac and cetuximab and all four combinations were studied in Colo-205 cells in cell culture flasks . The effects of nano-tetrac and cetuximab were adequately described as inhibition of the probability of successful replication . Modeling of all treatment arms simultaneously revealed an approximately additive effect of the combination . The effect of the combination treatment was slightly smaller than predicted by a purely noncompetitive interaction and slightly larger than predicted by a purely competitive interaction model . This suggests that there is a partial overlap between the mechanisms and pathways of action of nano-tetrac and cetuximab . That interpretation of the modeling results is supported by previous studies in our laboratory where we showed that tetrac interferes with crosstalk between the cell surface receptor for thyroid hormone and EGFR [24] and it can be assumed with confidence that nano-tetrac also interferes with this crosstalk . In addition , nano-tetrac , but not unmodified tetrac , decreases the expression of the EGFR gene [8] . For this study in cell culture flasks it was observed that cell counts in all treatment arms decreased noticeably and approximately in parallel after Day 6 ( Fig 9B ) which cannot be attributed to drug effect . Such observations further support the use of the perfusion bellows cell culture system which provides optimal nutrient uptake and oxygen transfer for the cells and will be utilized for future combination studies in colon cancer cells . The perfusion bellows cell culture studies we described provide useful pharmacodynamic information on the application of new drugs or combinations of various agents in vitro to human cancer cell lines . In combination with pharmacodynamic modeling and by including information about the expected pharmacokinetics of a drug , the perfusion bellows cell culture system permits study of the dose-response relationships of anti-neoplastic agents over a very wide concentration range in vitro , and can support translation from in vitro models to animal models and human clinical trials . Human glioblastoma cells ( U87MG ) , human breast cancer MDA-MB-231 cells ( MDAMB ) , human colon cancer Colo-205 cells , African green monkey kidney epithelial CV-1 cells and human embryonic kidney 293T cells were purchased from ATCC . Human follicular thyroid cancer FTC236 cells were generously provided by Dr . Orlo Clark ( University of California at San Francisco-Mt . Zion Medical Center , San Francisco , CA ) . U87MG cells were maintained for study in MEM ( Gibco , Carlsbad , CA ) supplemented with 10% fetal bovine serum ( FBS , Sigma Aldrich , St . Louis , MO ) . Colo-205 cells were maintained in RPMI ( Gibco ) supplemented with 10% FBS . MDA-MB , CV-1 and 293T cells were maintained in DMEM ( Gibco ) supplemented with 10% FBS . Follicular thyroid cancer cells were supported in 50% DMEM/50% Ham's F-12 ( Gibco ) plus 10 mU/ml of TSH ( Sigma Aldrich ) . Cells were cultured in a 5% CO2/95% air incubator at 37°C . Shown in Fig . 1 is a newly developed perfusion bellows cell culture system that is a disposable bioreactor capable of high density cell culture for studies of anti-cancer drugs . Each cell culture system is a compressible ( bellows ) 500 mL bottle that contains cell culture medium and specially-treated polymer flakes to which cells spontaneously attach and then proliferate . Through moving bellows and porous membranes the level of the medium in the bottle changes periodically . Consequently , the cells are alternately submerged in the culture medium and exposed to 5% CO2/95% air which creates a dynamic interface between air and medium on the plated cell surface that maximizes nutrient uptake and oxygen transfer . The system provides a low shear , high aeration and foam-free culture environment . Proprietary treatment of the surfaces of the flakes enables seating and the harvesting of cells and secreted proteins are readily isolated from the perfusate . In a non-perfusion bellows cell culture system that was also used , the medium in each bottle was replaced by fresh medium every 24 h . In the perfusion bellows cell culture system , medium was progressively refreshed over 24 h , so that one complete change of medium occurred over 24 h . To establish the cultures , 5×107 cells were seeded in perfusion and non-perfusion bellows bottles and incubated overnight at 37°C . Flakes were then harvested , trypsinized , and the cells were collected and counted . The number of cells that attached to flakes was 10–15×106 per bottle . For experiments , the perfusion bellows cell culture system was run for 2 d prior to starting the experiments . The cell numbers at this point were about 30–50×106 cells per bottle . Cultured cells were then exposed to 1% FBS-containing medium . Tetrac or nano-tetrac was added to the medium in the reservoir bottle to achieve the final concentrations reported for each experiment . Nano-tetrac utilized in the studies of proliferation of MDA-MB , U87MG , and Colo-205 cells was manufactured on contract by Azopharma ( Miramar , FL ) . Nano-tetrac for all other experiments was prepared at the Pharmaceutical Research Institute , Rensselaer , NY [9] . Unmodified tetrac was synthesized on contract by Peptido GmbH ( Bexbach , Germany ) . In LC/MS/MS experiments , medium samples ( 20 µL ) were injected onto an HP 1100 series HPLC system ( Agilent Technologies , Palo Alto , CA , USA ) , equipped with a narrow-bore Zorbax Eclipse XDB-C18 column ( 5 µm , 150×2 . 1 mm; Agilent ) . Separation was performed using a mobile phase of 0 . 1% ( v/v ) acetic acid ( A ) and 100% acetonitrile ( B ) , with a linear gradient of 20–60% B over 25 min . Flow rate was maintained at 0 . 2 mL min−1 and elution was monitored by a diode array detector ( 200–600 nm ) . The LC effluent was then introduced into a turbo ion-spray source on a Q/STAR-XL quadruple/time-of-flight ( TOF ) hybrid mass spectrometer ( Applied Biosystems , Foster City , CA , USA ) . Negative ESI mass spectra were acquired over the range m/z 100 to 400 . The electrospray voltage was set at −4 . 5 kV and the source temperature was maintained at 475°C . CID spectra were acquired using nitrogen as the collision gas under collision energies of 25–55 V . High purity nitrogen gas ( 99 . 995% ) was used as the nebulizer , curtain , heater and collision gas source . Total RNA was isolated as described previously [25]–[27] . First strand complementary DNAs were synthesized from 1 µg of total RNA , using oligo dT and AMV Reverse Transcriptase ( Promega , Madison , WI ) . First-strand cDNA templates were amplified for GAPDH , c-fos , PIG3 , c-Jun , and BAD mRNAs by polymerase chain reaction ( PCR ) , using a hot start ( Ampliwax , Perkin Elmer , Foster City , CA ) . Primer sequences were GAPDH ( 5′-AAGAAGATGCGGCTGACTGTCGAGCCACA-3′ [forward] and 5′- TCTCATGGTTCACACCCATGACGAACATG-3′ [reverse ) , c-fos ( 5′-GAATAAGATGGCTGCAGCCAAATGCCGCAA-3′[forward] and 5′-CAGTCA-GATCAAGGGAAGCACAGACATCT-3′ [reverse] ) , PIG3 ( 5′-TGGTCACAG-CTGGCTCCCAGAA-3′ [forward] and 5′-CCGTGGAGAAGTGAGGCAGAATTT-3′ [reverse] ) , c-jun ( 5′-GGAAACGACCTTCTATGACGATGCCCTCAA-3′ [forward] and 5′-GAACCCCTCCTGCTCATCTGTCACGTTCTT-3′ [reverse ) and BAD ( 5′-GTT-TGAGCCGAGTGAGCAGG-3′ [forward] and 5′-ATAGCGCTGTGCTGCCCAGA-3′ [reverse] ) . The PCR cycle was an initial step of 95°C for 3 min , followed by 94°C for 1 min , 55°C for 1 min , 72°C for 1 min , then 25 cycles and a final cycle of 72°C for 8 min . PCR products were separated by electrophoresis through 2% agarose gels containing 0 . 2 µg of ethidium bromide/mL . Gels were visualized under UV light and photographed with Polaroid film ( Polaroid Co . , Cambridge , MA ) . Photographs were scanned under direct light for quantitation and illustration . Results from PCR products were normalized to the GAPDH signal . Cells were harvested from flakes by trypsinization , washed with PBS , fixed in ice-cold 70% ethanol and stored in a freezer overnight . Cells were labeled to detect apoptosis with the In situ Cell Death Detection Kit , Fluorescein ( Roche Diagnostics Corporation , Roche Applied Science , Indianapolis , IN ) . The recommended procedures were used with modifications in permeabilization time and temperature to improve results . Fixed cells were centrifuged and washed once in PBS containing 1% bovine serum albumin ( BSA ) , then resuspended in 2 mL permeabilization buffer ( 0 . 1% Triton X-100 and 0 . 1% sodium citrate in PBS ) for 25 min at room temperature , followed by a wash in 0 . 5 mL PBS/1% BSA . Cells were resuspended in 50 µL TUNEL reaction mixture ( TdT enzyme and labeling solution ) and placed in an incubator for 60 min at 37°C in a humidified dark atmosphere . Labeled cells were washed in PBS/1% BSA , then resuspended in 0 . 5 mL ice-cold PBS/0/1% BSA Triton X-100 that contained 1 µg/mL 4′ , 6-diamidino-2-phenylindole ( DAPI ) for at 20 min . Cell samples were analyzed with a BD™ LSR II ( BD Biosciences , San Jose , CA ) , using BD FACSDiva™ software . Fluorescence histograms were gated on forward scatter ( FSC ) and side scatter ( SSC ) to exclude debris and clumped cells . Gating on height vs . area fluorescence of DAPI signal was set to eliminate clumped cells and to obtain the singlet population for analyzing the cell cycle phase ratios in G1 , S or G2/M . Immunoblot and nucleotide densities were measured with a Storm 860 phosphorimager , followed by analysis with ImageQuant software ( Molecular Dynamics , Sunnyvale , CA ) . Student's t test , with P<0 . 05 as the threshold for significance , was used to evaluate the significance of the hormone and inhibitor effects . Where cell counts were tested for statistical significance , the data were log-transformed prior to testing . For the cell count data , an α-adjustment to account for multiple comparisons was utilized according to the Holm t test . The concept of Loewe additivity [28] was applied to cell count data from combination treatments . For experiments involving cells counts at many time points , for multiple treatments , or both , multiple t tests were not an adequate method of analysis due to the large number of comparisons . In addition , multiple comparison tests treat the observations at each time point independently , whereas mathematical modeling , as described below , takes into account the full time course . Observed data are presented in the figures as individual data points or average ± standard deviation ( SD ) . The time course of cell counts of the several cancer cell lines treated with different concentrations of tetrac or nano-tetrac ( or a combination of tetrac with cetuximab or resveratrol , or nano-tetrac with cetuximab ) was modeled utilizing a naïve pooled approach in NONMEM VI ( version 6 . 2 ) . The pooled approach does not distinguish any potential unexplained variability between the bottles ( treatment arms ) from general assay error , e . g . , uncertainty in cell counts , but expresses both in the residual error . The perfusion bellows cell culture system experiments included one bottle per treatment arm with the multiple observations per time point being different cell counts of one sample for the tetrac experiments and the nano-tetrac with cetuximab combination study , and average cell counts from three studies for the nano-tetrac experiments . The population approach in NONMEM ( FOCE ) did not succeed in distinguishing inter-subject variability ( variability between bottles ) and unexplained random variability ( e . g . general assay error ) . The naïve pooled analysis in NONMEM was equivalent to a pooled analysis using the Maximum Likelihood approach in ADAPT , for example . S-ADAPT was also utilized as described below in order to make use of the MC-PEM algorithm and for additional model evaluation . All time points and treatment arms within each experiment were modeled simultaneously . A mechanism-based model [29] was adapted to describe the proliferation of cancer cells and the inhibition of proliferation by tetrac . This model assumes two populations of cells in different phases of the cell cycle: cells that are preparing for replication ( phase 1 ) and cells that are immediately ‘pre-replication’ ( phase 2 ) . Cells transition from phase 1 to phase 2 by a first-order growth rate constant , while replication from phase 2 to phase 1 is assumed to be fast ( Fig . 2 ) . The number of cells in phase 1 and 2 are described by:where C1 is the number of cells in phase 1 , C2 the number of cells in phase 2 , k21 the first order rate constant for replication ( transition from phase 2 to phase 1 ) , and k12 the first-order growth rate constant for transition from phase 1 to phase 2 . The k21 was assumed to be fast and therefore was fixed to 100 day−1 , which resulted in a ratio of k21/k12 of approximately 50 to 100 , depending upon the cell line . The total number of cells Ct is the sum of C1 and C2 . Rep is the replication efficiency factor which is described by:where Cmax is the maximum number of cells . Without tetrac ( or nano-tetrac ) , the replication efficiency factor approaches 2 , which reflects a 100% probability of successful replication . When Ct approaches Cmax , Rep approaches 1 , representing a 0% probability of net replication , that is , cells in reality still transition between the phases , but the number of cells does not increase further . The InhR describes the inhibitory effect of tetrac on the probability of successful replication:where ImaxR is the maximum effect of tetrac ( or nano-tetrac ) on the probability of successful replication and IC50R is the tetrac concentration needed to achieve a half-maximal effect . In the case of InhR< 0 . 50 , this effect results in cell killing , as it then follows that Rep • InhR< 1 . 0 . The latter case also illustrates that cells which do not replicate successfully are eliminated in this process . For some studies , inclusion of a decrease in ImaxR over time was necessary in order to adequately describe the data:where ImaxR0 is the ImaxR at time = 0 and kiR is a constant describing the decrease of ImaxR over time . Inhk describes the inhibitory effect of tetrac on the rate of growth:where Imaxk is the maximum effect of tetrac on rate of growth and IC50k is the tetrac concentration needed to achieve a half-maximal effect . Both IC50R and IC50k are measures for the sensitivity of the cancer cells to the effects of tetrac . A low IC50 corresponds to a high sensitivity of the cells to a particular drug effect , and vice versa . While the InhR describes an irreversible removal of cells from the cell cycle , Inhk only slows down the transitioning of cells through the cell cycle . The cells remain in state 1 for a longer period of time which represents growth and preparation for replication . This is reflected in a decreased slope of the growth curve . Although cells in state 1 and state 2 were not measured separately in the perfusion bellows cell culture system experiments reported here , the two effects were distinguishable and the parameters estimable . The effect on rate of growth decreases the slope of the growth curves whereas the effect on successful replication results in lower plateaus at the end of the growth curves for the treatment arms compared to control . As described below simulation estimation runs were performed to confirm the estimability of the parameters . The effects of nano-tetrac were modeled by the same equations as described above for unmodified tetrac . However the IC50 estimates for nano-tetrac are hypothetical concentrations that assume all of the tetrac bound to the nanoparticle is available for binding to the integrin receptor . A lag time for growth was included in order to describe the data successfully . The parameter k12 was low at the start of the experiment and increased over time:Here , k12max is the maximum growth rate constant and b and c are empirical constants . The residual variability was described by an additive error on log-scale . A model for non-competitive interaction was applied to the experiment on the effects of nano-tetrac , cetuximab , and their combination on Colo-205 cells . The effects of nano-tetrac ( InhRNPT ) and cetuximab ( InhRCET ) were described as:The effect of the combination was:which describes a non-competitive interaction [30] , [31] when ψ = 1 , that is both drugs act by completely separate pathways [32] , [33] . When ψ>1 , then the effect of the combination is less than would be expected from two drugs acting completely independent of each other . The decrease of cell counts in all treatment arms towards the end of the observation period in this study in cell culture flasks was modeled by a series of transit compartments . Model discrimination was based on comparison of the objective function in NONMEM , visual comparisons of observed and fitted cell counts over time , and observed vs . fitted plots . Simulation estimation experiments ( bootstraps ) were performed for the models of tetrac and nano-tetrac effects on U87MG and MDA-MB cells in order to explore the estimability of the model and the bias and uncertainty in the parameter estimates . The simulations were done in Berkeley Madonna ( v . 8 . 3 . 14 ) . The estimations were performed in both NONMEM ( pooled approach ) and the MC-PEM ( Monte Carlo parametric expectation maximization ) algorithm in parallelized S-ADAPT ( v . 1 . 56 ) . One hundred bootstrap datasets in NONMEM and fifty bootstrap datasets in S-ADAPT , each with 10 profiles per treatment arm , were run for each of the four experiments ( two cell lines and two formulations ) , assuming a very rich sampling schedule and an additive error on log-scale of 0 . 1 ( Bootstraps based on additive errors on log-scale of 0 . 02 , 0 . 05 , and 0 . 1 had been previously conducted for the model of tetrac effects in U87MG cells ) . As the bootstraps were performed in order to obtain a point estimate for the parameters and not to characterize their distribution , and also due to long run times , 50 to 100 bootstrap runs each were adequate . Those bootstraps based on the rich sampling schedule were conducted to evaluate the mathematical estimability of the model parameters under ideal experimental conditions , i . e . many sampling time points . One hundred bootstrap datasets each with 10 profiles per treatment arm were run in NONMEM for each of the four models with the sampling schedules that were actually used in the experiments and assuming an additive error on log-scale of 0 . 1 . The bootstraps based on the actual sampling schedules were performed to test whether the model parameters were well-estimable based on both the model and the experimental conditions . The median and 10% and 90% percentiles were calculated from each of those simulation estimation experiments .	Clinical treatment protocols for specific solid cancers have favorable response rates of 20%–25% . Cancer cells frequently become resistant to treatment . Therefore , novel anti-cancer drugs and combination regimens need to be developed . Conducting enough clinical trials to evaluate combinations of anti-cancer agents in several regimens to optimize treatment is not feasible . We showed that tetrac inhibits the growth of various cancer cell lines . Our newly developed in vitro system allowed studying the effects of tetrac over time in various human cancer cell lines . Our mathematical model could distinguish two effects of tetrac and may be used to predict effects of other than the studied dosage regimens . Human breast cancer cells were more sensitive to the effect on success of replication than the effect on growth rate , whereas the maximum possible effect was larger for the latter effect . Nanoparticulate tetrac , which does not enter into cells , had a larger effect than unmodified tetrac . The combinations of tetrac and resveratrol , tetrac and cetuximab ( Erbitux ) , and nano-tetrac and cetuximab showed approximately additive effects . Our in vitro perfusion system together with mathematical modeling may be useful for dose-finding , translation from in vitro to animal and human studies , and studying effects of other chemotherapeutic agents or their combinations .	[ "Abstract", "Introduction", "Results", "Discussion", "Materials", "and", "Methods" ]	[ "oncology", "oncology/breast", "cancer", "oncology/oncology", "agents", "oncology/neuro-oncology" ]	2011	Pharmacodynamic Modeling of Anti-Cancer Activity of Tetraiodothyroacetic Acid in a Perfused Cell Culture System
RNA-DNA hybrids are naturally occurring obstacles that must be overcome by the DNA replication machinery . In the absence of RNase H enzymes , RNA-DNA hybrids accumulate , resulting in replication stress , DNA damage and compromised genomic integrity . We demonstrate that Mph1 , the yeast homolog of Fanconi anemia protein M ( FANCM ) , is required for cell viability in the absence of RNase H enzymes . The integrity of the Mph1 helicase domain is crucial to prevent the accumulation of RNA-DNA hybrids and RNA-DNA hybrid-dependent DNA damage , as determined by Rad52 foci . Mph1 forms foci when RNA-DNA hybrids accumulate , e . g . in RNase H or THO-complex mutants and at short telomeres . Mph1 , however is a double-edged sword , whose action at hybrids must be regulated by the Smc5/6 complex . This is underlined by the observation that simultaneous inactivation of RNase H2 and Smc5/6 results in Mph1-dependent synthetic lethality , which is likely due to an accumulation of toxic recombination intermediates . The data presented here support a model , where Mph1’s helicase activity plays a crucial role in responding to persistent RNA-DNA hybrids . RNA-DNA hybrids can form during transcription when the nascent RNA base pairs with the template strand of the DNA . This results in a 3-stranded structure , referred to as an R-loop [1 , 2] . R-loops lead to DNA replication stress and hence increased rates of DNA breakage , unscheduled recombination , and genome rearrangements [3 , 4] . It is unclear exactly how R-loops cause replication stress , but it may be linked to their ability to promote local chromatin compaction [5] . R-loops can be removed by the RNase H enzymes ( RNase H1 and RNase H2 ) , which can degrade the RNA moiety of an RNA-DNA hybrid molecule . In yeast , RNase H1 is encoded by the RNH1 gene , while the RNase H2 enzyme is a trimeric complex made up of the gene products of RNH201 ( the catalytic subunit ) , RNH202 , and RNH203 [6 , 7] . Mutations in all three subunits of RNase H2 have been linked to the neurological auto-immune disorder , Aicardi-Goutières syndrome ( AGS ) [8] . In addition to RNase H-mediated hybrid degradation , the THO complex counteracts the accumulation of RNA-DNA hybrids . The THO complex is a transcription-coupled RNA processing complex that limits the formation of RNA-DNA hybrids in a co-transcriptional manner [9] by “capturing” the nascent RNA and ensuring that it gets efficiently exported . Furthermore , multiple helicases have been implicated in RNA-DNA hybrid removal via displacement of the RNA strand , including yeast Pif1 [10] and Sen1 [11] , human SETX ( the equivalent of yeast Sen1 ) [12] , AQR [13] , DDX19 [14] , DDX23 [15] , DDX1 [16] and DDX21 [17] . When R-loop removal factors are inactivated , many DNA repair genes ( namely those involved in homologous recombination ) become essential , underlining the propensity of R-loops to promote DNA damage as a result of replication stress [1 , 2] . RNA-DNA hybrids exist at telomeres in yeast and human cells and promote homology-directed repair [18–20] demonstrating their involvement in telomere maintenance . As telomeres shorten , RNA-DNA hybrids accumulate together with increased levels of the long non-coding RNA , telomeric repeat-containing RNA ( TERRA ) [21–25] , suggesting that telomeric RNA-DNA hybrids are made up of TERRA . The recent observations that the Fanconi Anemia repair pathway , which is involved in the repair of DNA inter-strand crosslinks and obstacles that impede replication fork progression , is crucial to dissolve R-loops that may block replication forks in human cells [26 , 27] suggest a relevant role of this DNA repair pathway in R-loop prevention and removal . Human FANCM has the ability to branch migrate replication structures , resolve RNA-DNA hybrids in vitro as well as to prevent RNA-DNA hybrid accumulation in vivo in human and chicken cells [27] . Although budding yeast cells lack a canonical Fanconi Anemia pathway , yeast Mph1 helicase stands out as a homolog of the FANCM protein [28] . In vitro , Mph1 is able to dissociate D-loops and promote replication fork reversal , similarly to FANCM [29] . Both the deletion and overexpression of yeast Mph1 are associated with increased replication stress and genome instability [30 , 31] , respectively indicating that its activity is important when facing replication stress , but must be tightly regulated in order to prevent toxicity . Interestingly , the loss of Mph1 has been reported to lead to a synthetic growth defect in the absence of RNH203 [32] , pointing to a possible role for Mph1 in R-loop regulation . Several studies have reported that the Smc5/6 complex is a negative regulator of Mph1 [33–35] . Two SMC subunits ( Smc5 and Smc6 ) and 6 non-SMC elements form this highly conserved complex , which structurally resembles the cohesin , condensin and the MRN complexes [36] . The deletion of MPH1 is able to rescue the temperature sensitive growth defect of smc5/6 mutant alleles as well as the accumulation of X-shaped recombination intermediates after treatment with DNA damaging agents [33–35] . Interestingly , the inactivation of the Smc5/6 complex [37 , 38] , as well as overexpression of Mph1 [39] , have been demonstrated to drive yeast into premature replicative senescence . Given the observations outlined above , we set out to investigate the role of Mph1 and its established negative regulator , Smc5/6 at RNA-DNA hybrids in vivo in the budding yeast Saccharomyces cerevisiae . In this study , we show that Mph1 forms nuclear foci when RNA-DNA hybrids accumulate , i . e . in rnh1 rnh201 mutants and at short telomeres . We demonstrate that Mph1 , and in particular its helicase function , suppresses the accumulation of RNA-DNA hybrids , at RNAPII-transcribed genes as well as at telomeres , and that in the absence of Mph1 , cells accumulate recombinogenic DNA damage in an RNA-DNA hybrid-dependent manner . Accordingly , Mph1 becomes essential when RNA-DNA hybrid removal is strongly impaired , as seen by the severe growth defect of the rnh1 rnh201 mph1 triple mutants . However , when the functionality of the Smc5/6 complex is compromised , Mph1’s helicase activity becomes deleterious in situations when RNA-DNA hybrids accumulate . We propose that Mph1 plays an important role in RNA-DNA hybrid metabolism and that its activity has to be tightly controlled by the Smc5/6 complex . Given the involvement of the human Fanconi Anemia M protein in preventing R-loop-mediated damage and RNA-DNA hybrid accumulation in vivo , we decided to address whether yeast cells deleted for the FANCM homolog MPH1 or other candidate helicases implicated in DNA repair are also involved in R-loop metabolism . Deletion mutants of SGS1 , SRS2 , MPH1 , RRM3 were assayed for the accumulation of Rad52 foci , a marker for DNA repair at sites of damage . Genotypes were considered as specifically accumulating RNA-DNA hybrid-induced damage when the increased number of Rad52 foci was abolished upon RNase H1 overexpression . The deletion of the RNase H enzymes ( rnh1 rnh201 ) and HPR1 , a member of the THO complex , where R-loop accumulation has been demonstrated to account for the majority of DNA damage [9] , served as positive controls . It is worth noticing that Rad52 foci were not fully suppressed in rnh1 rnh2 cells , which suggests that hybrids not easily accessible to RNase H1 may accumulate in these double mutants . Although all four single gene mutants tested led to a significant increase in Rad52 foci compared to wild type , only MPH1 deleted cells showed an RNase H1-sensitive increase ( Fig 1A ) . Therefore , we focused on achieving a more complete understanding of Mph1 and its role at RNA-DNA hybrids . To rule out yeast genetic background specific effects , we confirmed the accumulation of Rad52 foci in mph1 mutants in the W303 genetic background ( Fig 1B ) . We performed a western blot on an endogenously tagged Rad52 strain ( Rad52-TAP ) with and without RNase H1 overexpression to rule out an effect on Rad52 protein levels ( S1A Fig ) . This indicates that Mph1 has a role in the prevention of RNA-DNA hybrid formation or in minimizing the damage induced by RNA-DNA hybrids already formed . In order to get an indication of whether Mph1 may be acting directly at RNA-DNA hybrids , we tested whether it forms foci in mutants that are known to accumulate hybrids . We observed an increased frequency of endogenously expressed Mph1-GFP foci in rnh1 rnh201 double mutant cells , where hybrids accumulate to high levels ( Fig 1C ) . Mph1-GFP foci specifically accumulated in rnh1 rnh201 double mutant cells in the S- or G2 phase of the cell cycle and not in G1 ( Fig 1C ) , suggesting that Mph1 plays a role at RNA-DNA hybrids during DNA replication to prevent the accumulation of DNA damage . As Mph1 forms foci when RNA-DNA hybrids accumulate , we next investigated whether its function becomes essential when known RNA-DNA hybrid regulating factors are impaired . Indeed , it has been reported that the loss of MPH1 leads to growth impairment in the absence of RNH203 in the S288C background [32] . In the BY4741 and W303 genetic backgrounds , the mph1 rnh1 rnh201 triple mutants were either severely growth compromised or completely inviable , respectively ( Fig 1D and S1B Fig , for detailed growth curves and population doubling times see S1C Fig ) . In order to test if this essential function of Mph1 occurs during S phase , we created an allele of RNH1 ( S-RNH1 ) that is specifically expressed in S phase , by placing RNH1-TAP under the Clb6 promoter and fusing it to a Clb6 degron sequence [40] ( Fig 1E , S1D Fig ) . Compared to the rnh1 rnh201 mph1 triple mutant , the expression of Rnh1 in S phase did not have a growth defect in rnh201 mph1 mutants ( Fig 1F ) . Consistent with Mph1 preventing R-loop-induced damage , the slow growth of the rnh1 rnh201 mph1 mutants was associated with a slightly increased phosphorylation of Rad53 as well as induced expression of RNR3 ( Fig 1G ) , both being established markers for activation of the DNA damage checkpoint . To test whether Mph1’s helicase activity is essential to restore the growth of rnh1 rnh210 mph1 triple mutants , we reintroduced helicase-compromised Mph1 mutants , mph1-E210Q and mph1-Q603D [34] , on a plasmid by expressing them from the endogenous MPH1 promoter . Whereas wild type Mph1 fully suppressed the growth defect and checkpoint activation , neither of the helicase mutants was able to rescue the synthetic interaction ( Fig 1H and S1E Fig ) , despite being expressed at similar levels ( S1F Fig ) . As RNase H2 activity is capable of removing single ribonucleotides that have been misincorporated into DNA helix by the replicating DNA polymerase [41 , 42] , as well as consecutive RNA-DNA hybrids ( i . e . those formed in an R-loop ) , we wanted to determine at which type of RNA-DNA hybrid Mph1 was functioning . Therefore , we employed the RNH201-P45D-Y219A mutant , which is specifically defective in mono-ribonucleotide excision repair ( RER ) , but is proficient in removing longer hybrid stretches ( hereafter called RNH201-RED for ribonucleotide excision defective ) [43 , 44] . We transformed this plasmid into the rnh1 rnh201 mph1 cells and observed that its expression complements the synthetic growth defect to the same extent as wild type RNH201 ( S1G Fig ) . Consistently , the checkpoint activation , as monitored by Rnr3 expression , was also alleviated upon expression of both wild type RNH201 and RNH201-RED ( S1H Fig ) . We verified that the RNH201-RED allele is RER-defective in vivo by demonstrating that the expression of this mutant could not suppress the hydroxyurea ( HU ) sensitivity of pol2M644G rnh201 double mutants , which harbors high levels of misincorporated ribonucleotides [45] , whereas wild type RNH201 was able to complement ( S1I Fig ) . In summary , these data indicate that Mph1’s helicase activity becomes essential during DNA replication when consecutive RNA-DNA hybrids ( such as those present in R-loops ) accumulate and may not be required at misincorporated ribonucleotide insertions . As Mph1’s activity is required in rnh1 rnh201 double mutants we tested whether Mph1 is also essential in other mutants that have been reported to accumulate RNA-DNA hybrids . Unlike the loss of both RNase H1 and H2 functions , we found that deleting MPH1 does not result in growth impairment when either of the THO components , THP2 ( S2A Fig ) or HPR1 , are deleted ( S2B Fig ) , or in combination with the temperature-sensitive sen1-1 allele ( S2C Fig ) . Consistently , we also did not see evidence of DNA damage checkpoint activation as shown by both lack of detectable Rad53 phosphorylation and RNR3 induction , in mph1 hpr1 double mutants in the BY4741 background ( S2D Fig ) . Nonetheless , Mph1 foci were observed in hpr1 mutants and these foci were suppressed by RNase H1 overexpression ( S2E Fig ) , consistent with the notion that Mph1 foci are enriched at R-loops . These results suggest that Mph1 may only become essential when RNA-DNA hybrids reach very high levels as in RNase H-deficient double mutant cells ( see Fig 2A right ) . Alternatively , Mph1 may be required to act on a subset of RNA-DNA hybrids that is distinct from those affected by the THO and Sen1 proteins . The combined results of the synthetic growth defect between mph1 and rnh1 rnh201 mutants together with the fact that Rad52 foci accumulate in mph1 cells in an RNase H1-sensitive manner led us to speculate that RNA-DNA hybrids may accumulate in mph1 mutants . Since R-loops frequently lead to hyper-recombination , we sought to study recombination in the absence of Mph1 . For that , we used previously devised direct-repeat recombination assays based on two truncated copies of the LEU2 gene [46] . Unlike mutants of the THO complex or sen1-1 mutants , where R-loops accumulate , deletion of MPH1 results in only conservative changes , or no changes at all , in recombination , depending on the assay used ( S3A Fig , S3B Fig ) . These data , together with the RNase H-sensitive increase of Rad52 foci ( Fig 1A ) suggest that in the absence of the Mph1 helicase activity , the damage caused by RNA-DNA hybrids is not efficiently resolved into detectable recombination products and hence results in the accumulation of Rad52 foci . To assay for the presence of RNA-DNA hybrids , we performed indirect immunofluorescence on chromosome spreads . We detected an increase in the number cells with S9 . 6 foci in mph1 mutant cells , indicating increased levels of RNA-DNA hybrids ( Fig 2A ) . Cells containing more than 3 foci , were especially prevalent when MPH1 was deleted . hpr1 and rnh1 rnh201 cells served as positive controls , where hybrids are known to accumulate ( Fig 2A , right panel ) . To confirm the accumulation of RNA-DNA hybrids in mph1 mutants we used DNA-RNA immunoprecipitation ( DRIP ) to pull down RNA-DNA hybrids as previously described [26 , 47] . In agreement with the immunofluorescence data ( Fig 2A ) , RNA-DNA hybrids accumulated significantly within the protein coding genes , GCN4 ( Fig 2B , left panel ) and PDC1 ( right panel ) in mph1 mutants . The RNA-DNA hybrids signal could be strongly reduced upon in vitro RNase H1 treatment , thereby demonstrating the specificity of the DRIP signal . To investigate whether Mph1’s helicase activity is needed to prevent the accumulation of RNA-DNA hybrids , we also performed DRIP on helicase-dead Mph1 mutants ( mph1-E210Q and mph1-Q603D ) [36] , which accumulated RNA-DNA hybrids at the GCN4 ( Fig 2B , left ) and the PDC1 ( right ) loci similarly to the complete deletion . To better understand how Mph1 is involved in RNA-DNA hybrid regulation , we expanded our DRIP analysis to telomeres ( telomere 6R ) and the rDNA loci ( Fig 3 ) . Whereas loss of Mph1 function did not affect hybrid levels at the 18s rDNA locus , it led to an increase at telomere 6R , suggesting that Mph1 may act preferentially at RNA polymerase II transcribed loci . When DRIP was performed in the mph1 rnh1 rnh2 mutants ( only viable in BY4741 background , Fig 1 ) , we did not observe any further increase in RNA-DNA hybrids with respect to the single mph1 or double rnh1 rnh2 mutants ( Fig 3 ) . A similar epistasis was observed for Rad52 foci ( Fig 1A ) , suggesting that Mph1 may remove RNA-DNA hybrids via RNase H . The overexpression of Mph1 drastically increases the rate of replicative senescence in cells harboring short telomeres following the loss of telomerase activity [39] , implying that it may be functionally relevant at telomeres . Levels of the non-coding telomeric repeat-containing RNA ( TERRA ) increase as telomeres shorten in yeast and human cells [22 , 23 , 48] . Moreover , RNA-DNA hybrids ( presumably involving TERRA ) can be detected at telomeres and accumulate in the absence of the RNase H enzymes [18–20 , 49] . We allowed wild type and telomerase negative ( tlc1 ) cells to undergo approximately 60 population doublings and , as expected , observed an increase in TERRA levels at all tested telomeres ( Fig 4A ) . Upon performing DRIP on wild type and telomerase negative cells to monitor hybrid levels , we observed an increase in telomeric hybrids despite the fact that they were shorter ( Fig 4B ) . Importantly , in an independent experiment , we observed that the overexpression of RNase H1 abolished the increased DRIP signal in tlc1 cells ( S4A Fig ) , demonstrating the specificity of the RNA-DNA hybrid signal shown in Fig 4B . To verify that Mph1 was acting at short telomeres in an RNA-DNA hybrid dependent manner , we monitored Mph1-YFP foci formation as telomeres shorten in telomerase negative cells ( est2 ) . Importantly , we observed that Mph1-YFP foci accumulate as telomeres shorten ( Fig 4C and 4D ) . At the peak of senescence more than 50% of Mph1 foci co-localize with Cdc13 ( which represent dysfunctional telomeres ) and Rad52 foci , thereby indicating that they accumulate at critically short and dysfunctional telomeres ( Fig 4C–4E ) [50] . Importantly , Mph1’s ability to form foci was greatly reduced when RNase H1 was overexpressed ( Fig 4C and 4D ) . Strikingly , the number of Cdc13 foci was also decreased upon RNase H overexpression , suggesting that hybrids may be a source of DNA damage at telomeres upon shortening . By ChIP analysis we could see that loss of the C-terminal domain of Mph1 , which has been shown to interact with Rfa1 and Mte1 [30 , 51] , prevented Mph1 from localizing to telomeres in telomerase positive cells ( S4B Fig ) indicating that Mph1 location at telomeres may depend on its interaction with replication protein A ( RPA ) . Taken together , we conclude that Mph1 accumulates at short telomeres , in an RNA-DNA hybrid-dependent manner . Previous data indicates that the Smc5/6 complex is important to limit the accumulation of toxic recombination intermediates in the presence of MMS ( methyl methanesulfonate ) , and at natural pause sites [34 , 35 , 52–54] . A negative genetic interaction between the smc6 mutants and rnh201 as well as rnh202 was previously described in a high-throughput SGA screen [55] . Furthermore , there is ample evidence that the Smc5/6 complex is important for the accurate replication of the rDNA locus , a repetitive genomic region rich in RNA-DNA hybrids [56 , 57] . Based on these observations , we hypothesized that the Smc5/6 complex might be required for the accurate processing of DNA replication intermediates that arise when RNA-DNA hybrids are encountered by the replisome . To test this notion , we introduced the smc6-9 and smc5-6 temperature sensitive alleles into strains defective for either RNase H1 ( rnh1 ) or RNase H2 ( rnh201 ) activity , as well as in rnh1 rnh201 double mutants . The absence of RNH201 , but not RNH1 , resulted in a severe growth defect , when either SMC6 ( Fig 5A ) or SMC5 ( S5A Fig ) function was reduced . Next , we tested whether the deletion of MPH1 would rescue the above described synthetic growth defect between smc5/6 mutants and rnh201 . We observed that growth defects of both smc6-9 rnh201 and smc5-6 rnh201 cells were alleviated in the absence of MPH1 ( Fig 5B , and S5B Fig ) . In accordance with the synthetic growth defect occurring as a result of replication stress , we observed that smc6-9 rnh201 cultures accumulated cells with a 2N DNA content in comparison to both wild type and the respective single mutants ( S5C Fig ) . Importantly , the accumulation of cells with a 2N DNA content was reversed when MPH1 was additionally deleted ( S5C Fig ) . In order to confirm that DNA damage accumulated in smc6-9 rnh201 cells , we followed the formation of Rad52-mCherry foci ( Fig 5C ) . We observed a dramatic increase in the number of cells with Rad52-mCherry foci in smc6-9 rnh201 double mutants when compared to wild type and single mutant cells grown at semi-permissive temperature ( Fig 5C , bottom panel ) . Strikingly , the deletion of MPH1 reverted this accumulation to the levels observed in the respective single mutants ( Fig 5C ) . We speculated that Mph1-dependent recombination intermediates accumulate and may be contributing to cellular toxicity . Indeed , upon deleting either the recombination factors RAD52 or RAD51 in smc6-9 rnh201 mutants we observed an increased viability in the triple mutants as compared to the respective double mutants in a similar manner as to when MPH1 was deleted ( Fig 5D and S5D Fig ) . While the re-introduction of full-length MPH1 led to a severe growth defect in smc6-9 rnh201 mph1 cells , mph1-E210Q and mph1-Q603D helicase mutants behaved similar to the complete deletion of MPH1 ( Fig 5E ) . Moreover , introduction of the RNH201-RED allele complemented the synthetic growth defect , nearly to the same extent as wild type RNH201 ( S5E Fig ) . The overexpression of RNase H1 , was not able to alleviate the growth defects of the double mutants , which may indicate an RNase H2 specific sub-set of RNA-DNA hybrids that lead to toxicity ( S5F Fig ) . Finally , the deletion of RNH202 , an auxiliary component of RNase H2 also showed negative growth defects when combined with smc6-9 ( S5G Fig ) . A plasmid harboring full-length RNH202 complemented the growth defect as did an RNH202-ΔPIP mutant , where the proliferating cell nuclear antigen ( PCNA ) interaction motif had been mutated [43] ( S5G Fig ) . These data strongly suggest that the helicase activity of Mph1 needs to be counteracted by the Smc5/6 complex at consecutive RNA-DNA hybrids , and to a lesser extent at misincorporated ribonucleotides , in order to prevent the accumulation of toxic recombination products . Finally , we investigated whether the Smc5/6-mediated negative regulation of Mph1 function at RNA-DNA hybrids may also be conserved at short telomeres , where hybrids accumulate ( Fig 4B ) . Interestingly , Smc5/6 has recently been demonstrated to regulate TERRA levels [58] and we observed that Mph1 suppresses the accumulation of RNA-DNA hybrids at telomere 6R ( Fig 3 ) . In the absence of a functional Smc5/6 complex , telomerase negative cells ( est2 ) lose growth potential at very early population doublings ( Fig 5F ) consistent with the previously reported premature senescence phenotype [37 , 38] . Importantly , the additional loss of MPH1 alleviated , to a great extent , the premature senescence of smc6-9 est2 cells ( Fig 5F ) . These data led us to the conclusion that the Smc5/6 complex may be required to limit or process Mph1-mediated intermediates at short telomeres harboring RNA-DNA hybrids . Importantly , we did not detect increased levels of R-loops at either the GCN4 or PDC1 loci when either SMC5 or SMC6 were inactivated ( S5H Fig ) . In summary , our combined observations suggest that while Mph1 activity is required during replication through RNA-DNA hybrids , its unrestrained helicase activity may lead to the formation of lethal recombination intermediates at RNA-DNA hybrids . However , this toxic activity can be prevented by the presence of a functional Smc5/6 complex . We set out to identify additional helicases that are involved in RNA-DNA hybrid metabolism through our small-scale candidate genetic screen looking for mutants with an accumulation of DNA damage ( Rad52 foci ) that is abolished by overexpression of RNase H1 . Here we identified Mph1 as a factor preventing DNA damage at RNA-DNA hybrids . We analyzed the effects of Mph1 in RNA-DNA hybrid turnover in more detail and found that Mph1 not only forms foci when RNA-DNA hybrids accumulate , but is also instrumental in preventing RNA-DNA hybrids from accumulating . Furthermore , we observed that mph1 rnh1 rnh201 triple mutants are severely compromised for growth ( Fig 1D ) , pointing towards Mph1 acting in either an alternative and/or complementary pathway to RNase H1 and RNase H2 . The fact that Mph1 foci in rnh1 rnh201 cells specifically occurred in S/G2 , but not in G1 cells ( Fig 1C ) suggests that Mph1 plays a role at RNA-DNA hybrids after they are encountered by a replication fork . This is further supported by the fact that the expression of RNH1 specifically in S phase , is sufficient to reverse the lethality of rnh1 rnh201 mph1 mutants ( Fig 1F ) . Taking the in vitro enzymatic activities into account , we speculate that Mph1 may promote fork reversal ( see below ) , which can contribute to fork restart [59] , when replication forks stall at RNA-DNA hybrids . Restart of replication forks is especially crucial when DNA replication proceeds unidirectionally , i . e . at the telomere or in the rDNA , because converging replisomes do not exist to complete replication . In this regard , it is interesting that the SMC5/6 complex , the negative regulator of Mph1 , has been proposed to play a particularly important role at sites of unidirectional replication [36] . We report that mutants of the SMC5/6 complex show a synthetic growth defect with rnh201 and rnh202 , but not with rnh1 mutants . The synthetic growth defect can be rescued by either impairing homologous recombination or by deleting MPH1 . Interestingly , Rnh202 , as well as its human homolog , are non-catalytic subunits of RNase H2 that contain a conserved PIP-box motif ( a PCNA-interacting motif ) that may account for RNase H activity at the replication fork in human cells [60] . RNase H1 , in contrast , was not found to localize to replication foci during unperturbed replication . It is therefore hypothesized that when the replicative helicase encounters an RNA-DNA hybrid , PCNA provides a platform to recruit RNase H2 to the replication fork via the PIP-box motif . However , for the yeast RNase H2 PIP-box mutant no phenotype has been observed so far [43] . Moreover , we could rescue the growth defect of smc6-9 rnh202 mutants by introducing the PIP-box defective RNase H2 allele ( S5G Fig ) . This suggests that in yeast , RNase H2 may be recruited to the replication and repair machinery via a PCNA-independent mechanism . Mph1 is able to promote helicase-dependent replication fork reversal in vitro [29 , 61] . More recently the Smc5/6 complex has been implicated in specifically inhibiting the fork reversal activity of Mph1 in vitro [29] , while not perturbing its D-loop remodeling capacities . The genetic data presented here , i . e . mutations in the inhibitory Smc5/6 complex being toxic in RNase H2 mutants and that this growth defect can be rescued by deleting MPH1 , underlines the crucial regulation of Mph1 at sites of RNA-DNA hybrids , probably in the context of a replication fork . These results , together with previously published data have allowed us to propose a working model with respect to how Mph1 could act at replication forks stalled by RNA-DNA hybrids ( Fig 6 ) . When the replication fork encounters unresolved RNA-DNA hybrids , replication stress and fork stalling ensue ( Fig 6-1 ) . Either the accumulation of ssDNA at stalled forks or the ssDNA on the displaced strand of the R-loop may lead to the recruitment of Mph1 , which can interact with the ssDNA binding complex , RPA ( Fig 6-2 ) . This is in line with the recent observation that RPA is implicated in R-loop metabolism in vivo and subsequent stimulation of RNase H1 activity in vitro [62] . Consistently , we observed an increase in the number of cells with Mph1 foci when RNA-DNA hybrids accumulated in rnh1 rnh201 and hpr1 mutants . Moreover , shortened telomeres , which accumulate RNA-DNA hybrids , associated more frequently with Mph1 foci in an RNase H sensitive manner ( Fig 4C–4E ) . We hypothesize that the recruited Mph1 then remodels replication forks stalled at RNA-DNA hybrids to promote the subsequent removal of R-loops; perhaps via RNase H activity ( Fig 6-3 ) . The interaction with RNase H is supported by the genetic interactions between loss of MPH1 and RNase H functions , e . g . epistatic interactions in terms of Rad52-GFP foci ( Fig 1A ) and R-loop accumulation ( Fig 3 ) . Finally , the Mph1-dependent synthetic lethality between smc6-9 mutants and rnh201 support the notion that the Smc5/6 complex must be present to ensure that Mph1 does not create toxic homologous recombination ( Fig 5D ) intermediates , possibly due to uncontrolled Mph1 fork reversal at R-loops ( Fig 6-3 ) . Recently , it has been shown that FANCM , the Mph1 homolog in higher eukaryotes , is able to branch migrate and thereby unwind R-loop structures in vitro [27] . Moreover , chicken DT40 cells lacking a functional FANCM translocase domain accumulate RNA-DNA hybrids [27] . Although we cannot exclude that Mph1 may directly unwind R-loop structures in vivo , we favor a role for Mph1 at the replication fork , based on its biochemical properties and our genetic studies which implicate a replication role ( Fig 1C , 1E and 1F ) . Indeed , Mph1 recruitment to telomeres requires its C-terminal domain ( S4B Fig ) , which has been shown to interact with RPA and Mte1 . RPA specifically binds to ssDNA that has been unwound at the replication fork and Mte1 binds to branched DNA structures and recruits Mph1 to foci in S/G2 phase [63] . However , unlike deletion of MPH1 , deletion of MTE1 in smc6-9 rnh201 cells did not alleviate the growth defects of the double mutant ( S4C Fig ) . Thus , we conclude that Mte1 does not promote Mph1’s toxic activity at RNA-DNA hybrids in the absence of the Smc5/6 complex . Secondly , we hypothesize that if Mph1 were to unwind RNA-DNA hybrids in a more general manner , we would expect Mph1 to also be essential in sen1-1 and THO mutants , but we did not see such genetic interactions . Finally , we could only detect Mph1 foci in S/G2 phase cells and not in G1 cells , suggesting that Mph1 is acting at replication forks and not at RNA-DNA hybrids per se ( Fig 1C ) . Moreover , the S-Rnh1 allele alleviated the synthetic growth defect of rnh1 rnh201 mph1 mutants ( Fig 1F ) . Therefore , Mph1 appears to work as a double-edged sword at RNA-DNA hybrids . Whereas Mph1 gets recruited into foci in a hybrid dependent manner and can promote the resolution of R-loops , its helicase activity must be controlled ( by the Smc5/6 complex ) to avoid toxic recombination intermediates . This interpretation is strongly supported by the fact that Mph1 becomes essential in rnh1 rnh201 double mutants , but leads to lethality when RNA-DNA hybrids accumulate ( loss of RNH201 ) in the absence of Smc5/6 function . Interestingly , the depletion of FANCM in human cells leads to replication forks travelling faster over small distances [64] , but not when followed over longer time . This implies more replication fork pausing in the absence of FANCM . FANCM’s ATPase activity was necessary for slowing the replication fork and could therefore be a way of clearing RNA-DNA hybrids at the fork to ensure processive replication . Consistently , when the cells were challenged with CPT , which has been shown to lead to the accumulation of RNA-DNA hybrids [65 , 66] , FANCM was essential for replication fork progression and fork restart [64] . In summary , we have demonstrated that Mph1 and its regulation by Smc5/6 are critical at RNA-DNA hybrids . Further work is needed to determine whether Mph1 leads to fork reversal when the replication machinery is halted at RNA-DNA hybrids , and how Smc5/6 acts to prevent toxicity caused by Mph1 action on R loops . It will also be interesting to further understand how hybrids in different genomic contexts may require specific factors to ensure they are properly processed . This will especially be important in the context of diseases associated with faulty RNA-DNA hybrid processing such as AGS , ALS and AOA2 . Standard procedures for yeast strain construction and cell growth were used [67] . Strains used for microscopy are ADE2 LYS2 trp1-1 derivatives of W1588-4C [68] , a RAD5 derivate of W303-1A ( MATa ade2-1 can1-100 ura3-1 his3-11 , 15 leu2-3 , 112 trp1-1 rad5-535 ) [69] . CFP-tagged Mph1 was generated for expression from its native chromosomal locus with a 4-alanine linker as described [70 , 71] using primers listed in S1 Table and verified by sequencing . The HPR1 gene was deleted with KANMX6 in this strain to construct YBG722 . Other strains are derivatives of BY4741 ( MATa his3Δ0 ura3Δ0 leu2Δ0 met15Δ0 ) . The construction of RNH201 P45D Y219A , was described previously [25] . A complete list of strains and plasmids used in this study are listed in S2 Table and S3 Table , respectively . Yeast cells were grown and processed for fluorescence microscopy as described previously [72] . Fluorophores were cyan fluorescent protein ( CFP , clone W7 ) [73] , yellow fluorescent protein ( YFP , clone 10C ) [74] and red fluorescent protein ( RFP , clone yEmRFP; or mCherry ) [75] . Fluorophores were visualized on a Deltavision Elite microscope ( Applied Precision , Inc ) equipped with a 100x objective lens ( Olympus U-PLAN S-APO , NA 1 . 4 ) , a cooled Evolve 512 EMCCD camera ( Photometrics , Japan ) , and an Insight solid-state illumination source ( Applied Precision , Inc ) . Pictures were processed with Volocity software ( PerkinElmer ) . Images were acquired using softWoRx ( Applied Precision , Inc ) software . Spontaneous Rad52-YFP foci from mid-log growing cells carrying plasmid pWJ1344 were visualized and counted by fluorescence microscopy as described in [76] . For the recombination assays , cells were cultured in SC medium plates and grown for 3 to 4 days . Leu+ recombinants resulting from recombination in LYΔNS and TL-lacZ systems were selected on SC-Leu plates . Recombination frequencies were obtained by fluctuation tests as the median value of six independent colonies isolated from SC medium plates . The final frequency given for each strain and condition is the mean and SD of at least three median values , as described previously [77] . Proteins were extracted and analysed via western blotting according to Klermund et al , 2014 [78] . The Rad53 antibody ( EL7 . E1; gift from M . Foiani ) was used in a 1:16 dilution; the HA antibody ( clone 16B12; covance ) was used in a 1:2000 dilution . Anti-RNR3 is a polyclonal antibody from Agrisera antibodies and was used as a 1:300 dilution . RNA-DNA hybrid analysis ( for Fig 4B ) as well as protein chromatin immunoprecipitations at telomeres were performed as previously published [20] . Mph1-Myc protein was immunoprecipitated with an antibody against the Myc tag ( clone 9B11 , Cell Signaling/ NEB ) . Otherwise the protocol did not differ from the standard ChIP protocol referenced above . The following primers ( for sequence see S1 Table ) were used during the quantitative PCR step: oBL295 and oBL296 ( 500 nM final ) to amplify the 1L telomere , oLK57 and 58 ( 100 nM ) target the 15L telomere and oLK49 and oLK50 ( 300 nM ) are specific for the 6Y’ telomeres . Mid-log cultures grown in YPAD at 30°C or 32°C ( for temperature sensitive strains ) were collected . RNA-DNA hybrids were processed and analyzed as described [47] . For the negative control , half of the DNA was treated with 8μl RNase H ( New England BioLabs ) overnight at 37°C . Quantitative PCR was performed at the indicated regions using the SYBR Green PCR Master Mix ( Biorad ) and a 7500 Fast Real Time PCR System ( Applied Biosystems ) . The relative abundance of RNA-DNA hybrid immunoprecipitated in each region was normalized to the signal obtained in the inputs . Average and standard deviation of at least three independent experiments are shown . Chromosome spread from cells grown to mid-log phase in YPAD were prepared and labeled with the monoclonal antibody S9 . 6 and immunodetected with Cy3- conjugated goat anti-mouse antibody ( Jackson Laboratories , #115-165-003 as described [79] ) . Slides were mounted with 50 μl of VectaShield ( Vector Laboratories , CA ) with 1x DAPI and sealed with nail polish . For each replicate ( n>3 ) , between 150 and 250 nuclei were visualized and manually counted to obtain the fraction with detectable RNA-DNA hybrids . For Mph1-foci quantification , cells with EST2 deleted in the genome were propagated at 25°C in liquid dropout medium lacking uracil to preserve the plasmid pAP81 ( for ectopic expression of EST2 ) and determine foci number prior to telomerase loss for Mph1-YFP , Cdc13-CFP and Rad52-RFP in the strain SS283-23D . Senescence was induced by streaking cells on solid YPD medium and checking for loss of growth on SC-Ura . Selected colonies were inoculated in liquid synthetic complete medium supplemented with 100 μg/ml adenine ( SC+Ade ) and propagated at 25°C for approximately 100 more population doublings . We estimate this procedure involved approximately 30–50 population doublings from the point of colony formation on solid YPD until cells were analyzed by microscopy for the first time point . Samples were collected for monitoring population doubling time by measuring OD600 and for live cell microscopy analysis at regular intervals . Cell cultures were kept at OD600 between 0 . 2 and 0 . 9 through the course of the experiment . The senescence assay for the smc6-9 est2 mph1 mutant experiment was performed as published previously [39] . 15 mL yeast cultures were grown to an OD600 of 0 . 6–0 . 8 and RNA was extracted via the hot phenol method [80] . Reverse transcription and quantitative real time PCR to determine TERRA levels were performed as previously described [81] . For live cell microscopy experiments , the significance of the differences observed among cell populations was determined using one-tailed Fisher’s exact test . P-values with P < 0 . 05 were considered significant . Statistical analysis for the chromatin immunoprecipitation experiments was described previously [20] . Statistical tests ( Student’s t-test and Mann-Whitney U-test ) were calculated using GraphPad Prism software . In general , a P-value < 0 . 05 was considered statistically significant .	DNA damage can either occur exogenously through DNA damaging agents such as UV light and exposure to chemotherapeutics , or endogenously via metabolic , cellular processes . The RNA product of transcription , for example , can engage in the formation of RNA-DNA hybrids . Such RNA-DNA hybrids can impede replication fork progression and cause genomic instability , a hallmark of cancer . The misregulation of RNA-DNA hybrids has also been implicated in several neurological disorders . Recently , it has become evident that RNA-DNA hybrids may also have beneficial roles and therefore , these structures have to be tightly controlled . We found that Mph1 ( mutator phenotype 1 ) , the budding yeast homolog of Fanconi Anemia protein M , counteracts the accumulation of RNA-DNA hybrids . The inactivation of MPH1 results in a severe growth defect when combined with mutations in the well-characterized RNase H enzymes , that degrade the RNA moiety of an RNA-DNA hybrid . Based on the data presented here , we propose a model , where Mph1 itself has to be kept in check by the SMC ( structural maintenance of chromosome ) 5/6 complex at replication forks stalled by RNA-DNA hybrids . Mph1 acts as a double-edged sword , as both its deletion and the inability to control its helicase activity cause DNA damage and growth arrest when RNA-DNA hybrids accumulate .	[ "Abstract", "Introduction", "Results", "Discussion", "Materials", "and", "methods" ]	[ "chromosome", "structure", "and", "function", "nucleases", "molecular", "probe", "techniques", "enzymes", "dna-binding", "proteins", "enzymology", "dna-rna", "hybridization", "dna", "damage", "telomeres", "fungi", "dna", "replication", "molecular", "biology", "techniques"...	2017	The Smc5/6 complex regulates the yeast Mph1 helicase at RNA-DNA hybrid-mediated DNA damage
Metabolic homeostasis in metazoans is regulated by endocrine control of insulin/IGF signaling ( IIS ) activity . Stress and inflammatory signaling pathways—such as Jun-N-terminal Kinase ( JNK ) signaling—repress IIS , curtailing anabolic processes to promote stress tolerance and extend lifespan . While this interaction constitutes an adaptive response that allows managing energy resources under stress conditions , excessive JNK activity in adipose tissue of vertebrates has been found to cause insulin resistance , promoting type II diabetes . Thus , the interaction between JNK and IIS has to be tightly regulated to ensure proper metabolic adaptation to environmental challenges . Here , we identify a new regulatory mechanism by which JNK influences metabolism systemically . We show that JNK signaling is required for metabolic homeostasis in flies and that this function is mediated by the Drosophila Lipocalin family member Neural Lazarillo ( NLaz ) , a homologue of vertebrate Apolipoprotein D ( ApoD ) and Retinol Binding Protein 4 ( RBP4 ) . Lipocalins are emerging as central regulators of peripheral insulin sensitivity and have been implicated in metabolic diseases . NLaz is transcriptionally regulated by JNK signaling and is required for JNK-mediated stress and starvation tolerance . Loss of NLaz function reduces stress resistance and lifespan , while its over-expression represses growth , promotes stress tolerance and extends lifespan—phenotypes that are consistent with reduced IIS activity . Accordingly , we find that NLaz represses IIS activity in larvae and adult flies . Our results show that JNK-NLaz signaling antagonizes IIS and is critical for metabolic adaptation of the organism to environmental challenges . The JNK pathway and Lipocalins are structurally and functionally conserved , suggesting that similar interactions represent an evolutionarily conserved system for the control of metabolic homeostasis . System-wide coordination of cellular energy consumption and storage is crucial to maintain metabolic homeostasis in multicellular organisms . It is becoming increasingly apparent that endocrine mechanisms that are required for this coordination impact the long-term health of adult animals and significantly influence lifespan and environmental stress tolerance [1]–[3] . Insulin/IGF signaling ( IIS ) is central to this regulation , as loss of Insulin signaling activity impairs metabolic homeostasis , but induces stress tolerance and increases lifespan in a variety of model organisms [1]–[3] . Interestingly , environmental stress and cellular damage can systemically repress IIS activity , suggesting the existence of adaptive response mechanisms by which metazoans manage energy resources in times of need [4]–[8] . The mechanism ( s ) and mediators of this endocrine regulatory system are only beginning to be understood [1]–[3] , [9] . Studies in flies and worms have recently identified the stress-responsive Jun-N-terminal Kinase ( JNK ) signaling pathway as an important component of such an adaptive metabolic response to stress . JNK activation , which can be induced by a variety of environmental stressors , including oxidative stress , represses IIS activity , extending lifespan but limiting growth [4] , [10] . Interestingly , similar effects of JNK signaling are also observed in mammals , in which it represses Insulin signal transduction by various mechanisms , including an inhibitory phosphorylation of Ser-307 of the insulin receptor substrate , as well as activation of the transcription factor FoxO [11] , [12] . This inhibition contributes to Insulin resistance and the metabolic syndrome in obese mice [13] , suggesting that chronic inflammatory processes ( which result in activation of JNK signaling ) are central to the etiology of metabolic diseases in obese individuals [9] . Endocrine interactions between Insulin-producing and various Insulin-responsive tissues are likely to coordinate the adaptive metabolic response described above [1]–[3] . JNK-mediated activation of Foxo in Insulin Producing Cells ( IPCs ) of flies , for example , represses the expression of insulin-like peptide 2 ( dilp2 ) , regulating growth and longevity [4] . At the same time , Foxo activation in the fatbody results in lifespan extension , presumably by an endocrine mechanism that feeds back to IPCs [14] , [15] . Adipose tissue is increasingly being recognized as an important regulator of metabolic homeostasis . It secretes a variety of so-called adipokines , including the inflammatory cytokine TNF-alpha [9] . TNF-alpha activates JNK signaling , contributing to JNK-mediated insulin resistance in mouse models for obesity [9] , [13] . JNK activation in adipose tissue further induces expression of IL-6 , which specifically induces Insulin resistance in the liver [16] . While the chronic inhibition of insulin signaling by adipose-derived inflammatory cytokines thus has deleterious effects in obese individuals , it is likely that such endocrine interactions have evolved to govern metabolic homeostasis systemically in an adaptive manner [9] . Supporting this view , adipose tissue is an important regulator of lifespan in worms , flies , and mice , and it is emerging that systemic inhibition of Insulin signaling by adipose-derived factors is involved in this effect [1]–[3] . An endocrine role for adipose tissue in metabolic regulation has further been demonstrated in mice with adipose-specific deletion of the glucose transporter GLUT4 , in which secretion of the Lipocalin family member RBP4 from fat cells induces insulin resistance throughout the organism [4] , [17] , [18] . Such an endocrine system is expected to be adaptive , since it preserves glucose for only the most essential functions during starvation or environmental stress . At the same time , mis-regulation of this system is likely to contribute to metabolic diseases like type II diabetes . Accordingly , increased serum levels of RBP4 are found in obese and diabetic individuals [19] , and polymorphisms in the rbp4 locus are associated with type II diabetes [20] . The Lipocalins are a large family of mostly secreted proteins that bind small hydrophobic ligands [21] , [22] . Lipocalin family members are characterized by a low sequence similarity ( reflecting diversification of biological functions ) , but a highly conserved tertiary protein structure and similar exon/intron structures of their genes [23] , [24] . Recent studies implicate various Lipocalins in the regulation of systemic insulin action and of stress responses [18] , [25]–[28] . Interestingly , the neuroprotective Lipocalin ApoD is strongly induced in aging mice , rhesus macaques and humans , suggesting evolutionarily conserved regulation of this gene [29] , an it induces insulin resistance when overexpressed in the mouse brain [30] . The Drosophila genome contains three Lipocalin genes: NLaz , GLaz , and karl . Unlike the protein Lazarillo in more ancient insect lineages , which is GPI-anchored to the cell membrane of neurons [31] , all Drosophila Lipocalins are secreted extracellular proteins , like ApoD and all other vertebrate Lipocalins . Recent studies have identified an important role for GLaz in stress resistance and lifespan control as well as in the regulation of lipid storage [32] , [33] . While the function of NLaz remains unclear , in situ hybridization in Drosophila embryos shows that it is expressed in a subset of neuronal cells , and , interestingly , in the developing fat body [34] , indicating a potential role in the systemic regulation of metabolism . Here , we show that NLaz transcription is induced by oxidative stress and by JNK signaling in the fatbody , influencing metabolic homeostasis in the fly . Importantly , NLaz induces stress and starvation tolerance downstream of JNK signaling , and negatively regulates Insulin signaling , disrupting glucose homeostasis , repressing growth , and extending lifespan . Our results thus indicate that induction of NLaz mediates the antagonistic interaction between JNK and Insuling signaling in flies , acting as part of a stress response mechanism that adjusts metabolism and growth in response to environmental insults . Based on the ability of JNK signaling to antagonize IIS activity in flies and worms [4] , [10] , and on the starvation tolerance of flies with increased JNK signaling activity [35] , we hypothesized that this pathway plays a role in regulating metabolic homeostasis under physiological conditions . To start characterizing such a role , we analyzed the maintenance of nutrient stores under starvation conditions in wild-type flies and in flies mutant for the JNK activating Kinase Hemipterous ( JNKK/Hep ) . Interestingly , males hemizygous for the hep loss-of-function allele hep1 exhibited significantly reduced energy stores ( lipids and carbohydrates ) in ad libitum conditions compared to wild-type control flies , suggesting impaired metabolic homeostasis in these animals ( Figure 1A–C , Figure S1A ) . Accordingly , we found that in hep1 mutants , nutrient stores were rapidly depleted upon starvation . Interestingly , hep1 mutants also exhibited an accelerated and increased gluconeogenic response to starvation ( Figure 1D ) , measured by phosphoenolpyruvate carboxykinase ( PEPCK ) expression [36] , supporting the idea that JNK signaling mutants suffer a rapid decline in available free sugars upon starvation . Consistent with this view , hep1 hemizygotes are significantly more sensitive to starvation than wild-type controls ( Figures 1E and Figure S1B ) . Similarly , flies in which JNK signaling was repressed by ubiquitous over-expression of a dsRNA against the Drosophila JNK Basket ( Bsk ) were sensitive to starvation , confirming a loss of metabolic homeostasis in JNK loss-of-function conditions ( Figure 1F and Figure S1C ) . Interestingly , these findings recapitulate similar observations in IIS gain-of-function conditions , which lead to a decrease in steady-state metabolic stores and starvation sensitivity , as well as IIS loss-of-function conditions , which show the opposite results [37]–[41] . The effects of JNK signaling on metabolism thus support the perceived antagonism between JNK signaling and IIS in the regulation of energy homeostasis [4] , [10] . Since we had previously found that JNK represses dilp2 transcription in IPCs in response to oxidative stress ( [4] and Karpac et al . , submitted ) , we tested whether the starvation sensitivity in hep1 mutants correlates with elevated dilp2 transcription . Surprisingly , we found no difference in dilp2 ( nor dilp3 or dilp5 ) transcript levels in hep1 mutants compared to wild-type controls under ad libitum conditions , and no changes in dilp2 transcription in response to starvation ( Figure S1D , E , and data not shown; dilp2 transcript levels are insensitive to nutritional conditions , see [42] ) . These results suggest that , while JNK regulates dilp2 transcription to regulate systemic responses to stress , other targets of JNK might be mediating the control of metabolic homeostasis . Since mammalian JNK acts in adipose tissue to induce Insulin resistance , we focused on the fatbody as a potential site of action for JNK in flies . To identify potential mediators of JNK-induced metabolic changes , we tested the transcriptional response of a number of candidate genes to JNK activation . We focused on the putative adipokines and secreted regulators of Insulin signaling , dALS , IMP-L2 , GLaz , Karl , and NLaz , since these molecules or their mammalian homologues have been implicated in systemically governing metabolic homeostasis . Activation of JNK was achieved by over-expression of a constitutively active Hep ( Hepact ) in larvae using the TARGET system , which allows heat-inducible expression of UAS-linked transgenes [43] . We expressed Hepact ubiquitously ( using the ubiquitous driver T80-Gal4 ) , or specifically in the fatbody ( using the fatbody driver ppl-Gal4 , expressed both in larval and adult fatbody , [36] and Figure S8 ) , for a short period of time , increasing the likelihood of observing direct transcriptional effects of increased JNK activity . Transcript levels of potential JNK target genes were then assessed by real-time RT-PCR ( Figure 2 and Figure S2A ) . Among the tested molecules , we found that transcription of the Lipocalin NLaz was potently induced in both whole larvae as well as specifically in the fatbody in response to JNK activation , within a timeframe that resembles the induction of puc , a bona fide JNK signaling target gene ( Figure 2 ) . Another Lipocalin , Karl , was also induced by JNK signaling , albeit to a lesser extent ( Figure S2A ) while GLaz was not induced ( Figure S2A ) suggesting differential regulation of Drosophila Lipocalins by JNK . Since JNK can activate the transcription factor Foxo , we also tested whether Foxo regulates NLaz transcription . We found that JNK still induces NLaz in a Foxo mutant background ( Figure S2B ) and that over-expression of constitutively active Foxo ( FoxoTM ) was not sufficient to induce NLaz ( Figure S2C ) . NLaz thus appears to be a Foxo-independent downstream effector of JNK signaling . These results , and the known effects of the NLaz homologues ApoD and Rbp4 in mice [18] , [30] , suggest that NLaz or Karl might act downstream of JNK signaling to regulate metabolic homeostasis . To test this hypothesis , we measured carbohydrate and lipid levels in homozygous NLaz mutant flies ( using the knock-out allele NLazNW5 , derived from the NLazSceI allele [44] ) and isogenic controls . Similar to hep1 mutants , NLaz mutants exhibited reduced stores and rapid starvation-induced decline of glucose , trehalose , glycogen , and triglyceride levels ( Figure 3A–C and Figure S2D ) . Likewise , NLaz mutants showed an accelerated gluconeogenic response ( induction of PEPCK , [36] ) and were sensitive to starvation ( Figure 3D , E ) . Since NLaz is induced in response to starvation ( Figure S2E ) , and is expressed in the fatbody ( Figure 2B and [34] ) , we tested whether NLaz or Karl over-expression in the fatbody would be sufficient to protect the organism from starvation sensitivity . Indeed , we found that expression of NLaz using ppl-Gal4 promotes starvation tolerance ( Figure 3F and Figure S2E , F ) . In males , overexpression of NLaz results in increased glycogen stores while lipid levels decrease slightly and glucose levels remain normal ( Figure S9 ) , suggesting a shift in energy storage from lipids to glycogen . In females , on the other hand , Glycogen , Glucose and Lipids are increased when NLaz is overexpressed , demonstrating increased energy stores , accompanied by hyperglycemia ( Figure S9 ) . Expression of Karl , on the other hand , did not protect against starvation ( Figure S3A ) , suggesting that Drosophila Lipocalins , similar to vertebrate Lipocalins , are functionally specialized [21] , [22] , [30] . These results suggested that JNK-mediated induction of NLaz in the fatbody regulates metabolic homeostasis . Supporting this view , we found that fatbody expression of NLaz was sufficient to restore starvation resistance and Glucose and Triglyceride levels in hep1 mutants ( Figure 4 ) . Fatbody-derived NLaz is likely to be secreted ( see [34] , and Figure S7 ) , serving as systemic regulator of metabolic homeostasis . Supporting such a systemic role , we found that expression of NLaz in other tissues , such as pericardial cells and hemocytes of flies ( using the dorothy-Gal4 driver , dot-Gal4 [45] ) , also protects against starvation ( Figure S3B ) . A localized supply of NLaz is thus sufficient to exert its systemic protective function . JNK signaling promotes oxidative stress resistance in flies and worms . Similarly , reduced IIS activity also leads to stress tolerance , and the known crosstalk between these two pathways indicates that metabolic and stress responses are tightly linked , allowing organisms to balance protective and growth responses in accordance with available resources [1]–[3] , [46] . Based on these studies and on our findings described above , we reasoned that NLaz-mediated metabolic changes can influence stress tolerance of flies . To test this hypothesis , we first assessed whether NLaz transcription would be induced in response to oxidative stress , and found that NLaz expression is indeed elevated in flies exposed to the reactive-oxygen inducing compound Paraquat ( resembling the induction of puc; Figure 5A ) . We further tested stress sensitivity of NLaz mutants , and found that these flies exhibit increased sensitivity to Paraquat compared to wild-type control flies ( Figure 5B ) . Over-expression of NLaz both ubiquitously and in the fatbody , on the other hand , confers resistance to Paraquat as well as hyperoxic conditions , supporting a protective role for NLaz in the fly ( Figure 5C and Figure S4 ) . To test whether this function of NLaz was specific for oxidative stress , we assessed the sensitivity of NLaz over-expressing flies to infection with Enterococcus faecalis . E . faecalis infection is lethal to flies , but increased JNK as well as reduced IIS activity result in increased survival after infection [47] . Interestingly , NLaz over-expression in hemocytes ( which mediate immune responses in flies; [48] , [49] ) was not sufficient to promote defense against infection with E . faecalis ( Figure S4C ) , suggesting that NLaz acts specifically to protect against oxidative stress . Karl expression in hemocytes , on the other hand , is both sufficient and required for defense against E . faecalis infection ( Figure S4D ) . These results , in addition to the differential effect of these two Lipocalins on starvation tolerance suggest that NLaz and Karl control distinct and specific physiological responses downstream of JNK signaling . Stress protection by JNK signaling has been observed in pucE69 mutants , which exhibit elevated JNK signaling throughout the organism [35] , but also in flies over-expressing JNKK/Hep in neuronal tissue exclusively . This suggests that secreted molecules promote stress tolerance downstream of JNK signaling [4] , [10] . One endocrine mechanism by which JNK activity promotes stress tolerance is downregulation of dilp2 expression in IPCs ( [4] and Hull-Thompson et al . , in preparation ) . Interestingly , however , increased oxidative stress tolerance can also be observed when JNKK/Hep is specifically over-expressed in the fatbody ( Figure 5D ) . To test whether NLaz induction might mediate the endocrine effects of fatbody-specific JNK activation , we assessed stress tolerance of NLaz mutant flies in which JNKK/Hep was over-expressed in the fatbody . Remarkably , we found that lack of NLaz completely abolished the ability of JNKK/Hep expression to promote oxidative stress tolerance ( Figure 5D ) . These results indicate that Nlaz induction is an integral component of the JNK-mediated adaptation to environmental stress in Drosophila . Our results thus support a model in which Nlaz is required downstream of JNK signaling to promote metabolic homeostasis and tolerance to certain forms of stress . Since elevated JNK signaling is associated with increased longevity in the fly , we tested the effect of NLaz on lifespan . Consistent with our other findings , flies lacking NLaz function are short-lived relative to isogenic controls ( Figure 5E ) , while over-expression of NLaz with a ubiquitous driver increases lifespan ( Figure 5F and Figure S6 ) . These results further support the view that the effects of NLaz on metabolic homeostasis and stress tolerance are an important adaptive mechanism to preserve energy resources and optimize the fitness of the organism . Interestingly , the phenotypes we observed in flies over-expressing NLaz ( starvation tolerance , higher nutrient stores , oxidative stress resistance and extended lifespan ) are also phenotypes associated with reduced IIS activity in the fly [39]–[41] , suggesting that the effects of NLaz might be mediated by inhibition of IIS . To test this notion , we asked whether NLaz would affect other IIS-regulated processes in the fly . Indeed , we found elevated membrane localization of the reporter for PI3K activity , GFP-PH , in fatbody cells of NLaz mutant larvae , suggesting that Insulin signaling is increased in these cells in the absence of NLaz ( Figure 6A–C; PI3K activity is indicative of Insulin signaling activity in flies and GFP-PH is widely used as a reliable reporter for this activity [50] ) . These larvae exhibited similar , albeit less general metabolic defects as adult NLaz mutants , with unaffected Glucose and Triglyceride levels , but complete loss of Glycogen stores ( Figure 6D ) . To further test whether NLaz influences IIS activity , we measured the expression of selected Foxo target genes in response to inducible NLaz over-expression in larval fatbodies . For this analysis , we selected genes that are established Foxo target genes in Drosophila , namely thor , InR , dnaPolj , dLip4 , hsp22 , and l ( 2 ) efl [4] , [51]–[53] . A subset of these genes , InR , dLip4 and hsp22 showed significant changes in expression after induction of NLaz in the fatbody , indicating reduced IIS activity in these larvae ( Figure 6E ) . Further supporting a role for NLaz in repressing Insulin signaling , increased expression of NLaz in the fatbody also resulted in elevated hemolymph glucose levels in third-instar larvae ( Figure 6F ) , a phenotype that is associated with reduced IIS activity and that is reminiscent of defects in Glucose homeostasis observed in Insulin resistant vertebrates [41] . NLaz gene dose and expression in the fatbody further moderately affected overall size of the animal , resulting in larger animals when NLaz is mutated , and rescue of this overgrowth when NLaz is overexpressed in the fatbody of NLaz mutant flies ( Figure 6G ) . Together , these results suggest that fatbody-specific NLaz expression negatively regulates IIS activity in larvae . To test whether this effect of NLaz might contribute to the stress resistance and long lifespan observed in the adult , we assessed whether NLaz expression would also repress IIS activity in the adult fly . Indeed , induction of NLaz expression resulted in translocation of Foxo into the nucleus of fat body cells , suggesting decreased IIS activity ( Figure 7A ) . This was accompanied by the induction of the Foxo target gene Lip4 , but not other Foxo target genes , suggesting a context-dependent specific regulation of Foxo target genes in this background ( Figure 7B ) . To assess whether fatbody-specific expression of NLaz would affect IIS activity in other tissues , we observed GFP-PH fluorescence in the adult ovary , using GFP-PH localization in nurse cells as indicators for peripheral IIS activity . Indeed , NLaz expression resulted in significantly decreased GFP-PH signal at the cell membrane of nurse cells , further demonstrating negative regulation of IIS activity by secreted NLaz from the fatbody ( Figure 6C ) . The notion that NLaz acts upstream of IIS in metabolic regulation was also supported by the finding that over-expression of NLaz could not further promote starvation resistance of heterozygotes for the chico loss-of-function allele chico1 ( Figure 7D ) . Chico is the Drosophila homologue of Insulin receptor substrates , and chico1 heterozygotes exhibit decreased IIS activity , and are stress and starvation tolerant [38] , [39] . Expression analysis further indicates that NLaz acts downstream of insulin-like peptides , but upstream of chico , as NLaz expression is unaffected in chico1 heterozygous animals ( Figure S5B ) , and transcript levels of the three major insulin-like peptides , dilp2 , dilp3 , and dilp5 , were unaffected by NLaz loss-or gain-of-function conditions , or by over-expression of Hep in the fatbody ( Figure S5C , D ) . The repression of IIS by NLaz is thus not mediated by regulation of dilp transcription , but by downstream events that promote insulin resistance . Taken together , these results support the notion that NLaz represses Insulin signaling systemically both in larvae and in adult flies . Our findings support a role for JNK – mediated NLaz induction in the fatbody as a central part of an adaptive endocrine system that coordinates metabolism in response to environmental stress by regulating insulin sensitivity of peripheral tissues ( Figure 7E ) . Recent studies have highlighted the role of adipose-derived endocrine factors in such adaptive responses [3] , [9] . For example , reducing IIS activity or over-expressing Foxo specifically in adipose tissue leads to lifespan extension and stress tolerance in flies , mice and worms , presumably mediated by systemic repression of IIS [14] , [15] , [54] , [55] . Furthermore , amino acid deprivation of Drosophila fat body cells leads to marked decreases in PI3K activity in wing imaginal discs and in the epidermis [56] . In vertebrates , on the other hand , excessive JNK activation in adipose tissue induces insulin resistance in the periphery , promoting Type II diabetes [16] , [46] , [57] . Our results implicate NLaz as a mediator of such systemic repression of IIS activity by adipose tissue . JNK-mediated repression of IIS in flies is thus not only mediated by its function in IPCs , where it represses dilp2 transcription ( Wang et al . , 2005; Hull-Thompson et al . , in preparation ) , but also by adipose-specific induction of NLaz , which then inhibits IIS activity in insulin target tissues . This dual antagonism of IIS by JNK is intriguing , as it indicates that adaptive regulation of metabolism requires coordinated control of both insulin-like peptide production and peripheral insulin sensitivity ( Figure 7E ) . How the relative contribution of these effects regulates the organism's metabolic homeostasis , stress resistance and lifespan , is an interesting question that will require further investigation . Vertebrate Lipocalins have also been implicated in the modulation of insulin action , and recent studies suggest a protective role of these molecules under diverse stress conditions [18] , [26]–[28] , [30] . This function of Lipocalins thus emerges as an evolutionarily conserved adaptive mechanism , and our work integrates this mechanism into the known antagonism between JNK and IIS . Based on the evolutionary conservation of this antagonism it is tempting to speculate that vertebrate Lipocalins also act as effectors of JNK in the regulation of systemic insulin sensitivity , with important implications for potential therapeutic targeting of these molecules . While generally promoting metabolic homeostasis and stress tolerance , functional specialization of different Lipocalin family members is expected due to their high sequence divergence . Accordingly , our data show that the Lipocalins present in Drosophila differ in regulation and function . While NLaz and GLaz both regulate stress sensitivity , only NLaz was found to be regulated by JNK signaling . Regulation of Karl , on the other hand , does not influence starvation tolerance ( as NLaz does ) , but promotes resistance against infection by E . faecalis . Further investigation of this diversification of Lipocalin function promises to provide important insight into the systemic regulation of adaptation to diverse environmental challenges . Of particular interest will be to assess the role of Karl as a potential regulator of IIS during infection . Infection with Mycobacterium Marinum can result in significant repression of IIS activity , leading to phenotypes similar to wasting disease [58] . It is intriguing to speculate that excessive JNK-induced Karl expression may cause this pathology . In humans , dysregulation of Lipocalins has been correlated with obesity , insulin resistance , and type II diabetes [18] , [25] , [59] . The cause for this mis-regulation of Lipocalin expression remains unclear , however . Our results implicate JNK signaling , which is activated chronically in obese conditions , as a possible cause . The finding that mammalian lipocalin-2 , which impairs insulin action , is induced by the JNK activator TNFalpha , is especially intriguing [26] . Additional studies in vertebrates , as well as in the Drosophila model , will provide further insight into the physiological role of Lipocalins , their regulation by stress signaling , as well as their interaction with Insulin signaling . As Lipocalins are secreted molecules that bind hydrophobic ligands , it is further crucial to identify their physiological ligands in an effort to understand the mechanism ( s ) by which IIS activity is antagonized by Lipocalins . Such insight promises to provide a deeper understanding of the coordination of metabolic adaptation in metazoans as well as of the etiology of diabetes and other metabolic diseases . Fly lines were obtained as follows: OreR and Da-Gal4 from Bloomington stock center; hep1/FM6 , gift from S . Noselli; ppl-Gal4 , [36] , gift from Michael Pankratz; dot-Gal4 , gift from W . X . Li; UAS-Hep , gift from M . Mlodzik; UAS-bskRNAi , from Vienna Drosophila RNAi Center ( Transformant ID 34138 ) . All fly lines used in this paper were tested for Wolbachia infection and found to be negative for Wolbachia . Generation of the NLaz deletion strain ( NLazSceI ) is described in [44] . Isogenic knock-out and control lines were generated by outcrossing NLazSceI knock-out flies into a w1118-CS10 wild type strain . Sister lines containing either the wild type allele of NLaz ( line NLazCNW14 ) or the mutant allele ( line NLazNW5 ) were selected by PCR and subsequent SceI restriction digest . The line w1118-CS10 is a 10 generations outcross of w1118 into the CS background . P-element mediated transformation of w1118 mutant flies was used to generate pUASt-NLaz . The full-length NLaz cDNA was amplified using PCR ( as annotated in flybase ) , and it was inserted into pUASt via ligation into EcoRI and XbaI sites . Three different independent insertion lines of pUASt-NLaz were used in our experiments , producing identical results . Flies were fed a cornmeal and molasses based diet , and were reared at 25°C . For each experiment , care was taken to ensure flies developed at an equal larval density . Starvation experiments were performed by placing flies in empty vials or vials with water-soaked filters , as indicated in figure legends . For glucose , trehalose , glycogen , and triglyceride measurements , cohorts of 10 male flies were weighed prior to homogenization in 100 μL homogenization buffer ( 0 . 01 M KH2PO4 , 1 mM EDTA , pH 7 . 4 ) . Homogenates were spun for 2 min . at 3 , 000 rpm , and the supernatant was collected . 10 μL of homogenate was used in each of the following assays: Glucose: Homogenate was pipetted into 500 μL glucose reagent ( Glucose ( HK ) Assay Kit , Sigma ) . Samples were incubated at room temperature for 15 min . and absorbance was measured at 340 nm versus deionized water . Trehalose: 0 . 5 μL Trehalase from porcine kidney ( Sigma ) was added to homogenate . After 1 hour incubation at 37°C , glucose was measured as above , and the concentration of glucose prior to trehalose digestion was subtracted . Glycogen: 10 μL of starch assay reagent ( Starch assay kit , Sigma ) was mixed with homogenate . Samples were shaken at 60°C for 15 minutes . 10 μL of sample was used in glucose assay already described . Absorbance of glucose prior to digestion with starch assay reagent was subtracted from final absorbance . Triglycerides: Homogenate was added to 500 μL of activated triglyceride reagent ( Liquicolor Triglycerides , Stanbio ) and reaction was incubated at room temperature for 10 minutes . The absorbance was measured at 500 nm relative to activated triglyceride reagent . All metabolic measurements were normalized to fly weight . Flies were starved for 3–6 hours prior to re-feeding with 25 mM or 50 mM paraquat ( Methyl Viologen , Sigma ) in 5% sucrose . Paraquat solution was administered on soaked filters . Flies were kept in the dark once re-fed . For hyperoxia , 20–30 2-day old adult males , were maintained in vials containing standard food within a Plexiglas enclosure of 28×28×24 inches at room temperature ( 22–24°C ) . Oxygen ( 100% ) was passed through the box at a constant rate of 300 ml/min . Survival was assayed daily . Flies were collected within 24 hr of eclosion and were separated by sex at 2–3 days of age in groups of 20 . They were raised at 25°C under a 12 hr∶12 hr light cycle and transferred to fresh food vials every 2–3 days . To ensure identical genetic backgrounds , Da-Gal4 was out-crossed 10 times to w1118 . Similarly , UAS-NLaz4 and UAS-NLaz8 were generated and maintained in the w1118 background , minimizing effects of genetic background variations and hybrid vigor in the progeny of the crosses studied . Remaining effects of genetic background and hybrid vigor can be assessed in lifespan differences between the progenies of the two control crosses ( UAS-NLaz×w1118 and Da-GAL4×w1118 ) . Any GAL4-dependent modification of lifespan observed in the test cross ( Da-GAL4×UAS-NLaz ) , is thus due to the over-expression of the transgene . RNA was prepared from whole flies or larvae using Trizol reagent ( Invitrogen ) according to package instructions . Subsequently , Superscript reverse transcriptase ( Invitrogen ) and oligodT were used to generate cDNA . cDNA , diluted 1∶100 , served as template for real time PCR using SYBR green based detection on a BioRad MyIQ thermal cycler . All reactions were performed in triplicate , and melting curves were examined to ensure single products . Quantification was performed using the “delta-delta Ct” method to normalize to actin5C or rp49 transcript levels and to control genotypes . Average Ct values of technical replicates were used for normalization , and all data presented here are averages and standard-deviations from at least three independent experiments . Primer pairs utilized were as follows: Nuclear fluorescence was used to normalize membrane fluorescence . Fat bodies of 5 individual wandering third instar larvae were imaged using confocal microscopy and average fluorescence of membrane and nuclei was measured using the histogram function of NIH ImageJ . Intensity ratios were calculated for n = 6–10 individual cells from different fatbodies , and overall averages and standard deviations were calculated . Sibling flies were used for comparisons . Flies were weighed on a Mettler Toledo Ultrafine balance as cohorts of 10 flies in pre-weighed eppendorf tubes . Wings were photographed , and size was determined by quantifying the number of pixels within each wing with adobe Photoshop . S2 cells were maintained as adherent cultures at room temperature in Schneider's medium supplemented with 10% FBS . Cells were transfected with the pAHW plasmid ( Murphy Laboratory ) , into which the Lipocalin cDNA had been subcloned using gateway recombination , fusing the protein with a C-terminal 3xHA tag . Regulation of trafficking of the fusion protein by the N-terminal signal sequence was thus maintained . Transfection was performed with the Fugene HD ( Roche ) reagent in a 9∶2 ratio per the manufacturer's instructions . S2 cells were transfected with this construct and allowed to express the protein for 48 hours . The transfection medium was removed , and new medium was conditioned for 6 hours . Cells were separated from the conditioned medium by centrifugation ( 10′ , 5000 rpm ) , and lysed in standard lysis buffer with protease inhibitors . An aliquot of the cell lysate and one from the conditioned medium were assayed for protein concentration ( BCA assay , Pierce ) , and each adjusted to 1 ug/uL . 100 uL of these protein samples were mixed with SDS sample buffer , and denatured by heating to 95°C for 5 minutes . 10 ug of total protein from each sample were run on a denaturing SDS-page gel ( Nupage , Invitrogen ) , and transferred to PVDF membrane following standard protocols . The membrane was then probed for HA reactivity using an HRP-conjugated anti-HA antibody ( Roche ) , and detected using ECL West Dura substrate ( Pierce ) .	Metabolism of multicellular organisms has to adjust to environmental changes . Insulin signaling plays an important role in this regulation . Stress signals can repress Insulin signaling , curtailing growth to promote stress tolerance and extend lifespan . While this interaction allows managing energy resources under stress conditions , excessive JNK activity in adipose tissue of vertebrates has been found to promote type II diabetes . Thus , the interaction between stress and Insulin signaling has to be carefully regulated to ensure proper metabolic adaptation . Here , we identify a new regulatory mechanism by which stress signaling influences metabolism in fruitflies . We show that an evolutionarily conserved secreted protein , Neural Lazarillo ( NLaz ) , is induced in response to stress signals , and that it is required for metabolic regulation . NLaz mutant animals are more sensitive to stress and show significant metabolic deficiencies . Similarly , increased expression of NLaz inhibits growth , but increases stress and starvation tolerance . We show that these functions are mediated by an interaction with the Insulin signaling pathway . Our results show that the regulation of NLaz by stress signals is critical for metabolic adaptation of the organism to environmental challenges . Both the Insulin and JNK signaling mechanisms analyzed here are evolutionarily conserved , suggesting that similar interactions control metabolic adaptation in vertebrates .	[ "Abstract", "Introduction", "Results", "Discussion", "Materials", "and", "Methods" ]	[ "genetics", "and", "genomics/disease", "models", "diabetes", "and", "endocrinology/obesity", "genetics", "and", "genomics/gene", "function", "diabetes", "and", "endocrinology/type", "2", "diabetes" ]	2009	Control of Metabolic Homeostasis by Stress Signaling Is Mediated by the Lipocalin NLaz
Based on previous studies , historical records and risk factors , trachoma was suspected to be endemic in 31 health districts ( HDs ) in Guinea . To facilitate planning for the elimination of trachoma as a public health problem , national trachoma surveys were conducted between 2011 and 2016 to determine the prevalence of trachomatous inflammation—follicular ( TF ) and trachomatous trichiasis ( TT ) in all 31 endemic HDs . A total of 27 cross-sectional surveys were conducted , each using two-stage cluster sampling ( one survey in 2011 covered five HDs ) . Children aged 1–9 years and adults aged ≥15 years were examined for TF and TT , respectively , using the World Health Organization ( WHO ) simplified grading system . Indicators of household access to water , sanitation and hygiene ( WASH ) were also collected . A total of 100 , 051 people from 13 , 725 households of 556 clusters were examined , of whom 44 , 899 were male and 55 , 152 were female . 44 , 209 children aged 1–9-years and 48 , 745 adults aged ≥15 years were examined . The adjusted prevalence of TF varied between 1 . 0% ( 95%CI: 0 . 6–1 . 5% ) to 41 . 8% ( 95%CI: 39 . 4–44 . 2% ) , while the adjusted prevalence of TT ranged from 0 . 0% ( 95%CI: 0 . 0–0 . 2% ) to 2 . 8% ( 95%CI: 2 . 3–3 . 5% ) in the 27 surveys . In all , 18 HDs had a TF prevalence ≥5% in children aged 1–9 years and 21 HDs had a TT prevalence ≥0 . 2% in adults aged ≥15 years . There were an estimated 32 , 737 ( 95% CI: 19 , 986–57 , 811 ) individuals with TT living in surveyed HDs at the time of surveys . Trachoma is a public health problem in Guinea . 18 HDs required intervention with at least one round of mass drug administration and an estimated 32 , 737 persons required TT surgery in the country . The results provided clear evidence for Guinea to plan for national trachoma elimination . Trachoma , caused by ocular infection with Chlamydia trachomatis [1 , 2] , is the leading infectious cause of blindness worldwide . It affects those living in poverty , predominantly in rural areas , where access to water and sanitation is limited [3] . In 1998 , the World Health Assembly adopted Resolution 51 . 11 which targets the global elimination of blinding trachoma [4] . The World Health Organization ( WHO ) endorses the “SAFE” strategy for trachoma elimination: Surgery for trachomatous trichiasis ( TT ) , Antibiotic treatment for infection , Facial cleanliness and Environmental improvement to reduce transmission [5–7] . Prior to the implementation of the SAFE strategy , disease mapping is a critical first stage to determine the distribution of the disease and determine whether public health interventions are needed . The provision of accurate estimates of the prevalence of both active trachoma , mainly trachomatous inflammation–follicular ( TF ) , and TT enables national programs to plan and implement mass drug administration ( MDA ) and surgical services respectively [8] . Guinea is a country in West Africa with an estimated population of 12 million [9] and a total area of 245 , 857 km2 . The country is bordered on the west by the Atlantic Ocean , on the north by Guinea-Bissau and Senegal , on the east by Mali and Côte d'Ivoire , and on the south by Liberia and Sierra Leone . Guinea has a tropical climate of alternating rainy season and dry season of approximately six months each . It has four natural regions , including Lower Guinea or Maritime Guinea , Middle Guinea , Upper Guinea and Forest Guinea . Lower Guinea is a coastal plain which covers 18% of the national territory with a climate characterized by heavy precipitation ranging from 3 , 000 to 4 , 000 mm per year , and high humidity . Middle Guinea is a mountainous area which covers 22% of the national territory , with annual rainfall of 1 , 500 to 2 , 000 mm and a semi temperate climate . Upper Guinea is a region of plateaus and savanna woodland which covers 40% of the country . Precipitation ranges from 1 , 000 to 1 , 500 mm per year , with a dry and hot climate . Finally , Forest Guinea is a group of massifs and covers 20% of the national territory , characterized by an annual rainfall ranging between 2 , 000 and 3 , 000 mm with a humid climate . The country is split into eight administrative regions and 38 health districts ( HDs ) . Trachoma was suspected to be endemic in Guinea in 31 rural HDs , but not in 7 HDs in and around the capital Conakry , according to historical clinical records and a limited number of epidemiological surveys [10–12] . In 2001 an epidemiological survey was conducted in 10 of the 31 suspected HDs: nine in Upper Guinea ( Dabola , Dinguiraye , Faranah , Kankan , Kérouané , Kissidougou , Kouroussa , Mandiana and Siguiri ) and one HD in the forest region ( Beyla ) . The survey showed a prevalence of 33% of active trachoma ( including TF and trachomatous inflammation–intense [TI] ) among children under the age of 10 years , and 2 . 7% of TT among adults aged 15 and above [5 , 10 , 13] . In 2002 , a trachoma rapid assessment [14] was undertaken in five other suspected endemic districts in Middle Guinea ( Gaoual , Koubia , Koundara , Mali and Tougué ) and showed an average TF and TT prevalence of 23 . 0% and 1 . 1% respectively [10] . These survey results showed that the country was indeed endemic for trachoma and warranted major intervention to achieve the objective of trachoma elimination by the year 2020 . In 2011 , Guinea initiated an integrated national program to control and eliminate neglected tropical diseases ( NTDs ) , including trachoma . In order to plan for trachoma elimination activities , up-to-date accurate district-level estimates of trachoma prevalence were needed . From 2011 to 2016 , the national NTD program conducted trachoma baseline surveys . The results of these surveys are presented here along with discussion of their impact on intervention planning for trachoma elimination in Guinea . A series of cross-sectional , two-stage cluster sampling surveys were used to determine the prevalence of TF and TT in 31 HDs in Guinea , as recommended by WHO [8 , 15 , 16] . Surveys were conducted in multiple phases from 2011 to 2016 , according to the schedule of the integrated national NTD program , and used to develop a national map of trachoma prevalence . The timeline for specific districts were as follows: 2011: Baseline surveys were conducted in five HDs in Upper Guinea ( Dabola , Dinguiraye , Faranah , Kissidougou and Kouroussa ) . These five HDs were among the 10 HDs in the original 2001 epidemiological survey and had previously been shown to be highly endemic for trachoma [10] . The 2011 survey was conducted to confirm that prevalence was still high and to collect baseline data in these HDs before starting MDA . These five HDs were surveyed together as a single evaluation unit ( EU ) . 2012–13: Two HDs in 2012 ( Koundara in Middle Guinea and Yomou in Forest Guinea ) and nine HDs in 2013 ( Gaoual , Koubia , Mali and Tougué in Middle Guinea , Kankan , Kérouané , Mandiana and Siguiri in Upper Guinea and Beyla in Forest Guinea ) were surveyed . These 11 HDs were surveyed as 11 separate EUs . 2014–16: 11 HDs in Lower Guinea ( Boffa , Boké , Forécariah , Fria , Kindia and Télimélé ) and Middle Guinea ( Dalaba , Labe , Lélouma , Mamou and Pita ) were surveyed in 2014 . Due to the outbreak of Ebola virus disease , survey of the remaining four HDs in Forest Guinea ( Guéckédou , Macenta , N’zérékoré and Lola ) was delayed and completed between December 2015 and February 2016 . These 15 HDs were surveyed as 15 distinct EUs , using the systems and processes of the Global Trachoma Mapping Project ( GTMP ) [17] . For the 2011 survey , data on the total population of the villages belonging to the five targeted HDs were available from the community directed treatment with ivermectin ( CDTI ) report of 2010 [18] . From 2012 to 2014 , population data were estimated from the 1996 national census [19] , and for the 2016 survey , the population was projected from the 2014 national census [20] . The target population was children aged 1–9 years and adults aged ≥15 years for estimating the prevalence of TF and TT , respectively . Although these age groups were of primary importance of the surveys , all individuals over the age of 1 year living in selected households were screened for signs of trachoma , and data from all examined individuals were recorded . Any person who had resided in a sampled household for at least one month prior to the survey date was considered eligible for inclusion . People who refused to participate or were absent at the time of survey were not replaced . To estimate the EU-level prevalence of TF , a sample size of children aged 1–9 years was calculated for each EU , based on an expected TF prevalence of 10% [16 , 17 , 21] , with a 95% confidence level , an absolute precision of 3% , and a design effect ( to adjust for cluster sampling ) of 2 . 65 . A minimum sample size of 1 , 019 children aged 1–9 years was needed . To allow for 20% non-response , we tried to enroll 1 , 273 children aged 1–9 years in each EU . We did not calculate a sample size for TT prevalence in adults aged ≥15 years; instead , having determined the number of households required to recruit sufficient children to estimate TF prevalence , all consenting adults aged ≥15 years living in the same households were examined to estimate the TT prevalence in each EU . The sampling frames comprised all the villages in each EU and their respective populations . The protocol followed WHO recommendations [16] , with 20 villages and 30 households per village chosen from each EU . However , in certain EUs where there was a low average number of 1–9-year-old children per household , the number of villages to be surveyed in each EU was increased . In the first sampling stage , a list of all the villages was obtained for each EU , and 20 villages were selected systematically through a probability proportional to population size sampling strategy [16] . In the 2011 survey , which covered five HDs as a single EU , a total of 30 villages were purposively selected , with six drawn from each HD . In all cases , only villages with populations ranging from 300 to 5 , 000 inhabitants were eligible for selection . In the second sampling stage , all households from each selected village were listed , and 30 were randomly selected through the compact segment sampling method [14 , 21] . Within selected households , all consenting/assenting individuals over the age of one year were examined for trachoma signs . Heads of households were interviewed on key water , sanitation and hygiene ( WASH ) indicators using paper forms for 2011–2013 surveys and a standard Android phone-based questionnaire [22] developed by GTMP for the 2014–16 surveys . Graders were recruited from the existing pool of Ophthalmic Clinical Officers ( OCOs ) in Guinea . Training was provided to graders on examination of community residents in the sampled households for clinical signs of trachoma . The training of graders in 2014 and 2016 was performed by GTMP-certified grader-trainers . For the recognition of signs of trachoma , slides showing pictures of various forms of trachoma were used in classroom sessions . Classroom training was followed by a field test and certification . All the graders participating in the surveys had obtained a kappa for diagnosing TF of at least 0 . 7 in a formal inter-grader agreement test ( based on a sample of 50 children ) , compared to a GTMP certified grader trainer . Recorders were also trained to operate the Android devices and enter trachoma grading data from each eligible person as well as WASH data from each household visited . The recruitment of recorders was based on their knowledge and skills to operate and manage electronic devices at ease . Each survey team was comprised of one grader , one recorder and one community member from the sampled village ( acting as a guide/translator ) . All the teams conducted surveys district by district and household by household , and they were supervised by one supervisor . All residents in the household were enumerated , including registration of details of their age and gender . Household members aged one year and above , present at the time of survey and willing to participate in the surveys were examined by a grader with a 2 . 5x magnifying loupe and a torch or sunlight . The WHO simplified trachoma grading system was used [23 , 24] . Graders used gloves and alcohol-based hand gel to clean their hands before examining each participant to avoid spreading any potential infections , including Ebola . Prior to the eye examination , facial cleanliness for children aged 1–9 years was observed . Facial cleanliness was defined as the absence of nasal and ocular discharge . Basic treatment using drugs such as 1% tetracycline eye ointment and analgesics were provided in the field for those in need . Patients needing referral to a hospital were referred promptly . Field teams moved from one selected household to the next until all the 30 selected households in the village were surveyed . In the 2011–2013 surveys , data were collected and recorded on paper forms . The data included individual trachoma grading of both eyes of each subject , and household data , such as information on latrine and water sources . The data from 2011 survey were entered into Epi Info software ( CDC , Atlanta , US ) and those from 2012 and 2013 were entered into CS-PRO software ( Census Bureau and ICF Macro , US ) . From 2014 to 2016 , the GTMP data collection tools were used . Briefly , these were electronic data capture forms running on Android smartphones within the GTMP-LINKS application ( Task Force Links/Task Force for Global Health , Decatur , GA , USA ) [17 , 25] . Data collected in the field were transferred securely from the field to a central cloud-based reporting and data management system . Data from all surveys were transferred into SPSS ( IBM , version 23 ) for analysis . Descriptive statistics were used to examine the sample characteristics , the prevalence of trachoma , and the proportion of households with key WASH indicators . The adjusted prevalence of TF and TT was estimated according to the methods described previously [17] , i . e . adjusting the prevalence according to age and gender using the 2014 Guinean national population as the standard population [26] , except that the adjustment was made at the EU level rather than the village level . For TF prevalence in each EU , the proportion of 1–9-year-olds examined was adjusted by weighting the proportion of each 1-year age band examined by the proportion of that age band in the national 1–9-year-old population . For TT prevalence in each EU , the proportion of ≥15-year-olds examined was adjusted by weighting the proportion of each gender-specific 5-year age band examined by the proportion of that gender-specific age band in the national ≥15-year-old population . The 95% confidence intervals ( CIs ) of prevalence estimates were calculated using the Wilson score method without continuity correction [27] . The surveys were part of the routine disease surveillance activities of the national trachoma elimination program of the Ministry of Health ( MoH ) , Guinea . It was a standard public health measure and all procedures followed WHO recommendations . Protocols were approved by the national Ethics Committee of MoH , and , for the 2014–2016 surveys , by the research ethics committee of the London School of Hygiene & Tropical Medicine ( 6319 ) . Prior to examination , verbal informed consent or assent was obtained from all adult participants and for children from the head of the household . Those with active trachoma were provided with 1% tetracycline eye ointment . Table 1 summarizes the characteristics of the survey populations by EU . In total , 13 , 725 households were visited in 556 villages of 31 HDs . A total of 102 , 040 people were enumerated: 46 , 048 males and 55 , 992 females in 31 districts . Among these , a total of 100 , 051 people were examined ( 44 , 899 males and 55 , 152 females ) , representing a participation rate of 98 . 1% ( 97 . 5% for males and 98 . 5% for females ) ( Table 1 ) . The number of children aged 1–9 years examined was 44 , 209 and the number of the adults aged ≥15 years examined was 48 , 745 . The adjusted prevalence of TF in children aged 1–9 years is shown for each EU in Table 2 . There was a wide variation in the TF prevalence among EUs , ranging from 1 . 0% in Yomou district to 41 . 8% in Dabola , Dinguiraye , Faranah , Kissidougou and Kouroussa EU . The HD-level TF prevalence categories and the cluster TF prevalence distribution are shown in Fig 1 . Fourteen EUs ( encompassing eighteen HDs ) showed a TF prevalence above the MDA intervention threshold of 5% . Among these , nine ( 9 ) HDs had a TF prevalence between 5% and 9 . 9% ( Boffa , Boké , Forécariah , Fria , Koundara , Mali , Mamou , Pita and Telimélé in Middle and Lower Guinea ) . Four ( 4 ) HDs had a TF prevalence between 10% and 29 . 9% ( Kankan , Kérouané , Mandiana and Siguiri in Upper Guinea ) . Five ( 5 ) HDs had a TF prevalence of 30% or more ( Dabola , Dinguiraye , Faranah , Kissidougou and Kouroussa , all in Upper Guinea ) . Thirteen of the 31 HDs surveyed had a TF prevalence of less than 5% , including all the HDs of Forest Guinea . As shown in Table 2 , among 27 EUs ( 31 HDs ) , estimated prevalence of TT ranged from 0% ( 95% CI: 0–0 . 2% ) in Guéckédou HD to 2 . 8% ( 95% CI: 2 . 3–3 . 5% ) in Siguiri HD . Twenty-one ( 21 ) HDs had prevalence of TT of ≥ 0 . 2% ( Table 2 and Fig 2 ) . The total TT cases in the 31 HDs are estimated at 32 , 737 ( 95% CI: 19 , 986–57 , 811 cases ) . Dabola , Dinguiraye , Faranah , Kankan , Kissidougou , Kouroussa and Siguiri HDs had most TT cases . The key WASH indicators by EU are shown in Table 3 . Between 58% and 100% of households in the surveyed EUs reported drinking water source located less than 30 minutes or within 1 km walking distance . Tougué HD had the lowest proportion ( 58% ) of households with a drinking water source located less than 30 minutes . An average of 67% households had improved sanitation facilities ( ranging between 17 . 1% in Mali HD and 96% in Kankan HD ) . In the 2013 survey , more than 52% of toilets were poorly maintained ( detail not shown ) . Also in the 2013 survey , the TF prevalence was higher in the population living in households with poorly maintained toilets ( 38 . 9% ) , compared with 24 . 5% for households with well-maintained toilets ( χ2 = 65 . 404 , p<0 . 001 ) . The results of the national trachoma mapping revealed that trachoma was indeed a public health problem in Guinea . Among the 31 HDs suspected to be endemic for trachoma , 18 HDs had TF prevalence estimates above the WHO threshold of ≥5% and required district-level intervention with antibiotic MDA , as well as implementation of the , F and E components of SAFE . The total population in need of at least one round of MDA in these 18 HDs was approximately 5 . 5 million . Among the 18 HDs , nine HDs ( Boffa , Boké , Forécariah , Fria , Koundara , Mali , Mamou , Pita and Télimélé ) had TF prevalence estimates between 5% and 9 . 9% and therefore required implementation of one round of MDA before undertaking impact surveys . Four HDs ( Kankan , Kérouané , Mandiana and Siguiri ) had a TF prevalence of 10% to ≤29 . 9% requiring implementation of at least three rounds of MDA before undertaking impact surveys . Five districts ( Dabola , Dinguiraye , Faranah , Kissidougou and Kouroussa ) had TF prevalence estimates of ≥30% and therefore needed implementation of at least 5 rounds of MDA before impact surveys . The remaining 13 HDs had TF prevalence of below 5% thus implementation of antibiotic MDA is not required . The adjusted prevalence of TT in the 31 surveyed HDs varied from 0 . 0% to 2 . 8% . Twenty-one HDs had TT prevalence of ≥0 . 2% . There were an estimated 32 , 737 persons requiring TT surgical intervention . To eliminate trachoma as a public health problem , the health system has to ensure that the prevalence of TT unknown to the health system is reduced to less than 2 in 1000 people aged ≥15 years . In the 31 HDs , this means that the minimum number of TT patients that should be offered surgery to reach the elimination threshold , ignoring incident cases and mortality in those with TT , was 21 , 507 at the time of survey [8] . In general , HDs in Upper Guinea had the highest trachoma prevalence and the most TT cases requiring surgery in the country ( Figs 1 and 2 ) . The five HDs ( Dabola , Dinguiraye , Faranah , Kissidougou and Kouroussa , ) that showed high prevalence in the 2001 survey [13 , 28] , had this impression confirmed in the 2011 survey results . This part of Guinea borders with the Kayes , Koulikoro and Sikasso regions of Mali where high prevalence of trachoma has also been noted [29 , 30] . The HDs in Middle and Lower Guinea showed low trachoma prevalence , though there were some clusters in which more than 10% of examined children had TF ( Fig 1 ) . The presence here of some village-level TF proportions exceeding 10% should not cause alarm: an exponential distribution of cluster-level proportions is predicted to be present when an infectious disease is disappearing–and such a distribution has a tail with a few high-burden clusters [31] . None of the HDs in the Forest Guinea require antibiotic MDA . This is also consistent with the situation in Guinea’s neighboring countries , such as Cote d’Ivoire , Guinea Bissau , Liberia , Senegal and Sierra Leone , [13 , 32 , 33] . Overall , Upper Guinea is characterized by a dry season , lowest precipitation , and high year-round temperatures . Climatic and environmental factors may have influenced the distribution and prevalence of trachoma in Guinea via a number of mechanisms , including via influence on the distribution , abundance or seasonal activity of Musca sorbens , the principal eye-seeking fly implicated in trachoma transmission [30 , 34–37] . These mapping results provided evidence that has informed Guinea’s national plan to eliminate trachoma as a public health problem by 2020 . Antibiotic MDA started in 2013 in two districts and has since been scaled up to cover , by 2017 , all 18 HDs that require it . In addition to the implementation of MDA , these 18 HDs also required the implementation of the F and E components of SAFE for trachoma elimination . The lack of availability of toilets/latrines for the safe disposal of human faeces was found in Koubia , Koundara , Kissidougou and Mali HDs ( 50% , 46% , 33% and 17% respectively ) . It was noted that in all the 31 evaluated HDs , most water sources were within 1 km of the household . There was not , in the survey results , a clear link between the availability of toilets or the presence of drinking water less than 30 minutes and trachoma prevalence . A recent paper analyzed multi-country trachoma survey data which included 2014–2016 survey data from Guinea , found no apparent threshold between community-level water coverage and TF prevalence [38] . However , it was observed in the 2013 survey that more than half of all latrines were poorly maintained , and potentially , therefore , acting as M . sorbens breeding sites , or at the very least discouraging people from using them for their intended purpose . There were a number of limitations to our surveys . Firstly , although the same sampling methodology was used in all surveys , the indicators collected in different phases of surveys were somewhat inconsistent . Analyses of associations of trachoma with risk factors using data from across the country were therefore not performed . Instead , we have emphasized generation of WHO-recommended indicators for trachoma elimination , to guide national program decision-making . Secondly , the 2011 survey of five highly-endemic HDs was undertaken to confirm the results collected 10 years before , in order to ensure that antibiotic MDA intervention was still justified . These five HDs were not surveyed as individual EUs , but as one EU with six clusters per HD . The villages sampled in this 2011 survey were selected from known highly-endemic villages based on previous survey data , and therefore may have overestimated the true trachoma prevalence for these HDs . We have not reported HD-level estimates for these HDs , around which confidence intervals would be expected to be very wide . We note that no interventions against trachoma had taken place in these HDs between the 2001 and 2011 surveys , and that the EU-level prevalence estimate therefore matched our pre-survey expectations , justifying interventions . Thirdly , the surveys took place over several years . Projected populations in the survey years were used to estimate the number of prevalent TT cases . This may have caused inaccurate estimation of TT backlogs for program purposes . More work is currently being done to refine the ways that programs estimate TT prevalence . In conclusion , these surveys confirmed that trachoma was a public health problem in Guinea , with 18 HDs requiring intervention with at least one round of MDA and an estimated 32 , 737 people with TT requiring surgery . The data provide the evidence base for the Ministry of Health to plan for implementing the full WHO-endorsed SAFE strategy to eliminate trachoma as a public health problem from Guinea . By 2017 , Guinea achieved complete national antibiotic coverage of areas that require it for trachoma elimination purposes , and impact surveys are now being conducted to assess whether the country is on track to achieve the year 2020 goal .	Trachoma is the leading infectious cause of blindness worldwide . The World Health Organization ( WHO ) recommends that endemic countries implement the SAFE strategy ( surgery for trichiasis , antibiotic treatment , facial cleanliness and environmental improvement ) to achieve trachoma elimination by the year 2020 . Trachoma was suspected to be endemic in Guinea in 31 health districts except those in and around the capital Conakry , based on historical records and previous studies . To facilitate planning for the elimination of trachoma as a public health problem , Guinea conducted 27 separate trachoma surveys between 2011 and 2016 to determine the prevalence of trachomatous inflammation—follicular ( TF ) and trachomatous trichiasis ( TT ) in these 31 health districts . The results showed 18 health districts requiring intervention with at least one round of mass drug administration and an estimated 32 , 737 persons requiring TT surgery in the country . These data provided clear evidence for Guinea to plan for national trachoma elimination .	[ "Abstract", "Introduction", "Materials", "and", "methods", "Results", "Discussion" ]	[ "guinea", "antimicrobials", "medicine", "and", "health", "sciences", "european", "union", "drugs", "tropical", "diseases", "geographical", "locations", "microbiology", "surgical", "and", "invasive", "medical", "procedures", "health", "care", "bacterial", "diseases", "re...	2018	Baseline trachoma prevalence in Guinea: Results of national trachoma mapping in 31 health districts
Mendelian disorders are often caused by mutations in genes that are not lethal but induce functional distortions leading to diseases . Here we study the extent of gene duplicates that might compensate genes causing monogenic diseases . We provide evidence for pervasive functional redundancy of human monogenic disease genes ( MDs ) by duplicates by manifesting 1 ) genes involved in human genetic disorders are enriched in duplicates and 2 ) duplicated disease genes tend to have higher functional similarities with their closest paralogs in contrast to duplicated non-disease genes of similar age . We propose that functional compensation by duplication of genes masks the phenotypic effects of deleterious mutations and reduces the probability of purging the defective genes from the human population; this functional compensation could be further enhanced by higher purification selection between disease genes and their duplicates as well as their orthologous counterpart compared to non-disease genes . However , due to the intrinsic expression stochasticity among individuals , the deleterious mutations could still be present as genetic diseases in some subpopulations where the duplicate copies are expressed at low abundances . Consequently the defective genes are linked to genetic disorders while they continue propagating within the population . Our results provide insight into the molecular basis underlying the spreading of duplicated disease genes . Elucidating the molecular basis of human genetic disorders is one of the most important tasks in medical biology . The availability of the human genome sequence [1] , [2] has facilitated the identification of individual disease genes , e . g . in family pedigree analyses [3] as well as genome-wide association studies ( GWAS ) [4] , [5] . Exploring the characteristics of known disease genes and differences from non-disease genes using bioinformatics methods in recent studies has provided , for example , knowledge of their function [6] , evolutionary origin [7] , [8] , selective constraints [9]–[11] and network properties in the protein-protein interaction ( PPI ) network [12]–[14] , and insights into the genetics underpinning human inherited disorders , facilitating in silico identification of novel disease genes [9] , [11] . However , recent studies have revealed some controversial findings related to duplicated genes and no clear explanation has been given so far . For example , the accepted hypothesis was that disease genes tend to be singletons with fewer paralogs [15] since duplication can lead to functional redundancy [16]–[18] and thereby mask the effect of deleterious mutations [15] , [19]; however , disease genes were found surprisingly enriched in duplicates [8] . Moreover , the molecular mechanism by which the duplication statuses of disease genes contribute to their increased presence in the human genome is still unclear . Recently , it has been proposed that the presence of duplicates permits the accumulation of disease-causing mutations , the emergence of disease genes thus would be more likely to associate with duplicates [8] . Here we argue that this line of reasoning does not necessarily predict the enrichment of disease genes in duplicates even when the compensational capacity between duplicates is considered . For example , in duplicates ( i . e . more recent ones ) whose functional redundancy is resilient enough to mask some disease-causing mutations in one of the copies , the proportion of disease genes would be lower compared with that of overall singletons; however , for duplicates ( i . e . older ones ) whose compensation capacity is partial or no longer effective , they would be purged from the human genome at the same rate as singletons; combined together , the overall proportion of disease genes in duplicates would still be lower . Summarizing recent literature , we realized that the duplication-functional redundancy theory alone is perhaps insufficient in explaining the observed enrichment of disease genes in duplicates , and the contribution of additional factors should be explored and taken into consideration . In this work , we sought to provide a clear illustration on the evolutionary forces governing the propagation of disease genes in the human population by surveying exhaustively the characteristics of disease genes and comparing those with non-disease genes . We focused on monogenic disease genes ( MDs ) that have a clear association with and contribution to human genetic disorders , and tried to address the following questions . First , can the enrichment of disease genes in duplicates first revealed by Dickerson et al [8] be reproduced in an updated dataset and what are the contributions of multiple paralogs in multi-gene families ? Second , if disease genes indeed tend to have functional backups , is this supported by evidence showing a higher functional similarity between paralogs of disease genes than paralogs of non-disease genes ? A key factor being , if the functional divergence of disease genes is greater than that of non-disease genes , a lower or comparable proportion of disease genes in duplicates would be expected , mimicking a behavior that of singletons . Due to the divergence of the functional redundancy of duplicated genes [20] stratification of the genes according to their duplication age was necessary , otherwise resulting in false conclusions as shown in [21] . Third , what are the evolutionary factors acting on human disease genes within and/or across species that could contribute further to the functional compensation of duplicated disease genes ? And finally , what are the molecular mechanisms underlying the spreading of disease genes as duplicates or singletons in the human population ? In other words , how could the functional redundancy between duplicates actually increase their likelihood of being disease genes ? Initially , we investigated the duplications of human disease genes . Here , we considered three widely used approaches to detect duplicated genes in the human genome , including those based on simple homology ( FASTA ) , gene family evolution ( TreeFam ) and orthology ( eggNOG v3 ) ( see Methods ) which resulted in similar conclusions for all methods ( Figures S1 and S2 ) . As shown in Figure 1 , we found that 55% monogenic disease genes ( MDs ) were duplicates , a significantly higher fraction than in non-disease genes ( NDs; p = 2×10−8; Fisher's Exact Test ) ; similarly , we found 23% of the duplicates are also MDs , compared to 18% in singletons ( Figure 1B; see also Dataset S1 ) . Since duplicates are often found to be functionally compensating [16] , our results suggest disease genes are enriched in functional backups . Strikingly , we found that the number of paralogs in the same gene family did not have a significant impact on gene disease status ( Figures 1B , S1B and S2B ) , suggesting non-additive functional compensation from multiple gene family members . We next sought to find additional evidence for functional redundancy in duplicated disease genes by comparing with duplicated non-disease genes . Since the functional redundancy between duplicates decreases over time [20] , it is essential to compare duplicates of a similar age . We therefore first divided duplication pairs ( gene-closest paralog ) into distinct groups according to their duplication age , and then divided them into disease gene containing pairs , if at least one gene in a pair is disease-related ( MD-pairs ) , and non-disease gene containing pairs otherwise ( ND-pairs ) ( see Methods ) . Previous studies suggested that disease genes were under purifying selections compared with non-disease genes , by measuring the numbers of nonsynonymous substitutions per nonsynonymous site ( dN ) between human-mouse orthologs [11] . We confirmed these observations in our dataset using one-to-one orthologs between human and mouse , as well as those between human and macaque; the results are shown in Figure 4A and 4B , respectively . Furthermore , we found the selective constraints on disease duplicates are higher than on disease singletons ( genes that do not have homologs in the human genome ) , as shown in Figure 4C and 4D . The higher purifying selection on duplicated disease genes can also be observed within the human genome; as shown in Figure 4E , we found that MD- pairs always have lower dN values than ND-pairs of similar age . In summary , we have made two interesting observations regarding disease genes in duplicates . First , we have shown that human monogenic disease genes tend to frequently have functionally redundant paralogs , by comparing their characteristics to that of non-disease genes , stratifying both categories according to duplication age . Second , duplicates tend to harbor more disease genes than singletons , confirming the observation by an earlier study [8] , but contradicting theoretical expectations . What are possible explanations for these observations ? A possible scenario is that a disease gene and its duplicate are simultaneously required for certain functions; for example , they might be involved in the same protein complex . In this case , the two genes would be highly co-expressed and evolve similarly . However this is unlikely because the so-called “balance hypothesis” – both underexpression and overexpression of protein complex subunits would lower fitness of the host organism – [29] predicts that 1 ) duplicates are rarely involved in protein complexes and 2 ) the two duplicates from a common ancestor are rarely retained by the same complex unless all other members of the complex are also duplicated and the extra copies are also retained; otherwise the protein complex is imbalanced and evolutionarily deleterious [29] . We found that the first held true in MDs as well as NDs in human using a protein complex dataset from [30] , and comparing them with non-disease genes . Disease genes and their closest paralogs are significantly less likely to be involved in the same complexes ( p = 0 . 0002 , Odds Ratio = 0 . 57; Fisher's Exact Test ) . These results are consistent with a previous study in which only one gene out of a pair of duplicates was found to be associated with diseases [8] . A previous study suggested that duplicates associated with whole genome duplications ( WGDs ) are dosage balanced [31] and thus might not abide by the balance hypothesis . However , we found that pairs of WGD duplicates do not have a high likelihood to be in the same complexes compared with pairs of duplicates associated with small scale duplicates ( SSDs ) ( p = 0 . 22 , Fisher's Exact Test ) ; similar results could be obtained ( p = 0 . 63; Fisher's Exact Test ) using protein complex data from a genome-wide experimental survey on soluble proteins in human [32] . Thus , WGD is not a confounding factor for our observation . So how could functional redundancy actually promote the enrichment of disease genes in duplicates ? Here we propose a new model . We argue that functional compensation by duplication of genes would help mask the phenotypic effects of deleterious mutations , as previously suggested , and reduce the probability of purging the defect genes from the human population . The functional compensation could be further enhanced by the higher purifying selection on duplicated disease genes within and between species . However , due to the intrinsic expression stochasticity among individuals [33] , [34] , the deleterious mutations could present as genetic diseases in subpopulations where the duplicate copies express in low abundances . In other words , the corresponding genes would manifest as disease genes , while the mutant allele would remain in the population instead of being removed . This model is illustrated in more details in Figure 5 . Consequently , duplicates would be enriched in disease genes; the enrichment is weak , albeit significant , due to the complexity of gene regulation in the human genome . We obtained 21 , 731protein coding genes and the corresponding protein and coding sequences ( CDS ) from Ensembl [35] version 59 . In cases one gene coding for multiple proteins , the longest protein and the corresponding CDS is chosen as representative . All other gene annotations such as HGNC symbols , NCBI gene IDs and accession numbers were mapped to Ensembl gene identifiers to facilitate data integration . We downloaded the mapping data using Ensembl BioMart . We collected human disease genes from OMIM [36] and two recent literatures [37] , [38] . In each of the sources disease genes were divided into two categories , MDs – those associated with monogenic diseases , and PDs – those associated with polygenic diseases . We assigned genes associated with both types of diseases into the MD group; please note that changing this definition , for example by assigning this type of genes into the PD group did not change our main conclusions ( see Dataset S1 ) . All other genes that are not included in any of the three sources are considered non-disease genes ( NDs ) . We used three approaches to find duplicated genes in the human genome , including methods based simple homology search ( FASTA ) , gene family evolution ( TreeFam [39] ) and orthology ( eggNOG3 [40] using euNOG ) . Using the homology-based method , if two human genes had a bitscore higher than 80 in a FASTA [41] search at protein level , and the aligned region covers at least 50% of the shorter protein , they are considered as duplicates; please consult ref [21] for more information about the chosen cutoffs . Changing the cutoffs , for example by increasing the required proportion of the aligned regions for homology detection did not affect our results; see Figure S1 for more details . In the latter two methods , if a gene family or an orthologous group contains two or more human genes , these genes are duplicates . The numbers of duplicated genes identified by the three methods are 14 , 014 , 14 , 084 and 11 , 853 , respectively . We downloaded all gene families as well as their corresponding phylogenetic trees from TreeFam [39] ver8 . 0 . We excluded gene families that do not contain human genes , or contain genes from less than four different species , resulting in a set of 9 , 643 gene families . For each pair of duplicates in a gene family , we dated the ( putative ) duplication event by comparing the topology of the corresponding gene tree with that of a species tree . To compare with the TreeFam gene trees , we used a species tree downloaded from Ensembl ( http://www . ensembl . org/info/docs/compara; see also Figure S8 ) . As shown in Figure S7 , to date a duplication event of a pair of duplicated genes ( A2 and A3 in this case; see Figure S7A ) , we first located their last common ancestor ( LCA ) on the gene tree , and collected all the genes that are descendent to this LCA ( Figure S7A; in this case A_rat , A_mouse , A2_human and A3_human ) and their corresponding species ( in this case human , mouse and rat ) ; then we mapped these species on to the species tree ( Figure S7B ) and located the corresponding LCA; the age ( divergent time ) of the duplication event was then defined as the total branch length from this LCA to human on the species tree . The trees shown in Figures S7 and S8 were visualized and prepared using online tools , iTol [42] and EvolView [43] . Two rounds of whole genome duplication ( WGDs ) occurred during early chordate evolution [44] , [45] . Duplicated genes for which their duplication events can be dated back to that time are thus likely to associated with WGDs . Using similar criteria to [31] , we were able to identify in total 6 , 560 genes with their most recent duplication ( MRD ) ages dated after the split of human and Ciona intestinalis ( Ascidian ) , and before the split of human and fishes including Takifugu rubripes ( see also Figure S8 ) ; we found this number of WGD associated duplicates remarkably similar to that of [31] although different methods and numbers of species were used . We obtained the expression profiles of human genes in normal tissues from two sources [23] , [24]; we were able to map 12 , 436 and 17 , 553 probe-sets to Ensembl 59 gene IDs for the two expression datasets , respectively . Both datasets generated similar results . Therefore we showed the results based on [23] in the main text; results based on [24] are shown in Figure S3 . We downloaded GO annotations of human gene products from Ensembl BioMart and GO term hierarchy file ‘gene_ontology_ext . obo’ ( format version 1 . 2; Feb 2012 ) from the Gene Ontology database [46] . Genes ( gene products ) without GO annotations were excluded from further analyses . To compare functional redundancy based on semantic similarity of GO terms between any given two genes , we used the Bioconductor package GOSemSim [26]and restricted our analyses on leaf-GO terms in “molecular function” . Due to known biases towards a better annotation for disease genes ( see Results ) , we adopted a normalized version of GOSemSim as the following formula:where ‘x’ is the maximal number of GO terms associated with individual genes in a duplication pair , ‘min’ is the minimal number of GO terms associated with genes , ‘max’ is the maximal number of GO terms associated with genes; ‘+1’ is used to avoid zeros . We collected the protein-protein interaction data from several public databases , including STRING [47] ( version 9 , score> = 0 . 7 ) , HPRD [48] ( June 29 , 2010 ) , DIP [49] ( Feb 28 , 2012 ) , MINT [50] ( Feb 6 , 2012 ) , IntAct [51] ( Feb 7 , 2012 ) , and BioGRID [52] ( version 3 . 1 . 82 ) , and considered only physical bindings . In addition , we also included one experimental dataset [53] and one curated dataset from the literature [54] . In total , we obtained 80 , 202 interactions among 12 , 839 gene products . For each pair of duplicates in the human genome , we used a KaKs_Calculator [55] tool to calculate the dN ( the numbers of nonsynonymous substitutions per nonsynonymous site ) . We also downloaded dN values between human genes and their homologs in mouse and macaque from Ensembl [35] BioMart; we retained entries with “Homology Type” of “apparent_ortholog_one2one” or “ortholog_one2one” . In this study we applied three statistical tests to different types of datasets . 1 ) Fisher's Exact Test . We used it to test whether monogenic disease genes ( MDs ) are more likely to be duplicates compared with non-disease genes ( NDs ) . Since genes can be divided into four groups according to two kinds of classifications ( association with diseases and being duplicates ) , it is suitable to use Fisher's test . 2 ) Wilcoxon Rank Sum Test . We used this test to compare two sets of numerical values ( for example two sets of dN values for MD and ND genes respectively ) and access whether one tends to have higher values than the other; in this study it was often associated with boxplots . 3 ) Hypergeometric Distribution Test . To test whether duplicated MD genes tend to have higher functional redundancy with their most recent duplicates than that of ND genes of similar age , each of the two groups would be further divided into more than 10 age groups . We found in all cases , the majority of the MD groups had higher ( or lower ) mean values than the ND groups of the same age ( for example Figure 2A ) . To check whether such observations were significantly different from random expectation , we applied the Hypergeometric Distribution Test using the following function in R: phyper ( q , m , n , k ) , where m refers the number of cases where the mean values of the MD groups are higher ( or lower ) in the pool , n refers the number of cases where the mean values of the MD groups are lower ( or higher ) in the pool , k refers the number of cases randomly chosen from the pool of m + n , and q refers to the number of cases out of k where the mean values of the MD groups are higher ( or lower ) . In this study we set m = n = k = the number of valid age groups . All tests were performed using R ( http://www . r-project . org/ ) . All raw data and R scripts used in this study are available in Dataset S1 as an archive file; also included in this archive is a detailed instruction for the readers to reproduce our main results , including all the figures , supplementary figures , and statistical tests except Figure 5 , which was plotted manually .	Duplicated genes , as opposed to singletons , are genes that have additional copies in a genome due to evolutionary mechanisms such as whole genome duplication , homologous recombination or retrotransposition events . Duplicates can have similar functions and thus mask the phenotypic consequences when one copy is mutated . Conversely , the corresponding phenotypes would manifest themselves when mutations occur in singletons , since functional compensation is rare among non-duplicated genes . It would thus be expected that the primary source of monogenic diseases , diseases caused by mutations within a single gene , is singletons . However , the opposite was found to be true . Additionally , we found the functional similarity of duplicated disease genes to be greater than that of duplicated non-disease genes of an equivalent duplication age . So how could the stronger functional compensation among duplicates increase their likelihood to associate with diseases ? We propose that due to functional compensation in duplicates , disease-causing mutations are less likely to be removed from a human population in large scale since the phenotypes are masked; however , the functional compensation could be lost in a subpopulation , perhaps due to intrinsic variation in gene expression , and therefore lead to diseases . As a result , the duplicated disease genes are linked to genetic diseases , yet they continue to spread within the human population .	[ "Abstract", "Introduction", "Results", "Discussion", "Materials", "and", "Methods" ]	[ "genetics", "molecular", "genetics", "biology", "computational", "biology", "evolutionary", "biology", "genetics", "of", "disease", "evolutionary", "genetics" ]	2013	Human Monogenic Disease Genes Have Frequently Functionally Redundant Paralogs
Retinal vascular caliber provides information about the structure and health of the microvascular system and is associated with cardiovascular and cerebrovascular diseases . Compared to European Americans , African Americans tend to have wider retinal arteriolar and venular caliber , even after controlling for cardiovascular risk factors . This has suggested the hypothesis that differences in genetic background may contribute to racial/ethnic differences in retinal vascular caliber . Using 1 , 365 ancestry-informative SNPs , we estimated the percentage of African ancestry ( PAA ) and conducted genome-wide admixture mapping scans in 1 , 737 African Americans from the Atherosclerosis Risk in Communities ( ARIC ) study . Central retinal artery equivalent ( CRAE ) and central retinal vein equivalent ( CRVE ) representing summary measures of retinal arteriolar and venular caliber , respectively , were measured from retinal photographs . PAA was significantly correlated with CRVE ( ρ = 0 . 071 , P = 0 . 003 ) , but not CRAE ( ρ = 0 . 032 , P = 0 . 182 ) . Using admixture mapping , we did not detect significant admixture association with either CRAE ( genome-wide score = −0 . 73 ) or CRVE ( genome-wide score = −0 . 69 ) . An a priori subgroup analysis among hypertensive individuals detected a genome-wide significant association of CRVE with greater African ancestry at chromosome 6p21 . 1 ( genome-wide score = 2 . 31 , locus-specific LOD = 5 . 47 ) . Each additional copy of an African ancestral allele at the 6p21 . 1 peak was associated with an average increase in CRVE of 6 . 14 µm in the hypertensives , but had no significant effects in the non-hypertensives ( P for heterogeneity <0 . 001 ) . Further mapping in the 6p21 . 1 region may uncover novel genetic variants affecting retinal vascular caliber and further insights into the interaction between genetic effects of the microvascular system and hypertension . Changes in retinal vascular caliber provide unique information regarding the structure and state of the microvasculature in the eye , possibly reflecting pathophysiological processes in the microvascular systems elsewhere in the body . Narrowed retinal arteriolar caliber has been known to be predictive of hypertension [1] , [2] and coronary heart disease [3] , while wider retinal venular caliber is associated with higher blood pressure [4] , [5] , impaired fasting glucose , diabetes , dyslipidemia [6]–[8] , and risk of coronary heart disease [9] . In particular , because retinal vessels share embryological , anatomical and physiologic characteristics with cerebral vessels [10] , wider retinal venular caliber bas been closely linked to both subclinical and clinical cerebrovascular diseases , including lacunar infarction , white matter lesions , clinical stroke [9] , [11] , [12] and cerebral hypoxia [13] . Retinal vascular caliber has been observed to vary between racial/ethnic groups . In the Multi-Ethnic Study of Atherosclerosis , African Americans and Hispanics had wider retinal arteriolar and venular caliber compared to Whites and Asian Americans , even after controlling for cardiovascular risk factors [8] . In the Atherosclerosis Risk in Communities ( ARIC ) Study and Cardiovascular Health Study , African Americans had larger retinal arteriolar and venular calibers than European Americans while controlling for age , gender and mean arterial blood pressure [Wong TY , unpublished data , 2009] . The underlying reasons for this racial/ethnic difference are unclear and might be related to systemic , environmental , and measurement factors [14] , [15] . However , several lines of evidence provide support for genetic factors also being involved in the regulation of retinal vascular caliber . The heritability of retinal arteriolar and venular caliber was estimated to be 0 . 48 and 0 . 54 , respectively , in the Beaver Dam Eye Study [16] . Results from two twin studies also showed retinal vascular caliber may be primarily determined by genetic influence with the heritability of 0 . 57–0 . 70 for arteriolar caliber and 0 . 62–0 . 83 for venular caliber [17] , [18] . The observed racial/ethnic differences in retinal vascular caliber could not be fully explained by systemic and environmental factors alone , which prompted our hypothesis that differences in genetic background may partially account for differences in retinal vascular caliber across racial/ethnic populations . To identify chromosomal regions which may harbor genes that modulate retinal vascular caliber , we utilized admixture mapping , a technique that scans the genomes in recently admixed populations , such as African Americans , for regions which may contain variants that not only differ in frequencies between the two genetically diverse ancestral populations ( Europeans and West Africans in this case ) , but can also partially explain differences in phenotypes between populations [19]–[24] . Since the identification of ancestry-informative markers and the development of appropriate statistical methods for admixture analysis in African Americans [23] , admixture mapping and subsequent fine-mapping studies have has been successful in identifying determinant genetic variants for prostate cancer [25] , [26] , end stage renal disease ( ESRD ) [27] , white blood cell count [28] , [29] , and the circulating levels of interleukin 6 and interleukin 6 soluble receptor [30] . Differences in retinal vascular caliber between African Americans and European Americans make this an ideal phenotype to study with the admixture mapping approach . The main hypothesis of the present study was that some alleles affecting retinal vascular caliber are present at higher frequency in Africans than Europeans . We thus conducted a genome-wide admixture mapping scan for retinal arteriolar and venular caliber using approximately 1 , 365 ancestry informative markers in self-identified blacks from the ARIC study . In addition , hypertension is known as one of the most important risk factors for cardiovascular and cerebrovascular diseases . Previous studies indicated the importance of interactive effects between genes and vascular risk factors , in particular hypertension , on the occurrence of cardiovascular diseases [31]–[33] and cerebrovascular disorders [34]–[36] . It has been suggested that studies investigating the role of genetic components in vascular diseases would be more effective by analyzing interaction of genetic effects with conventional risk factors [35] . Therefore , a secondary purpose of our study was to test the hypothesis that the effect of genetic variants that conferred differences in retinal vascular caliber in African Americans was modified by hypertension . The correlations between retinal vascular caliber and PAA are shown on Figure 1 . CRVE was significantly correlated with PAA ( correlation coefficient , ρ = 0 . 071 , P = 0 . 003 ) , while the correlation between PAA and CRAE was weaker and not statistically significant ( ρ = 0 . 032 , P = 0 . 182 ) . To assess the potential interaction between genetic effects and hypertension , we examined the correlations limited to hypertensive individuals as an a priori subgroup analysis . We found that among hypertensive individuals , the correlations with PAA became stronger for both CRAE ( ρ = 0 . 084 , P = 0 . 008 ) and CRVE ( ρ = 0 . 094 , P = 0 . 003 ) . We carried out genome-wide admixture mapping scans for both CRAE and CRVE using case-only and case-control approaches , using up to 1 , 365 ancestry-informative SNP markers . Cases ( N = 261 ) and controls ( N = 260 ) for CRAE were defined as the extreme 15% of samples with narrowest caliber and the 15% with widest caliber , respectively , after adjustment for age , sex , study site , 6-year mean blood pressure , and fasting glucose level . For CRVE , the extreme 15% of samples with widest caliber was defined as cases ( N = 260 ) , while the extreme 15% with narrowest caliber as controls ( N = 261 ) . The mean CRAE was 46 . 2 µm ( ±15 . 3 ) narrower in cases compared to controls . On the other hand , the mean CRVE was 51 . 0 µm ( ±16 . 1 ) wider in cases compared to controls . Using 18 pre-specified European ancestry risk models , we did not detect significant admixture association with either CRAE or CRVE ( Table S1 ) . The genome-wide score , derived by averaging the evidence of association across all loci examined in the genome , was −0 . 73 for CRAE and −0 . 69 for CRVE ( Table 2 ) , which did not meet the thresholds of >2 for genome-wide significance [37] . To examine whether hypertension modified the effects of genetic ancestry on retinal vascular caliber , we performed admixture mapping scans by hypertension status . Cases and controls were defined as the extreme 15% ( after adjustment for the covariates as described above ) in the hypertensive subset and included about 150 subjects in each group . On average , CRAE was 46 . 5 µm ( ±15 . 5 ) narrower and CRVE was 51 . 6 µm ( ±16 . 7 ) wider in the cases , compared to the controls . We found genome-wide significant evidence of associations with CRVE in the hypertensive subset ( Table 2 ) . The genome-wide score in the case-only analysis was 2 . 31 , which meets our threshold of >2 for significance . The strongest admixture association for CRVE was observed at 6p21 . 1 ( 42 . 5 Mb on chromosome 6 in build 35 of the human genome reference sequence; Figure 2 ) , with the peak locus-specific LOD of 5 . 47 , again reaching our priori defined thresholds of >5 for significance [37] . To further correct for the multiple hypothesis testing in two subgroups , we divided our test statistic , which is the likelihood ratio that compares the model under a risk model to the null model by 2 . The likelihood ratio was 105 . 47 = 295120 . 9 . The likelihood ratio after corrected for two hypothesis testing was 295120 . 9/2 = 147560 . 5 , corresponding to a LOD score of 5 . 17 , which still exceeds the threshold of >5 for significance . No other locus exceeded a LOD score of 5 . To evaluate the stability of our results , we carried out a longer analysis with 10-times more iterations in our Markov Chain Monte Carlo run . We obtained a similar strength of signal with a genome-wide score of 2 . 28 and a peak locus-specific LOD of 5 . 43 at the same location . The risk model with the strongest score corresponded to a risk of 0 . 5 due to one copy of an European ancestral allele with the inverse risk for carrying zero copies ( see Methods for the set of risk models ) . Further refining the risk models , we obtain a genome-wide score of 3 . 66 and a locus-specific LOD of 6 . 85 in this region . The admixture-generated signal for CRVE in the hypertensive subset was further supported by the case-control analysis . At the 6p21 . 1 peak , the cases had a highly statistically significant increase in African ancestry compared to controls ( case-control Z score = −5 . 26 , P = 1 . 44×10−7; Figure 2 ) . The association was nominally genome-wide significant ( P = 2 . 88×10−4 ) after conservatively correcting for multiple hypothesis testing ( by multiplying by 2 , 000 because we tested 1 , 000 independent chromosomal chunks in two subgroups ) . We did not find any significant associations with CRAE in the hypertensive subset ( genome-wide score = 0 . 18 ) . The highest locus-specific LOD was 2 . 37 , arising from chromosome 5 , followed by the second highest LOD of 2 . 04 on chromosome 6 ( Figure S1 ) . Admixture scans were also carried out in the non-hypertensive subset , but we did not observe any evidence of association with either CRAE ( genome-wide score = −0 . 40 ) or CRVE ( genome-wide score = −0 . 21; Table 2 ) . To further assess the robustness of the admixture-generated signal , we extracted the local estimate of African ancestry at the 6p21 . 1peak , and tested for association with continuous CRVE by hypertension status . This enabled us to increase power by including all samples in a quantitative analysis . We carried out a series of linear regression analysis , with the normal-quantile transformed CRVE ( after adjusted for covariates as described above ) as a dependent variable and the local ancestry as an independent variable . As shown in Table 3 , local African ancestry alone was strongly associated with the transformed CRVE in the hypertensive subset ( P = 2 . 9×10−8; Model 1 ) . To assess whether continuous CRVE was associated with local ancestry at 6p21 . 1 due to their associations with global ancestry ( i . e . , PAA ) , we modeled the transformed CRVE as a function of local , global and regional ancestry . We found that the residual association of local ancestry with the transformed CRVE after adjustment for both global and regional ancestry remained significant ( P = 3 . 9×10−6 ) , indicating that there may be a gene in the region of 6p21 . 1 that is associated with CRVE above and beyond the fact that variants in this locus are highly differentiated between ancestral populations and thus correlated with global ancestry . The association was nominally genome-wide significant ( P = 7 . 8×10−3 ) after correcting for multiple hypotheses tested ( by multiplying by 2 , 000 ) . The results were similar when CRVE was additionally adjusted for other covariates ( Model 2 ) , including high-density lipoprotein ( HDL ) cholesterol , low-density lipoprotein ( LDL ) cholesterol and plasma triglyceride levels , body mass index ( BMI ) , smoking , and alcohol consumption . We estimated that each additional copy of an African ancestral allele at the 6p21 . 1 peak was associated with a CRVE increase of 0 . 37 Z-score units on average ( equivalent to ∼6 . 14 µm ) . In contrast , in the non-hypertensive subset , the local ancestry effect at the 6p21 . 1 peak on CRVE was weak and did not reach significance ( Table 3 ) . The effect of local ancestry showed significant heterogeneity ( P<0 . 001 ) between the hypertensive and non-hypertensive groups , which was in line with the results of the above dichotomous admixture scans . Given the significant statistical evidence for peak at 6p21 . 1 , we constructed 95% credible interval ( CI ) for the loci identified . The 95% CI spanned from 40 . 8 to 43 . 9 Mb on build 35 of the human genome reference sequence ( Figure S2 ) . The locus-specific LOD score and the association of the CRVE to local ancestry for the SNPs located near and within the 95% CI are presented in Table S2 . We used admixture mapping to search for genomic regions that may account for inter-individual variations in retinal vascular caliber . We found evidence for association with retinal venular caliber at 6p21 . 1 in hypertensive African Americans and observed concordant results when venular caliber was examined as a continuous variable , with higher levels of African ancestry being significantly associated with wider retinal venular caliber . The significant evidence of association with the local ancestry at the 6p21 . 1 peak was above and beyond the contribution of both global and regional ancestry . Methodologically , these results are interesting in that subset analysis was required in order to detect the association . We note that subsets analysis has previously been very successful in admixture mapping . For prostate cancer , the chromosome 8q24 locus was not detected until the analysis was limited to a subset of individuals with a younger age at diagnosis [25] . For ESRD , the admixture signal to be much stronger among non-diabetic ESRD ( mainly hypertensive ESRD ) only , compared to diabetic ESRD [27] . Subsequent fine mapping identified genetic variants strongly associated with both prostate cancer [26] and non-diabetic ESRD [27] . We hope to follow up the present analysis with successful fine-mapping as well . Our findings in persons with hypertension in the subgroup analysis imply that genes associated with hypertension may have exerted their effects on retinal vascular caliber . While we cannot exclude the possibility of a chance finding , we were able to show consistent results in the local ancestry analysis , which included the total population . Furthermore , from a physiological perspective , our finding that there is a genetic association with retinal venular caliber specifically in people with hypertension is sensible . Hypertension is known to have profound effects on the retinal microcirculation [38]–[40] , and may induce gene expression relevant to the modulation of retinal vessels ( see further discussion below ) [41] . Our findings are also in line with previous studies in other vascular diseases that indicated hypertension exaggerates the effects from genetic factors [31]–[36] . In hypertensive persons , we detected significant genetic association for retinal venular caliber , but not arteriolar caliber . Because retinal arterioles and venules likely possess distinct genetic determinants [42] , there may be no common genetic variants with a strong effect accounting for differences in retinal arteriolar caliber between European and African populations . Moreover , a diminished capability of retinal arterioles to remodel because of progressive sclerosis and rigidity of arteriolar vessel walls with age [43] , may have precluded a degree of change in arteriolar caliber equal to that observed in venular caliber . To our knowledge , there have been only two prior studies examining the genetic basis of retinal vascular caliber with a genome-wide approach , and both of them used linkage analysis [18] , [45] . By genotyping 385 microsatellite markers in the Beaver Dam Eye Study , Xing et al . [45] found several loci for CRAE and CRVE , with the most significant loci at 3q28 ( empirical P = 1 . 2×10−4 ) and 8q21 ( empirical P = 2 . 9×10−3 ) , respectively . A recent linkage analysis in the Australian Twins Eye Study identified 8p23 . 1 ( LOD = 2 . 24 ) and 2p14 ( LOD = 2 . 69 ) as suggestive loci for CRAE and CRVE , respectively [18] . Although findings of the two linkage analyses were not replicated in our study , possibly due to differences in study design and populations , all three studies provide lines of evidence that structural changes in the microvasculature of retina may have genetic determinants . One major concern of the present study is a potential measurement error on retinal vascular caliber itself , which has been suggested to account for some of the observed racial/ethnic difference in retinal caliber [15] . A recent report suggested that retinal pigmentation could be a source of error in the computer-assisted methods to measure retinal vascular caliber from photographs [15] . The study reported that arteriolar and venular calibers were significantly wider in East Asian children than their white counterparts . However , when the analysis was confined to children with dark brown iris ( a surrogate of retinal pigmentation ) only , the differences between racial groups were less pronounced . Nevertheless , measurement errors are less likely to bias results in the present study for the following reasons . First , in the ARIC study , the retinal vessel edge was not detected based on computer-generated pixel density curve , but located manually by graders [46] . We believe the manual grading of retinal vascular caliber would be less prone to bias than computerized grading schemes , yet this remains to be proved . Second , genetic ancestry was shown to be significantly correlated with human pigmentation [47] . Although we did not measure skin or retinal pigmentation in the ARIC study , the inclusion of global ancestry as a covariate in our local ancestry analysis ( Table 3 ) may thus provide an alternative way to adjust for the differences in retinal pigmentation . The 95% CI for the 6p21 . 1 locus , a ∼3 . 1 Mb region , contains genes that may have biological relevance to the development and modulation and of retinal vessels ( Figure S2 ) . One such gene is the vascular endothelial growth factor ( VEGF ) gene . VEGF is an endothelial-cell selective mitogen intimately associated with vasculogenesis , angiogenesis and vascular permeability . In the retina , VEGF plays a crucial role in the induction of retinal vasculature and its expression is regulated by hypoxia during embryonic development [48] . Moreover , animal experiments showed that VEGF acts as a survival factor for newly formed retinal vessels [49] , and it continues to be produced by retinal astrocytes and pericytes in the vicinity of retinal vessels in adults [50] . Interestingly , mechanical stretch on retinal vessel endothelium induced by systemic hypertension could increase the expression of VEGF and its receptor [41] . Further mapping work is needed to determine whether variants in the VEGF gene indeed contribute to variations in retinal vascular caliber . If proven , this may help explain why we detected the association at 6p21 . 1 only in hypertensive persons . In addition to VEGF , there are two other genes in the 95% CI that may potentially be associated with retinal phenotype: peripherin 2 ( PRPH2/RDS ) and guanylate cyclase activator 1A ( GUCA1A ) , both of which are also expressed in retina [51] , [52] . GUCA1A plays a role in the recovery of retinal photoreceptors from photobleaching [52] . PRPH2/RDS is mainly located in the outer segment of both rod and cone , and defects in this gene are associated with retinal degenerations [51] . VEGF appears to be the strongest candidate gene based on its known function . However , it does not exclude the possibility that the admixture-generated signal is due to other genes . The human leukocyte antigen ( HLA ) loci , located on chromosome 6p21 . 3 and about 10 Mb away from the 95% CI , are gene-rich and highly polymorphic [53] . The HLA region has been shown to play an important role in multiple disease susceptibility , particularly in autoimmune and infectious diseases [54] . HLA alleles are strongly associated with many neighboring SNPs , sometimes located at a considerable distance from the HLA allele with the linkage disequilibrium ( LD ) extending several Mb [55] . It remains to be determined whether the HLA alleles and the alleles in the 95% CI are in LD and may thus be associated with retinal venular caliber . In summary , using a genome-wide admixture mapping scan in 1 , 737 African Americans , we detected a risk locus influencing retinal vascular caliber in hypertensive individuals at 6p21 . 1 , where the association between local ancestry and retinal venular caliber was strong , suggesting the presence of a genetic effect beyond the effects of global ancestry . Follow-up fine mapping or haplotype tagging across the peak will be necessary to determine whether this region harbors genetic variations that may interact with hypertension to modulate retinal venular caliber . Understanding the genetic basis of retinal vascular caliber may provide novel insight into the development and remodeling of the microvasculature in the brain and elsewhere in the body . This study was conducted according to the principles expressed in the Declaration of Helsinki . All sample collections were carried out according to institutionally approved protocols for study of human subjects and written informed consent was obtained from all subjects . Subjects of the present study were from the 2 , 997 African-American participants of the ARIC study at the third examination . The ARIC study is a prospective epidemiologic study that examines clinical and subclinical atherosclerotic disease in a cohort of 15 , 792 persons ( including 4 , 266 African-Americans ) , aged 45 to 64 years at their baseline examination . Participants were selected by probability sampling from four U . S . communities: Forsyth County , NC ( 12% African American ) ; Jackson , MS ( 100% African American ) ; the northwest suburbs of Minneapolis , MN ( <1% African American ) ; and Washington County , MD ( <1% African American ) . The sampling procedure and methods used in ARIC have been described in detail elsewhere [56] . Participants self-reported their ethnicity . The baseline examination ( visit 1 ) took place from 1987 to 1989 , a second examination ( visit 2 ) from 1990 to 1992 , and a third examination ( visit 3 ) from 1993 to 1995 . Retinal vascular calibers were measured at visit 3 . Data from visit 3 were used for the present analysis . The final sample for the present analysis included 1 , 737 African Americans after excluding the following samples ( N = 1 , 267 ) : 1 ) African-American subjects who lived in Minneapolis , MN , or Washington County , MD , or 2 ) did not consent to genetic studies or did not have DNA samples available , 3 ) samples that were not genotyped successfully or that failed to pass quality control ( see “Elimination of problematic samples” ) , 4 ) subjects who had no retinal photographs , ungradable photographs or retinal vascular occlusions , and 5 ) subjects who had missing data on the covariates used in the main admixture mapping scans ( see “Admixture mapping” ) . The retinal photography procedure and its assessment have been described in detail elsewhere [46] . Briefly , a 45-degree retinal photograph of one randomly selected eye of each participant was taken at visit 3 following 5 minutes of dark adaptation . This photograph was centered on the region of the optic disc and the macula and was taken using an autofocus camera . Trained graders who were masked to participant measured the calibers of all arterioles and venules coursing through a specified area surrounding the optic disc according to a standardized protocol [46] . Individual vessel measurements were combined into summary indices: CRAE and CRVE . These indices represents average retinal arteriolar and venular caliber of the eye after taking into account the branching patterns . These measurements of retinal vascular calibers are reliable , with generally high intragrader and intergrader reliability coefficients ( 0 . 84 and 0 . 79 , respectively ) [46] , [57] . Current blood pressure was defined as measurements at the time of retinal photography at visit 3 , and 3- and 6- year past blood pressure was defined as measurement taken at visit 2 and visit 1 , respectively . Mean arterial pressure was defined as two thirds of diastolic plus one third of systolic blood pressure . Mean arterial pressure across the three visits was averaged to obtain the 6-year mean arterial pressure . BMI was calculated as weight ( in kg ) /height ( in meters ) squared . Blood collection and processing followed a standard protocol [58] . Fasting glucose was assessed by a modified hexokinase/glucose-6-phosphate dehydrogenase procedure . Total plasma cholesterol and triglyceride were measured using enzymatic methods [59] . LDL cholesterol was calculated [60] , and HDL cholesterol was measured after dextran-magnesium precipitation of non-HDL lipoproteins [58] . Cigarette smoking and alcohol consumption were ascertained from a questionnaire interview . Hypertension was defined as current systolic blood pressure ≥140 mm Hg , diastolic pressure ≥90 mm Hg , or self-reported use of medications for high blood pressure during the 2 weeks preceding the clinic examination at visit 3 . We genotyped a total of 1 , 536 SNPs included in the phase 3 admixture panel [28 , 61] . This panel was constructed by using the panel of ancestry informative markers previously published by Smith et al . ( phase 1 panel ) [24] , improving this panel by mining new ancestry informative markers from the data sets of Hinds et al . [62] and the Phase 2 International Haplotype Map [63] , and then validating them to confirm that they were indeed ancestry informative . Genotyping was performed by the Center for Inherited Disease Research ( CIDR , Johns Hopkins University , Baltimore ) , using the Illumina BeadLab platform [64] . The ARIC study has a rigorous quality control program , including blind duplicates . Many genotypes in duplicates were obtained using the Illumina BeadLab technologies in ARIC African-American participants , and CEPH and Yoruban samples . The mismatch rate among duplicate genotypes was 0 . 1% . We used previously published genotyping data to estimate the frequency of each SNP in West Africans and Europeans [24] , [25] , [64] . We used only those SNPs for which we were able to obtain data from both West African ( Yoruba ) and European American ( CEU ) populations from the International Haplotype Map . For SNPs in the phase 1 panel , we also added additional genotyping data from African and European samples , which was the same as the data collected in Smith et al . 2004 [24] . After genotyping , samples were eliminated based on the following criteria: 1 ) samples with low ( <94% ) call rate , 2 ) samples showing gender mismatch between self-reported data and genetically estimated gender based on 50 markers on the X chromosome , and 3 ) duplicate samples ( defined as >75% match in the genotypes between two samples . Moreover , we used built-in data checking programs in the ANCESTRYMAP [23] software to exclude samples with an apparent excess or deficiency of heterozygous genotypes compared with the expectation from the individuals' global ancestry . An apparent excess of heterozygous genotypes ( defined as the Z-score >10 ) usually indicates the individuals have parents with divergent ancestries ( for example , one parent who is entirely of European ancestry ) and such individuals nearly always have estimated European ancestry close to 0 . 5 [23] . To decrease the likelihood of false-positives in our admixture scans , we applied a series of filters that had the goal of detecting and removing any SNPs with problematic genotyping , as described previously [24 , 30 , 61 , 66] . First , SNPs were dropped if there were atypical clustering patterns , ill-defined clusters , or relatively low genotyping success rate ( 95% ) . This left us with 1 , 416 SNPs ( all with genotyping call rate >97% ) . We then applied three previously described filters to further eliminate SNPs from the analysis [66] . 1 ) We eliminated SNPs ( N = 15 ) if they did not meet the requirement for Hardy-Weinberg equilibrium ( P>0 . 01 ) in both ancestral West African and European populations . 2 ) We applied a “freqcheck” filter that examined whether the observed frequency of a SNP in African Americans was statistically consistent with being a mixture of the frequencies observed in the West Africans and European American samples that we used to represent the ancestral populations [23] . 3 ) Lastly , we applied a “ldcheck” filter that for each sample , iteratively eliminated SNPs that were less informative ( in terms of the information content about ancestry ) until none were within 200 Kbs of each other or in detectable LD with each other in the ancestral West African or European populations [23] . After imposing these requirements , 1 , 365 SNPs were left for analysis . Using the ANCESTRYMAP software [23] , we estimated a global ancestry for each individual , as indicated by PAA . ANCESTRYMAP uses a Markov Chain Monte Carlo approach to account for uncertainty in the unknown parameters ( including SNP allele frequencies in the West African and European ancestral populations , the number of generations since mixture , and the average proportion of ancestry inherited from ancestral populations ) that emerge from the Hidden Markov Model analysis . All Markov Chain Monte Carlo runs used 100 burn-in and 200 follow-on iterations , as recommended [23] , except for one longer run of 1 , 000 burn-in and 2 , 000 follow-on iterations , which we used to check the stability of our results . We used the ANCESTRYMAP software [23] to search for genomic regions associated with an increased percentage of either European or African ancestry . The main dichotomous admixture scans used the values of retinal vascular caliber adjusted using the following covariates: age , sex , study sites ( Forsyth County or Jackson ) , 6-year mean arterial blood pressure and fasting glucose level , because the latter two systemic factors were both significantly correlated with PAA ( both P<0 . 01 ) and known to be associated with retinal vascular calibers [4] , [7] , [8] . For the tests for associations of the local ancestry , we additionally adjusted for other cardiovascular risk factors . For the purpose of this dichotomous admixture analysis , study participants were ranked by the adjusted values for each trait . For CRAE , the 15% of individuals with the lowest values were defined as cases , and the 15% with the highest values as controls . For CRVE , conversely , the 15% with the highest values for were defined as cases , and the 15% with the lowest values as controls . Because ANCESTRYMAP uses Bayesian statistics , a prior distribution of risk models is required [23] . We tested 18 pre-specified European ancestry risk models to assess overall evidence of association by averaging across all models . The first 6 models used 0 . 4 , 0 . 5 , 0 . 67 , 1 . 5 , 2 . 0 and 2 . 5-fold risks of being a case due to inheritance of one copy of an European ancestral allele , with a risk of 1 for carrying zero copies of an European ancestral allele . The next 6 models used the same risk set as the first for carrying one copy of an European ancestral allele , whereas the risk of carrying zero copies were set to the reciprocal of the risks for carrying one copy . The last 6 models specified that inheritance of either one or two copies of an European ancestral allele had a risk of 1 , but carrying zero copies had risks of 0 . 4 , 0 . 5 , 0 . 67 , 1 . 5 , 2 . 0 and 2 . 5 . By convention used in the manuscript , a risk <1 . 0 for inheritance of one copy of an European ancestral allele at a given locus represents a risk model where European ancestry decreases risk relative to African ancestry . This set of models reflects the hypothesis that European ancestral alleles are less likely to confer risks but also tests for the alternative possibilities [23] . ANCESTRYMAP provided two scores to assess statistical significance: a locus-specific score and a case-control score [23] . A locus-specific score was obtained in cases ( case-only analysis ) by calculating the likelihood of the genotyping data at the SNPs at the locus under the risk model and comparing it to the likelihood of the genotyping data at the SNPs at the locus assuming that the locus is unassociated with the phenotype . The ratio of these two likelihoods is the “likelihood ratio” , and the log-base-10 of this quantity is the “LOD” score . A locus-specific LOD score of >5 has been recommended as criterion for genome-wide significance [37] . To obtain an assessment of the evidence for a risk locus anywhere in the genome , we averaged the likelihood ratio for association across all loci in the genome , and took the log10 to obtain a “genome-wide score” . We interpreted a genome-wide score >2 as significant [37] . A case-control score was calculated by comparing locus-specific deviations in European ancestry in cases versus controls at each locus across the genome . This score tests whether any deviation in ancestry from the genome-wide average is significantly different comparing cases with controls [23] . If there is no locus associated with phenotype , the case-control score is expected to be distributed approximately according to a standard normal distribution . For loci identified by the case-control score , the level of genome-wide significance was defined as a Z score >4 . 06 or <−4 . 06 , corresponding to an uncorrected nominal P<5×10−5 , or a corrected nominal P<0 . 05 after conservatively correcting for 1 , 000 hypotheses tested ( approximately equals the number of independent chromosomal chunks assigned to either African or European ancestry ) . The main admixture scans were based on a dichotomous phenotype ( i . e . , cases and controls ) in a subset of our samples . To check whether the results were consistent in our entire sample , we used the ANCESTRYMAP software to obtain local estimates of African ancestry at the admixture peak [23] , and then assessed the association of the local ancestry with retinal venular caliber as a continuous trait using linear regression models . Using ANCESTRYMAP , we also obtained regional estimates of ancestry based on the SNPs on chromosome 6p in the admixture panel . In addition to the five covariates used in the main admixture scans , CRVE was further adjusted for HDL cholesterol , LDL cholesterol and plasma triglyceride level , BMI , smoking , and alcohol consumption , all of which covariates have been shown to affect retinal vascular calibers [6]–[8] , [14] . We then performed a normal-quantile transformation for CRVE to ensure normality . In the linear regression models , the transformed CRVE was used as a dependent variable and the local estimates of ancestry as an independent variable . To determine whether there was evidence of residual association with local ancestry after adjustment for global and regional ancestry , we included each individual's PAA and estimated regional ancestry on chromosome 6p as covariates in the regression models . This enabled us to increase power by including all samples in a quantitative analysis , rather than using only a subset of samples with the highest 15% and lowest 15% values in the dichotomous admixture scans described above . To determine whether the association between the local ancestry and CRVE differed significantly by hypertension status , Z tests were used to compare the difference in the regression coefficients obtained from the hypertensive and non-hypertensive groups . To determine a 95% CI for the position of a trait locus , we obtained the likelihood ratio for association at each marker on the chromosome where we identified an association . This provided a Bayesian posterior probability for the position of the underlying causal variant assuming a flat prior distribution across the region for the position of the trait locus . We defined the CI as the central region of this peak containing 95% of the area .	Retinal vessels provide a window to microvascular systems elsewhere in the body . The diameter of retinal vessels varies between racial/ethnic groups , being generally wider in African Americans compared to European Americans . To determine whether genetic background may contribute to this observed difference , we scanned the entire genomes of 1 , 737 African Americans , searching for genomic regions where individuals with either wider retinal venular or narrower retinal arteriolar caliber have a difference from the average percentage of African ancestry . We find that the percentage of African ancestry is positively correlated with retinal venular caliber , particularly in the hypertensive individuals . We detect substantive evidence of association between excess African ancestry and wider retinal venular caliber on chromosome 6p21 . 1 in the hypertensives , but not in the non-hypertensives . The 6p21 . 1 region contains genes that are known to be involved in development and modulation and of retinal vessels . Our results suggest that genetic factors may contribute to the observed difference in retinal vascular caliber between African Americans and European Americans . Further fine-mapping studies of the genomic region may identify variants affecting retinal vascular caliber .	[ "Abstract", "Introduction", "Results", "Discussion", "Methods" ]	[ "ophthalmology/retinal", "disorders", "genetics", "and", "genomics/complex", "traits", "evolutionary", "biology/genomics", "cardiovascular", "disorders/peripheral", "vascular", "disease", "ophthalmology", "public", "health", "and", "epidemiology/epidemiology", "cardiovascular", "...	2010	Admixture Mapping Scans Identify a Locus Affecting Retinal Vascular Caliber in Hypertensive African Americans: the Atherosclerosis Risk in Communities (ARIC) Study
The World Health Organization ( WHO ) has targeted trachoma for elimination as a public health concern by 2020 . Mathematical modelling is used for a range of infectious diseases to assess the impact of different intervention strategies on the prevalence of infection or disease . Here we evaluate the performance of four different mechanistic mathematical models that could all realistically represent trachoma transmission . We fit the four different mechanistic models of trachoma transmission to cross-sectional age-specific Polymerase Chain Reaction ( PCR ) and Trachomatous inflammation , follicular ( TF ) prevalence data . We estimate 4 or 3 parameters within each model , including the duration of an individual’s infection and disease episode using Markov Chain Monte Carlo . We assess the performance of each models fit to the data by calculating the deviance information criterion . We then model the implementation of different interventions for each model structure to assess the feasibility of elimination of trachoma with different model structures . A model structure which allowed some re-infection in the disease state ( Model 2 ) was statistically the most well performing model . All models struggled to fit to the very high prevalence of active disease in the youngest age group . Our simulations suggested that for Model 3 , with annual antibiotic treatment and transmission reduction , the chance of reducing active disease prevalence to < 5% within 5 years was very low , while Model 2 and 4 could ensure that active disease prevalence was reduced within 5 years . Model 2 here fitted to the data best of the models evaluated . The appropriate level of susceptibility to re-infection was , however , challenging to identify given the amount and kind of data available . We demonstrate that the model structure assumed can lead to different end points following the implementation of the same interventions . Our findings are likely to extend beyond trachoma and should be considered when modelling other neglected tropical diseases . Trachoma remains the world’s leading infectious cause of blindness . 200 million people are reported to be at risk of infection , across 42 endemic countries [1] . The causative agent of infection is the bacterial pathogen Chlamydia trachomatis [2] . The World Health Organization through the Alliance for the Global Elimination of Trachoma by 2020 ( GET2020 ) is aiming to eliminate trachoma as a public health problem by 2020 . Two goals have been developed to assist endemic countries striving to achieve the elimination of trachoma as a public health problem . The first goal aims to reduce the prevalence of Trachomatous inflammation , follicular ( TF ) in children aged 1–9 years , to less than 5% by 2020 . Mathematical modelling has been successful in helping to formulate guidelines for the ongoing surveillance and control of a range of infectious diseases including malaria [3] , onchocerciasis [4] , lymphatic filariasis [5] and soil-transmitted helminths [6] . Furthermore , mathematical models can be used to provide guidance on suggested timelines to elimination or control , for a given set of initial conditions and available interventions . However , to generate informative and accurate predictions , models need to be informed by high quality epidemiological data , particularly in terms of the duration of infection and disease , as it is these states which are detectable through diagnostic tests . For trachoma , the control guidelines are based on the disease which occurs as a consequence of infection with C . trachomatis bacteria . Therefore , guidelines are based not on monitoring the causative agent of infection directly , but the longer-term disease associated with it [7] . It is understood that individuals can remain TF positive with detectable disease for far longer than they are Polymerase Chain Reaction ( PCR ) positive . Despite this , estimates of the duration of PCR detectable infection and the duration of disease are not commonly available and are rarely estimated [8] . Nonetheless , a good understanding of the time spent in these states is vital if accurate model projections on time to elimination of trachoma are to be developed . For example , in the absence of on-going sustained transmission in an endemic region , if one assumed that the duration of a disease episode was less than 1 month ( as estimated for individuals 15 years or older in the community [8] ) the expected time to reach the < 5% elimination threshold would be much shorter than if the assumed duration of disease was 2 years; thus , the assumed rate of recovery from disease is likely to have a large impact on the expected time to reach elimination targets . Mathematical models are developed and informed by the natural history of infection . For trachoma , the Susceptible , Infected , Susceptible ( SIS ) model structure has most commonly been used [9–15] where individuals in the S state are susceptible to infection and those in the I state are infected and infectious and , thus , are PCR detectable . Such models are fitted to PCR data collected during clinical trials of trachoma treatment [9 , 10 , 12] . However , current control guidelines are based on disease , not PCR detectable infection . Therefore , if models are to be informative in terms of whether the guidelines on TF prevalence will be achieved , it may be desirable to capture the dynamics of both PCR and TF positivity , although this has rarely been done [16 , 17] . The lack of modelling work in this field is likely to have been exacerbated by the limited longitudinal population level data that measure both PCR and TF positivity across multiple age groups , particularly following the implementation of interventions . In addition to the SIS compartmental model structure , several other variant structures may also be considered appropriate given the natural history of trachoma infection [18–21] . It has been reported for other infectious diseases that the structure of the model assumed can impact the estimated effort required to control and eliminate that disease [22–24] . Therefore , when modelling the transmission of trachoma to make projections on the feasibility of elimination , it is important to select not only the most parsimonious and statistically appropriate model , but also to understand how the assumed model structure may impact elimination projections . Here we compare four different model structures which could realistically all represent the natural history of trachoma infection as understood by epidemiologists through the interpretation of experimental data . We statistically fit each model to cross-sectional data on bacterial load , PCR and TF prevalence for three different age groups . We then evaluate the performance of each model structure using the Deviance Information Criterion ( DIC ) . With the parameter estimates obtained from each of the best performing models we assess the feasibility of and time to elimination , to understand if and how they differ . We used cross-sectional data on PCR and TF prevalence in individuals aged 1–4 , 5–14 and 15 years or older , collected from a hyperendemic community in Tanzania at one point in time , prior to the roll out of trachoma interventions [25] . Data on the mean bacterial load by age group were available from [25] . Data on age-specific bacterial load , PCR and TF prevalence were used to fit each of the models evaluated . We evaluate 4 different plausible natural histories of infection [9 , 11 , 13 , 26] and disease [7 , 16 , 18 , 19] that may occur following exposure to trachoma . The model structures we evaluate highlight the clinical and epidemiological observations made in the field and laboratory [19 , 21 , 27 , 28] . The first model , Model 1 , follows the structure represented in Fig 1a . Here susceptible individuals ( S ) become infected at a rate λ , they incubate infection in the ( I ) state , and progress at a rate σ to the infected infectious state ( ID ) where they test PCR positive and TF positive . Individuals leave ID at a rate ω and progress to the disease only state ( D ) , where they are only TF positive , and recover from the disease only state at a rate ρ and return to the susceptible state ( S ) . For Model 2 we assume the same structure as Model 1 ( Fig 1b ) , however we do not assume individuals in the D state are 100% immune to re-infection [19] . Instead we explore 3 levels of susceptibility to re-infection ( Γ ) : 20% , 50% and 80% . All other transitions are the same as described in Model 1 . In Model 3 ( Fig 1c ) we evaluate the structure previously postulated by Shattock et al [18] . Here , the first 3 states are identical to those described in Model 1 and 2 . However , for Model 3 we split the duration of time spent in the D state across two compartments . In the D state , individuals are immune to re-infection . They then progress to the PD state at a rate γ [19] . In this state individuals are susceptible to re-infection with the same susceptibility levels described in Model 2 . Model 4 ( Fig 1d ) , we introduce an additional infected state , the IO state , which comes after the incubating state , where individuals are not infectious ( I ) , but prior to the ID state . In the IO state individuals have a PCR detectable infection , but are not yet TF positive . From here individuals progress to IA at a rate η where they are PCR and TF positive , individuals recover from their infection and progress to the D state , where they are only TF positive , but as with Models 2 and 3 individuals could experience re-infection in the ( D ) state with the same susceptibility levels described in Models 2 and 3 [19] . All models follow the ‘ladder of infection’ structure [11 , 16 , 26] , whereby each subsequent infection leads to improved immunity following re-infection . In all 4 model structures we reflect improving immunity as an increase in the rate of recovery from infection and disease episodes , in addition to a reduction in infectivity with each successive infection . We assume that the infectivity of an individual is proportional to their bacterial load . In the model we reflect declines in bacterial load with repeated infection as reductions in an individual’s infectivity to others . This represents a trend in agreement with the data from trachoma endemic communities in which the bacterial load decreases with age [25 , 29 , 30] . For each model structure ( A-D , Fig 1 ) we have two sub-variants , the 4-parameter and the 3-parameter versions . These models pertain to two alternative sets of assumptions about how bacterial load and infectivity decline with consecutive infections . We assume for the 4 parameter model that infectivity is proportional to bacterial load and therefore declines exponentially with the number of prior infections . For the 3 parameter model we assume that infectivity declines linearly with the log of the bacterial load i . e . linearly with the number of prior infections . We chose exponential functions as fairly flexible low-parameter functions that , for the rates of recovery , would accomplish the goal of a ) rising from an initial value–for no and low numbers of infections–to b ) saturating at a high value for high numbers of infections . We note that we also tested the use of a log-logistic function instead of an exponential , however it was no better performing than the exponential function . Additional detail on the model parameters and state variables are presented in Table 1 . Details on the immunity functions and mathematical equations for each model are presented in S1 File . All parameter values and definitions are provided in Table 1 . We assume that the mean minimum duration of an infection episode was 10 weeks and the duration of a disease only episode was 1 week ( Table 1 ) , the same as those estimated for the oldest age group in Grassly et al [8] . We take estimates from the oldest age group to parameterise the minimum duration of an infection and disease episode . Immunity to trachoma is thought to develop through repeated infections . Therefore as those in the highest age group are most likely to have experienced the highest number of infections , we assumed that they would have the highest levels of immunity . It is inherently challenging to estimate immunity functions [32] and , given only 3 data points were available , the true values of any immunity parameters were likely to be unidentifiable . As such , exponential increases in the rate of recovery from infection and disease , with the number of prior infections experienced by an individual , were informed by Grassly et al [8] . Age-specific estimates of the duration of infection and disease , and were fixed for the purposes of model fitting . Each of the four model structures evaluated were fitted as 3 and 4 parameter models to the data . An additional factor: the relative susceptibility of the diseased , non-infected state for new infections was varied . One value was used for Model 1 , and 3 different values for each model structure 2 , 3 and 4 . Therefore a total of 10 different models were fitted for each parameterisation of the model structures . For the 4 parameter models we estimated the transmission rate parameter β per day−1 for the data , the duration of an individual’s first infection and disease episode in the ID and D states , and the rate at which infectivity changed with each successive infection for the bacterial load function ( Table 1 ) . For the 3 parameter model we estimated the first three parameters listed in the 4 parameter model , but assumed a constant linear decline in the log load of an individual’s bacterial load with each successive infection , thus , in the 3 parameter model we did not estimate the rate of increase in improved immunity with re-infection . Parameter estimation was performed using Markov Chain Monte Carlo ( MCMC ) . The chains were run for 10 , 000 iterations . The Robbins-Munro algorithm was implemented as part of the adaptive stage of the MCMC-Metropolis Hastings algorithm , to ensure the proposal distributions were adaptively tuned ensuring efficient exploration of the posterior [33] . Selection of the most parsimonious model and fit of each model to the data was assessed using the DIC [34] , therefore we assumed our posterior distribution was approximately multivariate normally distributed . Fits to the data for each model are presented in Table S2 in S1 File . Estimates from the 4 parameter model are provided in Table S3 in S1 File and estimates from the 3 parameter model are provided in Table S4 in S1 File . Uninformative uniform priors were specified for all parameters . MCMC diagnostics are presented in Table S5 in S1 File . We calculate the Gelman-Rubin statistic for 2 MCMC chains to assess convergence [35] and the Effective Sample Size ( ESS ) for each model fit . For each model structure ( described above ) we performed simulations to assess the potential impact of Facial cleanliness and Environmental improvements ( F and E ) within the community , along with the implementation of mass drug administration ( MDA ) . All simulations were started from endemic equilibrium . For communities with greater than 20% TF , 5 annual rounds of MDA were performed , and for those with TF 20% or less we performed 3 rounds of annual MDA . We also assessed the possible impact of F and E to reduce transmission . The true impact of F and E remains poorly quantified [36] , therefore we consider a range of reductions in transmission that may be possible ( between 0–50% ) . β was assumed to decline exponentially over the intervention period , to model an increasing uptake of transmission reduction interventions in the community over time . We model changes in β as an instantaneous drop when each annual round of MDA is performed as we assumed intensified health promotion activity would be conducted when MDA was distributed . Reductions in transmission which were only considered to occur through the implementation of F&E and were assumed to be maintained following the cessation of treatment . For each model structure we assessed the time taken and the feasibility of reducing TF prevalence to less than 5% within the community in children under 10 years old . A constant level of treatment coverage ( 80% ) between each round and across model comparisons was assumed , along with a fixed treatment efficacy of 85% [12 , 31] . A schematic of the movement of individuals between compartments following treatment is illustrated in Fig 1 . We assess the sensitivity of findings on the feasibility of elimination for the different model structures to variation in 6 different fixed parameters , these are: treatment efficacy , duration of first infection and disease episodes , maximum rate of recovery from infection , maximum rate of recovery from disease and the degree of age mixing in the population . These were assessed across both transmission settings and all levels of transmission reduction due to F and E . Across both transmission settings infection was more likely to re-emerge if the infectivity was assumed to decline exponentially ( Figs 3 , 4a–4c ) . However , in general , this functional form led to an overall better fit to the cross-sectional data ( Table 1 ) . This somewhat counterintuitive effect with exponentially declining infectivity ( explored in [37] ) results in the reproduction number associated with the full model increasing with each subsequent treatment round of MDA . This is as a result of the concentration of infectivity increasing with multiple rounds of MDA , as a higher number of individuals in the population have experienced fewer infections , resulting in individuals aggregating at higher infectivities as their progress along the ‘ladder of infection’ is slowed or halted due to MDA . When assuming an exponential decline in infectivity it was not possible to eliminate disease from the community with 40% TF prevalence with Models 2 or 3 , even with annual MDA for 5 years , this was only possible when some reduction in the transmission rate was also included ( Fig 3a–3c ) . It was only possible to eliminate disease when a linear decline in bacterial load was assumed under Models 2 and 4 with at least a 10% reduction in transmission and annual MDA for 5 years ( Fig 3d and 3f ) ; in all other structures and transmission reduction scenarios , infection re-bounded . It was not possible to reach the elimination threshold guideline at all with Model 3 ( Fig 3b and 3e ) . When TF baseline prevalence was 40% , under Models 2 and 4 , it was possible to eliminate infection within 5 years with annual MDA and an overall transmission reduction when assuming a linear decline in bacterial load ( Fig 3e and 3f ) , provided transmission reduction was greater than 10% . However , if an exponential decline in bacterial load was assumed ( Fig 3a–3c ) , it was only possible to eliminate infection in Model 2 with 5 annual rounds of MDA and 50% reduction in transmission , but this was not sufficient for Model 4 . Under no intervention conditions evaluated here was it possible to eliminate infection with Model 3 ( Fig 3b and 3e ) . Assuming TF prevalence was 20% we implemented 3 annual rounds of MDA and transmission reduction ( between 0–50% ) ( Fig 4 ) . Considering a linear decline in bacterial load it was possible with at least 10% transmission reduction to reduce disease prevalence below the target threshold in Model 2 and 4 . However , if there was no transmission reduction disease appeared to re-emerge ( Fig 4d and 4f ) . As with the previous prevalence levels it was not possible to achieve the target level of disease prevalence with Model 3 ( Fig 4e ) . When assuming an exponential decline in bacterial load , with all four model structures , it was possible to reduce the prevalence of disease to the target level of less than 5% . However , without subsequent rounds of MDA , this was not maintained in Model 3 and disease re-emerged ( Fig 4b ) . However , for Models 2 and 4 , 10% reduction in transmission respectively was sufficient to ensure that disease did not re-emerge in the community ( Fig 4a and 4c ) and suppression below the target level was maintained . Observing the rates of re-bound under different model structures we found that for Model 2 with modest levels of transmission reduction rapid re- emergence was observed for the 4 parameter model . This is in-part likely to be because for a given prevalence level we need a higher force of infection if immunity was high , resulting in faster rates of rebound until β was reduced sufficiently . For Model 3 under both parameter versions and all scenarios rapid rebound of disease was seen . This is likely to be because this structure includes 2 compartments which do not contribute to the overall force of infection . Therefore in order to obtain a fixed level of prevalence the value of β must be increased substantially in comparison to other model structures , thus making persistence of disease more likely under this structure . Limited evidence of re-bound was seen when evaluating Model 4 in comparison to other models this model had an overall longer duration of PCR detectable infection , therefore a lower value of β was needed to attain any given level of prevalence . This meant that when an intervention was applied , the lower overall force of infection resulted in a slower rate of rebound . The on average higher infectivity of individuals in the 3 parameter model in comparison to the 4 parameter model ( Table S4 in S1 File ) is likely to explain the slightly higher re-bound rates in the 3 parameter version of Model 4 , in comparison to the 4 parameter version . We conducted one-way univariate sensitivity analyses with 6 of the fixed parameters in the model to assess their impact on the models assessment of the feasibility of elimination ( Table S6 and Table S7 in S1 File ) . For the 4 parameter version of Model 1 and Model 2 when TF was 40% or 20% , variation in treatment efficacy and the minimum duration of infection had the largest impact on final TF prevalence . Here , higher treatment efficacy resulted in faster infection rebound , leading to a higher final TF prevalence . For Model 2 final TF prevalence was 70% compared to 61% when treatment efficacy was increased from 85% to 100% ( Table S6 in S1 File ) . In contrast , a 50% reduction in the minimum duration of infection resulted in fast infection re-bound resulting in a high final TF prevalence above the baseline results , until a 50% reduction in transmission was implemented ( Table S6 in S1 File ) . Decreasing the minimum duration of disease resulted in a final higher TF prevalence when transmission reduction < 50% was implemented . For the 3 parameter version of Model 1 and Model 2 final TF prevalence decreased with increasing treatment efficacy . ( Table S6 in S1 File ) . While reduction in the minimum duration of infection and disease episodes resulted in higher final TF prevalence levels when little or no transmission reduction was implemented . Final TF prevalence for Model 2 when transmission was reduced by 10% , was 60% when the minimum duration of infection was 5 weeks , but 8% for the 10 week baseline value ( Table S6 in S1 File ) . Across the 3 and 4 parameter versions of Model 3 , little to no variation in the final TF prevalence level was seen when sensitivity to the fixed parameters was conducted ( Table S6 and Table S7 in S1 File ) , and the inability to even come close to reducing or eliminating disease was seen across all parameter sets tested . Insensitivity of Model 3 to perturbations in the six different parameter sets , is likely to be a consequence of the high force of infection needed to obtain a fixed level of prevalence with this model structure ( as described in the model fitting results ) , which increases the persistence of infection and disease . Equally , individuals spend a long time in the TF +ve only state under this structure , therefore changes in infection rate parameters are less likely to have a profound impact on this model . For Model 4 , lower treatment efficacy resulted in a higher final TF prevalence than when the baseline value was used . Prevalence was 11% instead of 2% when treatment efficacy was 65% in the 4 parameter model when no transmission reduction was modelled ( Table S6 in S1 File ) . As with Models 1 and 2 , reductions in the maximum duration of infection and disease episodes at low levels of transmission reduction resulted in higher final TF prevalence for the 4 parameter model . If the maximum rate of recovery from infection was changed to 0 . 008 from 0 . 006 final TF prevalence was 40% , in comparison to 2% ( Table S6 in S1 File ) . However , when endemic prevalence of TF was 20% for the 4 parameter model , results were consistent across all variation in the parameter sets . For the 3 parameter versions of Model 4 results were consistent across all parameters sets tested when endemic prevalence of TF was 40% . However , when endemic TF prevalence was 20% for the 3 parameter model , an increase in final TF prevalence from the baseline was observed when the minimum duration of infection and disease episodes was reduction , particularly when no transmission reduction was implemented . Final TF prevalence was 14% instead of 4% when the minimum duration of infection was decreased from 10 weeks to 5 weeks ( Table S7 in S1 File ) . In this study we present the first attempt to fit a mechanistic epidemiological model of trachoma transmission to bacterial load data , PCR and TF prevalence data across 3 different age groups . We demonstrate that it is possible to fit to the age-structured PCR data well but the very high level of TF in the youngest age group analysed in this hyper-endemic setting proved challenging to capture with all model structures tested . In addition , we highlight that predictions about the future prevalence of TF within a community can depend on the model structure assumed . While a range of different model structures can describe the natural history of trachoma infection well , Model 2 , with re-infection in the D state ( TF positive , PCR negative ) , was statistically the best performing model under all conditions . Model 2 represented the most parsimonious model structure when assessed by the DIC score obtained through fitting to the dataset used here . The appropriate level of susceptibility to re-infection was , however , challenging to identify given the amount and kind of data available . We can only therefore confidently say that our model selection study suggests that individuals with active disease but no current infection remain susceptible to infection , but it does not suggest what their susceptibility to infection is , relative to those with neither active disease nor infection . We demonstrate that overall a better fit to the data , i . e . ensuring infection and disease age-specific prevalence were captured , was provided by an exponential reduction in bacterial load in comparison to a linear decline in load . However , the use of an exponential rather than linear bacterial load decline was also shown to suggest that more effort may be required to reduce the prevalence of TF in the long term , due to the persistently high levels of load associated with those who have experienced very few prior infections i . e . those likely to be the few remaining infected individuals when elimination is close . Estimates of the effort required and feasibility of elimination were markedly different under different model structures . Model 3 showed that the prospect of reaching TF less than 5% was very low , while with Model 2 annual MDA and transmission reduction together , in prevalence settings below 40% TF , ensured that the goal was reached within 5 years . To gain further understanding into the long-term transmission dynamics of trachoma and generate accurate elimination timelines , further insight into the duration of infection and disease episodes is required , ideally through at least one further study designed to measure these durations . Furthermore , our results highlight the importance of identifying and understanding the most appropriate and parsimonious structure to model trachoma transmission , this is essential if we wish to use mathematical models to help understand the transmission dynamics of trachoma and to model current and alternative intervention strategies . Our sensitivity analysis highlights that projections on the feasibility of elimination under different model structures were sensitive to a number of key parameters , particularly for Models 2 and 4 . Final TF prevalence after 7 years was most impacted by the assumed duration of an individual’s first infection and disease episodes , in addition to the efficacy of treatment . Suggesting that a more thorough understanding of these parameters would be valuable for future model forecasting . A small amount of variation in the final TF prevalence was observed when the maximum rates of recovery from infection and disease were perturbed , however the impact was not as profound as the outcomes from the aforementioned parameters . In general , we observed that as the modelled reduction in the transmission rate increased sensitivity of the model prediction of the final TF prevalence level decreased . However , the final TF prevalence outcomes from Model 3 appeared insensitive to perturbations across all parameter evaluated . We were consistently unable to capture the high prevalence of TF in the youngest age group , but were able to capture PCR prevalence for this age group . This suggests that the models evaluated may be missing or mis-specifying a key component of the epidemiological system . For example , it could be that the functional form used to describe the development of immunity has been mis-specified here ( as an exponential function ) or that age-specific prevalence ratios of PCR vs TF vary according to the transmission setting . However , particularly for Model 2 , prevalence of disease and infection were matched well for the two older age groups . However , it has been suggested that at the population level the relationship between TF and PCR positivity is approximately linear [38] , which can also be seen in our model projections . Since , we have only fitted to cross-sectional data from a single time-point from a single region and time point , we cannot disregard the possibility that there may be an anomaly in the data , and extrapolating our findings to a wider context should only be done with caution . Equally , the prevalence of infection within the adult age group may be considered high . However , in the absence of cross-sectional data collected across a wide range of age groups from different study sites , it is difficult to assess whether or not this observed infection prevalence is abnormally high or not . The models evaluated here have only been fitted to one high prevalence site , however trachoma transmission can be highly heterogeneous between neighbouring communities . Therefore it is possible that if we had used data from a different community we may have achieved different results . However our hope from fitting to data from this study site is that while certainly the baseline level of transmission within other neighbourhoods may not be as high , we would hope that some of our findings would be generalisable to other settings . Nevertheless the use of data from a single high prevalence study site is a clear limitation of the study . Statistical models have been shown to forecast prevalence of infection and disease well and can predict changes in prevalence over time [7 , 9 , 12] , however there is less flexibility within a statistical framework to explore the impact of novel or future alternative intervention strategies . Therefore , selecting an appropriate mechanistic model structure is important if we wish to more accurately model trachoma transmission and assess the possible impact of different intervention strategies in the lead up to 2020 . Furthermore , we demonstrate that our understanding on the feasibility of trachoma elimination varies under different model structures . In this study certainty about the appropriate model structure and susceptibility level to re-infection was hampered by a limited amount of data relating to the durations spent in different infection and disease states , in addition to longitudinal post-intervention follow-up data on infection and disease from a range of different communities and transmission settings . For example , if we knew the average duration an individual spends as PCR-positive but TF-negative we could parameterise our models with more certainty , a point that is even more important for the duration in which individuals remain only TF positive . However , in all of our models , PCR and TF positivity are inherently linked . We suggest that further validation of appropriate model structures can be provided through fitting different structures and different model types to longitudinal data from a range of different transmission settings , coupled with more large scale model and data comparisons , as we seek to develop models which help provide guidelines on time to elimination . Our findings may also be applicable to other NTDs where certain key parameters are not well known , where limited data exists and limited investigation has been done to validate the model structure being used to model transmission .	Trachoma is the world’s leading infectious cause of blindness . Mathematical models are used by researchers to examine the spread of infectious diseases and understand how they can be controlled . Such models are developed based on the natural history of infection . For trachoma we identify four different model structures which could all represent the natural history of trachoma infection . We fit each of the models to infection and disease prevalence data for 3 different age groups . We find that one of the models is able to fit the data better than others , however some factors about the model are difficult to identify due to limited data . The ease of eliminating disease within a community assuming the same interventions varied depending on the model structure assumed . Our results highlight that some models of trachoma fit to infection and disease data better than others , but that more data is needed to identify more specific aspects of the model structure . In addition we show that different model structures may give different results in terms of the effort required to control trachoma transmission .	[ "Abstract", "Introduction", "Materials", "and", "methods", "Results", "Discussion" ]	[ "medicine", "and", "health", "sciences", "pathology", "and", "laboratory", "medicine", "pathogens", "immunology", "tropical", "diseases", "microbiology", "mathematical", "models", "bacterial", "diseases", "age", "groups", "eye", "diseases", "neglected", "tropical", "dis...	2017	Improving our forecasts for trachoma elimination: What else do we need to know?
We consider the feasibility of reusing existing control data obtained in genetic association studies in order to reduce costs for new studies . We discuss controlling for the population differences between cases and controls that are implicit in studies utilizing external control data . We give theoretical calculations of the statistical power of a test due to Bourgain et al ( Am J Human Genet 2003 ) , applied to the problem of dealing with case-control differences in genetic ancestry related to population isolation or population admixture . Theoretical results show that there may exist bounds for the non-centrality parameter for a test of association that places limits on study power even if sample sizes can grow arbitrarily large . We apply this method to data from a multi-center , geographically-diverse , genome-wide association study of breast cancer in African-American women . Our analysis of these data shows that admixture proportions differ by center with the average fraction of European admixture ranging from approximately 20% for participants from study sites in the Eastern United States to 25% for participants from West Coast sites . However , these differences in average admixture fraction between sites are largely counterbalanced by considerable diversity in individual admixture proportion within each study site . Our results suggest that statistical correction for admixture differences is feasible for future studies of African-Americans , utilizing the existing controls from the African-American Breast Cancer study , even if case ascertainment for the future studies is not balanced over the same centers or regions that supplied the controls for the current study . A genetic association study estimating the main effects of single nucleotide polymorphisms ( SNPs ) or other genetic variants upon the risk of a rare or common disease in minority populations is a setting in which it is especially attractive to consider the use of existing genotype data as a supplementary or even a primary source of controls . DNA samples may be expensive and difficult to obtain , and response rates are often lower in minority populations [1] . Researchers might consider using an already available “stand in” population sample as controls , provided that genotype frequencies are equivalent to those in the population from which controls would be drawn . There are however two immediate concerns raised , one fundamental and the other technical in nature: The fundamental question is whether or not the controls are sampled from the same underlying population ( or populations ) as are the cases – or more generally the feasibility and “cost” ( generally loss of statistical power ) of controlling for case/control differences if they arise . The technical question is whether differences in genotyping , including differences in DNA preparation , and in the actual markers genotyped in cases and controls , i . e . when the platforms are not identical so that imputation is relied upon to make up the difference , may introduce false positive ( or false negative ) associations . Consider the problem of conducting a genetic association study aimed at discovering genetic variants related to the risk of a disease , where there already exists extensive genotyping data , perhaps publicly available , for members of similar populations . If the disease under consideration is rare ( so that genotype frequencies for controls may be expected to be the same as in the general population ) then it is intuitively appealing to consider using existing control data from studies of other rare diseases ( or population-based studies if they exist ) to reduce the cost or increase the statistical power of an association study . From the classical case-control literature , a study that uses 1∶ matching of controls to each of cases will be equivalent in power to a 1∶1 matched study with cases ( and an equal number of controls ) . Thus a study with a large number of controls , , for each case will have nearly twice the effective sample size of a 1∶1 matched study [2] . Note that 1∶m matched studies ( with m>1 ) are only cost effective if it is more costly or difficult to obtain additional cases than it is to obtain additional suitable controls since the increment in effective sample by ( for example ) doubling the number of case-control pairs ( in a 1∶1 matching ) is twice that of adding two new controls to each existing pair ( to achieve 1∶3 matching ) to a study even though the total number of participants is the same . If however the cost of adding an additional control is far less than that of adding an additional case ( because control data is already available ) then adding the almost “free” controls is highly attractive although the returns diminish as more and more controls are added , with the increment in effective sample size governed by m/ ( m+1 ) . This paper considers these issues from both theoretical and empirical perspectives . We apply a recent generalization [3]–[6] of the testing procedure of Bourgain et al [7] to the situation where population substructure ( in the broad sense ) , including admixture and relatedness between subjects , is estimated from marker data rather than being assumed to be known as the basis for our theoretical considerations of study power . We point out below the relationship between this procedure and that of the more widely known principal components technique [8] . Our empirical investigation of the use of existing controls utilizes data from a genome-wide association study ( GWAS ) of breast cancer in African American women , namely the African American Breast Cancer ( AABC ) study , in which cases and controls come from a total of 9 different studies widely distributed geographically throughout the United States . African Americans are a relatively understudied group ( compared to European Americans ) in studies of genetic susceptibility . African Americans are admixed with Europeans and ( in some cases ) with Hispanics ( themselves an admixed group ) and Native Americans [9] , [10] . We examine empirically the false positive rates that occur when cases from one geographical location or study within the AABC study are combined with controls from other AABC locations or studies , as well as the success ( and cost in terms of loss of effective sample size ) of adjustment for the observed population differences in global genetic ancestry when analyzing such illustrative data sets derived from the AABC study . We utilize an approach derived from that of Bourgain et al [7] which has been discussed extensively in recent papers [3]–[5] . This approach for accounting for relatedness between subjects in association tests adopts a “retrospective” approach towards the problem of testing for disease associations using marker data , in which ( as in the Armitage test ) the allele frequency of a variant is related to case-control status . In vector notation we have of observed values for a given SNP j . Here is the total number of subjects ( cases+controls ) in the study , and SNP values are coded as ( 0 , 1 , 2 ) for the number of copies of a specified allele , , carried by subject . ( This coding of SNP genotype implies that we are interested in additive models for the relationship between disease risk and genotype but the approach can readily be extended to other codings ) . The retrospective approach models the mean of as a function of case-control status . If we define the design matrix C to have rows ( 1 ci ) where ci is case-control status ( 0 or 1 ) for subject i then the mean , , of is written asRelatedness between subjects induces a covariance matrix for the number of copies of a given SNP of form ( 1 ) with specific to each SNP but with the same matrix for all SNPs . In fact , for known pedigree relationships , and unrelated founders , this matrix has diagonal elements equal to where is the inbreeding coefficient for subject and each off-diagonal element , , is twice the kinship coefficient for the relationship between subjects and [7] . It is worth noting that in general the topic of relatedness includes what is often considered to be population substructure . For example consider two large but isolated populations ( freely mixing within each population ) that have been separated for many generations . While a random sample of people from the same population ( with sample size small relative to the population size ) might be considered unrelated to each other when considering that population separately , when considering the two populations together people from one isolated population are considered to be related to each other , relative to those in the other population . In particular , genetic markers will , through a process of random drift and other factors , be able to distinguish members from the two populations , and this will be detectable when calculation of the matrix is performed . A standard method of simulating genetic markers for divergent populations stemming from the same ancestral population ( e . g . the Balding-Nichols model [11] ) can readily be shown to produce covariance matrices of the form of expression ( 1 ) . If and are both known then the best linear unbiased estimate ( BLUE ) of the regression vector is of weighted least squares formand the variance covariance matrix of the estimates is in the form ofThus inference on the significance of the allele frequency difference between cases and controls may be based upon the Wald test statistic ( 2 ) with the ( 2 , 2 ) element of In general of course , and are not known , except in the case of known pedigrees and unrelated founders , where can be computed from first principles . The estimation of using marker data has been considered by a number of authors and both method of moments [4] , [5] and maximum likelihood methods [3] have been considered . A method of moments estimator of can be concisely written [5] as ( 3 ) and the estimate of asOne value of this approach , which is exploited here , is that it is relatively easy to compute the power of the Wald test if we can hypothesize a form of the relatedness matrix . For a given form for ( below we consider several forms for both isolated population models and more complex admixed populations ) then for a given sample size , , a given allele frequency for a causal SNP , and a hypothesized difference in allele frequencies between cases and controls ( which can then be related to odds ratios in typical case/control analysis ) we can compute the non-centrality parameter of ( and hence the power of the test ) as ( 4 ) We illustrate the computation of this non-centrality parameter for a number of important special cases in the results section below . It is worth noting now , however , that the Bourgain test appears to be reasonably powerful compared to other procedures , and can sometimes be considered as a compromise between the principal components method [8] and genomic control [12] . We attempt to justify this last statement in the results section below . In addition to the Bourgain test we used several well known tools for addressing population structure in the AABC data . For example we computed eigenvectors through the use of the program EIGENSTRAT [8] . Briefly , each eigenvector explains a proportion of the genetic variation among samples in the analysis so that the leading eigenvector explains the greatest variation , followed by the second eigenvector , and so forth . The full set of eigenvectors form an orthonormal basis so that each eigenvector is scaled on the unit interval and linearly independent from all other eigenvectors . Note that the EIGENSTRAT procedure is operating on the same estimated matrix , , that we have described above . To assess ancestry within the AABC study in relation to reference populations from HapMap , we performed a principal components analysis based on ancestry informative markers that were genotyped in both the AABC study and the HapMap Phase 3 populations . The 2 , 546 ancestry informative markers ( contained within the Illumina 1M genotyping array which was used in the AABC scan ) were selected based on low inter-marker correlation and high correlation to a previously determined eigenvector that explained African and European ancestry . We quantified percent African ancestry for each of the nine AABC study populations by running the program STRUCTURE for each study population . The program implements a Markov Chain Monte Carlo algorithm that provides the posterior estimates of the proportion of ancestry from each of k clusters for each individual , where k is specified by the investigator . For each AABC study population , we assigned k = 3 , including genotypes from the same ancestry informative markers used in PCA genotyped in YRI , CEU , and JPT from HapMap Phase 3 . AABC included 9 epidemiological studies of breast cancer among African American women , which comprise a total of 3 , 153 cases and 2 , 831 controls . Below is a brief description of these studies . Genotyping in stage 1 was conducted using the Illumina Human1M-Duo BeadChip . Of the 5 , 984 samples from these studies ( 3 , 153 cases and 2 , 831 controls ) , we attempted genotyping of 5 , 932 , removing samples ( n = 52 ) with DNA concentrations <20 ng/ul by pico green assay . After clustering the genotype data we removed samples based on the following exclusion criteria: 1 ) unknown replicates ( ≥98 . 9% genetically identical ) that we were able to confirm ( only one of each duplicate was removed , n = 15 ) ; 2 ) unknown replicates that we were not able to confirm through discussions with study investigators ( pair or triplicate removed , n = 14 ) ; 3 ) samples with call rates <95% after a second attempt ( n = 100 ) ; 4 ) samples with ≤5% African ancestry ( n = 36 ) ( discussed below ) ; and , 5 ) samples with <15% mean heterozygosity of SNPs in the X chromosome and/or similar mean allele intensities of SNPs on the X and Y chromosomes ( n = 6 ) ( these are likely to be males ) . In the analysis , we removed SNPs with <95% call rate or minor allele frequencies ( MAFs ) <1% . To assess genotyping reproducibility we included 138 replicate samples; the average concordance rate was 99 . 95% ( >99 . 93% for all pairs ) . We also eliminated SNPs with genotyping concordance rates <98% based on the replicates . The final analysis dataset included 3 , 016 cases and 2 , 745 controls , with an average SNP call rate of 99 . 7% and average sample call rate of 99 . 8% . Hardy-Weinberg equilibrium ( HWE ) was not used as a criterion for removing SNPs for this analysis . We use the Balding Nichols model [11] for allele frequency differences between isolated populations . In this model allele frequencies for a SNP in modern data populations are distributed according to a beta distribution with the ancestral allele frequency of that SNP . In this model the variance of the modern day allele frequency is , thus is a parameter specifying the degree of separation between the modern day and ancestral population . As described in Rakovski and Stram 2009 [4] if genotypes are obtained for randomly sampled individual from two modern day isolated populations using this model and the separation of each modern day population from the ancestral population equals ( for ) statistic then the covariance matrix , between subjects for the jth SNP will have diagonal terms equal to for members of the first population , diagonal terms equal to for members of the second , off diagonal terms of , , or zero for pairs of individuals who are either both from the first population , both from the second population or from different populations respectively . Here is the frequency in the ancestral population of SNP . Consider now a study in which all cases come from one isolated population , and all controls from another . Assume for simplicity that ( both populations have the same degree of separation from their ancestral source ) , and that the number of cases and controls are both equal to so that total sample size is ( the calculations below can be readily altered for different matching fractions if necessary ) . Thus we can write the variance of the estimator of the case-control difference for SNP aswith the first column of C being a vector of 1's and the second column of C a vector of 1's and 0's indicating case-control ( and population ) status . Using a readily derived formula for the inverse of an matrix of compound symmetric formwe can easily write ( 5 ) Thus the non-centrality parameter , = of a test of association does not increase linearly in N , but rather is bounded above by the value . This can impose severe limitations on the power of any study in which there are such differences . To put this in perspective , consider two isolated populations which are each separated from their ancestral population with an F value of 0 . 0005 , and consider an allele that exhibits 40 percent frequency in the ancestral population . The variance of the difference between the two isolated modern day populations in the frequency of this allele is equal to so that we would expect by chance that there is a about a 1 . 5 percent difference in allele frequencies between cases and controls for such an allele . Consider now the detection , in a study of 5 , 000 cases and 5 , 000 controls , of a disease-causing allele of the same frequency associated with a 10 percent difference in allele frequencies between cases and controls ( = . 2 ) . The difference in allele frequencies is approximately 6 times larger than expected due to population differences , and can be seen to correspond to an odds ratio for disease , under a multiplicative risk model , of 1 . 5 per copy of the risk allele . From equation ( 2 ) the non-centrality parameter will be equal to 34 . 73 in this case; on the other hand if between cases and controls is 0 the non-centrality parameter will equal 208 . 33 . Thus a study that would , given no differences between cases and controls in population of origin , have overwhelming power ( > . 9999 ) to reject the null hypothesis at a genome wide level significance ( p< ) is , under this alternative , reduced to having power of only 56 percent after correcting for the differences in origins of cases and controls . The survey of European populations by Nelis et al [21] estimates fixation indices , , ( which can be equated to under the Balding Nichols model ) between populations in SNP allele frequencies which range from less than . 001 between neighboring populations to 0 . 023 for Southern Italy versus parts of Finland . Because our values as defined above are between present day and ancestral populations the fixation indices calculated between present day populations by Nelis et al need to be multiplied by ½ to be consistent with our definition of . Thus the example we have given ( ) corresponds only to the nearest neighbor populations in Europe and would appear to throw into doubt any thought of using control data not perfectly matched in ancestry to cases . While the calculations given above appear to be pessimistic regarding the usefulness of shared control data it is important to note that the completely isolated population model is naïve and makes assumptions not applicable to the study subjects for the AABC study or indeed for most modern populations . Therefore we broaden our discussion to admixed populations , specifically , when the DNA from both cases and controls come from groups that are admixed from the same two ancestral populations . We consider this in two parts , first deriving results for comparisons between “completely” admixed populations , i . e . where the two populations have different levels of admixture between the ancestral populations , but when there is no within-population heterogeneity in ancestry . Next we focus on the much more realistic setting of incompletely admixed populations serving as cases and controls . Our analysis focuses upon ( 1 ) estimating a more appropriate model for the distribution of ancestry in the data from the AABC data than the homogeneous “complete” admixture described above , ( 2 ) checking the adequacy of this model , enriching it if necessary , and ( 3 ) describing the implications of the model for the likely power to detect associations in studies in which all cases come from outside the AABC study populations , and controls are chosen from within the AABC . We can partly mimic such studies by making up “pseudo” case-control studies using the data from the different study sites within the AABC study . We have adopted a somewhat non-standard approach in relying upon the Bourgain test rather than principal components [8] or related methods [22]–[24] to control for population structure in a GWAS of a minority population with cases/controls drawn from multiple studies with different designs and recruitment approaches . We have done this mainly because we can give certain theoretical results for the Bourgain test when assuming specific forms for the true kinship matrix using this procedure . It is worth noting that the Bourgain test can be regarded as a random effects version of the usual principal components method . In particular the Bourgain model can be alternatively described as a model for the mean of conditional upon all eigenvectors of as ( 10 ) Now consider the coefficients as being independent random effects with mean zero and variances equal to times the associated eigenvalues , of . Averaging over all the random will yield the unconditional mean and variance covariance matrix . Our “smoothed” estimate , , of is motivated by expression ( 10 ) , and choosing a value of is analogous to choosing the number of eigenvectors to be used as fixed effects by EIGENSTRAT . The random effects framework also highlights the relationship between the Bourgain procedure and the genomic control method of Devlin and Roeder [25] . In genomic control one additional parameter that governs the dispersion of the test statistic is used to assess the association between and case-control status . In the smoothed version of the Bourgain test introduced here , we choose a total of such parameters . We have shown that if cases and controls come from genetically distinct populations but ones that have only recently diverged ( so that the parameter is very small ) then some limited power remains to detect true marker associations so long as the true value of is very large compared to the “typical” differences between cases and controls seen with the other markers . This is also analogous in interpretation to the genomic control method . However our explicit description of the upper bound on the noncentrality parameter of the association test for such a study clearly shows the limits of this design , and by implication , the limits of the genomic control procedure as well . Basically neither of these two methods behaves “properly” from a statistical point of view as sample size increases , i . e . the non-centrality parameter under an alternative hypothesis ( and hence power ) does not increase correspondingly . In genomic control the overdispersion parameter that the procedure corrects for increases with sample size , while for the Bourgain test the noncentrality parameter is bounded from above . For case-control studies involving two or more similarly admixed populations that differ in admixture fraction , the key issue in assessing the power of a study using cases from one population and controls from another is in determining the within-population heterogeneity of the admixture fractions , relative to the between population differences in average admixture . If the within-population heterogeneity is small then the situation is equivalent to the case of isolated populations , i . e . , there will be a bound on the power of a study to detect an effect with the bound determined by the upper limit on the noncentrality parameter as a function of as computed above . Despite the concerns raised in our theoretical considerations , in our assessment of the observed marker data from the AABC study we tentatively conclude that reuse of controls data from this study in future work may be statistically feasible . While there are clear differences in average admixture fraction between studies these are dwarfed by the within-study heterogeneity . Other signs of hidden structure in the AABC studies ( as evidenced by additional eigenvalues which are significant by the Tracy-Widom test ) do not appear to have a very large impact ( Figure 5 ) on the power of our hypothetical study using the CBCS and MEC cases and controls respectively . Control for the first few ( 1–200 in our case ) eigenvectors appears to dramatically reduce false positive associations with very little power loss ( about a 7 percent reduction in effective sample size ) relative to studies of homogeneous sets of cases and controls . We have used a specific set of ancestry informative markers in our analysis but the existence of genome-wide data for the AABC allows for considerable latitude in selecting SNPs to control for admixture , and even randomly selected SNPs , if enough are considered , can be used for admixture correction . Our use of the Bourgain test when considering the feasibility of a particular study design allows us to consider noncentrality parameters ( and hence power ) in particularly simple and helpful ways . While we have focused on the Bourgain method to correct for admixture differences in the AABC study our specific finding ( that little loss in power is anticipated when re-using control data from this study ) is likely to apply also to fixed-effects methods such as treating principal components or STRUCTURE estimates of percentage ancestry from ancestral populations as covariates . Our reasoning is based upon the close relationship between the principal components methods and the random effects rationale for the Bourgain test as given in equation ( 10 ) and also on the high correlation seen between STRUCTURE estimates of African ancestry in the AABC study and the first eigenvector from principal components .	This paper discusses and provides unique insight into an important problem raised by the current state of genetic studies into disease susceptibility , namely whether we can reuse genetic data from participants genotyped as controls in one study when cases ( people with a disease of interest ) are obtained from other studies , or whether each new study needs its own controls . We are interested in whether studies where cases and controls are sampled differently will give correct answers and are as powerful statistically as when new control data is also genotyped . Because of the huge investments made recently in large scale genotyping of cases and controls for various diseases , this is a timely question . This question is especially important in understanding the genetic causes of disease in as-yet relatively understudied population groups , such as African-Americans , in order to speed up progress when this is possible . We give theoretical results about the power of studies that reuse existing control genotypes based on statistical considerations . We also provide analysis of real data from a major study of the genetic causes of breast cancer in African-American women in order to shed practical light upon this issue .	[ "Abstract", "Introduction", "Methods", "Results", "Discussion" ]	[ "genetics", "and", "genomics/complex", "traits", "genetics", "and", "genomics", "genetics", "and", "genomics/population", "genetics" ]	2010	The Potential for Enhancing the Power of Genetic Association Studies in African Americans through the Reuse of Existing Genotype Data
NK cells are enriched in the liver , constituting around a third of intrahepatic lymphocytes . We have previously demonstrated that they upregulate the death ligand TRAIL in patients with chronic hepatitis B virus infection ( CHB ) , allowing them to kill hepatocytes bearing TRAIL receptors . In this study we investigated whether , in addition to their pathogenic role , NK cells have antiviral potential in CHB . We characterised NK cell subsets and effector function in 64 patients with CHB compared to 31 healthy controls . We found that , in contrast to their upregulated TRAIL expression and maintenance of cytolytic function , NK cells had a markedly impaired capacity to produce IFN-γ in CHB . This functional dichotomy of NK cells could be recapitulated in vitro by exposure to the immunosuppressive cytokine IL-10 , which was induced in patients with active CHB . IL-10 selectively suppressed NK cell IFN-γ production without altering cytotoxicity or death ligand expression . Potent antiviral therapy reduced TRAIL-expressing CD56bright NK cells , consistent with the reduction in liver inflammation it induced; however , it was not able to normalise IL-10 levels or the capacity of NK cells to produce the antiviral cytokine IFN-γ . Blockade of IL-10 +/− TGF-β restored the capacity of NK cells from both the periphery and liver of patients with CHB to produce IFN-γ , thereby enhancing their non-cytolytic antiviral capacity . In conclusion , NK cells may be driven to a state of partial functional tolerance by the immunosuppressive cytokine environment in CHB . Their defective capacity to produce the antiviral cytokine IFN-γ persists in patients on antiviral therapy but can be corrected in vitro by IL-10+/− TGF-β blockade . NK cells constitute a major cellular arm of the innate immune system and , as such , have been viewed as most relevant in the setting of the initial response to an acute infection . However , they may also be appropriately or inappropriately activated to exert effector function when persistent infection and its pathological sequelae become established . Their role may be particularly important in patients with CHB , in whom the virus-specific CD8 T cell arm of protection is markedly diminished and dysfunctional [1] , [2] . NK cells are greatly enriched in the liver , the site of HBV replication[3] , [4] . We have previously demonstrated an increase in activated CD56bright NK cells in the livers of patients undergoing flares of eAg-negative CHB . This subset can be induced to express TNF-related apoptosis-inducing ligand ( TRAIL ) , which is able to kill hepatocytes that have upregulated death-inducing TRAIL receptors , thereby contributing to liver inflammation in CHB[4] . The CD56bright subset can also be a potent source of cytokines such as IFN-γ[5] , [6] , a key cytokine shaping adaptive immunity and the delicate balance between protective and pathogenic responses . IFN-γ can clear HBV-infected hepatocytes through non-cytolytic mechanisms[7] , [8] . NK cell-derived IFN-γ could therefore constitute a vital antiviral mechanism in the liver , where hepatocytes are relatively resistant to the cytolytic mechanisms of perforin and granzyme production[9] . The intensity and quality of NK cell effector function is determined by the balance of activatory and inhibitory signals through their array of receptors ( NK-R ) , in addition to the influences exerted by the cytokine microenvironment . The TRAIL pathway of NK cell-mediated hepatocyte killing can be driven by the cytokines IFN-α and IL-8 , induced during flares of CHB[4] . Similarly , NK cells in HCV infection can be polarised towards cytolysis and expression of TRAIL as a result of exposure to endogenous[10] or therapeutic[11] IFN-α . Conversely , intrahepatic NK cell function can be down-regulated by the immunosuppressive cytokine IL-10 produced by Kupffer cells[12] . In addition , a role for IL-17 in curtailing NK cell function was recently demonstrated in disseminated vaccinia virus infection of mice with pre-existing dermatitis[13] . In this study we have investigated cytokine-driven modulation of IFN-γ production by NK cells in patients with CHB and explored the potential to restore their non-cytolytic antiviral function . To explore NK cell effector potential in the setting of persistent HBV infection , we first analysed the frequency of CD56bright ( CD16dim/neg ) and CD56dim ( CD16pos ) NK cell subsets in 64 patients with CHB compared to 31 healthy age-matched controls ( Table 1 ) . The proportion of circulating CD56bright NK cells was significantly increased in patients with CHB ( representative FACS plots Fig1a , summary data Fig1b ) , with a tendency to further increases in those with liver inflammation ( Fig1b ) . There was a trend for the percent of circulating NK cells to decrease in CHB ( Fig 1c ) but the absolute number of circulating CD56bright NK cells was still significantly increased ( p<0 . 05 data not shown ) . To determine whether there was a further enrichment of this immunoregulatory CD56bright NK cell subset at the site of viral replication , we compared the proportions in intrahepatic and circulating lymphocytes . In all eight patients with CHB from whom paired samples were available , the percent of CD56bright of total NK cells was higher in the intrahepatic compared to peripheral compartment ( Fig1d , e ) . Since NK cells make up a significantly greater proportion of intrahepatic than circulating lymphocytes in these patients ( Fig 1f ) , this corresponds to a substantial enrichment of CD56bright NK cells in the liver . We have previously shown that the CD56bright subset of NK cells can mediate hepatocyte apoptosis through expression of the death ligand TRAIL in flares of eAg-negative CHB[4] . In this cohort of patients we confirmed an increase in TRAIL expression ( largely on the CD56bright subset , Fig 2a representative plots ) in patients with either eAg+ or eAg- CHB who had evidence of liver inflammation ( Fig2a summary data ) . The CD56bright subset of NK cells can also be a potent source of IFN-γ[14] , a cytokine that has direct non-cytolytic antiviral effects on HBV replication [7] , [8] and can promote adaptive immune responses[6] . Despite the enrichment of CD56bright NK cells in CHB , we found that they had an impaired capacity to produce IFN-γ ( representative plots , Fig2b ) . There was a significant reduction in production of IFN-γ by NK cells from 46 patients with CHB compared to 29 healthy controls ( Fig2b ) . This reduction was seen irrespective of disease activity ( liver inflammation Fig2b , viral load or eAg status , data not shown ) or method of NK cell stimulation ( IL-12/IL-18 ( Fig2b ) , IL-12/IL-15 , K562 with IL-12/IL-18 or PMA/ionomycin , data not shown ) . Both the CD56bright subset and the CD56dim subset ( that has recently been recognised to also make a contribution to cytokine production[15] ) showed significantly impaired IFN-γ production ( FigS1a ) . Similarly , CD56bright and CD56dim NK cells in CHB showed a trend to produce less TNF-α , despite the strong stimulus required to reliably elicit this cytokine ( FigS1b ) . Simultaneous assessment of IFN-γ and TNF-a production showed a significant reduction in dual producing NK cells in CHB ( FigS1c ) . To assess NK cell cytolytic potential , we determined their capacity to degranulate as evidenced by CD107 expression following stimulation with K562 target cells and cytokines . There was no significant difference in NK cell degranulation potential in 33 patients with CHB compared to 21 controls ( Fig2c ) . Differential analysis by NK cell subset or by patient disease status did not show any differences ( data not shown ) . NK cells in CHB were therefore biased towards cytolytic and death-ligand mediated effector functions and defective IFN-γ production . To determine the potential of potent antiviral treatment to correct this bias in NK cell effector function , we studied a group of 22 patients with HBV viraemia well-suppressed on a combination of Lamivudine and Adefovir . Upon viral suppression and normalisation of liver inflammatory markers , there was no significant change in the percent of NK cells ( FigS2a ) , but the proportion of CD56bright NK cells decreased to levels observed in healthy controls ( Fig2d ) ; in line with this , NK cell TRAIL expression reduced to baseline levels ( Fig2d ) . However NK cell IFN-γ production was only partially augmented upon antiviral treatment ( mainly CD56dim subset , FigS2b ) and remained significantly lower than that in healthy controls ( Fig2d ) . Effector function of NK cells is tightly regulated by the cytokine milieu and their production of IFN-γ can be inhibited by immunosuppressive cytokines such as IL-10[12] , [16] and IL-17[13] . The levels of IL-17A were not elevated in sera from patients with CHB compared to controls ( Fig3a ) . In contrast , circulating concentrations of IL-10 were significantly increased in patients with active HBV disease ( Fig3b , c by CBA , confirmed by ELISA , data not shown ) , correlating with viral load ( r = 0 . 48 , p = 0 . 002 ) and ALT ( r = 0 . 37 , p = 0 . 03 ) . IL-10 levels showed a trend to decrease on antiviral treatment but remained significantly higher than in controls ( Fig3c ) , consistent with the limited restoration of NK cell IFN-γ production in these patients . To test whether IL-10 could induce the defect in NK cell IFN-γ production seen in CHB , we re-assessed NK cell effector function with or without the addition of exogenous IL-10 . IL-10 significantly suppressed NK-cell derived IFN-γ ( Fig3d ) , particularly in those patients in whom it was not already substantially reduced ( Fig3e , and in healthy controls , data not shown ) . By contrast , IL-10 had no effect on cytolytic ability or TRAIL phenotype ( Fig3f ) and did not affect the percent of NK cells ( FigS3a ) . The ability of IFN-α to further induce NK cell TRAIL expression in vitro[4] was also not abrogated by IL-10 ( data not shown ) . The effect of IL-10 was consistent but more modest on purified NK cells ( FigS3b ) , suggesting that some of its suppressive activity on NK cells is mediated indirectly via other constituents such as APCs . The contrasting effects of IL-10 on TRAIL and IFN-γ expression represented differential regulation of these effector functions in the same NK cells rather than the emergence of two distinct subsets . The small population of TRAIL-expressing NK cells present in healthy donors were at least as able to produce IFN-γ as the rest of the NK cell population ( FigS3c ) . The addition of exogenous IL-10 suppressed IFN-γ in NK cells regardless of their TRAIL expression ( FigS3c ) . In line with this , gating on the expanded population of TRAIL-expressing NK cells found in CHB demonstrated that their IFN-γ-producing capacity was no more reduced than that of the non-TRAIL-expressing fraction ( FigS3d ) . Since IL-10 was induced in CHB and exogenous IL-10 was able to mimic the selective suppression of NK cell effector function , we next investigated the potential to restore NK cell IFN-γ production by IL-10 blockade . Addition of antiIL10/IL10-R blocking mAbs restored the ability of both CD56bright and CD56dim NK cells from patients with active CHB to produce IFN-γ ( mean 2 . 5 fold increase , Fig4a , b , d ) . The majority of patients without biochemical evidence of liver inflammation ( and with low viral loads ) did not respond to this strategy ( Fig4c , d ) , in line with their lower levels of circulating IL-10 ( Fig3b ) . A subset of those patients failing to respond to IL-10 blockade did show recovery of NK cell IFN-γ production following blockade of both IL-10 and TGFβ , another immunosuppressive cytokine known to be able to inhibit NK cell production ( Fig4e , f ) . To investigate whether the suppression of NK cell IFN-γ was maintained at the site of HBV replication , paired liver and blood samples from eight patients with CHB were examined ( Table 2 ) . CD56bright NK cell IFN-γ production showed a trend to be even lower in the liver than the periphery of patients with CHB ( FigS4a ) . Levels of intrahepatic NK cell IFN-γ production did not significantly correlate with levels of ALT ( FigS4b ) , viral load or liver histology in this small sample of patients , only one of whom had histological evidence of significant liver inflammation ( Table 2 ) . Due to limited cell numbers , individual cytokine blockade could not be performed but dual IL-10/TGFβRII blockade reconstituted the proportion of NK cells able to produce IFN-γ ( %positive , Fig5a ) and increased their level of IFN-γ production ( MFI , Fig5b ) . The fold increase in the capacity of CD56bright NK cells to secrete IFN-γ upon IL-10/TGFβ blockade was greater in the liver than the periphery ( Fig5a , b ) . Accumulating evidence points to a contribution of NK cells in the battle to control persistent intracellular pathogens[6] , [17] , [18] . Although NK cells have been considered part of the innate immune response , recent data have suggested that they can possess properties previously ascribed to the adaptive arm , including the capacity to develop memory and tolerance[19] , [20] , [21] . In this study we show that NK cells can develop selective defects in antiviral function in the setting of chronic infection and inflammation , reminiscent of the hierarchical loss of effector function manifested by exhausted T cells[22] . Just as T cell defects have been attributed to excessive antigenic stimulation , functional impairment of NK cells has been ascribed to excessive stimulatory signals through the activating receptor NKG2D , resulting in its down-modulation[19] , [20] . This is a plausible mechanism in CHB since data from transgenic mice suggest that HBV can upregulate the intrahepatic expression of NKG2D ligands[23] . However , a recent study and our unpublished data do not support this mechanism , showing no down-regulation of NKG2D or consistent changes in other NK cell receptors that could account for the NK cell impairment seen in CHB[24] . Instead , our data suggest that the selective NK cell functional defects seen in this infection may be attributable to the immunosuppressive cytokine milieu . Our analysis of NK cell effector potential in a large cohort of patients with CHB revealed preservation of cytolytic capacity and an increase in TRAIL-bearing CD56bright NK cells . Despite this increase in the subset of NK cells that are usually the most potent source of cytokines[14] , there was a decrease in the overall NK cell capacity to produce IFN-γ . Such divergence of effector function is in line with the recent finding that cytokines are trafficked and secreted via completely different pathways to cytotoxic granules in NK cells[25] . Consistent with these distinct trafficking pathways , separate signalling pathways have been shown to control the release of cytokines and cytotoxic granules in NK cells[26] , [27] . Unique molecular switches are starting to be identified that couple NK cell receptor signalling with the generation of cytokines rather than cytotoxic functions[28] , [29] . It is therefore conceivable that a pathway specific to NK cell cytokine production is dysregulated in patients with CHB . The immunosuppressive cytokine IL-10 has been shown to specifically impair NK cell IFN-γ production[30] , in contrast with IL-17 and excessive NKG2D signalling , both of which result in down-modulation of all NK cell effector functions[13] , [20] . The liver is an immunotolerant organ , predisposed to the production of immunosuppressive cytokines; down-regulation of intrahepatic NK cell IFN-γ production has been linked to the local release of IL-10 by Kupffer cells[12] , [31] . We found that exposure of NK cells to IL-10 in vitro was able to recapitulate the selective reduction in IFN-γ production noted in patients with CHB . Furthermore , its blockade was able to restore the capacity of NK cells from patients with active HBV infection to produce IFN-γ . IL-10 was not able to inhibit cytotoxic degranulation and could not overcome the capacity of IFN-α to induce TRAIL , in line with the maintenance of these pathogenic functions of NK cells in CHB . IL-10 was consistently modestly elevated in the serum of patients with CHB , but would be expected to be at higher concentrations at the site of infection in the liver and in close proximity to the cells from which it is released . NK cells themselves can produce IL-10[14] , [32] to allow auto-suppression , but in the HBV-infected liver there are a number of other candidate cellular sources and there is likely to be a complex regulatory network involved in maintaining its production , as recently described in HIV infection[33] . We recently reported a transient induction of IL-10 in early acute HBV infection that was temporally associated with a transient suppression of the capacity of NK cells to produce IFN-γ , coincident with the increase in viraemia and production of viral antigens[16] . In our cohort of patients with CHB it was difficult to distinguish the influence of viraemia or liver inflammation , since both were increased in patients with elevated levels of IL-10 . Future study of a group of patients with high viral load but normal ALT ( immunotolerant phase ) could help to dissect the role of these factors . The fact that NK cell IFN-γ production and IL-10 levels were not significantly normalised by potent antiviral therapy suggests that the continued secretion of high levels of HBV proteins in these patients may play a role . In patients with low level CHB without evidence of liver inflammation , IL-10 was not elevated and its blockade alone could not rescue NK function , which instead required additional TGF-β blockade . TGF-β is another immunosuppressive cytokine that characterises the tolerising liver environment and has been shown to be increased in CHB[34] . TGF-β has been shown to be an alternative key regulator of the capacity of human NK cells to produce IFN-γ , suppressing IFN-γ and T-bet via Smad2/3/4[35] . The collective action of TGF-β and IL-10 may represent an important feedback mechanism to limit exuberant immune responses and tissue immunopathology in a vital organ like the liver . However , in the context of chronic infections , elevated levels may attenuate immune responses sufficiently to contribute to the failure of resolution of infection . A role for IL-10 in persistent viral infection has been highlighted recently by studies showing that blockade of the IL-10 receptor is associated with resolution of LCMV infection[36] , [37] . Genetic studies have also highlighted the importance of IL-10 in the antiviral response to HBV; polymorphisms of the IL-10 promoter resulting in elevated IL-10 production are associated with viral persistence , increased disease severity and progression[38] , [39] . Our data suggest that immunosupressive cytokines may polarise NK cells in CHB , having no effect on their expression of death ligands and cytolytic granules but inhibiting IFN-γ production . NK cells expressing death ligands like TRAIL would only be able to have a direct antiviral effect at the expense of liver damage . The decline in liver inflammation seen on antiviral treatment is compatible with the reduction in TRAIL-expressing CD56bright NK cells that we noted in this setting . However , potent antiviral therapy was unable to significantly restore the capacity of NK cells to produce IFN-γ , which would therefore retain an impaired capacity for non-cytolytic clearance of HBV from hepatocytes and boosting of adaptive immune responses . Our findings raise the possibility of immunotherapeutic targeting of IL-10 and TGF-β in CHB , with the caveat that these cytokines govern a critical balance between impeding pathogen clearance and restraining immunopathology . Clinical assessment and blood sampling were performed during routine hepatitis clinics , with written informed consent and local ethical board approval of the Royal Free Hospital , the Royal London Hospital and Camden Primary Care Ethics Review Board . All patients were anti-Hepatitis C- and anti-Human Immunodeficiency Virus-antibody negative and treatment naïve with the exception of a sub-group of 22 patients suppressed on a combination of Lamivudine and Adefovir . Patient characteristics are included in Table 1 . Paired peripheral blood and liver biopsy specimens ( surplus to diagnostic requirements ) were obtained from 8 CHB-infected patients ( Table 2 ) . Peripheral blood mononuclear cells ( PBMC ) were isolated by gradient centrifugation on Ficoll-Hypaque and frozen or immediately studied as described later . Sera were collected and frozen for later use . Intrahepatic lymphocytes were isolated as previously described[4] . For phenotypic analysis , PBMC isolated from HBV patients and healthy donors were stained with fluorochrome-conjugated antibodies to CD3-Cy5 . 5/PerCP , CD56-FITC , CD16-APC , and TRAIL-PE or isotype matched controls ( BD Biosciences , Cowley , U . K . ) . In selected experiments TRAIL expression was determined following overnight incubation with 50 ng/mL of rhIL-10 ( eBioscience ) . PBMC were acquired on a FACS Calibur flow cytometer ( Becton Dickinson ) and analysed using Flowjo analysis software ( Treestar ) . As previously described[16] , PBMC were incubated with 50 ng/mL of rhIL-12 ( Miltenyi ) and rhIL-18 ( R&D Systems , Abingdon , U . K . ) for 21 hours at 37°C . 1mM monensin ( Sigma-Aldrich , Gillingham , U . K . ) was added for the final 3 hours . Cells were fixed and permeabilised followed by intracellular staining for IFN-γ-PE ( R&D systems ) . Where indicated the same experiments were performed in the presence of rhIL-10 ( 50ng/mL ) , or blocking antibodies to anti-IL10 ( 5 µg/mL ) ( eBioscience ) and anti-IL-10R ( 10 µg/mL ) alone or in combination with antiTGFβRII ( 10 µg/mL ) ( BD Biosciences ) . NK IFN-γ production was determined by subtracting baseline IFN-γ production from that observed after cytokine or antibody treatment . NK cells from PBMC of a randomly selected group of patients were isolated ( >96% purity and viability ) ( Miltenyi Biotec , Germany , NK isolation kit ) to assess the effect of exogenous IL-10 on IFN-γ production . For TNF-α production , PBMC were stimulated with phorbol myristate acetate ( PMA ) ( 3 ng/mL ) and ionomycin ( 100 ng/mL ) ( Sigma ) for 3 hours; 1mM monensin ( Sigma-Aldrich , Gillingham , U . K . ) was added for the final 2 hours . Cells were then stained with the same antibody combination used for phenotyping prior to permealisation and intracellular staining for TNF-α . In selected experiments NK cell TNF-α and IFN-γ co-expression was assessed following PMA/I stimulation . As previously described[16] , PBMC were incubated with K562 cells ( 5∶1 E:T ratio ) for 3 hours at 37°C following overnight stimulation with a combination of rhIL-12/rhIL-18 or medium alone in the presence or absence of rh-IL10 . CD107a-PE antibody ( BD Biosciences , Cowley , U . K . ) was added at the time of stimulation with target cells and 1mM monensin was added during the last two hours of the incubation prior to staining and acquisition . CBA flex-sets were used for the determination of IL-10 , IL-17 ( BD Biosciences , Cowley , U . K ) according to manufacturers’ protocols for serum samples . Statistical significance was performed between paired samples using the Wilcoxon signed rank test and between HBV patients and healthy controls using the Mann-Whitney U test . Correlations between variables were evaluated with the Spearman rank correlation test . P<0 . 05 was considered to be significant for all tests .	Hepatitis B virus ( HBV ) infection is responsible for more than a million deaths annually as a result of the immune-mediated chronic liver damage it induces . One of the key immune players in the liver is the natural killer ( NK ) cell , which we have recently found can cause liver damage in HBV infection . Here we address the antiviral potential of NK cells in the HBV-infected liver and demonstrate that they have a specific impairment in their ability to produce the cytokine IFN-γ , which could limit their capacity to control HBV . We find that the potent antiviral drugs currently being used to treat HBV infection are unable to fully reverse this NK cell functional defect . We define a role for the immunosuppressive cytokine environment in HBV in down-regulating NK cell antiviral function , which can be restored by specific blockade of IL-10 and TGF-β . This work therefore highlights a mechanism contributing to the failure of immune control in chronic HBV infection , paving the way to new therapeutic options .	[ "Abstract", "Introduction", "Results", "Discussion", "Materials", "and", "Methods" ]	[ "virology/host", "antiviral", "responses", "infectious", "diseases/viral", "infections", "gastroenterology", "and", "hepatology/gastrointestinal", "infections", "immunology/innate", "immunity" ]	2010	Blockade of Immunosuppressive Cytokines Restores NK Cell Antiviral Function in Chronic Hepatitis B Virus Infection
Achieving facile specific recognition is essential for intrinsically disordered proteins ( IDPs ) that are involved in cellular signaling and regulation . Consideration of the physical time scales of protein folding and diffusion-limited protein-protein encounter has suggested that the frequent requirement of protein folding for specific IDP recognition could lead to kinetic bottlenecks . How IDPs overcome such potential kinetic bottlenecks to viably function in signaling and regulation in general is poorly understood . Our recent computational and experimental study of cell-cycle regulator p27 ( Ganguly et al . , J . Mol . Biol . ( 2012 ) ) demonstrated that long-range electrostatic forces exerted on enriched charges of IDPs could accelerate protein-protein encounter via “electrostatic steering” and at the same time promote “folding-competent” encounter topologies to enhance the efficiency of IDP folding upon encounter . Here , we further investigated the coupled binding and folding mechanisms and the roles of electrostatic forces in the formation of three IDP complexes with more complex folded topologies . The surface electrostatic potentials of these complexes lack prominent features like those observed for the p27/Cdk2/cyclin A complex to directly suggest the ability of electrostatic forces to facilitate folding upon encounter . Nonetheless , similar electrostatically accelerated encounter and folding mechanisms were consistently predicted for all three complexes using topology-based coarse-grained simulations . Together with our previous analysis of charge distributions in known IDP complexes , our results support a prevalent role of electrostatic interactions in promoting efficient coupled binding and folding for facile specific recognition . These results also suggest that there is likely a co-evolution of IDP folded topology , charge characteristics , and coupled binding and folding mechanisms , driven at least partially by the need to achieve fast association kinetics for cellular signaling and regulation . Cellular signaling and regulation are frequently mediated by proteins that , in part or as a whole , lack stable structures under physiological conditions [1]–[3] . Such intrinsically disordered proteins ( IDPs ) are highly prevalent in proteomes [4] and over-represented in diseases pathways [5] , [6] . For example , nearly one-third of eukaryotic proteins have been predicted to contain extended disordered regions [7] , and about 25% of disease-associated missense mutations can be mapped into predicted disordered regions [8] ( although cancer mutations appear to prefer ordered regions [9] ) . The prevalence of intrinsic disorder suggests that protein conformational heterogeneity could provide crucial functional advantages , for which many concepts have been proposed [10]–[14] . Understanding the physical basis of how intrinsic disorder mediates protein function ( and how such functional mechanism may fail in human diseases [15] ) is of fundamental significance and has attracted intense interests in recent years [16] . Important progresses have been made on characterizing the conformational properties of unbound IDPs and determining how these conformational properties contribute to efficient and reliable interactions [16]–[22] . A key recent recognition is that frequent requirement of protein folding for specific recognition of IDPs could lead to kinetic bottlenecks [23]–[25] . As predicted by the dual-transition-state theory [23] , the diffusion-limited encounter rate constant represents the upper bound for that of a coupled binding and folding interaction . Importantly , the upper bound can be achieved only if the IDP readily folds upon encounter , which requires folding rates on the order of 10 µs−1 or greater [23] . That is , IDPs need to achieve folding rates beyond the typical µs−1 “speed limit” estimated for folding of isolated proteins [26] to maximize association kinetics . Therefore , the putative functional advantages of intrinsic disorder , especially structural plasticity for specific interactions with numerous partners [27] , come with a potential cost of slow binding kinetics . Such kinetic bottleneck must be resolved for IDPs to be viable in cellular signaling and regulation . Interestingly , a recent survey of binding kinetic data revealed that IDP binding was not systematically slower than that of globular proteins [28] . The implication is that most IDPs do manage to fold rapidly upon nonspecific binding , and this is apparently consistent with the accumulating observations that IDP coupled binding and folding tends to follow induced folding-like baseline mechanisms ( i . e . , bind then fold ) [16] , [19] . Several factors could contribute to efficient folding of IDPs upon binding , in particular small interacting ( and folding ) domains and simple folded topologies with low contact orders . There also appears to be a delicate balance between pre-folding and conformational flexibility that allows an IDP to quickly fluctuate among accessible conformational states , especially upon encounter [16] , [29] , [30] . Nonetheless , it is not yet clear how in general IDPs may achieve fast folding at rates beyond the traditional µs−1 folding “speed limit” upon encountering their specific targets . An important characteristics of IDPs is that they are enriched with charged and polar residues [31] . Electrostatics can thus be expected to play key roles in IDP structure and function . For example , the charge content can modulate compaction and other conformational properties of free IDPs [32] , [33]; DNA search efficiency is controlled by charge composition and distribution in disordered tails of DNA-binding proteins [34] , [35] . It has been also observed or speculated in a few cases that electrostatics might be important for fast IDP recognition [36]–[39] . However , these discussions have been often based on the classic electrostatic steering effects [40] , and the actual underlying mechanisms of putative electrostatic acceleration were not known . Our recent computational and experimental study of the p27-Cdk2/cyclin A interaction revealed that long-range electrostatic forces could promote facile IDP recognition via an “electrostatically accelerated encounter and folding mechanism” [24] . Specifically , the measured p27/Cdk2/cyclin A association rate constants showed a strong salt-dependence , increased ∼12 fold when the ionic strength was reduced from 0 . 6 to 0 . 075 M . However , the salt-dependence is poorly described by an approximate Debye-Hückel relation [41] that mainly captures the electrostatic steering effects . Instead , simulations using a series of topology-based coarse-grained models suggested that long-range electrostatic forces exerted on a large number of charges on p27 did not only accelerate the encounter rate ( via the classical electrostatic steering effect [40] ) , but enhance the efficiency of p27 folding upon encounter by promoting native-like encounter topologies . Analysis of surface charges in a set of existing IDP complexes further revealed that the vicinity of IDP binding sites tended to be enriched with charges to complement those on IDPs [24] ( even though the IDP binding interface itself is more hydrophobic than the rest of the protein surface as previously observed [42] ) . Electrostatic forces are known to be a dominant long-range force that can guide protein orientation in protein-DNA interactions [43] , [44] and/or modulate early stages of protein folding [45]–[47] . One implication of enriched charges near IDP binding sites is thus that the electrostatically accelerated encounter and folding mechanism observed for p27 may be prevalent in signaling and regulatory IDPs . Nonetheless , the ability for long-range electrostatic forces to enhance folding upon binding can be surprising , as nonspecific interactions ( electrostatic or van der Waals ) have been generally expected to accelerate binding but slow down folding [48] , [49] . It has also been predicted that , while inter-chain electrostatic interactions facilitate binding of disordered chaperone Chz1 to histone variant H2A . Z-H2B , intra-chain electrostatic interactions could lead to premature collapse of Chz1 under low salt conditions and hinder the overall rate of forming the specific complex [50] . In the present work , we investigated the recognition mechanisms and the roles of long-range electrostatic interactions in forming of three IDP complexes , namely , p53-TAD1/TAZ2 , HIF-1α/TAZ1 , and NCBD/ACTR ( Table 1 ) . All these complexes have important biological functions . For example , tumor suppressor p53 is considered one of the most important proteins in cancer [51]; NCBD and TAZ1/2 are key regulatory domains of CBP , a key component of the general transcriptional machinery that plays critical roles in cell fate regulation [52] . For understanding IDP recognition , these systems involve more complex folded topologies than that of p27 in the p27/Cdk2/cyclin A complex . As shown in Fig . 1 , both HIF-1α/TAZ1 and NCBD/ACTR possess extensive binding interfaces , whereas the binding interface in p53-TAD1/TAZ2 is more localized . Importantly , while strong charge complementary exists near the binding interface ( as expected ) , the surface electrostatic potentials of the folded substrates do not show prominent features like those observed on Cdk2/cyclin A ( e . g . , see Fig . 1 of reference [24] ) to directly suggest that long-range electrostatic forces could promote native-like ( and thus more folding-competent ) encounter complexes . The NCBD/ACTR complex involves synergistic folding of two IDPs and thus offers a particularly interesting opportunity to understand whether and how electrostatic interactions may modulate the formation of nontrivial folded topologies . Amazingly , all three complexes associate with on-rates in excess of 107 M−1s−1 ( see Table 1 ) , a regime that is typically considered “diffusion-limited” and can only be accessed in the limit of ultrafast conformational transitions [40] . Series of topology-based coarse-grained models were first derived based on the complex structures to allow direct simulation of reversible binding and folding with tractable computational cost . Topology-based modeling is based on the theoretical framework of minimally frustrated energy landscapes for natural proteins [53] , and has been highly successful in predicting essential features of protein folding mechanisms [53]–[55] . Formation of stable IDP complexes such as those studied in this work should also satisfy minimal frustration , and thus topology-based modeling is applicable . Indeed , it has been successfully applied to several IDP complexes [56]–[60] , with many key predictions substantiated by independent experimental studies . Nonetheless , important differences do exist between IDPs and structured proteins in sequence compositions and binding interface characteristics [42] . We have previously demonstrated that traditional topology-based models need to be carefully calibrated to ensure proper balance among competing intramolecular and intermolecular interactions ( see Methods for detail on the calibration protocol ) [61] . We note that the importance of model calibration was also illustrated in a recent study of the HIF-1α/TAZ1 complex [59] . Table 2 summarizes the final calibrated models for all three complexes . The calculated residual helicity distributions of the unbound states are show in Fig . S1 . Three independent models were constructed for each complex: one without explicit charges ( mimicking high salt concentration with fully screened long-range electrostatic interactions ) , one with explicit charges ( mimicking low salt concentration with unscreened long-range electrostatic interactions ) , and a third one with explicit charges and 0 . 05 M salt ( mimicking physiological conditions ) . All models reproduce the experimental KD to the same order of magnitude , except that the no charge model for HIF-1α/TAZ1 yields a KD value about one order of magnitude too large . We note that calculated KD values can be very sensitive to small changes of in the scaling of intermolecular interactions during model calibration ( see Methods ) . It is computationally expensive to use REX simulations to systematically search for the parameter space , especially for models without explicit charges due to slower transitions . Nonetheless , by performing production simulations at the corresponding melting temperatures , remaining imperfections in the balance of various interactions should be further suppressed , allowing reliable comparative studies of the mechanistic roles of electrostatic interactions in coupled binding and folding . Free energy surfaces were constructed using various combinations of folding and binding order parameters to understand the baseline mechanisms of coupled binding and folding and to dissect the effects of long-range electrostatic forces . In particular , the fractions of native contacts formed have been shown to provide natural reaction coordinates for such mechanistic analysis [62] . Fig . 2 compares the free energy surfaces as a function of intra- and inter-molecular native contact factions for all three complexes , calculated using calibrated Gō-like models with and without explicit charges and/or salt ( see Table 2 ) . Both p53-TAD1 and HIF-1α recognitions follow induced folding-like mechanisms , where the peptides only gain structures after forming significant numbers of native intermolecular contacts . For example , Fig . 2A shows that p53-TAD1 does not start to fold until Qinter reaches ∼0 . 5 . Free NCBD is a molten globule with folded-like secondary structures [63] , and its synergistic folding with ACTR has been previously shown to involve multiple stages of selection and induced folding [25] , [60] , reminiscent of the “extended conformational selection” mechanism [30] . Nonetheless , neither protein gains significant secondary ( for ACTR ) or tertiary ( for NCBD ) structures until over 20% of native intermolecular contacts are formed ( Fig . 2G and 2J ) . Interestingly , formation of all three complexes involves intermediates , even though the intermediate in p53-TAD/TAZ2 interaction only become pronounced in the presence of nonspecific electrostatic forces ( see Fig . 2A vs 2C ) . Detailed examination of the simulation trajectories and various free energy surfaces using fractions of native contacts formed by different IDP segments ( e . g . , see Figs . S2 , S3 , S4 ) revealed the existence of multiple parallel pathways for forming HIF-1α/TAZ1 and NCBD/ACTR . While these mechanistic details are not the focus of the current work , they appear to be highly consistent with previous experimental and computational studies . For example , as shown in Fig . S2 , both the first and third helices of HIF-1α could initiate recognition , with the pathway initiated by the third helix binding being much more prevalent . Similar observations were also made in a separate computational study [59] . Specific recognition of NCBD/ACTR appears to be primarily initiated by the C-terminal segments of these two peptides ( Figs . S3 , S4 ) , which forms a key intermediate that was also suggested by an H/D exchange mass spectrometry study [64] . Kinetic data from a recent stop-flow study of the NCBD/ACTR interaction [65] are consistent with the prediction of induced folding as a baseline mechanism and have further confirmed the existence of parallel pathways and multiple folding intermediates . Representative snapshots along the dominant binding and folding pathways of p53-TAD1/TAZ2 and HIF-1α/TAZ1 are shown in Figs . S5 , S6 . Explicit inclusion of charges does not significantly perturb the baseline mechanisms of coupled binding and folding . As shown in Fig . 2 and Figs . S2 , S3 , S4 , long-range electrostatic forces do not lead to fundamental changes in any of the free energy surfaces examined . The baseline mechanisms for the formation of all three complexes remain induced folding-like . Furthermore , nonspecific electrostatic interactions do not change the relative prevalence of the parallel pathways that exist . For example , HIF-1α still initiates binding mainly through the third helix ( Fig . S2 ) ; synergistic folding NCBD and ACTR is still mainly initiated through their C-terminal segments ( Figs . S3 , S4 ) . The key effect of electrostatic forces appears to be substantial reductions in the free energy barriers that separate various basins . That is , even under the no salt condition , strong nonspecific electrostatic interactions do not appear to add to the ruggedness of coupled binding and folding free energy surfaces . An implication is that there exists a level of self-consistency between the charge distribution and folded topology in the bound states , despite a lack of apparent complementary between folding topologies and surface electrostatic potentials for these IDP complexes ( see Fig . 1 ) . Kinetics of coupled binding and folding was derived directly from production Langevin dynamics simulations performed using the calibrated Gō-like models at their corresponding Tm . The results , summarized in Table 2 , show that long-range electrostatic forces accelerate the reversible binding/unbinding transition rates for all three complexes . The overall electrostatic acceleration , estimated by comparing the average transition rates ( kTS ) calculated using models with and without explicit charges , ranges from ∼5 fold for HIF-1α to 10 fold for NCBD/ACTR . The magnitude of acceleration is similar to what was previously measured for other IDPs including p27 [24] and PUMA [39] ( both ∼10 fold ) . The presence of 0 . 05 M salt significantly attenuates the predicted electrostatic acceleration , to only about two fold . However , the effect of salt screening on electrostatic acceleration is likely over-predicted [24] , which is due to the Cα-only model used in this work and may be corrected with more detailed protein models [45] . Consistent with the kinetic analysis , there are significant reductions in the free energy barriers along Qinter ( see Fig . 3 ) , which has been shown to be a good binding reaction coordinate [61] . In addition , the magnitude of barrier reduction correlates well with the degree of rate acceleration calculated directly from Langevin dynamics simulations , with the largest barrier reduction observed for NCBD/ACTR and the smallest reduction observed from HIF-1α/TAZ1 . To further analyze the effects of electrostatic interactions on different stages of coupled binding and folding , the recognition process was divided into two generic steps , including an encounter step followed by an evolving ( folding ) step to final bound and folded state ( Eq . 1 in Methods ) . Such generic decomposition ignores the details of IDP-specific folding pathways , to allow on to focus on the net effects of electrostatic forces on the overall efficiency of IDP folding upon encounter . For this , three general states were identified during production simulations , including the unbound ( U ) , collision complex ( CC ) , and bound ( B ) states ( see Methods for specific criteria for state assignment ) . The mean first passage times ( MFPT ) and numbers of transitions ( ) among these states were then calculated . The results , summarized in Tables S1 , S2 , 3 , show that long-range electrostatic forces greatly reduce the average encounter time , from 0 . 72 to 0 . 03 ns for p53-TAD , from 0 . 37 to 0 . 20 ns for HIF-1α , and from 7 . 71 to 1 . 26 ns for NCBD . At the same time , long-range electrostatic forces also significantly enhance the efficiency of IDP folding upon encounter , allowing much larger fractions of the encounter complexes to eventually evolve to the bound states . For example , for NCBD/ACTR , only 16 out ∼2300 encounter events evolved to the bound state in absence of long-range electrostatic forces ( 0 . 7% ) ; whereas with explicit charges , there was ∼37% probability ( 108 out of 288 ) of forming the specific complex once the proteins were captured into the collision complex state ( Table S3 ) . For the HIF-1α/TAZ1 complex , the percentages of collision to specific complex transition are 0 . 4% without and 7% with explicit charges ( Table S2 ) ; for p53-TAD1/TAZ2 , the production percentages are 0 . 6% without and 60% with explicit charges ( Table S1 ) . It should be emphasized that nonspecific electrostatic interactions significantly stabilize the collision complexes , due to large and complementary net charges of the interacting proteins ( see Table 1 ) . As such , much fewer fully unbinding events were observed during production simulations using the charged models . This effect also led to more reversible transitions between the bound and collision complex states and thus an overestimation of the true folding efficiency of IDPs upon collision as estimated above . We also note that the collision complexes as defined in our analysis were not intended to represent so-called “encounter complexes” that have been often considered key intermediates of protein-protein association [66] , although encounter complexes are also believed to be mainly stabilized by nonspecific electrostatic interactions . The enhanced apparent efficiency of folding upon encounter appears to be frequently achieved at the cost of longer folding times . For example , the MFPTs of transitions from the collision complexes to the bound states increase from 0 . 26 to 3 . 94 ns for the p53-TAD1/TAZ2 complex ( Table S1 ) and from 8 . 14 to 44 . 56 ns for the NCBD/ACTR complex ( Tables S3 ) . The net effects on the kinetics of encounter and folding stages can be quantified by calculating three effective rate constants as defined in Eqns . 2–4 ( see Methods ) [28] . The results , summarized in Table 2 and plotted in Fig . 4 , clearly demonstrate that nonspecific electrostatic interaction enhance the encounter rates and reduce the escape rates of the collision complexes . Importantly , the effective evolution rates are always faster , by about three fold , in the presence of long-range electrostatic forces , despite longer MFPTs for the transitions from the collision complexes to the bound state observed for the p53-TAD1/TAZ2 and NCBD/ACTR complexes . The magnitude of electrostatic acceleration of folding upon encounter is similar to what was previously observed for folding and binding of p27 to the Cdk2/cyclin A complex [24] . Inspection of the conformational properties of the collision complexes provides further insights into the molecular basis for enhanced efficiency of IDP folding upon encounter due to long-range electrostatic forces . As shown in Fig . 5 , without nonspecific electrostatic interactions ( models without explicit charges ) , the initial contacts between two binding partners are largely random , and the distributions of IDP initial contact points on the substrate surface in the collision complexes are relatively uniform ( left column ) . In contrast , with the inclusion of explicit charges , the probabilities of IDP encountering near the native binding interface are dramatically increased . Coupled with reduced escape rates , this allows much higher efficiency of IDP folding upon encounter to achieve higher overall association rate constants ( Table 2 ) . The ability of long-range electrostatic forces to guide the recognition process is also reflected in the free energy surfaces as a function of binding RMSD of the IDP and center of mass separation between two peptides . As shown in Fig . 6 , long-range electrostatic forces generate a strong free energy gradient that extends over 10–15 Å away from the native bound positions , without creating over-stabilized misfolded states at short separation distances . It is intriguing that , even though both NCBD and ACTR are disordered in the unbound state , nonspecific long-range electrostatic forces between complementary charges on these two proteins can still manage to promote native-like topologies in the collision complexes . In particular , there is a much higher probability of NCBD and ACTR initiating contacts via the C-terminal helix of NCBD and the second helix of ACTR ( Fig . 5E–F ) . This is part of a key pathway of synergistic folding inherent to the NCBD/ACTR complex that was predicted by coarse-grained and atomistic simulations [25] , [60] and later substantiated by H/D exchange mass spectrometry [64] . Therefore , nonspecific electrostatic interactions appear to mainly augment existing folding pathways inherent to the folded topologies to facilitate efficient folding of IDPs upon encounter . Coupled with the previous observation that the vicinity of the IDP binding site tends to be enriched with charges to complement those on IDPs [24] , thee current results suggest that there is likely a co-evolution of IDP folded topology , charge characteristics , and coupled binding and folding mechanisms . Furthermore , the co-evolution is likely driven by the important need to achieve facile IDP recognition for cellular signaling and regulation . While fulfilling important functional constraints such as structural plasticity for binding numerous specific targets , protein intrinsic disorder can lead to potential kinetic bottlenecks to be viable in cellular signaling and regulation . Our previous work on the p27/Cdk2/cyclin A complex has revealed a mechanism where nonspecific electrostatic interactions not only enhance the protein-protein encounter kinetics but also promote folding-competent encounter topologies to increase the efficiency of IDP folding upon encounter [24] . Using carefully calibrated topology-based coarse-grained models , we have now further demonstrated that similar electrostatically accelerated encounter and folding mechanisms also underlie the formation of three IDP complexes with more complexed folded structures , namely , p53-TAD1/TAZ2 , HIF-1α/TAZ1 , and NCBD/ACTR . Importantly , these complexes lack apparent features on the electrostatic surface potentials to directly suggest the ability of nonspecific long-range electrostatic forces to promote native-like encounter topologies to enhance the IDP folding efficiency upon encounter . Nonetheless , there seems to exist a sufficient level of self-consistency between the charge distributions and folded topologies in the bound state to allow accelerated recognition in presence of nonspecific electrostatic interactions . Therefore , enriched charges on IDPs not only play key roles in modulating the conformational properties of the unbound state , but also likely play general and important roles in regulating efficient interactions of IDPs with specific partners . We note that IDPs are frequently regulated by post-translational modifications that add or remove charges . Improved mechanistic understanding of electrostatic forces in IDP recognition derived from the current work will thus help to dissect the profound impacts of post-translational modifications and disease-related mutations on IDP structure and interaction . Cα-only sequence-flavored Gō-like models [67] were first derived from the complex structures of p53-TAD1/TAZ2 , HIF1-α/TAZ1 and NCBD/ACTR ( see Table 1 ) using the Multiscale Modeling Tools for Structural Biology ( MMTSB ) Gō-Model Builder ( http://www . mmtsb . org ) [68] . The 3 zinc ions bound to TAZ1 in the HIF1-α/TAZ1 complex were modeled explicitly with distance restraints to the coordinating residues . All three models were then calibrated to balance the intrinsic folding propensity and the strength of intermolecular interactions using a previously described protocol [61] . Briefly , the strengths of intra-molecular native contact were uniformly scaled to reproduce the experimentally measured residual helicity of unbound IDPs , which are mainly based on NMR secondary chemical shift and/or circular dichroism analysis ( p53-TAD1 [69] , NCBD/ACTR [63] , and HIF1-α [70] ) . The residual helicity distributions calculated using the final models listed in Table 2 are provided in Fig . S1 . Then , the strengths of intermolecular contacts were adjusted , such that binding affinities calculated from replica exchange molecular dynamics ( REX-MD ) simulations approximately match the experimental values ( see Table 1 ) . Following the previously described procedure [24] , the calibrated sequence-flavored Gō-like models were then further modified by assigning proper explicit charges to all charged residues ( Lys , Arg , Glu and Asp ) as well as zinc ions in the HIF1-α/TAZ1 complex . The charged models were then re-calibrated to reproduce the experimental residual structure level ( Fig . S1 ) and binding affinity ( Table 2 ) . Such calibration is critical to avoid inherent bias for particular types of interactions , e . g . , intra- vs . inter-molecular or native vs . nonspecific electrostatic . Nonspecific electrostatic interactions were modeled using the Debye-Hückel potential to account for ionic screening . The dielectric constant was set at 80 . The complexes were simulated in cubic boxes with periodic boundary conditions imposed in CHARMM [71] , [72] . The box sizes are 100 , 100 and 105 Å for p53-TAD1/TAZ2 , HIF-1α/TAZ1 and NCBD/ACTR , respectively . Langevin dynamics was performed with 15 fs time steps and a friction coefficient of 0 . 1 ps−1 . SHAKE was used to fix all virtual bond lengths [73] . Non-bonded interactions were cut off at 25 Å . Unbound IDPs were simulated at 300 K for 750 ns to calibrate the intramolecular interactions . REX-MD was performed using the MMTSB Toolset [68] for calibration of the intermolecular interactions . For this , eight replicas spanning 270 to 400 K were used . The lengths of REX calibration simulations ranged from 1 . 05 µs ( for p53-TAD1/TAZ2 ) up to 10 µs ( for NCBD/ACTR ) , as needed for achieving sufficient convergence . Temperature weighted histogram analysis method ( WHAM ) [74] was used to compute the heat capacity ( CV ) curves and generate unbiased probability distributions for free energy and thermodynamic analysis . In particular , the dissociation constants ( KD ) were calculated from the bound and unbound probabilities at 300 K [61] , where the unbound state was defined as the state without any native intermolecular contacts formed . For NCBD/ACTR complex , the 1D free energy profile lack significant barriers between the unbound and partially bound intermediate states ( Fig . 3C , red trace ) . Therefore , the unbound probability was calculated as 1 – Pbound , where Pbound is the bound probability ( see below for the specific criteria of state assignments ) . Once calibrated , production simulations of 30–40 µs in lengths were performed using all models at the corresponding TM's ( see Table 2 ) . The TM value was first identified based on the CV curve and then fine tuned to ensure that similar probabilities of sampling the bound and unbound states were observed in the production simulation . All free energy profiles were calculated from the REX simulations and the kinetic analysis was performed based on the production simulations , unless otherwise stated . For calculation of contact fractions , a given native contact was considered as formed if the inter-Cα distance was within 1 . 0 Å of the distance in the native complex . Nonspecific intermolecular contacts are considered as formed when the inter-Cα distance is within 10 Å cutoff . Three general conformational states were defined for each complex , including the unbound ( U ) , collision complex ( CC ) and bound ( B ) states , to understand the effects of electrostatic forces on protein-protein encounter and subsequent folding upon encounter . The unbound state includes conformations with no specific or nonspecific contacts formed between IDP and substrate , and the collision complex state includes conformations with at least one nonspecific but no specific intermolecular contact formed . The bound states are defined as following: 1 ) for p53-TAD1/TAZ2: Ninter≥11; 2 ) for HIF-1α/TAZ1: Ninter≥26 for the no charge model , Ninter≥23 for the charged model , and Ninter≥24 for the charged model with 0 . 05 M salt; 3 ) for ACTR/NCBD: Ninter≥30 . Ninter is the total number of native intermolecular contacts formed . Note that slightly different criteria were used to define the bound state of HIF-1α/TAZ1 due to small shifts of the bound free energy basins calculated using different models ( see Fig . 3 ) . 15-ps running averages were used for assigning states , to avoid including fictitious transitions due to rapid small fluctuations in the calculated contact counts ( especially between the U and CC states ) . The overall on and off rates were calculated directly from the average lifetimes of the bound and unbound states ( see Table S4 ) . In addition , MFPTs and numbers of transitions among all three states were derived from the production simulation trajectories , and various rates were calculated as defined in Eqns . 2–4 . ( 1 ) ( 2 ) ( 3 ) ( 4 ) Here , kcap , kesc , and kevo are the capture , escape ( to the unbound state ) and evolution ( to the bound state ) rates of the collision complex , respectively; Nesc and Nevo are the numbers of escape and evolution transitions . Note that the MFPTs calculated correspond to the average times spent in an initial state before a transition to the final state . Ideally , the average lifetime of CC should be independent of whether the trajectory ends up in either the U or B state for a true three-state model as shown in Eq . 1 . However , the actual transitions between the CC and B states involve several intermediates that are not represented in Eq . 1 , and the effective MFPTs as calculated thus depend on both the initial and final states ( e . g . , see Tables S1 , S2 , S3 ) . Analytical expressions on similar MFPTs involved in amyloid fibril templating can be found a recent theoretical analysis by Schmit [75] . All molecular visualizations were prepared using VMD [76] .	Intrinsically disordered proteins ( IDPs ) are key components of regulatory networks that dictate various aspects of cellular decision-making . They are over-represented in major disease pathways , and are considered novel albeit currently difficult drug targets . Recognition of IDPs has extended the traditional protein structure-function paradigm , and various concepts have been proposed on how intrinsic disorder may confer crucial functional advantages . However , the physical basis of these concepts remains poorly established . In particular , while IDPs alone exist as ensembles of fluctuating structures , they frequently fold upon specific binding . Analysis of the physical timescales of protein folding and protein-protein encounter predicts that the requirement of peptide folding for specific binding could lead to a major kinetic bottleneck . In this work , carefully calibrated topology-based coarse-grained models were applied to directly simulate reversible folding and binding and investigate the recognition mechanisms of three IDP complexes . The results strongly support an electrostatically accelerated encounter and folding mechanism , where long-range electrostatic forces not only accelerate protein-protein encounter via “electrostatic steering” but also promote “folding-competent” encounter topologies to enhance the efficiency of IDP folding upon encounter .	[ "Abstract", "Introduction", "Results", "Discussion", "Methods" ]	[]	2013	Electrostatically Accelerated Encounter and Folding for Facile Recognition of Intrinsically Disordered Proteins
Between 20 and 30 bacteriologically confirmed cases of leprosy are diagnosed each year at the French National Reference Center for mycobacteria . Patients are mainly immigrants from various endemic countries or living in French overseas territories . We aimed at expanding data regarding the geographical distribution of the SNP genotypes of the M . leprae isolates from these patients . Skin biopsies were obtained from 71 leprosy patients diagnosed between January 2009 and December 2013 . Data regarding age , sex and place of birth and residence were also collected . Diagnosis of leprosy was confirmed by microscopic detection of acid-fast bacilli and/or amplification by PCR of the M . leprae-specific RLEP region . Single nucleotide polymorphisms ( SNP ) , present in the M . leprae genome at positions 14 676 , 1 642 875 and 2 935 685 , were determined with an efficiency of 94% ( 67/71 ) . Almost all patients were from countries other than France where leprosy is still prevalent ( n = 31 ) or from French overseas territories ( n = 36 ) where leprosy is not totally eradicated , while only a minority ( n = 4 ) was born in metropolitan France but have lived in other countries . SNP type 1 was predominant ( n = 33 ) , followed by type 3 ( n = 17 ) , type 4 ( n = 11 ) and type 2 ( n = 6 ) . SNP types were concordant with those previously reported as prevalent in the patients’ countries of birth . SNP types found in patients born in countries other than France ( Comoros , Haiti , Benin , Congo , Sri Lanka ) and French overseas territories ( French Polynesia , Mayotte and La Réunion ) not covered by previous work correlated well with geographical location and history of human settlements . The phylogenic analysis of M . leprae strains isolated in France strongly suggests that French leprosy cases are caused by SNP types that are ( a ) concordant with the geographic origin or residence of the patients ( non-French countries , French overseas territories , metropolitan France ) or ( b ) more likely random in regions where diverse migration flows occurred . Leprosy or Hansen’s disease is a chronic infectious disease caused by Mycobacterium leprae that the World Health Organization called to eliminate ( elimination being defined as the reduction of disease prevalence to <1 case per 10 000 population ) in the ‘90s [1] . The elimination of leprosy had been achieved at a global level in 2002 . However , the number of new cases reported worldwide remained stable around 200 000 since 2005 , and in 2013 , 44 countries still reported more than 100 cases annually [2] . One of the strategies proposed to reduce the disease burden is to detect leprosy cases as early as possible . In developing countries , the diagnosis remains exclusively based on clinical evidence ( e . g . pale or reddish patches on the skin and loss or decrease in sensation in the skin patches ) . The disease can be classified clinically , on the basis of the number of skin lesions , as paucibacillary ( 1 to 5 lesions ) or multibacillary ( more than 5 lesions ) [3] . In developed countries , bacteriological examination , histopathological or molecular methods such as PCR applied on slit-skin specimens can support the diagnosis [4] whereas these methods are more difficult to implement in developing countries . Autochthonous leprosy disappeared from all Northern and most Southern European countries . However , 20 to 30 leprosy cases are diagnosed every year at the French National Reference Center ( NRC ) for mycobacteria , mainly in immigrants from endemic countries or in people living in French overseas territories where leprosy is not eradicated [5] . Genotyping methods have been described to study the global and geographical distribution of distinct clones of M . leprae . Since the M . leprae genome is exceptionally well conserved and the bacillus is highly clonal , a correlation has been observed between the geographical origin of leprosy patients and SNP profiles [6] . We aimed at determining the SNP genotype of the M . leprae strains isolated in France . For this purpose , we applied the method described by Monot et al . [6] , as well as derived methods , for genotyping of the M . leprae isolates detected in France during the last 5 years . Biopsies containing M . leprae from 71 distinct patients received at the National Reference Center for mycobacteria between January 2009 and December 2013 were included in the study . These biopsies were sent from medical centers located in metropolitan France ( n = 25 ) , but also in French overseas territories ( New Caledonia ( n = 20 ) , Mayotte ( n = 19 ) , French West Indies ( Guadeloupe and Martinique ) ( n = 3 ) and French Polynesia ( n = 4 ) ) . The sex ratio ( male/female ) was 2 . 4 ( 50/21 ) . The average age was 40 ( range , 14–86 ) years . In all , 81 biopsies were examined that were comprised of one biopsy from 63 patients , three biopsies from two patients and two biopsies from six patients . The biopsies were stored at +4°C until DNA extraction which was performed within 3 days after the biopsy was received . Suspensions were prepared from skin biopsies as previously described [7 , 8] . Genomic DNA was extracted from 2-mL suspensions using the freeze-boiling technique [9] modified as follows: five cycles of heat-cold shocks ( 1 min at 100°C and 1 min at –196°C in liquid nitrogen ) were followed by 2 min of sonication ( Bransonic Model 1200 , Branson ) . The diagnosis of leprosy was confirmed by smear positivity for acid fast bacilli ( AFB ) , which were counted to establish an AFB count [10] , and/or the presence of M . leprae DNA , detected by RLEP PCR , as previously published [4 , 11] . In 78% of the biopsies M . leprae was detected with both methods . The RLEP PCR was slightly modified as follows: primers ( 0 . 4 μM each ) were added to beads containing dNTPS and Taq DNA polymerase ( Illustra puRe Taq Ready-To-Go PCR beads , GE Healthcare ) . DNA in the reaction mixture was denatured for 10 minutes at 94°C and amplified during 40 cycles of 1 min at 94°C , 1 min at 50°C and 1 min at 72°C with a final extension at 72°C for 10 min . DNA samples were stored at -80°C until further analysis . The samples were anonymized and used with Institutional Review Board approval for diagnosis of specimens received at Assistance Publique–Hôpitaux de Paris , Biology laboratories of Pitié-Salpêtrière Hospital . DNA samples used for SNP genotyping were those used for RLEP PCR , except in the case of 15 patients where DNA was newly extracted since no amplification was obtained with thawed DNA previously used for RLEP PCR . The SNP types at positions 14 676 , 1 642 875 and 2 935 685 ( reference TN strain numbering system ) were assessed using the original method of Monot et al . [6] and , when unsuccessful , using a modified protocol and primers in order to increase the efficiency of the original method [6] . Briefly , the following parameters were sequentially modified by: ( i ) decreasing the annealing temperature from 55°C to 48°C and the number of cycles from 45 to 30; ( ii ) designing new primers for nested PCR based on a combination of the original [6] and the new primers ( Table 1 ) . Crude extract ( 5 μL ) was added to 0 . 5 μL of Taq DNA polymerase ( 1 U/μL ) ( bioXact long DNA polymerase , Bioline ) in 2 . 5 μL of Taq buffer containing 5 μL of dNTP mixture ( 2 . 5 mM ) ( Sigma-Aldrich ) , 2 . 5 μL of 50 mM MgCl2 and 5 μL of each primer ( 4 μM ) . This reaction mixture was denatured at 94°C for 10 minutes followed by amplification consisting in 45 cycles of 1 min at 94°C for the first PCR and 30 cycles for nested PCR , 1 min at various annealing temperatures ( Table 1 ) , 2 min at 68°C and a final extension at 68°C for 10 min . In case of nested PCR , products of first PCR was applied to second PCR without dilution . After DNA amplification , unincorporated nucleotides and primers were removed by filtration through the membrane of a 96-well MultiScreen Filter Plate ( Merck Millipore Ltd . ) placed onto a MultiScreen HTS Vacuum Manifold connected to a vacuum pump . Sequencing reactions were carried out in a final volume of 10 μL , with 0 . 8 μL of Big Dye Sequencing RR-100 ( Big Dye Terminator Cycle Sequencing Ready reaction kit ( Applied Biosystems ) ) , 2 μL of BigDye sequencing buffer 5x , 0 . 6 μL of forward or reverse primers 4 μM and 2 μL of purified PCR product . The cycle sequencing was carried out on a thermal cycler using the following conditions: 1 min of initial denaturation at 96°C , 25 cycles of the following program segment: 10 sec at 96°C , 5 sec at 50°C , 4 min at 60°C . Sequencing products were purified with a Sephadex G-50 ( GE Healthcare ) gel filtration system . Sequences were compiled and analyzed using BioEdit software , as previously described [12] . Four SNP types were identified in this study . Applying the original method of Monot et al . [6] allowed their identification in 22/71 isolates , decreasing the annealing temperature and the number of cycles resulted in the characterization of an additional 22 isolates and nested PCR allowed the characterization of a further 23 isolates . Overall , 67/71 isolates ( 94% ) were successfully genotyped . For the 4 isolates unsuccessfully genotyped , even newly extracted DNA did not allow SNP genotyping . Thirty-five of the 67 patients ( 52% ) for whom the M . leprae strain was successfully genotyped were born in French overseas territories: five in Mayotte , 21 in New Caledonia , four in French Polynesia , four in French West Indies and one in La Réunion . Twenty-eight patients ( 42% ) were immigrants living in France: 16 from Comoros , two from Guinea , and one each from Mali , Ivory Coast , Benin , the Democratic Republic of the Congo ( DRC ) , the Republic of the Congo ( RC ) , Madagascar , Haiti , Brazil , India and Sri Lanka . The four remaining patients ( 6% ) were born in metropolitan France; two of them are living or have lived for a long time in RC or DRC , one has been travelling frequently in India and one was a humanitarian worker in many countries , including leprosy endemic areas . SNP type 1 was predominant ( 33 cases ) , followed by SNP type 3 ( 17 cases ) , SNP type 4 ( 11 cases ) and SNP type 2 ( 6 cases ) ( Fig 1 ) . Among the cases in which several biopsies were examined ( i . e . two biopsies from six patients and three biopsies from two patients ) , the SNP types were concordant in each individual case . The majority of the 35 patients from French overseas territories were from New Caledonia ( n = 21 ) and were infected with isolates belonging to SNP type 1 ( n = 3 ) , SNP type 2 ( n = 6 ) , SNP type 3 ( n = 11 ) and SNP type 4 ( n = 1 ) . The other patients of this group were from French Polynesia ( n = 4; SNP type 3 ) , Mayotte ( n = 5; SNP type 1 ) , the French West Indies ( n = 4 ) including Guadeloupe ( n = 1; SNP type 1 ) and Martinique ( n = 3; SNP type 4 ) or La Réunion ( n = 1; SNP type 3 ) . Among the 28 immigrant patients , those infected with SNP type 1 were born in RC ( n = 1 ) , Comoros ( n = 16 ) , India ( n = 1 ) , Sri Lanka ( n = 1 ) and Madagascar ( n = 1 ) . Patients with SNP type 3 were from Ivory Coast ( n = 1 ) and those with SNP type 4 were born in Mali ( n = 1 ) , Guinea ( n = 2 ) , Benin ( n = 1 ) , DRC ( n = 1 ) , Brazil ( n = 1 ) or Haiti ( n = 1 ) . The isolates from the four remaining patients who were born in metropolitan France were of SNP type 1 . Between 20 and 30 leprosy cases are still diagnosed each year at the French NRC . We investigated the correlation between the geographic origin of the patients and the SNP type of the M . leprae isolates collected over a period of five years . In general , the choice of the markers used for genotyping has been dependent on the timescale of the events of interest , with rapidly evolving markers ( VNTR markers ) for contact tracing over relatively short periods of time , and slower evolving markers ( SNPs ) for tracing dissemination of strains over hundreds to thousands of years [13] . The latter marker type was appropriate for our goal . By using several protocols , including nested PCR , the genotype of 94% of the isolates analyzed in the study has been determined . Amplifying DNA from clinical specimens for retrospective studies after freeze-thaw cycles can be difficult . Information on the yield ( success rate ) is usually lacking in publications reporting on M . leprae genotyping and where it was specified , it was of the order of 75% [14] or 94% [15] . Although using three different methods , i . e . ( i ) the original method of Monot et al . [6] , ( ii ) modified PCR conditions and ( iii ) nested PCR , we could not amplify DNA from four biopsies that were either AFB-negative or with low AFB counts ( < 8 ×104 AFB/ml ) . In contrast , the mean AFB count was 7×106/ml in the skin biopsy suspension containing isolates whose SNP type was successfully determined , in agreement with a study reporting that DNA amplification efficiency increases with increasing AFB counts [7] . Most leprosy patients diagnosed in France were born outside metropolitan France , either in another country ( n = 28 ) or in French overseas territories ( n = 35 ) , whereas only four patients were born in metropolitan France . France is a country of 65 million inhabitants living either in metropolitan France or in French overseas territories ( 2 million inhabitants ) , scattered over six continents . In the latter territories , leprosy is not eradicated and in 2011 the prevalence of leprosy ( per 10 000 inhabitants ) was 3 . 72 in Mayotte , 0 . 097 in La Réunion , 0 . 155 in Guadeloupe , 0 . 025 in Martinique , 0 . 7 in French Polynesia and 0 . 47 in New Caledonia [16] . Similar data have been reported in other Western European countries . In England , a review of 80 leprosy cases in Liverpool and Birmingham diagnosed between 1946 and 2003 showed that all patients except one acquired leprosy on the Indian subcontinent , in Africa or in South America . The only case of leprosy transmission within the UK concerned a 9-year-old child who acquired the disease in the 1940s from his father who had been infected in Brazil [17] . In Germany two cases were reported in 2006 of patients born in Pakistan or Sri Lanka [18] , whereas in Italy 59 cases were reported between 2003 and 2009 , all of patients who had immigrated from various countries [19] . In Spain , seven cases were described between 2004 and 2009 , six of immigrants and one of a Spanish patient who had worked for 25 years as a missionary in Venezuela [20] . WHO did not report European leprosy cases in its last report in 2013 [2] . The main results of the present study were ( i ) the demonstration that the SNP genotype of the isolates from leprosy patients born in metropolitan France correlated with the type ( s ) encountered in the countries these patients had visited or in which they had lived and where leprosy is not eradicated , ( ii ) the genotype determination of isolates originating from regions for which no data were available so far and ( iii ) the augmentation of data for regions covered by already published studies . For the patients who were born outside metropolitan France , i . e . in French overseas territories or other countries for which published records of M . leprae SNP types exist , the types correlated well with those encountered in the countries of birth of 31 patients ( Fig 1 ) . Published data in conjunction with the data of the present study are presented in Table 2 to provide a comprehensive overview of M . leprae SNP types classified according to geographical regions . There was a single incoherence , concerning the SNP type 3 isolate from a patient born in Ivory Coast , a type previously observed in North Africa and the Americas but so far not in West Africa where SNP type 4 is prominent . The latter type was the only one found in six patients born in Ivory Coast and 31 patients born in Mali ( Table 2 ) . We found a SNP type 4 for the first time in a patient born in New Caledonia where types 1 , 2 and 3 have been observed previously ( Fig 1 , Table 2 ) , likely reflecting diversity of populations and origins of migration flow , especially among the French overseas territories including the French West Indies where SNP type 4 is prominent . Furthermore , we determined SNP types of isolates from patients born in countries or regions not covered by previous studies ( n = 31 ) , i . e . Comoros , Haiti , Sri Lanka , Benin , Congo ( RC and DRC ) , La Réunion , Mayotte and French Polynesia . In a patient born in Haiti we found SNP type 4 which is not surprising since this type is prominent in several countries of South America ( i . e . Brazil , Bolivia , Uruguay , Venezuela and Mexico ( Table 2 ) ) . In a patient born in Sri Lanka , we found SNP type 1 , the major type found in India , a geographically close country ( Table 2 ) . In one patient born in Congo ( DRC ) and one patient born in Benin , we found SNP type 4 , in agreement with previous data describing this type in West African countries but also one case of SNP type 1 in Congo ( RC ) , a type found so far in Eastern Africa . However , it should be noted that the number of SNP typed M . leprae isolates from Africa , with the exception of Mali , is rather small ( Table 2 ) . In Mayotte and Comoros we found only SNP type 1 which is not surprising regarding the geographic localization of these islands between Madagascar and India where SNP type 1 is prominent . We found only SNP type 3 in La Réunion ( n = 1 ) and French Polynesia ( n = 4 ) ( Fig 1 ) . For Polynesia this is not surprising given the history of its settlement and its location in the Western Pacific region where SNP type 3 is common . As for La Réunion , the result can be explained by the immigration at the end of the 19th century of a large group of settlers , locally called “Sinwa” , from Southern China where SNP type 3 is prevalent ( Table 2 ) . The regions in which three of the four patients born in metropolitan France were infected are reasonably obvious: two lived in Congo , therefore the SNP type 1 found in both cases and previously described in Congo ( RC and DRC; Fig 1 ) supports the acquisition of leprosy in that region . One patient also infected with an SNP type 1 strain lived in India , another country where this type is frequently described ( Table 2 ) . As to where the last patient was infected , a humanitarian worker active in many countries , including leprosy endemic areas , we do not have enough information to support speculation . The phylogenic analysis of M . leprae strains isolated in France showed a good correlation between M . leprae genotypes and the geographical origin of the patients . These results are in agreement with the view that French leprosy cases are acquired in regions where leprosy is still endemic ( countries other than France ) or present ( French overseas territories ) but not in metropolitan France where leprosy has been eradicated . In conclusion , the SNP types of all M . leprae isolates from leprosy cases diagnosed in metropolitan France from 2009 to 2013 were consistent with what is known about their distribution in different regions of the world . The phylogenetic analysis also showed agreement between M . leprae genotype and geographical origin of the patients . The analysis further expanded our knowledge regarding the strain types from French overseas territories and their connection with migration flow . Previously published data combined with those reported here indicate that , in a globally stable situation , major changes in the prevalence of SNP types occur in regions where migration flow is greatest .	Leprosy is an old disease that is nearly eradicated from the European continent but not worldwide . The infectious agent , Mycobacterium leprae , has a highly conserved genome , and this property has been used to elucidate the route of its dissemination all over the world . At the French National Reference Center for mycobacteria , 20 to 30 leprosy cases are diagnosed every year , mainly in immigrants from endemic countries or in people living in French overseas territories . A phylogenetic analysis was conducted to investigate the relationship between M . leprae genotypes and the geographical origin of the patients .	[ "Abstract", "Introduction", "Materials", "and", "Methods", "Results", "Discussion" ]	[]	2015	New Insights into the Geographic Distribution of Mycobacterium leprae SNP Genotypes Determined for Isolates from Leprosy Cases Diagnosed in Metropolitan France and French Territories
In a screen for RNA mutagen resistance , we isolated a high fidelity RNA dependent RNA polymerase ( RdRp ) variant of Coxsackie virus B3 ( CVB3 ) . Curiously , this variant A372V is also resistant to amiloride . We hypothesize that amiloride has a previously undescribed mutagenic activity . Indeed , amiloride compounds increase the mutation frequencies of CVB3 and poliovirus and high fidelity variants of both viruses are more resistant to this effect . We hypothesize that this mutagenic activity is mediated through alterations in intracellular ions such as Mg2+ and Mn2+ , which in turn increase virus mutation frequency by affecting RdRp fidelity . Furthermore , we show that another amiloride-resistant RdRp variant , S299T , is completely resistant to this mutagenic activity and unaffected by changes in ion concentrations . We show that RdRp variants resist the mutagenic activity of amiloride via two different mechanisms: 1 ) increased fidelity that generates virus populations presenting lower basal mutation frequencies or 2 ) resisting changes in divalent cation concentrations that affect polymerase fidelity . Our results uncover a new antiviral approach based on mutagenesis . Amiloride and its derivatives are potassium-sparing diuretics used to treat hypertension and to prevent hypokalemia associated with congestive heart failure . These compounds act by inhibiting epithelial Na+ channels and the Na+/H+ , Na+/Ca2+ and Na+/Mg2+ antiport functions [1] , [2] . Due to its relatively low toxicity , the antiviral properties of amiloride is being explored . This compound inhibits the viroporins of coronaviruses , flaviviruses and retroviruses [3] , [4] , [5] , [6] . More recently , amiloride was shown to exert an antiviral effect on rhinovirus ( common cold ) and Coxsackie virus B3 ( CVB3 , viral myocarditis ) by directly affecting virus replication or release [7] , [8] . In this study , Harrison et al . isolated two viral RNA dependent RNA polymerase ( RdRp ) mutants of CVB3 that were more resistant to amiloride than wild type virus . The mechanism for this resistance remains unclear . We isolated one of these same CVB3 RdRp mutants in a screen for resistance to RNA mutagens with the goal of identifying higher fidelity RdRp variants . RNA mutagens such as ribavirin , 5-fluorouracil and 5-azacytidine are base analogs that are incorrectly inserted into the genome by the RdRp during replication and result in the accumulation of lethal mutations over several passages , a process referred to as lethal mutagenesis[9] , [10] , [11] , [12] , [13] . A similar screen previously identified a higher fidelity variant of poliovirus , suggesting that the intrinsic fidelity of RdRps can be modulated , despite their lack of proofreading functions [14] , [15] . Poliovirus RdRp fidelity variants have since proven to be useful tools for studying the role of genetic diversity in virus fitness and virulence , and have shown promise in improving vaccine attenuation and genetic stability [16] , [17] , [18] . In order to extend these observations to other medically relevant viruses , we performed a screen for high fidelity variants of CVB3 by selecting for resistance to the mutagenic base analogs , ribavirin and 5-azacytidine . Here we identify a CVB3 variant with higher fidelity that maps to a different region of the RdRp than the previously described position 64 of poliovirus [18] . Since the same mutation confers resistance to RNA mutagens and to amiloride , we hypothesized that amiloride has a previously unknown mutagenic activity . In this report we provide the first evidence for an indirect RNA mutagenic activity for the amiloride compounds and show that the amiloride resistant CVB3 RdRp variants resist this mutagenic activity through two different mechanisms . To determine the conditions in which to generate resistance to RNA mutagens , we treated CVB3 with various concentrations of ribavirin and 5-azacytidine ( AZC ) . Mutagen concentrations above 100 µM decreased viral viability by over 90% ( Figure 1A ) . In order to exert a strong enough selective pressure , without extinguishing the virus population during passage , we serially passaged a large virus population size ( 106 TCID50 ) in 50 µM of either mutagen . Every five passages , the viral RNA-dependent RNA polymerase ( RdRp ) region was sequenced ( Figure 1B ) . Wild type CVB3 acquired a new CtoU mutation resulting in an Alanine to Valine change at position 372 of the RdRp ( A372V ) after passage in both ribavirin ( by passage 10 ) and AZC ( by passage 20 ) , that did not arise in untreated control passages . The earlier emergence of this mutation under ribavirin treatment correlated with ribavirin's bias towards GtoA and CtoU transition mutations [19] , compared to AZC's bias for CtoG and GtoC transversions [20] . The role of A372V in resistance was confirmed by introducing the mutation back into the CVB3 wild type cDNA infectious clone and comparing the growth of A372V to wild type virus treated with 3 RNA mutagens with different nucleotide structures: ribavirin ( 300 µM ) , AZC ( 300 µM ) and 5-fluorouracil ( FU , 150 µM ) . Indeed , A372V was more resistant than wild type virus in each case ( Figure 1C ) . A similar profile of broad resistance to different base analogs was previously observed for the poliovirus G64S high fidelity variant [17] , suggesting that A372V is also a higher fidelity RdRp . To confirm this with genetic data , wild type and A372V virus stocks were prepared from infectious clones and used to infect cells that were treated with 400 µM ribavirin , or mock treated . At total cytopathic effect ( 48 hours after infection ) , the mutation frequencies and distributions within each population was determined by sequencing a 1 . 3 kb fragment of the capsid region from individual genomes ( Figure 1D ) . The untreated wild type virus population presented a mutation frequency of 4 . 5 mutations per 104 nucleotides ( Figure 1E and Table S1 ) . The A372V population , on the other hand , presented a mutation frequency that was approximately 2-fold lower than wild type ( P = 0 . 0225 ) , thereby confirming the higher fidelity of this variant . As expected , treatment with 400 µM ribavirin significantly increased the mutation frequency of both virus populations ( P<0 . 0001 ) ; however , the mutation frequency of A372V remained lower than wild type ( P = 0 . 039 ) . Furthermore , individual clones in the wild type virus population more often presented multiple mutations compared to A372V ( Table S1 ) . These results further confirmed the increased fidelity of the A372V RdRp . To provide further evidence for the increased fidelity of A372V RdRp , an in vitro biochemical assay was used to examine the relative levels of misincorporation for both wild type and A372V enzymes . Briefly , in vitro reactions were performed using equal amounts of each purified RdRp , saturating concentrations of nucleotide and a radiolabeled RNA primer-template substrate that permits nucleotide addition to be monitored by the extension of end-labeled primer ( Figure 2A ) [21] . By adding GTP to the reaction , the incorrect incorporation of this nucleotide can be monitored over time by the accumulation of n+1 product . As observed in Figure 2B–C , A372V RdRp consistently incorporated less GMP over time than wild type RdRp . Interestingly , the observed rate of correct nucleotide incorporation for wild type and A372V showed no significant difference , whereas the observed rate of GMP misincorporation showed a 2-fold difference between wild type and A372V RdRp ( Table 1 ) , confirming the increased fidelity of the A372V enzyme . Curiously , A372V was previously isolated in a screen for resistance to amiloride , a compound with no known mutagenic activity , that was shown to reduce CVB3 titers by partially inhibiting RNA synthesis [8] . Accordingly , A372V consistently replicated to higher titers than wild type at all concentrations of amiloride tested ( Figure 3A ) . A possible explanation for the dual resistance to both RNA mutagens and amiloride is that A372V produces more RNA genomes in the presence of these compounds and would thus titer higher than wild type . To determine whether higher A372V titers in the presence of amiloride were due to increased replicative capacity , one-step growth analysis was performed , in the presence or absence of amiloride . No significant differences in virus production kinetics were observed for wild type and A372V grown in the absence of amiloride ( Figure 3B ) , indicating that increasing fidelity of the CVB3 RdRp did not significantly impact virus multiplication . Furthermore , RNA synthesis of wild type and A372V had similar kinetics and were within the same order of magnitude , as determined by northern blot analysis ( Figure 3C ) . In fact , A372V virus produced slightly less RNA than wild type , yet consistently titered slightly higher because the RNA genomes made contain fewer deleterious mutations . Importantly , differences observed for wild type and A372V one-step growth kinetics in the presence of 400 µM amiloride were not statistically significant , with the exception of one time point ( 9 hours , p = 0 . 03 ) ( Figure 3D ) and northern blot analysis revealed that both viruses were similarly inhibited in RNA synthesis ( Figure 3E ) . Heightened replicative capacity was thus not responsible for the resistance of A372V to amiloride . Since it is unlikely that the same mutation confers resistance to two unrelated antiviral mechanisms , we explored whether amiloride has a previously unknown mutagenic activity . This possibility was not evident , since amiloride is not a base analog such as ribavirin , AZC and FU , whose mutagenic effects result from their direct misincorporation into genomes by the error-prone RdRp . We hypothesized that in addition to inhibiting RNA replication , amiloride compounds increase the virus mutation frequencies of the RNA genomes that are replicated . To address this , wild type and A372V virus infections were performed in the presence of either ribavirin or amiloride , or in the absence of either compound and the mutation frequencies of the resulting populations were determined ( Figure 4A ) . At 400 µM amiloride , the mutation frequency of wild type CVB3 increased from 4 . 5 to 9 . 4 mutations per 104 nucleotides ( P = 0 . 0005 ) . Similarly , A372V virus also increased mutation frequency , from 2 . 5 to 6 . 2 mutations per 104 nucleotides ( P = 0 . 0008 ) . However , since the basal mutation frequency of A372V was lower , this increase was better tolerated , explaining the high virus titers observed ( Figure 3A ) . Next , we determined whether the observed mutagenic activity was common to a wider range of amiloride compounds . Wild type virus was treated with the amiloride derivatives EIPA , MIA and benzamil and the mutation frequencies were determined as described above ( Figure 4B ) . Again , the mutation frequencies for wild type CVB3 increased from the basal 4 . 5 mutations per 104 nucleotides to 8 . 5 for EIPA ( P = 0 . 0028 ) and to 9 . 9 for MIA ( P<0 . 0001 ) . Although a tendency towards increase was observed for benzamil ( 6 . 3 mutations per 104 nucleotides ) , no statistically significant difference was established ( P = 0 . 105 ) . Furthermore , treatment of A372V with EIPA also increased the mutation frequency from 2 . 5 to 4 . 1 mutations per 104 nucleotides ( P = 0 . 05 ) and correlated the higher titers compared to wild type virus [8] . We then examined whether amiloride treatment biased the mutation profile towards specific substitutions , as is common for base analog RNA mutagens . In untreated populations , the transition mutations AtoG and TtoC ( UtoC on RNA genome ) were most common ( Figure 4C ) . Treatment of populations with ribavirin , which biases the mutation profile , resulted in the accumulation of GtoA and CtoT ( CtoU on RNA ) transition mutations . On the other hand , treatment with amiloride increased , but did not bias , the natural mutation profile . Taken together , our results provide the first evidence of a mutagenic activity for amiloride compounds and suggests that A372V resists this effect by increasing RdRp fidelity and lowering basal mutation frequency . The G64S variant of poliovirus is another higher fidelity RdRp variant whose fidelity altering determinant maps to a different region of the RdRp ( Figure S2 ) . Previous studies showed that G64S was resistant to ribavirin , AZC and FU; replicated with similar kinetics to wild type poliovirus in one-step growth curves and northern blot analysis; generated virus populations with lower basal mutation frequencies and presented a higher fidelity phenotype in biochemical incorporation assays [14] , [15] , [16] , [17] , [18] . This provided the unique opportunity to determine whether amiloride exerts a mutagenic activity on a different virus with a different fidelity increasing mutation . Wild type and G64S polioviruses were treated with amiloride , and the relative mutation frequencies were determined ( Figure 4D ) . The basal mutation frequency of wild type poliovirus was 5 . 8 mutations per 104 nucleotides sequenced and increased to 8 . 2 mutations per 104 nucleotides ( P = 0 . 039 ) upon amiloride treatment . The high fidelity G64S population had a basal mutation frequency of 4 . 0 mutations per 104 nucleotides that increased to 6 . 7 mutations per 104 nucleotides ( P = 0 . 021 ) in amiloride , but remained significantly lower than wild type virus ( P = 0 . 05 ) . In order to determine whether lower basal mutation frequencies correlated with resistance , amiloride treated and untreated populations were titrated ( Figure 4E ) . A higher percentage of viruses in the G64S population survived amiloride treatment ( 14 . 3 ± 2 . 0% ) compared to the wild type poliovirus population ( 1 . 9 ± 0 . 9% , P = 0 . 006 ) . Our results therefore confirm for two different high fidelity RdRps , that increased fidelity and lower mutation frequency confer resistance to amiloride . Along with A372V , Harrison et al . isolated a second RdRp variant , S299T , in their screen for amiloride resistance [8] . By contrast , our screens for RNA mutagen resistance failed to isolate S299T alongside A372V . We hypothesized then that S299T is either resistant to another amiloride-mediated antiviral effect ( inhibition of RNA synthesis , e . g . ) , or resistant to amiloride's mutagenic activity by a different mechanism . As for wild type and A372V , virus stocks were prepared from the cDNA infectious clone of S299T virus and studies were performed in parallel to wild type and A372V . Similar to A372V , S299T was more resistant than wild type at all concentrations of amiloride tested ( Figure 5A ) . The one step-growth studies performed in the absence of amiloride showed that S299T produced virus with kinetics similar to wild type virus ( Figure 5B ) . Likewise , northern blot analysis of these kinetic studies revealed that S299T did not produce higher levels of RNA in the absence of amiloride ( Figure 5C ) . However , although one step growth analysis of S299T grown in the presence of amiloride also did not reveal significant differences in virus titer compared to wild type CVB3 ( Figure 5D ) , northern blot analysis of RNA synthesized by 48 hours after infection revealed that the inhibitory effects of amiloride on S299T were not as dramatic as for wild type and A372V ( Figure 5E ) . Finally , to examine the potential effects of the S299T mutation on RdRp fidelity , infections were performed in the presence of ribavirin , amiloride , EIPA , or under mock-treatment and mutation frequencies were determined for each population . Unlike A372V , the mutation frequency of S299T did not suggest higher fidelity compared to wild type ( Figure 5F ) . Rather , it presented a significantly higher mutation frequency ( 7 . 0 mutations per 104 nucleotides , P = 0 . 0034 ) , suggesting that this variant encodes a lower fidelity polymerase . Indeed , treatment with 400 µM ribavirin increased the mutation frequency to 28 . 1 mutations per 104 nucleotides ( P<0 . 0001 ) , significantly higher than wild type ( P = 0 . 0207 ) and A372V ( P<0 . 0001 ) ( Figure 1E ) and in accordance with a lower fidelity polymerase . To confirm the lower fidelity phenotype conferred by the S299T mutation , an in vitro biochemical assay examining the relative levels of misincorporation with the purified S299T enzyme was performed . As shown in Figure 2 , S299T RdRp incorporated more GMP over time than wild type RdRp , consistent with the S299T mutation decreasing RdRp fidelity ( Table 1 ) . Unexpectedly , when treated with either amiloride or EIPA ( Figure 5F ) , this population did not undergo a significant increase in mutation frequency ( 7 . 9 mutations per 104 nucleotides , P = 0 . 442 and 7 . 5 , P = 0 . 859 , respectively ) , suggesting that S299T polymerase is completely unaffected by the mechanism by which amiloride increases mutation in wild type and A372V viruses . Hence , while A372V resists the mutagenic activity of amiloride by increasing RdRp fidelity; S299T partially resists RNA replication inhibition and this mutagenic effect entirely . Our results argue that increases in mutation frequency are a previously unknown antiviral mechanism of amiloride compounds , but the relative contribution of RNA synthesis inhibition and RNA mutagenesis is unclear . To examine the relative contribution of mutagenesis to the overall antiviral effect , we determined the mutation frequencies of wild type CVB3 virus populations treated with increasing concentrations of ribavirin or amiloride . These data were used to determine the fold increase in mutation frequency of wild type virus at increasing concentrations of drug compared to untreated control . Indeed , ribavirin treatment showed a dose-dependent increase in mutation frequency ( Figure 6A , dashed line , right y-axis ) that correlated with a decrease in virus titers ( solid line , left y-axis ) . These data agree with previous work showing mutagenesis to be the principal antiviral mechanism of ribavirin in tissue culture [10] . In contrast , treatment of CVB3 with increasing concentrations of amiloride ( at least above 100 µM ) did not produce a corresponding increase in mutation frequency ( Figure 6B , dashed line ) , although the decrease in virus yield was dose-dependent ( solid line ) . Northern blot analysis of RNA synthesis at these different amiloride concentrations revealed a dose dependent effect on RNA inhibition ( Figure 6C ) . These results suggest that replication inhibition is the principal cause of the dose-dependent drop in virus titers and that the mutagenic effects of amiloride are the result of an indirect , dose-independent effect on the polymerase . The dose dependent decrease in viral titer ( Figure 6A ) attributed to inhibition of RNA synthesis is proposed to result from a direct interaction of amiloride with the RdRp [8] . Given that dose dependence of mutation frequency was not observed for amiloride , we hypothesized that this mutagenic effect was the indirect result of the effect of amiloride on the cellular environment . Indeed , previous studies have shown amiloride treatment to alter the intracellular concentrations of free Na+ , Ca2+ , Mg2+ , which in turn can alter the equilibrium levels of other ions , including Mn2+ [1] , [2] , [22] . Mg2+ and Mn2+ are interesting candidates since they are cofactors essential for incorporation activity of polymerases such as viral RdRp . In our own cell culture conditions , we confirmed that in addition to the more commonly studied effects on Na+ , Mg2+ levels are also significantly perturbed in cells treated with amiloride ( P = 0 . 0009 ) ( Figure S1 ) . To examine the potential role of altered intracellular cation concentrations in the observed mutagenic activity , wild type , A372V and S299T viruses were grown in cell media supplemented with high concentrations of either NaCl , CaCl2 , MgCl2 or MnCl2 and the resulting mutation frequencies were determined . Growth of all three viruses in either NaCl or CaCl2 had no significant effect on mutation frequency ( Figure 6D–F ) . Increasing intracellular Mg2+ or Mn2+ concentrations on the other hand , resulted in a significant increase in mutation frequency for both wild type ( P<0 . 0001 for Mg2+ , P = 0 . 046 for Mn2+ , Figure 6D ) and A372V ( P = 0 . 021 for Mg2+ , P = 0 . 0073 for Mn2+ , Figure 6E ) viruses , the frequencies of A372V being significantly lower than those of wild type ( P<0 . 0001 ) as would be expected of the higher fidelity variant . In contrast , the mutation frequency of S299T remained unchanged ( P = 0 . 695 for Mg2+ , P = 0 . 769 for Mn2+ , Figure 6F ) . Importantly , one-step growth analysis of each virus grown in the presence of this concentration of MgCl2 confirmed that replication itself was not affected ( Figure S3 ) . Given the similar sensitivity and resistance profiles of each variant in both these and the amiloride treated populations , our results support the notion that amiloride induced mutagenesis results from changes in intracellular concentrations of the essential divalent cation cofactors . To gather further support for the link between amiloride mutagenesis and Mg2+/Mn2+ concentrations , wild type virus was serially passaged in the presence of either compound and the RdRp regions of the passaged virus populations were sequenced at the intervals indicated ( Figure 6G ) . Interestingly , wild type virus that was grown in the presence of MnCl2 had acquired ( passage 5 ) and fixed ( passage 7 onward ) the same high fidelity A372V mutation that was selected by RNA mutagen ( Figure 1 ) and amiloride [8] passage . Similarly , passage of wild type virus in MgCl2 selected the same high fidelity which was fixed in the population by passage 15 . In this report , we identify a previously undescribed mutagenic effect of amiloride treatment on CVB3 and poliovirus ( Figure 4 ) and confirm that amiloride inhibits RNA synthesis of Coxsackie B3 virus ( Figure 3E , 5E ) . What are the relative contributions of the two activities to the observed antiviral effect of amiloride compounds ? The dose dependent inhibition of RNA synthesis ( Figure 6C ) that correlates with the dose dependent reduction in wild type virus titers ( Figure 6B ) suggests that this is the principal antiviral activity for amiloride . This may reflect a direct interaction of amiloride with the RdRp , perhaps through blocking of the nucleotide entry tunnel or catalytic site , as was suggested by Harrison et al [8] . More detailed in vitro biochemical or in vitro replication assays should help determine the nature of this interaction . What is the molecular basis for amiloride induced mutagenesis ? The lack of dose dependence of the observed mutagenic effect at higher concentrations of amiloride ( Figure 6B ) that is typically observed for RNA mutagens such as ribavirin ( Figure 6A ) led us to seek an indirect mechanism for amiloride induced mutagenesis . Amiloride inhibits epithelial Na+ channels , Na+/H+ ion antiporters and other less characterized ion exchangers ( Na+/Ca2+ and Na+/Mg2+ ) which could in turn affect the equilibrium of other ions , such as Mn2+ [2] , [22] . Indeed , in our own experimental conditions we observed a significant alteration of intracellular Mg2+ concentrations , although 800 µM amiloride treatment was necessary because of detection limits ( Figure S1 ) . Since Mg2+ and Mn2+ are essential cofactors for polymerase activity and nucleotide insertion [23] , we explored a potential link between amiloride treatment , Mg2+/Mn2+ levels and CVB3 mutation frequency . Indeed , we found that increases in intracellular Mg2+ and Mn2+ correlate with increased virus mutation frequency ( Figure 6D–E ) . Whether amiloride inhibits Mg2+ and/or Mn2+ transporters directly ( which have not yet been identified in eukaryotes ) or indirectly through effects on Na+ channels and exchangers will be difficult to determine . Studies with other channel blockers of different structure may help to pinpoint this mechanism , although redundancy between channels may mask their effect . Nevertheless , the mutagenesis and resistance profiles of the A372V and S299T amiloride-resistant variants support this link between cation concentrations and amiloride's mutagenic effect . A372V is a higher fidelity RdRp variant that generates virus populations with lower basal mutation frequencies than wild type populations . In result , this virus population , although not impervious to the effects of mutagens , can better tolerate a moderate increase in mutation frequency that would otherwise lethally mutagenize the wild type population . Accordingly , A372V mutation frequencies increased following ribavirin treatment , amiloride treatment and treatment with high concentrations of Mg2+ and Mn2+ but in all cases , the frequencies were significantly lower than wild type populations under the same treatments . Unlike S299T , this variant is solely resistant to the mutagenic antiviral activity since its RNA synthesis is as inhibited by amiloride as wild type virus ( Figure 3E ) . It is perhaps for this reason that passage of wild type virus in Mg2+ and Mn2+ ( that mimics only the mutagenic , and not the RdRp inhibitory , activity of amiloride ) , resulted in the selection of the high fidelity A372V variant over S299T ( Figure 6G ) . The S299T variant seems to resist amiloride on both fronts: RNA synthesis is less inhibited ( Figure 5E ) and its mutation frequency is unchanged ( Figure 5F ) . Whether S299T's dual mechanisms of resistance to both of amiloride's antiviral activities are tightly coupled , or coincidental , remains to be determined . Nevertheless , a similar lack of mutagenic effect by Mg2+ and Mn2+ treatment ( Figure 6F ) provides further support for the role of divalent cations in amiloride-induced mutagenesis . However , this seems to come at the cost of decreasing overall RdRp fidelity , as evidenced by its hypersensitivity to the nucleoside mutagen ribavirin and in vitro biochemical data ( Figure 2 and 5F ) . Interestingly , low fidelity variants of poliovirus and FMDV also map to the same β9–α11 loop , a domain interacting with the active site where coordination of one of two divalent cation cofactors and the incoming nucleotide occurs [24] , [25] . Alterations to this domain may alter RdRp dependence on the relative availability of the Mg2+ cofactor , thereby resulting in a lower fidelity , Mg2+ concentration-insensitive RdRp or may change its preference for Mg2+ to Mn2+ . In poliovirus , a position 297 variant was shown to be a lower fidelity RdRp in vitro and a position 296 variant of FMDV also presented higher mutation frequencies [24] , [26] . In earlier studies , position 297 poliovirus variants were found to be dependent on Mn2+ for growth[27] . Biochemical assays have shown that poliovirus RdRp activity in the presence of Mn2+ results in higher mutation frequencies[28] , [29] and if S299T is Mn2+ dependent , it may explain the higher basal mutation frequency relative to wild type virus . The use of RNA mutagens to extinguish viral populations by hypermutation is a promising antiviral approach [11] , [12] , [13] , [19] , [30] , [31] , [32] . Lower fidelity variants ( such as S299T ) could more readily identify weakly mutagenic base analogs that could later be improved , while higher fidelity variants ( such as A372V ) would help test the efficacy of the strongest of mutagens , of nucleoside or non-nucleoside structure [33] , [34] , [35] . Since the description of lethal mutagenesis using base analogs , there is speculation as to whether other compounds can alter mutation frequency through direct or indirect action on the RdRp . It is important to note that the mutagenic activity of amiloride observed in our study was not as significant as that observed for nucleoside RNA mutagens . This raises a two-sided question: is the activity of amiloride strong enough to lethally mutagenize the virus population over a prolonged exposure or would the moderate increase in mutation frequency help the virus more rapidly evolve or escape immune responses ? Under these experimental conditions , the mutagenic effect was strong enough to force the virus to develop two mechanisms of resistance , suggesting that this mutagenic activity alone was sufficiently detrimental to the virus . Whether the antiviral mutagenic effect we observed here occurs at physiological intracellular amiloride concentrations in humans is not yet known and how this may affect the fidelity of cellular RNA polymerases remains to be studied . Nevertheless , our work uncovers a new target for drug discovery - compounds that induce reversible alterations of the intracellular concentrations of essential cation cofactors for viral RdRp . Our results should encourage screening of compound libraries for new molecules that , through direct interaction with RdRps or indirect effects on the intracellular environment , may alter the mutation frequency of RNA viruses by modulating polymerase fidelity . Finally , our study identifies two new CVB3 RdRp fidelity variants . To date , only position 64 variants of poliovirus were shown to exhibit increased RdRp fidelity [15] , [18] . Despite having similar RdRp structures [36] , [37] , [38] , A372V maps to a distant region of the polymerase ( Figure S2 ) . Our previous work and current findings ( higher fidelity A372V and lower fidelity S299T ) , along with the description of lower fidelity RdRp of poliovirus and Foot and Mouth Disease virus [24] , [26] , suggest that the intrinsic fidelity of RdRps is defined by multiple residues . The full extent of this fidelity network and whether it translates across RdRps from different virus families is yet to be determined . In addition to their utility in studies such as our current report , RdRp variants are valuable tools to study viral evolution and adaptation of RNA virus populations in vivo [14] , [16] , [17] . It will be interesting to determine how a virus population of increased or decreased genetic diversity will behave in the context of Coxsackie virus infection . HeLa ( Young ) and Vero cells were maintained in DMEM medium with 10% newborn calf serum . Plasmid bearing the cDNA of Coxsackie virus B3 ( Nancy ) strain was a kind gift of F . van Kuppeveld ( Radboud University , Nijmegen , Netherlands ) . Plasmids containing poliovirus cDNAs were previously described [17] . The following compounds were obtained from Sigma Aldrich: Ribavirin IUPAC 1-[ ( 2R , 3R , 4S , 5R ) -3 , 4-dihydroxy-5- ( hydroxymethyl ) oxolan-2-yl]-1H-1 , 2 , 4-triazole-3-carboxamide ) ; 5-fluorouracil IUPAC 5-fluoro-1H-pyrimidine-2 , 4-dione; 5-Azacitidine IUPAC 4-amino-1-β-D-ribofuranosyl-1 , 3 , 5-triazin-2 ( 1H ) -one; Amiloride IUPAC 3 , 5-diamino-6-chloro-N- ( diaminomethylene ) pyrazine-2-carboxamide; EIPA IUPAC 3-amino-6-chloro-N- ( diaminomethylidene ) -5-[ethyl ( propan-2-yl ) amino]pyrazine-2-carboxamide; MIA IUPAC 3-amino-5-[tert-butyl ( methyl ) amino]-6-chloro-N- ( diaminomethylidene ) pyrazine-2-carboxamide; benzamil IUPAC 3 , 5-diamino-N- ( N'-benzylcarbamimidoyl ) -6-chloropyrazine-2-carboxamide . All studies were performed on virus stocks ( wild type , A372V or S299T ) generated from cDNA infectious clones . The A372V and S299T variants were constructed using the Quikchange XL site directed mutagenesis kit ( Stratagene ) and the CVB3-Nancy infectious cDNA . Three infectious cDNA clones of each variant ( A372V and S299T ) were obtained , sequenced and used to generate three independent virus stocks . Three virus stocks of wild type virus were also prepared in this manner . Each of three stocks served as one of three triplicate samples in replication studies , and RNA mutagen and amiloride compound sensitivity assays . For mutation frequency data , the virus population generated by clone 1 for each variant was used throughout the study . Basal mutation frequencies were determined on 2 independently generated samples of each variant and no statistically significant differences in mutation frequencies were found . CVB3 cDNA plasmids were linearized with Sal I and poliovirus cDNA plasmids , with Eco RI . Linearized plasmids were purified with the Qiagen PCR purification kit . 2 . 5 µg of linearized plasmid was in vitro transcribed using T7 RNA polymerase ( Fermentas ) . 8 µg of transcript was electroporated into 4×106 Vero cells that were washed twice in PBS ( w/o Ca2+ and Mg2+ ) and resuspended in PBS ( w/o Ca and Mg ) at 107 cells/ml . Electroporation conditions were as follows: 0 . 4 mm cuvette , 25 µF , 700 V , maximum resistance , exponential decay in a Biorad GenePulser XCell electroporator . Cells were recovered in DMEM . 500 µl of p0 virus stock was used to infect 3×106 Vero cells in T25 flasks , to produce p1 virus . 250 µl of p1 virus was used to infect 1×107 Vero cells in DMEM-10% NCS in T75 flasks , to produce p2 virus . For each passage , virus was harvested at total cytopathic effect ( CPE ) by one freeze-thaw cycle and clarified by spinning at 10 K rpm for 10 minutes . Ten-fold serial dilutions of virus were prepared in 96-well round-bottom plates in PBS . Dilutions were performed in octuplate and 100 µl of dilution were transferred to 104 Vero cells plated in 100 µl of DMEM . After 5 days living cell monolayers were colored by crystal violet . TCID50 values were determined by the Reed and Muensch method . By plaque assay . HeLa cells were seeded into 6-well plates and virus preparations were serially diluted ( 10-fold ) in PBS . Cells were washed twice with PBS and infected with 250 µl of dilution for 30 minutes at 37°C , after which a semisolid overlay comprised of DMEM medium and 1 . 2% w/v Avicell ( FMC Biopolymer ) was added . 2 days after infection , cells were washed and stained with crystal violet 0 . 2% , and plaques were enumerated . HeLa cell monolayers were pretreated with 50 µM ribavirin , 50 µM 5-Azacytidine , 5 mM MgCl2 or 1 mM MnCl2 for 2 hours , then infected with 106 TCID50 of CVB3 . Blind serial passages were then performed on fresh mutagen-treated HeLa cell monolayers ( 250 µl of virus-containing supernatant ) . At the indicated passage intervals , viral RNA was extracted from purified virions with Trizol reagent ( Invitrogen ) and RT-PCR was performed ( Titan One-Step , Roche ) . PCR products were sequenced to identify consensus sequence changes within the RdRp region , between nucleotides 4701 and 7903 . HeLa cell monolayers in 6-well plates were pretreated for 2 hours ( ribavirin , AZC , FU , NaCl , CaCl , MgCl2 , or MnCl2 ) or 10 hours ( amiloride compounds ) with different concentrations of compound as indicated . We chose and verified concentrations of compounds that were not toxic to cells over a 72 hours period . For amiloride compounds , we chose and confirmed concentrations corresponding to virus inhibitory concentration ( IC50 ) values that were not toxic to cells , as determined by Harrison et al . [8] . Cells were then infected at an MOI = 0 . 01 with passage 2 virus . 48 hours post-infection , virus was harvested by one freeze-thaw cycle and virus titers ( TCID50 or plaque assay ) were determined . For replication studies , HeLa cells were either pretreated with 400 µM amiloride or mock treated and infected at an MOI of 10 . For one-step growth kinetics , transfected cells were frozen at different time points after infection and later titered by TCID50 assay . For Northern blot analysis , total RNA from infected cells was extracted by Trizol reagent ( Invitrogen ) and purified . 5 µg of total RNA were used per sample ( measured by Nanodrop ) . Gels were transferred onto a nitrocellulose membrane ( Whatman Turboblotter SuperCharge Nylon membrane kit ) , hybridized overnight with a dCTP-α32P labeled DNA probe corresponding to 200 bp of the RdRp , visualized on a Storm Phosphorimager and analyzed by ImageQuant . At total CPE , viral RNA in supernatants was extracted and RT-PCR amplified using the primers sets 878Forward and 2141Rev for CVB3 virus and 1337For and 2651Rev for poliovirus . The resulting PCR products were purified on column ( Nucleospin , Macherey-Nagel ) and TopoTA cloned ( Invitrogen ) . Blue/white screening was used on single colony transformants , positive clones were sequenced ( GATC Biotech ) . Sequence data was analyzed using the Lasergene software package ( DNAStar Inc ) . For statistical purposes , we retained sequence data only over the region for which every clone was represented ( 859 nucleotides for CVB3 and 884 nucleotides for PV ) . The number of mutations per 104 nucleotides sequenced was determined using the total mutations identified per population over the total number of nucleotides sequenced for that population multiplied by 104 . For each population , between 68 and 178 clones were sequenced representing between 58 , 000 to 153 , 000 nucleotides per sample ( see Table S1 ) . If the same mutation recurred in the same population , this was only counted once; however , this only occurred in two instances of one single repetition each and was not sufficient to change the mutation frequency values and statistics . A confluent layer of HeLa cells was treated with 800 µM amiloride for 10 hours and incubated with PBS containing 100 µM Ethylene glycol-bis ( β-aminoethyl ether ) -N , N , N' , N'-tetraacetic acid ( EGTA ) for 15 min , to chelate Ca2+ , before lysis in H2O-Tween 0 . 1% . The ratiometric magnesium indicator mag-fura-2 ( Invitrogen ) was added to a final concentration of 2 µM to 1 ml of sample and fluorescence was measured at 25°C using a Quanta-Master QM4CW spectrofluorometer ( PTI ) using a 1 cm path length quartz cuvette thermostated at 25°C . Excitation scans were performed from 250 to 490 nm with 1 nm steps , using 1 nm bandwidth; emission was monitored at 530 nm with 5 nm bandwidth . Continuous recordings of fluorescent intensities at 330 nm and 370 nm were transformed into 330/370 wavelength ratios . For one step growth curves , mutagen treatment assays and Mg2+concentration assays , the two-tailed paired student's t tests were used to determine significance with 95% confidence intervals . For mutation frequencies , χ2 tests and two-tailed Mann Whitney U tests were performed . χ2 tests compared the total number of mutations in a population to total nucleotides sequenced . Mann Whitney tests compared the ranked scores of number of mutations found in individual clones grouped by population . In all cases , Mann Whitney tests gave the more conservative P values and are the indicated here . Statistics were performed using Prism software ( GraphPad Inc ) . All RNA oligonucleotides were from Dharmacon Research , Inc . ; [γ-32P]-ATP ( 7000 Ci/mmol ) was from MP Biomedical; T4 polynucleotide kinase was from USB; ATP and GTP were from GE Healthcare; all other reagents were of the highest grade available from Sigma , Fisher or VWR . RNA oligonucleotides were purified by denaturing PAGE and end-labeled by using [γ-32P]ATP and T4 polynucleotide kinase as described previously[21] . Concentrations were determined by measuring the absorbance at 260 nm using a Nanodrop spectrophotometer and using the appropriate calculated extinction coefficient . Expression constructs for wild type , A372V and S299T CVB3 RdRp were created by using standard recombinant DNA protocols . DNA sequences were amplified using the appropriate CVB3 cDNA as template . Forward and reverse primers employed for amplification were selected based on the presence of unique restriction sites suitable for cloning of the 3D gene ( RdRp ) into the pSUMO expression plasmid[39] E . coli Rosetta cells were tansformed with the appropriate plasmid and these cells were used to produce an inoculum for large-scale growth . CVB3 3D gene expression was induced during exponential growth by addition of isopropyl-β-D-thiogalactopyranoside or by using auto-induction[40] . Induced cells were lysed in appropriate buffers , and the enzymes were purified to apparent homogeneity by using standard column chromatography resins and protocols[39] . Reactions were performed at 30°C . 32P-labeled primer extension assays: 2 µM CVB3 RdRp was mixed with 0 . 5 µM [32P]-primer-template substrate in 50 mM HEPES pH 7 . 5 , 5 mM MgCl2 , 60 µM ZnCl2 and 10 mM 2-mercaptoethanol . Reactions were initiated by the addition of ATP or GTP ( 1 or 5 mM ) in 50 mM HEPES pH 7 . 5 , 5 mM MgCl2 , 60 µM ZnCl2 , 10 mM 2-mercaptoethanol and 200 mM NaCl . Reactions were quenched at various times by addition of quench buffer ( 50 mM EDTA , 70% formamide , 0 . 025% bromphenol blue , and 0 . 025% xylene cyanol ) . CVB3 RdRp was diluted immediately prior to use in 50 mM HEPES pH 7 . 5 , 20% glycerol and 10 mM 2-mercaptoethanol . The volume of enzyme added to any reaction was always less than or equal to one-tenth the total volume . Products were resolved by denaturing PAGE . Stopped-flow fluorescence assay: Pre-steady state stopped-flow fluorescence experiments were performed using a Model SF-2001 stopped-flow apparatus ( Kintek Corp . , Austin , TX ) equipped with a waterbath . All reactions were performed at 30°C . 2 µM CVB3 3Dpol was mixed with 0 . 5 µM primer-template substrate containing 2-aminopurine ribonucleoside on the 5′ side of the templating base[41] in 50 mM HEPES pH 7 . 5 , 5 mM MgCl2 , 60 µM ZnCl2 and 10 mM 2-mercaptoethanol . Reactions were initiated by the addition of 1 mM ATP in 50 mM HEPES pH 7 . 5 , 5 mM MgCl2 , 60 µM ZnCl2 , 10 mM 2-mercaptoethanol and 200 mM NaCl . After mixing , reactant concentrations were reduced by 50% . Fluorescence emission was monitored by using a 370 nm cut-on filter ( model E370LP , Chroma technology corp . , Rockingham , VT . ) . The excitation wavelength used was 313 nm . For each experiment , at least four fluorescence traces were averaged . The relative fluorescence was plotted as a function of time and fit to a single exponential equation , , where F is the relative fluorescence intensity , kobs is the observed rate constant for nucleotide incorporation , t is the time and C is an offset . Quenched reaction mixtures were heated to 70°C for 2–5 min prior to loading 5 µl on a 20% denaturing polyacrylamide gel containing 1X TBE and 7 M urea . Electrophoresis was performed in 1X TBE at 85 watts . Gels were visualized by using a PhosphorImager and quantified by using the ImageQuant software ( Molecular Dynamics ) . Data were fit by nonlinear regression using the program , KaleidaGraph ( Synergy Software , Reading , PA ) .	RNA viruses have extreme mutation frequencies , due in large part to the erroneous nature of the viral RNA dependent RNA polymerases ( RdRp ) that replicate their genomes . Since RdRp lack proofreading and repair mechanisms , the use of base analogs as RNA mutagens to increase lethal mutations and extinguish the virus population is a promising antiviral strategy . Recently , a screen for resistance to this antiviral treatment identified a higher fidelity RdRp variant of poliovirus , indicating that RdRp fidelity can be modulated by single amino acid substitutions . To extend these observations to other viruses , we performed a similar screen using Coxsackie virus B3 ( CVB3 ) . We identified a new high fidelity RdRp variant which was also resistant to amiloride compounds that have no known mutagenic activity . Using wild type and RdRp fidelity variants of poliovirus and CVB3 , we show that amiloride compounds do have mutagenic activity and act on RNA virus populations indirectly , by altering intracellular ion concentrations that affect polymerase fidelity . Our results identify a new means of targeting viruses through increases in mutation frequency using non-nucleoside compounds that alter the cellular environment in which the virus replicates .	[ "Abstract", "Introduction", "Results", "Discussion", "Methods" ]	[ "virology/new", "therapies,", "including", "antivirals", "and", "immunotherapy", "virology/antivirals,", "including", "modes", "of", "action", "and", "resistance", "virology", "virology/mechanisms", "of", "resistance", "and", "susceptibility,", "including", "host", "genetic...	2010	Fidelity Variants of RNA Dependent RNA Polymerases Uncover an Indirect, Mutagenic Activity of Amiloride Compounds
Transposable elements ( TEs ) account for a large portion of the genome in many eukaryotic species . Despite their reputation as “junk” DNA or genomic parasites deleterious for the host , TEs have complex interactions with host genes and the potential to contribute to regulatory variation in gene expression . It has been hypothesized that TEs and genes they insert near may be transcriptionally activated in response to stress conditions . The maize genome , with many different types of TEs interspersed with genes , provides an ideal system to study the genome-wide influence of TEs on gene regulation . To analyze the magnitude of the TE effect on gene expression response to environmental changes , we profiled gene and TE transcript levels in maize seedlings exposed to a number of abiotic stresses . Many genes exhibit up- or down-regulation in response to these stress conditions . The analysis of TE families inserted within upstream regions of up-regulated genes revealed that between four and nine different TE families are associated with up-regulated gene expression in each of these stress conditions , affecting up to 20% of the genes up-regulated in response to abiotic stress , and as many as 33% of genes that are only expressed in response to stress . Expression of many of these same TE families also responds to the same stress conditions . The analysis of the stress-induced transcripts and proximity of the transposon to the gene suggests that these TEs may provide local enhancer activities that stimulate stress-responsive gene expression . Our data on allelic variation for insertions of several of these TEs show strong correlation between the presence of TE insertions and stress-responsive up-regulation of gene expression . Our findings suggest that TEs provide an important source of allelic regulatory variation in gene response to abiotic stress in maize . Transposable elements ( TEs ) , first described as “controlling elements” by Barbara McClintock [1] , are now known to make up the majority of angiosperm DNA [2]–[4] . TE insertions within genes may result in mutant alleles by changing the reading frame or splice pattern , frequently negatively affecting gene function . However , TEs also have the potential to contribute to regulation of gene expression , potentially playing an important role in responses to environmental stress [2] , [5]; McClintock initially referred to TEs as “controlling elements” based on their ability to influence the expression of nearby genes [1] , [6] . Several specific examples of TE influence on the expression of nearby genes have now been documented ( reviewed by [7]–[11] ) . TE insertions near genes may influence gene expression through several potential mechanisms , including inserting within cis-regulatory regions , contributing an outward reading promoter from the TE into the gene [12]–[15] , or providing novel cis-regulatory sequences that can act as enhancers/repressors by facilitating transcription factor binding [16] , or influencing the chromatin state of gene promoter regions [17]–[19] . Some TEs exhibit stress-responsive transcription or movement [20]–[25] . For example , expression of the tobacco Tnt1 element can be induced by biotic and abiotic stress [22]–[23] . The rice DNA transposon mPing can be activated in response to cold and salt stress [26]–[27] . The Arabidopsis retrotransposon ONSEN is transcriptionally activated by heat stress [16] , [28]–[29] . Tissue culture is a complex stress that can result in the activation of DNA transposons in maize and retrotransposons in rice [30]–[31] . There is also evidence that some of these TE responses to environmental conditions can affect the expression of nearby genes . Novel mPing MITE insertions in the rice genome in some cases resulted in up-regulation of nearby genes in response to cold or salt stress with no change in expression in control conditions [26]–[27] . The ONSEN retrotransposon insertions near Arabidopsis genes exhibit similar properties: alleles containing ONSEN insertions often show heat-responsive regulation while alleles lacking ONSEN are not up-regulated by heat stress [16] . These studies suggest that TEs can provide novel regulatory mechanisms and influence the response to environmental stress . Maize provides a good system for studying the potential influence of TEs on regulation of nearby genes . While TEs only account for ∼10% of the Arabidopsis genome [32] or ∼32% of the rice genome [33] , they contribute ∼85% to the maize genome [34]–[35] . Many TEs are located in pericentromeric regions and heterochromatic maize knobs [34] , [36] , but there are also many TE insertions interspersed between maize genes [37]–[39] . The majority of maize genes ( 66% ) are located within 1 kb of an annotated transposon [35] . In addition , allelic variation for the presence of TE insertions near genes is high in maize [39]–[41] , creating the potential for allelic regulatory differences at nearby genes . For example , polymorphic TE insertions in different haplotypes of the tb1 , Vgt1 and ZmCCT loci likely contribute to regulatory differences for these genes [42]–[44] . While there are good examples to suggest that specific TEs can influence the response of nearby genes to abiotic stress [16] , [26] it remains unclear how widespread this phenomenon is , how many genes are activated in such a TE-dependent manner , and whether multiple TE families are capable of controlling stress response . We identified a subset of TE families over-represented in the promoters of maize genes that exhibit stress-responsive up-regulation or activation of gene expression . Based on our data , as many as 20% of genes that showed increased expression in response to stress are located near a TE from one of these families . We find that stress-responsive TEs appear to provide enhancer-like activity for nearby promoters and allelic variation for TE insertions is strongly associated with variation in expression response to stress for individual genes . To test the hypothesis that genes responding to abiotic stress may be influenced by nearby TE insertions we focused our initial analyses on expression responses in the inbred B73 , for which a reference genome is available [35] . The TEs located within 1 kb of the transcription start site ( TSS ) of each gene were identified in the B73 reference genome . For each of 576 annotated TE families we determined whether genes located near the transposon were significantly enriched ( p<0 . 001 , >2 fold-enrichment and at least 10 expressed genes associated with the TE family ) for responsiveness to each of the stress conditions ( separate analyses for enrichment in up- or down-regulated genes for each stress ) relative to non-differentially expressed genes ( S3 Table ) . While the majority of transposon families are not associated with stress-responsive expression changes for nearby genes ( Fig . 2A–B; S3 Table ) , 20 TE families are significantly enriched for being located near genes with stress-responsive up-regulation and 3 TE families are associated with genes down-regulated in response to stress ( Fig . 2C; Table 1 ) . Examples of the expression changes for genes in different abiotic stresses are shown for two transposon families , ipiki and etug ( Fig . 2D ) . Genes located near ipiki are enriched for up-regulation following salt and UV stress while genes located near etug elements are enriched for heat-responsive up-regulation . Another striking example is the joemon TE family for which 59 of 68 expressed genes containing an insertion within 1 kb are activated following cold stress ( Table 1 ) . Although similar numbers of genes exhibit increased and decreased gene expression genome-wide following abiotic stress conditions , the majority of enriched TE family – stress combinations ( 28/31 ) are associated with up-regulated gene expression . For each of the stress conditions there were 4–9 TE families that are associated with up-regulation of gene expression . Some TE families are associated with altered expression in multiple stress treatments ( Table 1 , S4 Table; Fig . 2C ) and two of the TE families associated with down-regulation of gene expression under high salt stress were also associated with increased gene expression under UV stress . The TE families enriched for genes activated in response to stress include all major super-families of TEs: TIR DNA transposons , LTR gypsy-like ( RLG ) , copia-like ( RLC ) , or unknown ( RLX ) retrotransposons , and LINE elements ( Table 1 , ) . These TE families vary substantially for the number of genes that they are located near: from 30 to 3052 genes ( Table 1; S4 Table ) and are spread uniformly across the maize genome . The presence of these TEs near genes is not fully sufficient for stress-responsive expression . For each of the TE families identified , 26–87% of genes located near a TE insertion show stress responsive expression depending on the stress and the TE family . The expression levels for the TEs themselves was assessed for each of the treatments and in the majority of TE family – stress combinations ( 14 of 21 with expression data ) the TEs showed at least 2-fold increase in transcript levels in the stress treatment compared to control conditions ( Table 1 , S4 Table ) . There are several examples of TE families that exhibit increased levels of expression in a particular stress but the nearby genes are not enriched for stress-responsive expression ( S3 Table ) , suggesting that not all TEs that are influenced by a particular stress influence nearby genes . To understand what proportion of the transcriptome response to a specific abiotic stress may be explained by influences of specific TEs inserted near genes , up-regulated genes were classified according to whether they were located near a member of one of the stress-associated TE families ( 1 kb 5′ from TSS ) and whether they are up-regulated ( expressed under control and stress conditions ) or activated in response to stress ( only expressed following stress treatment ) . We found that a substantial portion of the transcriptome response to the abiotic stress could be associated with genes located near the set of 4–9 TE families that were identified as enriched for up-regulated genes ( Fig . 2E ) . In total , 5–20% of the genome-wide transcriptome response to the abiotic stress and as many as 33% of activated genes could be attributed to the genes located near one of these TE families ( Fig . 2E; S5 Table-6 ) . One possible mechanism by which these families of TEs could contribute to stress-responsive expression for nearby genes is that the TE may provide an outward-reading promoter that is stress-responsive . This model predicts that the orientation of the TE relative to the gene is important and that novel transcripts containing TE sequences fused to gene sequences would be present for up-regulated genes under stress conditions . In order to assess the importance of the orientation of the TE insertion relative to the gene , we compared the proportion of genes located on the same strand as a TE for genes up-regulated in response to stress and genes non-differentially expressed in response to stress for all TE families enriched for up-regulated genes ( S7 Table ) . While most families showed no significant difference in the proportion of genes on the same strand as the TE between the up-regulated and non-differentially expressed genes , a minority of families ( 4/20 ) showed significant enrichment . For example , 97% of the stress-responsive genes located near etug elements are on the same strand as the TE ( S7 Table ) . The potential for TEs to provide novel promoters in stress conditions was also assessed by creating de novo transcript assemblies for each of the treatment conditions ( S8 Table ) . These transcript assemblies were mapped to the reference genome to determine the transcriptional start sites in control and stress treatments . In particular , we focused on the 630 genes that had TE insertions at least 100 bp 5′ of the annotated transcription start site that had de novo transcript assemblies in both control and stress conditions . The location of the start site for the transcript assembly in control and stress conditions was compared to the location of the annotated start site and the location of the TE . There were a number of instances in which the transcript start site was located 5′ of the annotated site in both control and stress conditions and these likely reflect examples of incomplete annotation . There are 16 genes ( out of 630 with data ) that have a novel start site in the stress-treatment and not in the control that was located within or near the TE . There was not a significant enrichment for specific TE families among these 16 examples and these examples may simply reflect examples of inaccurate start site annotation without enough read depth in the control condition to identify the specific start site . These examples show that we could detect novel start sites but they suggest that it is rare for TEs to provide novel promoters in stress conditions . Alternative models include the possibility that the TE may contain cis-regulatory sequences that can act as binding sites for stress-induced transcription factors , or that the TE could influence the local chromatin environment in such a way that the region is more accessible under stress conditions . The analysis of TE distance from transcription start sites of stress-responsive genes suggests that in many cases the effect of TE on stress-responsive gene activation quickly diminishes as the distance increases beyond 500 bp – 1 kb ( S3A Fig . ) . The DREB/CBF transcription factors are often involved in transcriptional responses to abiotic stress in plants [46] . The consensus sequence for DREB/CBF binding ( A/GCCGACNT [47] ) was found in most of the TEs that were associated with stress-responsive expression for nearby genes , with the exception of elements that only exhibit UV stress response ( S3B Fig . ) . While we did not have evidence to distinguish between the possibilities that TEs provide either a sequence-specific binding site that might act as a stress-specific enhancer or influence the chromatin state in a non-sequence specific manner , our data are consistent with the TE insertions acting predominantly as local enhancers of expression rather than as novel promoters . Because individual TE copies are subject to frequent rearrangements and internal deletions , we investigated whether the presence of specific regions in each TE family were over-represented in insertions that confer stress-responsive expression . For six of the 20 TE families , this comparison revealed specific portions of the TE sequences enriched among insertions that convey stress-responsive expression . For example , naiba and etug insertions located near up-regulated genes are approximately four times as likely to contain a particular portion of the TE long terminal repeat ( LTR; p-value <0 . 001; S4 Fig . ) , and this same sequence is found in a subset of insertions of the related family , gyma , that are associated with up-regulated genes . While we did not have evidence to rule out the possibility that TEs influence the chromatin state in a non-sequence specific manner , these data indicate that the presence of particular regions of TE elements likely provide enhancer functions associated with gene expression responses to stress and help explain the variable effect of different insertions of the same family on stress-responsive expression . We assessed a number of properties of the TE-influenced stress-responsive genes in comparison with stress-responsive genes that are not associated with one of these TE families ( Table 2 ) . Stress-responsive genes located near the TE families tend to be substantially shorter in length with fewer introns . Analysis of developmental expression patterns for these genes using the B73 expression atlas [48] reveals that only 7% of the TE influenced genes are expressed in at least 5 tissues , compared to 41% of the non-TE influenced genes . The TE influenced genes are also less likely to be in the filtered gene set ( FGS ) , and the proportion of the TE influenced genes with syntenic homologs in other grass species is much lower than the proportion of non-TE influenced genes ( Table 2 ) . Each of these features was assessed separately for each of the TE families ( S7 Table ) and there is some variation for these properties among different families . These observations are compatible with the notion that TE insertions may in some cases function as enhancers that can drive expression of cryptic promoters in non-coding regions of the genome . This will result in stress-responsive production of transcripts that may be annotated as genes but may not produce functional proteins . However , 37% of TE influenced genes are included in the FGS that has been curated to remove transposon-derived sequences and a substantial proportion of the TE influenced genes are syntenic with genes from other species , have GO annotations , and could contribute to functional responses to stress ( Table 2 , S7 ) . These results suggest that many of TE influenced genes are not derived from TEs . We were particularly intrigued by the question of whether polymorphic insertions of TEs from families associated with stress-responsive expression of nearby genes might contribute to allelic variation for stress-responsive gene expression . The consistency of stress-responsive expression of TE-associated genes across the three inbred lines surveyed varied widely across TE families ( Fig . 3A; S5 Fig . ) . In order to assess whether insertions of TEs from the families associated with stress-responsive gene expression could contribute to allelic variation for gene expression regulation , we used whole-genome shotgun re-sequencing data from Mo17 and Oh43 [49] to find potential novel insertions of elements from the TE families identified in this study . We identified 23 novel ( not present in B73 ) high-confidence insertions of TEs from these families located within 1 kb of the TSS of maize genes and validated them by PCR ( S9 Table ) . Of the 10 genes with detectable expression in our RNAseq experiments , 7 showed stress-responsive up-regulation/activation associated with the TE-containing alleles ( Fig . 3B ) . This analysis was expanded to additional genotypes by using PCR to detect the presence/absence of the TE insertion in a diverse set of 29 maize inbred lines that were selected to represent diverse North American germplasm from the stiff stalk , non-stiff stalk , iodent , tropical , sweet corn and popcorn population groups . The relative expression of the gene in stress compared to control treatment was also determined in each inbred using quantitative RT-PCR ( S10 Table ) . For each of these genes we found that the alleles that lack the transposon insertion did not exhibit stress-responsive expression ( Fig . 4 ) , with the exception of one genotype for gene GRMZM2G108057 . In contrast , the majority of the alleles that contain the TE ( 60–88% ) exhibit stress-responsive up-regulation . Although for a single insertion we cannot rule out the possibility that differential expression is due to a different polymorphism on the same haplotype as the TE , the fact that we see TE-associated expression change in multiple genes for each of the TE families ( Table . S10 ) argues strongly against such an explanation in general . These data thus provide evidence that insertion polymorphisms for the TE families identified here can generate novel expression responses for nearby genes . Transposable elements are a major component of many eukaryotic genomes , and constitute the majority of plant nuclear DNA . TEs are usually considered as a deleterious or neutral component of these genomes . However , the interplay between TEs and genes may have important functional contributions to plant traits . There are clear examples of TE insertions that are linked to functionally relevant alleles in maize such as Tb1 [42] Vgt1 [43] and ZmCCT [44] . In these cases , a transposon insertion within a distant cis-regulatory sequence influences the regulation of adjacent genes . There are also examples of functionally relevant TE insertions in tomato , melons and citrus [50]–[52] that can influence gene expression , potentially through chromatin influences that generate obligate epialleles . Previous research in several plant species has suggested that at least some families of transposable elements may become transcriptionally activated following environmental stress . Tissue culture has been shown to result in activation of transposons and retrotransposons in a number of plant species [30]–[31] . There are also examples of transcriptional activation of TEs in response to specific abiotic stresses in tobacco [22] , rice [26]–[27] and Arabidopsis [16] , [28]–[29] . It is expected that the stress responsive expression of these TEs involves local enhancers that result in up-regulation of the TE promoter in response to stress . These local enhancers could also act upon other nearby promoters . There are a handful of examples in which transposon insertions have been linked to stress-responsive expression of nearby genes including the mPING insertions associated with cold-responsive expression in rice [26]–[27] and ONSEN insertions associated with heat-stress responsive expression in Arabidopsis [16] . If this is a common occurrence then we might expect it to be even more prevalent in a genome such as maize where many genes are closely surrounded by TEs . Our analysis suggests that a small number of TE families are associated with stress-responsive expression for nearby genes . While some TE families were associated with multiple stresses , we found a different subset of TE families for each abiotic stress that was evaluated . In most cases , these same TEs themselves were up-regulated in response to the stress treatment . However , we also noted that there were some TE families that themselves exhibit strong up-regulation but did not have apparent influences on a significant portion of nearby genes . Even though the majority of stress responsive regulation of gene expression is not associated with TEs , based on our data , up to 20% of genes up-regulated in response to stress and as many as 33% of genes activated in response to stress could be attributed to regulation by TEs . One of the alternative explanations would argue that only a small number of genes localized close to a TE are truly influenced by this TE insertion for their expression , while other up-regulated genes are secondary targets and are regulated by the TE influenced genes . Although some of the TE influenced genes we identified could be secondary targets , secondary target genes would not preferentially co-localize with TEs from specific families . The analysis of the nearby genes that were influenced by TEs suggests that many of them may not actually be protein coding genes . In one sense , this is an expected result . If an enhancer sequence is mobilized within the genome it will have the potential to influence expression from both gene promoter as well as cryptic promoters that may not be associated with coding sequences . The gene annotation efforts in maize have relied upon EST and RNA-seq expression data from a variety of conditions . In many cases the genes that were found to exhibit stress-responsive expression associated with TEs were only annotated as genes based upon evidence of their expression . We would expect that insertions of the TEs that provide stress-responsive enhancer activity would influence cryptic promoters not associated with genes in many cases , but would also affect the expression of nearby protein coding genes . The frequency of each appeared to vary among TE families , with some , like nihep , showing little difference between TE-influenced and non-TE-influenced up-regulated genes ( S7 Table ) . Overall , while TE influenced stress-responsive genes are enriched for short sequences with limited homology to sequences in other species , a significant proportion are longer , have several exons , are conserved in other species , and have GO annotations . A particularly interesting aspect of these results is the potential mechanism for creating novel cis-regulatory variation . Our understanding of how particular genes might acquire novel regulatory mechanisms is limited . In many cases SNPs within promoters or regulatory sequences have limited functional significance . Therefore , it is difficult to envision how a novel response to a particular environmental or developmental cue would arise . Variation in TE insertions has the potential to create novel regulatory alleles by providing binding sites for transcription factors or influencing chromatin . We provide evidence that allelic variation for stress-responsive expression can be created by the insertion of certain TEs . Variation in TE insertions would generate allelic diversity that could influence an organism's response to environmental conditions and would provide phenotypic variation that could be acted upon by selection . As with other types of variation , most examples of novel stress-responsive expression are likely to be neutral or deleterious and would not be expected to rise in allele frequency . However , a subset of novel stress-responsive expression patterns could be beneficial and become targets of natural or artificial selection contributing to gene regulation networks of environmental stress response . B73 , Mo17 , and Oh43 maize seedlings were grown at 24°C in 1∶1 mix of autoclaved field soil and MetroMix under natural light conditions in July 2013 . For cold stress , seedlings were incubated at 5°C for 16 hours . For heat stress , seedlings were incubated at 50°C for 4 hours . For high salt stress , plants were watered with 300 mM NaCl 20 hours prior to tissue collection . UV stress was applied in the growth chamber conditions using UV-B lamps for 2 hours prior to tissue collection . UV stress causes accumulation of DNA mutations but most of such mutations would either have no immediate effect on gene expression or would lead to decrease or abortion of expression of specific genes . Light conditions were the same for all stress and control conditions . Whole above ground tissue was collected for 14 day old seedlings at 9am and six seedlings were pooled together for each sample . Three replicates for heat and cold-treated B73 and Mo17 seedlings were grown 3 days apart . Three biological replicates of cold and heat stress and control conditions for B73 and Mo17 were prepared with eight plants pooled for each of the replicates . One biological replicate of high salt and UV stress conditions for B73 and Mo17 as well as all four stress and control conditions for Oh43 were prepared similarly . RNA was isolated using Trizol ( Life Technologies , NY , USA ) and purified with LiCl . All RNA samples were prepared by the University of Minnesota BioMedical Genomics Center in accordance with the TruSeq library creation protocol ( Illumina , San Diego , CA ) . Samples were sequenced on the HiSeq 2000 developing 10–20 million reads per sample . Transcript abundance was calculated by mapping reads to the combined transcript models of the maize reference genome ( AGPv2 ) using TopHat [53] . Reads were filtered to allow for only uniquely mapped reads . A high degree of correlation between replicates was observed ( r>0 . 98 ) . RPKM values were developed using ‘BAM to Counts' across the exon space of the maize genome reference working gene set ( ZmB73_5a ) within the iPlant Discovery Environment ( www . iplantcollaborative . org ) . Genes were considered to be expressed if RPKM>1 and differentially expressed if log2 ( stress/control ) > 1 or log2 ( stress/control ) <-1 . Statistical significance of expression differences was determined using DeSeq package for all fully replicated samples [45] . For each gene , transposons located within 1 kb of the transcription start site ( TSS ) were identified using the B73 reference genome annotation [35] and maize TE elements database [34] . TE distance from transcription start sites was determined using the closestBed tool from the BEDTools suite [54] where TEs upstream were given a positive distance value and TEs downstream were given a negative distance value . The transcriptional start site was defined as the 100-bp window intersecting the first base pair of a gene model from the maize genome gene set ( ZmB73_5b ) . The proportion of up-regulated , down-regulated , and non-differentially expressed genes that have an insertion of a TE element from a particular family was calculated for 576 TE families for four stress conditions . Fold-enrichment of up-regulated genes relative to all expressed genes ( the sum of up-regulated , down-regulated and non-differentially expressed genes ) and relative to all genes was calculated for all TE family/stress combinations . Given the total number of expressed genes associated with each TE family and the proportion of up- and down-regulated genes , the expected numbers of up- and down-regulated genes and non-differentially expressed genes were calculated and a multinomial fit test was conducted . TE families that had over 10 expressed genes associated with them , fold enrichment of up- or down-regulated genes over 2 , and p value <0 . 001 were considered “enriched” for up- or down-regulated , respectively . Similar analysis was conducted for working gene set and filtered gene set genes . The same set of “enriched” TE families was found for both groups of genes as well as when fold enrichment was calculated relative to all expressed genes or to all genes associated with TEs from a particular family . To assess expression changes in response to stress for TE families , the overlap tool from BEDTools suite [54] was used to obtain read counts per each TE accession . The output file from alignment ( BAM ) was mapped to TE positions listed in the TE GFF file downloaded from maizesequence . org . Each read was required to have 100% overlap with a given TE region . The reads mapping to more than 5 locations in the genome were omitted . The reads were then summed across the entire TE region and combined for each of the TE families . Tissue specific expression data is from the maize gene expression atlas [47] . Genes with RPKM of <1 were considered non-expressed . Orthologous and paralogous gene pairs were inferred from [55] . De novo assemblies for the control and each stress were performed for the B73 inbred line . Prior to assembly reads were cleaned using cutadapt version 1 . 4 . 1 [56] requiring a minimum read length of 30 . Reads were further cleaned with the FASTX toolkit version 0 . 0 . 14 ( http://hannonlab . cshl . edu/fastx_toolkit/ ) using the fastx_artifacts_filter and the fastq_quality_trimmer requiring a minimum read length of 30 and a minimum quality score of 20 . Read pairs for which one read was discarded during the read cleaning pipeline were discarded from further analyses . Within each treatment all reads across biological replicates were combined for treatment specific assemblies . The transcriptome assembly was conducted using Trinity version r20140413 [57] using default parameters and requiring a minimum transcript length of 200 . Each assembly was assessed based on the percentage of transcripts that could map back to the reference genome sequence and the percentage of input reads that could map to the final assembly . Transcripts were mapped to the maize v2 reference genome sequence ( http://ftp . maizesequence . org ) using GMAP version 2012-06-02 [58] with default parameters . Input reads were mapped back to the assembly using Bowtie version 0 . 12 . 9 [59] and TopHat version 1 . 4 . 1 [53] allowing a minimum and maximum intron size of 5 and 10 , 000 and the —no-novel-indels function . Assemblies were linked to stress differentially expressed genes based on the GMAP alignments . The start position for control and stress assembled transcripts were compared to identify transposable elements that act as either promoters or enhancers under stress conditions . Instances where the control assembled transcript starts within the gene model and the stress assembled transcript starts near or within the TE would provide evidence that the TE is acting as a promoter . Nonreference TE insertions were detected for Oh43 and Mo17 using relocaTE [60] , whole genome sequence from the NCBI SRA ( Oh43: SRR447831-SRR447847; Mo17: SRR447948-SRR447950 ) , and consensus TE sequences from the maize TE database [34] . Reads containing TEs were identified by mapping to consensus TE sequences , trimming portions of reads mapping to a TE , and mapping the remaining sequence to the reference genome . Nonreference TEs were identified when at least one uniquely mapped read supported both flanking sequences of the nonreference TE , overlapping for a characteristic distance that reflects the target site duplication generated upon integration ( five nucleotides for all LTR retrotransposons , nine nucleotides for DNA TIR mutator ) . Primers for six TE polymorphic genes up-regulated under stress conditions in Oh43 or Mo17 but not in B73 were designed using Primer 3 . 0 software [61] and PCR reactions were performed using Hot Start Taq Polymerase ( Qiagen , Ca , USA ) . Primer sequences are shown in S11 Table . cDNA synthesis and qPCR analysis were performed as described in [62] . Primers for 10 differentially expressed genes and two control genes ( GAPC and mez1 ) were designed using Primer 3 . 0 software [57] . Primer sequences are shown in S10 Table .	Transposable elements are mobile DNA elements that are a prevalent component of many eukaryotic genomes . While transposable elements can often have deleterious effects through insertions into protein-coding genes they may also contribute to regulatory variation of gene expression . There are a handful of examples in which specific transposon insertions contribute to regulatory variation of nearby genes , particularly in response to environmental stress . We sought to understand the genome-wide influence of transposable elements on gene expression responses to abiotic stress in maize , a plant with many families of transposable elements located in between genes . Our analysis suggests that a small number of maize transposable element families may contribute to the response of nearby genes to abiotic stress by providing stress-responsive enhancer-like functions . The specific insertions of transposable elements are often polymorphic within a species . Our data demonstrate that allelic variation for insertions of the transposable elements associated with stress-responsive expression can contribute to variation in the regulation of nearby genes . Thus novel insertions of transposable elements provide a potential mechanism for genes to acquire cis-regulatory influences that could contribute to heritable variation for stress response .	[ "Abstract", "Introduction", "Results", "Discussion", "Materials", "and", "Methods" ]	[ "genetics", "plant", "genetics", "biology", "and", "life", "sciences" ]	2015	Transposable Elements Contribute to Activation of Maize Genes in Response to Abiotic Stress
A better description of the clinical and laboratory manifestations of fatal patients with dengue hemorrhagic fever ( DHF ) is important in alerting clinicians of severe dengue and improving management . Of 309 adults with DHF , 10 fatal patients and 299 survivors ( controls ) were retrospectively analyzed . Regarding causes of fatality , massive gastrointestinal ( GI ) bleeding was found in 4 patients , dengue shock syndrome ( DSS ) alone in 2; DSS/subarachnoid hemorrhage , Klebsiella pneumoniae meningitis/bacteremia , ventilator associated pneumonia , and massive GI bleeding/Enterococcus faecalis bacteremia each in one . Fatal patients were found to have significantly higher frequencies of early altered consciousness ( ≤24 h after hospitalization ) , hypothermia , GI bleeding/massive GI bleeding , DSS , concurrent bacteremia with/without shock , pulmonary edema , renal/hepatic failure , and subarachnoid hemorrhage . Among those experienced early altered consciousness , massive GI bleeding alone/with uremia/with E . faecalis bacteremia , and K . pneumoniae meningitis/bacteremia were each found in one patient . Significantly higher proportion of bandemia from initial ( arrival ) laboratory data in fatal patients as compared to controls , and higher proportion of pre-fatal leukocytosis and lower pre-fatal platelet count as compared to initial laboratory data of fatal patients were found . Massive GI bleeding ( 33 . 3% ) and bacteremia ( 25% ) were the major causes of pre-fatal leukocytosis in the deceased patients; 33 . 3% of the patients with pre-fatal profound thrombocytopenia ( <20000/µL ) , and 50% of the patients with pre-fatal prothrombin time ( PT ) prolongation experienced massive GI bleeding . Our report highlights causes of fatality other than DSS in patients with severe dengue , and suggested hypothermia , leukocytosis and bandemia may be warning signs of severe dengue . Clinicians should be alert to the potential development of massive GI bleeding , particularly in patients with early altered consciousness , profound thrombocytopenia , prolonged PT and/or leukocytosis . Antibiotic ( s ) should be empirically used for patients at risk for bacteremia until it is proven otherwise , especially in those with early altered consciousness and leukocytosis . Dengue is the most prevalent mosquito-borne viral infection in the world [1] . Clinically , dengue ranges from asymptomatic , nonspecific febrile illness , classic dengue , to dengue hemorrhagic fever/dengue shock syndrome ( DHF/DSS ) [1] . Fatality rate and causes of fatality in dengue-affected patients greatly varied from one report to another [1]–[13] . While DSS was the major cause of fatality in patients with dengue illness reported in some series [1]–[13] , causes other than DSS were predominantly responsible for fatality reported in others [2] , [10] , [12]–[14] . However , only a small number of dengue-attributed mortality cases were included for analysis in each of these series [2] , [8] , [10]–[12] . A better description of the clinical and laboratory presentations of cases with fatal outcome may lead clinicians to an earlier recognition of the warning signs of severe dengue resulting in timely and improved management . To achieve this , the importance of continuous analysis of relevant findings in fatal patients from dengue-affected populations cannot be overemphasized . Among the major dengue epidemics in Taiwan over the past 3 decades , a large dengue outbreak caused by DENV-1 occurred in 1987–1988 in southern Taiwan , followed by another one caused by DENV-2 in 2002 in the same geographic area [15] . During the 2002 dengue epidemic in southern Taiwan , more than 5000 dengue cases were reported , and most of them were DHF that developed in adults [15] , [16]; of note , dengue-related fatality was found in 10 adults admitted at Kaohsiung Chang Gung Memorial Hospital ( KSCGMH ) , a 2500-bed facility serving as a primary care and tertiary referral center in this area . In this study , we retrospectively compared the clinical and laboratory features of dengue-affected adults who turned out to be fatal and who survived , and analyzed the fatal dengue cases aiming at understanding the causes of mortality and clarifying the clinical and laboratory evolutions preceding mortality . The data in this work were analyzed anonymously , and the study was conducted with a waiver of patient consent approved by the Institutional Review Board of KSCGMH ( Document No . 99-2671B ) . Patients with the diagnosis of dengue admitted to KSCGMH between June and December 2002 were potentially eligible for inclusion in this retrospective study . All clinically diagnosed dengue cases were serologically confirmed by at least one of the following criteria: ( i ) a positive reverse transcriptase-polymerase chain reaction ( RT-PCR ) , ( ii ) a positive enzyme-linked immunosorbent assay for specific immunoglobulin M antibody for dengue virus in the acute-phase serum , and ( iii ) at least fourfold increase in dengue-specific hemagglutination inhibition titers in the convalescent serum when compared to that in acute-phase serum [17] . The diagnosis of DHF was established based on the presence of fever , hemorrhage , thrombocytopenia ( <100×109 cells/L ) and clinical evidence of plasma leak ( i . e . , presence of hemoconcentration , pleural effusion , ascites and/or hypoalbuminemia ) indicating increased vascular permeability [17] . Hemoconcentration referred to >20% increase in hematocrit calculated as: ( maximum hematocrit - minimum hematocrit ) ×100/minimum hematocrit . The severity of DHF in serologically confirmed dengue patients was stratified based on the World Health Organization ( WHO ) criteria . Grade I referred to a positive tourniquet test result being the only hemorrhagic manifestation , while grade II referred to spontaneous bleeding . Grade III referred to a circulatory failure manifested by a rapid and weak pulse , as well as a narrowing pulse pressure ( ≦20 mmHg ) , whereas grade IV referred to a profound shock , with an undetectable pulse or blood pressure [17] . Grades III and IV of DHF were grouped as DSS [17] . All fatal DHF patients in this series resulted in dengue-related mortality which referred to the death that occurred within three weeks after hospitalization because of DHF . Hypothermia referred to a temperature <36°C detected at least twice from the ear drum of a dengue-affected patient . Massive gastrointestinal ( GI ) bleeding was defined as the passage of large amount of tarry or bloody stool coupled with hemodynamic instability and/or rapid decrease in hemoglobin level to ≦7 . 0 g/dL . Acute renal failure was defined as a rapid increase in serum creatinine ( Cr ) level ≧0 . 5 mg/dL when compared to that found at the patient's hospital presentation . Acute hepatic failure was defined as a raise in serum alanine aminotransferase ( ALT ) level ≧400 U/L ( reference value , <40 U/L ) . Leukocytosis was defined as a peripheral white cell count >12000/µL . Bandemia referred to the presence of band-form granulocytes in the peripheral blood . Profound thrombocytopenia referred to a platelet count <20000/µL . Prolongation of prothrombin time ( PT ) was defined as a PT≥3 seconds than a control , and prolongation of activated partial thromboplastin time ( APTT ) as an APTT ≥20% than a control . Concurrent bacteremia was defined as a positive bacterial growth from blood that was sampled for culture within 72 h after the patient was hospitalized for dengue . Demographic , clinical , laboratory and imaging information of the included DHF patients were retrieved from the retrospective review of their medical charts for analyses . Initial laboratory data referred to data detected from the dengue-affected patients upon their arrival at KSCGMH . Pre-fatal laboratory data were those detected from the blood specimens of the fatal patients sampled within the immediate 48 h before fatality . The 309 DHF patients included for analyses were separated into two groups: those who were fatal ( fatal group , N = 10 ) and those survived ( control group , N = 299 ) . The survived patients were those with detailed information available . We compared the demographic , clinical , imaging characteristics and initial laboratory data of the fatal patients and those of the controls , as well as the pre-fatal laboratory data and the initial laboratory data of the fatal patients . Mann-Whitney U test was used in comparison of continuous variables , while the Fisher's exact test was used to in assessment of dichotomous variables . A 2-tailed P<0 . 05 was considered statistically significant . A total of 714 adults with dengue illness were found at KSCGMH during the study period , and among them , 10 ( 8 men and 2 women; median age , 63 . 5 years [range , 33–78] ) with DHF ( 7 grade II DHF and 3 DSS ) turned out to be fatal , accounting for a dengue-related mortality rate of 1 . 3% ( details are shown in Table S1 ) . Of these fatal patients , the time lapses between dengue onset and hospital presentation ranged from 1 to 6 days ( median , 2 days ) , between hospital presentation to fatality 2 to 18 days ( median , 4 . 5 days ) , and between dengue onset to fatality 4 to 21 days ( median , 7 . 5 days ) . With the exception of patient 2 in whom the dengue diagnostic test was carried out from the blood specimen collected on the day 3 of his hospitalization , all patients had their blood sampled for dengue diagnosis within 24 h after admission . The median from dengue onset to the definitive diagnosis made was 5 days ( range , 4–11 days ) . Infection with DENV-2 in all fatal patients was confirmed by RT-PCR . Manifestations indicating plasma leak in these fatal patients included hemoconcentration ( patients 2 , 5–10 ) , presence of pleural effusion ( patients 1 , 3–6 , 8 and 10 ) and hypoalbuminemia ( patients 1 , 2 and 4 ) . Seven patients ( patients 1–4 , 6 , 7 and 9 ) with grade II DHF experienced shock resulting from bacterial sepsis ( patients 1 and 4 ) , concurrent bacterial sepsis and massive GI bleeding ( patient 9 ) , and massive GI bleeding ( patients 2 , 3 , 6 , and 7 ) . DSS alone was found in 3 ( patients 5 , 8 and 10 ) patients . Shock , regardless of cause , developed 1 to 16 days ( median , 3 days ) after their hospitalizations , and 4 and 17 days ( median , 6 . 5 days ) after dengue onset . Among the 3 DSS patients , DSS was recognized on day 3 ( patients 8 and 10 ) and day 6 ( patient 5 ) of their hospitalization , respectively . Patient 2 experienced 2 episodes of massive GI bleeding with hypovolemic shock on day 8 and day 16 of his hospital stay , respectively . Patient 4 with an underlying lung cancer suffered septic shock on day 15 of his hospitalization . The demographic , clinical and laboratory information of the fatal patients and controls is summarized in Tables 1 and 2 . A variety of clinical manifestations were found in each of these 10 fatal patients ( Table 1 ) . The leading ones , in decreasing order , were fever ( >38°C ) ( 90% ) , GI bleeding ( 90% ) , pleural effusion ( 70% ) , bone pain and cough ( each 60% ) . The previously reported early warning signs for severe dengue [11] , [17] , [18] , namely , persistent vomiting was found in 4 ( 40% ) patients , and sustained abdominal pain in 2 ( 20% ) . Pulmonary edema developed in 3 ( patients 4 , 5 and 8 ) patients; 2 of them with DSS experienced acute pulmonary edema emerged on day 5 ( patient 5 ) and day 6 ( patients 8 ) after dengue onset , respectively , while the other one ( patient 4 ) with lung cancer and hypoalbuminemia ( serum albumin , 1 . 4 g/dL [normal range , 3 . 0–4 . 5 g/dL] ) experienced septic shock on day 15 of hospitalization ( day 17 after dengue onset ) thus receiving fluid resuscitation , and pulmonary edema was found on following day . Acute renal failure was found in all of 8 patients ( patients 1 , 2 , 4–9 ) with data available , and acute hepatic failure in 4 ( patients 1 , 6–8 ) ( 57 . 1% ) of 7 patients with data available . Concurrent bacteremia was noted in 3 ( patients 1 , 4 and 9 ) ( 37 . 5% ) of 8 fatal patients from whom blood was sampled for bacterial culture within 72 h after their hospitalization ( Table S1 ) . Of these 3 bacteremic patients , one ( patient 1 ) experienced Klebsiella pneumoniae meningitis , while the other two experienced primary K . pneumonia ( patient 4 ) and Enterococcus faecalis bacteremia ( patient 9 ) , respectively . Of a total of 9 patients ( patients 2–10 ) with GI bleeding , 4 ( patients 3 , 7 , 9 and 10 ) ( 44 . 4% ) developed GI bleeding at their arrival; 5 ( patients 2 , 3 , 6 , 7 and 9 ) ( 55 . 5% ) experienced massive GI bleeding , and 3 ( patients 3 , 7 and 9 ) ( 33 . 3% ) developed massive GI bleeding within 24 h after admission . Among the 5 patients with massive GI bleeding , E . faecalis bacteremia was found in one ( patient 9 ) ; active bleeding was endoscopically found in another ( patient 3 ) with a gastric ulcer , and in the other ( patient 7 ) with hemorrhagic gastritis . Only patients 3 and 7 received endoscopic examination . Among the 5 patients with massive GI bleeding , acute renal failure developed in 2 ( patients 2 and 9 ) , and concurrent acute renal and hepatic failure in the other 2 ( patients 6 and 7 ) . Of the 5 fatal patients with consciousness disturbance , 4 ( patients 1 , 2 , 7 and 9 ) were found to developed altered consciousness within 24 hours and one ( patient 5 ) in the day 4 of his hospital stay . All of these 5 patients had blood sampled for bacterial culture , and one of them ( patient 1 ) had additional cerebrospinal fluid sampled for bacterial culture . Among the 4 patients with early altered consciousness , massive GI bleeding alone ( patients 7 ) , uremia and massive GI bleeding ( patient 2 ) , E . faecalis bacteremia and massive GI bleeding ( patient 9 ) , and K . pneumoniae meningitis and bacteremia ( patient 1 ) each were found in one . Hypernatremia ( serum sodium >170 meq/L [normal range , 134–148 meq/L] ) was additionally found in patient 2 in day 8 of his hospital stay . Altered consciousness abruptly developed in patient 5 on day 4 of his hospitalization which resulted from subarachnoid hemorrhage disclosed by a brain computed tomography , and a cerebral angiographic study was deferred because of his critical condition and acute renal failure in particular; his blood culture for bacteria was negative , and although hyperkalemia ( serum potassium , 7 . 9 meq/L [normal range , 3 . 6–5 . 0 meq/L] ) was found on day 7 , hemodialysis was not carried out as it was hemodynamically unstable . Neither hyperglycemia nor hypoglycemia was found in the 10 fatal patients in this series . Hyponatremia was not found in our series . Serum calcium level was not assayed in these fatal patients . Hypothermia was noted in 2 ( 20% ) patients ( patients 6 and 9 ) . One patient ( patient 9 ) with hypothermia detected at arrival experienced concurrent primary E . faecalis bacteremia , while the other ( patient 6 ) experienced abrupt change in temperature with a rapid switch from fever to hypothermia on day 4 of her hospital stay , and her blood bacterial culture was negative . All patients experienced respiratory failure that necessitated mechanical ventilatory support . The mean time from patient's hospital presentation to starting mechanical ventilation was 3 days ( range , 1–6 days ) , and the root causes of respiratory failure included massive GI bleeding ( patients 3 , 6 , 7 and 9 ) , sepsis ( patient 1 ) , DSS ( patients 8 and 10 ) , subarachnoid hemorrhage ( patient 5 ) , persistent drowsiness occurred within 24 h after hospitalization ( patient 2 ) and lung cancer with pleural effusion ( patient 4 ) . Intravenous fluid including 0 . 9% saline , Ringer's lactate , and 5% dextrose in 0 . 9% saline was administered at infusion rates ranging from 0 . 6 mL/Kg BW/h to 2 . 7 mL/Kg BW/h for the 10 fatal patients before development of shock and/or severe GI hemorrhage . In addition , transfusion of platelets and/or other blood component ( s ) ( i . e . , packed red blood cells and/or fresh frozen plasma ) was given for these fatal patients . Intravenous fluid replacement and blood transfusion were detailed in Table S1 . Prior to shock development , intravenous fluid supplements with 0 . 9% saline for the 3 patients with DSS was 1 . 6 mL/Kg BW/h ( patients 5 ) , 1 . 3 mL/Kg BW/h ( patient 8 ) and 0 . 8 mL/Kg BW/h ( patient 10 ) , respectively . Markedly elevated hemoglobin levels were found in patients with DSS on the day shock developed . Only platelet transfusion was given for these 3 patients before development of DSS . Among the 5 ( patients 2 , 3 , 6 , 7 and 9 ) patients with massive GI bleeding , intravenous fluid ( 0 . 9% saline or Ringer's lactate ) supplement was infused at rates ranging from 1 . 4 mL/Kg BW/h to 2 . 5 mL/Kg BW/h before development of hypovolemic shock , and 2 to 8 units of packed red blood cells were transfused on the day the massive GI bleeding emerged ( Table S1 ) . Because superimposing bacterial sepsis could not be excluded in these critically ill patients , all of them received intravenous antibiotic ( s ) within 72 h after admission . Upon hospitalization , the 3 patients ( patients 1 , 4 and 9 ) with concurrent bacteremia received empirical antibiotic ( s ) ( i . e . , ceftriaxone for patient 1 , piperacillin and gentamicin for patient 4 , and ceftriaxone and penicillin for patient 9 ) to which the subsequently isolated bacteria were susceptible in vitro . When it comes to causes of fatality in these10 fatal patients , intractable massive GI bleeding with hypovolemic shock was found in 4 ( 40% ) ( patients 2 , 3 , 6 and 7 ) , DSS alone in 2 ( 20% ) ( patients 8 and 10 ) , while DSS with subarachnoid hemorrhage ( patient 5 ) , K . pneumoniae bacteremia and meningitis with septic shock ( patient 1 ) , sepsis due to mechanical ventilation associated pneumonia ( patient 4 ) , as well as concurrent E . faecalis bacteremia and intractable massive GI bleeding with shock ( patient 9 ) each ( 10% ) were found in one patient . None of the fatal patients underwent autopsy . Pre-fatal leukocytosis was found in 6 ( patients 4 , 5 , 6 , and 8–10 ) ( 66 . 7% ) of the 9 patients ( patients 1 , 2 , and 4–10 ) with data available , and bandemia in 4 ( patients 1 , 2 , 5 and 7 ) ( 66 . 7% ) of the 6 patients ( patients 1 , 2 and 4–7 ) in whom the differential count of peripheral white blood cells was available . Of the 6 patients with development of pre-fatal leukocytosis , 2 ( patients 4 and 10 ) had leukopenia upon their arrival , 4 had their blood sampled for bacterial culture and E . faecalis bacteremia ( patient 9 ) was found in one patient ( 25% ) , all experienced GI bleeding , and 2 ( patients 6 and 9 ) ( 33 . 3% ) developed massive GI bleeding . Prolongation of pre-fatal PT was found in 6 ( 75% ) ( patients 2 , 4–6 , 8 and 9 ) of the 8 ( patients 1 , 2 , 4–6 , and 8–10 ) patients with data available . Of note , all the 6 patients with pre-fatal PT prolongation developed GI bleeding , and of them , 3 ( 50% ) experienced massive GI bleeding . Pre-fatal profound thrombocytopenia was found in 6 ( 60% ) fatal patients ( patients 1 , 3 , 5 and 8–10 ) ; of them , 5 ( patients 3 , 5 and 8–10 ) ( 83 . 3% ) developed GI bleeding , and 2 ( 33 . 3% ) ( patients 3 and 9 ) experienced massive GI bleeding . Pre-fatal hyperkalemia was found in only 1 ( patient 5 ) of the 6 ( patients 2 , 4 , 5 , 7–9 ) patients with data available . Significant differences in demographics and clinical manifestations between fatal patients and controls included male gender ( 80% vs . 44 . 1% , P = 0 . 047 ) , hypovolemic shock due to massive GI bleeding ( 50% vs . 0 . 7% , P<0 . 001 ) , concurrent bacteremia ( 37 . 5% vs . 3 . 9% . P = 0 . 010 ) , concurrent bacteremia with shock ( 25% vs . 1 . 3% , P = 0 . 022 ) , DSS ( 40% vs . 2 . 3% , P<0 . 001 ) , pulmonary edema ( 30% vs . 2 . 8%; P = 0 . 005 ) , acute renal failure ( 100% vs . 2%; P<0 . 001 ) , acute hepatic failure ( 57 . 1% vs . 4 . 4%; P<0 . 001 ) , hypothermia ( 20% vs . 0% , P = 0 . 001 ) , GI bleeding ( 90% vs . 16% , P<0 . 001 ) , massive GI bleeding ( 50% vs . 0 . 7%; P<0 . 001 ) , subarachnoid hemorrhage ( 10% vs . 0% , P = 0 . 032 ) and early altered consciousness ( 40% vs . 0% , P<0 . 001 ) ( Table 1 ) . Significant higher proportion of bandemia ( 37 . 5% vs . 1 . 8%; P = 0 . 001 ) from initial laboratory data between the fatal patients and the controls , and significant higher proportion of pre-fatal leukocytosis ( 66 . 7% vs . 10%; P = 0 . 020 ) and lower pre-fatal platelet count ( median , 17000 cells/µL vs . 35000 cells/µL; P<0 . 001 ) as compared to the initial laboratory data of the fatal patients were found ( Table 2 ) . The time interval from the dengue onset to patients' arrival at KSCGMH between the fatal and control groups did not differ significantly ( median , 2 days vs . 3 days; P = 0 . 055 ) ( Table 1 ) . The timing of admission in both the fatal and control groups allowed us to evaluate the critical evolutionary changes in the dengue-affected patients because critical events ( e . g . , dropped blood pressure and circulation collapse ) usually occur between day 3 and day 7 of the disease course [17] , [18] . Dengue case fatality rate was reported to vary from 0 . 5% to 5 . 0% [2]–[4] , [7] , [9] , [10] , [12] . However , once DSS developed , the case fatality may soar to as high as 12–44% [3]–[5] . Our series showed that of all dengue-related deaths , DSS alone accounted for only 20% , while intractable massive GI bleeding alone for 40% , and DSS with concurrent subarachnoid hemorrhage , intractable massive GI bleeding with concurrent bacteremia , bacterial sepsis with meningitis , and sepsis due to ventilator associated pneumonia each were responsible for 10% . DSS is characterized by severe plasma leak that leads to rapidly developed shock , and timely volume replacement is the cornerstone of therapy for the affected patients [17] . Notably , the volumes of intravenous fluid supplement prior to the full blown development of DSS in the 3 patients ( patients 5 , 8 , 10 ) in our series were obviously suboptimal [17] , [18]; in spite of the subsequent fluid resuscitation and blood/blood component transfusion , they died of profound shock and multi-organ failure between day 6 and day 7 after the onset of illness . The pulmonary edema developed in the 2 patients ( patients 5 and 8 ) with DSS on day 5 and day 6 after the dengue onset , respectively , was accompanied by a concurrent marked hemoconcentration ( see Table S1 for details ) suggested continuous fluid leakage from the intravascular compartment to the extravascular compartment and the lung alveolar space in particular , leading to profound shock and pulmonary edema . In contrast , the pulmonary edema developed on day 16 in patients 4 who had an underlying lung cancer with malignant pleural effusion obviously resulted from fluid overload . The latest WHO scheme classified dengue in terms of clinical severity as severe dengue ( i . e . , presence of severe bleeding , severe plasma leak and/or severe organ involvement ) or non-severe dengue; for practical reasons , patients with non-severe dengue were further separated into those with warning signs ( i . e . , abdominal pain , persistent vomiting , clinical fluid accumulation , mucosal bleeding , lethargy/restless , liver enlargement , and increase in hematocrit in concurrent with rapid decrease in platelet count ) and those without them [18] . Severe dengue patients with aggravated plasma leak and/or bleeding necessitate aggressive fluid resuscitation and additional blood transfusion as necessary , while non-severe dengue patients with warning sign ( s ) require strict observation , appropriate medical intervention and intravenous hydration , as they are at high risk for evolving into critical phase–severe dengue [18] . In addition to the aforementioned ones , our data suggest that leukocytosis , bandemia and hypothermia may be warning signs of severe dengue . From the pathophysiological point of view , leukocytosis and/or bandemia indicates a superimposing bacterial infection and/or other stressful stimuli [19] . Our data suggested that massive GI bleeding ( 33 . 3% ) and bacteremia ( 25% ) be the major causes of DHF patients' pre-fatal leukocytosis . Significantly , leukocytosis was found in the deceased patients before their death , and bandemia was found at the fatal patients' hospital presentation ( Table 2 ) ; the latter suggest that bandemia may be an early warning parameter of severe dengue . Mucosal bleeding may occur in any patient with dengue , and if the patient remains stable with fluid resuscitation/replacement , the mucosal bleeding should be considered a minor one [18] . Minor mucosal bleeding in dengue patients often results from diapedesis of erythrocytes around blood vessels with little inflammatory reaction [20] . If major bleeding occurs , it is usually from the GI tract [2] , [12] , [21] , [22] , and one of the risk factors for major GI bleeding is the existence of a peptic ulcer [21] , which is unfortunately not uncommonly develops in patients under stress [23] . One dengue series with 30 fatal dengue cases included disclosed that 80% of the fatal patients experienced GI bleeding , and severe bleeding with shock accounted for 30% of fatality [2] . The previously reported data [2] , [12] , [22] and ours suggest that even minor or moderate GI bleeding should be regarded as a warning sign of severe dengue and the patient in question needs close monitoring , as it potentially evolved into life-threatening intractable massive GI bleeding . Gastric ulcer and hemorrhagic gastritis each endoscopically found in one fatal patient with massive GI bleeding in our series raises the question of whether a H2-blocker or proton-pump inhibitor should be used in patients with severe DHF patients for prevention of massive GI bleeding . Further study is needed to answer this question . Of note , massive GI bleeding was found in 75% of patients who experienced early altered consciousness ( Table 1 ) ; 50% of patients with pre-fatal PT prolongation and 33 . 3% of patients with pre-fatal profound thrombocytopenia experienced massive GI bleeding . These data suggested that clinicians be alert to the potential development of severe GI bleeding when facing DHF patients with altered consciousness , and persistent PT prolongation and thrombocytopenia , and thereby initiate a timely management as necessary . Abdominal pain and persistent vomiting , the previously reported clinical warning signs of severe dengue [11] , [17] , [18] , did not differ between the fatal patients and controls in this series . In contrast , hypothermia significantly found in the fatal patients suggested that it should be considered a warning sign of severe dengue . Dengue-affected patients with hypothermia should therefore be intensively monitored , and aggressive workup is needed to clarify the potential cause ( s ) so that an effective treatment can be started timely . It is noteworthy that 50% of our patients presented with early altered consciousness suffered concurrent bacterial sepsis ( K . pneumoniae meningitis and E . faecalis bacteremia , respectively ) , highlighting the need for an immediate empirical antibiotic administration for dengue-affected patients with altered consciousness for the presumably superimposing bacterial sepsis until it is proven otherwise . The bacteria ( 2 K . pneumoniae and 1 E . faecalis isolates ) grew from culture of blood of 3 patients ( two of them each with the underlying hypertension and lung cancer ) sampled within 48 h after their admission were of normal intestinal flora . Our observation and previously reported concurrent bacteremia in patients with DHF caused by the members of Enterobacteriaceae [12] , [24] suggested that DHF patients are vulnerable to bloodstream invasion by microbes from the intestinal tract where they normally inhabit . These findings are consistent with the development of portal of entry for bacteria in bowels by disintegration of intestinal mucosal barriers in DHF patients reported previously [25] , [26] . Of the fatal bacteremic patients in this series , one patient with K . pneumoniae bacteremia and the other with simultaneous K . pneumoniae bacteremia and meningitis clearly experienced septic shock , while the shock in the patients who suffered massive GI bleeding and concurrently E . faecalis bacteremia might result from both hypovolemia and sepsis in view that E . faecalis has relatively low clinical virulence [27] . Nevertheless , our data suggest that when it comes to empirical use of antibiotic for suspicious concurrent bacteremia in dengue-affected patients , it is reasonable to cover bacteria from the intestinal tract . It is not surprising that acute renal failure ( 100% ) and acute hepatic failure ( 57 . 1% ) exclusively developed in fatal DHF patients in our series , as severe plasma leakage , massive bleeding and/or profound shock would lead to tissue hypo-perfusion , potentially rendering acute renal failure and hepatic failure [16] , [28] , [29] . There are some limitations in the present study . First , the fatalities in this series may be biased by patients' severity resulting from patient selection and referral pattern in a single medical center . Second , the lack of a standardized treatment protocol for severe dengue cases might bias patients' clinical outcomes in this retrospective analysis; this study thus addressed the pre-fatal clinical and laboratory evolutions in the deceased DHF patients , but not the appropriateness of treatment for them . Third , the small number of fatal cases made the statistical power quite small . In summary , our report highlights the causes of fatality other than DSS in patients with severe dengue , and suggested that in addition to those mentioned by the WHO 2009 scheme , hypothermia , leukocytosis , and bandemia may be warning signs of severe dengue . Early altered consciousness and GI bleeding/massive GI bleeding were significantly found among deceased DHF patients in this series . Dengue-affected patients should be closely monitored and appropriately treated once GI bleeding emerges , as it potentially evolved into massive GI bleeding; once massive GI bleeding develops , patients are at high risk for mortality , and this may be particularly true in patients with early altered consciousness , leukocytosis , profound thrombocytopenia and PT prolongation . Antibiotic ( s ) should be empirically added for patients at risk for developing concurrent bacteremia , especially in those with early altered consciousness and emergence of leukocytosis . Our data suggest that bandemia at hospital presentation may be a warning parameter for severe dengue , and monitoring the potential emergence of leukocytosis and persistence of thrombocytopenia may be helpful in evaluation of the progressive dengue severity . Further study is needed to confirm our observations . The findings of the suboptimal fluid resuscitations and blood/blood component transfusions in some of the fatal cases in this series underscores the importance of a timely effective volume replacement by fluid infusion and blood/blood component transfusion for patients with a severe dengue .	Fatality rate and causes of fatality in dengue-affected patients greatly varied from one reported series to another . A better understanding of the clinical and laboratory manifestations of fatal patients with dengue hemorrhagic fever ( DHF ) is important in alerting clinicians of severe dengue and improving management . In a retrospective analysis of 10 adults who died of and 299 survived ( controls ) DHF , dengue shock syndrome ( DSS ) alone was found in only 20% of dengue-related death , while intractable massive gastrointestinal ( GI ) bleeding was found in 40% , and DSS with concurrent subarachnoid hemorrhage , intractable massive GI bleeding with concurrent bacteremia , bacterial sepsis/meningitis , and sepsis due to ventilator associated pneumonia each were found in 10% . Early altered consciousness ( developed ≤24 h after hospitalization ) , GI bleeding/massive GI bleeding and concurrent bacteremia were significantly found among the deceased patients . Our data suggest that hypothermia , leukocytosis and bandemia at hospital presentation may be warning signs of severe dengue . Clinicians should be alert to the potential development of massive GI bleeding , particularly in patients with early altered consciousness , profound thrombocytopenia , prothrombin time prolongation and/or leukocytosis . Antibiotic ( s ) should be empirically used for patients at risk for bacteremia until it is proven otherwise , especially in those with early altered consciousness and leukocytosis .	[ "Abstract", "Introduction", "Materials", "and", "Methods", "Results", "Discussion" ]	[ "medicine" ]	2012	Fatal Dengue Hemorrhagic Fever in Adults: Emphasizing the Evolutionary Pre-fatal Clinical and Laboratory Manifestations
The production of antimicrobial reactive oxygen species by the nicotinamide dinucleotide phosphate ( NADPH ) oxidase complex is an important mechanism for control of invading pathogens . Herein , we show that the gastrointestinal pathogen Vibrio parahaemolyticus counteracts reactive oxygen species ( ROS ) production using the Type III Secretion System 2 ( T3SS2 ) effector VopL . In the absence of VopL , intracellular V . parahaemolyticus undergoes ROS-dependent filamentation , with concurrent limited growth . During infection , VopL assembles actin into non-functional filaments resulting in a dysfunctional actin cytoskeleton that can no longer mediate the assembly of the NADPH oxidase at the cell membrane , thereby limiting ROS production . This is the first example of how a T3SS2 effector contributes to the intracellular survival of V . parahaemolyticus , supporting the establishment of a protective intracellular replicative niche . Vibrio parahaemolyticus is a Gram-negative bacterium that inhabits warm marine and estuarine environments throughout the world [1] . This bacterium is recognized as the world’s leading cause of acute gastroenteritis associated with the consumption of contaminated raw or undercooked seafood [2] . In immunocompetent individuals , the illness is self-limiting with symptoms including diarrhea with abdominal cramping , nausea , vomiting , and low-grade fever [1] . However , for individuals with underlying health conditions , the bacterium can breach the gut barrier and cause septicemia corresponding to high mortality rates [3] . V . parahaemolyticus has also been reported to cause infection of seawater-exposed wounds , which in rare cases escalates to necrotizing fasciitis and septicemia [4] . The bacterium was also identified as the etiologic agent of acute hepatopancreatic necrosis disease ( AHPND ) , a shrimp illness that has recently emerged , causing a massive economic burden on the shrimp industry [5] . Among several virulence factors , including thermostable hemolysins ( TDH/TRH ) , polar and lateral flagella , and adhesins , V . parahaemolyticus encodes two Type III Secretion Systems ( T3SS1 and T3SS2 ) [6] . These are needle-like apparatuses used by the bacterium to inject proteins , termed effectors , into the host cell [7] . The first T3SS , T3SS1 , is present in all sequenced V . parahaemolyticus strains , including both environmental and clinical isolates , and is induced by culturing the bacteria in low Ca2+ , as in serum-free Dulbecco’s modified Eagle’s medium ( DMEM ) tissue culture growth medium [8] . While this system does not contribute to the bacterium’s enterotoxicity [9] , the T3SS1 effectors orchestrate a multifaceted and efficient death of the infected host cell [10] . V . parahaemolyticus more recently acquired the second T3SS , T3SS2 , through a lateral gene transfer event and this system is primarily associated with clinical isolates [6] . The T3SS2 becomes activated in the presence of bile salts [11 , 12] and is recognized as the principal virulence factor causing gastroenteritis [9] . We recently reported that during infection , T3SS2 promotes V . parahaemolyticus invasion of non-phagocytic cells [13 , 14] . We found that V . parahaemolyticus encodes VopC ( VPA1321 ) , a deamidase that constitutively activates the GTPases Rac and Cdc42 resulting in membrane ruffling and uptake of the bacterium into the cell [14 , 15] . Once inside the host cell , the bacterium is initially contained within an endosome-like vacuole [13] . Upon acidification of the vacuole , but prior to endosome fusion with the lysosome , V . parahaemolyticus breaks out of its vacuole and escapes into the cytosol [13] . V . parahaemolyticus then uses the cell as a protected replicative niche ( 100–300 bacteria/cell ) [13] . Although historically studied as an exclusive extracellular bacterium , these findings changed this long-standing view and established V . parahaemolyticus as a facultative intracellular bacterium . While the role of VopC to promote host cell invasion is well-defined [14] , the contribution of other T3SS2 effectors to the maintenance of the intracellular lifecycle of V . parahaemolyticus remains poorly understood . VopL ( VPA1370 ) , a T3SS2 effector , encodes three consecutive WASP-homology 2 ( WH2 ) domains intermixed with three proline-rich regions and a subsequent VopL C-terminal domain ( VCD ) ( Fig 1A ) [16–18] . WH2 domains are commonly found in nucleators of actin filaments; indeed , VopL’s in vitro nucleating activity is even more potent than that of the maximally-activated Arp2/3 complex [16] . Ectopic expression of VopL in epithelial cells causes a dramatic rearrangement of the actin cytoskeleton into filaments reminiscent of stress fibers [16] . Whether VopL nucleates actin from the barbed or pointed end remains a matter of disagreement [19 , 20] . Importantly , there is consensus that VopL promotes the nucleation of non-functional filaments in host cells , as opposed bona fide to filaments that can be recycled . As a result , VopL arrests actin monomers and the shortage of actin compromises the endogenous assembly of actin networks [19 , 20] . Despite the comprehensive characterization of VopL from a structural and a mechanistic standpoint over the last decade [16 , 21] , the contribution of this effector during a V . parahaemolyticus infection remained elusive . Herein , we show that VopL is required for the intracellular survival of V . parahaemolyticus . In the absence of VopL , intracellular bacteria filament , which indicates that the bacteria are under stress . We identified the stressor as an increase in exposure to reactive-oxygen species ( ROS ) . We found that the presence of VopL prevented filamentation by suppressing the generation of ROS by the nicotinamide adenine dinucleotide phosphate ( NADPH ) oxidase complex . To generate ROS , cytosolic and membranous subunits of the NADPH oxidase complex must come together at cell membranes [22] . VopL , by hijacking the actin cytoskeleton , impedes the translocation of NADPH cytosolic subunits to the cell membrane , thereby preventing the complete assembly of the enzymatic ROS complex . Thus , V . parahaemolyticus deploys the T3SS2 effector VopL to secure a safe replicative niche within the host cell . As discussed earlier , despite not contributing to V . parahaemolyticus’ enterotoxicity [9] , the T3SS1 becomes activated upon bacterial suspension in DMEM , causing the rapid death ( ~3h post-infection ) of tissue-cultured cells [10] , thereby masking the activity of the T3SS2 . To reveal the activity of T3SS2 we made use of the V . parahaemolyticus CAB2 strain , an isogenic strain derived from the clinical isolate RimD2210633 [14] . CAB2 contains a deletion for genes encoding two types of toxic factors . First , a deletion was made in exsA loci , resulting in the inactivation of the transcriptional activator for the T3SS1 [23] . Second , deletions were made for tdhAS , the two thermostable direct hemolysins ( TDH ) present in RimD2210633 , eliminating their cytotoxic activity [24] . The resulting strain , CAB2 , has been used in subsequent studies to assess the activity of T3SS2 and its effectors . Initially , we set out to investigate the contribution of VopL for the survival of CAB2 within Caco-2 cells , a colonic epithelial cell line . Bacterial survival can be assessed in two ways: first by determining the number of intracellular bacteria as a function of time post-invasion ( Fig 1B ) and second by visualization of intracellular CAB2 using confocal microscopy ( Fig 1C and S1 Fig ) . Shortly after the invasion of Caco-2 cells ( 2h post-invasion ) , CAB2 and its VopL-mutant counterpart ( CAB2ΔvopL ) exhibited comparable cell invasion and intracellular growth ( Fig 1B ) , displaying a patchy distribution inside their host cell ( S1A Fig ) . This activity was reminiscent of V . parahaemolyticus’ vacuolar localization early after invasion [13] . At 4h post-invasion , CAB2 and CAB2ΔvopL counts doubled ( Fig 1B ) , consistent with intracellular replication , which was indicated by a significant increase in bacterial load within Caco-2 cells ( S1B Fig ) . At 6h post-invasion , CAB2 exhibited an additional growth increase , while CAB2ΔvopL survival was substantially compromised as bacterial counts dropped more than twofold ( Fig 1B ) . Strikingly , confocal inspection of intracellular CAB2ΔvopL at 6h post-invasion revealed a dramatic change in bacterial morphology: CAB2 displayed characteristic V . parahaemolyticus’ rod-shape , while CAB2ΔvopL appeared significantly elongated ( Fig 1C ) . Expression of VopL in CAB2ΔvopL from a plasmid ( CAB2ΔvopL+pVopL ) increased intracellular growth to levels comparable to that of wild type bacteria ( Fig 1B ) and also rescued normal bacterial morphology ( Fig 1C and 1D ) . Elongation of CAB2ΔvopL was not a result of the cell body extension , but rather a deficiency in bacterial cell division . Closer inspection of CAB2ΔvopL revealed that the elongated cell body contained multiple nucleoids ( S1C Fig ) , which is consistent with continuous bacterial replication but ceased septation . This morphological phenotype is referred to as bacterial filamentation and represents an important strategy used by bacteria to survive during stressful situations [25] . Several bacteria have been reported to undergo filamentation as a protective mechanism against phagocytosis , as in the case of Uropathogenic Escherichia coli [26] , or against consumption of “inedible” filamentous bacteria by protists , as in the case of Flectobacillus spp . [27] . Filamentation can also be triggered in response to DNA-damaging stresses such as UV radiation , antibiotics , and reactive oxygen species ( ROS ) [25] . Given that CAB2ΔvopL deliberately invades Caco-2 cells ( via VopC ) , samples are not exposed to UV radiation , and intracellular bacteria are not exposed to gentamicin , as this antibiotic is not taken up by Caco-2 cells , we hypothesized that host generation of ROS could be responsible for the filamentous CAB2ΔvopL . To investigate whether ROS was causal for filamentation of intracellular CAB2ΔvopL , we assessed the generation of ROS inside Caco-2 cells by the nitroblue tetrazolium ( NBT ) assay [28 , 29] . In the NBT assay , the water-soluble tetrazolium dye is reduced by superoxide into blue insoluble formazan deposits , which are readily detectable by microscopic imaging [28 , 29] . Uninfected Caco-2 cells displayed little to no formazan precipitates ( S2 Fig ) . Importantly , the accumulation of formazan was substantially greater in cells infected with CAB2ΔvopL than in cells infected with wild type CAB2 ( S2 Fig ) , suggesting that VopL plays a role in suppressing the generation of ROS inside the host cell . The accumulation of formazan was observed in Caco-2 cells that contained intracellular bacteria as well as in cells that did not contain bacteria ( S2 Fig ) . These findings support that the presence of extracellular bacteria is sufficient to trigger the host ROS response ( possibly via pathogen-associated molecular patterns ( PAMPs ) such as lipopolysaccharides and flagella ) . While only a fraction of the host cells were invaded , all cells should be infected , and therefore , receive VopL via T3SS2 . Once delivered to infected cells , VopL turns off the ROS response . Next , we set out to investigate whether bacterial filamentation resulted from ROS production . While formazan deposits could be observed in the minority of Caco-2 cells invaded by rod-shaped , CAB2 bacteria ( Fig 2A and 2B ) , this precipitate was present in about 90% of the cells containing filamentous , CAB2ΔvopL bacteria ( Fig 2A and 2B ) . The tight correlation between filamentous bacteria and enriched deposits of formazan strongly implicates ROS as the stressor responsible for CAB2ΔvopL filamentation . ROS can be produced by NADPH oxidases , specialized enzymes whose sole function is the generation of ROS [22] . There are seven members of the NADPH oxidase ( NOX ) family , NOX1–5 and two dual oxidases ( DUOX1 and DUOX2 ) , which collectively produce ROS in a wide range of tissues where ROS participate in a variety of cell processes such as mitogenesis , apoptosis , hormone synthesis , and oxygen sensing [22 , 30] . NOX2 is a phagocyte-specific isoform , being highly expressed in neutrophils and macrophages where it plays an essential role in host defense against microbial pathogens [22] . NOX1 is the closest homolog of NOX2 , with whom it shares 56% sequence identity [22] . NOX1 is most abundant in the colon epithelium and is also expressed in a variety of cell lines , including Caco-2 cells [30 , 31] . At present , the physiological roles of colonic NOX1 are not fully understood . NOX1-derived ROS has been implicated in control of cell proliferation , mucosal repair after injury , and inflammatory response [30] . Importantly , evidence suggests a role for NOX1 as a host defense oxidase [32] . For instance , colon epithelial cells exhibited high NOX1-mediated ROS production in response to flagellin from Salmonella enteriditis [33] . In order to assess whether NOX1-dependent generation of ROS played a role for CAB2ΔvopL filamentation , we suppressed ROS generation using GKT136901 ( GKT ) , a direct and specific inhibitor of NOX1/4 ( NOX4 is primarily expressed in the kidney [22] ) [34] . GKT significantly attenuated the accumulation of formazan in Caco-2 cells infected with CAB2ΔvopL ( S3 Fig ) , confirming its suitability as an inhibitor of the ROS response . Importantly , this inhibitor reduced the number of host cells containing filamentous CAB2ΔvopL by more than twofold ( Fig 2C and 2D ) . As expected , GKT did not affected the intracellular growth of CAB2 , given that this strain exhibits minimal filamentation ( S4 Fig , Fig 1D ) . These findings strongly suggest that NOX1-generated ROS mediates bacterial filamentation . Our results thus far show that NOX1-generated ROS is causal for bacterial filamentation and that filamentation only occurs in the absence of VopL . Therefore , we hypothesized that VopL suppresses generation of ROS . To quantify NADPH oxidase-dependent production of ROS , we analyzed host cell release of superoxide , the product of NADPH oxidase-mediated reduction of molecular oxygen and the precursor of other ROS [35] . Detection of ROS generated by endogenous NOX1 in the colon , as well as in Caco-2 cells , is challenging to measure [33] . Indeed , under our experimental conditions we could not quantify superoxide in a sensitive manner during infection of Caco-2 cells , nor could we detect it upon cell stimulation with the PKC activator phorbol 12-myristate 13-acetate ( PMA ) ( S5 Fig ) . Thus , to further investigate the ability of VopL to control the generation of ROS by NADPH oxidases , we used a well characterized model cell system , the COSphox cell line , that has been used previously to biochemically analyze the production of ROS [36] . These cells stably express NOX2 ( gp91phox ) along with the other NOX2 complex subunits p22phox , p47phox , and p67phox [36] . Notably , the NOX1 enzymatic complex also includes p22phox , along with NOXO1 and NOXA1 , homologs of p47phox and p67phox , respectively [31 , 37] . Given the similarity in functioning of the NOX1 and NOX2 complexes , COSphox cells represent a suitable model of non-phagocytic cells with robust NOX-dependent production of ROS for the present study . Initially , we investigated CAB2 growth within COSphox cells , in the presence and absence of VopL . While CAB2 was able to efficiently replicate inside COSphox cells ( Fig 3A ) and displayed the bacterium’s normal rod-shape ( Fig 3B and S6 Fig ) , CAB2ΔvopL grew at approximately half the rate of wild type bacteria ( Fig 3A ) and , importantly , demonstrated a very dramatic filamentous phenotype ( Fig 3B and S6 Fig ) . Expression of VopL in CAB2ΔvopL from a plasmid ( CAB2ΔvopL+pVopL ) rescued intracellular growth to levels comparable to that of wild type bacteria ( Fig 3A ) and restored the rod-shaped bacterial morphology ( Fig 3B and 3C ) . These findings are in agreement with the observations made using Caco-2 cells and support a role for VopL in bacterial intracellular survival . As with Caco-2 cells , we investigated whether filamentation of intracellular CAB2ΔvopL resulted from ROS generation by COSphox cells . First , we compared superoxide-mediated accumulation of formazan deposits in COSphox cells containing rod-shaped ( CAB2 ) and filamentous ( CAB2ΔvopL ) bacteria . Formazan precipitates were greatly enriched in host cells containing filamentous bacteria ( Fig 4A ) , being present in about 75% of these cells , a fourfold increase in comparison to cells containing rod-shaped bacteria ( Fig 4B ) . Next , we assessed whether ROS generated in a NOX2-dependent manner was required for bacterial filamentation in COSphox cells . NOX2-dependent generation of ROS was inhibited by apocynin ( APO ) , which blocks ROS generation by preventing the complete assembly of the NOX2 enzymatic complex [22] . APO treatment of COSphox cells abrogated infection-elicited generation of superoxide ( S7 Fig ) and significantly attenuated CAB2ΔvopL filamentation ( Fig 4C ) , reducing the number of COSphox cells containing filamentous bacteria by 40% ( Fig 4D ) . Therefore , in agreement with our findings obtained with Caco-2 cell infection , ROS is an agent involved bacterial filamentation . Our results thus far show that VopL impairs the generation of ROS inside Caco-2 and COSphox cells ( given as a function of formazan accumulation in these cells ) . We next set out to quantify the production of ROS in the absence and presence of VopL . To quantify the production of ROS , we measured the extracellular release of superoxide by COSphox cells as a function of luminescence [35] . As a control for NOX2-dependent generation of superoxide , we stimulated COSphox cells with PMA . PMA activates the NOX2 complex via PKC-mediated phosphorylation of the p47phox subunit [38] . Cell stimulation with PMA led to a sustained generation of superoxide ( Fig 4E ) . Infection of COSphox cells with CAB2 induced the production of superoxide , which peaked at around 26 min post luminol addition and tapered off afterwards due to lack of continuous bacterial stimulus ( bacteria washed away prior to luminol addition ) ( Fig 4E ) . Importantly , bacteria-stimulated generation of ROS is substantially enhanced in the absence of VopL , as the peak in luminescence signal ( at 26 min ) during CAB2ΔvopL infection is 1 . 7 times higher than the luminescence peak resulting from CAB2 infection ( Fig 4E ) . Rescue of the CAB2ΔvopL strain with a VopL-expression plasmid lowered superoxide production to levels similar to that generated by the parental CAB2 strain ( Fig 4E ) . These results confirm our hypothesis that VopL suppresses NOX-generated ROS . As mentioned earlier , NOX2 is a multi-subunit complex; the latent complex is disassembled in resting cells and must become assembled at cell membranes with potential to generate ROS [22] . When at rest , the NOX2 complex regulatory subunits p67phox and p47phox are present in the cytosol as a heterotrimeric complex along with p40phox [22] . Rac , another complex subunit , is also present in the cytosol in its inactive , GDP-bound , form . Upon cell stimulation , all activated cytosolic components translocate to both cell plasma and phagocytic membranes where they interact with membranous subunits gp91phox ( NOX2 ) and p22phox and complete the assembly of a functional NOX complex [22] . Several pieces of evidence support a role for actin in NOX2 activity . For instance , addition of G-actin was shown to potentiate NOX activity in a cell-free system [39] . The p47phox and p67phox each contain a SH3 domain known to associate with the actin cytoskeleton [38] . In resting polymorphonuclear leukocytes ( PMN ) , p67phox is detected exclusively in the detergent-insoluble , cytoskeletal fraction [38] . Additionally , inhibition of actin polymerization by cytochalasin has been shown to modulate the translocation of NOX2 subunits from the cytosol to the plasma membrane upon stimulation of PMNs [40 , 41] . Despite the close homology between the NOX1 and NOX2 complexes ( the closest homologs within the NOX family ) , NOXO1 , the homolog of p47phox in the NOX1 complex , lacks the autoinhibitory region ( AIR ) domain present in p47phox [31 , 37] . As a result , NOXO1 , as well as its partner NOXA1 ( p67phox homolog ) , are constitutively associated with p22phox at the cell membrane [42] . Importantly , Rac1 is not constitutively localized to cell membranes . In fact , stimulated recruitment of Rac1 from the cytosol to cell membranes is crucial to the activation of NOXA1 , and thereby , the NOX1 complex [43] . Additionally , Rac1 promotes further recruitment of NOXA1 to cell membranes [43] . Given that VopL disrupts the normal assembly of the actin cytoskeleton , we hypothesized that this effector inhibited an actin-dependent step that is common to the activation of both NOX1 and NOX2 complexes . This step was hypothesized to be the recruitment of cytosolic subunits of NOX1 ( Rac1 ) and NOX2 ( p47phox , p67phox , Rac1 ) to cell membranes . Initially we investigated the NOX2 complex activation expressed in COSphox cells . To test our hypothesis , we transiently transfected COSphox cells with VopL and subsequently induced NOX2 activation using PMA . Cell stimulation with PMA caused p67phox to translocate from the cytosol to the plasma membrane and also caused an extensive rearrangement of the actin cytoskeleton with formation of membrane ruffles that co-localized with p67phox ( compare S8A and S8B Fig ) . As previously reported [16] , ectopic expression of wild type VopL ( WT VopL ) caused the formation of long actin strings reminiscent of stress fibers ( Fig 5A and S8C Fig ) . In cells transfected with WT VopL , PMA-stimulated actin ruffles were not formed ( Fig 5A ) and , importantly , the translocation of p67phox from the cytosol to the cell membrane was impaired ( Fig 5A ) . Quantification of enrichment of p67phox at the plasma membrane revealed a significantly smaller presence of this subunit at the membrane in the presence of VopL ( Fig 5B ) . To confirm that the inhibitory effect of VopL on p67phox translocation is dependent on the effector’s ability to manipulate the actin cytoskeleton , we also transfected cells with WH23-VopL , which contains point mutations at amino acids required for the actin binding activity of the WH2 domains [16] . We previously established that WH23-VopL is devoid of actin assembly activity in vitro and does not induce actin stress fiber formation in transfected cells [16] ( S8D Fig ) . Expression of WH23-VopL impaired neither PMA-stimulated membrane ruffling nor the cytosol-plasma membrane translocation of p67phox ( Fig 5C ) , which was highly enriched at the plasma membrane ( Fig 5D ) . During activation of the NOX2 complex , the stimulated recruitment of Rac to the membrane occurs independently from p47phox or p67phox [44] . Therefore , we also investigated whether VopL-mediated disruption of the actin cytoskeleton impaired translocation of Rac . To monitor Rac movement , COSphox cells were transiently transfected with EGFP-Rac1 with either WT or WH23-VopL . As was observed with p67phox , cells stimulated with PMA caused Rac1 to translocate from the cytosol to the plasma membrane ( compare S9A and S9B Fig ) . At the plasma membrane , Rac1 co-localized with actin ruffles ( S9B Fig ) . Importantly , expression of WT VopL completely inhibited PMA-stimulation recruitment of Rac1 to the plasma membrane ( S9C and S9E Fig ) , while in cells expressing WH23-VopL , the cytosol-plasma membrane translocation of Rac1 was unaffected ( S9D and S9E Fig ) . Next , we set out to determine if VopL could also inhibit stimulated recruitment of Rac1 to the plasma membrane in Caco-2 cells . Upon cell stimulation with PMA , Rac1 moved to the plasma membrane and co-localized with the actin ruffles ( compare Fig 6A and 6B ) . When cells were co-transfected with Rac1 and wild-type VopL , but not mutant WH23-VopL , Rac1 remained in the cytosol ( Fig 6C–6E ) . Therefore , VopL deploys a general mechanism to cripple the defenses of the host cell: it paralyzes the actin cytoskeleton , preventing assembly of both NOX1 and NOX2 complex , thereby inhibiting the generation of ROS during infection . VopL rearranges the actin cytoskeleton into linear strings of non-functional filaments that resemble stress fibers ( Fig 5A ) . By doing so , VopL retains p67phox and Rac1 in the cytosol , and consequently , impedes the activation of NOX2 . Therefore , we assessed whether manipulation of the actin cytoskeleton in the form of stable stress fiber-like structures could account for the inhibition of NOX2 assembly . Jasplakinolide is a potent inducer of actin polymerization and a stabilizer of actin filaments [45] . Treatment of COSphox cells with jasplakinolide induced stress fiber formation ( compare Fig 7A and 7B ) . As with VopL , jasplakinolide treatment decreased PMA-stimulated translocation of p67phox from the cytosol to the plasma membrane ( compare Fig 7C and 7D ) . However , in contrast to VopL transfected cells treated with PMA , actin ruffles are observed at the edges of cells treated with jasplakinolide and PMA ( Figs 5A and 7D , respectively ) . These findings further support our hypothesis that VopL manipulates the actin cytoskeleton to prevent NOX assembly and dampen ROS production . The T3SS2 effector VopL and its V . cholerae homologue VopF were discovered about a decade ago and their structure and biochemical activities are now well-characterized [16–19 , 46–48] . The coincidental discoveries that VopL/F produces non-functional filaments [19 , 20] and our findings that V . parahaemolyticus is a facultative intracellular pathogen [13] laid the groundwork for the elucidation of the biological role of VopL during infection . Here , we showed that NOX-derived generation of ROS plays an important role in controlling intracellular proliferation of V . parahaemolyticus . Specifically , ROS-dependent stress of the bacterium resulted in impairment of cell division and consequent filamentation of the bacteria . VopL is essential in preventing this deleterious event: by directly targeting the actin cytoskeleton and catalyzing the assembly of non-canonical actin filaments , VopL arrests the actin-dependent movement of cytosolic NOX subunits to cell membranes . As a result , VopL prevents the activation of the NOX complex and consequent production of ROS . Thus , this is the first report of how VopL aids V . parahaemolyticus infection: it secures a relatively “stress-free” environment within the host cell , enabling the bacterium to establish a successful replicative niche . Previous studies indicated that VopL does not play a significant role in bacterial colonization and fluid accumulation in the small intestine of rabbit models of V . parahaemolyticus infection [49 , 50] . Our present findings support that VopL could have an understated contribution to enterotoxicity , not obvious in the pathogenesis markers evaluated so far , prompting future investigations of a role for VopL in a diarrheogenic model . Distinct methods are used by other intracellular pathogens to inhibit ROS-mediated killing . Some pathogens scavenge ROS using extracellular polysaccharides , as in the case of Burkholderia cenocepacia and Pseudomonas aeruginosa [51 , 52] . Several other pathogens act on signaling machinery upstream of ROS and prevent activation of the NOX complex . [53–56] . Salmonella enterica Typhimurium excludes the NOX2 membranous subunits from the vacuole it inhabits in a T3SS-dependent manner [57] . Importantly , the virulence factors and mechanisms used by the vast majority of these pathogens remain unknown . The present work not only identified the virulence factor used by V . parahaemolyticus to suppress host ROS generation , but also revealed an unprecedented mechanism used by a microbial pathogen to do so ( Fig 8 ) . The pharmacological inhibition of NOX1 and NOX2 complexes , shown to fully suppress ROS generation in the case of apocynin , did not result in complete reduction of bacterial filamentation . These findings raise the possibility that additional factors may contribute to the arrest of bacterial division . Intracellular growth of certain Salmonella strains resulted in filamentous bacteria due to a defect in the bacterial histidine biosynthetic pathway [58] . Because V . parahaemolyticus only undergoes filamentation in the absence of VopL , under our experimental conditions , it is likely that host factors , rather than bacterial ones , elicit this morphological phenotype . Rosenberger and Finlay [59] reported an upregulation of MEK1 kinase during S . enterica Typhimurium infection of RAW 264 . 7 macrophages . MEK upregulation was causal for Salmonella filamentation , as occurrence of filamentous bacteria partially decreased in the presence of MEK inhibitors [59] . MEK and NOX activities operated in parallel to mediate Salmonella filamentation [59] . It is known that the actin cytoskeleton functions as a scaffold that mechanically modulates the activation of signaling pathways [60] . For instance , pharmacological inhibition of the actin polymerization reportedly inhibited ERK and AKT activation [61] . Interestingly , V . parahaemolyticus expresses VopA , a homolog of Yersinia spp . YopJ Ser/Thr acetyltransferase that has been shown to inhibit MAPK signaling pathways during infection [62–64] . Therefore , in addition to NOX assembly and MAPK signaling pathways , other pathways will be the subject of future investigation as further mediators of filamentation of the VopL-mutant . As discussed earlier , the T3SS2 is V . parahaemolyticus’ key virulence mechanism of enterotoxicity [9] . Importantly , T3SS2 is orthologous to the T3SS identified in several non-O1 , non-O129 clinical V . cholerae strains [65] . These strains lack cholera toxin and toxin-coregulated pilus but cause acute diarrheal diseases in a T3SS-dependent manner [46] . The non-O1 , non-O129 AM-19226 strain encodes VopL’s homolog VopF ( 32% sequence identity and 72% sequence similarity ) [46] . VopF induces actin-rich protrusion formation as opposed to actin stress fibers [46] . By disrupting the actin cytoskeleton , VopF causes depolarization of the epithelium and vopF-mutant strains present deficient epithelial colonization in vivo [47] . Therefore , despite the fact that VopL and VopF share close structural homology and use the same strategy to disrupt the actin cytoskeleton , these two effectors equip their bacteria with two distinct pathogenic mechanisms . Notably , the non-O1 , non-O139 strain 1587 encodes VopN , which shares similarity with both VopF and VopL [47] . VopN also nucleates actin filaments , and , like VopL , localizes to actin stress fibers [47] . While the physiological role of VopN remains elusive , we recently reported that the VopN-encoding strain 1587 invades epithelial cells in a T3SS-dependent manner [14] . Over many years of evolution , V . parahaemolyticus’ T3SS2 has maintained a repertoire of around a dozen effectors . One of these , VopC , has been shown to mediate cell invasion , and we now show a role for VopL in relieving free-radical stress for this intracellular pathogen . The activities of some of the other T3SS2 effectors have been studied and now the role that they play in this evolutionarily conserved invasive T3SS are ripe for future investigation . The V . parahaemolyticus CAB2 strain was derived from POR1 ( clinical isolate RIMD2210633 lacking TDH toxins ) , the latter being a generous gift from Drs . Tetsuya Iida and Takeshi Honda [66] . The CAB2 strain was made by deleting the gene encoding the transcriptional factor ExsA , which regulates the activation of the T3SS1 [14] . CAB2 was grown in Luria-Bertani ( LB ) medium , supplemented with NaCl to a final concentration of 3% ( w/v ) , at 30 °C . When necessary , the medium was supplemented with 50 μg/mL spectinomycin ( to select for growth of CAB2-GFP strains [9] ) or 250 μg/mL kanamycin . For in-frame deletion of vopL ( vpa1370 in RimD2210633 , GeneBank sequence accession number NC_004605 ) , the nucleotide sequences 1kb upstream and downstream of the gene were cloned into pDM4 , a Cmr Ori6RK suicide plasmid [14] . Primers used were 5’ GATCGTCGACATCAAATTGAATGCACTATGATC 3’ and 5’ GATCACTAGTAAAGAAGACCCCTTTATTGATTC 3’ for amplification of 1kb upstream region , and 5’ GATCACTAGTCTAGCGAGCACATAAAAAGC 3’ and 5’ GATCAGATCTTCCGGGGTGGTAAATGCTT3’ for 1kb downstream region . 1kb sequences were then inserted between SalI and SpeI sites ( upstream region ) or SpeI and BglII ( downstream region ) sites of the plasmid multiple cloning site . The resulting construct was inserted into CAB2 via conjugation by S17-1 ( λpir ) Escherichia coli . Transconjugants were selected for on minimal marine medium ( MMM ) agar containing 25 μg/mL chloramphenicol . Subsequent bacterial growth on MMM agar containing 15% ( w/v ) allowed for counter selection and curing of sacB-containing pDM4 . Deletion was confirmed by PCR and sequencing analysis . For reconstitution of CAB2ΔvopL , the sequence coding for vopL + FLAG tag was amplified using primers 5’ GATCCTGCAGATGCTTAAAATTAAACTGCCT 3’ and 5’ GATA GAATTC TTA CTTATCGTCGTCATCCTTGTAATC CGATAATTTTGCAGATAGTGC 3’ and then cloned into the pBAD/Myc-His vector ( Invitrogen , resistance changed from ampicillin to kanamycin ) between PstI and EcoRI sites . The 1kb nucleotide sequence upstream of vopC ( vpa1321 in RimD2210633 , accession numberNC_004605 . 1 ) was used as a promoter and cloned between XhoI and PstI sites using the primers 5’ GATC CTCGAG TATTCTTAATAAGTCAGGAGG 3’ and 5’GATC CTCGAG TATTCTTAATAAGTCAGGAGG3’ . The resulting construct was inserted into CAB2ΔvopL via triparental conjugation using E . coli DH5α ( pRK2073 ) . Transconjugants were selected for on MMM agar containing 250 μg/mL kanamycin . Reconstitution was confirmed by PCR . Empty pBAD plasmid ( without vopL gene insertion ) was introduced to CAB2 and CAB2ΔvopL strains for consistency in bacterial strain manipulation . Caco-2 cells ( ATCC , Manassas , VA ) were maintained in Minimal Essential Medium with Earl’s Balanced Salts ( MEM/EBSS , Hyclone , Logan , UT ) , supplemented with 20% ( v/v ) fetal bovine serum ( Sigma-Aldrich , St . Louis , MO ) , 1% ( v/v ) penicillin-streptomycin ( Thermo Fisher Scientific , Waltham , MA ) , and kept at 5% CO2 and 37°C . COSphox cells were grown in low-glucose Dulbecco’s Modified Eagle’s Medium ( DMEM , Thermo Fisher Scientific ) supplemented with 10% ( v/v ) fetal bovine serum , 1% ( v/v ) penicillin-streptomycin , 1% ( v/v ) sodium pyruvate ( Thermo Fisher Scientific ) , 0 . 8 mg/mL G418 ( Thermo Fisher Scientific ) , 200 μg/mL hygromycin ( Thermo Fisher Scientific ) , and 1 μg/mL puromycin ( Thermo Fisher Scientific ) , at 37°C with 5% CO2 [35] . Caco-2 and COSphox cells were seeded onto 24-well plates at a density of 2 . 5x105 ( Caco-2 ) or 1 . 5x105 cells/well ( COSphox ) and grown for 18–20 h . Overnight-grown bacterial cultures were normalized to an optical density at 600 nm ( OD600 ) of 0 . 3 and then grown in MLB supplemented with 0 . 05% ( w/v ) bile salts for 90 min at 37°C . Growth in the presence of bile salts allowed for induction of T3SS2 [11 , 12] . Mammalian host cells were subsequently infected with CAB2 strains at a multiplicity of infection ( MOI ) of 10 using culture medium devoid of antibiotics ( infection medium ) . To synchronize infection , cell plates were centrifuged at 200x g for 5 minutes . Infection was carried out for 2 h at 37°C , after which cells were washed with unsupplemented MEM/EBSS or DMEM and subsequently treated with infection medium containing 100 μg/mL gentamicin for 1–6 h . At the end of each time point , host cells were washed with 1x PBS for removal of extracellular dead bacteria and lysed with 0 . 5% ( v/v ) TX-100 . Cell lysates were serially diluted and plated on MMM agar for counting of colony forming units ( CFU ) as a measurement of intracellular bacterial survival/replication . To analyze bacterial filamentation , host cells were counted as containing filamentous bacteria when the cell predominantly contained bacteria longer than wild-type ( CAB2 ) size bacteria . Filamentous bacteria needed to be at least twice the size the size of wild-type CAB2 to be considered filamentous . Where indicated , samples were pre-treated with 10 μM GKT136901 ( GKT , Aobious , Gloucester , MA ) or added with 250 μM apocynin ( APO , Sigma-Aldrich ) at the remaining last 1 hour of infection or last hour of gentamicin incubation ( 6th hour ) . Caco-2/COSphox cells were left uninfected or infected as described above . Within 3 ( Caco-2 ) or 4 ( COSphox ) hours of gentamicin incubation , cell media was replaced with fresh media containing 1 mg/mL nitroblue tetrazolium ( NBT , Sigma-Aldrich ) and gentamicin for additional 3 ( Caco-2 ) or 1 ( COSphox ) hour . GKT or DMSO were added where indicated . Samples were then process for confocal analysis as described below . Caco-2 or COSphox cells were seeded onto 6-well plates at 5x105 cells/well and infected with CAB2 strains for 2 h as described above . Cells were trypsinized ( 0 . 25% trypsin/EDTA ) , centrifuged at 200x g for 5 min , and resuspended in Hank’s Balanced Salt Solution ( HBSS , ThermoFisher ) supplemented with Ca2+ and Mg2+ . Superoxide production was measured as a function of emission of luminescence using the Diogenes kit ( National Diagnostics Lab , Atlanta , GA ) and according to the manufacturer’s protocol [35] . Luminescence was monitored over 60 minutes using a FluoStar Optima plate reader . Caco-2 and COSphox cells were transiently transfected with 0 . 3 μg of either wild type ( WT ) VopL-Flag-psFFV or catalytically inactive VopL-WH2x3*-Flag-psFFV constructs [16] , 0 . 5 μg of pcDNA3-EGFP-Rac1 [67] ( plasmid # 12980 , Addgene , Cambridge , MA ) + 1 . 2 μg of empty psFFV using Fugene HD ( Promega ) for COSphox cells or Lipofectamine LTX with PLUS reagent ( ThermoFisher ) for Caco-2 cells for 20–24 h . Subsequently , COSphox/Caco-2 cells were treated with 0 . 4/1 . 0 μg/mL phorbol 12-myristate 13-acetate ( PMA , Sigma-Aldrich ) for 10/30 min at 37°C . For imaging , Caco-2 and COSphox cells were seeded respectively at 2 . 5x105 and 1-2x105cells/well onto 6-well plates containing UV-sterilized , poly-L-lysine-coated ( Sigma ) , glass coverslips . Following the infection and transfection protocols described above , samples were fixed with 3 . 2% ( v/v ) ρ-formaldehyde ( Thermo Fisher Scientific ) for 10 minutes at room temperature . Transfected COSphox cells were permeabilized with 0 . 5% ( w/v ) saponin ( Sigma ) for 10 minutes at room temperature and then blocked with 1% ( w/v ) bovine serum albumin ( BSA , Sigma-Aldrich ) in the presence of 0 . 1% saponin for 30 minutes at room temperature . In order to detect cells transfected with VopL , samples were subsequently incubated with anti-Flag antibody ( 1:100 dilution in 0 . 5% BSA , 0 . 1% saponin [Cell Signaling , #2368 , Danvers , MA] ) for 1 hour at room temperature , followed by incubation with anti-rabbit Alexa Fluor 488/A555 conjugated secondary antibody ( 1:500 dilution in 0 . 5% BSA , 0 . 1% saponin [Thermo Fisher Scientific , A-21441] ) for another 1h , room temperature . For detection of p67phox , samples were incubated with anti-p67phox at 1:50 dilution ( Santa Cruz , sc-7662 , Dallas , TX ) , followed by incubation with anti-goat Alexa Fluor 555 conjugated secondary antibody ( 1:500 dilution [Thermo Fisher Scientific , A-21432] ) . F-actin was stained with 2 units/mL of either rhodamine- or Alexa Fluor 680-phalloidin ( Thermo Fisher Scientific ) and DNA was stained with 1 μg/mL Hoechst A33342 ( Invitrogen , Carlsbad , CA ) . Coverslips were placed sample-side down on glass slides containing Prolong Gold anti-fade mounting media ( Thermo Fisher Scientific ) and imaged on Zeiss LSM710 and LSM800 confocal microscopes . Images were converted using ImageJ ( NIH ) . All data are given as mean ± standard deviation from at least 3 independent experiments unless stated otherwise . Each experiment was conducted in triplicate . Statistical analyses were performed by using unpaired , two-tailed Student’s t test with Welch’s correction . A p value of < 0 . 05 was considered significant .	The marine bacterium Vibrio parahaemolyticus is the world’s leading cause of food poisoning associated with the consumption of contaminated raw seafood . We recently discovered that during infection , V . parahaemolyticus invades cells from the host and uses a suite of effector proteins to convert the invaded cell into a niche for robust bacterial replication . In the present study , we describe how one of the effector proteins , VopL , contributes to this process by disrupting the actin cytoskeleton . Host cells produce reactive oxygen species ( ROS ) that cause damage to the pathogen’s DNA . This ROS production is dependent on a functional actin cytoskeleton . We observed that upon exposure to ROS , the mutant VopL-deficient V . parahaemolyticus underwent stress and as a result could not divide , exhibiting a filamentous morphology and concurrent replication impairment . This phenotype can be induced by exposure of the pathogen to ROS . In the presence of VopL , we observed an arrested assembly at the plasma membrane of nicotinamide dinucleotide phosphate ( NADPH ) oxidase complex , the enzymatic complex that catalyzes the generation of ROS . Paralysis of the actin cytoskeleton by VopL results in an inhibition of ROS production , thereby maintaining a relatively stress-free environment within the host cell for V . parahaemolyticus survival and replication .	[ "Abstract", "Introduction", "Results", "Discussion", "Materials", "and", "methods" ]	[ "medicine", "and", "health", "sciences", "pathology", "and", "laboratory", "medicine", "intracellular", "pathogens", "viral", "transmission", "and", "infection", "chemical", "compounds", "oxides", "pathogens", "biological", "cultures", "microbiology", "caco-2", "cells", ...	2017	T3SS effector VopL inhibits the host ROS response, promoting the intracellular survival of Vibrio parahaemolyticus
Schistosomes are trematode parasites of global importance , causing infections in millions of people , livestock , and wildlife . Most studies on schistosomiasis , involve human subjects; as such , there is a paucity of longitudinal studies investigating parasite dynamics in the absence of intervention . As a consequence , despite decades of research on schistosomiasis , our understanding of its ecology in natural host populations is centered around how environmental exposure and acquired immunity influence acquisition of parasites , while very little is known about the influence of host physiology , coinfection and clearance in the absence of drug treatment . We used a 4-year study in free-ranging African buffalo to investigate natural schistosome dynamics . We asked ( i ) what are the spatial and temporal patterns of schistosome infections; ( ii ) how do parasite burdens vary over time within individual hosts; and ( iii ) what host factors ( immunological , physiological , co-infection ) and environmental factors ( season , location ) explain patterns of schistosome acquisition and loss in buffalo ? Schistosome infections were common among buffalo . Microgeographic structure explained some variation in parasite burdens among hosts , indicating transmission hotspots . Overall , parasite burdens ratcheted up over time; however , gains in schistosome abundance in the dry season were partially offset by losses in the wet season , with some hosts demonstrating complete clearance of infection . Variation among buffalo in schistosome loss was associated with immunologic and nutritional factors , as well as co-infection by the gastrointestinal helminth Cooperia fuelleborni . Our results demonstrate that schistosome infections are surprisingly dynamic in a free-living mammalian host population , and point to a role for host factors in driving variation in parasite clearance , but not parasite acquisition which is driven by seasonal changes and spatial habitat utilization . Our study illustrates the power of longitudinal studies for discovering mechanisms underlying parasite dynamics in individual animals and populations . Schistosomes are a diverse , globally important group of trematode parasites that cause chronic inflammatory disease in humans , livestock and wildlife . They infect over 200 million people worldwide[1] , and inflict significant morbidity on already struggling human populations[2–4] . Several species of schistosomes infect livestock such as cattle , sheep and goats and cause significant economic loss and loss of valuable protein in regions where malnutrition is widespread[5–9] . Schistosomes are also parasites of wildlife including charismatic megafauna such as elephant , hippopotamus , rhinoceros , and African buffalo[10–13] . Although the effects of these parasites on wild animal populations are largely unknown , infection can cause severe pathology , and is a concern for conservation efforts[14 , 15] . Like most parasites , schistosome infection levels are highly heterogenic among individuals[16 , 17] . Based on decades of studies on human populations , variation in worm burdens is mainly attributed to the interplay of two factors: exposure and immunity[18] . Because schistosomes are transmitted to hosts through contact with aquatic habitats where their snail vectors thrive , exposure is determined by host behavior that brings the hosts into contact with snail-infested water sources and environmental factors that increase snail density ( e . g . [19] ) . Indeed , infection is typically clustered spatially into hotspots that are associated with water bodies , especially those that provide excellent habitat for snail vectors ( e . g . [1 , 17 , 20–22] ) . Decades of continued exposure to schistosomes slowly drives acquired immunity to reinfection; however , resistance is typically not complete[23–27] and is associated with increased eosinophils ( a white blood cell important in immune response to multicellular parasites ) [28] and specific IgE antibodies ( antibodies important in response to multicellular parasites ) [23–27] . Even with the same exposure levels , the development of immunity is heterogenic among individuals[29] , which could be driven by underlying genetic differences in immune genes[30] or other factors that affect immune function such as host condition or co-infection . Despite the rich literature linking environmental exposure and immunity with schistosome infection , other ecologic factors largely have been unexplored . Theoretical and empirical studies suggest that resource availability and the presence of co-infecting pathogens are likely to influence schistosome dynamics[31 , 32] . Resource availability ( i . e . food intake ) influences both the host immune response and pathogen fitness . Thus , increasing resources are predicted to have both costs and benefits for the pathogen[33–35] . For example , if host immunity is energetically costly , hosts in poor condition ( low resource acquisition ) would have a limited immune response and thus be more susceptible to infection while those in good condition ( high resource acquisition ) would have higher immune clearance[31 , 33 , 34] . Thus , in this scenario , low resource acquisition of the host would benefit the parasite . On the other hand , if parasite success is dependent on host resource acquisition , hosts in good condition will provide more resources for parasite growth and thus their parasites will be more successful[34 , 36] . This success could increase longevity of infection and also increase infective stages released into the environment; therefore , increasing within-host burdens over time . Thus , there is a dynamic interplay between resource availability , host immunity , and pathogen success . Indeed , this type of dynamic has been demonstrated in other host-pathogen systems , such as in gut nematodes of the Cuban tree-frog where resource restriction altered the host response to the pathogen , with the host employing a "tolerance" strategy when allowed to eat freely and using a "resistance" strategy when food-restricted [37] . Resource availability is therefore likely to have strong impact on schistosome dynamics because acquired immunity is a strong regulator of worm burdens and schistosomes are affected by host diet deficiency[38–42] . Multiple species of parasites within a host are also likely to influence the recruitment and disease course of new parasites , which in turn can impact disease dynamic patterns[43–45] . These parasites could compete directly within the host or through the immune system of the host , limiting or enhancing each other’s recruitment and maintenance[46–48] . Schistosomes are well-known to influence the disease course of other pathogens in experimental models , changing both host susceptibility and pathology due to other pathogens via effects on the host immune system[32 , 49 , 50] . However , less is known about how other pathogens affect schistosome burdens , even though schistosomes commonly occur with many other helminth species ( e . g . [51] ) . As part of their strategy to maintain a chronic infection , many helminths have multiple mechanisms to modulate the immunity of their hosts including general suppression through inhibition of Toll-like receptors ( innate immune sensing ) and the induction of regulatory T cells[52 , 53] . Furthermore , helminths may polarize the immune response in one direction , thereby preventing alternative responses ( e . g . Th1/Th2 polarization ) . This has been demonstrated in African buffalo where helminths alter the progression of bovine tuberculosis by skewing the host immune response to a Th2 dominated response , rather than a Th1 response that is more effective at suppressing tuberculosis infection[44] . Another gap in the natural history of schistosome infection is determining the factors that influence the natural loss of parasites[54] . This parameter is not typically considered in human populations because chemotherapy usually follows diagnosis . Additionally , the importance of worm loss is perhaps overlooked because of the estimated longevity of infections ( 3–9 years ) [55–57] . However , seasonal incidence of schistosomes in more temperate regions suggests a significant amount of worm loss on a shorter time scale[58 , 59] . Although largely over-looked , worm loss is an important parameter for host heath and disease dynamics . Worm burden directly influences disease pathology and the rate of worm loss is thought to enhance acquired immunity[29] . Thus , knowing the factors that drive the loss of established worms from a host has important implications . In this paper , we take a novel approach , focusing on dynamics of schistosome infection within individual hosts over time , by separating ACQUISITION and LOSS of schistosomes in a longitudinal study design . We dissect effects of immunology , co-infections , environmental factors ( spatial and seasonal ) , and host condition ( as a measure of resource acquisition ) on schistosome gain and loss in a free ranging African buffalo population . In southern Africa , African buffalo undergo extreme variation in resource availability due to seasonal variation in rainfall and temperature[60 , 61]; are known to be infected with numerous parasites and pathogens ( e . g . [62–66] ) ; and are long-lived , making longitudinal studies of parasite dynamics possible[67 , 68] . We ask: ( i ) What are the spatial and temporal patterns of schistosome infections in our study population; ( ii ) how do parasite burdens vary over time within individual hosts; and ( iii ) what host , co-infection , and environmental factors explain patterns of schistosome gain and loss in the buffalo ? We find that schistosome populations are seasonally dynamic and acquisition of worms is primarily driven by exposure , while loss of worms is complex and is driven by immunology , host physiology and coinfections . All animal procedures were approved by Oregon State University ( ACUP 4478 , ACUP 3267 ) and the University of Georgia ( A2010 10-190-Y3-A5 ) Institutional Animal Care and Use Committees ( IACUC ) , which follow the 8th Edition of the Guide for the Care and Use of Laboratory Animals ( Guide ) , NRC 2011; the Guide for the Care and Use of Agricultural Animals in Research and Teaching ( Ag Guide ) , FASS 2010; and the European Convention for the Protection of Vertebrate Animals Used for Experimental and Other Scientific Purposes , Council of Europe ( ETS 123 ) . Kruger National Park ( KNP ) is located in northeastern South Africa and comprises approximately 19 , 000 km2 , with an African buffalo ( Syncerus caffer ) population of approximately 30 , 000 animals[69] . We followed 200 buffalo in southern KNP between June 2008 and August 2012 to assess longitudinal variation in schistosome infection and determine the drivers of these changes–including the role of host nutrition , host immunity and coinfection with helminths ( Table 1 ) . Young adult female buffalo were captured from two herds and radio-collared in the southern portion of KNP , as part of a larger study on parasite interactions in free-ranging buffalo[44] . The first 100 buffalo were captured between June 23 and July 5 2008 ( Lower Sabie herd ) and the second 100 buffalo were captured between October 1 and October 8 2008 ( Crocodile Bridge herd ) . During the study , any animal that died or migrated out of the study area was replaced by a similarly aged individual ( n = 112 ) , resulting in a total of 312 individual buffalo that were followed for a period ranging from six months to four years ( median follow time was 3 years ) with 1750 data points . Animals were chemically immobilized with M99 ( etorphine hydrochloride ) and ketamine by darting from a helicopter or vehicle every 6 months for a maximum of 9 captures per individual . Following data collection , immobilization was reversed using M5050 ( diprenorphine ) and animals returned to their free-living existence . All immobilizations were performed by South African National Parks ( SANParks ) veterinarians and game capture staff . After immobilization , demographic data were collected including: season of capture , age , pregnancy and lactation status and a long-acting anthelmintic drug ( Panacur bolus -fenbendazole , Intervet UK ) was applied to 50% of the individuals . This bolus reduced nematode burdens[44] , but has no effect on schistosomes[70–72] . Season of capture was denoted as wet season ( November to April ) or dry season ( May to October ) . Pregnancy was assessed by rectal palpation[73] , which has a nearly 100% sensitivity rate after 51 days of gestation in Egyptian buffalo ( Bos bubalis;[74] ) , while lactation was evaluated via manual milking of all 4 teats[75] . Age was assessed from incisor emergence patterns for buffalo 2–5 years old animals , and from tooth wear of incisor one for buffalo 6 years and older[76] . Location was determined at the time of initial capture . Both herds , Crocodile Bridge and Lower Sabie , utilized areas in the southern section of Kruger National Park . The Crocodile Bridge herd was comprised of seven ( corner , mountain , malelane , mountain , powerline , randspruit , thicket ) sub-herds which maintained separate home ranges but were connected by frequent inter-subherd dispersal and migration , however for the herd analyses in this study appropriate sample sizes only existed for 4 subherds ( corner , mountain , powerline , randspruit ) [77] . Blood was taken from each animal via jugular venipuncture between 15–45 minutes after darting . Blood was collected into no-additive tubes for harvesting serum , into heparinized tubes for harvesting plasma , and into EDTA additive tubes for whole blood . All samples were placed on ice in a cooler within 5 minutes of collection for transportation back to the laboratory . Serum and plasma were collected from the top of the appropriate blood tubes after centrifugation for 10 minutes at 5 , 000 g to ensure the separation of cytological components . Whole blood was mixed and aliquoted for storage . Total time between sample collection and sample testing or storage in appropriate conditions was never greater than 8 hours , and typically ranged between 4–6 hours . Blood samples not processed immediately were stored in microcentrifuge tubes at -20°C until analysis . Schistosome infection was determined by a lateral flow assay to detect a specific schistosome antigen , circulating anodic antigen ( CAA ) in serum[78] . This antigen is produced by the schistosome and released into the blood[79] . The CAA concentration ( pg/ml ) was used as a proxy for adult worm burden in the buffalo , and has been shown to be highly sensitive and specific in human studies[80–84] , as well as in cattle [85 , 86] . We validated it for use in buffalo by comparing counts of adult schistosomes at cull to CAA level ( S1 Text ) . The species of schistosome was confirmed to be Schistosoma mattheei by sequencing a region of the large subunit ribosomal DNA and part of the mitochondrial DNA ( S2 Text ) . Feces was collected rectally from each animal , placed on ice and returned to the laboratory within 8 hours of collection . Gastrointestinal nematode infection was assessed using fecal egg counts . Fecal samples were processed on the day of collection using a modification of the McMaster method[87] . To identify the species of nematode worms present , feces were cultured to the infective larval stage and larvae were isolated using a modified Baerman apparatus , and then identified larve using PCR and sequencing[63] . The dominant helminths in this population of buffalo are gastrointestinal nematodes: Cooperia fuelleborni , Haemonchus bedfordi and Haemonchus placei[63] . The remaining feces was aliquoted and stored at -20°C or dried in a drying oven at 30°C for 24–48 hours . For a subset of the animals , captured at Lower Sabie in July and August 2009 , part of the fresh sample ( 3 g ) was processed via sedimentation to look for schistosome eggs[88]; however , due to the small amount of fecal material obtained via rectal collection , this was found to be an unreliable indicator of infection so was not continued . We expected host nutrition and immune response to play an important role in mediating schistosome dynamics within the host , so employed several parameters to assess each . Body condition was measured by visually inspecting and palpating four areas on the animal where fat is stored in buffalo: ribs , spine , hips and base of tail . Each area was scored from 1 ( very poor ) to 5 ( excellent ) and a body condition score calculated as the average of all four areas . This index is correlated with the kidney fat index[89] , and similar body condition indices have been used in other studies of African buffalo[90 , 91] . Fecal nitrogen , a measure of dietary protein[61 , 92] was assessed by the Agricultural Research Council ( Nelspruit , South Africa ) from dried feces[93] . Hematocrit , the volume percentage of red blood cells ( the primary resource for adult schistosomes ) in the blood , which is also correlated to body condition , was assessed by a hematology analyzer for whole blood on the day of capture[94] . Total plasma protein , a measure of animal nutritional status , was assessed using an Abaxis chemistry panel[95] . We measured cytokine levels and cellular immune responses that are associated with Th1 and Th2 immunity from blood . Adaptive immunity against schistosomes in humans is associated with a Th2 response with increased levels of IgE antibodies and eosinophils . Therefore , we expected that resistance to schistosomes would be associated with high levels of IL4 , total globulins , and eosinophils . We also expected markers associated with Th1 proinflammatory immunity ( IFNy ) to be low in these animals due to polarization toward Th2 responses . Finally , we evaluated nonspecific markers of innate immune function ( plasma BKA ) and inflammation ( haptoglobin ) . Schistosome infections in African buffalo were common , with an overall prevalence of 50% , and 79% of individuals exhibiting at least one positive test result ( CAA>20 ) over the course of the study . Schistosomes were highly aggregated among hosts ( Fig 1A ) , with most buffalo having very low concentrations at most time points ( 71% of the time animals had CAA concentrations below 100 ) , and a small number of hosts harboring the majority of worms ( Fig 1B ) . Schistosome burdens , and variability in burdens , increased with host age ( Fig 1C and 1D ) . The intensity of schistosome infections varied seasonally , and among years ( Fig 2 , Table 2 ) . Buffalo had higher schistosome burdens in the dry season than in the wet season . Overall , concentrations increased over the course of the study , with gains in schistosome numbers in the dry season only partially offset by losses during the following wet season . This increase likely reflects schistosome burdens in our study animals ratcheting up as they aged throughout the study: Replacing year with age in our analysis of spatio-temporal variation in schistosome burdens ( Table 3 ) improves model fit ( ANOVA p<0 . 05 ) . There was pronounced spatial variation in schistosome abundances in our study population . Animals had higher intensities of infection at Crocodile Bridge than at Lower Sabie ( Fig 3A , Table 2 ) , which was due to two subgroups of the Crocodile Bridge herd in the southeastern corner of the park with high burdens ( Fig 3B & Table 4 ) . Schistosome burdens varied dramatically over time within individuals ( Fig 4 ) . Animals both gained and lost schistosomes . The median decrease in CAA level was 12 ( 95% CI 9–15 ) , while the median increase in CAA level was 15 . 7 ( 95% 13 . 1–18 . 5 ) however the range was large ( Fig 4B ) . An increase in CAA concentration of 20 represents approximately 10 schistosome worms ( S1 Text ) with 11 animals having a decrease in of more than 50 worms in one 6-month period ( Fig 4B ) . Animals were more likely to gain schistosomes in the dry season , and lose them in the wet season ( model 4 , binomial dependent variable = gain or loss of schistosomes; B = 0 . 57 , SEM = 0 . 17 , p = 0 . 009 ) ; and poor body condition was weakly associated with a higher likelihood of losing , rather than gaining , schistosomes ( B = 0 . 31 , SEM = 0 . 16 , p = 0 . 06 ) . No other terms were included in the final model . We then considered what factors affected the magnitude of schistosome gain and loss ( change in concentration ) in individual hosts . The magnitude of schistosome gain was predicted only by season and the animal’s previous schistosome burden ( model 5 , Table 5 ) . Buffalo gained more schistosomes in the dry season , and animals with prior high burdens tended to gain more additional worms . In contrast , greater losses of schistosomes were observed in the wet season , and occurred in buffalo in poor body condition , with lower levels of IL4 , and those without gastrointestinal nematodes ( model 6 , Table 5 ) . The effect of nematodes was driven by one species , C . fuelleborni ( model 7 , Table 6 ) , with animals more likely to retain schistosomes if infected with this species . Additionally , the effect of IL4 on schistosome loss was strongest in animals that did not have C . fuelleborni while body condition had a much stronger effect on schistosome loss in those with this species of nematode ( Fig 5 ) . Pregnancy status , treatment with an anthelminthic bolus , fecal nitrogen , hematocrit , total white blood cell count , eosinophil count , total globulins , haptoglobin , plasma bactericidal ability , total plasma protein and age did not predict the size of an increase or decrease in schistosome number . Schistosome burdens varied seasonally and spatially in our study population of wild African buffalo . Conditions for schistosome exposure change radically between the wet and dry seasons . Buffalo , snail vectors , and larval schistosomes are water dependent and become concentrated at permanent water sources , such as slow moving rivers as the veld dries up[60 , 105] . During the wet season , buffalo spend less time watering in these permanent slow moving rivers , and can rely on other temporary water sources such as local pans where snail vectors are presumably less abundant[106] . Also , as water velocity in rivers increases , infection risk should decrease as snails and schistosome larvae are washed out . Indeed , infection dynamics of S . mattheei in snails have been found to be temporally variable and highly dependent on local conditions of both water flow and temperature which influence snail abundance and parasite development[58 , 107 , 108] . Seasonal change in schistosome prevalence in buffalo has been previously reported[109] , with prevalence highest in the late dry season ( 58% ) compared to the subsequent wet season ( 45% ) , a pattern consistent with our findings . In addition to seasonal patterns of infection , we found variation in schistosome burdens among the herds studied . Buffalo live in fission-fusion societies with defined geographic ranges and frequent exchange of individuals between herds[69] . The populations we describe have high inter-group dispersal rates[77] , and population genetic analyses suggest that there is little genetic sub-structuring among herds[110 , 111] . As such , it is unlikely that herd-level variation in schistosome burdens is based on genetic differences among herds . Instead , the observed spatial variation in worm burdens may be due to environmental heterogeneity that mediates exposure risk . We found the highest burdens in the herds that water along the eastern portion of the Crocodile River ( mountain and corner herds ) , compared to those that water upriver , along the western portion of the river ( powerline ) or along the Sabie River/Mlondozi dam ( thicket , Lower Sabie ) [77] . The eastern part of the river is characterized by numerous manmade weirs that slow river flow and alter river profile ( Govender , personal comm . ) creating superior habitat for the snail intermediate host[106] . Concordant with these findings , Pitchford et al . [109] found that the highest egg output from buffalo occurred in streams or rivers with man-made dams and dense riparian vegetation . Thus , altered water flow due to dams and weirs may have helped establish schistosome transmission foci due to increased populations of vector hosts and retention of schistosome infective stages ( e . g . [112] ) . In addition , there may be geographic variation in susceptibility to schistosomes , due to nutritional variability or localized differences in co-infecting parasite exposures . For instance , there is geographical variation in brucellosis[66] as well as some strongyle infections[93]; and the nutritional intake varies by locale with generally poorer body condition in the Crocodile Bridge herds[66] . Schistosome infections were highly dynamic in African buffalo . This longitudinal study showed that worm burdens in individual hosts fluctuated dramatically , but tended to increase over time , suggesting a lack of acquired immunity that is seen in other schistosome-host systems[29 , 54] . Schistosome burdens primarily increased during the dry season and decreased during the wet season , although typically not to baseline levels . The magnitude of schistosome clearance , with 11 animals losing more than approximately 50 worms , on this 6-month time scale , was not expected . In humans , schistosomes cause long-term , chronic infections lasting an average of 3–10 years[55 , 57 , 113 , 114] and even decades , due to molecular mimicry to prevent immune detection[29] and stem cells that repair damaged tissues—a trait schistosomes share with their planarian relatives that have incredible regeneration abilities[115] . Loss of adult worms or “self-cure” has been reported from Asian water buffalo , but is not well-known in humans[54 , 116 , 117] . The schistosome loss in African buffalo suggests a role for immune-mediated clearance of established schistosomes . It may be that in humans , loss of schistosomes also occurs on this short time scale , but is masked by continued recruitment of new parasites . Indeed , seasonal variation of schistosome burdens in humans has been reported from more temperate regions of their range where exposure is seasonal , suggesting significant worm loss in a 6-month period ( e . g . [58 , 59 , 118] ) . However , the factors that cause natural worm loss in humans are unexplored even though death of worms within a host is thought to drive the development of acquired immunity because antigens that are normally hidden in a living worm become exposed upon death[23 , 119] . Our results are among the first to determine the ecological and host related factors that are associated with schistosome loss in a longitudinal study design . Schistosome recruitment was primarily affected by variation in exposure but not variation in immunity . Of the host and environmental variables examined in this study , only season affected recruitment . Seasonal changes in vector and parasite infective stage abundance , and change in access to different water sources for the buffalo , likely underlie this pattern . Indeed , we have had difficulty finding snail vectors in the park during extreme dry conditions ( Beechler pers obs . ) . Schistosome recruitment was also predicted by previous burden . Buffalo that previously had high schistosome burdens were likely to acquire more in the subsequent dry season , suggesting that certain individuals are more likely to gain worms than others . However , no host immune factors were associated with schistosome acquisition . Several species of hosts are known to develop partial , antibody-mediated immunity after repeated exposures to schistosomes , ( e . g . humans , mice and cattle ) although its strength varies among host species[23 , 24 , 120] . In humans , this response is IgE-mediated , is correlated with high levels of Th2 cytokines , like IL4 , and removes migrating larvae via an antibody-dependent eosinophil response[29 , 121–123] . There was no evidence in our data that acquired immunity was providing protection against recruitment , as ( i ) IL4 and total globulins were not a significant driver of establishment and ( ii ) there was no age dependent effect such that older buffalo have lower prevalence or intensity of infection . On the contrary , worm burdens increased with host age . Taken together , these findings suggest that schistosome recruitment proceeds largely unhampered by acquired immunity . Nonetheless , it is possible that immune killing is localized in tissues such as the skin or lungs and thus systemic levels of immune factors are not reflective of local response . It is also possible that we “missed” acquired immunity in our study if it occurs in animals that are younger or older than those included in this analysis where the median age of buffalo at initial capture was 3 years . By contrast , schistosome clearance was affected by immunological and physiological factors , and co-infection by gastrointestinal nematodes , as well as season . The strongest drivers of worm clearance were circulating levels of IL4 and host body condition . Buffalo with low levels of IL4 , and buffalo in poor condition during the prior season , had fewer worms the following season . Although we predicted that a Th2 response would be associated with immune mediated clearing of worms , we found the opposite . Buffalo that had low IL4 levels during the previous collection period lost the most schistosomes by the next collection period . The pattern cannot be explained by schistosome immune polarization because , there was no association between simultaneous schistosome burden and IL4 . If this were the case high burdens would be associated with IL4 and schistosome loss would be associated with a drop in IL4 . Furthermore , The IL4 pattern does not appear to be due to a competing Th1 response as IFNy was eliminated during statistical model selection , and therefore not significant . It is possible that schistosome loss in buffalo is driven by Th2 immunity and due to the timing of our collection periods , we missed a peak in IL4 prior to schistosome loss , or that clearance occurs via an IL4-independent mechanism . Nevertheless , there remains a strong association between lower levels of IL4 during infection and worm loss . A seemingly parallel observation is that IL4 decreases to very low levels in water buffalo experimentally infected with its native schistosome , Schistosoma japonicum[124] . Although water buffalo are an important host for S . japonicum , they are not very permissive . Infections are short-lived , schistosome growth is reduced , and adaptive immunity develops over time[125 , 126] . However , in a more permissive host , yellow cattle , IL4 shows an increase after infection . Thus , comparing yellow cattle and water buffalo demonstrated that downregulation of IL4 is associated with worm expulsion and suggests that the immune response may help sustain infection[126] . Previous studies have indicated that an intact immune response , specifically CD4+ T cells are necessary for schistosome development and egg production in other host-schistosome models[127 , 128] . Therefore , we hypothesize that in buffalo , schistosomes require host signals that are associated with an Th2 response and are more likely to be lost when these host signals are not present . Co-infecting gastrointestinal nematodes also influenced schistosome burden in that there was a significant association between schistosome loss and presence of gastrointestinal nematodes , which was primarily due to the presence or absence of one species of nematode , C . fuelleborni , and also an interaction between these nematodes and IL4 . There was a negative association between schistosome loss and presence of C . fuelleborni such that buffalo with nematodes maintained more schistosomes and buffalo that did not have this nematode species lost more schistosomes . Several mechanisms may underlie this pattern . First , nematodes immunomodulate their hosts[129] , including Cooperia sp . [54] and a previous study in cattle has shown that Cooperia sp . and a related gastrointestinal nematode , Ostertagia ostertagi , together drive an increase in host susceptibility to lungworms[130] . Thus , nematode-driven changes in buffalo immune response could be preventing immune mediated schistosome loss . Interestingly , our data indicate that the lack of IL4 is an important driver of schistosome loss in the absence of C . fuelleborni , however in the presence of this nematode species , IL4 is less important , suggesting that C . fuelleborni alters the immune-schistosome interaction . However , intensity of C . fuelleborni was not related to IL4 concentration in this population[99] . Second , buffalo that are naturally resistant to nematodes may also be good at reducing their schistosome burdens . Previous work in another African buffalo population has shown that there may be a genetic basis to variation in gastrointestinal nematode burdens , where some animals are naturally more effective at maintaining low nematode burdens than others[131] . This work also suggests that this phenotype is associated with variation in the immune response , specifically levels of IFNy secretion[131] . If similar genetically-based resistance patterns hold in KNP buffalo , then we hypothesize that the immune response that allows individuals to be resistant to nematodes also aids in clearance of schistosomes . Third , the Cooperia-schistosome interaction could be partially mediated by host condition . Cooperia is associated with increases in body condition[99] , and , like the presence of C . fuelleborni , high body condition is associated with maintenance of schistosomes . However , buffalo condition was included ( and significant ) in the nematode-specific statistical analysis , suggesting that there is an additional effect of C . fuelleborni on schistosome infection . Schistosomes were strongly influenced by the resource acquisition of their hosts . Resources directly affected parasite success and did not appear to increase immune clearance of schistosomes . Body condition of the buffalo was negatively associated with worm loss so that buffalo with low body condition scores ( low resource acquisition ) lost more worms , and high condition scores retained more worms . These data support the hypothesis that the schistosomes are highly dependent on host resources for success . Schistosomes feed on the blood of the host , ingesting red blood cells and directly absorbing nutrients like glucose across the tegument[42] . Although this may seem like a relatively unlimited resource , experimental studies have shown that malnutrition of the host retards egg development of schistosomes[38–41]; thus , they are nutritionally dependent on the host . During the dry season , high quality food is scarce and many buffalo become nutritionally deficient[61 , 66 , 91] . Thus , it is reasonable to hypothesize that missing components in the diet of buffalo during this period can lead to the starvation of worms , which are lost between the dry season and wet season when nutritious food is most scarce . We did not see evidence of resource acquisition enhancing host immunity , which is consistent with the observation that downregulated immune responses ( i . e . decrease in IL4 ) was associated with worm loss . If our hypothesis that functioning Th2 immunity is necessary for schistosome maintenance is correct , then this adds a whole new angle to the theory regarding the effects of resource acquisition on pathogen dynamics , which as of yet has not considered the scenario where increased immune responses lead to parasite success ( i . e . high resource acquisition positively affects the pathogen both directly and via the host immune system ) . It is possible that CAA levels are linked to changes in worm feeding as well as worm loss since it is an antigen produced by feeding worms; however , the magnitude seen here is typically associated with worm loss[132 , 133] , not just reduced activity due to inadequate resources . Our study demonstrates that schistosome infections are surprisingly dynamic in a free-living mammalian host population . Our longitudinal data on infection profiles of individual hosts over time uncover prominent heterogeneity in parasite dynamics with evidence for “self-cure” of schistosomiasis , the magnitude of which depends on host immunity , body condition and co-infection by gastrointestinal nematodes . On the other hand , we found no evidence for a role of host immunity mediating schistosome recruitment . Schistosome dynamics in free-living buffalo thus appear to be driven by successive waves of worm recruitment during the dry season , partly balanced by self-cure within a 6-month time frame . The consequences of these patterns for host health and disease dynamics are yet to be explored; but schistosome burdens in buffalo ratchet up with host age , and older individuals are thus likely to play a disproportionate role in schistosome transmission , as well as bearing the burden of schistosome-associated disease . Our results are in sharp contrast to previous work on schistosomiasis , where acquired immunity to schistosome establishment results in lower burdens in older animals , adult schistosomes are rarely killed by the host’s immune system , and effects of heterogeneity in host body condition and co-infections on parasite population dynamics are rarely considered . As such , studies in natural host populations may be indispensable for understanding which hosts are pivotal to parasite transmission and disease prevention , and focusing interventions accordingly .	Schistosomes are a parasite of global importance , affecting over 200 million people worldwide , while also infecting livestock and wildlife . Despite decades of research on schistosomiasis in humans , little is known about what drives patterns of infection in untreated naturally occurring populations . We took advantage of a study in African buffalo to understand how geography , nutrition , immunity and coinfection drive schistosome acquisition and loss . The most striking outcome of our study was that schistosome burden varied seasonally within an individual , with some hosts able to completely clear infection , and others unable to do so . The ability of a buffalo to clear infection was affected by immune response and co-infection with other gastrointestinal parasites while host immunity and coinfection were not important in determining whether a buffalo became infected . These outcomes should be taken into consideration when designing control programs for human schistosomiasis .	[ "Abstract", "Introduction", "Materials", "and", "methods", "Results", "Discussion" ]	[ "schistosoma", "invertebrates", "medicine", "and", "health", "sciences", "helminths", "immunology", "vertebrates", "parasitic", "diseases", "animals", "nematode", "infections", "reptiles", "infectious", "diseases", "epidemiology", "crocodiles", "immune", "response", "eukary...	2017	Host immunity, nutrition and coinfection alter longitudinal infection patterns of schistosomes in a free ranging African buffalo population
RNA viruses exploit host cells by co-opting host factors and lipids and escaping host antiviral responses . Previous genome-wide screens with Tomato bushy stunt virus ( TBSV ) in the model host yeast have identified 18 cellular genes that are part of the actin network . In this paper , we show that the p33 viral replication factor interacts with the cellular cofilin ( Cof1p ) , which is an actin depolymerization factor . Using temperature-sensitive ( ts ) Cof1p or actin ( Act1p ) mutants at a semi-permissive temperature , we find an increased level of TBSV RNA accumulation in yeast cells and elevated in vitro activity of the tombusvirus replicase . We show that the large p33 containing replication organelle-like structures are located in the close vicinity of actin patches in yeast cells or around actin cable hubs in infected plant cells . Therefore , the actin filaments could be involved in VRC assembly and the formation of large viral replication compartments containing many individual VRCs . Moreover , we show that the actin network affects the recruitment of viral and cellular components , including oxysterol binding proteins and VAP proteins to form membrane contact sites for efficient transfer of sterols to the sites of replication . Altogether , the emerging picture is that TBSV , via direct interaction between the p33 replication protein and Cof1p , controls cofilin activities to obstruct the dynamic actin network that leads to efficient subversion of cellular factors for pro-viral functions . In summary , the discovery that TBSV interacts with cellular cofilin and blocks the severing of existing filaments and the formation of new actin filaments in infected cells opens a new window to unravel the way by which viruses could subvert/co-opt cellular proteins and lipids . By regulating the functions of cofilin and the actin network , which are central nodes in cellular pathways , viruses could gain supremacy in subversion of cellular factors for pro-viral functions . Plus-stranded ( + ) RNA viruses , which are important pathogens of plants , animals and humans , co-opt a number of host-coded proteins and lipids to facilitate the replication process [1–6] . These viruses also remodel host membranes and alter host cellular pathways to take advantage of host resources and to avoid recognition by host antiviral defenses . Characterization of an increasing number of host factors involved in ( + ) RNA virus replication has already revealed intriguing and complex interactions between various viruses and their hosts . Functional studies with selected host proteins have revealed a plethora of activities preformed by these host proteins during RNA virus infections [1 , 3 , 7–11] . In spite of the intensive efforts , our current cataloging of host factors is still far from complete and our current knowledge on the role of the identified host factors is incomplete . One of the advanced viral systems to study virus-host interactions is Tomato bushy stunt virus ( TBSV ) , a small ( + ) RNA virus , which can replicate in the model host yeast ( Saccharomyces cerevisiae ) [12–16] . TBSV replication requires two viral-coded proteins , namely the p33 and p92pol replication proteins . Albeit these proteins have overlapping sequences , they have different functions . p33 , which has RNA chaperone activity , has been shown to recruit the TBSV ( + ) RNA to the cytosolic surface of peroxisomal membranes , the sites of replication [17–20] . The p92pol has RNA-dependent RNA polymerase ( RdRp ) activity and binds to p33 to assemble the membrane-bound functional viral replicase complex ( VRC ) [14 , 19 , 21–24] . The activities of TBSV replication proteins , however , are affected by numerous host proteins [3 , 7–9 , 25] . Indeed , over 500 host genes/proteins that affected TBSV replication and/or recombination , have already been identified by using multiple genome-wide screens of yeast and global proteomics approaches [16 , 26–31] . Moreover , the tombusvirus VRC contains several host proteins [32–34] , including heat shock protein 70 ( Hsp70 ) , glyceraldehyde-3-phosphate dehydrogenase ( GAPDH ) , eukaryotic elongation factor 1A ( eEF1A ) , eEF1Bγ , the DDX3-like Ded1 , eIF4AIII-like RH2 and DDX5-like RH5 DEAD-box RNA helicases , and the ESCRT ( endosomal sorting complexes required for transport ) family of host proteins [25 , 33 , 35–39] . These proteins are required for VRC assembly or affect viral RNA synthesis [3 , 35 , 38 , 40–42] . The TBSV replication process also depends on phospholipids and sterols , which are actively recruited to the site of viral replication [43–47] . Previous genomic and proteomic screens have revealed that Cof1p ( cofilin in mammals ) interacts with the tombusvirus p33 replication protein , and a mutation in COF1 enhances TBSV RNA replication in yeast cells [30 , 48] , suggesting that COF1 could be an important host restriction factor . Cof1p is a major modulator of actin filament disassembly and an essential protein for yeast growth [49 , 50] . The major cellular function of Cof1p is to preferentially bind to ADP-actin subunits in actin filaments that results in twisting and severing the actin filaments [51 , 52] . Actin filament disassembly via Cof1p-induced depolymerization is required for remodeling of the actin cytoskeleton by providing free actin monomers as substrates for new filament formation [53–55] . Whether cofilin facilitates actin filament assembly or disassembly depends on the concentration of cofilin relative to actin . Overall , cofilins are conserved in eukaryotic cells and are essential from yeast to humans . Cofilins are known to affect many cellular pathways , including cell motility , cytokinesis , endocytosis , receptor functions , apoptosis , phospholipid metabolism , oxidative stress and gene expression via acting as a chaperone for nuclear actin [53 , 54 , 56 , 57] . Cofilins are also involved in several diseases , such as Alzheimer’s disease and ischemic kidney disease and other pathophysiological defects , such as infertility , immune deficiencies , inflammation , cancer , cognitive impairment [54 , 56 , 57] . Actin is highly conserved and abundant protein that exists in two forms in cells: globular monomeric ( G-actin ) and the active filamentous polymeric ( F-actin ) form . Actin undergoes multiple cycles of rapid nucleation , polymerization and disassembly , which is needed for remodeling the actin cytoskeleton . This organization of actin filaments is needed for vesicle transport , endocytosis , cell division and other functions in response to stimuli [58] . The cell frequently remodels the actin cytoskeleton with the help of ~100 highly conserved accessory proteins [58] . Regulation and activation of the accessory proteins , such as Arp2p and Arp3p actin nucleators and Cof1p , is needed to organize actin filaments in specific regions of the cytoplasm to carry out various functions . Viruses dramatically reorganize the infected cells to support viral replication . Since the actin network is a key element in cell organization and transport of cellular proteins , lipids and intracellular organelles , viruses might target the actin network to reprogram cellular pathways and aid VRC assembly [59] . In this work , we show that cofilin actin depolymerization factor is targeted by tombusviruses to promote virus replication . Using temperature-sensitive ( ts ) Cof1p mutants at the semi-permissive temperature , we find increased level of TBSV RNA accumulation in yeast cells . The in vitro activity of the tombusvirus replicase prepared from cof1-8ts yeast was also higher than the activity of the replicase prepared from wt yeast . Also , over-expression of Cof1p or a plant homolog ( Adf2 ) reduced TBSV accumulation in yeast . In addition , we show that TBSV replication depends on actin organization . Accordingly , inhibition of actin dynamic function via act1 mutations in yeast or pharmacological inhibitors in plant cells led to increased TBSV replication . We demonstrate that the actin network affects the ability of TBSV to recruit host proteins and sterols to the viral replication sites . Altogether , the emerging picture is that TBSV , via direct interaction between p33 and Cof1p , controls cofilin activities to obstruct the dynamic actin network to lead to efficient subversion of cellular proteins and sterols for pro-viral functions . In our previous high throughput screens , we observed that the actin network was represented by one of the highest number of genes identified for TBSV ( namely 18 genes ) . Particularly , we were interested in COF1 , since Cof1p has been shown to interact with the tombusviral p33 replication protein [30] . Since Cof1p is a key actin depolymerization factor [54 , 55] , binding of the tombusviral p33 replication protein to Cof1p might inhibit Cof1p interaction with ADP-actin and block actin disassembly and recycling of actin monomers to form new actin filaments . To test if the interaction between Cof1p and p33 replication protein is relevant during TBSV replication , we used a temperature-sensitive ( ts ) Cof1p mutant yeast in viral replication studies . Interestingly , the data from Northern blot analysis showed that cof1-8ts yeast supported TBSV replicon ( rep ) RNA accumulation at ~8-fold increased level at the semi-permissive temperature in comparison with yeast carrying wt COF1 , while the two strains showed comparable TBSV repRNA accumulation at the permissive temperature ( compare lanes 7–12 versus 1–6 , Fig 1A and 1B ) . These data indicate that Cof1p is an inhibitor of TBSV RNA replication in yeast host . The amount of p33 and p92pol was comparable in these yeast strains at the semi-permissive temperature , suggesting that Cof1p does not affect translation or stability of the viral replication proteins . Using the cof1-5ts mutant of Cof1p also showed ~3 . 5-fold increased TBSV repRNA accumulation at the semi-permissive temperatures ( S1A Fig ) , further supporting that Cof1p affects TBSV replication . To obtain additional evidence on the inhibitory role of Cof1p , we down-regulated Cof1p level in yeast using the titratable TET promoter [60] . We found that TBSV repRNA accumulation increased by ~2 . 5-fold when Cof1p was down-regulated ( lanes 13–18 , Fig 1C ) . Thus , TBSV replication is stimulated by reduced level of Cof1p in yeast . To test if Cof1p affects the viral replicase activity , we obtained cell-free extracts ( CFE ) from cof1-8ts and wt yeast , co-expressing p33 and p92pol . The CFE was programmed with the TBSV ( + ) repRNA in vitro , allowing one cycle of full TBSV repRNA replication [21] . We observed ~3-to-4-fold higher repRNA replication in CFE from cof1-8ts than in CFE from wt yeast ( Fig 1D , lanes 5–8 versus 1–4 ) . Interestingly , both the newly made ( + ) repRNA and double-stranded dsRNA ( correlating with the minus-stranded RNA [61] ) levels increased to a similar extent in CFE from cof1-8ts ( Fig 1D ) . Thus , Cof1p affects the early stage of TBSV replication , possibly VRC assembly prior to ( - ) -strand RNA synthesis . Altogether , the in vitro data are consistent with the results from yeast , supporting that Cof1p is a negative regulator of TBSV replication . Over-expression of the zz-His6-tagged Cof1p reduced TBSV repRNA accumulation by ~50% in both cof1-8ts and wt yeast at the permissive temperature ( 23°C , Fig 2A , lanes 1–4 and 5–8 ) . The inhibitory effect of Cof1p was even more pronounced at the semi-permissive temperature ( 32°C ) , resulting in ~30-fold reduction in TBSV repRNA level in cof1-8ts yeast over-expressing Cof1p ( Fig 2B , lanes 1–2 versus 5–6 ) . The amount of p33 and p92pol was comparable in these yeast strains at both temperatures . We also found a strong inhibitory effect on TBSV repRNA accumulation when the native ( nontagged ) Cof1p was over-expressed in cof1-8ts and wt yeast strains ( S2 Fig ) . To test if cofilin can directly inhibit TBSV replication , we purified recombinant Adf2 of Nicotiana tabacum , a plant homolog of the yeast Cof1p [62–64] . Adding purified recombinant NtAdf2 to yeast CFE ( as shown schematically in Fig 2C ) performing one full cycle of TBSV replication ( including both minus- and plus-strand synthesis ) resulted in up to 70% inhibition of TBSV RNA synthesis ( Fig 2D ) . This experiment showed that cofilin/ADF could directly inhibit TBSV replication in vitro , likely due to sequestering the p33 replication protein . To test if the higher level of TBSV RNA replication and the more active tombusvirus replicase from cof1-8ts yeast are due to the different subcellular localization or an altered distribution of the replication proteins , we used confocal laser microscopy imaging of live yeast cells . This study revealed that p33 replication protein formed a few large punctate structures in cof1-8ts yeast at the semi-permissive temperature in contrast with the more numerous , but smaller punctate structures in wt yeast ( Fig 3 ) . However , YFP-p33 was co-localized with the peroxisomal marker protein ( Pex13p ) in cof1-8ts yeast ( Fig 3A ) , suggesting that p33 is localized to peroxisome-derived membranes in both cof1-8ts and wt yeast cells . Altogether , the presence of large replication foci in cof1-8ts yeast suggests that tombusvirus replication proteins are likely more efficient in organizing the viral replication compartments containing many VRCs than in wt yeast cells . The formation of large p33-containing peroxisomal structures in cof1-8ts yeast expressing the Cof1 ts mutant ( Fig 3A ) were similar to those observed with phospholipid synthesis mutants that also supported TBSV replication at a high level [44 , 46 , 65] . These data suggest efficient assembly of VRCs when cofilin is mutated . To study if the presence of p33 changes the subcellular localization of Cof1p , we co-expressed RFP-tagged p33 and GFP-tagged Cof1p in cof1-8ts yeast cells at the semi-permissive temperature , followed by confocal laser microscopy . Interestingly , in the presence of p33 , Cof1p showed not only diffused localization in the cytosol , but also formed several punctate structures , which frequently co-localized with p33 ( Fig 3B , left images ) . In contrast , GFP-Cof1p was present mostly in the diffused form in the cytosol in cof1-8ts yeast in the absence of p33 ( Fig 3B , right panel ) . These results indicate that a fraction of Cof1p is co-localized with p33 at both early ( 5 hour ) and late time points , indicating that the co-expressed p33 can change the subcellular localization of Cof1p in cof1-8ts yeast . The viral replication protein-induced relocalization of Cof1p might inhibit the normal cellular function of Cof1p ( see Discussion ) . Since Cof1p is a key actin depolymerization factor [54 , 55] , and binds to the tombusviral p33 , it is plausible that p33 might inhibit Cof1p interaction with ADP-actin and block actin filament disassembly and recycling of actin monomers that is needed to form new actin filaments . Indeed , co-expression of p33 and p92 replication proteins in yeast led to the formation of a large actin-filament network ( called actin patches , which contain short , branched actin filaments with 1–2 minute turnover time ) in the cell , visible under confocal microscope ( Fig 3C ) . Formation of large cortical actin patches is indicative of lack of Cof1-driven disassembly of actin filaments and inhibition of dynamic actin function [52 , 58] . Indeed , blocking Cof1p function via applying semi-permissive temperature in cof1-8ts yeast or using the actin polymerization inhibitor , Cytochalasin-D , has led to similar actin organization ( large cortical actin patches ) to that induced by p33/p92 expression ( Fig 3C ) . The cortical actin patches were also influenced by expressing either p33 or p92 alone , albeit the effects were lesser than those observed with co-expression of p33 and p92 ( Fig 3C ) . These observations suggest that p33 and p92 interferes with the normal organization of actin in yeast , perturbing actin organization as efficiently as Cof1p mutation or chemical inhibition of actin polymerization . It is worthwhile to mention that p92 shares an identical N-terminal sequence with the full-length p33 and both p33 and p92 can affect actin organization under the experimental conditions ( Fig 3C ) . Based on these observations , we propose that TBSV replication proteins actively block actin organization via the interaction of p33/p92 and Cof1p . The best-characterized function of the multifunctional Cof1p is to bind to ADP-bound actin and depolymerize actin filaments in yeast cells [54 , 55] . To test if the actin-organization function of Cof1p is important for the inhibitory function of Cof1p on TBSV replication , we used a group of Cof1p mutants in the yeast-based TBSV replication assay . Interestingly , a Cof1p mutant that binds poorly to G- or F-actin ( cof1-22 ) [66] supported TBSV replication at ~3-fold higher level than wt ( S1B Fig ) . Also , those mutants that are defective in actin organization ( such as cof1-5ts and cof1-8ts ) supported TBSV replication at ~4-to-8-fold increased levels at the semi-permissive temperature ( Fig 1A and S1A Fig ) . In contrast , Cof1p mutants that have wt-like organization of actin cytoskeleton [such as cof1-6 , cof1-10 , cof1-11 , cof1-12 , cof1-13 and cof1-15; [66]] did not increase TBSV replication ( S1C Fig ) . We also tested if Cof1p mutants can interact with p33 in yeast using a split-ubiquitin assay . All the Cof1p mutants tested , including cof1-5ts , cof1-8ts and cof1-20 , interacted with p33 ( S1D Fig ) . Altogether , these data suggest that the ability of Cof1p to organize actin filaments in yeast is important for TBSV replication . The above data with Cof1p mutants suggested that Cof1p likely inhibits TBSV replication by organizing or remodelling actin filaments due to its ability to promote actin depolymerization and disassembly . To test this model , we used actin mutants , which are defective in actin organization under semi-permissive temperature [67] , in the yeast-based TBSV replication assay . Six yeast strains carrying actin ts mutants supported ~3-4-fold higher level of TBSV replication than wt yeast did at the semi-permissive temperature ( Fig 4A–4C ) . These yeast mutants produced p33 and p92 to a comparable level with wt yeast ( Fig 4A–4C ) . Overall , these actin mutants behaved similarly to selected cofilin mutants ( Fig 1 and S1 Fig ) by supporting high level of TBSV replication , thus suggesting that cofilin and the actin mutants might promote TBSV replication in some similar ways . Isolation of the membrane fraction of yeasts containing the VRCs , followed by testing for viral RNA synthesis in vitro , revealed that the viral replicase from three yeast strains expressing actin ts mutants supported 2-4-fold increased TBSV repRNA replication ( Fig 4D ) . The in vitro data are from normalized replication assays that are based on comparable level of p33 replication proteins . Altogether , the in vitro data also support that Act1p mutations have similar effects on TBSV replication to Cof1p mutants . To further demonstrate the active role of the actin network in TBSV replication , we applied Cytochalasin-D inhibitor in yeast , which blocks the polymerization of new actin filaments [59] . We found that the treatment with the inhibitor led to ~2 . 5-fold increase in TBSV replication ( Fig 4E ) . Similarly , ts mutation in Arp2 , which is an actin polymerization factor [58] , resulted in ~2 . 5-fold increase in TBSV replication at the semi-permissive temperature ( Fig 4F ) . These data confirmed that inhibition of new actin filament polymerization needed for dynamic actin remodelling in cells stimulates TBSV replication . To validate whether our findings with yeast cells on the role of Cof1p and actin in TBSV replication are valid in plant cells , first we tested the effect of Cytochalasin-D and Latrunculin-B inhibitors [68] of actin organization and new actin filament formation on TBSV replication . Treatments of N . benthamiana protoplasts with these inhibitors led to ~2-to-2 . 5-fold increase in TBSV genomic ( g ) RNA accumulation ( Fig 5A ) . These data suggest that dynamic actin network is inhibitory to TBSV replication in single plant cells . Treatment of whole leaves of N . benthamiana with Cytochalasin-D also resulted in increased TBSV RNA accumulation by up to ~5-fold in the treated leaves . This was followed by the appearance of symptom intensification in the treated plant compared to the control ( ethanol-treated ) plants ( Fig 5B ) . Treatment of whole leaves of N . benthamiana with CK-636 , which is an inhibitor of Arp2/3 [69] also resulted in ~2 . 5-fold increase in TBSV genomic ( g ) RNA accumulation ( Fig 5C ) . Similarly , knocking down NbAdf2 ( the homolog of the yeast Cof1p ) by VIGS in N . benthamiana also led to ~3-fold increase in TBSV gRNA accumulation in infected leaves ( Fig 6A ) . This ultimately led to the rapid death of TBSV-infected N . benthamiana plants when compared with the empty vector agroinfiltrated plants ( Fig 6A ) . Arabidopsis genome codes for four highly conserved ADF genes , which are 83-to-88% identical in nucleotide sequences [62–64] , making it highly likely that our VIGS approach knocked down the expression levels of all ADF genes . Electron microscopy ( EM ) images of plant cells from whole plants treated with Cytochalasin-D showed large areas of tightly packed virions in the cytosol ( Fig 7 ) . Comparable areas in the control plant cells contained fewer numbers of virions , which were packed less tightly together . However , it seems that the sizes of the virus-induced spherules ( containing VRCs ) were comparable in the Cytochalasin-D- treated and control plant cells ( albeit based on only a small number of spherules , Fig 7 ) . The increased number of virions in Cytochalasin-D-treated plant cells was in agreement with the higher accumulation level of TBSV RNAs in Cytochalasin-D-treated than in control plants ( Fig 5 ) . Based on all these data , the roles of cofilin ( i . e . Adf2 in plants ) and actin organization in replication of the TBSV genomic RNA have been firmly established in plants . To study the putative connection between the actin network and tombusvirus replication sites , we performed super-resolution microscopy on wt yeast cells actively replicating TBSV repRNA . The replication sites were detected with anti-p33 antibody , which also recognizes p92pol because of the overlapping sequence in the pre-readthrough region in p92 . The actin filaments were detected by phalloidin ( F-actin specific reagent ) attached to Atto-488 dye . Interestingly , several large tombusvirus replication sites were located in the vicinity of the actin filaments in wt yeast ( Fig 8A and S1 Video ) . Also , the actin filaments formed larger structures ( called “patches” with up to 2-fold increase in diameter ) in yeast cells replicating TBSV than in the absence of viral components ( Fig 8B ) . Because the actin cables are difficult to visualize in the tiny yeast cells , we also used plant cells transiently expressing RFP-tagged p33 replication protein . The actin cables were visualized via transgenically expressed GFP-tagged mTalin , which binds to actin filaments [70] . Confocal laser microscopy imaging revealed that the largest TBSV replication sites ( marked by p33-RFP ) , which are known to form on aggregated peroxisomal membrane surfaces , were formed where actin cables intersected ( Fig 9 ) . Individual cross-sections showed that the actin cables surrounded and sometimes crossed through the large viral replication sites/organelles ( Fig 9 , images on the right , and S3 Fig , which shows the structures in the presence of TBSV infection ) . Interestingly , we observed very similar arrangements between actin cables and viral replication sites ( marked by p36-RFP ) in case of Carnation Italian ringspot virus ( CIRV ) , a closely related tombusvirus that replicates on mitochondrial membrane surfaces ( Fig 10 and S4 Fig , which shows the structures in the presence of CIRV infection ) . Comparison of actin filament network revealed overall higher density and thicker and larger number of individual filaments in TBSV-replicating ( S3 Fig versus S5 Fig ) , and to a lesser extent , CIRV-replicating plant cells ( S4 Fig ) than in uninfected control cells ( S5 Fig ) . Thus , the actin network plays a role in formation of tombusvirus replication sites that resemble replication organelles in plant cells [71] . The close association between the actin network and tombusvirus replication proteins in both yeast and plant cells indicated a putative functional role for actin in tombusvirus replication . Since both cofilin and actin are essential for yeast and plant cells , in the above experiments we used approaches that temporarily and partially blocked the dynamic actin network , but still allowed the function of the stable actin network . To test the putative role of the actin network , we inhibited the functions of both dynamic and stable actin networks via expressing dominant negative mutant of myosin XI-K that is a motor protein involved in actin-based moving of cargos in the cytosol of plant cells [72 , 73] . Interestingly , expression of a dominant negative mutant of myosin XI-K strongly inhibited tombusvirus replication in N . benthamiana leaves ( Fig 6B ) . All these data support the functional role of the actin network in tombusvirus replication . Because TBSV replication greatly depends on the virus-driven retargeting of sterols , which become highly enriched at viral replication sites and facilitate the efficient assembly of VRCs [45 , 47] , we analyzed the sterol distribution in cofilin and actin mutant yeasts . Filipin-based staining of sterols , which can be visualized by fluorescent microscopy [45] , revealed the formation of large sterol-enriched compartments in cofilin and actin mutant yeasts when compared to wt yeast showing smaller sterol-enriched compartments ( Fig 11 ) . The difference in sterol enrichment and distribution was especially striking in the presence of viral components in cof1-8ts ( Fig 11E ) , act1-132ts and act1-121ts ( Fig 11B and 11C ) yeasts in comparison with the wt yeast at the semi-permissive temperature . We have shown previously that these internal ( not plasma membrane localized as observed in the absence of viral components in wt yeast , Fig 11A ) sterol-enriched compartments represent the viral replication sites [45] . In comparison with the wt yeast , the cofilin and actin mutant yeasts showed less even distribution of sterols in the plasma membrane , suggesting defect in normal sterol transport in case of the cofilin and actin mutants ( Fig 11 ) . Based on these observations , we suggest that TBSV could take advantage of the defective sterol transport mechanism in cofilin and actin mutant yeasts that likely allows the virus to easily hijack sterols and enrich them at the sites of viral replication that leads to highly efficient VRC assembly and increased viral RNA synthesis . To examine if the membrane fraction from the cof1-8ts or act1-132ts yeasts is more suitable for TBSV replication than from wt yeast at the semi-permissive temperature , we used a mix-and-match CFE approach as shown schematically in Fig 12A . The combination of membrane fraction from cof1-8ts with the soluble fraction from the wt yeast ( Fig 12B , lane 5 ) supported TBSV repRNA replication almost as efficiently as the nonfractionated CFE from cof1-8ts ( Fig 12B , lane 3 ) . On the contrary , the combination of membrane fraction from wt yeast with the soluble fraction from cof1-8ts supported TBSV repRNA replication at a ~4-fold reduced level ( Fig 12B , lane 4 ) , suggesting that the membrane fraction of cof1-8ts is responsible for the enhanced TBSV replication in vitro . We obtained similar results with the membrane fraction of act1-132ts ( Fig 12C , lane 5 ) , which was responsible for the enhanced level of TBSV repRNA replication in vitro ( Fig 12C ) . Altogether , these in vitro data are in agreement with the model that the intracellular membranes in cof1-8ts or act1-132ts yeasts are more suitable for TBSV replication , possibly because they are sterol-rich , in comparison with the intracellular membranes in wt yeast . TBSV controls intracellular sterol transport via hijacking oxysterol binding proteins ( OSBP-like or ORP , such as Osh3p , Osh5p , Osh6p and Osh7p ) and VAP protein ( VAMP-associated protein , such as yeast Scs2p or Vap27-1 , a plant ortholog ) via interaction with the p33 replication protein [45] . These cellular proteins help TBSV to stabilize membrane contact sites ( MCSs ) between the ER and the peroxisome , thus facilitating the efficient transport of sterols from the ER to the peroxisomal membranes where a large number of VRCs forms [45] . To test if TBSV could efficiently co-opt Osh6p and Vap27-1 [45] in yeast cofilin or actin mutants , we purified p33 replication proteins from detergent-solubilized membrane fractions , followed by Western blotting to measure the co-purified cellular Osh6p and Vap27-1 . We found ~2-to-2 . 5-fold increase in the co-purified cellular Osh6p in cof1-8ts , cof1-22ts ( Fig 13B ) , act1-121ts ( Fig 13A ) and act1-132ts ( Table 1 ) in comparison with wt yeast at the semi-permissive temperature . The Osh6p protein was also more efficiently co-purified with p33 in act1-105ts , and act1-112ts yeasts in comparison with wt yeast at the permissive or semi-permissive temperature ( S6 Fig ) . Interestingly , we also found ~2-to-2 . 5-fold increase in the co-purified cellular Vap27-1 , in cof1-8ts , cof1-22ts ( Fig 13D ) , act1-121ts ( Fig 13C ) and act1-132ts ( Table 1 ) in comparison with wt yeast at the semi-permissive temperature . Altogether , the more efficient co-purification of Osh6p and Vap27-1 in the cofilin and actin mutants suggests the enhanced or more stable formation of viral-induced MCSs and could explain the high enrichment of sterols at internal sites ( likely representing the replication sites ) in the cofilin and actin mutants . To further test if indeed the actin mutants affect sterol-enrichment at viral replication sites , we deleted Scs2p VAP protein in yeast ( double mutant act1-121ts scs2Δ yeast ) , followed by measuring TBSV repRNA replication . Interestingly , the double mutant supported a reduced level of repRNA replication , comparable to that observed in the single mutant yeast ( scs2Δ ) ( Fig 13E ) . These data suggest that the actin mutant could enhance TBSV replication only when functional VAP protein is present , thus when MCSs are formed , in act1-121ts yeast . The co-purification-based proteomics approach also revealed that co-opted host factors , such as Cdc34 ubiquitin-conjugating enzyme , Rpn11 deubiqutinase and Vps23p ESCRT factor , were co-purified with the tombusvirus replicase by ~2-fold more efficiently from act1-132ts yeast in comparison with wt yeast ( Table 1 ) . Other cellular host factors , such as Tef1p ( eEF1A ) , Tdh2p ( GAPDH ) , DDX3-like Ded1p and Vps4p AAA ATPase , were co-purified with p33 as efficiently from act1-132ts yeast as from wt yeast ( Table 1 ) . Based on these data , we suggest that the actin mutations also facilitate the subversion of some cellular host factors by TBSV into the VRCs . Using tombusviruses , we have discovered that the cellular cofilin actin depolymization factor is targeted by TBSV to obstruct the dynamic actin network . Down-regulation of Cof1p level or the use of Cof1p ts mutants at the semi-permissive temperature lead to increased level of TBSV RNA accumulation in yeast cells . The increased RNA replication is due to highly active VRCs , based on testing the in vitro activity of replicase in CFE prepared from yeast expressing cof1-8ts , which was ~3-fold more active than the replicase operating in wt CFE ( Fig 1D ) . Also , expression of NtAdf2 , a plant homolog of Cof1p [62 , 63] reduced TBSV replication in vitro ( Fig 2D ) . It seems that the inhibitory effect of Cof1p on TBSV replication is due to two different mechanisms . First , the interaction between Cof1p ( Adf2 homolog ) and p33 and possibly p92 could partially sequester the replication proteins , thus limiting their participation in VRC formation and TBSV RNA replication . Second , Cof1p also seems to inhibit TBSV replication via facilitating the dynamic actin re-organization in live yeast cells . The Cof1p-driven disassembly of existing actin filaments in combination with formation of new actin filaments might inhibit the continuous growth of the replication compartments , which could depend on stable MCSs for sterol transfer/enrichment at the sites of replication and steady supply of host factors to the replication sites ( Fig 14 ) . Accordingly , yeasts expressing Act1p ts mutants at the semi-permissive temperature supported increased level of TBSV RNA accumulation ( Fig 4 ) . Moreover , pharmacological inhibition of actin organization by Cytochalasin D or Latrunculin B in yeast , plant protoplasts or whole plants also resulted in increased level of TBSV RNA accumulation ( Figs 4–7 ) . Thus , actin organization , especially the formation of new actin filaments is a major restriction factor for virus replication and TBSV modulates the dynamic actin network to obstruct the formation of new actin polymers . How does dynamic actin organization limit TBSV replication in live cells ? Based on our current knowledge on TBSV replication , 15–20 cytosolic pro-viral host factors , peroxisomes/ER , phospholipids and sterols are required for efficient TBSV replication [9 , 16 , 17 , 46 , 47 , 77–79] . We predict that TBSV could easily co-opt these host factors for efficient VRC assembly when dynamic actin organization and remodeling of the actin network is inhibited . The obstructed actin network might inhibit the normal cellular distribution and function of host proteins and organelles , which might become more readily accessible to TBSV when new actin filament formation is hindered ( Fig 14 ) . Indeed , formation of large replication foci in combination with increased replicase activity is seen in cof1-8ts yeast at the semi-permissive temperature ( Fig 3 ) . Similar observation of the presence of large replication foci and increased replication was documented for yeast over-expressing Ino2p , a transcription factor regulating phospholipid biosynthesis [46] . Therefore , our model predicts that it is more difficult for the virus to recruit the host factors , sterols and peroxisomes for efficient VRC assembly when dynamic actin organization facilitates rapid transport of host components to their final destination within the cell , while disruption of actin organization via p33-Cof1p interaction and Cof1p re-localization ( in natural infections ) or via inhibitors , actin or Cof1p mutations enhances VRC assembly and TBSV replication ( Fig 14 ) . The actin network might also regulate the availability of given host factors , such as eEF1A , whose function is affected by actin through promoting GTP hydrolysis by the GTP-bound eEF1A and subsequent release of eEF1A from translation [80] that might increase the availability of eEF1A for viral functions [38 , 81] . Cofilin not only regulates actin dynamics , but also affects phospholipid metabolism via regulating phospholipase D activity [53 , 54 , 56 , 57] . However , previous genomics and proteomics screens with TBSV have not identified SPO14 or other phospholipase genes [9 , 26–30 , 32 , 34] . Therefore , the evidence is lacking for this mechanism during TBSV replication . Also , p33 interaction with Cof1p might interfere with apoptosis that requires re-localization of cofilin to the mitochondria [54] . Since other ( + ) RNA viruses also remodel subcellular membranes and rewire cellular pathways to facilitate virus replication and avoid antiviral responses [74] , it is possible that targeting cofilin and actin could be wide-spread among viruses . Accordingly , during HIV infection , the viral Nef protein inactivates cofilin molecules via its interaction with Pak2 cellular kinase that phosphorylates cofilin [82] . This results in reduced fibroblast and T cell motility , thus helping HIV to overcome T lymphocytes-mediated , chemotaxis-based antiviral response of the host . Also , HIV activates cofilin through chemokine receptor signaling to mediate entry into resting CD4 T cells [83] . These examples together with TBSV-mediated inhibition of cofilin/actin functions suggest that the actin network could be a major target for viruses to facilitate their replication in host cells . Cofilin and actin are also major targets of bacterial pathogens infecting mammalian cells , suggesting that the dynamics of actin network is key component and pathogenicity determinant [84] . Expression plasmids , S . cerevisiae strains , culturing conditions and TBSV repRNA measurements are presented in S1 Materials and Methods . Cytochalasin-D ( Santa Cruz Biotechnology ) [59] was added to overnight culture of BY4741 ( transformed with pHISGBK-CUP1-p33:ADH-DI-72 and pESC-CUP1-His-p92 ) at a final concentration of 0 . 03 and 0 . 06 μg/μl . The yeast was grown for further 24 h at 23°C and total RNA was extracted for Northern blot analysis . N . benthamiana leaves were first sap-inoculated with TBSV , and one day later , the same leaves were infiltrated with ethanol or Cytochalasin-D at 80 μg/ml concentration ( or as specified in Fig legends ) using a syringe to inhibit actin polymerization . Two days later , total RNA was isolated from the inoculated leaves and Northern hybridization was performed to measure tombusvirus RNA accumulation . In case of inhibitor treatment of the Arp2/3 complex , N . benthamiana leaves were infiltrated with DMSO ( 50 μM ) or CK-636 ( 25 or 50 μM ) ( ApexBio ) , and 12 h later , the same leaves were inoculated with TBSV-containing sap . One day later the infected leaves were again infiltrated with DMSO ( 50 μM ) or CK-636 ( 25 or 50 μM ) . One day later total RNA was isolated and Northern hybridization was performed as described above . Protoplasts were isolated from N . benthamiana leaves as previously described [87] . Freshly prepared protoplasts were treated with 70 or 130 μg/ml Cytochalasin-D or 40 μM of Latrunculin-B before electroporation with 1 μg of TBSV RNA . Protoplasts were incubated in the dark for 24 h at room temperature and the total RNA was extracted for Northern hybridization [87] . Expression plasmids pGD-T33-RFP or pGD-C36-RFP were transformed into Agrobacterium C58C1 separately , and transformed cells were selected on LB plate containing 50 μg/ml kanamycin , 100 μg/ml rifampicin and 5 μg/ml tetracycline . Transformed cells were grown in LB medium containing the antibiotics described above overnight , and suspended in MMA solution ( 10mM MES pH5 . 6 , 10mM MgCl2 , 200 μM acetosyringone ) at OD600 of 1 . 0 for 3–4 hours . Leaves of 8 weeks old transgenic Nicotiana benthamiana plants expressing GFP-mTalin ( mouse Talin ) [88] , which specifically binds to F-actin [70] ( a gift from Dr . Michael M . Goodin at University of Kentucky ) , were infiltrated with Agrobacterium in MMA solution . Leaf epidermal cells were observed under confocal laser microscope ( Olympus FV1000 microscope ) 2 days after infiltration for localization of both GFP-mTalin and p33-RFP/p36-RFP [89] . Note that GFP-mTalin transgenic Nicotiana benthamiana plants have normal growth when compared to wild type plants under green house conditions [88] . GFP-mTalin binds to F-actin , and its localization overlaps with phalloidin stained actin filaments [70] . Yeast strain BY4741 was transformed with pESC-T33/DI72 and pYES-T92 [89] . Transformed cells were pre-grown in synthetic complete medium lacking appropriate amino acids and containing 2% glucose overnight at 29°C . Then , yeast cells were grown in synthetic complete medium lacking appropriate amino acids containing 2% galactose at a starting 0 . 3 OD600 for 12–14 hours at 23°C . Cells were harvested , fixed with 3 . 7% formaldehyde and digested with zymolase 20T to remove cell wall as described in [43] . Treated cells were applied to poly-L-lysine–coated cover slips . The cells on the cover slips were sequentially immersed in methanol for 6 min and acetone for 30s at –20°C . Anti-p33 primary antibodies ( a gift from Dr . Herman B . Scholthof , Texas A&M University ) were diluted ( 1:400 ) in PBS ( pH 7 . 4 ) containing 0 . 05% Nonidet P-40 and 1% BSA and incubated overnight with the fixed cells at +4°C . Yeast cells were washed three times with PBS ( pH 7 . 4 ) with 1% BSA , then incubated subsequently with anti-mouse secondary antibody conjugated to Alexa Flour 647 for 1 h and ATTO488-phalloidin for 1 h in PBS ( pH 7 . 4 ) with 1% BSA , and washed three times with PBS ( pH 7 . 4 ) containing 1% BSA for at least 1 h to reduce background . Cells on the cover slip were subjected to super-resolution microscopic observation ( N-STORM Super Resolution Microscopy System from Nikon ) . To examine the distribution of ergoterol in BY4741 and act1ts mutants ( act1-121 and act1-132 ) , the yeast strains were co-transformed with plasmids pGBK-Hisp33-CUP1/Gal DI-72 and pGAD-Hisp92-CUP1 . Control untransformed and transformed yeasts were grown in SC minimal media supplemented with 2% galactose and 50 μM CuSO4 for 24 hours at 23°C , 27°C and 32°C . Cultures were fixed with 3% formaldehyde for 1 h at room temperature . Fixed cells were centrifuged and washed twice with distilled water . Washed cells were incubated with 5 mg/ml filipin complex ( Sigma Chemicals ) in the dark for 15 min at 23° . Filipin-based fluorescence was observed by spotting 3 μl of the cell suspensions onto poly-lysine microscope slides under UV light microscope using DAPI filter [45] . For co-purification of Osh6p protein with the membrane-bound p33 and p92 replication proteins [45] , yeast strains BY4741 , cof1-8ts , cof1-22ts , and the act1-121ts mutants were co-transformed with plasmids HpGBK-CUP1-FLAGp33/GAL1-DI-72 and LpGAD-CUP1-FLAGp92 and ( or HpGBK-CUP1-Hisp33/Gal1-DI-72 and LpGAD-CUP1-Hisp92 as a control ) plus the UpYC-NT plasmids expressing His6-tagged Osh6 protein from the GAL1 promoter ( UpYC-NT-Gal1-OSH6 ) . Transformed yeasts cultures were pre-grown in selective SC-ULH− + BCS medium supplemented with 2% glucose for 12 h at 23°C and then transferred to selective medium supplemented with 2% galactose for 24 h at either 23 or 32°C to induce His6-Osh6 protein expression from the GAL1 promoter . Then the cultures were supplemented with 50 μM CuSO4 to induce expression of FLAG-p33 and FLAG-p92 or His6-p33 and His6-p92 from the CUP1 promoter , and yeast cultures were grown for an additional 4 h at 23°C or 32°C . The cultures were centrifuged washed once with phosphate-buffered saline ( PBS ) , and then incubated in PBS buffer containing 1% formaldehyde for 1 h on ice to cross-link proteins [45] . Finally , yeast cultures were washed in PBS and proteins were FLAG-affinity purified as described previously [45] .	The actin network , which is a central node in cellular pathways , is frequently targeted by various pathogens to modulate cellular responses . In this paper , the authors show that TBSV interacts with cofilin actin depolymerization factor leading to inhibition of the dynamic function of the actin network in infected cells . This allows TBSV to utilize the existing actin filaments to efficiently recruit host proteins and lipids for viral replication and to build viral replication compartments for robust viral replication . Altogether , subversion of the actin network by TBSV is a key step for the virus to gain access to cellular resources required for virus replication .	[ "Abstract", "Introduction", "Results", "Discussion", "Materials", "and", "Methods" ]	[ "cell", "motility", "plant", "anatomy", "actin", "filaments", "plant", "cell", "biology", "microbiology", "membrane", "proteins", "fungi", "plant", "science", "cellular", "structures", "and", "organelles", "dynamic", "actin", "filaments", "contractile", "proteins", "a...	2016	Viral Replication Protein Inhibits Cellular Cofilin Actin Depolymerization Factor to Regulate the Actin Network and Promote Viral Replicase Assembly
In prokaryotes , clustered regularly interspaced short palindromic repeats ( CRISPRs ) and their associated ( Cas ) proteins constitute a defence system against bacteriophages and plasmids . CRISPR/Cas systems acquire short spacer sequences from foreign genetic elements and incorporate these into their CRISPR arrays , generating a memory of past invaders . Defence is provided by short non-coding RNAs that guide Cas proteins to cleave complementary nucleic acids . While most spacers are acquired from phages and plasmids , there are examples of spacers that match genes elsewhere in the host bacterial chromosome . In Pectobacterium atrosepticum the type I-F CRISPR/Cas system has acquired a self-complementary spacer that perfectly matches a protospacer target in a horizontally acquired island ( HAI2 ) involved in plant pathogenicity . Given the paucity of experimental data about CRISPR/Cas–mediated chromosomal targeting , we examined this process by developing a tightly controlled system . Chromosomal targeting was highly toxic via targeting of DNA and resulted in growth inhibition and cellular filamentation . The toxic phenotype was avoided by mutations in the cas operon , the CRISPR repeats , the protospacer target , and protospacer-adjacent motif ( PAM ) beside the target . Indeed , the natural self-targeting spacer was non-toxic due to a single nucleotide mutation adjacent to the target in the PAM sequence . Furthermore , we show that chromosomal targeting can result in large-scale genomic alterations , including the remodelling or deletion of entire pre-existing pathogenicity islands . These features can be engineered for the targeted deletion of large regions of bacterial chromosomes . In conclusion , in DNA–targeting CRISPR/Cas systems , chromosomal interference is deleterious by causing DNA damage and providing a strong selective pressure for genome alterations , which may have consequences for bacterial evolution and pathogenicity . Prokaryotes are constantly challenged with foreign genetic elements such as bacteriophages ( phages ) and plasmids [1] . These interactions are frequent and important on a global scale . For example , of the estimated 1031 phages on earth , approximately 1025 participate in infections of bacteria every second [2] affecting biogeochemical cycles such as the carbon cycle [3] . The strong selective pressure has resulted in the evolution of numerous mechanisms of ‘innate immunity’ in bacteria [1] , [4] , such as abortive infection systems [5] , and recent research has demonstrated that a prokaryotic ‘adaptive immune system’ exists . These ‘adaptive immune systems’ , termed Clustered Regularly Interspaced Short Palindromic Repeats ( CRISPRs ) , are a small RNA-based bacterial defence mechanism with some similarities to eukaryotic RNA interference and microRNAs ( for reviews see [6]–[11] ) . Simply , CRISPRs are an important part of an ‘immune system’ with genetic memory against extrachromosomal agents such as plasmids and phages . CRISPRs are found in ∼50% of sequenced bacteria and ∼85% of archaea [12] and are comprised of an AT-rich leader sequence of several hundred base pairs followed by short repeats interspersed with similar sized spacers of unique sequence . Spacers are derived from foreign nucleic acids and are important in the sequence-specific interference of phages and plasmids [9] , [13] . Closely associated with CRISPRs are the cas genes ( CRISPR associated ) 14–16 , which are necessary for resistance . CRISPR arrays and their associated cas genes are diverse , with CRISPR/Cas systems falling into three major types ( I–III ) , which are divided into further subtypes [17] , [18] . The characterised mechanism of CRISPR/Cas interference involves three phases; 1 ) resistance acquisition ( spacer incorporation into the CRISPR array [11] , [13] ) , 2 ) expression of cas genes and transcription and processing of the CRISPR arrays into small RNAs ( crRNAs ) [19]–[23] and 3 ) interference of either RNA [24]–[26] or DNA [27]–[29] at sites in the target element . Sequences in the invading element , from which spacers are derived and subsequently targeted , are termed protospacers . Adjacent to the protospacers , short motifs are present ( termed CRISPR motifs or protospacer adjacent motifs ( PAMs ) ) , that are important for both incorporation [30]–[32] and targeting [33]–[35] . The resistance mechanism is mediated by the Cas proteins , many of which have been shown to interact as ribonucleoprotein complexes [19] , [24] , [26] , [36]–[39] . CRISPR spacers that are homologous to database sequences are predicted to have targets in plasmids , phages and chromosomal genes . While CRISPR-interference of phages [13] and plasmids [28] has been proven , by comparison , the role of chromosomal targeting has received little attention . Initial analyses of CRISPR spacers showed that only ∼2% of spacers have identity to database sequences [9] . In lactic acid bacteria , of 104 spacers with 100% identity to databases , 73% matched with phage-related sequences , 5% with plasmids and 22% elsewhere within their own genome [40] . Another study showed that within archaea , 19% ( of 58 matches ) had identity elsewhere in the host chromosome [41] . This begs the question what role these chromosomal targeting spacers have . One early proposal was that CRISPRs might act as a gene regulation mechanism [7] , but this has not yet been shown . A small number of studies indicate that chromosomal targeting can be detrimental [42]–[44] and a bioinformatic analysis of chromosomal targeting led to the suggestion that chromosomal targeting is a case of autoimmunity [45] . The authors proposed that spacer incorporation from chromosomal protospacers is lethal and , as such , they observed a correlation between mutations that were predicted to interfere with the hypothesised toxicity and chromosomally-derived spacers . These hypotheses have yet to be tested in wet-lab experiments . In this study we have determined the effects of chromosomal targeting using both engineered and pre-existing spacers and tested the hypotheses proposed by Stern et al ( 2010 ) . To investigate CRISPR/Cas-mediated chromosomal targeting we utilised the potato phytopathogen Pectobacterium atrosepticum , which contains a single type I-F ( Ypest ) CRISPR/Cas system composed of Cas1 , a Cas2–Cas3 fusion , Csy1 , Csy2 , Csy3 and Cas6f ( originally termed Csy4 ) and 3 CRISPR arrays with CRISPR-4 type repeats [46] and 28 , 10 and 3 spacers [23] , [47] . Previously , we demonstrated this CRISPR/Cas system is transcribed and the CRISPR arrays are processed into crRNAs by Cas6f [23] . P . atrosepticum contains one spacer with a perfect match to a chromosomal gene within a horizontally acquired island ( HAI2 ) . Here , we provide direct experimental evidence that the targeting of chromosomal genes by CRISPR/Cas systems is toxic and show the various mechanisms that enable avoidance of this autoimmunity , including the dramatic reshaping of pathogenicity islands within the bacterial genome . This suggests that CRISPR/Cas systems have played a greater role in bacterial genome evolution than previously appreciated . Furthermore , these experiments provide an insight into functional details of type I-F systems and show that CRISPR/Cas systems can be engineered to delete specific regions of bacterial genomes and thus , can provide a tool for genome engineering . To test the effect of chromosome-targeting by CRISPR/Cas systems , a strategy was developed to engineer CRISPR arrays ( Materials and Methods ) ( Figure 1A ) . An array was engineered with the native type I-F CRISPR1 leader and three sense-orientation ( i . e . cannot target mRNA ) spacer-repeat units targeting the expI gene in P . atrosepticum ( Figure 1A; spacer and PAMs indicated in Table S1 ) . The expI gene encodes the N-acyl homoserine lactone synthase , which produces quorum sensing signals [48] . When the expI-targeting plasmid was transformed into P . atrosepticum , the efficiency was similar ( Figure S1 ) but transformants were almost undetectable compared with control plasmids containing either no spacers , or three spacers that do not match chromosomal targets ( a scrambled control ) ( Figure 1B ) . To determine if the toxicity was Cas-dependent , the effect of deleting the cas operon was assessed . The toxic effect of the chromosomal expI-targeting plasmid was abolished by deletion of the cas operon ( Figure 1B ) . Therefore , CRISPR/Cas systems with specific spacers that target the chromosome are detrimental to bacterial growth . The engineered CRISPR array included 780 bp of sequence 5′ of the first proximal repeat . Attempts to repress expression from these pBAD30-derived plasmids did not abolish toxicity ( Figure S2A ) , supporting our previous assignment of the CRISPR1 promoter within ∼180 bp of leader [23] . To develop tightly-controlled CRISPRs , a plasmid truncation series was produced with 780 , 180 , 52 and 16 bp 5′ of the first repeat . Only plasmids with 16 bp were controllable , which led to the identification of a putative CRISPR1 promoter within 52 bp of the leader ( Figure S2 ) . In the WT background , controlled induction of the expI-targeting plasmid resulted in a cessation of growth ( a plateau in OD600 ) compared with the scrambled control ( Figure 2A ) . Furthermore , no growth inhibition occurred in the cas mutant , or in the WT when grown under repressed conditions ( Figure 2B and 2C ) . A lacZ-targeting array was constructed and was also toxic in the WT but not the Δcas mutant ( Figure 2D ) . Note that P . atrosepticum utilises arabinose and grows to a higher OD600 when compared with the repressed ( glucose-grown ) controls . We hypothesized that a single spacer would be sufficient for targeting , since the 3× anti-lacZ plasmid contained only one spacer complementary to a protospacer with a consensus PAM ( protospacers contained 5′-protospacer-AC-3′ and 5′-protospacer-GT-3′ PAMs and the consensus 5′-protospacer-GG-3′ PAM [33] , [49] ( Table S1; the protospacer is defined as the target strand complementary to the crRNA and the PAM is denoted 5′-3′ on this strand [8] ) . In agreement , expression of CRISPRs containing only one spacer against either expI or lacZ inhibited bacterial growth ( Figure 2E ) . Controls using either repressed conditions or in the Δcas strain always demonstrated no effect of the chromosomal-targeting plasmids , as shown in Figure 2A–2D , and are therefore not shown for clarity here and in later figures . The growth inhibition , as measured by OD600 , was detected around mid-exponential phase ( 6–7 h ) for both targets , which corresponds with the increased expression of a native cas operon measured using a chromosomal cas-lacZ transcriptional/translational reporter ( Figure 2F ) . With plasmids containing one , two , three or eight identical chromosomal targeting spacers there was no apparent additional effect ( Figure 2G ) . The reduced OD600 resulting from a single spacer targeting expI , was reflected in a ∼105 reduction in viable count ( cfu/ml ) ( Figure 2H ) . The viable counts were assessed on media that repressed crRNA synthesis , indicating that most cells could not readily recover following chromosomal targeting , but that a subpopulation survived . The initial inoculum in these experiments was ∼107 cfu/ml and the final viable count following targeting was ∼104 cfu/ml compared with ∼109 cfu/ml for the negative controls . Together , these experiments demonstrate that a single spacer that targets the chromosome causes cas-dependent toxicity and a reduction in viable count . The toxicity following chromosomal targeting prompted an examination of morphological changes to the cells . LIVE/DEAD staining and fluorescence microscopy following 2 h of expression of a single spacer targeting expI led to the detection of elongated/filamentous cells ( Figure 3A ) . Interestingly , most filamentous cells were stained with SYTO9 , but not propidium iodide ( PI ) , indicating that , of the filamentous cells detected , they were still viable and maintaining membrane integrity ( Figure 3A ) . In all conditions including controls , a few cells were stained with PI , suggesting loss of viability in a subpopulation . Following chromosomal targeting , cells were also imaged by transmission electron microscopy ( TEM ) ( Figure 3B ) . Targeting of expI significantly increased the mean cell length to ∼10 µm compared with ∼2 µm with the non-targeting control , but some cells were over 20 µm ( Figure 3C ) . Therefore , chromosomal targeting caused a dramatic reduction in viable count and had a bacteriostatic effect on some cells , resulting in elongation . This cell elongation is likely to explain the slight increase in OD600 measurements prior to the plateau in growth ( e . g . see Figure 2A ) . Cellular filamentation in E . coli is indicative of DNA damage and induction of the SOS response [50] , which is consistent with a DNA target for type I-F systems . The detrimental effect of chromosomal targeting indicated that multiple mutational routes should lead to CRISPR/Cas avoidance . Such mutations have been predicted to include the cas genes , protospacer , PAM , and the repeats/crRNA processing [45] . We showed that mutation of the cas operon abrogates toxicity ( Figure 1 and Figure 2 ) . To test protospacer mutations , first the entire expI target gene ( including the protospacer ) was deleted from the chromosome . In this ΔexpI strain , the toxicity induced by a single expI spacer was abolished ( Figure 4A and 4B ) . Re-introduction of a single expI protospacer with an optimal 5′-protospacer-GG-3′ PAM restored toxicity , confirming that the protospacer enabled targeting ( Figure 4A and 4B ) . Previously , a ‘seed’ sequence of 8 nt in the spacer , adjacent to the 5′ handle , has been shown to be important for the initial binding of the crRNA to the target in the type I-E system [35] . In the type I-F system , the seed is less well defined , but a short ssDNA substrate of nucleotides 1–8 bound the Csy complex with highest affinity [38] . To test the role of the seed sequence , two seed sequence protospacer mutations were generated in the chromosome and tested for interference . A C3T mutation resulted in partial avoidance of targeting when measured by OD600 ( Figure 4B ) or by viable count ( an 100-fold reduction in viable count for the C3T PAM mutant compared with 104 to 105-fold reductions for WT protospacers ) . Next , we tested a C6T mutation , which did not affect targeting/toxicity , demonstrating that a level of mismatch is tolerated in the type I-F seed sequence ( Figure 4B ) . This result mirrors the tolerance observed at the identical position in the E . coli type I-E system [35] and is consistent with the existence of a discontinuous seed region in type I-F systems . PAM sequences are required for interference of plasmids and phages [34] , [35] . Evidence that the PAM is required in the type I-F systems was provided by the fact that re-introduction of an expI protospacer containing a single PAM nucleotide substitution mutation protected this strain from chromosomal targeting ( Figure 4B ) . This shows that a single mutation in the type I-F PAM is sufficient to escape targeting . Cady et al . recently identified G-1A PAM ( e . g . 5′-protospacer-AG-3′ ) phage escape mutants in the background of other protospacer mismatches [51] . Therefore , deletion or mutation of the protospacer target and PAM mutations can alleviate targeting and growth inhibition . Mutations in CRISPR repeats can inhibit pre-crRNA processing and crRNA generation and hence interfere with chromosomal targeting . Mutation ( s ) were introduced in both repeats flanking a single expI spacer ( Figure 4C ) . Firstly , a G20A mutation , that was previously shown to abrogate Cas6f ( Csy4 ) -dependent endonucleolytic cleavage [22] , abolished targeting ( Figure 4D ) . Next , we predicted that a C18A mutation would destabilise the crRNA stem-loop secondary structure , abolish processing and chromosomal targeting . Indeed , the C18A mutant was non-toxic ( Figure 4D ) . By introducing a compensatory mutation ( C18A/G8U ) , toxicity was restored , demonstrating that the repeat RNA stem-loop secondary structure was important but the sequence was not essential . Interestingly , a recent report showed the C18A/G8U mutation in P . aeruginosa subtly affected RNA binding and cleavage by Cas6f [52] . Apparently , this 2-fold reduced cleavage is sufficient for crRNA generation and toxicity ( Figure 4D ) . To test the involvement of Cas6f in crRNA generation , and hence toxicity , targeting in a Δcas6f mutant was assessed . We previously showed that deletion of cas6f abolished crRNA generation in P . atrosepticum [23] . As expected , chromosomal targeting was absent in the Δcas6f strain ( Figure 4D ) . In summary , particular mutations of repeats or the endoribonuclease provides protection from CRISPR/Cas-mediated chromosomal targeting . Spacer 6 in CRISPR2 has a 100% match to eca0560 in the P . atrosepticum genome within an ∼100 kb horizontally acquired island named HAI2 ( Figure 5A–5C ) [23] . The function of ECA0560 , a TraG-family protein , is unknown but it is highly conserved in Integrative Conjugative Elements ( ICE ) [53] , such as HAI2 , and is predicted to be involved in their mobility . HAI2 contains the cfa gene cluster involved in the biosynthesis of coronafacic acid , a polyketide phytotoxin important for plant pathogenicity in potato [54] . Since we demonstrated chromosomal targeting is toxic , we hypothesised that this spacer is non-toxic due to mutations that might interfere with the targeting mechanism . Clearly , the cas genes are functional , given our engineered assays ( see Figure 2 ) . However , the repeats adjacent to spacer 6 contained mutations ( Figure 5B ) and the PAM was not the type I-F consensus ( Figure 5C ) [33] . Firstly , we examined if the G1A and A13U repeat mutations affected targeting . The repeat mutations were “repaired” by cloning spacer 6 between two WT CRISPR1 consensus repeats . When expression of spacer 6 with consensus repeats was induced in the WT , no toxicity occurred ( Figure 5C and 5D ) . This suggested that the repeat mutations were not the cause of tolerance to this spacer . Indeed , in an in vitro assay , Cas6f cleaved pre-crRNA transcripts covering repeats either side of either spacer 2 ( control ) or spacer 6 ( Figure S3 ) , indicating that these mutations do not inhibit endonucleolytic processing to yield spacer 6 crRNAs . Together , these results show that the inability of this spacer to target the chromosome was not due to repeat mutations . Next , the role of the PAM was assessed . The protospacer had a non-consensus type I-F PAM of 5′-protospacer-TG-3′ ( Figure 5C ) compared with the consensus of 5′-protospacer-GG-3′ [33] ( Figure 5D ) . To test if the 5′-protospacer-TG-3′ PAM accounted for the lack of targeting , a single spacer was engineered against a 5′-protospacer-GG-3′ in eca0560 ( Figure 5E ) . This engineered spacer caused a toxic effect on P . atrosepticum when compared with the 5′-protospacer-TG-3′ non-consensus PAM and a no spacer control ( Figure 5D ) . This result is in agreement with our single nt mutation introduced in the PAM of the expI-targeting spacer , which abolished targeting ( Figure 4 ) and with two other recent studies [49] , [51] . In summary , these results show that targeting HAI2 is toxic to P . atrosepticum , but that a non-optimal PAM sequence present in the protospacer of CRISPR2 spacer 6 has allowed evasion from interference . Our demonstration that CRISPR/Cas systems can target host genomes and cause profound growth inhibitory effects that are avoided by a range of mutations led us to ask whether chromosomal targeting can drive genome evolution due to spontaneous target site deletion . Specifically , we tested if targeting of the HAI2 pathogenicity island could result in complete loss , or internal deletions within the island . We used a strain with a KmR cassette in eca0573 , a gene within HAI2 , which provided a marker to screen for island loss . Expression of the engineered crRNA targeting eca0560 ( in HAI2 ) led to growth inhibition ( e . g . Figure 5D ) , but when cultures were left for 36 h , suppressor mutants arose . Twenty isolates from 600 survivor colonies ( 12 independent experiments ) were sensitive to kanamycin , suggesting that loss of the eca0573 KmR marker had occurred . In control experiments with a non-targeting plasmid ( pBAD30 ) all 600 isolates retained kanamycin resistance . We assumed that the loss of KmR was due to specific targeting of HAI2 . However , it was possible that general DNA damage and stress , caused by chromosomal interference , promoted the loss of the kanamycin marker . When we targeted expI ( eca0105 ) elsewhere in the genome , instead of eca0560 , none of the 600 survivors had lost the kanamycin resistance marker in eca0573 , supporting that avoidance of specific targeting occurs by protospacer deletion and flanking DNA sequences . Next , we examined how the mutants had avoided CRISPR-targeting . HAI2 inserts into the P . atrosepticum genome by site-specific recombination between the attP ( plasmid ) site in circularised pHAI2 and the attB ( bacteria ) site in the phenylalanine tRNA gene in the chromosome . The resulting linearised form of HAI2 is flanked by attL ( left ) and attR ( right ) sites , which are composites of the original attP and attB sites . By using combinations of primers that assess the presence of attB , attP , attL , attR , eca0560 ( target gene ) and cas1 ( control ) , the presence or absence of the entire HAI2 or the target gene was determined ( Figure 6A ) . Two major classes of mutants were present within the 20 survivors ( see Figure 6B–D ) . There were 13 class I mutants that had lost the entire pathogenicity island ( ΔHAI2 ) ( Figure S4A ) , whereas 7 class II mutants were identified , which had lost kanamycin resistance and the target gene eca0560 , but retained attL and attR , suggesting an internal HAI2 mutation ( Figure S4B ) . The PCR result for a representative from both classes is shown in Figure 6B . The attB PCR product from multiple ΔHAI2 strains was sequenced ( Figure S4C ) , demonstrating that HAI2 had been lost by a precise excision event , but that the excised pHAI2 form ( i . e . attP ) had been eliminated from these strains ( Figure 6B and 6C ) . In WT strains there is a low frequency of HAI2 excision ( ∼10−6 ) [53] , which can be detected as faint attB and attP PCR products . However , when HAI2 is lost entirely , attB is strongly amplified and no attP product is detected , testament to the loss of this pathogenicity island . We also mapped the deletions in the class II mutants that retained a portion of the island ( attL and attR ) but lacked eca0560 and kanamycin resistance . By using extensive combinations of primer pairs for genes in different parts of HAI2 ( Figure S5 ) , we could determine which regions were still present and which were absent ( Figure 6E ) . These analyses demonstrated 5 different deletions amongst these 7 mutants . Mutant 10 has the largest deletion and has lost up to 78 genes including the coronafacic acid ( cfa ) cluster . Surprisingly , all 7 mutants retain the ability to generate excised derivatives of pHAI2 , as detected in an attP PCR , despite lacking many genes including several belonging to the syntenic core [55] ( Figure S4B ) . All mutants also lack the type IV pilus genes indicating that they are unlikely to be self-transmissible [56] . To map the deleted region precisely , primers on either end of the predicted deletion sites were used in PCR and the resulting products sequenced . In this manner , mutants 2 , 5 and 14 had the exact deletion junction sequenced . These mutants all contained a deletion from 596727 to 637003 within the published P . atrosepticum SCRI1043 sequence [54] , which corresponded to deletion within eca0522 to a site within eca0573 . Part of the KmR insertion ( 289 bp ) was still present , which included a 5′-TTGGCAC-3′ heptanucleotide sequence at the site of deletion that might have facilitated the recombination/DNA repair with eca0522 following crRNA interference of eca0560 ( Figure 6D and 6E ) . The resulting strains have deleted 40 , 277 bp , 51 entire genes and 2 partial genes . The question remained what the impact would be if non-mobile regions of the genome were targeted . To test targeting of non-mobile regions , a strain with a single WT expI protospacer immediately adjacent to a cat cassette was used ( RBV01 ) . This strain was targeted with the complementary expI crRNA ( pE1-16 ) and survivors were screened for loss of the linked chloramphenicol resistance gene . Of 624 survivors screened , 23% were CmS , indicating that deletion of cat and other genomic regions does occur to enable the evasion of chromosomal targeting . The higher frequency of CmS ( 23% ) compared with KmS ( 3% ) survivors in the expI and eca0560 targeting experiments might be a result of the linkage distance between the markers and the target site ( immediately adjacent for expI and ∼10 kb for eca0560 ) . To test a markerless system , the lacZ gene ( eca1490 ) was targeted with a single crRNA ( pL1-16 ) . Ten survivors that were white on X-gal plates were analysed by PCR in more detail ( Figure S6 ) . All mutants retained cas1 ( eca3679; control ) , yet had lost the lacZ gene and >50 kb of chromosomal sequence 5′ of lacZ including lacY ( eca1489 ) , two large non-ribosomal peptide synthetases ( NRPS ) ( eca1488 and eca1487 ) and genes eca1486-eca1482 ( Figure S6 ) . The extent of these deletions was not characterized , but it is apparent that targeting other regions of the chromosome can also result in large changes in genomic content . The role of this NRPS region is unknown , but has similarities to others that produce secondary metabolites required for pathogenicity . In summary , the toxic effect elicited by CRISPR/Cas-mediated chromosomal targeting provides a strong selective pressure for the loss of the protospacer target . This can result in large-scale genomic changes which include precise excision and loss of pathogenicity islands , their modification , or deletion of other regions of the chromosome . Therefore , CRISPR systems are likely to have played an additional and greater role in the evolution of bacterial genomes than previously thought . In this study we set out to examine the effects of chromosomal targeting by CRISPR/Cas systems . A bioinformatic approach to this question had previously led to the suggestion that incorporation of spacers that match the chromosome is accidental , resulting in a detrimental interference effect and the selection for mutants that have inactivated targeting [45] . We showed directly that CRISPR/Cas targeting of the chromosome is toxic and that mutations that disrupt the CRISPR/Cas mechanism enable cell survival . Importantly , we demonstrate that the negative fitness cost associated with chromosomal targeting can provide a strong selective advantage for strains lacking the target DNA . This selective pressure can result in large-scale genomic changes , including the deletion and remodelling of pathogenicity islands and hence , CRISPR/Cas can influence bacterial genome evolution that may lead to changes in virulence . Furthermore , this strong selection provides a tool for the deletion of targeted regions of bacterial genomes . A novel method was developed to generate tightly-controlled crRNA expression vectors for cloning of multiple spacer sequences ( Figure 1 ) . By using these vectors to express crRNAs against three independent non-essential host genes in P . atrosepticum , we provide direct experimental evidence that chromosomal targeting by CRISPR/Cas systems is highly detrimental to bacterial growth and viability . Toxicity is a CRISPR/Cas-dependent process since deletion of the cas operon , the absence of the targeting crRNA or the presence of non-targeting ( scrambled ) spacers all led to a non-toxic effect ( Figure 1 , Figure 2 , Figure 3 ) . Expression of chromosome-targeting crRNAs resulted in a non-reversible ∼105 reduction in viable counts ( Figure 2H ) , but some cells continued to grow and elongate ( Figure 3 ) . The filamentation phenotype was reminiscent of defects caused during the SOS response triggered by DNA damage [50] , in accordance with DNA as the target of type I-F systems . Although there is no direct evidence , DNA is the likely target for type I-F CRISPR/Cas systems . For example , in vitro , the P . aeruginosa type I-F ribonucleoprotein Csy complex bound DNA that was complementary to the crRNA spacer [38] . However , in another study it was proposed that RNA is targeted [57] , but recent work by the same group demonstrated plasmid and phage interference [51] , consistent with a DNA target . The spacers in our study were designed sense to the mRNA of non-essential genes and therefore , could only base pair with the template strand of DNA . Therefore , the toxic and cell elongation phenotypes observed upon chromosomal targeting provide further evidence that DNA is targeted by type I-F systems . Our molecular and genetic analyses have directly tested the bioinformatic hypotheses put forth by Stern et al . ( 2010 ) that chromosomal targeting is detrimental rather than regulatory . This model predicts that chromosomal targeting results from the accidental incorporation of host DNA as spacers and only cells that acquire mutations that deactivate targeting will survive . We unambiguously show that mutation of the cas genes , PAM , protospacer and repeats alleviate the severe fitness cost associated with self-targeting . However , it is tempting to speculate that there could be evolutionary advantages of self-targeting . In particular environments , chromosomal targeting might result in enhanced adaptive plasticity by enabling a subpopulation of cells to remove or remodel genomic regions associated with reduced fitness . Our experiments have also provided insight into mechanistic details of the type I-F systems . For example , we demonstrated a role for the protospacer , specifically the seed sequence , and the PAM in interference and that single nucleotide mutations in the seed sequence can have varying effects on targeting by the type I-F system ( Figure 4 ) . Interference is typically viewed as an all-or-none phenomenon , and in the type I-E system , a single C3A protospacer mutation enabled phage M13 infection [35] . We showed a C3T mutation led to partial interference , which either highlights differences between type I-E and I-F systems , or might indicate that intermediate effects exist , depending on the mismatched nucleotide or the greater sequence context . An intermediate phage resistance was reported with multiple protospacer mutations in the type I-F system of P . aeruginosa [51] . The role of the repeats and the Cas6f endoribonuclease on chromosomal targeting was also determined . Mutation of cas6f , a CRISPR repeat C20A mutation , known to abolish processing [22] , or a C18A repeat mutation all led to protection from self-targeting . In contrast , single repeat mutations G1A and A13U or a C18A/G8U double mutant were all functional , confirming a degree of flexibility of Cas6f proteins for their substrate RNA [52] . These results highlight that a more detailed understanding of specific endoribonucleases is required before assuming that defective processing will result from repeat mutations [45] . Bioinformatic studies have provided some insight into the effects of chromosomal targeting . Some E . coli and Salmonella strains contain type I-F arrays , but lack the type I-F cas genes . In these cases , spacers with homology to the type I-F genes are present in the orphan arrays and two non-mutually exclusive theories have been suggested . One theory is that self-cas targeting and subsequent deletion of the cas genes occurred [58] , which is supported by our data and might help explain why roughly half of bacteria lack CRISPR/Cas . Alternatively , these orphan arrays might target and inhibit acquisition of plasmids encoding type I-F cas genes [59] , [60] . An anti-CRISPR/Cas role was also proposed in archaea , where multiple spacers matched ORFs in other genomes and a subset had similarity to cas genes from other CRISPR/Cas systems [61] . However , it is also feasible that some of these ORFs were previously chromosomal and have been eliminated by CRISPR/Cas genomic targeting . A small number of experimental studies have suggested chromosomal targeting causes a toxic effect . In Pelobacter carbinolicus , one spacer in the type I-E system matches the hisS gene [42] , which contains the consensus 5′-protospacer-CTT-3′ PAM . Transformation with a plasmid containing an anti-hisS spacer into a G . sulfurreducens strain , which contained the P . carbinolicus hisS gene , was reduced compared with a vector control [42] . Also in a type I-E system , expression of engineered crRNAs targeting a lambda prophage in E . coli led to an ∼2 log reduction in viable count compared with control spacers [43] and has been used as a positive selection for CRISPR-inactive mutants [62] . In Sulfolobus solfataricus , transfection with arrays targeting the host β-galactosidase gene caused a growth reduction and resulted in elimination of the artificial arrays by recombination with the native CRISPRs [44] . In these studies , the data indicates that targeting host ( or integrated; e . g . hisS and lambda ) genes could be toxic . Our research has extended these studies in a number of ways . Most significantly , we investigated the mechanism of toxicity and show it is a highly specific CRISPR/Cas-dependent process that requires Cas proteins , a PAM , particular complementarity between the protospacer and spacer in the crRNA . Furthermore , we show the toxicity can result in cellular filamentation , indicative of DNA damage , which ultimately can result in genome evolution in bacterial survivors . For chromosomal interference to cause mutations and genome evolution , DNA must be targeted and result in toxicity . Therefore , our results are relevant for the majority of CRISPR/Cas systems , most of which are thought to target DNA . However , one subtype , the type III-B ( Cmr ) system , targets RNA [24]–[26] and as such is unlikely to cause toxicity/mutation , except when essential genes are targeted . The DNA damage-induced SOS response can cause higher mutation rates [63] , which raises the possibility of further mutational side effects from CRISPR/Cas chromosomal targeting . The SOS response also triggers induction of many prophages , such as lambda [64] , and it is interesting to speculate that chromosomal targeting might also cause phage-mediated bacterial cell suicide . The negative impact of acquiring spacers from the chromosome suggests that mechanisms exist to avoid this . Indeed , we have shown mutations can result in tolerance to chromosomal spacers; however , this is not an efficient system and would lead to defective CRISPR/Cas systems at a high rate . In fact , this may provide some explanation as to why only half of all bacteria surveyed contain CRISPR/Cas . How do bacteria control the accidental incorporation of detrimental chromosomally-derived spacers ? In experiments with the E . coli type I-E system , acquisition of chromosomally-derived spacers was generally PAM-dependent , but was rare relative to acquisition from a plasmid when adjusted for the number of possible PAM targets [30] . The mechanism for reducing spacer acquisition from the chromosome is unknown but is a critical question that remains to be addressed [11] . What are the mechanisms for CRISPR/Cas-mediated generation of precise ( i . e . class I ) and imprecise ( i . e . class II ) chromosomal deletions ? For generation of class I mutants ( i . e . ΔHAI2 ) we propose two non-mutually exclusive models; 1 ) as a selector of naturally occurring deletions and 2 ) as a sequence-specific mutator . Firstly , in the rare HAI2-excising cells ( 10−6 ) , attB is formed via site-specific recombination and the excised pHAI2 will be subject to degradation due to the presence of the CRISPR/Cas target site , yielding ΔHAI2 strains . Secondly , CRISPR/Cas has a mutator role through chromosomal DNA damage ( observed toxicity , viable counts and cell elongation ) , likely caused by the nuclease activity of Cas3 [65]–[67] . A double strand break ( DSB ) in the island , induced by CRISPR/Cas , might be repaired by precise island removal via integrase and/or excisionase-dependent site-specific recombination between attL and attR , generating attB ( a ΔHAI2 strain ) . A linear DNA fragment would be released with attP flanked by any remaining DNA from either end of the island ( following CRISPR degradation ) . This DNA would be unstable and lost upon cell division or through degradation by nucleases . Based on our data we hypothesise a DNA degradation and alternative end joining ( A-EJ ) model for generation of the imprecise chromosomal deletions ( i . e . class II ) . CRISPR/Cas targeting leads to Cas3-mediated DNA damage [65]–[67] . The Cas3 nuclease activity and/or end resection by RecBCD [68] , leads to extensive degradation of host DNA . In mutants 2 , 5 and 14 , a region of 7 nt homology was identified at the repair site , which is too short for homologous recombination . However , A-EJ repairs DSBs by using regions of microhomology ( 1–9 nt ) [68] and as such , could provide a model for the partial mutants . We have developed a tool for the precise removal of genomic islands ( and potentially other genomic regions ) for analysis of their function in the host bacterium . During the revision of this manuscript , the type II Cas9 system was also shown to enable editing of bacterial genomes [69] . The HAI2 island that was modified in this study shares similarity in structure and sequence to the SPI-7 pathogenicity island of Salmonella enterica serovar Typhi [54] , [70] , but also contains important plant pathogenicity determinants for P . atrosepticum [54] . HAI2 is therefore a good example of the mosaic nature that is common in genomic islands [71] . Indeed , all class II mutants had deleted major portions of HAI2 , yet retained the ability to excise and circularise ( Figure 6 ) , suggesting that CRISPR/Cas could be an important factor driving the mosaicism and evolution of genomic islands . A recent study in Streptococcus agalactiae demonstrated that CRISPR can inhibit acquisition of an ICE introduced by conjugation [72] . Our study shows that ICEs can be deleted or modified after their acquisition and hence , CRISPR/Cas might act at later stages to remove mobile elements from bacterial genomes . We hypothesise that upon exposure to HAI2 in the past , P . atrosepticum acquired a spacer targeting this ICE , but due to its integration in the chromosome it only survived due to a single nt mutation in the PAM that alleviates targeting . Other selective pressures might have existed to retain HAI2 , such as the coronofacic acid pathogenicity determinants present on the island and the existence of a pemIK toxin-antitoxin system . The plant host can also influence the loss or transfer and selection for genomic islands , as exemplified by the in planta dynamics of the PPHGI-1 island in Pseudomonas syringae pv . phaseolicola [73] , [74] . HAIs are important in the evolution of bacteria , due to their transfer of virulence genes or other ecologically important traits [75] and are considered part of the accessory genome , which can constitute 10% of the entire chromosome [76] . A simple method for positive selection of their deletion promises advances into the study of their functional roles . In conclusion , we demonstrated that chromosomal targeting is highly detrimental to bacterial growth and that mutations that inactivate the CRISPR/Cas systems allow survival . The negative impact of this targeting resulted in the selection of strains containing dramatic mutations in the target site , such as those deleted for part of ( e . g . ∼40 kb ) , or an entire ( ∼100 kb ) pathogenicity island . Therefore , CRISPR/Cas systems can play a significant role in the evolution of bacterial genomes that may influence pathogenicity . We propose that chromosomal targeting has resulted in widespread changes to bacterial genomes , a prediction prompting further bioinformatic studies . All strains and plasmids used in this study are given in Table S1 and Table S2 , respectively . P . atrosepticum SCRI1043 [54] was grown at 25°C and E . coli at 37°C in Luria Broth ( LB ) at 180 rpm or on LB-agar plates containing 1 . 5% ( w/v ) agar . When required , medium was supplemented with the following: ampicillin ( Ap; 100 µg/ml ) , chloramphenicol ( Cm; 25 µg/ml ) , kanamycin ( Km; 50 µg/ml ) , tetracycline ( Tc; 10 µg/ml ) , D-glucose ( 0 . 2% w/v ) and L-arabinose ( 0 . 1% w/v ) . Bacterial growth was measured in a Jenway 6300 spectrophotometer at 600 nm ( OD600 ) . All experiments were repeated in at least three biological replicates . All oligonucleotides are from Invitrogen or IDT and are listed in Table S4 . All strains and plasmids were confirmed by PCR and DNA sequencing was performed at the Allan Wilson Centre , NZ . Plasmid DNA was prepared using Zyppy Plasmid Miniprep Kits ( Zymo Research ) according to manufacturer's guidelines . DNA from PCR and agarose gels was purified using the GE Healthcare Illustra GFX PCR DNA and Gel Band Purification Kit . Restriction enzymes and T4 ligase were from Roche or NEB . Plasmids were designed that would enable the cloning of any spacer and repeat sequence into an artificial CRISPR1 array that contains one repeat , yielding a CRISPR with two repeats and one spacer . The enzyme BbsI was chosen because it cuts at a distance from its recognition site and hence , it was used to cut in the repeat and could be used multiple times to introduce subsequent spacer-repeat sequences ( Figure 1 ) . Firstly , either 780 , 180 , 52 or 16 bp of the CRISPR1 leader and first repeat region was amplified with forward primers AM05 , TGO9 , RVO1 or RVO2 and the reverse primer TGO10 , the products were digested with XmaI and SalI and ligated to pBAD30 previously digested with the same enzymes . The resulting CRISPR spacer entry plasmids were named pC1-780 , pC1-180 , pC1-52 and pC1-16 ( Table S3 ) . These plasmids were then digested with BbsI and products derived from the primer pairs that contained engineered spacer-repeat units ( Table S1 and Table S4 ) were cloned into this site to incorporate one exact new spacer and repeat . This procedure was repeated up to three times , but could in theory be performed indefinitely to construct artificial arrays . Electrocompetent P . atrosepticum cells were prepared as follows . Ten ml P . atrosepticum cultures were grown overnight in LB and used to inoculate flasks of LB , which were grown until the OD600 reached 0 . 4–0 . 6 . The cells were pelleted by centrifugation at 4°C at 2219×g , resuspended in ice-cold H2O and centrifuged as above . The H2O wash was repeated and then cells were washed using ice-cold 10% glycerol . Cells were resuspended in 10% glycerol , divided into 50 µl aliquots and stored at −80°C . For transformations , DNA was adjusted to working concentrations of 50 ng/µl ±5 ng/µl and 50 ng was added to 50 µl of competent P . atrosepticum cells . The cells and DNA were incubated on ice for 10 min then electroporated ( 1 mm electro-cuvettes , 1800 V , capacitance 25 µF and resistance 200 ohms ) . Bacteria were recovered in 1 ml LB for 2 h at 25°C and then plated on LB containing the appropriate supplements and grown at 25°C . Transformation efficiency was calculated as transformants/ng of DNA and normalised to control plasmids . Strains containing inducible pBAD30-based plasmids expressing single spacers from CRISPR1 arrays that targeted lacZ , expI or eca0560 were used to inoculate 10 ml LB containing 0 . 2% glucose and Ap and incubated overnight . These cultures were used to inoculate individual 25 ml cultures of LB with either 0 . 2% glucose or 0 . 2% arabinose ( in 250 ml flasks ) at a starting OD600 of 0 . 05 . Cultures were grown at 25°C with shaking at 180 rpm and growth ( OD600 ) was measured . Where indicated , the viable count was determined by measuring the cfu/ml . Briefly , 1 ml samples of culture were taken , serially diluted and 10 µl of each dilution was plated onto LB agar containing Ap and 0 . 2% glucose . The plates were incubated at 25°C and colonies were counted . A plasmid for the construction of an expI deletion mutant was generated as follows . Firstly , approximately 500 bp of sequence 5′ and 3′ of expI was amplified by PCR using primer pairs PF314 and PF315 and PF316 and PF317 . Secondly , these products were used as template in an overlap extension PCR with primers PF314 and PF317 . This product was digested with KpnI and XbaI and cloned into pBluescriptII SK+ cut with the same enzymes , yielding plasmid pTA163 . Next , the cat gene and promoter were amplified by PCR using primers TGO74 and TGO75 with pACYC184 as template . This product was digested with HindIII and SalI and ligated with pTA163 cut with the same enzymes , giving plasmid pTA164 . Finally , the 5′ 500 bp-cat-500 bp 3′ fragment was subcloned from pTA164 into pKNG101 on a BamHI/XbaI fragment , yielding plasmid pTA165 . E . coli CC118λpir carrying pTA165 was used in an allelic exchange protocol as previously described [77] , [78] via triparental mating with the helper E . coli strain HH26 , pNJ5000 and P . atrosepticum as the recipient , resulting in strain PCF81 ( ΔexpI::cat ) . Engineered variant protospacers were introduced into P . atrosepticum as described below . Plasmid pTA164 was used as the template in a PCR with RV19 and PF317 to amplify a 1 . 5 kb fragment containing the cat gene and the 500 bp 3′ expI flanking sequence . RV19 contained the protospacer sequence for the expI-1 spacer with a 5′-protospacer-GG-3′ PAM . The resulting PCR fragment was cloned using XhoI and XbaI restriction sites into pTA164 to make pRX38 . To generate other protospacer variants , forward primers that introduced the expI-1 protospacer sequence with a 5′-protospacer-TG-3′ PAM ( RV20 ) , a C6T mutation ( RV21 ) or a C3T mutation ( RV29 ) were used in PCRs with PF317 and pRX38 as template . The resulting PCR fragments were cloned using XhoI and XbaI restriction sites into pTA164 to make plasmids pRX39 ( 5′-protospacer-TG-3′ PAM ) , pRX40 ( C6T ) and pRX34 ( C3T ) . Finally , the 2 kb BamHI/XbaI 5′ 500 bp-cat-500 bp 3′ fragments containing the variant protospacers from pRX38 , pRX39 , pRX40 and pRX34 were cloned into pKNG101 to make pRX41 , pRX42 , pRX43 and pRX35 , respectively . Allelic exchange was performed with these suicide plasmids , as described above for PCF81 , and resulted in strains RBV01 ( GG PAM ) , RBV02 ( 5′-protospacer-TG-3′ PAM ) , RBV03 ( C6T ) and RBV04 ( C3T ) . The LIVE/DEAD BacLight Bacterial Viability Kit ( Invitrogen ) was used to stain cells , which were then observed with an Olympus BX51 microscope . Samples were pelleted at 13000 rpm for 5 min then resuspended in distilled water and pelleted again to wash away the growth media at least once and finally resuspended in 500 µl of distilled water . One hundred µl of washed culture ( approximately 1×107 cells ) was taken and stained with 1 . 5 µl of the flurophore SYTO9 ( green fluorescent , 480–500 nm ) , and 1 . 5 µl of propidium iodide ( PI ) ( red fluorescent , 490–635 nm ) . Cells were stained for 20 min in the dark , 8 µl samples were mounted on slides and imaged under the oil immersion 100× lens . For TEM , carbon-coated copper grids were prepared using standard methods as previously described [79] . Grids were analysed using the Phillips CM100 BioTWIN transmission electron microscope . Cell measurements were performed on TEM images of at least 60 randomly selected cells from each treatment . The CRISPR1 repeat mutations , G20A , C18A and C18A/G8U were introduced into the sequence of the pE1-16 array using primer pairs RV22 and RV25 , RV23 and RV26 and RV24 and RV27 , respectively . Products were amplified by PCR with 8% DMSO due to the CRISPR repeat secondary structures in primers RV22 , RV23 and RV24 . These products were cloned into pBAD30 with XmaI and BbsI making the plasmids , pE1-16 G20A , pE1-16 C18A , and pE1-16 C18A/G8U , respectively . A KmR transposon mutation in eca0573 on HAI2 ( JTC101 ) was isolated in an independent random mutagenesis ( Chang and Fineran; unpublished data ) and provided a marker to screen for island loss . Strains of JTC101 with either a vector control ( pBAD30 ) or plasmids targeting expI ( pE1-16 ) or eca0560 ( pTraG1-16 ) were grown in 10 ml LB with Ap and 0 . 2% arabinose for 36 h with Ap added to the media at 12 h intervals to ensure plasmid maintenance . By plating on LBA with Ap and 0 . 2% arabinose we isolated approximately 600 colonies from at least 12 independent cultures per strain that were resistant to CRISPR/Cas-associated HAI2 targeting . These survivors were patched onto LBA plates containing Ap and 0 . 2% arabinose with or without Km . Mutants that failed to grow on the plates containing Km were isolated and subjected to PCR to check for loss of the transposon and to ascertain if parts of HAI2 were lost due to CRISPR/Cas targeting . The same approach was used for targeting of expI using plasmid pE1-16 in strain RBV01 , which contains a single expI protospacer immediately adjacet to a CmR cassette . In addition , lacZ was targeted by plasmid pL1-16 in the WT background . The primers used are listed on Table S4 and details are given in the results .	Bacteria have evolved mechanisms that provide protection from continual invasion by viruses and other foreign elements . Resistance systems , known as CRISPR/Cas , were recently discovered and equip bacteria and archaea with an “adaptive immune system . ” This adaptive immunity provides a highly evolvable sequence-specific small RNA–based memory of past invasions by viruses and foreign genetic elements . There are many cases where these systems appear to target regions within the bacterial host's own genome ( a possible autoimmunity ) , but the evolutionary rationale for this is unclear . Here , we demonstrate that CRISPR/Cas targeting of the host chromosome is highly toxic but that cells survive through mutations that alleviate the immune mechanism . We have used this phenotype to gain insight into how these systems function and show that large changes in the bacterial genome can occur . For example , targeting of a chromosomal pathogenicity island , important for virulence of the potato pathogen Pectobacterium atrosepticum , resulted in deletion of the island , which constituted ∼2% of the bacterial genome . These results have broad significance for the role of CRISPR/Cas systems and their impact on the evolution of bacterial genomes and virulence . In addition , this study demonstrates their potential as a tool for the targeted deletion of specific regions of bacterial chromosomes .	[ "Abstract", "Introduction", "Results", "Discussion", "Materials", "and", "Methods" ]	[ "bacteriology", "bacterial", "physiology", "microbial", "mutation", "microbial", "physiology", "microbial", "evolution", "microbial", "pathogens", "biology", "microbiology", "bacterial", "evolution" ]	2013	Cytotoxic Chromosomal Targeting by CRISPR/Cas Systems Can Reshape Bacterial Genomes and Expel or Remodel Pathogenicity Islands
Bardet-Biedl Syndrome ( BBS ) is a heterogeneous syndromic form of retinal degeneration . We have identified a novel transcript of a known BBS gene , BBS3 ( ARL6 ) , which includes an additional exon . This transcript , BBS3L , is evolutionally conserved and is expressed predominantly in the eye , suggesting a specialized role in vision . Using antisense oligonucleotide knockdown in zebrafish , we previously demonstrated that bbs3 knockdown results in the cardinal features of BBS in zebrafish , including defects to the ciliated Kupffer's Vesicle and delayed retrograde melanosome transport . Unlike bbs3 , knockdown of bbs3L does not result in Kupffer's Vesicle or melanosome transport defects , rather its knockdown leads to impaired visual function and mislocalization of the photopigment green cone opsin . Moreover , BBS3L RNA , but not BBS3 RNA , is sufficient to rescue both the vision defect as well as green opsin localization in the zebrafish retina . In order to demonstrate a role for Bbs3L function in the mammalian eye , we generated a Bbs3L-null mouse that presents with disruption of the normal photoreceptor architecture . Bbs3L-null mice lack key features of previously published Bbs-null mice , including obesity . These data demonstrate that the BBS3L transcript is required for proper retinal function and organization . Visual impairment and blindness have far reaching implications for society . Hundreds of individually rare , but collectively common Mendelian disorders can cause blindness . One of these disorders is a heterogeneous syndromic form of retinal degeneration , Bardet-Biedl Syndrome ( BBS , OMIM 209900 ) . This pleiotropic disorder is characterized by retinal degeneration , obesity , polydactyly , renal abnormalities , hypogenitalism and cognitive impairment [1]–[4] . Additionally , BBS is associated with an increased incidence of hypertension , diabetes mellitus and heart defects [1] , [2] , [5] . Although there is variability in the ocular phenotype between individuals , BBS patients typically present with early and progressive photoreceptor degeneration , leading to both central and peripheral vision loss by the third decade of life [1] , [6]–[13] . To date , 14 genes ( BBS1-14 ) have been implicated in BBS [14]–[28] . Analysis of mouse models of BBS ( Bbs1M390R/M390R , Bbs2−/− , Bbs4−/− and Bbs6−/− ) reveals that these mice have major components of the human phenotype including retinal degeneration , obesity , renal cysts and neurological deficits [29]–[32] . Multiple lines of evidence suggest that BBS phenotypes involve cilia dysfunction in a range of tissues , including the retina . The vertebrate retina contains photoreceptors , highly polarized cells with a modified cilium ( connecting cilium ) that joins the photosensitive outer segment ( OS ) to the protein synthesizing inner segment ( IS ) . The connecting cilium transports cellular components from the IS to the OS that are necessary for the structure and function of the OS [33] , [34] . Intraflagellar transport ( IFT ) proteins are important in this intraphotoreceptor transport process as they play a key role in both assembly and maintenance of photoreceptor cells [35]–[37] . Loss of IFT genes in vertebrates leads to abnormal OS development , retinal degeneration and mislocalization of photopigments [36] , [38] , [39] . Retina phenotypes observed in Bbs-null mice are similar to those seen with loss of IFT genes , indicating that BBS proteins play a role in transporting proteins through the connecting cilium into the OS of the photoreceptor . For instance , characterization of the retinal phenotype in the mouse model has shown that photoreceptor death is preceded by mislocalization of rhodopsin [29]–[32] , [40] . Recent work with two independently generated Bbs4-null mice indicates that Bbs4 proteins play an important role in establishing both correct structure as well as proper transport of phototransduction proteins [40] , [41] . In the zebrafish model system , individual knockdown of bbs genes results in defects in the ciliated Kupffer's Vesicle ( KV ) and delayed retrograde transport within the melanosome [26] , [42] , [43] . Moreover , work in Caenorhabditis elegans has shown that bbs1 , bbs3 , bbs5 , bbs7 and bbs8 localize to the basal body of ciliated cells and are involved in IFT [23] , [44] . Taken together , these data strongly support a role for BBS proteins in intracellular transport and cilia; thus further substantiating a critical role for BBS genes in the specialized connecting cilium of the photoreceptor for cell maintenance and function . BBS3 is a member of the Ras superfamily of small GTP-binding proteins , which is subdivided into ADP-ribosylation factor ( ARF ) and ARF-like ( ARL ) subgroups [45] . The precise function of ARL proteins is unknown , but it has been proposed that they play a role in membrane and/or vesicular trafficking [45] . Work in C . elegans indicates that ARL6 is specifically expressed in ciliated cells and undergoes IFT along the ciliary axoneme [17] . Here we report the identification of a second transcript of BBS3 , BBS3L , and determine that the BBS3L protein product plays an important role in eye structure and function . BBS3L is evolutionally conserved and is unique among BBS gene products as it is expressed predominantly in the eye , suggesting a specialized role in vision . We have established both mouse and zebrafish models to study the function of BBS3L , and determined that BBS3L is specifically required for retinal organization and function . Expressed sequence tag ( EST ) data for human BBS3 was compared to the known coding region of the gene . Although most of the ESTs were virtually identical to the BBS3 reference sequence , a few were found to contain 13 extra base pairs . Interestingly , all ESTs that contained this alternative sequence originated from retina or whole eye libraries , suggesting that this second longer transcript , BBS3L , has an expression pattern that is limited to the eye . BBS3L results from differential splicing that leads to the inclusion of a 13 base pair exon and a shift in the open reading frame generating different C-terminal regions ( Figure 1A ) . The striking conservation of the C-terminal region of the long isoform in human , mouse and zebrafish strongly suggests that bbs3L has functional relevance ( Figure 1B ) . To determine if the bbs3 and bbs3L transcripts have similar tissue-specific expression in other species , RT-PCR of zebrafish and mouse tissues was performed . Zebrafish bbs3 is expressed in all adult tissues examined , while bbs3L expression is limited to the eye ( Figure 1C ) . Similar tissue expression patterns for Bbs3 and Bbs3L were seen in the mouse with the addition of low levels of Bbs3L mRNA expression in the brain ( Figure S1 ) . A developmental profile in zebrafish embryos reveals that while bbs3 is expressed throughout development , bbs3L is not expressed until 48 hours post fertilization ( hpf ) . This is a time when retinal neuroepithelial cells are exiting the cell cycle and differentiating into photoreceptor cells , the light sensing cells of the retina [46] ( Figure 1D ) . To determine the functional role of bbs3L in development and to distinguish the individual roles of the two bbs3 protein products , we utilized antisense oligonucleotide mediated gene knockdown ( morpholinos , MO ) in zebrafish . Two independent MOs were utilized: one targeting the splice junction specific to the long transcript ( bbs3 long MO ) and the other a previously described MO targeting both transcripts ( bbs3 aug MO ) through blocking of the translational start site [43] ( Figure 2A ) . RT-PCR was used to determine the knockdown efficiency of the bbs3 long MO on staged embryos and demonstrated knockdown of the long transcript through at least 5 days post fertilization ( dpf ) ( Figure 2B ) . Knockdown of bbs function in zebrafish generates two prototypical defects: reduction of the size of the Kupffer's vesicle ( KV ) as well as retrograde transport defects [26] , [42] , [43] . As previously demonstrated , alterations in the formation of the ciliated KV was the earliest observable phenotype resulting from knockdown of both bbs3 transcripts by the bbs3 aug MO [43] . At the 8–10 somite stage ( 12–14 hpf ) in wild-type and control injected embryos the KV has formed in the posterior tailbud . The KV diameter is approximately 50 µm and is larger than the width of the notochord ( Figure 2C and 2D ) . Injection of the bbs3 aug MO resulted in a reduction of KV size to a width less than that of the notochord ( Figure 2E ) . Knockdown of both bbs3 transcripts by the aug MO results in a statistically significant increase in embryos with KV defects ( Fisher's exact test , p<0 . 001 ) ( Figure 2F ) . Of note , injection of the bbs3 long MO does not lead to KV defects ( Figure 2F ) . The second prototypical phenotype observed in bbs MO-injected embryos ( morphants ) is delayed trafficking of melanosomes . Zebrafish are able to adapt to their surroundings through intracellular trafficking of melanosomes within melanophores in response to light and hormonal stimuli [47]–[50] . To test the rate of this movement , 5-day old zebrafish were dark adapted , to maximally disperse the melanosomes ( Figure 2G and 2H ) and then treated with epinephrine to chemically stimulate the retrograde transport of melanosomes [42] , [43] , [51] ( Figure 2I ) . Wild-type and control injected embryos show rapid movement of melanosomes to a perinuclear location averaging 1 . 4 minutes , whereas bbs3 aug MO injected embryos demonstrated a statistically significant delay averaging 2 . 1 minutes ( ANOVA with Tukey , p<0 . 01 ) ( Figure 2J ) . In contrast , the rate of melanosome movement in bbs3 long knockdown embryos is statistically the same as control embryos , averaging 1 . 5 minutes ( Figure 2J ) . To test for MO-specificity as well as differential function of the two bbs3 transcripts , human RNAs of both BBS3 and BBS3L were used . Embryos co-injected with a combination of MO and RNA were evaluated for suppression of MO induced KV and melanosome transport defects . Co-injection of BBS3 RNA with the aug MO did not rescue the KV defect but was sufficient to suppress the melanosome transport delay; however , co-injection of the aug MO with BBS3L RNA was not able to suppress either MO-induced defect ( Table S1 ) . Myc-tagged BBS3 and BBS3L RNA injection and Western blot analysis confirmed expression of the protein out to 5 dpf ( data not shown ) . Taken together , our data demonstrates that bbs3L plays a role independent from KV formation and melanosome transport and that human BBS3 can partially compensate for the loss of zebrafish bbs3 . Since BBS patients develop retinitis pigmentosa and the bbs3L transcript is differentially expressed in the eye , we sought to functionally test the role of bbs3 in vision . The zebrafish retina develops rapidly; at 60 hpf the retina is fully laminated and by 3 dpf zebrafish larvae are visually responsive [52]–[54] . Zebrafish elicit a characteristic escape response when exposed to rapid changes in light intensity and this startle response can be used as an assay for vision function [53] . In this assay , the behavior of a 5-day old larvae was monitored in response to short blocks of a bright , stable light source [53] ( t = 0 , Figure 3A ) . The typical response , a distinct C-bend and a change in swimming direction , is scored over a series of 5 trials , timed 30 seconds apart ( t = 139ms , Figure 3A ) . Uninjected embryos respond on average 3 . 09 times ( Figure 3B , Table 1 and Video S1 ) . Cone-rod homeobox ( crx ) gene knockdown was used as a control for vision impairment as loss of this gene is known to affect photoreceptor formation in zebrafish [55] , [56] . crx knockdown embryos respond an average of 1 . 28 times ( ANOVA with Tukey , p<0 . 01 ) . Knockdown using either the bbs3 aug or bbs3 long MO resulted in a statistically significant ( ANOVA with Tukey , p<0 . 01 ) reduction in the number of responses ( 1 . 81 and 1 . 77 times respectively ) compared to controls , indicating vision impairment ( Figure 3B , Table 1 and Video S2 ) . These data support a key role for bbs3L in vision function . To functionally test the specific role of both bbs3 and bbs3L in vision , rescue experiments were performed . To investigate whether bbs3 could compensate for loss of bbs3L , wild-type human BBS3 RNA was co-injected with the bbs3 long MO . Although BBS3 RNA was sufficient to suppress the melanosome transport delays associated with bbs3 aug morphant embryos ( Table S1 ) , BBS3 RNA was insufficient to rescue the vision impairment induced by loss of only bbs3L ( Figure 3B and Table 1 ) . Conversely , co-injection of BBS3L RNA with the bbs3 aug MO was sufficient to rescue the vision defect ( ANOVA with Tukey , p<0 . 01 ) ( Figure 3B and Table 1 ) . Based on these rescue experiments , bbs3L is necessary and sufficient for vision function . Previous work has demonstrated that Bbs1 M390R knockin , Bbs2 , Bbs4 and Bbs6 mutant mice initially form photoreceptors; however , the photoreceptors subsequently show a mislocalization of rhodopsin , a photopigment protein , to the cell bodies of the outer nuclear layer ( ONL ) and undergo progressive photoreceptor degeneration [29]–[32] , [40] . By gross histology , wild-type , bbs3 aug and bbs3 long morphant zebrafish embryo retinas displayed a fully laminated retina at 5 dpf ( data not shown ) . Ganglion cell outgrowth and optic nerve formation was evaluated using the ath5:GFP [Tg ( atoh7:GFP ) ] transgenic line , a marker of ganglion cell and axon outgrowth [57] . We found that gross retinal ganglion axon trajectories were not perturbed in bbs3 aug or long morphants ( data not shown ) . While the overall architecture of the retina appeared morphologically normal at 5 dpf , we investigated photopigment localization in bbs3 morphants . Photopigments are known to localize to the outer segment of the zebrafish photoreceptor; therefore , we assessed opsin localization using an antibody specific to green cone opsin [58] . In the wild-type retina , green opsin is found in the outer-segment of the green cone photoreceptor ( Figure 4A ) . In bbs3 aug and bbs3 long morphants green opsin expression was not restricted to the outer segments of the photoreceptors; rather , green opsin was also detected in the cell bodies of the outer nuclear layer throughout the entire retina ( Figure 4B and 4E ) . To determine whether there is a functional difference between BBS3 and BBS3L in its ability to rescue the green opsin localization in the photoreceptors of MO-injected embryos rescue experiments were performed . The first question we addressed was if BBS3L RNA was sufficient to rescue green opsin localization in morphant embryos . Expression of wild-type human BBS3L RNA led to improved green opsin localization in both bbs3 aug and bbs3L morphant embryos ( Figure 4D and 4G ) . The percentage of cells mislocalizing green opsin was quantified and indeed BBS3L RNA was able to statistically rescue the green opsin defect in bbs3 aug morphants ( Fisher's exact test , p<0 . 01 ) ( Figure 4H and Table 1 ) . We next investigated whether BBS3 could compensate for loss of bbs3L in the zebrafish retina . Co-injection of wild-type human BBS3 RNA failed to rescue green opsin localization in bbs3 aug and bbs3L morphant embryos ( Figure 4C , 4F , and 4H and Table 1 ) . These data are consistent with the vision startle response rescue data and supports the hypothesis that BBS3L has an eye specific role . Moreover , these data support a specific role for bbs3L in the retina and for localization of proteins within the photoreceptor cell . A polyclonal antibody against a central region of the mouse Bbs3 peptide , which is conserved across human and mouse , was generated to recognize both isoforms of Bbs3 . Cellular localization of Bbs3 was assessed in donor human and mouse retinal tissue . Immunohistochemistry was performed on transverse cryosections from adult human and adult mouse eyes using the Bbs3 antibody . Staining revealed expression of Bbs3 ( green ) in the ganglion cell layer and the nerve fiber layer as well as the photoreceptor cells of both mouse ( Figure 5A ) and human retinal tissue ( Figure 5D ) . Additionally , peanut agglutinin ( PNA , red ) was used as a marker for cone outer segments in both mouse ( Figure 5B ) and human retinal sections ( Figure 5E ) . The merge represents the co-localization of Bbs3 ( green ) and PNA ( red ) in the photoreceptor cells of both mouse ( Figure 5C ) and human ( Figure 5F ) . The specificity of the BBS3 antibody for immunohistochemistry was confirmed through peptide blocking of the antibody on wild-type mouse retina ( Figure S2 ) . To characterize the effects of loss of Bbs3L on mammalian photoreceptors a targeted knockin of the long form of Bbs3 was carried out by altering the splice donor and acceptor sites flanking exon 8 , leading to the exclusion of exon 8 upon homologous recombination ( Figure 6A ) . This approach leads to the preservation of Bbs3 expression in the Bbs3L-null mice . RT-PCR confirmed the generation and transmission of the Bbs3L allele in +/− and −/− mice ( Figure S3 ) . Unlike previously generated BBS knockout mice , which are obese by 7 months of age , Bbs3L−/− mice do not become obese ( data not shown ) [29]–[32] . This supports the idea that Bbs3L function is restricted to the retina and is consistent with the zebrafish knockdown studies . Gross histological examination of 8-month-old wild-type and homozygous ( Bbs3L−/− ) mutant mice revealed that while all cell layers were present ( Figure 6B and 6E ) , the inner segments of the photoreceptors were disrupted in a majority of the mutant mice as compared to wild-type ( Figure 6C and 6F ) . In wild-type mice , the inner segment layer is arranged in a parallel array; while in the Bbs3L-null mice the parallel arrangement of the IS was eccentric with individual inner segments randomly oriented . Additionally , immunohistochemistry with the Bbs3 antibody ( green ) , which recognizes the endogenous Bbs3 protein that is still present , and rhodopsin ( red ) in Bbs3L−/− mice further confirms inner segment disorganization in Bbs3L−/− mutant mice compared to wild-type ( Figure 6D and 6G ) . The present study identifies and characterizes the eye-enriched transcript BBS3L using both the zebrafish and mouse model systems . While typical BBS genes are ubiquitously expressed and lead to multiple phenotypes in human , mice and zebrafish , BBS3L expression is restricted to the eye and serves as a useful tool for understanding the specific pathophysiology of BBS proteins in blinding diseases . By knockdown in zebrafish , we find that bbs3L is required for visual function and localization of the photopigment green cone opsin; however , bbs3L is dispensable for the cardinal features of BBS in zebrafish , including reduced KV and delayed melanosome transport . Moreover , BBS3L RNA , but not BBS3 RNA , is sufficient to rescue both the vision defect as well as green opsin localization . These data provide strong evidence that bbs3L is specifically required for retinal organization and function . Immunohistochemistry using an antibody that recognizes both Bbs3 and Bbs3L indicates strong expression of the protein in the ganglion cell layer , nerve fiber layer and photoreceptor cells in both human and mouse retinas . By using this antibody on Bbs3L-null mouse retinas , we can deduce that Bbs3 is expressed in both the photoreceptors and ganglion cells . This is consistent with expression data indicating that Bbs3L is enriched in the retina . We have previously demonstrated that knockdown of bbs genes in the zebrafish leads to KV defects and melanosome transport delays [26] , [42] , [43] . As previously reported , knockdown of bbs3 using the aug morpholino yields both KV and melanosome transport defects; however , knockdown of only bbs3L does not affect the KV or melanosome transport . The lack of these cardinal features is not surprising given that bbs3L is not expressed at the KV stage and that in adult zebrafish the long transcript is only expressed in the eye . Since bbs3 and bbs3L are identical except for the splicing of the last exon , we cannot technically knockdown bbs3 alone without affecting bbs3L . However , based on rescue data , bbs3 knockdown alone appears to be responsible for both the KV and melanosome transport defects seen with the aug morpholino . Importantly , bbs3 and bbs3L do not seem to be functionally interchangeable . Forced expression of BBS3L RNA , at a time and place where the endogenous transcript is not present , does not rescue the cardinal features of BBS in the zebrafish that result from knockdown of both transcripts . Moreover , over-expression of BBS3 in the whole embryo cannot restore vision loss resulting from the knockdown of only bbs3L . It should be noted that melanosome transport is evaluated after the vision startle assay; therefore , we know that over-expressed BBS3 is functional at the time of the vision assay . Although bbs3 may have some effect on vision that is below the detection level of our assay , we have demonstrated that bbs3L function is both necessary for vision and sufficient to rescue vision loss in the zebrafish . Similar to the zebrafish results , a Bbs3L-null mouse lacks the observed phenotypes of previously published Bbs-null mice , such as obesity [29]–[32] . The effect of Bbs3 in the mouse retina may be more significant as Bbs3L-null mice present with only a variable mild disruption of the normal architecture . This indicates that in the mouse retina , Bbs3 is able to partially compensate for loss of Bbs3L . Moreover , the difference in phenotype between zebrafish and mouse could potentially be due to the ratio of cones and rods found in each model system . One hypothesis is that bbs3L plays a major functional role in cones , but only a minor role in rods . At the stages examined in the zebrafish , cones are the only functional photoreceptors in the retina , whereas mice have a rod-dominated retina [59] , [60] . These attributes are important to consider when looking at the role of BBS in human disease progression , as humans rely on their fovea , a specialized cone-dominant structure in the center of the macula , for visual acuity . Continued characterization of the Bbs3L-null mouse may shed more light on this difference between the mouse and zebrafish system , as well as elucidate a more definitive role for BBS3L in the retina . Taken together , these date demonstrate that the BBS3L transcript is specifically required for retinal organization and function . While we have identified a second transcript of BBS3 , a gene known to cause BBS , we would not expect patients with mutations affecting only BBS3L to present with BBS . Based on our findings in both a zebrafish and mouse model of BBS3L , patients with mutations in BBS3L alone would present with a non-syndromic retinal disease , characterized by photoreceptor dysfunction and death . Indeed , recent homozygosity mapping of a consanguineous Saudi family has identified a missense mutation in BBS3 that leads to non-syndromic RP [61] . Functional characterization of this mutation in the zebrafish may provide additional clues to the role of BBS3 in the eye . Thus this eye specific transcript , BBS3L , will serve as a useful tool for understanding the pathophysiology of other blinding diseases . In addition , our data indicate that expression of BBS3L , rather than BBS3 , would be needed for gene therapy aimed at treatment of blindness in BBS3 patients . All animal work in this study was approved by the by the University Animal Care and Use Committee at the University of Iowa . Expressed sequence tag ( EST ) data for human and mouse BBS3 was downloaded from NCBI and compared to the known coding region as represented by the NCBI reference sequence ( NM_177976 . 1 and NM_032146 . 3 for human and NM_019665 . 3 for mouse ) .	Retinitis pigmentosa ( RP ) , a disorder of retinal degeneration resulting in blindness , occurs due to mutations in dozens of different genes encoding proteins with highly diverse functions . To date , there are no effective therapies to delay or arrest retinal degeneration . RP places a large burden on affected families and on society as a whole . We have studied a syndromic form of RP known as Bardet-Biedl Syndrome ( BBS ) , which leads to degeneration of the photoreceptor cells and is associated with non-vision abnormalities including obesity , hypertension , diabetes , and congenital abnormalities of the kidney , heart , and limbs . In this study we utilized two model systems , the zebrafish and mouse , to evaluate the function of a specific form of BBS ( BBS3 ) . We have identified a novel protein product of the BBS3 gene and demonstrated that functional and structural abnormalities of the eye occur when this form of BBS3 is absent . This finding is of significance because it indicates that BBS3 mutations can lead to non-syndromic blindness , as well as blindness associated with other clinical features . This work also indicates that treatment of BBS3 blindness will require replacement of a specific form of the BBS3 gene .	[ "Abstract", "Introduction", "Results", "Discussion", "Materials", "and", "Methods" ]	[ "developmental", "biology", "ophthalmology/retinal", "disorders", "developmental", "biology/morphogenesis", "and", "cell", "biology", "genetics", "and", "genomics/gene", "expression", "cell", "biology", "molecular", "biology/rna", "splicing", "genetics", "and", "genomics/dise...	2010	Identification and Functional Analysis of the Vision-Specific BBS3 (ARL6) Long Isoform
The segmentation of the vertebrate body is laid down during early embryogenesis . The formation of signaling gradients , the periodic expression of genes of the Notch- , Fgf- and Wnt-pathways and their interplay in the unsegmented presomitic mesoderm ( PSM ) precedes the rhythmic budding of nascent somites at its anterior end , which later develops into epithelialized structures , the somites . Although many in silico models describing partial aspects of somitogenesis already exist , simulations of a complete causal chain from gene expression in the growth zone via the interaction of multiple cells to segmentation are rare . Here , we present an enhanced gene regulatory network ( GRN ) for mice in a simulation program that models the growing PSM by many virtual cells and integrates WNT3A and FGF8 gradient formation , periodic gene expression and Delta/Notch signaling . Assuming Hes7 as core of the somitogenesis clock and LFNG as modulator , we postulate a negative feedback of HES7 on Dll1 leading to an oscillating Dll1 expression as seen in vivo . Furthermore , we are able to simulate the experimentally observed wave of activated NOTCH ( NICD ) as a result of the interactions in the GRN . We esteem our model as robust for a wide range of parameter values with the Hes7 mRNA and protein decays exerting a strong influence on the core oscillator . Moreover , our model predicts interference between Hes1 and HES7 oscillators when their intrinsic frequencies differ . In conclusion , we have built a comprehensive model of somitogenesis with HES7 as core oscillator that is able to reproduce many experimentally observed data in mice . Somitogenesis is an embryonic process that provides the basis for the mesodermal segmentation of the vertebrate body . Somites are derivatives of the presomitic mesoderm ( PSM ) , a mesenchymal tissue that is formed during gastrulation and maintained by proliferation of cells in the tail bud . They are epithelial balls of cells that separate from the anterior end of the PSM to both sides of the neural tube . In mice , approximately every two hours one pair of somites is formed until proliferation in the tail bud stops and a species-specific number of somite pairs has been generated [1] . Fundamental to somitogenesis is the formation of a segmental boundary between the last formed somite and the unsegmented PSM . Before a boundary becomes morphologically visible , wave-like gene expression patterns propagate from the posterior to the anterior end of the PSM with the same periodicity as somites are formed [2] . Most prominent among these cycling genes are those involved in the Delta/Notch ( D/N ) pathway , like Lfng and the helix-loop-helix transcription factors Hes1 , Hes5 , Hes7 and Heyl . They are induced by the NOTCH intracellular domain ( NICD ) , which is cleaved off from the NOTCH receptor upon binding to DELTA or JAGGED ligands at adjacent cells and acts subsequently as co-transcription factor in the nucleus of NOTCH expressing cells [3] . NICD shows a cycling and wave-like expression in the PSM [4] . D/N signaling and Hes7 oscillation are essential for somitogenesis [5] , [6] . For example , loss of NOTCH1 function resulted in delayed and disorganized somitogenesis [7] . Similarly , in mice lacking the NOTCH ligand DELTA-LIKE 1 ( DLL1 ) or the down-stream effector HES7 somites are not properly segmented and display a disrupted rostral-caudal polarity [8] , [9] . In contrast , oscillating expression of Lfng in the posterior PSM seems to be dispensable for the formation of somites that later give rise to sacral and tail vertebrae [10] , [11] . Other genes required for normal somitogenesis belong to the Fgf and Wnt/β-catenin signaling pathways . Both Fgf8 and Wnt3a are transcribed in the growth zone of the tail bud but not in the more anterior region of the PSM . A slow decay of Fgf8 mRNA leads to a graded expression of FGF8 protein levels from the posterior to the anterior end of the PSM [12] . Likewise , a posterior to anterior gradient of nuclear β-catenin is observed [13] . A third gradient of retinoic acid ( RA ) is established in the reverse direction and thought to suppress Fgf8 expression [14] . Genes downstream of the Fgf pathway cycle in phase with respect to D/N oscillations , whereas genes belonging to Wnt/β-catenin signaling cycle in anti-phase [15] . Experimental manipulations of the Fgf or Wnt/β-catenin pathway also impair somite formation [13] , [16] , and inhibition of casein kinase 1 , which is downstream of Wnt , lengthens the period of the somitogenesis clock [17] . In Mesp2 deficient embryos , somite boundary formation is lost [18] . MESP2 induces the expression of Epha4 [19] . In chick , the EPHA4 receptor binds to the ephrin B2 ligand on cells across the future boundary and thereby triggers furrow formation and cell epithelialization at the gap between the forming somite and the PSM [20] . Mesp2 is expressed periodically by joint binding of NICD and the T-box transcription factor TBX6 in its promoter region resulting in a narrowing stripe of Mesp2 expression at the anterior PSM [21] . While Tbx6 expression is static and restricted to the PSM and is rapidly down-regulated as the somites form , NICD expression is dynamic , forming a wave moving in anterior direction through the PSM and contracting antero-posteriorly in width as it nears the anterior end [4] . Based on the promoter information for Mesp2 and additional evidence that FGF8 suppresses posterior Mesp2 expression , Oginuma et al . formulated a model that describes how dynamic NICD induces Mesp2 expression patterns in the PSM [22] . Later , they simulated Mesp2 and TBX6 expression with a system of differential equations in a computer model of a one-dimensional array of cells [23] . Other models employ a modulo function on an Fgf gradient [24] to generate the NICD wave , or a Boolean variable for NICD , which is repressed by the action of a Hes7 oscillator with an empirically adjusted oscillation period [25] . We aim to develop an integrated model that depicts the causal chain leading from processes in the growth zone – unfortunately still incompletely known – to the dynamic gene expression patterns in the PSM wave zone and to segmentation of the PSM into somites . In particular , we are driven by the following questions: These questions are connected to our central question: what is the core oscillator driving somitogenesis and how does it work ? While this manuscript was in the review process , Hester et al . published such an integrated model for chick somitogenesis , in which repression of D/N signaling by LFNG serves as the core oscillator [26] . Here , we propose an extended theoretical gene regulatory network ( GRN ) for Mesp2 oscillation that is based on a Hes7 feedback oscillator [27] driven by dynamic NICD expression with LFNG as modulator/enhancer of D/N signaling . Additionally , we assume a negative feedback of HES7 on Dll1 expression and hypothesize an influence of the Wnt3a signaling gradient on the decay of NICD . Figure 1 shows a schematic representation of the vertebrate segmentation process and the underlying gene/protein expression patterns that we consider to be essential . By incorporating the extended GRN in a multi-cell simulation program of the growing PSM that allows real-time observation of gene expression in thousands of virtual cells [28] , we are able to model the dynamics of NICD expression and link D/N signaling to the Mesp2/Ephrin system . As result we observe dynamic wave-like expression of NICD and Dll1 in silico as seen in vivo [4] , [29] as well as periodic Mesp2 expression along the growing PSM [4] . Introducing a gene “Epha4” downstream of MESP2 we obtain regular formation of “Epha4” expression maxima . The usefulness of our in silico system is demonstrated by the elimination of Hes7 or Ripply2 from the GRN , which results in non-oscillatory NICD and Mesp2 expression patterns moving from anterior to posterior or , in the case of absent RIPPLY2 , in a double-striped Mesp2 expression as observed similarly in respective mouse models [6] , [30]–[32] . Furthermore , we show the occurrence of beat [33] in oscillating gene expression resulting from the interference of two genetic oscillators with different eigenfrequencies . A lot of experimental information in biology resides in pictures derived from experiments showing their results by in situ staining . These results lead to hypotheses formulated e . g . in the form of GRNs . In the field of somitogenesis , Gonzalez et al . [34] has built a database of all relevant experiments and examined whether the GRNs discussed so far can explain the observed results . They found gaps in our understanding and tried to fill them with hypothetical interactions . Still missing is a computational validation whether the postulated GRN can really explain observed gene expression patterns . Therefore , one would need a gene- and cell-based model of the process in question . Our model is intended as a first step in this direction . Somitogenesis is comprised of several subprocesses such as the growth of the PSM , oscillatory gene expression , synchronization of the oscillators , boundary formation between PSM and the next forming somite , somite polarization , and somite epithelialization . Several mathematical models exist for some of the processes , which provide a basic understanding of the described phenomena . However , each model has its own assumptions and simplifications , so it is not clear whether existing partial models are ‘consistent and integrable with one another’ [26] . Furthermore , there are experimentally generated phenotypes , which can be fully understood only by the interaction of several parts of the system , each of them modeled separately until now ( for an example , see our in silico Hes7 knock-out experiment below . ) We intended to build a comprehensive model of somitogenesis , in which most processes generated by the action of a GRN i . e . by integrating the differential equations describing the processes . However , the proliferation in the growth zone and the deactivation of Fgf8 and Wnt3a expression when cells leave the growth zone were put in ‘by hand’ and are controlled by the program . We introduced EPH4A as a marker , which has the only task to trigger the shape of a cell to indicate boundary formation at the anterior PSM . This process is effected by the program when a certain threshold of EPH4A protein concentration is reached in the simulation by the GRN . We designed our program with the intention that a user can easily change the numerical values of the rate constants , the Hill functions , or take out genes . The resulting expression pattern of a chosen gene product can be followed in real time . We therefore attached a detailed graphical user interface ( GUI ) to our program ( see the screenshots in the user manual provided as supplementary Text S1 . ) To model gene expressions in the PSM we use essentially the same methodology as recently described in Tiedemann et al . 2007 [28] , i . e . a gene- and cell-based simulation program that numerically solves differential equations describing a gene regulatory network in each cell and displays the actual concentration of a selected gene product by color intensity ( virtual in situ staining ) . The following improvements were made to the program: A growth zone of several cell layers now extends the rectangular geometry of the PSM . During growth of the PSM one cell of each column along the growth direction is randomly chosen for mitosis . Thereby the program allocates a new instance of a Java ‘cell’ object , which inherits all concentration values of its mother cell . A new ‘cell’ is created at the location of its ‘mother cell’ and then pushes stepwise all other cells of the respective column towards the posterior end of the PSM . The corresponding movements and position changes are computed by the program . The growth zone of the PSM is defined by the program to encompass the last n layers ( user defined , default value: n = 15 ) , i . e . planes perpendicular to the growth direction . If a cell leaves the growth zone Fgf8 and Wnt3a expression is shut down by the program , as we have no GRN modeling this process . The diffusion of FGF8 and WNT3A protein is not simulated in detail since we assume the gradients to be mostly determined by the intracellular decay of the respective mRNAs . So we assume a very short diffusion range for FGF8 and WNT3A [35] , i . e . each cell receives only the proteins from its nearest neighbors . The concentrations are averaged and act immediately on the targets of the respective signal transduction pathway . This means , we assume the intervening processes are fast compared to the mRNA decay , which determines the dynamics and scale of the gradients . Furthermore , we introduce distinct variables for cellular and nuclear concentrations of proteins and the respective mRNAs . The distinction in compartments is made for the oscillatory factors HES1/7 , MESP2 , NICD and LFNG , but not for the slow-changing concentrations of proteins and mRNAs of Notch1 , Tbx6 , Fgf8 and Wnt3a . For the DLL1 ligand and the NOTCH receptor we set separate variables in the cytoplasm and membrane compartments . We validated our model system by the in silico elimination of Hes7 , which results in a constant , receding stripe of Mesp2 expression moving in anterior to posterior direction with the growing PSM ( Figure 4 , Video S6 ) . This result is consistent with experimentally observed data in mice deficient for Hes7 [6] , [25] . A constantly anterior to posterior moving Mesp2 expression in the PSM can be also observed when the influence of HES7 on the Dll1 promoter is eliminated . Although Hes7 expression is oscillatory it shows no discernible wave in the PSM , because constant DLL1 and NOTCH1 expression result in constant production of NICD . So Mesp2 is consequently moving within the borders set by TBX6 and FGF8 expression in the growing PSM ( data not shown ) . A similar Mesp2 expression pattern was observed in vivo in embryos expressing NICD throughout the PSM [30] . We introduced a term for constant cytoplasmic NICD production in our simulation and observed again a constant , receding stripe of Mesp2 expression moving in anterior to posterior direction ( data not shown ) . When we eliminated Ripply2 from the GRN we observed two stripes of Mesp2 expression ( Figure 5 ) . Similar expression patterns for Mesp2 were observed in mice deficient for Ripply2 [31] , [32] . By the usage of the chemical compound SU5402 to inhibit Fgf signaling , Niwa et al . observed a broadened stripe of Mesp2 as result of a precociously expression in the next clock cycle [25] . To simulate an inhibition of Fgf signaling we reduced the FGF8 protein production rate during a simulation run by 50% after 600 minutes and observed a similar expanded Mesp2 expression ( Figure 6 , Video S7 ) . Genetic oscillators of the Hes1/7 type are understood to result from a negative feedback with delays . These delays can be incorporated into a mathematical description in various ways . Either directly as delayed time arguments describing the duration of transcription , translation and transport processes [39] , [65] , resulting in delay-differential-equations , or , alternatively , as chains of transport equations [64] , possibly enhanced by the introduction of nonlinearities like Michaelis-Menten or Hill-type functions [42] describing , for example , saturated decay processes . A third possibility is the explicit modeling of intracellular diffusion of proteins and mRNA in the cytoplasm of a cell [66] . Although delay models require only two equations per gene and the delays , for instance caused by splicing of Hes7 introns [65] , are easier to measure than in our compartment model , we think that delay equations used so far have two drawbacks . First , within delays many steps in gene-expression processes are hidden . As probably most processes like splicing , transport between nucleus and cytoplasm , protein and mRNA decay in the eukaryotic cell are controlled by signal transduction pathways , several modes of cross-talk [67] are neglected . Of course , it is not meaningful to model each step in transcription and translation by one differential equation . The transport steps and the saturated decay implemented in our model should be understood as representative for all these cellular processes . So , a cross-over model restricting the delays to transcription and translation would probably be the best model for future simulations . Second , our model uses nonlinearities and saturation functions to consider proteasome-effected degradation of transcription factors . Proteasomes are located in the cytoplasm as well as in the nucleus [68] , [69] . We introduce nonlinearities to allow a possible saturation of the proteasome machinery in the nucleus but neglect this possibility in the cytoplasm . However , if one simulates explicitly phosphorylation and ubiquitination of a protein before its destruction in the proteasome [70] , one could assume saturated decay also in the cytoplasm [43] . For further information concerning different processes involving NICD in the nucleus and cytoplasm see [71] , [72] . Because degradation processes in an eukaryotic cell are affected by protein complexes , i . e . molecular machines like proteasomes for protein disposal or exosomes for mRNA destruction , nonlinear descriptions could be appropriate for other processes as well . Every machine has a saturation threshold that can be overwhelmed by substrate molecules when their concentration is too high . Another important issue , which needs more effort , is the role of transcription cofactors and their influence on the Hill coefficients of associated transcription factors . The higher the Hill coefficient and the degree of cooperation , the higher is the propensity to oscillations [42] . Many genes were found to oscillate in the PSM of mouse embryos [15] , most of them are downstream of the Fgf , Wnt and D/N –pathways . Genes downstream of the Fgf pathway cycle with respect to D/N oscillations , whereas genes belonging to Wnt signaling cycle in anti-phase . Some of the downstream genes in both pathways act as inhibitors along the signal transduction cascade , forming negative feedback loops and allow for oscillations . Two scenarios are conceivable: an oscillator in one pathway acts as master clock and controls the others , or alternatively , no master clock exists and all cycling genes are equally important in the oscillatory network . A model for the latter case was developed by Goldbeter and Pourquie [44] . Genes of all three pathways generate three oscillators , which are coupled and synchronized by genetic interactions between them . We will argue for the first scenario with the D/N pathway as the central driver . First , there is evolutionary conservation . Comparing cycling gene expression in mouse , chick , and zebrafish , Krol et al . [73] found that ‘conservation of cyclic genes for all three species is limited to orthologs of the HES/her transcriptional repressors’ . For example , comparing the Wnt pathway between chick and mouse only Axin2 oscillates in both species [73] , and in anolis lizards , only the orthologs of Hes1 , Hes7 , Dll1 , and Dll3 are dynamic whereas other genes like Lfng are not oscillating [74] . Second , several experimental data point to the same direction . In mice , most of the FGF8 targets , which oscillate in phase with D/N cycling genes , are controlled either directly by NICD as co-transcription factor , as in the case of Dusp6 and Spry4 , or indirectly via HES7 like Spry2 and Dusp4 [6] , [36] , [75] . Moreover , if one considers also the activation of Snail and Nrarp by NICD [76] , [77] , it seems reasonable to suppose that D/N signaling controls all genes that cycle in phase to D/N oscillations [15] . D/N signaling controls also Nkd1 , which interact with components of the Wnt pathway [78] . One of the cycling genes in this pathway is Axin2 , which have a negative feedback on Wnt signaling and could be a critical component of a Wnt oscillator [79] . However , it was shown that Axin2 deficient mice show no somitogenesis phenotype [80] . Of course , all these arguments do not prove conclusively that D/N is the central oscillator in mice . However , it supports the assumptions made in our model , which sets D/N signaling at the center . The complex D/N oscillator consists of two interlocking negative feedback loops . On the one hand , Lfng , which is dynamically expressed in the PSM , modulates D/N signaling by glycosylation of the NOTCH receptor . It is induced by NICD and suppressed by HES7 and was therefore considered to be a core molecular mechanism of the somitogenesis clock [81] , [82] . On the other hand , the negative feedback of Hes7 , which is driven by NICD and FGF8 , onto itself , represses also Lfng expression . There are in silico models that take the Lfng negative feedback loop as the core of the somitogenesis clock [26] , [44] , while others emphasize the role of Hes7 [25] . Niwa et al . , for example , assumed a direct suppression by HES7 on NICD expression [25] . However , both models cannot explain , in our opinion , some crucial experiments . LFNG protein is secreted from a cell to terminate its action in the secreting cell but probably not as mechanism for cell-cell communication [83] . The ‘Lfng only’ models cannot describe the uniformly receding Mesp2 expression observed in the PSM of Hes7 knock-out embryos [6] , [25] . These models have to postulate a direct influence of NICD on Dll1 expression in the same cell , otherwise an oscillation in one cell could not be transmitted to neighboring cells , i . e . there would be no information transfer between cells , which is needed for the postulated role of D/N signaling in synchronizing cellular oscillators of the PSM [84] . Furthermore , Lfng expression seems not to be required for proper somitogenesis in the late tail bud phase [10] , [11] , and Hes7 expression is still dynamic in Lfng deficient mice [6] . In addition , Lfng shows a constant expression in zebrafish and Medaka [85] . Contrary to these findings , Oginuma et al . recently reported a requirement for oscillating Lfng expression in the posterior PSM for proper somitogenesis [23] , and Niwa et al . observed a damping in the amplitude of HES7 oscillations in Lfng deficient compared to wild type mice [25] . A model that postulates a direct interaction of HES7 on NICD has difficulties to explain the fact that Dll1 was found to be cycling in the PSM [29] . One would have to assume a direct interaction of NICD on Dll1 or a longer causal chain , in which , for example , NICD controls Nkd1 , which controls Wnt signaling that controls Dll1 . Here , we show that a simple model analogous to the zebrafish network is able to describe the dynamic NICD expression in the PSM as it was shown by Morimoto et al . [4] . Retinoic acid ( RA ) is important for synchronizing and balancing the development of left and right halves of the PSM [98] . RA is expressed in the somites and forms an anterior to posterior gradient opposing the posterior to anterior gradient of Fgf8 . The opposing action of both gradients was modeled by Goldbeter et al . [99] and results in a sharp expression cut-off at the anterior boundary of the FGF8 gradient . Because there is not enough information about the control of RA in somites , which would allow us to incorporate RA signaling in our GRN , it is currently not included in our simulations . However , we modeled an inhibitory action of FGF8 on Mesp2 expression with a high hill coefficient to ensure a steep drop-off at the anterior end of FGF8 expression . Apart from these considerations , our model is intended to model somitogenesis in the tail bud phase . Cunningham et al . [100] have shown that during mouse embryogenesis from E9 . 5 to E13 . 5 Mesp2 expression and somitogenesis was not changed in Raldh2 knock-out mice , which are unable to produce RA in the somites . They concluded ‘that as early as E9 . 5 Raldh2 is not required to limit the anterior extent of the caudal Fgf8 expression zone . ’ While it is known , for example , that Dusp6 is controlled by FGF8-ERK1/2 via the Ets family of transcription factors [101] , it is yet unknown which of the FGF8 downstream factors activate Hes7 and inhibit Mesp2 [22] , [36] . Once this is known , we could extend our model by more detailed modeling of Fgf8 and Wnt3a signaling and the corresponding downstream genes , some of which show also cycling expression [15] due to the various negative feedbacks that are discussed for these signal transduction pathways [102] . For a single-cell-simulation this was done by Goldbeter and Pourquie [44] , and for a 2-D simulation of chicken somitogenesis by Hester et al . [26] . Recently , Niwa et al . [25] observed that the expression of pERK and DUSP4 , which are downstream of FGF8 , is not constantly receding but periodically covers and frees the MESP2 expressing region of the PSM . This could be a result of the negative feedback of DUSP4 , driven by HES7 oscillations , on Fgf8 signaling and be important for the control of Mesp2 expression in the future somite [25] . Furthermore , besides refining the Mesp2 , Ripply2 , and Tbx6 expression patterns by incorporating the mechanisms proposed by Takahashi et al . [24] a much more detailed modeling of the Ephrin/Eph receptor system with forward as well as backward signaling components [103] should be considered , as also other genes in this network cycle or are controlled by HES7 [36] , [73] . In our simulations , all cells start synchronously in the growth zone and remain synchronized because daughter cells inherit the oscillation phase of their mother cells . So , synchronization of cellular oscillators by D/N coupling is not needed . We achieved a partial solution of the synchronization problem in an earlier version of our model with a growth zone comprising only one single cell layer . When we coupled the HES1 transcription factor as inhibitor to the Dll1 promoter while Notch signaling activates Hes1 , we observed similar synchronized cellular oscillations in neighboring cells as for a delay differentiation model of zebrafish [39] , [86] or the Hes7 oscillator of mouse [58] . However , we achieved synchronization only when we consider the binding of NICD as dimer [38] and when the relative phases of Hes oscillations in all cells generated in the one cell-layered growth zone differ not more than 25% ( Figure S9 ) . Uriu et al . have demonstrated that one has to include random cell movements into the PSM to achieve faster synchronization [104] . Such a random cell motility gradient downstream of FGF8 was recently described for chick embryos [105] . Furthermore , the assumption of D/N signaling to function exclusively through nearest neighbor communication could be reconsidered . In Drosophila , for example , dynamic filopodia transmit D/N signals during bristle formation [106] . Similar cellular extensions bearing Delta ligands were observed in zebrafish [107] . An even more radical deviation from the common view on D/N signaling might be considered by the finding that DLL4 incorporated in endothelial exosomes might be transferred and integrated into the cell membrane of distant cells [108] . Somitogenesis describes not only border formation , but also comprises the mesenchymal to epithelial transition in the outer layer of cells of the forming somite . Snail and Zeb2 are genes that are well known to be involved in epithelial-to-mesenchymal transition ( EMT ) by controlling E-cadherin [109] but also show cyclic expression in the PSM [29] , [76] . Finally , somitogenesis requires an extra-cellular matrix ( ECM ) of fibronectin surrounding the PSM [110] . Jülich et al . demonstrated that in zebrafish reverse signaling by EphrinB2a is sufficient to initiate Itgα5 clustering , alleviates non-cell-autonomous transinhibition , which prevents fibronectin activation within the PSM , but induces fibronectin matrix assembly along somite borders [111] . Taken together , these findings reinforce the need to include ECM and integrin signaling into the modeling of somitogenesis . However , the translation of gene expression patterns into the control of cell adhesion and cell shape , as well as the interaction with the extracellular matrix would require the development of a model for the cell skeleton and other features , which should not only be realistic enough but also computable in a simulation with thousands of cells . The influence of the NICD wave on the dynamics of Mesp2 and Ripply2 expression , activated by TBX6 and repressed posteriorly by FGF8 , was recently simulated in several mathematical models . To reproduce NICD expression as experimentally observed the wave was generated either by a wave function [23] , by a modulo function on an Fgf gradient [24] , or by a Boolean variable for NICD which is repressed by the action of a Hes7 oscillator with an empirically adjusted oscillation period [25] . Here , like in Hester et al . [26] , we propose a model in which the NICD wave in the PSM results from a GRN . However , in our model of the mouse core oscillator for somitogenesis a negative feedback of the HES7 oscillator on the expression of Dll1 leads to periodic expression of NICD , while Lfng is considered not as central part of the core oscillator but as a modulator of the Hes7-D/N oscillator . Its oscillation slows down because the degradation of NICD is influenced by a Wnt3a gradient in the posterior PSM . Since quantitative information on rate constants for production and degradation of mRNAs and proteins is mostly missing , we classify the genes in our model with fast or slow dynamics . However , our model is able to reproduce a great deal of experimentally observed data in mice . Encouraged by the agreement between experiment and our simulation , we hypothesize that HES7 binds to the Dll1 promoter . To prove this hypothesis experimentalists could perform in vivo chromatin immunoprecipitation ( ChIP ) analyses according to a protocol previously described by Bessho et al . 2003 [5] . In the following , we will summarize the achievements but also the deficiencies of our model in bullet-point form: Achievements Deficiencies However , our model can answer at least theoretically some of the questions listed in a recent review [79] concerning the nature of the somitogenesis oscillator and its interaction with the Wnt3a gradient . Furthermore , we now have a model for the generation of dynamic NICD expression in the PSM , which is also robust to our parameter variations . In addition , we can connect our model to the model of somite border formation as formulated by Oginuma et al . [22] , which can explain important aspects of somitogenesis . Furthermore , our programming framework makes it easy to expand the model in future works . Finally , we are pleading for Hes7 as the central pacemaker of the somitogenesis clock with Lfng having only a modulatory role . This scenario can explain some observations other models cannot , and is in accordance with the evolutionary conservation of the somitogenesis clock . This plea is made not only by arguments , but also by demonstrating its viability in computer simulations . However , in the end the in vivo experiment has to decide whether HES7 binds to the Dll1 promoter – as our model assumes – or not . The GRN is represented by 38 differential equations . The equations , the specific rate constants for the decay , production , as well as import of proteins and mRNAs for each gene are given in the supporting information ( Table S1 ) . The program is written in Java with the 3D-extension . For rendering the concentrations of the different gene products their values have to be normalized to lie in a range between zero and one . The differential equations are numerically solved with a fourth order Runge-Kutta-algorithm . Videos were made using camstudio software . The program writes all data of user selected cells to file . Plots were performed with the GNU-plot software . This is especially useful if one does not know beforehand the concentration range of a gene product that is selected to view in the simulation . The program can be downloaded at www . helmholtz-muenchen . de/en/ieg/downloads/simulation11 . As the program requires the Java 3D API , which for legal reasons cannot be packaged into the downloadable file by us , users must ensure that it is installed on their computers before trying to run the program . The source code is available upon personal request .	Somitogenesis is a process in embryonic development establishing the segmentation of the vertebrate body by the periodic separation of small balls of epithelialized cells called somites from a growing mesenchymal tissue , the presomitic mesoderm ( PSM ) . The basic mechanisms are often discussed in terms of the clock-and-wave-front model , which was proposed already in 1976 . Candidate genes for this model were found only in the last fifteen years with the cyclically expressed Hairy/Hes genes functioning as the clock and posteriorly expressed Fgf , Tbx6 , and Wnt genes establishing the gradient ( s ) . In addition , the Delta/Notch signal transduction pathway seems to be important for boundary formation between forming somites and the remaining PSM by inducing Mesp2 expression just behind a future somitic boundary . Although many in silico models describing partial aspects of somitogenesis already exist , there are still conflicts regarding the mechanisms of the somitogenesis clock . Furthermore , a simulation that fully integrates clock and gradient was only recently published for chicken . Here , we propose a cell- and gene-based computer model for mammalian somitogenesis , simulating a gene regulatory network combining clock ( Hes1/7 ) and gradient ( Tbx6 , Fgf8 , Wnt3a ) with Delta/Notch signaling resulting in dynamic gene expression patterns as observed in vivo finally leading to boundary formation .	[ "Abstract", "Introduction", "Model", "Results", "Discussion", "Materials", "and", "Methods" ]	[ "systems", "biology", "developmental", "biology", "embryology", "biology", "computational", "biology", "signaling", "networks", "pattern", "formation" ]	2012	From Dynamic Expression Patterns to Boundary Formation in the Presomitic Mesoderm
Multiple Sclerosis ( MS ) is an autoimmune disease associated with abnormal expression of a subset of cytokines , resulting in inappropriate T-lymphocyte activation and uncontrolled immune response . A key issue in the field is the need to understand why these cytokines are transcriptionally activated in the patients . Here , we have examined several transcription units subject to pathological reactivation in MS , including the TNFα and IL8 cytokine genes and also several Human Endogenous RetroViruses ( HERVs ) . We find that both the immune genes and the HERVs require the heterochromatin protein HP1α for their transcriptional repression . We further show that the Peptidylarginine Deiminase 4 ( PADI4 ) , an enzyme with a suspected role in MS , weakens the binding of HP1α to tri-methylated histone H3 lysine 9 by citrullinating histone H3 arginine 8 . The resulting de-repression of both cytokines and HERVs can be reversed with the PADI-inhibitor Cl-amidine . Finally , we show that in peripheral blood mononuclear cells ( PBMCs ) from MS patients , the promoters of TNFα , and several HERVs share a deficit in HP1α recruitment and an augmented accumulation of histone H3 with a double citrulline 8 tri-methyl lysine 9 modifications . Thus , our study provides compelling evidence that HP1α and PADI4 are regulators of both immune genes and HERVs , and that multiple events of transcriptional reactivation in MS patients can be explained by the deficiency of a single mechanism of gene silencing . Multiple Sclerosis ( MS ) is a progressive inflammatory disease of the central nervous system in which leukocytes and antibodies attack myelin sheaths , resulting in demyelination and ultimately destruction of the axons [1] . Many lines of evidence point at inappropriate activation of T cells as an initiating event of the pathological process , although the mechanism at the root of this T cell activation is still poorly defined ( for a recent review see [2] ) . In MS patients , activation of the T cell population is associated with increased expression of a series of cytokines [3] , [4] . The abnormally abundant expression of the genes encoding these regulators of the immune system may be a consequence , but also possibly a cause of the activation of the T cells . It is therefore essential to explore the mechanisms that keep these genes in check in normal cells and that may be defective in MS patients . Interestingly , in MS and other autoimmune diseases including Rheumatoid Arthritis and Systemic Lupus Erythematosus , transcription of Human Endogenous RetroViruses ( HERVs ) is also increased in T cells [5] , [6] , [7] . HERVs are abundant vestigial retroviral sequences that in healthy cells are largely silenced by the epigenetic mechanisms repressing most repeated DNA sequences . These mechanisms include DNA methylation and histone H3 lysine 9 ( H3K9 ) methylation . DNA methylation favors chromatin compaction by promoting recruitment of histone deacetylases ( HDACs ) or alternatively by directly reducing the affinity of transcription factors to their cognate DNA binding sites [8] . Consistent with this , deletion of the de novo DNA methyltransferase Dnmt1 results in massive re-expression of ERVs in the mouse embryo [9] . DNA methylation at ERV promoters is particularly high in differentiated mouse cells [10] , while it may be partially dispensable in mouse embryonic stem ( ES ) cells [11] . In these cells , the major source of silencing appears to be H3K9 methylation [12] , [13] , [14] . This histone modification is recognized by a number of proteins containing either chromo- [15] , [16] , [17] , [18] , [19] , [20] , MBT- [21] , PHD- [22] , or Tudor- [21] , [23] domains . Mainly HP1 proteins have been detected on mouse ERV promoter sequences [24] , [25] , although their role in the repression of these sequences in mouse ES cells is still at debate [26] . HP1 proteins are particularly interesting in the context of MS because in addition to their possible function in the silencing of repeated DNA [27] , [28] , they are present on the promoters of a number of genes involved in immune defense , including the immunomodulatory cytokine TNFα [29] , the interleukins IL1β [30] , [31] , IL6 [32] , and IL8 [33] , and several interferon-inducible genes [34] . They also participate in the regulation of the HIV1 long terminal repeat ( LTR ) that shares several regulatory mechanisms with immune genes [35] , [36] , [37] . Consistent with their role in the transcriptional control of inducible genes , the binding of HP1 proteins to chromatin is subject to regulation . In particular , the methylation mark on histone H3 can be removed by histone demethylases [38] . A more transient regulation of HP1 binding may occur by modification of residues neighboring H3K9 , including phosphorylation of serine 10 and acetylation of lysine 14 [39] , [40] , [41] . The histone H3 arginine 8 ( H3R8 ) located immediately upstream of H3K9 , is also subject to modifications [42] that theoretically could interfere with HP1 binding , although this has never been investigated . This arginine can be either methylated [43] or converted into the non-coding amino acid citrulline [44] , [45] , [46] , [47] , [48] . Citrullination of H3R8 is catalyzed by the calcium-dependent peptidylarginine deiminase PADI4 . This enzyme , that is the only member of its family to enter the nucleus , also citrullinates histone H3 on arginines 2 , 17 , and 26 , as well as histones H2A and H4 on their respective arginine 3 [44] , [45] , [46] , [49] . Many reports describe PADI4 as a regulator of transcription . On p53 targets [50] , [51] , and estrogen-regulated genes , including pS2 [44] , [45] , [52] , it functions as a repressor either by interfering with activating arginine-methylation events [44] , or by favoring recruitment of HDACs [52] . Inversely , PADI4 also associates with a number of transcriptionally active promoters and functions as an activator of c-Fos via a mechanism that involves facilitated phosphorylation of the ETS-domain protein Elk-1 [53] . In Asians , hyperactivity of PADI4 has been associated with Rheumatoid Arthritis [54] , [55] , [56] , and contributes to the generation of antibodies directed against citrullinated proteins during the development of this disease [57] . An earlier study has also reported increased nuclear localization of this enzyme in white brain matter of MS patients [58] The transcriptional deregulation of both cytokines and HERVs in MS patient T cells prompted us to investigate a possibly co-regulation of these two types of transcription units by HP1 proteins . At cytokine genes and HERVs we examined in tissue culture cells , transcriptional repression required HP1α , while PADI4 functioned as an activator by destroying the HP1α binding site on the tail of histone H3 . Consistent with this , we observed that in circulating blood cells from MS patients , recruitment of HP1α to the promoter of the master cytokine TNFα and to HERV sequences is significantly reduced , while citrullination of H3R8 at these positions is increased . Taken together , our data strongly suggest that increased citrullination of histone H3 can antagonize gene-specific chromatin-mediated silencing in T cells and thereby participate in increased cytokine expression during the normal inflammatory response and in MS patients . While several reports describe an implication of HP1 proteins in the regulation of genes involved in immune defense [29] , [30] , [31] , [32] , [33] , [34] , the role of these proteins in the silencing of HERVs in human cells needed to be clarified . We carried out these experiments in MCF7 cells , a breast tumor-derived cell line frequently used to examine expression of HERVs ( see for example [59] ) . Chromatin Immunoprecipitation ( ChIP ) assays demonstrated that in these cells , HP1α accumulates on HERV-K , HERV-W , and HERV-H promoters at levels similar to those observed on Satellite-2 sequences ( Figure 1A ) . As expected , HP1α was also detected on the promoters of cytokines TNFα and IL8 . Consistent with this , depletion of HP1α with small interfering RNAs ( siRNAs ) resulted in increased expression of the HERVH/env62 , HERVH/env59 , HERVH/env60 , ERVWE1 , HERVK/env102 , TNFα , IL8 , and IL16 , a cytokine also relevant for MS [60] ( Figure 1B , Figure S1A and S1B ) . In these experiments , expression of IL23 [61] was unaffected , while the control estrogen-responsive pS2 gene was repressed rather than activated . Reactivation of HERVs and TNFα was also observed upon depletion of HP1β and HP1γ , two other members of the HP1 family ( Figure 1C ) . HP1α binds tri-methylated histone H3 lysine 9 ( H3K9me3 ) . The neighboring H3R8 residue is one of the 3 arginines recognized by the anti-H3cit ( 2 , 8 , 17 ) antibody used to show increased histone H3 citrullination in MS patients [58] . This raised a possibility of interference between citrullination of H3R8 and HP1α binding to histone H3 . We therefore explored whether citrullination of histone H3R8 occurs in vivo on histone H3 tails already tri-methylated on K9 . To this end , we generated an antibody recognizing the double citrullination-methylation modification H3cit8K9me3 ( Figure 2A ) . The specificity of this antibody was verified by dot blots using synthetic peptides mimicking modified histone tails ( Figure 2B ) and by testing the ability of the same peptides to compete with the binding of the antibody to cellular targets in fixed breast cancer-derived MCF7 cells ( Figure 2C ) . In the later assay , the anti-H3cit8K9me3 antibody yielded an immunofluorescent staining very similar to that obtained with anti-H3cit ( 2 , 8 , 17 ) antibody ( Figure 2D ) . We next focused our attention on PADI4 , the partially nuclear peptidylarginine deiminase responsible for the citrullination of histone H3 . To determine whether this enzyme can citrullinate H3R8 when H3K9 is methylated , we generated HEK293-derived cell lines expressing either wild-type ( WT ) , hyperactive [62] , or hypoactive [63] versions of PADI4 under the control of a ponasterone-inducible promoter ( Figure 2E ) . Induction of PADI4 synthesis and activity with ponasterone and the ionophore A23187 , respectively , allowed detection of the H3cit8K9me3 double modification when the cells were expressing WT or hyperactive versions of PADI4 . Under these conditions , we also observed a general increase in histone H3 citrullination , but no change in the levels of H3K9me3 , indicating that H3R8 citrullination and H3K9 tri-methylation are not antagonistic . Taken together , these experiments demonstrate that the H3cit8K9me3 double modification exists in vivo and that its formation is favored by increased PADI4 activity . We next investigate the impact of the H3R8 citrullination on the binding of HP1α to histone H3 tails tri-methylated on K9 . When histone H3 peptides were spotted on membrane , HP1α bound a peptide carrying the single K9me3 modification , but not a peptide with a double cit8K9me3 modification ( Figure 3A ) . We also tested the ability of these peptides to interfere with the binding of HP1α to its endogenous target sites . For this , we took advantage of the fact that recombinant GST-HP1α protein incubated on fixed permeabilized cells distributes in a pattern indistinguishable from that of the endogenous protein [64] . In this assay , while H3R8K9me3 peptide competed with the cellular sites for GST-HP1α binding , H3cit8K9me3 peptide did not ( Figure 3B ) . As expected , H3cit8K9 and H3R8K9 peptides also failed to compete for GST-HP1α binding . Surface plasmon resonance allowed us to quantify the effect of H3R8 citrullination and indicated a more than 200-fold decrease in the affinity of HP1α for H3cit8K9me3 compared to H3R8K9me3 ( Kd 313±28 µM and 1 . 39±0 . 06 µM , respectively; Figure 3C–3D ) . We noted also that citrullination of H3R8 exerted a 10-fold higher effect on HP1α binding than did its methylation ( Kd 29 . 6±1 . 3 µM ) . To document that citrullination compromises transcriptional repression of HERVs and cytokines , we finally used siRNAs against PADI4 in the MCF7 cells known to express relatively high levels of PADI4 [65] . Depletion of this protein had an effect inverse to that of HP1α depletion and resulted in decreased levels of HERV transcripts ( Figure 3E , Figure S1A and S1C ) . Levels of Satellite 2 transcripts ( but not α-Satellite transcripts ) were also decreased , suggesting a broad yet selective effect of citrullination on the silencing of repeats . PADI4 depletion also decreased expression of the immune genes TNFα , IL16 , and IL8 , as well as IL23 and IL1A , but not TGFß1 ( Figure 3F ) . The control pS2 gene known to be negatively regulated by PADI4 [44] , [45] , was , as expected , moderately stimulated . MCF7 cells are estrogen-responsive and an estradiol ( E2 ) treatment combined with an ionophore increases total levels of both PADI4 ( Figure S2A and [65] ) and H3cit8K9me3 modification ( Figure 4A , compare lanes 1 and 2 ) . In contrast , reduced PADI activity can be obtained with the specific inhibitor Cl-amidine [66] . This drug affects the nuclear PADI4 as illustrated by the decreased levels of H3Cit8K9me3 in MCF7 cells ( Figure 4A; compare lanes 1 and 3 ) . Thus , treatment with either E2/A23187 or Cl-amidine allowed us to control endogenous nuclear PADI activity at will . As in the PADI4-depletion experiments , treatment of the MCF7 cells with Cl-amidine reduced expression of HER V-H/env62 , ERVWE1 , and the selected cytokine genes , but not α-Satellite and TGFβ1 ( Figure 4B , black bars ) . Inversely , augmenting PADI4 activity by treating the MCF7 cells with E2 and ionophore resulted in a Cl-amidine-sensitive increase in expression of the same HERVs and cytokines ( Figure 4B , grey and white bars & S2B ) . Finally , we performed ChIP to follow the impact of endogenous nuclear PADI activity on the citrullination of histone H3 and the recruitment of HP1α to the LTRs of HERV-H and ERVWE1 ( HERV-W/LTR ) and the promoter of TNFα . These assays confirmed that stimulation of PADI4 activity with E2 and ionophore locally increases levels of citrullinated histone H3 , as detected with either anti-H3cit ( 2 , 8 , 17 ) or anti-H3cit8K9me3 antibody ( Figure 4C–4D and Figure S2C ) , while the levels of HP1α recruitment were markedly decreased ( Figure 4E , black and grey bars ) . These levels of HP1α occupancy were partially restored by further treating the cells with Cl-amidine ( Figure 4E , white bars ) , illustrating that this inhibitor can overcome the detrimental effect of excessive PADI activity . The level of HP1α recruitment at the pS2 control promoter region was low and was not substantially affected by changes in citrullination levels ( Figure 4C–4E ) . Taken together , these results demonstrate that HP1α and citrullination antagonistically regulate several immune genes and HERVs , and that this regulation is druggable . An inflammatory response can be induced in Jurkat T cells stimulated with an ionophore and the phorbol ester PMA . The stimulation of the Jurkat cells correlates with an eviction of HP1α from the promoter region of TNFα and IL8 ( Figure 5A ) , and also from HERV-H/LTR62 and HERV-W/LTR ( Figure S3A ) . The treatment also results in an approx . 6-fold increase of PADI4 accumulation ( Figure 5B ) and is expected to increase PADI activity as a consequence of the ionophore-induced calcium influx . We therefore used this system to determine whether increased PADI activity is associated with normal transcriptional activation of immune genes . Stimulation of the Jurkat cells resulted in a rapid and very transient recruitment of PADI4 to the promoters of TNFα and IL8 , and at HERV-H/LTR62 and HERV-W/LTR ( Figure S3B ) . This recruitment correlated with increase levels of H3Cit8K9me3 at these positions ( Figure 5C and Figure S3C ) . Finally , inhibition of PADI activity with Cl-amidine reduced the kinetic and the abundance of TNFα and IL8 mRNA accumulation in Jurkat cells stimulated by ionomycin and PMA ( Figure 5D and ) . Together , these observations suggest that PADI activity participates in the modification of the epigenetic landscape at the promoter of immune genes upon stimulation of T cells , and thereby facilitate the transcriptional activation of these genes . To investigate whether levels of HP1α recruitment and H3cit8K9me3 double modification at HERV and cytokine promoters were affected in MS , we collected PBMCs from 18 families , each family consisting of one MS patient and a genetically related healthy control ( Table S1 ) . The patients suffered from either relapsing-remitting ( n = 10 ) , or secondary progressive ( n = 8 ) MS . As PBMCs yield only minute amounts of chromatin , our analysis was restricted to the LTRs of the unique HERV-H locus LTR59 [67] and the unique HERV-W locus ERVWE1 [68] , and to the promoter of the cytokine TNFα . Consistent with previous observations [4] , [5] , [6] , [7] , transcription of these loci was significantly augmented in the MS patients ( Figure 6A–6B ) . ChIP assays revealed that recruitment of HP1α to the TNFα and the examined HERV promoters was significantly reduced in the MS patients compared to their genetically related healthy controls , while recruitment to a control promoter ( RPLP0 ) was unchanged ( Figure 6C ) . We also observed a significant correlation between HP1α levels on the TNFα promoter and HP1α levels on the LTRs of the HERVs , further suggesting that a single pathway regulates HP1α binding to both types of transcription units ( Figure S4 ) . In ChIP assays , the anti-H3cit8K9me3 antibody was functional , but with a relatively poor sensitivity . Therefore , our analysis was restricted to the 9 families ( 9 patients and their respective healthy relatives ) from whom we had the most abundant material . In these samples , levels of H3cit8K9me3 at the promoters of HERV-W/ERVWE1 and TNFα were significantly increased in the patients when compared to the genetically related healthy controls ( Figure 6D–6E ) . On HERV-H/LTR59 , levels of H3cit8K9me3 also appeared increased in the patients , but the p value associated with this data ( 0 . 06 ) is above the significance level of 0 . 05 . We finally questioned whether PADI4 could be linked to the increased levels of H3cit8K9me3 . To this end , we examined PBMCs collected from the 18 families described above . Analysis of these samples by RT-PCR showed that PADI4 mRNA levels were significantly elevated in MS patients compared to the genetically related healthy controls ( approx . 1 . 5-fold , Figure 6F ) . Altogether , these experiments showed that in the patients , increased expression HERV-W/ERVWE1 and TNFα transcripts and decreased recruitment of HP1α at their promoter region is accompanied by a local increase in H3R8 citrullination and a moderate up-regulation of PADI4 expression . In this report , we show that dependence on HP1α-mediated silencing is a common denominator between cytokines and HERVs , both expressed at abnormally high levels in T cells from MS patients , and we suggest that a decreased efficiency of the HP1-mediated silencing may participate in the pathological deregulation of these transcription units . In this context , we find that one source of defective HP1-mediated silencing is citrullination of H3R8 . This histone modification reduces the affinity of the chromo domain of the HP1 proteins to the methylated histone H3K9 residue and thereby defines a novel mechanism regulating HP1-binding to chromatin . Using an antibody specifically recognizing H3cit8K9me3 , we show that this double modification is induced in the presence of elevated levels of PADI4 , the only known nuclear peptidylarginine deiminase . Interestingly , when an inflammatory response is induced in Jurkat T cells , expression of PADI4 is increased and levels of H3cit8K9me3 rise at the promoters of the immune genes IL8 and TNFα . Under these conditions , inhibiting PADI activity with the chemical inhibitor Cl-amidine results in reduced kinetic and amplitude in the activation of the two immune genes . These observations show that PADI activity and citrullination of histone H3 are required for normal activation of immune genes and define PADI4 as a novel regulator of cytokine expression . We speculate that H3cit8 , together with other histone modifications such as H3S10 and H3S28 phosphorylation participate in creating an epigenetic landscape favorable for the transcriptional activation of a subset of immune genes . PADI4 activity could also be artificially increased in MCF7 cells treated with estradiol and an ionophore . This allowed us to show that abnormally elevated levels of PADI activity result in transcriptional stimulation of several immune genes . Consistent with this , PBMCs collected from MS patients ( and compared to healthy relatives ) showed in average increased expression of TNFα , increased levels of H3cit8K9me3 at the promoter of this gene , and increased expression of PADI4 . These observations strongly suggest that inappropriate activity of PADI4 can participate in the deregulation of immune genes relevant for MS ( see model Figure 7 ) . We here note that the estrogen/ionophore treatment inducing PADI4 expression in MCF7 cells also stimulated production of this enzyme in Jurkat T cells ( data not shown ) . We therefore speculate that PADI4 could be involved in the activating effect of estrogen on TNFα expression observed in T cells under some conditions ( [69] and references therein ) and could thereby play a role in the higher incidence of MS in females [70] . Other enzymes affecting the affinity of HP1 proteins for chromatin may also be good candidates for an implication in MS . For example , Jak2 that is expressed at increased levels in MS Th17 cells [71] also cause exclusion of HP1α from chromatin by phosphorylating H3Y41 , a residue contacted by the C-terminal region of the HP1 proteins [72] . Along the same lines , we note that levels of arginine methylation of myelin basic protein MBP is increased in MS patients [73] , while we find that methylation of H3R8 reduces affinity of HP1α for the neighboring methylated H3K9 approximately 10-fold . Possibly , the same arginine methylases may be involved in the modification of both MBP and histones . In addition , PADI4 has earlier been described as involved in arginine demethylation , although methylated arginines are rather poor substrates for this enzyme in vitro [44] , [45] , [47] . Methylation and citrullination may therefore allow for a gradual activation of HP1α target genes in response to external stimuli . The fact that PADI4 is a regulator of cytokines that can be either positively regulated by cellular stimuli or negatively regulated by specific inhibitors provides yet unexplored avenues to the control of inflammation , and in the case of MS , molecules such as Cl-amidine may potentially allow restoring chromatin-mediated repression of over-activated cytokine genes . While HP1 proteins are best described as heterochromatic silencers and suppressors of variegation , our observations confirm that these proteins are also highly relevant for the transcriptional control of inducible genes that require a transient phase of silencing . The sharing of regulatory mechanisms between euchromatic cytokine genes and repeated sequences such as HERVs suggests that many bridges may exist between active and inactive chromatin , and that there is a continuum and not a clear-cut boarder between euchromatin and heterochromatin . Therefore , probing the status of heterochromatic silencing as well as its defects may provide much new insight on the transcriptional programs in which cells are engaged . Blood samples were collected from each participant after informed consent as approved by the local Danish ethical committee . The study population consisted of 36 subjects , encompassing 18 MS patients clinically diagnosed for MS and fulfilling the diagnostic criteria of Poser et al . , 1993 [75] and 18 unaffected ( healthy ) first or second degree relatives , one for each of the MS patients . The participants were from a homogenous population ( Caucasian , Northern European descent ) . Venous blood was drawn and processed on the same day in our laboratory . The clinical and demographic data of each participant are summarized in Table S1 . The mean age of both the MS patients and their unaffected relatives was 52 years . The gender ratios for MS patients ( 11 female/7 male ) and unaffected relatives ( 9 female/9 male ) were also comparable . Peripheral blood mononuclear cells ( PBMCs ) were prepared by standard Isopaque-Ficoll centrifugation . The separated cells were cryo-preserved in RPMI with addition of 20% inactivated human serum ( HS ) and 10% DMSO , at −135°C until use . For the assays , PBMCs were thawed and cultured for 24 h in RPMI-1640 with 10% inactivated human serum , and 100 Uml-1 penicillin-streptomycin at 37°C in a 5% CO2 incubator prior to use . For each family , PBMCs from the patient and the control individual were analyzed at the same time . The data were analyzed by using the software XLSTAT ( version 2010 . 5 . 06 , www . xlstat . com ) . When indicated in the text , Wilcoxon signed rank test [76] was used to determine whether a significant ( p<0 . 05 ) difference . Anti-H3cit8K9me3 antibody was produced in rabbits using a peptide coupled to KLH with the following sequence: ARTKQTA ( cit ) ( Kme3 ) STGGKAPRC . Anti-PADI4 ( ChIP: P4749; immunoblots: ab50332 ) , anti-H3cit ( 2 , 8 , 17 ) ( ab5103 ) , anti-H3 ( ab1791 ) , and anti-H3K9me3 ( ab8898 ) antibodies were purchased from Abcam . Anti-HP1α ( ChIP: 1H5; immunoblots: 2G9 ) and anti-Brg1 ( 2E12 ) were from Euromedex . Calcium ionophore A23187 ( used on HEK293 and MCF7 cells ) and ionomycin ( only ionophore tolerated by Jurkat cells ) , and 17-ß-estradiol ( E2758 ) were purchased from Sigma . DNA was labeled with 4′ , 6-diamidino-2-phenylindole ( DAPI; Invitrogen ) at a concentration of 150 ng . ml−1 . The peptide ARTKQTARKSTGGKAPRC was used for competition , overlay , and surface plasmon experiments , either unmodified or with K9me3 , R8meK9me3 , cit8K9me3 , or cit8 modifications . Peptides were carefully quantified by amino acid analysis , and the presence of the modifications was confirmed by mass spectrometry . ChIP was carried out essentially as previously described [74] , with minor alterations . MCF7 or Jurkat cells , or PBMCs were cross-linked in phosphate-buffered saline ( PBS ) containing 1% formaldehyde ( Sigma ) for 10 min at room temperature . The crosslinking reaction was quenched with PBS containing 125 mM glycine , followed by three washes with ice-cold PBS . The chromatin was fragmented by sonication to produce average DNA lengths of 0 . 5 kb . After ChIP , the eluted DNAs were detected by quantitative PCR using the primers listed in Table S2 . Levels of histone modifications are expressed as % of H3 , and levels of HP1α are expressed relatively to the signal obtained for ChIP using non-immune IgGs . Values are averaged from 3 independent experiments . Real-time SPR assays were performed at 25°C in PBS . GST-HP1α was covalently coupled to a CM5 sensor chip , using a Biacore 2000 instrument and an Amine Coupling Kit ( GE Healthcare ) , achieving three different immobilization densities ( Rimmo ) of 3500 , 6500 , and 15000 resonance units ( RU; 1RU ≈1 pg . mm−2 ) . On the remaining flow cell , 5700 RU of GST were immobilized to prepare a reference surface . A series of 10 concentrations of peptides ( 50 nM–25 µM for H3R8K9me3 , 200 nM–100 µM for the H3cit8K9me3 , H3cit8K9 , and unmodified peptides ) were injected for 2 min over the GST-HP1α and GST surfaces at a flow rate of 50 µL . min−1 . After following the dissociation for 5 min , the surfaces were regenerated with a 3-min wash of 2 M NaCl , and two 15-sec washes with 10 mM glycine-HCl ( pH 1 . 5 ) and 0 . 05% SDS . The association and dissociation profiles were double-referenced using the Scrubber 2 . 0 software ( BioLogic Software ) ( i . e . both the signals from the reference GST surface and from blank experiments using PBS instead of peptide were subtracted ) . The steady-state SPR responses ( Req ) were plotted against the peptide concentration ( C ) and fitted according to the following equation: ( 1 ) where Kd is the equilibrium dissociation constant of the peptide/GST-HP1α interaction and Rmax the maximal binding capacity of GST-HP1α . The percentage of bound HP1α sites was determined as follows: ( 2 ) MCF-7 and HEK293 cells were cultured in Dulbecco's modified Eagle's medium ( DMEM , Gibco BRL ) , and Jurkat cells were cultured in RPMI-1640 , all with 10% decomplemented fetal bovine serum ( FBS ) and 100 U . ml−1 penicillin-streptomycin at 37°C in a 5% CO2 incubator . MCF7 cells were treated with 200 nM of estradiol ( E2 ) for 24 h , washed three times with 1× PBS and then incubated for 30 minutes in Locke's solution ( 10 mM HEPES . HCl , pH 7 . 3 , 150 mM NaCl , 5 mM KCl , 2 mM CaCl2 , and 0 . 1% glucose ) supplemented with 5 µM A23187 ( C7522 , Sigma ) . Jurkat cells were treated with phorbol myristate acetate ( PMA ) at 40 nM . Ionophores were used at 1 µM . PADI4 inhibitor , Cl-amidine ( from Bertin Pharma ) was dissolved in 1× PBS as a 50 mM stock solution . MCF7 and Jurkat cells were treated with 200 µM Cl-amidine in complete cell culture medium for exactly 72 h . The cDNAs coding for either WT PADI4 , hyperactive PADI4 ( Mut+ ) [62] , or hypoactive PADI4 ( Mut− ) [63] were inserted into the pi_tk_hygro vector for retroviral delivery [77] . Two days after transfection using FuGENE ( Roche ) , the medium of packaging HEK293 cells was filtered on a 0 . 45 µm filter ( Millipore ) and supplemented with polybrene ( AL-118 , Sigma ) at 100 µg . ml−1 . This medium was used to infect host HEK293 cells ( 3 consecutive infections ) followed by selection with hygromycin ( H3274 , Sigma ) at a final concentration of 100 µg . ml−1 . HP1 small interfering RNAs ( siRNAs ) were described previously [35] . PADI4 ( J-012471-05 ) and Glyceraldehyde-3-phosphate dehydrogenase control siRNAs were purchased from Dharmacon . Cells were harvested 72 h after transfection with DharmaFECT 1 ( T 2001-03 ) . Total RNA from PBMCs and MCF7 cells was extracted with RNeasy ( Qiagen ) and quantified with an ND-1000 ( Nanodrop ) . After DNase treatment ( Roche ) , reverse transcription was performed using SuperScript III ( Invitrogen ) and random hexanucleotides according to the manufacturer's instructions . Complementary DNA was quantified by RT-qPCR as previously described [35] . PCR primers are listed in Table S2 . Proteins were extracted as described previously [35] and detected by Western blotting . Acid extraction of histones was conducted as described by Shechter , et al . , [78] . Immunofluorescent labeling was performed in MCF7 cells after sequential treatment with 200 nM of estradiol ( E2 ) for 24 h in culture media followed by 5 µM A23187 for 30 min in Locke's solution . Cells were permeabilized in ice cold CSK ( 20 mM PIPES pH 6 . 8 , 200 mM NaCl , 600 mM sucrose , 6 mM MgCl2 , and 2 mM EGTA ) 0 . 5% Triton X-100 ( v/v ) , 0 . 1 mM PMSF for 30 sec . Cells were fixed with CSK-3 . 7% paraformaldehyde at room temperature ( RT ) for 10 min , then blocked in PBS-0 . 05% Tween-20 ( v/v ) , 10% ( v/v ) FBS for 30 min at RT . In peptide competition experiments , primary antibodies were pre-incubated with 1 µg of indicated histone H3 peptides or without peptide for 30 min . Cells were incubated with antibodies at 4°C overnight and then coverslips were washed three times in PBS-BSA 0 . 5% ( w/v ) . Coverslips were incubated with secondary antibodies for 1 h at room temperature protected from light , before being washed three times in PBS-BSA 0 . 5% ( w/v ) , and once in PBS , before final staining with DAPI . Imaging was conducted on an Axiovert 200 M microscope ( Zeiss ) coupled with an Apotome with Axiovision 4 . 7 ( Zeiss ) .	In patients suffering from Multiple Sclerosis ( MS ) , T lymphocytes express abnormally high levels of a subset of cytokines . The same cells also transcribe a series of vestigial retroviral sequences normally silenced by chromatin factors . In this study , we have searched for regulatory mechanisms shared between the cytokines and the retroviral sequences . We find that the repressor protein HP1α is present on the promoter of both types of transcription units in normal cells and that the recruitment of this protein to these promoters is decreased in MS patients . Furthermore , we show that the delocalization of HP1α from these promoters can be caused by citrullination of histone H3 , and we provide evidence indicating that levels of this histone modification is augmented in MS patients . Together our data provide a possible explanation for the simultaneously increased transcriptional activity of cytokines and endogenous retroviruses in MS-patient T lymphocytes and suggest that inhibitors of the enzyme responsible for the increased citrullination of histone H3 could help restore normal levels of cytokine activity in the patients .	[ "Abstract", "Introduction", "Results", "Discussion", "Materials", "and", "Methods" ]	[ "biology" ]	2012	Citrullination of Histone H3 Interferes with HP1-Mediated Transcriptional Repression
Computational models of classical conditioning have made significant contributions to the theoretic understanding of associative learning , yet they still struggle when the temporal aspects of conditioning are taken into account . Interval timing models have contributed a rich variety of time representations and provided accurate predictions for the timing of responses , but they usually have little to say about associative learning . In this article we present a unified model of conditioning and timing that is based on the influential Rescorla-Wagner conditioning model and the more recently developed Timing Drift-Diffusion model . We test the model by simulating 10 experimental phenomena and show that it can provide an adequate account for 8 , and a partial account for the other 2 . We argue that the model can account for more phenomena in the chosen set than these other similar in scope models: CSC-TD , MS-TD , Learning to Time and Modular Theory . A comparison and analysis of the mechanisms in these models is provided , with a focus on the types of time representation and associative learning rule used . Classical conditioning theories aim to understand how associations between stimuli are learned . Ever since Pavlov [1] the process of association formation has been understood to depend crucially on the temporal relations between stimuli [2 , 3 , 4] . Yet , classical conditioning theories have so far struggled to work when time is taken into account as an attribute of the stimulus representation . The study of time as a mental representation is the object of a separate area of study known as interval timing . Interval timing theories have produced a rich variety of time representations [5 , 6 , 7 , 8 , 9] , and therefore are a natural place to look for ways to integrate time into classical conditioning . In this paper we first analyse previous efforts in this direction before introducing a new hybrid classical conditioning and timing model . The process of association formation is understood to be of fundamental survival value for both human and non-human animals . Prediction , which forms the core of classical conditioning , allows the organism to adapt to significant events in its surroundings . A prototypical experiment in classical conditioning , a type of associative learning , involves a neutral stimulus and an unconditioned stimulus ( US ) which is capable of eliciting an unconditioned response ( UR ) . After repeated pairings of both stimuli in a specified order and temporal distance , the neutral stimulus comes to elicit a response similar to the UR . This response is called the conditioned response ( CR ) and the neutral stimulus is said to have become a conditioned stimulus ( CS ) . Classical conditioning theories typically conceptualize this process as the formation of a link ( association ) between the internal representations of CS and US . Their basic building blocks are [10 , 11]: ( a ) the representations of stimuli , and ( b ) a learning rule to update the association weights between these representations . Although most theories do not attempt to find neurophysiological correlates , these constructs are nonetheless commonly assumed to be instantiated by ( a ) neural activity in the form of spike rates , and ( b ) synaptic plasticity [12 , 13 , 14] . These have found some support in the neuroscientific literature , particularly studies of the role of dopamine in reward prediction [15 , 16 , 17 , 18] . However it is important to note that there is still no widely accepted complete neural mechanism for classical conditioning and that most theories stay at the computational level of explanation . Stimulus representations are generally thought of as neural activation that is elicited by the stimulus , which may linger for a short time as a ‘trace’ after stimulus offset . Representations are commonly one of two types: molar or componential . Molar ( or elemental ) trace theories treat the stimulus as a single conceptualized unit whose activity is usually assumed to peak quite early following stimulus onset , and then gradually decrease [19 , 20 , 21 , 22 , 23 , 24] . In contrast , componential trace theories break down the CS representation into smaller units , each capable of being associated with the US , with some units more active early during the CS and others late , but all leaving a trace after activation [25 , 26 , 27 , 28] . Learning rules may be classified according to different criteria . An important period in the recent history of the field gave rise to one of these criteria . Prior to 1970’s conditioning used to be rooted in the stimulus-response tradition , which attributed crucial importance to the temporal pairing , or contiguity , of stimuli for the development of associations . The linear operator learning rule [19] is one of the products of that period . In the late 1960’s and early 1970’s important experimental discoveries using compound stimuli , that is , a stimulus formed by combining other individual stimuli , showed the contiguity view to be incomplete [29 , 30] . These compound experiments indicated that the formation of associations also depended on the reinforcement history of the individual elements forming the compound stimulus . This led to the development of new learning rules [31 , 32 , 33] capable of combining individual reinforcement histories in compounds , which the linear operator rule cannot . The first , and arguably still the most influential , of these learning rules is the Rescorla-Wagner [RW , 31] . It has become famous for being the first model able to provide an account for the blocking effect [34] , where a novel CS does not become associated with the US if it is reinforced only in compound with a previously conditioned CS . The CR is usually not a single event . Organisms time their responses so that they emerge gradually during the duration of the CS and reach maximum frequency or intensity around the time of reinforcement . Interval timing theories have attempted to provide an account for this timing of the CR . One of the fundamental properties of timing behaviour is that it is approximately timescale invariant , i . e . the whole response distribution scales with the interval being timed [35 , 36] . One of the consequences of timescale invariance is that the coefficient of variation , that is the standard deviation divided by the mean , of the dependent measure of timing is approximately constant . A number of timing models have put forth explanations for timescale invariance and other timing properties ( how time is encoded , how it is stored in memory and how it gets translated into behaviour ) by recourse to an internal pacemaker . The most influential pacemaker-based timing theory to date is Scalar Expectancy Theory [SET , 5 , 37] . The pacemaker is supposed to mark the passage of time by emitting pulses . These pulses can be gated to an accumulator via a switch which closes at the start of a relevant interval and opens when the interval is finished . The accumulator count is kept in working memory . At the end of the interval the current count is transferred to a long-term reference memory . Behaviour is guided by the action of a comparator which actively compares the count in working memory to the one retrieved from reference memory . In spite of the considerable overlap , interval timing and classical conditioning are not easily integrated . Most conditioning theories are trial-based , that is they consider the trial as the unit of time . A trial is generally taken to be the state where a CS is present ( or CSs in compound ) and which may or may not contain a US ( or USs ) . The most influential model in this category is the Rescorla-Wagner [RW , 31] . In order to account for different stimulus durations , trial-based theories like RW must resort to some sort of time discretization , usually by subdividing the trial into ‘mini-trials’ . Each mini-trial is treated as a trial in its own right , which are then used to update associative links . This gives rise to the problem of deciding on a particular discretization . Also , given that humans experience time passing as a continuous flow , it is unlikely that animals discretize their conditioning experience in such a way . A more realistic approach to timing is taken by real-time theories . These theories attempt to formalize the concept of a continuous flow of time . The Temporal Difference model [TD , 38 , 39] was one of the earliest and still most influential real-time classical conditioning model . It may be thought of as a real-time version of RW . When used with stimulus representations such as the Complete Serial Compound [CSC , 40] , Microstimuli [MS , 28 , 41] and the Simultaneous and Serial Configural-cue Compound [SSCC , 42] it is capable of reproducing some timing phenomena like the gradual increase in anticipatory responding that occurs before a signalled reinforcer , and the lower response rates observed during longer CSs . However , only MS-TD has a time representation capable of approximating the most fundamental property of timing , timescale invariance . Another issue with the stimulus representations for TD is that their approach to timing resembles the strategy used by trial-based models , i . e . they all split the stimulus into a number of smaller units or states , the number of which being directly proportional to the duration of the stimulus . Given that conditioning is observed in a timescale that ranges from milliseconds to hours [43 , p . 189] this can lead to a very high number of units being required . The stimulus as a whole no doubt is a complex entity , and the brain may be employing a large number of neurons to represent it , but to dedicate so many resources only for timing might not be the most energy-efficient strategy . Also , TD and its stimulus representations do not usually account for a change in timing that is not tied to reinforcement . Animals time the occurrence of different events , such as onset and offset of stimuli [see for example 44] , but TD usually only allows for the timing of rewards . On the other hand , timing models have made even fewer attempts at integrating aspects of classical conditioning . A notable exception is the Learning to Time [LeT , 7 , 45] model . It represents the passage of time by transitioning between internal states according to a stochastic pacemaker , an idea borrowed from an earlier timing model called the Behavioural Theory of Time [6] . Learning takes place by associating reinforcement presentation with the current internal state according to the linear operator , a standard classical conditioning rule . LeT offers an account of the basic dynamics of association formation , but it cannot explain cue-competition phenomena like blocking . In a blocking procedure , a CS is first paired with a US until a CR is acquired . The same CS is then presented together with a novel CS and both are paired with the US for a few trials . If the novel CS is now presented alone it elicits little or no responding , and so it is said to be blocked by the first CS . LeT’s learning rule , the linear operator , has largely been supplanted by RW in classical conditioning modelling because it cannot explain cue-competition phenomena . Like TD , LeT also employs a representation that requires as many units as time-steps , making it a resource-intense model . Modular Theory [MoT , 46 , 47] is a timing model which because of its explicit goal of integrating timing and learning may be called a hybrid theory . MoT has introduced novelties that allow it to account for some aspects of the dynamics of classical conditioning that LeT cannot . Its architecture is different than the connectionist one ( states or units connected by modifiable links ) assumed by RW , TD and LeT . Instead , it uses a more cognitive architecture , with separate information processing stages that deal with perception , memory and decision . It postulates two separate memories: a pattern memory which stores CS durations , and a strength memory which stores the associative strength between each pattern memory and the US . This separation allows MoT to deal with more complex situations involving the dynamics of learning during acquisition and extinction . However , MoT also relies on the linear operator to update its strength memory , which , like LeT , prevents it from accounting for cue-competition phenomena . Although the models mentioned above , namely TD , LeT and MoT , have accomplished a great deal in terms of bringing together timing and conditioning , they each have their different strengths and weaknesses as we have touched above . In this paper we introduce a model that tries to address some of these weaknesses while preserving the strengths . More specifically , the model has the following strengths . It represents time in real-time . Like MoT and unlike LeT and TD , its time representation does not require an arbitrary large number of units or states . Similarly to TD but unlike LeT and MoT , it uses a learning rule that preserves the main features of RW which allow it to account for compound phenomena . It can time the onset and offset of all stimuli , not only of rewards , and store a memory for each . It includes two update rules: one for timing that is updated by time-markers , and another for associations that is updated by the US . Hence , simple stimulus exposure causes the model to learn and store its duration . This capability is not present in models that depend only on an associative learning rule to also learn about time , such as TD and LeT . This new model is essentially a way to connect one of the most influential classical conditioning theories , the Rescorla-Wagner model [31] , with a recently developed timing theory called Timing Drift-Diffusion Model [TDDM , 48 , 49] . The TDDM is based on the drift-diffusion model , widely used in decision making theory , and it provides an adaptive time representation that has commonalities with pacemaker-based models like SET and LeT [50] . These models postulate the existence of a pacemaker that emits pulses at a regular rate , which are then counted to mark the passage of time . To preserve timescale invariance they either postulate a specific type of noise in the memory saved for intervals and a ratio-based decision process ( SET ) or adapt the rate of pulses ( LeT ) . The TDDM takes the latter route but sets a fixed threshold on pulse counting . To emphasize the unification of these two theories we call our proposal the Rescorla-Wagner Drift-Diffusion Model ( RWDDM ) . We evaluate RWDDM based on how well it can simulate the behaviour of animals in a number of experimental procedures . Many classical conditioning phenomena have been identified which collectively represent a significant challenge for any single model to explain . A recent list [51] has compiled 12 categories , which include acquisition , extinction , conditioned inhibition , stimulus competition , preexposure effects , temporal properties , among others . Of particular interest to a theory of timing and conditioning are phenomena that involve elements of both timing and conditioning . As we detail later , we have searched the literature for documented effects that can challenge the main mechanisms embodied in RWDDM . We proceed by first introducing the new model . We compare its formalism with four models that have similar scope , namely CSC-TD , MS-TD , MoT and LeT . In the results section we present the phenomena we will simulate , followed by the results of our simulations , and compare them to the current explanations given by LeT , MoT and TD . Among the theories capable of providing an account of both timing and conditioning , arguably four stand out for their scope or influence . They are CSC-TD , MS-TD , LeT and MoT . TD has been developed primarily as a learning model , without the explicit intention of addressing timing . It may be visualized as a real-time rendition of the RW rule . Its basic learning algorithm , is given by: V t ( x t ) = ∑ i w t ( i ) x t ( i ) , ( 11 ) δ t= λ t - ( V t ( x t - 1 ) - γ V t ( x t ) ) , ( 12 ) w t + 1= w t + α δ t e t ( 13 ) where Vt is the US prediction at time t , formed by a linear combination of the weights w ( i ) and the CS representation values x ( i ) . This update algorithm is performed at each time step , and not only at the end of a trial like RW and RWDDM . Another important difference is that eq ( 12 ) computes a difference between the current US value and the temporal difference between predictions . Hence , δt > 0 if the US is higher than this temporal difference in prediction , and δt < 0 if the US is lower . The constant 0 < γ < 1 is termed a discount factor . Eq ( 13 ) updates the weights for the next time step . The vector et stores eligibility traces , which are functions describing the activation and decay of representations xt . The three most common eligibility traces used are: accumulating traces , bounded accumulating and replacing traces . These three types accumulate activation in the presence of the CS and discharge slowly in its absence , the first accumulates with no upper bound , the second only until the upper bound and the third is always at the upper bound whilst the CS is present [39 , pp . 162-192] . The richness of TD’s timing account relies on the choice of CS representation x . The Complete Serial Compound representation [CSC , 40] postulates one CS element x ( i ) per time unit of CS duration . Each element is only switched on at its activation time unit , and then decays afterwards following its choice of eligibility trace e ( i ) ( usually an exponential decay function ) . This componential representation , which increases in size linearly with CS duration , should be contrasted with RWDDM’s molar representation ( eq ( 8 ) ) which requires only one element . CSC may be called a time-static representation , whilst RWDDM is a time-adaptive representation , with a rule to change its structure based on a change in time ( eqs ( 6 ) and ( 8 ) ) . CSC-TD also lacks any mechanism to explain timescale invariance of the response curve , which is present in RWDDM . A modification of CSC has recently been developed , the Simultaneous and Serial Configural-Cue Compound [SSCC , 42] . SSCC-TD formalizes the idea that when multiple stimuli are presented together in time , a configural cue–a novel stimulus that is unique to the current set of present stimuli–is formed . SSCC follows on the CSC representation , but , unlike any other TD model , it allows for the representation of compounds and configurations of stimuli . Because SSCC-TD is a real-time model , it also allows for the simulation of CR timing during compounds and configurations . However , its approach to timing is still the same as CSC , i . e . it breaks down the stimuli into a series of elemental units which are activated in series . Therefore , with respect to timing only we will consider SSCC to belong to the family of CSC representations . The Microstimuli representation [28 , 41] introduced a more realistic description of time . Unlike CSC , it uses a fixed number of elements x ( i ) per stimulus . The ith microstimulus is given by: x t ( i ) = 1 2 π exp ( - ( y t - i / m ) 2 2 σ 2 ) · y t ( 14 ) where m is the total number of microstimuli , y is an exponentially decaying time trace set at 1 at CS onset . It will be noted that a microstimulus is a Gaussian curve modulated by the decaying trace yt . The set of microstimuli generated by the CS will then give rise to partially overlapping Gaussians , with decreasing heights and increasing widths across time . The fact that only a fixed number of microstimuli are required per CS is an improvement to the potentially large numbers of elements in CSC . The MS representation tries to capture the idea that as time elapses , the stimulus leaves a more diffuse and faint impression . However , even though it is more realistic than CSC , it still lacks a mechanism to produce exact timescale invariance . Learning to Time is primarily a theory of interval timing which can also account for some aspects of conditioning . Here we will deal with its most recent version in [45] , which differs somewhat from the earlier version in [7] . Its CS representation resembles CSC in postulating a long series of elements ( or states ) that span the whole stimulus duration . Unlike CSC , it transitions from state to state at a rate that varies from trial to trial , and that is normally distributed . Hence , time during a trial is represented as a noiseless linear increase from states n = 1 , 2 , 3 , … ( one per time-step ) at a fixed rate . This linear time representation resembles the linear accumulator in RWDDM , except that the latter has noise built into the linear accumulator , whilst LeT assumes noise only at the intertrial level . Each state n is associated with the US via an associative link . At the end of a trial , the strength w of these links are updated as follows: Note that unlike RWDDM’s associative update rule , eqs ( 15 ) to ( 17 ) do not include a summation term . This places a severe limitation on the ability of LeT to deal with compound conditioned stimuli . LeT’s strength lies on its being able to explain timescale invariance of the response curve . Machado and colleagues [45] showed that it is possible to derive timescale invariance using only the assumption of intertrial normality of state transition rate . Finally , LeT assumes that responses are emitted at a constant rate if the current active state has associative strength w ( n ) greater than a threshold θ . The fact that responding depends on the associative strength of the current state , and that this strength only changes with US associations , prevents LeT from accounting for changes in timing that are not related to US occurrence . For example , there is evidence that animals learn the timing of a preexposed CS [70] and are sensitive to changes in timing during extinction [71] , two situations that do not involve the occurrence of a US . Modular Theory is another primarily timing theory that can also deal with some aspects of conditioning . It treats the onset of a stimulus as signalling a time expectation to reinforcement . Its time representation T is , like LeT , an accumulator that increases linearly with time t , T = ct , where c is a constant . When reinforcement is delivered the current reading from the accumulator is stored in what is called pattern memory . Pattern memory is updated at each trial n according to m ( n ) = m ( n - 1 ) + α ( T * - m ( n - 1 ) ) ( 18 ) where α is a learning rate and T* is reinforcement time . Eq ( 18 ) may be contrasted to ( 6 ) from RWDDM . The main difference is that pattern memory in MoT stores a moving exponential average of intervals , whilst the slope in RWDDM stores a moving exponential harmonic average of intervals . However , both models are similar in that they can potentially time the occurrence of any event , not only rewards . MoT’s pattern memory and RWDDM’s slope can be made , for example , to adapt to mark the end of stimuli that are not necessarily paired with a reward . A stochastic threshold b is used to mark response initiation . The threshold distribution is set so as to yield timescale invariance of the response curve . Its mean , B , is a fixed proportion of the value in pattern memory , B = km ( n ) , where k is the proportionality constant , and its standard deviation is γB , where γ is the coefficient of variation of B . Hence , the coefficient of variation of the threshold , i . e . of response initiation , is constant for all intervals , which is the timescale invariance of the response curve . RWDDM derives timescale invariance of response curve from noise in the accumulator ( eq ( 2 ) , not from the threshold . This account of time from MoT is an instantiation of Scalar Expectancy Theory , arguably one of the most successful timing models to date . Being a purely timing theory , SET does not address associative learning directly , so it does not have a rule for changes in association between stimuli . MoT bridges this gap by adding a rule to update what is termed strength memory , w ( n ) . Strength memory holds the associative strength between stimulus and reinforcement . The rule consists of a linear operator: Δ w ( n ) = { β e ( 0 - w ( n - 1 ) ) if US is absent , β r ( 1 - w ( n - 1 ) ) if US is present , ( 19 ) with β a constant that can determine different rates of update for acquisition ( βr ) and extinction ( βe ) . Eq ( 19 ) may be compared with ( 9 ) . Note that , unlike RWDDM , eq ( 19 ) does not contain the summation term from RW based rules . MoT also includes a rule for response rate that is more realistic than RWDDM’s given by ( 10 ) . It is partly derived from an empirical analysis of real-time responding in animals . We refer the interested reader to [46] for a fuller description . We will only mention here that MoT generates a two-state response pattern , low and high . The transition between states is determined by the crossing of threshold B , and the high state is proportional to strength memory w ( n ) . Other theories exist which are similar in scope to CSC-TD , MS-TD , LeT and MoT . Two notable examples are the Componential version of the Sometimes Opponent Process model [C-SOP , 72] and the Adaptive Resonance Theory—Spectral Timing Model [ART-STM 26] . C-SOP builds a CS representation based on two sets of elements , or components , one that includes elements activated as a function of time and another whose elements are randomly activated . Associative strength for each element is updated using the standard trial-based RW rule . Simulations in [72] have demonstrated that C-SOP can produce some degree of timescale invariance . ART-STM is a neural net with an input layer and one hidden layer , which allows it to explain nonlinear conditioning phenomena ( such as negative pattern ) that a single-layer RW neural net cannot . It employs a CS representation that is very similar to the microstimuli used in MS-TD , so it also shows a degree of timescale invariance . Other theories could be mentioned [for two inuential examples see 52 , 23 , 53] but we will limit the analysis to CSC-TD , MS-TD , LeT and MoT for two reasons: a ) these four models collectively embody most of the conditioning and timing mechanisms used in modelling these areas , and b ) our goal here is not to provide a comprehensive review , but rather focus on the mechanisms that are shared by our proposed model and the others . Table 1 summarizes the main mechanisms/features of the models described above . In terms of the type of time representation , it may be observed that the models fall roughly into two categories: ( a ) those that employ a chain of units or states activated sequentially ( CSC-TD , MS-TD , LeT ) , and ( b ) those that employ an accumulator ( MoT and RWDDM ) . Those in category ( b ) may be considered more economical both computationally and biologically , as they don’t require a number of units that increase with time . In terms of what the representations can time , two categories may be discerned: ( a ) those that time only rewards ( CSC-TD , MS-TD and LeT ) , and ( b ) those that can time any stimuli ( MoT and RWDDM ) . Models in category ( b ) have more flexibility to create a temporal map involving all stimuli present , including those not signalling reward . In terms of timescale invariance , the models are basically divided between those that can account for it ( MS-TD , LeT , MoT and RWDDM ) and the one that cannot ( CSC-TD ) . Finally , in terms of the type of associative learning rule used , models are divided between those that use a RW-type rule ( CSC-TD , MS-TD , RWDDM ) and those that use the linear operator ( LeT and MoT ) . The ones that use RW are wider in scope , being able to account for cue-competition phenomena , which form the core of classical conditioning . The main innovation of RWDDM over its predecessors is the combination of a noisy linear accumulator for timing with the RW rule for associative learning . As Table 1 shows , linear accumulator theories are the only ones in our sample of the models that can fully account for timescale invariance . But because they rely on the linear operator rule , they cannot account for cue-competition and other compound stimuli phenomena in conditioning . Therefore RWDDM extends the application of the linear accumulator to compound stimuli , covering a wider range of conditioning phenomena . In summary , the model we propose is , to the best of our knowledge , the only one that unites the flexibility , computational economy and timescale invariance of the linear accumulator as a time representation , to the RW associative learning rule , which accounts for many more conditioning phenomena than the linear operator . In the next section we evaluate the models against a number of phenomena in conditioning and timing . A conditioned response emerges gradually over the course of several trials where the CS signals the arrival of a US . If a measure of CR strength ( such as rate or magnitude ) is plotted against the number of trials , the shape and rate of this acquisition curve will depend largely on the CR and organism , but it usually follows a negatively accelerated curve [1 , 43] . Pavlov [1] believed timing of the CR would emerge only later in acquisition , through a process he described as inhibition of delay whereby the initial part of the CS would become inhibitory . Recent and more detailed analyses suggest that an estimate for the time to reinforcement is acquired very early in training , possibly even after one or two trials , although the expression of such estimation may not be observable until later in training [73 , 74 , 75 , 76] . If the CS no longer signals reinforcement , CR strength gradually decreases over the course of these extinction trials , until it finally disappears . If the CS is made to signal the US again , the CR returns , a process that is called reacquisition . It is a consistent finding that reacquisition is faster than acquisition [77 , 46 , 43 , p . 185] . Learning is loosely defined as an enduring change in behaviour as a result of experience . Acquisition of a CR is the most basic demonstration that classical conditioning is a form of learning . As such , all classical conditioning models provide an account of it . When a previously conditioned stimulus is no longer followed by reinforcement , the conditioned response gradually decreases . An important theoretical question for hybrid timing/conditioning models concerns what happens to the timing of responses in extinction . Using the peak procedure Ohyama and colleagues [78] found that although the maximum ( peak ) response rate decreased in extinction , peak time and sensitivity ( measured by the coefficient of variation ) remained virtually unchanged . Drew and colleagues [79] investigated the behaviour on extinction by changing CS duration between acquisition and extinction . Groups where the CS changed to a shorter or longer duration were compared to another where the duration did not change . They found that CS duration had little effect on the rate of extinction , with all groups taking about the same number of trials to achieve CR extinction . However , when the CS used in extinction was considerably longer ( 4 times ) than the one acquired , extinction was facilitated . Guilhardi and Church [71] performed a similar experiment ( experiment 2 ) and observed that when stimulus duration is changed from acquisition to extinction , the pattern of responding during extinction gradually shifts to the new duration over extinction trials . Following the same procedure , Drew and colleagues [80] also used partial reinforcement to slow down the rate of acquisition , and thus observe if response patterns really do shift gradually to the new duration . They confirmed that when CS duration was increased from acquisition to extinction , the within-trial response peak shifted gradually to the right over the course of extinction . When the CS was shortened , the results were not conclusive . Also , when CS duration was changed from training to extinction , the speed of extinction increased , but this appeared to be explained at least in part by the shifting of response patterns . In summary: a ) peak timing and CV are not altered in extinction when using a peak procedure , b ) changing the CS duration from training to extinction causes the within-trial response peak to shift to the new duration , and c ) changing the CS duration in extinction can speed up extinction , but this may be due to the shifting of the response peak and not to changes in associative strength . These results pose a challenge to the models analysed here . Out of CSC-TD , MS-TD , LeT and MoT , only MoT has a mechanism that would allow it to account for time change in extinction . When a subject is exposed to repeated and non-reinforced presentations of a stimulus it has never encountered before , this procedure is called preexposure . If reinforcement is subsequently paired with the preexposed CS , the initial rate of CR acquisition is usually lower compared to acquisition to a nonpreexposed stimulus , a phenomenon called latent inhibition [81] . The asymptotic level of conditioning , however , is not normally affected by preexposure [82] . Latent inhibition is an important representative of a class of phenomena involving latent effects . Collectively , these phenomena demonstrate that something is learned about the stimulus even when it does not signal reinforcement . Therefore , latent inhibition cannot be accounted by the Rescorla-Wagner model , since the theory only applies when there are changes in associative strength . A question relevant for real-time conditioning models is what happens to timing when a preexposed stimulus is conditioned . To answer this question , Bonardi and colleagues [70] used CSs of variable and fixed durations ( the variable duration CS had the same mean as the duration of the fixed CS ) to vary the temporal conditions between preexposure and conditioning phases . Latent inhibition was observed even when the temporal information from the two phases was different . Crucially , timing , as measured by the response gradient within a trial , appeared to improve in the preexposed CS even when the temporal information was different between the two phases . As alluded to above , latent inhibition cannot be accounted by the associative learning update rule used in RWDDM , the Rescorla-Wagner . However , we show here that RWDDM is compatible with the Pearce-Hall rule [33 , 83] , one of the most widely used models for explaining latent inhibition and other latent learning effects . We demonstrate that this modification maintains the basic framework of the RWDDM , and that it can account for latent inhibition and improved timing with preexposure . None of the other models analysed here can account for latent inhibition without modifications . Improved timing with preexposure could be accounted by Modular Theory , but not by the the current version of the other models . Arguably , the most important compound conditioning phenomenon is blocking . It is part of a class of cue competition and compound phenomena discovered in the late 1960s which challenged the view that conditioning was driven by the pairing , or contiguity , of CS-US . These results suggested that conditioning with compound stimuli was influenced by the reinforcement histories of the elements forming the compound [29 , 30] . This led to the development of a new generation of models that could account for those findings [31 , 32 , 33] . The rule we use , the Rescorla-Wagner , provides an explanation for blocking that is based on the summation term in eq ( 1 ) . In a blocking procedure a CS is first paired with a US in phase 1 of training . During phase 2 a novel CS is presented in compound with phase 1 CS and paired with the US for just a few trials . Subsequently , when tested alone the novel CS elicits less responding than if it had been trained in compound with another novel stimulus [34] . The previously reinforced CS is said to block the novel CS . The temporal information encoded by each CS has an effect on the amount of blocking observed . Schreurs and Westbrook [84] varied the ISI in the pre-training and compound phases , and observed less blocking when the durations were different in both phases than when they were the same . Barnet and colleagues [85] performed a similar experiment but with forward and simultaneous conditioning varying between phases , and also found that blocking was stronger when blocked and blocking CSs had the same temporal history . Jennings and Kirkpatrick [86] used compounds where the elements had different durations . They observed that a long blocking CS could block a co-terminating short Cs , but a short blocking CS failed to block a co-terminating long CS ( see rows 1 and 3 in Fig 7 ) . Amundson and Miller [87] performed four blocking experiments using trace conditioning . In two of them the blocking CS trace duration changed between phases , and blocking was not observed . In the other two experiments the trace duration was held fixed between phases , and the blocking and blocked CSs were presented serially and not in a compound ( see rows 2 and 4 of Fig 7 ) . Blocking was observed when the blocking CS followed the blocked CS , but not in the reverse condition . The studies reviewed above appear to show that changing the ISI of the blocking CS between phases may attenuate blocking . Another finding is the apparent asymmetry of blocking when the ISI of the blocking CS is kept constant between phases . Rows 1 and 2 of Fig 7 suggest that a long blocking ISI can block a short blocked ISI . Rows 3 and 4 suggest that a short blocking ISI does not block a long blocked ISI . As mentioned above , RWDDM can account for blocking because it uses the RW rule . The summation term in eq ( 1 ) formalizes the widely held view that a given US can only confer a limited amount of associative strength which CSs must compete for . Different theories exist that take other approaches to blocking [see for example 32 , 88 , 89] but among the ones analysed here ( for their ability to handle timing also ) only CSC-TD and MS-TD are equipped to deal with it . We show next that RWDDM can account for the blocking of a short CS by a long CS , and that by making the reasonable assumption of second-order conditioning it can also account for the lack of blocking of a long CS by a short CS . CSC-TD and MS-TD are also capable of providing an account of both blocking conditions . Learning occurs not only when a CS signals the occurrence of a US , but also when a CS signals the omission of a US . It is commonly assumed that the excitation caused by the former is counteracted by an inhibition produced by the latter . This is again formalized by the summation term in the RW rule . Conditioned inhibition is thus one of the phenomena that , together with blocking and other compound phenomena , challenged the contiguity interpretation of classical conditioning . A conditioned inhibition procedure involves reinforced trials with a CS , say A+ , intermixed with non-reinforced trials with a compound AB- . Conditioned responding develops during A+ trials but not during AB- . Hence , conditioned inhibition is a key conditioning phenomenon since it is also a form of discrimination learning . Conditioned inhibition poses higher technical challenges for a model of learning and timing as responses cannot be directly observed . To assess conditioned inhibition two types of measures are used [91]: summation and retardation tests . There are different procedures that can generate inhibition , so we refer here specifically to the inhibition produced by alternating A+ with AB- trials . CSA is called a training excitor , and CSB an inhibitor . In summation tests , this inhibitor is then presented together with a different excitor , and the inhibitor is said to pass the test if there is a decrement in responding compared to the excitor alone . In retardation tests , the inhibitor by itself is now paired with the US , and it is said to pass the test if acquisition is slower than with a neutral stimulus . Denniston and Miller [91] reviewed a series of studies that varied the durations of the training excitor and that between the inhibitor and the training excitor . The studies showed that conditioned inhibition is observed when the temporal relations between training and testing are preserved , and not otherwise . However , the studies reviewed by [91] used as measure of conditioned inhibition the time to resume drinking ( licking suppression ) when presented with the inhibitor . Williams and colleagues [92] investigated inhibition caused by reinforcement omission in excitatory conditioning , a more direct measure than licking suppression . In their experiments the inhibitor stimulus signalled the omission of one of two USs ( at 10 or 30 seconds ) that had been associated with the excitor stimulus . Using summation tests they found that the inhibitor would suppress responding only at the specific time of predicted US omission . Retardation tests confirmed that the time of US omission is encoded by the inhibitor . We show here that RWDDM can account for inhibition and its time specificity . CSC-TD and MS-TD are also equipped to deal with these results . MoT and LeT do not currently have the necessary mechanisms to explain inhibition . The two related phenomena described here are important in that they appear to challenge the summation effect . A common observation is that a compound of two previously conditioned CSs usually produces more responding than its individual components [93 , 43 , p . 204] . However , failure to obtain summation is also common [94 , 95] , and the precise conditions when it is observed or not is still a current topic of debate [see 24 , for a discussion] . Here we consider two cases in which summation was not observed and that RWDDM can offer a possible explanation . Aydin and Pearce [96] used an autoshaping procedure to condition pigeons to stimuli of 30 second duration . They observed little or no summation in compound trials , but a response curve with a consistent shift to the left . This earlier start of responding was observed even when one of the components was a neutral preexposed CS . The shift of the response curve to the left was termed disinhibition of delay . Meck and Church [44] performed an analogue experiment using the peak procedure . They trained rats to associate a light and a sound ( both of 50 second duration ) individually to a reinforcement , and then used a peak procedure to investigate what happens to timing in their compound . Like [96] they also found no summation and a shift to the left in the compound . Furthermore , rats also stopped responding earlier in the compound peak trials . Taken together , these results appear to show that in some cases summation is not observed , and responding in the compound starts earlier than in the component CSs . One possible explanation for this effect is that the subject fails to recognize the two individual components of the compound , what is known as generalisation decrement . If this is the case then it would be a performance effect , and not a learning phenomenon . We cannot rule this out , but we show that RWDDM’s trial variability in time estimation provides a plausible mechanism to explain this effect . The only other models in our analysis set that can account for this are MoT and LeT . The interval between CS onset and US onset is called Inter Stimulus Interval ( ISI ) . In general , measures of CR strength such as response frequency and amplitude decrease with longer ISIs [97 , 57 , 43] . Response timing is commonly analysed by using fixed interval ( FI ) schedules of reinforcement , which rely on a fixed ISI . It is a well established result that the peak in the response curve decreases with longer FIs [98 , 99] . However , the entire response curve approximately scales with FI . This is obtained by plotting different FI response curves as the proportion of maximum response strength versus the proportion to FI , a normalization procedure . The resultant normalized curves roughly superimpose [100 , 101 , 9 , 36] . This is sometimes called scalar timing , and it is one of the manifestations of the more general property of timescale invariance . CSC-TD does not have a mechanism to explain either timescale invariance or the ISI effect . Its more recent development , MS-TD , can approximately reproduce both timescale invariance and the ISI effect . LeT is also a timescale invariant model , but does not appear to show the decrease in response peak as a function of FI . MoT , at least in its earlier version [47] , can reproduce both the ISI effect and timescale invariance . Procedures where a stimulus signals reinforcement at more than one location in time are called mixed FI or two-valued interval schedules . A mixed FI involves only one CS which could be of short or long duration , and the subject has no way of knowing which duration it is currently experiencing until the US is delivered . Catania and Reynolds [98] conditioned pigeons in a mixed FI and reported a pattern of responding during the long CS that resembles a combination of two distinct FIs ( with two peaks ) when the separation between the intervals was in the ratio 8:1 but not at smaller proportions . Cheng and colleagues [103] found a similar result ( experiment 2 ) when the intervals were in 5:1 proportion , and Leak and Gibbon [104] showed that with intervals in the 8:1 proportion the scalar property ( measured by the CV ) holds approximately even for three-valued interval schedules . Whitaker and colleagues [105] ran three experiments with Mixed FIs in rats and found two peaks with the same CV when the proportion between the durations was greater than 4:1 , but not for smaller proportions . They also found that the peak height at the short duration was higher than at the long duration in most cases . Whitaker and colleagues [106] used intervals in the very small proportion 2:1 and still found two peaks that became more distinct when the short interval was presented more often than the long . These results are interesting because they challenge in particular models of timing . They have served to provide evidence in favour of SET , and against BeT and the first version of LeT [104] . Subsequently , they provided motivation for the development of the current version of LeT [45] . LeT can now account for the multiple response peaks in Mixed FIs , and their superimposition , but it cannot produce peaks with decreasing heights . Modular Theory has the necessary mechanisms to account for all the features of the data above . The TD models , MS and CSC , could both account for multiple peaks , but their account of superimposition would vary , with MS being superior than CSC . We show next that RWDDM can account for all features of the data in Mixed FIs . Schedules of reinforcement specify the conditions of reinforcement delivery . There are a number of different types of schedules , some are based on the time elapsed between reinforcements , some on the number of responses emitted between reinforcements , but there can be other possibilities . Of particular interest for a timing and conditioning model are the two most commonly used time-based schedules: variable and fixed interval . Variable Interval schedules of reinforcement ( VI ) consist in the delivery of a US following a CS that varies in duration from trial to trial . The CS durations are usually derived from an arithmetic or geometric sequence . In contrast , Fixed Interval schedules of reinforcement ( FI ) use a CS of fixed duration in all trials . Skinner and Ferster [107] reported that VIs tend to produce behaviour with a constant rate throughout the trial , whilst FIs produce scalloped curves with a pause following each reinforcement and a rapid increase in rate until the next reinforcement . Catania and Reynolds [98] performed a detailed analysis of behaviour under VIs and found that response rate declined with the average reinforcement rate . Within a trial response frequency increased with time , following approximately a negatively accelerated curve . When normalized by maximum response rate and time to reinforcement , these curves showed a considerable degree of superimposition . Matell and colleagues [108] trained rats on a VI in which intervals were sampled from an uniform distribution U ( 15 , 45 ) , and then tested using a peak procedure . They compared the VI response peak curve to the peak curve from a control group trained on an FI 30 ( the mean of the VI distribution ) . Although the two curves were not significantly different statistically , the VI response peak curve peaked slightly earlier and was slightly higher than the control group . Jennings and colleagues [109] compared timing performance between VI and FI in three experiments , but found VI timing only in a VI where the average interval was 30 seconds . The other experiments from the same paper produced results more in agreement with the earlier work by [107] showing a constant rate of responding during VI trials . Taken together , these studies appear to show that timing may sometimes be present during VI schedules . In this case , animals appear to be learning the average of the interval distribution . Here we demonstrate with simulations that RWDDM can account for such findings . The only other model in our analysis set that can account for this result is Modular Theory . Although animals are able to time different durations simultaneously , as seen in mixed FIs , paradoxically under certain circumstances a type of temporal averaging can be observed . This is a relatively new and important phenomenon , which challenges in particular theories of timing to propose a mechanism that can explain such averaging . When rats are trained using two distinct stimulus modalities , a visual stimulus ( a light ) and an auditory ( a tone ) , each signalling reinforcement at a different time , responding during compound presentations of both stimuli peaks roughly in the middle of both durations [110] . This intermediate response curve to the compound superimposes with the two other single stimulus curves when normalized , suggesting that the animal is timing only one average duration . The type of average being computed appears to be modulated by the reinforcement probabilities associated with each stimulus duration , with the weighted geometric average fitting the data better than a weighted arithmetic average or a non-weighted average [111 , 112 , 113] . Significantly , temporal averaging in rats is only consistently observed when the auditory stimulus signals the short interval and the visual stimulus signals the long interval [111 , 114] . Even when each stimulus is associated with a different response option ( light reinforced with a left nosepoke , tone with a right ) rats still tend to mix the temporal information during compound trials [115] . We do not make a strong claim about RWDDM’s ability to explain this data . Rather , we show that it has the necessary elements from which an account can begin to be formulated . MoT also has similar elements from which an account can be built . CSC-TD , MS-TD and LeT do not appear to be equipped to deal with this phenomenon . Table 4 summarizes the results from the simulations . RWDDM was able to reproduce the main features of the data in 8 out of the 10 experiments . In the other 2 the model was able to partially account for the data . To allow for comparison we have offered qualitative predictions for the other 4 models in Table 4 . It is important to note that for most of the 10 phenomena analysed here simulations using these models are not available in the literature . Although we have tried our best to provide predictions based on our understanding of these models , we have not actually simulated them . Therefore it is possible that in some cases a model may produce results that we did not foresee if the right set of parameters is found or some of the assumptions are relaxed . It is also possible that some simple modifications might allow the models to explain the data . We endeavoured to point out some such modifications that seem likely to work when discussing the simulation results above , but we do not make predictions based on them because the purpose here is only to provide a comparison of the current mechanisms of each model and therefore encourage future work on model improvement . With that in mind , Modular Theory has fared best after RWDDM , being able to account for 7 out of the 10 experiments . MS-TD and CSC-TD shared the second place with 3 out of 10 . LeT came in last , able to account for 2 experiments . The last column of Table 4 identifies the main mechanisms responsible for successfully accounting for each phenomenon . In this paper we introduced a new real-time model for classical conditioning and timing . The model combines elements from two theories , the Rescorla-Wagner conditioning model and the TDDM interval timing theory . We have simulated the model on 10 conditioning phenomena selected from the literature , which collectively represent a particular challenge for any single model to explain . The model was successful in accounting for 9 , and can be made to account for the rest if simple modifications are made . The mechanisms used by other models of similar scope were evaluated to see if they could also account for the data . The model that got closer to this level of success in this set of phenomena was Modular Theory . This was due to MoT and RWDDM having a significant overlap in terms of mechanisms . Both models use an accumulator to mark the passage of time . Both models require only a single associative unit per stimulus that adapts to the temporal information conveyed by the stimulus . Their main difference is that MoT still uses the linear operator rule which precludes it from explaining blocking and other compound phenomena , whilst RWDDM uses the RW rule which can account for those phenomena . The same limitation is faced by TILT , a recent model that we did not analyse but that shows promising results and has desirable timing properties . RWDDM may be improved in several ways . It is quite likely that the asymptote of learning may not be described by the simple inverse relationship to reinforcement time that we assumed . In some of the experiments modelled here , response peak seemed to decrease slower with ISI than our inverse relationship predicted . Functions other than Gaussians might be used to represent the CS , which could better fit the data in the case of latent inhibition for example . These and other theoretical issues may be better elucidated by new experiments involving compound stimuli and a manipulation of their durations , such as the experiments with blocking , compound peak procedure and temporal averaging analysed here . We have also adopted the P-H rule in one experiment , but have not explored its application in the others . Making the P-H rule an integral part of RWDDM would add one more parameter but it would also allow RWDDM to account for other preexposure and attentional effects that the rule is designed to account . This is not a difficult modification , and we have already shown it to be feasible . RWDDM may be regarded , like TD , as a real-time extension of RW . Unlike TD and LeT , it does not require a number of associative units that grows linearly with time . It adds to RW the powerful timing mechanism of TDDM . But also , by making a link with a version of DDM , it shows that it may be possible to arrive at a unified account of timing , conditioning and decision making .	How does the time of events affect the way we learn about associations between these events ? Computational models have made great contributions to our understanding of associative learning , but they usually do not perform very well when time is taken into account . Models of timing have reached high levels of accuracy in describing timed behaviour , but they usually do not have much to say about associations . A unified approach would involve combining associative learning and timing models into a single framework . This article takes just this approach . It combines the influential Rescorla-Wagner associative model with a timing model based on the Drift-Diffusion process , and shows how the resultant model can account for a number of learning and timing phenomena . The article also compares the new model to others that are similar in scope .	[ "Abstract", "Introduction", "Model", "Results", "Discussion" ]	[ "learning", "classical", "conditioning", "conditioned", "response", "social", "sciences", "neuroscience", "learning", "and", "memory", "simulation", "and", "modeling", "cognitive", "psychology", "cognition", "memory", "research", "and", "analysis", "methods", "learning", ...	2017	A Rescorla-Wagner drift-diffusion model of conditioning and timing
Sporotrichosis is a polymorphic chronic infection of humans and animals classically acquired after traumatic inoculation with soil and plant material contaminated with Sporothrix spp . propagules . An alternative and successful route of transmission is bites and scratches from diseased cats , through which Sporothrix yeasts are inoculated into mammalian tissue . The development of a murine model of subcutaneous sporotrichosis mimicking the alternative route of transmission is essential to understanding disease pathogenesis and the development of novel therapeutic strategies . To explore the impact of horizontal transmission in animals ( e . g . , cat-cat ) and zoonotic transmission on Sporothrix fitness , the left hind footpads of BALB/c mice were inoculated with 5×106 yeasts ( n = 11 S . brasiliensis , n = 2 S . schenckii , or n = 1 S . globosa ) . Twenty days post-infection , our model reproduced both the pathophysiology and symptomology of sporotrichosis with suppurating subcutaneous nodules that progressed proximally along lymphatic channels . Across the main pathogenic members of the S . schenckii clade , S . brasiliensis was usually more virulent than S . schenckii and S . globosa . However , the virulence in S . brasiliensis was strain-dependent , and we demonstrated that highly virulent isolates disseminate from the left hind footpad to the liver , spleen , kidneys , lungs , heart , and brain of infected animals , inducing significant and chronic weight loss ( losing up to 15% of their body weight ) . The weight loss correlated with host death between 2 and 16 weeks post-infection . Histopathological features included necrosis , suppurative inflammation , and polymorphonuclear and mononuclear inflammatory infiltrates . Immunoblot using specific antisera and homologous exoantigen investigated the humoral response . Antigenic profiles were isolate-specific , supporting the hypothesis that different Sporothrix species can elicit a heterogeneous humoral response over time , but cross reaction was observed between S . brasiliensis and S . schenckii proteomes . Despite great diversity in the immunoblot profiles , antibodies were mainly derived against 3-carboxymuconate cyclase , a glycoprotein oscillating between 60 and 70 kDa ( gp60-gp70 ) and a 100-kDa molecule in nearly 100% of the assays . Thus , our data broaden the current view of virulence and immunogenicity in the Sporothrix-sporotrichosis system , substantially expanding the possibilities for comparative genomic with isolates bearing divergent virulence traits and helping uncover the molecular mechanisms and evolutionary pressures underpinning the emergence of Sporothrix virulence . Recent years have seen a global burden of fungal infections in warm-blooded hosts , many of which have the potential to trigger epidemics or cause endemic diseases [1] . Sporotrichosis is a chronic infection of humans and animals caused by Sporothrix species in the order Ophiostomatales , which is mainly composed of saprophytic fungi . Remarkably , a pathogenic group consisting of four species ( Sporothrix brasiliensis , Sporothrix schenckii , Sporothrix globosa , and Sporothrix luriei ) has emerged as an important threat to the health of several warm-blooded hosts around the globe [2] . As the genetic distance increases from this pathogenic group , basal phylogenetic groups of Sporothrix are usually associated with a decrease in infectivity for mammals . Therefore , rare agents of sporotrichosis are found outside the clinical group , such as Sporothrix chilensis , Sporothrix mexicana , Sporothrix pallida and Sporothrix stenoceras [3–7] . Among the clinical agents , S . brasiliensis is by far the most virulent , followed by S . schenckii and S . globosa [8 , 9] , but all of them are capable of disease in mostly healthy patients [10] . A myriad of virulence factors , such as thermotolerance , adherence factors , and melanin produced during the host-pathogen interplay , help S . brasiliensis invade mammalian hosts , cause sporotrichosis , and evade host defenses . In one study of experimental sporotrichosis , Arrillaga-Moncrieff et al . [9] employed a disseminated model of infection to reveal striking differences related to the species , source , and genetic background , suggesting that the pathogenesis of sporotrichosis could be species-specific . Similarly , Fernandes et al . [8] confirmed differential virulence in the S . schenckii clade ( pathogenic clade ) and proposed a correlation among protein secretion , immunogenicity , genetic diversity , and virulence . Despite great efforts in understanding the pathogenesis of sporotrichosis in light of recent taxonomic changes in Sporothrix [8 , 9 , 11–13] , only a few studies have used a subcutaneous model of infection [14–17] . This subcutaneous route of inoculation reflects the natural route of transmission of Sporothrix spp . and holds true for traumatic inoculation with soil and plant material contaminated with Sporothrix propagules or animal-driven inoculation [3] . A major difference between these sources of contamination is related to the morphotype inoculated; the environmental inoculum is expected to involve hypha and conidia [18] , whereas animal-driven inoculation is expected to involve deep inoculation of yeasts into the subcutaneous tissue [19] . Hitherto poorly explored animal-driven inoculation deserve consideration for future developments in the system S . brasiliensis-sporotrichosis . Brito et al . [14] experimentally infected BALB/c mice by inoculating two strains of S . schenckii ( sensu lato ) subcutaneously into the left hind footpad . After challenge , animals presented with weight loss , cutaneous lesions , signs of inactivity , and different survival rates , mimicking human disease . Histological analysis revealed lesions in the organs , with inflammatory infiltrate and granuloma in the liver and inflammatory reaction in the area where inoculated . In recent years , human and feline sporotrichosis due to S . brasiliensis has emerged as a major health problem in Brazil , initially affecting the metropolitan region of Rio de Janeiro , but quickly spreading to several urbanized areas , some as far as 2 , 000 km from the epicenter of the initial epidemics [20–22] . In the present study , we developed a subcutaneous murine model of infection mimicking the alternative route of transmission that includes both animal horizontal transmission ( e . g . , cat-cat ) and zoonotic transmission ( e . g . , cat-human ) in which a high load of S . brasiliensis yeast cells may be inoculated through feline bites and scratches . Afterwards , we explored this system to compare virulence levels among 11 isolates of S . brasiliensis and allied species based on the fungal tissue burden , survival assay , histopathological aspects of affected organs , weight loss , protein secretion , immunogenicity , and genetic diversity . The study was performed in strict accordance with recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health and approved by the Institutional Ethics in Research Committee of the Federal University of São Paulo ( protocol number 8123190914 ) . Sporothrix isolates were obtained from the Federal University of São Paulo ( UNIFESP ) , São Paulo , Brazil . These isolates were characterized previously at the species level by phylogenetic analysis of the calmodulin-encoding gene [8 , 23] , species-specific PCR [24] , and rolling circle amplification ( RCA ) [25] . We selected a set of S . brasiliensis ( n = 11 ) , S . schenckii ( n = 2 ) , and S . globosa ( n = 1 ) based on the following criteria: clinical origins ( human and feline ) , geographic region , idiomorphic mating type ( MAT1-1 and MAT1-2 ) , and genetic diversity ( Table 1 ) . It is important to emphasize that although S . brasiliensis presents with little diversity during outbreaks [26 , 27] our efforts were focused to choose the most genetically deviating ones . Reference strains were included in all experiments . The Ss06 ( S . globosa ) , Ss39 ( S . schenckii ) , Ss126 ( S . schenckii ) , and Ss54 ( S . brasiliensis ) isolates were used as controls for low , medium , and high virulence levels as described previously [8] . Attenuation of virulence may occur more rapidly in some strains than others when subjected to successive in vitro subculturing and therefore impact multiple comparisons as proposed here . To prevent any bias among Sporothrix spp . isolates at the start of in vitro culturing , all isolates were passed through BALB/c and then re-isolated as a monosporic culture prior to challenge experiments [8] . Sporothrix isolates were maintained in Sabouraud dextrose agar slants ( Difco , Detroit , USA ) and cultivated for 7 days at 25°C prior to use . Approximately 2×106 conidia ( 90% viable cells ) were used to inoculate 500-ml flasks containing 150 ml of Brain Heart Infusion Broth ( Difco , Detroit , USA ) . The cultures were incubated at 37°C in a rotary shaker ( Multitron II—Infors HT , Switzerland ) with constant orbital agitation ( 100 rpm ) for 4 days . Yeast cells were collected by centrifugation at 3 , 500 g for 15 min ( 4°C ) and then washed three times in phosphate buffered saline ( PBS ) . The yeast cell concentration was adjusted to 5×106 cells ( 25 μl ) and cell viability assessed by trypan blue staining and by plating dilutions of the suspension on BHI plates as described previously [8 , 30] . Only samples with ≥85% viability were employed for experimental infection . Male BALB/c mice weighing 25–30 g ( 6 to 8 weeks old ) were obtained from Federal University of São Paulo ( UNIFESP ) for the virulence assays; 90 mice were used for colony-forming unit ( CFU ) assays and 150 for survival assays . Animals were housed in temperature-controlled rooms at 23–25°C , five per cage , in standard boxes with ad libitum access to food and water [8] . Animals were divided into 15 groups of 10 mice each and inoculated subcutaneously into the left hind footpad ( one group for each Sporothrix isolate and one negative control group ) . The mice were anesthetized with 0 . 2 mg/kg xylazine and 20 mg/kg ketamine and subcutaneously inoculated with 25 μl containing 5×106 yeast cells/animal . The control group received 25 μl of PBS only . The infected mice were observed daily and their mortality recorded over the following 16 weeks [8] . The percent weight loss was determined by measuring each animal’s weight every week post-inoculation ( up to 16 weeks ) and comparing it to the animal’s weight on the day of inoculation . Animals were divided into 15 groups of 6 mice each ( one group for each Sporothrix isolate and one negative control group ) . The mice were anesthetized with 0 . 2 mg/kg xylazine and 20 mg/kg ketamine and subcutaneously inoculated with 5×106 yeast cells/animal . The control group received 25 μl of PBS only . Twenty days post-infection , the animals were sacrificed by CO2 anesthesia and the liver , spleen , kidneys , lungs , heart , brain , and footpad aseptically removed . The organs and footpad were separated , weighed , and homogenized in sterile PBS using a tissue grinder . Samples ( 100 μl ) of each homogenate were seeded on Petri dishes containing BHI agar and incubated at 37°C . Colonies were counted from day 15 to 20 . The results were expressed as CFU/g tissue [30] . Serum was collected from the mice and stored at –20°C for Western blot as described elsewhere [8] . CFU assay results were compared among groups and analyzed by analysis of variance ( ANOVA ) followed by post-hoc Tukey . Significance was set at P≤0 . 05 . For survival assays , data were analyzed by Kaplan-Meier survival plots followed by the log-rank test . For weight loss assays , data were analyzed by paired t-test . P≤0 . 05 was considered significant . All analyses were performed using GraphPad Prism version 6 for Windows . Post-mortem tissues , including liver , spleen , kidneys , lungs , heart , brain , and the infected footpad , were collected 20 days post-infection , fixed in 10% formaldehyde , and embedded in paraffin . The footpads were descaled in 7% nitric acid ( 48 h ) , fixed with 10% formaldehyde , and embedded in paraffin . The organs and footpads of animals in the PBS group were collected as negative controls . Embedded tissue sections ( 3-μm-thick ) were stained with hematoxylin and eosin ( H&E ) and periodic acid–Schiff ( PAS ) for observation by light microscopy using an Olympus BX51 microscope ( Tokyo , Japan ) equipped with an Olympus SC100 camera . Exoantigens from the mycelial phase of 14 isolates of the Sporothrix species ( Table 1 ) were obtained in Sabouraud broth at 25°C using a rotary shaker ( Multitron II—Infors HT , Switzerland ) with constant orbital agitation ( 100 rpm ) for 10 days as described previously [31] . Whole cell extracts of Sporothrix yeast cells ( S . brasiliensis CBS 132990 and S . schenckii CBS 132974 ) were obtained as described elsewhere [32] . Protein extracts from isolates CBS 132990 ( Ss54 ) and CBS 132974 ( Ss118 ) were selected as references for immunoblot assays because they had been successfully used to diagnose sporotrichosis in ELISA and immunoblot [33 , 34] . The protein concentrations were determined for all antigenic preparations using the Bradford method [35] . Exoantigens and whole cellular extracts were kept at -80°C until use . In order to evaluate the diversity of proteins secreted by different Sporothrix isolates , 14 exoantigens ( 5 μg/lane ) were separated by SDS-PAGE using 10% gels [36] and silver-stained [37] . The relative molecular weights of the fractions were estimated using standard broad-range molecular weight markers ( Protein Benchmark , Invitrogen ) . Individual bands were recorded , converted into binary data , and the profile of each Sporothrix exoantigen compared using Jaccard's similarity coefficient . Dendrogram analyses were performed in SYSTAT 13 ( Systat Software , San Jose , CA ) . Immunoblotting was carried out by resolving the 14 exoantigens and whole cellular proteins from two extracts ( Ss54 = CBS 132990 and Ss118 = CBS 132974 ) on 10% SDS-PAGE , followed by electrotransfer to nitrocellulose membrane ( 0 . 2 μm; Bio-Rad ) at 20 V for 30 min in transfer buffer ( 25 mM Tris base , 192 mM glycine , 20% methanol , pH 8 . 3 ) [38] using a Trans-Blot SD semi-dry device ( Bio-Rad ) . Free binding sites were blocked overnight with PBS blocking buffer ( 1% bovine serum albumin supplemented with 0 . 05% [vol/vol] Tween 20 , 5% [wt/vol] skim milk , pH 7 . 6 ) at 4°C . The membrane was cut lengthwise into 0 . 5-cm strips , and each blot was probed with its respective antisera at a dilution of 1:200 ( PBS-Tween 20 , 0 . 005% ) for 1 h at room temperature . The membranes were washed three times with Tris-buffered saline ( pH 7 . 5 ) containing 0 . 05% [vol/vol] Tween-20 for 10 min . Immunoreactive proteins were detected by incubation with peroxidase-conjugated goat anti-mouse IgG ( 1:1000 dilution ) as secondary antibody ( Sigma-Aldrich , USA ) for 1 h at room temperature . Next , the membranes were washed with Tris-buffered saline ( pH 7 . 5 ) containing 0 . 05% [vol/vol] Tween-20 and the signal detected using an enhanced chemiluminescence detection kit ( GE Healthcare ) . Blots were imaged in a transilluminator ( Uvitec Cambridge ) . Allience 4 . 7 software was used to take several images at different exposures , from 2 s each to a total of 10 images over 2 s . Proteins ( 300 μg ) of S . brasiliensis Ss54 ( = CBS 132990 ) were precipitated using the 2D clean-up kit ( GE Healthcare , Piscataway , NJ , USA ) following the manufacturer's recommendations . Proteins were mixed with IPG rehydration buffer ( 7 M urea , 2 M thiourea , 2% CHAPS , 1 . 2% DeStreak , 2% vol/vol isoelectric focusing [IEF] buffer pH 4–7 , and trace bromophenol blue ) to a final volume of 250 μl . The strips ( pH 4–7 , 13 cm ) were allowed to rehydrate at 30 V for 12 h , focused ( Ettan IPGphor III system; GE Healthcare , USA ) , equilibrated , apposed to the second dimension ( 10% ) gels , and run as described [39] . Proteins were developed with silver [37] or directly transferred to nitrocellulose membrane ( 0 . 2 μm; Bio-Rad ) in the case of immunoblot analysis . For immunoblotting , membranes were probed against mouse antisera at a dilution of 1:200 ( PBS-Tween 20 , 0 . 005% ) for 2 h at room temperature in order to evaluate seroreactive spots related to 3-carboxymuconate cyclase as described earlier by our group [34 , 39] . Immunoblot conditions and immunodetection were essentially as described above ( SDS-PAGE and immunoblotting assay ) . After challenge , lesion patterns in mice mimicked those of cats and humans , with intensities varying in an isolate-dependent manner from minor injuries to suppurating subcutaneous nodules that progressed proximally along lymphatic channels ( Fig 1 ) . Isolates Ss39 ( S . schenckii ) , Ss126 ( S . schenckii ) , Ss34 ( S . brasiliensis ) , and Ss67 ( S . brasiliensis ) were able to develop minor injuries in the left hind footpad of infected mice . S . brasiliensis isolates Ss54 , Ss66 , Ss99 , Ss174 , Ss226 , Ss252 , Ss261 , and Ss265 induced the development of large lesions , limiting the movement of the animals due to swelling and increased footpad size . In addition , S . brasiliensis isolates Ss252 , Ss54 , Ss99 , Ss174 , and Ss226 caused a partial loss of the fingers to complete footpad loss in some inoculated animals . Lesions were verified in the tails , back , and skin of 14 . 2% of infected animals . At the end of the study , animals that developed large lesions exhibited sequelae , usually in the footpad and the joint next to the region of the footpad ( inoculum site ) . However , some isolates , such as the non-virulent Ss06 ( S . globosa ) and Ss104 ( S . brasiliensis ) , did not result in lesions in the footpads . Notwithstanding , these non-virulent strains ( in a subcutaneous route ) were able to disseminate when subjected to an intravenous route during passages in BALB/c mice to restore the virulence , suggesting that virulence is affected by the chosen injection route . The fungal tissue burden in the liver , spleen , kidneys , lungs , heart , brain , and footpad is shown in Fig 2 . All Sporothrix isolates were recovered from the left hind footpad , except for S . globosa ( Ss06; P < 0 . 001; Fig 2G ) . The dissemination power among Sporothrix isolates followed a subcutaneous route . After the footpad , the liver was the most affected organ ( n = 8 ) , followed by the lungs ( n = 3 ) , brain ( n = 3 ) , spleen ( n = 2 ) , heart ( n = 2 ) , and kidneys ( n = 2 ) . Across the main isolates in the pathogenic clade , strains belonging to the S . brasiliensis clade were usually more virulent than those belonging to S . schenckii and S . globosa . Among S . brasiliensis strains , Ss174 ( MAT1-1 ) and Ss226 ( MAT1-2 ) were significantly more virulent ( P < 0 . 01 ) with high fungal burden and dissemination power , closely followed by Ss34 ( MAT1-1 ) , Ss54 ( MAT1-1 ) , Ss66 ( MAT1-2 ) , Ss99 ( MAT1-2 ) , Ss252 ( MAT1-2 ) , Ss261 ( MAT1-1 ) , and Ss265 ( MAT1-1 ) , whereas Ss67 ( MAT1-2 ) and Ss104 ( MAT1-2 ) were the least virulent strains , causing self-limited infection ( footpad only ) with lower invasiveness , similar to isolates Ss39 ( MAT1-2 ) and Ss126 ( MAT1-2 ) of S . schenckii ( Fig 2 ) . This suggests that the invasiveness of S . brasiliensis is strain-dependent , and mating type idiomorphs being directly associated with virulence in the Sporothrix spp . is unlikely . The isolate Ss06 ( S . globosa , MAT1-1 ) was considered non-pathogenic because it was not able to colonize any organ . No fungal load was observed in organs from animals that received sterile PBS ( Fig 2 ) . The survival time of each mouse was recorded ( Fig 3 ) , and differences in the median survival time among strains were analyzed . All mice challenged with highly invasive strains ( Ss226 [weeks 2–9] , Ss174 [weeks 11–13] , and Ss252 [weeks 6–16] ) died between 2 and 16 weeks post-infection ( Fig 3 ) , although the numbers of CFUs were significantly greater in mice infected with Ss226 ( P < 0 . 001 ) ( Fig 2 ) . On the other hand , mice challenged with strains Ss06 , Ss34 , Ss39 , Ss54 , Ss66 , Ss67 , Ss99 , Ss104 , Ss126 , Ss261 , and Ss265 were still alive after 16 weeks . The animals were evaluated for percentage weight loss during the survival experiments . We used weight loss as a virulence measure because it is correlated with mortality rate . Remarkably , mice inoculated with highly virulent isolates Ss174 ( P < 0 . 0001 ) and Ss226 ( P = 0 . 0006 ) induced substantial and chronic weight loss ( -5% and -15% of their body weight , respectively ) of infected animals ( Fig 4A ) . In addition , mice infected with S . brasiliensis Ss34 , Ss54 , Ss67 , Ss104 , Ss252 , Ss261 , and Ss265 presented with moderate weight loss ( P < 0 . 0001 ) compared to the non-infected group , and the mice began to regain weight between 2 and 5 weeks post-inoculation ( Fig 4B ) . In contrast , S . schenckii isolates Ss126 ( P = 0 . 0063 ) and Ss39 ( P = 0 . 0519 ) did not induce weight loss in infected mice and the later did not reach the threshold ( P < 0 . 05 ) for significance compared to the PBS group ( Fig 4C ) . At the site of inoculation , histopathological parameters necrosis , suppurative inflammation , and mononuclear inflammatory infiltrate were noted in all Sporothrix-inoculated animals . All histopathological parameters increased in severity in S . brasiliensis isolates Ss174 and Ss226 . However , a few strains were able to disseminate and induce granuloma ( Ss99 , Ss174 , Ss226 , and Ss261 ) and mononuclear/polymorphonuclear inflammatory infiltrates ( Ss226 , Ss252 , ad Ss265 ) in the liver of infected animals ( Fig 5 ) . In the heart , an inflammatory infiltrate was observed in mice inoculated with Ss252 . Fig 5 shows some of the differences in the histopathological parameters in the course of infection with two polar S . brasiliensis strains . Table 2 summarizes the virulence characteristics , weight loss , and histopathological features of Sporothrix isolates . The protein secretion profiles of the Sporothrix strains analyzed by SDS-PAGE were heterogeneous , but important differences were observed regarding localization and the intensity of several bands ( Fig 6A and 6B ) . More than 11 bands were stained ( Fig 6D ) , and the proteins present in all strains corresponded to molecular weights of 140 , 120 , 110 , 92 , 88 , 72 , 65 , 52 , 38 , 32 , and 28 kDa . Each strain exhibited a major band at 32 kDa ( Fig 6C ) . To further examine the contribution of immunogenicity , an immunoblotting reaction employing the Sporothrix polyclonal antiserum revealed that IgG antibodies produced by the mice recognized 17 fractions overall , ranging from 22 to 130kDa in the whole cell extracts and from 38 to 100 kDa in the exoantigen preparations . Notably , immunodominant molecules included 3-carboxymuconate cyclase ( KP233225 , a classical glycoprotein oscillating between the 60 and 70 kDa fractions ) and against a 100-kDa molecule ( KP247558 ) in nearly 100% of the cases ( Fig 7 ) . Moreover , the antigenic profiles were isolate-specific , supporting the hypothesis that different Sporothrix can elicit a heterogeneous humoral response over time , but a high level of cross reaction was observed between S . brasiliensis and S . schenckii whole cell extracts and the same antiserum ( Fig 7A ) , supporting that antigenic epitopes are conserved among different species embedded in the pathogenic clade . In a scenario were gp60-70 were the main antigen recognized via 1D immunoblot we explored 2D-immunoblot to confirm the identity of this antigen . In doing so , we used the same protein extract describe by Rodrigues et al . [39] , which were resolved by 2DGE followed by immunoblot . Thereafter membranes were probed against a pool of sera of mice experimentally infected in this study . Multiple distinct seroreactive spots in the same gel were previously identified as the same protein ( Fig 8A ) and represent charge ( isoforms ) and molecular mass variants ( glycoforms ) [34 , 39]; examples of these variants are circumscribed within a red area in Fig 8B–8I ) , confirming that isoforms of 3-carboxymuconate cyclase are recognized by IgG antibodies during experimental infection of S . brasiliensis . Small variations were noted in immunoblot profiling considering the recognition pattern using the crude proteome in 1D ( crude Ss54; Fig 7A ) and the precipitated proteome used for 2DGE ( Ss54; Fig 8B–8I ) . Indeed , the 2DGE analysis of fungal proteins is quite difficult due to the high concentration of contaminant molecules . To remove interfering compounds from our crude extract ( salts , lipids , detergents , nucleic acids and phenolic compounds ) and improve 2DGE we used 2-D Clean-Up Kit as previously recommended [32 , 39] . A pivotal challenge in studying the pathogenesis of Sporothrix species is to develop a quantifiable approach that reproduces both the pathophysiology and symptomology of sporotrichosis . Mice are often used to assess the virulence of Sporothrix species ( BALB/c , C57BL/6 , OF-1 ) , but simple models varying from cultured mammalian cells [40] to the caterpillar Galleria mellonella [41 , 42] have been employed successfully . We selected the murine system because endothermy seems to protect mammals from most Sporothrix spp . [26 , 43] . We chose a set of human and animal-derived isolates of S . brasiliensis that are of both medical and veterinary interest and have been the main focus of extensive research in the last few years due to the global emergence of sporotrichosis [20 , 43] . Mice challenged subcutaneously with S . brasiliensis developed typical lesions of sporotrichosis in the footpad 1 week post-infection; they very similar to those observed in human and animal sporotrichosis , with an ulcerous-crust appearance , validating our murine model . Moreover , histopathology showed granulomatous inflammation and a pyogenic process with the presence of yeasts similar to those observed by Brito et al . [14] . Previous models of murine sporotrichosis employed a subcutaneous route of infection and compared the virulence levels among several clinical isolates of Sporothrix , confirming differences among them [14 , 16] , especially interspecific variation as observed between S . brasiliensis and S . schenckii [15] . Previous epidemiological studies have shown that recombination is most likely to occur in S . schenckii which allows generation of greater genetic diversity [27] . On the other hand , this phenomenon is expected to be absent or less frequent in natural populations of S . brasiliensis resulting in highly clonal population structure during outbreaks [27] . Remarkably , our results support the heterogeneity of virulence levels in S . brasiliensis isolates with low , medium , and high virulence , as well as confirm the absence of virulence in S . globosa as reported in previous studies of a systemic route of infection [8 , 9] . Another striking finding is that S . brasiliensis ( Ss34 , Ss174 , and Ss226 ) can disseminate across the central nervous system , even when we employed a subcutaneous route . These results involving brain dissemination parallel available data from a systemic model based on CFUs [9] , histological analysis , or a molecular approach using species-specific PCR [24] . Such atypical manifestation in the central nervous system may draw the attention of physicians to the occurrence of more severe manifestations of disease caused by Sporothrix in endemic areas [44 , 45] . Moreover , animal-born isolates ( Ss54 , Ss174 , Ss226 , and Ss252 ) presented greater virulence than the remaining S . brasiliensis isolates of human origin . This emerging virulence could explain the recent appearance of atypical and more severe cases of sporotrichosis during zoonotic episodes in areas where S . brasiliensis is endemic [46 , 47] . Based on a systemic route of infection , Arrillaga-Moncrieff et al . [9] and Fernandes et al . [8] compared the virulence of different species of the Sporothrix and found that S . brasiliensis was the most virulent , followed by S . schenckii and S . globosa , which had low or no pathogenicity . In our study , it was clear that the dissemination power is related to the route of infection , as the same isolates previously found to be highly virulent ( Ss54 ) , medium ( Ss126 ) , and non-virulent ( Ss06 ) in a systemic model of infection [8] presented with a lower capacity for dissemination when submitted to a subcutaneous route ( Fig 2 ) . We emphasize that the subcutaneous route of infection reflects the natural route of transmission . Moreover , when a cat transmits the fungus by scratching or biting , the contamination is caused by the yeast inoculum and not by conidia [20] , a morphotype recognized to be more virulent to mammals [48] . This may enhance the alternative , feline-driven transmission route [3] . Therefore , we used 5 × 106 yeast cells as inoculum considering the high yeast load present in the skin lesions and exudates of cats [19] . Within the intraspecific diversity S . brasiliensis , we identified isolates Ss226 and Ss174 as highly virulent . The ability of these strains to cause earlier death than remaining Sporothrix could be related to their capacity to induce liver damage , thereby impairing or leading to earlier loss of liver function . Histological parameters revealed granuloma formation and mononuclear/polymorphonuclear infiltration in isolates that induced dramatic weight loss and death . These data reinforce the high invasiveness of some strains of S . brasiliensis compared to our controls , S . schenckii and S . globosa . On the other hand , we recognize Ss104 as a low virulent S . brasiliensis strain . Remarkably , this isolate was recovered from a human case nearly 2 , 000 km from the epicenter of the cat-transmitted epidemic in the metropolitan area of Rio de Janeiro , Brazil [27] . The minimum genome size of isolate Ss104 was estimated to be 31 . 3 Mb by pulsed field gel electrophoresis [23] , comprising at least six chromosomal bands ( 7 . 0 , 6 . 7 , 5 . 8 , 5 . 4 , 3 . 5 , and 2 . 9 Mb ) , which deviates genetically from the highly virulent clone of S . brasiliensis implicated in the Rio de Janeiro epidemic that usually presents with five chromosomal bands and has an average genome size of 25 . 7 Mb . Moreover , gene synteny highlighted differences in isolate Ss104 , including gene translocations and a considerable amount of repetitive DNA sequences [23] , which may be related to differing genomic organization and explain the atypical phenotype of virulence . Among human pathogenic fungi , such as Blastomyces spp . [49] , Histoplasma capsulatum [50] , and Paracoccidioides spp . [51] , isolates with a divergent genetic repertoire have been demonstrated to present with differential regulation of critical virulence factors , affecting pathogenicity to the mammalian host . Sporothrix is a heterothallic fungus with a single mating type locus that produces two alleles , MAT1-1 and MAT1-2 . This system requires two compatible partners for mating to occur [29] , though the sexual cycle in clinical Sporothrix has not yet been observed in nature or under laboratory conditions . Interestingly , S . brasiliensis populations associated with feline epizooties have been suggested to reproduce clonally with the predominance of MAT1-1 or MAT1-2 in South and Southeast Brazil , respectively [29] . Therefore , we explored the virulence profiles among distinct mating type idiomorphs to test whether there is a connection between mating type and virulence in Sporothrix . Interestingly , the most virulent isolates harbor opposite mating types ( i . e . , Ss174 = MAT1-1; Ss226 = MAT1-2 ) , suggesting that both are similarly pathogenic with high dissemination power , induction of weight loss , and induction of death . This is consistent with the severity of the disease observed among animals naturally infected with S . brasiliensis in both epidemics occurring in South and Southeast Brazil [20–22] . Mating type has been found to be a source of variation in virulence in certain serotypes of Cryptococcus neoformans [52] and Aspergillus fumigatus [53] , including a deviating ratio between a/α or MAT1-1/MAT1-2 recovered from clinical cases . On the other hand , mating type appears to have no influence in Histoplasma capsulatum , in which both mating types are evenly distributed in the soil ( 1:1 ) , deviating in clinical samples with the majority of MAT1-1 ( 7:1 ) [54 , 55] , but no significant difference was found between strains MAT1-1 and MAT1-2 in a murine model of infection [56] , similar to our data in clinical Sporothrix . Sporothrix-mammal interactions may result in different outcomes varying from self-limited to severe disseminated infection . Such variation may be regarded as a function of the host's immune response delivered against Sporothrix . Remarkably , the consequence of this complex interaction may reduce the amount of continuing damage caused by the microbe to a level that is insignificant [57] . The cell-mediated immunity during sporotrichosis involves interactions between CD4+ T cells , macrophages , dendritic cells , neutrophils and the release of soluble mediators which are protective and promote clearance of the infection . Of these cell populations , CD4+ T cells and macrophages play an essential role for resolution of sporotrichosis , since athymic nude mice clearly demonstrate that loss of CD4+ T cells , but not CD8+ T cells , renders mammals susceptible to Sporothrix infection [16] . In this scenario , it is tempting to hypothesize that the different lesion patterns and disease severity observed in our study may reflect distinct lymphocyte activation profiles by the different strains used . To further reinforce this notion , it was recently demonstrated that S . brasiliensis and S . schenckii are differentially recognized by human peripheral blood mononuclear cells , depending on the morphotype ( conidia , germlings or yeasts ) and cell wall composition [58] . To assess the humoral response against Sporothrix species , mice have been widely used in experimental sporotrichosis , as well as serum derived from naturally infected humans [59 , 60] and cats [33 , 34] . Cellular and humoral immune responses triggered upon Sporothrix introduction into the subcutaneous tissue may play important roles in the development and severity of sporotrichosis [61–63] . We reported previously that Sporothrix molecules are able to induce an important humoral response during human [39] , murine [8] , and feline sporotrichosis [34] . Here , the notion that divergent strains elicit variant-specific humoral immunity is supported by the finding that most S . brasiliensis strains induced antibodies against a great diversity of antigens during murine sporotrichosis , a scenario similar to the ongoing epizootics in Rio de Janeiro . S . brasiliensis-infected cats also produce a diversity of antibodies that seem to be divergent from one diseased cat to other [34] , but this diverse repertoire of antibodies generates high levels of cross reactivity when probed against S . brasiliensis and S . schenckii antigenic preparations , supporting a conservation of epitopes across the main pathogenic species [34 , 39] . Judging from these immunogenic profiles , our results may also shed light on recent evolutionary divergence in the clinical clade , as this recognition pattern is absent in the non-virulent ancestral S . mexicana [39] , a species inserted in the S . pallida complex with attenuated virulence to mammals [9 , 11] . Despite the diversity in the clinical clade , we found that 3-carboxymuconate cyclase ( Accession number KP233225; gp60 in S . brasiliensis and gp70 in S . schenckii ) is the major antigenic target in the early IgG response of murine , feline [34] , and human sporotrichosis [39] whereby , after infection , more and more gp60-70-specific antibodies can be found [63] . Classically , gp60-70-specific antibodies are associated with protective immunity in which passive immunization with either polyclonal monospecific or monoclonal antibodies raised against 3-carboxymuconate cyclase can reduce the fungal burden of infected animals , probably by inhibiting adhesion to the components of the extracellular matrix and enhancing the opsonization of yeasts [17 , 64–68] . Judging from a highly polymorphic humoral response raised against strain-specific immunity , we highlight gp60-70 as a common molecule recognized by all antisera of Sporothrix-infected mice . Therefore , we propose gp60-70 as an interesting candidate to match all pathogenic Sporothrix , including those associated with the most severe disease , as in S . brasiliensis , probably inducing cross-protection against a wide range of antigenic variants . The relevance of 3-carboxymuconate cyclase as a target in the humoral response has been shown in several studies [15 , 63 , 64 , 66 , 67 , 69 , 70] . Other important Sporothrix vaccine targets include a 100-kDa antigenic molecule . A 100-kDa molecule has also been found in the human immunoproteome ( Accession number KP247558 ) and identified as endoplasmic signal peptidase . Similar studies exploring the humoral response in sporotrichosis have demonstrated an antigen-specific IgG response against a peptide hydrolase ( 44 kDa ) and an enolase ( 47 kDa ) . Specific antibodies were able to significantly enhance phagocytosis , inhibit adhesion , affording in vivo protection [71] . Taken together , these results are helping build an important panel of candidate vaccine antigens with the potential for immunization in areas of hyperendemicity . Thus , understanding the diversity of antigenic variation may help in designing vaccines that represent the worldwide repertoire of polymorphic Sporothrix antigens . In conclusion , our data broaden the current view on S . brasiliensis epidemics driven by populations with predominantly low levels of genetic diversity [27] . We demonstrate that the virulence process is strain-specific . Mimicking natural infection via traumatic inoculation , we demonstrated that the route is crucial to understanding the pathogenesis of sporotrichosis . We also demonstrated the presence of a panel of immunodominant molecules that could be used as antigens in anti-Sporothrix vaccines . These advances have opened up important possibilities for comparative genomic studies , characterization of new Sporothrix antigens , and confirming vaccine targets to tackle the progress of sporotrichosis in endemic areas .	Sporotrichosis is polymorphic infection that is acquired after traumatic implantation of Sporothrix propagules into host tissues . In Brazil , large epizooties occur in domestic cats and consequently massive zoonotic transmission has been observed since the 2000s without signs of decreasing over time . In this study , we explored a subcutaneous murine model of sporotrichosis mimicking the alternative route of transmission considering animal horizontal transmission ( cat-cat ) and zoonotic transmission ( cat-human ) in which S . brasiliensis yeasts are inoculated . Surprisingly , we found that virulence in the clonal offshoot S . brasiliensis is strain-dependent where highly virulent isolates can disseminate from the site of inoculation ( footpad ) to vital organs of infected animals , inducing dramatic weight loss , which correlates with host death . Such variation may be regarded as a function of the host's immune response delivered against Sporothrix . Additionally , humoral immune responses produced IgG against isoforms of 3-carboxymuconate cyclase ( gp60-gp70 ) and a 100-kDa molecule . Our study demonstrated that increased or attenuated virulence occurs in S . brasiliensis considering the route of infection , morphotype inoculated and the mammalian immune response . In addition to improving our understanding of the pathogenesis of the emerging animal-born S . brasiliensis , this study has key implications for future developments in serological diagnosis and vaccination efforts .	[ "Abstract", "Introduction", "Methods", "Results", "Discussion" ]	[ "humoral", "immunity", "body", "weight", "medicine", "and", "health", "sciences", "yeast", "infections", "pathology", "and", "laboratory", "medicine", "weight", "loss", "molecular", "probe", "techniques", "pathogens", "immunology", "microbiology", "immunoblotting", "ani...	2017	Exploring virulence and immunogenicity in the emerging pathogen Sporothrix brasiliensis
Multi-drug tolerance is a key phenotypic property that complicates the sterilization of mammals infected with Mycobacterium tuberculosis . Previous studies have established that iniBAC , an operon that confers multi-drug tolerance to M . bovis BCG through an associated pump-like activity , is induced by the antibiotics isoniazid ( INH ) and ethambutol ( EMB ) . An improved understanding of the functional role of antibiotic-induced genes and the regulation of drug tolerance may be gained by studying the factors that regulate antibiotic-mediated gene expression . An M . smegmatis strain containing a lacZ gene fused to the promoter of M . tuberculosis iniBAC ( PiniBAC ) was subjected to transposon mutagenesis . Mutants with constitutive expression and increased EMB-mediated induction of PiniBAC::lacZ mapped to the lsr2 gene ( MSMEG6065 ) , a small basic protein of unknown function that is highly conserved among mycobacteria . These mutants had a marked change in colony morphology and generated a new polar lipid . Complementation with multi-copy M . tuberculosis lsr2 ( Rv3597c ) returned PiniBAC expression to baseline , reversed the observed morphological and lipid changes , and repressed PiniBAC induction by EMB to below that of the control M . smegmatis strain . Microarray analysis of an lsr2 knockout confirmed upregulation of M . smegmatis iniA and demonstrated upregulation of genes involved in cell wall and metabolic functions . Fully 121 of 584 genes induced by EMB treatment in wild-type M . smegmatis were upregulated ( “hyperinduced” ) to even higher levels by EMB in the M . smegmatis lsr2 knockout . The most highly upregulated genes and gene clusters had adenine-thymine ( AT ) –rich 5-prime untranslated regions . In M . tuberculosis , overexpression of lsr2 repressed INH-mediated induction of all three iniBAC genes , as well as another annotated pump , efpA . The low molecular weight and basic properties of Lsr2 ( pI 10 . 69 ) suggested that it was a histone-like protein , although it did not exhibit sequence homology with other proteins in this class . Consistent with other histone-like proteins , Lsr2 bound DNA with a preference for circular DNA , forming large oligomers , inhibited DNase I activity , and introduced a modest degree of supercoiling into relaxed plasmids . Lsr2 also inhibited in vitro transcription and topoisomerase I activity . Lsr2 represents a novel class of histone-like proteins that inhibit a wide variety of DNA-interacting enzymes . Lsr2 appears to regulate several important pathways in mycobacteria by preferentially binding to AT-rich sequences , including genes induced by antibiotics and those associated with inducible multi-drug tolerance . An improved understanding of the role of lsr2 may provide important insights into the mechanisms of action of antibiotics and the way that mycobacteria adapt to stresses such as antibiotic treatment . Mycobacterium tuberculosis appears to generate specific and coordinated transcriptional responses to antibiotic treatment [1 , 2] . Several broad categories of genes are induced by antibiotics , including a number involved in stress responses and others linked to specific metabolic pathways that are inhibited by antibiotics [1] . The functional roles of antibiotic-induced transcriptional changes are poorly understood . Some changes are likely to be adaptive in that they induce antimicrobial tolerance or are important for intrinsic drug resistance [3–5] . Other changes are likely to be detrimental to the cell and may be ultimately linked to cell death . An improved understanding of the functional role of antibiotic-induced genes may be gained by studying the factors that regulate their expression . Histone-like proteins are reasonable candidates for regulators of antibiotic responses in bacteria because they assist in the control of stationary and exponential phase cell growth and regulate genes that respond to environmental changes [6 , 7] . Whether histone-like proteins influence the transcriptional response to antibiotics is not known . The M . tuberculosis iniBAC operon ( iniB or Rv0341 , iniA or Rv0342 , and iniC or Rv0343 ) encodes an important example of antibiotic-regulated genes . This operon is specifically induced by antibiotics such as isoniazid ( INH ) and ethambutol ( EMB ) that inhibit cell wall biosynthesis [2] , but it is not induced by other stresses , such as hydrogen peroxide and heat shock , or by factors such as lysozyme that digest the cell wall [8] . It has recently been demonstrated that M . bovis BCG strains overexpressing M . tuberculosis iniA grow and survive longer than control strains upon exposure to inhibitory concentrations of either INH or EMB , a condition analogous to classical antibiotic tolerance [3] . The goal of the current study was to identify genes required for transcriptional repression of the iniBAC promoter ( PiniBAC ) and to determine whether this regulatory pathway is part of a broader regulatory network in mycobacteria . Here , we demonstrate that the lsr2 gene product downregulates transcription of the M . tuberculosis iniBAC genes as well as another INH-induced pump , efpA . We found that Lsr2 exhibits properties similar to other bacterial histone-like proteins , suggesting that it regulates gene expression by controlling chromosomal topology . This study represents the first report to our knowledge of a gene that regulates antibiotic-induced transcription in M . tuberculosis and suggests that Lsr2 has important regulatory functions in mycobacteria . Previous studies have found an association between antibiotic-mediated induction of the iniBAC genes and multi-drug tolerance in BCG [3] . We performed transposon mutagenesis studies to identify repressors of iniBAC transcription and gain a better understanding of the regulation of drug tolerance . Mutagenesis was performed in NJS20 , an M . smegmatis Mc2155 strain that contained a single copy of the M . tuberculosis PiniBAC fused to a lacZ reporter inserted into attP [8] . NJS20 normally exhibits a subtle light tan-blue color when cultured on media containing X-gal ( Figure 1A ) . We found five strongly blue transposon mutants that had an unusual round and shiny colony morphology ( Figure 1A and 1B ) . This morphology had been previously reported for lsr2 transposon mutants [9] . Three of the blue colonies were selected and found to contain transposon insertions into two separate sites of the M . smegmatis lsr2 ( MSMEG6056 ) gene ( http://cmr . tigr . org/tigr-scripts/CMR/GenomePage . cgi ? org_search =&org=gms ) . We investigated the role of lsr2 in repressing basal and antibiotic-induced PiniBAC activity . An NJS20 ( lsr2::Tn5370 ) lsr2 transposon mutant ( NJS20 . 1 ) was cultured to mid log phase , EMB or control media was added to each culture for 24 h , and lacZ expression was measured . A second NJS20 strain with an intact lsr2 gene containing a random transposon insertion ( NJS20 . w ) was selected from the same transposon library to serve as a control . The results of these experiments showed that levels of β-galactosidase activity were approximately five times higher in the NJS20 . 1 lsr2 transposon mutant strain than in the control strain in the absence of antibiotic treatment ( Figure 2A ) . PiniBAC was also induced to higher levels in NJS20 . 1 than in NJS20 . w after treatment with EMB ( Figure 2B ) . The NJS20 . 1 lsr2 transposon mutant was then complemented by overexpressing the M . tuberculosis lsr2 gene ( Rv3597c ) in pMP167 , creating strain NJS20 . 1c . In the absence of antibiotic treatment , complementation restored the wild-type phenotype and lowered basal lacZ expression to levels not significantly different than those of the wild-type NJS20 . w control ( Figure 2A ) . The colony morphology also reverted to normal in the complemented strain ( Figure 1A ) . The complemented strain showed significantly less PiniBAC induction than the NJS20 . w transposon mutant control in the presence of EMB treatment . In other words , the presence of lsr2 on a multi-copy plasmid actually repressed EMB-mediated PiniBAC induction below that of the wild-type strain ( Figure 2B ) . These results indicate that the lsr2 gene controls both basal and antibiotic-induced levels of iniBAC expression . It has been shown previously that M . smegmatis strains overexpressing iniA are somewhat more resistant to EMB than controls [3] . We postulated that the NJS20 . 1 lsr2 mutant would also be more resistant to EMB due to de-repression of iniA . NJS20 . 1 was indeed more resistant to EMB than either NJS20 . w or NJS20 . 1c using the proportions method of susceptibility testing [10] ( Figure 3A ) . Complementation restored EMB susceptibility to the level of the control strain . We tested all strains with ciprofloxacin in the same manner as the EMB assays to address the possibility that the lsr2 activity is specific to cell wall antibiotics ( Figure 3B ) . All strains were equally susceptible to ciprofloxacin at all concentrations , suggesting that the role of lsr2 is limited to antibiotics that target the cell wall . Changes in colony morphology and antibiotic susceptibility can be associated with changes in cell wall permeability . Therefore , it was possible that the observed differences in PiniBAC induction could be due to increased entry of EMB into the cell . We measured the cell wall permeability of NJS20 . 1 to both hydrophilic and hydrophobic compounds by examining permeability to glycerol [carbonyl-C14] and chenodeoxycholic acid [carbonyl-C14] , respectively . No change in the intracellular levels of either compound was noted in NJS20 . 1 compared to control NJS20 . w ( unpublished data ) , indicating that disruption of lsr2 does not lead to substantial permeability changes . Microarray studies were performed to further investigate the role of lsr2 in transcriptional regulation under baseline and antibiotic-inducing conditions . A new M . smegmatis Δlsr2 strain that contained a complete unmarked deletion of lsr2 ( strain NJS22 ) was created to perform these studies . Three different comparisons were made: Expression of NJS20 was compared to NJS22 expression to identify genes that were up- or downregulated by deletion of lsr2 ( comparison 1 ) ( Table S1 ) . NJS20 grown in 7H9 media was compared to NJS20 cultured with EMB to identify genes that were induced by EMB in wild-type M . smegmatis ( comparison 2 ) ( Table S2 ) . NJS20 treated with EMB was compared to NJS22 treated with EMB to identify genes that were induced by EMB in a Δlsr2 background and to identify the EMB-induced genes that were further upregulated ( “hyperinduced” ) by lsr2 deletion ( comparison 3 ) ( Table S3 ) . All genes with statistically significant changes ( p < 0 . 05 ) in gene expression were included in each analysis ( Figure 4A ) . Comparison 1 revealed that 344 genes were upregulated and 286 genes were downregulated by lsr2 deletion . The increased ratio of up- versus downregulated genes was even more remarkable when the analysis was restricted to genes with statistically significant expression changes of 1 . 5-fold ( 146 up- versus 103 downregulated ) or 2-fold ( 41 up- versus 19 downregulated ) and is consistent with our hypothesis that lsr2 has broad and principally repressive effects on transcription . As predicted by the M . tuberculosis PiniBAC reporter assay , M . smegmatis iniA expression was significantly increased in NJS22 compared to NJS20 , although absolute upregulation was only 1 . 3 . Other broad categories of genes that were upregulated in condition 1 included genes involved in cell wall processes , metabolism , and transport ( Table S1 ) . Interestingly , stress response genes were not strongly represented among the genes upregulated in this comparison . The microarray studies allowed us to search for DNA sequences that might represent binding sites for Lsr2 or an associated protein . However , no consensus sequences were identified in alignments of up to 400 bp upstream of the 15 most strongly induced genes . In contrast , these regions were found to be unusually adenine-thymine ( AT ) –rich ( 43 . 2% AT compared to a mean of 32 . 6% AT in the M . smegmatis genome ) . Mapping the induced and repressed genes over the entire chromosome revealed a number of chromosomal regions with large clusters of highly upregulated genes ( Figure 4B ) . The upregulated genes within each cluster did not appear to comprise single operons ( Figure 4C ) . As with the 20 most highly induced genes , the regions 400 bp upstream of the upregulated genes shown in Figure 4C were unusually AT-rich ( 41 . 4% AT for region 1 and 40 . 5% AT for region 2 ) . These results are consistent with the hypothesis that lsr2 encodes a protein with relatively non-specific rather than sequence-specific DNA-binding properties that preferentially binds to AT-rich sequences in a manner similar to that of some other histone-like proteins [11] . Our reporter studies had shown that the M . tuberculosis PiniBAC was upregulated by lsr2 disruption in NJS20 . 1 , induced by EMB in NJS20 , and hyperinduced by EMB in NJS20 . 1 . We examined conditions 1–3 to identify the complete complement of M . smegmatis genes that exhibited this expression pattern . As predicted , we found that iniA was significantly upregulated/induced in all three conditions . Interestingly , only ten other genes had similar expression patterns ( Figure 4A; Table 2 ) ( p = 0 . 0001 that this number of genes were not present in all three conditions by chance ) . This group was overrepresented by genes involved in cell wall biosynthesis , transport , or other cell wall functions , providing a link to iniA , which appears to encode for a pump-associated protein in M . tuberculosis [3] . One hundred and twenty-one genes were induced in both condition 2 and 3 ( p = 0 . 0001 ) , indicating that many of the genes that are induced by EMB in wild-type M . smegmatis are hyperinduced in a Δlsr2 background . These results support the hypothesis that lsr2 is involved in controlling the level of expression of a subset of cell wall–active antibiotic-induced genes . Interestingly , only 21 genes were upregulated by condition 1 and induced in condition 2 ( p = 0 . 81 ) , while 41 were upregulated by condition 1 and induced in condition 3 ( p = 0 . 0001 ) . These results indicate that many of the genes controlled by lsr2 are not related to EMB treatment , suggesting that lsr2 is involved in the control of a broad range of cellular processes . The observation that lsr2 participates in repression of the M . tuberculosis PiniBAC in M . smegmatis suggested that lsr2 might perform a similar function in M . tuberculosis . We were unable to generate an M . tuberculosis Δlsr2 strain using allelic exchange methods to study this question directly [12] . Although this result cannot be taken as proof for gene essentiality , it is consistent with Himar1-based transposon mutagenesis studies , which indicate that lsr2 is essential in M . tuberculosis H37Rv [13] , and with the results of another transposon mutant screen in M . tuberculosis [14] in which Lsr2 insertions were only detected at the extreme 3-prime end of the gene ( R . McAdam , personal communication ) . We then decided to study the effect of lsr2 overexpression in M . tuberculosis in strain H327Rv by overexpressing M . tuberculosis lsr2 using the multi-copy plasmid pMV261::lsr2 ( creating strain NJT18 ) . NJT18 overexpressed lsr2 approximately 70-fold , as confirmed by quantitative PCR ( unpublished data ) . We cultured NJT18 and a wild-type H37Rv control strain containing the pMV261 vector ( H37Rv ( pMV261 ) ) to mid log phase , incubated the cultures with INH at a final concentration of 1 . 0 ug/ml ( or no antibiotic control ) for 24 h , and then measured expression of iniB , iniA , and iniC by quantitative PCR . We found that overexpression of lsr2 downregulated INH-mediated induction of the iniBAC genes in NJT18 compared to the H37Rv ( pMV261 ) control ( Figure 5 ) . These results are consistent with our discovery that overexpression of lsr2 in M . smegmatis repressed EMB-mediated induction of PiniBAC , and they confirm the role of lsr2 in repressing gene expression in M . tuberculosis . We examined the effect of lsr2 overexpression on kasA , efpA , and inhA expression in order to determine whether lsr2 acted specifically on the iniBAC operon or whether it had a more global effect ( Figure 5 ) . The kasA and efpA genes were examined because both genes are induced by INH . Furthermore , efpA has been annotated as a efflux pump [15] , suggesting that it might have functions analogous to the iniA-associated pump . Expression of inhA was studied as a control because INH does not induce this gene . We found that inhA expression was not affected by lsr2 overexpression . This indicates that lsr2 overexpression does not cause generalized repression of all gene expression in M . tuberculosis . INH-mediated induction of efpA and kasA were modestly downregulated in the lsr2 overexpression strain , although the downregulation of kasA induction did not appear to be statistically significant . Inactivation of lsr2 resulted in a remarkable change in colony morphology in M . smegmatis ( Figure 1 ) . A similar observation has been made previously by Chen et al . [9] . Colony morphology is often associated with a change in the cell wall structure [16–18] . Chen et al . noted that disruption of lsr2 in M . smegmatis was associated with the disappearance of two apolar lipids . We analyzed the lipid composition of the NJS20 . 1 lsr2 mutant in this study compared to that of the wild-type M . smegmatis Mc2155 strain by thin layer chromatography ( TLC ) . The apolar and polar lipids from wild-type M . smegmatis and NJS20 . 1 were extracted and analyzed by one-dimensional and two-dimensional TLC . In contrast to the previous observations in [9] , no difference was observed in the apolar fractions of the two strains ( unpublished data ) . However , a new spot was observed in the polar fractions of NJT20 . 1 ( Figure 6 ) . Similar results were noted after [1-14C]-acetate labeling of actively dividing cells . These experiments suggest that this new compound is a glycolipid because it was visualized with orcinol , a reagent for detecting glycolipids , and the compound migrated like a glycolipid [19] . Furthermore , the compound integrated [1-14C]-acetate , indicating that it contains fatty acids . Transposon mutants may be complicated by polar effects , although this phenomenon can usually be controlled for by complementation experiments . We considered the possibility that the discrepancy between our results and those of Chen et al . could have been due to differences in the location of the transposon insertion . However , we obtained identical results when we repeated the lipid analysis with the Δlsr2 strain NJS22 ( unpublished data ) . We prepared recombinant Lsr2 and then performed electrophoretic mobility shift assays ( EMSAs ) to characterize the ability of Lsr2 to specifically bind PiniBAC . A large mobility shift was observed when Lsr2 was incubated with a 227-bp PCR amplicon of the PiniBAC ( Figure 7A ) . Titration of Lsr2 against two concentrations of PiniBAC ( 5 and 50 fmole ) showed a dissociation constant ( Kd ) of approximately 1 μM ( Figure 7B ) . However , Lsr2 appeared to bind to DNA non-specifically , because a 200-bp PCR amplicon of the M . tuberculosis 16S rRNA gene produced similar mobility shifts and exhibited a similar Kd ( unpublished data ) . Furthermore , competition analysis with unlabeled PiniBAC and poly dI-dC DNA ( Figure 7C ) ( and 1 kb ladder; unpublished data ) demonstrated an equal or better ability to compete for Lsr2 binding . M . tuberculosis Lsr2 has a predicted mass of approximately 12 kDa and a pI of 10 . 69 These properties suggested that Lsr2 might have features in common with bacterial histone-like proteins [20] even though BLAST and iterative PSI-BLAST searches did not reveal any significant similarities . The mobility shifts observed in the EMSA assays indicated a complex that was much larger than would be expected by the association of a single 12-kDa Lsr2 molecule with its DNA target ( Figure 7A ) . The formation of large protein–DNA complexes has also been reported with histone-like proteins [21–23] . We performed cross-linking studies between Lsr2 and PiniBAC DNA to test the specificity of the interaction between Lsr2 and DNA and to rule out the possibility that these complexes were caused by electrostatic binding artifacts . Treatment with 0 . 1% sodium dodecyl sulfate ( SDS ) caused the Lsr2–DNA complexes to dissociate , resulting in a loss of the original gel shift ( Figure 7D ) . However , the shift was recovered by the addition of 0 . 1% of glutaraldehyde to the samples prior to SDS treatment . Lsr2 was then incubated with PiniBAC in the presence of glutaraldehyde for various times ( Figure 8 ) . Lsr2 multimers were detectable ( in the form of multiple bands ) as early as 1 min; longer incubation times produced very large complexes . Similar results were obtained after incubating Lsr2 with 1 kb ladder molecular weight marker DNA under similar conditions ( unpublished data ) . These results demonstrate that Lsr2 directly interacts with a broad range of DNA sequences , resulting in the formation of large oligomeric complexes . Histone-like proteins have been reported to preferentially bind supercoiled DNA compared to linear DNA [20 , 22 , 24] . We incubated various amounts of Lsr2 with supercoiled and linear pCV125 vector to examine binding preference in an agarose-based gel shift assay . Identical experiments were also performed with a pCV125 vector containing the PiniBAC sequence ( pG21898–12 ) to determine whether Lsr2 preferentially bound to PiniBAC under either of these conditions . A gel shift was only observed in the presence of the supercoiled plasmid . The results were similar whether or not the pCV125 plasmid contained PiniBAC sequences ( Figure 9 ) . Histones and histone-like proteins are typically able to protect DNA from degradation by DNase [20 , 23] . This property was tested in Lsr2 by first incubating ΦX174 DNA with Lsr2 and then treating the complex with 0 . 02 or 1 . 0 unit of DNase I for 1 min . Lsr2 was inactivated in each sample by treatment with protease K followed by boiling in SDS prior to analysis by gel electrophoresis to prevent a confounding gel shift of the Lsr2-treated DNA . We found that DNase I digested ΦX174 DNA into small fragments averaging less than 100 bp in size in the absence of Lsr2 pretreatment ( longer periods of DNase I digestion completely digested the DNA ) . In contrast , DNase I activity was substantially inhibited by pretreatment with Lsr2 ( Figure 10 ) . Lsr2 pretreatment followed by digestion with 0 . 02 unit of DNase I produced a wide range of DNA fragments ranging from approximately 150 bp to the size of the supercoiled vector . Lsr2 pretreatment also conferred some protection against the activity of the higher concentration of DNase I , resulting in DNA fragments with an average size of approximately 150 bp ( Figure 10 ) , while this same concentration of DNase I completely digested the DNA sample in the absence of Lsr2 pretreatment ( unpublished data ) . Heat-treated Lsr2 retained the ability to inhibit DNase , which is consistent with the known heat stability of histone-like proteins [25 , 26] . This protective effect could be the result of a histone-like interaction between Lsr2 and the DNA target of DNase . Alternately , Lsr2 could be inhibiting DNase due to a direct interaction between the Lsr2 and DNase proteins . DNase could not be co-eluted with Lsr2 bound to a nickel matrix when this experiment was performed to test for protein–protein interactions . These results suggest that Lsr2 does not interact directly with DNase . We also tested the ability of Lsr2 to inhibit transcription in vitro as has been reported for other histones and histone-like proteins [27 , 28] . We used a standard in vitro T7 promoter–based transcription assay of a pGEM vector for these experiments because Lsr2 did not appear to bind specifically to M . tuberculosis DNA sequences . In the absence of Lsr2 pretreatment , transcription of the pGEM vector produced the expected 1 . 0- and 2 . 3-kb mRNA transcripts ( imidazole was added to these reactions to control for the presence of imidazole in the buffer containing Lsr2 ) . The expected mRNA transcripts were also present in transcription reactions containing 200 ng of Lsr2; however , 600 ng of Lsr2 completely inhibited transcription ( Figure 11 ) . In order to confirm that transcription was not being inhibited by non-specific effects of an added protein , we repeated these experiments after identical amounts of M . tuberculosis ESAT 6 protein were added to the transcription reaction . In contrast to Lsr2 , ESAT 6 did not inhibit transcription ( unpublished data ) . DNA relaxation assays are often used to characterize the ability of histones and histone-like proteins to introduce supercoils into relaxed DNA in the presence of topoisomerase I [7 , 22] . We relaxed supercoiled ΦX174 DNA with topoisomerase I , then added Lsr2 and measured its ability to re-introduce supercoils . Lsr2 produced a small degree of additional supercoils to the relaxed DNA in this assay , consistent with histone-like activity . A small amount of linear DNA was also produced , suggesting that Lsr2 has nuclease properties ( Figure 12A ) . These results could indicate that Lsr2 has only a modest histone-like ability to introduce supercoils into DNA . However , it was possible that Lsr2 also inhibited the interaction between topoisomerase I and the DNA target , which is also necessary for the introduction of supercoils in this assay . To differentiate between these two possibilities , we repeated the DNA relaxation assay , this time simultaneously adding Lsr2 and topoisomerase I to the supercoiled ΦX174 DNA . We found that topoisomerase I produced substantially less relaxed DNA when it was co-incubated with Lsr2 , especially when higher amounts of Lsr2 were used ( Figure 12B ) . Inhibition of topoisomerase I appears to be a novel activity that has not been reported for other bacterial histone-like proteins . We have shown that Lsr2 is a histone-like protein with broad downregulatory and ( to a lesser extent ) upregulatory activity in M . smegmatis . The lsr2 gene also appears to regulate the degree of EMB-mediated induction in a large set of genes that are induced in wild-type M . smegmatis by EMB . In M . tuberculosis , lsr2 appears to downregulate antibiotic-mediated induction of the M . tuberculosis iniBAC and efpA genes . Our results suggest that lsr2 has a role in regulating the drug tolerance phenotype that is associated with iniA overexpression in M . smegmatis and BCG . Lsr2 is also likely to have other regulatory functions , including those associated with cell wall biosynthesis , transport , and responses to antibiotic treatment . The lsr2 genes of M . leprae , M . tuberculosis , and M . smegmatis share an unusually high degree of homology ( 87% identity and 91% similarity for M . leprae compared to M . tuberculosis; 87% identity and 90% similarity for M . smegmatis compared to M . tuberculosis ) , suggesting an important biological role in these species . Lsr2 was first reported to be one of the major seroreactive proteins in M . leprae patients , and was further characterized as a seroreactive protein in both leprosy and tuberculosis [29 , 30–32] . Our discovery that Lsr2 binds DNA may explain the high immunoreactivity observed in these prior studies . We postulate that Lsr2 exists as a complex with mycobacterial DNA in extracellular fluid , where it serves as a potent adjuvant by simulating TLR-9 in macrophages and dendritic cells [33] . However , the importance of this immune response in immunopathogenesis or host immunity to tuberculosis or leprosy remains unclear . The lsr2 gene has been previously reported to be induced by a number of stress conditions , including starvation , heat shock , and INH treatment [34–36] . Lsr2 production was also found to be induced in cultures supplemented with iron [37] . The association between induction of lsr2 expression and these stress responses may provide additional clues to its role in cellular regulation . Histones have been shown to broadly regulate transcription in eukaryotic cells through their influence on chromosomal topology [20] . The histone-like proteins of bacteria represent a diverse group of molecules that share the common property of small size and strong positive charge . Bacterial histone-like proteins have been associated with regulation of various cell stresses or responses to environmental changes [7 , 38 , 39] . Some histone-like proteins in Escherichia coli have been shown to directly affect antibiotic resistance by controlling expression of efflux pumps [40] . Disruption of hupA ( one of the two genes encoding the HU protein ) in E . coli K-12 had recently been shown to cause morphological changes similar to those we observed in NJS20 . 1 and NJS22 [41] . Very little is known about histone-like proteins in mycobacteria . The M . smegmatis hlp gene encodes a histone-like protein that is induced by cold-shock [42] and anaerobic-induced dormancy [43] . The hlp gene was also found to be important for invasion of M . leprae into peripheral nerves , and it has been hypothesized that it acts as an adhein during mycobacterial infections [44] . MDP1 , the M . tuberculosis and BCG homolog of hlp , has recently been shown to bind DNA and inhibit transcription in vitro in a manner similar to that of Lsr2 . MDP1 is also induced in stationary phase cultures , and appears to participate in the binding of M . tuberculosis to alveolar epithelial cells [45–47] . However , unlike lsr2 , MDP1 has homologies to other histone-like proteins , such as hlp and the E . coli HU protein; furthermore , the regulatory roles ( if any ) of MDP1 , and other histone-like proteins in M . tuberculosis , are not known . In addition to its size and pI , Lsr2 shares many properties with other bacterial histone-like proteins . Our microarray studies identified clusters of genes with AT-rich 5-prime untranslated sequences that were induced in the Δlsr2 strain . This finding closely parallels the activity of the histone-like protein H-NS , which transcriptionally silences clusters of laterally acquired genes in Salmonella by binding to AT-rich sequences [11] . We showed that Lsr2 forms large multimeric complexes with DNA ( with a preference to supercoiled forms ) , protects against DNase I treatment , and introduces a modest degree of supercoiling into relaxed plasmids , properties consistent with histone-like proteins . Lsr2 also appears to inhibit in vitro transcription and topoisomerase I . Co-elution studies did not detect any interactions between Lsr2 and DNase , suggesting that Lsr2 exerts its suppressive effect by interacting with DNA rather than by directly inhibiting proteins . Given the other similarities of Lsr2 to histone-like proteins , it is likely that Lsr2 inhibits RNA polymerase and topoisomerase activity by causing topological changes to the DNA targeted by these enzymes . This inhibition may be related to the ability of Lsr2 to form large oligomeric complexes with DNA . It is possible that the activity of Lsr2 is modulated in vivo by other cellular proteins and by local variations in chromosomal sequences; however , this remains to be determined . A PSI-BLAST analysis of the Lsr2 sequence reveals a nuclease motif , which is consistent with the weak nuclease activity that was noted in some of our experiments and may be related to its function . The Lsr2 sequence is unique , exhibiting no significant similarities to any histone-like protein . Thus , Lsr2 represents a novel class of histone-like proteins . Our discovery that Lsr2 is involved in regulating a subset of INH- and EMB-inducible genes suggests at least one significant function . Investigations of the cellular responses to antibiotic treatments ( as distinct from investigations of antibiotic resistance mechanisms ) are in their infancy . Although deletion of the histone-like protein gene hns in E . coli was found to de-repress expression of multi-drug transporters related to TolC and confer multi-drug resistance [40] , this work represents the first investigation to our knowledge of the regulation of antibiotic-induced genes in M . tuberculosis . Studies of cellular responses to antibiotics may be crucial to understanding the mechanisms by which bacteria survive or die in the presence of antibiotics . For example , we previously demonstrated that the iniBAC genes are induced by INH , EMB , and a number of other antibiotics that act by inhibiting cell wall biosynthesis in M . tuberculosis [8] . Induction of iniA was shown to confer multi-drug tolerance through the action of a multi-drug resistance–like pump [3] . Antibiotic tolerance can occur through other mechanisms such as overproduction of various inhibitors , enzymes , and regulatory proteins in non-mycobacterial bacteria [4 , 5 , 48–50] . It is possible that a multi-functional lsr2 also regulates these and other pathways in mycobacteria . Deletion analysis of lsr2 in M . tuberculosis would be particularly useful for studies of its function . Unfortunately , we have been unable to delete lsr2 from this species , although deletion was easily accomplished in M . smegmatis . The lsr2 gene may be essential in M . tuberculosis . lsr2 was suggested to be essential by Himar1-based transposon mutagenesis [13] , and all of the lsr2 transposon mutants characterized by McAdam et al . [14] contained insertions at the extreme 3-prime end of the gene where the transposon would be unlikely to affect functional capacity . If confirmed by future studies , the finding that this gene is essential is consistent with our hypothesis that lsr2 regulates important cellular pathways in M . tuberculosis . The lsr2 gene has been found to be essential for biofilm formation in M . smegmatis . This study also showed that an M . smegmatis lsr2 transposon mutant had altered colony morphology and contained two previously unidentified apolar lipids that were novel mycolate-containing compounds [9] . We found the same altered colony morphology but did not detect any novel apolar lipids in our analysis of the M . smegmatis ( lsr2::Tn5370 ) strain NJS20 . 1 or in the Δ lsr2 strain NJS22 . However , we did detect a new compound , possibly a glycolipid , in the polar fraction of both NJS20 . 1 and NJS22 . The dissimilarity between these two lipid analyses could be due to a difference in the way the strains were grown or harvested for the TLC analysis . Despite the apparent contradiction between the two studies , both investigations suggest that lsr2 is involved in regulating a wide range of cellular processes . In summary , lsr2 encodes a histone-like DNA-binding protein that appears to be essential for controlling responses to certain types of antibiotic stress . Lsr2 has also been linked to other types of stress responses and other cellular functions in mycobacteria . An improved understanding of the role of lsr2 and of the stress responses associated with this gene may provide important insights into the mechanisms of action of antibiotics and the way that mycobacteria adapt to certain types of stresses such as antibiotic treatment . This knowledge could in turn be used to design more effective antibiotic treatments for both drug-susceptible and drug-resistant M . tuberculosis . E . coli DH5α was the host for all plasmid constructions . Experiments with M . smegmatis either used Mc2155 or , in the case of the transposon mutagenesis experiments , NJS20 , a Mc2155 strain containing the pG21898–12 plasmid integrated into Mc2155 chromosome at attP ( Table 1 ) [8] . The pG21898–12 plasmid is a reporter construct that contains the M . tuberculosis PiniBAC fused to lacZ [8] . Experiments with M . tuberculosis used strain H37Rv . E . coli was cultured at 37 °C in Luria-Bertani medium with the addition of hygromycin B ( 200 μg/ml; Sigma , http://www . sigmaaldrich . com ) or kanamycin ( 40 μg/ml; Sigma ) where appropriate . M . smegmatis strains were grown at 37 °C on a rotary shaker in Middlebrook 7H9 medium ( Difco , http://www . vgdusa . com/DIFCO . htm ) containing 0 . 05% Tween 80 , 0 . 02% glycerol , 10% ADC ( Sigma ) [51] , and 25 ug/ml kanamycin or 40 μg/ml apramycin as appropriate . Transposon mutants of NJS20 were cultured in the presence of hygromycin B ( 50 μg/ml ) . M . tuberculosis strains were cultured at 37 °C on a rotary shaker in Middlebrook 7H9 medium ( Difco ) containing 0 . 05% Tween 80 , 0 . 02% glycerol , and 10% ADC with 12 . 5 μg/ml kanamycin added as appropriate . The complete lsr2 gene ( nucleotides 4040981–4041319 ) was amplified by PCR from H37Rv chromosomal DNA using primers pMV261-lsr2F and pMV261-lsr2R ( Table 1 ) , digested with PvuII and ClaI , and then cloned into pMV261 [52] ( which encodes for kanamycin resistance ) to create pMV261::lsr2 , or into pMP167 [53] ( which encodes for apramycin resistance ) at the PstI/ClaI sites to create pMP167::lsr2 ( Table 1 ) . M . tuberculosis lsr2 ORF ( nucleotides 4040981–4041319 ) was inserted into the NdeI/XhoI sites of pET-30 , creating pET::lsr2 . The pET::lsr2 plasmid was transformed into BL-21 E . coli competent cells . Plasmids pCV125 , pG21898–12 , and pBluescript have been described elsewhere [8] . Transposon mutagenesis was performed in the M . smegmatis iniBAC promoter reporter strain NJS20 as described [54] using the minitransposon vector pJSC84 , which contains inverted repeats flanking a hygromycin cassette [54] . The construct was packaged in a TM4 temperature-sensitive phage , and transfected into NJS20 . Cells were grown in 7H9 to mid log phase , prewarmed to the non-permissive temperature of 37 °C , and then mixed with 1010 pfu/ml ( multiplicity of infection 10 ) . The cell–phage mixture was incubated at the non-permissive temperature of 30 °C for 30 min , plated on 7H10 agar containing IPTG and β-galactosidase , and then incubated at 37 °C for 2–3 d . Strain NJS22 , an M . smegmatis strain containing a complete unmarked deletion of lsr2 , was created using the sacB counter selection method as described previously [55] . Briefly , DNA sequences from 6154142 to 6155817 and from 6156110 to 6157718 in M . smegmatis Mc2155 were PCR amplified using primer pairs F1-lsr2KO - R13-lsr2KO and F2-lsr2KO - R2 lsr2KO , respectively ( Table 1 ) , and cloned into the p2NIL vector [55] , followed by insertion of a PacI cassette containing sacB and lacZ . Blue colonies containing single crossover events were identified on X-gal/kanamycin media . Double crossover events from these blue colonies were selected on 2% sucrose/X-gal media . Deletion mutants were confirmed by real-time PCR for the lsr2 gene using primers Flsr2SC and Rlsr2SC ( Table 1 ) . Assays were performed as described [8] using o-nitrophenyl-β-D-galactopyranoside ( 4 mg/ml; Sigma ) to detect the presence of β-galactosidase activity . β-galactosidase units were calculated using the formula 1 , 000 × OD420/time ( minutes ) × 0 . 5 × OD590 . cDNA probes for microarray experiments were generated as previously described [56] . One microgram of mRNA in a mixture containing 6 μg of random hexamers ( Invitrogen , http://www . invitrogen . com ) , 0 . 01 M dithiothreitol , an aminoallyl-deoxynucleoside triphosphate mixture containing 25 mM each dATP , dCTP , and dGTP , 15 mM dTTP , and 10 mM amino-allyl-dUTP ( aa-dUTP ) ( Sigma ) , reaction buffer , and 400 units of Powerscript reverse transcriptase ( Clontech , http://www . clontech . com ) was incubated at 42 °C overnight . The RNA template then was hydrolyzed by adding NaOH and EDTA to a final concentration of 0 . 2 and 0 . 1 M , respectively , and incubating at 65 °C for 15 min . Unincorporated aa-dUTP was removed with a Minelute column ( Qiagen , http://www . qiagen . com ) . The probe was eluted with a phosphate elution buffer ( 4 mM KPO4 [pH 8 . 5] , in ultrapure water ) , dried , and resuspended in 0 . 1 M sodium carbonate buffer ( pH 9 . 0 ) . To couple the amino-allyl cDNA with fluorescent labels , nhs-Cy3 or nhs-Cy5 ( Amersham , http://www . amersham . com ) was added at room temperature for 2 h . Uncoupled label was removed using the Qiagen Minelute PCR purification . Microarray studies of each condition were performed with three separate RNA samples obtained from separate cultures using a dye-flip protocol ( two microarrays for each RNA sample ) . Epoxy- or aminosiline-coated slides were prehybridized in 5x SSC ( 1x SSC is 0 . 15 M NaCl plus 0 . 015 M sodium citrate ) ( Invitrogen ) , 0 . 1% SDS , and 1% bovine serum albumin at 42 °C for 60 min . The slides then were washed at room temperature with distilled water , dipped in isopropanol , and spun dry . Equal volumes of the appropriate Cy3- and Cy5-labeled probes were combined , dried , and then resuspended in a solution of 40% formamide , 5x SSC , and 0 . 1% SDS . Resuspended probes were denatured to 95 °C prior to hybridization . The probe mixture then was added to the microarray slide and allowed to hybridize overnight at 42 °C . Hybridized slides were washed sequentially in solutions of 1x SSC and 0 . 2% SDS , 0 . 1x SSC and 0 . 2% SDS , and 0 . 1x SSC at room temperature , then dried , and scanned with an Axon GenePix 4000 scanner ( http://www . moleculardevices . com ) . Individual TIFF images from each channel were analyzed with TIGR Spotfinder ( http://www . tm4 . org ) . Microarray data were normalized by Iterative Log Normalization using TIGR MIDAS software ( http://www . tm4 . org ) . Filtration of data was done in Microsoft Excel . Median spot intensity values were used to calculate the log2 ratio and fold change . Genes considered significantly differentially expressed by microarray analysis were identified by a one class t-test analysis comparing flip dye pairs of microarray experiments . The p-value was calculated based on permutation with random sampling using an overall alpha value ≤ 0 . 05 . Unequal group variance was assumed by Welch assumption . Mid log phase cultures of M . smegmatis strains were grown in 7H9 media to an OD600 of 0 . 8 . Either glycerol [carbonyl-C14] ( Amersham Phar-macia Bio , http://www . gelifesciences . com ) or chenodeoxycholic acid [carbonyl-C14] ( American Radiolabeled Chemical , http://www . arcincusa . com ) were added at a concentration of 0 . 5 μCi/ml and 2 . 0 μCi/ml , respectively , to the cultures to test the permeability of glycerol and chenodeoxycolic acid . The cultures were again incubated at 37 °C with gentle shaking . At different time points , 200 ml of each culture were removed and applied to glass fiber filters ( Schleicher & Schuell , http://www . whatman . com ) . A vacuum was applied using a vacuum filtration unit . Samples were washed twice with water and once with ethanol . Filters were dried at room temperature for 10 min and the amount of radioactivity was measured by scintillation counting . Purified recombinant Lsr2 protein was obtained by fusing the M . tuberculosis lsr2 gene with the 6-histidine tag at the NH2 terminus using the pET-30 vector . The 12-kDa–tagged rLsr2 protein was purified to near-homogeneity by nickel affinity chromatography ( Amersham Pharmacia Bio ) under non-denaturing conditions using an imidazole gradient . Eluted fractions were assessed in a 12% polyacrylamide gel , revealing a single band of the expected molecular weight . Western blot analysis using anti–His tag antibody ( Amersham Pharmacia Bio ) confirmed the presence of a single band corresponding to Lsr2 . The 227-bp promoter region of PiniBAC ( nucleotides 409379–409560 ) was amplified by PCR reaction using α-dCTP in the reaction . Radiolabeled PiniBAC , poly dI-dC , or 1 kb ladder were incubated with different concentrations of Lsr2 protein for 20 min on ice in a 10-μl reaction cocktail containing 20 mM Tris-HCl ( pH 7 . 0 ) , 0 . 01% BSA , 2 mM DDT , and 10 mM NaCl . Protein–DNA complexes were resolved by electrophoresis through a 5% polyacrylamide gel for 1 h at 4 °C and then examined by autoradiography . Different amounts of recombinant Lsr2 were incubated with PiniBAC DNA in a mixture containing 20 mM Tris-HCl ( pH . 7 . 0 ) , 0 . 01% BSA , 2 mM DDT , and 10 mM NaCl . Cross-linking was performed by the addition of 0 . 1% of glutaraldehyde ( Sigma ) to the mix . Aliquots were taken and loaded onto a 10% polyacrylamide gel at different time points . ΦX174 ( Promega , http://www . promega . com ) plasmid DNA ( 0 . 5 μg ) was incubated for 20 min on ice with or without Lsr2 ( 200 ng ) in a final volume of 50 ul . Either 0 . 02 or 1 unit of DNase I ( New England Biolabs , http://www . neb . com ) was then added , and the mixture was incubated at 37 °C for 1 min . DNase was inactivated by incubating at 75 °C for 15 min . The samples were then treated with 6% SDS and 4 mg/ml of protease K for 30 min at 37 °C , and then analyzed on a 1% agarose gel . Recombinant Lsr2 ( 1 ug ) was incubated with a nickel Sepharose matrix ( GE Healthcare , http://www . gehealthcare . com ) for 1 h at room temperature . The unbound Lsr2 was removed by washing the matrix three times with 0 . 1 M PBS , 20 mM MgCl2 , and 20 mM NaCl . DNase I ( 5 ug; New England Biolabs , http://www . neb . com ) was added to the nickel matrix–Lsr2 and incubated overnight at 4 °C . After washing using 0 . 1 M PBS , 20 mM MgCl2 , and 20 mM NaCl , Lsr2 was eluted from the nickel matrix using 0 . 1 M PBS , 20 mM MgCl2 , 20 mM NaCl , and 20 mM imidazole . Eluted proteins were assessed in each fraction by Coomassie blue–stained SDS-PAGE . Briefly , 0 . 5 μg of pGEM was incubated with either control imidazole buffer or either 200 or 600 ng of Lsr2 in a final volume of 20 ul , followed by in vitro transcription according to the manufacturer's recommendation using the Riboprobe in vitro Transcription Systems ( Promega ) . Samples were treated with 6% SDS and 4 mg/ml protease K for 30 min at 37 °C and then analyzed on a 1% agarose gel . The assay was carried out using a standard protocol [22] . Briefly , 200 ng of ΦX174 RT DNA ( Promega ) was incubated in the presence of different units ( 2–12 units ) of topoisomerase I ( Invitrogen ) , with and without Lsr2 , and incubated at 37 °C for 30 min . After incubation , 6% SDS and proteinase K ( 4 mg/ml; Invitrogen ) were added to the mixture , which was incubated for 15 min at 37 °C . Samples were run on a 0 . 7% agarose gel and then stained with ethidium bromide ( Sigma ) for analysis . Total RNA was extracted from M . tuberculosis strains . cDNA synthesis and quantitative PCR with molecular beacons were performed as described previously [57] . The molecular beacons and primers used to study expression of kasA and inhA were described previously [57] , and the molecular beacons and primers used to study expression of lsr2 , iniB , iniA , iniC , and efpA are shown in Table 1 . M . smegmatis strains ( 50 ml ) were grown in Mueller-Hinton broth ( Difco ) at 37 °C up to late log phase . The cultures were spun and the cell pellets were washed once with PBS buffer . After discarding the supernatants , the cell pellets were lyophilized . The dried pellets were treated as described previously [19] . Briefly , the cell pellet ( 0 . 2 g ) was resuspended in methanol/0 . 3% aqueous NaCl solution ( 2 ml; 10/1 , v/v ) and the suspension was extracted twice with petroleum ether ( 1 ml ) for 15 min at room temperature . The petroleum ether phases were combined and dried under nitrogen to yield the apolar lipid fraction , which was resuspended in dichloromethane . The methanol/saline fraction was heated at 100 °C for 5 min , cooled , and treated with chloroform/methanol/0 . 3% aqueous NaCl solution ( 2 . 5 ml; 9/10/3 , v/v/v ) at room temperature for 45 min . After centrifugation , the solvent fraction was removed and the residue was extracted with chloroform/methanol/0 . 3% aqueous NaCl solution ( 1 ml; 5/10/4 , v/v/v ) at room temperature for 15 min and spun ( 3000g , 2 min ) . The solvent fractions were combined , mixed with chloroform/0 . 3% aqueous NaCl solution ( 3 ml , 1/1 , v/v ) , vortexed , and spun . The upper phase was discarded and the lower phase was dried under nitrogen to yield the polar lipids fraction , which was solubilized in chloroform/methanol ( 2/1 , v/v ) . The lipid fractions were analyzed by one-dimensional and two-dimensional TLC using silica gel 60F-254 TLC plates ( Alltech , http://www . alltech . com ) . The following solvent systems were used to run one-dimensional TLC: hexane/ethyl acetate ( 9/1 , v/v ) ; hexane/ethyl acetate ( 1/1 , v/v ) ; ethyl acetate; chloroform/methanol ( 95/5 , v/v ) ; chloroform/methanol/water ( 90/10/1 , v/v/v ) ; and chloroform/methanol/water ( 60/30/6 , v/v/v ) . For two-dimensional TLC , the solvent systems used were as follows: either first dimension , chloroform/methanol/water ( 100/14/0 . 8 , v/v/v ) , second dimension , chloroform/acetone/methanol/water ( 50/60/2 . 5/3 , v/v/v/v ) ; or first dimension , chloroform/methanol/water ( 60/30/6 , v/v/v ) , second dimension , chloroform/acetic acid/methanol/water ( 40/25/3/6 , v/v/v/v ) . The lipids were visualized by spraying with a 10% sulfuric acid solution in ethanol or with an orcinol solution ( Alltech ) . The M . smegmatis strains were grown in Middlebrook 7H9 supplemented with 0 . 2% glycerol , 10% ADS enrichment , and 0 . 2% Tween 80 to log phase ( OD600 = 0 . 8 ) , and labeled with [1-14C]-acetate ( 15 μCi ) for 1 h . After centrifugation , the lipids were extracted as described above .	Understanding the cellular processes stimulated when Mycobacterium tuberculosis is treated with antibiotics may provide clues as to why months of therapy and use of several drugs simultaneously are required to prevent antibiotic resistance . Antibiotic treatment “turns on” or induces certain M . tuberculosis genes . These genes are of special interest because they appear to help M . tuberculosis survive the stress of antibiotic treatment . Our study of the regulation of antibiotic-induced genes , including iniBAC , in two mycobacterial species revealed that a small protein called Lsr2 controls iniBAC and other antibiotic-induced genes , especially ones related to the cell wall . Lsr2 binds to DNA in a relatively non-specific manner and appears to inhibit certain enzymes that must interact with DNA as part of their function . These properties differentiate Lsr2 from classical regulators of gene expression that bind to specific DNA sequences , and suggest that Lsr2 is a novel histone-like protein . These proteins regulate genes by changing the way DNA is shaped , and , indeed , we found that Lsr2 can change the shape of DNA by introducing a small number of coils into its structure . Our results suggest that Lsr2 is a major regulator of antibiotic-induced responses in mycobacteria .	[ "Abstract", "Introduction", "Results", "Discussion", "Materials", "and", "Methods" ]	[ "infectious", "diseases", "in", "vitro", "microbiology", "molecular", "biology", "genetics", "and", "genomics", "eubacteria" ]	2007	Transcriptional Regulation of Multi-Drug Tolerance and Antibiotic-Induced Responses by the Histone-Like Protein Lsr2 in M. tuberculosis
Positive-strand RNA viruses are the largest genetic class of viruses and include many serious human pathogens . All positive-strand RNA viruses replicate their genomes in association with intracellular membrane rearrangements such as single- or double-membrane vesicles . However , the exact sites of RNA synthesis and crucial topological relationships between relevant membranes , vesicle interiors , surrounding lumens , and cytoplasm generally are poorly defined . We applied electron microscope tomography and complementary approaches to flock house virus ( FHV ) –infected Drosophila cells to provide the first 3-D analysis of such replication complexes . The sole FHV RNA replication factor , protein A , and FHV-specific 5-bromouridine 5'-triphosphate incorporation localized between inner and outer mitochondrial membranes inside ∼50-nm vesicles ( spherules ) , which thus are FHV-induced compartments for viral RNA synthesis . All such FHV spherules were outer mitochondrial membrane invaginations with interiors connected to the cytoplasm by a necked channel of ∼10-nm diameter , which is sufficient for ribonucleotide import and product RNA export . Tomographic , biochemical , and other results imply that FHV spherules contain , on average , three RNA replication intermediates and an interior shell of ∼100 membrane-spanning , self-interacting protein As . The results identify spherules as the site of protein A and nascent RNA accumulation and define spherule topology , dimensions , and stoichiometry to reveal the nature and many details of the organization and function of the FHV RNA replication complex . The resulting insights appear relevant to many other positive-strand RNA viruses and support recently proposed structural and likely evolutionary parallels with retrovirus and double-stranded RNA virus virions . Positive-strand RNA [ ( + ) RNA] viruses contain messenger-sense , single-stranded RNA in their virions; they represent over a third of known virus genera; and they include many important human , animal , and plant pathogens [1] . A common , if not universal , feature of ( + ) RNA virus replication is the association of their RNA replication complexes with infection-specific host intracellular membrane rearrangements [2–19] . Characterizing the features of these membrane-associated RNA replication complexes should identify general principles and mechanisms of ( + ) RNA virus replication and could lead to broadly applicable control strategies for ( + ) RNA viruses including , e . g . , hepatitis C virus and the SARS coronavirus . For many ( + ) RNA viruses—including alphaviruses [5] , other members of the alphavirus-like superfamily [15] , rubiviruses [7 , 20] , flaviviruses [21] , tombusviruses [22] , and others [4 , 23–25] —RNA replication occurs in association with ∼50–70-nm diameter membranous vesicles or spherules that form in the lumen of specific secretory compartments or organelles . The similarity of these structures suggests that RNA replication by such otherwise distinct viruses involves important conserved features related to membranes . For some viruses , the localization of viral replicase proteins [11 , 17 , 23 , 26–28] or viral RNA synthesis [5 , 15 , 29] suggest that such spherules may contain or comprise the viral RNA replication complex . For brome mosaic virus ( BMV ) and some other viruses , two-dimensional ( 2-D ) electron microscopy ( EM ) reveals that a fraction of such spherules have interiors that appear to be connected to the cytoplasm by membranous necks [15 , 25 , 28] . However , limitations inherent in random sectioning and 2-D analysis prevent standard EM from resolving many issues crucial to understanding spherule structure and function , such as the range of spherule diameter and volume , and whether all spherule interiors are connected to the cytoplasm or if some bud free from their adjacent bounding membranes . To resolve these and other issues central to the mechanism of RNA replication , we used EM tomography ( EMT ) to provide the first , to our knowledge , three-dimensional ( 3-D ) ultrastructural study of the membrane-bound RNA replication complexes of a ( + ) RNA virus . EMT generates high-resolution , 3-D images or tomograms by digitally processing a series of 50–100 electron micrographs collected as a specimen is tilted in 1–2 ° increments on an axis perpendicular to the electron beam [30] . Similar 3-D EMT analyses have been crucial to reveal many important features of complex cellular organelles such as the Golgi apparatus [31–34] , endoplasmic reticulum [33 , 34] , and mitochondria [35–37] . We chose flock house virus ( FHV ) , the best characterized member of the Nodaviridae , as a ( + ) RNA virus with advantageous features for such studies . FHV has been used as a model to study RNA replication [8 , 9 , 38–40] , virion structure and assembly [41 , 42] , and genomic packaging [42–46] . FHV has a 4 . 5-kb bipartite RNA genome in which RNA2 ( 1 . 4 kb ) encodes the capsid precursor [47] whereas RNA1 ( 3 . 1 kb ) encodes an RNA silencing inhibitor [48 , 49] and a multifunctional RNA replication factor , protein A [40 , 50 , 51] . Protein A , the only FHV protein needed for RNA replication , is directed by an N-terminal targeting and transmembrane sequence to outer mitochondrial membranes , where it colocalizes by immunofluorescence with the sites of viral RNA synthesis [8 , 38] . Gradient flotation and dissociation assays showed that protein A behaves as an integral transmembrane protein [38] . Additionally , protease digestion and selective permeabilization after differential epitope tagging demonstrated that protein A is inserted into the outer mitochondrial membrane with the N terminus in the inner membrane space or matrix , while the majority of the protein A sequence is exposed to the cytoplasm [38] . Protein A also self-interacts in vivo in ways that are important for RNA replication [52] . Like many other ( + ) RNA viruses [4 , 5 , 7 , 15 , 20–25] , FHV infection induces the formation of ∼50-nm membranous vesicles or spherules , which , for the case of FHV , are found between the mitochondrial outer and inner membranes [8] . Here we use EMT and multiple complementary approaches to provide 3-D visualization of a ( + ) RNA virus replication complex . Among other findings , the results show that FHV spherules are compartments or mini-organelles for viral RNA synthesis , which form by invagination of the outer mitochondrial membrane and communicate with the cytoplasm through ∼10-nm diameter necks . The results further indicate that each spherule contains , on average , ∼100 membrane-spanning , self-interacting protein A molecules and that FHV-infected cells contain 2–4 genomic RNA replication intermediates per spherule . These observations define a new level of understanding of the nature , structure , and organization of a viral RNA replication complex , including principles that are likely relevant to many other ( + ) RNA viruses . Protein A is the only FHV protein needed for RNA replication and so must co-localize with viral RNA replication complexes . Prior immunofluorescence and immunogold labeling EM localized protein A to the outer mitochondrial membrane in FHV-infected cells [8] . However , in those prior attempts at immunogold labeling , fixation conditions needed to preserve spherule ultrastructure abolished protein A antigenicity for the polyclonal antibody used , hence blocking protein A localization relative to spherules . To overcome this , we identified a monoclonal antibody against protein A [9] that was able to detect protein A under fixation conditions that sufficiently retained spherule ultrastructure . Immunogold EM with this protein A monoclonal antibody revealed that nearly all protein A was in or on mitochondrial spherules in FHV-infected cells ( Figure 1 ) . Over 900 gold particles in 25 different electron micrographs were counted and 88% ± 5% of the specific gold labeling density above background ( see Materials and Methods ) was associated with spherules . Cytoplasmic labeling , presumably including protein A being translated and/or trafficked in the cytoplasm , was just 2% ± 7% above background labeling levels . The remaining 10% ± 5% of immunogold label was associated with mitochondria but not discernable spherules , including gold particles on the cytoplasmic face of the outer mitochondrial membrane where some protein A might have been localized that was not , or not yet , internalized into spherules . The clustering pattern of immunogold particles in a subset of spherules may be due to nonuniform epitope exposure or signal amplification by secondary antibodies . To avoid over-weighting the calculations due to such clustering effects , we also analyzed the same 25 micrographs counting each cluster as one event . The resulting count of clusters gave a very similar pattern to the one described above ( 94% spherule associated ) . By immunofluorescence microscopy , we found that 5-bromouridine ( BrU ) -labeled FHV RNA synthesis occurs exclusively at outer mitochondrial membranes in infected Drosophila cells [8] . To localize more precisely FHV RNA synthesis in relation to spherules , we incubated mitochondria isolated from uninfected and FHV-infected Drosophila cells with a nucleotide mix including 5-bromouridine 5'-triphosphate ( BrUTP ) and performed immunogold labeling EM with an antibody recognizing BrU incorporated into RNA , but not unincorporated BrUTP . For mitochondrial preparations from FHV-infected cells , spherules were the major site of immunogold labeling ( Figure 2A and 2B ) . Of 221 gold particles examined , 70% ± 18% were on spherules . The remaining 30% of gold particles that fell outside of spherules may include mature RNA products released from spherules and nonspecific background labeling . For mitochondria from uninfected cells , background labeling levels were independent of the addition or omission of BrUTP and averaged 15% of the total immunogold labeling of BrUTP-treated mitochondria from FHV-infected cells . We found that using isolated mitochondria was advantageous for the BrUTP-labeling experiments because of low transfection efficiencies of BrUTP into whole Drosophila cells . Nevertheless , we were able to obtain some immunogold labeling results using intact Drosophila cells , which also showed that spherules were the major sites of BrUTP-labeling ( Figure 2C ) . Gold particles in the intermembrane space of the mitochondrion in the lower right are well within the distance ( 20 nm ) from spherules that may be spanned by the primary and secondary antibodies linking the immunogold particles to their target epitopes [53] . Having shown that spherules were the sites of protein A accumulation and FHV RNA synthesis , we applied 3-D EMT to provide a new level of analysis of spherule morphology and topology . As noted in the Introduction , the 3-D nature of EMT overcomes many serious limitations of 2-D EM analysis to reveal possible connections to surrounding membranes and compartments , complete dimensions , and other fundamental characteristics not accessible from conventional transmission EM analyses of random sections . For example , along the z-axis parallel to the electron beam , standard transmission EM projects a 50–70-nm section into a single view , whereas EMT allows computationally dissecting an entire ∼250-nm-thick sample volume into successively viewable planes spaced with a resolution of just a few nanometers [54] . To produce 3-D reconstructions of FHV-infected cells including modified mitochondria , Drosophila S2 cells were harvested 12 h post infection ( hpi ) and fixed , embedded , and sectioned as described under Materials and Methods . For each reconstruction , a tilt series of 60 images was collected by rotating a 250-nm-thick section of resin-embedded sample in 2 ° increments between −60 ° to +60 ° relative to the plane perpendicular to the beam , and was digitally processed to produce a tomographic reconstruction . Using Drosophila cells from three independent FHV infection experiments , five independent reconstructions were generated using a single-tilt series technique ( Figure 3C–3D and additional unpublished data ) and one reconstruction was performed using a double tilt technique ( Figure 3A and 3B ) to improve tomographic resolution further [55] . Representative results are shown in the figures . For one such tomogram , Figure 3A shows the image of a computationally dissected , 2 . 2-nm-thick virtual section , revealing an FHV-modified mitochondrion containing spherules in the mitochondrial intermembrane space . This 2-D image shows a typical view of randomly sectioned , FHV-modified mitochondria , in which some spherules appear to be light bulb–shaped invaginations attached to the outer membrane by small diameter necks ( white arrowheads ) , whereas others appear to be free vesicles in the intermembrane space ( asterisks ) . Figure 3B shows another virtual section from the same tomogram , displaced down the perpendicular z-axis by ∼15 nm to a point where those spherules that appeared to be free vesicles in Figure 3A ( asterisks ) now show necked attachments to the outer membrane . To determine if all spherules were attached to the outer mitochondrial membrane , or if a population of spherules budded free of this membrane , we followed individual spherules through dozens of successive 2 . 2-nm-spaced adjacent planes perpendicular to the electron beam ( a “z-series” of sections ) . When all six reconstructions were examined in this way , all ∼500 spherules in all ∼8 FHV-modified mitochondria examined were found to be connected to the outer mitochondrial membrane by a membranous neck observable in some plane of the sample . The red arrowhead in Figure 3A points to a channel through the spherule neck that connects the interior of a spherule to the cytoplasm . Thus , all spherules are necked invaginations of the outer mitochondrial membrane whose interiors remain connected to the cytoplasm , and sections in which a given spherule appears to be a free vesicle simply represent planes that did not pass through the smaller diameter neck linking the spherule membrane to the mitochondrial outer membrane . This is illustrated more dynamically in Video S1 , which animates the progression through a z-series of sections of the tomogram of Figure 3A and 3B . Figure 3C–3D shows two virtual sections from another tomogram , which are displaced ∼150 nm down the perpendicular z-axis from each other . As shown in a video through this z-series ( Video S2 ) , mitochondrion 1 curves significantly in the space between these two sections , such that the plane of Figure 3C sections mitochondrion 1 spherules parallel to an axis through the spherule necks , whereas the parallel plane of Figure 3D sections the spherules on another part of the mitochondrion 1 surface tangential to the axes through their necks . These two perpendicular views of similar spherules on the same mitochondrion are notable because Figure 3C strongly resembles images of spherules induced by alphaviruses , nodaviruses , etc . [8 , 28] , whereas Figure 3D resembles images of apparently distinct “vesicle packets” described for flaviviruses [3] . Thus , some apparently distinct membrane rearrangements and vesicle structures observed in connection with RNA replication by different ( + ) RNA viruses may represent related structures distinguished in part by the perspective from which they were viewed . To generate 3-D surface maps of the virus-induced membrane rearrangements associated with FHV RNA replication , we manually traced the inner and outer mitochondrial membranes ( including spherules ) over ∼100 adjacent , 2 . 2-nm-spaced virtual sections of selected tomographic reconstructions , and we used a computer-generated mesh overlay to join these tracings into continuous surfaces ( Figure 4 ) . Figure 4A shows part of the relationship between the electron density of the mitochondrion in Figure 3A and its 3-D map , and Video S3 provides a much more dynamic visualization of this relationship and the complete 3-D map . For clarity , the cytoplasmic faces of outer mitochondrial membranes are colored blue , spherule membranes are white , and inner mitochondrial membranes are yellow . Figure 4B shows a close-up view of a portion of the 3-D map in Figure 4A that demonstrates the connection of the spherules to the outer mitochondrial membrane . This and other similar maps confirmed as noted above that the spherule membranes ( white ) are continuous with the outer mitochondrial membrane ( blue ) . Figure 4C is a 90 ° rotation of Figure 4B that shows a view looking down on the surface of an FHV-modified mitochondrion , with the outer membrane ( blue ) rendered translucent to reveal the spherules beneath ( Video S4 ) . The necked channels connecting the interior of each spherule to the cytoplasm ( red arrowhead ) are clearly visible as circular openings in the outer membrane . For 150 individual spherules in four mitochondria from four cells and three experiments , we measured the interior diameters of these neck channels as the distance between the two lipid bilayers , from inner leaflet to inner leaflet , at the point where the tomographic plane sliced through the center of the neck . The resulting distribution of neck diameters is shown in Figure 5A . The average diameter of the neck channel was 10 . 5 ± 1 . 8 nm ( Figure 5A ) , which is more than large enough to allow import of ribonucleotides and export of RNA products ( diameter < 2 nm ) . Surface-rendered , 3-D maps of the two mitochondria from Figure 3C are shown in Figure 4D , illustrating also the inner mitochondrial membrane ( yellow ) . Using such surface-rendered maps ( Figure 4 and other unpublished data ) , we also measured the interior volume and membrane surface area of 175 spherules . As illustrated in Figure 5C , the spherule volumes spanned a range of ∼15 , 000 to 50 , 000 nm3 . A range of spherule sizes is seen in Figure 4E , which is a rotated view of mitochondrion 2 in Figure 4D , with the outer membrane removed . The average spherule interior volume was ∼33 , 000 nm3 ( Figure 5C ) , and the average interior spherule membrane surface area was ∼6 , 000 nm2 ( Figure 5B ) . Since both protein A ( Figure 1 ) and nascent FHV RNA ( Figure 2 ) localized predominantly or exclusively to spherules , the relative numbers of protein A , RNA replication templates , and spherules could provide important insights into the structure and organization of FHV RNA replication complexes . Accordingly , we measured the number of molecules of protein A and FHV RNAs per cell in Drosophila S2 cells at 4 , 8 , 12 , and 24 hpi with FHV . The numbers of positive- and negative-strand genomic RNAs per cell were measured by quantitative Northern blotting calibrated with known amounts of in vitro transcripts ( Figure 6A and 6B ) . The number of protein A molecules per cell was measured by quantitative Western blotting calibrated with known amounts of co-electrophoresed , purified protein A standards ( Figure 6E ) . Starting before 4 hpi and continuing thereafter in FHV-infected Drosophila cells , the primary mode of viral RNA synthesis is ( + ) RNA synthesis from negative-strand RNA [ ( − ) RNA] templates ( Figure 6A–6D ) . The number of ( + ) RNA1 and ( + ) RNA2 per cell increased from ∼40 , 000 molecules of each RNA species at 4 hpi to ∼2–3 million each by 24 hpi ( Figure 6C ) . Such ( + ) RNA products primarily accumulate in the cytoplasm for translation and encapsidation , and only a minor fraction of ( + ) RNAs fractionate with the membrane-associated RNA replication complex ( P . Van Wynsberghe , P . Ahlquist , unpublished data ) . By contrast to positive-strand export and accumulation in the cytoplasm , FHV ( − ) RNAs appear to function only as RNA replication intermediates and are completely membrane-associated ( P . Van Wynsberghe , P . Ahlquist , unpublished data ) . ( − ) RNA thus is a key measure of a minimal RNA replication complex , because every mature RNA replication complex , active in ( + ) RNA synthesis , must contain at least one ( − ) RNA template . Therefore , the number of ( − ) RNAs gives an estimate of the maximal number of replication complexes per cell . ( − ) RNA1 accumulation plateaued by 8 hpi at ∼16 , 000 copies per cell ( Figure 6D ) . ( − ) RNA2 accumulation increased throughout the first 24 hpi , although more slowly after 12 hpi , reaching ∼50 , 000 molecules per cell by 24 hpi ( Figure 6D ) . The number of protein A molecules plateaued by 8 hpi ( Figure 6F ) , which is consistent with prior results that protein A synthesis occurs early in infection and then declines [47] . Intriguingly , the peak level of protein A was ∼2 million molecules per cell ( Figure 6F ) . Protein A was thus present at dramatically higher levels than ( − ) RNA templates were . The ratio of protein A to ( − ) RNAs was relatively consistent over all time points examined , with averages throughout infection of 118 ± 23 and 64 ± 20 protein A copies per ( − ) RNA1 and ( − ) RNA2 , respectively ( Figure 6G ) . To understand the organization of the replication complex in relation to the spherules better , we compared the number of spherules per cell with the number of protein A and ( − ) RNA molecules per cell . To measure the number of spherules per cell , we collected FHV-infected Drosophila S2 cells at 12 hpi , processed them for transmission EM , and imaged 25 randomly sectioned cell profiles . All spherules in each imaged cell section were counted and divided by the cell section volume , which was calculated by measuring the cell area using ImageJ ( National Institutes of Health ) and multiplying by the effective section thickness ( see Materials and Methods ) . The number of spherules per cell was calculated by multiplying the resulting density of spherules by the average volume of the almost perfectly round , 10 μm-diameter Drosophila S2 cells [56] ( and our independent , matching measurements ) . These calculations revealed the average number of spherules per cell at 12 hpi to be ∼20 , 000 ± 11 , 000 ( Table 1 ) . The ratio of protein A per cell to spherules per cell revealed that on average , there are ∼100 copies of protein A per spherule ( Table 1 ) . Further comparison to the Figure 6 data shows that , on average there are ∼1 ( − ) RNA1 and ∼2 ( − ) RNA2 molecules per spherule ( Table 1 ) . The implications of these results for the organization of replication complexes are considered further in the Discussion . Protein A is a transmembrane protein in outer mitochondrial membranes [38] and is ∼90% localized within spherules ( Figure 1 ) . Therefore , protein A must line the interior membrane surface of spherules . If protein A is similar to typical globular proteins , its volume would be ∼183 nm3 , based on the protein A molecular weight of 112 kDa [57] and the average partial specific volume of typical proteins [58] . If globular , protein A then would have a diameter of ∼7 nm and cover a surface area of ∼40 nm2 . Thus , the average spherule interior membrane surface area of 6 , 000 nm2 ( Figure 5B ) provides enough space to accommodate at most ∼150 protein A molecules , under a perfect close-packing arrangement . Therefore , the measured value of ∼100 protein A molecules per spherule ( Table 1 ) is near saturation for the spherule interior membrane surface area . We modeled 50 7-nm-diameter spheres representing protein A adjacent to the membrane surface within a tomographic model of half a typical spherule ( Figure 7 ) to demonstrate how protein A may pack into the spherules . The resulting near-full occupancy of the interior membrane surface area by protein A ( Figure 7 ) and the nature of protein A as a transmembrane protein whose self-interaction is required for RNA replication [38 , 52] imply that the ∼100 copies of protein A form an inner network or shell within the spherule ( Figure 7B ) . Such a shell would explain the formation and maintenance of the high-energy membrane deformation of spherules . A shell of these dimensions appears reasonable , given that the main shell of a reovirus core is 60 nm in diameter and is composed of 120 copies of a slightly larger protein λ1 ( 142 kDa ) [59] . The distribution of FHV spherule size spans a defined range of ∼30–45-nm intramembrane diameter , suggesting some flexibility in the assembly of the protein A shell . Other examples of high-density protein shells of flexible size and shape include the capsids of retroviruses , influenza , and retrotransposons [60–62] . For example , one species of Ty retrotransposon forms virus-like capsids that have a 30–50-nm range of diameters , similar to FHV spherules , and contain on average 300 copies of a 381–amino acid protein subunit [62] , a protein content very close to the FHV spherule average of 100 copies of 998–amino acid protein A . Endocytic vesicles , secretory transport vesicles , and synaptic vesicles are further examples of protein-induced membrane vesicles that each have a range of variable sizes ( 50–100-nm diameter ) , despite being formed by regular arrays of uniform proteins [63 , 64] . Because ∼90% of protein A , the FHV RNA polymerase ( Figure 1 ) , and ∼70% of newly synthesized FHV RNA ( Figure 2 ) are spherule-associated , and essentially all FHV ( − ) RNA templates are membrane associated ( P . Van Wynsberghe , P . Ahlquist , unpublished data ) , it appears likely that ( − ) RNAs , and thus any double-stranded RNAs ( dsRNAs ) are within spherules . Sequestration of dsRNA within such a compartment may allow the virus to avoid , minimize , or delay dsRNA-induced host–cell defense responses such as protein kinase , RNA activated ( PKR ) and RNase L [65] or RNA interference ( RNAi ) [66] . Such dsRNA localization is consistent with earlier observations of virus-induced membrane spherules containing fibrils with salt-dependent nuclease sensitivity [25 , 67] . The ∼100 protein A molecules per spherule ( Table 1 ) would consume ∼18 , 300 nm3 of interior volume , leaving ∼14 , 000 nm3 within an average spherule to accommodate FHV RNA . Based on 0 . 655 nm3 per hydrated nucleotide for the crystal structure of duplex RNA [43 , 68 , 69] , the volumes of FHV RNA1 , RNA2 , and RNA3 would be 2035 , 917 , and 254 nm3 , respectively . Thus , in addition to ∼100 protein A molecules , a spherule of average size has enough interior space to contain at most four single-stranded RNA ( ssRNA ) or two dsRNA copies of all three FHV RNA species . Given this maximal occupancy , the estimate from biochemical data of an average of one ( − ) RNA1 and two ( − ) RNA2 templates per spherule ( Table 1 ) , together with at least one nascent ( + ) RNA progeny strand for each , appears fully reasonable . Currently , it is not known if FHV RNA1 and RNA2 are replicated in separate or common spherules . If RNA1 and RNA2 were in separate spherules ( i . e . , 50% of spherules containing RNA1 and 50% containing RNA2 ) , then the ratios of ( − ) RNA1 and ( − ) RNA2 to total spherules ( Table 1 ) imply each RNA1-containing spherule would have two replication intermediates , and each RNA2-containing spherule would have approximately four replication intermediates . Because RNA1 ( 3 . 1 kb ) is twice as long as RNA2 ( 1 . 4 kb ) , the total RNA content in both cases then would be nearly equal . If RNA1 and RNA2 were together in the same spherule , then each spherule would hold on average three replication intermediates ( one RNA1 and two RNA2 ) . The possibility of spherules containing both species of RNAs is intriguing , considering the interactions of FHV RNAs required for replication: FHV subgenomic RNA3 , which is templated from RNA1 , transactivates RNA2 replication and , in turn , RNA3 replication is suppressed by the resulting progeny RNA2 [70] . RNA3 , and not its protein product , is responsible for transactivating RNA2 [70] . However , it is also possible that RNA3 is produced in one spherule during RNA1 replication and then exported to the cytoplasm prior to transactivating RNA2 . Membrane spherules similar to those of FHV are induced by many other ( + ) RNA viruses including alphaviruses [5] , other members of the alphavirus-like superfamily [15] , rubiviruses [7 , 20] , flaviviruses [21] , tombusviruses [22] , and others [4 , 23–25] . Among these , one of the best-studied with regard to the localization and stoichiometry of RNA replication complex components is BMV . BMV and FHV differ in many important respects including that BMV encodes a much larger complement of RNA replication proteins [15 , 19] . Nevertheless , although the understanding that we present here for FHV RNA replication complexes is more advanced in many ways , the known characteristics of BMV RNA replication complexes are strikingly similar to those for FHV . Both BMV and FHV induce spherules of similar dimensions where viral RNA synthesis and viral replication proteins are localized [15] . BMV replication protein 1a , which is sufficient to induce spherules [15] , is also a strongly membrane-associated [71] , self-interacting [72 , 73] protein that is present at high copy number per spherule [15] . Similarly , whereas the ultrastructural organization of hepatitis C virus RNA replication complexes has not been defined , recent results suggest that these may also involve a dramatic excess of nonstructural protein copies per ( − ) RNA [74] . In addition to the many ( + ) RNA viruses whose RNA replication is associated with spherules , other ( + ) RNA viruses induce various , apparently distinct membrane rearrangements [4 , 17 , 21 , 26 , 27 , 75 , 76] . Although some or most of this variability reflects real ultrastructural differences , at least some of the perceived differences may be due to differences in perspective under conventional 2-D imaging . Our tomography results demonstrated that equivalent FHV spherules appeared to vary in morphology and topological relation to adjacent membranes when viewed in two dimensions from different perspectives ( Figures 3C and 3D ) . A greater understanding of the 3-D nature of membrane rearrangements associated with RNA replication by other ( + ) RNA viruses may reveal shared features or common underlying principles . Based on results with BMV , Schwartz et al . identified potential parallels between the assembly , structure , and function of membrane-associated RNA replication complexes and the cores of reverse-transcribing and dsRNA virus virions , including the sequestration of genomic RNA templates within a virus-induced compartment for replication [15 , 19] . The results presented here for FHV validate and extend these parallels by showing that all FHV spherules are membrane invaginations topologically equivalent to a budding , enveloped virion ( Figure 4 ) , and that self-interacting , transmembrane protein A is present at levels sufficient to coat the inner spherule membrane in a multi-subunit shell similar to the capsids of retrovirus and dsRNA virus cores ( Figure 7 ) . As with dsRNA viruses , hepadnaviruses , and retroviruses , the high copy of protein A per spherule suggests that there may be threshold effects in replication protein expression to initiate replication . Further analysis of the structure , interactions , and function of FHV RNA replication complexes should provide additional insights into the basic mechanisms of ( + ) RNA virus replication and potentially identify new approaches for antiviral interference . Drosophila S2 cells were grown at 28 °C in Gibco Drosophila serum-free media ( SFM ) . Cells were dislodged by gentle scraping , pelleted , and resuspended at 107 cells/ml . FHV was added at a multiplicity of infection of 10 for all experiments . The cells and virus were incubated at 26 °C on a rotary shaker at 1 , 000 revolutions per minute ( rpm ) for 1 h to let the virus attach . After the hour incubation , the cells were plated onto a tissue culture dish and further incubated at 28 °C . Mitochondria were isolated from Drosophila cells as described by Echalier [77] . Briefly , cells were recovered by scraping and centrifugation and resuspended in a hypotonic buffer that contained 20 mM N-2-hydroxyethylpiperazine-N'-2-ethanesulfonic acid ( HEPES; pH 7 . 4 ) , 1 mM EGTA , and a protease inhibitor cocktail ( 1 mM phenylmethanesulphonylfluoride , 5 μg/ml pepstatin A , 1 μg/ml chymostatin , 10 mM benzamidine , 10 μg/ml leupeptin , and 0 . 5 μg/ml bestatin ) . After a 10 min incubation at room temperature , an equal volume of double isotonic buffer was added that consisted of the hypotonic buffer plus 0 . 5 M mannitol . The cells were lysed for 7 min using a pre-chilled Potter-Elvehjem homogenizer fitted with a Teflon pestle ( Kimble-Kontes; http://www . kimble-kontes . com/ ) and attached to a stirrer motor spinning at 250 rpm . The lysate was transferred to a Dounce homogenizer fitted with a type B glass pestle ( Kimble-Kontes ) and disrupted manually for 100 strokes on ice . Unbroken cells and nuclei were removed by two 10 min centrifugation steps at 500g at 4 °C . Mitochondria were pelleted by centrifugation at 3700g for 10 min at 4 °C , resuspended in an isotonic buffer containing 0 . 25 M mannitol , and washed by a second centrifugation at 7000g . BrUTP incorporation on the isolated mitochondria was performed at 28 °C for 1 h as described previously [15] . Drosophila cells were infected with FHV as above . At 8 hpi , cells were treated with 20 μg/ml actinomycin D for 30 min . FuGENE 6 ( Roche; http://www . roche . com ) was diluted 10-fold in phosphate buffered saline pH 7 . 4 and mixed with BrUTP and actinomycin D to final concentrations of 10 mM and 20 μg/ml , respectively . The FuGENE/BrUTP/actinomycin D mix was incubated for 15 min at room temperature then added to the cells and incubated at 4 °C for 15 min . After the 4 °C incubation , the cells were moved to 28 °C for a 15-min labeling period and then immediately fixed and processed for EM . Mouse monoclonal antibodies against FHV protein A have been described previously [9] . MAb clone 2–1 . 1 . 2 . 4 . 8 , which recognizes the protein A epitope between amino acids 230 and 399 , was used for immunogold EM labeling . BrUTP immunolabeling fixation was performed as described previously [15] , except that samples were embedded in LR Gold resin . Samples were sectioned and placed on nickel grids . Sections were blocked with a goat-blocking solution ( Aurion; http://www . aurion . nl ) , and incubated for 1 h with an anti-BrU antibody ( PRB-1; Molecular Probes; http://probes . invitrogen . com ) , diluted 1:100 in an incubation solution containing 100 mM phosphate-buffered saline pH 7 . 4 and 0 . 1% BSA-c ( Aurion ) . Grids were washed six times in incubation solution without antibody , then incubated for 2 h with a goat-anti-mouse antibody conjugated to an ultrasmall gold particle ( Aurion ) that was diluted 1:100 in incubation solution , and washed six times again with incubation solution . Silver enhancement was performed for 30 min using R-GENT SE-EM ( Aurion ) . Protein A immunogold EM was performed in the same manner using the mouse monoclonal antibody at a dilution of 1:100 . Background labeling was determined using uninfected control cells . Labeling density was determined by calculating the surface area of spherules , mitochondria , and cytoplasm using the point-hit method [78] . Specific labeling was determined by subtracting the background labeling density . Northern blotting was done as described previously [8] . The number of molecules of FHV RNAs was determined by comparison with a serial dilution of a known amount of in vitro transcripts representing a known amount of ( + ) RNA or ( − ) RNA molecules . RNA levels were quantitated with ImageQuant software ( Molecular Dynamics; http://www . mdyn . com/ ) . Western blotting was done as described previously [8] . The number of molecules of protein A was determined by comparison with a purified protein A standard . To generate the standard for quantitation , protein A was expressed in Escherichia coli as described previously [8] . To purify protein A , the hydrophobic transmembrane domain of protein A was deleted ( amino acids 8–89 ) , replaced with a C-terminal His6 tag , and purified by talon column ( Clontech; http://www . clontech . com ) affinity chromatography . To further purify protein A , we performed preparative electrophoresis using a BioRad mini-prep cell . A 6% , 9 . 5-cm gel was run at 200 V for 9 h with an elution speed of 150 μl/min . Fractions containing the purified , truncated protein A standard were collected and quantitated based on comparison with known standards of bovine serum albumin and β-galactosidase . We quantitated protein levels with Lumi-imager software ( Roche ) . For conventional transmission EM , cells were fixed and embedded as previously described [8] . For electron tomography , cells were fixed in 2% parafomaldehyde and 2 . 5% glutaraldehyde in 0 . 1 M sodium cacodylate , pH 7 . 4 , post-fixed in 1% osmium tetroxide with 0 . 8% potassium ferrocyanide in sodium cacodylate buffer , stained with 2% uranyl acetate , dehydrated in a graded series of ethanol , and embedded in Durcupan ACM resin . To calculate the total number of FHV-induced mitochondrial spherules per cell , FHV-infected Drosophila S2 cells were collected at 12 hpi , processed for transmission EM , and sectioned into 70-nm-thick slices . For each of 25 randomly selected cells imaged in these sections , we then counted all observable spherules with diameters larger than 20 nm . The number of spherules counted for each cell then was divided by the relevant section volume , which was calculated by measuring the cell area using ImageJ ( National Institutes of Health ) and multiplying by the effective section thickness . The effective section thickness is a correction used to avoid overcounting spherules with centers outside of the 70-nm physical section , which would otherwise be counted twice if adjacent sections were analyzed . As previously used to calculate synaptic vesicles per cell [79] , this effective section thickness is the thickness that would encompass the centers of all counted spherules . In this case , the effective section thickness was 116 nm , based on adding 23 nm ( the distance from the spherule center to a radius-perpendicular plane bisecting the spherule to yield a 20-nm-diameter section ) to each face of the 70-nm section . The number of spherules per cell was calculated by multiplying the resulting density of spherules by the average volume of the almost perfectly round , 10 μm-diameter Drosophila S2 cells [56] ( and our independent , matching measurements ) . Three separate FHV infections produced samples for six independent tomograms . To survey the preservation quality and FHV-infection efficiency of the Drosophila cells , thin-sectioned material ( ∼80 nm thick ) was examined using a JEOL 1200FX electron microscope . 3-D reconstructions of portions of the cell containing FHV-infected mitochondria were generated using current techniques of electron tomography [80] . Sections were cut with a thickness of ∼250 nm from blocks exhibiting well-preserved ultrastructure . These sections were stained for 30 min in 2% aqueous uranyl acetate , followed by 30 min in lead salts . Fiducial cues consisting of 20-nm colloidal gold particles were deposited on both sides of the section . For each reconstruction , a series of images was collected with a JEOL 4000EX intermediate-voltage electron microscope operated at 400 kV . The specimens were irradiated before initiating a tilt series in order to limit anisotropic specimen thinning during image collection . Pre-irradiation in this manner subjected the specimen to the steepest portion of the nonlinear shrinkage profile before images were collected . Six tilt series were collected: five single-tilt and one double-tilt . “FHV2” was the highest resolution single-tilt reconstruction and “FHV6” was the high-resolution double-tilt reconstruction; the majority of the analyses were conducted on these two reconstructions . The single-tilt series were recorded at 40 , 000 magnification with an angular increment of 2 ° from −60 ° to +60 ° about an axis perpendicular to the optical axis of the microscope using a computer-controlled goniometer to increment accurately the angular steps . These single-axis tilt series were collected using a CCD camera with pixel dimensions 1 , 960 × 2 , 560 . The pixel resolution was 0 . 55 nm . The illumination was held to near parallel beam conditions and optical density maintained constant by varying the exposure time . The IMOD package was used for generating the reconstructions [81] . Double-tilt tomography was performed by first collecting two tilt series of the same cellular region around orthogonal axes . After the first tilt series was complete using an angular increment of 2 ° from −66 ° to +66 ° , the specimen grid was rotated 90 ° , and the second tilt series was acquired from −60 ° to +66 ° . The IMOD software suite was used for fiducial mark tracking and alignment . The positions of 30 gold particles were tracked in both tilt series . After alignment , the tomographic reconstruction was generated by a projective algorithm [82] . Volume segmentation was performed by manual tracing in the planes of highest resolution with the program Xvoxtrace [83] . The mitochondrial reconstructions were visualized using Analyze ( Mayo Foundation , Rochester , MN , United States ) , ImageJ ( National Institutes of Health ) , or the surface-rendering graphics of Synu ( National Center for Microscopy and Imaging Research , San Diego , CA , United States ) as described by Perkins et al . 2001 [84] . These programs allow one to step through slices of the reconstruction in any orientation and to track or model features of interest in three dimensions . Measurements of structural features were made from planes within the reconstructed volume with the program ImageJ ( National Institutes of Health ) or within segmented volumes by the programs Synuarea and Synuvolume ( National Center for Microscopy and Imaging Research ) . Some 3-D maps , images , and videos were created using the software Amira ( Mercury TGS; http://www . tgs . com ) .	Whereas cells store and replicate their genomes as DNA , most viruses have RNA genomes that replicate by using virus-specific pathways in the host cell . The largest class of RNA viruses , the positive-strand RNA viruses , replicate their genomes on intracellular membranes . However , little is understood about how and why these viruses use membranes in RNA replication . The well-studied flock house virus ( FHV ) replicates its RNA on mitochondrial membranes . We found that the single FHV RNA replication factor and newly synthesized FHV RNA localized predominantly in numerous infection-specific membrane vesicles inside the outer mitochondrial membrane . We used electron microscope tomography to image these membranes in three dimensions and found that the interior of each vesicle was connected to the cytoplasm by a single necked channel large enough to import ribonucleotide substrates and to export product RNA . The results suggest that FHV uses these vesicles as replication compartments , which may also protect replicating RNA from competing processes and host defenses . These findings complement results from other viruses to support possible parallels between genome replication by positive-strand RNA viruses and two distinct virus classes , double-stranded RNA and reverse-transcribing viruses .	[ "Abstract", "Introduction", "Results", "Discussion", "Materials", "and", "Methods" ]	[ "viruses", "virology" ]	2007	Three-Dimensional Analysis of a Viral RNA Replication Complex Reveals a Virus-Induced Mini-Organelle
Laribacter hongkongensis is a newly discovered Gram-negative bacillus of the Neisseriaceae family associated with freshwater fish–borne gastroenteritis and traveler's diarrhea . The complete genome sequence of L . hongkongensis HLHK9 , recovered from an immunocompetent patient with severe gastroenteritis , consists of a 3 , 169-kb chromosome with G+C content of 62 . 35% . Genome analysis reveals different mechanisms potentially important for its adaptation to diverse habitats of human and freshwater fish intestines and freshwater environments . The gene contents support its phenotypic properties and suggest that amino acids and fatty acids can be used as carbon sources . The extensive variety of transporters , including multidrug efflux and heavy metal transporters as well as genes involved in chemotaxis , may enable L . hongkongensis to survive in different environmental niches . Genes encoding urease , bile salts efflux pump , adhesin , catalase , superoxide dismutase , and other putative virulence factors—such as hemolysins , RTX toxins , patatin-like proteins , phospholipase A1 , and collagenases—are present . Proteomes of L . hongkongensis HLHK9 cultured at 37°C ( human body temperature ) and 20°C ( freshwater habitat temperature ) showed differential gene expression , including two homologous copies of argB , argB-20 , and argB-37 , which encode two isoenzymes of N-acetyl-L-glutamate kinase ( NAGK ) —NAGK-20 and NAGK-37—in the arginine biosynthesis pathway . NAGK-20 showed higher expression at 20°C , whereas NAGK-37 showed higher expression at 37°C . NAGK-20 also had a lower optimal temperature for enzymatic activities and was inhibited by arginine probably as negative-feedback control . Similar duplicated copies of argB are also observed in bacteria from hot springs such as Thermus thermophilus , Deinococcus geothermalis , Deinococcus radiodurans , and Roseiflexus castenholzii , suggesting that similar mechanisms for temperature adaptation may be employed by other bacteria . Genome and proteome analysis of L . hongkongensis revealed novel mechanisms for adaptations to survival at different temperatures and habitats . Laribacter hongkongensis is a recently discovered , Gram-negative , facultative anaerobic , motile , seagull or S-shaped , asaccharolytic , urease-positive bacillus that belongs to the Neisseriaceae family of β-proteobacteria [1] . It was first isolated from the blood and thoracic empyema of an alcoholic liver cirrhosis patient in Hong Kong [2] . In a prospective study , L . hongkongensis was shown to be associated with community acquired gastroenteritis and traveler's diarrhea [3] , [4] . L . hongkongensis is likely to be globally distributed , as travel histories from patients suggested its presence in at least four continents: Asia , Europe , Africa and Central America [4]–[6] . L . hongkongensis has been found in up to 60% of the intestines of commonly consumed freshwater fish , such as grass carp and bighead carp [4] , [7] , [8] . It has also been isolated from drinking water reservoirs in Hong Kong [9] . Pulsed-field gel electrophoresis and multilocus sequence typing showed that the fish and patient isolates fell into separate clusters , suggesting that some clones could be more virulent or adapted to human [8] , [10] . These data strongly suggest that this bacterium is a potential diarrheal pathogen that warrants further investigations . Compared to other families such as Enterobacteriaceae , Vibrionaceae , Streptococcaceae , genomes of bacteria in the Neisseriaceae family have been relatively under-studied . Within this family , Neisseria meningitidis , Neisseria gonorrhoeae and Chromobacterium violaceum are the only species with completely sequenced genomes [11]–[13] . In view of its potential clinical importance , distinct phylogenetic position , interesting phenotypic characteristics and the availability of genetic manipulation systems [14]–[17] , we sequenced and annotated the complete genome of a strain ( HLHK9 ) of L . hongkongensis recovered from a 36-year old previously healthy Chinese patient with profuse diarrhea , vomiting and abdominal pain [4] . Proteomes of L . hongkongensis growing at 37°C ( body temperature of human ) and 20°C ( average temperature of freshwater habitat in fall and winter ) [9] were also compared . The complete genome of L . hongkongensis is a single circular chromosome of 3 , 169 , 329 bp with a G+C content of 62 . 35% ( Figure 1 ) . In terms of genome size and number of predicted coding sequences ( CDSs ) , rRNA operons and tRNA genes ( Table 1 ) , L . hongkongensis falls into a position intermediate between C . violaceum and the pathogenic Neisseria species . A similar intermediate status was also observed when the CDSs were classified into Cluster of Orthologous Groups ( COG ) functional categories , except for genes of RNA processing and modification ( COG A ) , cell cycle control , mitosis and meiosis ( COG D ) , replication , recombination and repair ( COG L ) and extracellular structures ( COG W ) , of which all four bacteria have similar number of genes ( Figure 2 ) . This is in line with the life cycles and growth requirements of the bacteria . C . violaceum is a highly versatile , facultative anaerobic , soil- and water-borne free-living bacterium and therefore requires the largest genome size and gene number . The pathogenic Neisseria species are strictly aerobic bacteria with human as the only host and therefore require the smallest genome size and gene number . L . hongkongensis is a facultative anaerobic bacterium that can survive in human , freshwater fish and 0–2% NaCl but not in marine fish or ≥3% NaCl and therefore requires an intermediate genome size and gene number . The L . hongkongensis genome lacks a complete set of enzymes for glycolysis , with orthologues of glucokinase , 6-phosphofructokinase and pyruvate kinase being absent ( Table S1 ) . This is compatible with its asaccharolytic phenotype and is consistent with other asaccharolytic bacteria , such as Campylobacter jejuni , Bordetella pertussis , Bordetella parapertussis and Bordetella bronchiseptica , in that glucokinase and 6-phosphofructokinase are also absent from their genomes [18] , [19] . On the other hand , the L . hongkongensis genome encodes the complete sets of enzymes for gluconeogenesis , the pentose phosphate pathway and the glyoxylate cycle ( Table S1 ) . Similar to C . jejuni , the L . hongkongensis genome encodes a number of extracellular proteases and amino acid transporters . These amino acids can be used as carbon source for the bacterium . The genome encodes enzymes for biosynthesis of the 21 genetically encoded amino acids and for biosynthesis and β-oxidation of saturated fatty acids ( Tables S2 and S3 ) . The L . hongkongensis genome encodes a variety of dehydrogenases ( LHK_00527–00540 , LHK_01219–01224 , LHK_02418–02421 , LHK_00801–00803 , LHK_01861 , LHK_02912–02913 and LHK_00934 ) that enable it to utilize a variety of substrates as electron donors , such as NADH , succinate , formate , proline , acyl-CoA and D-amino acids . The presence of three terminal cytochrome oxidases may allow L . hongkongensis to carry out respiration using oxygen as the electron acceptor under both aerobic conditions [type aa3 oxidase ( LHK_00169–00170 , LHK_00173 ) ] and conditions with reduced oxygen tension [type cbb3 ( LHK_00995–00996 , LHK_00998 ) and type bd ( LHK_02252–02253 ) oxidases] . The genome also encodes a number of reductases [fumarate reductase ( LHK_02340–02342 ) , nitrate reductase ( LHK_02079–02085 ) , dimethylsulfoxide ( DMSO ) reductase ( LHK_02496–02498 ) and tetrathionate reductase ( LHK_01476–01478 ) ] , which may help carry out respiration with alternative electron acceptors to oxygen ( fumarate , nitrate , DMSO and tetrathionate ) under anaerobic conditions . This is supported by the enhanced growth of L . hongkongensis under anaerobic conditions in the presence of nitrate ( data not shown ) . Further studies are required to confirm if the bacterium can utilize other potential electron acceptors . There were 441 transport-related proteins ( 13 . 6% of all CDSs ) in the L . hongkongensis genome , comprising an extensive variety of transporters , which may reflect its ability to adapt to the freshwater fish and human intestines , and freshwater environments . According to the Transporter Classification Database ( TCDB ) ( http://www . tcdb . org/ ) , all seven major categories of transporters are present in L . hongkongensis . Primary active transporters ( class 3 transporters ) were the most abundant class of transporters , accounting for 43 . 3% ( 191 CDSs ) of all annotated CDSs related to transport , among which 104 belong to the ATP-binding cassette ( ABC ) transporter superfamily and 41 were oxidoreduction-driven transporters . Electrochemical potential-driven transporters ( class 2 transporters ) were the second most abundant class of transporters , accounting for 27 . 9% ( 123 CDSs ) of all annotated CDSs related to transport , most of which ( 117 CDSs ) are various kinds of porters including major facilitator superfamily ( MFS ) ( 19 CDSs ) , resistance-nodulation-cell division ( RND ) superfamily ( 22 CDSs ) , amino acid-polyamine-organocation family ( 8 CDSs ) , dicarboxylate/amino acid∶cation symporter ( DAACS ) family ( 5 CDSs ) and monovalent cation∶proton antiporter-2 family ( 3 CDSs ) , and various heavy metal transporters which may be involved in detoxification and resistance against environmental hazards . Three different types of class 2 transporters , belonging to the DAACS , tripartite ATP-independent periplasmic transporter and C4-dicarboxylate uptake C family , are likely involved in the transport of malate , which can be used as the sole carbon source for L . hongkongensis in minimal medium [unpublished data] . The remaining class 2 transporters were ion-gradient-driven energizers belonging to the TonB family ( 6 CDSs ) . The third most abundant class of transporters was the channels and pores ( class 1 ) , with 39 CDSs including 12 α-type channels , 26 β-barrel porins . Among the 12 α-type channels , four were mechanosensitive channels which are important for mediating resistance to mechanophysical changes . The remaining transporters belong to four other classes , namely group translocators ( class 4 , 9 CDSs ) , transport electron carriers ( class 5 , 16 CDSs ) , accessory factors involved in transport ( class 8 , 9 CDSs ) and incompletely characterized transport system ( class 9 , 54 CDSs ) . In line with their asaccharolytic nature , the genomes of L . hongkongensis and C . jejuni do not contain genes that encode a complete phosphotransferase system . The five families of multidrug efflux transporters , including MFS ( 6 CDSs ) , RND ( 8 CDSs ) , small multidrug resistance family ( 2 CDSs ) , multidrug and toxic compound extrusion family ( 2 CDSs ) and ABC transporter superfamily ( 5 CDSs ) , were all present in L . hongkongensis , which may reflect its ability to withstand toxic substances in different habitats [20] . 20 CDSs were related to iron metabolism , including hemin transporters , ABC transporters of the metal type and ferrous iron , iron-storage proteins and the Fur protein responsible for iron uptake regulation . In contrast to C . violaceum which produces siderophores for iron acquisition , but similar to the pathogenic Neisseria species , proteins related to siderophore formation are not found in L . hongkongensis genome . In addition to a TonB-dependent siderophore receptor ( LHK_00497 ) , a set of genes ( LHK_01190 , LHK_01193 , LHK_01427–1428 ) related to the transport of hemin were present , suggesting that L . hongkongensis is able to utilize exogenous siderophores or host proteins for iron acquisition , which may be important for survival in different environments and hosts . Except the first strain of L . hongkongensis isolated from the blood and empyema pus of a patient which represented a non-motile variant , all L . hongkongensis strains , whether from human diarrheal stool , fish intestine or environmental water , are motile with polar flagella . The ability to sense and respond to environmental signals is important for survival in changing ecological niches . A total of 47 CDSs are related to chemotaxis , of which 27 encode methyl-accepting chemotaxis proteins ( MCPs ) and 20 encode chemosensory transducer proteins . While most MCPs are scattered throughout the genome , the transducer proteins are mostly arranged in three gene clusters ( Figure S1 ) . At least 38 genes , in six gene clusters , are involved in the biosynthesis of flagella ( Figure S2 ) . Enteric bacteria use several quorum-sensing mechanisms , including the LuxR-I , LuxS/AI-2 , and AI-3/epinephrine/norepinephrine systems , to recognize the host environment and communicate across species . Unlike the genomes of C . violaceum and the pathogenic Neisseria species which encode genes involved in LuxR-I and LuxS/AI-2 systems respectively , the L . hongkongensis genome does not encode genes of these 2 systems . Instead , the AI-3/epinephrine/norepinephrine system , which is involved in inter-kingdom cross-signaling and regulation of virulence gene transcription and motility , best characterized in enterohemorrhagic E . coli [21] , [22] , is likely the predominant quorum-sensing mechanism used by L . hongkongensis . Several human enteric commensals or pathogens , including E . coli , Shigella , and Salmonella , produce AI-3 [23] . A two-component system , QseB/C , of which QseC is the sensor kinase and QseB the response regulator , has been found to be involved in sensing AI-3 from bacteria and epinephrine/norepinephrine from host , and activation of the flagellar regulon transcription [21] . While the biosynthetic pathway of AI-3 has not been discovered , two sets of genes , LHK_00329/LHK_00328 and LHK_01812/LHK_01813 , homologous to QseB/QseC were identified in the L . hongkongensis genome , suggesting that the bacterium may regulate its motility upon recognition of its host environment . The presence of two sets of QseB/QseC , one most similar to those of C . violaceum and the other most homologous to Azoarcus sp . strain BH72 , is intriguing , as the latter is the only bacterium , with complete genome sequence available , that possesses two copies of such genes . Before reaching the human intestine , L . hongkongensis has to pass through the highly acidic environment of the stomach . In the L . hongkongensis genome , a cluster of genes , spanning a 12-kb region , related to acid resistance , is present . Similar to Helicobacter pylori , the L . hongkongensis genome contains a complete urease gene cluster ( LHK_01035–LHK_01037 , LHK_01040–LHK_01044 ) , in line with the bacterium's urease activity . Phylogenetically , all 8 genes in the urease cassette are most closely related to the corresponding homologues in Brucella species ( α-proteobacteria ) , Yersinia species ( γ-proteobacteria ) and Photorhabdus luminescens ( γ-proteobacteria ) , instead of those in other members of β-proteobacteria , indicating that L . hongkongensis has probably acquired the genes through horizontal gene transfer after its evolution into a distinct species ( Figure S3 ) . Upstream and downstream to the urease cassette , adi ( LHK_01034 ) and hdeA ( LHK_01046 ) were found respectively . Their activities will raise the cytoplasmic pH and prevents proteins in the periplasmic space from aggregation during acid shock respectively [24] , [25] . In addition to the acid resistance gene cluster , the L . hongkongensis genome contains two arc gene clusters [arcA ( LHK_02729 and LHK_02734 ) , arcB ( LHK_02728 and LHK_02733 ) , arcC ( LHK_02727 and LHK_02732 ) and arcD ( LHK_02730 and LHK_02731 ) ] of the arginine deiminase pathway which converts L-arginine to carbon dioxide , ATP , and ammonia . The production of ammonia increases the pH of the local environment [26] , [27] . Similar to other pathogenic bacteria of the gastrointestinal tract , the genome of L . hongkongensis encodes genes for bile resistance . These include three complete copies of acrAB ( LHK_01425–01426 , LHK_02129–02130 and LHK_02929–02930 ) , encoding the best studied efflux pump for bile salts , and two pairs of genes ( LHK_01373–01374 and LHK_03132–03133 ) that encode putative efflux pumps homologous to that encoded by emrAB in E . coli [28] . Furthermore , five genes [tolQ ( LHK_00053 ) , tolR ( LHK_03174 ) , tolA ( LHK_03173 ) , tolB ( LHK_03172 ) and pal ( LHK_03171 ) ] that encode the Tol proteins , important in maintaining the integrity of the outer membrane and for bile resistance , are also present [29] . In the L . hongkongensis genome , a putative adhesin ( LHK_01901 ) for colonization of the intestinal mucosa , most closely related to the adhesins of diffusely adherent E . coli ( DAEC ) and enterotoxigenic E . coli ( ETEC ) , encoded by aidA and tibA respectively , was observed ( Figure S4 ) [30] , [31] . aidA and tibA encode proteins of the autotransporter family , type V protein secretion system of Gram-negative bacteria . All the three domains ( an N-terminal signal sequence , a passenger domain and a translocation domain ) present in proteins of this family are found in the putative adhesin in L . hongkongensis . Moreover , a putative heptosyltransferase ( LHK_01902 ) , with 52% amino acid identity to the TibC heptosyltransferase of ETEC , responsible for addition of heptose to the passenger domain , was present upstream to the putative adhesin gene in the L . hongkongensis genome ( Figure S4 ) . In addition to host cell adhesion , the passenger domains of autotransporters may also confer various virulence functions , including autoaggregation , invasion , biofilm formation and cytotoxicity . The L . hongkongensis genome encodes a putative superoxide dismutase ( LHK_01716 ) and catalases ( LHK_01264 , LHK_01300 and LHK_02436 ) , which may play a role in resistance to superoxide radicals and hydrogen peroxide generated by neutrophils . The same set of genes that encode enzymes for synthesis of lipid A ( endotoxin ) , the two Kdo units and the heptose units of lipopolysaccharide ( LPS ) are present in the genomes of L . hongkongensis , C . violaceum , N . meningitidis , N . gonorrhoeae and E . coli . Moreover , 9 genes [rfbA ( LHK_02995 ) , rfbB ( LHK_02997 ) , rfbC ( LHK_02994 ) , rfbD ( LHK_02996 ) , wbmF ( LHK_02799 ) , wbmG ( LHK_02800 ) , wbmH ( LHK_02801 ) , wbmI ( LHK_02790 ) and wbmK ( LHK_02792 ) ] that encode putative enzymes for biosynthesis of the polysaccharide side chains are present in the L . hongkongensis genome . In addition to genes for synthesizing LPS , a number of CDSs that encode putative cytotoxins are present , including cytotoxins that act on the cell surface [hemolysins ( LHK_00956 and LHK_03166 ) and RTX toxins ( LHK_02735 and LHK_02918 ) ] and those that act intracellularly [patatin-like proteins ( LHK_00116 , LHK_01938 , and LHK_03113 ) ] [32] , [33] . Furthermore , a number of CDSs that encode putative outer membrane phospholipase A1 ( LHK_00790 ) and collagenases ( LHK_00305–00306 , LHK_00451 , and LHK_02651 ) for possible bacterial invasion are present . To better understand how L . hongkongensis adapts to human body and freshwater habitat temperatures at the molecular level , the types and quantities of proteins expressed in L . hongkongensis HLHK9 cultured at 37°C and 20°C were compared . Since initial 2D gel electrophoresis analysis of L . hongkongensis HLHK9 proteins under a broad range of pI and molecular weight conditions revealed that the majority of the proteins reside on the weakly acidic to neutral portion , with a minority on the weak basic portion , consistent with the median pI value of 6 . 63 calculated for all putative proteins in the genome of L . hongkongensis HLHK9 , we therefore focused on IPG strips of pH 4–7 and 7–10 . Comparison of the 2D gel electrophoresis patterns from L . hongkongensis HLHK9 cells grown at 20°C and 37°C revealed 12 differentially expressed protein spots , with 7 being more highly expressed at 20°C than at 37°C and 5 being more highly expressed at 37°C than at 20°C ( Table 2 , Figure 3 ) . The identified proteins were involved in various functions ( Table 2 ) . Of note , spot 8 [N-acetyl-L-glutamate kinase ( NAGK ) -37 , encoded by argB-37] was up-regulated at 37°C , whereas spot 1 ( NAGK-20 , encoded by argB-20 ) , was up-regulated at 20°C ( Figures 3 , 4A and 4B ) . These two homologous copies of argB encode two isoenzymes of NAGK [NAGK-20 ( LHK_02829 ) and NAGK-37 ( LHK_02337 ) ] , which catalyze the second step of the arginine biosynthesis pathway . The transcription levels of argB-20 and argB-37 at 20°C and 37°C were quantified by real time RT-PCR . Results showed that the mRNA level of argB-20 at 20°C was significantly higher that at 37°C and the mRNA level of argB-37 at 37°C was significantly higher that at 20°C ( Figure 4C and 4D ) , suggesting that their expressions , similar to most other bacterial genes , were controlled at the transcription level . When argB-20 and argB-37 were cloned , expressed and the corresponding proteins NAGK-20 and NAGK-37 purified for enzyme assays , their highest enzymatic activities were observed at 37–45°C and 45–50°C respectively ( Figure 4E ) . Moreover , NAGK-20 , but not NAGK-37 , was inhibited by 0 . 25–10 mM of arginine ( Figure 4F ) . L . hongkongensis probably regulates arginine biosynthesis at temperatures of different habitats using two pathways with two isoenzymes of NAGK . L . hongkongensis and wild type E . coli ATCC 25922 , but not E . coli JW5553-1 ( argB deletion mutant ) , grew in minimal medium without arginine , indicating that L . hongkongensis contains a functional arginine biosynthesis pathway . NAGK-20 is expressed at higher level at 20°C than 37°C , whereas NAGK-37 is expressed at higher level at 37°C than 20°C . Bacteria use either of two different pathways , linear and cyclic , for arginine biosynthesis . Similar to NAGK-20 of L . hongkongensis , NAGK of Pseudomonas aeruginosa and Thermotoga maritima , which employ the cyclic pathway , can be inhibited by arginine as the rate-limiting enzyme for negative feedback control [34]–[37] . On the other hand , similar to NAGK-37 of L . hongkongensis , NAGK of E . coli , which employs the linear pathway , is not inhibited by arginine [35] , [36] . We speculate that L . hongkongensis can use different pathways with the two NAGK isoenzymes with differential importance at different temperatures of different habitats . Phylogenetic analysis of NAGK-20 and NAGK-37 showed that they were more closely related to each other than to homologues in other bacteria ( Figure 5 ) . The topology of the phylogenetic tree constructed using NAGK was similar to that constructed using 16S rRNA gene sequences ( data not shown ) . This suggested that the evolution of argB genes in general paralleled the evolution of the corresponding bacteria , and argB gene duplication has probably occurred after the evolution of L . hongkongensis into a separate species . The requirement to adapt to different temperatures and habitats may have provided the driving force for subsequent evolution to 2 homologous proteins that serve in different environments . Notably , among all 465 bacterial species with complete genome sequences available , only Thermus thermophilus , Deinococcus geothermalis , Deinococcus radiodurans , Roseiflexus castenholzii and Roseiflexus sp . RS-1 possessed two copies of argB , whereas Anaeromyxobacter sp . Fw109-5 and Anaeromyxobacter dehalogenans 2CP-C possessed one copy of argB and another fused with argJ ( Figure 5 ) . The clustering of argB in two separate groups in these bacteria suggests that argB gene duplication has probably occurred in their ancestor , before the divergence into separate species . The prevalence of T . thermophilus , Deinococcus species and Roseiflexus species in hot springs suggested that this novel mechanism of temperature adaptation may also be important for survival at different temperatures in other bacteria . Further experiments on differential expression of the two isoenzymes at different temperatures in these bacteria will verify our speculations . Traditionally , complete genomes of bacteria with medical , biological , phylogenetic or industrial interests were sequenced only after profound phenotypic and genotypic characterization of the bacteria had been performed . With the advance in technology and bioinformatics tools , complete genome sequences of bacteria can be obtained with greater ease . In this study , we sequenced and analyzed the complete genome of L . hongkongensis , a newly discovered bacterium of emerging medical and phylogenetic interest , and performed differential proteomics and downstream characterization of important pathways . In addition , putative virulence factors and a putative novel mechanism of arginine biosynthesis regulation at different temperatures were discovered , further characterization of which will lead to better understanding of their contributions to the survival and virulence of L . hongkongensis , the Neisseriaceae family and other bacteria . A similar “reverse genomics” approach can be used for the study of other newly discovered important bacteria . The genome sequence of L . hongkongensis HLHK9 was determined with the whole-genome shotgun method . Three shotgun libraries were generated: one small-insert ( 2–4 kb ) library and one medium-insert ( 5–6 kb ) library in pcDNA2 . 1 , and a large-insert ( 35–45 kb ) fosmid library in pCC2FOS . DNA sequencing was performed using dye-terminator chemistries on ABI3700 sequencers . Shotgun sequences were assembled with Phrap . Fosmid end sequences were mapped onto the assembly using BACCardI [38] for validation and support of gap closing . Sequences of all large repeat elements ( rRNA operons and prophages ) were confirmed by primer walking of fosmid clones . The nucleotide sequence for the complete genome sequence of L . hongkongensis HLHK9 was submitted to Genbank under accession number CP001154 . Gene prediction was performed by Glimmer [39] version 3 . 02 , and results post-processed using TICO [40] for improving predictions of translation initiation sites . Automated annotation of the finished sequence was performed by a modified version of AutoFACT [41] , supplemented by analysis by InterProScan [42] . Manual curation of annotation results was done with support from the software tool GenDB [43] . In addition , annotation of membrane transport proteins was done by performing BLAST search of all predicted genes against the curated TCDB [44] . Ribosomal RNA genes were annotated using the online RNAmmer service [45] . Putative prophage sequences were identified using Prophage Finder [46] . Frameshift errors were predicted using ProFED [47] . CRISPRs ( Clustered Regularly Interspaced Short Palindromic Repeats ) were searched by using PILER-CR [48] , CRISPRFinder [49] and CRT ( CRISPR recognition tool ) [50] . Single colony of L . hongkongensis HLHK9 was inoculated into brain heart infusion ( BHI ) medium for 16 h . The bacterial cultures were diluted 1∶100 in BHI medium and growth was continued at 20°C for 20 h and 37°C for 6 h , respectively , with shaking to OD600 of 0 . 6 . After centrifugation at 6 , 500×g for 15 min , cells were lysed in a sample buffer containing 7 M urea , 2 M thiourea and 4% CHAPS . The crude cell homogenate was sonicated and centrifuged at 16 , 000×g for 20 min . Immobilized pH gradient ( IPG ) strips ( Bio-Rad Laboratories ) ( 17 cm ) with pH 4–7 and 7–10 were hydrated overnight in rehydration buffer containing 7 M urea , 2 M thiourea , 4% CHAPS , 1% IPG buffer pH 4–7 ( IPG strip of pH 4–7 ) and pH 6–11 ( IPG strip of pH 7–10 ) ( GE Healthcare ) and 60 mM DTT with 60 µg of total protein . The first dimension , isoelectric focusing ( IEF ) , was carried out in a Protean IEF cell electrophoresis unit ( Bio-Rad Laboratories ) for about 100 , 000 volt-hours . Protein separation in the second dimension was performed in 12% SDS-PAGE utilizing the Bio-Rad Protean II xi unit ( Bio-Rad Laboratories ) . 2D gels were stained with silver and colloidal Coomassie blue G-250 respectively for qualitative and quantitative analysis , and scanned with ImageScanner ( GE Healthcare ) . ImageMaster 2D Platinum 6 . 0 ( GE Healthcare ) was used for image analysis . For MALDI-TOF MS analysis , protein spots were manually excised from gels and subjected to in-situ digestion with trypsin , and peptides generated were analyzed using a 4800 Plus MALDI TOF/TOF Analyzer ( Applied Biosystems ) . Proteins were identified by peptide mass fingerprinting using the MS-Fit software ( http://prospector . ucsf . edu ) and an in-house sequence database of L . hongkongensis HLHK9 proteins generated using the information obtained from the complete genome sequence and annotation . Only spots with at least two-fold difference in their spot volume between 20°C and 37°C and those uniquely detected at either temperature were subjected to protein identification by MALDI-TOF MS analysis . Three independent experiments for each growth condition were performed . L . hongkongensis HLHK9 cells were grown in minimal medium M63 [51] supplemented with 20 mM L-malate as carbon source and 19 mM potassium nitrate as nitrogen source , and 1 mM each of vitamin B1 and vitamin B12 . The pH of all media was adjusted to 7 . 0 with KOH . Essentiality of arginine for growth of L . hongkongensis HLHK9 was determined by transferring the bacterial cells to the modified M63 medium with or without 100 mM of L-arginine . Escherichia coli ATCC 25922 and JW5553-1 ( argB deletion mutant ) [52] were used as positive and negative controls respectively . All cultures were incubated at 37°C with shaking for 5 days . Growth in each medium was determined by measuring absorbance spectrophotometrically at OD600 . The experiment was performed in duplicate . mRNA levels of argB-20 and argB-37 in L . hongkongensis HLHK9 cells grown in 20°C and 37°C were compared . Total RNA was extracted from culture of L . hongkongensis HLHK9 ( OD600 of 0 . 6 ) grown in conditions described in proteomic analysis by using RNeasy kit ( Qiagen ) in combination with RNAprotect Bacteria Reagent ( Qiagen ) as described by the manufacturer . Genomic DNA was removed by DNase digestion using RNase-free DNase I ( Roche ) . The total nucleic acid concentration and purity were estimated using A260/A280 values measured by NanoDrop ND-1000 spectrophotometer ( NanoDrop Technologies ) . Bacteria were harvested from three independent replicate cultures . cDNA was synthesized by RT using random hexamers and SuperScript III kit ( Invitrogen ) as described previously [53] , [54] . cDNA was amplified by TaqMan PCR Core Reagent kit ( Applied Biosystems ) in an ABI Prism 7000 Sequence Detection System ( Applied Biosystems ) . Briefly , 2 µl of cDNA was amplified in a 25 µl reaction containing 2 . 5 µl of 10× TaqMan buffer A , 5 . 5 mM of MgCl2 , 0 . 2 mM of each deoxynucleoside triphosphates ( dNTPs ) , 0 . 8 µM of each primer , 0 . 8 µM of gene-specific TaqMan probe with a 5′-[6-carboxyfluorescein ( 6-FAM ) ] reporter dye and a 3′-[6-carboxytetramethylrhodamine ( TAMRA ) ] quencher dye , 2 . 5 U of AmpErase Uracil N-glycosylase ( UNG ) and 0 . 625 U AmpliTaq Gold polymerase ( Applied Biosystems ) . Primers and TaqMan probes were designed using Primer Express software , version 2 . 0 ( Applied Biosystems ) ( Table S4 ) . Reactions were first incubated at 50°C for 2 min , followed by 95°C for 10 min in duplicate wells . Reactions were then thermal-cycled in 40 cycles of 95°C for 15 s and 60°C for 1 min . Absolute standard curve method was used for determination of transcript level for each gene . Standard curves were made by using serial dilutions from plasmids containing the target sequences with known quantities . Housekeeping gene RNA polymerase beta subunit , rpoB , was used as an internal control . Triplicate assays using RNAs extracted in three independent experiments confirmed that transcript levels of rpoB were not significantly different ( P>0 . 05 ) at 20°C compared with 37°C ( data not shown ) . The transcript levels of argB-20 and argB-37 were then normalized to that of rpoB . Triplicate assays using RNAs extracted in three independent experiments were performed for each target gene . The phylogenetic relationships among NAGK-20 and NAGK-37 of L . hongkongensis HLHK9 and their homologues in other bacteria with complete genomes available were analyzed . Phylogenetic tree was constructed by the neighbor-joining method using Kimura's two-parameter correction with ClustalX 1 . 83 . Three hundred and eleven positions were included in the analysis . Cloning and purification of ( His ) 6-tagged recombinant NAGK proteins of L . hongkongensis HLHK9 was performed according to our previous publications , with modifications [53] , [55] . To produce plasmids for protein purification , primers ( 5′- GGAATTCCATATGCTGCTTGCAGACGCCC -3′ and 5′- GGAATTCCATATGTCAGGCTGCGCGGATCAT -3′ for argB-20 and 5′- GGAATTCCATATGGTTATTCAATCTGAAGT -3′ and 5′- GGAATTCCATATGTCAGAGCGTGGTACAGAT -3′ for argB-37 ) were used to amplify the genes encoding NAGK-20 and NAGK-37 , respectively , by PCR . The sequence coding for amino acid residues of the complete NAGK-20 and NAGK-37 was amplified and cloned , respectively , into the NdeI site of expression vector pET-28b ( + ) ( Novagen ) in frame and downstream of the series of six histidine residues . The two recombinant NAGK proteins were expressed and purified using the Ni2+-loaded HiTrap Chelating System according to the manufacturer's instructions ( GE Healthcare ) . Purified NAGK-20 and NAGK-37 were assayed for N-acetyl-L-glutamate kinase activity using Haas and Leisinger's method [56] , with modifications . The reaction mixtures contained 400 mM NH2OH⋅HCl , 400 mM Tris⋅HCl , 40 mM N-acetyl-L-glutamate , 20 mM MgCl2 , 10 mM ATP and 2 µg of enzyme in a final volume of 1 . 0 ml at pH 7 . 0 . After incubation at 25°C , 30°C , 37°C , 45°C , 50°C , 55°C or 60°C for 30 min , the reaction was terminated by adding 1 . 0 ml of a stop solution containing 5% ( w/v ) FeCl3⋅6H2O , 8% ( w/v ) trichloroacetic acid and 0 . 3 M HCl . The absorbance of the hydroxamate⋅Fe3+ complex was measured with a spectrophotometer at A540 [57] . Inhibition of the kinase activities of NAGK-20 and NAGK-37 were examined with and without 0 . 25 , 0 . 5 , 0 . 75 , 1 , 2 . 5 , 5 , 10 , and 20 mM of L-arginine and incubated at 37°C for 30 min . One unit of N-acetyl-L-glutamate kinase is defined as the amount of enzyme required to catalyze the formation of 1 µmol of product per min under the assay conditions used . Each assay was performed in duplicate . Results were presented as means and standard deviations of three independent experiments .	Laribacter hongkongensis is a recently discovered bacterium associated with gastroenteritis and traveler's diarrhea . Freshwater fish is the reservoir of L . hongkongensis . In order to achieve a rapid understanding on the mechanisms by which the bacterium adapts to different habitats and its potential virulence factors , we sequenced the complete genome of L . hongkongensis , compared its gene contents with other bacteria , and compared its gene expression at 37°C ( human body temperature ) and 20°C ( freshwater habitat temperature ) . We found that the gene contents of L . hongkongensis enable it to adapt to its diverse habitats of human and freshwater fish intestines and freshwater environments . Genes encoding proteins responsible for survival in the intestinal environments , adhesion to intestinal cells , evasion from host immune systems , and putative virulence factors similar to those observed in other pathogens are present . We also observed , in gene expression studies , that L . hongkongensis may be using different pathways for arginine synthesis regulated at different temperatures . Phylogenetic analysis suggested that such mechanisms for temperature adaptation may also be used in bacteria found in extreme temperatures .	[ "Abstract", "Introduction", "Results/Discussion", "Materials", "and", "Methods" ]	[ "genetics", "and", "genomics/genome", "projects", "microbiology" ]	2009	The Complete Genome and Proteome of Laribacter hongkongensis Reveal Potential Mechanisms for Adaptations to Different Temperatures and Habitats
Patterns of social mixing are key determinants of epidemic spread . Here we present the results of an internet-based social contact survey completed by a cohort of participants over 9 , 000 times between July 2009 and March 2010 , during the 2009 H1N1v influenza epidemic . We quantify the changes in social contact patterns over time , finding that school children make 40% fewer contacts during holiday periods than during term time . We use these dynamically varying contact patterns to parameterise an age-structured model of influenza spread , capturing well the observed patterns of incidence; the changing contact patterns resulted in a fall of approximately 35% in the reproduction number of influenza during the holidays . This work illustrates the importance of including changing mixing patterns in epidemic models . We conclude that changes in contact patterns explain changes in disease incidence , and that the timing of school terms drove the 2009 H1N1v epidemic in the UK . Changes in social mixing patterns can be usefully measured through simple internet-based surveys . Seasonal changes in patterns of social contacts have a marked influence on the spread of infectious diseases . In particular , the patterns of school terms and holidays affect the incidence of infections with a significant impact on school-age children , including measles , pertussis , and influenza [1]–[6] . Mathematical models can be used to explain and attempt to predict the spread of infectious diseases; however until recently a lack of data about social contact patterns has restricted the applicability of these models . In 2008 results were published from the POLYMOD study , a social contact survey involving participants in 8 European countries [7]; this study described patterns of social mixing , quantifying the tendency of people to mix with others of a similar age , and showing that the highest levels of contact were between children . These data have been used to model close-contact infectious diseases , and have been found useful in explaining observed patterns of incidence [2] , [5] , [7]–[10] . Important factors are still missing from available datasets . One such factor is good information about how social contact behaviour varies over time . On an individual level , there is day-to-day variation in social behaviour [11] , and incidence data suggest that there are population-level changes resulting from events such as school holidays [1]–[4] , [6] . As part of the POLYMOD study , some data were collected during the school holidays , demonstrating significant changes in contact patterns during holiday periods [12] . Studies focusing on school-age children have confirmed that children make substantially fewer contacts on average during the holidays and at weekends than when at school [13]–[16] . However , there is a general lack of information about temporal changes in contact patterns , in particular quantifying the impact of school holidays on contact behaviour within the population as a whole . In the absence of these data , mathematical models of disease spread have been obliged to make a range of plausible assumptions about how to model the impact of school holidays [2]–[4] , [6] , [17]–[19] . Here , we present the results of a longitudinal population-level social mixing survey and use these data to parameterise an age-structured model of H1N1v incidence . In April 2009 , H1N1v influenza emerged in the Americas . Over the next few months , this virus spread around the globe , causing millions of cases worldwide . The UK experienced two distinct peaks in incidence , one in July 2009 , and another in October 2009 [1] , [2] . Serological data collected during the epidemic suggest that , in some parts of the UK , over 40% of children aged 5–14 were infected before the end of the first wave of infection [20] , with an estimated cumulative incidence over the second epidemic wave in this group of 59% [21]; these serological data suggest that the great majority of cases were not captured in incidence estimates derived from clinical surveillance [1] , even though such estimates may give a good indication of incidence trends . The UK flusurvey ( www . flusurvey . org . uk ) was developed as an internet-based tool to augment existing influenza surveillance [22] , [23] , most of which depends on recording healthcare usage by symptomatic individuals [1] , [24] , [25] and so misses individuals with influenza-like-illness ( ILI ) who do not seek medical attention . The UK flusurvey is an attempt to record ILI incidence that does not depend on ill individuals seeking healthcare [23] . As well as estimating incidence trends [26] , flusurvey data have been used to estimate the effectiveness of influenza vaccination [27] . Flusurvey participants were also asked about their social contact behaviour . Here we describe the results of the social contact survey carried out during the H1N1v epidemic in the UK . We describe changes in contact patterns that took place during the pandemic; in particular , we quantify the impact of school holiday periods . In order to test whether this internet-based social contact survey captures epidemiologically-relevant patterns of social interactions , we use the measured dynamic contact patterns in a simple mathematical model of influenza spread , and explore their ability to explain the observed patterns of incidence . We find that a relatively simple model , parameterised by our age-structured mixing data , gives a good match with observed patterns of incidence . The contact survey was completed 9 , 261 times by 3 , 338 individuals , many completing it multiple times . 104 surveys were excluded from further analysis because of missing age information; the analysis that follows is based on the remaining 9 , 157 reports . The data can be found in the Supporting Information , Dataset S1 . As expected , the majority of reports were completed by adults during the school term time ( Table S1 in Text S1 ) . We have therefore not further subdivided the school-aged groups or to attempted to distinguish between different holiday periods ( e . g . summer holiday and autumn half term holiday ) . Fig . 1 shows the impact of school holidays on the social mixing patterns of the population . Both for conversational and for physical contacts the most obvious change was in the number of interactions between school-aged children . School holidays had a much smaller effect on the number of contacts made by or with other age groups . There was a large , highly significant , reduction during the school holidays in the daily average number of conversational contacts made by those aged 5–18 ( from 41 . 2 during term time to 24 . 8 during the holidays , p = 0 . 001; Table 1 ) . Older age groups reported a small , but statistically significant , change . There were fewer physical contacts than conversational contacts reported , and the reduction in the number of physical contacts reported by school children during school holidays ( from 11 . 0 to 8 . 9 ) was not statistically significant . Models parameterised using these measured mixing patterns were fitted to estimated incidence curves ( Fig . 2 ) . While models parameterised using both conversational and physical contact patterns broadly capture observed incidence , the patterns of conversational contacts appear to provide a better fit to incidence data than patterns of physical contacts . In particular , models parameterised using physical contact patterns cannot capture the timing and depth of the trough in incidence at the end of the summer holidays . The model fits are similar whether using Health Protection Agency ( HPA ) or flusurvey-adjusted incidence estimates . An outbreak would have grown more slowly had it begun during the school holidays than during term time . During the holidays , in the absence of prior immunity , the initial growth rate of the epidemic , R , would have been approximately 35% lower than during term time ( 25% lower in the model using patterns of physical contact ) – falling from 1 . 57 to 1 . 07 . Prior immunity reduced initial growth rate by approximately 10% , to 1 . 42 in term time and 0 . 91 in the holidays ( Fig . 3 ) . Estimated parameter values are reasonably consistent across the models used ( Table S2 in Text S1 ) , aside from the transmission rate per encounter , τ , which , as expected , is larger in the models using physical contact patterns . The value of the rescaling factor is estimated to be between 9 and 15 . The models suggest that around 30% of adults and over 50% of school-aged children had acquired immunity by the end of the outbreak ( Table S3 in Text S1 ) . Both models and incidence estimates indicate that incidence during school term time was dominated by those aged under 18 , whereas during holidays the majority of cases were in adults ( Fig . 4 ) [1] . The good agreement between the models and the data supports the usefulness of the mixing data obtained . Mixing matrices generated from bootstrapping the original dataset suggest that the substantial change in contacts between term time and holidays is necessary for the model to be able to fit the incidence data , with low-difference bootstrap matrices resulting in models that fit the observed data less well ( Figs S1 , S2 ) . Substantial and significant changes in social contact patterns take place during school holidays . The greatest change is seen in school-aged children , who make approximately 40% fewer conversational contacts ( 95% CI 22–59% ) each day during the school holidays than during term time . These changes in social contact patterns have a large impact on the spread of infections . As the incidence patterns of the 2009 H1N1v epidemic in the UK show , incidence began to fall at the start of the holiday period and began to rise again when schools reopened . Models incorporating these dynamic contact patterns capture the observed dynamics of influenza , suggesting that the social contact patterns reported here are closely correlated to those relevant to the spread of influenza . The large fall in contacts during school holidays generates the observed decline in cases seen during the summer of 2009 . The models highlight the impact of prior immunity on epidemic behaviour , and suggest that , had the first cases arrived in the population during the school holidays , existing immunity in the population would have been sufficient to prevent the epidemic from taking off until schools reopened . This work supports previous studies that suggest that school holidays are associated with significant changes in mixing patterns and in epidemic behaviour . The impact of holidays appears larger than some studies suggest [12] , though not as large as others [13]–[16] . Different survey tools are likely to give different results: in contrast to surveys that use a detailed contact diary-based approach [7] , [12]–[14] , [28] , [29] the method used here did not require participants to give additional details about each of the people they met , and thus there was no time-saving incentive towards recording fewer contacts; on the other hand , listing one by one all encounters may provide an aid to recall . Several different methods of collecting social contact data have been used in other studies , including self-completed paper contact diaries [7] , [11] , [28] , [29] , network studies [8] , [30] , [31] , electronic contact diaries [32] , online contact diaries [29] and automated electronic proximity sensors [33] . All have been found to be useful , and none to be perfect . Perfect recall of all encounters is unlikely , especially for short-duration encounters [30] . Some studies have found electronic self-reported contact data to perform similarly to paper diaries [29] , while others have found that more encounters are reported when using paper diaries [32] . In our study , we collected aggregate numbers of contacts ( by age group and social setting ) , in order to reduce the time required to complete the surveys; a previous study suggests that this approach gives similar results to contact diaries if , as in our case , the recall period was short [28] . In common with other contact surveys [7] , [11] , data about the contact patterns of young children could be reported on their behalf by their parents , which may limit its reliability . Collecting contact data from young children is challenging though not impossible [13] , [15] , [31] , and although our survey was designed to be straightforward to complete it was not possible to devise something that would be equally suitable for all age groups . School closure has been suggested as an intervention to control infection , an idea that models have helped to explore [17]–[19] . Although this work demonstrates that scheduled school holidays have a large impact on transmission , school closure as a public health intervention may not have the same effect on social mixing patterns , since child care arrangements during unplanned , short-notice , closures may differ from those during school holidays . Unsurprisingly , there is only limited information available on this subject [15] , [16] , [34] . Furthermore , as was seen in the UK , it is likely that the epidemic would take off again once schools re-opened; thus school closure is more likely to be useful as a way to delay transmission than to prevent it altogether . The models developed here suggest that a large fraction of the UK population was infected during the 2009 H1N1v epidemic . The same conclusion resulted from serological sampling that reported seropositivity by the end of the first wave of over 45% in children aged 5–14 in the regions of the UK first affected by the epidemic [20] , and a cumulative incidence of 59% in this group over the second wave [21] . Interpretation of serological data is difficult , since not all those infected are expected to have seroconverted by the time of the sampling and blood samples used in these studies are not sampled at random [20] , [21] . However , the models presented here , available serological data [20] , [21] , and other modelling work [2] all suggest that the original incidence figures dramatically underestimated the true number of infections . The models suggest that estimated influenza incidence only includes around 7–11% of all people infected . A number of factors may account for this , including mild or asymptomatic infections that would not have been diagnosed as ILI , imperfect test sensitivity , or poor estimates of the fraction of individuals with ILI who seek medical attention . Ideally , there would be perfect incidence data to which to fit epidemic models . However , incidence estimates are not perfect , and serological surveys cannot give fine-scaled information about weekly incidence patterns . Here , we have used models appropriate to the level of incidence and behavioural data available and fitted models to incidence estimated in two different ways , in both cases drawing similar conclusions . The social contact data used here are , likewise , imperfect . Participants in the flusurvey are not a random sample of the UK population , and we are unable to control for all biases in this self-selecting sample [23] . It would be interesting to be able to look at variations in contact patterns at a finer temporal resolution , such as comparing different holiday periods or detecting other temporal variations , but in this case the sample size , particularly of school age children , is not large enough to make this feasible . We cannot reasonably justify splitting up the most interesting and important groups – school-aged children – any further into , for example , primary and secondary school groups . It is planned to continue the UK flusurvey in future years , and it is hoped that wider recruitment will allow these issues to be explored more fully in due course . We found that patterns of conversational encounters provided a better fit to incidence data than patterns of physical encounters . Some other studies have found that models using patterns of physical encounters provide a better fit to serological profiles [8] , [9] though other studies do not find a difference between using physical and conversational encounter patterns [10] . Of course , fitting models to serological profiles that are the result of many years of potential exposure is not the same as fitting to short term incidence data . We found that the relatively small school-holiday change in numbers of physical encounters was unable to explain the sharp decline in incidence associated with the summer holiday period , an effect that may be less important when considering cumulative exposure over many years . Or it may simply be the case that conversational encounters provide a better proxy for interactions that led to the transmission of H1N1v than physical encounters . The mathematical model of influenza transmission used here is extremely simple , with a population categorised into broad age groups roughly corresponding to normal patterns of work and school attendance . The model ignores geographical differences in transmission and incidence across the UK . The novel aspect of the model is that it makes use of measured changes in patterns of social contacts taking place between these groups as a result of the opening and closing of schools . The model is parameterised by data collected from an internet-based survey completed by a subset of the population of interest at the time of the epidemic . Despite the caveats , the survey reported here is , to our knowledge , the only large-scale longitudinal study of population-level social contacts to have been carried out . We have shown that internet-based contact surveys can be used in large-scale studies . The fact that the contact data can be used in models to capture observed incidence patterns suggests that we have succeeded in quantifying epidemiologically relevant longitudinal social contact patterns . Participation in this opt-in study was voluntary , and all analysis was carried out on anonymised data . The study was approved by the ethics committee of the London School of Hygiene and Tropical Medicine . The UK flusurvey was launched in July 2009 , based on similar systems used elsewhere in Europe [22] . It ran from July 2009 until March 2010 . Members of the public were encouraged to register via the flusurvey website and reported their symptoms ( or lack of symptoms ) each week . On registration , participants completed a background survey recording information about themselves including age , gender , and vaccination history . Participation in all parts of the flusurvey was entirely voluntary . Participants were prompted to continue to take part with a weekly email reminder . Further details about the flusurvey can be found in [23] and in Text S1 . Participants could also take part in a contact survey . This could be completed as often as participants chose; they were reminded of it each week , but its completion was not heavily advocated since the principal interest was in measuring incidence and behavioural response to infection [23] , [26] . The contact survey was a simplified version of that used in other contact studies [7] , [12]–[14] , [16]: participants were asked two main questions: “How many people did you have conversational contact with yesterday ? ” and “How many people did you have physical contact with yesterday ? ” In each case , participants were asked to report the numbers of people they met in 4 different age groups ( 0–4; 5–18; 19–64; 65+ ) , roughly corresponding to normal school and work attendance , and three different social settings ( Home , Work/School/College , Other ) . Participants were asked to approximate larger numbers of contacts using in the following categories: 16–24; 25–49; 50–99; 100 or more; while we would have liked to collect precise numbers , it was decided that this would present an unrealistic recall challenge for participants . For larger numbers of contacts , in the analyses that follow the number of contacts was approximated by midpoint of these categories aside from the category “100 or more” , which was approximated by 150 . Further details can be found in Text S1 . Participants were categorised into the same age groups as contacts ( 0–4; 5–18; 19–64; and 65+ ) ; time period was categorised as term time or school holidays . To explore the influence of school holiday periods on the number and age distribution of contacts , accounting for multiple reports from participants who completed the contact survey multiple times , we used a population averaged negative binomial regression model with robust standard errors [35] , [36] . Analyses were carried out separately for each age group of participants . Time period and gender were considered as explanatory variables , but gender was found not to be a significant factor and was subsequently omitted from the analyses . Analyses were carried out in Stata 11 . Because the weekly survey reminder email was sent to participants each Wednesday , and the contact survey asked about “yesterday's” contacts , most reports related to Tuesdays . Therefore , although a small number of surveys were completed on other days , day of the week was not included as a variable in the analysis . A dynamic , differential-equation , age-structured , Susceptible-Exposed-Infectious-Recovered ( SEIR ) model [3] was used to investigate whether measured changes in contact patterns could explain the observed epidemic dynamics . In this model , susceptible individuals become infected at a rate proportional to the number of contacts they have with infected individuals . Each contact ( whether made during term time or holidays ) has the same rate of transmission , τ; thus the rate at which a susceptible individual in age group i acquires infection is given by , where Ij is the number of infectious individuals in group j , nj the size of group j , and Bi , j the number of contacts per unit time each individual in group i makes with individuals in group j . When infected , an individual enters the exposed ( latent ) class , during which she is infected but not yet infectious . She then enters the infectious class at rate ν , then recovers at rate g . Because we consider events taking place over only a few months , ageing is not included . The model is described by the following set of differential equations:where Si , Ei , Ii , and Ri are respectively the number of susceptible , exposed , infected , and recovered individuals in group i . The contact matrix {Bi , j} is time dependent , representing differences in mixing patterns between school term times and holidays , taking values BTi , j during term time and BHi , j during the school summer holiday . The initial growth rate of the epidemic , R , was calculated as the dominant eigenvalue of the next generation matrix M , with elements {Mi , j = ( τ/g ) Bj , iSi/ni} ( where , in the early stages of the epidemic Si = ni in the absence of immunity ) [3] . The model was fitted to weekly incidence data based on individuals with ILI who sought medical attention [1] . Combined with laboratory testing of swabs taken from a subset of those who sought medical attention , these data are thought to give a good estimate of the number of cases of H1N1v with ILI who sought medical attention . To estimate the total number of H1N1v cases these observed cases must be scaled up to account for those individuals with influenza who do not seek medical attention . We fit the model to two different estimates of weekly influenza incidence: one calculated by the HPA , using a scaling factor that was informed by flusurvey data made available to the HPA during the early part of the 2009 H1N1v pandemic [1]; the other using subsequent analysis of healthcare-seeking behaviour recorded by flusurvey users with ILI [27] . In reality , both estimates only provide approximations to true incidence trends . The advantage of the latter , flusurvey-adjusted , estimate is that it uses directly measured differences in healthcare-seeking behaviour between different age groups , and changes in this behaviour over time . A large , unknown , number of people infected with influenza were either asymptomatic or displayed only mild symptoms , and as such would not have been recorded as ILI [2] , [21] , even if they had sought medical attention . In common with other modeling work , to account for this under-recording we apply a rescaling factor to the case estimates . Previous modeling work considered a rescaling factor of 7 . 5 , 10 , and 12 . 5 , and concluded that a rescaling factor of 10 was reasonable [2]; here , we seek a more precise value for this parameter . Two models were used: one using social contact pattern data relating to conversational encounters , and a second using data about physical encounters . Weekly incidence as predicted by the model was fitted to estimated incidence data using a least-squares fit . Five model parameters were estimated: the transmission rate , the rescaling factor , the start of the epidemic and the beginning and end of the school holidays . Because of the rescaling factor included in the model , we fit to the shape of the incidence curve not its absolute value . The best-fitting parameter sets ( Table S2 in Text S1 ) were used to calculate the initial growth rate of the epidemic , R , for an outbreak beginning during term time and for one beginning during the school holidays , in the presence and in the absence of pre-existing immunity . Calculated values of R can be found in Table S5 in Text S1 . To explore the role of variability in the collected contact data , 1000 bootstrap copies of the dataset were generated , matching the original dataset in the number of responses from each age group in term time and holiday periods . These bootstrapped datasets were used to estimate a range of contact matrices describing term time and school holiday mixing patterns . It is not the absolute number of contacts but rather the change between holiday and term time contact patterns that is important for understanding the observed incidence; therefore , bootstrapped matrices were ranked according to the ratio of the term time and holiday epidemic growth rates . Models were fitted using those bootstrapped datasets that resulted in contact matrices that generated the 5th and 95th percentiles of this ratio ( referred to as “low-difference bootstrap” and “high-difference bootstrap” respectively ) . Serological testing in England indicated that a large number of people , particularly older people , had prior immunity to H1N1v [20] . In common with other interpretations [2] , [20] , we have assumed that a haemagluttination inhibition titre at or above 1∶32 provides immunity , and that the fraction of the population in each age group with levels greater than this before the epidemic is immune to further H1N1v infection . Values used in the models can be found in Table S4 in Text S1 . To match the availability of serological data , the model population is parameterised to represent the population of England . For simplicity , we use a latent period of one day and an infectious period of 1 . 8 days for H1N1v influenza in the UK , derived from previous modeling work by Baguelin et al [2] . Contact rates between age groups are taken directly from the flusurvey contact survey , and can be found in Text S1 .	Changes in patterns of social mixing can result in changes in epidemic behaviour; this was observed during the 2009 influenza pandemic , in which the epidemic declined during school holidays and grew during term time . Until now , little information has been available to quantify how people's mixing patterns change over time . Here , we present the results of an internet-based survey of social mixing patterns that was carried out in the UK throughout the 2009 pandemic . We show that school holidays resulted in a substantial drop in the number of social contacts made each day , particularly between children . To test whether these measured patterns of social mixing could explain the observed epidemic , we used our mixing data in a simple mathematical model of influenza spread . We found that changing social contact behaviour could explain levels of infection in the community , and conclude that the timing of school terms was responsible for the shape of the influenza epidemic .	[ "Abstract", "Introduction", "Results", "Discussion", "Methods" ]	[ "medicine", "infectious", "diseases", "computer", "science", "public", "health", "and", "epidemiology", "mathematics", "epidemiology", "applied", "mathematics", "biology", "computational", "biology", "population", "biology", "computerized", "simulations", "viral", "diseases...	2012	Measured Dynamic Social Contact Patterns Explain the Spread of H1N1v Influenza
Olfactory neuropiles across different phyla organize into glomerular structures where afferents from a single olfactory receptor class synapse with uniglomerular projecting interneurons . In adult Drosophila , olfactory projection interneurons , partially instructed by the larval olfactory system laid down during embryogenesis , pattern the developing antennal lobe prior to the ingrowth of afferents . In vertebrates it is the afferents that initiate and regulate the development of the first olfactory neuropile . Here we investigate for the first time the embryonic assembly of the Drosophila olfactory network . We use dye injection and genetic labelling to show that during embryogenesis , afferent ingrowth pioneers the development of the olfactory lobe . With a combination of laser ablation experiments and electrophysiological recording from living embryos , we show that olfactory lobe development depends sequentially on contact-mediated and activity-dependent interactions and reveal an unpredicted degree of similarity between the olfactory system development of vertebrates and that of the Drosophila embryo . Our electrophysiological investigation is also the first systematic study of the onset and developmental maturation of normal patterns of spontaneous activity in olfactory sensory neurons , and we uncover some of the mechanisms regulating its dynamics . We find that as development proceeds , activity patterns change , in a way that favours information transfer , and that this change is in part driven by the expression of olfactory receptors . Our findings show an unexpected similarity between the early development of olfactory networks in Drosophila and vertebrates and demonstrate developmental mechanisms that can lead to an improved coding capacity in olfactory neurons . The discontinuous glomerular map at the first relay for olfactory information in vertebrates and insects ( olfactory bulb and antennal lobe , respectively ) is an important model for developmental mechanisms by which neurons assemble into functional neural networks [1]–[6] . This is specially so for the adult olfactory system of Drosophila , since identification of its odorant receptor genes ( OR ) [7]–[9] has made it possible to dissect the organization of the olfactory circuit , and confirms previous ideas of common design principles in vertebrate and insect olfactory systems [10]–[12] . The study of developmental mechanisms that lead to the formation of olfactory circuits in mice and in adult Drosophila has shown that different strategies are used in the two organisms . In mice , olfactory sensory neurons ( OSNs ) lead the process of glomerulus formation and influence the dendritic development of mitral and tufted cells ( the projection neurons of the olfactory bulb ) [13]–[16] . In marked contrast to this , development of the adult olfactory system in Drosophila begins with the positioning of projection neuron dendrites in glomerular-sized territories before the arrival of OSN axons [2] . The adult PNs develop independently of the adult OSNs [17] , and the initial positioning of their dendrites depends partly on signals provided by pre-existing larval OSNs [18] . Thus , larval OSNs play a significant role in patterning the adult olfactory network . The role of activity in the development of the two systems is also different . While blocking activity or synaptic transmission in OSNs during development in mice or zebrafish shows that activity is essential for development and refinement of the olfactory map [19] , [20] , similar experiments have failed to show any such developmental effects in adult Drosophila [21] . In none of these systems , however , has the pattern of neuronal activity during development been documented and this means that the results are difficult to interpret because the patterns of activity that are being blocked are unknown . The larval olfactory system of Drosophila shares organizational principles and all the experimental advantages of its adult counterpart , but is numerically much simpler . It consists of 21 OSNs with their cell bodies grouped in an anterior ganglion ( Dorsal organ ganglion , DOG ) . These neurons send dendrites to the dorsal organ ( DO ) , where odour volatiles are detected , and axons into the CNS , where they terminate in the antennal lobe ( AL ) . Each of the OSNs expresses a different OR and sends its axons to a different glomerulus . Thus , unlike the adult or indeed vertebrate systems , in the larva there is no convergence of OSN axons , and every OSN constitutes a single class . Each glomerulus is innervated by one PN , establishing an olfactory map like the one present in vertebrates but with 1∶1 connectivity [22]–[24] . Despite the role of the larval olfactory system in patterning the adult olfactory circuit [18] and a growing number of studies using the larval system as a model olfactory network [25]–[27] , almost nothing is known about its developmental origins , let alone the way in which pre- and postsynaptic neurons come together to form a functional olfactory network . Indeed , the only information we have at present concerns the precursors of the OSNs and PNs [1] , [28] . Here we describe the development and regulated assembly of the larval olfactory circuit in Drosophila from its earliest beginnings in the embryo to functional maturity at hatching . We use a combination of genetic and dye injection techniques to determine the sequence of events leading to olfactory wiring . We combine genetic and laser ablation techniques to show that the development of the larval PNs , unlike their adult counterparts , depends on the ingrowing axons of embryonic OSNs . In addition we investigate the emergence of spontaneous patterns of activity in embryonic OSNs as they develop and show that this activity plays a role in restricting OSN glomerular territories . Our results reveal an unexpected degree of similarity between the development of the olfactory systems in vertebrates and the Drosophila larva . To study the development of the olfactory system , we used a combination of genetic labelling with the Gal4/UAS system [29] and single cell dye injection . We visualized OSNs and PNs at early stages , before any contact is made , by using acj6-Gal4 , which is expressed in all embryonic OSNs [30] and PNs ( personal observation ) . We find that OSNs are born in contact with the brain ( Figure 1A–E ) , and from stage 13 ( about 10 h AEL ) onwards they extend short axonal projections into the brain ( Figure 1E ) . The OSN axons elongate during development as the process of head involution [31] displaces them away from their initial lateral-ventral position towards their final more dorsal and anterior position . Within the brain , OSN axons grow a short distance until they reach the location of the future AL ( Figure 1A–F ) . The PNs extend axons towards higher brain centres before they sprout their dendrites ( Figure 1C , 1F , and 1K–M ) . OSN axon terminals first make contact with the most proximally located region of the PN axonal bundle at late stage 15 ( about 12 h AEL ) ( Figure 1F and 1J ) . To get a detailed insight into this process , we performed single cell dye injections of OSNs and PNs at seven different developmental stages ( 13 h , 14 h , 15 h , 16 h , 17 h , 18 . 5 h , and 21 h AEL ) . In total we injected 73 OSNs ( on average 10 . 42 for each stage , and a minimum of six ) and 107 PNs ( on average 17 . 83 for each stage , and a minimum of seven ) . At 12–13 h AEL , when OSN axon terminals contact the dendrite-less axons of PNs for the first time , all OSN terminals have prominent growth cones ( Figure 1H and 1J , n = 11 ) , which in 50% of the filled OSNs are still present at 14 h AEL ( n = 10 ) , but not at 15 h AEL or later ( compare Figure 1H–J to Figure 2B–I ) . Injections of pairs of OSNs with two different dyes at 13 h and 14 h AEL , when filopodia are still present , reveals that even at these early stages OSN terminals occupy distinct territories , which may represent their final positions within the larval AL ( Figure 1H ) . At 14 h AEL none of the PNs injected had sprouted dendrites yet ( n = 14 , Figure 1L–L′ and 1M–M′ , and Figure S1 ) . At 15 h AEL only 43% of the PNs injected had a dendrite ( n = 21 ) , while by 16 h AEL all PNs injected showed dendritic growth ( n = 18 , Figure 2K ) . Thus , unlike the Drosophila adult , in the embryo , PNs only extend dendrites after they have been extensively contacted by axonal terminations of OSNs . We now followed the further maturation of the OSN terminals and PN dendrites and axons until hatching at 21 h AEL . OSN terminals are very variable during embryogenesis ( Figure 2B–G ) , but by 18 . 5 h AEL the terminals begin to mature , they are more condensed , although they still have filopodia and overshoot their glomerular territories ( Figure 2D and 2H ) . By hatching , terminals are less variable and more compact ( Figure 2E and 2I ) . PN dendrites with short filopodia grow and branch in restricted glomerular-sized territories from the very beginning rather than growing in an exploratory fashion followed by pruning ( Figure 2K–N ) . The structure of PN dendrites is very variable during development and continues to be so until hatching ( Figure 2K–N ) . This makes it difficult to identify unequivocally when PN dendrites acquire their mature structure . The output sites of PNs , at their terminals in the MB and LH , are still immature at 16 and 17 h AEL with prominent growth cones ( Figure 2K′ and 2L′ ) . It is only at 18 . 5 h AEL that axons of PNs start acquiring their characteristic first instar larval morphologies , without growth cones and innervating one or two glomeruli in the MB ( Figure 2M′ , compare to Figure 2N′ ) . Taken together our results suggest that 18 . 5 h AEL is the stage at which pre- and post-synaptic components of the larval olfactory system first begin to acquire their mature morphologies . Since larval OSNs have been reported to fire action potentials spontaneously [27] , we wondered when this activity starts , and whether it plays a role in the development of the olfactory network during embryogenesis . We therefore developed a technique to record extracellularly from embryonic OSN cell bodies . Using this technique , we recorded simultaneously from a random sample of up to 6 of the 21 OSNs in the DOG ( Figure 3A ) . In each recording , electrophysiological activity was allocated to individual OSNs using the newly developed spike sorting software Spikepy ( http://code . google . com/p/spikepy/ , Figure 3B–D , Materials and Methods ) . To confirm the activity we recorded was indeed originating from OSNs , we expressed the light activatable channel , Channelrhodopsin-2 ( ChR-2 ) [32] in all OSNs ( Orco-Gal4;UAS-CD8GFP/UAS-ChR2 ) . We found that when OSNs were activated by exposure to 470 nm light , there was an increase in the firing rate of all units in our recordings ( Figure 3B and 3E; for each individual unit the mean firing rate in the 10 s during light stimulation was significantly higher than the mean firing rate in the 10 s before light stimulation , p<0 . 001 ) , confirming that the activity we record does indeed derive predominantly from OSNs . We now asked when OSN activity begins by recording from the OSNs of embryos at different stages . We found that only 42% of the embryos showed OSN action potentials at 15 h AEL , even when stimulated with ChR2 ( n = 12 ) , while from 16 h AEL and onwards , we recorded spontaneous activity in every embryo we tested . Thus , we conclude that spontaneous spiking in OSNs begins at about 15 h AEL . Interestingly , action potentials recorded extracellularly at these early stages ( 15 h AEL ) have a different shape , and a longer time course than the ones recorded at later stages , and the shape changes progressively over development ( Figure 3F ) . The spontaneous firing rate is also lower at earlier stages and increases progressively as development proceeds ( Figure 3F , firing rate at 15 h , 0 . 1±0 . 02 Hz , n = 25 units from four different embryos; 16 h , 0 . 59±0 . 07 Hz , n = 37 units from eight different embryos; 18 . 5 h , 1 . 13±0 . 31 Hz , n = 21 units from six different embryos; first larva: 2 . 53±0 . 52 Hz , n = 24 units from four different larvae ) . The olfactory system of Drosophila larvae is thought to code for the presence of particular odours using a rate coding strategy , combined with a population code [26] . This means that in response to an odour each OSN responsive to that odour codes for its presence by changing the frequency at which it fires action potentials ( rate coding ) , rather than by precise timing of spikes ( temporal coding ) . Signal detection theory suggests that discriminability between the presence and absence of a stimulus ( stimulus detectability ) is governed by the absolute value of the difference between the mean of the spike counts with and without stimulus divided by the square root of the summed variances of these spike counts ( equation 1 ) . Thus stimulus detectability is enhanced by low variability in the spike train , because this results in low variability in spike count ( low variance ) over the counting window . In our data we observe that spontaneous firing patterns at early stages ( 15 h and 16 h AEL ) are more bursty than at later stages ( 18 . 5 h AEL and L1 ) ( Figure 4A ) , with inter-burst intervals ( IBIs ) varying between approximately 0 . 5 and 2 min , similar to the IBIs found in other developing systems [33] . We wondered whether variability in the spike train ( and therefore discriminability ) would change during the course of development . As a commonly used and straightforward measure of spike train variability , we use the coefficient of variation of interspike intervals ( CV ) , which is defined as the ratio of the standard deviation to the mean of the interspike intervals ( ISIs ) ; thus , the lower the CV , the lower the variability in the spike train [34] , [35] . We find that as development proceeds , the CV decreases progressively , with a statistically significant step between 16 h and 18 . 5 h AEL ( CV , for 16 h = 1 . 59±0 . 13 , for 18 . 5 h = 1 . 18±0 . 079 , p = 0 . 03; Figure 4B ) . Thus , the pattern of OSN activity changes over development , reducing spike train variability , which in turn might serve to increase odour detectability . Because olfactory receptors are responsible for the high spontaneous firing rate of larval OSNs [27] , we reasoned that the developmental change in the pattern of spontaneous activity in OSNs might be due to the onset of OR functioning . To test this , we recorded OSN activity from Orco mutants at 16 h AEL ( first time when all embryos showed spontaneous activity ) and first instar larvae . Orco mutants do not traffic the specific ORs to the membrane and are therefore anosmic [36] , and their spontaneous firing rate in third instar larvae and adults is diminished but not abolished [27] . As expected , the spontaneous firing rate of Orco mutant first instar larvae , but not 16 h AEL embryos , is reduced by half when compared to controls ( firing rate , for 16h_control = 0 . 58±0 . 07 Hz , n = 37 units from eight different embryos; 16h_Orco_mut = 0 . 68±0 . 13 Hz , n = 28 units from four different embryos; p = 0 . 55 , for L1_control = 2 . 53±0 . 52 Hz , n = 24 units from four different larvae; L1_Orco_mut = 1 . 23±0 . 25 Hz , n = 24 units from four different larvae; p = 0 . 03 , Figure 4C ) . We then analyzed the CV of the ISIs and found that indeed spike train variability is significantly increased in Orco mutant first instar larvae , and 16 h AEL embryos , when compared with controls ( 16h_control = 1 . 59±0 . 13 , 16h_Orco_mut = 2 . 35±0 . 22 , p = 0 . 01; L1_control = 1 . 08±0 . 08 , L1_Orco_mut = 1 . 72±0 . 21 , p = 0 . 01; Figure 4B ) , and accordingly its activity patterns are more bursty ( Figure 4A ) . We conclude that the reduction in the variability of the spike train during development requires , at least in part , Orco expression and is therefore likely to be attributable to the onset of OR function ( see Discussion ) . Having identified the principal steps in the morphological and physiological development of the olfactory circuit , we decided to look for regulatory mechanisms operating to ensure integrated assembly of pre- and postsynaptic elements in the antennal lobe . Since PNs only extend dendrites several hours after they have been extensively contacted by OSNs , it seemed possible that the growth of PN dendrites might be regulated by the presence of OSN terminals . To address this question , we laser-ablated all the OSNs on one side in intact embryos before the stage at which OSNs and PNs make contact ( Figure 5A ) . We then allowed the embryos to develop until hatching and dissected them as first instar larvae . We targeted the ablations by expressing GFP in all sensory neurons with the Gal4 line PO163 [37] . The presence of the label allowed the success of the ablation to be assessed immediately after the operation and later in the first instar dissected animals . The ablations were very specific and only OSNs were ablated , while other closely positioned sensory neurons , such as taste neurons , remained intact ( Figure 5B ) . We visualized PNs in these animals using Q-system , a Gal4-independent expression system; specifically , we used GH146-QF , which is expressed exclusively on PNs [38] . Interestingly , our experiments show that PNs require presynaptic innervation during development to survive . In 22% of cases ( n = 27 ) , there were no PNs on the ablated side , as compared with complete survival of all PNs on the control side , visualized with both GH146-QF and PO163-Gal4 , which is fortuitously also expressed in PNs ( Figure 5C and 5D–D″ ) . In those cases where some PNs survived on the ablated side , they were found contacting other brain regions , most commonly the subesophageal ganglion ( SOG ) , presumably due to its proximity to the AL ( Figure 5E ) , but occasionally PNs attracted innervation from other neighbouring axon bundles ( Figure S2 ) . In every case the AL structure was completely lost from the ablated side , whether it was visualized with Nc82 antibody staining or as a conspicuous gap in DAPI labelling ( Figure 5D and 5E ) . Thus the survival of PNs appears to require presynaptic innervation during development , but this innervation does not need to be specifically from OSNs , and other axonal terminals can also support PN survival . Interestingly , when PNs do survive , their dendrites are normally longer than controls ( Figure 5E ) , suggesting they elongate until they find presynaptic partners , with the implication that OSNs would normally give PN dendrites a stop growth signal . Our recordings from developing OSNs show that spontaneous activity starts at the time of PN dendrite extension ( 15 h AEL ) . This together with the finding that PNs require presynaptic innervation during development for survival suggested that PN dendrite development or survival might depend on the spontaneous firing of OSNs . We therefore silenced OSNs by expressing UAS-Kir2 . 1 using the Orco-Gal4 line , which is expressed in all OSNs before the onset of spontaneous activity ( Figure S3 ) . Expression of Kir2 . 1 has been shown to silence several types of Drosophila neurons [39] , [40] , and when expressed in adult OSNs , all odour responses are abolished [36] , but its use in larval OSNs has not been reported . We therefore tested the effects of our manipulation by recording extracellularly as described above from control first instar larvae ( Orco-Gal4;UAS-GFP ) and larvae expressing Kir2 . 1 in OSNs ( Orco-Gal4;UAS-Kir2 . 1 ) . We found that in eight out of eight control larvae we could record action potentials . However , as expected , action potentials could not be recorded from any of eight Kir-expressing animals ( Figure 6A and B , Materials and Methods ) . We then examined PNs using GH146-QF;QUAS-mTomato , in first instar larvae in which OSNs had been silenced by Kir expression . We found that the gross morphology of PNs in these larvae appeared normal when compared to controls ( Figure 6C and D ) . However , quantification of PN dendritic occupancy within the AL as quantified by pixel intensity plots ( see Materials and Methods ) revealed a small but significant ( Figure 6E , p<0 . 0001 , n = 12 ) increase in PN dendrite occupancy within the AL when all OSNs had been silenced , as compared to controls . Thus , we conclude that OSN activity is dispensable for PN survival , dendrite extension , and maintenance and that the requirement for innervation that we have demonstrated is therefore likely to be contact mediated and activity independent . However , we demonstrate that in the absence of OSN activity , PN dendrites overgrow , which supports our previous observation that OSNs provide a stop growth signal to PN dendrites , probably through both contact and activity-dependent interactions . We next investigated whether spontaneous activity in embryonic OSNs might have a more subtle role in the wiring of the olfactory circuit . We sought to answer this question by genetically silencing and visualizing a subset of OSNs . Because we found published OSN lines that are specific for a subset of OSNs , such as OR specific lines , only begin to be expressed around 18 . 5 h AEL ( Figure S3 ) , we screened for new Gal4 lines that would be expressed in subsets of OSNs from earlier stages . We find that Lim3b-Gal4 is expressed reliably in four of the 21 larval OSNs from 17 h to 21 h AEL ( Figure S4 ) , with an onset of expression in two of these OSNs at 16 h AEL . We used this line to drive expression of a new UAS-myr-mRFP insertion line [41] , which is brighter than previously available lines , and thus allows us to visualize fine neuronal processes . We crossed this Lim3b-Gal4;UAS-mRFP line with UAS-Kir2 . 1 or UAS-DorK to silence the four OSNs . DorK ( Drosophila open rectifier K channel ) has previously been shown to silence Drosophila neurons in a similar way to Kir2 . 1 [42] , and we decided to use it as an alternative and independent way to silence OSNs . We find that the silenced OSNs in first instar larvae have the immature-like morphologies with broad axonal terminals and multiple filopodia that are more characteristic of earlier developmental stages than newly hatched larvae ( Figure 7E and 7H ) . Quantification of ALs with this phenotype in control and experimental animals showed that the difference is significant ( percentage of AL with immature OSN terminals: control = 19% , n = 16; Lim3b::Kir2 . 1 = 69% , n = 23; Lim3b::DorK = 70% , n = 27; p ( control , Lim3b::Kir2 . 1 ) = 0 . 003; p ( control , Lim3b::DorK ) = 0 . 001 , Figure 7A–K and 7L ) . Silenced terminals also appeared to be expanded within the AL when compared to controls , occasionally even extending beyond the synaptic region of the AL ( Figure 7D and 7F ) . We reconstructed the volumes of the AL and OSN terminals ( see Materials and Methods ) and found that silenced OSNs indeed occupied a larger percentage of the AL synaptic volume than controls ( p ( control-kir ) = 0 . 02; p ( control-DorK ) = 0 . 04; Figure 7M ) . These data suggest that spontaneous activity of OSNs is essential for them to develop normal axon terminals and that its absence either triggers or fails to suppress an exploratory growth programme . Our analysis to this point has been restricted to the role of OSN spontaneous activity under competitive conditions , in which only four of the 21 OSNs are silenced while the rest have normal levels of activity . To test for possible activity-dependent interactions among neighbouring OSN terminals , we documented the morphological and volumetric characteristics of the same four Lim3b positive terminals when all OSNs had been silenced . To achieve this , we generated a LexA-Kir2 . 1 line that when combined with Orco-LexAOp [43] silences all OSNs , leaving the Gal4/UAS system available for visualization of the same OSNs that we analyzed before ( Figure 7I ) . Under these conditions , when activity is blocked in all OSNs , the terminals of the four Lim3b positive OSNs remain immature in the first instar larva ( p ( control_Orco::Kir ) = 0 . 001; n ( Orco::Kir ) = 16; Figure 7I–J and 7L ) , comparable to the condition where all other OSNs are unaffected . We therefore conclude that this immature phenotype is cell-autonomous and independent of interactions between neighbouring terminals . However , OSNs that develop in an AL in which all OSN activity has been blocked occupy a volume within the AL that is not significantly different from that of controls ( p ( control_Orco:: Kir ) = 0 . 25 , Figure 7M ) . Thus acquisition of a mature morphology by OSN terminals requires spontaneous activity in the cells concerned , regardless of whether neighbouring cells are active or not . However , the restriction of terminal volumes within the antennal lobe is a competitive process in which silent endings appear to have a growth advantage over neighbouring active terminals . A key finding in this study is the interdependence of OSNs and PNs for the proper development of the larval AL . Although at early stages of embryogenesis OSN and PN axons approach the site of the future AL independently of each other , once PN dendrites penetrate the emerging AL , interactions with OSN regulate the patterning of connectivity . Embryonic development of the Drosophila AL begins with OSN terminals targeting distinct territories that probably represent the origins of AL glomeruli . At this stage PN axons turn away from this site and continue growing towards higher brain centres . By the time that growth cones of OSN axons contact the proximal region of PNs axons , the PNs have not yet extended any dendrites . Hours later , PN extend dendrites directed towards particular territories within the emerging AL , possibly guided by the same cues that direct OSN terminal targeting . The early arrival of OSNs in the future region of the AL before PN dendrite extension suggested a possible role for OSNs in the development of the AL . Indeed , we found that PNs require presynaptic innervation for their survival , although innervation does not necessarily have to come from OSNs . Additionally , there is no specific requirement for OSN terminals in promoting sprouting of PN dendrites since in the absence of OSNs , surviving PNs have dendrites . These dendrites are normally longer than controls , suggesting they elongate until they find presynaptic partners , with the implication that OSNs normally give PN dendrites a stop growth signal . We show that this effect is both contact and activity dependent , because PNs in animals where all OSNs had been silenced have overgrown dendrites that do not extend beyond the AL . A similar effect has been found in the dendrites of motorneurons in Drosophila embryos , where the removal of presynaptic terminals induces an overgrowth of postsynaptic motorneuron dendrites that anticipates the dendritic overgrowth induced by the lack of pre-synaptic activity at later developmental stages [44] . Independently of whether PNs survive or not , in all cases the AL is lost when OSNs are ablated . Loss of the AL has also occurred on an evolutionary scale in terrestrial isopods , which in the process of colonising the land have secondarily lost their olfactory sensilla in the main olfactory appendage , together with the corresponding olfactory deutocerebral structures ( second neuromere of the supraesophageal ganglion where the olfactory lobe is located ) . Furthermore , in some species the tritocerebrum ( posteriorly adjacent neuromere to the deutocerebrum ) seems to have acquired additional neuropile structures [45] . Our findings show that there is an interdependence in the development of the Drosophila embryonic olfactory system that results in the loss of deutocerebral olfactory structures ( the AL ) in response to the ablation of OSNs . At the same time the finding of occasional ectopic tritocerebral and subesophageal innervation of PNs indicates a possible developmental route for the evolutionary acquisition of additional tritocerebral structures . Our results contrast with previous studies in adult Drosophila , which show that PNs pioneer development of the adult AL independently from adult OSN development . Why is development of the olfactory system in Drosophila different during embryogenesis and metamorphosis ? Interestingly , experiments in other embryonically developing olfactory systems , in both vertebrates and invertebrates , also demonstrate an essential role for OSN ingrowth in the development of their first olfactory centres . Experiments in Xenopus where OSNs were removed unilaterally at early embryonic stages showed that an olfactory bulb fails to develop on the ablated side , but is present on the control side [46] . Similarly , an experiment in cockroaches where most , but not all , OSNs were unilaterally removed during embryogenesis before they innervate the AL showed that the deafferented lobe was severely disrupted , its characteristic glomeruli were missing , and it was markedly reduced in volume . Furthermore , as with our findings , PNs in these partially deafferented lobes were sparsely branched and had elongated dendrites instead of their characteristic uniglomerular tufts [47] . In contrast , when OSNs were ablated early in adult development in insects ( Manduca [48] and Drosophila adult [17] ) an AL still formed , and PN dendrites arborized in their glomerular territories . We conclude that the differences we find in the development of the Drosophila larval and adult olfactory systems probably arise from fundamental differences between embryonic development and metamorphosis . In embryos ( vertebrate or Drosophila ) there is no preexisting network to guide development , whereas during metamorphosis the adult olfactory system makes use of cues derived from the larval olfactory system [18] . Thus its wiring seems to rely more on external cues and less on interactions among its network components than the wiring of the larval network . Our method allows spontaneous activity to be recorded from OSNs developing in vivo in the Drosophila embryo . Although it has been assumed that OSNs in mice and insects may be active during development ( see below ) [19] , [21] , and there is a previous report of activity recorded from the antennal nerve of Manduca during adult development [49] , ours is , to our knowledge , the first systematic description of the onset and developmental maturation of normal patterns of spontaneous activity in OSNs . Our results reveal three important features about the development of activity patterns in OSNs: As in other developing systems [50] , [51] , the earliest action potentials generated by OSNs are different from mature ones , with smaller amplitude and a longer duration . Such changes in spike shape seem to be a general feature of emerging activity as ionic conductances are acquired and mature . At early stages we record intermittent bursts of activity in the OSNs . Activity patterns that consist of spontaneous bursts are common to many developing neural networks , including the auditory [52] , visual [53] , motor [54] , [55] , and olfactory systems ( this study and [50] and their time course is remarkably similar across different neural systems , with inter-burst intervals varying between 0 . 5 and 2 mi [33] like those we report here ) . Such activity may be an inevitable consequence of cells acquiring mature excitable properties , but it is also possible that the generality of these activity patterns , and the diversity of mechanisms by which they are generated and terminated , is an indication of an essential and significant role in the development of neural networks [33] . As development proceeds , variability of the spike train diminishes , which is predicted according to information theory to increase signal ( odour ) detectability . A previous in vitro study of locust frontal ganglion neurons showed that there is a transient period during the wiring process when activity is irregular , but as the network matures , regularity increases [56] . As far as we know , ours is the first direct statistical analysis of the transition from immature to mature spike-trains in vivo and allows us to suggest that the coding capabilities of the network improve as it develops . It seems likely that a change towards patterns that would be expected to increase signal detectability , and thus network functionality , would be a general feature in neural networks as they mature . The mechanisms by which this immature activity is generated , shaped , and terminated vary from system to system [33] . In the embryonic OSNs , the transition from irregular spike-trains to continuous discharge may require the expression of olfactory receptors ( OR ) , because in larvae mutant for the co-receptor Orco , necessary for OR function , this transition does not occur normally . Since Orco is expressed before the onset of spontaneous activity ( Figure S3 ) , we suggest that the change in the pattern of OSN spontaneous activity is likely to be driven , at least in part , by the onset and level of expression of specific ORs . However , this might not be the only factor shaping spontaneous activity patterns over development , and other factors such as expression of other ion channels may also play a role . This might explain why 16 h AEL Orco mutants have indistinguishable levels of activity when compared with controls , yet the variability in their spike train is significantly increased . Previous studies have suggested that spontaneous activity is essential for the normal development of vertebrate OSNs , but that there is no such requirement in insects [19] , [21] , [49] , [57] . However , we find that there is a role for OSN activity in the development of the larval olfactory network . OSN activity regulates the morphology of OSN terminals independently of activity in neighbouring axons , and without activity terminals appear immature and occupy larger territories . This is similar to what has been described in zebrafish and mouse OSN terminals devoid of activity [19] , [57] . There is also a report of a similar phenotype found in the AL of third instar Drosophila larvae after synaptic release was blocked in a large subset of OSNs [58] . Our results show that while immature terminal morphology is a cell autonomous phenotype that is independent of activity levels in neighbouring OSN axons , the expansion of OSN terminals is limited by interactions among the OSN terminals . Interestingly a similar process has been found to regulate the morphology and terminal expansion of retinotectal axons [59] . Thus the control of axonal terminal extension via activity-dependent interactions may be a general process in the wiring of nervous systems . The nature of inter-axonal interactions that limit terminal growth remains unknown and is one example of how future work using amenable experimental systems such as the one provided by the larval olfactory network in Drosophila larvae may reveal general mechanisms operating during the assembly of neural circuitry . Eggs were collected from flies kept on apple juice agar supplemented with yeast paste and maintained at 25°C . For ChR2 expressing flies , we supplemented the yeast paste with all-trans retinal ( Sigma-Aldrich ) to a final concentration of 1 mM . Fly stocks and recombinant chromosomes were generated using standard procedures . For embryonic stages before 13 h AEL , flies were left to lay eggs on agar plates for 14 h at 25°C and whole mount stainings were performed . Whole mount immunostainings followed standard protocols . Primary antibodies used were: goat anti-GFP 1∶500 Abcam and mouse 22c10 1∶20 DSHB ( Developmental Studies Hybridoma Bank , USA ) secondary antibodies were Alexa 488 and Cy3 conjugated ( 1∶800 , Invitrogen and Jackson Lab , respectively ) . Specimens were cleared and mounted in Vectashield ( Vector Laboratories ) under n°1 coverglasses , and for the different orientations they were rotated under the coverglass before the confocal scans . Embryos were dechorionated with bleach for 5 min and selected during two short windows , either at the three-part gut stage ( 13 h AEL ) or when their main dorsal tracheae begin to fill with air ( 18 . 5 h AEL ) . These embryos were placed on a 25°C agar plate in an incubator for the number of hours required for each time point . PNs were filled with Lucifer Yellow as described in [44] . Primary antibodies used were: rabbit anti-Lucifer Yellow 1∶1 , 000 , Invitrogen; goat anti-GFP 1∶1600 , and AbCam; mouse Nc82 [60] , 1∶50 , DSHB . Secondary antibodies were Alexa 488 , Cy3 , and Cy5 conjugated ( 1∶800 , Invitrogen and Jackson Lab , respectively ) . OSNs were filled with the lipophilic tracer dyes DiI and DiO , because the limited diffusion of LY did not allow visualization of the terminal axons of OSN in the AL . Embryos of the appropriate stage were dissected on poly-Lysine-coated coverslips up to 15 h AEL and on Sylgard-coated coverslips for embryos older than 15 h AEL as in [41] , but opening the front of the animal to get access to the OSNs . Injections proceeded as described for Lucifer yellow injections , but OSNs were filled intracellulary with DiI or DiO dissolved in dry EtOH ( 2 mg/ml ) . The dye was allowed to diffuse for 2–4 h at 4°C , and specimens were mounted as described for PN preparations . For ablation experiments , embryos were dechoronionated with bleach for 5 min and selected at stage 14 based on features as described in [31] and OSNs were visualized under a 60× dipping lens with a LSR ultraview scan head mounted on a Leica DM6000B spinning disk microscope . OSNs were ablated with a Micropoint laser ablation system from Spectra-Physics ( USA ) mounted on this microscope and using a computer interface from Metamorph software ( molecular devices ) . Once OSNs were ablated , embryos were placed in individual apple juice agar plates and left to develop for 24 h . Brain dissections and antibody stainings were as for the Lucifer Yellow injections . No antibody was used to visualize RFP fluorescence , and the images show native RFP fluorescence after fixation and antibody staining in the other two channels . Images were collected using a Leica SP5 confocal laser scanning microscope . Image z-stacks were processed using ImageJ 1 . 39 s software ( U . S National Institute of Health , Bethesda , Maryland , USA , http://rsbweb . nih . gov/ij/ ) , and figures were generated using Photoshop CS2 ( Adobe Systems , San Jose , CA ) . For calculating AL and OSN volumes in Figure 7 , the “Segmentation editor” plugin of ImageJ was used . To calculate the data in Figure 6E , images were imported in Amira ( http://www . amira . com/ ) , and AL volumes were marked using the segmentation editor , and used as a mask in the PN dendrite channel , where intensity histograms using a bin number of 256 were calculated . These data were exported to Excel where intensity histograms were normalized for total pixel number for every stack and subsequently exported to “R project” ( R Foundation for Statistical Computing , Vienna , Austria , 2005 . http://R-project . org ) for plotting . For delimiting the AL volume the contour of the Nc82 staining of the AL was used . The pLOT-EGFP-Kir construct was created by amplifying the EGFP-Kir cDNA from pUAST-EGFP-Kir vector [39] by Polymerase Chain Reaction ( PCR ) using the following Gateway primers: attB1-ATG GTG AGC AAG GGC GAG GAG CTG T and attB2-TCA TAT CTC CGA TTC TCG CCG TAA G . The PCR product was introduced into pDONRTM221 ( Invitrogen ) via Gateway cloning to create the pEntry-vector . In the subsequent Gateway cloning reaction , this vector was combined with the pLOT-W vector [61] to fuse the EGFP-Kir channel downstream of the LexA operon . DNA was purified using a Qiagen Midi Kit and transgenic lines were generated by BestGene Inc . ( Chino Hills , CA , USA ) . Embryos of the appropriate stage were dissected as for OSN dye injections in physiological saline composed of ( in mM ) : 135 NaCl , 5 KCl , 5 CaCl2-2H2O , 4 MgCl2-6H2O , 5 TES ( 2-[[1 , 3-dihydroxy-2- ( hydroxymethyl ) propan-2-yl]amino]ethanesulfonic acid ) , 36 Sucrose , adjusted to pH 7 . 15 with NaOH . Borosilicate glass capillaries were pulled with a Sutter Instruments P-87 puller and fire polished to achieve a final tip of approximately 5 µm . The recording electrodes were back filled with physiological saline . The recording electrode was placed close to the DOG under an Olympus BX51WI microscope with a 60× water immersion objective using a hydraulic micromanipulator ( Narishige ) , and suction was applied with a syringe until a seal was obtained and action potentials could be recorded . Normally between one and eight cells were sucked into the pipette . A wire in the bath acted as reference electrode . Voltage signals were amplified with a differential AC amplifier ( AM-Systems , Sequim , WA ) . Signals were low-pass filtered at 10 KHz and digitalized at 20 KHz . Data were digitized using a Power Lab 4/30 and recorded with LabChart5 ( AD instruments ) . Data acquired with LabChart5 were exported as MATLAB files . Recordings were divided into 5 min segments and imported into Spikepy , an open-source spike sorting software ( http://code . google . com/p/spikepy/ ) . At least one segment of 5 min was analyzed per animal . When recordings were very long , a maximum of four representative segments were analyzed . Units were separated using Spikepy with procedures that were adopted with the aim of reliably distinguishing cells rather than picking up the maximum number of cells . Briefly , the software filters the data to remove part of the baseline noise , applies an amplitude threshold for detecting the spikes , then extracts spike features for spike sorting using the full spike shape , and then clusters the spikes using the K-means method . Thus , Spikepy discriminates spikes based on shape and amplitude . It then outputs the results as graphs as shown in Figures 3 and 4 , and as a MATLAB file containing information on each cluster , among other parameters . MATLAB files generated by Spikepy were opened in MATLAB , and firing rates , PSTH , and CVs were calculated . For experiments referring to the absence of action potentials in Orco-Gal4::UAS-Kir2 . 1 animals , recordings were performed alternatively from control and experimental animals ( n = 8 for each condition ) . Our success rate of recording action potentials in control animals was of 100% , however in the Orco::Kir2 . 1 animals , no action potentials could be recorded , even after repeated trials . Data were analyzed and plotted using “R project” ( R Foundation for Statistical Computing , Vienna , Austria , 2005; http://R-project . org ) . Data were analyzed statistically using the Shapiro-Wilk test to assess for normality followed by a Student's t test or a Wilcoxon rank-sum test as appropriate . The exceptions are the data in Figure 7L , which are categorical and therefore analyzed with Fisher's exact test , and data in Figure 6E , which were analyzed using a Kolmogorov-Smirnov test: ( 1 )	The mechanisms underlying the patterning of connectivity in the insect olfactory system are radically different from those found in vertebrates , but to date most studies in insects have focused on the development of the adult olfactory network . Here , for the first time , we report how larval olfactory circuitry is formed in the embryo of the fruitfly Drosophila . By labelling developing sensory neurons and interneurons from the earliest stages to maturity , we find that the patterning of the antennal lobe in Drosophila , like the olfactory bulb in mouse , is pioneered by ingrowing sensory afferents , and that interneuronal development depends on the terminals of these pioneering afferents . We also find that antennal lobe patterning depends on contact and activity-mediated interactionsbetween its component cells , as it does in vertebrates . Finally , we report the results of electrophysiological recordings in developing embryos , the first of their kind in any developing olfactory network . We conclude that fundamental mechanisms of circuit assembly and patterning are conserved between Drosophila and vertebrates .	[ "Abstract", "Introduction", "Results", "Discussion", "Materials", "and", "Methods" ]	[ "developmental", "biology", "biology", "neuroscience" ]	2012	Embryonic Origin of Olfactory Circuitry in Drosophila: Contact and Activity-Mediated Interactions Pattern Connectivity in the Antennal Lobe
In all sexual organisms , adaptations exist that secure the safe reassortment of homologous alleles and prevent the intrusion of potentially hazardous alien DNA . Some bacteria engage in a simple form of sex known as transformation . In the human pathogen Neisseria meningitidis and in related bacterial species , transformation by exogenous DNA is regulated by the presence of a specific DNA Uptake Sequence ( DUS ) , which is present in thousands of copies in the respective genomes . DUS affects transformation by limiting DNA uptake and recombination in favour of homologous DNA . The specific mechanisms of DUS–dependent genetic transformation have remained elusive . Bioinformatic analyses of family Neisseriaceae genomes reveal eight distinct variants of DUS . These variants are here termed DUS dialects , and their effect on interspecies commutation is demonstrated . Each of the DUS dialects is remarkably conserved within each species and is distributed consistent with a robust Neisseriaceae phylogeny based on core genome sequences . The impact of individual single nucleotide transversions in DUS on meningococcal transformation and on DNA binding and uptake is analysed . The results show that a DUS core 5′-CTG-3′ is required for transformation and that transversions in this core reduce DNA uptake more than two orders of magnitude although the level of DNA binding remains less affected . Distinct DUS dialects are efficient barriers to interspecies recombination in N . meningitidis , N . elongata , Kingella denitrificans , and Eikenella corrodens , despite the presence of the core sequence . The degree of similarity between the DUS dialect of the recipient species and the donor DNA directly correlates with the level of transformation and DNA binding and uptake . Finally , DUS–dependent transformation is documented in the genera Eikenella and Kingella for the first time . The results presented here advance our understanding of the function and evolution of DUS and genetic transformation in bacteria , and define the phylogenetic relationships within the Neisseriaceae family . Transformation in bacteria is a complex process involving uptake of naked extracellular DNA followed by homologous recombination ( HR ) . Different reproductive barriers have evolved in diverse transformation-competent bacteria , which distinguish in favour of acquisition and recombination of homologous DNA sequences and discriminate against heterologous and potentially hazardous DNA [1] . In particular , interspecies recombination with heterologous DNA in single cellular organisms could cause gene disruptions and/or disturb sensitive cellular processes , which could in turn have adverse phenotypic consequences . Adaptations that may contribute to sexual isolation and at the same time promote genetic stability include restriction modification systems , fratricide in streptococci and cannibalism in Bacillus subtilis , quorum-sensing , biofilm formation and HR regulation and suppression [2] , [3] , [4] , [5] . Transformation in Neisseria sp . and members of the Pasteurellaceae family is unique in the requirement for short uptake sequences in the transforming DNA , named DNA Uptake Sequences ( DUS ) and Uptake Signal Sequences ( USS ) , respectively [6] , [7] . The genomes of these organisms harbour thousands of DUS and USS , constituting up to 1% of their entire chromosomes [8] , [9] , [10] . DUS has accumulated in the core genome , i . e . the set of common genes , of N . meningitidis , N . gonorrhoeae and N . lactamica and was found to maintain its sequence identity from frequent recombination [11] . DUS was first identified in N . gonorrhoeae as a 10-mer ( 5′-GCCGTCTGAA-3′ ) and has been documented functional in transformation of meningococci and gonococci [7] , [12] , [13] . Later , a revised 12-mer DUS ( 5′-AT-GCCGTCTGAA-3′ , here named AT-DUS ) was shown to elevate transformation further [14] , [15] . High level expression of the competence and minor pilin protein ComP has been shown to increase DUS-specific uptake , and a definite association between DUS and ComP was published recently [16] , [17] , [18] . A linear relationship between the number of DUS and the ability to competitively inhibit the uptake of radio-labelled DNA in N . gonorrhoeae has been documented , suggesting initial surface binding of DUS [19] . An additive effect of DUS has been documented also in transformation experiments in N . meningitidis , although no linear relationship between the number of DUS and transformation frequencies was evident [20] . Importantly , DNA binding and uptake assays do not fully correlate with the outcome of transformation assays , indicating that more than one level of DUS specificity exist [15] . Recently an influence of DUS location relative to homologous and recombinogenic regions of transforming DNA was demonstrated , suggesting that DUS may initiate DNA processing by a yet undefined way [20] . Two versions of USS have been described in Pasteurellaceae: version A ( 5′-AAGTGCGGT-3′ ) , named Hin-USS , is found in Haemophilus influenzae and Actinobacillus actinomycetemcomitans ( now named Aggregatibacter actinomycetemcomitans ) and USS version B ( 5′-ACAAGCGGT-3′ ) , named the Apl-USS subtype , is found in Actinobacillus pleuropneumoniae [21] . DUS-like repeat sequences have been described for N . subflava and N . sicca [22] and recently also in N . elongata [23] . Different variants of DUS are here termed DUS dialects alluding to their role as nucleotide ‘words’ in genetic ‘communication’ and in concordance with the previous use of the term ‘dialects’ in genetic contexts [24] . Even though DUS seems to have disseminated in the genus Neisseria , virtually nothing is known about DUS repeats in the family Neisseriaceae genera Kingella , Eikenella and Simonsiella . To fill this knowledge gap , the work presented here examines DUS specificity and dialects within the family Neisseriaceae [11] , [25] . The results reveal the presence of eight DUS dialects in different branches of the robust Neisseriaceae phylogenetic tree . In transformation assays , the DUS sequence divergence negatively influences inter-species transfer of DNA . A DUS core of only three nucleotides is present in all dialects and is strictly required for transformation . Assays with radiolabelled DNA show species specific relevance of DUS dialects for both binding and uptake of DNA . This work supports the idea that DUS specificity is a highly efficient barrier to interspecies transformation , that has great impact on the evolution of the Neisseriaceae . Neisseriaceae genomes were obtained from online databases ( i . e . , the Human Microbiome Project [26] and other initiatives ) and searched for overrepresented/highly-repeated sequences . Several very overrepresented 10-mers were identified in different genomes that displayed high degrees of similarity to the canonical DUS sequence first described in N . gonorrhoeae [7] , [12] , [13] . Eight distinct and abundant variants of DUS were identified and are shown in Figure 1 , five of which are potential DUS as they were not previously functionally confirmed . The DUS variants are called dialects in concordance with previous use of the term for describing variants of short DNA motifs that probably are strongly affected by some DNA template-dependent processing proteins [24] , and each DUS dialect was given a name according to the nomenclature scheme described in materials and methods . Every DUS dialect was found in exceptionally high numbers in their respective genomes and the exact occurrences are presented in Table S1 together with the number of degenerate DUS in which one nucleotide position were permitted to vary . AT-DUS was found in all available N . gonorrhoeae and N . meningitidis genomes ( 10-mer: n≈1900 ) . In addition , AT-DUS was found as the most overrepresented repeat in the genomes of N . lactamica ( n≈2200 ) , N . cinerea strain ATCC 14685 ( n = 943 ) and N . polysaccharea strain ATCC 43768 ( n = 2183 ) . AG-DUS was identified in Neisseria sp . oral taxon 014 ( n = 3236 ) , N . subflava strain NJ9703 ( n = 2871 ) , N . flavescens ( n = 1196 and n = 2767 in strains NRL30031 H210 and SK114 , respectively ) , N . mucosa strain C102 ( n = 2964 ) , N . bacilliformis strain ATCC BAA-1200 ( n = 4265 ) , N . weaveri strains ATCC 51223 and LMG 5135 ( n≈2850 ) , and N . elongata subsp . glycolytica strain ATCC 29315 ( n = 3273 ) . AG-mucDUS was the most prevalent repeat in N . mucosa strain ATCC25996 ( n = 1543 ) , N . sicca ( n = 3770 ) and N . macacae strain ATCC 33926 ( n = 3729 ) ( Table S1 ) . A previous study showed that in N . subflava strain ATCC 19243 , a 7 kb long sequence harbouring folP ( GeneBank AJ581792 . 1 ) contained 7 mucDUS and 1 AG-DUS [22] , whereas the genome of the N . subflava NJ9703 strain investigated here contained mainly the AG-DUS . The folP fragment is absent in the equivalent position in N . subflava NJ9703 and elsewhere in the genome . A BLAST search of all available Neisseriaceae genomes with the 7 kb fragment showed that the mucDUS positions around folP in N . subflava ATCC 19243 were present in N . sicca . The genome of N . wadsworthii 9715 displayed a distinct DUS dialect , wadDUS ( n = 2426 ) , which is identical to AT-DUS with a T insertion after position +3 ( Figure S1A ) . In the genome of K . oralis ATCC 51147 , yet another new dialect was discovered , the kingDUS ( n = 5918 ) . The occurrence of nearly six thousand kingDUS in a single small genome ( 2 . 4 Mb ) is the highest density of any DUS dialect detected so far . By allowing a single nucleotide divergence in the kingDUS , the number of kingDUS-similar sequences increased by 21% to a total 7153 hits for the K . oralis genome . The completion , closure and annotation of the K . oralis genome may eventually alter the absolute numbers of kingDUS present , but approximately 2 , 5% of this particular genome will still remain occupied by the kingDUS which is very high compared to the approximate 1% DUS occupancy in N . meningitidis and N . gonorrhoeae genomes . The genome of S . muelleri ATCC 29453 displayed the simultaneous presence of two DUS dialects . The kingDUS ( n = 2257 ) described above and a new dialect simDUS ( n = 2292 ) were in the S . muelleri genome detected in nearly equal numbers with a total count of 4549 . SimDUS differed from the kingDUS by an A/T transversion at position +3 ( Figure 1 ) . The genome sequences of K . kingae ATCC 23330 and K . denitrificans ATCC 33394 also revealed a new dialect , king3DUS , which differed from the kingDUS in an A/C transversion in position +9 ( Figure 1 ) and a G/A transition in position −1 ( Figure 1 ) . The king3DUS was present in 2787 and 3603 copies in the genomes of K . kingae ATCC 23330 and K . denitrificans ATCC 33394 , respectively . Finally , the most divergent dialect of DUS relative to the AT-DUS was identified in the genomes of E . corrodens ATCC 23834 ( n = 3269 ) and Neisseria shayeganii 871 ( n = 2245 ) , termed eikDUS . Notably , eikDUS was the only DUS with an A in position +4 ( Figure 1 ) . All the different dialects of DUS were conserved in positions +6 , +7 , +8 ( CTG ) as well as +10 ( A ) as demonstrated in Figure 1 . Based on the available genome sequences of genus Neisseria , no genome was devoid of any dialect of DUS . In the family Neisseriaceae , however , five genomes were found not to contain an abundant repeat that was an obvious DUS; these were the genomes of Laribacter hongkongensis HLHK9 , Lutiella nitroferrum 2002 , Pseudogulbenkiania sp . NH8B , Chromobacterium sp . C-61 and Chromobacterium violaceum ATCC12472 . Their respective over-represented 10-mers are listed in Table S1D . We noticed , however , that the DUS core sequence 5′-CTG-3′ was found as the reverse complement sequence 5′-CAG-3′ in the most over-represented 10-mer sequences from Laribacter hongkongensis and C . violaceum ( Table S1D ) . Until now , a core genome phylogenetic tree for members of the Neisseriaceae was made only for the human genus Neisseria species [27] , [28] , [29] , for all the available N . meningitidis genomes [25] and for collections of Neisseria strains [30] , [31] . Here , a phylogenetic tree encompassing 23 representative members of the family Neisseriaceae was generated based on their common core genome containing 474 coding sequences ( Figure 2 ) . The 16SrDNA phylogenetic tree [32] , [33] made for the Neisseriaceae differed from the core genome based tree ( Figure 2 ) . Notably , the DUS dialect distribution in the two trees differed considerably , and the core genome tree branches reflected the presence of different dialects in a congruent manner . Also a phylogenetic tree based on ComP , a recently reported DUS-specific binding protein [18] , displays high degree of congruence with different DUS dialects , although some deviations are apparent ( Figure S7 ) . The robust phylogeny finds that N . shayeganii 871 is closely related to E . corrodens ATCC 23834 , and S . muelleri ATCC 29453 is located among the three different Kingella species . Neisseria sp . oral taxon 014 is in the 16SrDNA tree wrongly placed close to the cluster containing N . lactamica . N . mucosa C102 is more closely related to N . subflava and N . flavescens than to N . mucosa ATCC 25996 . The latter strain is the one in which the mucDUS was first described [28] and is located on the same branch in the 16SrDNA phylogenetic tree as the type strain N . mucosa ATCC 19696 , based on the available partial sequence ( data not shown ) . This observation separated N . mucosa C102 from the N . mucosa ATCC 25996 reference strain . C . violaceum served as the outgroup in Figure 2 , based on its suitable genomic distance from the other Neisseriaceae members . The evolutionary history of DUS in Neisseriaceae may be traced and depicted as follows: The DUS-based transformation system evolved after the split from the shared common ancestor with C . violaceum . Neither could a ComP be indentified by BLAST searches of the genome of C . violaceum . Among the DUS-containing bacteria , the eikDUS-group separated first from the main branch , and is also the only group with an A in position +4 of the DUS . N . shayeganii strain 871 clusters with E . corrodens and may erroneously have been taxonomically assigned to the genus Neisseria . Thereafter , the kingDUS- and king3DUS-groups branched off from the canonical DUS-group and S . muelleri might have been in the process of separating itself from the Kingella group . N . wadsworthii separated from the AT-DUS-group by a change in DUS specificity evident from the insertion of a T in position +3 of AT-DUS . N . macacae , N . sicca and N . mucosa ATCC 25996 were separated from the AT-DUS- and AG-DUS-groups by the C/T transition in position +2 . The new DUS dialects identified here exhibit several divergent positions ( Figure 1 ) and we became interested in studying the discrete impact of the nucleotides that constitute a functional DUS . In a transversion mutation approach , the contribution of each individual nucleotide of the well-characterized AT-DUS was tested in quantitative transformation of N . meningitidis strain MC58 and the results are shown in Figure 3A . Also the effects of single transversion mutations in all twelve AT-DUS positions on DNA binding and uptake were measured and the results are summarized in Figure 3B . Any alteration of AT-DUS significantly reduced the transformation frequency ( paired t-test , seven experiments , p≤0 . 02 ) , although to a variable extent . The negative control lacking DUS does not transform at all . Our previous finding [14] was confirmed in that the two semi-conserved nucleotides in positions −2 and −1 at the 5′ end of the DUS , constituting the revised 12-mer AT-DUS , positively contribute to transformation efficacy , since both their respective transversions performed less than the complete AT-DUS . Furthermore , the transversions in individual positions of the 10-mer DUS were found to impair transformation performance . When the G in position +1 was transversed , the performance in transformation was reduced to 50% relative to the performance of the complete signal . Alterations in position +2 and +3 reduced the relative performance down to 20% and 28% , respectively . Alterations of position +4 ( 5% ) and +5 ( 2% ) had a more than one log reduction in relative transformation performance . The C , T and G at the 3′ half of the DUS ( positions +6 , +7 and +8 ) was shown to be particularly important for the DUS effect , since transformation was nearly abolished when the nucleotides in these positions were altered . A G/C transversion in position +8 gave rise to a total of only 2 CFU in seven experiments emphasizing the near complete loss of DUS-function . This functionally important 5′-CTG-3′ core is conserved in all dialects of DUS ( Figure 1 ) . The two adenines at the 3′ end of DUS ( position +9 and +10 ) display minor contributions to the overall effect of DUS , and mutants perform at around 50% of the full AT-DUS . The A in position +10 is also conserved in all DUS dialects but contributes less significantly to the functionality of DUS than the 5′-CTG-3′ core . The distance-dependent gradual influence of the bases around the short 5′-CTG-3′ core sequence may reflect the strength of molecular interactions between DUS and the electropositive stripe on the surface of ComP [18] that warrants further investigations . The effect of single transversion mutations on DNA binding and uptake in Figure 3B shows that DNA binding is high in N . meningitidis strain MC58 and that only DNA uptake is significantly affected . All the individual alterations of DUS bind better than the negative control lacking DUS . Relative DNA uptake was greatly reduced for the DNA without DUS and DUS with mutations in positions +4 to +8 , being approximately 2% of bound DNA for the mutations in the core positions +6 to +8 ( Figure 3B ) . In another strain , N . meningitidis 8013 , DNA binding is lower overall and is together with the uptake negatively affected by the alterations of DUS . Again it is the alterations 5′-CTG-3′ that most dramatically affects DNA uptake and/or binding in both strains tested ( Figure S4 ) . Potential commutation , defined as the interchange of DUS-linked genetic information , between different Neisseriaceae was first investigated by employing different DUS dialects in quantitative transformation experiments of N . meningitidis strain MC58 , and the results are shown in Figure 4A and Table 1 . Inversely , K . denitrificans , E . corrodens and N . elongata were tested for their respective DUS dependency by using PCR products of the rpsL gene conferring streptomycin resistance flanked by their own DUS or other DUS dialects . As shown in Table 1 , the transformation frequency was always highest for their autologous DUS variant . The AG-DUS differs from AT-DUS in just a single nucleotide in position -1 , and has 90% efficacy in N . meningitidis MC58 . AG-mucDUS differs from AT-DUS in a T/G transversion in position −1 and a C/T transition in position +2 . A 50% difference in transforming abilities of DUS and the 10-mer mucDUS in N . meningitidis was previously shown , although without statistical significance [22] . The 12-mer AG-mucDUS , which occurs 165 times in the genome of N . meningitidis strain MC58 , displayed here a 66% reduced transformation efficacy relative to that of AT-DUS ( Figure 4A and Table 1 ) . For the more drastic C/A transversion in the second position ( +2 ) in AT-DUS , a DUS-like sequence that occurred only once in the entire MC58 genome , the relative transformation was reduced by about 80% ( Figure 3A ) . The AT-mucDUS was found 19 times , but the AG-mucDUS was found 88 times in the MC58 genome indicating previous interspecies transfer from the AG-mucDUS group . These transformation assays in N . meningitidis MC58 showed inter assay variations ( Figure S3 ) but the Kendall's W test showed a very high concordance of gained orders ( Kendall's W = 0 . 9145 , χ2 = 44 . 8112 , df = 7 , p<0 . 0001 ) . AT-DUS containing DNA was completely unable to transform E . corrodens , the most phylogenetic distant DUS-containing species with the most divergent DUS dialect of the Neisseriaceae . E . corrodens was however readily transformed with its autogenic eikDUS documenting DUS-favoured transformation in the Eikenella genus for the first time . Also K . denitrificans transformed very poorly with AT-DUS and showed significant transformation with the autogenic king3DUS demonstrating DUS-favoured transformation in the Kingella genus for the first time . Low but significant transformation was achieved with AT-DUS in N . elongata harbouring the very similar AG-DUS . No biological transformation data was generated for Neisseria mucosa and N . sicca harbouring AG-mucDUS since the two strains tested , Neisseria mucosa type strain ATCC 19696 and N . sicca ATCC 29259 , were not transformable with PCR-generated DNA or isogenic genomic DNA , both conferring streptomycin resistance . Although the transformation efficiency is a measure of the final biological outcome , it is not useful for a quantitative measure of DNA binding to the cell and the DNA uptake . To assess the latter parameters in regard to the DUS dialects , the levels of binding and uptake of radiolabelled DNA was measured in different strains and species . In N . meningitidis MC58 , binding of DNA with different DUS dialects was only reduced 1 . 7-fold and 3 . 3-fold for AG-DUS and AG-mucDUS , respectively , but about 60-fold for AG-kingDUS and about 95-fold for AG-eikDUS compared to AT-DUS ( Figure 4B ) . DNA uptake was around 60% of bound DNA for AT-DUS , AG-DUS and AG-mucDUS but only around 10% of bound DNA for AG-kingDUS and AG-eikDUS . Another N . meningitidis strain , serogroup C strain 8013 , was also tested and showed reduced overall sequence specific and dialect dependent DNA binding but did not show sequence specific DNA uptake ( Figure S4 and Figure S5 ) . ComP in these two meningococcal strains are identical , indicating that more factors influence DUS-dependent DNA uptake in these strains . N . mucosa ATCC 25996 , N . elongata subsp . glycolytica ATCC 29315 , K . oralis ATCC 51147 and E . corrodens ATCC 23834 were also tested for DNA binding and uptake ( Figure 5 ) . N . mucosa showed a DNA binding of >2% of added DNA while all other species tested displayed values <0 . 3% . Only N . elongata showed a DUS dialect-dependent DNA binding with 0 . 3% for its own AG-DUS and 0 . 01% for AG-eikDUS ( Figure 5B ) . The binding and uptake performance of individual DUS-dialects in N . elongata mirrors those of N . meningitidis strain 8013 ( Figure S5 ) . The counts for K . oralis and E . corrodens were below 100 cpm and differences in performance of the different DNA templates were accordingly small . However , it is noteworthy that DNA binding of the autogenous DUS is significantly ( p≤0 . 05 ) higher than the negative control in K . oralis . Similarly , the DNA uptake of the autogenous DUS is significantly ( p≤0 . 05 ) higher than the negative control in E . corrodens . DNA uptake was around 60% of bound DNA for N . elongata , K . oralis and E . corrodens but only around 0 . 4% for N . mucosa . The results suggest that the investigated strains of these four bacterial species may not , or only to a small extent , carry out DUS-specific uptake of DNA ( Figure 5 ) contrasting the clear transformation data ( Table 1 ) . There are highly overrepresented sequences in the genomes of bacteria in general and the skewed occurrence of di- , tri- and tetra-mers has been particularly well documented [39] . These repeat distributions have proven valuable for classification [40] . The crossover hotspot instigator ( Chi ) sequence differs between bacterial species and new Chi sequences have been identified by a bioinformatics search for motifs [41] . Although uptake sequences and Chi sequences both are closely linked to homologous recombination , Chi and USS are distinctly different sequences in H . influenzae with no functional overlap [42] . As previously demonstrated [40] , the DUS sequence is the most abundant repeat in the genomes of N . meningitidis and N . gonorrhoeae with about 1900 occurrences in the 2 . 2–2 . 3 Mb genomes ( Table S1 ) . In the Neisseria sp . containing the AG-DUS dialect , the counts were generally higher , around 3000 . The most frequent DUS dialect was found for the kingDUS in K . oralis ATCC 51147 with nearly 6000 kingDUS within its 2 . 4 Mb genome sequence . Despite the high numbers of accurate DUS hits , the numbers of DUS with a single nucleotide divergence were considerably higher in all species ( Table S1 ) , revealing a potential for the activation of even more DUS positions . The difference in total DUS count and ratio of DUS to genome size may reflect an ultimate saturation-state of DUS or indicate that this state has not yet been reached . Also , if DUS specificity was for some reason lost , the DUS could be degenerating slowly but progressively , as observed for pseudogenes . Bacterial genomes with a high number of DUS had a relatively low number of DUS with a single divergence ( DUS+1mut ) and vice versa . For example , Neisseria bacilliformis ATCC BAA-1200 had 4265 DUS and 4914 DUS+1mut ( ratio 1∶1 . 15 ) while Neisseria cinerea ATCC 14685 had 943 DUS and 1372 DUS+1mut ( ratio 1∶1 . 45 ) ( Table S1 ) . These differences could reflect differences in DUS dependency , which is known to vary in different N . gonorrhoeae strains , and may therefore also vary between species and their respective dialects [15] . Future studies will seek to address the influence of sequence variation and regulation of ComP and its antagonist PilV [17] in this regard . DUS saturation of the chromosome may also be opposed by factors such as the degree of interference with coding ability for intragenic DUS . Notably , all DUS dialects , except king3DUS , harboured a stop codon ( UGA ) in one reading frame , which imposes an obvious limitation on the liberty of positioning DUS . By exploiting two different DUS dialects simultaneously some flexibility may be achieved in regard to which amino acids that are encoded by intragenic DUS . The genome sequence of S . muelleri harboured both the kingDUS and the simDUS , allowing for the variation of Q↔L and S↔C at the protein level when a DUS is found within a coding sequence . However , other explanations for the co-occurrence of two DUS-dialects in a single genome are high frequency of commutation between species ( simDUS and kingDUS differ in a single nucleotide only ) , or consecutive habitats in different mammalian hosts with access to variant DUS dialects . DUS specificity may be altered by mutations in comP or by the acquisition of alleles encoding a novel DUS dialect , or simply by the presence of two DUS-specific proteins with different affinities , which could have originated from a simple gene duplication . However , only a single copy of comP was identified in the S . muelleri genome . It is also noteworthy that 1294 occurrences of the simDUS and kingDUS in S . muelleri are arranged as overlapping pairs in a dyad symmetry structure ( Figure S6 ) , which may indicate a dimer-based mechanism of DUS recognition . No preference for a single reading frame or positioning inside or outside of coding sequences was obvious when using the preliminarily annotated genome sequence from S . muelleri . A similar symmetry is found also in Kingella species where king3DUS pairs with king2DUS . The latter showed positive influence on transformation ( data not shown ) but is not a commonly found DUS by itself ( Figure S6 ) . DUS have been found to locate in permissive regions of the core genomes of N . meningitidis , N . gonorrhoeae and N . lactamica , and intragenic DUS positions are common allowing them to be transcribed [11] . Intergenic regions , on the other hand , are particularly permissive and DUS sequences have been found to associate with transcriptional terminators by having frequently adopted an inverted paired organization , able to form stem-loop structures on ssDNA [7] , [14] , [43] . This inverted pair organization was found in high numbers in all genomes harbouring DUS dialects , suggestive of their association to transcriptional terminators ( Table S2 ) . In contrast to the simDUS and kingDUS arrangement in the S . muelleri genome , the individual DUS in an inverted pair DUS do not overlap . Another interesting observation is the occurrence of peregrine DUS , exemplified by the mucDUS in Table S1A . N . meningitidis and N . gonorrhoeae genomes had a very consistent mucDUS count of about 160 and 110 , respectively , while Neisseria sp . oral taxon 014 st . F0314 had the highest mucDUS count ( 467 ) , and the N . weaveri strains had the lowest counts ( 15 ) in the canonical DUS group . The N . meningitidis genome for example contained 8% mucDUS in addition to the canonical DUS , while the genomes of the mucDUS-containing group of bacteria harboured between 7% and 12% DUS . These numbers likely reflect recent commutation between these two groups sharing the same ecological niche . The exchange of highly selectable markers between N . meningitidis and the commensal Neisseriae is well established [44] , [45] , [46] , [47] , [48] , [49] . The amount of mucDUS relative to the canonical DUS was particularly low in the N . weaveri genomes ( 0 . 5% , Table S1A ) , which is interesting since N . weaveri is a canine commensal and only an opportunistic pathogen to humans [31] , [50] , [51] . In contrast to a previous report on the abundance of mucDUS in N . subflava strain ATCC 19243 , we found that AG-DUS is the most abundant dialect in N . subflava NJ9703 [22] . Possibly , N . subflava strain ATCC 19243 acquired the folP sequence fragment from N . sicca and therefore harbours the mucDUS in this region , or alternatively , N . subflava strain ATCC 19243 is more closely related to N . sicca than to N . subflava NJ9703 . We identified more canonical DUS in the N . elongata subsp . glycolytica strain ATCC 29315 , 3273 as opposed to 2142 than in the former study by Higashi et al . [23] , and more mucDUS , 174 as opposed to 117 . These differences could possibly be due to recent updates of the genome sequence files available . Comparison of the predominantly mucDUS-containing bacteria is difficult , as Higashi et al . did not specify which strains were analysed , both were , however , reported to contain >3400 copies of the mucDUS . In contrast to this observation , we identified only 1543 mucDUS in N . mucosa strain ATCC 25996 while N . mucosa C102 had only 155 mucDUS and 2964 canonical DUS . These discrepancies indicate that strain C102 may erroneously be assigned N . mucosa , also since both the core genome phylogeny ( Figure 2 ) and that of ComP demonstrated the close genetic relationship between this strain and N . subflava and N . flavescens . Neisseriaceae are highly recombinogenic yielding a polyphyletic family structure , and resolving the family into distinct species was achieved by including large amounts of sequence data in the analysis . Initially , such analyses were based on sequence divergence of a single gene ( 16SrDNA ) or on a small number of housekeeping genes as in multi locus sequence typing ( MLST ) [52] . The evolution of distinct DUS dialects in this phylogenetically compact family is a striking example of how preference for homologous DNA in highly transformable bacteria affects evolution . Differences between the dialects are expected to be mirrored in the amino acid sequence ( s ) of the recently confirmed DUS-specific binding protein ComP , and warrants further functional investigation . The congruence between DUS-dialect and phylogeny and the presence of ComP suggests that those dialects that remain to be confirmed functional DUS are true DUS . This is also further emphasized by the exceptional overrepresentation and conservation of each dialect in their respective genomes . The most plausible hypothesis explaining these observations is DUS-dependent bias in frequent transformation/recombination [11] . The differential influence of each nucleotide in AT-DUS on N . meningitidis transformation and DNA binding/uptake was tested by employing donor DNA harbouring altered DUS . Single nucleotides were altered to be the transverse ( purine↔pyrimidine ) and the least common nucleotide at that position in the N . meningitidis genome . A similar analysis , based on the uptake of radioactive labelled DNA , has previously been reported in H . influenzae [53] , [54] . Here the first steps of transformation were investigated by a DNA binding and uptake assay . The quantitative transformation method employed here measures the outcome of both uptake and recombination of DNA . This gradual analysis is important since it has been documented that DUS may influence multiple steps during transformation [15] . The most significant 5′-CTG-3′ core identified in transformation was conserved in all dialects of DUS ( Figure 1 ) . In contrast to the transformation experiments with N . meningitidis MC58 , DNA binding did not display differential binding of AT-DUS and mutated AT-DUS versions , but clearly showed that binding discriminated against the DUS-less negative control ( Figure 3B ) . This observation suggests that DNA binding in this strain is not very strict in terms of DUS specificity , and that DUS and single nucleotide mutated DUS can contribute to binding . Also the observation that the DUS dialects most similar to AT-DUS bound better than the more distant dialects emphasizes this point ( Figure 4B ) . It has been hypothesized that DUS specificity may function at more than one level during transformation [15] and one may speculate that initial binding by ComP [17] , [18] could display weak DUS specificity and that the influence of the core 5′-CTG-3′ first become influential during uptake or later during the transformation process . The DNA uptake data from N . meningitidis ( Figure 3B and Figure 4B ) are corroborating the transformation data in that AT-DUS outperforms the other dialects tested . The differences in relative uptake of the close and distant DUS dialects in N . meningitidis ( Figure 4B ) suggest that DNA uptake is not only a relative function of binding , but may be influenced by DUS specificity . The transformation performance of individual DUS-dialects from separate phylogenetic branches ( Figure 2 ) was tested in N . meningitdis , N . elongata , E . corrodens and K . denitirificans ( Table 1 ) . The degree of similarity between the DUS-dialect of the recipient species , AT-DUS in N . meningitidis , and that in the donor DNA , AG-DUS , AG-mucDUS , AG-kingDUS and AG-eikDUS , directly correlate with the level of transformation . The potential for high levels of commutation when DUS dialects are similar is reflected in reports describing interchange of DNA between pathogenic and commensal Neisseria in vivo [45] , [55] . This correlation is also evident in transformations of the AG-DUS species N . elongata , since AT-DUS outperforms AG-mucDUS . These observations suggest further that a nucleotide change in position −1 of the DUS is less influential than a change in position +2 in concordance with the results in N . meningitidis ( Figure 3A ) . Transformations in the genera Eikenella and Kingella show strict autologous DUS-dependency in transformation indicating that AT-DUS is too divergent to allow transformation . It is well established that general sequence divergence between recipient chromosome and transforming DNA is strongly affecting homologous recombination , the last step in transformation [3] . Based on these observations one may anticipate that the phylogenetic distance correlates with the potential for commutation since DUS dialect distribution is reflected in the orthology of the Neisseriaceae . Furthermore , no significant transformation of N . meningitidis was observed when transforming DNA carried USS , which is the DUS of the Pasteurellaceae . Since H . influenzae and N . meningitidis share the same habitats , and are likely to encounter each other's DNA in e . g . oropharyngeal biofilms , the establishment of a functional barrier to commutation between these species may be important for the preservation of genome integrity . Genetic exchange between N . meningitidis and H . influenzae is rare [56] while the frequent commutation within the Pasteurellaceae is well documented [21] , [57] . N . elongata subsp . glycolytica was previously shown to be transformable with a GT-mucDUS , but with an 8-fold reduced efficacy compared to a GT-DUS [23] . In our analysis , using a similar reporter construct , this factor was higher for the AG-mucDUS when compared to the AT-DUS ( 25-fold ) and when compared to the AG-DUS ( 35-fold ) ( Table 1 ) . These differences could relate to the employment of the non-ideal GT-DUS in the initial study . The DNA binding and uptake assays show that N . elongata subsp . glycolytica , like N . meningitidis , binds DNA in a DUS-specific manner with preference for the most similar DUS sequence corroborating the transformation results discussed above . The influence of AG-DUS and AT-DUS in transformation of N . mucosa or N . sicca could not be tested since these strains were incompetent for transformation . The molecular basis for this remains unexplored , but the DNA uptake data show that transformation deficiency can be linked to the reduced ability to take up DNA , suggestive of a malfunction in this initial step of transformation . It is curious that N . mucosa binds DNA exceptionally well in a DUS-independent manner and this observation warrants further investigation . The functionality of AG-mucDUS in transformation has also been verified in other laboratories ( N . Weyand , personal communication , [22] ) but mucDUS-dependent transformation of species in the mucDUS-group has not yet been demonstrated . Recent observations in our lab confirm that also wadDUS is a true DUS affecting transformation in N . wadsworthii ( unpublished data ) . The simDUS found in S . muelleri and kingDUS found in K . oralis are the only DUS that remains to be functionally verified , but the presence of comP genes in their respective genomes [18] , the high overrepresentation of each DUS dialect and intragenomic DUS conservation in addition to their high similarity to the other DUS strongly suggests that they are , or at least have been , genuine DUS . E . corrodens and K . denitrificans have previously been shown competent for transformation with homospecific DNA [58] , [59] . Here , we established that this specificity is DUS-dependent . The DNA binding and uptake results did not reflect the differences observed in transformation since DNA binding was relatively uniform irrespective of the DUS-dialect used . However , it must be noted that binding was very low in both K . oralis and E . corrodens although a weak but statistically significant preference for autologous DUS in DNA binding or uptake , respectively , is evident ( Figure 5 ) . Here , we expanded the number of bacteria that utilize a DUS-dependent mechanism for transformation of homologous DNA . Eight distinct dialects of DUS in the family Neisseriaceae were described , and the ability to overcome transformation barriers was assayed by both transformation and DNA binding and uptake for five of these . Furthermore , the evolution of DUS dialects corresponds with the evolution of the distinct core genomes of each phylogenetic clade . The DUS signal was analyzed by single nucleotide mutational analysis and an essential three nucleotide core sequence was found to be strictly required for transformation . This functional DUS core is conserved in all eight DUS dialects . The level of commutation was found to correlate with the phylogenetic distance and also to the similarity of the DUS sequences themselves . Future studies will explore the evolution of DUS dialects in regard of the recently confirmed association between DUS and ComP . Plasmids containing the dialects and transversion variants of DUS were based on the plasmid p0-DUS [14] . Oligonucleotides listed in Table S3 were used to amplify the pilG::ermC fragment from p0-DUS by PCR whereby the oligomer OH3 was always used as reverse primer . The PCR products were digested with XhoI and SacII and inserted into the multiple cloning site of the vector pBluescript II SK+ ( Stratagene , USA ) . E . coli strain ER2566 ( NEB , USA ) was used for cloning and the strain XL-1 Blue ( Stratagene , USA ) was used for large scale purification of plasmids due to higher yields . Plasmids were purified using the QIAGEN Plasmid Plus Midi Kit ( Qiagen , Germany ) . Plasmids are listed in Table S4 . The DNA was diluted to a concentration of 100 ng/µl in 10 mM Tris , pH 8 , and stored at −20°C until used . Bacteria used in this study are listed in Table S4 . Escherichia coli strains were grown on LB medium . E . coli XL-1 blue was used for the quantitative production of the pDV plasmids . N . meningitidis and N . elongata were grown on blood agar plates , on GC medium plates or in liquid GC medium supplemented with IsoVitaleX . K . denitrificans and E . corrodens were grown on blood or chocolate agar plates or , when in liquid , in brain heart infusion broth . Antibiotics were added to the media when appropriate . Quantitative transformation was performed as previously described [14] using plasmid or genomic DNA carrying an antibiotic resistance marker . Briefly , for N . meningitidis , cells grown over night at 37°C were suspended in 5% CO2 saturated GC medium containing IsoVitaleX and 7 mM MgCl2 . 5 µl of DNA ( 100 ng/µl ) were provided in 15 ml tubes , 500 µl cell suspension was added , shortly mixed by vortexing and incubated at 37°C for 30 min without agitation . Each sample was diluted by adding 4 . 5 ml GC medium and incubated for 4 . 5 h at 37°C on a tumbler ( 60 rpm ) . The cultures were then mixed and serial dilutions prepared in GC medium . Of each undiluted sample , 50 µl aliquots were spread on blood agar plates containing 8 µg ml−1 erythromycin or 50 µg ml−1 streptomycin and 50 µl of the 10−7 dilution were spread on blood agar plates without antibiotics . At least 2 agar plates were inoculated from each sample and experiments were repeated at least three times . Colonies were counted following over night incubation in a 5% CO2 atmosphere at 37°C . Individual transformation frequencies were calculated as the number of antibiotic-resistant colony forming units ( CFU ) per total CFU . The absolute transformation frequencies varied between experiments , possibly due to the difficulty in reproducing bacterial suspensions with identical fractions of competent bacteria . This problem was also reported earlier [22] , and was resolved here by the use of relative values based on an internal standard . A Kendall's W test showed a very high concordance of gained orders ( Kendall's W = 0 . 9346 , χ2 = 85 . 0506 , df = 13 , p<0 . 0001 ) . The absolute transformation frequencies are plotted in Figure S2 . In order to transform N . elongata , K . denitrificans and E . corrodens , small alterations of the protocol were required . The N . elongata subsp . glycolytica strain employed here was the type strain originally isolated by Henriksen et al . [68] . Agglutinating P+ colonies of N . elongata were pre-selected since previous experiments had documented a positive correlation between agglutination and transformation [69] . The type strain of K . denitrificans and E . corrodens strain 31745 were used for quantitative transformation . The E . corrodens strain 31745 is a reference strain that was previously shown to be transformable [58] . To generate donor DNA for the quantitative transformation experiments , spontaneous streptomycin-resistant mutants of N . elongata , K . denitrificans and E . corrodens were isolated and the genomic DNA from these strains used as a template for a PCR of the rpsL gene using primers listed in Table S5 . The rpsL genes were sequenced , confirming the Lys to Arg mutation at amino acid position 43 conferring streptomycin resistance [70] . rpsL PCR products were adjusted to 50 ng/µl and 10 µl thereof was used in transformation of 500 µl cell suspension . K . denitrificans was transformed in Brain heart infusion medium supplemented with 7 mM MgCl2 and the second incubation time was extended to 5 hours . For radiolabeling the pDV plasmids ( Table S4 ) were linearized by digestion with ScaI , purified using QIAquick columns ( Qiagen ) and treated with exonuclease III ( Fermentas ) . After heat inactivation 10 µg of the partially single stranded DNA were incubated with 3 µM dNTPs , 20 µCi [α-32P]dCTP and 10 units Klenow fragment ( 3′→5′ exo– ) ( NEB ) for 90 min at 37°C . The fill-in reaction was finished after increasing the dNTP concentration to 30 µM and additional incubation for 30 min . The products were purified as before and showed specific activities of 4×105 to 2×106 CPM µg−1 . The Neisseriaceae were grown in liquid medium to an optical density at 660 nm of approximately 1 . MgCl2 was added to 7 mM and one ml aliquots were incubated with DNA with about 5×106 CPM activity and rotated at 37°C for 45 min . The samples were then split into two 500 µl samples and 12 . 5 units Benzonase ( Merck ) was added to one of these . After additional incubation for 15 min the cells were washed three times by centrifugation for 3 min at 5000× g and resuspension in liquid medium including 7 mM MgCl2 . The final pellets were resuspendet in 3 ml scintillation fluid ( Ultima Gold MV , PerkinElmer ) and measured twice for 3 min in a Tri-Carb 2900TR ( PerkinElmer ) using an energy window LL-UL = 50–1700 . DNA binding and uptake are reported as the percentage of DNA added and percentage of cell-bound DNA , respectively , and the results are presented as the means of for 3 to 5 replicates .	Through computational and biological methods , this work analyzes the function and evolution of short DNA sequences called DNA uptake sequences ( DUS ) that regulate genetic transformation of bacteria in the family Neisseriaceae . Previous studies show that DUS affects transformation favourably . Here , for the first time , we document the existence of eight distinct DUS dialects that display differences in their respective nucleotide sequence that limits genetic “communication” between species . This suggests that each DUS dialect represents a barrier to horizontal gene transfer of heterologous DNA contributing to genetic isolation . Single nucleotide analysis of DUS was used to identify a three nucleotide core sequence common to all DUS dialects and essential for transformation . The discovery of multiple DUS dialects emphasizes that homologous recombination , allelic reassortment , and bacterial “sex” play an important role in the evolutionary past of Neisseriaceae species .	[ "Abstract", "Introduction", "Results", "Discussion", "Materials", "and", "Methods" ]	[ "bacteriology", "organismal", "evolution", "genome", "evolution", "population", "genetics", "microbiology", "gene", "function", "prokaryotic", "models", "model", "organisms", "microbial", "evolution", "divergent", "evolution", "bacterial", "pathogens", "sequence", "analysis...	2013	Dialects of the DNA Uptake Sequence in Neisseriaceae
The over-replicating wMelPop strain of the endosymbiont Wolbachia pipientis has recently been shown to be capable of inducing immune upregulation and inhibition of pathogen transmission in Aedes aegypti mosquitoes . In order to examine whether comparable effects would be seen in the malaria vector Anopheles gambiae , transient somatic infections of wMelPop were created by intrathoracic inoculation . Upregulation of six selected immune genes was observed compared to controls , at least two of which ( LRIM1 and TEP1 ) influence the development of malaria parasites . A stably infected An . gambiae cell line also showed increased expression of malaria-related immune genes . Highly significant reductions in Plasmodium infection intensity were observed in the wMelPop-infected cohort , and using gene knockdown , evidence for the role of TEP1 in this phenotype was obtained . Comparing the levels of upregulation in somatic and stably inherited wMelPop infections in Ae . aegypti revealed that levels of upregulation were lower in the somatic infections than in the stably transinfected line; inhibition of development of Brugia filarial nematodes was nevertheless observed in the somatic wMelPop infected females . Thus we consider that the effects observed in An . gambiae are also likely to be more pronounced if stably inherited wMelPop transinfections can be created , and that somatic infections of Wolbachia provide a useful model for examining effects on pathogen development or dissemination . The data are discussed with respect to the comparative effects on malaria vectorial capacity of life shortening and direct inhibition of Plasmodium development that can be produced by Wolbachia . Wolbachia pipientis is an intracellular maternally inherited bacterial symbiont of invertebrates that is very common in insects , including a number of mosquito species [1] , [2] . It can manipulate host reproduction in several ways , including cytoplasmic incompatibility ( CI ) , whereby certain crosses are rendered effectively sterile . Females that are uninfected produce infertile eggs when they mate with males that carry Wolbachia , while there is a ‘rescue’ effect in Wolbachia-infected embryos such that infected females can reproduce successfully with any males . Therefore uninfected females suffer a frequency-dependent reproductive disadvantage . Wolbachia is able to rapidly invade populations using this powerful mechanism [3]–[5] . A strain of Wolbachia called wMelPop has been identified that over-replicates in somatic tissues and roughly halves the lifespan of laboratory Drosophila melanogaster [6] . A transinfection of wMelPop from Drosophila into the mosquito Aedes aegypti also leads to a similarly shortened lifespan in the lab , as well as inducing strong CI , which has made it a very promising candidate for the development of new strategies for controlling mosquito-borne diseases [7] . All mosquito-borne pathogens require an extrinsic incubation period before they can be transmitted that is relatively long ( ∼9 days for malaria ) compared to mean mosquito lifespan in the field; therefore , a reduction in the number of old individuals in the population will reduce disease transmission [8]–[11] . We recently found that the presence of wMelPop also produces a major upregulation of a large number of immune genes in Ae . aegypti and inhibits the development of filarial nematode worm parasites [12] . We hypothesized that the two effects are functionally related – higher levels of immune effectors in wMelPop-infected mosquitoes render them better able to kill parasites [12] . Homologs of some of the Ae . aegypti genes that are upregulated in the presence of wMelPop have been previously shown to have the ability to regulate development of Plasmodium parasites in Anopheles , for example a transgene encoding cecropin-A/a synthetic cecropin-B of Hyalophora cecropia; the NF-κB-like transcription factor Rel2 controlling the Imd pathway; and TEP ( Thioester containing ) opsonization proteins [13]–[20] . It has recently been shown that the wMelPop-infected Ae . aegypti line has impaired ability to transmit an avian malaria , Plasmodium gallinaceum [21] . It is possible that these effects of wMelPop could be particular to the Ae . aegypti transinfection; however , if comparable upregulation of orthologous immune genes , and inhibition of Plasmodium development are also seen in the important Anopheles vectors of human malaria , it may provide a stimulus to the development of new Wolbachia-based malaria control strategies . To address this question we used Anopheles gambiae , the most important vector of malaria in Africa , which like Ae . aegypti is not naturally infected with Wolbachia . The creation of stably inherited lines of An . gambiae is likely to require a long period of microinjection and selection , as had to be performed for Ae . aegypti [7] . However , in advance of the successful creation of an An . gambiae stable transinfection , the effects of the presence of wMelPop on immunity and malaria transmission can be tested using an established wMelPop-infected An . gambiae cell line [22] and the ability to create somatic lifetime infections of Wolbachia in adult female mosquitoes by intrathoracic inoculation [23] , [24] . The wMelPop strain forms disseminated somatic infections in its natural Drosophila host [6] , in common with some but not all Wolbachia strains [25] . Given that a ) Plasmodium parasites will travel solely through somatic tissues on their journey to the salivary glands , and b ) that many of the known antimalarial immune effectors are humoral/systemic , we consider that the creation of somatic infections of Wolbachia via adult inoculation represents a useful model for stably inherited germline-associated infections . To examine this hypothesis further , we also created somatic wMelPop infections in Ae . aegypti , in order to compare the magnitude of the effects on mosquito immunity and filarial nematode parasite development with those observed in the stably wMelPop-transinfected line . Given that a stable wMelPop infection of An . gambiae does not yet exist , it was necessary to create transient somatic infections by intrathoracic innoculation with purified Wolbachia . RNA from these transinfected females was then tested for expression levels of six immune genes , and upregulation of all these genes was observed compared to buffer injected and E . coli - injected controls ( Figure 1 ) . Of these genes , LRIM1 and TEP1 ( whose products have been shown to interact in the opsonisation response ) have previously been shown to have an important inhibitory or antagonistic effect on Plasmodium development [18]–[20] . Importantly , injected mosquitoes were left for eight days before Plasmodium challenge or qRT-PCR , and therefore the pulse of immune gene upregulation caused by the injury itself or by the E . coli challenge would be expected to have already passed [15] . The wMelPop infected cell line MOS55 [22] showed upregulation of all six selected immune genes compared to an uninfected cell line created by tetracycline curing of infected MOS55 ( Figure 2 ) . These data add confidence to the hypothesis that it is the presence of wMelPop itself that is inducing immune gene upregulation , and by extension Plasmodium inhibition , and that these effects are not artefacts of the intrathoracic injection process . The degree of upregulation was different for some genes in the cell line than observed for the somatic in vivo transinfection . However these differences would be expected given that many immune genes are primarily expressed in particular cell types/organs in adult mosquitoes , such as the fat body cells or in the case of TEP1 , the haemocytes [18] , and the cellular composition of this larval-derived cell line is unknown . Three Plasmodium berghei challenge experiments were conducted on transiently Wolbachia-infected A . gambiae females compared to buffer injected , uninjected , and in one case E . coli-injected controls ( Figure 3a–c ) . In all three experiments highly significant reductions in intensity of oocyst infection in the wMelPop transinfected females were observed compared to the other treatments , while there were no significant differences between any of the control treatments within each experiment . Mean P . berghei intensities were reduced in the wMelPop-infected mosquitoes by between 75% and 84% compared to the corresponding buffer injected control groups . A further experiment confirmed the lack of any significant differences in intensity between the E . coli-injected , buffer injected and uninjected controls ( data not shown ) . In order to obtain evidence for a causal link between the immune upregulation and the Plasmodium inhibition phenotypes , TEP1 knockdown was undertaken by injection of dsRNA at the same time as Wolbachia injection . Significantly higher oocyst numbers were observed compared to the control where dsLacZ was injected at the same time as Wolbachia ( Figure 3d ) . This experiment provides evidence for a significant contribution of Wolbachia-induced TEP1 upregulation to the Plasmodium inhibition phenotype . We assessed the utility of the transient wMelPop somatic infection model by comparing the effects on host immunity and pathogen development with those observed in stable inherited infections of wMelPop . To do this we utilized a filarial nematode-susceptible line of another mosquito species , Ae . aegypti , in which we have previously carried out Brugia pahangi challenges on a stable wMelPop-transinfected line [7] , [12] . We created somatic wMelPop infections using exactly the same methodology as carried out for An . gambiae , and after eight days challenged them with B . pahangi or carried out qRT-PCR . The somatic Wolbachia infection also induced upregulation of selected immune genes ( PGRPS1 , CECD , CLIPB37 , CTL ) ( Figure 4a ) . The scale of upregulation was considerably lower than observed in the comparable Ae . aegypti stable transinfection as previously reported [12] . Likewise , challenge of the somatically wMelPop infected females with B . pahangi did produce a significant reduction in the numbers developing to the L3 ( infectious ) stage compared to the controls ( Figure 4b ) , as was previously observed in the stable inherited wMelPop infected line , which showed >50% reduction in mean numbers of L3 compared to the Wolbachia-uninfected control at the same microfilarial challenge density [12] . Using quantitative PCR comparing three groups of two mosquitoes with the single copy genes ftsZ ( Wolbachia ) and Actin5C ( Ae . aegypti ) for normalization , we estimated that there were approximately 176±70 times more wMelPop cells in the stably infected line compared to the somatic infections . This may explain this reduced effect on gene upregulation . Therefore we conclude that intrathoracic inoculation can be a valuable way to test the effects of Wolbachia on host immunity and pathogen transmission . Although extrapolations to different mosquito species and parasites must be made with care , it does seem likely that the effects observed for somatic Wolbachia infections using the methodology reported here are likely to be smaller than for a stable inherited infection , and thus that the estimations made may be conservative . An experiment to test whether the immune upregulation observed in wMelPop-infected mosquitoes affects the density of the Wolbachia itself was conducted using the stable inherited infection of wMelPop in an Ae . aegypti Refm background [7] , [12] . Wolbachia ftsZ gene expression ( used as a proxy for Wolbachia density ) was found to be higher in dsRel2-injected than in dsLacZ-injected mosquitoes at both day six and day ten post-injection ( Figure 4c ) . These data suggest that the immune effectors controlled by the Imd ( Rel2-controlled ) pathway can influence Wolbachia densities . The very high rate of maternal transmission observed in wMelPop-infected Ae . aegypti [7] , despite chronic immune upregulation , means that the biological significance of this density difference is unknown , although potentially it could act to limit wMelPop pathogenicity to some degree . More comprehensive experiments addressing this question will make use of transgenic immune knockdown lines infected with wMelPop , which are currently being produced , and are expected to enable the effects of stronger and more long lasting immune pathway knockdown to be investigated . The data reported strongly support the hypothesis that wMelPop can inhibit the development of Plasmodium in Anopheles malaria vector mosquitoes . The An . gambiae/P . berghei combination , although not one that occurs in nature , does represent a tractable and well studied model for which considerable information is already available about Plasmodium killing mechanisms; however we recognize the challenge experiments will ultimately need to be repeated with the far less tractable human parasite P . falciparum once a stably inherited Wolbachia transinfected line of An . gambiae has been created . The densities of P . berghei used in laboratory challenges such as these can be high compared to those of P . falciparum that would occur in nature , although the mean intensities recorded in these studies lie within the range recorded for P . falciparum in the field . The significant reductions in intensity we recorded in laboratory experiments are considered likely to translate to significant reductions in oocyst prevalence/transmission in a real-life setting . The knockdown experiment provided evidence for a major role of TEP1 , and by extension LRIM1 whose products interact as part of the same opsonization pathway [20] , in the inhibition of P . berghei development . This is the first time a direct link between the Wolbachia pathogen inhibition and immune upregulation phenotypes has been made . A more detailed and exhaustive investigation of the relative contributions of different components of the Anopheles immune system to Plasmodium killing can be made once stable inherited Wolbachia infections have been established . Taken together with the recent report of reduction in P . gallinaceum development in wMelPop-infected Ae . aegypti [21] , the data increase the desirability of creating stably inherited wMelPop transinfections in important malaria vectors . The potential combination of lifespan shortening and direct inhibition of Plasmodium development in the mosquito would represent a very attractive control strategy , since both of these phenotypes are critical components of malaria vectorial capacity . A simple model exploring relative contributions of these two parameters to vectorial capacity is shown in Figure 5 . Though lifespan reduction and Plasmodium inhibition can each substantially reduce the vectorial capacity of a mosquito population , together they act synergistically to reduce transmission . Depending on the scale of lifespan reduction that would be observed under field conditions , which is as yet unknown , the Plasmodium inhibition effect could dramatically increase the efficacy of the wMelPop infection in reducing malaria transmission . Other Wolbachia strains might also show malaria inhibition effects , particularly if they reach high somatic densities and/or induce large-scale immune stimulation . Here we show that the use of transient somatic infections of Wolbachia by adult female inoculation followed by pathogen challenge is a valuable means to test likely effects on immunity and transmission . This is significant as it allows comparison and selection of strains for the most desirable properties prior to the lengthy , and technically very challenging , process of creating stably inherited Anopheles transinfections . If other Wolbachia strains can be identified which also inhibit Plasmodium transmission , they would represent an attractive alternative to wMelPop if they do not shorten lifespan to the same extent , since they are therefore likely to have much lower fitness costs . Only the wMelPop strain has to date been found to produce a strong life-shortening phenotype . Laboratory estimates suggest that transinfection of wMelPop in Aedes aegypti can reduce fitness by around 50% [7] . This would appear to make it difficult for this strain of Wolbachia to spread by means of CI through natural populations [26] , particularly where populations are fragmented . However , fitness estimates made in relatively benign laboratory conditions , where a comparatively large fraction of the population become old , can overestimate the relative costs of infection . In the field most mosquitoes die early and few live long enough to experience higher Wolbachia-induced mortality ( although those that do are significant to disease control , if they would otherwise have lived long enough to transmit the infection ) . As shown in Figure 5 reductions in longevity and Plasmodium inhibition together determine vectorial capacity and it will also be important to understand the joint effects of the two phenotypes on mosquito fitness in the field . Detailed knowledge of the demographics of the target species is also important [27] . Selective pressures acting on the host would likely modulate the life-shortening phenotype over time , but this may not occur rapidly enough to prevent a sustained period of disease control . Wolbachia is now known to inhibit the dissemination or development of a variety of insect pathogens and insect-borne pathogens – various Drosophila pathogenic viruses , dengue and chikungunya viruses of humans , and filarial nematode parasites in addition to Plasmodium [12] , [21] , [28]–[31] . Some of these pathogen-inhibition phenotypes have been reported in Drosophila species that naturally harbour Wolbachia , in other words they are not restricted to species such as Ae . aegypti or An . gambiae in which Wolbachia forms a novel transinfection . On a broader level these Wolbachia cases can be added to various other examples where bacterial symbionts have been shown to provide protective effects against one or more pathogens [32] , [33] , although the mechanisms involved are likely to be diverse . Parallels can also be drawn with the effects of entomopathogenic fungi , which can both reduce Anopheles lifespan and directly inhibit Plasmodium development [34]–[36] . Pathogen inhibition represents a new and increasingly significant component of our understanding of the effects of Wolbachia in insects , and provides excellent prospects for the development of novel malaria control strategies . All procedures involving animals were approved by the ethical review committee of Imperial College and by the United Kingdom Government ( Home Office ) , and were performed in accordance with United Kingdom Government ( Home Office ) and EC regulations . Wolbachia wMelPop was purified from the infected An . gambiae cell line MOS55 [22] , [37] as previously described [23] , [24] . This protocol has previously been shown to allow Wolbachia replication in the recipient An . gambiae [24] . Cells obtained from one 75 CM2 flask were re-suspended in 100 µL of Schneider medium without antibiotics ( optical density , OD = 0 . 09 ) . 69 nL of this Wolbachia suspension ( or 69 nL Schneider for the controls ) were microinjected into the thorax of young An . gambiae females of the G3 strain or Ae . aegypti females of the Refm strain [38] using an Nanoject microinjector ( Drummond ) . The mosquitoes were supplied with 10% sucrose ad libitum and left to recover for at least eight days prior to qRT-PCR or challenge experiments . A similar OD of 0 . 1 for E . coli was used to inject another set of controls . Gene expression levels were monitored using qRT-PCR . Total RNA was extracted with Trizol reagent from groups of ten An . gambiae or Ae . aegypti females maintained at 26°C and 70% relative humidity , and cDNAs were synthesised from 1 µg of total RNA using SuperScript II enzyme ( Invitrogen ) . qRT-PCR was performed on a 1 to 20 dilution of the cDNAs using dsDNA dye SYBR Green I . Reactions were run on a DNA Engine thermocycler ( MJ Research ) with Chromo4 real-time PCR detection system ( Bio-Rad ) using the following cycling conditions: 95°C for 15 minutes , then 45 cycles of 95°C for 10s , 59°C for 10s , 72°C for 20s , with fluorescence acquisition at the end of each cycle , then a melting curve analysis after the final one . The cycle threshold ( Ct ) values were determined and background fluorescence was subtracted . Gene expression levels of target genes were calculated , relative to the internal reference gene Actin5C or RS17 for Ae . aegypti and RS7R for An . gambiae . Primers were designed using Vectorbase ( www . vectorbase . org ) mosquito gene sequences/orthology criteria , and the wMel genome sequence [39] , since wMel and wMelPop are closely related [40] . Primer pairs used to detect target gene transcripts are listed in Table 1 . The density of Wolbachia in somatic and stable infections of Ae . aegypti was estimated using both qPCR and qRT-PCR . DNA was extracted using the Livak method and qRT-PCR or qPCR equipment and protocols were the same as those described above . The single copy genes ftsZ ( Wolbachia ) and Actin5C and S7 ( Ae . aegypti ) were used to estimate relative numbers of Wolbachia normalized against the mosquito genome . General parasite maintenance was carried out as previously described [41] . P . berghei ANKA 2 . 34 parasites were maintained in 4–10-week-old female Theiler's Original ( TO ) mice by serial mechanical passage ( up to a maximum of eight passages ) . Hyper-reticulocytosis was induced 2–3 days before infection by treating mice with 200µL i . p . phenylhydrazinium chloride ( 6mg/ml in PBS; ProLabo UK ) . Mice were infected by intraperitoneal ( i . p . ) injection and infections were monitored on Giemsa-stained tail blood smears . In four independent experiments , individual 4–10 week old Theiler's Original ( TO ) mice were treated with 200µL i . p . phenylhydraziuium chloride ( PH; 6mg/ml in PBS; ProLabo UK ) to induce hyper-reticulocytosis . Three days later mice were injected by intraperitoneal ( i . p . ) injection with 106 parasites of P . berghei ANKA 2 . 34 as described previously [41] . Three days post mouse infection , batches of 100 starved Anopheles gambiae strain G3 females , eight days post injection with Wolbachia , buffer , E . coli or uninjected controls , were allowed to feed on the infected mice . 24h after feeding , mosquitoes were briefly anesthetized with CO2 , and unfeds removed . Mosquitoes were then maintained on fructose [8% ( w/v ) fructose , 0 . 05% ( w/v ) p-aminobenzoic acid] at 19–22°C and 50–80% relative humidity . At day 10 post-feeding , mosquito midguts were dissected , and oocyst numbers ( intensity ) and prevalence recorded . The Kruskal-Wallis test was used to compare oocyst counts ( intensity of infection ) and Fisher's exact test for prevalence ( percentage of mosquitoes containing at least one oocyst ) . T7-tailed primers ( see Table 1 ) were used to amplify fragments of the TEP1 and REL2 gene from female cDNA template or the LacZ gene from E . coli total DNA . dsRNA was synthesized using the T7 Megascript kit ( Ambion ) and adjusted to a concentration of 3 or 4 µg/µl in RNAse free water for dsREL2 and dsTEP1 respectively . For REL2 KD 69nl of dsRNA were injected per female mosquito , For TEP1-wolbachia KD 69 nl of a mix of 2 parts dsRNA to 1 part of purified wMelPop in Schneider's medium ( OD 0 . 3 ) were injected into the thorax of CO2 anesthetized female An . gambiae mosquitoes ( total ∼200 per group ) . Five days after injection ( in order to still fall within the gene knockdown period ) , mosquitoes were fed on a Plasmodium infected mouse . Ae . aegypti mosquitoes of the filaria-susceptible Refm strain were fed on sheep blood containing 23 B . pahangi microfilaria per µL eight days post Wolbachia innoculation , plus buffer-injected controls of the same age; any females that did not feed properly were removed . Dissections were carried out 10 days after the infective blood meal under a dissecting stereomicroscope . Kruskal-Wallis tests were used to compare counts of B . pahangi L3 ( infective stage larvae ) .	Malaria is one of the world's most devastating diseases , particularly in Africa , and new control strategies are desperately needed . Here we show that the presence of Wolbachia bacteria inhibits the development of a malaria parasite in the most important Anopheles mosquito species of Africa . In addition we show that the presence of Wolbachia results in the switching on of immune genes that are known to affect development of many species of malaria parasite . When added to the lifespan-shortening effects of this particular strain of Wolbachia , and the general ability of Wolbachia to spread through insect populations , our study provides a stimulus for the development of Wolbachia-based malaria control methods . It also provides new insights into the wide range of effects of Wolbachia in insects .	[ "Abstract", "Introduction", "Results", "Discussion", "Materials", "and", "Methods" ]	[ "genetics", "and", "genomics/disease", "models", "genetics", "and", "genomics/genetics", "of", "the", "immune", "system" ]	2010	Wolbachia Stimulates Immune Gene Expression and Inhibits Plasmodium Development in Anopheles gambiae
Is visual cortex made up of general-purpose information processing machinery , or does it consist of a collection of specialized modules ? If prior knowledge , acquired from learning a set of objects is only transferable to new objects that share properties with the old , then the recognition system’s optimal organization must be one containing specialized modules for different object classes . Our analysis starts from a premise we call the invariance hypothesis: that the computational goal of the ventral stream is to compute an invariant-to-transformations and discriminative signature for recognition . The key condition enabling approximate transfer of invariance without sacrificing discriminability turns out to be that the learned and novel objects transform similarly . This implies that the optimal recognition system must contain subsystems trained only with data from similarly-transforming objects and suggests a novel interpretation of domain-specific regions like the fusiform face area ( FFA ) . Furthermore , we can define an index of transformation-compatibility , computable from videos , that can be combined with information about the statistics of natural vision to yield predictions for which object categories ought to have domain-specific regions in agreement with the available data . The result is a unifying account linking the large literature on view-based recognition with the wealth of experimental evidence concerning domain-specific regions . The discovery of category-selective patches in the ventral stream—e . g . , the fusiform face area ( FFA ) —is one of the most robust experimental findings in visual neuroscience [1–6] . It has also generated significant controversy . From a computational perspective , much of the debate hinges on the question of whether the algorithm implemented by the ventral stream requires subsystems or modules dedicated to the processing of a single class of stimuli [7 , 8] . The alternative account holds that visual representations are distributed over many regions [9 , 10] , and the clustering of category selectivity is not , in itself , functional . Instead , it arises from the interaction of biological constraints like anatomically fixed inter-region connectivity and competitive plasticity mechanisms [11 , 12] or the center-periphery organization of visual cortex [13–17] . The interaction of three factors is thought to give rise to properties of the ventral visual pathway: ( 1 ) The computational task; ( 2 ) constraints of anatomy and physiology; and ( 3 ) the statistics of the visual environment [18–22] . Differing presuppositions concerning their relative weighting lead to quite different models of the origin of category-selective regions . If the main driver is thought to be the visual environment ( factor 3 ) , then perceptual expertise-based accounts of category selective regions are attractive [23–25] . Alternatively , mechanistic models show how constraints of the neural “hardware” ( factor 2 ) could explain category selectivity [12 , 26 , 27] . Contrasting with both of these , the perspective of the present paper is one in which computational factors are the main reason for the clustering of category-selective neurons . The lion’s share of computational modeling in this area has been based on factors 2 and 3 . These models seek to explain category selective regions as the inevitable outcome of the interaction between functional processes; typically competitive plasticity , wiring constraints , e . g . , local connectivity , and assumptions about the system’s inputs [12 , 26–28] . Mechanistic models of category selectivity may even be able to account for the neuropsychology [29 , 30] and behavioral [31 , 32] results long believed to support modularity . Another line of evidence seems to explain away the category selective regions . The large-scale topography of object representation is reproducible across subjects [33] . For instance , the scene-selective parahippocampal place area ( PPA ) is consistently medial to the FFA . To explain this remarkable reproducibility , it has been proposed that the center-periphery organization of early visual areas extends to the later object-selective regions of the ventral stream [13–15 , 17] . In particular , the FFA and other face-selective region are associated with an extension of the central representation , and PPA with the peripheral representation . Consistent with these findings , it has also been argued that real-world size is the organizing principle [16] . Larger objects , e . g . , furniture , evoke more medial activation while smaller objects , e . g . , a coffee mug , elicit more lateral activity . Could category selective regions be explained as a consequence of the topography of visual cortex ? Both the eccentricity [15] and real-world size [16] hypotheses correctly predict that houses and faces will be represented at opposite ends of the medial-lateral organizing axis . Since eccentricity of presentation is linked with acuity demands , the differing eccentricity profiles across object categories may be able to explain the clustering . However , such accounts offer no way of interpreting macaque results indicating multi-stage processing hierarchies [17 , 34] . If clustering was a secondary effect driven by acuity demands , then it would be difficult to explain why , for instance , the macaque face-processing system consists of a hierarchy of patches that are preferentially connected with one another [35] . In macaques , there are 6 discrete face-selective regions in the ventral visual pathway , one posterior lateral face patch ( PL ) , two middle face patches ( lateral- ML and fundus- MF ) , and three anterior face patches , the anterior fundus ( AF ) , anterior lateral ( AL ) , and anterior medial ( AM ) patches [2 , 36] . At least some of these patches are organized into a feedforward hierarchy . Visual stimulation evokes a change in the local field potential ∼ 20 ms earlier in ML/MF than in patch AM [34] . Consistent with a hierarchical organization involving information passing from ML/MF to AM via AL , electrical stimulation of ML elicits a response in AL and stimulation in AL elicits a response in AM [35] . In addition , spatial position invariance increases from ML/MF to AL , and increases further to AM [34] as expected for a feedforward processing hierarchy . The firing rates of neurons in ML/MF are most strongly modulated by face viewpoint . Further along the hierarchy , in patch AM , cells are highly selective for individual faces and collectively provide a representation of face identity that tolerates substantial changes in viewpoint [34] . Freiwald and Tsao argued that the network of face patches is functional . Response patterns of face patch neurons are consequences of the role they play in the algorithm implemented by the ventral stream . Their results suggest that the face network computes a representation of faces that is—as much as possible—invariant to 3D rotation-in-depth ( viewpoint ) , and that this representation may underlie face identification behavior [34] . We carry out our investigation within the framework provided by a recent theory of invariant object recognition in hierarchical feedforward architectures [37] . It is broadly in accord with other recent perspectives on the ventral stream and the problem of object recognition [22 , 38] . The full theory has implications for many outstanding questions that are not directly related to the question of domain specificity we consider here . In other work , it has been shown to yield predictions concerning the cortical magnification factor and visual crowding [39] . It has also been used to motivate novel algorithms in computer vision and speech recognition that perform competitively with the state-of-the-art on difficult benchmark tasks [40–44] . The same theory , with the additional assumption of a particular Hebbian learning rule , can be used to derive qualitative receptive field properties . The predictions include Gabor-like tuning in early stages of the visual hierarchy [45 , 46] and mirror-symmetric orientation tuning curves in the penultimate stage of a face-specific hierarchy computing a view-tolerant representation ( as in [34] ) [46] . A full account of the new theory is outside the scope of the present work; we refer the interested reader to the references—especially [37] for details . Note that the theory only applies to the first feedforward pass of information , from the onset of the image to the arrival of its representation in IT cortex approximately 100 ms later . For a recent review of evidence that the feedforward pass computes invariant representations , see [22] . For an alternative perspective , see [11] . Though note also , contrary to a claim in that review , position dependence is fully compatible with the class of models we consider here ( including HMAX ) . [39 , 47] explicitly model eccentricity dependence in this framework . Our account of domain specificity is motivated by the following questions: How can past visual experience be leveraged to improve future recognition of novel individuals ? Is any past experience useful for improving at-a-glance recognition of any new object ? Or perhaps past experience only transfers to similar objects ? Could it even be possible that past experience with certain objects actually impedes the recognition of others ? The invariance hypothesis holds that the computational goal of the ventral stream is to compute a representation that is unique to each object and invariant to identity-preserving transformations . If we accept this premise , the key question becomes: Can transformations learned on one set of objects be reliably transferred to another set of objects ? For many visual tasks , the variability due to transformations in a single individual’s appearance is considerably larger than the variability between individuals . These tasks have been called “subordinate level identification” tasks , to distinguish them from between-category ( basic-level ) tasks . Without prior knowledge of transformations , the subordinate-level task of recognizing a novel individual from a single example image is hopelessly under-constrained . The main thrust of our argument—to be developed below—is this: The ventral stream computes object representations that are invariant to transformations . Some transformations are generic; the ventral stream could learn to discount these from experience with any objects . Translation and scaling are both generic ( all 2D affine transformations are ) . However , it is also necessary to discount many transformations that do not have this property . Many common transformations are not generic; 3D-rotation-in-depth is the primary example we consider here ( see S1 Text for more examples ) . It is not possible to achieve a perfectly view-invariant representation from one 2D example . Out-of-plane rotation depends on information that is not available in a single image , e . g . the object’s 3D structure . Despite this , approximate invariance can still be achieved using prior knowledge of how similar objects transform . In this way , approximate invariance learned on some members of a visual category can facilitate the identification of unfamiliar category members . But , this transferability only goes so far . Under this account , the key factor determining which objects could be productively grouped together in a domain-specific subsystem is their transformation compatibility . We propose an operational definition that can be computed from videos of transforming objects . Then we use it to explore the question of why certain object classes get dedicated brain regions , e . g . , faces and bodies , while others ( apparently ) do not . We used 3D graphics to generate a library of videos of objects from various categories undergoing rotations in depth . The model of visual development ( or evolution ) we consider is highly stylized and non-mechanistic . It is just a clustering algorithm based on our operational definition of transformation compatibility . Despite its simplicity , using the library of depth-rotation videos as inputs , the model predicts large clusters consisting entirely of faces and bodies . The other objects we tested—vehicles , chairs , and animals—ended up in a large number of small clusters , each consisting of just a few objects . This suggests a novel interpretation of the lateral occipital complex ( LOC ) . Rather than being a “generalist” subsystem , responsible for recognizing objects from diverse categories , our results are consistent with LOC actually being a heterogeneous region that consists of a large number of domain-specific regions too small to be detected with fMRI . These considerations lead to a view of the ventral visual pathway in which category-selective regions implement a modularity of content rather than process [48 , 49] . Our argument is consistent with process-based accounts , but does not require us to claim that faces are automatically processed in ways that are inapplicable to objects ( e . g . , gaze detection or gender detection ) as claimed by [11] . Nor does it commit us to claiming there is a region that is specialized for the process of subordinate-level identification—an underlying assumption of some expertise-based models [50] . Rather , we show here that the invariance hypothesis implies an algorithmic role that could be fulfilled by the mere clustering of selectivity . Consistent with the idea of a canonical cortical microcircuit [51 , 52] , the computations performed in each subsystem may be quite similar to the computations performed in the others . To a first approximation , the only difference between ventral stream modules could be the object category for which they are responsible . To make the invariance hypothesis precise , let gθ denote a transformation with parameter θ . Two images I , I′ depict the same object whenever ∃θ , such that I′ = gθ I . For a small positive constant ε , the invariance hypothesis is the claim that the computational goal of the ventral stream is to compute a function μ , called a signature , such that \| μ ( g θ I ) - μ ( I ) \| ≤ ϵ . ( 1 ) We say that a signature for which Eq ( 1 ) is satisfied ( for all θ ) is ϵ-invariant to the family of transformations {gθ} . An ϵ-invariant signature that is unique to an object can be used to discriminate images of that object from images of other objects . In the context of a hierarchical model of the ventral stream , the “top level” representation of an image is its signature . One approach to modeling the ventral stream , first taken by Fukushima’s Neocognitron [53] , and followed by many other models [54–58] , is based on iterating a basic module inspired by Hubel and Wiesel’s proposal for the connectivity of V1 simple ( AND-like ) and complex ( OR-like ) cells . In the case of HMAX [55] , each “HW”-module consists of one C-unit ( corresponding to a complex cell ) and all its afferent S-units ( corresponding to simple cells ) ; see Fig 1B . The response of an S-unit to an image I is typically modeled by a dot product with a stored template t , indicated here by ⟨I , t⟩ . Since ⟨I , t⟩ is maximal when I = t ( assuming that I and t have unit norm ) , we can think of an S-unit’s response as a measure of I’s similarity to t . The module corresponding to Hubel and Wiesel’s original proposal had several S-units , each detecting their stored template at a different position . Let g x ⃗ be the translation operator: when applied to an image , g x ⃗ returns its translation by x ⃗ . This lets us write the response of the specific S-unit which signals the presence of template t at position x ⃗ as ⟨ I , g x ⃗ t ⟩ . Then , introducing a nonlinear pooling function , which for HMAX would be the max function , the response C ( I ) of the C-unit ( equivalently: the output of the HW-module , one element of the signature ) is given by C ( I ) = max i ( ⟨ I , g x → i t ⟩ ) ( 2 ) where the max is taken over all the S-units in the module . The region of space covered by a module’s S-units is called its pooling domain and the C-unit is said to pool the responses of its afferent S-units . HMAX , as well as more recent models based on this approach typically also pool over a range of scales [56–58] . In most cases , the first layer pooling domains are small intervals of translation and scaling . In the highest layers the pooling domains are usually global , i . e . over the entire range of translation and scaling that is visible during a single fixation . Notice also that this formulation is more general than HMAX . It applies to a wide class of hierarchical models of cortical computation , e . g . , [53 , 58–60] . For instance , t need not be directly interpretable as a template depicting an image of a certain object . A convolutional neural network in the sense of [61 , 62] is obtained by choosing t to be a “prototype” obtained as the outcome of a gradient descent-based optimization procedure . In what follows we use the HW-module language since it is convenient for stating the domain-specificity argument . HW-modules can compute approximately invariant representations for a broad class of transformations [37] . However , and this is a key fact: the conditions that must be met are different for different transformations . Following Anselmi et al . [37] , we can distinguish two “regimes” . The first regime applies to the important special case of transformations with a group structure , e . g . , 2D affine transformations . The second regime applies more broadly to any locally-affine transformation . For a family of transformations {gθ} , define the orbit of an image I to be the set OI = {gθ I , θ ∈ ℝ} . Anselmi et al . [37] proved that HW-modules can pool over other transformations besides translation and scaling . It is possible to pool over any transformation for which orbits of template objects are available . A biologically-plausible way to learn the pooling connections within an HW-module could be to associate temporally adjacent frames of the video of visual experience ( as in e . g . , [63–68] ) . In both regimes , the following condition is required for the invariance obtained from the orbits of a set of template objects to generalize to new objects . For all gθ I ∈ OI there is a corresponding gθ′ t ∈ Ot such that ⟨ g θ I , t ⟩ = ⟨ I , g θ ′ t ⟩ ( 3 ) In the first regime , Eq ( 3 ) holds regardless of the level of similarity between the templates and test objects . Almost any templates can be used to recognize any other images invariantly to group transformations ( see S1 Text ) . Note also that this is consistent with reports in the literature of strong performance achieved using random filters in convolutional neural networks [69–71] . Fig 1A illustrates that the orbit with respect to in-plane rotation is invariant . In the second regime , corresponding to non-group transformations , it is not possible to achieve a perfect invariance . These transformations often depend on information that is not available in a single image . For example , rotation in depth depends on an object’s 3D structure and illumination changes depend on its material properties ( see S1 Text ) . Despite this , approximate invariance to smooth non-group transformations can still be achieved using prior knowledge of how similar objects transform . Second-regime transformations are class-specific , e . g . , the transformation of object appearance caused by a rotation in depth is not the same 2D transformation for two objects with different 3D structures . However , by restricting to a class where all the objects have similar 3D structure , all objects do rotate ( approximately ) the same way . Moreover , this commonality can be exploited to transfer the invariance learned from experience with ( orbits of ) template objects to novel objects seen only from a single example view . The theory makes two core predictions: Learned invariance to group transformations should be transferable from any set of stimuli to any other . For non-group transformations , approximate invariance will transfer within certain object classes . In the case of 3D depth-rotation , it will transfer within classes for which all members share a common 3D structure . Both core predictions were addressed with tests of transformation-tolerant recognition based on a single example view . Two image sets were created to test the first core prediction: ( A ) 100 faces derived from the Max-Planck institute face dataset [72] . Each face was oval-cropped to remove external features and normalized so that all images had the same mean and variance over pixels ( as in [73] ) . ( B ) 100 random greyscale noise patterns . 29 images of each face and random noise pattern were created by placing the object over the horizontal interval from 40 pixels to the left of the image’s center up to 40 pixels to the right of the image’s center in increments of 5 pixels . All images were 256 × 256 pixels . Three image sets were created to test the second core prediction: ( A ) 40 untextured face models were rendered at each orientation in 5° increments from −95° to 95° . ( B ) 20 objects sharing a common gross structure ( a conical shape ) and differing from one another by the exact placement and size of smaller bumps . ( C ) 20 objects sharing gross structure consisting of a central pyramid on a flat plane and two walls on either side . Individuals differed from one another by the location and slant of several additional bumps . The face models were generated using Facegen [74] . Class B and C models were generated with Blender [75] . All rendering was also done with Blender and used perspective projection at a resolution of 256 × 256 pixels . The tests of transformation-tolerant recognition from a single example were performed as follows . In each “block” , the model was shown a reference image and a set of query images . The reference image always depicted an object under the transformation with the median parameter value . That is , for rotation in depth of faces , it was a frontal face ( 0° ) and for translation , the object was located in the center of the visual field . Each query image either depicted the same object as the reference image ( target case ) or a different object ( distractor case ) . In each block , each query image was shown at each position or angle in the block’s testing interval . All testing intervals were symmetric about 0 . Using a sequence of testing intervals ordered by inclusion , it was possible to investigate how tolerance declines with increasingly demanding transformations . The radius of the testing interval is the abscissa of the plots in Figs 2 and 3 . For each repetition of the translation experiments , 30 objects were randomly sampled from the template class and 30 objects from the testing class . For each repetition of the depth-rotation experiments , 10 objects were sampled from template and testing classes that were always disjoint from one another . Networks consisting of K HW-modules were constructed where K was the number of sampled template objects . The construction followed the procedure described in the method section below . Signatures computed by these networks are vectors with K elements . In each block , the signature of the reference image was compared to the signature of each query image by its Pearson correlation and ranked accordingly . This ranked representation provides a convenient way to compute the ROC curve since it admits acceptance thresholds in terms of ranks ( as opposed to real numbers ) . Thus , the final measure of transformation tolerance reported on the ordinate of the plots in Figs 2 and 3 is the mean area under the ROC curve ( AUC ) over all choices of reference object and repetitions of the experiment with different training / test set splits . Since AUC is computed by integrating over acceptance thresholds , it is a bias free statistic . In this case it is analogous to d′ for the corresponding 2AFC same-different task . When performance is invariant , AUC as a function of testing interval radius will be a flat line . If there is imperfect invariance ( ϵ-invariance ) , then performance will decline as the radius of the testing interval is increased . To assess imperfect invariance , it is necessary to compare with an appropriate baseline at whatever performance level would be achieved by similarity in the input . Since any choice of input encoding induces its own similarity metric , the most straightforward way to obtain interpretable results is to use the raw pixel representation as the baseline ( red curves in Figs 2 and 3 ) . Thus , a one layer architecture was used for these simulations: each HW-module directly receives the pixel representation of the input . The first core prediction was addressed by testing translation-tolerant recognition with models trained using random noise templates to identify faces and vice versa ( Fig 2 ) . The results in the plots on the diagonal for the view-based model ( blue curve ) indicate that face templates can indeed be used to identify other faces invariantly to translation; and random noise templates can be used to identify random noise invariantly to translation . The key prediction of the theory concerns the off-diagonal plots . In those cases , templates from faces were used to recognize noise patterns and noise was used to recognize faces . Performance was invariant in both cases; the blue curves in Fig 2 were flat . This result was in accord with the theory’s prediction for the group transformation case: the templates need not resemble the test images . The second core prediction concerning class-specific transfer of learned ϵ-invariance for non-group transformations was addressed by analogous experiments with 3D depth-rotation . Transfer of invariance both within and between classes was assessed using 3 different object classes: faces and two synthetic classes . The level of rotation tolerance achieved on this difficult task was the amount by which performance of the view-based model ( blue curve ) exceeded the raw pixel representation’s performance for the plots on the diagonal of Fig 3 . The off-diagonal plots show the deleterious effect of using templates from the wrong class . There are many other non-group transformations besides depth-rotation . S1 Text describes additional simulations for changes in illumination . These depend on material properties . It also describes simulations of pose ( standing , sitting , etc ) -invariant body recognition . How can object experience—i . e . , templates—be assigned to subsystems in order to facilitate productive transfer ? If each individual object is assigned to a separate group , the negative effects of using templates from the wrong class are avoided; but past experience can never be transferred to new objects . So far we have only said that “3D structure” determines which objects can be productively grouped together . In this section we derive a more concrete criterion: transformation compatibility . Given a set of objects sampled from a category , what determines when HW-modules encoding templates for a few members of the class can be used to approximately invariantly recognize unfamiliar members of the category from a single example view ? Recall that the transfer of invariance depends on the condition given by Eq ( 3 ) . For non-group transformations this turns out to require that the objects “transform the same way” ( see S1 Text for the proof; the notion of a “nice class” is also related [76 , 77] ) . Given a set of orbits of different objects ( only the image sequences are needed ) , we would like to have an index ψ ¯ that measures how similarly the objects in the class transform . If an object category has too low ψ ¯ , then there would be no gain from creating a subsystem for that category . Whenever a category has high ψ ¯ , it is a candidate for having a dedicated subsystem . The transformation compatibility of two objects A and B is defined as follows . Consider a smooth transformation T parameterized by i . Since T may be class-specific , let TA denote its application to object A . One of the requirements that must be satisfied for ϵ-invariance to transfer from an object A to an object B is that TA and TB have equal Jacobians ( see S1 Text ) . This suggests an operational definition of the transformation compatibility between two objects ψ ( A , B ) . Let Ai be the ith frame of the video of object A transforming and Bi be the ith frame of the video of object B transforming . The Jacobian can be approximated by the “video” of difference images: JA ( i ) = ∣Ai − Ai+1∣ ( ∀i ) . Then define the “instantaneous” transformation compatibility ψ ( A , B ) ( i ) : = ⟨JA ( i ) , JB ( i ) ⟩ . Thus for a range of parameters i ∈ R = [−r , r] , the empirical transformation compatibility between A and B is ψ ( A , B ) : = 1 \| R \| ∑ i = - r r ⟨ J A ( i ) , J B ( i ) ⟩ . ( 4 ) The index ψ ¯ that we compute for sets of objects is the mean value of ψ ( A , B ) taken over all pairs A , B from the set . For very large sets of objects it could be estimated by randomly sampling pairs . In the present case , we were able to use all pairs in the available data . For the case of rotation in depth , we used 3D modeling / rendering software [75] to obtain ( dense samples from ) orbits . We computed the transformation compatibility index ψ ¯ for several datasets from different sources . Faces had the highest ψ ¯ of any naturalistic category we tested—unsurprising since recognizability likely influenced face evolution . A set of chair objects ( from [78] ) had very low ψ ¯ implying no benefit would be obtained from a chair-specific region . More interestingly , we tested a set of synthetic “wire” objects , very similar to those used in many classic experiments on view-based recognition e . g . [79–81] . We found that the wire objects had the lowest ψ ¯ of any category we tested; experience with familiar wire objects does not transfer to new wire objects . Therefore it is never productive to group them into a subsystem . The above considerations suggest an unsupervised strategy for sorting object experience into subsystems . An online ψ-based clustering algorithm could sort each newly learned object representation into the subsystem ( cluster ) with which it transforms most compatibly . With some extra assumptions beyond those required for the main theory , such an algorithm could be regarded as a very stylized model of the development ( or evolution ) of visual cortex . In this context we asked: Is it possible to derive predictions for the specific object classes that will “get their own private piece of real estate in the brain” [8] from the invariance hypothesis ? The extra assumptions required at this point are as follows . Cortical object representations ( HW-modules ) are sampled from the distribution D of objects and their transformations encountered under natural visual experience . Subsystems are localized on cortex . The number of HW-modules in a local region and the proportion belonging to different categories determines the predicted BOLD response for contrasts between the categories . For example , a cluster with 90% face HW-modules , 10% car HW-modules , and no other HW-modules would respond strongly in the faces—cars contrast , but not as strongly as it would in a faces—airplanes contrast . We assume that clusters containing very few HW-modules are too small to be imaged with the resolution of fMRI—though they may be visible with other methods that have higher resolution . Any model that can predict which specific categories will have domain-specific regions must depend on contingent facts about the world , in particular , the—difficult to approximate—distribution D of objects and their transformations encountered during natural vision . Consider the following: HW-modules may be assigned to cluster near one another on cortex in order to maximize the transformation compatibility ψ ¯ of the set of objects represented in each local neighborhood . Whenever a new object is learned , its HW-module could be placed on cortex in the neighborhood with which it transforms most compatibly . Assume a new object is sampled from D at each iteration . We conjecture that the resulting cortex model obtained after running this for some time would have a small number of very large clusters , probably corresponding to faces , bodies , and orthography in a literate brain’s native language . The rest of the objects would be encoded by HW-modules at random locations . Since neuroimaging methods like fMRI have limited resolution , only the largest clusters would be visible to them . Cortical regions with low ψ ¯ would appear in neuroimaging experiments as generic “object regions” like LOC [82] . Since we did not attempt the difficult task of sampling from D , we were not able to test the conjecture directly . However , by assuming particular distributions and sampling from a large library of 3D models [74 , 78] , we can study the special case where the only transformation is rotation in depth . Each object was rendered at a range of viewpoints: −90° to 90° in increments of 5 degrees . The objects were drawn from five categories: faces , bodies , animals , chairs , and vehicles . Rather than trying to estimate the frequencies with which these objects occur in natural vision , we instead aimed for predictions that could be shown to be robust over a range of assumptions on D . Thus we repeated the online clustering experiment three times , each using a different object distribution ( see S2 Table , and S6 , S7 , S8 , S9 , and S10 Figs ) . The ψ-based clustering algorithm we used can be summarized as follows: Consider a model consisting of a number of subsystems . When an object is learned , add its newly-created HW-module to the subsystem with which its transformations are most compatible . If the new object’s average compatibility with all the existing subsystems is below a threshold , then create a new subsystem for the newly learned object . Repeat this procedure for each object—sampled according to the distribution of objects encountered in natural vision ( or whatever approximation is available ) . See S1 Text for the algorithm’s pseudocode . Fig 4 shows example clusters obtained by this method . Robust face and body clusters always appeared ( Fig 5 , S8 , S9 , and S10 Figs ) . Due to the strong effect of ψ ¯ , a face cluster formed even when the distribution of objects was biased against faces as in Fig 5 . Most of the other objects ended up in very small clusters consisting of just a few objects . For the experiment of Figs 4 and 5 , 16% of the bodies , 64% of the animals , 44% of the chairs , and 22% of the vehicles were in clusters consisting of just one object . No faces ended up in single-object clusters . To confirm that ψ-based clustering is useful for object recognition with these images , we compared the recognition performance of the subsystems to the complete system that was trained using all available templates irrespective of their cluster assignment . We simulated two recognition tasks: one basic-level categorization task , view-invariant cars vs . airplanes , and one subordinate-level task , view-invariant face recognition . For these tests , each “trial” consisted of a pair of images . In the face recognition task , the goal was to respond ‘same’ if the two images depicted the same individual . In the cars vs . airplanes case , the goal was to respond ‘same’ if both images depicted objects of the same category . In both cases , all the objects in the cluster were used as templates; the test sets were completely disjoint . The classifier was the same as in Figs 2 and 3 . In this case , the threshold was optimized on a held out training set . As expected from the theory , performance on the subordinate-level view-invariant face recognition task was significantly higher when the face cluster was used ( Fig 5B ) . The basic-level categorization task was performed to similar accuracy using any of the clusters ( Fig 5C ) . This confirms that invariance to class-specific transformations is only necessary for subordinate level tasks . We explored implications of the hypothesis that achieving transformation invariance is the main goal of the ventral stream . Invariance from a single example could be achieved for group transformations in a generic way . However , for non-group transformations , only approximate invariance is possible; and even for that , it is necessary to have experience with objects that transform similarly . This implies that the optimal organization of the ventral stream is one that facilitates the transfer of invariance within—but not between—object categories . Assuming that a subsystem must reside in a localized cortical neighborhood , this could explain the function of domain-specific regions in the ventral stream’s recognition algorithm: to enable subordinate level identification of novel objects from a single example . Following on from our analysis implicating transformation compatibility as the key factor determining when invariance can be productively transferred between objects , we simulated the development of visual cortex using a clustering algorithm based on transformation compatibility . This allowed us to address the question of why faces , bodies , and words get their own dedicated regions but other object categories ( apparently ) do not [8] . This question has not previously been the focus of theoretical study . Despite the simplicity of our model , we showed that it robustly yields face and body clusters across a range of object frequency assumptions . We also used the model to confirm two theoretical predictions: ( 1 ) that invariance to non-group transformations is only needed for subordinate level identification; and ( 2 ) that clustering by transformation compatibility yields subsystems that improve performance beyond that of the system trained using data from all categories . These results motivate the the next phase of this work: building biologically-plausible models that learn from natural video . Such models automatically incorporate a better estimate of the natural object distribution . Variants of these models may be able to quantitatively reproduce human level performance on simultaneous multi-category subordinate level ( i . e . , fine-grained ) visual recognition tasks and potentially find application in computer vision as well as neuroscience . In [42] , we report encouraging preliminary results along these lines . Why are there domain-specific regions in later stages of the ventral stream hierarchy but not in early visual areas [2 , 3] ? The templates used to implement invariance to group transformations need not be changed for different object classes while the templates implementing non-group invariance are class-specific . Thus it is efficient to put the generic circuitry of the first regime in the hierarchy’s early stages , postponing the need to branch to different domain-specific regions tuned to specific object classes until later , i . e . , more anterior , stages . In the macaque face-processing system , category selectivity develops in a series of steps; posterior face regions are less face selective than anterior ones [34 , 83] . Additionally , there is a progression from a view-specific face representation in earlier regions to a view-tolerant representation in the most anterior region [34] . Both findings could be accounted for in a face-specific hierarchical model that increases in template size and pooling region size with each subsequent layer ( e . g . , [41 , 42 , 84 , 85] ) . The use of large face-specific templates may be an effective way to gate the entrance to the face-specific subsystem so as to keep out spurious activations from non-faces . The algorithmic effect of large face-specific templates is to confer tolerance to clutter [41 , 42] . These results are particularly interesting in light of models showing that large face templates are sufficient to explain holistic effects observed in psychophysics experiments [73 , 86] . As stated in the introduction , properties of the ventral stream are thought to be determined by three factors: ( 1 ) computational and algorithmic constraints; ( 2 ) biological implementation constraints; and ( 3 ) the contingencies of the visual environment [18–22] . Up to now , we have stressed the contribution of factor ( 1 ) over the others . In particular , we have almost entirely ignored factor ( 2 ) . We now discuss the role played by anatomical considerations in this account of ventral stream function . That the the circuitry comprising a subsystem must be localized on cortex is a key assumption of this work . In principle , any HW-module could be anywhere , as long as the wiring all went to the right place . However , there are several reasons to think that the actual constraints under which the brain operates and its available information processing mechanisms favor a situation in which , at each level of the hierarchy , all the specialized circuitry for one domain is in a localized region of cortex , separate from the circuitry for other domains . Wiring length considerations are likely to play a role here [87–90] . Another possibility is that localization on cortex enables the use of neuromodulatory mechanisms that act on local neighborhoods of cortex to affect all the circuitry for a particular domain at once [91] . There are other domain-specific regions in the ventral stream besides faces and bodies; we consider several of them in light of our results here . It is possible that even more regions for less-common ( or less transformation-compatible ) object classes would appear with higher resolution scans . One example may be the fruit area , discovered in macaques with high-field fMRI [3] . Lateral Occipital Complex ( LOC ) [82] These results imply that LOC is not really a dedicated region for general object processing . Rather , it is a heterogeneous area of cortex containing many domain-specific regions too small to be detected with the resolution of fMRI . It may also include clusters that are not dominated by one object category as we sometimes observed appearing in simulations ( see Fig 4 and S1 Text ) . The Visual Word Form Area ( VWFA ) [4] In addition to the generic transformations that apply to all objects , printed words undergo several non-generic transformations that never occur with other objects . We can read despite the large image changes occurring when a page is viewed from a different angle . Additionally , many properties of printed letters change with typeface , but our ability to read—even in novel fonts—is preserved . Reading hand-written text poses an even more severe version of the same computational problem . Thus , VWFA is well-accounted for by the invariance hypothesis . Words are frequently-viewed stimuli which undergo class-specific transformations . This account appears to be in accord with others in the literature [92 , 93] . Parahippocampal Place Area ( PPA ) [94] A recent study by Kornblith et al . describes properties of neurons in two macaque scene-selective regions deemed the lateral and medial place patches ( LPP and MPP ) [95] . While homology has not been definitively established , it seems likely that these regions are homologous to the human PPA [96] . Moreover , this scene-processing network may be analogous to the face-processing hierarchy of [34] . In particular , MPP showed weaker effects of viewpoint , depth , and objects than LPP . This is suggestive of a scene-processing hierarchy that computes a representation of scene-identity that is ( approximately ) invariant to those factors . Any of them might be transformations for which this region is compatible in the sense of our theory . One possibility , which we considered in preliminary work , is that invariant perception of scene identity despite changes in monocular depth signals driven by traversing a scene ( e . g . , linear perspective ) could be discounted in the same manner as face viewpoint . It is possible that putative scene-selective categories compute depth-tolerant representations . We confirmed this for the special case of long hallways differing in the placement of objects along the walls: a view-based model that pools over images of template hallways can be used to recognize novel hallways [97] . Furthermore , fast same-different judgements of scene identity tolerate substantial changes in perspective depth [97] . Of course , this begs the question: of what use would be a depth-invariant scene representation ? One possibility could be to provide a landmark representation suitable for anchoring a polar coordinate system [98] . Intriguingly , [95] found that cells in the macaque scene-selective network were particularly sensitive to the presence of long straight lines—as might be expected in an intermediate stage on the way to computing perspective invariance . Is this proposal at odds with the literature emphasizing the view-dependence of human vision when tested on subordinate level tasks with unfamiliar examples—e . g . [72 , 79 , 99] ? We believe it is consistent with most of this literature . We merely emphasize the substantial view-tolerance achieved for certain object classes , while they emphasize the lack of complete invariance . Their emphasis was appropriate in the context of earlier debates about view-invariance [100–103] , and before differences between the view-tolerance achieved on basic-level and subordinate-level tasks were fully appreciated [104–106] . The view-dependence observed in experiments with novel faces [72 , 107] is consistent with the predictions of our theory . The 3D structure of faces does not vary wildly within the class , but there is still some significant variation . It is this variability in 3D structure within the class that is the source of the imperfect performance in our simulations . Many psychophysical experiments on viewpoint invariance were performed with synthetic “wire” objects defined entirely by their 3D structure e . g . , [79–81] . We found that they were by far , the least transformation-compatible ( lowest ψ ¯ ) objects we tested ( Table 1 ) . Thus our proposal predicts particularly weak performance on viewpoint-tolerance tasks with novel examples of these stimuli and that is precisely what is observed [80] . Tarr and Gauthier ( 1998 ) found that learned viewpoint-dependent mechanisms could generalize across members of a homogenous object class [106] . They tested both homogenous block-like objects , and several other classes of more complex novel shapes . They concluded that this kind of generalization was restricted to visually similar objects . These results seem to be consistent with our proposal . Additionally , our hypothesis predicts better within-class generalization for object classes with higher ψ ¯ . That is , transformation compatibility , not visual similarity per se , may be the factor influencing the extent of within-class generalization of learned view-tolerance . Though , in practice , the two are usually correlated and hard to disentangle . In a related experiment , Sinha and Poggio ( 1996 ) showed that the perception of an ambiguous transformation’s rigidity could be biased by experience [108] . View-based accounts of their results predict that the effect would generalize to novel objects of the same class . Since this effect can be obtained with particularly simple stimuli , it might be possible to design them so as to separate specific notions of visual similarity and transformation compatibility . In accord with our prediction that group transformations ought to be discounted earlier in the recognition process , [108] found that their effect was spared by presenting the training and test objects at different scales . Many authors have argued that seemingly domain-specific regions are actually explained by perceptual expertise [24–27 , 109] . Our account is compatible with some aspects of this idea . However , it is largely agnostic about whether the sorting of object classes into subsystems takes place over the course of evolution or during an organism’s lifetime . A combination of both is also possible—e . g . as in [110] . That said , our proposal does intersect this debate in several ways . Our theory agrees with most expertise-based accounts that subordinate-level identification is the relevant task . The expertise argument has always relied quite heavily on the idea that discriminating individuals from similar distractors is somehow difficult . Our account allows greater precision: the precise component of difficulty that matters is invariance to non-group transformations . Our theory predicts a critical factor determining which objects could be productively grouped into a module that is clearly formulated and operationalized: the transformation compatibility ψ ¯ . Under our account , domain-specific regions arise because they are needed in order to facilitate the generalization of learned transformation invariance to novel category-members . Most studies of clustering and perceptual expertise do not use this task . However , Srihasam et al . tested a version of the perceptual expertise hypothesis that could be understood in this way [111] . They trained macaques to associate reward amounts with letters and numerals ( 26 symbols ) . In each trial , a pair of symbols were displayed and the task was to pick the symbol associated with greater reward . Importantly , the 3-year training process occurred in the animal’s home cage and eye tracking was not used . Thus , the distance and angle with which the monkey subjects viewed the stimuli was not tightly controlled during training . The symbols would have projected onto their retina in many different ways . These are exactly the same transformations that we proposed are the reason for the VWFA . In accord with our prediction , Srihasam et al . found that this training experience caused the formation of category-selective regions in the temporal lobe . Furthermore , the same regions were activated selectively irrespective of stimulus size , position , and font . Interestingly , this result only held for juvenile macaques , implying there may be a critical period for cluster formation [111] . Our main prediction is the link between transformation compatibility and domain-specific clustering . Thus one way to test whether this account of expertise-related clustering is correct could be to train monkeys to recognize individual objects of unfamiliar classes invariantly to 3D rotation in depth . The task should involve generalization from a single example view of a novel exemplar . The training procedure should involve exposure to videos of a large number of objects from each category undergoing rotations in depth . Several categories with different transformation compatibilities should be used . The prediction is that after training there will be greater clustering of selectivity for the classes with greater average transformation compatibility ( higher ψ ¯ ) . Furthermore , if one could record from neurons in the category-selective clusters , the theory would predict some similar properties to the macaque face-processing hierarchy: several interconnected regions progressing from view-specificity in the earlier regions to view-tolerance in the later regions . However , unless the novel object classes actually transform like faces , the clusters produced by expertise should be parallel to the face clusters but separate from them . How should these results be understood in light of recent reports of very strong performance of “deep learning” computer vision systems employing apparently generic circuitry for object recognition tasks e . g . , [62 , 112] ? We think that exhaustive greedy optimization of parameters ( weights ) over a large labeled data set may have found a network similar to the architecture we describe since all the basic structural elements ( neurons with nonlinearities , pooling , dot products , layers ) required by our theory are identical to the elements in deep learning networks . If this were true , our theory would also explain what these networks do and why they work . An HW-architecture refers to a feedforward hierarchical network of HW-layers . An HW-layer consists of K HW-modules arranged in parallel to one another ( see Fig 1B ) . For an input image I , the output of an HW-layer is a vector μ ( I ) with K elements . If I depicts a particular object , then μ ( I ) is said to be the signature of that object . The parameters ( weights ) of the k-th HW-module are uniquely determined by its template book T k = { t k 1 , ⋯ , t k m } . ( 5 ) For all simulations in this paper , the output of the k-th HW-module is given by μ k ( I ) = max t ∈ T k ( ⟨ I , t ⟩ ∥ I ∥ ∥ t ∥ ) . ( 6 ) We used a nonparametric method of training HW-modules that models the outcome of temporal continuity-based unsupervised learning [42 , 67] . In each experiment , the training data consisted of K videos represented as sequences of frames . Each video depicted the transformation of just one object . Let G0 be a family of transformations , e . g . , a subset of the group of translations or rotations . The set of frames in the k-th video was Otk = {gtk ∣ g ∈ G0} . In each simulation , an HW-layer consisting of K HW-modules was constructed . The template book Tk of the k-th HW-module was chosen to be T k : = O t k = { g t k \| g ∈ G 0 } . ( 7 ) Note that HW-architectures are usually trained in a layer-wise manner ( e . g . , [57] ) . That is , layer ℓ templates are encoded as “neural images” using the outputs of layer ℓ − 1 . However , in this paper , all the simulations use a single HW-layer . One-layer HW-architectures are a particularly stylized abstraction of the ventral stream hierarchy . With our training procedure , they have no free parameters at all . This makes them ideal for simulations in which the aim is not to quantitatively reproduce experimental phenomena , but rather to study general principles of cortical computation that constrain all levels of the hierarchy alike .	Domain-specific regions , like the fusiform face area , are a prominent feature of ventral visual cortex organization . Despite decades of interest from a large number of investigators employing diverse methods , there has been surprisingly little theoretical work on “why” the ventral stream may adopt this modular organization . In this study we propose a computational account of the role played by domain-specific regions in ventral stream function . It follows from a new theoretical analysis of the recognition problem which highlights the importance of building representations that are robust to class-specific transformations . These results provide a unifying account linking neuroimaging and neuropsychology-based ideas of domain-specific regions to the psychophysics and electrophysiology-oriented literature on view-based object recognition and invariance .	[ "Abstract", "Introduction", "Results", "Discussion", "Methods" ]	[]	2015	The Invariance Hypothesis Implies Domain-Specific Regions in Visual Cortex
A literature survey and analysis was conducted to describe the epidemiology of dengue disease in Brazil reported between 2000 and 2010 . The protocol was registered on PROSPERO ( CRD42011001826: http://www . crd . york . ac . uk/prospero/display_record . asp ? ID=CRD42011001826 ) . Between 31 July and 4 August 2011 , the published literature was searched for epidemiological studies of dengue disease , using specific search strategies for each electronic database . A total of 714 relevant citations were identified , 51 of which fulfilled the inclusion criteria . The epidemiology of dengue disease in Brazil , in this period , was characterized by increases in the geographical spread and incidence of reported cases . The overall increase in dengue disease was accompanied by a rise in the proportion of severe cases . The epidemiological pattern of dengue disease in Brazil is complex and the changes observed during this review period are likely to have been influenced by multiple factors . Several gaps in epidemiological knowledge regarding dengue disease in Brazil were identified that provide avenues for future research , in particular , studies of regional differences , genotype evolution , and age-stratified seroprevalence . PROSPERO registration number: CRD42011001826 . Dengue disease is an escalating public health problem [1] . Approximately 2·5 billion people live in over 100 endemic countries , predominantly in tropical areas where dengue viruses ( DENV ) can be transmitted [2] . DENV are arboviruses that are transmitted to humans by infected Aedes aegypti ( Linnaeus ) mosquitoes – the primary vector . Infection with any one of four DENV serotypes ( DENV-1 , -2 , -3 , or -4 ) can produce a spectrum of illness ranging from a mild , non-specific febrile syndrome , to classic dengue fever ( DF ) , or severe disease forms , such as dengue haemorrhagic fever ( DHF ) and dengue shock syndrome ( DSS ) , that can be fatal . The World Health Organization ( WHO ) estimates that >50 million dengue infections and >20 , 000 dengue-related deaths occur annually [1] , [3] , [4] . A recent disease distribution model has estimated there to be 390 ( 95% credible interval 284–528 ) million dengue infections per year , of which 96 million are apparent ( i . e . , cases manifest any level of clinical or sub-clinical severity ) [3] . During 2001–2007 , >4 million cases were notified in the Americas , and during 1995–2002 , >75% of these cases were reported from Brazil [5] , [6] . Ae . aegypti was eradicated from Brazil as a result of a Pan American Health Organization ( PAHO ) programme to control the spread of yellow fever . Additionally , DENV transmission was also suppressed in the Americas during the eradication programme . South American countries became re-infested with Ae . aegypti after the programme was discontinued and this , combined with the co-circulation of multiple DENV serotypes , led to the spread of dengue disease across the continent [5] , [7]–[9] . In 1982 , there was a dengue outbreak in a small city in the northern region of Brazil ( Boa Vista/Roraima ) , which was quickly brought under control and the virus did not spread [10] . In 1986 , the re-emergence of DENV-1 in Rio de Janeiro state [11] resulted in over 60 , 000 reported cases in 1987 and the subsequent spread of DENV increased national public health concerns [12]–[14] . Since the late 1980's the incidence of dengue disease continued to increase; 204 , 000 cases were reported nationally in 1999 [15] , [16] . By 2000 , DENV transmission was reported in 22/27 Brazilian states , and the mosquito vector was present in all states [17] . Much of Brazil is affected by a tropical wet and dry climate with high temperatures , high humidity and seasonal variations in rainfall; climate patterns that can provide appropriate conditions for breeding and survival of the Ae . aegypti mosquito . The country is divided into five regions ( North , Northeast , Central-West , Southeast , and South ) comprising 26 states and the federal district containing the capital city , Brasília . In 2000 there were nearly 170 million inhabitants of Brazil , increasing to more than 190 million in 2010 [18] , the majority of whom live in the large cities of the Southeast and Northeast regions [19] . The National System for Surveillance and Control of Diseases ( SNVS ) of Brazil , operates as part of the national health system ( Sistema Único de Saúde , or SUS ) . All reported cases from public health services or private health providers are included in the notification database ( Sistema de Informacoes de Agravos de Notificacao [SINAN] ) , which is openly accessible via the internet [20] . Until 2011 , the SNVS adopted the case definitions outlined in WHO guidelines [21] , [22] . In 1997 , the WHO categorized symptomatic dengue disease as: undifferentiated fever , DF and , DHF [21] . DHF was further classified into four severity grades , with grades III and IV being defined as DSS . However , difficulties in applying the criteria for DHF [23] , led the WHO to suggest a new classification based on levels of severity: non-severe dengue disease with or without warning signs , and severe dengue disease [22] . During 2000–2011 , both surveillance and hospitalization reporting systems in Brazil used DF and DHF; the surveillance system used an additional classification designated ‘DF with complications’ ( DFC ) [24] . Importantly , the articles included in this literature analysis that were based on secondary data used these surveillance sources . Our objectives of this literature search and analysis were to describe the epidemiology of dengue disease ( national and regional incidence [by age and sex] , seroprevalence and serotype distribution and other relevant epidemiological data ) in Brazil during 2000–2011 , and to identify gaps in epidemiological knowledge requiring further research . Between 31 July 2011 and 4 August 2011 , we searched databases of published literature ( Table 1 ) for epidemiological studies of dengue disease in Brazil . Search strategies for each database were described with reference to the expanded Medical Subject Headings ( MeSH ) thesaurus , encompassing the terms ‘dengue’ , ‘epidemiology’ , and ‘Brazil’ . Google and Yahoo searches ( limited to the first 50 results ) were used to identify national and international reports and guidelines , congress abstracts , and grey literature ( e . g . , Ministry of Health data , lay publications ) . To reduce selection bias , peer-reviewed contributions in English , Portuguese , or Spanish published between 1 January 2000 and 4 August 2011 were included; no limits by sex , age , ethnicity of study participants , or by study type were imposed . Single-case reports and articles only reporting data prior to 1 January 2000 were excluded . Unpublished reports were included if they were identified in one of the sources listed above . Data from grey materials supplemented that from peer-reviewed literature . Publications not identified in the target databases by the search strategy ( e . g . , locally published papers ) and unpublished data sources meeting the inclusion criteria ( e . g . , theses , Ministry of Health data ) were included if recommended by members of the literature review group . Editorials and data from literature reviews of previously published peer-reviewed studies were excluded . Duplicates and articles not satisfying the inclusion criteria were removed following review of the titles and abstracts . A further selection was made based on review of the full text from the first selection of references . Included publications were summarised using a data extraction instrument developed as a series of spreadsheets . Due to the expected heterogeneity of eligible studies in terms of selection , and number and classification of cases , a meta-analysis was not conducted . For the purposes of the analysis we defined national epidemics as those years with an incidence/100 , 000 above the 75th percentile for the period . A trend analysis was conducted on the national incidence and case number data . During the period 2000–2010 , the incidence of dengue disease in Brazil varied substantially , reaching a peak in 2010 of >1 million cases ( 538/100 , 000 inhabitants ) and the lowest value was approximately 72 , 000 cases in 2004 ( 63 . 2/100 , 000 inhabitants ) ( Table 2 , Figure 2A–C , Table S2 ) [6] , [15] , [16] , [26]–[31] . Despite the yearly variations and cyclical epidemics , trend analysis of the incidence of dengue in Brazil in the period 2000–2010 showed an overall increase in incidence over time that was not statistically significant ( β = 12·9/cases per 100 , 000 , p = 0·49 ) . Analysis of the number of cases of dengue disease over the review period shows a growth trend that was not statistically significant ( β = 47·984 cases/year , p = 0·25 ) . Nevertheless , the trend analysis suggests a worsening of the problem over time . There were three national epidemics ( years with incidence above the 75th percentile for the period [279 . 95] ) in 2002 , 2008 and 2010 . In 2002 there were 684 , 527 to 794 , 219 probable cases of DF , in 2008 , 637 , 663 to 806 , 036 cases [16] , [26] , [27] , and in 2010 there were over 1 million reported cases ( Table 2; Figure 2A ) [26] . A trough occurred in 2004 ( 71 , 847 to 113 , 000 cases ) [16] , [26] , [27] , [31] , representing <10 times the number reported in the peak year , 2010 ( Table 2; Figure 2A ) . The number of reported severe cases also varied by year and the annual proportion of DF manifest as DHF was 0 . 1–0 . 5% over the review period . In 2000 , the annual number of DHF cases was between 40 and 4502 [6] , [15] , [16] , [26] , [27] . The number of DHF cases during 2000–2010 ( >18 , 000 ) is striking when compared with data from the previous decade: during the 1990s <1000 cases of DHF were reported [26] . The years in which numbers of DHF cases peaked reflected the national epidemic years for dengue disease described above , with high numbers of DHF cases in 2002 and 2008 ( Figure 2B ) . The 2008 national epidemic of DF/DHF continued with elevated incidence into 2009/2010 , with nearly 4000 cases of DHF reported in 2010 [26] . The proportion of severe cases reported is typical of countries in the Americas , but is low compared with Asia where the reported incidence of DHF is much greater [32] . In similar-sized populations , attack rates for severe dengue disease are 18 times greater in Southeast Asia than in the Americas [32] . However , differences in health surveillance system reporting guidelines and variations in case management practices may contribute to the differences in the number of cases reported , and limit the ability to make valid comparisons [33] . In Brazil , DHF cases are defined by strict application of all four criteria from the 1997 WHO guidelines , which is not the case elsewhere [1] . Similarly , hospitalizations related to dengue disease increased over the survey period to >94 , 000 hospitalizations in 2010 ( Figure 2C ) [26] . The incidence of dengue-related hospitalization was 31·6/100 , 000 population during the 2002 national epidemic , approximately 40·8/100 , 000 during the 2008 national epidemic , and 49·7/100 , 000 during the 2010 national epidemic [26] . These increases in hospitalization rates during epidemic years might suggest an increase in the severity of dengue disease in Brazil , although an increased awareness during epidemics and a lower threshold for hospitalization might also account for these increases . The number of dengue-related deaths followed the same patterns as the other epidemiological indices of dengue disease . In 2010 , of 13 , 909 cases classified as DFC and 3807 classified as DHF , there were 370 and 308 fatal cases , respectively . The overall number of DHF- or DFC-related deaths was 678 compared with only 19 in 2004 ( Figure 2C ) [26] . A seasonal pattern of dengue disease was observed in those studies with available seasonal case distribution data . The highest incidences occurred during January–June [34]–[38] , corresponding to the period of highest rainfall and humidity , providing suitable conditions for Ae . aegypti breeding and survival . The study by Goncalves Neto et al . [35] showed that 83·3% of dengue disease cases occurred during the rainy season and demonstrated a positive Pearson correlation with the amount of rainfall ( r = 0·84 ) and relative humidity ( r = 0·76 ) and a negative correlation with temperature ( r = −0·78 ) . We found published regional data for part of the study period from four of the five Brazilian regions [6] , [28] , [34] , [35] , [39]–[51] . No published data were recovered for the North region . The available data show that incidence rates varied greatly throughout the country ( data not shown; Table S3 ) . In a study of 146 Brazilian cities in October 2006 , incidence rates ( per 100 , 000 population ) in the 61 cities that reported >500 dengue disease cases ranged between 24·70 ( Sao Paulo ) and 6222·71 ( Campo Grande ) [52] . By the end of 2006 , 25 of the 27 states had reported local dengue epidemics [15] . The geographic distribution of the Ae . aegypti vector has widened over the 11-year review period , involving an increasing number of municipalities ( Figure 2D ) and this has resulted in a broader regional distribution of dengue disease . In most regions the dengue disease incidence followed national trends ( Figure 2E ) . In the early years of the survey , the Southeast and Northeast regions were most affected by DENV infections , whereas from 2009 more cases were reported from studies within the Central-West region . Incidence rates reported in the South region were consistently lower than in other regions . The distribution of reported cases of dengue disease during the 2010 national epidemic was different from that in the 2002 and 2008 national epidemics with high attack rates observed over larger areas of Brazil [26] . These regional variations in dengue disease incidence are unsurprising given the geographically diverse nature of Brazil with its large variations in climate and population density . A change in the age distribution of dengue disease over the survey period was evident from the available data . Young adults were most affected by DF and DHF during 2000–2007 and 2000–2005 , respectively ( i . e . , DHF was coincident with the highest incidence of DF ) . However , in 2006 the incidence of DHF among children aged <5 years increased ( 0·47/100 , 000 ) and was higher than among those aged 10–19 years and 20–39 years ( 0·36/100 , 000 and 0·46/100 , 000 , respectively ) [9] . During 1998–2006 , most DHF cases were in the 20- to 40-year age group , whereas in 2007 >53% of DHF cases occurred in children <15 years of age [53] . In 2007 , a large proportion of cases of dengue-related hospitalizations ( 40 . 8% ) occurred among those aged <10 years . Furthermore , children aged 5–9 years and 10–14 years showed marked increases in hospitalization rates ( 68·2 and 60·6/100 , 000 population , respectively ) during the 2008 national epidemic , compared with during the 2002 national epidemic ( 15·9 and 23·1/100 , 000 population , respectively ) [26] . These hospitalization data are in agreement with the distribution of hospitalizations for dengue disease according to age for 2002–2011 ( Figure 3 ) [26] , which suggests a change in age pattern in 2007–2008 ( a reduction in the first quartile age ) although data from 2009 suggest this change may have been transient . The median age of death from DF was approximately 38 years in 2002 and fell to 30 years between 2007 and 2009 [26] . Regional age-related data from eligible studies are sparse and inter-regional comparisons are difficult ( Table 3 ) [35] , [39]–[42] , [44] , . The most comprehensive data are for 2001–2008 from Ceará state , Northeast region [39] . In 2001 , the highest incidence of cases occurred in those aged 20–59 years , whereas in the 2008 national epidemic , those mostly affected were aged <10 years . These data reflect the national changes in age distribution of dengue disease . Slightly more women than men are affected by dengue disease throughout Brazil [36] , which is similar to the sex distribution of reported cases in other Latin American countries [9] . During 2001–2010 the male∶female ratio of reported cases ranged from 0·75–0·82 [9] , [26] . Regional data were more variable . In 2000 the ratio was 1·09 in the city of São Luís [35] , and 0·5 in the City of Santos in 2010 [54] . Women with dengue disease were slightly older than men ( mean age 33·7 years versus 30·2 years , respectively; p = 0 . 019 ) [37] . Despite some gaps , our literature survey and analysis provides a comprehensive overview of the evolving epidemiology of dengue disease in Brazil over the period 2000–2011 . This study has several important strengths . Our survey was thorough; we screened >700 articles to identify relevant publications and we developed a comprehensive data extraction instrument to facilitate the capture of all relevant data . Nevertheless , the lack of comprehensive and continuous data for the survey period limits our ability to make comparisons and draw firm conclusions over the years , across regions , and among different ages . For example , age-stratified data were not reported systematically and age range boundaries differed by study . Therefore , although we can suggest trends in age distribution , it is not possible to directly compare data from the selected publications . The inclusion of publications in three languages reduced selection bias in our literature review and analysis . However , despite the inclusion of PhD dissertations and theses there is a bias towards published articles . An assessment of quality of evidence was not carried out and potential weaknesses of some studies such as inadequately described case selection , small sample sizes , and unspecified statistical methods were not reasons for exclusion . Consequently , any limitations of the original studies are carried forward into our review . Many of the studies relied on data reported by passive surveillance systems , which can vary between regions and over time [33] and may misrepresent the number of cases due to changes in reporting behaviour and misclassifications . Our literature survey and analysis identified several knowledge gaps , which indicate potential avenues for future study . In particular , there are gaps relating to the regional incidence of dengue disease in Brazil , national and regional age-related data , and national and regional serotype information . Further epidemiological studies may help to clarify and define regional differences . The large increase in the number of DHF cases and the shift in age distribution of DHF towards younger age groups that occurred during the 2007–2008 national epidemic warrant explanation . One possibility is that the change in circulating DENV serotypes over time may have affected the pattern of dengue disease epidemiology in Brazil [78] . Age-stratified seroprevalence studies will improve assessment of the level of transmission and inapparent infection , as well as providing information relating to the age shift . Further studies into the risk factors for dengue disease and its severity are also important . For example , in Southeast Asia , DENV infection has been more widespread for a longer period of time than in the Americas , creating a large group of individuals likely to experience a second or third infection [32] . These secondary infections carry an increased risk of severe dengue disease . The data in this review do not address the Southeast Asian experience and further examination as to whether this phenomenon is replicated in Brazil is required . In addition , few studies in the review specifically measured the effects of urbanization in Brazil , with effects only inferred from studies of other socio-demographic factors . The diversity of ethnic backgrounds within the population suggests that further genetic studies are warranted to determine whether ethnicity affects the clinical expression of dengue disease and the risk for severe outcomes . Studies are also required to clearly define associations with other diseases if comorbidity screening is to be used to identify patients at a greater risk of developing DHF . We acknowledge that there are gaps in our epidemiological knowledge of dengue disease in Brazil , due , in part ( as in many other countries ) to the inherent weaknesses of passive surveillance systems . The majority of infections are clinically non-specific consequently dengue disease is often mis-diagnosed during inter-epidemic periods [8] . The findings presented here are in broad agreement with those of Honório et al . [79] , who found only 23·3% of infections were symptomatic , and with Lima et al . [80] , who showed that the number of cases reported for the Southeast region of Brazil under-represented the number of infected individuals . This was also found in studies conducted in other countries [81] . Only when an epidemic occurs is the full spectrum of the disease recognised . Consequently , the disease is likely to be under-reported during inter-epidemic periods but over-reported during epidemics [82] . Overall , we believe the national surveillance data under-estimate the true incidence of DENV infections . However , extensive representative serological surveys are required to estimate the true rate of infection and transmission and , thus , despite its drawbacks , passive reporting is important for the identification of disease trends over time . Our review and analysis of the epidemiology of dengue disease in Brazil during the past decade suggests an overall increase in the distribution and severity of dengue disease . During the last decade ( 2000–2010 ) , a total number of 8 , 440 , 253 cases were reported ( the highest figure in the history of dengue disease in this region ) with the highest number of severe cases ( 221 , 043; 2 . 6% ) and fatal cases ( 3058; 0 . 036% of the total reported cases and 1 . 38% of the severe cases ) [83] . The 1588 cases of severe dengue disease and 163 deaths reported as of epidemiological week 8 in 2011 , represent 67% and 73% , respectively , of the total cases registered in the Americas [84] . The co-circulation of multiple DENV serotypes and high dengue disease endemicity may be responsible for the increased occurrence of severe forms of dengue disease and increases in the numbers of dengue disease-related hospitalizations . In addition , the increase in the number of severe cases of dengue disease and a shift in age group predominance of severe forms observed during 2007/08 confirm that dengue disease must remain a public health priority in Brazil . Even though the studies included in this literature review have improved our understanding of the epidemiology of dengue disease in Brazil , further studies are required to clarify the epidemiological pattern and to understand regional epidemiological differences , the diversity of genotypes of circulating serotypes and the extent of herd immunity by age group . Our review has highlighted the main epidemiological characteristics of dengue in Brazil in the first decade of this century and revealed that the epidemiological pattern of dengue disease in Brazil is complex . The changes observed are likely to have been the result of multiple factors , which still require elucidation .	Dengue disease is the most prevalent arthropod-borne viral disease in humans and is a global and national public health concern in Brazil . We conducted this review to consolidate and describe the existing evidence on the epidemiology of dengue disease in Brazil , between 2000 and 2011 , to gauge the recent national and regional impact of dengue disease and provide a basis for setting research priorities and prevention efforts . We used well-defined methods to search and identify relevant research , according to predetermined inclusion criteria . Despite control measures , the increased territorial distribution of the mosquito vector and the co-circulation of multiple dengue virus serotypes have resulted in increases in the incidence and distribution of dengue disease . The number of disease-related hospitalizations and deaths has also increased . Efforts to control the increasing disease incidence have been unsuccessful . This review of dengue disease epidemiology will help enhance knowledge and future disease management . Despite the high volume of research retrieved , we have identified several avenues for future research , in particular studies of regional differences , genotype evolution and age-stratified seroprevalence that will improve our knowledge of dengue disease , contribute to a more accurate estimate of global disease incidence , and also inform evidence-based policies for dengue disease prevention .	[ "Abstract", "Introduction", "Methods", "Results", "and", "Discussion" ]	[]	2013	Epidemiological Trends of Dengue Disease in Brazil (2000–2010): A Systematic Literature Search and Analysis
The ability to find and consume nutrient-rich diets for successful reproduction and survival is fundamental to animal life . Among the nutrients important for all animals are polyamines , a class of pungent smelling compounds required in numerous cellular and organismic processes . Polyamine deficiency or excess has detrimental effects on health , cognitive function , reproduction , and lifespan . Here , we show that a diet high in polyamine is beneficial and increases reproductive success of flies , and we unravel the sensory mechanisms that attract Drosophila to polyamine-rich food and egg-laying substrates . Using a combination of behavioral genetics and in vivo calcium imaging , we demonstrate that Drosophila uses multisensory detection to find and evaluate polyamines present in overripe and fermenting fruit , their favored feeding and egg-laying substrate . In the olfactory system , two coexpressed ionotropic receptors ( IRs ) , IR76b and IR41a , mediate the long-range attraction to the odor . In the gustatory system , multimodal taste sensation by IR76b receptor and GR66a bitter receptor neurons is used to evaluate quality and valence of the polyamine providing a mechanism for the fly’s high attraction to polyamine-rich and sweet decaying fruit . Given their universal and highly conserved biological roles , we propose that the ability to evaluate food for polyamine content may impact health and reproductive success also of other animals including humans . Animals make use of all of their senses while searching and evaluating food . For most , smell and taste are the major modalities to assess the quality and nutritional value of food . While odors help the animal to track down food over long distances , short-range evaluation using the sense of taste is ultimately crucial for the decision whether to feed or to lay an egg [1 , 2] . In general , animals prefer calorie-dense and sweet foods to bitter foods ( see for instance [3] ) . In addition , animals need to consume food containing other important nutrients . Among the vital dietary constituents are polyamines [4] . Polyamines , most notably putrescine , spermine , and spermidine , are essential for basic cellular processes such as cell growth and proliferation , and are of specific importance during reproduction [5 , 6] . While polyamines can be generated by endogenous biosynthesis or microbes in the gut , a significant fraction comes from the diet [4] . Animal products , soybeans , or certain fruits are rich sources of polyamines [4] . In cells , these polycations are bound to nucleic acids , proteins , and phospholipids , where they participate in fundamental cellular processes such as DNA replication , RNA translation , and mitosis [6 , 7] . Polyamine deficiency can have fatal consequences on reproductive success [5]; and low polyamine levels have been linked to neurodegenerative diseases and ageing [8] . Yet , high polyamine concentrations are found in cancer cells suggesting that their excess could be unhealthy [6] . Notably , the enzymes that generate endogenous polyamines decline with ageing [5–7] . And the exogenous supply through high polyamine diets can have beneficial effects on ageing , memory loss , and reproduction in a variety of model species and humans [4 , 8–11] . Given their beneficial but also detrimental roles , food industry has consequently measured the amount of polyamines in many food items [12] . Whether animals can directly evaluate polyamine content for instance through their taste organ is not known . The smell of polyamines , however , can be detected by animals , including humans; it is strongly pungent and at higher concentrations unpleasant to humans . There is circumstantial evidence indicating that insects may detect the smell and taste of polyamines . For instance , Calliphora blowflies are attracted to decaying flesh and lay their eggs exclusively into corpses . This attraction could be due to the carrion smell of cadaverine [13] . Drosophila flies are notorious for their attraction to overripe and fermenting fruit [14 , 15] . The amount of polyamines increases dramatically in fruit within days after harvest [16] and during fermentation [17] . Therefore , for these flies , polyamines are candidate molecules for the detection of beneficial feeding and egg-laying sites . This may also hold true for females of other insects . For instance , females of the dengue fever vector , the mosquito Aedes aegypti typically lay their eggs in batches in standing waters such as flowerpot plates with decaying organic and polyamine-rich materials [18 , 19] . Although an olfactory receptor , trace amine-associated receptor 13c ( TAAR13c ) , for one polyamine , cadaverine , was recently described in zebrafish [20] , no taste receptor has been identified so far . Chemosensory receptors for the detection of polyamines remain uncharacterized in insects . To detect chemosensory stimuli animals use highly specialized families of receptor proteins that are present in sensory neurons on peripheral taste or smell organs [21] . Given their importance in animal life , a large effort goes into identification of these receptors and their putative odors or tastes . D . melanogaster has proven to be a useful model in matching olfactory and gustatory receptors ( GRs ) to their ligands and has contributed much to our understanding of chemosensory coding in the nervous system [22] . Insects possess three classes of olfactory receptors ( ORs ) : the classical ORs , the more recently described but evolutionarily older family of ionotropic receptors ( IRs ) , and a few GRs [23–26] . Each olfactory sensory neuron ( OSN ) is located in a sensillum on either antenna or maxillary palp and expresses a specific type or very small combination of receptors , which are tuned to a narrow group of molecules . All OSNs that express the same receptor project their axons to one of ~50 glomeruli in the antennal lobe ( AL ) in the central brain [27] . This highly conserved architecture allows the translation of a nonspatial sensory cue into a highly organized spatial map and provides the logic for odor coding [21] . Upon additional local processing at the level of the AL , the odor information is sent via projection neurons ( PNs ) to two main higher brain centers , the mushroom body and the lateral horn [28] . While many of the ~45 ORs have been deorphanized , ligands for a number of IRs remain uncharacterized [22 , 29] . Previous work showed that most of the IR OSNs express one of the putative coreceptors IR8a or IR25a [30 , 31] . Among the deorphanized IRs is IR92a , the receptor for ammonia and small amines [32] . The behavioral role of most IRs , however , remains elusive with few exceptions such as IR84a and IR64a [29 , 33 , 34] . Gustatory receptor neurons ( GRNs ) , in contrast to OSNs , are found on many peripheral as well as internal organs [35] . On the external sensory organs , GRN-containing sensilla are mainly found on the labellum , the legs , and the wing margins [35] . The labellum carries ~60 morphologically distinct sensilla with four GRNs each that are tuned to distinct flavors such as sweet , salty , water ( appetitive ) or bitter , and acidic ( noxious ) . While GRs form the best-characterized family of taste receptors to date [22 , 36] , more recent members of the IRs have been implicated in the sensation of tastants [37–39] . For instance , IR76b was shown to be essential for the detection of appetitive concentrations of salt [40] . Interestingly , IR76b is also expressed in GRNs that do not detect salt , but a role for these neurons has not been assigned yet . Finally , increasing evidence suggests that dietary amines can be tasted , but receptors have not been identified yet [37] . Peripheral GRNs project to the central brain or in the case of GRNs on tarsae or wings , also to the ventral nerve cord [35] . In the central brain , the distinct patterns of innervation in an area called the subesophageal zone ( SEZ ) by bitter and sweet neurons indicated a taste map similar to but less structured than the map found in the AL of the olfactory system [41] . In contrast to the olfactory system , however , higher order processing of tastes is still not well understood [42] . We have analyzed at the sensory , molecular , and behavioral levels how polyamines guide insect preference behavior . First , we demonstrate that a polyamine-rich diet significantly increases the number of progeny of flies . Second , we show that flies can find and evaluate polyamine-rich feeding and egg-laying sites using their senses of smell and taste . We characterize the polyamine receptors and demonstrate an essential role for specific IRs in olfactory and gustatory organs . Altogether , our data characterize sensory receptors for polyamines and their behavioral role in insects and indicate that the ability to sense polyamines promotes reproductive success and survival . Given the evidence that polyamines are vital molecules during reproduction , we asked whether males and females feeding on polyamine-rich food would produce more offspring compared to flies on standard fly food ( see Materials and Methods ) . Therefore , we crossed single males to single females in two different conditions—on standard fly food and fly food that had been supplemented with putrescine or cadaverine solution ( ~2 . 5 mmol polyamine/l of food ) . After 4 d , the parental generation was discarded , and the number of eggs laid was quantified . In addition , once these eggs had developed into flies , these were counted again . Females on polyamine-enriched food laid ~3 times the amount of eggs compared to females on standard food . Similarly , fly pairs fed a high-polyamine diet had ~3 times more offspring than the couples on standard fly food ( Fig 1A ) . Thus , it indeed appeared that polyamine-enriched food was beneficial for reproduction of flies similarly as what has been suggested for other species such as humans . D . melanogaster flies are notorious for being attracted to , and for laying their eggs into , decaying fruits [14] . We asked whether this preference was rooted in the need to consume polyamines , present in fermenting fruit . First , we quantified the attraction of male and female flies to fruits at different stages of maturity with a laboratory choice assay , the T-maze . We found that flies show a strong aversion to green bananas , are indifferent to yellow bananas , but are highly attracted to the same batch of overripe bananas 5–7 d later ( Fig 1B ) [43] . Next , we tested whether Drosophila’s attraction to decaying fruit could , in part , be attributed to increased concentrations of polyamines produced during ripening and decay [16] . Running the same T-maze assays with different polyamines , we found that female and male flies were strongly attracted to the odor of spermine ( sperm . ) , spermidine ( sperd . ) , diaminopropane ( prop . ) , putrescine ( put . ) , cadaverine ( cad . ) , diaminohexane ( hex . ) , diaminoheptane ( hept . ) , diaminooctane ( oct . ) , and diaminodecane ( dec . ) ( Fig 1C ) . The responses were dose-dependent with 1 mM ( ~10 ppm ) eliciting the strongest attraction compared to lower as well as higher concentrations ( S1A Fig; 1 μM–1 M ) . This concentration roughly corresponded to the amount of putrescine found in fermented banana ( ~0 . 9 mmol/kg ) or fresh oranges ( ~1 . 3 mmol/kg ) [12] . We observed the same attraction when single female flies were assayed in the T-maze , suggesting that individual flies perceive and are attracted to the odor ( S1B Fig ) . To uncover the neural basis of the fly’s attraction to this class of important nutrients , we sought to identify the receptor for polyamine sensation . Preference to a chemical in the T-maze is typically mediated by the olfactory system . Indeed , flies with surgically removed antennae , the main olfactory organ , lost their T-maze preference for polyamines ( S1C Fig ) . Previous reports indicated that OR- and IR-expressing OSNs respond to putrescine in single sensillum recordings ( SSR ) [23 , 44–46] . From these two receptor classes , the entire OR system can be impaired at once by mutating the obligatory OR coreceptor ( Orco ) [47] . Orco mutant flies maintained normal attraction to putrescine ( S1D Fig ) , excluding this family of receptors from our search as suggested before [46] . We then analyzed the requirement of IRs . To suppress all IR-mediated chemosensation , we relied on atonal ( ato ) mutants , as they fail to develop IR-expressing coeloconic sensilla [23 , 48 , 49] . ato mutant flies did not show any preference for putrescine or cadaverine in the T-maze ( S1E Fig ) . We concluded that IRs mediate attraction to volatile polyamines . To identify specific IRs , we carried out a small genetic screen using loss of function of single IRs . From all IR mutants tested , including the two putative coreceptors IR8a and IR25a , only IR76b mutants showed a significant reduction in polyamine attraction in the T-maze ( Fig 1D , S1F Fig ) . To confirm a requirement for IR76b , we silenced the activity of IR76b neurons by expressing the inward-rectifier potassium channel Kir2 . 1 [50] . Flies of the genotype IR76b-Gal4;UAS-Kir2 . 1 showed a strong impairment in attraction to putrescine or cadaverine ( Fig 1E ) . It has been suggested that IRs may form functional heteromers in olfactory neurons , similarly to ORs [30] . Furthermore , IR76b as judged by reporter expression using a previously characterized IR76b-Gal4 transgene [40] was expressed in multiple types of OSNs with axons innervating four glomeruli strongly and three weakly in the AL ( Fig 1F; [46] ) . This strengthened the notion that another receptor might be used in conjunction with IR76b as previously hypothesized [46] . Among the OSNs previously shown to respond to putrescine [46] is a subset of IR41a-expressing neurons , housed in ac2 sensilla [31 , 46] . We blocked the activity of IR41a neurons with Kir2 . 1 ( IR41a-Gal4;UAS-Kir2 . 1 ) and found that similar to IR76b neuron silencing , these flies showed no attraction to polyamines ( Fig 1E ) . Using double labeling with IR41a-Gal4 and IR76b-QF , which labels the same neurons as IR76b-Gal4 ( S2 Fig and see also Silbering et al . [46] ) , we found that a small subset of OSNs innervating a ventral and central ( VC5 ) glomerulus coexpresses IR41a and IR76b ( Fig 1F ) . To obtain more direct evidence of a requirement of IR76b in IR41a neurons , we re-expressed IR76b in the IR76b mutant background selectively in IR41a or in all IR76b neurons ( Fig 1G ) . As expected , re-expression of IR76b in IR76b neurons fully rescued the flies’ attraction to polyamine odor ( IR76b-Gal4;UAS-IR76b;IR76b1 , Fig 1G ) . Importantly , the same rescue was observed when we re-expressed IR76b selectively in IR41a neurons ( IR41a-Gal4;UAS-IR76b;IR76b1 , Fig 1G ) . In a reciprocal experiment to test the role of IR41a in these neurons , we used RNAi to knockdown IR41a in IR76b-expressing neurons ( IR76b-Gal4;UAS-IR41a-i ) and assayed the effect in the T-maze . Knockdown of IR41a using two different RNAi transgenes reduced attraction to polyamines significantly as compared to control flies ( Fig 1H ) . These data , taken together , provide strong evidence that IR41a/IR76b coexpressing neurons are necessary and sufficient to mediate polyamine attraction . To strengthen the evidence for a role of IR41a/IR76b OSNs in the detection of polyamine odor , we used in vivo calcium imaging as a proxy of neuronal activity of these OSNs . To this end , we expressed the genetically encoded calcium indicator GCaMP6f [51] under the control of IR76b-Gal4 or IR41a-Gal4 ( IR-Gal4;UAS-GCaMP6f ) and recorded increases in intracellular calcium levels in response to the polyamine odor stimulus at the level of the axon terminals in the AL in the brain ( Fig 2A ) . When GCaMP was expressed exclusively in IR41a neurons , the innervated glomerulus strongly responded to putrescine in a concentration-dependent manner ( Fig 2B–2D; see also [46] ) . Similarly , only a single glomerulus , the one innervated by IR41a/IR76b neurons , responded strongly to polyamine odor when GCaMP was expressed in all IR76b neurons ( Fig 2E–2H ) . An independent IR76b , but not IR41a , neuron-innervated glomerulus , by contrast , did not respond significantly to polyamines ( S3A–S3C Fig ) . Notably , the response was very long lasting ( > 15 s , Fig 2D and 2H ) . Photoionization detector ( PID ) measurements suggested that odor might have been released into the airstream for up to 4 s with a 500 ms stimulus , because polyamine leaves a trace in the delivery line , which is cleaned out by air . This could , in part , explain this long-lasting response . Alternatively , this prolonged response could be a feature of some IR neurons as has been observed for other odors [32] . To test a requirement for IR76b in the observed odor response , we analyzed IR76b1 mutant flies . The response of the IR41a glomerulus was strongly reduced in IR76b loss of function mutants ( IR76b-Gal4;UAS-GCaMP6f;IR76b1 ) confirming the essential role of IR76b for polyamine odor detection ( Fig 2I–2K ) . These experiments suggest that IR41a and IR76b function in the same neurons as polyamine receptors used by flies to detect polyamine-rich food sources such as overripe fruit . Our data indicated that flies are attracted to a source of volatile polyamine such as overripe fruit through specific olfactory neurons . Furthermore , we showed that females on polyamine-enriched food laid more eggs and had more offspring than females on standard food ( see Fig 1A ) . We therefore asked if females use polyamines as a hallmark of beneficial egg-laying sites . To analyze this , we quantified the number of eggs laid on a polyamine-rich but otherwise plain , sugar-free substrate ( polyamine and 1% agarose ) versus a control substrate ( 1% agarose ) in an oviposition assay ( Fig 3A and 3B ) . Surprisingly , and in contrast to their olfactory preference , female flies avoided the polyamine-rich substrate and laid the majority of eggs on the polyamine-free site ( Fig 3B , S4A–S4C Fig ) . Single females made the same choice as groups of females and laid their eggs away from polyamines showing that this aversion was not a consequence of overcrowding ( Fig 3C , S4D Fig ) . While it was previously suggested that female flies avoid laying their eggs directly into feeding substrates [52] , the apparent dislike of the beneficial polyamine-rich substrates as oviposition sites was surprising . This behavior could reflect the rather artificial assay conditions , where flies chose between polyamine-rich and polyamine-free but an otherwise taste- or odorless substrate . Because in a decaying fruit , polyamines appear together with other food odors and tastes , we reasoned that females might find them more appealing for oviposition when combined to other chemosensory cues . We tested this by mixing either of two polyamines , putrescine or cadaverine , with apple juice and gave the flies the choice to lay their eggs either on apple juice alone or on the mixture . While flies prefer to lay their eggs on apple juice compared to pure polyamine , they strongly preferred the mixture of apple juice and polyamines to apple juice alone ( Fig 3D , S4E Fig ) . Similarly , flies laid significantly more eggs onto a substrate that contained sugar and putrescine than on sugar alone ( S4F Fig ) . Thus , polyamines are not only beneficial for egg-laying and increase the number of progeny , they also provide and enhance appealing landmarks for oviposition . The data highlighted that females use combinatorial sensory cues to decide where to lay an egg . We therefore examined which sensory modalities contributed to this short-range decision during oviposition . To facilitate the dissection of sensory mechanisms , we returned to assays using polyamine alone ( agarose + polyamine versus agarose ) and first determined the role of different sensory organs in polyamine choice during egg-laying . We found that ablation of antennae had no effect on oviposition avoidance to putrescine showing that IR41a/IR76b OSNs were dispensable once the female had closed in on an egg-laying substrate ( Fig 3E , S4G Fig ) . We therefore turned to the other chemosensory modality , gustation . Taste organs are specified during development by the transcription factor Poxn . In Poxn mutants , most taste organs with the exception of some taste organs in the pharynx [53] are transformed into mechanosensory organs , including those found on the labellum , the legs , and the wings [54 , 55] . Compared to wild type females , the taste-impaired Poxn mutant females completely lost their aversion to oviposit onto polyamine-rich substrate and laid their eggs in equal numbers of both sites of the assay ( Fig 3F , S4H Fig ) . This phenotype was rescued to wild type levels when Poxn was re-expressed using a full-genomic Poxn construct that rescued all taste neurons ( Fig 3F , S4H Fig ) [55] . Therefore , we concluded that the ultimate choice for egg-laying choice depends on GRNs , but does not involve pharyngeal taste neurons . We next sought to determine the external taste organs most critical for the female’s egg-laying decision . To this end , we ablated single taste organs and assayed oviposition behavior . Ablation of the lower segments of front , middle , or hind legs had no effect on the oviposition choice of wildtype females and the majority of eggs were laid on the polyamine-free site ( Fig 3E , S4G Fig ) . Similarly , females with clipped wings behaved normally to polyamines ( Fig 3E , S4G Fig ) . By contrast , ablation of the labellum resulted in equal egg-laying on both sides of the plate and complete loss of oviposition preference , strongly suggesting , together with the Poxn mutant data , that polyamine-based oviposition choice requires taste neurons on the fly’s labellum ( Fig 3E , S4G Fig ) . As we cannot ablate all legs simultaneously , it remains possible that tarsal IR76b neurons also contribute to some extent to the choice . Nevertheless , these other IR76b neurons cannot compensate for the lack of labellar neurons . Thus , gustatory neurons on the labellum are essential to mediate the short-range polyamine choice behavior during egg-laying . Taste sensilla contain several types of neurons including bitter- , salt- , and sugar-sensitive neurons [37] . How flies taste any amine including polyamines is unknown [37] . We thus sought to determine which taste receptors mediate the gustatory perception of polyamines . Some GRNs express IRs [37 , 56] . In contrast to the odor-detecting IR41a , IR76b is expressed in the gustatory system including GRNs of the labellum . These IR76b neurons project to the SEZ [40] ( Fig 3G ) . Given the requirement of IR76b olfactory neurons in attraction to polyamines , we tested the involvement of their gustatory counterparts in oviposition behavior . Four loss of function mutants of IR76b ( e . g . , IR76b1 ) , as well as the Gal4-mediated neuronal silencing of IR76b neurons ( IR76b-Gal4;UAS-Kir2 . 1 ) completely abolished polyamine avoidance during egg-laying ( Fig 3H , S4I Fig ) . By contrast , mutants for other IRs ( e . g . , IR8a and IR25a ) or ORCO or silencing of sugar-sensitive neurons ( GR5a and GR64f ) by Kir2 . 1 had no effect on polyamine avoidance of pure polyamine-agar substrate during oviposition ( Fig 3I , S4I–S4K Fig ) . These data implicate labellar IR76b taste neurons into the detection of polyamine taste . Polyamines are strongly bitter-tasting compounds to humans [57] . We tested whether this sensation could trigger the aversion of polyamines in the absence of fruit or sugar ( see above ) . When we silenced bitter-sensing GR66a taste neurons [58] during egg-laying , the aversive ( pure ) polyamine became attractive to these females ( GR66a-Gal4;UAS-Kir2 . 1 ) , which then strongly preferred to lay their eggs on the polyamine-rich otherwise plain substrate ( Fig 3I , S4K Fig ) . These results confirm that polyamines are highly attractive egg-laying substrates and suggest that GR66a neurons may inhibit or counteract such attractiveness . This result also provides a mechanistic explanation for why polyamines in the context of fruit or sugar are highly attractive to flies , because sweet sensation can quench the perception of bitter ( [59] see Fig 3D , S4E and S4F Fig ) . Silencing of both IR76b and GR66a neurons simultaneously ( GR66a-Gal4 , IR76b-Gal4;UAS-Kir2 . 1 ) abolished this preference for the polyamine-rich substrate . This further demonstrates that the attraction to polyamine indeed depends on IR76b ( Fig 3I , S4K Fig ) . Given that IR76b and GR66a are both expressed in taste organs , we carried out double-labeling experiments using genetic reporters to analyze their relative position on the labellum and their axonal projections into the SEZ ( Fig 3G , S5 Fig ) . We first confirmed that IR76b-QF and IR76b-Gal4 used in the behavioral assays were coexpressed in gustatory neurons . Although the relative expression levels of the two reporters showed some variation , they were by and large coexpressed ( S2 Fig ) . We therefore used the IR76b-QF reporter to analyze potential expression in GR bitter neurons ( GR66a-Gal4 ) . Our reporter expression data suggested that these receptors are not expressed in the same neurons in the labellum as the respective axon populations innervated distinct regions of the SEZ ( Fig 3G , S5 Fig , e . g . , SEZ panel ) [40] . These results suggest that two different taste neuron populations on the labellum are required to taste and evaluate polyamine-rich egg-laying substrates . To strengthen this evidence and to confirm that IR76b receptors were indeed required to taste polyamines , we carried out rescue experiments in IR76b1 mutant females . Re-expression of IR76b in IR76b receptor neurons ( IR76b-Gal4 , UAS-IR76b;IR76b1 ) resulted in a full rescue of oviposition behavior ( Fig 3J , S4L Fig ) . By contrast , re-expression of IR76b selectively in IR41a OSNs ( IR41a-Gal4 , UAS-IR76b;IR76b1 ) did not rescue oviposition choice confirming that this behavior depended on taste neurons ( Fig 3J , S4L Fig ) . Finally , we re-expressed IR76b in GR66a neurons ( GR66a-Gal4 , UAS-IR76b;IR76b1 ) and observed no rescue of the choice behavior confirming that IR76b receptors critical for polyamine taste do not reside in GR66a bitter neurons ( Fig 3J , S4L Fig ) . Hence , the fly receives and integrates two types of information , quality and valence , from one molecule , using two types of taste neurons . Given our data on polyamine and apple juice/sugar , such integration could potentially follow a mechanism that was recently demonstrated for sugar neurons , which are indirectly inhibited by bitter neurons via GABAergic inhibitory neurons of the SEZ [59] . This type of multimodal taste integration would allow the fly to measure the relative levels of polyamines and sugars during the evaluation of food . Our behavioral data provides strong evidence that polyamine sensing in the context of oviposition requires IR76b receptor in taste neurons on the female’s labellum . To test more directly whether taste neurons responded to polyamines , we monitored the response of IR76b neurons to putrescine using in vivo calcium imaging with GCaMP6f ( IR76b-Gal4; UAS-GCaMP6f; Fig 4A ) . Labellar stimulation with a putrescine solution led to a significant increase in GCaMP-fluorescence in primarily two regions of the SEZ innervated by at least two distinct sets of IR76b taste neurons ( Fig 4A–4E ) . One of these regions ( region of interest [ROI] 1 ) responded more strongly at for egg-laying behavior relevant concentrations ( 1 mM ) ( Fig 4C and 4D ) . By contrast , ROI 2 responded to higher concentrations of polyamine ( 100 mM; Fig 4C and 4E ) . We also observed calcium increases in ROI 1 in IR76b neuron axon terminals upon stimulation with salt consistent with a previous report implicating IR76b in the detection of low salt concentrations ( S6A Fig , 50 mM NaCl , [40] ) . Given that IR76b neurons are also found on the legs , we tested their responses to polyamines . To this end , we recorded GCaMP-fluorescence directly from the neurons on the tarsae ( S6B–S6D Fig ) . Our results suggest that polyamine-sensitive tarsal taste neurons exist on all legs and therefore could potentially contribute—although probably to a lesser extent—to the egg-laying choice . Of note , we only observed a significant response to high ( 100 mM ) but not to lower polyamine concentrations ( 1 mM ) . To gain more evidence for a role of IR76b as polyamine receptor , we recorded changes in GCaMP fluorescence in IR76b1 mutant flies and found that the response to polyamine was absent in both the ROI 1 and ROI 2 areas ( Fig 4F–4I ) . Similarly , the response to salt was also significantly reduced ( S6E Fig ) . These results demonstrate that IR76b receptor is required for polyamine detection by GRNs . Together with our behavioral data , these results show that IR76b , in addition to its function as salt receptor [40] , is a taste receptor for polyamines . We wondered which type of IR76b-expressing labellar GRN detects polyamines . Labellar taste neurons are housed in three types of taste bristles , short ( S ) , intermediate ( I ) , and long ( L ) , as well as in taste pegs [35] ( Fig 4J ) . L-type GRNs expressing IR76b promote the attraction to low levels of salt [40]; and IR76b expression as judged by the Gal4 reporter ( IR76b-Gal4;UAS-mCD8GFP ) can be observed in L-type sensilla [40] . In addition , we found that the reporter was strongly expressed in taste pegs ( Fig 4J ) . Consistent with this , the innervation of the ROI 1 area in the SEZ by IR76b GRN axons resembled the innervation previously observed for GRNs in taste pegs ( e . g . , GRNs responding to carbonation , [60] ) . As the ROI 1 area responded most strongly to behaviorally relevant concentrations of polyamines , it is conceivable that IR76b peg neurons are required for oviposition choice behavior . To test which neurons responded to the taste , we used tip recordings and recorded putative responses of the taste neurons housed in sensilla . In particular , we asked whether L-type sensilla were activated by polyamines . We found that none of the recorded L-type sensilla showed a response to putrescine consistent with our hypothesis that a different GRN type detects polyamines such as the peg neurons ( Fig 4K and 4L ) . We next tested the responses of S-type sensilla . These sensilla responded significantly to polyamines ( Fig 4K and 4L ) , but IR76b did not mediate these responses , as tip recordings on IR76b1 mutants still showed strong responses of S-type sensilla to putrescine ( Fig 4L ) . Given that bitter receptors are prominent in these sensilla , we hypothesize that bitter receptors such as GR66a might instead mediate these responses consistent with their involvement in oviposition behavior . We conclude that the fly’s taste sensation of polyamines requires IR76b in labellar GRNs , most likely in peg neurons but not in L-type sensilla that mediate the response to salt [40] . IR76b is also not required in S-type sensilla , which presumably mediate the bitter taste of these compounds . Finally , it appears that amongst IR76b taste neurons , some are involved in the sensation of polyamines while others sense salt , indicating the involvement of up-to-now unidentified partner receptors for salt and polyamines , respectively . Our data suggest that female flies sense and interpret polyamine odor and taste . To address how flies use and integrate this multisensory input elicited by a single compound during egg-laying site selection , we video-monitored the time flies spend on polyamine-rich compared to control substrate in the setup of the oviposition assay ( Fig 5A ) . To simplify the assay and the interpretation , we carried out the experiments on polyamine-rich substrates devoid of additional odors or tastes . We found that wild type female flies spent significantly more time on the polyamine site ( positive position index; solid lines ) , although they preferred to lay their eggs on the control site ( negative oviposition index; dashed lines ) ( Fig 5B , S7A Fig ) . Notably , flies showed the strongest positional preference for polyamine at the beginning of the experiment , but this preference steadily declined in parallel to an increase of eggs laid ( Fig 5B ) . We investigated which of the identified receptor neurons and thus sensory modalities are used for position and oviposition , respectively . Taste neuron-deficient Poxn mutants confirmed that oviposition aversion was dependent on taste; Poxn mutants showed no aversion and laid their eggs on both sides of the assay ( Fig 5C , S7B and S7C Fig ) . By contrast , positional attraction to polyamines remained intact in these flies ( Fig 5C ) . Of note , Poxn mutant females laid significantly less eggs in the first three hours that were observed in this assay ( S7B Fig ) . However , in the longer assay as shown above ( S4H Fig ) , the total number of eggs was comparable to controls suggesting that a lack of the sense of taste may initially inhibit the female’s willingness to deposit her eggs . We found that silencing IR41a OSNs ( IR41a-Gal4;UAS-Kir2 . 1 ) selectively affected position preference and not oviposition avoidance , suggesting that positional preference required olfactory detection of polyamines ( Fig 5D , S7D Fig ) . Furthermore , females with silenced IR76b neurons ( IR76b-Gal4;UAS-Kir2 . 1 ) , which included the IR41a/IR76b OSN population as well as the IR76b taste neurons , were completely indifferent to polyamines in position and oviposition ( Fig 5D , S7D Fig ) . They also showed a similar phenotype as Poxn mutant females and appeared to hold their eggs longer before deciding to lay them consistent with a deficient taste system ( S7D Fig and S4K Fig ) . Silencing of GR66a bitter taste neurons ( GR66a-Gal4;UAS-Kir2 . 1 ) had no effect on position behavior , but as reported above , reversed oviposition avoidance to preference ( Fig 5D , S7D Fig ) . Egg numbers were comparable to controls ( S7D Fig ) . Furthermore , concurrent inactivation of IR76b and bitter tasting neurons ( Gr66a-Gal4 , IR76b-Gal4;UAS-Kir2 . 1 ) completely abolished positional and oviposition preference ( Fig 5D , S7D Fig ) . In summary , these data show that odor- and taste-induced behaviors do not depend on each other but rather happen in a parallel or sequential manner . Polyamine odors could be long-range attractive cues for female flies to navigate to putative oviposition sites . Upon arrival on or in close proximity of such sites ( a decaying fruit or the confined space of our oviposition assay ) , the sense of smell is dispensable , and females refine their choice for oviposition with their taste organs , likely integrating polyamine inputs with other tastes ( e . g . , sugar ) . Having shown that Drosophila flies use polyamines to identify egg-laying sites , we next investigated a more general role of these compounds beyond the vinegar fly . Ae . aegypti mosquitoes transmit the dangerous disease dengue fever and cause about 25 , 000 deaths per year [61] . ORs are among the potential targets of measures for pest control . Because Aedes mosquito adults live in human households and are often found attached to sheets , curtains , etc . [62] , insecticide treatments of adults are often limited by their proximity to humans . Targeting breeding sites might therefore be more practical and efficient [63] . Using laboratory oviposition assays , we asked whether Aedes females are attracted to the odor of polyamines for egg-laying ( Fig 6A ) . Single females were released into a caged area and given the choice between laying into a cup that smelled of polyamines and an odorless cup . They were prevented from tasting the compounds in this assay and forced to use their sense of smell . We found that egg-laying female mosquitoes deposit significantly more eggs into water that smelled of putrescine or cadaverine compared to control ( Fig 6B and 6C ) . This attraction was concentration-dependent , and mosquitoes were most attracted at 1μM and 10 μM of putrescine or cadaverine ( Fig 6B and 6C ) . By contrast , concentrations as of 100 μM started to become repulsive , and females preferred to lay their eggs into the nonsmelling cup ( Fig 6B and 6C ) . These experiments suggest that also other insects such as mosquitoes use polyamines to find feeding and egg-laying sites . These polyamine-based behaviors and possibly the detection mechanisms might be conserved in other species . Polyamines represent an important component of animal nutrition [6 , 12] . Deficiency as well as excess of polyamines can be detrimental to health and reproduction [6] . Therefore , species might have undergone a selection to choose food with levels of polyamines that meet their physiological needs . Our results provide important biological significance for the preference of D . melanogaster for polyamine-rich food such as fermented fruit or fresh oranges ( see also [14 , 15] ) . In particular , we show that a polyamine-enriched diet increases the number of offspring produced by a fly couple . Females feeding on polyamine-supplemented food lay significantly more eggs . Mechanistically , we speculate that the diverse and conserved roles of polyamines during cell cycle progression , differentiation , and autophagy among others are responsible for the beneficial effect [5] . Interestingly , in addition to uptake through the diet , polyamine-synthesis enzymes ( i . e . , ornithine decarboxylase ) are selectively up-regulated after mating in the spermatheca of female mosquitoes , a tissue involved in sperm storage , egg-production , and laying in mosquito and Drosophila , consistent with a role in fertility and reproduction [64] . Conception and proper embryonic development depend on polyamines also in humans [5] . Polyamines also form a substantial part of the male ejaculate ( i . e . , spermine and spermidine ) and infertile men show lower levels of spermine , spermidine , and putrescine in their semen [5] . Interestingly , polyamine-synthesizing enzymes in the cell decay with age , spurring studies on the beneficial effects of polyamines in the diet . Polyamine supplementation might counteract age-related loss of fertility in men and women , but also other ageing-induced deficits such as loss of memory or even lifespan [4 , 8 , 10 , 11] . On the other hand , an excess of these compounds in the cell , and therefore possibly in the diet has been linked with the occurrence and progression of cancer and other diseases suggesting that polyamine intake should be carefully regulated [6] . With the characterization of sensory mechanisms underpinning the attraction to polyamines , we can begin to analyze how fundamental physiological needs shape sensory processing and ultimately impinge on feeding and reproductive behavior . Flies detect the odor and taste of polyamines . In addition , our data suggests that two types of gustatory neurons evaluate quality and valence of polyamines separately . The female to identify egg-laying sites uses these two taste modalities , polyamine taste and bitter taste . In this context and presumably during feeding , sugar in the fruit appear to override the bitter taste of polyamines , which translates at the behavioral level to a preference for polyamine-rich compared to polyamine-poor fruity substrates . This claim is supported by data showing that female flies strongly prefer to lay their eggs into pure polyamine substrates when their bitter sense is silenced genetically . Therefore , several modalities contribute to polyamine choice behavior . Such multimodal detection of polyamines might ensure that the animal only consumes polyamines in the context of a suitable food source and at beneficial concentrations . A similar interaction was suggested between sugar and bitter taste sensation in the fly , indicating that concentrations might be generally estimated by assessing the relative amounts of tastants [59 , 65 , 66] . Multimodal taste experiences are essential to judge food quality also for humans . Therefore , sugar is frequently used to quench bitter tastes in food and to make medication more palatable [67] . In addition to integrating two types of taste modalities , flies also appear to use their sense of smell to find polyamine-rich foods . Our tracking data shows that these two kinds of information , smell and taste , are used sequentially consistent with odor being a long-range and taste a short-range signal . Flies that have found the source of the polyamine using their sense of smell do no longer require it to make the decision on where to lay eggs . However , whether odor and tastes are integrated in a more complex odor/taste environment of polyamines and other cues remains to be investigated . Our behavioral data shows that flies are attracted to all polyamines tested including putrescine , spermine , and cadaverine . Notably , they appear to prefer concentrations that are typically found in fermented foods , overripe fruit , or oranges , also a favorite of flies [15] . The receptor IR76b is required for the detection of the odor as well as the taste . Mutants for IR76b show significantly reduced attraction to polyamines . In this context , IR76b seems to work with another receptor , IR41a , which is specific to a very small subset ( ~7 ) of antennal OSNs . Although IR76b is more broadly expressed than IR41a in the olfactory system , it is unclear whether it plays the role of a coreceptor like IR8a or IR25a [31] . IR25a , which appears to be coexpressed with IR76b and IR41a , could play this role in polyamine detection . However , IR25a mutant flies show no decrease in polyamine attraction compared to controls . Thus , it is possible that IR76b and IR41a work as functional receptor heteromers , a configuration that is necessary and sufficient to form functional chemoreceptors [30 , 31] . Notably , the effect of the IR76b mutation in calcium imaging of OSNs appeared significantly stronger than the effect of the mutation in odor-guided behavior ( see Fig 1 and Fig 2 ) , although IR76b and the IR41a glomerulus appeared to be necessary and sufficient to mediate polyamine odor attraction ( see Fig 1 ) . This difference could be due to the slightly different conditions in behavior and imaging . On the one hand , animals are freely moving during behavior and might experience odor plumes rather than constant streams . On the other hand , animals are exposed for longer periods of times to the odor in the T-maze as compared to the imaging experiments . These effects might also explain a discrepancy between our behavioral data and a previous study implicating the IR coreceptor IR8a in the detection of putrescine using single sensillum electrophysiology [31] . As mentioned above the same IR8a loss of function mutants as used by Abuin et al . [31] did not show any significant reduction in polyamine odor attraction in behavioral assays . In the gustatory system , IR76b expressed in labellar GRNs is necessary and sufficient to mediate polyamine choice behavior . In contrast to the olfactory system , a mutation in IR76b results in complete loss of preference behavior to polyamines as well as a loss of calcium responses in GRNs . Calcium imaging and tip recording along with expression data indicate that polyamines are not recognized by IR76b expressed in L-type or S-type sensilla on the labellum , but instead might be detected by peg taste neurons that express IR76b . Furthermore , IR76b taste neurons on the leg , although not essential for egg-laying decisions , responded to polyamines . Based on our data , we can exclude and infer the involvement of certain types of IR76b taste neurons for polyamine detection , but further experiments will be required to reveal their exact identity and position . A previous study showed that IR76b in L-type neurons was required for the fly’s attraction to low levels of salt [40] . How can the same receptor mediate two or more different taste modalities ? The easiest explanation might be the involvement of at least one coreceptor for either salt or polyamine . Given that IR76b expressing L-type sensilla do not respond to polyamines , this appears to be a likely scenario at least for the detection of polyamines . Up to now , our candidate approaches , however , have not identified such a coreceptor . The other possibility is that the same receptor has putative binding sites for both of these ligands . In fact , ORs with few exceptions such as the CO2 receptors [24 , 26] detect multiple odorants . Nevertheless , salt and polyamines appear to be very different types of ligands . The activation of IR76b receptor by salt seems to depend on a particular amino acid located in the transmembrane domain , which is required for ion conductance in ionotropic glutamate receptors ( iGluRs ) [40] . The authors propose that this amino acid will keep the channel in a constitutively open or partly open position , which allows sodium entry when GRNs contact salt . The structural similarity of IRs and iGluRs might provide hints for how polyamines could activate IRs . Polyamines are released from presynaptic terminals and can thus interact with the extracellular domains of synaptic iGluRs [68] . Such interactions appear to modulate the activity of some iGluRs ( e . g . , [69] ) . For instance , spermine can potentiate the activity of NMDA ( N-methyl-D-aspartate ) receptor by binding at a site within the extracellular domains of one of the subunits , the NR1 subunit [69–71] . Comparison of the putative structures of IR41a and IR76b and the structure of the NR1 subunit shows that these receptors share a high structural similarity ( S8 Fig ) . It is thus conceivable that polyamine activation of IR41a and IR76b follows a similar mechanism as polyamine potentiation of NMDAR [72] . Structure–function analysis guided by studies on the NMDAR will help to test this model . It is certainly exciting to speculate that the binding and modulation by polyamines has been acquired early on in the evolution of iGluRs and further optimized in specific IRs . D . melanogaster stocks were raised on conventional cornmeal-agar medium at 25°C temperature and 60% humidity and a 12 hr light:12 hr dark cycle . Following fly lines were used to obtain experimental groups of flies in the different experiments: The majority of the lines were obtained from Bloomington ( http://flystocks . bio . indiana . edu/ ) or the VDRC stock center ( http://stockcenter . vdrc . at ) . The Poxn lines were a gift by Werner Boll and IR76b-QF was a gift by Craig Montell .	Animals , including humans , evaluate food by its smell and taste . Odors and tastes not only signal the presence of food , they also reveal details about the type and amount of nutrients contained in it . A preference for certain foods frequently reflects the specific metabolic needs of an animal . Among the important but less known compounds that animals consume with their diet are polyamines . These pungent smelling molecules are essential for reproduction , development , and cognition . Interestingly , they are also produced by the cell and body , but their levels decline with age . A diet high in polyamines can improve age-related memory deficits and loss of fertility . We have used the model fly Drosophila melanogaster to unravel if and how animals detect polyamines in their food and environment , and which role this detection plays in their food choice behavior . Polyamine levels are particularly high in the fly’s favorite food and egg-laying substrate , overripe and decaying fruit . We found that food supplemented with polyamines indeed improves the reproductive success of a fly couple . We show that Drosophila is highly attracted to polyamines and uses them to identify promising egg-laying and feeding sites . It detects them through an ancient clade of receptor proteins on its olfactory and taste organs . We speculate that other animals can also detect polyamines and use their smell and taste to identify sources of these beneficial nutrients .	[ "Abstract", "Introduction", "Results", "Discussion", "Materials", "and", "Methods" ]	[ "smell", "invertebrates", "medicine", "and", "health", "sciences", "legs", "social", "sciences", "limbs", "(anatomy)", "neuroscience", "animals", "reproductive", "physiology", "animal", "models", "drosophila", "melanogaster", "model", "organisms", "drosophila", "research"...	2016	Ionotropic Chemosensory Receptors Mediate the Taste and Smell of Polyamines
One outcome of interspecific hybridization and subsequent effects of evolutionary forces is introgression , which is the integration of genetic material from one species into the genome of an individual in another species . The evolution of several groups of eukaryotic species has involved hybridization , and cases of adaptation through introgression have been already established . In this work , we report on PhyloNet-HMM—a new comparative genomic framework for detecting introgression in genomes . PhyloNet-HMM combines phylogenetic networks with hidden Markov models ( HMMs ) to simultaneously capture the ( potentially reticulate ) evolutionary history of the genomes and dependencies within genomes . A novel aspect of our work is that it also accounts for incomplete lineage sorting and dependence across loci . Application of our model to variation data from chromosome 7 in the mouse ( Mus musculus domesticus ) genome detected a recently reported adaptive introgression event involving the rodent poison resistance gene Vkorc1 , in addition to other newly detected introgressed genomic regions . Based on our analysis , it is estimated that about 9% of all sites within chromosome 7 are of introgressive origin ( these cover about 13 Mbp of chromosome 7 , and over 300 genes ) . Further , our model detected no introgression in a negative control data set . We also found that our model accurately detected introgression and other evolutionary processes from synthetic data sets simulated under the coalescent model with recombination , isolation , and migration . Our work provides a powerful framework for systematic analysis of introgression while simultaneously accounting for dependence across sites , point mutations , recombination , and ancestral polymorphism . Hybridization is the mating between species that can result in the transient or permanent transfer of genetic variants from one species to another . The latter outcome is referred to as introgression . Mallet [1] recently estimated that "at least 25% of plant species and 10% of animal species , mostly the youngest species , are involved in hybridization and potential introgression with other species . " Introgression can be neutral and go unnoticed in terms of phenotypes but can also be adaptive and affect phenotypes . Recent examples of adaptation through hybridization include resistance to rodenticides in mice [2] and mimicry in butterflies [3] . Detecting regions with signatures of introgression in eukaryotic genomes is of great interest , given the consequences of introgression in evolutionary biology , speciation , biodiversity , and conservation [1] . With the increasing availability of genomic data , it is imperative to develop techniques that detect genomic regions of introgressive descent . Let us consider an evolutionary scenario where two speciation events result in three extant species A , B , and C , with A and B sharing a most recent common ancestor . Further , some time after the splitting of A and B , hybridization occurs between B and C ( that is , sexual reproduction of individuals from these two species ) . This scenario is depicted by the phylogenetic network in Fig . 1 . Immediately upon hybridization , approximately half of the hybrid individual's genome comes from an individual in species B , whereas the remainder comes from an individual in species C . However , in homoploid hybridization , where the hybrid offspring has the same ploidy level as the two parental species , hybridization is often followed by back-crossing ( further mating between the hybrid population and either of the two parental populations ) . Repeated back-crossing , followed by the effects of genetic drift and natural selection , results in genomes in the hybrid individuals that are mosaics of genomic material from the two parental species , yet not necessarily with a 50–50 composition . Thus , detecting introgressed regions requires scanning across the genome and looking for signals of introgression . In a comparative framework , detecting introgressed regions can be achieved by evolutionary analysis of genomes from the parental species , as well as genomes from introgressed individuals . In such an analysis , a walk across the genomes is taken , and local genealogies are inspected; incongruence between two local genealogies can be taken as a signal of introgression [4] . ( Here , we focus on topological incongruence; see [5] for a related discussion on local variation of coalescence times . ) However , in reality , the analysis is more involved than this , owing to potentially confounding signal produced by several factors , a major one of which is incomplete lineage sorting ( ILS ) . As recombination breaks linkage across loci in the genome , the result is that independent loci might have different genealogies by chance , which is known as ILS . ILS is common to several groups of eukaryotic taxa where species diverged with insufficient time for all genomic loci to completely sort , resulting in a scenario where introgression and ILS effects need to be distinguished [3] , [6]– . Fig . 1 illustrates this issue , where local genealogies across recombination breakpoints differ due to ILS , but also differ inside vs . outside introgressed regions . While other factors , such as gene duplication and loss [10] , could potentially add to the complexity of the phylogenetic and genomic patterns , we focus here on introgression and ILS . Recently , new methods were proposed to detect introgression in the presence of ILS . Durand et al . 's statistic allows for a sliding-window analysis of three-taxon data sets , while accounting for introgression and ancestral polymorphism [11] . However , this statistic assumes an infinite-sites model and independence across loci . Yu et al . [5] proposed a new statistical model for the likelihood of a species phylogeny model , given a set of gene genealogies , accounting for both ILS and introgression . However , this model does not work directly from the sequences; rather , it assumes that gene genealogies have been estimated , and computations are based on these estimates . Further , the model assumes independence across loci . Of great relevance to our work here is an array of hidden Markov model ( HMM ) based techniques that were introduced recently for analyzing genomic data in the presence of recombination and ILS [12]–[14]; however , these methods do not account for introgression . A recent extension [15] was devised to investigate the effects of population structure and migration . Finally , Saguaro is a recent method that combines HMMs with artificial neural networks to annotate genomic regions into different classes based upon local phylogenetic incongruence [16] . The classes are meant to categorize local genealogies , but the method is not aimed at elucidating the cause of incongruence . In this paper , we devise a novel model based on integrating phylogenetic networks with hidden Markov models ( HMMs ) . The phylogenetic network component of our model captures the relatedness across genomes ( including point mutation , recombination , ILS , and introgression ) , and the HMM component captures dependence across sites and loci within each genome . Using dynamic programming algorithms [17] paired with a multivariate optimization heuristic [18] , the model can be trained on genomic data , and allows for the identification of genomic regions of introgressive descent . We applied our model to chromosome 7 genomic variation data from three mouse data sets . Our analysis recovered an introgression event involving the rodenticide resistance gene Vkorc1 , which was recently reported in the literature [2] . Based on the analysis , 9% of sites within chromosome 7 are in fact of introgressive origin , which is a novel finding in that previously only a localized region ( that included Vkorc1 ) had been identified , with no further regions scanned . When applied to the negative control data set , our model did not detect any introgression , further attesting to its robustness . Our software is publicly available as part of the open-source PhyloNet distribution [19] . The method and software will enable new analyses of eukaryotic data sets where introgression is suspected , and will further help shed light on the Tree of Life—or , Network of Life . Let be a set of aligned genomes , and denote the site in the alignment ( if we view the alignment as a matrix where the rows are the genomes and the columns are the sites , then is the column in the matrix ) . Since the genomes are aligned , every has evolved down a local genealogy , and since we assume that hybridization has occurred , each local genealogy has evolved within the branches of a parental tree . This is illustrated in Fig . 2 . It is important to note that for each , any tree could be the local genealogy . That is , if we denote by the set of rooted binary trees on leaves , then for each , it is the case that , for every tree along with its branch lengths . However , the set of parental species trees is always constrained by the actual evolutionary history of species . For example , in Fig . 2 , only the two shown trees and are the possible parental species trees . Given a set of aligned genomes , each of length , and a set of parental species trees , we define a set of random variables each of which takes values in the set . We are now in position to define the problem for which we provide a solution: ( 1 ) for every and . Once this problem is solved and the method is run on a set of aligned genomes , we will be able to deduce the evolutionary history of every site , thus answering questions such as ( 1 ) which regions in the genomes are of introgressive descent ( these would be the ones whose parental species tree , for the example in Fig . 2 , is ; ( 2 ) is there recombination within introgressed regions ( these would be indicated by switching among local genealogies in a region yet all genealogies evolved within ) ; and , ( 3 ) what is the distribution of lengths of introgressed regions . Let us consider the scenario of Fig . 2 , where only one individual is sampled per species . We propose a hidden Markov model ( HMM ) for modeling the evolution of the three genomes . The HMM for this simple case would consist of 7 states: a start state , and six additional states: ( ) , corresponding to three possible local genealogies within parental tree , and ( ) , corresponding to three possible local genealogies within parental tree . We denote by and the local genealogies to which states and correspond , respectively; see Fig . 3 . In this model , transition between two states or two states corresponds to switching across recombination breakpoints . The probabilities of such transitions have to do with population parameters ( e . g . , population size , recombination rates , etc . ) . Transition from a state to an state indicates entering a introgressed region , while transition from an state to a state indicates exiting an introgressed region . The probabilities of such transitions have to do , in addition , with introgression and evolutionary forces ( back-crossing , selection , etc . ) . Each state emits a triplet of letters that corresponds to a column in the three-genome sequence alignment . The probability of emitting such a triplet can be computed using a standard phylogenetic substitution model [20] . Following the approaches of [12] , [21] , the transition probabilities in our model do not represent parameters in an explicit evolutionary model of recombination and introgression . Our choice was made to ease analytical representation and to permit tractable computational inference . We contrast our choice with alternative approaches: examples include ( in order of increasing tractability of computational inference at the cost of more simplifying assumptions ) methods incorporating the coalescent-with-recombination model [22] , the sequentially Markovian coalescent-with-recombination model [14] ( which adds the single assumption that coalescence cannot occur between two lineages that do not share ancestral genetic material ) , and the discretized sequentially Markovian coalescent-with-recombination model [23] ( which additionally discretizes time ) . Assuming that the probability of a site ( or locus ) in the genome of B being introgressed ( in this case , inherited from C ) is , we follow the model of [5] and use this parameter to constrain the transition probabilities . Furthermore , we capture topological changes in local genealogies due to recombination using parameters —the probability of switching from a local genealogy congruent with its containing parental tree to one that is incongruent—and —the probability of switching from a gene genealogy incongruent with its containing parental tree to one that is congruent . Finally , we model incomplete lineage sorting by allowing every local genealogy with the probability of observing it given its containing parental tree [24] . For example , assume a site is emitted by state and consider the next site . If the next site is in an introgressed region , the HMM should switch , with probability , to an state . If the next site is not in an introgressed region , then the HMM should stay in the states , with probability , and the next HMM state depends upon whether or not the two sites are separated by a recombination breakpoint that causes a change in local genealogy incongruence ( with respect to the containing parental tree ) : if they are , then the HMM should switch from state to a different state ( ) with probability ; otherwise , the HMM should stay in state with probability . Thus , the transition probability from to any other ( ) state is and to any ( ) state is , where is either or depending on whether or not the HMM transition corresponds to a change in local genealogy incongruence , is the probability of genealogy 's topology given the parental tree in , and is the probability of genealogy 's topology given the parental tree . The quantities are computed under the coalescent using the technique of [24] . If we denote by the set of ( non-start ) states , then a transition from the start state to a state occurs according to the the normalized gene tree probability For such that and correspond to the same parental tree , let . Furthermore , for , let . Then , the full transition probability matrix , with rows labeled from top to bottom , and similarly for columns ( from left to right ) , is Given that and for every pair of indices and , it follows that the entries in each row of the matrix add up to . Further , the HMM always starts in state ; that is the initial state probability distribution is given by for state and for every other state . Once in a state , the HMM emits an observation , which is a vector in the genomic sequence alignment . Emissions occur according to a substitution model ( we used the generalized time-reversible ( GTR ) model [25] ) , yielding the emission probability where are the branch lengths of the gene tree associated with state . ( It is straightforward to extend our model to other substitution models , including models nested within the GTR model and the GTR+ model , where is an additional parameter for rate variation across sites . ) Modeling a phylogenetic network in terms of a set of parental trees fails for most cases [26] . For example , if two individuals are sampled from species B in Fig . 1 , then one allele of a certain locus in one individual may trace the left parent ( to C ) , while another allele of the same locus but in the other individual may trace the right parent ( to A ) . Neither of the two parental trees in Fig . 3 can capture this case . Similarly , if one individual is sampled per species , but multiple introgression events occur or divergence events follow the introgression , the concept of parental trees collapses [5] . To deal with the general case—where multiple introgressions could occur , multiple individuals could be sampled , and introgressed species might split and diverge ( and even hybridize again later ) —we propose the following approach that is based on MUL-trees [5] . The basic idea of the method is to convert the phylogenetic network into a MUL-tree and then make use of some existing techniques to complete the computation on instead of on . A MUL-tree [27] is a tree whose leaves are not uniquely labeled by a set of taxa . Therefore , alleles of individuals sampled from one species , say , can map to any of the leaves in the MUL-tree that are labeled by . For network on taxa , we denote by the set of alleles sampled from species ( ) , and by the set of leaves in that are labeled by species . Then an allele mapping is a function such that if , and , then [5] . Fig . 4 shows an example of converting a phylogenetic network into a MUL-tree along with all allele mappings when a single allele is sampled per species . The branch lengths and inheritance probabilities are transferred from the phylogenetic network to the MUL-tree in a straightforward manner ( see [5] for details ) . Now , two changes to the PhyloNet-HMM given for the simple case above are required . While in the simple case above , we used two classes of states ( the and states ) , in the general case , the PhyloNet-HMM will contain classes of states , where is the number of all possible allele mappings . As above , the transitions within a class of states corresponds to local phylogeny switching due to recombination and ILS , whereas transitioning between classes corresponds to introgression breakpoints . Second , the probability of observing a genealogy's topology given a containing parental tree is now computed using the method of [5] , since the methods of [24] , [28] are not applicable to MUL-trees . We used a hill-climbing heuristic to infer model parameters that maximize the likelihood of the model . Here , the model consists of Notice that the values are completely determined by the parental tree branch lengths and gene tree topology; hence , they are not free parameters in this model . The standard forward and backward algorithms [17] were used to compute the model likelihood for fixed . We used Brent's method [18] as a univariate optimization heuristic during each iteration of the hill-climbing search heuristic . To reduce the possibility of overfitting during optimization , branch length parameters were optimized for each topologically distinct parental tree , and similarly for each topologically distinct unrooted gene genealogy ( since we use a reversible substitution model ) . States therefore "shared'' branch length parameters based on topological equivalence of parental trees and gene genealogies . To evaluate the effectiveness of our optimization heuristic , we utilized different starting points for the model inference phase . We found that our heuristics were robust to the choice of starting point since the searches all converged to the same solution ( data not shown ) . We found that the choice of starting point only affected search time . After model parameter values were inferred , Viterbi's algorithm [17] was used to compute optimal state paths and , thus , annotations of the genomes . More formally , using Viterbi's algorithm , we computed Further , we used the forward and backward algorithms to conduct posterior decoding and assess confidence for the states on a path : where is the probability of the observed sequence alignment up to and include column , requiring that ( computable with the forward algorithm ) ; is the probability of the last columns ( is the total number of columns in the alignment ) , requiring that ( computable with the backward algorithm ) ; and , is the probability of the alignment ( computable with either the forward or backward algorithms ) . In the Results section , we show results based on both the optimal path , , as well as posterior decoding , as the latter provides the probabilities in Eq . ( 1 ) in the problem formulation above . To evaluate the performance of PhyloNet-HMM in scenarios where the true history of evolutionary events are known , we simulated data under the coalescent model [29] with recombination , isolation , and migration [22] using ms [30] . The specific model used for our simulation ( Fig . 5 ) is based upon the consensus phylogeny for the species in our empirical study [31] , to which we added migration processes . It is important to note that the model differs in one aspect compared to the one in the empirical study: the empirical data sets were sampled so that one Mus musculus sample had a very low chance of being introgressed , whereas both M samples in the simulation may be involved in introgression . The simulation conditions were based upon consensus estimates from relevant prior literature ( summarized in Table 1 ) . We used a divergence time between in-group taxa of 1 . 5 Mya , generation time of 2 generations per year , and an effective population size of 50 , 000 , which implies divergence time between the M and S populations . The outgroup population split from the ancestral population of A and B at time . We used a cross-over rate , corresponding to cM/Mb ( compare with the cM/Mb reported for mice and the cM/Mb reported for humans [32] ) . We explored multiple migration scenarios hypothesizing either no migration ( ) or migration at one of two different rates ( or ) . For scenarios including migration , we utilized two different sets of relatively recent migration times ( either between and or between and ) compared to the divergence time between A and B . Finally , substitutions occurred according to , corresponding to substitutions/site/year based on the estimate above ( compared with substitutions/site/year reported by [33] ) . A simulation condition consisted of a setting for each simulation parameter ( in units , as required by ms [30] ) . For each condition , we repeated simulation to produce twenty replicate datasets per condition . The simulation of an individual dataset proceeded in two steps . First , ms was used to simulate local gene genealogies given the the coalescent model specified by the simulation condition . Then , using seq-gen [34] , DNA sequence evolution was simulated on each local genealogy under the Jukes-Cantor model of substitution [35] . Sequences were simulated with total length of 100 kb distributed across the local genealogies . Our study utilizes six mice that were either newly sampled or from previous publications . Details for the six mice are listed in Table 2 . Newly sampled mice were obtained as part of a tissue sharing agreement between Rice University and Stefan Endepols at Environmental Science , Bayer CropScience AG , D-40789 Monheim , Germany and Dania Richter and Franz-Rainer Matuschka at Division of Pathology , Department of Parasitology , Charité-Universitätsmedizin , D-10117 Berlin , Germany ( reviewed and exempted by Rice University IACUC ) . The M . m . domesticus data set was constructed as follows . We included a wild M . m . domesticus sample from Spain , part of the sympatry region ( i . e . , where the species co-occur geographically ) between M . m . domesticus and M . spretus . To help maximize genetic differences as part of the design goals of our pipeline , we also selected a "baseline'' M . m . domesticus sample that originated from a region as far from the sympatry region as possible . Thus , we chose a mouse from the country of Georgia in Asia where M . spretus does not occur , and , presumably , M . m . domesticus there are ancestral to those M . m . domesticus that are part of derived populations in Western Europe , including Spain , and that encountered M . spretus during their westward dispersal . We utilized two M . spretus samples . The samples came from different parts of the sympatry region in Spain . The M . m . musculus control data set contained two wild M . m . musculus samples from China and the above two M . spretus samples . The Mouse Diversity Array was used to obtain the empirical data used in our study [36] . Data for previously published samples were obtained from [31] , [37] , [38] . Since the probe sets in these studies differed slightly , we used the intersection of the probe sets in our study . A total of 535 , 988 probes were used . We genotyped all raw reads using MouseDivGeno version 1 . 0 . 4 [38] . We utilized a threshold for genotyping confidence scores of 0 . 05 . We phased all genotypes into haplotypes and imputed bases for missing data using fastPHASE [39] . Less than 15 . 1% of genotype calls were heterozygous or missing and thus affected by the fastPHASE analysis . The genotyping and phasing analyses were performed with a larger superset of samples . The additional samples consisted of the 362 samples used in [38] that were otherwise not used in our study . After genotyping and phasing was completed , we thereafter used only the samples listed in Table 2 in the Appendix . Genomic coordinates and annotation in our study were based on the GRCm38 . p2 reference genome ( GenBank accession GCA_000001635 . 4 ) . MouseDivGeno also makes use of data from the MGSCv37 reference genome ( GenBank accession GCA_000001635 . 1 ) . We evaluated the performance of PhyloNet-HMM using simulated data sets . Here , we focus on results concerning inferred probabilities ( computed using Eq . ( 2 ) ) on simulations with different migration processes . In Fig . 6 , we plot the percentage of sites for which ( is computed using Eq . ( 2 ) ) as a function of the migration rate . For the isolation-only model ( ) , the method effectively infers no introgression for any of the sites . For the isolation-with-migration models ( ) , the inferred percentages of introgressed sites were greater than zero and increased as a function of the migration rate . A potentially more informative comparison would be between the inferred percentages of introgressed sites and the percentages of sites in the simulation that involved migrant lineages . However , the simulation software that we used does not support annotating lineages in this way , nor is it a simple task to modify it to achieve this goal . ( Furthermore , as noted above , we were unable to exactly simulate evolution under the evolutionary scenario in the empirical study since the simulation software did not permit us to constrain lineage evolution so that one of the samples from population A was not introgressed . ) On the other hand , for all simulated sites , the simulation software outputs the simulated gene genealogy under which the site evolved , along with branch lengths in coalescent units . This output from simulation can be used to obtain lower bounds on the true percentage of introgressed sites . Specifically , if a site evolved under a gene genealogy where one of the two A lineages and any subset of the B lineages are monophyletic and the lineages have a simulated coalescence time greater than and smaller than , then migration must have occurred for those lineages to coalesce in that time span , based on the model used for simulation ( Fig . 5 ) . As shown in Fig . 6 , for all simulated model conditions , the introgression frequency reported by PhyloNet-HMM is greater than or equal to lower bounds on the true introgression frequency , obtained using this observation . Clearly , when the duration of the migration period increases , the variation in the estimates of our method increases , which results in a pattern that seemingly does not change from migration rate to . However , it is important to note that the extent of variability in this case precludes making a conclusion on the lack of increase in the percentage of sites . Nonetheless , the important message here is that the estimates of our method start varying more as the duration of the migration period increases . We also found that the probability of observing a gene genealogy conditional on a containing parental tree differed between the two parental trees ( results not shown ) . Under all simulation conditions , the inferred gene tree distribution ( conditional on the containing parental tree ) had multiple genealogies with non-trivial posterior decoding probabilities , suggesting that within-row transitions were capturing switching in local genealogies due to ILS . That is , the simulated data sets clearly had evidence of incongruence due to both introgression and ILS . Finally , Fig . 7 and Fig . 8 show that in training our PhyloNet-HMM model on the simulated data , base frequencies were accurately estimated at 0 . 25 ( which are the base frequencies for all four nucleotides we used in our simulations ) and substitution rates were estimated generally between and ( we used in our simulations ) . Further , the results were robust to the migration rates and durations of migration periods . We applied the PhyloNet-HMM framework to detect introgression in chromosome 7 in three sets of mice , as described above . Each data set consisted of two individuals from M . m . domesticus and two individuals from M . spretus . Thus the phylogenetic network is very simple , and has only two leaves , with a reticulation edge from M . spretus to M . m . domesticus; see Fig . 9 ( a ) . As we discussed above , the evolution of lineages within the species network can be equivalently captured by the set of parental trees in Fig . 9 ( b-c ) . Since in each data set we have four genomes , there are 15 possible rooted gene trees on four taxa . Therefore , for each data set , our model consisted of 15 states , 15 states , and one start state , for a total of 31 states . We use our new model and inference method to analyze two types of empirical data sets . The first type includes individuals of known introgressed origin , and our model recovers the introgressed genomic region reported in [2] ( Fig . 10 ) . On the other hand , the second type consists of "control" individuals collected from geographically distant regions so as to minimize the chances of introgression ( though , it is not possible to rule out that option completely ) . Our model detected no regions of introgressive descent in this dataset ( Fig . 11 ) . We ran PhyloNet-HMM to analyze the M . m . domesticus data set , which consisted of samples from a putative hybrid zone between M . m . domesticus and M . spretus ( Fig . 10 ) . The data set covered all of chromosome 7 , the chromosome containing the Vkorc1 gene . Vkorc1 is a gene implicated in the introgression event and the spread of rodenticide resistance in the wild [2] . Based on the pattern of recovered parental trees , the PhyloNet-HMM analysis detected introgression in the vicinity of the Vkorc1 gene from approximately 123 . 0 Mb to 130 . 8 Mb , reproducing the findings of [2] . The presence of the introgression in the M . m . domesticus sample from mainland Spain but not the one from the country of Georgia suggests that the putative introgression may be polymorphic; preliminary results on additional Spanish samples ( not shown ) support this hypothesis . The analysis also uncovered recombination and incomplete lineage sorting in the region , as evidenced by incongruence among the rooted gene genealogies that were ascribed to loci . The PhyloNet-HMM analysis detected introgression in 8 . 9% of sites in chromosome 7 , containing over 300 genes . Notably , the analysis located similar regions in other parts of chromosome 7 which were not investigated by prior studies such as [2] . Examples include the region from 107 . 7 Mb to 108 . 9 Mb and the region from 115 . 2 Mb to 117 . 6 Mb . It is worth mentioning that the method does detect ILS within introgressed regions and outside those regions as well; yet , it does not switch back and forth between these two cases repeatedly ( which is an issue that plagues methods that assume independence across loci ) . As described by our model above , if we sum the transition probabilities from any state to all states , we obtain a value for . We performed this computation for each state , and took the average of all estimates based on each of the 15 states . Our model estimates the value of as . This can be interpreted as the probability of switching due to introgression , and can shed light on introgression parameters . The posterior decoding probabilities , based on Eq . ( 2 ) , for all positions in chromosome 7 , are shown in Fig . 10 ( a ) . Clearly , the introgressed regions indicated by green bars in Fig . 10 ( d ) have very high support ( close to 1 ) , particularly the region around the Vkorc1 gene . To further validate our approach , we repeated our scans on the M . m . musculus control data set ( Fig . 11 ) , which contained two sets of genomes of mice that are not known to hybridize . The first set of mice consisted of the M . spretus samples from the previous scan , and the second set of mice consisted of geographically and genetically distinct samples from M . m . musculus , which is not known to hybridize with M . spretus in the wild . PhyloNet-HMM did not detect introgression on the control data set . The analysis recovered signatures of ILS , though , based on local incongruence among inferred rooted gene genealogies . In this paper , we introduced a new framework , PhyloNet-HMM , for comparative genomic analyses aimed at detecting introgression . Our framework allows for modeling point mutations , recombination , and introgression , and can be trained to tease apart the effects of incomplete lineage sorting from those of introgression . We implemented our model , along with standard HMM algorithms , and analyzed an empirical data set of chromosome 7 from mouse genomes where introgression was previously reported . Our analyses detected the reported introgression with high confidence , and detected other regions in the chromosome as well . Using the model , we estimated that about 9% of the sites in chromosome 7 of an M . m . domesticus genome are of introgressive descent . Further , we ran an empirical analysis on a negative control data set , and detected no introgression . On simulated data , we accurately detected introgression ( or the lack thereof ) and related statistics from data sets generated under both isolation-with-migration and isolation-only models . We described above how to extend the model to general data sets with arbitrary hybridization and speciation events , by using a MUL-tree technique . However , as larger ( in terms of number of genomes ) data sets become available , we expect the problem to become more challenging , particularly in terms of computational requirements . Furthermore , while the discussion so far has assumed that the set of states is known ( equivalently , that the phylogenetic network is known ) , this is not the case in practice . This is a very challenging problem that , if not dealt with carefully , can produce poor results . In this work , we explored a phylogenetic network corresponding to a hypothesis provided by a practitioner . In general , the model can be "wrapped" by a procedure that iterates over all possible phylogenetic network hypotheses , and for each one the model can be learned as above , and then using model selection tests , an optimal model can be selected . However , this is prohibitive except for data sets with very small numbers of taxa . As an alternative , the following heuristic could be adopted instead: first , sample loci across the genome that are distant enough to guarantee that they are unlinked; second , use trees built on these loci to search for a phylogenetic network topology using techniques described in [40]; third , conduct the analysis as above . Of course , the phylogenetic network identified by the search might be inaccurate , in which case the use of an ensemble of phylogenetic networks that are close to that one in terms of optimality may be beneficial .	Hybridization is the mating between individuals from two different species . While hybridization introduces genetic material into a host genome , this genetic material may be transient and is purged from the population within a few generations after hybridization . However , in other cases , the introduced genetic material persists in the population—a process known as introgression—and can have significant evolutionary implications . In this paper , we introduce a novel method for detecting introgression in genomes using a comparative genomic approach . The method scans multiple aligned genomes for signatures of introgression by incorporating phylogenetic networks and hidden Markov models . The method allows for teasing apart true signatures of introgression from spurious ones that arise due to population effects and resemble those of introgression . Using the new method , we analyzed two sets of variation data from chromosome 7 in mouse genomes . The method detected previously reported introgressed regions as well as new ones in one of the data sets . In the other data set , which was selected as a negative control , the method detected no introgression . Furthermore , our method accurately detected introgression in simulated evolutionary scenarios and accurately inferred related population genetic quantities . Our method enables systematic comparative analyses of genomes where introgression is suspected , and can work with genome-wide data .	[ "Abstract", "Introduction", "Materials", "and", "Methods", "Results/Discussion", "Conclusions" ]	[ "genetics", "biology", "and", "life", "sciences", "genomics", "evolutionary", "biology", "computational", "biology" ]	2014	An HMM-Based Comparative Genomic Framework for Detecting Introgression in Eukaryotes
Although the SLX4 complex , which includes structure-specific nucleases such as XPF , MUS81 , and SLX1 , plays important roles in the repair of several kinds of DNA damage , the function of SLX1 in the germline remains unknown . Here we characterized the endonuclease activities of the Caenorhabditis elegans SLX-1-HIM-18/SLX-4 complex co-purified from human 293T cells and determined SLX-1 germline function via analysis of slx-1 ( tm2644 ) mutants . SLX-1 shows a HIM-18/SLX-4–dependent endonuclease activity toward replication forks , 5′-flaps , and Holliday junctions . slx-1 mutants exhibit hypersensitivity to UV , nitrogen mustard , and camptothecin , but not gamma irradiation . Consistent with a role in DNA repair , recombination intermediates accumulate in both mitotic and meiotic germ cells in slx-1 mutants . Importantly , meiotic crossover distribution , but not crossover frequency , is altered on chromosomes in slx-1 mutants compared to wild type . This alteration is not due to changes in either the levels or distribution of double-strand breaks ( DSBs ) along chromosomes . We propose that SLX-1 is required for repair at stalled or collapsed replication forks , interstrand crosslink repair , and nucleotide excision repair during mitosis . Moreover , we hypothesize that SLX-1 regulates the crossover landscape during meiosis by acting as a noncrossover-promoting factor in a subset of DSBs . Genomic DNA is subjected to a variety of endogenous and exogenous sources of damage . To repair this DNA damage , the structure-specific endonucleases function to cleave branched DNA structures such as Y forks , 5′-flaps , 3′-flaps , stem loops , bubbles , replication forks ( RFs ) and Holliday junctions ( HJs ) . SLX-1 , a structure-specific nuclease that is highly conserved in eukaryotes , harbors both GIY-YIG-type endonuclease and PHD-type zinc finger domains . Furthermore , SLX1 requires the regulatory subunit SLX4 to perform its nuclease activities in both yeast and humans [1]–[5] . Specifically , the budding yeast Slx1-Slx4 complex , co-purified from Escherichia coli , shows endonuclease activity towards Y forks , 5′-flaps , RFs and HJs [1] . The immunoprecipitation products of fission yeast Slx1-TAP cleave stem loops and HJs [2] . Finally , the human SLX1-SLX4 complex , purified from both human cells and E . coli , cleaves Y forks , 3′-flaps , 5′-flaps , RFs , stem loops and HJs [3]–[5] . However , it is not known whether these endonuclease activities of SLX1 are conserved in other species such as flies and worms . slx1 was first identified in a synthetic-lethal screen for genes required for the viability of cells lacking Sgs1 , the budding yeast RecQ helicase [6] . Sgs1 functions as an anti-recombinase by unwinding and dissolving toxic recombination intermediates , thereby maintaining genome stability [7] . slx1 deletion ( slx1Δ ) mutants do not exhibit lethality , DNA damage sensitivity or sterility in either budding or fission yeast [2] , [6] , [8] . However , they exhibit defects in the completion of rDNA replication [2] , [9] , [10] . Moreover , in humans , the siRNA based-depletion of SLX1 increases both the endogenous and exogenous levels of DNA damage resulting from exposure to ionizing radiation ( IR ) , camptothecin ( CPT ) and DNA interstrand crosslinking agents [3]–[5] , [11] . In humans , SLX4 forms a complex with three structure-specific nucleases , SLX1 , XPF-ERCC1 and MUS81-EME1 [3]–[5] . Moreover , individuals carrying mutations in SLX4 exhibit symptoms of Fanconi anemia , a syndrome characterized by chromosomal instability in humans [12] . A mouse knockout of Slx4 also shows chromosomal instability phenotypes similar to those of Fanconi anemia in humans [13] . In budding yeast , Slx4 binds to Slx1 and Rad1XPF in a mutually exclusive manner [6] , [14] , [15] . We showed that HIM-18 , the SLX4 homolog in C . elegans , interacts with SLX-1 and XPF-1 [16] . Furthermore , we found that HIM-18/SLX-4 ( referred to herein as HIM-18 ) and XPF-1 are required for wild type levels of meiotic crossover formation [16] . However , whether SLX-1 is required for meiotic crossover formation remains unknown . Moreover , whether SLX1 , XPF and MUS81 function as structure-specific endonucleases either in the same or in different complexes in vivo is also unclear . Here we show that Caenorhabditis elegans SLX-1 cleaves branched DNA substrates in a HIM-18/SLX-4-dependent manner in vitro . Furthermore , slx-1 ( tm2644 ) mutants , which encode for a catalytically inactive ( nuclease negative ) protein , show accumulation of recombination intermediates in both mitotic and meiotic cells as well as increased sensitivity to DNA damage inducing agents . These results suggest that SLX-1 is required for DNA repair by processing repair intermediates through its nuclease activity . However , while double Holliday junction resolution is required for crossover formation during meiosis , meiotic crossover frequencies were not reduced in slx-1 ( tm2644 ) mutants , and instead , crossover distribution was altered compared to wild type . Therefore , although SLX-1 is not an essential nuclease for crossover formation via double Holliday junction resolution between homologous chromosomes during meiotic recombination , our studies reveal that SLX-1 plays a role in regulating crossover distribution . This regulation is not mediated through changes in either the levels or distribution of DNA double-strand breaks ( DSBs ) along chromosomes . Instead , we propose a model in which SLX-1 regulates meiotic crossover distribution such that they occur at the terminal thirds of chromosomes by promoting DSB repair via a noncrossover pathway along the mid-section of chromosomes . We previously showed that C . elegans SLX-1 interacts with HIM-18 in a yeast two-hybrid assay [16] . To examine whether these two components directly interact with one another , we transiently transfected HEK-293T cells with epitope-tagged HIM-18 and SLX-1 and performed immunoprecipitation experiments . As shown in Figure 1A–1C , full-length HA-SLX-1 associates with full-length Myc-HIM-18 , but not with a control Myc tagged protein ( Myc-GFP ) under the tested conditions . To further characterize the interaction between SLX-1 and HIM-18 , we co-expressed the N-terminal domain of SLX-1 that contains its nuclease domain ( HA-SLX-1-N; residues 1–272 , Figure 1A ) with Myc-HIM-18 in 293T cells . In contrast to what was observed with HA-SLX-1 ( full-length ) , Myc-HIM-18 does not associate with HA-SLX-1-N , consistent with yeast two-hybrid results [3] . We also ascertained that the C-terminal domain of SLX-1 that contains its PHD domain ( SLX-1-C; residues 273–443 ) does not interact with HIM-18 ( data not shown ) . These results suggest that the interaction between HIM-18 and SLX-1 is direct and involves both the N-terminal and C-terminal domains of SLX-1 . To examine the roles of HIM-18 and SLX-1 during recombination , we assessed if the HIM-18/SLX-1 complex displayed endonucleolytic activity towards synthetic DNA substrates . HA-HIM-18-CCD ( conserved C-terminal domain , residues 548–718 ) and Myc-SLX-1 were co-expressed in 293T cells and immuno-purified with an α-HA antibody ( the CCD domain of HIM-18 was used since it expressed at a higher level than full-length HIM-18 ) . The HA-HIM-18/Myc-SLX-1 complex was incubated with either a radiolabeled replication fork ( RF ) or a Holliday Junction ( HJ ) substrate , and the reaction products were separated by native gel electrophoresis and visualized by autoradiography . HA-HIM-18/Myc-SLX-1 exhibited endonucleolytic activity against both RFs and HJs at a level that was comparable to the human HA-SLX4-ΔN complexes purified from 293T cells ( Figure 1D and 1E ) [5] . Under similar conditions , neither HIM-18 nor SLX-1 alone displayed appreciable catalytic activity against RFs and HJs ( Figure 1D and 1E ) , indicating that SLX-1 is the catalytically active component of the HIM-18/SLX-1 complex . To further characterize its processing activity , we analyzed the substrate preference of HIM-18/SLX-1 and compared its activity against 5′-flap , HJ , and RF substrates . As shown in Figure 1F–1G , time-course experiments revealed that HA-HIM-18/Myc-SLX-1 preferred the RF substrate to the 5′-flap or HJ substrates under the conditions tested . We also observed that HIM-18/SLX-1 had substantially lower activity against the 3′-flap substrate compared to RFs , 5′-flaps , or HJs ( Figure S1 ) . Next , we determined the cleavage sites of each DNA substrate ( Figure 2 ) . We determined that HA-HIM-18/myc-SLX-1 cleaved the RF substrate on strand 1 at 2 nucleotides 3′ to the branch point ( Figure 2A and 2D ) . However , in the case of RFs , the cleavage efficiency against the double stranded region ( strand 1 ) was significantly higher than against the same region of the 5′-flap ( Figure 2B ) . In addition , it should be noted that although human HA-SLX4-ΔN purified from 293T cells had a cleavage specificity against RF that was significantly different to that of HA-HIM-18/myc-SLX-1 , recombinant human SLX1/SLX4 ( SBD ) generated RF cleavage products that included the 32 nucleotide product ( two nucleotides 3′ to the branch point ) that was also produced by the C . elegans protein complex ( Figure 2A ) . HA-HIM-18/myc-SLX-1 purified from 293T cells cleaved the 5′-flap on strand 3 , precisely at the junction between the 5′single-stranded arm and the double stranded region of the substrate similar to 5′-flap cleavage by Saccharomyces cerevisiae Slx1-Slx4 [1] . Cleavage also occurred on strand 1 ( i . e . the double-stranded region of the 5′-flap ) , although with much less efficiency than strand 4 , at 2 nucleotides 3′ to the branch point ( Figure 2B and 2D ) . Since HIM-18/SLX-1 displayed significant cleavage activity against HJs ( Figure 1D and 1E ) , we sought to determine if it was in fact a canonical HJ resolvase . A characteristic of bona fide HJ resolvases such as bacterial RuvC , human GEN1 and human SLX1/SLX4 , is the ability to cleave opposing strands of a HJ in a symmetric manner to generate products that can be directly ligated [3]–[5] , [17] , [18] . To test this possibility , we performed cleavage assays against a HJ substrate radiolabeled at either strands 1 or 3 and analyzed the reaction products by denaturing gel electrophoresis . As shown in Figure 2C and 2D , HA-HIM-18/myc-SLX-1 cleaved the HJ substrate at a unique site on strand 1 , which was two nucleotides 3′ to the branch point . On strand 3 , however , HA-HIM-18/myc-SLX-1 displayed substantially lower cleavage activity compared to strand 1 and cut the HJ at two sites , respectively 1 and 3 nucleotides 3′ to the branch point . This was in contrast to the human HA-SLX4-ΔN complex purified from 293T cells , which cleaves the HJ with perfect symmetry on strands 1 and 3 two nucleotides 3′ to the branch point ( Figure 2C and 2D ) [5] . These data indicate that the C . elegans HIM-18/SLX-1 complex , though displaying cleavage activity against a HJ substrate , does not appear to function as a bona fide HJ resolvase and is reminiscent of S . cerevisiae and Schizosaccharomyces pombe Slx1-Slx4 endonucleases [1] , [2] under the tested conditions . The slx-1 ( tm2644 ) mutant , obtained from the Japanese National Bioresource Project , carries a 205 bp deletion encompassing parts of intron 4 and exon 5 which removes the splice acceptor site of the downstream exon 4 ( Figure 3A ) . RT-PCR using a primer set located between the start and stop codons of the slx-1 gene revealed that there are several splice variants containing premature stop codons in slx-1 ( tm2644 ) mutants ( Figure 3C ) . Sequence analysis of the RT-PCR products revealed that this alternative splicing does not occur in wild type ( Figure 3B ) . SLX-1 harbors both a GIY-YIG nuclease domain and a PHD finger domain ( Figure 3B ) . 42% of the splice variants lack both the nuclease and PHD finger domains ( Figure 3D ) , whereas 23% contain an intact PHD finger domain ( exons 6 and 7 ) and 35% contain an intact nuclease domain ( exons 3 and 4 ) ( Figure 3D ) . Both exons 3 and 4 carry the conserved catalytic sites of the nuclease ( R202 in exon 3 and E243 in exon 4 ) . Moreover , both catalytic sites have been shown to be essential for the nuclease activity of SLX1 both in fission yeast ( R34 and E74 ) [2] and in humans ( R41 and E82 ) [3] . Importantly , the nuclease activity of SLX-1 requires a physical interaction with HIM-18 , which is a homolog of human SLX4 ( Figure 1D and 1E ) . However , yeast two-hybrid and immunoprecipitation assays revealed that SLX-1N1–272 , which is potentially expressed in slx-1 ( tm2644 ) mutants and stems from the only splice variant still carrying the nuclease domain , does not bind to HIM-18 and lacks a nuclease activity ( Figure 1A–1E and Figure 3E ) . Therefore , these results suggest that slx-1 ( tm2644 ) mutants are loss-of-function for SLX-1's nuclease activity . To investigate whether slx-1 plays a role in either mitotic development or meiosis , and examine the genetic interactions between slx-1 and other genes implicated in processing recombination intermediates during these cell division programs , we measured the brood size , embryonic lethality , larval arrest and the incidence of males observed among slx-1 , him-6/BLM , xpf-1 and gen-1 mutant offspring ( Table 1 ) . A decreased brood size is suggestive of increased sterility , whereas either increased embryonic lethality or larval arrest are suggestive of mitotic defects . Finally , a high incidence of males ( Him phenotype ) is indicative of increased X chromosome nondisjunction and correlates with meiotic defects , whereas a combination of increased embryonic lethality accompanied by a high incidence of males is suggestive of increased aneuploidy resulting from meiotic missegregation of both autosomes and the X chromosome , respectively [19] . slx-1 mutants showed a 32% reduction in brood size ( P<0 . 0001 ) , a 4 . 6-fold increase in embryonic lethality ( P = 0 . 0006 ) , and a 2-fold increase in larval arrest ( P = 0 . 00197 ) compared to wild type , supporting a role for slx-1 in mitotic repair . Moreover , a 1 . 2-fold increase in larval arrest ( P = 0 . 0417 ) was observed in slx-1;him-18 double mutants compared to him-18 single mutants . This marginally lower significance cutoff most likely indicates that SLX-1 is dependent on HIM-18 during larval development . However , we cannot rule out the possibility that SLX-1's function is not completely HIM-18-dependent during this developmental process . Next , we investigated the genetic interaction of SLX-1 with HIM-6 , a C . elegans homolog of yeast Sgs1 and the human Bloom syndrome helicase , which is required for double Holliday junction dissolution [7] . slx-1;him-6 double mutants showed synthetic lethality compared to either single mutant ( P<0 . 0001 ) . These results suggest that SLX-1 and HIM-6 function in parallel or alternate pathways , similar to budding yeast [6] and flies [20] . Finally , we examined the genetic interactions between slx-1 and the structure-specific endonucleases xpf-1 and gen-1 , which are a component of the HIM-18 complex and a Holliday junction resolving enzyme , respectively [16] , [21] . slx-1;xpf-1 double mutants showed a 2 . 4-fold increase ( P<0 . 0001 ) in embryonic lethality and a 3 . 6-fold increase in larval arrest ( P = 0 . 0029 ) compared with xpf-1 single mutants . These results suggest that SLX-1 and XPF-1 may have overlapping roles during mitotic development . Interestingly , slx-1;xpf-1;him-18 triple mutants exhibited similar phenotypes to xpf-1;him-18 double mutants , supporting our observation that the nuclease activity of SLX-1 is dependent on HIM-18 and therefore an slx-1 mutation causes no obvious plate phenotypes in an xpf-1;him-18 background . Analysis of slx-1;gen-1 double mutants revealed a 69% reduction in brood size ( P<0 . 0001 ) and a 5 . 5-fold increase in larval arrest ( P = 0 . 0079 ) compared with slx-1 single mutants . These results suggest that SLX-1 and GEN-1 share similar , albeit independent , mitotic roles at least during larval development . Homologous recombination provides for the repair of both spontaneous DSBs , stemming from stalled or collapsed replication forks at S phase , and programmed DSBs , produced by SPO-11 during prophase of meiosis I . The organization of nuclei in a temporal and spatial gradient in the C . elegans germline facilitates the identification and analysis of specific stages of both mitotic and meiotic nuclei . Specifically , nuclei at the distal tip end are undergoing mitotic proliferation ( zones 1 and 2 ) , nuclei at the transition zone are in the leptotene/zygotene stages of meiosis ( zone 3 ) , followed by nuclei in early pachytene ( zone 4 ) , mid pachytene ( zone 5 ) and late pachytene ( zones 6 and 7 ) ( Figure 4A ) . To investigate whether SLX-1 is required for the maintenance of genomic integrity in the C . elegans germline , we monitored the levels as well as the kinetics of appearance and disappearance of RAD-51 , a protein involved in strand invasion/exchange during DSB repair ( Sung , 1994; Colaiacovo et al . 2003 ) . Quantification of RAD-51 foci revealed that these were elevated in both mitotic ( zones 1 and 2 ) and meiotic ( zones 3 , 6 and 7 ) nuclei in slx-1 mutants compared to wild type ( Figure 4A and Figure S2 ) . Moreover , the mitotic RAD-51 foci persist through late meiotic prophase ( late pachytene stage; zone 6 ) as observed in slx-1;spo-11 double mutants which lack the formation of programmed meiotic DSBs ( Figure 4A and Figure S2 ) . However , simple subtraction of the number of RAD-51 foci in slx-1; spo-11 from the number observed in slx-1 , as a means of approximating the dynamics of SPO-11-dependent DSBs , also reveals an increase of meiotic recombination intermediates during late pachytene in slx-1 mutants . Further support for a role for SLX-1 in germline DNA repair stems from our analysis of germ cell apoptosis , which was increased 2 . 3-fold in slx-1 mutants compared to wild type ( Figure 4B ) . Increased germ cell apoptosis was previously shown to occur when an inability to repair DNA damage results in the activation of a DNA damage checkpoint in late pachytene [22] . Taken together , these results suggest that SLX-1 is required for the proper repair of both stalled/collapsed replication forks and meiotic DSBs . To further investigate which kinds of DNA damage are repaired by SLX-1-HIM-18 in vivo , we performed a series of DNA damage sensitivity assays by exposing slx-1 mutants to γ-irradiation , which produces DSBs , nitrogen mustard ( HN2 ) , which produces DNA inter-strand crosslinks , camptothecin ( CPT ) , which results in single-strand nicks , and UVC , which causes cyclobutane pyrimidine dimers ( Figure 5 ) . After treatment with g-irradiation , no statistically significant , reduction in hatching ratio was observed in slx-1 mutants compared to wild type ( P = 0 . 5512 and P = 0 . 1455 , respectively at 50 and 100 Gy ) ( Figure 5A ) . In contrast , hatching ratios were significantly reduced in him-18 ( P<0 . 0001 and P<0 . 0001 ) and slx-1;him-18 ( P = 0 . 0004 and P = 0 . 0002 ) double mutants after doses of either 50 or 100 Gy of irradiation ( Figure 5A ) . These results suggest that HIM-18 plays a role in the repair of exogenously induced DSBs , which is independent of the nuclease activity of SLX-1 . Exposure to HN2 revealed that slx-1 and slx-1;him-18 mutants share similar hypersensitivity to HN2 compared to wild type at both 100 µM and 200 µM , while him-18 mutants showed more severe hypersensitivity compared to slx-1 and slx-1;him-18 mutants ( Figure 5B ) . These results could be explained by the fact that SLX-1 function is HIM-18-dependent . Therefore , in him-18 mutants ICL repair may be further affected by the presence of inactive SLX-1 . Exposure to CPT revealed hypersensitivity among slx-1 mutants compared to wild type at both 500 nM and 1000 nM doses ( Figure 5C ) . Moreover , him-18 and slx-1;him-18 mutants showed more severe hypersensitivity compared to slx-1 mutants . Since CPT inhibits the removal of topoisomerase I , thereby forming nicked sites after replication [23] , these results suggest that SLX-1 is either required for efficient removal of the TOP1-CPT complex or resolution of a HJ intermediate during the re-establishment of a replication fork . Extrapolating from observations made in other species [1] , [5] , loss of HIM-18 may reduce the nuclease activity of SLX-1 , MUS-81 and potentially XPF-1 . This would explain why him-18 and slx-1;him-18 mutants show similar hypersensitivity to CPT . Notably , slx-1 , him-18 and slx-1;him-18 mutants showed hypersensitivity to UV ( Figure 5D ) , although the loss of either SLX1 or SLX4 orthologs does not result in hypersensitivity in budding yeast [8] , fission yeast [2] , flies [24] , mouse [13] and human cells [3] , [5] . Whereas mutants in several other DNA repair genes such as mus81 and sgs1 in S . cerevisiae [6] , [25] have exhibited UV sensitivity , but have no proven direct role in nucleotide excision repair ( NER ) , one can not discard the possibility that in C . elegans , unlike in other species , SLX-1 and HIM-18 may be required for NER . To investigate whether the higher levels of RAD-51 foci are due to either an increased number of DSBs or a delay of the repair process in slx-1 mutants , we quantified DSB levels by RAD-51 immunostaining in rad-54 mutants , in which DSB repair is blocked and DSB-bound RAD-51 is proposed to be trapped in the germline [26] . In wild type , RAD-51 foci start to increase in nuclei at the entrance into meiosis ( zone 3 ) , peak at 3 . 7 foci/nucleus at mid pachytene ( zone 5 ) and are practically all gone by late pachytene ( zone 7 ) ( Figure 4A , Figure 6B and Table S2 ) . In contrast , in rad-54 mutants , higher levels of mitotic RAD-51 foci were observed ( 0 . 8 and 1 . 0 compared to 0 . 1 at zones 1 and 2 , P<0 . 0001 , respectively ) , and meiotic RAD-51 foci peaked at 75 foci/nucleus at diplotene ( −7 oocyte ) , only being completely absent by late-diakinesis ( −1 oocyte ) ( Figure 6A–6B and Table S2 ) . These three observations: 1 ) elevated mitotic RAD-51 foci; 2 ) peak of RAD-51 foci at diplotene; 3 ) RAD-51 foci only being completely lost by late diakinesis , are different from those described in [26] . They concluded that RAD-51 foci peak at 12/nucleus at early , mid and late pachytene stages in rad-54 mutants . However , both our studies as well as those of others have since revealed a higher number of RAD-51 foci in this mutant background ( 18–30 and 26–63 foci at mid and late pachytene , respectively [27] , [28] , this current study ) . Further support for the number of DSBs we observed in the rad-54 background stems from our analysis of the level of RPA-1-YFP foci in brc-2 mutants , in which DSB repair is blocked and replacement of RPA-1 ( replication protein A ) by RAD-51 at the resected DSB ends is inhibited in the germline [29] ( Figure S4 ) . We observed a similar number of RPA-1-YFP foci ( 59 . 1 ) in brc-2 mutants to that of RAD-51 foci ( 62 . 9 ) in rad-54 mutants in late pachytene nuclei ( zone 7 ) . Importantly , we confirmed that the elevated levels of RAD-51 foci we observe in rad-54 mutants at mid and late pachytene , where the events of repair we are focused on take place , are not already at a possible maximum thus obscuring our ability to utilize this mutant background to identify further increases in DSB levels . Specifically , following the formation of additional DSBs by γ-irradiation in rad-54 mutants , we observed 45 foci at 50 Gy compared to 29 foci at 0 Gy in zone 5 ( P<0 . 0001 ) ( Figure S3 ) . slx-1rad-54 double mutants exhibited levels of RAD-51 foci similar to those observed in rad-54 mutants , although RAD-51 foci levels accumulated with slightly faster kinetics than in rad-54 mutants . Furthermore , it is known that elevated levels of meiotic DSBs rescue him-17 mutants , which are deficient in meiotic DSB formation [30] , [31] . slx-1;him-17 double mutants did not rescue the him-17 mutant phenotype ( Figure 6C–6D ) . Taken together , these results suggest that the total levels of DSBs are wild type in slx-1 mutants . In C . elegans , as in many other species , crossover formation is biased towards the terminal thirds of autosomes [32] , [33] . To measure the levels and distribution of DSBs along chromosomes , and determine whether they are altered in slx-1 mutants , we performed three-dimensional traces of chromosomes in late pachytene nuclei by visualizing chromosome axes with an antibody to the meiotic cohesin REC-8 and quantified the levels and distribution of recombination intermediates along these chromosome axes with a RAD-51 antibody , comparing rad-54 and slx-1 rad-54 double mutants ( Figure 6E–6F ) . To distinguish the X chromosome from the autosomes , we identified the X chromosome pairing center end with an anti-HIM-8 antibody [34] . We did not detect a biased distribution of RAD-51 foci along either the arms or the central region of the chromosomes in rad-54 mutants . Therefore , this even distribution of DSBs along the lengths of the chromosomes suggests the existence of mechanisms that inhibit crossover formation after the induction of DSBs at the central region of the chromosomes . Interestingly , both the levels and distribution of RAD-51 foci along chromosome axes were similar between rad-54 and slx-1 rad-54 mutants in both autosomes and the X chromosomes ( Figure 6F ) . These results suggest that SLX-1 does not alter either the overall levels or the distribution of DSBs along either the X chromosomes or the autosomes . To investigate whether SLX-1 and HIM-18 are required for meiotic crossover formation , we first observed crossover frequencies in slx-1 , him-18 and slx-1;him-18 mutants on both chromosomes IV and X by using the snip-SNP method [16] , [35] . Crossover frequencies were not significantly different between wild type and slx-1 mutants on either chromosome ( Figure 7A and 7B ) . However , reduced crossover frequencies were detected in slx-1;him-18 double mutants compared to him-18 single mutants . These data coincide with the observation of a lack of a Him phenotype among slx-1 mutants , whereas slx-1;him-18 mutants show a more severe Him phenotype compared to him-18 single mutants ( Table 1 ) . Taken together , these results suggest that SLX-1 is not required to make interhomolog crossovers in normal meiosis , but is partially required on both autosomes and X chromosomes in a him-18 background . Meiotic DSBs , with the exception of the subset designated to be repaired as future interhomolog crossovers , are repaired either by interhomolog noncrossover or intersister pathways [36] , [37] . We examined intersister repair by monitoring chromosome morphology in diakinesis oocytes of syp-2 and rec-8 mutants , where either interhomolog interactions or sister chromatid cohesion are impaired , respectively [38]–[40] . We did not observe any additive cytological defects in both slx-1;syp-2 and slx-1;rec-8 double mutants compared with syp-2 or rec-8 single mutants ( Figure S5 ) . While these results suggest that SLX-1 may not function during intersister repair , confirmation awaits additional analysis of sister chromatid separation at anaphase where either lagging chromosomes or chromosome bridges might be detected if there are defects in the intersister resolution of dHJs . We next used the snip-SNP method to assess crossover distribution along chromosomes III , IV , V and X . In slx-1 mutants , the frequency of crossovers detected in the center ( intervals B–C ) of the autosomes ( III , IV and V ) is higher ( 3 . 1- , 4 . 2- and 2 . 7-fold increases , respectively ) than in wild type , whereas it is reduced in the arm regions ( intervals A–B and C–D ) ( Figure 7C and Table S3 ) . Using a pair of morphological markers we further confirmed the occurrence of a higher crossover frequency at the central region of chromosome II in slx-1 mutants compared to wild type ( Figure 7D ) . Crossover distribution along the X chromosome is also different between wild type and slx-1 mutants . At the right portion of the central region of the X chromosome ( B'–C ) , the frequency of crossovers in slx-1 is higher ( 2 . 2-fold increase , P = 0 . 0348 ) than in wild type . These results suggest that SLX-1 is required for proper crossover distribution along both autosomes and the X chromosome . To further examine the role of SLX-1 in crossover distribution regulation , we measured crossover distribution on chromosomes IV and X in him-18 and slx-1;him-18 double mutants . Crossover levels are increased at the center of chromosome IV in both him-18 and slx-1;him-18 double mutants compared with wild type ( 1 . 9- and 2 . 5-fold increases , respectively ) . Moreover , while the observed increases are similar between slx-1 and slx-1;him-18 , only a moderate increase is detected in him-18 mutants on chromosome IV ( Figure 7C and Table S3 ) . Given that SLX-1 is nearly catalytically dead with regard to its nuclease activity in the absence of HIM-18 in vitro ( Figure 1D–1E ) , the nuclease activity of SLX-1 may not work in him-18 mutants in vivo . Instead , it is possible that other functions of SLX-1 , for example the PHD finger-dependent recognition of epigenetic marks , might still work in him-18 mutants . Taken together , these results suggest that the nuclease activity of SLX-1 may be important to maintain proper crossover distribution . It has been proposed that only one of the various DSB sites along a chromosome is specifically designated as a future interhomolog crossover site [41] . ZHP-3 , a homolog of S . cerevisiae Zip3 protein , has been proposed to mark crossover precursor sites during late pachytene stage in C . elegans [16] , [42] . To investigate whether crossover designation properly occurs in slx-1 mutants , we compared the numbers of ZHP-3-GFP foci present in pachytene nuclei in wild type and slx-1 mutants . The average number of ZHP-3-GFP foci observed in slx-1 mutants is 80% of those in wild type ( Figure 7E ) . Specifically , while nearly 90% of late pachytene nuclei contain six ZHP-3 foci in wild type , only 45% of nuclei contain six and 55% have less than five ZHP-3-GFP foci in slx-1 mutants ( Figure 7F ) . Given that crossover levels are indistinguishable between wild type and slx-1 mutants , these results suggest that a crossover pathway that is not associated with ZHP-3 foci exists in slx-1 mutants . Alternatively , it could be possible that SLX-1 is required for proper crossover designation . In budding yeast , the substrate preference observed for recombinant Slx1-Slx4 is 5′-flaps>Y forks>RFs>mobile HJs>3′-flaps>fixed HJs ( the latter involves an asymmetric cut given the non-ligatable processed substrate that is then detected ) ( Flott and Brill 2003 ) . In fission yeast , the Slx1 immunoprecipitation product cuts stem loops and HJs ( symmetric cut ) [2] . Finally , in humans , SLX1/SLX4 exhibits preference for 5′flaps and HJs>RFs>3′-flaps [3] , [5] . In this current study , we determined that in C . elegans , the preference of SLX-1-HIM-18 is for RFs>5′-flaps>HJs>3′-flaps ( Figure 1F–1G and Figure 2D ) . The slight difference observed in the order of substrate preference in C . elegans compared to those in yeast and humans is thought to originate from either the difference of the growth temperature or the difference of the length of the N-terminal domain of SLX-1 ( 168 amino acids compared to 9 amino acids in both the yeast and human orthologs ) ( Figure 1A ) . Future analysis may require performing the in vitro nuclease assay at 20°C , which is the optimum temperature for growth of C . elegans , and using an N-terminal truncated SLX-1 . Similar to other organisms , the nuclease activity of SLX-1 depends on HIM-18 ( Figure 1D–1E ) . This endonuclease activity potentially affects the homologous recombination machinery during ICL-repair , break-induced replication and NER . During prophase of meiosis I , homologous recombination occurs between homologous chromosomes . Double Holliday junction resolution is important for crossover formation , however the HJ resolvase activity of the SLX-1-HIM-18 complex is not required for meiotic crossover formation ( Figure 7B ) . Potentially , other structure-specific endonucleases such as XPF-1 , which interacts with HIM-18 [16] , MUS-81 and GEN-1 [21] , may act coordinately to resolve dHJs during meiotic recombination . We showed that slx-1 mutants were hypersensitive to several kinds of DNA damaging agents ( Figure 4C and Figure 8A ) . Notably , both slx-1 and him-18 mutants showed hypersensitivity to UV . This phenotype is different from that observed in budding yeast , fission yeast , flies , mice and humans [2] , [3] , [5] , [8] , [13] , [24] . SLX-1 has a GIY-YIG nuclease domain also present in UvrC in E . coli where it is important for making an incision 3′ to the damage site ( cyclobutane pyrimidine dimer , CPD ) during the NER process [43] , [44] . Additionally , the N-terminal domain of C . elegans SLX-1 is longer than that of its homologs in other organisms , so it is possible that the N-terminal region confers the NER function of SLX-1 . XPF is largely known as a repair factor for the NER pathway , including in C . elegans [45] , [46] . XPF exists in two types of complexes in human cells , one is a 2M Dalton complex containing SLX1 , SLX4 , MUS81 , EME1 and ERCC1 , the other is the XPF-ERCC1 heterodimer [4] . Therefore , it remains to be determined whether XPF-1 , SLX-1 and HIM-18 make single or heterologous complexes during different DNA damage responses or at different stages of the cell cycle . It is thought that the dual incisions surrounding an interstrand crosslink are performed by MUS81 , XPF or FAN1 [47]–[52] . In this study , we raise the possibility that SLX-1 might have the activity required for the incision based on the HN2 hypersensitivity observed in slx-1 mutants ( Figure 5B and Figure 8A ) . Surprisingly , the mutation in slx-1 partially suppressed the HN2 induced DNA damage sensitivity observed in him-18 mutants ( Figure 5B ) . It has been reported that SLX1 represses the nuclease activity towards RFs and 3′-flaps of MUS81 and XPF in human cells [4] . Therefore , in the him-18 mutant , SLX-1 might repress the incision activity of MUS81 and XPF or other nucleases such as FAN1 . Given that HIM-18 carries sites potentially recognized by kinases and ubiquitin/SUMO conjugating enzymes [16] , that both human and yeast Slx4 are phosphorylated by ATM/ATR [53] , and that human SLX4 is phosphorylated by PLK1 [5] , it is possible that post-translational modifications of HIM-18 might modulate its ability to regulate the nuclease activity of the components of the HIM-18 complex either directly or indirectly . In addition to HN2 hypersensitivity , both accumulation of RAD-51 foci in late pachytene ( zone 6 ) and germ cell apoptosis in him-18 mutants are partially rescued by the slx-1 mutation ( Figure 4A and 4B ) . One possible explanation is that SLX-1 might be deregulated in the absence of HIM-18 in these cases . In him-18 mutants , inactive SLX-1 might inhibit these repair pathways . In slx-1;him-18 double mutants , since there is no inhibition by SLX-1 , the him-18 phenotype is partially suppressed . Further studies will reveal the regulation of the HIM-18 complex in each DNA repair pathway . We showed that SLX-1 is not required for wild type levels of crossover formation during meiotic recombination ( Figure 7B ) . It is believed that there are two possible pathways that can lead to crossover formation , one is double Holliday junction resolution [54] and the other is a “nick/counternick” mechanism [55] . A possible explanation for why crossover frequency is not changed in slx-1 mutants is that SLX-1 and other structure specific nucleases , such as GEN-1 , MUS-81 and XPF-1 , are partially redundant and can compensate for each other with regards to the activity of crossover formation . This is supported in part by the observations that gen-1 mutants are fertile [21] ( Table 1 ) , mus-81 mutants do not enhance X chromosome non-disjunction [16] and xpf-1 mutants exhibit only a mild reduction in crossover levels [16] . Moreover , slx-1 enhances the developmental defects observed in xpf-1 and gen-1 mutants , supporting the idea that functions of SLX-1 are partially redundant with those of GEN-1 and XPF-1 ( Table 1 ) . Furthermore , a recent study suggests that MUS81 , SLX4 and GEN1 can compensate for lack of the BLM helicase in human cells by resolving HJs in somatic Bloom's syndrome cells [56] . Therefore , further investigation of the genetic interactions between these structure-specific endonucleases may reveal whether there are redundancies for the activities of Holliday junction resolution during meiotic recombination . Another aspect to consider is that excess crossovers generated in wild type following X-ray exposure are dependent on MUS-81 in C . elegans , although MUS-81 is not required for physiological crossover formation during meiosis [57] . mus-81 and slx-1 mutants share a couple of phenotypes that are not observed in either xpf-1 or gen-1 mutants , such as the elevated levels of RAD-51 foci observed during mitotic proliferation ( Figure 4A and Figure S2 ) [16] and the synthetic lethality with him-6 ( Table 1 ) ( Saito et al . , unpublished results ) . Therefore , it remains to be determined whether SLX-1 is also required for crossover formation under an excess of DSBs resulting from IR treatment in a manner similar to mus-81 mutants . How is crossover distribution regulated in C . elegans meiosis ? Recently , it was reported that crossover distribution is shifted from the arms to the center of chromosomes in xnd-1 mutants , a phenotype reminiscent to that we observed in our current analysis of slx-1 mutants [58] . Hyperacetylation of histone H2A lysine 5 is one of the features of the xnd-1 mutants . However , the acetylation levels of H2AK5 are similar to those of wild type in slx-1 mutants ( Figure 7 ) . Therefore , the hyperacetylation of H2AK5 is not the cause of the change of crossover distribution in slx-1 mutants . Based on the analysis of both DSB and crossover distribution in slx-1 mutants , we propose a model in which SLX-1 inhibits crossover formation at the center of the chromosomes during meiotic recombination ( Figure 9B ) . While crossover formation is biased to the arms in wild type , surprisingly we found that DSBs are more evenly distributed along chromosome axes in wild type . Only one of the DSBs , presumably the one marked by ZHP-3 , is converted into an interhomolog crossover at one of the arm regions . All other DSBs are repaired either by intersister repair or interhomolog noncrossover formation . Interestingly , the number of ZHP-3 foci is reduced to 80% of wild type levels in slx-1 mutants , and nevertheless the total number of crossovers is similar between wild type and slx-1 mutants ( Figure 7B , 7E , 7F ) . These data suggest that there is a pathway not associated with ZHP-3 foci to make a crossover in slx-1 mutants . Whether this pathway leads to crossover formation at the center region of the chromosomes and whether these ZHP-3 foci-independent crossovers depend on MUS-81 , which is known to make ZHP-3 foci-independent crossovers when there is an excess of DSBs [57] , are issues that remain to be solved . In yeast , the crossover/noncrossover decision is made very early , prior to or during the formation of stable single-end invasion ( SEI ) intermediates , and therefore earlier than dHJ formation [59] . Once a dHJ is formed , it is usually converted into a crossover in yeast meiosis . However , it is not known whether this tendency is conserved in other organisms . We hypothesize that DSBs introduced at the mid-section of chromosomes are converted into a noncrossover product either via a synthesis-dependent strand annealing ( SDSA ) pathway or via symmetric resolution of a dHJ in wild type C . elegans . Based on the HJ cleavage activities we observed for SLX-1 , SLX-l might have a role in converting a dHJ into a noncrossover product via same sense resolution of a dHJ arising at the mid section of chromosomes during meiotic recombination ( Figure 9B ) . In addition to its nuclease domain , SLX-1 has a PHD finger . This type of domain is largely known to be involved in chromatin remodeling , transcriptional control and ubiquitin/SUMO E3 ligase activity . Chromosome arms , where crossovers happen at a higher frequency , are marked by methylated histone H3 lysine 9 ( H3K9me ) during early embryogenesis and the L3 larval stage in C . elegans [60] , [61] . However , it is not yet known whether this kind of epigenetic mark is also observed during meiotic recombination or whether other kinds of epigenetic marks delimitate the chromosome arms and the central region in C . elegans . One possibility is that the PHD finger of SLX-1 is involved in epigenetic change/read and somehow divides the arms and central regions along chromosomes ( Figure 9B ) . It will be important to investigate what kinds of epigenetic marks may be read by the PHD finger of SLX-1 , and how this may regulate crossovers . Taken together , our analysis indicates that SLX-1 is required for several kinds of mitotic DNA repair pathways and reveals a role for this protein in the regulation of meiotic crossover distribution thereby promoting the maintenance of genomic integrity . Importantly , our study leads us to propose a model in which SLX-1 functions as a noncrossover promoting factor at the crossover cold regions during meiotic recombination . C . elegans strains were cultured at 20°C under standard conditions [32] . The N2 Bristol strain was used as the wild-type background . The following mutations and chromosome rearrangements were used in this study: LGI: slx-1 ( tm2644 ) , rad-54 ( ok615 ) , hT2[bli-4 ( e937 ) let- ? ( q782 ) qIs48] ( I; III ) ; LGII: xpf-1 ( e1487 ) , dpy-10 ( e128 ) , unc-4 ( e120 ) , mIn1[dpy-10 ( e128 ) mIs14] ( II ) ; LGIII: him-18 ( tm2181 ) , brc-2 ( tm1086 ) , xnd-1 ( ok709 ) , qC1[dpy-19 ( e1259 ) glp-1 ( q339 ) qIs26] ( III ) ; LGIV: Ppie-1::zhp-3::gfp , spo-11 ( ok79 ) , him-6 ( ok412 ) , rec-8 ( ok978 ) , nT1[unc- ? ( n754 ) let- ? qIs50] ( IV; V ) , nT1[qIs51] ( IV; V ) ; LGV: him-17 ( ok424 ) , syp-2 ( ok307 ) [16] , [19] , [32] , [42] , [62]–[67] . Transgenes: opIs263 [Prpa-1::RPA-1-YFP::3′-UTR] [68] . The slx-1 allele , tm2644 , is predicted to encode for a catalytically inactive ( nuclease-negative ) protein . We also tried to knockdown slx-1 both by RNAi and by generating slx-1 ( tm2644 ) /Df trans-heterozygotes . RNAi utilizing feeding clones generated from either SLX-1 cDNA ( Vidal ORFeome library; [69] , [70] ) or genomic DNA ( Ahringer RNAi library; [71] , [72] ) , did not result in depletion in either wild type or slx-1 ( tm2644 ) mutant backgrounds . Attempts to generate a slx-1/Df trans-heterozygote for classical genetic characterization of the slx-1 ( tm2644 ) allele were hindered because the only available deficiency encompassing that gene , sDf4 , involves a free chromosome duplication that interferes with this analysis . HIM-18 and SLX-1 open reading frames in either pENTR or pDONR derivatives were transferred to the indicated expression vectors ( pDEST-myc or pDEST-HA ) using the Gateway cloning system ( Invitrogen ) and sequence validated . 293T cells were grown in Dulbecco Modified Eagle medium ( DMEM ) supplemented with 10% ( v/v ) FBS ( Invitrogen ) , 100 units of penicillin per ml , and 0 . 1 mg streptomycin per ml . Antibodies against HA ( 16B12; Covance ) and Myc ( 9E10; Santa Cruz ) epitopes were utilized . For protein interaction studies , the indicated proteins were expressed in HEK293T cells using Lipofectamine ( Invitrogen ) and after 24–48 h , cells were lysed in 50 mM Tris-HCl pH 7 . 5 , 150 mM NaCl , 0 . 5% NP-40 , 10 mM NaF , 1 mM EDTA+protease inhibitors ( ROCHE ) , and cleared lysates used for immunoprecipitation with the indicated antibodies . Immune complexes were washed 4–5X with lysis buffer , re-suspended in SDS laemli buffer and were subjected to polyacrylamide gel separation and immunoblotting with the indicated antibodies . Recombinant GST-h . SLX1/His-h . SLX4 ( SBD ) was purified as previously described [5] . 5′32P-labeled DNA substrates ( 5′-flap , 3′-flap , Holliday Junction , and Replication Fork ) were prepared as previously described [5] . The sequences used for the preparation of labeled substrates are presented in Table S1 . Radiolabeled substrates were incubated with the indicated immune complexes expressed and purified as described above . Immune complexes were washed 3X in cleavage buffer ( 50 mM Tris pH 8 . 0 , 5 mM MgCl2 , 40 mM NaCl , 1 mM DTT , 100 µg/ml BSA ) prior to initiating the reaction . After 30 min at 37°C , reaction mixtures were treated with 25 mM EDTA and 1% Proteinase K in 10% SDS prior to electrophoresis on either 8% polyacrylamide gels ( native ) or 12% polyacrylamide-urea gels ( denaturing ) . Reaction products were visualized by autoradiography and quantified with ImageJ software . Quantitative analysis of RAD-51 foci was performed as in [38] except that all seven zones composing the germline were scored . 2–3 germlines were scored for each genotype . The average number of nuclei scored per zone for a given genotype was as follows , ± standard deviation: zone 1 , n = 79±24; zone 2 , n = 93±38; zone 3 , n = 103±34; zone 4 , n = 93±28; zone 5 , n = 82±18; zone 6 , n = 64±14; and zone 7 , n = 58±13 . Statistical comparisons between genotypes were performed using the two-tailed Mann-Whitney test , 95% confidence interval ( C . I . ) . C . elegans gonads were fixed and stained with rabbit α-RAD-51 ( SDIX ) ( 1∶20 , 000 ) , mouse α-REC-8 ( Abcam ) ( 1∶100 ) and guinea pig α-HIM-8 ( 1∶100 ) . Chromosome axes were traced in 3D along the REC-8 signal and straightened by using either Priism [73] or softWoRx ( Applied Precision ) . For each chromosome axis , positions of the RAD-51 foci were measured with softWoRx . Statistical comparisons between genotypes were performed using the two-tailed Mann-Whitney test , 95% confidence interval ( C . I . ) . To assess ionizing radiation ( IR ) sensitivity , animals ( ∼19 hours post L4 stage ) were treated with 0 , 50 or 100 Gy of IR from a Cs137 source at a dose rate of 1 . 86 Gy/min . For UVC sensitivity , animals were placed on the UV stratalinker 2400 ( Stratagene ) and exposed to 0 , 300 or 600 J/m2 . For nitrogen mustard ( HN2 ) sensitivity , young adult animals were treated with 0 , 100 or 200 µM of HN2 ( mechlorethamine hydrochloride; Sigma ) in M9 buffer containing E . coli OP50 with slow shaking in the dark for 19 hours . Treatment with camptothecin ( CPT; Sigma ) was similar , but with doses of 0 , 500 or 1000 nM . Following treatment with IR , UVC , HN2 or CPT , animals were plated to allow recovery for 3 hours . For all damage sensitivity experiments , 21 animals were plated 7 per plate and hatching was assessed for 4 hours after the recovery . After 1 . 5 days , hatched worms and dead eggs were counted . Each damage condition was replicated at least three times in independent experiments . 22–24 hour post-L4 hermaphrodites were stained with acridine orange ( AO ) for 2 hours and mounted under coverslips in 5 µl of a 15 mM sodium azide solution on 1 . 5% agarose pads . Apoptotic nuclei stained with AO were observed in the late pachytene region of the germline with a Leica DM5000 B fluorescence microscope . Between 33 and 47 gonads were scored for each genotype . Statistical comparisons between genotypes were performed using the two-tailed Mann-Whitney test , 95% C . I . Meiotic crossover frequencies were assayed utilizing single-nucleotide polymorphisms ( SNP ) markers as in [74] , except that +/+ worms were used as a control . PCR and DraI restriction digests of single worm lysates were performed as described in [75] . The following DraI SNP primers were utilized: A ( uCE3-637 ) , B ( CE3-127 ) , C ( snp_Y39A1 ) , D ( uCE3-1426 ) for chromosome III , A ( uCE4-515 ) , B ( pkP4055 ) , C ( snp_F49E11 ) , D ( pkP4099 ) for chromosome IV , A ( pkP5076 ) , B ( snp_Y61A9L ) , C ( pkP5129 ) , D ( snp_Y17D7B ) for chromosome V , and A ( pkP6143 ) , B ( pkP6105 ) , B' ( snp_F11A1 ) , C ( pkP6132 ) , D ( uCE6-1554 ) for the X chromosome . Statistical analysis was performed using the two-tailed Fisher's Exact test , 95% C . I . , as in [26] ( Table S4 ) . Recombination analysis using visible markers was performed as in [76] and recombination frequencies were calculated as in [77] . The yeast two-hybrid assay was performed according to [78] . cDNA of HIM-18 full length , SLX-1 full length , SLX-1N1–272 and SLX-1C273–443 were cloned into the Gateway donor vector ( pDONR223 ) . Each construct was then subcloned into 2 µ Gateway destination vectors pVV213 ( activation domain ( AD ) , LEU2+ ) and pVV212 ( Gal4 DNA binding domain ( DB ) , TRP1+ ) . AD-Y and DB-X fusions were transformed into MATa Y8800 and MATa Y8930 yeast strains , respectively . These yeast strains have three reporter genes: GAL2-ADE2 , met2::GAL7-lacZ and LYS2::GAL1-HIS3 . MATa Y8800 and MATα Y8930 were mated on YPD plates and diploids carrying both plasmids were selected on SC-Leu-Trp plates . The interactions were assessed by growth on -His+1 mM 3-AT plates at 30°C .	Crossover formation between homologous chromosomes is important for generating genetic diversity in subsequent generations , as well as for promoting accurate chromosome segregation during meiosis , which is a specialized cell division program that results in the formation of haploid gametes ( sperm and eggs ) from diploid parental germ cells . In the nematode Caenorhabditis elegans , a single off-centered crossover is formed on the chromosome arms between every pair of homologous chromosomes . Crossover formation at the central region of the chromosomes is suppressed by unknown mechanisms . By using high-resolution 3-D microscopy , we found that , while crossover distribution is biased to the arm regions along the chromosomes , DNA double-strand breaks ( DSBs ) , which initiate the homologous recombination repair process , are evenly distributed along the chromosomes . These results suggest the existence of mechanisms that inhibit crossover formation after induction of DSBs at the central region of the chromosomes . In this study , our findings lead us to hypothesize that SLX-1 , a structure-specific endonuclease , inhibits crossover formation at the central region of the chromosomes , probably via its resolution activity of the Holliday junctions , which are four-stranded recombination intermediates , to produce noncrossover products .	[ "Abstract", "Introduction", "Results", "Discussion", "Materials", "and", "Methods" ]	[ "biology" ]	2012	SLX-1 Is Required for Maintaining Genomic Integrity and Promoting Meiotic Noncrossovers in the Caenorhabditis elegans Germline
Cardiovascular disease ( CVD ) is the leading cause of death worldwide . Recent genome-wide association ( GWA ) studies have pinpointed many loci associated with CVD risk factors in adults . It is unclear , however , if these loci predict trait levels at all ages , if they are associated with how a trait develops over time , or if they could be used to screen individuals who are pre-symptomatic to provide the opportunity for preventive measures before disease onset . We completed a genome-wide association study on participants in the longitudinal Bogalusa Heart Study ( BHS ) and have characterized the association between genetic factors and the development of CVD risk factors from childhood to adulthood . We report 7 genome-wide significant associations involving CVD risk factors , two of which have been previously reported . Top regions were tested for replication in the Young Finns Study ( YF ) and two associations strongly replicated: rs247616 in CETP with HDL levels ( combined P = 9 . 7×10−24 ) , and rs445925 at APOE with LDL levels ( combined P = 8 . 7×10−19 ) . We show that SNPs previously identified in adult cross-sectional studies tend to show age-independent effects in the BHS with effect sizes consistent with previous reports . Previously identified variants were associated with adult trait levels above and beyond those seen in childhood; however , variants with time-dependent effects were also promising predictors . This is the first GWA study to evaluate the role of common genetic variants in the development of CVD risk factors in children as they advance through adulthood and highlights the utility of using longitudinal studies to identify genetic predictors of adult traits in children . Cardiovascular disease ( CVD ) affects over 79 million people in the United States [1] , and is the leading cause of death worldwide [2]–[4] . Identifying the genetic determinants of CVD can lead to more effective diagnostics , prognostics , therapeutics , and , ultimately , preventive strategies . The best chance for prevention would be to identify risk at the earliest possible age . Genome-wide association ( GWA ) leveraging cross-sectional phenotypic data has been a particularly useful approach to identifying loci that influence many of the quantitative risk factors of CVD [5]–[10] , however the use of cross sectional data does not provide insight into how such risk factors develop over time . Longitudinal studies , particularly those that begin in childhood , allow for the identification of risk profiles of susceptible individuals before disease onset . The Bogalusa Heart Study ( BHS ) is a longitudinal study focused on the early natural history of CVD . The BHS began in 1973 and includes up to 9 phenotypic screenings in childhood ( 4–17 years of age ) and up to 10 adult ( 18–48 years of age ) cross-sectional screenings . We have conducted a longitudinal genome-wide association study on a subset of the total sample of unrelated individuals with a large number of measurements ( mean number of measurements = 8 , range = 4–13 ) and are of European Ancestry ( N = 525 ) . We conducted a genome-wide association study of longitudinal measures of 12 traits measured from childhood through adulthood on participants of the BHS of European ancestry: anthropomorphic ( height , weight , and waist circumference ) , blood pressure ( BP ) ( diastolic and systolic BP ) , heart rate , blood lipids ( low density lipoprotein cholesterol ( LDL ) , high density lipoprotein cholesterol ( HDL ) , total cholesterol ( TC ) , and triglycerides ) , and metabolic traits ( glucose and insulin ) . Genotyping was performed on the Illumina Human610 and HumanCVD BeadChips [11] for a total of 545 , 821 SNPs passing QC and allele frequency filters ( see Materials and Methods ) . Imputation was performed using the CEU HapMap 2 as a reference population with the computer program MACH v . 1 . 0 . 16 ( http://www . sph . umich . edu/csg/yli/mach/ ) [12] , providing genotype estimates for an additional 1 , 622 , 114 SNPs . For each SNP , we tested whether it had an average linear effect over time ( SNP effect ) , and whether it entered into a time-dependent effect ( SNPxAGE interaction effect ) , such that the genotype is associated with variation in the linear trajectory of the trait from childhood through adulthood . Both SNP and SNPxAGE effects were calculated using linear mixed models as implemented in the R nlme package [13] , adjusting for age and gender . Table 1 lists all regions showing SNP effect associations ( P<10−6 ) and Table 2 lists all regions showing association ( P<10−6 ) with SNPxAGE effects . We analyzed the regions surrounding the top associations for consistency with recombination hotspots and LD relationships ( Figure S1 ) and provide Manhattan plots of each trait association ( Figure S2 ) . From both sets of analyses , there were 5 novel associations with a P-value less than 5×10−8 and 6 novel regions where there were at least 10 genotyped or imputed SNPs with P<10−5 . The most significant association ( rs7890572 , P = 3 . 8×10−10 ) was observed with a linear triglyceride trajectory effect ( i . e . , SNPxAGE effect ) on the X chromosome within the IL1RAPL1 gene and near the gene encoding glycerol kinase ( GK ) , in which mutations have been implicated in pseudo-hypertriglyceridemia , caused by high levels of glycerol creating measurement artifacts in the triglyceride assay [14] . A novel association of potential biological interest involved a SNP effect on insulin levels with variation in the CHN2 locus ( rs3793275 , P = 5 . 8×10−9 ) , a beta-chimerin that has recently been described as part of a fusion gene also containing the insulin receptor that was shown to be responsible for severe insulin deficiency [15] . This SNP is also associated with glucose trajectories in our dataset ( SNPxAGE; P = 1 . 5×10−7 ) . In the 7q11 region , 25 SNPs are associated ( P<10−5 ) with diastolic BP ( SNP effect; peak SNP rs709595 , P = 7 . 0×10−7 ) . The calcitonin gene-related peptide receptor ( CRCP ) is approximately 200 kb from the top SNP , but contains SNPs that are in LD with the top SNP ( see Figure S1 ) . The calcitonin gene-related peptide is a vasodilator [16] and its receptor CRCP has been previously implicated in hypertension in a small candidate gene association study of hypertension in Japanese individuals [17] . In addition to novel associations , there were three regions showing SNP associations that have been previously identified in GWA studies: rs853773 [18] near G6PC2 was associated with a glucose SNP effect ( P = 7 . 0×10−7 ) , rs247616 [5] near CETP was associated with an HDL SNP effect ( P = 6 . 6×10−7 ) , and the APOE e2 SNP rs7412 [19] was associated with a genome-wide significant LDL SNP effect ( P = 1 . 6×10−8 ) . A region near APOA5 that had been previously implicated in triglyceride levels showed a significant SNPxAGE effect on triglycerides in our study ( rs12280753; P = 1 . 8×10−8 ) . Although the nearest gene to rs12280753 is not APOA5 , this SNP was also the most strongly associated SNP in previous studies of adult triglyceride levels [5] , [10] , [20] . We pursued replication of these findings in genotyped individuals within the Young Finns ( YF ) cohort , consisting of 2 , 442 Finnish individuals tracked from childhood through middle adulthood ( ages 3–45 ) with three measures in young individuals ( ages 3–24 ) and two measures in older individuals ( ages 24–45 ) . These individuals have been genotyped on a custom-built Illumina genotyping chip ( 670K ) . Using the same analysis methods , we tested whether the top SNP was associated in the YF study ( Table 3 ) . Imputed genotype dosages were used when direct genotype data was not available . For the APOE-e2 SNP rs7412 , which is not in HapMap or on the 670K chip , we used the SNP with the next strongest association in the BHS ( rs445925 ) . There were two SNPs that significantly replicated beyond the multiple testing threshold ( P<0 . 05/51 = 1×10−3 ) : the rs247616 SNP at CETP ( P = 1 . 7×10−18 ) , and rs445925 at APOE ( P = 4 . 1×10−15 ) . There was no trend to replicate the direction of effect between the studies: within the SNP effects , there were 12/21 ( 57% , chi-square P = 0 . 51 ) markers that showed the same direction of effect , while within SNPxAGE effects , there were 14/30 ( 47% , chi-square P = 0 . 72 ) . The samples were combined and P-values were calculated for the combined BHS and YF data , using study as a covariate ( Table 3 ) . The associations at rs247616 at CETP with HDL-cholesterol ( P = 9 . 7×10−24 ) and rs445925 at APOE with LDL-cholesterol ( P = 8 . 7×10−19 ) were strongly significant , but no other regions in the combined BHS/YF data reached genome-wide significance of P<5×10−8 . Genetic variants will be most useful for trait prediction when they are associated with a trait above and beyond other known risk factors . In addition , the ability to predict adult trait levels in children , before disease onset , can lead to a disease prevention strategy . In longitudinal studies starting in childhood and going into adulthood , we can ask whether genetic loci are associated with the adult trait level above and beyond the trait level seen in the first measure taken in childhood . To test this hypothesis , we evaluated whether our associated markers were likely to be predictive of adult levels of the traits , after adjustment for trait levels in childhood . To account for variation in data collection , we also included the age at each of these measures as well as gender as covariates in the analysis . Within the BHS , variants that were characterized as SNPxAGE effects were more likely to be predictive of adult values after correcting for childhood values , which is expected since these variants were characterized in BHS initially ( Table 4 ) . In the YF study , however , we also saw more SNPxAGE variants associated with adult levels given childhood levels ( Table 4 ) . There were 6 variants that were associated with adult levels in the YF study at P<0 . 05 , with 2 corresponding to the genome-wide significant SNP effects and 4 corresponding to BHS SNPxAGE variants . Only the association of rs445925 with LDL-cholesterol was strong enough to withstand multiple corrections . Further analysis of this observation is warranted in a larger cohort . We assessed whether associations that have been described in previous adult cross-sectional GWA studies exhibit consistent effects over time and whether the effect sizes observed in children through middle-aged adults are consistent with those previously described . We identified 169 SNP-trait associations ( see Materials and Methods ) for which we had directly genotyped or imputed genotype data . We first estimated our power to detect each previous association at alpha = 0 . 05 under a more structured , but similar study design ( i . e . , 8 equally spaced measurements ) , given the previously reported effect size and allele frequency . Under this model , we would expect to have detected 40/169 ( 24% ) associations at P<0 . 05 , and we observed a similar number of SNP effects in the BHS data ( 32/169; 19% ) . We evaluated the associations across all traits together by comparing how well the previously reported effect size was recapitulated in the BHS GWA ( Figure 1A ) . For consistency across studies and traits , if an effect size was not already expressed in terms of percent standard deviation ( %SD ) , we converted the previously reported effect size into %SD and compared the previous effect size to the SNP effect . The previously reported effect size was a strong predictor of the SNP effect ( slope = 0 . 47 , P = 1 . 2×10−21 ) , suggesting that SNPs that have been previously identified in adult cross-sectional GWA studies are good predictors of time-averaged effects in the BHS sample . We also determined whether the same previously identified SNPs were likely to show effects on a trait over time ( SNPxAGE effects ) . Under a simple model that assumed that all of the effect in adults is due to a locus that has no effect in childhood , we estimated power to detect such an interaction effect in a similarly structured study with 8 repeated measures . Given these assumptions , we would have expected to see 24/169 ( 14% ) SNPxAGE associations . We observed 6/169 ( 3 . 6% ) SNPs that showed SNPxAGE effects at P<0 . 05 , indicating that effects seen in SNPs described in adult GWA studies are not due primarily to differences in effects over time , although larger studies will be required to definitively characterize this . We considered whether a composite genotype score would better predict overall CVD risk factor trajectories or time-dependent effects than any single locus . For each person and each trait , we created a score by summing the expected effect in percent standard deviation of each allele that the person carried . We then determined whether the score was associated with the trait's average value and trajectory by using this score as a predictor for each trait in a linear mixed model , adjusting for age and gender . We assessed the score's average effect across time ( score effect ) and whether or not there was a time-dependent effect ( scoreage effect ) . The traits HDL , LDL , total cholesterol , triglycerides , and height showed strongly significant score effects , while only triglycerides showed a scoreage effect ( Table 5 ) . Longitudinal data was visualized by color-coding the individuals according to the decile of their overall score and the average linear trend of each group was plotted ( LDL , Figure 1B and others in Figure S3 ) . These results indicate that the cumulative effects of SNPs that are identified in large adult cross-sectional studies are generally age-independent effects , with an exception in triglycerides , which was the only trait to show a significant scoreage effect . We additionally tested whether previously identified variants were predictive of adult levels after adjusting for childhood levels ( Table 6 ) . We saw that 25/169 ( 14 . 8% ) showed association at P<0 . 05 . These observations in the BHS data suggest that even though results from existing GWA studies demonstrate age-independent effects , they can be predictive of trait values in adults . We identified seven associations at P<5×10−8 showing either time-averaged or time-dependent effects on CVD risk factors in the BHS , two of which have been previously characterized . Of all associations with P<10−6 , we were able to strongly replicate the association in the YF with HDL-cholesterol at CETP with a combined P = 9 . 7×10−24 , and LDL-cholesterol at APOE with a combined P = 8 . 7×10−19 . Differences that exist between the cohorts , such as birth year ( 15 year difference ) , and environmental differences could have influenced replication of the remaining SNPs . Larger discovery studies will also have better resolution and power to accurately estimate longitudinal effect sizes , likely allowing for more robust replication . We evaluated the longitudinal effects of markers that have been previously identified in adult GWA studies . We found that previously identified markers showed time-averaged effects consistent with their reported effect size . This argues that the linear mixed model is an effective tool for modeling time-averaged effects in a GWA setting and that adult GWA studies may be capturing variation that tends to have consistent effects over time . Using a scoring approach , the overall signal from previously identified markers tended to have strong associations with time-averaged effects , but except in the case of triglycerides , did not show time-dependent effects . Previously identified markers were also likely to be associated with adult trait levels above and beyond childhood levels . Although we primarily describe time-averaged effects for previously identified markers , there may be more subtle time-dependent effects that larger studies will be better able to capture . It is important to note that although we focused on analysis of linear trends over time , a linear model may not best capture these trends . Other approaches could be explored further such as non-linear models when there is an a priori expectation of trait trajectory , or model free approaches . These additional models could lead to additional variations that influence trajectories , or more precise estimations of effect size . Longitudinal studies are particularly suited to capturing effects that vary over time . Genetic variation that shows a time-dependent effect may help predict those that will go onto develop disease before they show symptomatic traits . The discovery of variants associated with SNPxAGE interaction effects could thus be used to screen young individuals who are pre-symptomatic and provide the opportunity for preventive measures decades before disease onset . We explored how well the markers that we identified predicted adult traits after correcting for childhood traits and suggest further study of variants with SNPxAGE effects as possibly better predictors of adult trait levels above and beyond childhood levels . These results are consistent with the idea that longitudinal studies may be a useful tool to better capture time-dependent variation that could ultimately be better predictive of future outcomes . The study was approved by the institutional review board and the ethics committee of each institution . Written informed consent was obtained from each participant in accordance with institutional requirements and the Declaration of Helsinki Principles . All subjects in the BHS gave informed consent at each examination , and for those under 18 years of age , consent of a parent/guardian was obtained . Study protocols were approved by the Institutional Review Board of the Tulane University Health Sciences Center . Between 1973 and 2008 , 9 cross-sectional surveys of children aged 4–17 years and 10 cross-sectional surveys of adults aged 18–48 years ( Figure S4 ) , who had been previously examined as children , were conducted for CVD risk factor examinations in Bogalusa , Louisiana . This panel design of repeated cross-sectional examinations has resulted in serial observations from childhood to adulthood . By linking the 19 surveys , 12 , 163 individuals have been examined , with 37 , 317 observations . In the ongoing Longitudinal Aging Study funded by NIH and NIA since 2000 , there are 1 , 202 subjects who have been examined 4–14 times from childhood to adulthood and have DNA available for GWA genotyping . Based on the analysis of identity-by-state ( IBS ) sharing from whole genome genotyping data , we focus on a subset of 525 genotyped individuals who are of European ancestry and unrelated ( 229 male , 296 female ) . The average number of measurements per individual is 8 ( range 4–13 ) . The YF cohort is a Finnish longitudinal population study sample on the evolution of cardiovascular risk factors from childhood to adulthood [21] . The first cross-sectional study was conducted in 1980 in five centers and included 3 , 596 participants in the age groups of 3 , 6 , 9 , 12 , 15 , and 18 , who were randomly chosen from the national population register . After baseline in 1980 these subjects have been re-examined in 1983 and 1986 as young individuals , and in 2001 and 2007 as older individuals . Genotype data for the present analysis ( DNA collected in 1980 , 2001 and 2007 ) was available for 2 , 442 individuals . In the latest follow-up in 2001 , a total of 2 , 283 participants ( of which DNA is available from 2 , 265 participants ) were examined for numerous study variables , including serum lipoproteins , glucose , insulin , obesity indices , blood pressure , life-style factors , smoking status , alcohol use and general health status . Previously identified markers were obtained through the NHGRI database [25] ( accessed 5/20/09 ) . Marker associations , alleles , and allele frequencies were verified with those reported in the original papers and corrected if required . Markers were used if the alleles at the locus provided unambiguous orientation or if the allele frequencies were different enough between A/T and C/G SNPs to distinguish which allele was the associated allele . We thus excluded any A/T or C/G SNPs with a minor allele frequency >0 . 4 and required that the allele frequency in the previously reported study be within 10% of the allele frequency in the BHS . We excluded studies of non-European Ancestry origin . One SNP per cytogenic region was used for each phenotype: the SNP with the smallest previously reported p-value was used . Effect size was translated to percent standard deviation . If the effect size was reported in an absolute measure ( e . g . cm for height ) , then the standard deviation from the BHS study was used . Standard deviation was calculated from the standard error of the SNP association reported in the linear mixed model . For glucose , cholesterol , and triglycerides measures , units were converted to mg/dl before converting to %SD . A risk value was calculated for each individual based on the imputed genotype and previously reported effect size , converted to %SD . The %SD was multiplied by the allelic dosage for each SNP and summed over all the associated SNPs for each phenotype . The resulting risk value was then used as a predictor for the BHS individuals . GWA was performed using linear mixed model regression with fixed covariates of age and sex , random slope , and random intercept . Genotypes were coded as 0 , 1 , or 2 when the SNP was genotyped and by dosage ( scale 0–2 ) when imputed . Analysis was performed within the nlme package in R [13] . Covariance structures were determined by testing all spatial covariance structures ( exponential , Gaussian , linear , rational quadradics , and spherical ) with covariates and a sample of SNPs , and picking the structure that best fit the data as measured by the lowest AIC ( Akaike Information Criteria ) value . SNP and SNPxAGE interaction effects were estimated separately . Although the default nlme optimizer tended to have difficulty converging , we obtained good results by using the optim optimizer on data where all missing data was removed . The number of SNPs that converged and for which we obtained results is listed in Table S1 . Analyses were performed on a compute cluster with 600 , 000 tests taking ∼3 hrs on 64 processors . If genomic inflation factors were inflated or deflated , we reran the GWA using the first four MDS components as covariates . If the inflation factor was still less than 0 . 90 or greater than 1 . 05 , we removed the analysis . In addition , we filtered body mass index ( BMI ) SNP , BMI SNPxAGE , and weight SNP analyses completely from the analysis due to a combination of consistently inflated or deflated genomic inflation factors or a long list of highly associated SNPs . Power was calculated using GPower 3 [26] . We used the MANOVA repeated measures module with 8 repeated measures with a correlation of 0 . 5 between them , similar to the correlations seen in this study . We estimated power for between-factor and between-within interaction effects . Effect size ( f ) was calculated asand R2 was calculated from the allele frequencies as reported in the original associations ( p and q ) and the effect size in terms of %SD [27] .	We have studied the association between genetic factors on a whole genome level and cardiovascular disease ( CVD ) risk factors in a population of individuals studied from childhood through adulthood . The longitudinal study design has enabled the investigation of genetic variation influencing trait values over time . We have identified DNA variants that are associated with CVD trait values consistently over time , and a second set of variants that are associated with CVD trait values in a time-dependent manner . We also show that variants previously identified in adult populations have consistent effects within our population and that these effects are usually similar across childhood through adulthood . The discovery of time-dependent variants that influence CVD trait values over time can potentially be used to screen young individuals who are pre-symptomatic and provide the opportunity for preventive measures decades before disease onset .	[ "Abstract", "Introduction", "Results", "Discussion", "Materials", "and", "Methods" ]	[ "genetics", "and", "genomics/genetics", "of", "disease", "genetics", "and", "genomics/complex", "traits", "cardiovascular", "disorders" ]	2010	Longitudinal Genome-Wide Association of Cardiovascular Disease Risk Factors in the Bogalusa Heart Study
Mechanistic mathematical modeling of biochemical reaction networks using ordinary differential equation ( ODE ) models has improved our understanding of small- and medium-scale biological processes . While the same should in principle hold for large- and genome-scale processes , the computational methods for the analysis of ODE models which describe hundreds or thousands of biochemical species and reactions are missing so far . While individual simulations are feasible , the inference of the model parameters from experimental data is computationally too intensive . In this manuscript , we evaluate adjoint sensitivity analysis for parameter estimation in large scale biochemical reaction networks . We present the approach for time-discrete measurement and compare it to state-of-the-art methods used in systems and computational biology . Our comparison reveals a significantly improved computational efficiency and a superior scalability of adjoint sensitivity analysis . The computational complexity is effectively independent of the number of parameters , enabling the analysis of large- and genome-scale models . Our study of a comprehensive kinetic model of ErbB signaling shows that parameter estimation using adjoint sensitivity analysis requires a fraction of the computation time of established methods . The proposed method will facilitate mechanistic modeling of genome-scale cellular processes , as required in the age of omics . In the life sciences , the abundance of experimental data is rapidly increasing due to the advent of novel measurement devices . Genome and transcriptome sequencing , proteomics and metabolomics provide large datasets [1] at a steadily decreasing cost . While these genome-scale datasets allow for a variety of novel insights [2 , 3] , a mechanistic understanding on the genome scale is limited by the scalability of currently available computational methods . For small- and medium-scale biochemical reaction networks mechanistic modeling contributed greatly to the comprehension of biological systems [4] . Ordinary differential equation ( ODE ) models are nowadays widely used and a variety of software tools are available for model development , simulation and statistical inference [5–7] . Despite great advances during the last decade , mechanistic modeling of biological systems using ODEs is still limited to processes with a few dozens biochemical species and a few hundred parameters . For larger models rigorous parameter inference is intractable . Hence , new algorithms are required for massive and complex genomic datasets and the corresponding genome-scale models . Mechanistic modeling of a genome-scale biochemical reaction network requires the formulation of a mathematical model and the inference of its parameters , e . g . reaction rates , from experimental data . The construction of genome-scale models is mostly based on prior knowledge collected in databases such as KEGG [8] , REACTOME [9] and STRING [10] . Based on these databases a series of semi-automatic methods have been developed for the assembly of the reaction graph [11–13] and the derivation of rate laws [14 , 15] . As model construction is challenging and as the information available in databases is limited , in general , a collection of candidate models can be constructed to compensate flaws in individual models [16] . For all these model candidates the parameters have to be estimated from experimental data , a challenging and usually ill-posed problem [17] . To determine maximum likelihood ( ML ) and maximum a posteriori ( MAP ) estimates for model parameters , high-dimensional nonlinear and non-convex optimization problems have to be solved . The non-convexity of the optimization problem poses challenges , such as local minima , which have to be addressed by the selection of optimization methods . Commonly used global optimization methods are multi-start local optimization [18] , evolutionary and genetic algorithms [19] , particle swarm optimizers [20] , simulated annealing [21] and hybrid optimizers [22 , 23] ( see [18 , 24–26] for a comprehensive survey ) . For ODE models with a few hundred parameters and state variables multi-start local optimization methods [18] and related hybrid methods [27] have proven to be successful . These optimization methods use the gradient of the objective function to establish fast local convergence . While the convergence of gradient based optimizers can be significantly improved by providing exact gradients ( see e . g . [18 , 28 , 29] ) , the gradient calculation is often the computationally most demanding step . The gradient of the objective function is usually approximated by finite differences . As this method is neither numerically robust nor computationally efficient , several parameter estimation toolboxes employ forward sensitivity analysis . This decreases the numerical error and computation time [18] . However , the dimension of the forward sensitivity equations increases linearly with both the number of state variables and parameters , rendering its application for genome-scale models problematic . In other research fields such as mathematics and engineering , adjoint sensitivity analysis is used for parameter estimation in ordinary and partial differential equation models . Adjoint sensitivity analysis is known to be superior to the forward sensitivity analysis when the number of parameters is large [30] . Adjoint sensitivity analysis has been used for inference of biochemical reaction networks [31–33] . However , the methods were never picked up by the systems and computational biology community , supposedly due to the theoretical complexity of adjoint methods , a missing evaluation on a set of benchmark models , and an absence of an easy-to-use toolbox . In this manuscript , we provide an intuitive description of adjoint sensitivity analysis for parameter estimation in genome-scale biochemical reaction networks . We describe the end value problem for the adjoint state in the case of discrete-time measurement and provide an user-friendly implementation to compute it numerically . The method is evaluated on seven medium- to large-scale models . By using adjoint sensitivity analysis , the computation time for calculating the objective function gradient becomes effectively independent of the number of parameters with respect to which the gradient is evaluated . Furthermore , for large-scale models adjoint sensitivity analysis can be multiple orders of magnitude faster than other gradient calculation methods used in systems biology . The reduction of the time for gradient evaluation is reflected in the computation time of the optimization . This renders parameter estimation for large-scale models feasible on standard computers , as we illustrate for a comprehensive kinetic model of ErbB signaling . We consider ODE models for biochemical reaction networks , x ˙ = f ( x , θ ) , x ( t 0 ) = x 0 ( θ ) , ( 1 ) in which x ( t , θ ) ∈ R n x is the concentration vector at time t and θ ∈ R n θ denotes the parameter vector . Parameters are usually kinetic constants , such as binding affinities as well as synthesis , degradation and dimerization rates . The vector field f : R n x × R n θ ↦ R n x describes the temporal evolution of the concentration of the biochemical species . The mapping x 0 : R n θ ↦ R n x provides the parameter dependent initial condition at time t0 . As available experimental techniques usually do not provide measurements of the concentration of all biochemical species , we consider the output map h : R n x × R n θ ↦ R n y . This map models the measurement process , i . e . the dependence of the output ( or observables ) y ( t , θ ) ∈ R n y at time point t on the state variables and the parameters , y ( t , θ ) = h ( x ( t , θ ) , θ ) . ( 2 ) The i-th observable yi can be the concentration of a particular biochemical species ( e . g . yi = xl ) as well as a function of several concentrations and parameters ( e . g . yi = θm ( xl1 + xl2 ) ) . We consider discrete-time , noise corrupted measurements y ¯ i j = y i ( t j , θ ) + ϵ i j , ϵ i j ∼ N ( 0 , σ i j 2 ) , ( 3 ) yielding the experimental data D = { ( ( y ¯ i j ) i = 1 n y , t j ) } j = 1 N . The number of time points at which measurements have been collected is denoted by N . Remark: For simplicity of notation we assume throughout the manuscript that the noise variances , σ i j 2 , are known and that there are no missing values . However , the methods we will present in the following as well as the respective implementations also work when this is not the case . For details we refer to the S1 Supporting Information . We estimate the unknown parameter θ from the experimental data D using ML estimation . Parameters are estimated by minimizing the negative log-likelihood , an objective function indicating the difference between experiment and simulation . In the case of independent , normally distributed measurement noise with known variances the objective function is given by J ( θ ) = 1 2 ∑ i = 1 n y ∑ j = 1 N y ¯ i j - y i ( t j , θ ) σ i j 2 , ( 4 ) where yi ( tj , θ ) is the value of the output computed from Eqs ( 1 ) and ( 2 ) for parameter value θ . The minimization , θ * = arg min θ ∈ Θ J ( θ ) , ( 5 ) of this weighted least squares J yields the ML estimate of the parameters . The optimization problem Eq ( 5 ) is in general nonlinear and non-convex . Thus , the objective function can possess multiple local minima and global optimization strategies need to be used . For ODE models multi-start local optimization has been shown to perform well [18] . In multi-start local optimization , independent local optimization runs are initialized at randomly sampled initial points in parameter space . The individual local optimizations are run until the stopping criteria are met and the results are collected . The collected results are visualized by sorting them according to the final objective function value . This visualization reveals local optima and the size of their basin of attraction . For details we refer to the survey by Raue et al . [18] . In this study , initial points are generated using latin hypercube sampling and local optimization is performed using the interior point and the trust-region-reflective algorithm implemented in the MATLAB function fmincon . m . Gradients are computed using finite differences , forward sensitivity analysis or adjoint sensitivity analysis . A näive approximation to the gradient of the objective function with respect to θk is obtained by finite differences , ∂ J ∂ θ k ≈ J ( θ + a e k ) - J ( θ - b e k ) a + b , ( 6 ) with a , b ≥ 0 and the kth unit vector ek . In practice forward differences ( a = ϵ , b = 0 ) , backward differences ( a = 0 , b = ϵ ) and central differences ( a = ϵ , b = ϵ ) are widely used . For the computation of forward finite differences , this yields a procedure with three steps: In theory , forward and backward differences provide approximations of order ϵ while central differences provide more accurate approximations of order ϵ2 , provided that J is sufficiently smooth . In practice the optimal choice of a and b depends on the accuracy of the numerical integration [18] . If the integration accuracy is high , an accurate approximation of the gradient can be achieved using a , b ≪ 1 . For lower integration accuracies , larger values of a and b usually yield better approximations . A good choice of a and b is typically not clear a priori ( cf . [34] and the references therein ) . The computational complexity of evaluating gradients using finite differences is affine linear in the number of parameters . Forward and backward differences require in total nθ + 1 function evaluations . Central differences require in total 2nθ function evaluations . As already a single simulation of a large-scale model is time-consuming , the gradient calculation using finite differences can be limiting . State-of-the-art systems biology toolboxes , such as the MATLAB toolbox Data2Dynamics [7] , use forward sensitivity analysis for gradient evaluation . The gradient of the objective function is ∂ J ∂ θ k = ∑ i = 1 n y ∑ j = 1 N y ¯ i j - y i ( t j , θ ) σ i j 2 s i , k y ( t j ) , ( 7 ) with s i , k y ( t ) : [ t 0 , t N ] ↦ R denoting the sensitivity of output yi at time point t with respect to parameter θk . Governing equations for the sensitivities are obtained by differentiating Eqs ( 1 ) and ( 2 ) with respect to θk and reordering the derivatives . This yields s ˙ k x = ∂ f ∂ x s k x + ∂ f ∂ θ k , s k x ( t 0 ) = ∂ x 0 ∂ θ k s i , k y = ∂ h i ∂ x s k x + ∂ h i ∂ θ k ( 8 ) with s k x ( t ) : [ t 0 , t N ] ↦ R n x denoting the sensitivity of the state x with respect to θk . Note that here and in the following , the dependencies of f , h , x0 and their ( partial ) derivatives on t , x and θ are not stated explicitly but have the to be assumed . For a more detailed presentation we refer to the S1 Supporting Information Section 1 . Forward sensitivity analysis consists of three steps: Step 1 and 2 are often combined , which enables simultaneous error control and the reuse of the Jacobian [30] . The simultaneous error control allows for the calculation of accurate and reliable gradients . The reuse of the Jacobian improves the computational efficiency . The number of state and output sensitivities increases linearly with the number of parameters . While this is unproblematic for small- and medium-sized models , solving forward sensitivity equations for systems with several thousand state variable bears technical challenges . Code compilation can take multiple hours and require more memory than what is available on standard machines . Furthermore , while forward sensitivity analysis is usually faster than finite differences , in practice the complexity still increases roughly linearly with the number of parameters . In the numerics community , adjoint sensitivity analysis is frequently used to compute the gradients of a functional with respect to the parameters if the function depends on the solution of a differential equation [35] . In contrast to forward sensitivity analysis , adjoint sensitivity analysis does not rely on the state sensitivities s k x ( t ) but on the adjoint state p ( t ) . The calculation of the objective function gradient using adjoint sensitivity analysis consists of three steps: Step 1 and 2 , which are usually the computationally intensive steps , are independent of the parameter dimension . The complexity of Step 3 increases linearly with the number of parameters , yet the computation time required for this step is typically negligible . The calculation of state and output trajectories ( Step 1 ) is standard and does not require special methods . The non-trivial element in adjoint sensitivity analysis is the calculation of the adjoint state p ( t ) ∈ R n x ( Step 2 ) . For discrete-time measurements—the usual case in systems and computational biology—the adjoint state is piece-wise continuous in time and defined by a sequence of backward differential equations . For t > tN , the adjoint state is zero , p ( t ) = 0 . Starting from this end value the trajectory of the adjoint state is calculated backwards in time , from the last measurement t = tN to the initial time t = t0 . At the time points at which measurements have been collected , tN , … , t1 , the adjoint state is reinitialised as p ( t j ) = lim t → t j + p ( t ) + ∑ i = 1 n y ∂ h i ∂ x T y ¯ i j - y i ( t j ) σ i j 2 , ( 9 ) which usually results in a discontinuity of p ( t ) at tj . Starting from the end value p ( tj ) as defined in Eq ( 9 ) the adjoint state evolves backwards in time until the next measurement point tj−1 or the initial time t0 is reached . This evolution is governed by the time-dependent linear ODE p ˙ = - ∂ f ∂ x T p . ( 10 ) The repeated evaluation of Eqs ( 9 ) and ( 10 ) until t = t0 yields the trajectory of the adjoint state . Given this trajectory , the gradient of the objective function with respect to the individual parameters is ∂ J ∂ θ k = - ∫ t 0 t N p T ∂ f ∂ θ k d t - ∑ i , j ∂ h i ∂ θ k y ¯ i j - y i ( t j ) σ i j 2 - p ( t 0 ) T ∂ x 0 ∂ θ k . ( 11 ) Accordingly , the availability of the adjoint state simplifies the calculation of the objective function to nθ one-dimensional integration problems over short time intervals whose union is the total time interval [t0 , tN] . Algorithm 1: Gradient evaluation using adjoint sensitivity analysis % State and output Step 1 Compute state and output trajectories using Eqs ( 1 ) and ( 2 ) . % Adjoint state Step 2 . 1 Set end value for adjoint state , ∀t > tN: p ( t ) = 0 . for j = N to 1 do Step 2 . 2 Compute end value for adjoint state according to the jth measurement using Eq ( 9 ) . Step 2 . 3 Compute trajectory of adjoint state on time interval t = ( tj−1 , tj] by solving Eq ( 10 ) . end % Objective function gradient for k = 1 to nθ do Step 3 Evaluation of the sensitivity ∂J/∂θk using Eq ( 11 ) . end Pseudo-code for the calculation of the adjoint state and the objective function gradient is provided in Algorithm 1 . We note that in order to use standard ODE solvers the end value problem Eq ( 10 ) can be transformed in an initial value problem by applying the time transformation τ = tN − t . The derivation of the adjoint sensitivities for discrete-time measurements is provided in the S1 Supporting Information Section 1 . The key difference of the adjoint compared to the forward sensitivity analysis is that the derivatives of the state and the output trajectory with respect to the parameters are not explicitly calculated . Instead , the sensitivity of the objective function is directly computed . This results in practice in a computation time of the gradient which is almost independent of the number of parameters . A visual summary of the different sensitivity analysis methods is provided in Fig 1 . Besides the procedures also the computational complexity is indicated . The implementation of adjoint sensitivity analysis is non-trivial and error-prone . To render this method available to the systems and computational biology community , we implemented the Advanced Matlab Interface for CVODES and IDAS ( AMICI ) . This toolbox allows for a simple symbolic definition of ODE models ( 1 ) and ( 2 ) as well as the automatic generation of native C code for efficient numerical simulation . The compiled binaries can be executed from MATLAB for the numerical evaluation of the model and the objective function gradient . Internally , the SUNDIALS solvers suite is employed [30] , which offers a broad spectrum of state-of-the-art numerical integration of differential equations . In addition to the standard functionality of SUNDIALS , our implementation allows for parameter and state dependent discontinuities . The toolbox and a detailed documentation can be downloaded from http://ICB-DCM . github . io/AMICI/ . For the comparison of different gradient calculation methods , we consider a set of standard models from the Biomodels Database [37] and the BioPreDyn benchmark suite [27] . From the biomodels database we considered models for the regulation of insulin signaling by oxidative stress ( BM1 ) [38] , the sea urchin endomesoderm network ( BM2 ) [39] , and the ErbB sigaling pathway ( BM3 ) [40] . From BioPreDyn benchmark suite we considered models for central carbon metabolism in E . coli ( B2 ) [41] , enzymatic and transcriptional regulation of carbon metabolism in E . coli ( B3 ) [42] , metabolism of CHO cells ( B4 ) [43] , and signaling downstream of EGF and TNF ( B5 ) [44] . Genome-wide kinetic metabolic models of S . cerevisiae and E . coli ( B1 ) [45] contained in the BioPreDyn benchmark suite and the Biomodels Database [15 , 45] were disregarded due to previously reported numerical problems [27 , 45] . The considered models possess 18-500 state variable and 86-1801 parameters . A comprehensive summary regarding the investigated models is provided in Table 1 . To obtain realistic simulation times for adjoint sensitivities realistic experimental data is necessary ( see S1 Supporting Information Section 3 ) . For the BioPreDyn models we used the data provided in the suite , for the ErbB signaling pathway we used the experimental data provided in the original publication and for the remaining models we generated synthetic data using the nominal parameter provided in the SBML definition . In the following , we will compare the performance of forward and adjoint sensitivities for these models . As the model of ErbB signaling has the largest number of state variables and is of high practical interest in the context of cancer research , we will analyze the scalability of finite differences and forward and adjoint sensitivity analysis for this model in greater detail . Moreover , we will compare the computational efficiency of forward and adjoint sensitivity analysis for parameter estimation for the model of ErbB signaling . The evaluation of the objective function gradient is the computationally demanding step in deterministic local optimization . For this reason , we compared the computation time for finite differences , forward sensitivity analysis and adjoint sensitivity analysis and studied the scalability of these approaches at the nominal parameter θ0 which was provided in the SBML definitions of the investigated models . For the comprehensive model of ErbB signaling we found that the computation times for finite differences and forward sensitivity analysis behave similarly ( Fig 2a ) . As predicted by the theory , for both methods the computation time increased linearly with the number of parameters . Still , forward sensitivities are computationally more efficient than finite differences , as reported in previous studies [18] . Adjoint sensitivity analysis requires the solution to the adjoint problem , independent of the number of parameters . For the considered model , solving the adjoint problem a single time takes roughly 2-3-times longer than solving the forward problem . Accordingly , adjoint sensitivity analysis with respect to a small number of parameter is disadvantageous . However , adjoint sensitivity analysis scales better than forward sensitivity analysis and finite differences . Indeed , the computation time for adjoint sensitivity analysis is almost independent of the number of parameters . While computing the sensitivity with respect to a single parameter takes on average 10 . 09 seconds , computing the sensitivity with respect to all 219 parameters takes merely 14 . 32 seconds . We observe an average increase of 1 . 9 ⋅ 10−2 seconds per additional parameter for adjoint sensitivity analysis which is significantly lower than the expected 3 . 24 seconds for forward sensitivity analysis and 4 . 72 seconds for finite differences . If the sensitivities with respect to more than 4 parameters are required , adjoint sensitivity analysis outperforms both forward sensitivity analysis and finite differences . For 219 parameters , adjoint sensitivity analysis is 48-times faster than forward sensitivities and 72-times faster than finite differences . To ensure that the observed speedup is not unique to the model of ErbB signaling ( BM3 ) we also evaluated the speedup of adjoint sensitivity analysis over forward sensitivity analysis on models B2-5 and BM1-2 . The results are presented in Fig 2b and 2c . We find that for all models , but model B3 , gradient calculation using adjoint sensitivity is computationally more efficient than gradient calculation using forward sensitivities ( speedup > 1 ) . For model B3 the backwards integration required a much higher number of integration steps ( 4 ⋅ 106 ) than the forward integration ( 6 ⋅ 103 ) , which results to a poor performance of the adjoint method . One reason for this poor performance could be that , in contrast to other models , the right hand side of the differential equation of model B3 consists almost exclusively of non-linear , non-mass-action terms . Excluding model B3 we find an polynomial increase in the speedup with respect to the number of parameters nθ ( Fig 2b ) , as predicted by theory . Moreover , we find that the product nθ ⋅ nx , which corresponds to the size of the system of forward sensitivity equations , is an even better predictor ( R2 = 0 . 99 ) than nθ alone ( R2 = 0 . 83 ) . This suggest that adjoint sensitivity analysis is not only beneficial for systems with a large number of parameters , but can also be beneficial for systems with a large number of state variables . As we are not aware of any similar observations in the mathematics or engineering community , this could be due to the structure of biological reaction networks . Our results suggest that adjoint sensitivity analysis is an excellent candidate for parameter estimation in large-scale models as it provides good scaling with respect to both , the number of parameters and the number of state variables . Efficient local optimization requires accurate and robust gradient evaluation [18] . To assess the accuracy of the gradient computed using adjoint sensitivity analysis , we compared this gradient to the gradients computed via finite differences and forward sensitivity analysis . Fig 3 visualizes the results for the model of ErbB signaling ( BM3 ) at the nominal parameter θ0 which was provided in the SBML definition . The results are similar for other starting points . The comparison of the gradients obtained using finite differences and adjoint sensitivity analysis revealed small discrepancies ( Fig 3a ) . The median relative difference ( as defined in S1 Supporting Information Section 2 ) between finite differences and adjoint sensitivity analysis is 1 . 5 ⋅ 10−3 . For parameters θk to which the objective function J was relatively insensitive , ∂J/∂θk < 10−2 , there are much higher discrepancies , up to a relative error of 2 . 9 ⋅ 103 . Forward and adjoint sensitivity analysis yielded almost identical gradient elements over several orders of magnitude ( Fig 3b ) . This was expected as both forward and adjoint sensitivity analysis exploit error-controlled numerical integration for the sensitivities . To assess numerical robustness of adjoint sensitivity analysis , we also compared the results obtained for high and low integration accuracies ( Fig 3c ) . For both comparisons we found the similar median relative and maximum relative error , namely 2 . 6 ⋅ 10−6 and 9 . 3 ⋅ 10−4 . This underlines the robustness of the sensitivitity based methods and ensures that differences observed in Fig 3a indeed originate from the inaccuracy of finite differences . Our results demonstrate that adjoint sensitivity analysis provides objective function gradients which are as accurate and robust as those obtained using forward sensitivity analysis . As adjoint sensitivity analysis provides accurate gradients for a significantly reduced computational cost , this can boost the performance of a variety of optimization methods . Yet , in contrast to forward sensitivity analysis , adjoint sensitivities do not yield sensitivities of observables and it is thus not possible to approximate the Hessian of the objective function via the Fisher Information Matrix [46] . This prohibits the use of possibly more efficient Newton-type algorithms which exploit second order information . Therefore , adjoint sensitivities are limited to quasi-Newton type optimization algorithms , e . g . the Broyden-Fletcher-Goldfarb-Shanno ( BFGS ) algorithm [47 , 48] , for which the Hessian is iteratively approximated from the gradient during optimization . In principle , the exact calculation of the Hessian and Hessian-Vector products is possible via second order forward and adjoint sensitivity analysis [49 , 50] , which possess similar scaling properties as the first order methods . However , both forward and adjoint approaches come at an additional cost and are thus not considered in this study . To assess whether the use of adjoint sensitivities for optimization is still viable , we compared the performance of the interior point algorithm using adjoint sensitivity analysis with the BFGS approximation of the Hessian to the performance of the trust-region reflective algorithm using forward sensitivity analysis with Fisher Information Matrix as approximation of the Hessian . For both algorithms we used the MATLAB implementation in fmincon . m . The employed setup of the trust-region algorithm is equivalent to the use of lsqnonlin . m which is the default optimization algorithm in the MATLAB toolbox Data2Dynamics [7] , which was employed to win several DREAM challenges . For the considered model the computation time of forward sensitivities is comparable in Data2Dynamics and AMICI . Therefore , we expect that Data2Dynamics would perform similar to the trust-region reflective algorithm coupled to forward sensitivity analysis . We evaluated the performance for the model of ErbB signaling based on 100 multi-starts which were initialized at the same initial points for both optimization methods . For 41 out of 100 initial points the gradient could not be evaluated due numerical problems . These optimization runs are omitted in all further analysis . To limit the expected computation to a bearable amount we allowed a maximum of 10 iterations for the forward sensitivity approach and 500 iterations for the adjoint sensitivity approach . As the previously observed speedup in gradient computation was roughly 48 fold , we expected this setup should yield similar computation times for both approaches . We found that for the considered number of iterations , both approaches perform similar in terms of objective function value compared across iterations ( Fig 4a and 4b ) . However , the computational cost of one iteration was much cheaper for the optimizer using adjoint sensitivity analysis . Accordingly , given a fixed computation time the interior-point method using adjoint sensitivities outperforms the trust-region method employing forward sensitivities and the FIM ( Fig 4c and 4d ) . In the allowed computation time , the interior point algorithm using adjoint sensitivities could reduce the objective function by up to two orders of magnitude ( Fig 4c ) . This was possible although many model parameters seem to be non-identifiable ( see S1 Supporting Information Section 4 ) , which can cause problems . To quantify the speedup of the optimization using adjoint sensitivity analysis over the optimization using forward sensitivity analysis , we performed a pairwise comparison of the minimal time required by the adjoint sensitivity approach to reach the final objective function value of the forward sensitivity approach for the individual points ( Fig 4e ) . The median speedup achieved across all multi-starts was 54 ( Fig 4f ) , which was similar to the 48 fold speedup achieved in the gradient computation . The availability of the Fisher Information Matrix for forward sensitivities did not compensate for the significantly reduced computation time achieved using adjoint sensitivity analysis . This could be due to the fact that adjoint sensitivity based approach , being able to carry out many iterations in a short time-frame , can build a reasonable approximation of the Hessian approximation relatively fast . In summary , this application demonstrates the applicability of adjoint sensitivity analysis for parameter estimation in large-scale biochemical reaction networks . Possessing similar accuracy as forward sensitivities , the scalability is improved which results in an increased optimizer efficiency . For the model of ErbB signaling , optimization using adjoint sensitivity analysis outperformed optimization using forward sensitivity analysis . Mechanistic mathematical modeling at the genome scale is an important step towards a holistic understanding of biological processes . To enable modeling at this scale , scalable computational methods are required which are applicable to networks with thousands of compounds . In this manuscript , we present a gradient computation method which meets this requirement and which renders parameter estimation for large-scale models significantly more efficient . Adjoint sensitivity analysis , which is extensively used in other research fields , is a powerful tool for estimating parameters of large-scale ODE models of biochemical reaction networks . Our study of several benchmark models with up to 500 state variables and up to 1801 parameters demonstrated that adjoint sensitivity analysis provides accurate gradients in a computation time which is much lower than for established methods and effectively independent of the number of parameters . To achieve this , the adjoint state is computed using a piece-wise continuous backward differential equation . This backward differential equation has the same dimension as the original model , yet the computation time required to solve it usually is slightly larger . As a result , finite differences and forward sensitivity analysis might be more efficient if the sensitivities with respect to a few parameters are required . The same holds for alternatives like complex-step derivative approximation techniques [51] and forward-mode automatic differentiation [28 , 52] . For systems with many parameters , adjoint sensitivity analysis is advantageous . A scalable alternative might be reverse-mode automatic differentiation [28 , 53] , which remains to be evaluated for the considered class of problems . For the model of ErbB signaling we could show that adjoint sensitivity based optimization outperforms forward sensitivity based optimization , which is the standard in most systems biology toolboxes . With the availability of the MATLAB toolbox AMICI the adjoint sensitivity based approach becomes accessible for other researchers . AMICI allows for the fully automated generation of executables for adjoint or forward sensitivity analysis from symbolic model definitions . This way , the toolbox is easy-to-use and can easily be integrated with existing toolboxes . Also other MATLAB toolboxes for computational modeling , e . g . AMIGO [6] , Data2Dynamics [7] , MEIGO [54] and SBtoolbox2 [55] could be extended to exploit adjoint sensitivity analysis . In addition to adjoint sensitivity analysis , these MATLAB toolboxes could exploit forward sensitivity analysis available via AMICI , as AMICI yields computation times comparable to those of tailored numerical methods such as odeSD [56] ( S1 Supporting Information Section 5 ) or Data2Dynamics [7] . Moreover AMICI comes with detailed documentation and is already now used by several research labs . Our study of the model of ErbB signaling suggests that for the available data , a large number of parameters remains non-identifiable . While novel technologies provide rich dataset , we expect that non-identifiability will remain a problem . In particular if merely relative measurements are available , as the case for many measurement techniques , additional unknown scaling factors need to be introduced . These scaling factors are , in combination with initial conditions and total abundances , often the source of practical and structural non-identifiabilites [18] . Fortunately , for a broad range of biological questions , these information are not necessary and also state-of-the-art methods optimization seem to work reasonably well in the presence of non-identifiabilities . For the considered model of EreB signaling , we were able to achieve a significant decrease in the objective function value , despite the non-identifiability of parameters . This demonstrates that gradient based optimization is still feasible for large-scale problems . Yet , we believe that convergence of the optimizer could be improved by regularizing the objective function by integrating prior knowledge , possibly in a Bayesian framework [57] , from databases such as SABIO-RK [58] or BRENDA [59] . Beyond the use in optimization , gradients computed using adjoint sensitivity analysis will also facilitate the development of more efficient uncertainty analysis methods . Riemann manifold Langevin and Hamiltonian Monte Carlo methods [60 , 61] exploit the first and second order local structure of the posterior distribution and profit from more efficient gradient evaluation . The same holds for novel emulator-based sampling procedures [62] and approaches for posterior approximation [63] . By exploiting the proposed approach , rigorous Bayesian parameter estimation for models with hundreds of parameters could become a standard tool instead of an exception [64 , 65] . In conclusion , adjoint sensitivity analysis will facilitate the development of large- and genome-scale mechanistic models for cellular processes as well as other ( multi-scale ) biological processes [66] . This will complement available statistical analysis methods for omics data [67] by providing mechanistic insights and render a holistic understanding feasible .	In this manuscript , we introduce a scalable method for parameter estimation for genome-scale biochemical reaction networks . Mechanistic models for genome-scale biochemical reaction networks describe the behavior of thousands of chemical species using thousands of parameters . Standard methods for parameter estimation are usually computationally intractable at these scales . Adjoint sensitivity based approaches have been suggested to have superior scalability but any rigorous evaluation is lacking . We implement a toolbox for adjoint sensitivity analysis for biochemical reaction network which also supports the import of SBML models . We show by means of a set of benchmark models that adjoint sensitivity based approaches unequivocally outperform standard approaches for large-scale models and that the achieved speedup increases with respect to both the number of parameters and the number of chemical species in the model . This demonstrates the applicability of adjoint sensitivity based approaches to parameter estimation for genome-scale mechanistic model . The MATLAB toolbox implementing the developed methods is available from http://ICB-DCM . github . io/AMICI/ .	[ "Abstract", "Introduction", "Methods", "Results", "Discussion" ]	[ "applied", "mathematics", "simulation", "and", "modeling", "algorithms", "optimization", "genomic", "databases", "mathematics", "genome", "analysis", "research", "and", "analysis", "methods", "genome", "complexity", "biological", "databases", "differential", "equations", ...	2017	Scalable Parameter Estimation for Genome-Scale Biochemical Reaction Networks
Chromatin insulators are genetic elements implicated in the organization of chromatin and the regulation of transcription . In Drosophila , different insulator types were characterized by their locus-specific composition of insulator proteins and co-factors . Insulators mediate specific long-range DNA contacts required for the three dimensional organization of the interphase nucleus and for transcription regulation , but the mechanisms underlying the formation of these contacts is currently unknown . Here , we investigate the molecular associations between different components of insulator complexes ( BEAF32 , CP190 and Chromator ) by biochemical and biophysical means , and develop a novel single-molecule assay to determine what factors are necessary and essential for the formation of long-range DNA interactions . We show that BEAF32 is able to bind DNA specifically and with high affinity , but not to bridge long-range interactions ( LRI ) . In contrast , we show that CP190 and Chromator are able to mediate LRI between specifically-bound BEAF32 nucleoprotein complexes in vitro . This ability of CP190 and Chromator to establish LRI requires specific contacts between BEAF32 and their C-terminal domains , and dimerization through their N-terminal domains . In particular , the BTB/POZ domains of CP190 form a strict homodimer , and its C-terminal domain interacts with several insulator binding proteins . We propose a general model for insulator function in which BEAF32/dCTCF/Su ( HW ) provide DNA specificity ( first layer proteins ) whereas CP190/Chromator are responsible for the physical interactions required for long-range contacts ( second layer ) . This network of organized , multi-layer interactions could explain the different activities of insulators as chromatin barriers , enhancer blockers , and transcriptional regulators , and suggest a general mechanism for how insulators may shape the organization of higher-order chromatin during cell division . The physical organization of eukaryotic chromosomes is key for a large number of cellular processes , including DNA replication , repair and transcription [1]–[6] . Chromatin insulators are genetic elements implicated in the organization of chromatin and the regulation of transcription by two independent modes of action: ‘enhancer blocking’ insulators ( EB insulators ) interfere with communications between regulatory elements and promoters , whereas ‘barrier’ insulators prevent the spread of silenced chromatin states into neighboring regions [7]–[9] . Recently , insulator elements have been implicated in chromosome architecture and transcription regulation through their predicted binding to thousands of sites genome-wide . For instance , insulators were shown to regulate transcription of distinct gene ontologies , to separate distinct epigenetic chromatin states , and to recruit H3K27me3 domains to Polycomb bodies [10]–[13] . In Drosophila , five insulator families have been identified , that differ by their DNA-binding protein ( insulator binding protein , or IBP ) : Suppressor of Hairy-wing [Su ( Hw ) ] [14] , boundary element-associated factor ( BEAF32 ) [15] , Zeste-white 5 ( Zw5 ) [16] , the GAGA factor ( GAF ) [17] , and dCTCF [18] , a distant sequence homologue of mammalian CTCF . Two BEAF32 isoforms exist ( BEAF32A and BEAF32B ) . In this paper , we will only consider BEAF32B ( which will be referred to as BEAF32 ) as: ( i ) BEAF32B represents more than 95% of the binding peaks detected by chip-seq in cell lines [11] , ( ii ) BEAF32A binding does not play a role in the insulating function of BEAF [19] , and ( iii ) BEAF32A expression is not essential for the development of embryos in adult flies [20] . IBPs are often necessary but not sufficient to ensure insulation activity at a specific locus , and several insulator co-factors have been shown to be additionally required . Particularly , Centrosomal Protein 190 ( CP190 ) [21] , a protein originally described for its ability to bind to the centrosome during mitosis [22] , was shown to play a crucial role in the insulation function of various IBPs [10] , [23] , [24] . Insulator proteins often associate in clusters of overlapping binding sites more often than would be expected by chance , suggesting that these factors often bind as a complex to the same genetic locus . For instance , BEAF32 , dCTCF and CP190 binding sites most often cluster with at least another factor ( ∼70 , ∼77 and >90% , respectively ) [25] . In addition , insulators show a large compositional complexity , as demonstrated by the frequencies of binding of different combinations of insulator associated proteins: CP190 associates with its most common partner BEAF32 ( ∼50% ) , but also to a lesser extent to dCTCF and Su ( HW ) ( 25 and 20% , respectively ) , while BEAF32 , dCTCF , and CP190 cluster together in >15% of CP190 binding sites [10] , [25] , [26] . This compositional complexity may be key to understanding the locus-specific functions of insulators . A critical feature of Drosophila and vertebrate insulators is their ability to form specific long-range DNA interactions ( hereafter LRIs ) [27]–[34] . Three-dimensional loops have been implicated in all levels of chromatin organization ranging from kb-size loops to larger intra-chromosomal loops hundreds of kb in size [6] , [35] , [36] . To date , it is unclear what factors provide the physical interactions required for the formation and regulation of LRIs . In addition to binding the specific insulator sequences , IBPs have been proposed to be sufficient to bridge two distant DNA molecules [10] , [37] . However , other factors such as CP190 , Mod ( mdg4 ) , or cohesin have been implicated in the formation of LRIs [10] , [38]–[40] . The observation that most CP190 binding sites co-localize with insulator binding proteins ( >90% ) [10] , [25] prompted the hypothesis that CP190 is a common regulator of different insulator classes [10] , [40] . CP190 is composed of a BTB ( bric-a-brac , tramrack , and broad complex ) /POZ ( poxvirus and zinc-finger ) domain , four predicted C2H2 zinc-finger motifs , and an E-rich , C-terminal region . Importantly , CP190 has been recently shown to preferentially mark chromatin domain barriers [13] . These barriers are also heavily bound by other insulator proteins , such as BEAF32 , dCTCF and to a lesser extent Su ( HW ) , and have been shown to often form LRIs [12] . Overall , these data suggest a role for CP190 in participating in the three dimensional folding of the genome by the formation of long-range interactions . Surprisingly , a second factor , called Chromator , was also shown to be overrepresented at physical domain barriers [13] . During mitosis , Chromator forms a molecular spindle matrix with other nuclear-derived proteins ( Skeletor and Megator ) [41] . In contrast , during interphase Chromator localizes to inter-band regions of polytene chromosomes [42] , [43] and plays a role in their structural regulation as well as in transcriptional regulation [44] . Chromator can be divided into two main domains , a C-terminal domain containing a nuclear localization signal , and an N-terminal domain containing a chromo-domain ( ChD ) required for proper localization to chromatin during interphase [45] . Here , we investigate the molecular associations between different components of insulator complexes ( BEAF32 , CP190 and Chromator ) by biochemical and biophysical means . We developed a unique assay to determine what factors are necessary and essential for the formation of long-range DNA interactions , and show that BEAF32 is necessary but not sufficient to bridge long-range interactions . In contrast , addition of CP190 or Chromator is sufficient to mediate LRI between specifically-bound BEAF32 nucleoprotein complexes . This ability of CP190 and Chromator to establish LRI requires specific contacts between BEAF32 and their C-terminal domains , and dimerization through their N-terminal domains . In particular , the BTB/POZ domains of CP190 form a strict homodimer , and its C-terminal domain interacts with several IBPs . We propose a general model for insulator function in which BEAF32/dCTCF/Su ( HW ) provide DNA specificity ( first layer proteins ) whereas CP190/Chromator are responsible for the physical interactions required for long-range contacts ( second layer ) . The multiplicity of interactions between insulator binding and associated proteins could thus explain the different activities of insulators as chromatin barriers , enhancer blockers , and transcriptional regulators . BEAF32 co-localizes genome-wide with CP190 and Chromator , but the molecular mechanisms underlying this co-localization are unknown . To investigate whether this observed co-localization was due to direct protein-protein interactions , we heterologously expressed and purified BEAF32 , CP190 , Chromator and several protein subdomains . BEAF32 was expressed as a MBP ( Maltose-Binding Protein ) fusion protein ( Figure 1A–B ) , since wild-type BEAF32 was mainly insoluble . CP190 , Chromator , their C-terminal domains ( CP190-C and Chromator-C , respectively ) , and CP190-BTB/POZ were heterologously expressed as His-tagged fusions ( Figure 1A–B , and Materials and Methods ) . After purification , proteins were >95% pure and were specifically recognized by the corresponding antibodies ( Figure 1B , and Materials and Methods ) . A typical example of co-localization of these factors can be found at the Tudor-SN locus , a genomic region that shows a strong localization pattern for BEAF32 , CP190 , and Chromator but not for dCTCF or Su ( HW ) ( Figure 1C ) , and contains six specific binding sites for BEAF32 ( CGATA motifs ) [19] . To directly test whether BEAF32 was able to specifically bind to this genomic site , we PCR-amplified a 447 bp DNA fragment from Tudor-SN that contained six CGATA motifs ( hereafter DNAtudor , Figure 1C ) . First , we used an electric mobility shift assay ( EMSA ) in which a plasmid containing the DNAtudor insertion was restricted and used as a substrate ( Figure 2A ) . The restriction reaction produced three different DNA fragments of 750 , 1627 and 4025 bp , the second of which contained the 447-bp DNAtudor insertion , and was the only DNA fragment harboring specific CGATA motifs . The specific binding of factors to these different DNA fragments was assessed by quantifying the disappearance of unbound DNA species , as bound species often produced smeared bands due to rapid association/dissociation of proteins from DNA at low affinities and due to the low resolution of the gel matrix . The binding of BEAF to DNA was specific , as only the DNAtudor-containing band was preferentially shifted by addition of BEAF32 ( Figure 2A ) . Secondly , to quantify the affinity and specificity of DNA binding by BEAF32 , we implemented a fluorescence anisotropy-based assay that directly measures the binding of proteins to DNA . The binding of proteins , such as BEAF32 , to short fluorescently-labeled DNA fragments decreases the rotational diffusion of the DNA molecule and increases the fluorescence anisotropy of the attached fluorophore ( Figure 2B ) [46] . BEAF32 binds with a moderate apparent affinity to non-specific DNA ( 58 bp DNA fragment with no CGATA motif , hereafter DNANS ) , and the binding isotherm can be well described by a simple single-site model ( Eq . 1 , Text S1 , KD = 165±30 nM , Figure 2C ) . In contrast , BEAF32 binds to a specific DNA fragment of the same length ( 58 bp; DNA fragment containing three CGATA motifs from Tudor-SN , hereafter DNAS ) with a higher affinity and displaying a degree of cooperativity ( Figure 2C ) . The binding isotherm cannot be fitted by a single-site model , thus we turned to a Hill model ( Eq . 2 , Text S1 ) with a resulting apparent affinity of KD = 68±5 nM and a Hill coefficient of n = 3±0 . 4 . In addition , the change in fluorescence anisotropy signal was larger for DNAS ( 32±2 anisotropy units ) than for DNANS ( 12±5 anisotropy units ) , indicating that BEAF32-DNAS makes a larger complex . Overall , these results indicate a cooperative binding of BEAF32 to CGATA motifs , suggesting oligomerization of BEAF32 at genomic sites containing multiple CGATA motifs . These results were consistent with competitive inhibition experiments ( Supplementary Figure S1A ) . Equivalent fits of the DNANS binding isotherm to a Hill model produced a Hill coefficient of 0 . 9±0 . 3 , consistent with cooperative binding of BEAF32 to DNAS being due to the presence of CGATA motifs . CP190 contains a BTB/POZ domain and is predicted to possess four classical C2H2 zinc-finger motifs that could be involved in direct DNA binding . It is unclear whether CP190 can directly associate to DNA , or rather relies on its binding to other factors to target specific binding sites [21] , [47] . To address this question , we investigated the ability of CP190 to bind to Tudor-SN . This locus displays CP190 binding by Chip-chip [25] , [48] ( Figure 1C and Supplementary Table S6 ) and may thus contain moderate affinity sites for CP190 . By EMSA , we observed that CP190 associated equally well to all DNA fragments , with no specificity shown for the DNAtudor-containing fragment ( Figure 2D ) . Next , we tested the binding specificity of CP190 by fluorescence anisotropy , using two different dsDNA fragments ( DNAS and DNANS ) . DNAS should contain the potential CP190 moderate affinity sites giving rise to the in vivo binding of CP190 to Tudor-SN , while DNANS is a DNA fragment of the same length but with a random sequence serving as a control for specificity . In agreement with EMSA , fluorescence anisotropy experiments showed moderate DNA binding affinity but no specificity ( KD = 109±5 nM , n = 2±0 . 3 for CP190 on both DNAS and DNANS , Figure 2E ) . These results are supported by competition experiments ( Supplementary Figure S1B ) , and are in agreement with similar experiments showing that CP190 fails to show any specificity when using a dsDNA fragment containing the predicted binding sequence of CP190 [25] ( Supplementary Figure S6 ) . Overall , these results are consistent with the specificity of in vivo binding of CP190 to Tudor-SN being mediated by other factors . Next , we tested whether the C-terminal domain of CP190 was involved in the ability of CP190 to bind DNA non-specifically by determining the DNA binding properties of CP190-C , a protein construct that contains neither BTB/POZ nor the zinc-finger motif ( Figure 1A ) . CP190-C was not able to bind DNAS ( Figure 2H ) , consistent with the non-specific association of CP190 to DNA being mediated by the N-terminal domain of CP190 . Binding competition experiments of pre-bound BEAF32-DNAS are inconsistent with CP190-BTB/POZ being involved in DNA binding ( Supplementary Figure S1D ) , but further experiments will be required to determine the contribution of the different domains in the N-terminus of CP190 to DNA association . In addition , we cannot exclude the possibility that other factors or post-translational modifications may partially affect the mechanism of DNA binding by CP190 . However , the ubiquitous co-localization of CP190 with factors displaying specific DNA-binding activities ( BEAF32 , dCTCF , Su ( HW ) ) ( >90% ) [25] suggests that the presence of CP190 at specific loci is mediated in most cases by other proteins . From these experiments , we cannot exclude the possibility that CP190 may bind specifically to other genomic sites . The ability of Chromator to associate to DNA has not been described so far , although its association to chromatin has been suggested to require its Chd-containing N-terminal domain ( Yao et al , 2012 ) . Despite the presence of high affinity in vivo sites for Chromator in Tudor-SN , our EMSA and fluorescence anisotropy experiments showed that Chromator binds DNA non-specifically ( Figure 2F–G ) and with a lower affinity than BEAF32 or CP190 ( KD = 360±30 nM and n = 2±0 . 2 , see Figure 2G and Supplementary Figure S1C ) . Chromator-C did not present any DNA binding activity ( Figure 2H ) , suggesting that Chromator binding to DNA requires its N-terminal domain or uncharacterized post-transcriptional modifications . Next , we investigated whether BEAF32 directly interacts with CP190 and Chromator by using several complementary approaches . First , we employed co-immunoprecipitation ( co-IP ) to detect protein-protein interactions with heterologously purified proteins . A guinea pig-anti-Chromator-antibody was covalently linked to a column and a mix of purified BEAF32 and Chromator were incubated in the column for 60 min , eluted and analyzed by western blotting ( Figure 3A , see full bands of all co-IPs in Supplementary Fig . S2C–I ) . Western-Blot analysis of the elution clearly showed the specific interaction between BEAF32 and Chromator ( Figure 3A , middle column ) , whereas neither BEAF32 nor Chromator were found to bind to an IgG-antibody column ( Figure 3A , right column ) . Importantly , BEAF32 did not bind to an anti-Chromator column in the absence of Chromator ( Supplementary Figure S2B ) . To investigate what domain of Chromator is involved in interactions with BEAF32 , we performed co-IP experiments in which a mix of BEAF32 and Chromator-C were incubated in a column covalently bound by antiChromator antibody , and the elution analyzed by western blotting . Interestingly , BEAF32 is specifically retained in the Chromator-C column , consistent with BEAF32/Chromator interactions being mediated by the C-terminal domain of Chromator ( Figure 3B ) . Additionally , Chromator is retained in a CP190 column , an interaction that seems to be specifically mediated by CP190-BTB/POZ ( Supplementary Figure S2K ) . Similar co-IP experiments were performed to test putative BEAF32-CP190 interactions . A rabbit-anti-CP190-antibody was covalently linked to a resin and incubated with a purified mix of BEAF32 and CP190 or CP190-C . BEAF32 binds efficiently to both CP190 and CP190-C ( Figure 3C–D ) , but failed to interact with CP190-BTB/POZ ( Supplementary Figure S7 ) . BEAF32 is not recognized by CP190 antibodies and was not retained by an anti-CP190 column ( Supplementary Figures S2A–B ) . These results indicate that BEAF32/CP190 interactions are mediated by the C-terminal domain of CP190 , although we cannot discard an additional contribution of the zinc-finger domains of CP190 to this interaction . Interestingly , both BEAF32 and CP190 were retained in an anti-Chromator antibody column ( Figure 3E ) , consistent with binary interactions between BEAF32 and CP190/Chromator and with interactions between CP190 and Chromator . To test whether these interactions are physiologically relevant , we performed Co-IP experiments using S2 nuclear extracts ( see Materials and Methods ) . Interactions between BEAF32 , Chromator and CP190 were clearly detected while either using anti-Chromator or anti-CP190 ( Figures 3F–G , respectively ) antibodies . Overall , these results suggest that BEAF32 , Chromator and CP190 are part of the same molecular complex . However , further work is necessary to determine the architecture and stoichiometry of this complex . Next , we investigated whether interactions among BEAF32 , CP190 and Chromator lead to the formation of higher-order DNA interactions . First , we used EMSA to test whether BEAF32-CP190/Chromator sub-complexes bind to the 447 bp DNAtudor fragment ( Figure 1C ) . BEAF32 binding to DNAtudor ( Figure 4A , Lane 1 , band I ) produced a discrete shift corresponding to a BEAF32/DNAtudor complex ( Figure 4A , lane 2 , band II ) . Consistent with previous results , neither CP190/Chromator ( as the concentrations used here were lower than the KD ) , nor their C-terminal fragments were able to bind DNAtudor under these conditions ( Figure 4A , band I , lanes 3 , 5 , 7 and 9 , respectively ) . Interestingly , a second band with lower electrophoretic mobility appeared only when BEAF32 and either CP190 or Chromator were simultaneously present ( Figure 4A , lanes 4 and 6 , band III ) . Furthermore , this complex did not form when CP190 was replaced by CP190-C ( Figure 4A , compare lanes 4 and 8 ) , and exhibited a similar intensity when Chromator-C was used instead of full-length Chromator ( lane 10 in Figure 4A ) . Control experiments where BEAF32 was replaced by MBP showed that the formation of the protein-DNA complexes leading to bands II and III required the presence of BEAF32 ( Figure 4B ) . Band II thus corresponds to a complex formed by BEAF32 and DNAtudor , while band III indicates the presence of specific interactions between DNA-bound BEAF32 and CP190 , Chromator , and Chromator-C . To characterize the different complexes formed by BEAF32 , CP190 and Chromator , we turned to fluorescence correlation spectroscopy ( FCS ) . FCS uses the fluctuations in the number of freely diffusing fluorescently-labeled molecules within a confocal volume to characterize their diffusion time [49] , [50] ( Figure 5A ) . Thus , the formation of protein-DNA complexes can be monitored by the increase in the apparent size ( related to the diffusion time ) of a fluorescently-labeled dsDNA fragment upon protein binding . In our case , we used the 58 bp DNAS fragment ( harboring three CGATA motifs , 5′-Cy3B labeled , see Materials and Methods ) as a fluorescent reporter to quantitatively monitor the formation of BEAF32/CP190/Chromator complexes ( Figure 5D ) . Identical results were obtained with an atto655-DNAS probe ( Supplementary Figure S3 ) . Incubation of DNAS with saturating concentrations of BEAF32 ( 400 nM ) led to an increase in its apparent diffusion time from 0 . 53±0 . 03 to 0 . 85±0 . 04 ms , obtained by fitting our measurements to a 3-D diffusion model with a triplet state ( Eq . 3 , Text S1 ) . This shift is consistent with the binding of BEAF32 to DNAS leading to the production of a molecular complex ( hereafter B32S complex ) with an increased apparent size ( Figure 5B ) . The addition of low concentrations of CP190 ( 50 nM ) to DNAS produced a small increase in the diffusion time ( from 0 . 53±0 . 03 to 0 . 65±0 . 09 ms ) , consistent with the sub-affinity concentrations used . In contrast , CP190-C did not change the diffusion time of DNAS ( Figure 5B ) , in agreement with our previous results showing no DNA-binding activity for this domain of CP190 ( Figure 2H ) . Next , we investigated whether CP190 binds to B32S complexes . We observed that the incubation of pre-formed B32S complexes with low-concentrations of CP190 ( 50 nM ) led to a considerable increase in the size of complexes ( Figure 5C ) . This low CP190 concentration ( below its affinity ) was used to enhance the specificity of CP190/BEAF32 interactions and limit the direct binding of CP190 to DNAS . Conversely , the addition of CP190-C to B32S slightly decreased the apparent size of the complex ( Figure 5C ) . To ensure that this small decrease in diffusion time was not due to the dissociation of BEAF32 from DNAS , we performed fluorescence anisotropy experiments . The anisotropy of pre-formed B32S complexes was independent of the concentration of CP190-C , but decreased to the anisotropy of free DNAS upon addition of high salt concentrations ( Supplementary Figure S4 ) . These results indicate that the decrease in diffusion time observed in B32S/CP190-C complexes is not due to the dissociation of BEAF32 from DNAS , but to the change in the shape of the complex upon CP190-C binding . Overall , these results are consistent with either CP190 binding a B32S complex or triggering long-range inter-segment interactions between two B32S complexes . To discriminate between these two models , we turned to fluorescence cross-correlation spectroscopy ( FCCS ) . FCCS measures the correlated fluorescence intensity fluctuations of two spectrally-distinct , fluorescently-labeled molecules to quantitatively determine whether they are in the same molecular complex [50] , [51] . When two DNA fragments labeled with different colors are part of the same molecular complex , their fluorescence fluctuations will be correlated ( LRI ) , whereas no cross-correlation will be observed if the diffusion of the two DNA fragments is independent ( no LRI , Figure 5D ) . We used a 50/50 mixture of DNAS labeled with Cy3B and atto655 . Since these two fluorophores can display a significant level of crosstalk between detection channels , introducing apparent cross-correlation in the absence of interaction , we used pulsed interleaved excitation ( PIE-FCCS ) [52] , [53] a technique that eliminates this artifactual effect and allows quantitative fluorescence cross-correlation measurements . The cross-correlation signals were measured for DNAS , B32S , and solutions of pre-formed B32S complex incubated with either CP190 or CP190-C and fitted with Eq . 4 ( Text S1 ) . Neither DNAS , nor B32S showed cross-correlation ( Figure 5E ) , demonstrating the inability of BEAF32 alone to mediate long-range intermolecular interactions between CGATA motifs . In agreement with our previous observations ( lack of band III in lane 8 , Figure 4A ) , addition of CP190-C to B32S did not trigger the formation of intermolecular complexes , suggesting that the E-rich domain of CP190 is not sufficient to generate LRIs in vitro . In contrast , these complexes were formed in the presence of full-length CP190 , demonstrated by the appearance of a clear cross-correlation signal ( 18±6% , Figure 5E ) . From the DNA labeling efficiencies of Cy3B- and atto655-labeled oligonucleotides ( ∼57 and 97% , respectively ) , and the fact that a maximum of 50% of the bridged DNA can be observed in the cross-correlation amplitude ( since Cy3B-Cy3B or atto655-atto655 complexes do not produce a cross-correlation signal ) , we can conclude that 65±22% of the B32S-atto655 complexes take part in LRIs mediated by CP190 . Importantly , under these conditions CP190 alone was not able to generate LRIs ( Figure 5E ) , and addition of neither full-length CP190 nor CP190-C affected the specific binding of BEAF32 to DNAS ( Supplementary Figure S4 ) . Thus , while CP190-C interacts with BEAF32 , the N-terminal domain of CP190 appears necessary for the formation of inter-segment LRIs mediated by BEAF32-bound DNA in vitro ( as CP190-C is not sufficient to mediate these interactions ) . In agreement with this model , the competition of pre-formed B32S-CP190-B32S complexes with the purified , isolated CP190-BTB/POZ domain ( Figure 1A ) led to the disappearance of cross-correlation signal ( Figure 5F ) , but not to the displacement of BEAF32 from DNAS ( Supplementary Figure S1D ) . Overall , the FCCS data strongly suggest that the BTB/POZ domain of CP190 is involved in the direct protein-protein interactions required for the establishment of long-range contacts . To directly test this hypothesis , we solved the crystal structure of the CP190-BTB/POZ . BTB/POZ motifs are widespread in eukaryotes ( 350 BTB/POZ-containing proteins in the human genome ) . Despite a low degree of primary sequence conservation ( as low as 10% ) , the various structures reported in the literature are very similar ( root mean square deviation , or RMSD ∼1–2 Å ) with the overall architecture being composed of a cluster of five alpha helices capped on one end by three beta sheets . BTB/POZ motifs have been found to homodimerize , heterodimerize , and in rare cases to promote tetramerization . These different types of oligomerization states depend primarily on the surface residues involved in oligomerization and have been well documented elsewhere [54] , [55] . The CP190-BTB/POZ domain crystallized as a stable and symmetric homodimer , in agreement with gel-filtration analysis . The overall structure is similar to classic BTB/POZ-ZF transcriptional factors where the N-terminal BTB/POZ domain is followed by several Zinc-Fingers domains ( Figure 6A , Materials and Methods and Supplementary Table S1 ) . The dimerization interface is stabilized by a swapped β-strand that forms a long groove where extended polypeptidic segments can bind in order to recruit other protein partners . The dimer interface ( 1902 Å2/monomer according to PISA 1 . 47 [56] ) is composed of numerous hydrophobic interactions mainly from alpha helices α1 and α2 ( i . e . W12 , F15 , F16 , F23 , L47… ) . The native homodimeric organization is also reinforced by the N-terminal strand ( residues Glu2 to Asp10 ) being swapped: the β1 stand of a monomer interacts with the β5 strand of the other monomer . The sequence conservation among CP190 orthologs from insects ( 55 sequences analyzed using CONSURF [57] ) show little conservation besides the domain core , the dimerization interface and the peptide binding groove ( Figure 6A ) . Interestingly , this suggests that CP190-BTB/POZ does not form higher order macromolecular assemblies by itself while partner recruitment requires homo-dimerization . Importantly , we found that CP190-BTB/POZ forms strict homo-dimers ( Figure 6A ) , consistent with the ability of CP190 to form LRIs . Finally , we used FCS and PIE-FCCS to test whether Chromator was able to mediate LRIs between two B32S complexes in vitro . The addition of 100 nM Chromator to DNAS generated a small but noticeable change in the diffusion time ( 0 . 53±0 . 03 to 0 . 59±0 . 03 ms ) ( Figure 5G ) , consistent with our previous results ( Figures 2F and 2G ) indicating that Chromator interacts non-specifically with DNA . In contrast , incubation of DNAS with Chromator-C did not induce any change in the diffusion time of the probe ( Figure 5G ) , in agreement with anisotropy experiments ( Figure 2H ) . Interestingly , addition of Chromator ( but not of Chromator-C ) to B32S considerably changed the diffusion time of the complex ( Figure 5H ) , suggesting an interaction between Chromator and the B32S complex . Similarly to the results obtained for CP190 , addition of Chromator to pre-formed B32S complexes led to a cross-correlation amplitude of 13±4% , corresponding to a total of 47±14% of the B32S-atto655 complexes bridged by Chromator interactions ( Figure 5I ) . The formation of these complexes was not observed when Chromator-C was added to B32S , nor when Chromator was added to DNAS in the absence of BEAF32 ( Figure 5I ) . Overall , these results are consistent with interactions between the N-terminal domains of Chromator being required for the bridging function of Chromator , with Chromator-C providing the main direct interactions to BEAF32 . We cannot discard , however , the possibility that Chromator-N may also partially interact with BEAF32 . Recently , genome-wide approaches have been used to investigate the roles of different insulator types in genome organization . Insulators enriched in both BEAF32 and CP190 are implicated in the segregation of differentially expressed genes and in delimiting the boundaries of silenced chromatin [25] . Notably , BEAF32 and CP190 are often found to bind jointly to the same genetic locus ( >50% of CP190 binding sites contain BEAF32 ) [10] , [25] . However , the molecular origin of this genome-wide co-localization was unknown as there was no direct proof of interaction between these proteins . Here , we showed , for the first time to our knowledge , that BEAF32 is able to interact specifically with CP190 in vitro and in vivo . In particular , we observed that this interaction is mediated by the C-terminal domain of CP190 , with no implication of the C2H2 zinc-finger or the BTB/POZ domains , consistent with previous studies showing that the N-terminus of CP190 was not essential for its association with BEAF32 in vivo [58] . BEAF32 interacts specifically and cooperatively with DNA fragments containing CGATA motifs , consistent with previous observations [19] . In contrast , the binding of CP190 to DNA showed lower affinity and no specificity and required its N-terminal domain ( containing four C2H2 zinc-fingers ) . Overall , these data suggest that one pathway for CP190 recruitment to DNA genome-wide requires specific interactions of its C-terminal domain with BEAF32 . Other factors , such as GAF [38] , are likely also involved in the recruitment of CP190 to chromatin , explaining why RNAi depletion of BEAF32 does not lead to the dissociation of CP190 from an insulator binding class containing high quantities of BEAF32 and CP190 [47] . We cannot discard that post-translational modifications in CP190 may also allow it to bind DNA directly and specifically , providing a second pathway for locus-specific localization . In addition to acting as chromatin barriers , insulators have been typically characterized for their ability to block interactions between enhancers and promoters through the formation of long-range contacts [7] , [16] , [27] , [28] , [59]–[62] . Here , we developed a fluorescence cross correlation-based assay that allowed us , for the first time to our knowledge , to investigate the ability of BEAF32 , CP190 and their complex to bridge specific DNA fragments , mimicking LRIs . We show that specific LRI can be stably formed between two DNA fragments containing BEAF32 binding sites , solely in the presence of both BEAF32 and CP190 . Interestingly , LRI are displaced by competition in trans with the BTB/POZ domain of CP190 , and LRIs are not observed in the presence of BEAF32 and CP190-C . Thus , both protein domains are required for the bridging activity of CP190 . These data strongly suggest that the C-terminal domain is responsible for BEAF32-specific contacts whereas the N-terminal domain of CP190 is involved in the formation of LRI through CP190/CP190 contacts ( Figure 6B ) . The role of the N-terminal domain of CP190 in protein-protein interactions is consistent with previous studies showing that N-terminal fragments of CP190 containing the BTB/POZ domains co-localize with full-length CP190 in polytene chromosomes [58] . BTB/POZ are a family of protein-protein interaction motifs conserved from Drosophila to mammals , and present in a variety of transcriptional regulators . BTB/POZ are found primarily at the N-terminus of proteins containing C2H2 zinc-finger motifs [63]–[65] , and can be monomeric , dimeric , or multimeric [54] , [55] . In fact , a recent study proposed that isolated CP190-BTB/POZ domains can exist as dimers or tetramers in solution [66] . The oligomerization behavior of CP190-BTB/POZ could have important implications for the role and mechanism by which CP190 bridges LRIs . Here , we showed that the BTB/POZ domains of CP190 forms homo-dimers with a large , conserved interaction surface ( Figure 6A ) , consistent with these domains being responsible for the formation of the direct protein-protein interactions required for the establishment of long-range contacts . Interestingly , the oligomerization of CP190-BTB/POZ into homo-dimers implies a binary interaction between two distant DNA sequences , imposing important constraints for the mechanisms of DNA bridging by CP190 . In addition to interacting with BEAF32 , CP190 is able to directly interact with other insulator binding proteins , such as dCTCF , Su ( HW ) , and Mod ( Mdg4 ) [21] , [39] , [66]–[68] , or with the RNA interference machinery [69] . These interactions are usually mediated by the C-terminal domain of CP190 , but a role for the C2H2 zinc-finger or the BTB/POZ domains in providing specific protein-protein contacts cannot be discarded [70] . In fact , an interesting feature of several homo-dimeric BTB/POZ domains is their ability to recruit a multitude of protein partners using a single protein-protein binding interface . For instance , several transcriptional co-repressors ( BCOR , SMRT and NCor ) are able to bind with micromolar affinity ( 2∶2 stoichiometry ) to the BTB/POZ domain of BCL6 , despite their low sequence homology [71] , [72] . In this case , the mechanism of binding involves the formation of a third strand by the N-terminus of co-repressors folding onto the two strands exchanged by the BCL6-BTB/POZ monomers on their interface , with the rest of the minimal domain of interaction ( 10 residues ) winding up along the lateral groove of the BCL6-BTB/POZ dimer ( peptide binding groove in Figure 6A ) . In the case of CP190 , the sequence and structural features of the conserved peptide binding groove within insect CP190-BTB/POZ domains suggest that the dimer interface of CP190 may act as a protein-protein interaction platform . Thus , the ability of BTB/POZ domains to form dimers and the promiscuous binding of CP190 to different insulator binding proteins ( Su ( HW ) , dCTCF [39] , [67] , and BEAF32 ) suggest not only that insulators share protein components [73] , but also that CP190 may bridge long-range contacts involving distinct factors at each end of the DNA loop ( Figure 6B ) . This model is consistent with previous proposals [73] , and with the requirement of both C- and N-terminal domains of CP190 for fly viability [58] . Importantly , it provides a rationale for CP190 being a common factor between insulator binding proteins . CP190 frequently binds with additional insulator binding proteins ( ∼85% ) , with BEAF32 and dCTCF being the most common partners ( ∼50% and ∼25% , respectively ) , and Su ( Hw ) amongst the least frequent partner ( ∼20% ) [10] , [25] . Importantly , BEAF32 does not show clustering with either dCTCF or Su ( HW ) in the absence of CP190 ( <0 . 5% or ∼0 . 1% , respectively ) [10] , suggesting that the clustering of two insulator binding proteins requires CP190 . The ability of CP190 to mediate LRIs between sites harboring different insulator binding proteins raises important questions: Are these LRIs specific ? How is this specificity regulated ? Are other factors or post-translational modifications involved in this selectivity ? Future research will be needed to address these important questions . Chromator localizes to inter-band regions of polytene chromosomes [42] , [43] and binds to the barriers of physical domains genome-wide [13] , however the mechanism leading to these localization patterns has been lacking . Previous studies showed that BEAF32 and Chromator co-localize at some genomic sites , and suggested that these proteins may participate in the formation of a single complex [@Gan:2011hy] . Here , we showed for the first time that BEAF32 directly and specifically interacts with Chromator in vivo and in vitro . This interaction is mediated by the C-terminal domain of Chromator , thus the ChD domain does not seem to be directly involved in interactions with BEAF32 . Our results show that Chromator possesses a reduced affinity for DNA and binds with no sequence specificity to loci displaying strong Chromator binding peaks at the site tested ( Tudor-SN locus , Figures 2G and 1C ) . Thus , we suggest that specific interactions between BEAF32 and Chromator may be responsible for its recruitment to polytene inter-band regions and domain barriers . Significantly , most BEAF32 binding sites genome-wide ( >90% , Figure 6C and Supplementary Figure S5A ) contain Chromator , suggesting an almost ubiquitous interaction between the two factors . Interestingly , Chromator also co-localizes with the JIL-1 kinase at polytene inter-band regions and the two proteins directly interact by their C-terminal domains [41] . JIL-1 is an ubiquitous tandem kinase essential for Drosophila development and key in defining de-condensed domains of larval polytene chromosomes . Importantly , JIL-1 participates in a complex histone modification network that characterizes active , de-condensed chromatin , and is thought to reinforce the status of active chromatin through the phosphorylation of histone H3 at serine 10 ( H3S10 ) [74]–[76] . Thus , BEAF32 could be responsible for the recruitment of the Chromator/JIL-1 complex to active chromatin domains to prevent heterochromatin spreading ( Figure 6D ) [@Gan:2011hy] . This mechanism would be consistent with the observation that BEAF32 localizes primarily to de-condensed chromatin regions in polytene chromosomes [15] , is implicated in the regulation of active genes [10] , [11] , [25] , [77] and delimits the boundaries of chromatin silencing [25] . CP190 is a common partner of BEAF32 , dCTCF , and Su ( HW ) , and has been thus proposed to play a role in the formation of long-range interactions at these insulators [10] , [67] . On the other hand , both CP190 and Chromator have been recently shown to be massively overrepresented at barriers between transcriptional domains [12] , [13] . In this paper , we show , for the first time , that only when CP190 or Chromator are present can long-range interactions between BEAF32-bound DNA molecules be generated . We provide strong evidence that the formation of in vitro LRI requires three ingredients: ( 1 ) binding of BEAF32 to its specific DNA binding sites; ( 2 ) specific interactions between the C-terminal domains of CP190/Chromator and BEAF32; and ( 3 ) homo- interactions between CP190/Chromator molecules mediated by their N-terminal ends . To further investigate the roles of CP190 and Chromator in the formation of LRIs , we aggregated together statistically relevant contacts containing specific combinations of insulator factors from Hi-C data from embryos [13] ( Figure 6E , and Materials and Methods ) . This analysis shows a relatively high correlation between the presence of BEAF32 and both CP190 and Chromator in sites displaying a high proportion of interacting bins between distant BEAF32 sites ( Figure 6E ) , as compared with neighboring sites ( 16 . 9% of interacting bins for Chromator and CP190 sites; Wilcoxon test: p-value ∼1e-7 ) . Thus , CP190 and Chromator may play a role at a subset of genetic loci by mediating and/or stabilizing interactions between BEAF32 and a distant locus bound by BEAF32 or a different insulator binding protein . Interestingly , the binding of BEAF32 to CGATA sites as multimers , and the existence of CP190-Chromator interactions suggest that long-range interactions at a single locus could involve hybrid/mixed complexes comprising at least these three factors . These observations suggest a general model for insulator function in which BEAF32/dCTCF/Su ( HW ) provide DNA specificity ( first layer proteins ) whereas CP190/Chromator are responsible for the physical interactions required for long-range contacts ( second layer ) . Direct or indirect interactions of first layer insulator proteins with additional factors ( e . g . JIL-1 , NELF , mediator ) are very likely involved in directing alternative activities ( e . g . histone modifications , regulation of RNAPII pausing ) to specific chromatin loci . This model provides a rationale for the compositional complexity of insulator sequences [25] and for the multiplicity of functions often attributed to insulators ( e . g . enhancer blocker , chromatin barrier , transcriptional regulator ) . Ultimately , a characterization of the locus-specific composition of insulator complexes and their locus-specific function may be required to obtain a general picture of insulator function . In mammals , CTCF is the only insulator protein identified so far , but other factors , such as cohesin have been identified as necessary and essential for the formation of CTCF-mediated long-range interactions [28] , [30] , [32] . Mammalian CTCF contains eleven zinc-fingers , and it has been shown that different combinations of zinc-fingers could be used to bind different DNA sequences [78] . Thus , in mammals CTCF may play the role of first layer insulator protein , whereas other factors such as cohesin or mediator may play the role of second layer insulator proteins [31] . This model proposing different functional roles for insulator factors could also explain the mechanism by which insulators are able to help establish and reinforce the transcriptional state of chromatin domains throughout cell division . First layer proteins remain bound to chromatin at all stages of the cell cycle [15] , [79] . In contrast , both CP190 and Chromator are chromatin-bound during interphase but display a drastic redistribution during mitosis: CP190 strongly binds to centrosomes while Chromator co-localizes to the spindle matrix [22] , [43] . Thus , the dissociation and cellular redistribution of second layer insulator proteins during cell division would be responsible for the massive remodeling of chromosome architecture occurring during mitosis , and for the re-establishment of higher-order contacts at the onset of interphase . In contrast , first layer insulator proteins would act as anchor points for the re-establishment of higher-order interactions after mitosis , and for the maintenance of the transcriptional identity of physical domains . Thus , our model suggest distinct roles for insulator binding proteins and co-factors in actively shaping the organization of chromatin into physical domains during the cell cycle . This model is consistent with recent genome-wide data suggesting that , overall , first layer insulator proteins remain bound to their binding sites during mitosis , whereas second layer insulator proteins tend to show a large change in binding patterns [79] , [80] . Further genome-wide and microscopy experiments will be needed to quantitatively test this model . DNA plasmids were propagated in E . coli DH5a or in DB3 . 1 cells ( depending on vector used ) . Proteins were expressed and purified from E . coli BL21 ( DE3 ) -pLysS cells ( Invitrogen ) as described elsewhere [81] . Details on vectors , primers , protein constructs and protein purification procedures can be found in Text S1 and in Supplementary Tables S2 , S3 . A 447 bp genomic region containing the Tudor-SN locus was subcloned into pTST101 to make pTST101-447pos ( oligonucleotides are shown in Supplementary Table S4 ) . pTST101-447pos was digested by NdeI , HindIII , and SalI resulting in three linear fragments , including DNAtudor ( 1627 bp long dsDNA fragment containing the 447 bp Tudor-SN locus ) and two additional dsDNA fragments ( 750 and 4025 bp ) . Restricted pTST101 ( 1 . 7 nM ) was incubated with increasing amounts of purified BEAF32 , CP190 or Chromator in 150 mM NaCl , 30 mM Tris/HCl pH 7 . 4 , 5 mM mercaptoethanol . A gel loading buffer ( 50% glycerol , 50 mM Tris/HCl pH 7 . 4 ) was added and the DNA-protein mixture was directly analyzed in a 1% TAE agarose gel . DNA was labeled using Sybersafe ( Invitrogen ) and visualized on a gel imaging system ( Image Station 4000 MM Pro–Carestream Molecular Imaging ) . No difference in binding specificity was observed when DNA competitors ( e . g . dIdC ) were added to the protein-DNA mix . For super-shift assays , the 447 bp Tudor-SN locus ( chromosome 3L: 264375–264822 ) was PCR amplified from S2 Drosophila genomic DNA . Purified proteins were added to the DNA in a reaction mixture in a total volume of 20 µl and incubated for 10 min on ice . A gel loading solution ( 50% glycerol , 50 mM Tris/HCl pH 7 . 4 ) was added and the DNA-protein mixture was directly analyzed on a 2% TAE agarose gel . Fluorescence anisotropy experiments used short , 5′-Cy3B labeled DNA fragments ( DNAS and DNANS , Eurogentec , oligonucleotide sequences are shown in Supplementary Table S5 ) . Anisotropy measurements were carried out using a Tecan Safire II micro plate reader fluorimeter and a Corning 384 Low Flange Black Flat Bottom plate . All measurements were carried out in 30 mM Tris/HCl pH 7 . 5 , 0 . 01 mg/ml BSA , 0 , 004% Tween20 , 100 mM NaCl , 20 µM ZnSO4 , 5 mM mercaptoethanol in a final volume of 60 µl . DNA binding studies were performed by adding increasing amounts ( 0–800 nM ) of purified proteins to 2 . 5 nM of Cy3B or atto-655 5′-labeled 58-bp dsDNA . Dissociation measurements were performed by adding large amounts ( up to 1000 nM ) of unlabeled DNAS or NaCl ( 350 mM final ) . Further details can be found in Text S1 . Reaction buffers and DNA substrates ( at a final DNA concentration of 2 . 5 nM ) were the same as those used for fluorescence anisotropy ( oligonucleotide sequences are shown in Supplementary Table S5 ) . Fluorescence correlation and cross-correlation experiments were carried out on a custom-built setup allowing Pulse Interleaved Excitation ( PIE ) with Time Correlated Single Photon Counting ( TCSPC ) detection as described elsewhere [53] . It is important to note that our measurements allow us to detect only 50% of the complexes involved in bridging , as complexes containing two DNA molecules with the same color do not contribute to the cross-correlation amplitude . Further details on PIE-FCS and the models used to fit data can be found in Text S1 . Drosophila S2 cells ( DGRC ) were grown in Schneider cell medium supplemented with 10% calf serum . 3×106 cells were centrifuged for 10 min at 1000 g and 4°C . All subsequent steps were performed on ice . Cells were washed twice in PBS and resuspended in hypotonic lysis buffer ( 10 mM Tris/HCl pH 7 . 5 , 10 mM KCl , 1 . 5 mM MgCl2 , complete EDTA-free protease inhibitors ( Roche ) ) , and washed again twice with hypotonic buffer . After 30 min on ice , lysed cells were pushed through a 25G needle . In addition , lysates were washed with hypotonic buffer and centrifuged at 1000 g . Nuclei were resuspended in nuclear lysis buffer ( 300 mM KCl , 50 mM Tris/HCl Ph 7 . 5 , 10% glycerol , 1% Triton ×100 , and protease inhibitors ) with benzonase ( Novagen , 71206 ) and incubated for 30 min on a rotating wheel at 4°C . Next , nuclear lysates were centrifuged at 14000 g for 15 min at 4°C . The supernatant was transferred to a clean tube . This resulted in 200 µl of nuclear extract with a total protein concentration of ∼20 mg/ml . This protocol was adapted from Hart et al . [82] . Purified proteins/S2 nuclear extracts were separated on a 10–12% SDS-Polyacrylamide-gel and electro-blotted for 1 h at 100 mV onto a nitrocellulose membrane ( Protran* Nitrocellulose Membrane Filters , Whatman* ) . Next , membranes were blocked ( 3% BSA in TBST ) for 1 h and subsequently washed ( 1% BSA in TBST ) before incubation for 1 h with polyclonal purified primary antibody ( guinea-pig-anti-Chromator/rabbit-anti-CP190 or mouse-anti-BEAF32 from DSHB ) . Several washing steps ( 1% BSA in TBST ) followed before the incubation with HRP-labeled secondary antibody ( goat anti-guinea pig IgG-HRP Conjugate Thermo scientific , Goat anti-Mouse IgG ( H+L ) -HRP conjugate Pierce , goat anti-rabbit IgG ( H+L ) -HRP Conjugate Biorad ) for 40 min . After further washing steps the membrane was developed using Pierce ECL Western Blotting Substrate and imaged ( Image Station 4000 MM Pro – Carestream Molecular Imaging ) . Purified polyclonal antibodies ( anti-Chromator ( 60 µg ) , anti-CP190 ( 60 µg ) , control goat-IgG ( 90 µg ) were immobilized ( 2 h , room temperature ) on 100 µl Amino Link Plus Coupling agarose-bead-slurry ( Pierce Co-Immunoprecipitation Co-IP Kit ) following the manufacturer instructions . Different concentrations of heterologous purified proteins or 100 µl of S2 nuclear extract ( 20 mg/ml ) including protease inhibitor ( Roche , EDTA free ) were added for control goat-IgG , guinea-pig-anti-Chromator , or rabbit-anti-CP190 immobilized agarose beads in IP-Lysis buffer ( part of the Cp-IP Pierce kit , total volume 400 µl ) and incubated on a rotary wheel for 1–3 h at 4°C in a final volume of 400 µl . Depending on the bait protein used , the bead-antibody-protein-complex was washed several times with 400 µl IP lysis-buffer , followed by PBS including 200–1000 mM NaCl until no protein could be detected in the washing step . Elution was carried out after incubating the protein-bead complex for 3 min in elution buffer at pH 2 . 8 . Eluted proteins were analyzed by Western-blot-analysis . Aggregation plots were obtained from genome-wide data from Sexton et al . [13] , and were constructed by following the strategy developed by Jee et al . [83] . First , interacting Hi-C DpnII bins containing genomic features of interest ( BEAF32 , CP190 or Chromator ) were identified . BEAF32 binding sites were considered as anchors and CP190 , Chromator or both sites as targets [83] . Second , only LRI at distances between 15 and 60 kbp and containing BEAF32 in the anchor and CP190/Chromator in the target were further considered . The lower limit was set to 15 kbp , as significantly high background levels occur for bins at distances <15 kbp . The upper limit ( 60 kbp ) was set to be smaller than the average size of topological domains [13] . Third , Hi-C interaction profiles were binned in 500 bp windows +/−5 kbp around the target site . Next , target sites were aligned , aggregated together , and normalized ( blue solid lines , Figure 6E ) . Internal controls ( grey lines , Figure 6E ) were obtained by using the same procedure but for target sites that did not contain any of the features ( CP190 or Chromator ) . This procedure generated background interaction levels reflecting the chromatin context of the anchor site . Frequencies of interactions were statistically tested by Wilcoxon tests . For the analysis of ChIP-chip data ( Venn diagrams ) , publicly available . gff3 files were downloaded from the modENCODE website ( http://data . modencode . org/ ) corresponding to CP190 , BEAF32 and Chromator/Chriz ChIP-chip experiments performed in BG3 and S2 cells [48] , [84] ( datasets 274 , 275 , 278 , 279 , 280 , 921 , 924 ) . Overlaps between binding sites were calculated with the intersectBed function of the BEDTools software [85] . Venn diagrams were generated with the vennDiagram package in R . Crystallization trials was carried out by the sitting-drop technique using the classic , PEG , PACT and AmSO4 suites ( Quiagen , France ) and low-profile microplates ( Grenier , France ) at room temperature . 0 . 5 µl protein solution was mixed with an equal volume of reservoir solution . Several conditions yielded crystals . Optimizations were done with the hanging-drop vapor diffusion technique . 1 µl protein solution was mixed with 1 µl of reservoir . We obtained well diffracting crystals ( 2 . 03 Å ) using 0 . 8 M NaH2PO4 , 0 . 8 M KH2PO4 , 0 . 1M Hepes/pH 7 . 5 . Crystals were soaked in 30% glycerol for cryoprotection and diffraction data were collected under cryogenic conditions on our laboratory anode and at the European Synchrotron Radiation Facility ( ESRF , Grenoble ) . Image data were processed and scaled using the programs MOSFLM ( Leslie , 1999 ) and SCALA of the CCP4 suite [86] . The crystal belonged to space group P3221 with unit cell parameters a = b = 84 . 98 Å , c = 40 . 87 Å , α = β = 90° and γ = 120° . The structure of CP190-BTB/POZ was solved by molecular replacement with an in-house dataset at 2 . 3 Å resolution using the program PHENIX ( phenix . autoMR ) [87] and a combination of five partial models extracted from the server TOME [88] used to gather potential templates through fold-recognition . Structure refinement and rebuilding were performed with COOT [89] , PHENIX ( phenix . refine ) [87] and REFMAC ( Murshudov et al , 1997 ) from the CCP4 suite [86] using a dataset recorded at the ESRF at 2 . 0 Å resolution . Data collection and refinement statistics are summarized in Supplementary Table S1 . The structure has been deposited with the Protein Data Bank ( PDB 4U77 ) .	Chromatin insulators mediate specific long-range DNA interactions required for the three dimensional organization of the interphase nucleus and for transcription regulation , but the mechanisms underlying the formation of these interactions is currently unknown . In this manuscript , we investigate the molecular associations between different protein components of insulators ( BEAF32 , CP190 and Chromator ) by biochemical and biophysical means , and develop a novel biophysical assay to determine what factors are necessary and essential for the formation of long-range DNA interactions ( LRI ) . Importantly , we show that CP190 and Chromator are able to mediate LRIs between specifically-bound BEAF32 nucleoprotein complexes . This ability of CP190 and Chromator to establish LRI requires specific contacts between BEAF32 and their C-terminal domains , and dimerization through their N-terminal domains . In particular , the BTB/POZ domains of CP190 form a strict homodimer . We propose a general model for insulator function in which BEAF32/dCTCF/Su ( HW ) provide DNA specificity , whereas CP190/Chromator are responsible for the physical interactions required for long-range contacts . This network of organized , multi-layer interactions could explain the different activities of insulators , and suggest a general mechanism for how insulators may shape the organization of higher-order chromatin during cell division .	[ "Abstract", "Introduction", "Results", "Discussion", "Materials", "and", "Methods" ]	[ "genome-wide", "association", "studies", "biochemistry", "genomics", "cell", "biology", "nucleic", "acids", "chromosome", "biology", "genome", "analysis", "gene", "expression", "genetics", "biology", "and", "life", "sciences", "epigenetics", "computational", "biology", ...	2014	Chromatin Insulator Factors Involved in Long-Range DNA Interactions and Their Role in the Folding of the Drosophila Genome
Demographic models built from genetic data play important roles in illuminating prehistorical events and serving as null models in genome scans for selection . We introduce an inference method based on the joint frequency spectrum of genetic variants within and between populations . For candidate models we numerically compute the expected spectrum using a diffusion approximation to the one-locus , two-allele Wright-Fisher process , involving up to three simultaneous populations . Our approach is a composite likelihood scheme , since linkage between neutral loci alters the variance but not the expectation of the frequency spectrum . We thus use bootstraps incorporating linkage to estimate uncertainties for parameters and significance values for hypothesis tests . Our method can also incorporate selection on single sites , predicting the joint distribution of selected alleles among populations experiencing a bevy of evolutionary forces , including expansions , contractions , migrations , and admixture . We model human expansion out of Africa and the settlement of the New World , using 5 Mb of noncoding DNA resequenced in 68 individuals from 4 populations ( YRI , CHB , CEU , and MXL ) by the Environmental Genome Project . We infer divergence between West African and Eurasian populations 140 thousand years ago ( 95% confidence interval: 40–270 kya ) . This is earlier than other genetic studies , in part because we incorporate migration . We estimate the European ( CEU ) and East Asian ( CHB ) divergence time to be 23 kya ( 95% c . i . : 17–43 kya ) , long after archeological evidence places modern humans in Europe . Finally , we estimate divergence between East Asians ( CHB ) and Mexican-Americans ( MXL ) of 22 kya ( 95% c . i . : 16 . 3–26 . 9 kya ) , and our analysis yields no evidence for subsequent migration . Furthermore , combining our demographic model with a previously estimated distribution of selective effects among newly arising amino acid mutations accurately predicts the frequency spectrum of nonsynonymous variants across three continental populations ( YRI , CHB , CEU ) . Demographic models inferred from genetic data play several important roles in population genetics . First , they complement archeological evidence in understanding prehistorical events ( such as the number and timing of major continental migrations ) which have left no written record [1] , [2] . Second , they facilitate the search for genetic regions that have been targets of non-neutral forces , such as recent natural selection , by guiding our expectations as to how much sequence and haplotype variation one expects to see in a given genomic region ( and , more importantly , the variance around these expectations ) [3] . Finally , existing demographic models can guide sampling design for subsequent population or medical genetic studies . Given their many uses , it is not surprising that many studies have inferred demographic models for populations of humans and other species [4]–[15] . The process of inferring a demographic model consistent with a particular data set typically involves exploring a large parameter space by simulating the model many times , often using coalescent-theory based Monte Carlo approaches . For computational reasons , many of the demographic inference procedures developed thus far have focused on single population models or models with multiple populations but no subsequent migration after subpopulations split ( i . e . , [4]–[6] , [16] , [17] , but also see [10] , [18] ) . Methods that do consider multiple populations with migration often assume independent non-recombining regions [7] , [19] and do not often scale to genomic size data sets . Approaches for jointly considering recombination and migration often use a restricted set of summary statistics [9] of the data , which limits their statistical power . Finally , complex demographic inferences that make use of many summary statistics are often very computationally intensive [8] , [10] , [18] , which precludes thorough investigation of their statistical properties . Here , we develop and apply a computationally efficient diffusion-based approach to the problem of demographic inference , based on the multi-population allele frequency spectrum ( AFS ) ( i . e . , the joint distribution of allele frequencies across diallelic variants ) [10] , [17] , [18] , [20] , [21] . Given a genetic region sequenced in multiple individuals from each of populations , the resulting AFS is a P-dimensional matrix . Each entry of this matrix records the number of diallelic genetic polymorphisms in which the derived allele was found in the corresponding number of samples from each population . For example , if diploid individuals from two populations were sequenced , with 10 individuals from population 1 and 5 from population 2 , the AFS would be a 21-by-11 matrix ( indexed from 0 ) . The [2 , 0] entry would record the number of polymorphisms for which the derived allele was seen twice in population 1 but never seen in population 2 , while the [20] , [5] entry would record polymorphisms for which the derived allele was homozygous in all individuals from population 1 and seen 5 times in population 2 . If all polymorphic sites possess only two alleles and can be considered independent , the AFS is a complete summary of the data . Many of the statistics commonly used for population genetic inference , such as and Tajima's , are summaries of the AFS ( see [18] , [22] ) . Efficient techniques exist for simulating the AFS of a single population [4] , [5] , [23] . The joint AFS between two populations has been used by several recent studies [10] , [11] , [18] , [24] , but these have all relied upon very computationally intensive coalescent simulations . Here we approximate the joint multi-population AFS by numerical solution of a diffusion equation , and our implementation supports up to three simultaneous populations . Because the diffusion approach neglects linkage , our comparison with the data is through a composite likelihood function . Such likelihoods are consistent estimators under a wide range of population genetic scenarios for selectively-neutral data , but do not correctly capture variances [25] . ( Lower recombination induces higher linkage and higher variance in the entries of the AFS . ) As we demonstrate below , the efficiency of our diffusion approach enables both conventional and parametric bootstrap resampling of the data , allowing us to accurately estimate confidence intervals for parameter values and critical values for hypothesis tests [26] , accounting for any degree of linkage found in the data . This bootstrap procedure overcomes the traditional concerns with composite likelihood as a philosophy for inference in population genetics . To demonstrate the utility of our approach , we apply our method to two epochs in human history , using single nucleotide polymorphism ( SNP ) data from the Environmental Genome Project ( EGP ) [27] , the largest public database of human resequencing data . We first study the expansion of humans out of Africa , jointly modeling the history of African , European , and East Asian populations . We then study the settlement of the New World , jointly modeling European , East Asian , and admixed Mexican populations . In both cases , we quantify the uncertainty of our parameter inferences and test hypotheses about migration ( bootstrapping to account for linkage ) . In particular , we infer an earlier divergence between African and Eurasian populations than previous studies , because our inferences account for the substantial migration between these populations . Our methods also find no evidence for multiple migrations between East Asia and the New World . While similarly complex models for human continental populations have been studied [8] , to our knowledge , our analysis is the first in which the full joint AFS is used for inference and in which uncertainty and goodness-of-fit have been quantified . An important advantage of the diffusion approach is the ease with which selection can be incorporated . As an illustrative application , we also predict the distribution of protein-coding variation between populations . In agreement with the data , we find that less nonsynonymous variation is shared between populations than might be expected based only on patterns of shared noncoding variation . While no model can capture the full complexity of any species' genetic history , the models presented refine our understanding of the expansion of humanity across the globe . None of the methodology is specific to humans , and we expect our method will find wide application to demographic inference of other species . To efficiently simulate the AFS , we adopt a diffusion approach . Such approaches have a long and distinguished history in population genetics , dating back to R . A . Fischer [28]–[30] . The diffusion approach is a continuous approximation to the population genetics of a discrete number of individuals evolving in discrete generations . An important underlying assumption is that per-generation changes in allele frequency are small . Consequently , the diffusion approximation applies when the effective population size is large and migration rates and selection coefficients are of order . If we have samples from populations , the numbers of sampled sequences from each population are . ( For diploids , is typically twice the number of individuals sampled from population 1 . ) Entry of the AFS records the number of diallelic polymorphic sites at which the derived allele was found in samples from population 1 , from population 2 , and so forth . ( If ancestral alleles cannot be determined , then the “folded” AFS can be considered , in which entries correspond to the frequency of the minor allele . ) We model the evolution of , the density of derived mutations at relative frequencies in populations at time . ( All run from 0 to 1 . ) Given an infinitely-many-sites mutational model [31] and Wright-Fisher reproduction in each generation , the dynamics of for an arbitrary finite number of populations are governed by a linear diffusion equation: ( 1 ) The first term models genetic drift , and the second term models selection and migration . Figure 1A illustrates the effects of different evolutionary forces on components of . Time is in units of , where is the time in generations and is a reference effective population size . The relative effective size of population is . The scaled migration rate is , where is the proportion of chromosomes per generation in population that are new migrants from population . ( Thus migration is assumed to be conservative [32] ) . Finally , the scaled selection coefficient is , where is the relative selective advantage or disadvantage of variants in population . Boundary conditions are no-flux except at two corners of the domain , where all population frequencies are 0 or 1; these are absorbing points corresponding to allele loss or fixation . Because the diffusion equation is linear , we can solve simultaneously for the evolution of all polymorphism by continually injecting density at low frequency in each population ( at a rate proportional to the total mutation flux ) , corresponding to novel mutations . Changes in population size and migration alter the parameters in Equation 1 , while population splits and mergers alter the dimensionality of . For example , if new population 3 is admixed with a proportion from population 1 and from population 2 then ( 2 ) where denotes the Dirac delta function . To remove population 2 , is integrated over : . Given , the expected value of each entry of the AFS , , is found via a P-dimensional integral over all possible population allele frequencies of the probability of sampling derived alleles times the density of sites with those population allele frequencies . For SNP data obtained by resequencing , these probabilities are binomial , so ( 3 ) In some cases of ascertained data [33] , the resulting bias can be corrected by modifying the above equation [11] , [34] . Let correspond to the parameters of a demographic model we wish to estimate from the observed multi-population allele frequency spectrum , which we denote . Assuming no linkage between polymorphisms , each entry in the AFS is an independent Poisson variable [20] , with mean ( which depends on ) . We can , therefore , construct a likelihood function using standard statistical theory: ( 4 ) So is the product of Poisson likelihoods , one for each entry in the AFS . In words , our approach consists of calculating the expected allele frequency spectrum using a particular demographic model ( and set of parameter values for that demographic model ) using our diffusion approach . We then maximize the similarity between and the observed AFS over the parameter values that can take on . Competing demographic models can be chosen from using standard statistical theory such as the nested likelihood ratio test or information criteria such as the Akaike or Bayesian Information Criteria . For linked polymorphisms , is a composite likelihood . Such likelihoods are consistent estimators under a wide range of neutral population genetic scenarios [25] , but simulations incorporating linkage are necessary to estimate variances and define critical values for hypothesis testing and model selection . In our applications , we estimate variances using simulations from the coalescent simulator ms [35] . Solving the multi-population diffusion equation is substantially more demanding than the single-population case [23] . This is primarily because the boundary conditions are more complex , and the numerical grid of population frequencies must be much coarser to be computationally tractable , because it is of dimensions . For example , a previous single-population study [23] used a uniform grid of order values between 0 and 1 . Extending this grid to a three-population simulation would require an infeasible array of size . Instead , we use a nonuniform grid and extrapolation to enable accurate computation using of order 100 values along each dimension , for a final array size of order . We solve the diffusion equation on a regular nonuniform grid , using a finite difference scheme [36] inspired by the method of Chang and Cooper [37] ( Text S1 ) . Mutations in population arise at frequency . The diffusion approximation applies when , but the minimum frequency in our numerical simulation is that of the first grid point , denoted . To overcome this , we extrapolate our results to an infinitely fine grid . We use a quadratic extrapolation on the logarithm of the AFS entry , modeling the bias introduced by the finite initial grid point as ( 5 ) Here is an AFS element calculated at grid size and is the extrapolated value . Given three evaluations at different grid sizes , we solve for and use this value when calculating likelihoods . This vastly increases both the speed and accuracy of our calculation ( Supplementary Figure 3 in Text S1 ) . While higher-order extrapolations may improve accuracy in some cases , they may also be more sensitive to numerical noise . Our empirical experience is that a quadratic approximation provides a good compromise between accuracy , efficiency , and robustness . The computational cost for a single likelihood evaluation scales as where is the number of grid points used . In our experience , for stability and accuracy should somewhat larger than the largest population sample size . Although our theoretical framework extends to an arbitrary number of populations , the exponential scaling of computation with limits our current applications to three simultaneous populations . Importantly , our likelihood calculation is deterministic and numerically smooth , so numerical derivatives can be used in optimization . We use the the quasi-Newton BFGS method [36] , which converges in order steps , where is the number of free parameters . Our implementation of these methods , , is written in cross-platform Python and C , making use of the NumPy [38] , Scipy [39] , and Matplotlib libraries [40] . It is distributed under the open-source BSD license . All calculations herein were performed with version 1 . 1 . 0 . We estimated parameter uncertainties by both conventional bootstrap ( fitting data sets resampled over loci ) and parametric bootstrap ( fitting simulated data sets ) . To generate simulated data we used the coalescent program ms [35] , a region-specific recombination rate , and the detailed EGP sequencing strategy ( Text S1 ) . The confidence intervals reported in Table 1 and Table 2 derive from a normal approximation to the bootstrap results . For the conventional bootstrap , confidence intervals were calculated as . For the parametric bootstrap , biased-corrected intervals were calculated as . The maximum-likelihood value is denoted , while and denote the mean and standard deviation of the bootstrap results . Aside from the growth rates , all our model parameters are positive by definition , so in those cases we used their logarithms when calculating confidence intervals . Pearson's goodness-of-fit test was performed using all 213−2 = 9259 bins in the AFS . Results are similar if we restrict our analysis to entries in which the expected value is greater than 1 or greater than 5 . We used the National Institute of Environmental Health Science's Environmental Genome Project SNPs database [41] , which results from direct Sanger resequencing of environmental response genes in several populations . We considered all diallelic SNPs in 5 . 01 Mb of sequence from noncoding regions of 219 autosomal genes ( Supplementary Table 8 in Text S1 ) . These data have been the subject of many publications , including [17] , [23] , [27] , [42] . As an assessment of quality , additional high-coverage short-read sequencing has recently been performed across 8 samples in this data set . Over 26 , 000 sites , the SNP concordance between this next-generation sequencing and the original Sanger sequencing averages 99 . 5% ( D . Nickerson , personal communication ) . Given the high quality of this data set , we do not incorporate sequencing error into our modeling . We believe such correction will be essential in future applications to less accurate short-read sequencing data , as inference based on the frequency spectrum is sensitive to rare alleles . To estimate the ancestral allele , we aligned to the panTro2 build of the chimp genome [43] . Like other methods based on the unfolded AFS , our analysis is sensitive to errors in identifying the ancestral allele . We statistically corrected the AFS for ancestral misidentification [17] , using a context-dependent substitution model [44] . This procedure has been shown to perform better than aligning to multiple species [17] . To account for missing data and ease qualitative comparisons between populations , we projected all spectra down to 20 samples per population [5] ( Text S1 ) . The human-chimp divergence in the data is 1 . 13% . We assumed a divergence time of 6 My [45] and a generation time of 25 years . This yielded an estimated neutral mutation rate of per site per generation , which is comparable to direct estimates [46] . There is some controversy as to the appropriate generation time to assume in human population genetic studies [47] , [48] . In particular , the human generation time may differ between cultures and may have changed during our biological and cultural evolution . The bootstrap uncertainties reported in Table 1 and Table 2 do not include systematic uncertainties in the human-chimp divergence or generation times . The generation time , however , formally cancels when converting between genetic and chronological times . In our prediction of the distribution of nonsynonymous polymorphism , the distribution of selective effects assumed was a negative-gamma distribution with shape parameter and scale [49] . The AFS was calculated by trapezoid-rule integration over this distribution , using 201 evaluations logarithmically spaced over . All demographic parameters , including the scaled mutation rate , were set to the maximum-likelihood values from our Out of Africa analysis . In Figure 1 , we provide examples of the AFS under different demographic scenarios . Figure 1B illustrates the isolation-with-migration model for which the spectra are calculated . The expected spectrum at zero divergence time is shown in Figure 1C . Figure 1D shows the expected spectrum at various divergence times under various demographic scenarios . Qualitatively , correlation between population allele frequencies declines with increasing divergence time , depopulating the diagonal of the AFS . On the other hand , migration prolongs and sustains correlation . Less obviously , AFS entries corresponding to shared low-frequency alleles distinguish between increased migration and reduced divergence time ( Supplementary Figure 1 in Text S1 ) . Additionally , differences in genetic drift between populations with different effective sizes result in asymmetries in the AFS . These qualitative features of the AFS are also evident in human data . Detailed modeling allows us to quantify our inference regarding the type , timing , and strength of demographic events that are consistent with the data . The computer program implementing our method is named ( Diffusion Approximations for Demographic Inference ) . It is open-source and freely available at http://dadi . googlecode . com . Figure 1E compares with a coalescent approach to evaluating the likelihood of frequency spectrum data . The coalescent simulator ms [35] was used to generate a simulated data set from the model in Figure 1B , with parameters , , , , , scaled total recombination rate , and 20 samples per population . Coalescent-based estimates of the expected AFS were generated by averaging ms simulations , each run with and . These estimates were scaled to for comparison with the simulated data set . ( This procedure is substantially faster than simulating with larger and . ) Each estimate took approximately 7 . 2 seconds of computation . The histogram in Figure 1E shows the resulting distribution of estimated likelihoods of the data . Shown by the red line in Figure 1E is the result from our diffusion approach ( with grid sizes ) , which took approximately 2 . 0 seconds of computation . The yellow line is the likelihood from coalescent simulations , illustrating the high accuracy of our diffusion approach . ( Note that the coalescent approach we consider here is not necessarily optimal . We are , however , unaware of any such approach that is competitive in computational speed with the diffusion method . ) The computational advantage of the diffusion method is even larger when placed in the context of parameter optimization . Unlike the coalescent approach , there is no simulation variance , so efficient derivative-based optimization methods can be used . As examples , consider our applications to human data , which involve 20 samples per population . On a modern workstation , fitting a single-population three-parameter model took roughly a minute , while fitting a two-population six-parameter model took roughly 10 minutes . The fits of three-population models with roughly a dozen parameters typically took a few hours to converge from a reasonable initial parameter set . This speed allows us to use extensive bootstrapping to estimate variances , overcoming the limitations of composite likelihood . Our analysis of human expansion out of Africa used data from three HapMap populations: 12 Yoruba individuals from Ibadan , Nigeria ( YRI ) ; 22 CEPH Utah residents with ancestry from northern and western Europe ( CEU ) ; and 12 Han Chinese individuals sampled in Beijing , China ( CHB ) . Because approaches based on the frequency spectrum are sensitive to miscalling of the ancestral state , we statistically corrected for ancestral misidentification using an approach that accounts for a myriad of mutation and context-dependent biases ( such as CpG effects ) [17] . To ease qualitative comparison among populations and account for missing data , we projected the data down to 20 sampled chromosomes per population [5] . Because this data set is of very high quality ( >99% concordance of sequenced SNPs with next-generation sequencing of the same individuals to high coverage; see Methods ) , we do not explicitly correct for sequencing errors here . We were left with 17 , 446 segregating diallelic SNPs from effectively 4 . 04 Mb of sequence . Figure 2A shows the resulting AFS . For ease of visualization , the top row of Figure 2C shows the two-population marginal spectra . There are many possible three-population demographic models one could consider for these populations . To develop a parsimonious yet realistic model , we first considered the marginal AFS for each population and each pair of populations . Previous analyses found that the YRI spectrum is well-fit by a two-epoch model with ancient population growth [5] , [17] , and we found this as well ( Supplementary Figure 6 in Text S1 ) . Previous analyses of the CEU and CHB populations found that both populations went through bottlenecks [5] , [11] concurrent with divergence [11] . Such models qualitatively fit our marginal CEU-CHB spectrum ( Supplementary Figure 7 in Text S1 ) . Combining these demographic features yields the model illustrated in Figure 2B . The maximum likelihood values for the 14 free parameters are reported in Table 1 . Qualitatively , the resulting model reproduces the observed spectra well , as seen in the second and third rows of Figure 2C . ( The correlation between adjacent residuals is due in part to our projection of the data down from a larger sample size ( Supplementary Figure 8 in Text S1 ) . ) Allowing for asymmetric gene flow yielded very little improvement in fit , as did allowing for growth in the Eurasian ancestral population or allowing the CEU and CHB bottleneck and divergence times to differ ( data not shown ) . Our composite likelihood function assumes that polymorphic sites are independent . Because it thus overestimates the number of effective independent data points , confidence intervals calculated directly from the composite likelihood function will be too small . To control for linkage , we performed both conventional and parametric bootstraps . Because our sequenced genes are typically well separated , they can be treated as independent , and our conventional bootstrap resampled from the 219 sequenced loci . For the parametric bootstrap , simulated data sets that incorporate linkage and the EGP's sequencing strategy were generated with ms [35] . Table 1 reports parameter 95% confidence intervals from both the conventional and bias-corrected parametric bootstraps . The parametric bootstraps yield slightly smaller confidence intervals than the conventional bootstrap , suggesting that some variability in the data has not been accounted for by our simulations . This variability may involve small varied selective forces on the sequenced regions or slight relatedness between sampled individuals . The parametric bootstrap results additionally show that our method possesses very little bias in parameter inference ( Supplementary Figure 9 in Text S1 ) . As seen in Table 1 , the times for growth in the African ancestral population and divergence of the Eurasian ancestral population ( and ) have particularly wide confidence intervals , likely a consequence of the high inferred migration rate between the African and Eurasian ancestral populations . shows high correlation with the ancestral population size , while shows no strong linear correlation with any other single parameter ( Supplementary Figure 11 in Text S1 ) . We found that 92 out of our 100 conventional bootstrap fits yield , supporting the contention that the CHB population suffered a more severe bottleneck than the CEU population [11] ( Supplementary Figure 11 in Text S1 ) . We used several metrics to assess our model's goodness-of-fit , in additional to visual inspection of the residuals seen in Figure 2C . Figure 2D compares the decay of linkage disequilibrium ( LD ) in the data and in the parametric bootstrap simulations . The agreement seen is notable because our demographic inference used no LD information in building and fitting the model . This LD comparison thus serves as independent validation of both our model and bootstrap simulations . We also asked whether the likelihood found in the real data fit is atypical of fits to simulated data . Out of fits to 100 simulated data sets , 2 produced a smaller likelihood ( worse fit ) than the real data fit ( Figure 2E ) , yielding a p-value of ≈0 . 02 . One can craft examples in which a likelihood-based goodness-of-fit test fails to exclude very poor models [50] . Thus we also applied Pearson's goodness-of-fit test , a more robust and standard method for data that is in Poisson-distributed bins , such as the AFS [36] . In our case , we must use our parametric bootstraps to assess the significance of the sum-of-squared-residuals test statistic , because many entries in the AFS are small and because they are not strictly independent . Figure 2E shows the bootstrap-derived empirical distribution of . Two of the bootstraps yielded a larger ( worse fit ) than the real data fit , giving a p-value of ≈0 . 02 , identical to that from the likelihood-based test . ( The two simulations that yield a higher than the real fit are not the same two that yield a lower , suggesting that these tests are somewhat independent . ) In some cases specific frequency classes of SNPs , such as rare alleles , may be of particular interest . In Supplementary Table 5 in Text S1 , we provide comparisons of the joint distribution of rare alleles seen in the data with that from our simulations . These comparisons indicate that our model also reproduces well this interesting region of the frequency spectrum . Finally , in Figure 4 we compare the model and data using larger bins of SNPs specific to particular populations or segregating at high or low frequency . In all cases the model agrees within the uncertainty of the bootstrapped data . Taken together , these tests suggest that our model provides a reasonable , though not complete , explanation of the data , lending credence to our demographic estimates . The inferred contemporary migration parameters ( , and ) are small , raising the question as to whether they are statistically distinguishable from zero . Figure 2F shows that the improvement in fit to the real data upon adding contemporary migration to the model is much larger than would be expected if there were no such migration , implying that the contemporary migration we infer is highly statistically significant . Omitting ancient migration ( ) reduced fit quality even more , indicating that the data also demand substantial ancient migration ( data not shown ) . To study the settlement of the Americas , we used the previously considered 22 CEU and 12 CHB individuals , plus an additional 22 individuals of Mexican descent sampled in Los Angeles ( MXL ) . Data were processed as in our Out of Africa analysis , yielding 13 , 290 segregating SNPs from effectively 4 . 22 Mb of sequence . Figure 3A shows the resulting AFS , while Figure 3C shows the marginal spectra . A model in which the CEU and CHB diverge from an equilibrium population did not reproduce the AFS well ( Supplementary Figure 13 in Text S1 ) . Interestingly , a model allowing a prior size change in the ancestral population better fit the AFS but very poorly fit the observed LD decay ( Supplementary Figure 13 in Text S1 ) . Thus , reproducing the AFS does not guarantee reproduction of LD , at least given a historically unrealistic model . To develop a more realistic model , we endeavored to include the effects of Eurasian divergence from and migration with the African population . Computational limits precluded us from considering all 4 populations simultaneously , so we dropped the African population from the simulation upon MXL divergence ( Figure 3B ) . Table 2 records the maximum-likelihood parameter values inferred for this model . Because this fit did not include African data , we could not reliably infer demographic parameters involving the African population . Thus , for this point estimate we fixed the Africa-related parameters , , , , , , and to their maximum-likelihood values from Table 1 . Figure 3C compares the model and data spectra . The residuals show little correlation , with the possible exception that the model may underestimate the number of high-frequency segregating alleles . Parameter confidence intervals are reported in Table 2 . To account for our uncertainty in those parameters derived from the Out of Africa fit , for each conventional bootstrap fit we used a set of Africa-related parameters randomly chosen from the sets yielded by our Out of Africa conventional bootstrap . For the parametric bootstrap , we used the maximum-likelihood point estimates . Again , we see that the conventional bootstrap confidence intervals are comparable to , although slightly wider than , the parametric bootstrap intervals . Several parameters in this analysis have direct correspondence with our Out of Africa analysis . Of particular note , the confidence intervals for the CEU-CHB divergence time overlap . In assessing goodness of fit , Figure 3D shows that this model does indeed reproduce the observed pattern of LD decay . Unlike in our Out of Africa analysis , however , here the LD decay was used to choose the form of the model ( although not its parameter values ) , so this is not a completely independent assessment of fit . Of our 100 parametric bootstrap fits , 13 yielded a worse likelihood than the real fit ( Figure 3E ) , for a p-value of ≈0 . 13 . Applying Pearson's test , we find that 23 of 100 bootstrap fits yield a higher ( worse ) than the fit to the real data , for a p-value of ≈0 . 23 , similar to that of the likelihood analysis . Comparing distributions of rare alleles , our model typically reproduces the observed distribution well , although it may be somewhat overestimating the proportion of alleles that are rare or absent in the CHB population ( Supplementary Table 7 in Text S1 ) . In sum , our model appears to be a reasonable explanation of this data , somewhat better than in our Out of Africa analysis . An essential feature of the Mexican-American individuals considered here is that they are typically admixed from Native American and European ancestors . The ≈50% average European admixture proportion we inferred for the MXL population is consistent with previous estimates for Los Angeles Latinos [51] . We have no direct data from the Native American populations ancestral to MXL , but our model does account for their divergence from East Asia . A model neglecting this divergence ( by setting to zero ) fit the data substantially worse and yields an unrealistically high average European admixture proportion into MXL of 68% . Not only are Mexican-American individuals admixed , their admixture proportions also vary , and this subtlety is not directly accounted for in our analysis . To assess its effect on our results , we first roughly estimated the ancestry proportion of each individual , using essentially a maximum-likelihood version [18] of the algorithm used in structure [52] ( Text S1 ) . ( Methods based on “admixture LD” , which identify breakpoints between regions of Native American and European ancestry , may be more powerful [53] . However , the strategy used by the EGP of sequencing widely spaced genes will resolve few of these breakpoints , limiting the applicability of these methods . ) We then performed additional parametric bootstrap analyses , using simulations with a distribution of individual ancestry chosen to mimic that seen in the data and , to further test the method , with an extremely wide distribution . These simulations showed that variation in individual ancestry does not bias our parameter inferences ( Supplementary Figure 19 in Text S1 ) . Remarkably , it does not even change our statistical power . This is evidenced by the fact that these bootstrap simulations yielded confidence intervals identical to our original simulations without variation in ancestry proportion ( Supplementary Figure 19 in Text S1 ) . Nevertheless , future studies may profit by incorporating individual ancestry information [18] , perhaps inferred from admixture LD . Finally , our model allowed us to assess the role recurrent migration from Asia played in the settlement of the New World [2] . When we added CHB-MXL migration to our model , we found that the maximum likelihood migration rate was per generation . As shown in Figure 3F , the resulting improvement in likelihood is typical ( p-value≈0 . 45 ) of fits including CHB-MXL migration to data simulated without it . Our data and analysis thus yielded no evidence of recurrent migration in the settlement of the New World . Note , however , that this simple test does not necessarily rule out more complex scenarios , in which migration may vary over time . Polymorphisms that change protein amino acid sequence are of medical interest because they are particularly likely to affect gene function [54] . Correspondingly , they are often subject to natural selection . Diffusion approaches are particularly useful for studying such nonsynonymous polymorphism , because they easily incorporate selection . Although the diffusion approximation assumes that sites are unlinked , nonsynonymous segregating sites are rare enough that this is often a reasonable approximation [49] . As an illustration , we used our Out of Africa demographic model to predict the distribution of such variation between continental populations . To do so , we must specify a distribution for the selective effects of nonsynonymous mutations that enter the population . For this we adopted a negative gamma distribution whose parameters were recently inferred [49] . The resulting distribution of segregating variation is shown in Figure 4A . ( To ease comparison , we have assumed the same scaled mutation rate as in the neutral case of Figure 2C . ) As expected , selection sharply reduces the amount of segregating polymorphism . Figure 4B shows the proportion of variants within various classes . Also as expected , selection shifts nonsynonymous variation toward lower frequencies , raising the proportion of singletons and lowering the proportion at frequency greater than 10% . Less obviously , it also reduces the proportion of variation that is shared between populations . In the neutral case , 43% of polymorphism is predicted to be present in more than one population , while in the selected case only 35% is . Thus genetic inferences from coding polymorphism may be less transferable between populations than might be expected from neutral patterns of allele sharing . In the data considered here , there are about 400 nonsynomymous polymorphisms segregating in the three populations considered . This is too few for a detailed goodness-of-fit test of our predicted distribution . ( Although see Supplementary Figure 20 in Text S1 for a direct AFS comparison . ) Nevertheless , we observe that our predictions shown in Figure 4B all lie within the bootstrap 95% confidence intervals from the data . Our diffusion approximation to the joint allele frequency spectrum is a powerful tool for population genetic inference . Although the diffusion approximation neglects linkage between sites , our method's computational efficiency allows us to use extensive bootstrap simulations to account for the effects of linkage . ( Let us reiterate that linkage does not affect the expected allele frequency spectrum of neutral sites , so our diffusion-based approach is estimating the same AFS that coalescent simulations are estimating , but in a small fraction of the time ) . We applied our method to human expansion out of Africa and settlement of the New World , using public resequencing data from the Environment Genome Project . The flexibility of the diffusion approach also allowed us to consider the distribution of non-neutral variation , which is difficult to address with other approaches . Although no model can capture in detail the complete history of any population , the models presented here help refine our understanding of human expansion across the globe . Our demographic results are in most respects broadly consistent with previous analyses of human populations . In particular , single-population analyses have also inferred African population growth and European and Asian bottlenecks [4]–[6] . Also , the migration rates we infer are similar to those inferred by Schaffner et al . [8] but somewhat smaller than those of Cox et al . [15] . On the other hand , Keinan et al . [11] inferred no significant migration between CEU and CHB . Finally , our estimate of a New World founding effective population size in the hundreds is compatible other inferences [14] . Perhaps our most interesting demographic results are the inferred divergence times . Other studies [11] , [12] have estimated divergence times between Europeans and East Asians similar to the ≈23 kya we infer . Interestingly , archeological evidence places humans in Europe much earlier ( ≈40 kya ) [1] . Our inferred divergence time of ≈22 kya between East Asians and Mexican-Americans is somewhat older than the oldest well-accepted New World archeological evidence [2] . The divergence we infer may reflect the settlement of Beringia , rather than the expansion into the New World proper [14] . Finally , the divergence time of ≈140 kya we infer between African and Eurasian populations is consistent with archeological evidence for modern humans in the Middle East ≈100 kya [1] , but it is much older than other inferences of ≈50 kya divergence from mitochondrial DNA [1] . This discrepancy may be explained by our inclusion of migration in the model . Migration preserves correlation between population allele frequencies , so an observed correlation across the genome can be explained by either recent divergence without migration or ancient divergence with migration . In fact , the African-Eurasian migration rate we infer of ≈25×10−5 per generation is comparable to the ≈100×10−5 inferred from census records between modern continental Europe and Britain [55] . One difficulty in interpreting our divergence times is that the sampled populations may not best represent those in which historically important divergences occurred . For example , the Yoruba are a West African population , so the divergence time we infer between Yoruba and Eurasian ancestral populations may correspond to divergence within Africa itself . Future studies of more populations [56]–[58] will help alleviate this difficulty . Another difficulty is that the genic loci we study here may not be ideal for demographic inference . Although we consider only noncoding sequence in fitting our historical model , selection on regulatory or linked coding sites may skew the AFS [59] . In fact , the EGP data have been shown to differ in some ways ( e . g . Tajima's ) from intergenic regions [58] . Nevertheless , we use the EGP data because it is currently the largest public resource of noncoding human genetic variation , and we fit a neutral model because disentangling the small expected effects of selection on these sites from demographic effects will require additional data . The rapidly declining cost of sequencing will give future studies access to many more loci that are likely to be less influenced by selection . Importantly , the computational burden of our method is independent of the amount of sequence used to construct the AFS . Additional loci will also increase power to discriminate between models and incorporate more detail . The AFS encodes substantial demographic information . It is has been shown , however , that an isolated population's AFS does not uniquely and unambiguously identify its demographic history [60]; we expect a similar result to hold for multiple interacting populations . Moreover , the AFS does not capture all the information in the data . As illustrated by the alternative New World models we considered , patterns of linkage disequilibrium encode additional information . Future studies may profit from coupling our efficient AFS simulation with methods that address other aspects of the data . We have developed a powerful diffusion-based method for demographic inference from the joint allele frequency spectrum . We applied our method to human expansion out of African and the settlement of the New World , developing models of human history that refine our knowledge and raise intriguing questions . We also applied our method to predict the distribution of nonsynonymous variation across populations , and this prediction is consistent with the available data . Our methods and the models inferred from it offer a foundation for studying the history and evolution of both our own species and others .	The demographic history of our species is reflected in patterns of genetic variation within and among populations . We developed an efficient method for calculating the expected distribution of genetic variation , given a demographic model including such events as population size changes , population splits and joins , and migration . We applied our approach to publicly available human sequencing data , searching for models that best reproduce the observed patterns . Our joint analysis of data from African , European , and Asian populations yielded new dates for when these populations diverged . In particular , we found that African and Eurasian populations diverged around 100 , 000 years ago . This is earlier than other genetic studies suggest , because our model includes the effects of migration , which we found to be important for reproducing observed patterns of variation in the data . We also analyzed data from European , Asian , and Mexican populations to model the peopling of the Americas . Here , we find no evidence for recurrent migration after East Asian and Native American populations diverged . Our methods are not limited to studying humans , and we hope that future sequencing projects will offer more insights into the history of both our own species and others .	[ "Abstract", "Introduction", "Methods", "Results", "Discussion" ]	[ "computational", "biology/genomics", "genetics", "and", "genomics/population", "genetics" ]	2009	Inferring the Joint Demographic History of Multiple Populations from Multidimensional SNP Frequency Data
To characterize the repair pathways of chromosome double-strand breaks ( DSBs ) , one approach involves monitoring the repair of site-specific DSBs generated by rare-cutting endonucleases , such as I-SceI . Using this method , we first describe the roles of Ercc1 , Msh2 , Nbs1 , Xrcc4 , and Brca1 in a set of distinct repair events . Subsequently , we considered that the outcome of such assays could be influenced by the persistent nature of I-SceI-induced DSBs , in that end-joining ( EJ ) products that restore the I-SceI site are prone to repeated cutting . To address this aspect of repair , we modified I-SceI-induced DSBs by co-expressing I-SceI with a non-processive 3′ exonuclease , Trex2 , which we predicted would cause partial degradation of I-SceI 3′ overhangs . We find that Trex2 expression facilitates the formation of I-SceI-resistant EJ products , which reduces the potential for repeated cutting by I-SceI and , hence , limits the persistence of I-SceI-induced DSBs . Using this approach , we find that Trex2 expression causes a significant reduction in the frequency of repair pathways that result in substantial deletion mutations: EJ between distal ends of two tandem DSBs , single-strand annealing , and alternative-NHEJ . In contrast , Trex2 expression does not inhibit homology-directed repair . These results indicate that limiting the persistence of a DSB causes a reduction in the frequency of repair pathways that lead to significant genetic loss . Furthermore , we find that individual genetic factors play distinct roles during repair of non-cohesive DSB ends that are generated via co-expression of I-SceI with Trex2 . Chromosome double-strand breaks ( DSBs ) can be repaired by a number of mechanisms with a variety of mutagenic consequences [1] . In the context of ionizing radiation ( IR ) therapy or chemotherapy that utilizes DSB-inducing agents , such DNA damage in non-tumor cells could result in oncogenic mutations that cause secondary malignancies [2] . Thus , characterizing the factors and pathways that influence DSB repair will be important to develop therapeutic approaches that may limit the risk of secondary tumors , and to understand the etiology of genome rearrangements associated with primary cancer development . DSB repair pathways show a varying propensity for genetic loss . A relatively precise form of repair is homology-directed repair ( HDR ) that uses the identical sister chromatid as a template for Rad51-mediated strand invasion and nascent DNA synthesis [1] . In contrast , end-joining ( EJ ) pathways are variably mutagenic , depending on the extent of end-processing and the fidelity of end-pairing . For instance , EJ via the V ( D ) J recombination nonhomologous end-joining ( NHEJ ) machinery has the potential to be precise , especially when DSB ends can be ligated without significant processing [3] . However , Ku-independent EJ ( Alternative-NHEJ , Alt-NHEJ ) often leads to deletion mutations , which are predominantly associated with short stretches of homology ( microhomology ) at repair junctions [4] , [5] . Similar to Alt-NHEJ is single-strand annealing ( SSA ) , which also causes deletions with homology at repair junctions , but involves extensive regions of homology [6] . In addition , for each of these pathways , loss of correct end-pairing during the repair of multiple simultaneous DSBs can lead to chromosomal rearrangements . For instance , EJ between distal ends of two tandem DSBs ( Distal-EJ ) results in loss of the chromosomal segment between the DSBs . To characterize the genetic factors that influence these pathways , one approach involves analyzing repair of site-specific DSBs in mammalian cells , such as those generated by the rare-cutting endonuclease I-SceI . For instance , using this approach , HDR , SSA , and Alt-NHEJ were shown to be promoted by CtIP and Nbs1 [7]–[10] , which are factors implicated in the formation of ssDNA via end resection [9] , [11] . As well , the strand exchange factors Rad51/Brca2 were found to promote HDR and suppress SSA [12] , [13] , and a number of additional genetic factors have been found to promote HDR [14] . Other studies have addressed the influence of factors involved in NHEJ during V ( D ) J recombination , including Ku and Xrcc4-Ligase IV . For example , Ku/Xrcc4-deficient cells show higher HDR [15] , and Ku-deficient cells show elevated SSA and Alt-NHEJ [5] . In addition , Ku and Xrcc4 have been shown to promote EJ that restores the I-SceI site , measured as EJ between distal ends of two tandem I-SceI-induced DSBs ( S+DEJ ) [16] , [17] . To further address the process of DSB repair pathway choice in mammalian cells , we have developed this two-part study . In the first part , we provide a detailed characterization of the roles of Ercc1 , Msh2 , Nbs1 , Xrcc4 , and Brca1 during individual repair events . From these studies , we provide evidence that individual genetic factors may not be specific for particular pathways of repair , but rather promote a mechanistic step that is common among distinct repair pathways . Regarding particularly distinct findings , we present evidence that Msh2 promotes HDR , whereas Ercc1 is particularly required for repair events that require removal of a nonhomologous segment . Moreover , these experiments provide essential reagents for the development of the second part . In the second part of this study , we have addressed whether the outcome of these repair assays could be affected by the persistent nature of I-SceI-induced DSBs . Namely , since precise EJ restores the I-SceI site , chromosomal I-SceI sites are prone to repeated cutting by the I-SceI endonuclease , which has been referred to as the persistent nature of endonuclease-generated DSBs [18]–[21] . To address this aspect of repair , we expressed a 3′ exonuclease , Trex2 [22] , [23] , to partially degrade the 3′ overhangs generated by I-SceI , and thereby promote EJ products that have lost the I-SceI site . Since these EJ products are resistant to further cutting by I-SceI , we suggest that Trex2 expression can limit the persistence of I-SceI-induced DSBs . Using this approach , we find that Trex2 expression strongly decreases the frequency of Distal-EJ in favor of EJ events that maintain proximal end-pairing . Trex2 expression also causes a significant decrease in Alt-NHEJ and SSA . In contrast , HDR is not inhibited by Trex2 expression . These results indicate that limiting the persistence of DSBs can suppress repair pathways that are prone to genetic loss . As well , using this Trex2 approach , we find that individual genetic factors play distinct roles during repair of non-cohesive DSB ends . To investigate the genetic requirements of individual DSB repair pathways , as well as the effect of the persistence of a DSB on repair , we have developed a series of reporters for discrete repair events . In each case , we generate an I-SceI-induced DSB within a chromosomally integrated inactive GFP cassette , where the structure of each reporter is designed such that repair of the DSB by a specific pathway results in restoration of the GFP+ cassette . For instance , three reporters were designed to measure distinct end-joining ( EJ ) events , as described previously [7] , and summarized below . First , the EJ5-GFP reporter measures end-joining between distal ends of two tandem I-SceI-induced DSBs ( Figure 1A [7] ) . This Distal-EJ product results in loss of a fragment between the two I-SceI sites ( puro gene ) , and thereby restores the juxtaposition of the promoter next to the remainder of the GFP cassette . This repair product was previously referred to as total-NHEJ , but Distal-EJ is a more precise description of these repair events , since proximal-EJ would lead to maintenance of the fragment between the two I-SceI sites , and not lead to a GFP+ cassette . Such Distal-EJ can result in either reconstitution of the I-SceI site ( S+DEJ ) or generation of an I-SceI-resistant site . In previous work with this reporter , Ku70 was shown to be essential for S+DEJ events , but completely dispensable for I-SceI-resistant EJ events [7] . As well , the repair junctions of I-SceI-resistant EJ events were shown to predominantly exhibit microhomology ( 90% ) [7] . The findings that I-SceI-resistant EJ products are elevated in Ku-deficient cells , and show evidence of microhomology , suggest that these events are one measure of Alt-NHEJ . However , Ku70 may play an important role during a subclass of I-SceI-resistant EJ events that involve minimal microhomology [16] . Another reporter , EJ2-GFP , specifically measures such Alt-NHEJ events ( Figure 1B , [7] ) . This reporter involves a single I-SceI-induced DSB within a disrupted GFP coding sequence , where a discrete set of Alt-NHEJ events restores a functional GFP cassette . The predominant GFP+ product utilizes 8 nucleotides ( nt ) of microhomology that flank the DSB , which results in a 35 nt deletion . Other Alt-NHEJ events with different deletion sizes can also restore the GFP+ cassette , though these products are less frequent ( 15% of total products ) . Importantly , the GFP+ repair events measured with EJ2-GFP have been shown to be suppressed by Ku70 [7] , which further indicates that this reporter measures Alt-NHEJ . Finally , the SA-GFP reporter measures SSA between two GFP fragments that share 266 nt of homology and are separated by 2 . 7 kb , where an I-SceI site is present in the downstream GFP fragment ( Figure 1C ) . Notably , while this SSA event involves a significant stretch of homology , such repair is suppressed by the homologous strand-exchange factor RAD51 [8] . This finding that the GFP+ product from SA-GFP is suppressed by RAD51 , combined with the relatively low frequency of HDR associated with crossing-over and/or long gene conversion tracts [24] , [25] , suggests that such rare HDR events do not likely contribute significantly to the formation of the GFP+ product in SA-GFP [8] . Distinct from the above reporters for EJ , the DR-GFP reporter is designed to measure HDR ( Figure 2A ) , where a gene fragment ( iGFP ) serves as a template for RAD51-mediated HDR of an I-SceI-induced DSB in an upstream SceGFP cassette [12] . However , these GFP segments differ by only 11 point mutations ( see Figure S1A [26] ) ; therefore , DR-GFP does not require removal of a large segment during repair . In contrast , HDR of complex lesions , such as a series of inter-strand crosslinks ( ICLs ) , and HDR between divergent sequences , could require removal of a significant chromosomal segment to complete repair [27] . To begin addressing this aspect of HDR , we developed another reporter , DRins-GFP ( Figure 2B ) , which is designed to require removal of a nonhomologous segment during HDR . Specifically , this reporter contains a 464 nt insertion of mouse genomic sequence ( intron segment of the Rb gene ) placed downstream of the I-SceI site in SceGFP ( Ins464SceGFP ) . Removal of this insertion would be critical for resolution of the HDR product , but also may be important to disrupt attempts to strand invade the insertion sequence at the Rb locus . To analyze this reporter , both DR-GFP and DRins-GFP were integrated into the pim1 locus of wild-type ( WT ) mouse ES cells . We used ES cells for this study because of the prevalence of specific mutant cell lines , but also because of the relevance of stem cells in regenerative medicine and the etiology of cancer [28] . Subsequently , we transfected these cell lines with an expression vector for I-SceI , and determined the efficiency of HDR by FACS analysis of GFP+ cells . For completion , we also included cell lines with the reporters in Figure 1 in these experiments , and confirmed the structure of the GFP+ product for DRins-GFP ( Figure 2C ) . Regarding a direct comparison between the HDR reporters , we found that HDR of the DRins-GFP reporter was significantly less efficient than for DR-GFP ( 8-fold , p<0 . 0001 , Figure 2C ) . This result indicates that HDR is impaired by the insertion , which also suggests that HDR repair of the DRins-GFP reporter may have unique mechanistic requirements relative to HDR of DR-GFP . Though , as an alternative interpretation , attempts to strand invade the insertion at the Rb locus could contribute to the low efficiency of HDR of the DRins-GFP reporter . Regarding the possibility of distinct mechanistic requirements between these HDR events , we considered the notion that HDR repair of the DRins-GFP reporter may share a common mechanistic step with SSA , thereby providing a bridge between HDR and SSA . Namely , HDR of DRins-GFP is similar to SSA repair of SA-GFP in that both require removal of an extended segment , whereas HDR of DR-GFP does not . So , we hypothesized that HDR repair of DRins-GFP may share end-processing steps with SSA repair of SA-GFP . Such a processing step could involve extensive 5′ to 3′ resection , and/or cleavage of the insertion as an unpaired 3′ ssDNA tail . In particular , we addressed the hypothesis that 3′ ssDNA tail removal , via Ercc1 , may be a common step between SSA and HDR of the DRins-GFP reporter . Ercc1 forms a complex with Xpf and is involved in endonucleolytic cleavage of 3′ ssDNA [29] , which supports a role for Ercc1 during processing of 3′ ssDNA . Furthermore , Ercc1 has been shown to promote SSA [8] , as well as EJ deletion products during joining of plasmid substrates [30] . To test the above hypothesis , we integrated DRins-GFP into an Ercc1-deficient mouse ES cell line ( Ercc1−/− ) , in which both alleles of Ercc1 were targeted with selection cassettes near the 3′ end of the gene [31] . Then , we transfected this cell line with an expression vector for I-SceI , along with either a complementation vector for Ercc1 , or the associated empty vector ( EV ) . As well , we performed this set of transfections on a set of Ercc1−/− cell lines with each of the other reporters in Figure 1 and Figure 2 , many of which have been described previously [7] . Expression of Ercc1 via the complementing vector was confirmed by immunoblotting ( Figure S1B ) . Subsequently , we quantified the fold-effect of the complementation vector on the efficiency of repair , as compared to parallel transfections with EV . We have found that quantifying such fold-complementation provides the most consistent means for determining the influence of a given genetic factor . Importantly , we have not observed any clear effects on viability or proliferation resulting from complementation in any of the genetic analysis in this study ( unpublished observations ) . In any case , such variations are rare in mouse ES cells , given their high rate of proliferation , lack of a p53-dependent G1/S checkpoint , and short gap phases ( G1/G2 ) [32] . As an alternative , we have included the overall frequency of repair for each of the below experiments , to allow for a direct comparison across different cell lines , which yields the same basic conclusions as the complementation experiments ( Figure S3 ) . From these experiments ( Figure 2D ) , we found that Ercc1 complementation showed a significant increase in the efficiency of HDR of the DRins-GFP reporter ( 2 . 9-fold ) , and showed the same effect on SSA ( 2 . 9-fold ) . In contrast , consistent with previous results [7] , Ercc1 played a minor role in Alt-NHEJ ( EJ2-GFP , 1 . 5-fold ) , and insignificant roles in HDR of the DR-GFP reporter and Distal-EJ ( DR-GFP , 1 . 2-fold; EJ5-GFP , 1 . 3-fold ) . These results indicate that Ercc1 is particularly important for DSB repair involving processing of long nonhomologous segments , rather than SSA per se . The above analysis with Ercc1 provides an example of how a genetic factor may not be specific for an individual repair pathway , but rather promotes a mechanistic step that may arise during multiple repair events . To provide further evidence for this notion , we next present a similar analysis with other genetics factors . In addition , we will be including many of the reagents from this genetic analysis during our later description of experiments involving expression of Trex2 . Since Msh2 is important for the mechanistic step of mismatch detection during mismatch repair [33] , we wondered whether this factor might also be important for other pathways of repair in mammalian cells . We analyzed the five reporters described in Figure 1 and Figure 2 using Msh2−/− ES cells [34] , and the complementation approach described for Ercc1 . Notably , expression of Msh2 from the complementing vector was confirmed by immunoblotting ( Figure S1B ) . From these experiments ( Figure 2E ) , we found that Msh2-complementation promotes HDR of both the DR-GFP and DRins-GFP reporters ( 2-fold ) . In contrast , we found that Msh2-complementation had no effect on the overall efficiency of Alt-NHEJ , or Distal-EJ , which is consistent with previous studies in hamster ( CHO ) cells [35] . Furthermore , we find that Msh2-complementation has no clear effect on SSA , which is distinct from the role of Ercc1 in mammalian cells shown above , and the role of MSH2 during SSA in yeast [36]–[38] . This distinction between Ercc1 and Msh2 during HDR is further developed in experiments with Trex2 , in that Trex2 expression promotes HDR in Msh2−/− but not Ercc1−/− cells ( see below ) . In summary , we find that Msh2 is specifically important for HDR , and shows distinct roles during DSB repair compared to Ercc1 . We continued with an analysis of the role of Nbs1 during repair . Previously , an Mre11-complex ( Mre11-Rad50-Nbs1 ) interacting factor , CtIP [9] , [11] , was shown to promote HDR , SSA , and Alt-NHEJ , but was found to be dispensable for Distal-EJ [7] . As well , Nbs1 and Mre11 have recently shown to promote Alt-NHEJ [10] , [39]–[41] . We sought to further investigate the role of Nbs1 during EJ , perform a comparative analysis of the role of Nbs1 during multiple pathways of repair , and develop reagents used in the below Trex2 experiments . For this analysis , we used a double-targeted Nbs1n/h mouse ES cell line that was generated in a previous study , in which the Nbs1 gene was targeted at both alleles by neo ( n ) and hyg ( h ) cassettes , such that these cells were previously shown to lack any Nbs1 protein [42] . However , this result contradicts the notion that the MRE11-complex appears essential for viability of mouse ES cells [43] . Also , while the targeting constructs were designed to remove exon 6 , only one such double-targeted clone was isolated [42] , raising the possibility that one allele may involve an aberrant targeting event that merely causes a decrease in Nbs1 expression , similar to an ES cell line deficient in the Blm helicase [44] . Accordingly , we tested whether the Nbs1n/h cell line still expresses intact full-length Nbs1 , but at a substantially lower level . For this , we performed immunoblot analysis using an anti-Nbs1 antibody on whole cell extracts from Nbs1n/h cells , and found an immunoblot signal at the correct size for Nbs1 that co-migrated with the Nbs1 signal in WT ( see Figure 3B ) . Importantly , the Nbs1 immunoblot signal in Nbs1n/h cells was substantially lower than WT ( at least 5-fold reduction , see Figure 3B ) . The difference between this analysis and the previous study showing no Nbs1 immunoblot signal , using the identical cell line [42] , may reflect variations in the sensitivity of immunoblotting . Nevertheless , the Nbs1n/h cell line is clearly deficient in wild-type levels of Nbs1 , which can be complemented with transient expression of Nbs1 ( see Figure 3B ) . Nbs1n/h cells were previously shown to exhibit reduced HDR and SSA [42] , where Alt-NHEJ was not directly addressed in this study . To test the role of Nbs1 during multiple pathways , we generated Nbs1n/h mouse ES cell lines with an integrated copy of each reporter in Figure 1 and Figure 2B . The parental Nbs1n/h cells and the DR-GFP Nbs1n/h cell line were obtained directly from the laboratory that generated these reagents [42] . Using these cell lines , we evaluated the fold-effect of complementation of Nbs1 on repair in the Nbs1n/h cells , using the same approach as described for Ercc1 . From these experiments ( Figure 3A ) , we found that HDR and SSA were both promoted by Nbs1-complementation ( DR-GFP , 2 . 3-fold; DRins-GFP , 1 . 8-fold; and SA-GFP , 2 . 7-fold ) , consistent with the previous study with these cells [42] . As well , from comparison of SA-GFP repair frequencies between Nbs1n/h and WT cells , the role of Nbs1 during SSA is even more pronounced ( Figure S3C ) . With respect to the EJ reporters in the Nbs1n/h cells ( Figure 3A ) , we found that Alt-NHEJ ( EJ2-GFP ) was promoted by Nbs1 complementation ( 1 . 5-fold ) , whereas Distal-EJ ( EJ5-GFP ) was unaffected . For another measure of Alt-NHEJ , using EJ5-GFP , we quantified the relative ratio of I-SceI-restoration ( S+DEJ ) to I-SceI-resistant EJ products during Distal-EJ ( see Figure 1A ) . With this analysis , a defect in Alt-NHEJ would be expected to cause an increase in the proportion of S+DEJ events . To quantify this repair event , we amplified a region surrounding the I-SceI site in the EJ5-GFP reporter using sorted GFP+ cells , followed by I-SceI digestion analysis . During such analysis , we ensure that all our experiments are performed under conditions for complete I-SceI digestion [7] , [15] , which includes limiting the amount of amplification product [45] , as well as performing digestion analysis of control amplification products with an intact I-SceI site ( see Materials and Methods ) . Using this method , we found that S+DEJ events are increased in Nbs1n/h cells relative to WT ( 1 . 6+/−0 . 1-fold , p<0 . 0001 , Figure 3C , Figure S1C ) . We also found that transient complementation of Nbs1 in Nbs1n/h cells reduced S+DEJ products back to near WT levels ( Figure 3C , Figure S1C ) . Thus , Nbs1 appears to promote Alt-NHEJ , but is dispensable for S+DEJ . Thus , we suggest that Nbs1 is important for a number of repair events that require access to homology . We next addressed the role of Xrcc4 during repair , which is a factor that binds to Ligase IV and promotes both its stability and function during NHEJ [46] . In previous studies , Xrcc4−/− mouse ES cells have been shown to exhibit elevated levels of HDR [15] . We extended the analysis of these Xrcc4−/− ES cells [47] , using the reporters and complementation approach described above , where expression of Xrcc4 from the complementing vector was confirmed by immunoblotting ( Figure S1B ) . In particular , we performed these experiments to address the role of Xrcc4 during SSA , and to establish reagents used for the below analysis of EJ using Trex2 . From these experiments ( Figure 3D ) , we found that Xrcc4 complementation resulted in a significant inhibition of HDR ( DR-GFP , 1 . 8-fold; DRins-GFP , 2 . 7-fold ) , SSA ( SA-GFP , 2 . 8-fold ) , Alt-NHEJ ( EJ2-GFP , 2 . 9-fold ) , and Distal-EJ ( EJ5-GFP , 1 . 5-fold ) . To characterize the nature of EJ events in Xrcc4-deficient cells , we determined the efficiency of I-SceI-restoration ( S+DEJ ) during Distal-EJ , using amplification analysis of GFP+ sorted cells from the EJ5-GFP transfections , as described above for Nbs1 . From this analysis , we found that the efficiency of S+DEJ is reduced in Xrcc4−/− ES cells relative to WT cells ( 2 . 8+/−0 . 2-fold , p<0 . 0001 , Figure 3E , Figure S1C ) . As well , the Xrcc4−/− cells show an additional class of smaller I-SceI-resistant products , indicative of extensive deletions during EJ ( Figure 3E ) . Next , we performed this EJ analysis on GFP+ sorted cells following co-expression of I-SceI with Xrcc4 in Xrcc4−/− cells . From this experiment , we found that Xrcc4 expression suppressed the formation of extensive deletion products , suggesting that transient complementation of Xrcc4 can restore its end-protection functions . In contrast , co-expression of I-SceI and Xrcc4 caused only a partial restoration of the efficiency of S+DEJ in Xrcc4−/− cells ( 1 . 5-fold increase relative to Xrcc4−/− , p = 0 . 0008 , Figure 3E , Figure S1C ) . This result may reflect an inability to completely restore the ligase functions of Xrcc4-Ligase IV by transient complementation . However , even comparing Xrcc4−/− versus WT for the efficiency of S+DEJ , we find that Xrcc4 is not absolutely required for this repair event . Thus , other ligase complexes may be able to complete the S+DEJ event , particularly since this product could be stabilized by the microhomology of the cohesive I-SceI overhangs [46] . In summary , these data indicate that Xrcc4 plays some role in S+DEJ events , and suppresses SSA , Alt-NHEJ , HDR , and Distal-EJ . We suggest that suppression of HDR , SSA , and Alt-NHEJ could result from the end-protection function of Xrcc4 [48] , which may limit end resection during these pathways . In contrast , the finding that Xrcc4-complementation suppresses Distal-EJ may reflect a role for Xrcc4 is supporting EJ between proximal ends . For comparison with Nbs1 and Xrcc4 , we also determined the effect of Brca1-deficiency on repair of the EJ5-GFP reporter . Also , we introduce this cell line here , as it is used below for additional EJ experiments with Trex2 ( see below ) . Specifically , we integrated EJ5-GFP into mouse ES cells that are homozygous for an exon 11-deletion allele ( Brca1−/− ) , which encodes a protein with a substantial internal deletion [49] , [50] . The size of Brca1 has made transient complementation unfeasible , such that we have been limited to a comparison of repair versus WT . The reporters DR-GFP and SA-GFP have already been analyzed in this Brca1−/− cell line , showing a 5 . 3-fold and a 1 . 8-fold decrease relative to WT ES cells , in HDR and SSA , respectively [8] . Using the Brca1−/− EJ5-GFP cell line , we expressed I-SceI and subsequently determined the frequency of Distal-EJ . As well , we quantified the relative efficiency of S+DEJ versus I-SceI-resistant Distal-EJ products in GFP+ sorted cells . From these experiments , we found that the total frequency of Distal-EJ ( %GFP+ ) was increased in Brca1−/− versus WT ES cells ( 2-fold , p<0 . 0001 , Figure S3E ) . As well , from quantification of S+DEJ from GFP+ cells , we found a significant decrease in this repair event in Brca1−/− cells compared to WT ES cells ( 3-fold decrease , Figure 3F , Figure S1C ) . Thus , Brca1 promotes S+DEJ , which may indicate that Brca1 is important for EJ of cohesive ends . Based on this notion , Brca1 could feasibly promote S+EJ at proximal ends , which may account for the suppression of Distal-EJ . While these proximal S+EJ events cannot be quantified , this model is supported by other reports showing a role for Brca1 during EJ of plasmid substrates with cohesive ends [51] , [52] , and are developed with the below Trex2 experiments . We next considered the possibility that the outcome of these studies on repair may be affected by the unstable nature of EJ products that restore the I-SceI site , which are prone to repeated cutting by I-SceI . This property of endonuclease-generated DSBs has been referred to as the persistent nature of such DSBs in previous studies [18]–[21] . Thus , we developed a method to promote the formation of I-SceI-resistant EJ products , and thereby limit the persistent nature of I-SceI-induced DSBs . We then used this approach to address how the relative persistence of DSBs may affect the mutagenic consequences of such damage . For this , we co-expressed I-SceI with a protein that we predicted would catalyze partial degradation of the 3′ ssDNA 4 nt overhangs generated by I-SceI , and hence promote formation of EJ products that are resistant to cleavage by I-SceI . Specifically , we expressed mammalian Trex2 , which is a potent non-processive 3′ to 5′ exonuclease [22] , [23] , [53] . We first determined whether Trex2 expression promotes EJ products that are resistant to cleavage by I-SceI , using the EJ5-GFP reporter . In these experiments , transfection of the Trex2 expression vector leads to at least a 10-fold increase of Trex2 mRNA above WT , largely due to the relatively low endogenous level of Trex2 expression in these cells , based on quantitative RT-PCR ( data not shown ) . Following transfection of I-SceI along with Trex2 or EV , we quantified the formation of I-SceI-resistant EJ products . Regarding this analysis of two tandem I-SceI induced DSBs , three different sets of ends can be paired during EJ . Two of these end-pairs result in retention of the intervening puro cassette: pairing of the proximal ends that flank the 3′ I-SceI site , and pairing of the proximal ends that flank the 5′ I-SceI site . In contrast , pairing of the distal ends of the 5′ and 3′ I-SceI sites ( Distal-EJ ) results in loss of the intervening puro gene . To quantify formation of I-SceI-resistant EJ products for each of these end-pairs , we amplified the region surrounding each EJ event ( Figure 4A ) , and subjected the amplification products to I-SceI digestion analysis . From this analysis , we found that Trex2 expression results in the formation of I-SceI-resistant EJ products between proximal ends of the 3′ I-SceI site ( 24%+/−8% and 27%+/−5% of total amplified product in WT and Trex2null ES cells [22] , respectively ( Figure 4B and 4C ) . In addition , we found a similar effect of Trex2 expression on EJ between proximal ends of the 5′ I-SceI site ( 30%+/−7 I-SceI-resistant products in WT ES cells , Figure 4D ) . In contrast , in the absence of Trex2 expression , these I-SceI-resistant proximal-EJ products were not detectable ( see S+EV , Figure 4B–4D ) . Regarding Distal-EJ , Trex2 expression caused a substantial increase in the fraction of I-SceI-resistant products , in that S+DEJ products were undetectable in the GFP+ repair events from cells transfected with Trex2 ( Figure 4E ) . Thus , Trex2 expression promotes the formation of I-SceI-resistant EJ products in EJ5-GFP , between proximal ends at both the 5′ and 3′ I-SceI sites , as well as during Distal-EJ . We next addressed whether the exonuclease activity of Trex2 is involved in its ability to promote I-SceI-resistant EJ products . To begin with , we characterized the repair junctions of the Trex2-mediated I-SceI-resistant EJ products at the 3′ I-SceI site , by cloning these products and sequencing individual clones . From this analysis , we found sequences that are consistent with exonucleolytic processing of the 3′ overhangs ( Figure 4F ) . For example , the most abundant product ( 6/11 , 54% ) shows mutation of the 3′ overhang ATAA/TATT to AA/TT . Notably , only one product ( 1/11 , 9% ) showed any evidence of microhomology ( 1 nt . microhomology , ATAA/TATT to A/T ) . Thus , the structures of these EJ products are consistent with the known non-processive 3′ to 5′ exonuclease activity of Trex2 [22] , [23] , [53] . In addition , we characterized a mutant form of Trex2 ( H188A ) , which has been shown to lack exonuclease activity , but retains significant DNA binding activity ( reduced only 60% from Trex2-WT ) [54] . For this , we co-transfected expression vectors for Trex2-H188A and I-SceI into WT ES cells with EJ5-GFP , using identical conditions as the previous experiments with wild-type Trex2 . From these experiments , we found that the Trex2-H188A mutant caused no detectable formation of I-SceI-resistant EJ products at the 3′ I-SceI site ( see Figure 4B ) . Along these lines , we also wanted to address whether Trex2 expression caused an overall increase in DNA damage , as assessed by immunoblotting of a marker for chromosome breaks , γH2AX [55] . We found that transfection of Trex2 had no affect on the level of γH2AX , as compared to spontaneous γH2AX levels from parallel EV transfections ( Figure S2A ) , which is consistent with previous reports showing expression of wild-type Trex2 does not cause an increase in chromosome breaks [53] . Given that Trex2-mediated EJ products do not involve substantial amounts of microhomology ( see Figure 4F ) , we hypothesized that these repair events might be dependent upon Xrcc4 , since Xrcc4-Ligase IV is particularly effective at ligating substrates that are not stabilized by annealing [56] . To test this , we co-transfected the Trex2 and I-SceI expression vectors in Xrcc4−/− cells with the EJ5-GFP reporter . We then quantified the formation of I-SceI-resistant EJ products at the 3′ I-SceI site , as described for WT cells ( see Figure 4B ) . From these experiments , we reproducibly found no detectable level of I-SceI-resistant proximal EJ products from Trex2 expression in Xrcc4−/− cells ( Figure 4G ) , where such products were readily detected in WT cells ( see Figure 4B ) . This result indicates that EJ of ends processed by Trex2 is dependent upon Xrcc4 , which may reflect a critical role for Xrcc4-Ligase IV during ligation of ends that do not contain substantial microhomology . Consistent with this notion , Xrcc4 is much less important for I-SceI-restoration ( see Figure 3E ) , while the 4 nt of microhomology from the I-SceI overhangs might allow EJ by other ligase complexes [46] . In total , these data support the notion that the exonuclease activity of Trex2 catalyzes partial degradation of I-SceI DSB overhangs , thereby promoting the formation of I-SceI-resistant EJ products . However , it is certainly possible that Trex2 additionally could be recruiting other factors to facilitate the EJ process . In any case , co-expression of Trex2 and I-SceI appears to result in I-SceI-resistant EJ products . Since these products cannot be repeatedly cut by I-SceI , we suggest that Trex2 expression can limit the persistent nature of I-SceI-induced DSBs . We then considered whether expression of Trex2 affects the relative efficiency of distinct repair events , beginning with the reporters described in Figure 1 . From these experiments ( Figure 5A and 5B ) , we found that co-expression of Trex2 with I-SceI in WT ES cells caused a striking decrease in the efficiency of Distal-EJ ( 4 . 2-fold ) , as well as a significant decrease in SSA and Alt-NHEJ ( SA-GFP , 2 . 8-fold; EJ2-GFP , 2-fold ) . In contrast , expression of the Trex2-H188A nuclease-deficient mutant caused no statistical difference in such repair ( Figure S2B ) . These results indicate that limiting the persistence of DSBs via Trex2 causes a reduction in Distal-EJ , SSA , and Alt-NHEJ , each of which result in significant deletion mutations . Next , we analyzed the effect of Trex2 on the HDR reporters shown in Figure 2 , using the co-transfection approach described above . From these experiments ( Figure 5A and 5B ) , we found Trex2 expression caused no effect on HDR of DR-GFP , and a minor increase on HDR of DRins-GFP ( 1 . 5-fold ) , where expression of the Trex2-H188A mutant showed no effect on HDR in either reporter ( Figure S2B ) . The increase in HDR for DRins-GFP may be due to Trex2-mediated removal of the I-SceI-overhangs , which would remove some of the mismatched base-pairs between the 5′ DSB end and the template for repair ( see Figure S1A ) . This overhang processing may be particularly important for DRins-GFP , since this reporter may be specifically affected by the mismatched base-pairs between the 5′ end of the DSB and iGFP , since the terminus of the 3′ end of the DSB is not homologous to iGFP ( see Figure 2B ) . In any case , these results suggest that Trex2 expression does not inhibit HDR , which is distinct from the effects on Distal-EJ , SSA , and Alt-NHEJ . We next investigated whether repair of DSB ends modified by Trex2 show distinct genetic requirements , focusing on Distal-EJ and HDR . For this , we determined the effect of Trex2 expression on the EJ5-GFP and DR-GFP reporters in each of the DNA repair mutant cell lines described earlier in this study . For each of these cell lines , we first determined whether Trex2 expression promotes I-SceI-resistant EJ products between proximal ends of the 3′ I-SceI site in EJ5-GFP . As described in Figure 4G , this Trex2-mediated EJ product was not detected in Xrcc4−/− cells . However , for the other cell lines ( Ercc1−/− , Msh2−/− , Brca1−/− , and Nbs1n/h ) , we found that Trex2 expression causes the formation of this I-SceI-resistant EJ product to a level that is indistinct from WT ( Figure S2C ) . Thus , for each of the cell lines except Xrcc4−/− , Trex2 expression promotes the formation of I-SceI-resistant EJ products that are not prone to repeated cutting , which likely limits the persistence of I-SceI-induced DSBs . Subsequently , we quantified the effect of Trex2 expression on the frequency of Distal-EJ and HDR for each of these lines , as determined for WT ES cells in Figure 5A . Beginning with Ercc1 , we found that Trex2 expression in Ercc1−/− cells affected Distal-EJ and HDR in a manner indistinguishable from WT ( Figure 6A and 6B , respectively ) . In contrast , each of the other cell lines showed distinct effects of Trex2 expression on Distal-EJ and/or HDR . We found that Nbs1n/h cells showed a much more mild affect of Trex2 expression on Distal-EJ ( 1 . 8-fold compared to 4 . 2-fold in WT , Figure 6A ) . Regarding HDR , Trex2 expression in the Nbs1n/h cells showed a significant decrease in this pathway ( 2-fold , Figure 6B ) . In Brca1−/− cells , Trex2 caused an inhibition of Distal-EJ that was similar to WT ( Figure 6A ) , but showed a significant decrease in HDR ( 2-fold , Figure 6B ) . Thus , with Trex2 expression , which likely results in a less persistent DSB , Nbs1 and Brca1 show an increased role in promoting HDR , and Nbs1 is important for limiting the frequency of Distal-EJ . In contrast , with Msh2−/− cells , we found that Trex2 expression caused an elevation of HDR ( 1 . 6-fold , Figure 6B ) , and a reduction in Distal-EJ that is similar to WT ( Figure 6A ) . In this case , since Trex2-mediated processing of the 3′ I-SceI overhangs may remove a few of the mismatches between SceGFP and iGFP ( see Figure S1A ) , this result indicates that such processing is particularly important for HDR in Msh2−/− cells . As well , since Trex2 expression did not cause an increase in HDR in Ercc1−/− cells ( Figure 6B ) , these results further support the notion that Ercc1 and Msh2 play distinct roles during HDR , as described in Figure 2 . Finally , we also addressed how Trex2 expression may affect DSB repair in Xrcc4−/− cells . As described above ( see Figure 4G ) , Trex2 expression in these cells does not result in I-SceI-resistant EJ products between proximal ends , suggesting that Xrcc4 is required for EJ of proximal Trex2-modified ends . Regarding distal ends , we found that Trex2 expression caused a decrease in the frequency of Distal-EJ in Xrcc4−/− cells ( 6 . 9-fold , Figure 6A ) . These results indicate that Trex2-modified ends are not efficiently repaired by EJ between either proximal or distal ends in the absence of Xrcc4 . Accordingly , such products may be more likely to be processed by end resection , and hence be repaired by other pathways . In support of this notion , we find that Trex2 expression caused a substantial increase in HDR and Alt-NHEJ in Xrcc4−/− cells ( DR-GFP , 2 . 3-fold; EJ2-GFP , 1 . 8-fold , Figure 6B and 6C ) . Furthermore , the suppression of SSA by Trex2 was substantially reduced in Xrcc4−/− cells compared to WT ( 1 . 3-fold versus 2 . 8-fold , respectively , Figure 6C ) . This minor decrease in SSA may reflect a bias towards HDR and/or Alt-NHEJ of Trex2-processed DSB ends in Xrcc4−/− cells . In summary , Xrcc4−/− cells appear deficient for EJ of Trex2-processed ends , relying more on other repair pathways that likely require end resection , particularly HDR and Alt-NHEJ . The findings of the genetic analysis reinforce the notion that some factors are not specific for individual repair pathways per se , but rather may promote a particular mechanistic step that arises during multiple repair events . For example , Msh2 , which binds to mismatched bases and promotes their removal during mismatch repair [33] , also appears to function during HDR of DSBs , as measured by the DR-GFP and DRins-GFP reporters . Accordingly , Msh2 may be important for removing mismatches between the DSB ends and the template for repair ( see Figure S1A ) . Such mismatch removal may occur during strand exchange and/or prior to strand extension . In support of this notion , expression of Trex2 , which could remove the mismatches within the 3′ overhangs ( see Figure S1A ) , promotes HDR in Msh2-deficient cells . Thus , the role of Msh2 during mismatch detection and removal [33] may be important for multiple repair pathways and types of DNA damage . As another example , we find that Ercc1 promotes repair events that require the removal of an extended nonhomologous insertion , rather than a particular repair pathway . Namely , Ercc1 promotes both SSA , as well as an HDR event that requires removal of a nonhomologous insertion . These results are consistent with the known biochemical function of Ercc1/Xpf in cleaving nonhomologous 3′ ssDNA tails [29] , which is also important for nucleotide excision repair [57] , EJ of plasmid substrates [30] , and gene targeting [58] . So , we suggest that this activity of Ercc1 may be important for the removal of nonhomologous segments during multiple repair pathways . Similarly , we find that Nbs1 is important for a number of repair events that require access to homology , similar to previous results with the Mre11-complex interacting factor , CtIP [7] . Given that these factors are implicated in ssDNA formation via end resection [9] , [11] , these results suggest that Nbs1/CtIP-mediated end resection might be a common step among HDR , SSA , and Alt-NHEJ . However , the role of Nbs1 during repair could also reflect functions during DNA end tethering and/or activation of the DNA damage response via ATM [59] , [60] . In contrast to Nbs1 , we find that Xrcc4 suppresses repair that requires access to homology ( HDR , SSA , and Alt-NHEJ ) . Regarding mechanism , the end-protection function of Xrcc4 [48] may suppress ssDNA formation via end resection , and hence access to homology . As an additional possibility , the EJ functions of Xrcc4-Ligase IV [3] may also be important to effectively compete with repair pathways that require access to homology . In either case , bypass of Xrcc4 function is likely a common mechanistic step during HDR , SSA , and Alt-NHEJ . Considering the EJ functions of Xrcc4 , we find that the non-cohesive ends formed by Trex2 cannot be efficiently repaired by EJ in Xrcc4-deficient cells , using either proximal or distal ends . These findings likely reflect the unique capability of Xrcc4-Ligase IV during ligation of DSB ends that are not stabilized by annealing [3] , [56] . As well , this defect in EJ of non-cohesive ends is consistent with the IR hypersensitivity and V ( D ) J recombination defects of the Xrcc4−/− ES cells [47] . Furthermore , we find that Trex2 expression causes an increase in HDR and Alt-NHEJ in Xrcc4−/− cells . Thus , Xrcc4-deficient cells show an increased reliance on HDR and Alt-NHEJ for repair of non-cohesive DSB ends . Since HDR and Alt-NHEJ are promoted by CtIP [7] , whose functions appear limited to the S/G2/M phases of the cell cycle [9] , [61] , Xrcc4−/− cells would be expected to show an enhanced ability to repair DSBs formed these cell cycle phases . Consistent with this notion , the IR hypersensitivity of Xrcc4-deficient cells is diminished when these cells are exposed to IR in late S phase [62] . In summary , we find that Xrcc4-deficient cells show defects in EJ repair of non-cohesive DSB ends , as well as an increased reliance on HDR and Alt-NHEJ for repair of such DSB ends . Finally , while Brca1 is similar to Nbs1 in promoting HDR and SSA [8] , we also found that Brca1 supports S+DEJ and suppresses the total frequency of Distal-EJ , which suggests that Brca1 could be important for I-SceI-restoration EJ . In contrast , Brca1 is not important for EJ repair of Trex2-processed ends , which lack significant microhomology . These findings raise the possibility that Brca1 may be particularly important for EJ of cohesive ends that do not require end resection . Consistent with this notion , previous studies have shown a role for Brca1 during EJ repair of plasmid substrates with cohesive ends [51] , [52] . Thus , the function of Brca1 during repair cannot be limited to promoting access to homology via ssDNA formation [11] , [63] , which is also supported by findings that Brca1 may associate with a number of multi-subunit complexes [64] , and includes additional functions apart from E3 ligase activity [65] . In summary , with this genetic analysis , we have provided some distinct findings on the role of individual factors during repair of both cohesive and non-cohesive DSB ends . We addressed how the persistence of a DSB affects the frequency of mutagenic repair events . For this , we used expression of Trex2 , which we find promotes the formation of I-SceI-resistant EJ products , which we suggest limits the persistent nature of I-SceI-induced DSBs . While Trex2 is likely promoting these products directly through its exonuclease function , it is certainly possible that Trex2 could additionally be recruiting other factors to facilitate the formation of I-SceI-resistant EJ products . In either case , with this approach , we have found that limiting the persistence of a DSB reduces the frequency of deletion mutations caused by Distal-EJ , SSA , and Alt-NHEJ . Regarding the effect on Distal-EJ , this result suggests that the relative persistence of DSBs can affect the fidelity of end-pairing during EJ . Persistent breaks could lead to a failure of proximal end-pairing by a number of mechanisms , depending on which factors perform this pairing function . As one example , the DNA tethering activity of the Mre11-complex [59] , [60] may support proximal end-pairing during EJ [21] . In this model , persistent DSBs could signal a direct disruption of the Mre11-complex tethering activity , which could lead to the loss of proximal end-pairing . Alternatively , the Mre11-complex may not be able to sustain correct end pairing under the conditions of a persistent DSB . Consistent with such models , we find that Nbs1 has no effect on Distal-EJ of relatively persistent DSBs ( I-SceI expression alone ) . In contrast , we find that Nbs1 is important to inhibit Distal-EJ of relatively less persistent DSBs ( expression of both I-SceI and Trex2 ) . Thus , Nbs1 may promote correct end-pairing during EJ , but in a manner that is less efficient for persistent DSBs . In contrast to Nbs1 , Xrcc4 and Brca1 are important for inhibition of Distal-EJ of persistent DSBs ( I-SceI expression alone , see Figure 3D , Figure S3E ) . In summary , we suggest that DSB persistence may affect the relative roles of factors and complexes involved in end-pairing during EJ , where the Trex2 approach described here may facilitate future investigation into this process . In addition to affecting Distal-EJ , expression of Trex2 also caused a significant inhibition of SSA and Alt-NHEJ , but not HDR . Considering one model , the Trex2 protein may directly inhibit end resection , perhaps by blocking access of a resection factor to the DSB . However , such direct inhibition does not explain the differential effect of Trex2 on SSA and Alt-NHEJ versus HDR . As well , this model is inconsistent with the findings that Trex2-H188A does not affect repair , as this protein lacks the exonuclease activity , but shows only a 60% reduction in DNA binding activity [54] . Perhaps more likely , Trex2 expression limits the persistence of I-SceI-induced DSBs , which decreases the probability that end resection will be initiated , but in a manner that diminishes Alt-NHEJ and SSA , but not HDR . This differential effect between the pathways may be related to the unique requirement for the sister chromatid during HDR , which is the preferred template even if an intrachromosomal repeat is present [66] . Thus , considering this model , one of the earliest mechanistic steps following a DSB could be attempts to detect the presence of the sister chromatid . If the sister chromatid is found , this event could trigger Xrcc4-bypass and promotion of end resection via CtIP and the Mre11-complex [9] , [11] . Given the presence of the sister chromatid , such end resection would likely be followed by efficient strand invasion and HDR . This model is supported by our findings that factors implicated in end resection , Nbs1 and Brca1 [11] , [63] , show an elevated importance for HDR of a less persistent DSB ( i . e . when Trex2 is expressed ) . Although , this result could also reflect a role for Nbs1 during direct detection of the sister chromatid , given the DNA tethering capabilities of the Mre11-complex [59] , [60] . To summarize this model , sister chromatid detection would precede EJ to trigger end resection , such that the persistent nature of a DSB may not be particularly relevant for the frequency of HDR . Furthermore , in considering this model , we note that the persistence of a DSB has been shown to differentially affect HDR versus SSA in another set of findings . Specifically , a previous study compared repair of both I-SceI-generated DSBs and IR-induced DSBs , where the I-SceI DSBs would be expected to be more persistent than IR DSBs [67] . In this study , I-SceI-generated DSBs were found to stimulate both Rad51-dependent and Rad51-independent repair pathways , which are measures of HDR and SSA , respectively . In contrast , less persistent IR DSBs showed a strong preference for Rad51-dependent repair ( HDR ) . Thus , this previous study is consistent with the notion that the persistence of I-SceI-generated DSBs may be more important for SSA than HDR . As well , it is notable that HDR of the DRins-GFP reporter is also not inhibited by Trex2 expression . This reporter is similar to DR-GFP in that it requires strand invasion with a homologous template , but is similar to SSA in that it requires Ercc1-dependent removal of an insertion . Thus , the Trex2/DRins-GFP result further supports the notion that strand invasion may be the mechanistic step of HDR that is relatively unaffected by the persistence of a DSB . Regarding another consideration with this reporter , the finding that HDR is less efficient for DRins-GFP than DR-GFP may suggest that limiting efficient strand invasion to one end of the DSB may suppress HDR . These data raise the possibility that strand invasion of both DSB ends may be required for efficient HDR , which is evocative of the classical double-strand break repair model [68] . Finally , since a number of investigators have been developing meganucleases to initiate gene targeting [69] , we suggest that co-expression of such meganucleases with Trex2 may provide a means to maintain efficient homologous targeting by HDR , while simultaneously suppressing repair events that are genome destabilizing . In general , we suggest that co-expression studies of meganucleases with Trex2 will lead to additional insight into the pathways that support genome maintenance . The DRins-GFP reporter is a derivative of pim-DR-GFP#6 [70] , where a 464 nt . BglII/AvrII intronic fragment of the mouse Rb gene [24] was cloned downstream of the I-SceI site . Complementation/expression cassettes for each gene were cloned into pCAGGS-BSKX [12] . The ERCC1 and Nbs1 complementation vectors have been described previously [8] , [42] , the Msh2 insert was derived from pHA801 [34] , the XRCC4 insert was derived from clone GI:16740906 , ATCC#10659357 . The mouse Trex2 coding sequence is present within a single exon [22] , and thus was generated from PCR amplification of mouse ES genomic DNA for cloning into pCAGGS-BSKX [12] , using these primer sequences: 5′cagctctaggcctcattgtt and 5′agagcctggatgaatggatg . The expression vector of the Trex2-H188A mutant was generated by site-directed mutagenesis of the above expression vector with the Quikchange method ( Stratagene ) using the primer 5′gctgaacccagtgctgccgcttcagcagaaggtgatgtgc along with the complementary primer . Chromosomal integration of reporters into mouse ES cells was performed by electroporation using a pulse of 700–730 V 10 µF . Electroporation cuvettes contained 107 cells in 0 . 8 ml of Optimem ( Invitrogen ) , along with 20–30 µg of linearized plasmid for random integration and 70 µg of linearized plasmid for Pim1 targeting . Culturing of mouse ES cells on gelatin , and targeting of reporters to the Pim1 locus was performed as previously described [7] . The reporters targeted to Pim1 are DR-GFP , DRins-GFP , EJ2-GFP , EJ5-GFP into WT and Xrcc4−/− , and EJ5-GFP into Trex2null and Brca1−/− . Otherwise , individual reporters were introduced by random integration using the linked puro gene by selecting for clones in 1–2 µg/ml puromycin , where an intact copy of the reporter was confirmed by Southern blotting , as described previously [7] , [26] . To measure repair , 105 cells were plated onto 12 well plates , and transfected the next day with 3 . 6 µl of Lipofectamine 2000 ( Invitrogen ) mixed with 0 . 8 µg of pCBASce , along with 0 . 4 µg of either empty vector ( pCAGGS-BSKX ) , or the relevant complementation/expression vector . Transfection was performed in 1 ml of antibiotic-free media for 4 hours , after which the transfection media was replaced with regular media . The percentage of GFP positive cells was quantified by flow cytometric analysis ( FACS ) 3 d after transfection on a Cyan ADP ( Dako ) , from cells suspended and fixed in phosphate-buffered formaldehyde . Amplification of PCR products from sorted GFP+ cells , associated restriction digests , and quantification of bands were performed as previously described [7] , [15] , where KNDRF and KNDRR are shown as p3 and p2 respectively , primer p1 is EJ5purF: 5′agcggatcgaaattgatgat , primer p4 is EJ5purR: 5′ cttttgaagcgtgcagaatg , and DRins-GFP amplifications use KNDRF and DRRT6: 5′aggttcagggggaggtgt . To ensure complete I-SceI digestion , PCR products were purified using a GFX column ( GE ) , and digested for 1 h ( 37°C ) with 5 U of I-SceI ( NEB ) , followed by an adding another 5 U of I-SceI and 1 h of digestion . With this protocol , we always ensure complete cutting with a control PCR template that contains an intact I-SceI site ( see Figure 4A ) , and further ensure that our experimental samples contain less or equal amounts of PCR product as these controls , to avoid any possibility of problems with excess substrate affecting complete cutting [45] . Repair frequencies are the mean of a minimum of four transfections where error bars represent the standard deviation from the mean . In most cases , repair frequencies are shown relative to samples co-transfected with I-SceI and an empty vector ( EV ) . For this calculation of fold-complementation , the percentage of GFP+ cells from each sample was divided by the mean value of the EV samples treated in the parallel experiment . Statistical analysis was performed using the unpaired t-test . Transfections were performed as in the repair assays , and 2 d after transfection , protein was isolated by repeated freeze/thawing in NETN buffer ( 20 mM Tris pH 8 , 100 mM NaCl , 1 mM EDTA , 0 . 5% IGEPAL , 1 mM DTT ) with Protease Inhibitor Cocktail ( Roche ) . Protein was separated on 4–12% SDS-PAGE , and probed with anti-NBS1 antibody ( Bethyl labs , A301-284A ) and HRP-conjugated anti-rabbit ( Santa Cruz Biotechnology , sc-2004 ) , or probed with HRP-conjugated anti-GAPDH ( Abcam , ab9482 ) , and developed with ECL ( GE ) .	A deleterious lesion in DNA is a break of both strands , or a chromosome double-strand break ( DSB ) . DSBs can arise during normal cellular metabolism , but are also a consequence of many forms of cancer therapy . If DSBs are not repaired prior to cell division , entire segments of a chromosome can be lost . Several pathways ensure that DSBs are repaired , though some pathways are prone to causing mutations and/or chromosomal rearrangements , each of which can contribute to cancer development . In the first part of this study , we describe the roles of individual genetic factors in distinct repair pathways of DSBs generated by the I-SceI endonuclease . From these studies , we find that some factors can function in multiple repair pathways . In the second part of this study , we present a method for partially degrading the cohesive DSB overhangs that are generated by I-SceI , which we find facilitates repair products that are not prone to being re-cut by the endonuclease . As a consequence , we have limited the persistence of such breaks , which we find causes a reduction in repair pathways that lead to significant genetic loss . As well , we use this method to characterize the role of individual genetic factors during the repair of non-cohesive DSB ends .	[ "Abstract", "Introduction", "Results", "Discussion", "Materials", "and", "Methods" ]	[ "molecular", "biology/dna", "repair", "molecular", "biology/recombination", "genetics", "and", "genomics/chromosome", "biology" ]	2009	Limiting the Persistence of a Chromosome Break Diminishes Its Mutagenic Potential
Ellis-van Creveld ( EvC ) syndrome is a skeletal dysplasia , characterized by short limbs , postaxial polydactyly , and dental abnormalities . EvC syndrome is also categorized as a ciliopathy because of ciliary localization of proteins encoded by the two causative genes , EVC and EVC2 ( aka LIMBIN ) . While recent studies demonstrated important roles for EVC/EVC2 in Hedgehog signaling , there is still little known about the pathophysiological mechanisms underlying the skeletal dysplasia features of EvC patients , and in particular why limb development is affected , but not other aspects of organogenesis that also require Hedgehog signaling . In this report , we comprehensively analyze limb skeletogenesis in Evc2 mutant mice and in cell and tissue cultures derived from these mice . Both in vivo and in vitro data demonstrate elevated Fibroblast Growth Factor ( FGF ) signaling in Evc2 mutant growth plates , in addition to compromised but not abrogated Hedgehog-PTHrP feedback loop . Elevation of FGF signaling , mainly due to increased Fgf18 expression upon inactivation of Evc2 in the perichondrium , critically contributes to the pathogenesis of limb dwarfism . The limb dwarfism phenotype is partially rescued by inactivation of one allele of Fgf18 in the Evc2 mutant mice . Taken together , our data uncover a novel pathogenic mechanism to understand limb dwarfism in patients with Ellis-van Creveld syndrome . Ellis-van Creveld syndrome ( EvC ) , a chondroectodermal dysplasia and mesoectodermal dysplasia , is an autosomal recessive congenital disease [1] . EvC patients generally bear a variety of defects such as shorter limbs and ribs , postaxial polydactyly , as well as dysplastic nails and teeth . Previous genetic studies have shown that more than two thirds of EvC patients carry mutations in either EVC or EVC2 [2 , 3] . Interestingly , LIMBIN , originally identified as the causative gene for chondrodysplastic dwarfism in Japanese Brown cattle , was later discovered as the cattle orthologue of EVC2 [4] . A recent study reported spontaneous mutations in EVC2 in Tyrolean Grey cattle [5] . In both cases , EVC2 homozygous mutant cattle suffered from chondrodysplastic dwarfism , suggesting evolutionarily conserved functions of EVC2 during appendicular bone development . Appendicular bone development occurs through endochondral ossification , a process during which a series of locally produced factors are interacting to ensure the correct length and shape of each skeletal element [6 , 7] . Fibroblast Growth Factor ( FGF ) signaling , mediated by FGF18 produced in the perichondrium , is involved in this regulatory network [8 , 9] . In the growth plate , FGF signaling inhibits chondrocyte proliferation through STAT1-mediated p21 expression [10 , 11] and inhibits chondrocyte differentiation through MEK/pERK-mediated signaling [12] . In addition to FGF signaling , Indian Hedgehog and parathyroid hormone-related protein ( PTHrP ) signaling plays a central role in the regulation of chondrocyte proliferation , differentiation and maturation . Indian Hedgehog is synthesized by pre-hypertrophic chondrocytes [13] and stimulates PTHrP synthesis in chondrocytes at the distal end of the growth plate [14] . PTHrP promotes proliferation of chondrocytes and prevents them from progressing to pre-hypertrophic differentiation directly [14] and through suppression of FGF receptor 3 ( Fgfr3 ) expression [15] . Once proliferating chondrocytes are far away enough from the source of PTHrP , they differentiate into pre-hypertrophic chondrocytes , start synthesizing Indian Hedgehog , and then further mature into hypertrophic cells . Recently , primary cilium has been identified as a cellular organelle essential for Hedgehog signaling transduction [16] . Upon Hedgehog signaling induction , Smoothened ( SMO ) moves into the cilium [17] . At the same time , a protein complex consisting of glioma-associated oncogene ( GLI ) and Suppressor of Fused ( SUFU ) moves into the cilium , where it is quickly dissociated [18–20] . GLI proteins subsequently traffic out of the cilium and move into the nucleus , where they function as transcriptional activators of Hedgehog responsive genes [18] . Recent studies have also demonstrated that EVC and EVC2 interact with each other and are mutually required for localization at the base of the primary cilium [21–23] . Protein structure analysis coupled with biochemical analysis demonstrated that both EVC and EVC2 are N-terminus-anchored type I transmembrane proteins [24] . Despite that there is no functional domains identified in EVC or EVC2 , they intracellularly localize at the base of the primary cilium through interaction with EFCAB7 and IQCE [25] . Upon induction of Hedgehog signaling , the EVC/EVC2 complex interacts with SMO at the base of cilium , which affects GLI2 accumulation in the ciliary tips [21 , 23] . Consistent with in vitro mechanistic studies , diminished Hedgehog signaling was observed in the growth plates of mice lacking either Evc or Evc2 [23 , 26 , 27] . Since the Hedgehog-PTHrP loop has a central role in regulating chondrocyte proliferation and maturation , thereby determining the length of the appendicular skeleton , disrupted Hedgehog signaling in the growth plate is the speculated reason for dwarfism in the EvC syndrome [21 , 23] . Previous genetic studies have indicated that Hedgehog signaling plays critical roles during many processes during mouse embryonic development . For example , a loss-of-function mutation in Shh leads to holoprosencephaly , neural tube defect , and a variety of midline defects , such as cyclopia , cleft palate and cleft lips [28 , 29] , and a single digit and other limb skeletal defects [30] . A loss-of-function mutation in Indian Hedgehog ( Ihh ) was shown to severely impair endochondral ossification , resulting in extremely short and undermineralized limb bones [13] and in early closure of the growth plates [31] . Despite the indispensable functions implicated by previous studies for EVC/EVC2 in transducing Hedgehog signaling , except for short limbs , the aforementioned severe defects are observed neither in EvC patients nor in Evc and Evc2 mutant mice . This discrepancy between the function of EVC/EVC2 in transducing Hedgehog signaling and phenotypic observations in Evc or Evc2 mutant mice prompted us to ask to what extent does the loss of function of Evc/Evc2 reduce Hedgehog signaling , and why does Evc/Evc2 loss of function specifically impact limb development . To answer these questions , we investigated Hedgehog signaling and FGF signaling levels during defective endochondral ossification in Evc2 mutant mouse lines [27] . We demonstrate that a nonsense mutation in Evc2 that mimics mutations seen in EvC patients , leads to compromised but not abrogated Hedgehog signaling . In addition , we found that FGF signaling is significantly elevated in the Evc2 mutant growth plate due to increased expression of Fgf18 in the perichondrium . We successfully demonstrated that inactivation of one allele of Fgf18 in the Evc2 mutant mice partially rescued the dwarfism phenotype . We conclude that both reduced Hedgehog signaling and elevated FGF signaling play a critical role in the pathogenesis of the unique form of dwarfism that characterizes Evc2 mutants . Our findings explain differences between proposed functions of EVC/EVC2 based on biochemical approaches and symptoms found in the EvC patients , and thus provide insight for better options to treat dwarfism found in EvC patients . We generated Evc2 mutant mice by introducing a premature stop codon , along with an IRES-LacZ cassette , into exon12 ( equivalent to human exon 14 ) to mimic one of the nonsense mutations identified in human patients [32] . Compared with control littermates , homozygous mutant mice showed a decrease in body length as well as in appendicular bone length at 4 weeks of age ( S1A Fig ) . They did not show a difference in body length or body weight at birth ( S1B and S1C Fig ) , but they already had shorter limb bones [27] . Evc2 mutant mice were thus born with disproportionately short limbs . Staining of heterozygous mutant tibia growth plates for β-Gal activity indicated that Evc2 was expressed in chondrocytes as well as the perichondrium ( S1D Fig ) , which is consistent with RNA in situ hybridization results [4] . Histological analysis of humeral growth plates indicated shorter hypertrophic and proliferating chondrocyte zones , and fewer hypertrophic chondrocytes in Evc2 mutant embryos at E18 . 5 ( Fig 1A , 1E , 1F , 1G and 1H ) . Similar characteristics were observed at E16 . 5 ( Fig 1B , 1E , 1F , 1G and 1H ) , and E14 . 5 ( Fig 1C , 1E , 1F and 1G ) . On the other hand , histologic analysis at E12 . 5 , when chondrocytes were just starting to differentiate from condensed mesenchyme , showed no difference in the length of cartilage primordia ( Fig 1D , 1I and 1J ) . The same tendency was also observed in other limb skeletal elements , such as ulna , radius , femur and tibia [27] . These observations suggest that mutation of Evc2 leads to dwarfism by affecting growth plate chondrocyte proliferation and/or maturation , but not by affecting mesenchymal condensation or differentiation of condensed mesenchymal cells into chondrocytes . It has been reported that EVC2 is a ciliary protein [24 , 26] . To examine whether Evc2 mutation leads to loss of ciliary EVC2 , we visualized EVC2 protein in embryonic growth plates using an antibody recognizing the N-terminus of EVC2 . As expected , EVC2 was localized at the base of the cilia in control growth plates , but was undetectable in cilia of Evc2 mutant growth plates ( S2A and S2B Fig ) . Similarly , our previous work [27] indicated that Evc2 mutant primary chondrocytes do not have ciliary EVC2 or EVC . These findings strongly suggest that the truncation mutation at exon12 of Evc2 leads to abrogation of ciliary localization of both EVC2 and its interaction partner EVC . Hedgehog signaling , mediated by Indian Hedgehog in the growth plate , plays an important role in directing chondrocyte proliferation and hypertrophic maturation . Consistent with previous studies in Evc and Evc2 mutant mice [23 , 26] , we also detected decreased Hedgehog signaling in Evc2 mutant growth plates [27] . To evaluate the remaining Hedgehog signaling levels , we dissected out E16 . 5 tibia cartilage from control and Evc2 mutant growth plates for RNA isolation . qRT-PCR for Gli1 , Ptch1 and Pthrp ( Fig 2A ) , which are direct targets of Hedgehog signaling , indicated that Hedgehog signaling was significantly reduced , up to 40% of its normal level , in Evc2 mutants . These observations are consistent with our previous report on the Hedgehog signaling level using Gli1-lacZ reporter mice [27] . On the other hand , Ihh expression in Evc2 mutant growth plates remained at the control level ( Fig 2A ) . The results from in situ hybridization were consistent with expression analysis from qRT-PCR . Despite increased signal intensity of Ihh expression detected in Evc2 mutants , the expression area is more restricted in Evc2 mutant growth plate ( S3A Fig ) , which is consistent with previous reports [23 , 26] . To test whether this reduced Hedgehog signaling reflected impaired response of Evc2 mutant chondrocytes to the Hedgehog ligand , we isolated primary chondrocytes and examined their response to Hedgehog signaling induction by a smoothened agonist ( SAG ) . Similarly to what we observed in vivo , quantification of Gli1 mRNA levels indicated that Hedgehog signaling was reduced to 60% and 40% in SAG-treated limb and rib primary chondrocytes from Evc2 mutant mice , respectively , compared to SAG-treated control cells ( Fig 2B ) . The accumulation of GLI proteins in ciliary tips is regarded as a hallmark of Hedgehog signaling induction [18] . To analyze the ciliary localization of GLI2 , we co-stained E16 . 5 tibia growth plates for GLI2 with acetylated tubulin , a ciliary marker . In contrast with previous observations in cultured cells [21] , we found decreased GLI2 accumulation in ciliary tips in both resting and proliferating chondrocytes of Evc2 mutant growth plates ( Fig 2C , 2D and 2F ) , while the percentage of cilia with GLI2 staining remained the same in control and Evc2 mutants ( Fig 2E ) . Induction of Hedgehog signaling in primary chondrocytes also resulted in diminished accumulation of GLI2 ( S4A Fig ) , SUFU ( S4C Fig ) and KIF7 ( S4E Fig ) in ciliary tips , with no effect on ciliary accumulation of SMO ( S4G Fig ) . To further confirm that tissues/cells bearing Evc2 mutation can still respond to Hedgehog signaling to some extent , we treated Evc2 mutant tibiae with the Smoothened agonist , SAG , ex vivo and observed significant increases in tibia length compared with untreated Evc2 mutant tibiae ( Fig 2H ) . All aforementioned data thus concurred that the mutation of Evc2 leads to compromised but not abrogated Hedgehog signaling , likely due to impaired accumulation of Hedgehog components in ciliary tips . We sought to further dissect the mechanisms by which Evc2 loss of function specifically impacts limb development . The aforementioned decrease in length of the hypertrophic chondrocyte zones in Evc2 mutants is characteristic of Achondroplasia , the most common form of dwarfism in humans caused by gain of function mutations in FGF receptor 3 ( FGFR3 ) [33 , 34] To examine whether Evc2 mutant growth plates have altered FGF signaling , we first examined their phospho-ERK level by immunohistochemistry . In E16 . 5 Evc2 mutant tibiae , we detected elevated phospho-ERK in both resting and proliferating chondrocytes , but not in hypertrophic chondrocytes ( Fig 3A and 3B ) . Quantification indicated an increase of about 50% in Evc2 mutant growth plates compared to controls ( Fig 3E ) . Beside elevating ERK phosphorylation , FGF signaling is also known to slow down cell cycle progression through STAT1-mediated p21 expression [10 , 11] . Immunofluorescence for STAT1 demonstrated less plasma membrane-associated and more nuclear STAT1 in resting and proliferating chondrocytes in Evc2 radii than in controls ( Fig 3C , 3D and 3F ) . On the other hand , as a negative control , in the perichondrium , there was no nuclear STAT1 detected in Evc2 mutants , which is consistent with a previous report that only FGFR3 ( expressed in resting and proliferating chondrocytes ) but not FGFR2 ( expressed in perichondrium ) can induce STAT1 nuclear translocation [10] . To confirm the elevation of FGF signaling in Evc2 mutants , we examined the expression of FGF signaling targets in the growth plate . Spry2 , Spry3 , and Spry4 were all significantly overexpressed in Evc2 mutant growth plates ( Fig 4A ) . An increase of Spry3 expression was also detected in the embryonic limbs by in situ hybridization ( S5A Fig ) . It is known that in chondrocytes , STAT1 activation up-regulates p21 expression [10 , 11] . Consistently , we detected an elevated level of p21 mRNA in embryonic growth plate cartilage in Evc2 mutants ( Fig 4B ) . In conclusion , the mutation of Evc2 led to an elevation of FGF signaling that is likely to contribute to the dwarfism phenotype . To dissect the mechanism leading to elevated FGF signaling in Evc2 mutants , we first evaluated the expression levels of FGF receptors 1 , 2 and 3 in the growth plate . Consistent with our observation that elevated FGF signaling was only detected in resting and proliferating chondrocytes , we detected elevated expression for Fgfr3 ( Fig 4C ) , which is exclusively expressed in resting and proliferating chondrocytes [35] , but did not detect any expression change for Fgfr2 , which is specifically expressed in perichondrium [36] or Fgfr1 , primarily expressed in osteoblasts [36] . The increased Fgfr3 expression is also supported by the in situ hybridization of Fgfr3 in the embryonic limbs ( S5B Fig ) . It was previously observed that PTH/PTHrP treatment could repress Fgfr3 expression in a chondrocyte cell line in vitro [37] and in growth plate chondrocytes in vivo [15] . As a direct target of Hedgehog signaling in the growth plate , Pthrp expression was decreased in Evc2 mutants ( Fig 2A ) , an effect that may lead to derepression of Fgfr3 expression . To test this potential regulatory network , we evaluated the impact of PTH on the expression of Fgfr3 in wild type primary chondrocytes . PTH ( 1–34 ) binding to the PTH/PTHrP receptor can elicit downstream signaling as a substitute for PTHrP in chondrocytes [38] . Treatment of primary chondrocytes with PTH for 24 h led to a 75% decrease in Fgfr3 expression ( S6 Fig ) . These results thus suggest that elevation of Fgfr3 expression , occurring as a secondary effect of compromised Hedgehog signaling , may contribute to the increase in FGF signaling observed in Evc2 mutant growth plates . FGF18 , an FGF ligand documented to regulate endochondral ossification , is produced by the perichondrium surrounding the growth plates [8 , 9 , 39–41] . To assess Fgf18 expression in Evc2 mutants , we isolated perichondrium and detected a two-fold increase in Fgf18 mRNA level in Evc2 mutants compared to controls ( Fig 4F ) . On the other hand , we did not detect changes in the expression levels of Fgf9 or Fgf23 , FGFs that also regulate skeletal development [41 , 42] . Elevated Fgf18 expression in the perichondrium was corroborated by Fgf18 RNA in situ hybridization in tibia growth plates and immunohistochemistry of GFP in Evc2 mutant carrying an Fgf18GFP:CreEr allele ( Fig 4D and 4E ) . Taken together , our data suggest that both elevated Fgfr3 expression , as a secondary effect of compromised Hedgehog signaling , and elevated Fgf18 expression in the perichondrium contribute to elevating FGF signaling in Evc2 mutant growth plates . To examine if increased endogenous FGF signaling in Evc2 mutant growth plates is a potential cause of the dwarfism phenotype , we set up tibia ex vivo cultures and promoted bone growth by suppressing endogenous FGF signaling . Tibiae from control and Evc2 mutants grew at similar rates during a 7-day culture without FGF signaling inhibitor treatment ( 1 . 13±0 . 03 and 1 . 16±0 . 03 , Fig 5A ) . At a low concentration of U0126 ( 20 μM ) , an inhibitor of MEK ( activated by FGFR and activator of ERK ) , tibiae from control and mutant mice grew faster than untreated samples ( 1 . 38±0 . 03 and 1 . 34±0 . 03 , respectively , p<0 . 01 , Fig 5A ) , but there was no statistically significant difference between the two genotypes ( Fig 5A , # ) . In contrast , at a higher concentration of U0126 ( 40 μM ) , Evc2 mutant tibiae grew faster ( 1 . 39±0 . 02 ) than controls ( 1 . 31±0 . 02 ) relative to untreated samples ( Fig 5B , * , p<0 . 05 ) . Similar growth patterns were observed when SU5402 , a specific inhibitor for FGF receptor kinase activity , was applied ( 10 and 20 μM , Fig 5C and 5D ) . Histologic analysis indicated that in both control and Evc2 mutant tibiae ( Fig 5E ) , the U0126 and SU5402 treatments led to an increase in the length of growth plates and hypertrophic chondrocyte zones . These data thus suggest that higher levels of endogenous FGF signaling in Evc2 mutant growth plates contribute to the dwarfism in Evc2 mutants . Taken together , all results presented so far suggest that in addition to compromised Hedgehog signaling , elevated FGF signaling also contributes to the shortening of Evc2 mutant growth plates . In Evc2 mutants , compromised Hedgehog signaling is possibly due to impaired ciliary accumulation of GLI2; while higher FGF signaling is likely due to ( 1 ) elevated Fgfr3 expression in the proliferative zone as a result of compromised Hedgehog signaling , and ( 2 ) elevated Fgf18 expression in the perichondrium . Thus , in a chondrocyte-specific deletion of Evc2 , we would expect to exclude the impact of increased expression of Fgf18 in the perichondrium . Aggrecan enhancer-driven , tetracycline-inducible Cre ( ATC ) is a transgenic allele containing an Aggrecan gene enhancer with internal tetracycline regulatory elements that allows Tetracycline-dependent expression of Cre recombinase specifically in growth plate and other chondrocytes , but not in the perichondrium [43] ( S7B Fig ) . To specifically delete Evc2 in chondrocytes we generated ATC; Evc2 floxed mice . Examination of E18 . 5 bone showed a significant decrease in Gli1 and Pthrp expression in growth plate chondrocytes ( Fig 6G and 6K ) , indicating decreased Hedgehog signaling in these cells . At the same time , we detected no change in Spry3 expression ( readout for FGF signaling ) in the growth plate ( Fig 6I ) and no change in Fgf18 expression in the perichondrium ( Fig 6J ) , indicating that neither Fgf18 expression nor FGF signaling was altered , despite elevated Fgfr3 expression ( Fig 6H ) . In situ results for Gli1 and Spry3 also support the notions of compromised Hedgehog signaling while no difference for FGF signaling in tibia at E18 . 5 ( S8 Fig ) . These Evc2 conditional mutants showed only a moderate decrease in tibia length ( Fig 6D ) compared to Evc2 germ line mutants ( Fig 6A and 6B ) . At E18 . 5 , these mutants only displayed a 10% decrease in the total length of tibiae ( Fig 6C and 6D ) , 20% decrease in the length of the hypertrophic chondrocyte zone ( Fig 6C and 6D ) , 20% decrease in the number of hypertrophic chondrocytes ( Fig 6C and 6D ) , and a 13% decrease in the length of the proliferating chondrocyte zone ( Fig 6C and 6D ) , but no difference in the total length of the growth plate ( Fig 6C and 6D ) . To further demonstrate that Evc2 loss of function in the perichondrium is critically involved in the elevation of FGF signaling and pathogenesis of dwarfism , we used the Dermo1 ( Twist2 ) Cre allele [44] that demonstrates recombination in both chondrocytes and the perichondrium ( S7A Fig ) . In Evc2 floxed mice carrying Dermo1Cre , we detected a significant decrease in Gli1 and Pthrp expression ( Fig 6L and 6P ) and an increase in Fgfr3 expression ( Fig 6M ) in the growth plate , as expected . At the same time , we detected increased Spry3 expression in the growth plate ( Fig 6N ) and increased Fgf18 expression in the perichondrium ( Fig 6O ) , indicating elevated FGF signaling . We also detected a dramatic decrease of tibia length ( Fig 6E and 6F ) , which was similar to that observed in Evc2 germ-line mutants ( Fig 6A and 6B ) . More specifically , at E18 . 5 , we detected a 21% decrease in the overall length of tibiae ( Fig 6E and 6F ) , 52% decrease in the length of the hypertrophic chondrocyte zone ( Fig 6E and 6F ) , 44% decrease in the number of hypertrophic chondrocytes ( Fig 6E and 6F ) , 36% decrease in the length of the proliferating chondrocyte zone ( Fig 6E and 6F ) , and a 21% decrease in the total length of the growth plate ( Fig 6E and 6F ) . Taken together , these data demonstrate that compromised Hedgehog signaling mediated by Evc2 mutation only partially contributes to the dwarfism in Evc2 mutants; additionally , elevated FGF signaling , mediated by Evc2 mutation in the perichondrium , plays a critical role in the pathogenesis of dwarfism in Evc2 mutants . To further demonstrate that elevated FGF signaling plays critical roles during the pathogenesis of dwarfism in Evc2 mutant mice , we inactivated one allele of Fgf18 in Evc2 mutant mice ( Evc2 ex12/ex12; Fgf18 LacZ/+ ) . Compared to Evc2 ex12/ex12 mutants , removal of one copy of Fgf18 allele partially rescued the dwarfism in the Evc2 ex12/ex12; Fgf18 LacZ/+mutant ( Fig 7A ) . More specifically , the length of Evc2 mutant tibia is 75% of control , while the length of Evc2 ex12/ex12; Fgf18 LacZ/+ tibia is about 82% of control ( Fig 7B , n = 6 , p<0 . 05 ) . Similarly , the length of the growth plate , length of the hypertrophic zone , number of hypertrophic chondrocytes and the length of the proliferating zones observed in Evc2 ex12/ex12 mutant tibia are all partially rescued in Evc2 ex12/ex12; Fgf18LacZ/+ embryos . We did not see overt differences between wild type and Fgf18 heterozygous mutants as previously reported [40] . These results demonstrate that elevated FGF signaling mediated by elevated Fgf18 expression in the perichondrium critically contributes to the pathogenesis of dwarfism in Evc2 mutant mice . In this work , Evc2 mutant mice were generated that mimic one of nonsense mutations identified in EvC human patients . Although the mutation is different from that of previously reported Evc2 mutant mouse line [23] , in which Evc2 was deleted from exon 1 , our strategy too resulted in abrogation of EVC2 protein in cilia . Using an antibody recognizing the N-terminus of EVC2 , we could not detect any ciliary EVC2 ( S2A and S2B Fig ) [27] , and as a result , EVC also lost its ciliary localization [27] . This observation is consistent with previous studies that deletion of 83 amino acids or more in the C terminus of EVC2 leads to complete loss of ciliary localization [21 , 23] . Thus , in our Evc2 mutant , the ciliary function of the EVC/EVC2 complex is completely abolished , just as it is when Evc2 is deleted from exon1 . Consistently , the length of each appendicular bone is reduced in our Evc2 mutant mice as much as in the previously reported mutants [23] . Previous in vitro biochemical and molecular biological studies indicated that the EVC/EVC2 complex is essential during Hedgehog signaling [21 , 23] . However , congenital defects due to abnormally diminished Hedgehog signaling , such as neural tube and midline defects , are not present in EvC patients or in Evc/Evc2 mutant mice . On the other hand , our in vivo and in vitro studies suggest that an inactivating mutation of Evc2 leads to compromised but not abrogated Hedgehog signaling . Therefore , we favor the idea that compromised but not abrogated Hedgehog signaling in Evc2 mutant embryos leads to limited aspects of developmental abnormalities in the processes that require Hedgehog signaling . Similarly , although decreased Hedgehog signaling in the limb bud was detected using a Ptch1-lacZ reporter , there was no digit pattern defect in Evc2 mutant mice [23] . During limb development , loss of function of Indian Hedgehog leads to extremely short limbs and distorted growth plates [13] , which are apparently more severe than the phenotypes found in Evc [26] and Evc2 mutant mice ( [23] and this work ) . Chondrocyte-specific postnatal knockout of Indian hedgehog leads to early closure of growth plates , as a result of diminished Pthrp expression from as early as postnatal day 15 [31] . Similarly , diminished Hedgehog signaling caused by ablation of primary cilia specifically in cartilage also leads to early closure of growth plate by P15 [45] , which is observed in neither Evc nor Evc2 mutant mice for up to 6 weeks [26] . All these facts strongly suggest that in Evc and Evc2 mutant mice there is still a substantial level of Hedgehog signaling remaining . This notion is supported by previous observations that there is still substantial level of Gli1 expression detectable in Evc or Evc2 mutant growth plates [23 , 26] . These phenotypic observations coincide with our finding that Evc2 mutant primary chondrocytes from embryonic limbs and ribs still retain 40% to 60% of wild type levels of Hedgehog signaling ( Fig 2B ) . Since FGF signaling was shown to regulate ciliary length [46] , no studies have yet addressed how primary cilia regulate cellular responses to FGF ligands . In primary chondrocytes and MEFs , we did not detect differential responses of control and Evc2 mutant cells to FGF ligands . On the other hand , we detected both elevated Fgfr3 expression ( Fig 4C ) in the growth plate as well as elevated Fgf18 expression in the perichondrium of Evc2 mutants ( Fig 4D , 4E and 4F ) . Previous reports [15 , 37] and our current work suggest that PTHrP negatively regulates Fgfr3 expression ( S6 Fig ) . Therefore , elevated Fgfr3 expression in the growth plate is possibly due to decreased Pthrp expression caused by compromised Hedgehog signaling . However , elevated Fgfr3 expression itself appears insufficient to elevate FGF signaling , since the chondrocyte-specific deletion of Evc2 that we achieved using ATC resulted in compromised Hedgehog signaling and elevated Fgfr3 expression ( Fig 6H ) , but neither an elevation of Fgf18 expression in the perichondrium ( Fig 6J ) nor in an increase in FGF signaling ( Fig 6I ) . Therefore , elevated Fgf18 expression in the perichondrium likely plays a major role in upregulating FGF signaling in the Evc2 mutant growth plate . It is still possible that elevated Fgfr3 has an additive impact in the presence of elevated Fgf18 expression on the final outcome of FGF signaling . Our work also indicated that elevated expression of Fgf18 in the perichondrium was a consequence of the Evc2 mutation in the perichondrium , since elevation of Fgf18 expression was dependent upon Evc2 deletion in these cells ( Fig 6J and 6O ) . The molecular mechanism of how loss of Evc2 in the perichondrium leads to elevated Fgf18 expression is currently under investigation . ATC-dependent Evc2 conditional mutants allowed us to exclude the impact of FGF signaling and thereby to evaluate how a compromised Hedgehog-PTHrP feedback loop results in dwarfism . Despite a mild decrease in the length of their tibiae ( Fig 6C and 6D ) , these mutants had growth plate defects quite different from those of Evc2 germ-line mutants ( Fig 6A and 6B ) . These results clearly demonstrate that the affected Hedgehog-PTHrP feedback loop mediated by Evc2 mutation in chondrocytes is not sufficient to impact the length of proliferating chondrocyte zone , hypertrophic chondrocyte zone and growth plate as severely as observed in Evc2 germ-line mutants ( Fig 6A and 6B ) . In support of this idea , Dermo1Cre-dependent Evc2 conditional mutants exhibited a more dramatic decrease in tibia length ( Fig 6E and 6F ) and their growth plate phenotype was similar to that of Evc2 germ-line mutants . Therefore , both elevated FGF signaling and a compromised Hedgehog-PTHrP feedback loop likely contribute critically to the pathogenesis of dwarfism in Evc2 mutants . Consistent with our findings , decreased proliferating and hypertrophic chondrocyte zones were also observed in Evc mutant mice [26] , suggesting a shared mechanism leading to dwarfism in Evc and Evc2 mutant mice . In conclusion , the results obtained from in vitro and in vivo suggest a model wherein Evc2 inactivating mutations as well as EvC syndrome causing mutations partially compromise but do not abrogate Hedgehog signaling . The resulting compromised Hedgehog-PTHrP feedback loop only partially contributes to the dwarfism . In addition , our new findings have exposed a novel regulatory mechanism in which Evc/Evc2 inactivating mutations in the perichondrium leads to a significant elevation of FGF signaling . Both effects , i . e . , compromised Hedgehog-PTHrP feedback loop and elevated FGF signaling , likely synergize to render the severe dwarfism that characterizes the EvC syndrome . This model thus also suggests that the available therapeutic solutions being tested for Achondroplasia could be used to relieve , at least partially , the severity of dwarfism in EvC patients . The generation of Evc2 mutant mice and Evc2 floxed mice was reported elsewhere [27] . Fgf18LacZ mutant mice were reported previously [40] . Fgf18GFP:CreER mice contain a splice acceptor ( SA ) GFP:CreERT2 insertion into the first intron of Fgf18 ( D . M . O . , I . H . H . unpublished ) . To obtain Evc2 homozygous mutant embryos , timed mating between two heterozygous Evc2 germ-line knockout mice was carried out . Noon of the date when the vaginal plug was observed was designated embryonic day 0 . 5 ( E0 . 5 ) . Evc2 floxed mice were bred with mice carrying doxycycline inducible Aggrecan enhancer-driven , tetracycline-inducible Cre ( ATC ) [43] or Dermo1Cre [44] mice to generate conditional deletions of Evc2 . For chondrocyte-specific deletion mediated by ATC , doxycycline-supplemented chow diet ( Harlan TD01306 ) was provided to pregnant females from E9 . 5 . All mouse experiments were performed in accordance with University of Michigan guidelines and federal laws covering the humane care and use of animals in research . All animal procedures used in this study were approved by the Institutional Animal Care and Use Committee ( IACUC ) at the University of Michigan ( Protocol #PRO00005716 ) . Limbs were dissected out from embryos , fixed in 4% paraformaldehyde ( PFA ) , embedded in paraffin , sectioned and stained with hematoxylin and eosin ( H&E ) according to standard procedures . For skeletal staining , dissected limbs were skinned and stained with alcian blue and alizarin red according to [47 , 48] . For immunohistochemistry , limbs were fixed in 4% PFA overnight at 4°C and cryo-protected in 30% sucrose in PBS solution before embedding in OCT . Specimens were cut into 10-μm sections and incubated overnight at 4°C with antibody against EVC2 ( Y20 , 1:50 , Santa Cruz Biotechnology , Santa Cruz , CA , USA ) , GLI2 ( H300 , 1:50 , Santa Cruz Biotechnology , Santa Cruz , CA , USA ) , SMO ( N19 , 1:50 , Santa Cruz Biotechnology , Santa Cruz , CA , USA ) , SUFU ( H300 , 1:50 , Santa Cruz Biotechnology , Santa Cruz , CA , USA ) , EVC ( HPA008703 , 1:50 , Sigma , St . Louis , MO , USA ) , KIF7 ( Ab95884 , 1:100 , Cambridge , MA , USA ) , acetylated tubulin ( T6793 , 1:1000 , Sigma , St . Louis , MO , USA ) , gamma tubulin ( T5326 , 1:1000 , Sigma , St . Louis , MO ) , pERK ( 4695 , 1:50 , Cell signaling , Danvers , MA 01923 ) , or STAT1 ( Ab3987 , 1:100 , Cambridge , MA , USA ) . Sections were then incubated with corresponding Alexa Fluor 488 conjugated secondary antibody for 1 h at room temperature before mounting with ProLong Gold Anti-fade Reagent with DAPI ( P36935 , Life Technologies , Grand Island , NY , USA ) . All fluorescence images were acquired at room temperature by confocal microscopy ( Nikon C1 ) through Nikon EZ-C1 3 . 91 and processed by Adobe Photoshop CS6 . For primary chondrocyte isolation , rib or long bone cartilage was dissected from E18 . 5 embryos and digested with collagenase A ( Roche , Indianapolis , IN , USA ) . Chondrocytes released will be subsequently cultured in DMEM ( Life Technology , Grand Island , NY , USA ) with 10% FBS ( Atlanta Biologicals , Flowery Branch , GA , USA ) . Experiment will be carried out using cells within 5 passages . For immunocytochemistry , cultured primary chondrocytes were starved in 0 . 5% serum for 36 h before treatment with 100 nmol of SAG ( Chemicon , Billerica , MA , USA ) for 4 h . Cells were then fixed in 4% PFA and permeabilized in PBS with 0 . 1% Triton X-100 ( Sigma , St . Louis , MO , USA ) before incubation with primary antibody at 4°C for overnight and with fluorescent secondary antibody . Mounting was done with ProLong Gold Anti-fade Reagent containing DAPI . RNA was isolated from primary chondrocytes using TRIzol ( Life Technologies , Grand Island , NY , USA ) according to manufacturer’s instructions . For RNA isolation from embryonic growth plates , long bones were dissected out at E16 . 5 . Growth plates were collected from tibiae and placed into TRIzol for homogenization according to manufacturer’s instructions . For isolation of perichondrium cells , growth plates were dissected out from embryonic tibiae and digested with 1 unit /ml Dispase [48] . For reverse transcription , 1 μg of total RNA was reverse-transcribed using SuperScript Reverse Transcriptase ( Life Technologies , Grand Island , NY , USA ) . Quantitative real-time PCR was performed using Applied Biosystems ViiA7 , with the following taqman probes: Mm00494645_m1 for Gli1 , Mm99999915_g1 for Gapdh , Mm00439612_m1 for Ihh , Mm00436026_m1 for Ptch1 , Mm00436057_m1 for Pthrp , Mm00433294_m1 for Fgfr3 , Mm00438941_m1 for Fgfr2 , Mm00438932_m1 for Fgfr1 , Mm00432448_m1 for Cdkn1a ( P21 ) , and Mm00433286_m1 for Fgf18 . Tibiae were dissected out from hindlimbs of E16 . 5 embryos and placed into 24-well plates with 1 ml medium ( alpha MEM ( Life Technologies , Grand Island , NY , USA ) , 0 . 5% FBS , penicillin and streptomycin ( Life Technologies , Grand Island , NY , USA ) . Media were changed every other day , and the full length of tibia was measured at day 1 ( D1 ) and D7 . The ratios of the lengths at D7 over the lengths at D1 was calculated and compared between controls and mutants . RNA in situ hybridization was carried out as previously described [49] using a digoxygenin-labeled Fgf18 probe [39 , 40] . Briefly , embryonic tissues were immediately fixed in 4% PFA , before cryo-protected in 30% sucrose in PBS . Then , 20 μm sections were treated with proteinase K , post-fixed with 4% PFA , before treated with acetic anhydride solution ( Sigma , St . Louis , MO , USA ) . Sectioned tissues were hybridized with RNA probe in hybridization solution containing 5X SSC , 50% formamide , 1mg/mg tRNA ( Sigma , St . Louis , MO , USA ) , 0 . 1mg/ml Heparin ( Sigma , St . Louis , MO , USA ) at 65°C . Sectioned tissues were then digested with RNase A ( Roche ) and washed in post-hybridization washing solution containing 0 . 2X SSC , before incubation with alkaline phosphatase conjugated mouse anti-digoxygenin for overnight . Purple color for positive signal was developed through incubation sections with BM Purple for AP substrate precipitating ( Roche ) .	Ellis-van Creveld ( EvC ) syndrome is a congenital skeleton disorder characterized by short limbs . Recent studies indicated that EVC and EVC2 , the proteins encoded by two causative genes of EvC syndrome , play important function in transducing Hedgehog signaling , a signaling pathway critical for embryonic development . The defective Hedgehog signaling in chondrocytes is therefore the speculated reason for dwarfism in EvC patients . However , despite the apparent skeletal abnormalities observed in EvC patients , other tissues that require Hedgehog signaling are relatively normal . To understand how skeletal development is specifically affected in EvC syndrome , we analyze the limb skeletogenesis using Evc2 mutant mice . Our data demonstrated that mutation in Evc2 only moderately affected Hedgehog-PTHrP feedback loop in the growth plate , which only partially contributes to the dwarfism . Additionally , the elevated Fibroblast Growth Factor ( FGF ) signaling , another signal important for embryonic development , critically contributes to the pathogenesis of limb dwarfism . We identified that loss of EVC2 function in the perichondrium , the tissue surrounding growth plate chondrocytes , is critical to develop the dwarfism in the Evc2 mutant mice . Overall , our data uncover a novel pathogenic mechanism to understand limb dwarfism in EvC patients and suggest that therapies for Achondroplasia caused by elevated FGF signaling may be applicable for relieving dwarfism found in EvC syndrome .	[ "Abstract", "Introduction", "Results", "Discussion", "Materials", "and", "Methods" ]	[ "growth", "plate", "tibia", "medicine", "and", "health", "sciences", "chondrocytes", "dwarfism", "achondroplasia", "developmental", "biology", "bone", "connective", "tissue", "cells", "skeleton", "cartilage", "morphogenesis", "cellular", "structures", "and", "organelles",...	2016	Elevated Fibroblast Growth Factor Signaling Is Critical for the Pathogenesis of the Dwarfism in Evc2/Limbin Mutant Mice
Drug-drug interactions ( DDIs ) are a common cause of adverse drug events . In this paper , we combined a literature discovery approach with analysis of a large electronic medical record database method to predict and evaluate novel DDIs . We predicted an initial set of 13197 potential DDIs based on substrates and inhibitors of cytochrome P450 ( CYP ) metabolism enzymes identified from published in vitro pharmacology experiments . Using a clinical repository of over 800 , 000 patients , we narrowed this theoretical set of DDIs to 3670 drug pairs actually taken by patients . Finally , we sought to identify novel combinations that synergistically increased the risk of myopathy . Five pairs were identified with their p-values less than 1E-06: loratadine and simvastatin ( relative risk or RR = 1 . 69 ) ; loratadine and alprazolam ( RR = 1 . 86 ) ; loratadine and duloxetine ( RR = 1 . 94 ) ; loratadine and ropinirole ( RR = 3 . 21 ) ; and promethazine and tegaserod ( RR = 3 . 00 ) . When taken together , each drug pair showed a significantly increased risk of myopathy when compared to the expected additive myopathy risk from taking either of the drugs alone . Based on additional literature data on in vitro drug metabolism and inhibition potency , loratadine and simvastatin and tegaserod and promethazine were predicted to have a strong DDI through the CYP3A4 and CYP2D6 enzymes , respectively . This new translational biomedical informatics approach supports not only detection of new clinically significant DDI signals , but also evaluation of their potential molecular mechanisms . Drug-drug interactions ( DDIs ) are a major cause of morbidity and mortality and lead to increased health care costs [1]–[3] . DDIs are responsible for nearly 3% of all hospital admissions [4] and 4 . 8% of admissions in the elderly [1] . And with new drugs entering the market at a rapid pace ( 35 novel drugs approved by the FDA in 2011 ) , identification of new clinically significant drug interactions is essential . DDIs are also a common cause of medical errors , representing 3% to 5% of all inpatient medication errors [5] . These numbers may actually underestimate the true public health burden of drug interactions as they reflect only well-established DDIs . Several methodological approaches are currently used to identify and characterize new DDIs . In vitro pharmacology experiments use intact cells ( e . g . hepatocytes ) , microsomal protein fractions , or recombinant systems to investigate drug interaction mechanisms . The FDA provides comprehensive recommendations for in vitro study designs , including recommended probe substrates and inhibitors for various metabolism enzymes and transporters [6] . The drug interaction mechanisms and parameters obtained from these in vitro experiments can be extrapolated to predict in vivo changes in drug exposure . For example , a physiologically based pharmacokinetics model was developed to predict the clinical effect of mechanism based inhibition of CYP3A by clarithromycin from in vitro data [7] . However , in vitro experiments alone often cannot determine whether a given drug interaction will affect drug efficacy or lead to a clinically significant adverse drug reaction ( ADR ) . In vivo clinical pharmacology studies utilize either randomized or cross-over designs to evaluate the effect on an interaction on drug exposure . Drug exposure change serves as a biomarker for the direct DDI effect , though drug exposure change may or may not lead to clinically significant change in efficacy or ADRs . The FDA provides well-documented guidance for conducting in vivo clinical pharmacology DDI studies [6] . If well-established probe substrates and inhibitors are used , involvement of specific drug metabolism or transport pathway can be demonstrated by in vivo clinical studies . For example , using selective probe substrates of OATPs ( pravastatin ) and CYP3A ( midazolam ) and probe inhibitors of OATPs ( rifampicin ) and CYP3A ( itraconazole ) , it was shown that hepatic uptake via OATPs made the dominant contribution to the hepatic clearance of atorvastatin in an in vivo clinical PK study [8] . However , due to overlap in substrate selectivity , an in vivo DDI study alone will often not provide mechanistic insight into the DDI . Finally , in populo pharmacoepidemiology studies use a population-based approach to investigate the effect of a DDI on drug efficacy and ADRs . For example , the interactions between warfarin and several antibiotics were evaluated for increased risk of gastrointestinal bleeding and hospitalization in a series of case-control and case-crossover studies using US Medicaid data [9] . Indeed , epidemiological studies using large clinical datasets can identify potentially interacting drugs within a population , but these studies alone are insufficient to characterize pharmacologic mechanisms or patient-level physiologic effects . The aforementioned in vitro , in vivo , and in populo research methods are complementary in characterizing new drug-drug interactions . Yet these methods are all limited by their relatively small scale . Such studies usually focus on a few drug pairs for one or a limited number of metabolizing enzymes or transporters a time . Performing large scale screening for novel drug interactions requires higher throughput strategies . Literature mining and data mining have become powerful tools for knowledge discovery in biomedical informatics , and are particularly useful for hypothesis generation . A recent notable example in clinical pharmacology is the successful detection of novel DDIs through mining of the FDA's Adverse Event Reporting System [10] . In this study , pravastatin and paroxetine were found to have a synergistic effect on increasing blood glucose . This finding was validated in three large electronic medical record ( EMR ) databases . While a ground-breaking success , this approach provides little evidence regarding the mechanism of the interaction . In this paper , we present a novel approach using literature mining for screening of potential DDIs based on mechanistic properties , followed by EMR-based validation to identify those interactions that are clinically significant . We focus on clinically and statistically significant DDIs that increase the risk of myopathy . Our initial drug dictionary consisted of 6937 drugs . Of these , 1492 drugs were validated as FDA approved drugs ( Figure 1 ) . Among these 1492 drugs , our text mining approach identified 232 drugs , as either CYP substrates or inhibitors ( Table S1 ) . Recall rate ( i . e . the proportion of true positives identified by the text mining method among all the true positives ) and accuracy ( i . e . the proportion of true positives among the text mined results ) were used to evaluate the text mining performance . The recall rate of this text mining analysis was 0 . 97 , with the information retrieval ( IR ) step being rate-limiting . In the information extraction ( IE ) step , the two initial curators agreed on 78% of cases . The third curator was able to establish DDI relevance and extract information in the 22% of cases which were in disagreement . The third curator also confirmed 100% accuracy among 20% of randomly chosen abstracts that the first two curators had agreed upon . Therefore , the accuracy of our text mining analysis reached 100% . These drugs' metabolism and inhibition enzymes were experimentally determined by probe substrates and inhibitors recommended by the FDA Drug-Drug Interaction guidelines . Their categorizations are reported in Table S1 . Out of the 149 CYP substrates identified , 102 ( 68% ) , were substrates of CYP3A4/5 . This was consistent with the literature that about half of the drugs on the market which undergo metabolism are metabolized by CYP3A [11] . A total of 59 drugs were found to undergo metabolism by more than one CYP enzyme . We also identified 123 CYP inhibitors , with CYP3A4/5 , CYP2D6 , CYP2C9 , CYP1A2 , and CYP2C19 having comparable numbers of inhibitors , ( 48 , 39 , 39 , 39 , 31 respectively ) . Fewer inhibitors were identified for other enzymes . Fifty inhibitors were found to inhibit more than one enzyme . Among 232 drugs with known metabolism and/or inhibition enzyme information ( Figure 1 ) , 13 , 197 drug interaction pairs were predicted based on their pertinent CYP enzymes ( Figure 2 ) . Among these 13 , 197 predicted DDIs , 3670 DDI pairs were prescribed as co-medications in actual patients within the Common Data Model ( CDM ) dataset . In other words , these 3670 predicted DDI pairs may have potential real-world clinical implication . Among those 3670 predicted DDI pairs from in vitro studies , text mining identified 196 pairs with published clinical drug-drug interaction study results . These in vivo studies tested whether a substrate drug's exposure ( i . e . systemic drug concentration ) was increased when co-administrating with an inhibitor . The recall rate of this text mining analysis was 0 . 94 . The accuracy of this text mining analysis reached 100% , after manual IE from two curators and validation from the third . Among these 196 in vivo validated DDI pairs , 123 of them were found to have significant DDIs ( Figure 2 ) , i . e . drug exposure increased significantly ( P<0 . 05 ) , and it increased by more than 2 fold . The additional 73 pairs were considered not to be clinically significant DDI's . In our CDM dataset , there were medication records on 817 , 059 patients . Among these patients , 59 , 572 ( 7 . 2% ) experienced myopathy events ( Table 1 ) . Two major subcategories of myopathy: myalgia and myositis/muscle weakness accounted for more than 95% of the cases . There were 53 rhabdomyolysis cases . In the cohort of individuals suffering a myopathy event , the average age was 40 . 2 year ( SD = 23 years ) ; 59 . 1% were female , and the average medication frequency was 3 . 8 ( SD = 2 . 5 ) . However , 65 . 8% of the race data were missing . In our initial data analysis , we found that females had higher myopathy risk than males ( 8 . 6% vs 5 . 4% , p<2e-16 , Table 2 ) ; and each one year increase in age was associated with 0 . 15% higher myopathy risk ( p<2e-16 ) . These results were consistent with the literature [12] . The 3670 DDI pairs identified in the CDM database were tested using the additive model , i . e . whether an inhibitor would increase the myopathy risk of the substrate compared to the substrate alone . Both age and sex were justified in the logistic regression . The p-value threshold was chosen as 0 . 05/3670 = 0 . 0000136 after Bonferroni justification , with OR greater than 1 . There were 124 and 287 significant DDI pairs for CYP2D6 and CYP3A4/5 enzymes , respectively ( Figure 3 and Table S2 ) . The other enzymes had fewer significant DDI pairs . Pathway enrichment analysis suggested similar results , i . e . CYP2D6 and CYP3A4/5 enzymes had more significant DDI pairs than the other enzymes , p = 8E-8 and 4E-2 respectively . Although this DDI analysis was confounded by the other co-medication variables , it was indeed a global description of DDI effects from various CYP enzymes . This global analysis provided us a picture of the metabolism enzymes that were most important in understanding the increased myopathy risk associated with DDIs . In order to remove the effect of myopathy risk of the inhibitor itself , a synergistic DDI test was conducted to determine whether substrate and inhibitor together have higher risk than the combined additive risk when the substrate or inhibitor is taken alone . Both age and sex were justified as covariates . DDI pairs were removed if either one of the drugs was prescribed to treat symptoms of myopathy . We set the significance threshold as p = 0 . 0000136 , as justified the multiple primary hypotheses on 3670 predicted DDI pairs . Table 3 presents the five significant synergistic DDI pairs: ( loratadine , simvastatin ) , ( loratadine , alprazolam ) , ( loratadine , duloxetine ) , ( loratadine , ropinirole ) , and ( promethazine , tegaserod ) . Their relative risks were ( 1 . 69 , 1 . 86 , 1 . 94 , 3 . 21 , 3 . 00 ) respectively , the p-values were ( 2 . 03E-07 , 2 . 44E-08 , 5 . 60E-07 , 2 . 60E-07 , 2 . 60E-07 , 8 . 22E-07 ) respectively , and their associated enzymes were primarily CYP3A4/5 and CYP2D6 . Additional analyses of myopathy were performed for these five DDI pairs . In the first myopathy analysis , the total number of medications ordered during the drug exposure window was added as a covariate in the logistic regression . This variable was used as a surrogate marker for the comorbidities of a patient . The average number of medications used by individuals during the drug exposure window was 3 . 6 with SD = 2 . 4 . Table 4 presents the five DDI effects on myopathy after adjusting for the total number of medications . Compared to table 3 , all the single drug myopathy risks and drug combination risks were reduced after justifying for the number of co-medications . The DDI evidence became even more significant ( p-values less than 3e-12 ) , and risk ratios became even bigger , between 2 . 72 and 7 . 00 . The medication frequency itself was also associated with increased myopthay risk . The addition of one co-medication was associated with an increased myopathy risk between 0 . 6% and 0 . 9% in testing the 5 DDI pairs . All p-values are less than 2e-16 . In the second myopathy analysis , only the first myopathy events were considered , because co-medications administered after the first myopathy event but before the follow-up myopathy events were potential confounders . In other words , it was difficult to justify whether the co-medication drug exposure resulted from the myopathy or caused myopathy . Table S3 presents the data analysis for the DDI pairs: ( loratadine , simvastatin ) , ( loratadine , alprazolam ) , ( loratadine , ropinirole ) , ( loratadine , duloxetine ) , and ( promethazine , tegaserod ) . Their relative risks are ( 1 . 34 , 1 . 38 , 1 . 38 , 1 . 81 , 1 . 70 ) respectively , the p-values are ( 3 . 20E-03 , 2 . 1E-05 , 9 . 4E-04 , 3 . 1E-03 , 2 . 3E-03 ) respectively . This analysis based on first myopathy event with these five selected DDI pairs confirmed the trend of our previous synergistic DDI analysis . Unlike DDI signal detection from AERS by Dr . Altman's group [10] , we enriched our EMR signal detection by focusing on CYP-mediated DDIs that were mined and predicted from PubMed abstracts . There are multiple recent publications on drug interaction text mining . Two automatic literature mining systems were developed to predict drug interactions based on their associated metabolism enzymes [13] , [14] . An evidential approach was developed to differentiate in vitro and in vivo DDI studies , curate drug metabolism and inhibition enzymes , and predict DDIs based on their pertinent enzymes [15] . Our text mining approach took advantage of these two methods , i . e . metabolism based DDI prediction; and emphasized the text mining performance more stringently . The IR step of our method is an automatic algorithm , which has high recall rate ( 0 . 97 ) ; while the IE step is a manual curation step , with high precision ( 100% ) . In addition , we implemented CYP enzyme probe substrates and inhibitors from the FDA guidance into the literature mining method . This strategy supplies information on the potential mechanism for the predicted DDIs . Our current text mining method focuses on pharmacokinetic-based drug interaction literature , i . e . reported substrate drug exposure changed by drug interaction . Text mining which focuses on pharmacodynamics ( PD ) DDI literature has been recently discussed [16] , [17] . PD DDI literature reports the drug efficacy or side-effect changes , but it usually does not report drug exposure change . Among the 13197 predicted DDIs from in vitro PK study literature mining , 3670 of them may have clinical relevance , i . e . they were taken as co-medications by at least some of the 2 . 2 million patients in our clinical dataset . However , only 196 of them ( 5 . 3% ) have been tested in clinical pharmacokinetic DDI trials . Among these 196 clinically tested DDIs , 123 of them ( 62 . 7% ) showed significant substrate drug exposure increase when co-administrated with the inhibitor . This striking finding calls for further evaluation of those predicted DDIs that have not been subjected to rigorous study . As a matter of fact , all five DDI pairs which showed an increased myopathy risk in our pharmaco-epidemiology study lack clinical pharmacokinetic studies . The FDA labels of all 7 of the drugs which comprise the five significant DDI pairs report myopathy related side effects ( Table S4 ) . This evidence confirms the myopathy risk for each individual drug . In order to understand the mechanisms of each interaction , we further explored literature regarding those agents . In Figure 4 and Table S5 , we integrated information on the metabolism and inhibition enzymes of those 7 drugs from a full-text based literature review of reported in vitro studies of the drugs . Table 5 presented the DDI potency prediction for the five DDI pairs . Loratadine ( substrate ) and simvastatin ( inhibitor ) were predicted to have a strong DDI through the CYP3A4/5 enzyme . Tegaserod ( substrate and inhibitor ) and promethazine ( substrate and inhibitor ) were predicted to have strong DDI through the CYP2D6 enzyme . Their interactions are mixed inhibition and auto-inhibition . The other four drug pairs were predicted to have moderate DDIs: loratadine ( inhibitor ) and omeprazole ( substrate ) interact through both the CYP2C19 and CYP3A4/5 enzymes; loratadine ( inhibitor ) and alprazolam ( substrate ) interact through CYP3A4/5; loratadine ( substrate ) and duloxetine ( inhibitor ) interact through the CYP2D6 enzyme; and loratadine ( inhibitor ) and ropinirole ( substrate ) interaction is through CYP3A4/5 . Two DDI data analysis strategies were implemented to identify drug-drug interactions associated with an increased risk for myopathy . The first approach employed an additive model coupled with a CYP metabolism pathway enrichment analysis . This strategy stems from the newly formed discovery nature of bioinformatics research , i . e . to search for commonality among many hypothesis tests . The second strategy employed a synergistic model coupled with extensive confounder justification . This strategy follows the more stringent pharmaco-epidemiology considerations , which heavily controls for false positives . Unlike the additive model , the synergistic model can justify the myopathic risk effect from an inhibitor in the presence of other potential confounders . Therefore , the additive model would potentially identify more false positive DDIs . However , the additive model is more powerful than the synergistic model in identifying the true positive DDIs . Many more DDIs were identified by the additive model based DDI analysis than by the synergistic strategy . Because pathway enrichment analysis allows more flexibility toward false positive DDIs , the additive model identified CYP3A4/5 and CYP2D6 enzymes as they have the enriched DDI pairs . Although the synergistic model DDI analysis only inferred five significant DDI pairs , upon additional literature review , it was found that these pairs also showed mechanistic involvement of CYP2D6 and CYP3A4/5 enzymes . The consistency of the mechanistic interpretations of the two separate DDI analysis strategies delivers an encouraging message: the bioinformatics approach and the pharamco-epidemiology approach are complementary and mutually supportive . Our synergistic DDI test is a very stringent approach , compared to the additive approach used by the other investigators [9] , [18] , [19] . We recognize that our synergistic DDI test may exclude some true DDIs . It assumes that all myopathy is the result of drug administration , and patients who don't take the DDI drugs won't have myopathy . However , there is a background rate of myopathy in patients that is not due to either of the two drugs in a specific DDI . If the patients who don't take drugs have a baseline risk of myopathy , the relative risk estimated through our synergistic DDI test will be smaller than the true relative risk . In our follow-up sensitivity analysis , medication frequency was justified in the DDI analysis . This factor would also account for a portion of baseline myopathy risk . Another potential approach to estimate the baseline myopathy risk is to identify a control patient group that matches the demographics , co-morbidity , and co-medication distributions of the group exposed to the DDIs . This approach deserves further investigation . Like many pharmaco-epidemiology studies using observational data , our analysis of the DDI effect on myopathy has several limitations . Creating an accurate phenotypic definition using billing codes may be unreliable , with both false-positives and false-negatives likely to occur . Our dataset also lacked clinical notes from which more detailed symptom data could be extracted . Further research including validation with manual chart review is necessary to establish optimal phenotypic definitions for myopathy , as well as more granular definitions for myotoxicity and rhabdomyolysis . Further research including validation with manual chart review is necessary to establish optimal phenotypic definitions for myopathy , as well as more granular definitions for myotoxicity and rhabdomyolysis using a combination of ICD9 codes , lab tests , and clinical notes . Another limitation of our analysis is that it is subject to several potential population bias introduced by the EMR database itself . Our retrospective observational data do not allow for controlling many potential covariates that a traditional prospective study offers . In particular , the race data is not complete in our database . It is also equally challenging to design a prospective study to validate our results from a pharmaco-epidemiology study . Clinical pharmacokinetic studies or further in vitro metabolism/inhibition studies of the selected DDI pairs found to increase myopathy may provide further validation of an interaction between the drugs . We are also looking forward to validating our results in another large EMR database . Our text mining and DDI prediction is CYP metabolism enzyme based . Therefore , our interpretation of the five significant drug interactions focuses only on CYP drug-drug interaction mechanisms . However , this does not preclude the involvement of other DDI mechanisms , such as drug transporter interactions or pharmacodynamic interactions . In a recent GWAS study , expression of the OATP1B1 transporter was shown to predict myopathy risk associated with simvastatin [20] . Therefore , it is possible that loratadine interacts with simvastatin through this or other transporter mechanisms . Studies are currently underway to further characterize the mechanisms of the five identified DDIs . The concomitant use of CYP3A metabolized statins ( atorvastatin , lovastatin and simvastatin ) with strong CYP3A inhibitiors ( e . g . ketoconazole and itraconazole ) reportedly increases risk of statin-induced myopathy . In addition , case reports of increased myopathy in transplant recipients being treated with tacrolimus or cyclosporine [21] argue for the avoidance of this combination . The interaction between statins and fibrates , specifically gemfibrozil , leading to increased risk of myopathy is well recognized [22] . Gemfibrozil is a substrate of CYP3A but not a potent inhibitor . Thus , it is likely that this interaction occurs through pharmacodynamic , not pharmacokinetic , based interactions . Although these interactions are widely reported , we found no increased risk of myopathy with concomitant use of ketoconazole , itraconazole , tacrolimus , or gemfibrozil within the CDM database . Their related myopathy risks of these DDIs are reported in Table 6 . This finding is likely due to the limitation of our data analysis , in which we define concomitant drug administration by prescription orders that occur within a predefined timeframe . As these drug interactions are well-known , it is likely that although the two drugs may have been ordered within the predetermined time window , the individual may have discontinued one medication before starting the second . For some drugs that are used short-term , e . g . ketoconazole , it will be difficult to identify true concomitant use from prescription records . As a matter of fact , among these statin DDI pairs in Table 6 , less than 110 patients took both drugs within the pre-defined one month interval in each pair . This limited our power to detect significant DDIs to less than 15% , if we anticipate a 1 . 5-fold RR of DDI myopathy . Provided that medication data in our CDM is relatively new , between 2004 and 2009 , it is likely that clinicians were aware of potential interactions and thus suggested patients avoid co-administration of these interacting drugs . As described in the introduction , an in vitro , an in vivo , or an in populo pharmacologic study alone cannot cover the whole spectrum of mechanistic and clinically significant DDI research . These studies usually focus on a few drug pairs for one or a limited number of metabolizing enzymes or transporters at a time . In this paper , we combined a literature discovery approach and a large EMR database validation method for novel DDI prediction and clinical significance assessment . The scale of our research covered all FDA approved drugs . The literature based discovery approach predicted new DDIs and their associated CYP-mediated metabolism enzymes . The clinical significance of these interactions was then assessed in large database of electronic medical records . This translational bioinformatics approach successfully identified five DDI pairs associated with increased myopathy risk . Compared to traditional in vitro , in vivo , and in populo DDI studies , our proposed translational bioinformatics approach covers a broader spectrum and identifies risk on a larger scale . It certainly motivates more in vitro studies to investigate alternative DDI mechanisms and more clinical pharmacokinetics study to investigate the clinical significance of these DDIs . The Indiana Network for Patient Care ( INPC ) is a heath information exchange data repository containing medical records on over 11 million patients throughout the state of Indiana . The Common Data Model ( CDM ) is a derivation of the INPC containing coded prescription medications , diagnosis , and observation data on 2 . 2 million patients between 2004 and 2009 . The CDM contains over 60 million drug dispensing events , 140 million patient diagnoses , and 360 million clinical observations such as laboratory values . These data have been anonymized and architected specifically for research on adverse drug reactions through collaboration with the Observational Medical Outcomes Partnership project [23] . This CDM model is a de-identified eletronic medical record database . All the research work has IRB approval . Our drug dictionary consists of 6 , 837 drugs names that include all brand/generic/drug group names . They were primarily derived from DrugBank [24] . We then excluded non-approved and experimental drugs , and focused only on FDA approved therapeutic agents , which left 1492 unique drug generic names for the mining purpose ( Figure 1 ) . The INPC CDM data set has 54490 unique drug “Concept IDs” . A Concept ID in the CDM typically maps to an RxNorm clinical drug ( e . g . , simvastatin 20 mg ) or ingredient ( simvastatin ) . Some Concept IDs may contain multiple drug components ( e . g . , lisinopril/hydrochlorothiazide ) . Our drug dictionary was mapped to CDM Concept ID's using regular expression matching and manual review . In total , 1293 unique drugs identified from DrugBank were mapped successfully , while 199 drugs could not be matched . The unmatched drugs were categorized as follows: banned drugs , illicit drugs , organic compounds , herbicide/insecticides , functional group derivatives , herbal extract , DrugBank drugs not covered by CDM , and literature only drug names . In our CDM dataset , 817059 patients had medication records available . Literature mining was conducted on 10 CYP enzymes: ( CYP1A2 , CYP2A6 , CYP2B6 , CYP2C8 , CYP2C9 , CYP2C19 , CYP2D6 , CYP2E1 , CYP3A4/CYP3A5 ) ( Figure 5 ) . Please note that these CYPs cover all the major ones , but not all of the CYPs . A probe substrate of enzyme E is defined as being selectively metabolized by enzyme E; while a probe inhibitor of enzyme E selectively inhibits enzyme E's metabolism activity . CYP probe drugs and inhibitors for the DDI text mining approach were selected as those drugs well-established as probes or inhibitors by DDI researchers and defined in the FDA guidance [6] . The in vitro CYP enzyme substrate and inhibitor text mining and the DDI prediction was divided into the following steps . In vivo DDI text mining was conducted on those predicted DDI pairs from in vitro DDI text mining ( Figure s1 ) . It is broken down the following steps . Demographic variables , age and sex , were justified in the DDI association analyses . The total number of different medications ordered during the one month drug exposure window was used as a covariate in the logistic regression . It serves as a surrogate of the patients' overall health status , and justifies for myopathy effects from medications other than the hypothesized DDI drug pair . It is recognized that an individual patient can experience multiple myopathy events . Our drug-condition model considered two situations: all myopathy events and the first myopathy event . The advantage of selecting the first myopathy event is that it is not confounded with other medications taken between the first and the follow-up myopathy events . However , limiting the data to first myopathy even reduces the sample size , and thus the power to identify a DDI . DDI pairs , in which at least one drug was prescribed to treat symptoms of myopathy ( e . g . narcotic and non-steroidal analgesics ) , were excluded from the DDI tests . However , the patients prescribed these drugs are kept in the data analysis .	Drug-drug interactions are a common cause of adverse drug events . In this paper , we developed an automated search algorithm which can predict new drug interactions based on published literature . Using a large electronic medical record database , we then analyzed the correlation between concurrent use of these potentially interacting drugs and the incidence of myopathy as an adverse drug event . Myopathy comprises a range of musculoskeletal conditions including muscle pain , weakness , and tissue breakdown ( rhabdomyolysis ) . Our statistical analysis identified 5 drug interaction pairs: ( loratadine , simvastatin ) , ( loratadine , alprazolam ) , ( loratadine , duloxetine ) , ( loratadine , ropinirole ) , and ( promethazine , tegaserod ) . When taken together , each drug pair showed a significantly increased risk of myopathy when compared to the expected additive myopathy risk from taking either of the drugs alone . Further investigation suggests that two major drug metabolism proteins , CYP2D6 and CYP3A4 , are involved with these five drug pairs' interactions . Overall , our method is robust in that it can incorporate all published literature , all FDA approved drugs , and very large clinical datasets to generate predictions of clinically significant interactions . The interactions can then be further validated in future cell-based experiments and/or clinical studies .	[ "Abstract", "Introduction", "Results", "Discussion", "Methods" ]	[ "medicine", "clinical", "pharmacology", "pharmacoepidemiology", "drugs", "and", "devices", "statistics", "text", "mining", "drug", "information", "mathematics", "pharmacology", "biostatistics", "information", "technology", "drug", "metabolism", "adverse", "reactions", "phar...	2012	Literature Based Drug Interaction Prediction with Clinical Assessment Using Electronic Medical Records: Novel Myopathy Associated Drug Interactions
The ILEP Nerve Function Impairment in Reaction ( INFIR ) is a cohort study designed to identify predictors of reactions and nerve function impairment in leprosy . The aim was to study correlations between clinical and histological diagnosis of reactions . Three hundred and three newly diagnosed patients with World Health Organization multibacillary ( MB ) leprosy from two centres in India were enrolled in the study . Skin biopsies taken at enrolment were assessed using a standardised proforma to collect data on the histological diagnosis of leprosy , leprosy reactions and the certainty level of the diagnosis . The pathologist diagnosed definite or probable Type 1 Reactions ( T1R ) in 113 of 265 biopsies from patients at risk of developing reactions whereas clinicians diagnosed skin only reactions in 39 patients and 19 with skin and nerve involvement . Patients with Borderline Tuberculoid ( BT ) leprosy had a clinical diagnosis rate of reactions of 43% and a histological diagnosis rate of 61%; for patients with Borderline Lepromatous ( BL ) leprosy the clinical and histological diagnosis rates were 53 . 7% and 46 . 2% respectively . The sensitivity and specificity of clinical diagnosis for T1R was 53 . 1% and 61 . 9% for BT patients and 61 . 1% and 71 . 0% for BL patients . Erythema Nodosum Leprosum ( ENL ) was diagnosed clinically in two patients but histologically in 13 patients . The Ridley-Jopling classification of patients ( n = 303 ) was 42 . 8% BT , 27 . 4% BL , 9 . 4% Lepromatous Leprosy ( LL ) , 13 . 0% Indeterminate and 7 . 4% with non-specific inflammation . This data shows that MB classification is very heterogeneous and encompasses patients with no detectable bacteria and high immunological activity through to patients with high bacterial loads . Leprosy reactions may be under-diagnosed by clinicians and increasing biopsy rates would help in the diagnosis of reactions . Future studies should look at sub-clinical T1R and ENL and whether they have impact on clinical outcomes . Diagnosing leprosy and the immunological reactions that complicate this disease is not always straightforward . Leprosy skin lesions can have a very variable appearance and the presence of inflammation which is associated with immune reactions is not always obvious . The INFIR cohort study was set up in North India in 2000 to study risk factors for leprosy reactions in a cohort of newly diagnosed patients with multibacillary ( MB ) leprosy . Patients with definite clinical evidence of MB leprosy were recruited to the cohort . All patients had a skin biopsy on recruitment . Previous publications relating to this cohort have reported on clinical and neurophysiological aspects of the study . Important clinical findings included the observations that there was a high level of nerve damage in these patients and that new nerve damage and clinical reactions were detectable in 28% of patients at recruitment . [1] The serological studies showed that LAM IgG1 antibody levels were significantly elevated in patients with skin reactions and nerve function impairment ( NFI ) . [2] The immuno-histological studies in skin biopsies have shown a significant association between the presence of three cytokine proteins , TNF-α , iNOS and TGF-β , and Type 1 Reactions ( T1R ) in skin and nerve damage . [3] This paper reports on the correlation between the clinical and the histological findings . The clinical manifestations of leprosy are determined by the immune response of the patient to Mycobacterium leprae . There is a spectrum of immune responses . Patients with tuberculoid leprosy ( TT ) have well developed cell mediated immunity and localised skin or nerve lesions , whilst patients at the other end of the spectrum have lepromatous leprosy ( LL ) which is associated with absent cell mediated immunity , mycobacterial proliferation and numerous skin and nerve lesions . Between these two extremes are the borderline types of leprosy in which patients have decreasing levels of cell mediated immunity and increasing numbers of lesions as they move from borderline tuberculoid ( BT ) to borderline lepromatous ( BL ) . Patients are assigned a Ridley-Jopling classification on the basis of the morphology , type and number of skin lesions , and nerve involvement , supplemented by the bacterial index ( BI ) [4] and histological examination of the skin lesion wherever possible . The Ridley-Jopling types are Tuberculoid ( TT ) , Borderline Tuberculoid ( BT ) , Borderline Borderline ( BB ) , Borderline Lepromatous ( BL ) , Lepromatous Lepromatous ( LL ) , Pure Neural ( PN ) and Indeterminate ( I ) . The Ridley-Jopling classification links immune status and clinical manifestations . Patients with borderline leprosy are immunologically unstable and at risk of developing leprosy reactions and NFI . The Leprosy Unit at the World Health Organization ( WHO ) has developed a simpler classification for use in the field and for assigning patients to treatment regimens . Patients are classified on the number of skin and nerve lesions and skin smear positivity/negativity . Patients with up to five skin lesions and/or one nerve involved and smear negativity are classified as having paucibacillary leprosy ( PB ) and are treated with six months of PB multi-drug therapy ( MDT ) and patients with more than five skin lesions and/or more than one nerve involved and/or smear positivity are classified as having multi-bacillary ( MB ) leprosy and receive 12 months of MB MDT . [5] In referral centres both classification systems may be used . Previous studies comparing clinical and histological diagnoses have found variability between the two ways of classifying patients . Moorthy et al [6] assessed 372 skin biopsies in patients in India and found agreement between the clinical and histological diagnoses in only 62 . 6% of cases . Pardillo et al [7] in a study in the Philippines found substantial under-diagnosis of BB/BL and BL disease with 38% of these patients having fewer than five lesions and so being classified as PB . In the INFIR cohort study patients were recruited using the WHO classification but then assigned a Ridley-Jopling classification using clinical criteria and then had a histological classification made from the skin biopsy . This allows us to compare the clinical and histological classifications . Classification of leprosy patients is important because if under-diagnosis of MB patients occurs , then patients with a significant bacterial load will be under-treated and be at risk of relapse . Conversely , patients who have low bacterial loads but more than five lesions will be classified as having MB type leprosy and will be over-treated . T1R are delayed hypersensitivity reactions and are clinically important because acute peripheral nerve damage occurs during these episodes . T1R are clinically defined by the presence of new erythema in skin lesions and new loss of nerve function in peripheral nerves . Histologically , oedema and inflammation are seen in leprosy granulomas in skin and nerve biopsies . [8] , [9] The clinical definition of Type 1 and Type 2 Reactions has developed by consensus with little testing of the accuracy of clinical diagnosis of T1R . A previous study in India compared the diagnostic rates for T1R by clinicians and histopathologists [10] and showed that clinicians had a higher rate of diagnosing reactions than histopathologists . In that study the clinical diagnosis of a T1R was accompanied by histological changes in 60% cases . The structure of the INFIR cohort with all patients having a biopsy taken at baseline enables us to examine the diagnoses of T1R clinically and histologically and to calculate sensitivity and specificity rates for these different diagnostic tools . We predicted that the clinicians would have a higher rate of diagnosis of T1R than the histopathologist . We also predicted that there would be reactions diagnosed on histological examination that had not been apparent clinically . Erythema Nodosum Leprosum ( ENL ) is an immune-mediated common complication of LL , occurring in about 50% of patients with LL , and presenting with skin lesions ( red , painful and tender subcutaneous lesions ) , fever and systemic inflammation that may affect the nerves , eyes , joints , testes and lymph nodes . [11] , [12] The diagnosis of ENL has also evolved by consensus with different case definitions being used . [13] The case definition for ENL in the INFIR cohort was based on detecting skin lesions and systemic signs/symptoms of inflammation in patients with BL/LL classification . ENL is diagnosed histologically when a vasculitis with neutrophil polymorph cells infiltrating the lesions are present . [8] A study in Pakistan comparing the clinical and histological diagnosis of ENL found that 36% of patients with clinically diagnosed ENL did not have the typical cell infiltrates associated with vasculitis in their skin biopsies . [14] Thus , there are also discrepancies between clinical and histological diagnoses of ENL . In the INFIR cohort study we tested the correlation between the clinical and histological diagnoses of ENL at entry into the cohort . The INFIR cohort study allowed us to compare the diagnosis of leprosy and both types of reaction in a cohort of newly diagnosed patients . We report here on the following comparisons: This was a cohort study of 303 newly registered MB patients . The patients were followed up monthly for one year and every second month during the second year . Recruitment of subjects took place in The Leprosy Mission ( TLM ) hospitals in Naini and Faizabad , specialist leprosy referral centres in Uttar Pradesh , India . The histopathological analysis was done at the LEPRA Society Blue Peter Research Centre in Hyderabad , Andhra Pradesh . The study population comprised newly registered MB patients requiring a full course of MDT . A detailed description of the study design methods , clinical definitions , documentation and the status of the cohort at baseline have been published . [1] , [15] Patients were initially classified by the Ridley-Jopling scale clinically before the local skin smear result was available . When the histological diagnosis became available all the classification diagnoses were reviewed together with slit skin smear data and reconciled to give a final diagnosis which was then used for subsequent analysis . For the Ridley-Jopling classification the histological diagnosis took precedence over the clinical classification so that a patient classified clinically as BT but with BL histology would have a final BL classification . All the data obtained in the study , including the clinical , neurophysiological , serological and histopathological data , were entered on computer locally and subsequently merged into a single Microsoft Access database . All patients had an elliptical incision skin biopsy taken from an active skin lesion at enrolment . If the patient developed a Type 1 or Type 2 reaction a second skin biopsy was taken from a typical active lesion . The biopsies were split in half , one portion being fixed in 10% buffered Formalin and the other snap frozen in liquid nitrogen and then transported to the Blue Peter Research Centre at Hyderabad for processing and analysis . The skin biopsies were processed and embedded in paraffin and serially sectioned in the saggital plane at 5 µm thickness on a Leica microtome . Sections were stained with Haematoxylin and Eosin stain ( H & E stain ) to study morphology , and modified Fite Faraco stain to identify acid fast bacilli ( AFB ) . AFB was graded according to Ridley scale of 0 to 6+ as Bacillary Index of Granuloma ( BIG ) . The biopsy assessments were done using a standardised set of definitions for histological features and recorded on a proforma . A single pathologist ( SS ) reviewed the H & E and Fite stained sections and assessed the diagnosis of leprosy , assigned each case a Ridley-Jopling classification and assessed the presence of leprosy reaction . This was confirmed when evidence of nerve inflammation and/or AFB were seen and a granulomatous inflammation consistent with leprosy was present . [4] The following morphological features were assessed on all sections: A purposely selected sample of 66 slides was sent to a second pathologist who used the same scoring system for diagnosis of leprosy and reactions . The selection covered the full range of leprosy types and reactions . The paired assessments made by SS and by an independent assessor blinded to the assessment by SS were compared . There was perfect or good agreement on the Ridley-Jopling classification in all 51 biopsies ( this excludes biopsies that showed non-specific inflammation ) . For the BI assessment the Kappa was 0 . 5 and for the granuloma fraction assessment the Kappa was 0 . 6 indicating good agreement . SS diagnosed T1R in 20/66 biopsies and MJ in 11/66 . There were four T1R diagnosed by MJ but not SS and 13 diagnosed by SS but not MJ . Comparing the ENL diagnoses showed that SS diagnosed 6/66 as having ENL and MJ 4/66 . They only agreed on one ENL diagnosis . A sub group of patients at risk of developing T1R was identified . This excluded patients with ENL and the patients with no significant lesion ( NSL ) . Patients with ENL were excluded because they were very unlikely to have both T1R and ENL together at baseline . Patients with NSL were excluded in this analysis because the comparison involved a histological comparison and their biopsy showed so little inflammation that assessing the histological features of reaction was not possible . Similarly the group at risk of developing ENL included LL cases plus any BL cases in ENL at time of leprosy diagnosis . LL cases with no ENL at time of diagnosis were therefore included in both groups . The NSL group was excluded from both groups . No financial incentives were given to participants . However , travel expenses were refunded on occasion and where relevant , lost earnings of daily labourers compensated . The study adhered to the International Ethical Guidelines for Biomedical Research Involving Human Subjects [16] . Permission for the study was obtained from the Indian Council of Medical Research and the Research Ethics Committee of the Central JALMA Institute for Leprosy in Agra gave ethical approval . Written consent was obtained from individual study subjects before inclusion in the study , using a standard consent form . Three hundred and three patients were recruited and 299 had biopsies that were adequate for examination . Four biopsies were too small or too superficial with inadequate dermis . Table 1 compares the initial clinical classification which was made against the histological diagnosis . BT was the main clinical diagnosis , comprising 59 . 5% of the patients , but 41% were reclassified after histological diagnosis: 24 ( 13% ) to BL , two to LL and 32 and 15 to indeterminate and NSL respectively . Patients in the BL group were re-classified after histological diagnosis , with 17 going to BT , nine to LL and two and three to indeterminate and NSL respectively . The LL group had the highest rate of revised diagnoses with only 17 cases being diagnosed and confirmed ( 54% ) . Eleven ( 35% ) cases were reclassified to BL , two to BT 6% and one to Indeterminate . Two BT and nine BL cases were re-assigned to the LL category . The PN category encompassed the whole spectrum . Although the PN cases had no apparent skin lesions , histological evidence of leprosy was found in 9/13 skin biopsies ( one BT , three BL and five indeterminate ) . Assessing the agreement between diagnoses was calculated as the number of positive assessments with agreement as a percentage of the total number of positive assessments for BT 68 . 6% , BL 54 . 2% and LL 57 . 6% . Four histological categories were recognised for this group ( see above ) so that patients whose leprosy was resolving could be identified . Sixty one ( 20 . 4% ) patients had a histological diagnosis of indeterminate or NSL seen on biopsy . Thirty two biopsies showed signs of indeterminate leprosy , with the potential to either progress or heal . Eight biopsies were classed as indeterminate/resolved , indicating that early leprosy had been present and was now healing . In 12 biopsies there was no evidence of leprosy . Ten biopsies were classed as NSL/resolved indicating that a lesion had been present but was resolving ( Table 2 ) . Thus 20 . 4% of this cohort of MB leprosy patients had skin biopsies with only minimal inflammation and in half of these cases there was evidence of resolution . Clinically these patients had been classified: 47 as BT , four as BL , one as LL , four as BT ( PN ) and five as BL ( PN ) . Table 3 shows the clinical signs of leprosy in this group which showed both significant numbers of skin lesions , ranging from 23 . 2 ( mean ) in the indeterminate group to 16 . 5 in the NSL/resolved group; and significant numbers of thickened nerves , ranging from 4 . 4 ( mean ) in the indeterminate group to 2 . 75 in the NSL group . One patient had tender nerves . These clinical signs could also be consistent with active or recently active leprosy . Four patients developed NFI or reaction ( one motor NFI , one neuritis , two sensory NFI ) during follow-up . Type 1 and Type 2 reactions were diagnosed clinically and the diagnosis reviewed on the database by WvB and DNL . Reactions were also diagnosed histologically . Of the cohort , 265 patients were at risk of developing T1R and 43% , 39% and 9% of the clinically diagnosed BT , BL and LL patients being diagnosed by the histopathologist as having a T1R in their baseline biopsy . The histopathologist also rated his diagnostic certainty for a T1R . A T1R was diagnosed definitely in 96 cases and also in a further 22 biopsies of which 17 were rated probable and six possible T1R . Table 4 compares the clinical and histological diagnoses of T1R , differentiating between clinical reactions with skin only , nerve only and nerve plus skin involvement . We found substantial disparity between clinical and histological diagnoses of reactions ( Table 4 ) . Clinicians diagnosed reactions in 96 patients ( 36 skin only , 41 nerve only and 19 skin and nerve ) . Reactions were diagnosed histologically in 113 patients ( 26 skin only , 14 nerve only and 13 skin and nerve ) . The skin comparisons are the most useful here since all patients had a skin biopsy but only a subset of patients had a nerve biopsy . For 109 patients there was clinical and histological agreement of no reaction and agreement on 53 in reaction . There were 60 patients with a histological diagnosis of T1R but no clinical diagnosis and 43 with a clinical diagnosis but not a histological . Using histological diagnosis as a gold standard , the sensitivity of clinical diagnosis is 34 . 5% and the specificity 89 . 4% . One hundred and thirteen patients had a clinical diagnosis of BT leprosy and 49 had T1R diagnosed ( 17 skin only , 26 nerve only , six skin and nerve ) ; T1R was diagnosed histologically in 69 patients ( 55 skin only , 12 nerve only and five skin and nerve ) . The sensitivity of clinical diagnosis is 53 . 1% and the sensitivity is 61 . 9% . Sixty seven patients had BL leprosy and 36 had a clinical diagnosis of T1R ( 15 skin only , 11 nerve only and 10 skin and nerve ) . Histological diagnoses were made in 31 patients giving clinical diagnosis a sensitivity of 61 . 1% and a specificity of 71 . 0% . The clinical information on patients with a histological reaction diagnosis was explored to see whether the skin diagnosis was a marker for pathology elsewhere . Seventeen patients had evidence of an immune mediated process going on elsewhere . Of the seven BL cases , six had new sensory NFI and one had ENL; of the 10 BT cases , five had sensory NFI , three both sensory and motor NFI , and two had other processes ( one motor NFI , one ENL ) . There were 44 patients who had a clinical diagnosis of T1R and no evidence of ongoing nerve damage . The hypothesis that these reaction diagnoses might be indicators for a reaction about to happen was tested by looking at the subsequent reaction history of these patients . There were 74 patients in this group , who had been diagnosed with a T1R histologically but not clinically . Of these , 30 had a reaction during follow-up and in 15 cases they had a T1R diagnosed both clinically and histologically . These reactions were spread out over the follow-up period but with a peak in the first three months of follow-up when six T1Rs occurred . It is therefore possible that reactions are being pre-diagnosed by the pathologist but only in a small number of patients . There were 28 patients with LL who were at risk of ENL . ENL was diagnosed clinically in two LL patients at entry to the study , while 13 patients had histological evidence of ENL on their skin biopsy taken at baseline ( Table 5 ) . Two patients with BL leprosy were also diagnosed with ENL . The certainty of the histological diagnosis of ENL was nine definite , four probable and three possible . Using histological diagnosis as a gold standard , this gives clinical diagnosis a sensitivity of 15 . 4% and a specificity of 100% . Patients ( LL and BL types ) had single and multiple clinical episodes of ENL . Fourteen patients had 24 episodes of ENL , five occurred at baseline as single events , and seven patients had multiple episodes in the two year follow up . We have no data about episodes of ENL after the close of the two year follow-up . Only two of the 12 patients diagnosed with histological but not clinical ENL at baseline subsequently developed ENL . The key finding from this cohort study is that reactions are more frequent than is clinically evident . We have also shown that leprosy manifests in a range of clinical and histological pathologies , and that there are significant numbers of patients both with indeterminate disease and with healing and resolving disease at the site of the biopsy . Patients were carefully evaluated at recruitment and their diagnoses reviewed after the histological results were available . There were significant differences in the diagnosis of T1R by clinicians and pathologists , with the clinicians diagnosing fewer T1R . This is surprising and contrasts with a previous piece of work on the diagnosis of T1R in India when histopathologists diagnosed fewer T1R than clinicians . [10] T1R can be difficult to diagnose; it can be difficult to differentiate clinically between BT and BT in reaction . Histologically , the diagnosis depends on demonstrating granuloma and dermal oedema and these signs can also be variable . Clinicians and pathologists were probably looking for different factors to make the diagnosis of a T1R . . There is variation between pathologists in the agreement on the diagnosis of reactions . In this INFIR study we have found marked differences between the assessments that two pathologists give to the diagnosis of reactions . [3] In the previous study we found that finding expression of HLA-DR in the epidermis significantly increased the rate of histological diagnosis , and staining for this marker could also be applied to these biopsies . It is also important to ensure that pathologists involved in clinic-pathological studies have pre-study training to agree on the evaluation of diagnostic criteria especially for reactions . Studies should be planned that involved clinicians and pathologists in a real-time review of leprosy patients with suspected reactions and their biopsy findings so that diagnostic criteria are established that link the diagnoses of clinicians and pathologists more closely . It is also important to determine whether there are clinical consequences , such as new nerve impairment for patients with sub-clinical reactions . It may be that patients with subclinical reactions would benefit from steroid treatment . ENL was also diagnosed differently by clinicians and pathologists in this cohort . The finding of ENL in 17% of the skin biopsies from LL patients and 7% from the BL patients shows that ENL is a continuing problem . This ENL was being diagnosed at baseline , whereas one would expect a higher rate of ENL after several months of treatment . The changes seen in the biopsies here when the diagnosis of ENL was made were typical , with infiltration of polymorphs into the lesions and a vasculitis . It can be difficult to diagnose ENL clinically especially when it is present in a mild form . Still it is surprising that 80% of the histologically diagnosed ENL episodes at baseline did not have clinical signs of reaction . It might be difficult to detect ENL in a newly diagnosed LL case when the LL skin lesions are active . This study suggests that subclinical ENL may be important . This finding needs to be validated in other studies and also with patients at risk of ENL followed closely to determine the clinical effects of sub clinical ENL . This highlights the importance of training doctors and health workers to specifically ask patients with LL and BL type disease about symptoms of ENL such as new nodular lesions , bone pain , orchitis and fever . ENL is important to diagnose because it may cause morbidity to eyes , bones and testes . These data show that the leprosy WHO classification MB group is very heterogeneous and comprises patients with all types of leprosy with the exception of single lesion tuberculoid and so includes patients with Indeterminate , BT , BB , BL , LL and PN leprosy . Patients were entered into this cohort when the enrolling clinician felt certain that the patients had a clinical diagnosis of MB leprosy . Of the BT patients 80% had no mycobacteria detectable on either slit skin smear or in their biopsies . The BT disease seen in these patients is immunologically active and we have reported elsewhere that staining for cytokines and inflammatory markers in these biopsies shows a high level of immunological activation with abundant production of the pro-inflammatory cytokines TNFa , iNOS , and TGFb . [3] High rates of BT leprosy in Indian patients have been reported before . Moorthy et al [6] found that BT leprosy was clinically diagnosed in 54% of their cohort but present in 72% of biopsies . It is also surprising that in a cohort designed to recruit new untreated MB cases that there should be 17 . 9% patients with histological evidence of indeterminate or resolving leprosy . There are several possible explanations that should be considered , inadequate biopsies ( including those taken from a non-active lesion ) , self healing of early lesions and undeclared previous treatment . Early leprosy lesions often have minimal inflammation and in the Karonga study in Malawi , NSL inflammation was found in 17% of biopsies taken from a cohort of 664 patients with suspected leprosy . [17] The biopsy might also have missed the active inflammation . In the study of Moorthy et al , indeterminate leprosy was reported clinically in 3 . 5% and found histologically in 6 . 7% of patients . It may be postulated that this high rate of self healing is part of a picture of local high endemicity where there is a high rate of infection with M . leprae and a high self healing rate . There can be very high rates of local infection in the Indian sub-continent . In Mumbai , Shetty et al [18] have found local case detection rates of leprosy as high as 9 . 42/10 , 000 . The reclassifications that occurred between the BL and BT group are not surprising because the skin lesions may appear similar . Furthermore patients may spontaneously upgrade from a BL to BT phenotype without having an overt T1R . Conversely , patients with BT leprosy may be moving silently towards the BL and even LL phenotypes and this is reflected in the results . We found 24 patients initially classified as BT whose skin biopsies showed BL leprosy and two biopsies that showed LL . There was also significant under-diagnosis of LL disease in this cohort . Most of the misdiagnoses relating to LL disease were in the BL group and differentiating between these two types is not always straightforward . Groenen et al [19] found that adding in a category of diffuse infiltration and nodules improved the diagnosis of LL cases in the MB leprosy classification by ensuring that diffuse infiltration is not overlooked . This data re-iterates the value of doing skin biopsies in leprosy especially when reactions are suspected . One simple lesson from this study is that the MB classification is useful because these patients have a high rate of leprosy reactions , and resources and follow-up should be focused on these patients . The findings from this cohort illustrate how complex diagnosis and classification of leprosy reactions can be and how important regular discussion and review of patients and their biopsies between pathologists and clinicians can be . It is also important that referral centres should have a ready access to pathologists who are experienced in leprosy diagnosis and this should be recognised when planning and funding such centres . It is very important that both clinicians and pathologists be aware of the local patterns of presentation and are able to detect changes in these patterns . It would also be useful to quiz these patients further to establish whether any of them have received anti-leprosy drugs from another source such as a private practitioner . It is also important that there should be teaching and training about classification of leprosy patients and this is another function of referral centres that needs to be developed . This study has shown that leprosy continues to present in a range of forms and that early self healing disease is present as well as the more advanced forms . In this report we have focused on the changes seen in skin biopsies and have found that there is a significant under-diagnosis of both T1R and ENL , when comparing clinical diagnosis against histological evidence . This has important clinical implications for training and service delivery . Health workers need to be trained to suspect reactions , robust referral systems for evaluating patients with suspected reactions need to be developed and programme managers need to ensure that there are adequate supplies of steroids for the treatment of reactions in both field stations and referral centres .	Leprosy affects skin and peripheral nerves . Although we have antibiotics to treat the mycobacterial infection , the accompanying inflammation is a major part of the disease process . This can worsen after starting antibacterial treatment with episodes of immune mediated inflammation , so called reactions . These are associated with worsening of nerve damage . However , diagnosing these reactions is not straightforward . They can be diagnosed clinically by examination or by microscopic examination of the skin biopsies . We studied a cohort of 303 newly diagnosed leprosy patients in India and compared the diagnosis rates by clinical examination and microscopy and found that the microscopic diagnosis has higher rates of diagnosis for both types of reaction . This suggests that clinicians and pathologists have different thresholds for diagnosing reactions . More work is needed to optimise both clinical and pathological diagnosis . In this cohort 43% of patients had Borderline Tuberculoid leprosy , an immunologically active type , and 20% of the biopsies showed only minimal inflammation , perhaps these patients had very early disease or self-healing . The public health implication of this work is that leprosy centres need to be supported by pathologists to help with the clinical management of difficult cases .	[ "Abstract", "Introduction", "Materials", "and", "Methods", "Results", "Discussion" ]	[ "medicine", "bacterial", "diseases", "infectious", "diseases", "leprosy" ]	2012	Comparing the Clinical and Histological Diagnosis of Leprosy and Leprosy Reactions in the INFIR Cohort of Indian Patients with Multibacillary Leprosy
The Mi-2/nucleosome remodeling and histone deacetylase ( NuRD ) complex is a multiprotein machine proposed to regulate chromatin structure by nucleosome remodeling and histone deacetylation activities . Recent reports describing localization of NuRD provide new insights that question previous models on NuRD action , but are not in complete agreement . Here , we provide location analysis of endogenous MBD3 , a component of NuRD complex , in two human breast cancer cell lines ( MCF-7 and MDA-MB-231 ) using two independent genomic techniques: DNA adenine methyltransferase identification ( DamID ) and ChIP-seq . We observed concordance of the resulting genomic localization , suggesting that these studies are converging on a robust map for NuRD in the cancer cell genome . MBD3 preferentially associated with CpG rich promoters marked by H3K4me3 and showed cell-type specific localization across gene bodies , peaking around the transcription start site . A subset of sites bound by MBD3 was enriched in H3K27ac and was in physical proximity to promoters in three-dimensional space , suggesting function as enhancers . MBD3 enrichment was also noted at promoters modified by H3K27me3 . Functional analysis of chromatin indicated that MBD3 regulates nucleosome occupancy near promoters and in gene bodies . These data suggest that MBD3 , and by extension the NuRD complex , may have multiple roles in fine tuning expression for both active and silent genes , representing an important step in defining regulatory mechanisms by which NuRD complex controls chromatin structure and modification status . Since its discovery in the late 1990's by a number of investigators , the Mi-2/nucleosome remodeling and histone deacetylase ( NuRD ) complex has been proposed to regulate chromatin structure and promote transcriptional repression via its intrinsic nucleosome remodeling and histone deacetylation activities [1] . Although this model provided a useful experimental framework , like all models , it has been challenged by subsequent data . In particular , the depiction of NuRD's principal function as a regulator of a silent chromatin state has been questioned by genetic and molecular studies from Georgopoulos and colleagues [2] , [3] , [4] and Hendrich and colleagues [5] , which provide compelling evidence that NuRD can have both positive and negative impacts on gene expression . zNuRD complex contains six core subunits which are invariably encoded by 2 or more gene paralogs , prompting the hypothesis that combinatorial assembly of subunits may contribute to functional specificity [6] . The smallest complex subunit can be either MBD2 or MBD3 [7] , members of the family of proteins that possess the methyl CpG binding domain ( MBD ) fold . While MBD2 is a bona fide methyl CpG binding protein , mammalian MBD3 has lost the ability to selectively interact with methylated DNA [8] . Recently , however , it has been suggested that MBD3 specifically binds another modified form of DNA , 5-hydroxymethylcytosine ( 5-hmC ) , leading to NuRD recruitment at 5-hmC marked loci in embryonic stem cells [9] . Other investigators find that NuRD complexes are involved in different aspects of the transcription cycle , coupling its action to enhancers [10] or to recruitment of polycomb complexes [11] . Currently , the field lacks consensus on the localization of MBD3 ( and by extension NuRD complex ) in the genome as well as its roles in modulating chromatin biology to facilitate gene regulation [12] . Chromatin immunoprecipitation coupled to massively parallel sequencing , ChIP-seq , represents ‘gold standard’ methodology to identify sites of enrichment of a particular protein ( or modified protein ) across the genome . High quality ChIP-seq data are dependent on a number of technical factors , including antibody quality , fixation conditions ( where ChIP is performed from fixed chromatin ) , chromatin shearing or cleavage with nuclease , and stringency of wash conditions for immune complexes . Progress over the last decade has established the principle that conditions that produce high quality ChIP data for one protein may not necessarily be effective for others . Performing ChIP on chromatin regulators , including NuRD complex , is particularly challenging [13] . Given the technical issues , the lack of concordance of recent NuRD ChIP-seq studies may not be surprising . It does , however , highlight a need for independent studies to provide additional data for comparison . Here we have analyzed human MBD3 localization across the genome using two complementary genomic approaches . We first used DNA adenine methyltransferase identification ( DamID ) , a methodology independent of antibodies , chromatin shearing and other technically challenging features of ChIP . We also performed ChIP-seq studies of endogenous human MBD3 following optimization of antibody , cross-linking and chromatin shearing . The results of these two techniques were in excellent agreement , suggesting they may not be unduly influenced by technical issues . The resulting , composite location analysis revealed unexpected association of MBD3 with the active fraction of the genome . MBD3 localized preferentially to active promoters characterized by histone marks associated with open chromatin and with genomic regions bearing the properties of enhancers , supporting the emerging evidence that models depicting NuRD's principal functions as a component of repressive chromatin require re-evaluation . In addition , MBD3 coated gene bodies of actively transcribed genes , extending to the transcript end site . Conversely , we also observed association of MBD3 with promoters of genes marked by repressive histone marks , albeit not with the frequency of active promoters , highlighting the likelihood that NuRD action is not unidirectional and is context dependent . Finally , the largest category of loci bound by MBD3 has no obvious association with known chromatin features , underscoring the relative dearth of knowledge on NuRD localization and function . We initiated location analysis of MBD3 in human cells using the DamID technique [14] . To facilitate biological comparisons , we chose two human breast cancer cell lines with different biological properties . MCF-7 cells provide a model for the luminal class of breast tumors , MDA-MB-231 ( MDA-231 ) cells model basal tumors [15] . As a prelude to location analysis , we first confirmed that the cell lines chosen express comparable levels of endogenous MBD3 , and that the exogenously expressed MBD3-Dam fusion protein was expressed , nuclear , and could be incorporated into NuRD complex ( Figure 1 , Figure S1 ) . We prepared two biological replicates of Dam alone and MBD3-Dam fusion for each cell line and performed 2 color hybridization using human promoter arrays . The raw data were processed as described in Methods and the ratio of MBD3-Dam to Dam alone displayed in genome browser format . Visual inspection of the data indicated multiple areas where the two cell types displayed highly similar patterns of localization as well as regions where localization differed by cell type ( Figure 1C ) . We determined local enrichment in the two cell lines by defining peaks using a conventional peak-calling algorithm ( see Methods ) . Peak localization was consistent across biological replicates ( greater than 75% overlap ) and differed across cell type ( less than 50% overlap ) suggesting the data were of high quality . We identified putative NuRD target genes as those containing a peak within 3 kb of the annotated transcription start site , resulting in 7 , 064 putative targets in MCF-7 cells , 9 , 310 in MDA-231 cells . Comparison of target genes across cell type indicated many common targets and many cell-type specific targets ( Figure 2A ) . Collectively , these data indicate that MBD3 , and by extension NuRD complex , has a cell-type specific localization pattern , suggesting cell-type specific functions . To assess the relationship of MBD3 localization to gene expression and histone modification patterns , we merged the biological replicates for each cell type , binned promoters into 20 bins from −7 to +3 kb relative to transcription start site ( TSS ) , and calculated occupancy for MBD3 and for H3K4me3 ( assessed by ChIP-seq ) as described in Methods . We displayed the results in a heatmap with genes ordered based on MBD3 density; scores for H3K4me3 and for gene expression were displayed in the same order . We observed a striking association of MBD3 DamID score with H3K4me3 density by ChIP-seq and with gene expression – regardless of cell type ( Figure 2B ) . Most MBD3 bound genes were highly expressed and carried high levels of H3K4me3 in both MCF-7 and MDA-231 cells , suggesting MBD3 preferentially associates with open chromatin regions at actively transcribed genes . To describe the binding pattern of MBD3-Dam with promoters , we constructed a composite analysis across all promoters ( −3 to +3 kb relative to TSS ) . MBD3 was prominently enriched in two distinct peaks , each about 1 . 5 kb from TSS , with a prominent dip at TSS in both cell lines ( Figure 2C ) . To ascertain whether this enrichment pattern was associated with promoter type , we subdivided our MBD3 data into three promoter classes [16] . Regardless of cell line , we observed that MBD3 preferentially associates with CpG rich promoters with little obvious accumulation evident at CpG poor promoters ( Figure 2D ) . To develop an understanding of the association of MBD3 with genic regions not represented on the promoter arrays , we performed MBD3 DamID using a tiling array covering human chromosomes 6 , 7 , and 8 . We constructed a gene model ( see Methods ) , observing a gradient of MBD3 density from a peak around TSS that gradually declines across the gene body . A second peak of MBD3 accumulation was noted roughly concurrent with the transcript end site ( TES ) ( Figure 2E ) . The association of MBD3 with overlapping , but distinct , genes in MCF-7 and MDA-231 cells suggested that the MBD3-NuRD complex may play a role in cell-type specific patterns of transcription . To address this issue , we performed Functional Analysis using Broad Institute's Molecular Signature Data Base ( MSigDB v 3 . 0 ) . We specifically asked whether the MBD3-NuRD target genes ( as defined for Figure 2A ) were enriched for gene expression patterns diagnostic of the luminal and basal transcriptional programs in breast cancer ( Figure 2F , Table S1 ) . Functional analysis demonstrated significant enrichment of luminal discriminatory genes , but not basal discriminatory genes , in the MCF-7 ( luminal ) specific MBD3-bound gene set . MDA-231 ( basal ) specific MBD3-NuRD putative targets displayed the opposite pattern . These data suggest that MBD3 , and by extension the MBD3-NuRD complex , may have a biological role in breast cancer subtype specification . The DamID technique , while extremely useful , has relatively low resolution , relies on the presence of the GATC motif , and our experiments provided data only on the regions tiled in the microarray platform chosen . Therefore , we opted to pursue MBD3 ChIP-seq to obtain a higher resolution map for MBD3 that included genic and intergenic regions not represented on the arrays . We first optimized conditions for ChIP of endogenous MBD3 , finding that fixation conditions are critical to the success of robust MBD3 ChIP . We applied a two-step crosslinking method , similar to our previous conditions [17] , [18] , crosslinking with disuccinimidyl glutarate followed by formaldehyde ( see Methods ) . Using these optimized conditions , we prepared two biological replicates of MBD3 ChIP in MCF-7 and in MDA-231 cells . Precipitated DNA was analyzed by massively parallel sequencing . After initial filtering and mapping of the sequence data , we merged the biological replicates for each cell line prior to further analysis ( see Methods ) . Visual assessment in genome browser format indicated multiple regions where enrichment appeared similar in both cell lines as well as many regions where enrichment was cell-type specific ( Figure 3A ) . We assessed local enrichment of MBD3 more rigorously by defining peaks using SICER [19] . We detected 35 , 165 and 23 , 880 peaks in MCF-7 and MDA-231 cells , respectively; peak overlap between the two cell lines was similar to the pattern observed in the DamID analysis ( Figure 3B ) . Finally , we formally compared peaks detected in ChIP-seq to DamID and observed excellent concordance ( Table S2 , Figure S2 ) . Like the DamID data , higher MBD3 density by ChIP was observed at a large number of promoter regions . As with DamID , we visualized the MBD3 localization data by binning promoters ( 500 bp bins , −7 to +3 kb relative to TSS ) and ordering genes by MBD3 density . We once again observed a strong association of MBD3 with actively transcribed genes marked by H3K4me3 ( Figure 4A ) . As was the case with DamID , metagene analysis indicated prominent peaks of MBD3 localization flanking a prominent dip in density at the TSS with a preferential association with CpG rich promoters , although the resolution of ChIP-seq was clearly superior ( Figure 4B , 4C ) . The ChIP-seq data permitted us to analyze regions not covered in depth by the DamID analysis . We assessed the overlap of MBD3 peaks with genomic features , finding that MBD3 overlapped with promoters ( −3 to +3 kb relative to TSS ) and with TSS with high frequency ( Figure 4D ) . As was the case with DamID , we also observed frequent peak overlap with TES's . MBD3 tended to be more frequently associated with genes ( exons and introns ) than with intergenic regions , although localization between annotated genes is a common event ( Figure 4D ) . Metagene analysis ( Figure 4E ) indicated that MBD3 peaks were found in abundance at or near the TSS with a 5′ to 3′ gradient across the gene body and a second , less pronounced , peak concurrent with the TES . As was the case with DamID data , genes marked by MBD3 in MCF-7 cells were enriched in transcripts defining the luminal gene expression pattern in breast cancer; genes marked by MBD3 in MDA-231 cells were enriched in transcripts defining the basal transcriptional program ( Figure 4F , Table S3 ) . Collectively , the ChIP-seq data indicate that MBD3 is predominantly found at actively transcribed genes with CpG island promoters and that the protein coats the gene body , extending to the TES . We utilized publicly available data from the ENCODE project [20] collected in MCF-7 cells to ascertain the nature of chromatin bound by MBD3 . Focusing first on transcript 5′ ends , we noted that transcripts in which an MBD3 peak overlaps the promoter region ( −3 to +3 kb relative to TSS ) were associated with a peak of H3K4me3 91 . 5% of the time ( Figure 5A ) . MBD3 bound promoters were associated with H3K9me3 very infrequently , approximately 4 . 7% of the time ( although this does represent a substantial portion of H3K9me3 bound promoters −42 . 6% ) . A significant number of MBD3 bound promoters were packaged in chromatin characterized by the presence of H3K27me3 ( 10 . 3% of MBD3 associated promoters , 48 . 3% of H3K27me3 associated promoters ) in agreement with data from Hendrich and colleagues in murine ES cells [11] . Given the prevalence of MBD3 peaks in intergenic regions , we asked whether these loci bore chromatin signatures indicative of function . We plotted the distributions of MBD3 and histone modifications ( H3K4me3 and H3K27ac ) relative to the MBD3 peak center . Non-TSS MBD3 colocalized with H3K27ac but not with H3K4me3 , while MBD3 colocalized with both H3K4me3 and H3K27ac around TSS ( Figure 5B ) . Thus , non-TSS MBD3 bound regions , on average , bear the chromatin signature of active enhancers , a genomic feature that has been associated with NuRD complex in ES cells [10] . Next , we quantified the overlap of MBD3 with various patterns of histone marks across the genome by dividing the human genome into 1 kb windows and calculating the enrichment for several chromatin marks in each window ( see Methods ) . Overall , we noted at least five different patterns of histone modification characteristic of MBD3 bound genomic regions ( Figure 5C ) . Regions encompassing the transcription start and containing histones modified by H3K4me3 ( pattern 1; active promoters ) were very abundant in this dataset , accounting for 18 . 4% of MBD3 enriched regions . Colocalization with H3K27me3 and TSS ( pattern 2; inactive promoters ) was present with some frequency ( 2 . 3% of all MBD3 enriched regions ) , while colocalization with H3K9me3 ( pattern 3; inactive promoters ) was rare ( about 1 . 4% ) in these data . Regions enriched in both MBD3 and H3K27ac , but not overlapping TSS ( pattern 4; active enhancers ) , were the most frequent pattern observed in our dataset ( 38 . 6% ) . Surprisingly , approximately one quarter ( 22 . 5% ) of MBD3 enriched genomic regions ( pattern 5 ) were found in chromatin lacking any of these patterns of histone marks , suggesting that much remains to be learned regarding NuRD complex enrichment relative to local chromatin features . To clarify whether MBD3 localization has any correlation with DNA modification in this system , we queried DNA methylation status within MBD3 bound regions . In murine ES cells , Yildirim and colleagues suggested a causal relationship between 5-hmC modification and NuRD localization [9] . We assessed the methylation status of MBD3 bound CpG islands ( using ENCODE reduced representation bisulfite sequencing data in MCF-7 cells ) . The majority of CpG islands overlapping a peak of MBD3 were hypomethylated with 60% of these islands falling within the lowest decile of DNA methylation ( Table 1 , Figure S3 ) . MBD3 was enriched at islands falling in the lowest two deciles of DNA methylation and excluded from the highest 3 ( Table 1 ) . These data indicate that MBD3 binds preferentially to unmethylated CpG islands in human breast cancer cells and agrees with recent biochemical studies indicating that MBD3 has no measurable biochemical preference for methylated cytosine [21] , [22] . Distal enhancer elements physically interact with promoter regions and these interactions have a major role in gene regulation [23] . A substantial portion of MBD3 bound loci have chromatin features consistent with function as enhancers . The spatial architecture of the nucleus and proximity of distal regulatory elements to promoters relative to occupancy of a given protein is conveniently measured by ChIA-PET [24] . Given the high frequency with which we observed MBD3 peaks at the TSS of active genes , we assessed the relationship of MBD3 occupancy to RNA polymerase II and to distal regulatory DNA by querying a Pol II ChIP-PET data set in MCF-7 cells . We observed frequent occurrence of MBD3 enrichment at both ends of Pol II ChIA-PET pairs . An exemplar locus , GATA3 is depicted in Figure 6A . We used 3C technology [25] to validate whether a selected subset of MBD3 bound intergenic peaks coinciding with Pol II ChIA-PET ends , including loci in the vicinity of GATA3 , NR2F2 , NRIP1 , and MASTL in MCF-7 cells , are in proximity to their respective promoters in three-dimensional space . At all 4 loci queried , we detected an interaction between a distal MBD3 bound peak and the promoter region ( Figure 6B , Figure S4 ) . To extend these observations to the level of the entire dataset , we plotted MBD3 density at TSS for genes with a Pol II ChIA-PET pair anchored at TSS , ranking genes by MBD3 abundance . Display of MBD3 abundance at the distal ChIA-PET pair in the same order revealed that genes with high level MBD3 at TSS tend to have high MBD3 at the distal region defined by Pol II ChIA-PET as being in proximity in three-dimensional space ( Figure 6C ) . These data show that some intergenic MBD3/H3K27ac peaks are in physical proximity to core promoters in three dimensional space , consistent with action as enhancers . Because MBD3 is a component of NuRD complex which contains chromatin remodeling factors ( CHD3 and CHD4 ) , we hypothesized that MBD3 regulates nucleosome occupancy at its binding sites . To test this hypothesis , we mapped the nucleosome positions in MBD3 depleted and control MCF-7 cells . We verified that MBD3 was efficiently depleted ( Figure S5 ) . Native chromatin from control and MBD3 depleted cells were digested with micrococcal nuclease and the resulting mononucleosome-sized DNA fragments were collected and subjected to massively parallel sequencing . We identified about 150 million nucleosomes for each group and performed a metagene analysis centered on TSS ( Figure 6D , Table S4 ) . Control cells showed a regular nucleosome organization consistent with previous publications; a nucleosome depleted region ( NDR ) is observed around the TSS and well-positioned nucleosomes flank this NDR . In MBD3 depleted cells , nucleosome phasing was similar to that in control cells but nucleosome occupancy was decreased - particularly at the NDR , and the −1 , +1 , +2 , +3 , and +4 nucleosomes . To test whether this effect is MBD3 dependent , we ranked promoters based on MBD3 occupancy ( Figure S6 ) . The changes at the NDR and +1nucleosome did not correlate with MBD3 occupancy and may be indirect . However , the occupancy pattern at −1 , +2 , +3 , and +4 nucleosomes differed substantially upon MBD3 depletion at promoters in the highest quartile while these same nucleosomes did not change upon MBD3 depletion in the lowest quartile . These data indicate that MBD3 regulates nucleosome organization , particularly near promoters and in gene bodies that have high MBD3 occupancy . Chromatin regulators are critical integration points wherein biological signals are converted into alterations in gene expression . These protein machines are essential to normal cell function , to development and to differentiation [26] . A large number of chromatin regulators are mutated in cancer , highlighting the importance of their function in normal cells [27] . Critical to understanding the biology of these regulators is determination of their sites of accumulation , and presumably of their action , within the genome . Here , we have utilized two complementary techniques to address the localization of MBD3 , and by extension NuRD complex , arriving at a robust and reliable location map . These data associate MBD3 with previously undescribed genomic features , including extensive colocalization with the bodies of active genes . Further , an abundant category of MBD3 localization observed was not associated with any particular genomic feature , histone or DNA mark we analyzed , suggesting that the catalog of functions for NuRD is not yet exhaustive . Finally , we assessed the contribution of MBD3 to chromatin organization , finding that MBD3 regulates nucleosome organization near promoters and within gene bodies - consistent with its localization . The use of multiple techniques for location analysis of chromatin associated factors provides an opportunity to control for common technical problems inherent to a single protocol . DamID involves expression of a fusion protein that modifies DNA in its genomic vicinity with detection relying on creation of novel restriction sites . It suffers from poor spatial resolution relative to ChIP and the necessity of exogenous expression of a fusion protein that must faithfully recapitulate the biological properties of the unmodified factor . Chromatin immunoprecipitation relies on biochemical fractionation of chromatin , in many cases following fixation . High quality ChIP results are reliant on antibody affinity and specificity as well as on crosslinking conditions and biochemical fractionation methods . Our results using these two techniques are in excellent agreement , suggesting they are converging on a robust genomic location map for MBD3 in the system chosen . Given the poor concurrence and quality issues ( Table S5 ) of recent NuRD ChIP-seq data [12] , the convergence of location determined by independent techniques provides clarity to the question of where the enzyme is enriched in the genome . Localization of MBD3 at promoter regions of active genes marked by H3K4me3 was unexpected given biochemical experiments documenting the failure of NuRD to productively interact with H3K4 methylated peptides [28] , [29] . Importantly , the current data are in excellent agreement with CHD4 ChIP-seq experiments performed in thymocytes by Katia Georgopoulos and colleagues [4] and with ChIP-seq of exogenously expressed MBD3 in HeLa cells [30] and in murine ES cells [31] . MBD3 enrichment with active histone marks at regions with the characteristics of enhancers underscores the surprising association of NuRD with the active fraction of the genome , in agreement with reports on CHD4 localization in K562 cells [13] and in murine ES cells [10] . Given that the half-life of acetylated histones in the active fraction of the genome is less than 5 minutes [32] , it is not surprising that histone deacetylases , including NuRD , may accumulate there [33] . Presumably , the dynamic equilibrium between the acetylated and deacetylated state for histones , or other chromatin associated factors , is an important determinant of promoter/enhancer function and its regulation . While association of MBD3 with active promoters was abundant in our data , we also observed accumulation in regions with local chromatin marks diagnostic of transcriptional repression . A significant number of promoters marked by H3K27me3 in MCF-7 cells also bore a peak of MBD3 , in agreement with ChIP-seq for CHD4 in murine ES cells [11] . Somewhat surprisingly , we did not observe substantial colocalization of MBD3 with H3K9me3 ( although we did observe that a substantial proportion of promoters bound by H3K9me3 are also bound by MBD3 ) , despite extensive biochemical data documenting specific interaction of the PHD finger domains of CHD4 with this mark [34] . While we do not completely understand the nature of this discrepancy , it may reflect the propensity of this histone mark to be localized in the repetitive fraction of the genome . The biochemical data also indicate high affinity interaction of CHD4's PHD fingers with H3K9 acetylation [35] , which agrees very nicely with our ChIP-seq data . It is interesting to speculate that association of the PHD1/2 domain of CHD4 with different modifications , both of which change the physical properties of a single lysine residue on histone H3 , may be instrumental in directing NuRD to regions of the genome with completely opposing functional states . The methyl CpG binding domain family is intimately tied to cytosine modification [36] . MBD3 is a most interesting member of this family , being a bona fide methyl-CpG binding factor in some , but not all , taxa [37] . Whether MBD3 can sense cytosine modification remains a matter of some contention in the literature; some investigators describe interactions with 5-hydroxymethyl C [9] , others do not [21] , [22] , [31] . Here , we describe enrichment for MBD3 at CpG islands that have extremely low levels of cytosine modification as measured by reduced representation bisulfite sequencing which does not distinguish between methylation and hydroxymethylation . We interpret this data as supporting the model that mammalian MBD3 does not recognize cytosine methylation or hydroxymethylation [21] , [31] and is preferentially bound at CpG islands with low levels of cytosine modification . Collectively , MBD3 localization supports roles for NuRD complex in regulation of chromatin structure and/or protein modification status at promoters of active genes , at enhancers , at stably repressed genes , and in bodies of actively transcribed genes . Functional data reported here document a role for MBD3 , and by extension NuRD , in nucleosome organization , a critical determinant of chromatin structure . Further , they predict novel functions that remain to be described . These predictions clearly indicate that models describing NuRD as a static corepressor are inadequate in the face of emerging genomic data . Rather , it seems likely that NuRD is involved at multiple levels in modulation of epigenetic features to facilitate chromatin biology . Recently , whole-exome sequencing revealed high-frequency deletion of a short segment of chromosome 19 containing the MBD3 locus in uterine serous carcinoma and frequent point mutations in CHD4 in serous endometrial tumors , suggesting fundamental functions of NuRD in primary tumors [38] , [39] . Future challenges for the field include defining modes of local enrichment at specific genomic features as well as functional studies to describe the nature and extent of enzymatic and non-enzymatic actions of NuRD complex on the chromatin fiber . MCF-7 , MDA-MB-231 , and 293T cells were obtained from the American Type Culture Collection and cultured at 37°C , 5% CO2 in Dulbecco's Modified Eagle Medium/Nutrient Mixture F-12 ( DMEM/F-12 ) Media containing 10% fetal bovine serum supplemented with penicillin-streptomycin . DamID lentiviral vectors , pLgw-RFC1-V5-EcoDam and pLgw-V5-EcoDam were kindly provided by Dr . Bas van Steensel , Netherlands Cancer Institute . Human MBD3 cDNA ( BC043619 ) was amplified and cloned into pENTR/D-TOPO and then recombined by an LR-reaction into destination vector pLgw-RFC1-V5-EcoDam ( pLgw-MBD3-V5-EcoDam ) . Retroviral knockdown constructs , pSMP-Luc ( Addgene plasmid 36394 ) and pSMP-MBD3_3 ( Addgene plasmid 36373 ) were obtained from Addgene . All constructs were verified by DNA sequencing . Lentivirus production and infection were performed as previously described [40] , using pLgw-MBD3-V5-EcoDam ( MBD3-Dam ) or pLgw-V5-EcoDam ( Dam-only ) . Retrovirus production and infection were performed as described [41] , using pSMP-Luc or pSMP-MBD3_3 . The DamID experiments were carried out as previously described [42] . Briefly , MCF-7 or MDA −231 cells were seeded into 6-well plates . Seventy-two hrs after infection , genomic DNA was isolated using Qiagen DNeasy tissue kit . Genomic DNA ( 2 . 5 µg ) was digested with Dpn I followed by adaptor ligation . The ligated product was digested with Dpn II and amplified by PCR . One microgram amplified product was labeled using Dual-color DNA Labeling kit ( Nimblegen ) according to manufacturer's protocol and then hybridized to Nimblegen 2 . 1M Deluxe promoter array or human 2 . 1 M Whole-Genome Tiling array ( Array 5 of 10 covering chromosomes 6–8 ) and washed following the manufacturer's directions . The slides were scanned using a DNA microarray scanner ( G2565BA; Agilent Technology ) and the images were processed with the Nimblegen software . A two-step normalization approach was used , where the first step is designed to correct for GC bias and dye bias within a chip ( intrachip correction ) and the second step corrects for variations across chips ( interchip correction ) . The first step was within-chip normalization . First , all probes were binned according to their GC content . The GC content was computed as a ratio of C and G nucleotides to the total number of nucleotides in the probe sequence . The overall variability in GC content values was used to compute bin width according to zero-stage rule [43] , [44] . These bin widths are proven to be approximate L2 optimal; i . e . , they minimize mean integrated square error . The bins with fewer probes were then merged so that each bin contains at least 500 probes . Within each bin , Lowess regression was used to predict log-transformed cy5 values as a smooth function of log-transformed cy3 values [45] , [46] . The scaled ( median of absolute residuals is used for scaling ) difference between observed and predicted log ( cy5 ) values were used as normalized signal . The second step was between-chip normalization . Once the data were corrected for dye and GC bias as described in the first step , we employed quantile normalization independently for each histone mark and DNA methylation to correct for between-chip variation . The differentially bound “peak” regions for each cell type comparisons were identified using modified ACME algorithm that allows for spooling data across replicates [46] . This algorithm like ACME , depends on three user-specified tuning parameters: window size ( w ) , signal threshold ( s ) and p-value threshold ( p ) . To identify MBD3 bound peaks , we first compute the number ( x ) of signal values within window of size w ( centered at probe ) that are greater than 100sth percentile across all replicates . Next , we compute enrichment p-value for probe using hypergeometric distribution as followingwhere N denotes total number of probes , k denotes number of probes in window and r denotes the number of replicates . Finally , the peaks are identified as runs of enrichment p-values that are less than p-value threshold ( p ) . The analysis presented here correspond to signal threshold ( s = 0 . 95 ) , window size ( w = 2000 ) and p-value threshold ( p = 0 . 001 ) with peaks containing less than six probes excluded . For each set of differentially bound and cell-specific gene signatures we performed Fisher's exact test to assess enrichment of gene-sets from Molecular Signature Database ( MSigDB , version3 . 0 , Broad Institute ) and other published Cancer gene sets . The resulting significance p-values were subjected to Benjamin-Hochberg ( FDR ) multiple test correction . We employed chromosome bound circular permutations test described in [46] to assess whether the observed overlap between two sets of genomic intervals ( e . g . peaks ) is significantly enriched or depleted compared to expected overlap . To perform overlap analysis , we used disjoined probes from the promoter-array as a unit of overlap . The ‘member status’ ( ie . whether the disjoined probe belongs to a genomic region of interest or not ) of disjoined promoter probes in one of genomic interval set was permuted 20 , 000 times using chromosome bound circular permutation as following . For each permutation a randomly generated number drawn from uniform distribution between 1 and number of disjoined probes from the analyzed chromosome was used to shift the membership status of all disjoined probes within a chromosome . The permutations resumed at the beginning of chromosome when the shift of status exceeded number of disjoined probes available in a chromosome . The odds-ratio of enrichment/depletion , corresponding significance p-value and 95% confidence interval were calculated by comparing observed overlaps and average overlap across all 20 , 000 permutations . The above approach was utilized to assess enrichment/depletion of overlap in significant DamID peaks ( modified ACME peaks with at least 6 promoter probes ) and MBD3 ChIP-seq peaks ( peaks with fold change> = 2 and false discovery rate< = 0 . 00001 ) . We also utilized the above approach to assess enrichment/depletion of overlap in significant cell specific DamID peaks ( uncommon peaks in MCF-7 and MDA-231 cells ) with promoter regions ( 3 kb upstream and 3 kb downstream of TSS ) of Luminal vs . Basal breast cancer differentially expressed genes from Charafe et al [47] . ChIP experiments for MBD3 were performed as previously described with several modifications [17] , [18] . MCF-7 or MDA-231 cells were crosslinked with 1 . 5 mM disuccinimidyl glutarate ( Thermo Scientific ) for 45 min at room temperature followed by 1% formaldehyde ( Sigma ) for 10 min at room temperature . Fixation was quenched by 125 mM glycine . Cells were washed twice with ice-cold PBS and stored at −80°C . Cells were thawed on ice and resuspended with 270 µL lysis buffer [50 mM Tris-HCl ( pH 8 . 0 ) , 10 mM EDTA , 1% SDS and protease inhibitor] . After incubation on ice for 5 min , cells were sonicated with Bioruptor for 6 cycles of 30 sec on and off at HIGH setting . Samples were centrifuged at 16 . 1 kg for 5 min at 4°C . Supernatant was collected into new tube and pellet was resuspended with 150 µL lysis buffer . The resuspended pellets were sonicated with Bioruptor for 4 cycles of 30 sec on and off at HIGH setting . Samples were centrifuged at 16 . 1 kg for 5 min at 4°C and supernatant were combined . 100 µL supernatant was diluted 10 times with IP buffer [16 . 7 mM Tris-HCl ( pH 8 . 0 ) , 1 . 2 mM EDTA , 0 . 01% SDS , 1 . 1% TritonX-100 , 167 mM NaCl , 5 mg/mL BSA and protease inhibitor] and incubated with 4 µg of control rabbit IgG ( Santa cruz ) or anti-MBD3 antibody ( Abcam; ab91458 ) at 4°C overnight . Samples were further incubated with 20 µL of Dynabeads Protein A and G at 4°C for 1 . 5 hr . The beads were washed once with IP buffer , twice with RIPA buffer [25 mM Tris-HCl ( pH 7 . 6 ) , 150 mM NaCl , 1% NP-40 , 1% sodium deoxycholate , 0 . 1% SDS] , twice with high salt RIPA buffer ( 1∶9 dilution of 5M NaCl with RIPA buffer ) , twice with LiCl buffer [20 mM Tris-HCl ( pH 8 . 0 ) , 250 mM LiCl , 1% NP-40 , 1% sodium deoxycholate , 2 mM EDTA] and twice with TE buffer [10 mM Tris-HCl ( pH 8 . 0 ) , 0 . 1 mM EDTA] . Washed beads were resuspended with elution buffer ( 1% SDS and 0 . 1 M sodium bicarbonate ) and incubated at 65°C for 45 min . Eluted DNA was adjusted to 300 mM NaCl and incubated with RNase at 65°C for 4 hr followed by incubation at 65°C for 1 hr with proteinase K . DNA was purified with a MinElute PCR purification kit ( QIAGEN ) . Sequencing libraries were prepared with Truseq DNA kit or Nextera XT kit ( following the manufacturer's protocols ) . MBD3 ChIP-seq and Input libraries were sequenced using Illumina GAIIx technologies at the NIH Intramural Sequencing Center ( MCF-7 MBD3 ChIP , MCF-7 Input and MDA-231 Input ) or using MiSeq technologies at the NIEHS Epigenomics Core Facility ( MDA-231 MBD3 ChIP and replicate Input ) . Standard Illumina CASAVA 1 . 8 utilities were used to generate . fastq output files . All libraries were sequenced as single end 36mers . ChIP experiments for H3K4me3 in MDA-231 cells were carried out using a Magna ChIP kit ( Millipore ) following the manufacturer's suggested protocols . For each ChIP experiment , 5×106 MDA −231 cells were crosslinked with 1% formaldehyde at 37°C for 10 min . The anti-H3K4me3 ( Upstate , no . 04-745 ) antibody was used . The ChIP-seq libraries were prepared from 10 ng of both Input and ChIP DNA samples using a ChIP-seq sample preparation kit ( Illumina ) according to the manufacturer's protocol . MDA-231 H3K4me3 ChIP-seq and Input libraries were sequenced using Illumina GAII technologies at the University of Missouri DNA Core Facility . For these libraries Illumina CASAVA 1 . 3 utilities were used to generate . fastq output files . All libraries were sequenced as single end 36mers . Public MCF-7 data used for analyses included: H3K4me3 ( GSM945269 ) , H3K9me3 ( GSM945857 ) , H3K27ac ( GSM945854 ) , H3K27me3 ( GSM970218 ) , Reduced Representation Bisulfite ( GSM683787 ) , and triplicate expression data ( GSM425734 , GSM425735 , GSM425736 ) and ( GSM425737 , GSM425738 , GSM425739 ) . Sequenced reads from MDB3 ChIP-seq and Input libraries were combined for replicate samples and filtered based on a mean base quality score <20 . Filtered reads were then aligned to the human reference genome ( UCSC assembly hg19 , GRCh37 ) using the Bowtie short-read alignment program ( v0 . 12 . 8 employing parameters −v 2 , −m 1 ) to retain reads mapped to unique genomic locations with at most 2 mismatches . Only non-duplicate reads were used in subsequent peak calling analyses and the generation of coverage tracks . To make the coverage tracks , aligned reads were extended at the 3′ end to a length of 300 bases ( the expected genomic fragment size ) , and bigWig files were generated to visualize aggregate genomic coverageMBD3 peaks for each cell type were identified using SICER with a FDR threshold of 0 . 001 and the following parameters ( redundancy threshold = 1 , window size = 200 , gap size = 600 , fragment size = 300 ) . Basic quality control of MBD3 ChIP-seq data ( ours and others ) is provided in Table S5 . hg19 RefSeq transcript model definitions were downloaded from UCSC . Transcripts covering duplicated genomic loci [have exactly same transcription start site ( TSS ) and end site ( TES ) ] were excluded yielding 31 , 723 RefSeq transcripts . ChIP-seq peaks were defined as associated with genomic features if they intersect at least 1 bp . Genomic features queried included exons , introns , TSS , TES , promoters ( +/− 3 kb from TSS ) , gene bodies ( TSS to TES ) , and intergenic regions ( not in gene body ) . From the set of 31 , 723 RefSeq transcripts , we selected full-length gene bodies ( i . e . from the transcript start to the 3′ transcript end ) larger than 2 kb . Read counts were averaged in 70 bp windows for regions +/− 5 kb from TSS . Within the gene bodies , reads were averaged in windows equal to 1% of the gene length . All window read counts were normalized by the total number of bases in each window and the length of the window . The CpG contents and the ratio of observed versus expected CpG dinucleotides were determined in a 1 , 001 bp window around TSS ( +/− 500 bp ) as described [48] . Promoter categories were then classified into 3 categories as follows: HCPs ( high-CpG promoters ) contain a 500 bp area with CpG ratio above 0 . 75 and GC content above 55%; LCPs ( low-CpG promoters ) do not contain a 500 bp area with a CpG ratio above 0 . 48; and ICPs ( intermediate CpG promoters ) are neither HCPs nor LCPs . Heatmaps represent transcript expression , binding profiles of MBD3 and H3K4me3 in MCF-7 and MDA-231 cells at promoter regions ( −7 to +3 kb relative to TSS ) and are sorted by MBD3 binding profiles . For DamID data , we iteratively filtered genes as followed: 1 ) Downloaded hg18/refseq gene table from UCSC browser site ( with date stamp October 20 , 2009 ) . 2 ) 31 , 451 genes from chr1-22 and chrX and chrY were retained . 3 ) 24 , 983 genes with non-missing gene expression signal were retained . 4 ) 15 , 292 genes with at least one microarray probe for each TSS bin were retained . 5 ) 11 , 872 genes with unique TSS were retained . 6 ) 8 , 207 genes with no 7 kb upstream and no 3 kb downstream neighbors were used in heatmap . For ChIP-seq data , we filtered genes as followed: 1 ) Transcripts with duplicated genome locus were eliminated . 2 ) Transcripts were selected with a MAD ( Mean Absolute Deviation ) higher than 1 . 0 , to further eliminate transcripts with low variability around promoter regions in MBD3-binding profiles; 25 , 569 transcripts ( MCF-7 ) and 23 , 605 transcripts ( MDA-231 ) were retained and used in the heatmap . Raw sequencing datasets for MBD3 and histone modifications H3K4me3 , H3K27me3 , H3K27ac , and H3K9me3 from ENCODE were filtered , mapped , deduplicated , and extended ( to 300 bp ) as described above . RPKM thresholds for establishing presence or absence of each mark were determined following the method described by Whyte et al for detecting super-enhancers [49] , but evaluating over all non-overlapping 1 kb genomic bins rather than called peaks . If one or more 1 kb bin ( evaluated with step size 100 ) in the region +/− 3 kb of an annotated RefSeq TSS had RPKM above the established threshold , presence of MBD3 or histone mark was assigned for that TSS . Comparison of MBD3 signal with H3K4me3 , H3K27ac , H3K9me3 , or H3K27me3 at all TSS is shown by Venn diagram . Called MBD3-binding peaks in MCF-7 cells were separated according to whether they overlap with a RefSeq promoter region ( TSS +/− 3 kb ) . Then a region with 3 kb flanking the center of peaks was extracted accordingly . Reads from MBD3 , H3K4me3 and H3K27ac profiles were averaged in 60 bp windows for the above region and normalized by the total mapped reads . CpG islands annotation was downloaded from UCSC hg19 genome , and CpG islands were separated according to whether they overlap at least 1 bp with the MBD3-binding peaks . Reduced Representation Bisulfite sequencing data in MCF-7 cells ( GSM683787 ) was downloaded and the methylation levels of the CpG islands associated with MBD3-binding peaks were calculated as the follows: percentage of methylation = total sequenced reads being methylated/total reads being sequenced . The interaction dataset , part of the Pol II Chromatin Interaction Analysis with Paired-End Tag data ( Pol II ChIA-PET ) , GSM970209 , were downloaded from GEO database . Then the data were filtered to keep the PET pair ends that have at least one end overlapped with the TSS loci ( +/− 3 kb ) . Among the retained PET pair ends , each end in the PET pair ends was examined as follows: 1 ) Assign the nearest hg19 Refseq gene if this end is overlapped with TSS loci ( +/− 3 kb ) ; 2 ) Assess if this end was overlapped with the MBD3 peaks; 3 ) Count the reads from MBD3 ChIP-seq data within the region that is 500 bp flanking the center of this end . The reads were normalized to the total mapped reads . The data are displayed as follows: 1 ) Sort the read number at TSS end of the filtered ChIA-PET pair ends in a descending order; 2 ) Plot the sorted read number in upper panel: y-axis is the sorted read number and x-axis is the sorted PET pairs; 3 ) Plot the read number in lower panel: y-axis is the corresponding read number in the other end and x-axis is the sorted PET ends in the same order as in the upper panel . 3C experiment was performed as previously described [25] . Briefly , MCF-7 cells were crosslinked by 1% formaldehyde for 10 min at room temperature and fixation was quenched by 125 mM glycine . Cells were resuspended with lysis buffer ( 10 mM Tris-HCl , 10 mM NaCl , 0 . 2% NP40 ) and incubated on ice for 15 min . The pellet was resuspended with 1 . 2×NEB2 buffer and incubated at 65°C for 15 min and followed by digestion with EcoRI overnight . Digested chromatin was ligated with T4 DNA ligase ( NEB ) at 16°C for 2 hr . Purified 3C template ( 200 ng ) was amplified with iQ SYBR mix ( BioRad ) . Mononucleosomal DNA were prepared from MCF-7 cells ( Luc or MBD3 knockdown ) as previously described with some modifications [50] . Briefly , nuclei were collected from 5×105 cells and digested by MNase ( 1 . 25 , 2 . 5 , or 5 units ) for 5 min at 37 °C . Mononucleosomal DNA were collected by gel size selection and pooled . Sequencing libraries were prepared from 1 µg of pooled DNA with TruSeq DNA Sample Preparation Kit . The resulting libraries were sequenced on a HiSeq 2000 ( Illumina ) as paired end 101mers . To ensure that low quality reads were excluded from the analysis , the raw sequence reads were filtered to remove any entries for which either read in a pair had mean base quality score <20 . Filtered reads were aligned to the human genome ( GRCh37/hg19 ) via Bowtie ( v0 . 12 . 8 with parameters -m 1 –best –strata –chunkmbs 1024 -I 0 -X 1000 ) [51] . Multiple sequencing lanes from the same MNase library were merged prior to downstream analysis , and duplicate reads per library were removed using MarkDuplicates . jar from the Picard tools package ( v1 . 86 ) ( http://picard . sourceforge . net ) . To capture only sequenced reads corresponding to the size of a mononucleosome , filtering was applied to retain only fragments 120–180 bases in length; 90–95% of mapped , deduplicated read pairs per library passed this filter . Each mapped read pair was then converted to a single point at the midpoint of the fragment . The two replicate libraries per condition were merged , and ‘bedGraph’ files were generated ( via BEDtools , version 2 . 17 . 0 [52] ) for visualization of genomic distribution of mapped fragments . Metagene plots of nucleosome occupancy were generated with mapped MNase-seq fragments downsampled to the same read count per condition ( N = 156 , 822 , 505 ) . Counts per position were aligned relative to each annotated RefSeq TSS , with the RefSeq loci split into quartiles based on the number of MBD3 ChIP-seq reads mapped to the region +/− 3 kb of the given TSS . The positional counts in each dataset ( shMBD3 or shLuc per quartile ) were normalized by the mean score over the region +/− 5 kb of TSS , then smoothed with a moving average ( period = 50 ) . The microarray data and sequence data from this study are available at the NCBI Gene Expression Omnibus ( GEO ) under accession number , GSE44737 , GSE 44776 , and GSE51097 .	Chromatin structure is tightly regulated by multiple mechanisms; its dysregulation is associated with developmental abnormalities and disease . The Mi-2/nucleosome remodeling and histone deacetylase ( NuRD ) complex is proposed to regulate chromatin structure by changing the location and/or the chemical properties of the fundamental building block of chromatin , the nucleosome . NuRD has been shown by genetics to be important for normal development , yet the detailed mechanism of how NuRD regulates chromatin structure is still unclear . Here , we study the localization and function of MBD3 , a component of NuRD , in two human breast cancer cell lines using two independent genomic technologies . Our data demonstrate that existing models , which associate NuRD with transcriptional repression , are not completely correct . Rather , MBD3 showed cell-type specific localization at active genes . Moreover , we found a previously unidentified localization of MBD3 across gene bodies and identified a regulatory role for MBD3 in nucleosome organization . Our data provide a reliable starting point from which to address mechanisms by which NuRD controls chromatin structure and nuclear biology .	[ "Abstract", "Introduction", "Results", "Discussion", "Methods" ]	[]	2013	MBD3 Localizes at Promoters, Gene Bodies and Enhancers of Active Genes
Insecticide-based methods represent the most effective means of blocking the transmission of vector borne diseases . However , insecticide resistance poses a serious threat and there is a need for tools , such as diagnostic tests for resistance detection , that will improve the sustainability of control interventions . The development of such tools for metabolism-based resistance in mosquito vectors lags behind those for target site resistance mutations . We have developed and validated a simple colorimetric assay for the detection of Epsilon class Glutathione transferases ( GST ) -based DDT resistance in mosquito species , such as Aedes aegypti , the major vector of dengue and yellow fever worldwide . The colorimetric assay is based on the specific alkyl transferase activity of Epsilon GSTs for the haloalkene substrate iodoethane , which produces a dark blue colour highly correlated with AaGSTE2-2-overexpression in individual mosquitoes . The colour can be measured visually and spectrophotometrically . The novel assay is substantially more sensitive compared to the gold standard CDNB assay and allows the discrimination of moderate resistance phenotypes . We anticipate that it will have direct application in routine vector monitoring as a resistance indicator and possibly an important impact on disease vector control . Prevention of mosquito-borne diseases depends in large part on vector control and usually involves the use of insecticides . Insecticide-based methods include insecticide-impregnated bed nets , indoor or aerial sprays and water treatments . Pyrethroids and the organochlorinated insecticide DDT ( 1 , 1 , 1-dichloro-2 , 2-bis ( p-chlorophenyl ) ethylene ) are the preferred choice for Indoor Residual Spraying ( IRS ) and have been used extensively for many decades for the control of disease vectors . Despite environmental concerns , DDT remains one of the cheapest and most effective long-term weapons against vector borne diseases in several stable endemic areas [1] . Although wide scale insecticide implementation has led to impressive decreases in vector borne disease transmission , the emergence and spread of insecticide resistance poses a serious threat and there is a need for new tools that will improve the sustainability of current control interventions [2] . Understanding resistance mechanisms and developing simple diagnostic tests for the early detection of insecticide resistance is an important prerequisite for the application of resistance management strategies . Insecticide resistance in disease vectors has been attributed to increased rates of insecticide detoxification or mutations in the target sites [3] . Increased rates of glutathione transferase ( GST ) - mediated DDT dehydrochlorination confers resistance to DDT in several mosquito species , such as Aedes aegypti , the major vector of dengue and yellow fever worldwide , and Anopheles gambiae , the major malaria vector in sub-saharan Africa [4] , [5] . This DDT detoxification reaction is catalysed by the Epsilon class GST , GSTE2-2 in An . gambiae , An . cracens and Ae . aegypti mosquitoes from different geographical origins [4] , [5] , [6] . Detection of metabolism – based insecticide resistance is more complex than screening for specific mutations known to cause target site resistance . Current techniques for measuring elevated GSTE2-2 levels in mosquitoes , such as real time PCR or specific ELISA based on antibodies are elaborate or require the use of expensive equipment and consumables , and are therefore not accessible to laboratories on a limited budget [4] , [5] . Biochemical assays for detecting metabolic resistance generally employ generic substrates that are recognised by most or all members of the enzyme families . For example , GST activity is usually measured using 1-chloro-2 , 4-dinitrobenzene ( CDNB ) , 1 , 2-dichloro-4-nitrobenzene ( DCNB ) , and , more recently , monoclorobimane [7] , [8] . Unlike assays to detect elevated esterase activity , which can be read by eye , the current GST assays require a spectrophotometer that can measure absorbance in the UV range , or fluorimeter with multiple emission/excitation channels [8] , [9] , limiting their applicability in the field . Potentially greater sensitivity and specificity could be achieved if substrates that were specifically recognised by the enzyme ( s ) responsible for insecticide metabolism were employed . A colorimetric assay for GSTs with alkyl transferase activity , capable of catalysing the release of iodine from haloalkene substrates , has been recently described [10] . Using a modified version of this assay , Dowd et al . [11] screened a large number of recombinant mosquito GSTs for alkyltransferase activity with several haloalkene substrates , to identify potential enzyme biosensors for detecting insecticides . Recombinant epsilon GSTs , but not the delta or sigma GSTs , which are the most abundant in insects [12] , showed a remarkable ability to utilise iodoethane as a substrate and produce a dark blue colour , which can be measured spectrophotometrically or visually [11] . Here , we have adapted the alkyl transferase/iodoethane -based colorimetric assay to measure GST activity associated with DDT resistance in individual Ae . aegypti mosquitoes . Six Ae . aegypti mosquito strains were used in this study: The standard laboratory reference strain ( New Orleans ) was kindly provided by the Center for Disease Control and Prevention ( CDC ) , Atlanta , USA , the susceptible Ivory Coast strain was collected from Cote d'Ivoire , the Iquitos strain originating from Peru and the Solidaridad , Isla Mujeres and Merida strains , from Mexico , were kindly provided by Prof . William Black ( Colorado State University , USA ) . All strains were reared under standard conditions ( 28°C±2°C , 80% RH ) at the Liverpool School of Tropical Medicine . Bioassays were performed on 1–3 day old adults using the World Health Organization ( WHO ) adult susceptibility test papers – DDT 4% [7] . The time causing 50% mortality ( LT50 ) was obtained 24h after the exposure . Cloning into a pET3a vector , expression in Escherichia coli BL21 ( DE3 ) plysS , and purification of Ae . aegypti recombinant Epsilon GSTs were conducted as described previously [5] , [13] . The eluted enzyme was concentrated using a Vivaspin 15R concentrator and exchanged using a PD-10 column into 50 mM sodium potassium phosphate ( pH 7 . 4 ) , 10 mM dithiothreitol , and 40% glycerol according to the manufacturer's instructions and samples were stored at −80°C , until used . Mosquitoes were homogenised in 0 . 1M Tris-HCl , pH 8 . 2 ( 20 µl per individual ) , the mixture was centrifuged at 14 , 000×g for 10 min at 4°C , and the supernatant was used as the enzyme source for the biochemical assays . Standard GST spectrophotometric assays were performed by monitoring the formation of the conjugate of CDNB or 1 , 2-dichloro-4-nitrobenzene ( DCNB ) , and reduced glutathione ( GSH ) [9] . The iodide-releasing reaction was carried out as previously described [10] and optimised by Dowd et al . [11] , with GSH ( 2 . 5 mM ) and iodoethane ( 2 . 5 mM ) in 0 . 1M phosphate buffer pH 8 . 2 and enzyme source in a total volume of 100 µl at 25°C . The reaction was incubated at 30 min , or for different periods of time depending on the reaction rate studied during optimisation stages . Blue colour developed immediately after addition of 50 µl starch solution ( 0 . 25 g partially hydrolysed potato starch in 25 ml of Milli-Qwater and boiled in a microwave oven until all starch has dissolved ) and 100 µl acidified peroxide solution ( 2% H2O2 in 2 mM HCl ) . The blue colour was quantified spectrophotometrically at 610 nm using a VERSAmaxTM microplate spectrophotometer ( Molecular Devices , Sunnyvale , CA , USA ) , or estimated visually by eye . A standard curve was prepared from different concentrations of KI in 0 . 1M Tris–HCl buffer , pH 8 . 2 . Specific activities towards iodoethane were calculated from the linear range of the enzymatic reaction , and a plot of absorbance at 610 nm against potassium iodide concentration . They are expressed as µmole of iodide released /min/mg . All measurements were made in triplicate . Protein concentrations were measured using Bio-Rad protein assay reagent with bovine serum albumin as the protein standard [14] . Mosquito extracts ( 0 . 060 mg total protein ) were analysed with SDS-polyacrylamide gel electrophoresis ( 10% acrylamide running gel and 4% acrylamide stacking gel ) and electroblotted onto polyvinylidene difluoride membrane . The membrane was probed for 2 hours with an anti-AaGSTE2-2 antibody at 1∶1000 dilution in 3% milk-PBS-Tween solution and for 1 hour with a peroxidase-labelled anti-rabbit antibody at 1∶10000 dilution . Immunoreactive proteins were visualised using a horseradish peroxidase sensitive ECL chemiluminescent Western blotting kit ( GE Healthcare ) . We recently showed that , unlike other GST members tested , epsilon GSTs can very efficiently utilise the haloalkene iodoethane as a substrate [11] . In order to determine the amount of mosquito protein required to measure GST activity in the visual range ( colour change ) and set the linear limits of the colorimetric assay , we tested different amounts of mosquito extracts ( 0 . 010–0 . 120 mg of total protein ) at time points between 5 and 60 min . The minimum amount of protein extract that gave a visible colour range in any mosquito strain , after 30min incubation period , was 0 . 030 mg ( Figure 1A ) . No visible colour change was observed for the reference susceptible strain New Orleans , even when much higher amounts of protein ( and longer incubation times up to 60 min , data not shown ) were included in the assay ( Figure 1A ) . The product/colour formation is linear for at least 30 min , when approximately 0 . 060 mg mosquito homogenate ( equivalent to ¾ of an individual Ae . aegypti female ) was assayed ( Figure 1B ) . The linear range of the reaction was not affected by temperature fluctuations between 25 and 35°C ( data not shown ) . The LT50 values of six Ae . aegypti mosquito strains following exposure in 4% DDT were determined ( Figure 2A ) . The susceptible New Orleans and Ivory Coast strains showed LT50 values of 20 min or less , whilst the Iquitos and Solidaridad strains exhibited LT50 values of 76 min and 100 min , respectively; accurate LT50 values could not be determined for Merida and Isla Mujeres strains , due to the very high levels of resistance ( LT50>300 min ) . To confirm the association of the AaGSTE2-2 enzyme with the resistance phenotype , we performed Western blot analysis , using crude mosquito homogenates probed with anti-AaGSTE2-2 antiserum . A single band of approximately 25 kDa was detected in all strains , with intensity levels highly correlated with the LT50 values/DDT resistance data ( Figure 2B ) . Using the optimised colorimetric assay , we determined the specific GST activity in adult females from several Ae . aegypti strains . As shown in Table 1 , there is a >15-fold difference in alkyl transferase activity between the highly DDT resistant Merida and Isla Mujeres strains , and the susceptible Ivory Coast strain . The alkyl transferase activity of the Iquitos and Solidaridad strains , which showed moderate resistance levels , was 4- and 12-fold higher , respectively , compared with the Ivory Coast strain ( Table 1 ) . A highly significant correlation was observed between the LT50s and the enzymatic activities obtained by the iodoethane/colorimetric assay ( R2 = 0 . 97 , P<0 . 01 ) . The difference in alkyltransferase activity between the different strains can be easily visualised by eye ( Figure 2C ) , via the effort of multiple individuals . This correlation between resistance phenotype and specific activity does not hold for the model substrates CDNB and DCNB . For CDNB there was significantly higher activity in the four resistant strains compared to the two susceptible stains ( Table 1 ) but no difference in activity between the moderately and highly resistant groups . For DCNB , the relationship was even less clear . For example no significant difference was observed between the moderate resistant strain Solidaridad and the New Orleans susceptible strain ( Table 1 ) . By screening a large number of recombinant mosquito GSTs for alkyltransferase activity with several substrates , Dowd et al . [11] showed that mosquito epsilon GSTs , AaGSTE2-2 and AaGSTE4-4 , can utilise the haloalkene iodoethane as substrate but that this substrate was not recognised by delta or sigma class GSTs . To determine whether the ability to catalyse the release of iodine from iodoethane was a general property of epsilon GSTs , we expressed six family members and measured their specific activity against this substrate . As shown in Table 2 , the highest activity was obtained with the DDTase AaGSTE2-2 ( 10 . 3 µmole iodide/min/mg ) . Other members of the Epsilon class also recognised this substrate but their specific activities were lower ( 0 . 03 to 4 . 3 µmole iodide/min/mg ) . AaGSTE8-8 exhibited the lowest activity , possibly due to the low amino acid identity ( approximately 30% ) of this gene with other members of this class [15] . The respective CDNB activities of the recombinant epsilon GSTs are also shown in Table 2 for comparison . The DDTase AaGSTE2-2 has lower or similar specific activity with CDNB compared to other members of the family and hence this substrate cannot specifically recognise GSTs implicated in insecticide resistance . We have developed a simple colorimetric assay for the specific detection of GST activity associated with DDT resistance in Ae . aegypti . The colorimetric assay is substantially more sensitive in detecting DDT resistance in Ae . aegypti , compared to the gold standard CDNB assay currently being used in routine mosquito resistance monitoring studies [7] . The differences in GST activities among strains with high , moderate or negligible resistance were over 15-fold for iodoethane , but only 1 . 5–4 . 3-fold for CDNB and DCNB substrates . In contrast to iodoethane , the latter general substrates failed to discriminate moderate resistance phenotypes ( Table 1 ) . This increased sensitivity of the novel colorimetric assay provides greater potential for the identification of resistance at early stages , a crucial pre-requisite for the implementation of evidence-based resistance management tactics . Unlike UV/spectrophotometric CDNB and DCNB assays , the alkyltransferase/iodoethane assay produces a dark blue colour that is both highly correlated with AaGSTE2-2-overexpression-based DDT resistance and can be estimated by eye at least semi-quantitatively ( Figure 2 ) . This novel assay can be performed by non-qualified personnel , without sophisticated equipment . It is robust at temperatures between 25–35°C , with a wide linear range of quantification , and a sensitivity which allows the measurement of GST activity in a single mosquito . The cost of the assay is less than 0 . 05 USD per mosquito , while the shelf life of the substrate iodoethane is at least 1 year at 4°C . Here , we have focused on Ae . aegypti , as DDT resistance is extremely high in many populations of this species in dengue endemic regions [16] . However , the assay can be adapted for measuring GSTE2-2/DDTase – based DDT resistance in other mosquito species , such as the major malaria vector An . gambiae . This was not tested here , as there were no suitable resistant strains available . Nevertheless , given that iodoethane is a very good substrate also for the orthologue enzyme AgGSTE2-2 ( data not shown ) and this enzyme is the key enzyme responsible for DDT resistance in this species [4] , there is no reason to believe that this assay will not work for Anopheles mosquitoes too . In conclusion , we describe a simple colorimetric test for the detection of the GSTE2-2/DDTase- based resistance in mosquitoes . It combines the most desirable features of specificity and sensitivity with the low cost and ease of use required for a routine test in endemic countries . We anticipate that the assay will have direct application in routine vector monitoring as a resistance indicator and help improve the sustainability of insecticide based control strategies .	Aedes mosquitoes transmit many human viral pathogens including dengue , yellow fever and chikungunya . Most of these pathogens have no specific treatment or vaccine and hence their control is reliant on controlling the mosquito vectors , which usually involves the use of insecticides . In order to prevent the alarming prospect of mosquito control failure due to the rapid selection and spread of insecticide resistance in several mosquito populations worldwide , it is essential that effective resistance management strategies are implemented and adhered to . The development of simple diagnostic tests for the early identification and monitoring of resistance is an important prerequisite for this task . Here , we describe the development of a simple colorimetric test for the detection of GSTE2-2/DDTase-based resistance in individual mosquitoes . The novel assay combines the most desirable features of specificity and sensitivity with the low cost and ease of use required for a routine test in endemic countries . It can have direct application in routine vector monitoring as a resistance indicator and help improve the sustainability of insecticide based control strategies .	[ "Abstract", "Introduction", "Materials", "and", "Methods", "Results", "Discussion" ]	[ "infectious", "diseases/viral", "infections", "ecology/population", "ecology", "biochemistry/experimental", "biophysical", "methods" ]	2010	A Simple Colorimetric Assay for Specific Detection of Glutathione-S Transferase Activity Associated with DDT Resistance in Mosquitoes
The family Arenaviridae comprises three genera , Mammarenavirus , Reptarenavirus and the most recently added Hartmanivirus . Arenaviruses have a bisegmented genome with ambisense coding strategy . For mammarenaviruses and reptarenaviruses the L segment encodes the Z protein ( ZP ) and the RNA-dependent RNA polymerase , and the S segment encodes the glycoprotein precursor and the nucleoprotein . Herein we report the full length genome and characterization of Haartman Institute snake virus-1 ( HISV-1 ) , the putative type species of hartmaniviruses . The L segment of HISV-1 lacks an open-reading frame for ZP , and our analysis of purified HISV-1 particles by SDS-PAGE and electron microscopy further support the lack of ZP . Since we originally identified HISV-1 in co-infection with a reptarenavirus , one could hypothesize that co-infecting reptarenavirus provides the ZP to complement HISV-1 . However , we observed that co-infection does not markedly affect the amount of hartmanivirus or reptarenavirus RNA released from infected cells in vitro , indicating that HISV-1 does not benefit from reptarenavirus ZP . Furthermore , we succeeded in generating a pure HISV-1 isolate showing the virus to replicate without ZP . Immunofluorescence and ultrastructural studies demonstrate that , unlike reptarenaviruses , HISV-1 does not produce the intracellular inclusion bodies typical for the reptarenavirus-induced boid inclusion body disease ( BIBD ) . While we observed HISV-1 to be slightly cytopathic for cultured boid cells , the histological and immunohistological investigation of HISV-positive snakes showed no evidence of a pathological effect . The histological analyses also revealed that hartmaniviruses , unlike reptarenaviruses , have a limited tissue tropism . By nucleic acid sequencing , de novo genome assembly , and phylogenetic analyses we identified additional four hartmanivirus species . Finally , we screened 71 individuals from a collection of snakes with BIBD by RT-PCR and found 44 to carry hartmaniviruses . These findings suggest that harmaniviruses are common in captive snake populations , but their relevance and pathogenic potential needs yet to be revealed . The first member of the family Arenaviridae , lymphocytic choriomeningitis virus ( LCMV ) , was identified and isolated already in the 1930s [1] . During the following four decades several novel members of the family were identified including human pathogens such as Junin ( JUNV ) , Machupo ( MACV ) , and Lassa ( LASV ) viruses [1] . For decades , arenaviruses were known as rodent-borne viruses with the exception of Tacaribe virus ( TCRV ) , which was isolated from a bat [1] . In the early 2010s , three independent groups identified novel arenaviruses as the potential causative agents for boid inclusion body disease ( BIBD ) [2–6] . BIBD is characterized by intracellular cytoplasmic inclusion bodies ( IB ) within almost all cell types of affected snakes [4 , 7 , 8] . The IB mainly ( or solely ) consist of arenavirus nucleoprotein ( NP ) [4 , 8] , and BIBD was recently successfully reproduced by experimental reptarenavirus infection [9] . The identification of the “snake arenaviruses” prompted the establishment of two new genera , Mammarenavirus and Reptarenavirus , within the family Arenaviridae [1] . Snakes with BIBD often , if not always , carry L and S segments of several reptarenavirus species [10 , 11] . Furthermore , infected snakes usually harbor more L than S segments , which significantly hampers the taxonomic classification of reptarenaviruses [10–12] . The International Committee on Taxonomy of Viruses ( ICTV ) Arenaviridae study group has recommended that the PAirwise Sequence Comparison ( PASC , available at ( https://www . ncbi . nlm . nih . gov/sutils/pasc/viridty . cgi ? textpage=overview ) tool should be used for genus and species determination [1] . The PASC tool classifies arenaviruses to the same genus if the nucleotide sequence identity in the S segment is >29–40% and >30–35% in the L segment [1 , 13] . When analyzing some of the virus isolates of our first paper on BIBD [4] , we found a virus genome with coding strategy similar to arenaviruses and named the isolate Haartman Institute snake virus-1 ( HISV-1 ) [10] . Analysis of the HISV-1 genome with the PASC tool showed that HISV-1 represents a novel arenavirus genus , and in 2018 the third genus , Hartmanivirus , was established in the family Arenaviridae [14] . Arenaviruses are RNA viruses with a single-stranded , bisegmented , negative-sense RNA genome and an ambisense coding strategy [1] . The large ( L ) genome segment encodes matrix/Z protein ( ZP ) and RNA-dependent RNA polymerase ( RdRp ) and the small ( S ) segment encodes glycoprotein precursor ( GPC ) and nucleoprotein ( NP ) [1] . Arenaviruses replicate in the cytoplasm of the infected cells , the genome replication and transcription requires both RdRp and NP [15] . Initially , the ZP was also thought to contribute to the latter processes [16] but later studies have demonstrated that ZP rather acts to suppress both [15 , 17] . All structural proteins of arenaviruses have essential roles in the arenavirus life cycle: RdRp is required for genome replication , GPC for spike formation to gain cell entry , NP for genome packaging and replication , and ZP for budding and regulation of replication [15 , 18] . Additionally , the NPs of all mammarenaviruses but TCRV inhibit type I interferon ( IFN-I ) induction [19] at multiple steps of the signaling pathway [20] . Likewise , the ZPs of mammarenaviruses that are pathogenic in humans inhibit IFN-I signaling by targeting RIG-I and MDA5 [20] . The ZPs also interact with cellular components such as PML ( promyelocytic leukemia protein ) , eIF4E ( eukaryotic translation initiation factor 4E ) , and the ESCRT ( endosomal sorting complexes required for transport ) system required for budding [20] . When assembling the genome of HISV-1 we observed that the L segment lacks an open-reading frame ( ORF ) for the ZP [10] . At that point , we did not have a pure HISV-1 isolate and we could thus neither confirm the latter finding nor could we investigate whether HISV-1 would survive without co-infecting reptarenavirus ( es ) . Herein , we report the isolation and characterization of a pure HISV-1 cell culture isolate demonstrating that infectious virions are produced despite the lack of ZP . We also identified three additional hartmanivirus species and provide the complete coding regions for their genomes along with a number of nearly complete reptarenavirus genome segments . We identified hartmaniviruses in snakes with BIBD , but could also detect hartmanivirus infection in apparently healthy snakes , suggesting that these viruses are not directly linked to BIBD pathogenesis . Even though we could not associate hartmanivirus infection with pathological changes in vivo , we observed cytopathic effects of HISV-1 infection in vitro . Previously we used next-generation sequencing ( NGS ) for the characterization of reptarenavirus isolates [4] . This led to the identification of HISV-1 , a putative representative for novel arenavirus genus , in addition to several reptarenavirus S and L segments [10] . Even though we had fairly high coverage of the S ( 11–27092 fold ) and L ( 357–4493 fold ) segments of HISV-1 , we were unsure whether these represented the full length segments since the L segment comprised an ORF for the RdRp [10] , but the ORF for the ZP found in other arenaviruses was missing . Since the original HISV-1 preparation contained also a reptarenavirus ( UHV-2 , University of Helsinki virus-2 ) , we used serial dilution to obtain single virus isolates of both UHV-2 and HISV-1 ( S1 Fig ) . Successful production of a clean HISV-1 isolate indicated that even though an ORF for the ZP , which functions as the matrix protein , cannot be found in the L segment , HISV-1 is able to replicate without a co-infecting reptarenavirus . For most subsequent comparative experiments we used UHV-2 as the reference reptarenavirus , since it was the co-infecting reptarenavirus of the original HISV-1 isolate . For a few experiments we used UGV-1 ( University of Giessen virus-1 ) instead , as this is a reptarenavirus grows to high titers in our cell culture model . For the sequence analyses we chose to use the type species of each arenavirus genera . To confirm the lack of an ORF for the ZP in the L segment of HISV-1 , we isolated RNA from a batch of purified HISV-1 and sequenced the ends of the S and L segments using T4 RNA ligase to generate cyclic RNAs . The latter then served as templates for RT-PCR over the genome ends ( S2 Fig ) and yielded several sequences that indeed covered the genome ends of both the S and L segment . In addition to the S and L segment specific primer pairs , we also successfully applied primer combinations , i . e . L segment forward-S segment reverse and S segment forward-L segment reverse , in RT-PCRs , which suggested that T4 RNA ligation also produced S and L segment chimeras . This was confirmed by sequencing . The genome end sequencing then confirmed that the L segment of HISV-1 indeed lacks the matrix protein/ZP found in other arenaviruses . Subsequent NGS and de novo genome assembly for the purified HISV-1 preparation did not identify additional genome segments . The consensus sequence of the S segment revealed a nucleotide insertion ( a stretch of 7 instead of 6 adenines at 575–581 ) in the GP1 ORF of our original submission [10] , which led to incomplete “in silico” translation of the GPC . Obtaining the full length genome segments of HISV-1 allowed us to compare the genomes of the three arenavirus genera , Mammarenavirus , Reptarenavirus , and Hartmanivirus . Since this is the first time that full length genome segments of all arenavirus genera are available , we decided to perform a bioinformatics-based comparison of their genomes and proteomes , and selected the type species of each genus for these “in silico” comparisons . The S segments of all three genera , schematically depicted in Fig 1A , are identical in their coding strategy and similar in size . However , while the L segments of mammarenaviruses and reptarenaviruses share the same coding strategy and are similar in size , the L segment of hartmaniviruses lacks the ZP ORF ( Fig 1A ) . As shown in Fig 1B , the genome ends–represented by 21 terminal nucleotides–of all arenavirus species have the ability to form a panhandle structure . The panhandles of LCMV and Golden Gate virus ( GGV ) L segments comprise 18 consecutive complementary nucleotides , while the corresponding region in the HISV-1 L segment contains two non-paired nucleotides . Comparison of the genome segment ends revealed a conserved CGCACxGxGxA motif at the 5´ end of mammarena- , reptarena- , and hartmanivirus S and L segments ( shown in bold in Fig 1C ) . Similarly , the 3´ends show conserved nucleotides , GCGUGxCxCCU ( shown in bold in Fig 1C ) , complementary to those found at the 5´ end with the underlined residue making an exception . While the predicted panhandle structures may differ even between viruses of the same species , the overall panhandle structure is maintained throughout arenavirus genera by the aforementioned conserved nucleotides . The terminal complementarity of the RNA is essential for replication and transcription [21] which , at least partially , may explain the conservation of the segment ends throughout the family . The non-conserved sites at both ends are speculated not to contribute to sequence-specific interaction with the RdRp [22] , thus explaining the observed variation at these sites . The proteomes and the amino acid identities between the corresponding proteins ( based on MAFFT alignment ) of the three genera are presented in S2 Table . The major difference in the proteomes is the lack of ZP in hartmaniviruses . While the ZPs of reptarenaviruses and mammarenaviruses share only 16% amino acid identity , their functions are assumed to be similar [2] . Interestingly , reptarenavirus ZPs have an N-terminal transmembrane helix ( TM ) [2 , 4] , whereas the N-terminus of mammarenaviruses is myristoylated [2] . The structural proteins of HISV-1 are all 20 to 23% identical to their LCMV and GGV counterparts . The RdRp and NP of GGV are slightly closer to LCMV ( 28% and 32% identical ) than to HISV ( 20% and 20% identical ) . However , the GPCs of HISV-1 and LCMV ( 23% identity ) are more similar than the GPCs of LCMV and GGV ( 16% identity ) . See S2 Table for more detail . The GPC of each arenavirus genera are schematically presented in Fig 2A . By prediction the GPCs contain several N-glycosylation sites: 7 in HISV-1 ( 4 in GP1 and 3 in GP2 ) and LCMV ( 5 in GP1 and 2 in GP2 ) and 9 in GGV ( 7 in GP1 and 2 in GP2 ) ( Fig 2A ) . The cleavage between GP1 and GP2 is mediated by subtilisin-kexin isozyme-1/site-1 protease ( SKI-1/S1P ) for mammarenaviruses ( Fig 2A ) . Using ELM ( eukaryotic linear motif resource , http://elm . eu . org/ , [23] ) we identified a potential SKI-1/S1P cleavage site in GGV GPC , and the GPC alignment of known reptarenavirus species suggests that the site is conserved ( Fig 2A ) . For HISV-1 we could not detect a SKI-1/S1P cleavage site but rather identified a potent furin cleavage site in the same region . By comparing the region to GPCs of the other hartmanivirus species ( described later in the manuscript ) we could show that the furin cleavage site is preserved among the hartmaniviruses found thus far ( Fig 2A ) . A more thorough investigation of the GPCs indeed shows similarities between HISV-1 and LCMV . To begin with , both HISV-1 and LCMV have long ( 55 and 58 residues , respectively ) signal peptides ( SP ) that by prediction are myristoylated at position 2 , while the SP of GGV is only 23 residues and lacks the myristoylation site ( Fig 2A ) . For mammarenaviruses the SP remains in the virion and is known as a stable signal peptide ( SSP ) that interacts with GP2 via an intermolecular zinc-binding motif [24] ( Fig 2C ) . Sequence comparison between mammarenavirus and hartmanivirus SSPs and GP2s ( Fig 2B ) shows conservation at the histidine and cysteine residues required for the SSP-GP2 interaction . Additionally , the GP2s of both mammarenaviruses and hartmaniviruses have a relatively long cytoplasmic tails ( 48 and 42 residues respectively ) while the GP2 tail of reptarenaviruses only comprises a few ( two in GGV ) residues ( Fig 2A and 2B ) . The above suggests that there might be differences in the organization of the spikes complex between reptarenaviruses and the other two arenavirus genera . Of note , the GP2 is the most conserved reptarenavirus protein , sharing 68–99% sequence identity between all the known species and 87–99% identity when CASV and UHV-1 are excluded . The GP2s of both mammarenaviruses and hartmaniviruses appear to be more variable . To monitor and characterize HISV-1 infection in cell culture , we produced an antiserum against the HISV-1 NP which we refer to as anti-HISV NP antiserum throughout the manuscript . We used the C-terminal half of the NP as the antigen , since this strategy had produced a broadly reactive anti-UHV-1 NP antiserum [10] referred to as anti-UHV NP antiserum throughout the manuscript . To test the anti-HISV NP antiserum , we performed infection and co-infection experiments with UHV-2 and HISV-1 on cultured boid kidney cells ( I/1Ki ) , and analyzed the infected cells by western blotting , immunofluorescence ( IF ) staining , and immunohistology ( IH ) ( Fig 3 ) . The western blot shows that the anti-HISV NP antiserum does not cross-react with UHV-2 NP , and vice versa ( Fig 3A ) . The IF staining of infected cells concurs with the western blotting results , and also indicates that neither of the antisera reacts with cellular proteins at the concentrations applied ( Fig 3A ) . A comparison of the IF staining patterns in infected cells showed that HISV-1 infection produces large fluorescent foci ( i . e . infection of several cells in close proximity to each other ) , whereas UHV-2 infection resulted in scattered individual positive cells ( 100x magnification in Fig 3B ) . At a higher magnification , UHV-2 infected cells were found to exhibit the punctate NP staining pattern typical for reptarenavirus infected cells with inclusion bodies ( IB ) [4 , 25] , whereas HISV-1 NP appeared to be more diffusely distributed in the cells ( 400x magnification in Fig 3B ) . A similar staining pattern was seen in HISV-1 infected I/1Ki cell pellets , indicating that the antiserum is suitable for immunohistology ( IH ) ( Fig 3C ) . Since we did not see a marked difference in the amount of viral NP in western blots when comparing single-virus infection to co-infection ( Fig 3A ) , we decided to study the effect of co-infection on virus replication by quantifying the amount of viral RNA released from infected cells , using qRT-PCR . We initially used UHV-2 and HISV-1 for the experiment , since these viruses originated from the same isolate . Fig 4A shows representative results of one of the three consecutive experiments . We found that the amount of UHV-2 RNA released in co-infection was not reduced as compared to the single virus infection ( Fig 4A ) , indicating that co-infection does not markedly affect the replication rate of reptarenaviruses and hartmaniviruses . To provide further evidence for this observation , we performed a similar co-infection experiment using HISV-1 and UGV-1 . We decided to use UGV-1 , the virus of which S segment we most often find in snakes with BIBD ( other authors call this S segment S6 [11] ) . Also , UGV-1 grows to high titers in our cell culture model . We obtained results comparable to those obtained with UHV-2 ( Fig 4A and 4B ) , indicating that reptarenaviruses and hartmaniviruses do not interfere with each other’s replication during co-infection . We also tried infecting cultured mammalian cell lines ( Baby hamster kidney , BHK-21; African green monkey kidney , Vero E6 ) with HISV-1 at 30°C and 37°C , however , we could not detect viral antigen or replication . As indicated by IF staining ( Fig 3B ) the reptarenavirus NP forms large IB in infected cells , which we and others have described in detail [2 , 4] . However , the IF staining suggested that such IB would not be present in HISV-1 infected cells . Therefore , we performed an ultrastructural study of HISV-1 infected snake cells . Indeed , we could not demonstrate IB in HISV-1 infected cells by electron microscopy ( EM ) . Instead , the HISV-1 infected cells generally exhibited extensive cytoplasmic vacuolization and blebbing of the plasma membrane ( PM ) ( Fig 5A to 5C ) . We also observed tubular structures within the cells ( Fig 5B ) , which appeared to contain electron electron-dense material similar to the blebs at the PM ( Fig 5B to 5D ) . For comparison , an example of reptarenavirus infection-associated electron dense IB is presented in Fig 5E . We assumed that the electron-dense structures observed in HISV-1 infected cells would contain and/or consist of viral proteins . Indeed , immuno-EM revealed that both the tubular structures and the blebs at plasma membrane contain HISV-1 NP ( Fig 6 ) ; the latter often appeared to be continuous between two adjacent cells ( Fig 6E , inset ) suggesting direct cell-to-cell contact . We found HISV-1 NP also associated with vacuoles ( Fig 6B and 6C ) and in the nucleai/along the nuclear membrane ( Fig 6G ) , which is in line with the IF staining ( Fig 3B ) . The HISV-1 NP was found to be abundant in the tubular structures with the infected cells ( Fig 6D , 6F and 6G ) . Viral antigen expression in the blebs at the PM and in the intracellular tubular structures together with the detection of large foci of infected cells in IF staining could suggest that cell-to-cell spreading plays a greater role in the spread of hartmaniviruses than of reptarenaviruses . For comparison , the NP of the reptarenavirus ( UGV-1 ) accumulated in the cytoplasmic inclusion bodies ( Fig 7 ) . Since we could not find evidence of ZP at genome level , we decided to analyze purified HISV-1 virions in comparison to those of purified reptarenavirus ( UGV-1 ) . We used UGV-1 for this approach , as it produces abundant virions in our cell culture model . For the structural study , we purified the virions by density gradient ultracentrifugation ( Fig 8A ) . The major band in the main fractions containing HISV-1 ( F6-F8 in Fig 8A ) represents the NP , as demonstrated by western blotting ( Fig 8B ) . We then compared the SDS-PAGE patterns of purified UGV-1 and HISV-1 to find more evidence on the lack of ZP in hartmaniviruses . The NP appears as the most abundant protein for both UGV-1 and HISV-1 with an approximate size of 70 kDa by SDS-PAGE mobility ( Fig 8B ) . The exact sizes of GP1 and GP2 are not known , however , by comparing the concentrated supernatants produced by the same cell line infected with reptarenavirus vs . hartmanivirus , the indicated bands most likely represent the viral glycoproteins ( Fig 8B ) . A doublet band at around 16–17 kDa is evident in the UGV-1 preparation , which is slightly larger than the calculated molecular weight , 12 . 7 kDa , of the ZP . The observed difference between predicted molecular weight ( M . W . ) and SDS-PAGE mobility could be due to palmitoylation , since the ZP is by prediction palmitoylated at 3 sites ( http://csspalm . biocuckoo . org/online . php ) . As expected , the corresponding bands are missing from the HISV-1 preparation ( Fig 8B ) , providing further evidence that hartmaniviruses lack the ZP . The result also suggests that the lack of ZP is not complemented by a similar ( -sized ) cellular protein . We then studied the HISV-1 virions in EM under negative staining . The virions appeared pleomorphic to roundish with an approximate diameter of 120–150 nm and most of the particles seemed not to be intact ( top panels in Fig 8C ) . We then analyzed HISV-1 virions in cryo-EM , which provided , in agreement with the SDS-PAGE analysis , no evidence of continuous density lining the inside of the membrane ( Fig 8C ) . A comparison of the appearance of HISV-1 and UGV-1 virions in cryo-EM confirmed that the overall morphology and size ( roughly 120–150 nm in diameter ) of the virions is similar ( Fig 8C ) . We recently studied several Boa constrictor clutches born to parental animals with BIBD , aiming to demonstrate or rule out vertically transmission of reptarenaviruses [12] . We used NGS to initially define the reptarenavirome of each clutch , and confirmed our findings by species-specific RT-PCRs . However , for one of the five clutches we used only RT-PCR to demonstrate the virus transmission , since we already had analyzed three clutches from the same breeder by NGS [12] . Some of our species-specific RT-PCRs designed based on the NGS results of other clutches did not work optimally with the samples from clutch #5 [12] , which prompted us to study some of the animals ( snakes 2 . 1–2 . 5 , Table 1 ) by NGS and de novo genome assembly . We found a hartmanivirus , designated as Veterinary Pathology Zurich virus-1 ( VPZV-1 ) , accompanied by three reptarenavirus L and two S segments in the liver of the father of this clutch ( snake 2 . 1 , Table 1 ) and in some of the 12 to 20-month-old offspring ( snakes 2 . 2–2 . 5 , Table 1 ) as well as one additional reptarenavirus L segment in a pooled sample of the offspring ( 2 . 2 , Table 1 ) . The GenBank accessions for the hartmaniviruses and reptarenaviruses identified in this study are provided in S1 Table . These findings suggest that also hartmaniviruses can be vertically transmitted , although we cannot entirely rule out the transmission after birth for these juvenile snakes . We also identified VPZV-1 by NGS in liver and brain samples of an adult snake from a different breeder ( 2 . 6 , Table 1 ) , again together with several reptarenavirus L and S segments . By studying some of the cell culture isolates of an earlier study [4] ( animal 2 . 7 , Table 1 ) with the NGS approach , and found a virus which according to ICTV criteria represents a genetically distinct lineage of the VPZV species ( designated ad VPZV-2 ) . This virus was accompanied by three reptarenavirus L and one S segments . In another snake ( 1 . 4 , Table 1 ) from the breeder of snake 2 . 6 , we found a virus that by sequence comparison represents the same species as HISV-1 and thus named it HISV-2 . The next hartmaniviruses we identified in a pooled blood sample of snakes ( 3 . 1 , Table 1 ) with confirmed BIBD . According to ICTV criteria the identified viruses represent yet another hartmanivirus species , and were designated as Old Schoolhouse virus-1 and -2 ( OScV-1 and -2 ) . Subsequently , we found another representative of OScV-2 in a pooled blood sample ( 3 . 2 , Table 1 ) of snakes with BIBD from another breeding colony . Finally , we identified a putative representative of a fourth hartmanivirus species ( 4 . 1 , Table 1 ) , which we named Dante Muikkunen virus-1 ( DaMV-1 ) , in a snake with mild neurological signs suspected to be associated with BIBD . In addition to identifying DaMV-1 and reptarenavirus segments , we also found the snake to carry a novel deltavirus , which we describe outside this report [26] . Congruently with PASC analysis ( S3 Table ) , the phylogenetic analyses suggested that the novel hartmaniviruses clustered according to their tentative species designations on the basis of their L and S segments ( Fig 9 ) . The phylogenetic analysis of the RdRp amino acid sequences of the representatives of all known arenavirus species suggested that the genus Hartmanivirus forms an outlying group separate from mammarenaviruses and reptarenaviruses , whereas recently found Wenling frogfish arenaviruses [27] form an outlying group separate from the hartmani- , reptarena- and mammarenaviruses ( Fig 9 ) . To gain some information on the prevalence of hartmaniviruses in larger populations , we designed primers based on the L segments of OScV-1 and -2 , and used RT-PCR to screen a set of 71 blood samples collected from the same breeding colony , in which we initially detected these viruses . Interestingly , 44/71 snakes were RT-PCR positive for OScV-1 and/or -2; of these , one snake was positive for both OScV-1 and -2 . Thirty-four snakes in the collection were diagnosed with BIBD as they exhibited IBs in blood cells , of these , 23 had hartmanivirus infection . The fact that close to 70% of snakes with BIBD had an accompanying hartmanivirus infection indicates that the role of hartmanivirus infection in the pathogenesis of BIBD needs to be further investigated . Production of the anti-HISV NP antibody enabled us to study the tissue and cell tropism of hartmaniviruses , using immunohistology . We initially screened the tissues of the snake from which UHV-2 and HISV-1 originate ( animal 1 . 1 , Table 1 ) . In comparison to reptarenavirus NP , which can be detected in most tissues as cytoplasmic IB [28] , HISV NP expression was very limited and most consistent in the brain . The neurons exhibited a diffuse , finely granular , cytoplasmic and/or axonal reaction ( Fig 10A ) , which clearly differed from the reptarenavirus NP expression pattern ( Fig 10B ) . Additionally , HISV-1 NP expression was occasionally seen in a range of other cell types: smooth muscle cells in the lung and respiratory epithelial cells ( Fig 10C ) , endothelial cells and medial smooth muscle cells in arteries ( Fig 10D ) , ependymal cells in the brain ( Fig 10E ) , and axons of peripheral nerves ( Fig 10F ) . Examination of another three snakes that harbored HISV-1 or -2 ( animals 1 . 2–1 . 4 ) revealed a similar HISV-1 NP expression pattern , though with some variation in the range of cell types ( Table 1 ) . In animal 1 . 4 , infected with HISV-2 , a broader range of epithelial cells was found to be occasionally positive ( pulmonary epithelial cells ( Fig 10G ) , glandular epithelial cells in trachea and stomach ( Fig 10H ) , intestinal epithelial cells , acinar epithelial cells in the exocrine pancreas ( Fig 9I ) and tubular epithelial cells in the kidney ( Fig 10J ) ) . Also , smooth muscle cells in the muscular layers of stomach and intestine as well as dendritic cells in the spleen were found to express the viral antigen ( Figs 9 and 10 ) . In animal 1 . 3 , the liver exhibited HISV-1 NP expression in sinusoidal endothelial cells . In none of the hartmanivirus negative animals was there any evidence of NP expression ( Table 1 ) . We tested some of the snakes found to be infected with the above HISV-like viruses ( VPZV-1: 2 . 1 , 2 . 3–2 . 6; DaMV-1: 4 . 1 ) by IH for HISV-1 NP . The antibody appeared to cross react only minimally , as the reaction was restricted to a few individual cells in two animals ( Table 1 ) . We then tested six snakes that had been HISV-1 negative by RT-PCR ( 5 . 1–5 . 5 ) or hartmanivirus negative by NGS ( 5 . 6 ) for the expression of HISV-NP , all with a negative result ( Table 1 ) . After a most recent revision the family Arenaviridae currently comprises three genera Mammarenavirus , Reptarenavirus and Hartmanivirus [14] . Until now the genus Hartmanivirus only contained HISV-1 [14] , a virus that was only characterized at nucleotide level [10] . The genome of HISV-1 appeared to lack ORF for the ZP , i . e . the matrix protein of mammarenaviruses and reptarenaviruses [10] . We had isolated HISV-1 in a permanent boid kidney cell culture , however the isolate contained two viruses , UHV-2 and HISV-1 . Hence , we could not confirm the lack of ZP in the initial report . It was also unclear whether HISV-1 would survive without a co-infecting reptarenavirus . Herein , we report the production of a pure HISV-1 isolate , which represents the type species of the genus Hartmanivirus . The successful generation of a pure HISV-1 isolate indicates that hartmaniviruses can grow in the absence of a co-infecting reptarenavirus and allowed us to study the physical properties of HISV-1 virions at nucleotide , protein and structural level . We then used HISV-1 as the model to characterize hartmanivirus infection at both in vitro and in vivo level . Furthermore , we can expand the genus Hartmaniviruses to four known species , by obtaining complete or near complete genome segments for three new hartmanivirus species . Finally , by screening a snake collection for two hartmaniviruses , we provide first evidence that hartmanivirus infections are rather common in captive snakes . Comparison of the HISV-1 genome to those of viruses in the other genera of the family Arenaviridae shows similarities , for example in the genome ends , but also a striking difference , namely the lack of an ORF for the ZP , the matrix protein present in the other arenaviruses . We did not find additional genome segments for HISV-1 when we performed NGS on a pure preparation of HISV-1 , indicating that HISV-1 comprises an S and L segment like the viruses of other arenavirus genera . We tried to seek evidence for ZP or a ZP surrogate by analyzing a pure HISV-1 preparation by SDS-PAGE and cryo-EM , but again found no evidence . The fact that we found three further hartmanivirus species ( with similar coding strategy ) provides further support that the lack of ZP is a general feature of the genus Hartmanivirus . The ZP drives the budding of mammarenaviruses which is mediated by proline-rich late domain motifs PTAP or PPPY [19 , 29] . Curiously , the ZP of the bat-borne Tacaribe virus is devoid of the late domains but does still efficiently mediate budding [30] . The ZPs of reptarenaviruses also lack the late domain , which could imply that the putative budding function of reptarenavirus ZP resides in a yet unknown motif . Interestingly , a sequence comparison reveals a conserved late motif , PPPY , in the reptarenavirus NPs , which is not found in the NPs of hartmaniviruses and mammarenaviruses . Also , the C-terminus of reptarenavirus NPs has been reported to contain late domain like motifs [2] . Thus one could speculate that also the NP associates with the budding of reptarenaviruses . Experimental studies are needed to demonstrate the role of individual proteins in the budding of reptarenaviruses , but these are beyond the scope of this report . The fact that hartmaniviruses lack a ZP , but nonetheless efficiently produce infectious particles suggests that the budding function resides in some other structural protein . Alternatively , the virus could induce a cellular protein that aids to virus budding , however we found no evidence of ZP sized proteins in purified HISV-1 preparations . Hantaviruses are similar to hartmaniviruses with regards to their proteins; they also encode RdRp , GPC and NP , but lack a matrix protein . The cytoplasmic tail of the Gn glycoprotein of hantaviruses is suggested to act as a matrix protein surrogate [31] . Thus , one could speculate that the GP2 tail of hartmaniviruses contains a motif that mediates budding . And indeed , it harbors a conserved P-Y/F-P-H-Y-P stretch ( Fig 2B ) , which by ELM ( eukaryotic linear motif resource , http://elm . eu . org/ [23] ) prediction binds to the apoptosis-linked gene 2 ( ALG-2 ) protein which bridges ALIX ( ALG-2-interacting protein X ) and ESCRT-I complex via interactions with ALIX , TSG101 and VPS37 [32] . Thus , hartmaniviruses might utilize the interaction with ALG-2 to gain access to the ESCRT pathway for egress , analogously to HIV-1 [33] . Our immuno-EM data suggest that the NP of hartmaniviruses is included in the buds that form along the plasma membrane of infected cells . One could thus speculate that at some point the NP of reptarenavirus ancestors obtained the budding function via emergence of late domains . This would have made the GP2 tail redundant for budding , thus explaining the lack of it in reptarenaviruses . Alternatively , it has been proposed that the GP2 of reptarenaviruses evolved from a recombination event with a filovirus or retrovirus that provided the new gene [2] . Whatever the chain of events , it seems that the ZP could have emerged around the same time to regulate replication and to bridge between the NP and the GPs to facilitate efficient genome packaging . The ZP may also have emerged before speciation of mammarenaviruses and reptarenaviruses , perhaps initially without the late domains . These hypotheses provide interesting topics for further studies , and identification of arenaviruses from other animals could help to shed light on the evolution of arenaviruses . Comparison of the GPCs between arenavirus genera revealed that mammarenaviruses and hartmaniviruses harbor both SSP and a cytoplasmic tail in their GP2 , features missing from the GPC of reptarenaviruses . The N-terminal halves of the mammarena- and hartmanivirus SSPs are more similar than the C-terminal halves . The N-terminus of the SSPs contains a myristoylation motif/site followed by a hydrophobic stretch until a conserved lysine residue in mammarenavirus or RGR motif in hartmanivirus SSPs ( Fig 2B ) . The SSP of mammarenaviruses is suggested to span the viral membrane twice , leaving the conserved lysine residue on the virion surface [24] . Even though the C-terminal halves of hartmanivirus SSPs are less hydrophobic mammarenavirus SSPs , we hypothesize the SSPs to have a similar topology . Supporting the above , we identified a conserved cysteine residue close to the SSP C-terminus ( Fig 2B ) , which participates in formation of an intersubunit zinc-finger structure between two conserved histidine and four cysteine residues of the GP2 [24] ( Fig 2C ) that are also conserved in both mammarena- and hartmaniviruses . We hypothesize these to indicate similar spike structure between mammarena- and hartmaniviruses ( Fig 2C ) . The fact that phylogenetic analysis , as discussed below , suggests hartmaniviruses to represent the ancestors of mamm- and reptarenaviruses renders the observed differences in the GPC ORF interesting . Because the SSP and GP2 cytoplasmic tail are found in mammarena- and hartmaniviruses but not in reptarenaviruses , the reptarenaviruses seem to have lost these features at some point during their evolution , perhaps in a suggested recombination event with filo- or retroviruses [2] . The fact that GP2s of reptarenaviruses are more conserved than GP2s of mammarena- and hartmaniviruses would support both adaptation to a new niche or the speculative recombination event . In line with the previous studies [10 , 27] the phylogenetic analysis suggested that the hartmaniviruses form a basal lineage for both reptarena- and mammarenaviruses while the Wenling frogfish viruses form an outgroup to the other arenaviruses ( Fig 8 ) . Therefore , the phylogeny of arenaviruses resembles , but does not recapitulate the evolution of the respective host species . This suggests that in addition to the apparent co-evolution between arenaviruses and their hosts [27] at least one host-switch event has occurred during the evolution of arenaviruses , potentially from snakes to mammalia . Discovery of arenaviruses from other reptiles or from amphibia would shed more light on the extent of co-evolution and frequency of cross-species transmission events among arenaviruses . Our IF studies on hartmanivirus ( HISV-1 ) , reptarenavirus ( UHV-2 and UGV-1 ) and co-infected cell cultures showed that the distribution of NP within the infected cells varies distinctly between the two genera . We also observed that HISV-1 rapidly induced the occurrence of large foci of infected cells in our cell culture system , while UHV-2 infected cells are scattered and mostly individual . We interpreted this as evidence of more pronounced cell-to-cell spreading of HISV-1 . Potential support of this interpretation is the observed bleb formation of the plasma membrane with HISV-1 ( Fig 5A–5D; Fig 6A and 6E ) , but not with reptarenavirus ( UHV-2 ) infection ( Fig 7 ) . The membrane blebs contained electron dense material which we assumed to be of viral origin . Indeed , immuno-EM showed that the membrane blebs contain HISV-1 NP ( Fig 6E ) . The latter also accumulated along cytoplasmic nanoscale tubules ( Fig 6D , 6F and 6G ) . Interestingly , influenza viruses were recently shown to utilize tunneling nanotubules ( involved in intercellular communication ) to transfer viral proteins and genome from infected to naive cells [34] . It is thus tempting to speculate that the NP-loaded ( and likely also RNA containing ) membrane blebs and cytoplasmic tubules in HISV-1 infected cells are indications that hartmaniviruses employ a similar strategy . Further studies are needed to address the above hypotheses . The comparative investigation of snakes with BIBD by IH for reptarenavirus and HISV-1 NP provides evidence that hartmaniviruses have a more restricted cell tropism than reptarenaviruses . The mammalian kidney cells ( BHK-21 and Vero E6 ) commonly used for propagation of mammarenaviruses were not permissive for HISV-1 . However , both cell lines are permissive for reptarenaviruses when cultured at 30°C , [4 , 25] , which could indicate broader tissue tropism of reptarenaviruses . The frequency and extent to which hartmaniviruses were detected in neurons of infected snakes suggests a pronounced neurotropism , it therefore seems worth testing if mammalian neuronal cells would be more permissive for HISV-1 . Also , further studies are needed to demonstrate the presence or absence of hartmaniviruses in snake secretions . Coincidentally , we found hartmanivirus in “clutch 5” of our previous study [12] , and could demonstrate that the father and some of the juvenile offspring were carriers of the same virus . These results are indicative of vertical transmission . We identified HISV-1 accidentally while aiming to obtain full length genomes for reptarenavirus isolates [10] . Similarly , the viruses identified herein were in the vast majority found in snakes with BIBD . Our earlier observation was that identification and full genome sequencing of reptarenaviruses works very well from brain-derived total RNA . However , in a previous study we noticed that the brain might display only a fraction of the reptarenavirus S and L segments found in the blood [12] . Due to this we have recently focused on studying liver- and/or blood-derived RNA for NGS studies , which has also led to identification of more hartmaniviruses . IH analysis of tissues from snakes with hartmanivirus infection showed that viral antigen is not abundantly expressed in the brain , which suggests that the amount of RNA in the brain is indeed low . So far we have only studied snakes with either suspected or confirmed BIBD , in which the hartmaniviruses were always accompanied by reptarenaviruses . However , by producing a pure isolate of HISV-1 , we could demonstrate that hartmaniviruses do not require reptarenavirus co-infection for their infectious cycle in vitro . We further show that reptarena- and hartmanivirus co-infection does not negatively affect the replication of either virus . While we could not associate hartmanivirus infection with any pathological changes , further studies are needed to confirm if hartmanirviruses are apathogenic in snakes . Future studies are also needed to identify the natural host ( s ) of hartmaniviruses . Also , the recent discovery of a three segmented arenavirus in fish [27] indicates that more arenaviruses are yet to be found with potential to alter the understanding of arenavirus evolution . The samples included in this study originated from animals submitted by their owners either to the Institute of Veterinary Pathology , Vetsuisse Faculty , University of Zurich , Switzerland , or to the Department of Veterinary Biosciences , Faculty of Veterinary Medicine , University of Helsinki , Finland , for a diagnostic post mortem examination . An Animals Scientific Procedures Act 1986 ( ASPA ) schedule 1 ( appropriate methods of humane killing , http://www . legislation . gov . uk/ukpga/1986/14/schedule/1 ) procedure was applied to euthanize the snakes . Full diagnostic post mortem examination , blood sampling and diagnostic testing of collected samples were performed with full owners' consent . Ethical permissions for euthanasia and diagnosis-motivated necropsies ( both routine veterinary procedures ) were not required due to suspicion of a lethal disease , BIBD . The study was performed on tissues or full blood of 23 snakes that were suspected to suffer from BIBD . A further 68 blood samples from snakes in a private breeding collection which previously had animals dying with BIBD were screened for hartmanivirus infection upon the owner’s request . All animals were captive snakes from breeding collections in Germany and Switzerland , ranging in age from juvenile to more than 12 years ( Table 1 ) . The Boa constrictor kidney cell line , I/1Ki , was used for virus propagation and virus isolation attempts as described [4] . A virus preparation containing HISV-1 and UHV-2 described in [10] was used as the source of pure isolates . The isolation strategy is depicted in S1 Fig . Briefly , 10-fold dilution series of the virus stock were prepared on I/1Ki cells grown on a 96-well plate , and at 14 days post infection ( dpi ) the cells inoculated with virus dilutions 1:107 and 1:108 were transferred onto a 24-well plate . The cell culture medium was collected at 7 and 14 dpi , and the pooled cell culture supernatants were analyzed by virus species specific RT-PCR as described [10 , 12] . For production of HISV-1 stock , a 75-cm2 flask of semi-confluent I/1Ki was inoculated with 500 μl of the pooled supernatant from the 24-well plate , the cell culture medium was collected and replaced at 2–3 day intervals until 14 dpi , and the pooled supernatants were filtered through a 0 . 45 μm syringe filter ( Millipore ) and stored at -80°C for further use . Large quantities of HISV-1 were produced by inoculating semi-confluent I/1Ki 75-cm2 flasks with 1 ml of 1/50-1/200 diluted HISV-1 stock , followed by supernatant collection as described above . Viruses were concentrated by pelleting through a sucrose cushion and more pure virus preparations were obtained by sucrose density gradient ultracentrifugation as described [4 , 35 , 36] . For cryo-electron microscopy ( cryo-EM ) the virus-containing fractions were pooled and dialyzed against phosphate-buffered saline ( PBS ) . For transmission electron microscopy and immunocytology , cells were infected and harvested six days after inoculation , pellets prepared and fixed as described [4] . For co-infection experiments I/1Ki cells were inoculated either with equal amounts of HISV-1 and UHV-2 , or HISV-1 and UHV-2 alone as controls ( multiplicity of infection > 1 ) . The infection vs . co-infection experiments were done in duplicate . Trizol and Trizol LS isolation reagent ( Life Technologies ) in combination with QiaGEN RNeasy Mini Kit ( Qiagen ) was used for RNA isolation as described [12] . RNA isolation from cell culture supernatants was done with either the QIAamp Viral RNA Mini Kit ( Qiagen ) or the GeneJET RNA Purification Kit ( Thermo Fisher Scientific ) following the manufacturer’s instructions . No carrier RNA was used during RNA isolation for samples analyzed by NGS . RT-PCR served to detect viral RNA from cell culture supernatants as described [10 , 12] . HISV-1 RNA isolated from pelleted virus was treated with T4 polynucleotide kinase ( Thermo Fisher Scientific ) according to the manufacturer’s protocol , purified using the QiaGEN RNeasy Mini Kit ( Qiagen ) , and circularized with T4 RNA ligase ( Thermo Fisher Scientific ) . The RNA circularization reaction ( 2 h at 25°C ) mix ( 20 μl in DEPC-treated water ) included: 2 μl of 10X reaction buffer , 5 μl of isolated RNA , 1 μl of T4 RNA ligase , 0 . 5 μl RNAse inhibitor ( 40 μ/μl ) , 10% PEG 8000 , and 100 μM ATP . The reaction ligase was inactivated by heating the reaction mix to 70°C for 10 min . The circularized RNA was reverse transcribed using RevertAid H Minus Reverse Transcriptase ( Thermo Fisher Scientific ) following the manufacturer’s protocol for specific primers . The S segment primers were 5´-CTCCATTTACTCGAACAAGCTCAC-3´ and 5´-CAGGTTAAATTCATTGTTGGAGCA-3´ , the L segment primers were 5´-GCACAACAATCTTTCTGCGAT-3´ and 5´-CAGGGCTTTGTTTTGTCCAG-3´ . Phusion Flash High-Fidelity PCR Master Mix ( Thermo Fisher Scientific ) was used for PCR amplification . The reaction mix consisted of: 1 μl of cDNA , 10 μl of Master Mix , 1 μl of forward and reverse primer ( 10 μM stocks ) , and 7 μl of molecular grade water . The cycling conditions were: 1 ) 10 s at 98°C; 2 ) 1 s at 98°C; 3 ) 5 s at 60°C; 4 ) 7 s at 72°C; 5 ) 1 min at 72°C; steps 2 to 4 were repeated 35 times . The PCR products were separated by agarose gel electrophoresis , purified with the QIAquick gel extraction kit ( Qiagen ) following the manufacturer’s instructions , and cloned into plasmid using Zero Blunt TOPO PCR Cloning Kit ( Thermo Fisher Scientific ) following the manufacturer’s recommendations . Plasmid minipreps were purified using the GeneJET Plasmid Miniprep Kit ( Thermo Fisher Scientific ) and the purified plasmids were sent for sequencing ( with M13 forward and reverse primers ) to Microsynth ( Zurich , Switzerland ) . To study the secondary structures formed by the genome ends of LCMV ( strain Armstrong 53b , S segment GenBank accession NC_004294 , L segment NC_004291 ) , GGV ( S segment NC_018483 , L segment NC_018482 ) , and HISV-1 ( S segment KR870017 , L segment KR870031 ) we used DuplexFold Web Server of RNA structure at RNAstructure ( Web Servers for RNA Secondary Structure Prediction , available at https://rna . urmc . rochester . edu/RNAstructureWeb/Servers/DuplexFold/DuplexFold . html ) [37 , 38] . We applied standard parametrization for the predictions , the folding free energies for the models chosen for presentation are: LCMV S segment -30 . 9 kcal/mol; LCMV L segment , -41 . 4 kcal/mol; GGV S segment -33 . 4 kcal/mol; GGV L segment -35 . 6 kcal/mol; HISV S segment -34 . 3 kcal/mol; and HISV L segment , -29 . 9 kcal/mol . NGS and de novo assembly was done as described [12 , 39] . The sequences were aligned with Clustal Omega algorithm [40] implemented in EMBL-EPI webserver [41] . The phylogenetic trees were constructed using Bayesian Monte Carlo Markov Chain ( MCMC ) method implemented in BEAST version 2 . 4 . 7 [42] using LG or HKY-G-I substitution models for amino acid and nucleotide sequences , respectively . The analyses were run for 50 million states and sampled every 5000 steps . They were carried out on the CSC server ( IT Center for Science Ltd . , Espoo , Finland ) . Posterior probabilities were calculated with a burn-in of 10% and checked for convergence using the Tracer version 1 . 6 . Full-length NP ( amino acids 1–582 ) , and N- ( aa 1–295 ) and C-terminal ( aa 296–582 ) parts of it were PCR cloned for E . coli expression using primers ( forward for full length and N-terminal portion 5´-CACCATGTCCTTGAACAAGGACCTT-3´; reverse for N-terminal portion 5´- TCTGTCGCTGGTGCAACC-3´; forward for C-terminal portion 5´- CACCATGATCTCATCTCAAAACATACC-3´; reverse for full length and C-terminal portion 5´- GTTGTTCATTATGTAGTTGAA-3´ ) designed according to the Champion pET101 Directional TOPO Expression Kit with BL21 Star ( DE3 ) One Shot Chemically Competent E . coli manual ( Thermo Fisher Scientific ) . Protein production and purification was done as described [39] . Full length HISV NP could not be recovered by this strategy , but the N- and C-terminal portions were recovered in moderate and good amount , respectively . The purified C-terminal portion of HISV NP was dialyzed against PBS . A rabbit polyclonal antiserum against C-terminal portion HISV NP ( anti-HISV NP-C ) was produced by Biogenes GmbH ( Berlin , Germany ) . IF staining was done on cells grown on 24-well Glass Bottom Plates ( In Vitro Scientific ) as described [39] . The primary antibodies , anti-UHV NP-C or anti-HISV NP-C antisera , were used at 1:2 , 000 dilution in PBS . Routine protocols , described in [43] , were utilized for SDS-PAGE and western blotting , the results were recorded with the Odyssey Infrared Scanning System ( LI-COR ) . A Taqman qRT-PCR assay for quantifying the S and L segments of UHV-2 and HISV-1 served to monitor the growth of UHV-2 and HISV-1 in cell culture . The primers and probes were: UHV-2 S segment forward primer ( FWD ) 5´-GCAAAACAGAACTGCTGCAGTC-3´ , reverse primer ( REV ) 5´-TGCGATACAGACATAATTAGAGACATTG-3´ , and probe 5´-6-Fam ( carboxyfluorescein ) -GTCACCATGTGTCCCTCAGAACTCATTCA-3´-BHQ-1 ( Black Hole Quencher ) ; UHV-2 L segment FWD 5´-TTGGGGAGTTTGTTACCAATGT-3´ , REV 5´- GTGGGCCCAAATAACAAACCT-3´ , and probe 5´-6-Fam- CTCTCTCGGACCTCCCACTTGTTCCTTTATG-3´-BHQ-1; UGV-1 S segment forward primer ( FWD ) 5´- CAAGAAAAACCACACTGCACA-3´ , reverse primer ( REV ) 5´- AACCTGTTGTGTTCAGTAGT-3´ , and probe 5´-6-Fam ( carboxyfluorescein ) - CTCGACAAGCGTGGGCGGAGG-3´-BHQ-1 ( Black Hole Quencher ) ; UGV-1 L segment FWD 5´- TCATAAAAGCTTCTAGCTATTCTTTTCAT-3´ , REV 5´- CAAGTTGGAGGCCCAAGAG-3´ , and probe 5´-6-Fam- TGAAGTCTCCTCCAAGACCCTGGTTATCAG-3´-BHQ-1; HISV-1 S segment FWD 5´-CTCAAAATCTTACCGAAGTTGTATGTAC-3´ , REV 5´-CACTTTCCCTTTTGGATCTTTG-3´ , and probe 5´-6-Fam-GTGACGACCAAGTGTCGGGTCACAC-3´-BHQ-1; HISV-1 L segment FWD 5´-GAGTCTTTGTTTGATAATGGTGGTT-3´ , REV 5´-ATTGAAGACTACAGAACCATATC-3´ , and probe 5´-6-Fam-TCATTTGATTCAAGTGTTCTCAGGGCA-3´-BHQ-1 ( Metabion International Ag ) . RNA isolation for Taqman assays was done with the GeneJET RNA purification kit ( Thermo scientific ) with carrier RNA following the manufacturer’s protocol . Taqman Fast Virus 1-step master mix ( Thermo scientific ) was used for qRT-PCR , 10 μl reactions were run with the AriaMX real-time PCR System ( Agilent ) in duplicate with recommended cycling conditions: 1 ) 50°C 5 min; 2 ) 95°C 20 s; 3 ) 95°C 3 s; 4 ) 60°C 30 s ( steps 3 and 4 were repeated 39 times ) . Pellets from HISV-1 or UGV-1 infected I/1Ki cells were fixed in 1 . 5% glutaraldehyde , buffered in 0 . 2 M cacodylic acid buffer , pH 7 . 3 for 12 h at 5°C and embedded in epoxy resin . Toluidin blue stained semithin ( 1 . 5 μm ) sections and , subsequently , ultrathin ( 100 nm ) sections were prepared and the latter contrasted with lead citrate and uranyl acetate and examined with a Philips CM10 transmission electron microscope at 80kV . For immuno-EM , cell pellets were fixed in 2 . 5% glutaraldehyde in 0 . 5xPBS and epoxy resin embedded . Thin sections ( 100 nm ) were prepared and incubated for 30 min at RT in PBS with 1% BSA , followed by overnight incubation with rabbit anti-HISV NP antiserum ( diluted 1:1 , 000 in PBS with 1% BSA ) at 4°C . After washing with PBS , sections were incubated with 18 nm gold-conjugated goat anti-rabbit IgG antibody ( Milan Analytica AG , Rheinfelden , Switzerland; diluted 1:20 in PBS with 1% BSA ) for 2 h at RT . Sections were then contrasted and examined as descrived above . Immunohistology with anti-HISV NP antiserum at 1:6 , 000 dilution was performed on formalin-fixed , paraffin embedded ( FFPE ) tissue sections , following previously published protocols [4 , 12 , 39] . Refer to Table 1 for animals examined . Consecutive sections incubated with a non-reactive rabbit polyclonal antibody instead of the specific primary antibody served as negative controls . A further section of each block was stained for reptarenavirus NP as described [4] . A FFPE pellet prepared from each an HISV- and reptarenavirus-infected cell sample served as positive control for the immunohistological stains . All snakes were also examined for any histopathological changes , using hematoxylin-eosin stained consecutive sections . The virus sequences obtained in this study are made available via GenBank , the accession numbers are provided in S1 Table .	From the 1930s to 2015 arenaviruses were known as mainly rodent-borne viruses , which occasionally infect humans , causing a severe disease . After isolation of novel arenaviruses from snakes , the family Arenaviridae now comprises three genera , Mammarenavirus , Reptarenavirus , and Hartmanivirus . Here we characterize a pure isolate of the putative hartmanivirus type species , Haartman Institute snake virus-1 ( HISV-1 ) and report three new hartmanivirus species: Old Schoolhouse virus , Dante Muikkunen virus and Veterinary Pathology Zurich virus . The genomes of all these hartmaniviruses lack the matrix protein found in other arenaviruses , and using HISV-1 as a model we demonstrate that hartmaniviruses can nonetheless form infectious particles . Although HISV-1 induces cytopathic changes in cell culture , there was no evidence of a cytopathogenic effect in infected snakes , where hartmaniviruses have a restricted cell tropism . We further identified hartmanivirus infection in over 60% of snakes in a single breeding collection , indicating that hartmaniviruses are relatively common in boid snakes . Several animals of the collection studied were diagnosed with BIBD , and it is not clear whether hartmaniviruses play a role in the development of BIBD in reptarenavirus-infected snakes . The co-infection with viruses from two genera of the same virus family provides an interesting model for future studies .	[ "Abstract", "Introduction", "Results", "Discussion", "Materials", "and", "methods" ]	[ "reverse", "transcriptase-polymerase", "chain", "reaction", "blood", "serum", "medicine", "and", "health", "sciences", "body", "fluids", "pathology", "and", "laboratory", "medicine", "pathogens", "biological", "cultures", "microbiology", "vertebrates", "animals", "viruses...	2018	Characterization of Haartman Institute snake virus-1 (HISV-1) and HISV-like viruses—The representatives of genus Hartmanivirus, family Arenaviridae
Gene expression data has been used in lieu of phenotype in both classical and quantitative genetic settings . These two disciplines have separate approaches to measuring and interpreting epistasis , which is the interaction between alleles at different loci . We propose a framework for estimating and interpreting epistasis from a classical experiment that combines the strengths of each approach . A regression analysis step accommodates the quantitative nature of expression measurements by estimating the effect of gene deletions plus any interaction . Effects are selected by significance such that a reduced model describes each expression trait . We show how the resulting models correspond to specific hierarchical relationships between two regulator genes and a target gene . These relationships are the basic units of genetic pathways and genomic system diagrams . Our approach can be extended to analyze data from a variety of experiments , multiple loci , and multiple environments . Epistasis has traditionally been discussed in two distinct contexts , corresponding to the disciplines of classical molecular genetics and quantitative genetics . In each case , the term describes an interaction between alleles at two or more loci . However , the methods for detecting epistasis and interpretations of the underlying biology have kept historical divisions in place despite calls for synthesis [1] . This is largely because the two fields traditionally study different types of traits in different experimental populations . The classical epistasis experiment compares a double-mutant with two associated single-mutants . Epistasis is present if the observed double-mutant phenotype is categorized as being the same as a single-mutant phenotype . This implies a specific type of interaction in which an allele at one locus masks the effect of variation at the second locus . This relationship is described as the first locus being epistatic to the second , and can be interpreted as one gene acting upstream of the other . This hierarchical interpretation has been used to construct biological pathways via a series of epistatic gene pairs . However , this approach is limited by the necessity of easily observed and categorized phenotypes [2] . In contrast , quantitative genetics examines traits that vary continuously and cannot easily be categorized . Such trait distributions result from the cumulative effects of many genes . Each additional gene increases the possible combination of alleles , and the number of possible phenotypes grows exponentially . An individual's phenotype is the sum of the allelic effects at each gene and the effect of the environment . Epistasis is defined as a deviation from these additive gene effects [3] . A quantitative genetic model can include multiple loci and multiple interactions . Epistasis in this sense describes a functional relationship between genes in the context of a trait , but it includes both hierarchical relationships and nonhierarchical relationships and there is no way to distinguish between these . Any genetic effect is only relevant to the population being studied due to the presence of genetic background . Background is genetic variation that is unobserved in the population and cannot be modeled . The classical experiment is performed using genetically homogenous laboratory strains so there is no background . Quantitative genetics studies diverse populations and background variation is almost always present . The implication of this is that epistasis may be detected in one experiment but not in another . This has led to criticisms that epistasis in the quantitative genetic sense is a statistical construct rather than a true representation of biology . In fact , both approaches seek to illustrate underlying molecular architecture and each has its strengths . A hierarchical interpretation of epistasis is attractive as increased focus is placed on genetic pathways and systems diagrams . However , quantitative approaches are necessary to accommodate continuous data types such as gene expression , metabolite concentrations , and fitness . Recent literature suggests that such approaches are being adopted . For example , while early large-scale fitness profiles in yeast deletion mutants [4] , [5] were scored categorically , St Onge et al [6] measured fitness in 650 double-deletion yeast strains and employed a novel quantitative analysis . The rise in genomic techniques has broken down one of the traditional barriers discussed above: the same traits are now being used in both classical and quantitative settings [7] . Gene expression is perhaps the most prevalent example . Instead of a single phenotypic trait value , a vector of expression measurements describes each individual . Expression profiling in single-deletion yeast strains found that 34% of mutants showed twenty or more differentially expressed genes [2] . Expression quantitative trait locus ( eQTL ) mapping uses a linear modeling approach to associate genetic variation with gene expression traits [8]–[12]; Storey et al . [13] found over thirty percent of traits were jointly linked to two loci in yeast . When gene expression correlates with a complex phenotype , the corresponding traits may reflect the molecular basis of that trait at a level intermediate between genotype and phenotype . Some studies suggest that epistasis is pervasive among expression traits [14]–[16] and such traits may have more QTLs than classical traits [13] , [17] . Since gene expression is being used in both classical and quantitative contexts , it is a valuable framework in which to compare the ability to detect epistasis and interpret the nature of relationships between genes . We propose a framework for estimating and interpreting epistasis using expression traits . Our goal is to accommodate the continuous nature of the data , yet still preserve a hierarchical interpretation of epistasis . Such interpretations are well established for classical epistasis experiments [18] , but have only recently been studied for complex data [19] . We refine the classical interpretations by explicitly modeling gene expression . Gene effects and interactions are estimated using a linear model , in a manner comparable to eQTL mapping . Our method selects the best-fit regression model for each trait , which describe the order and the nature of gene function . Such relationships are the basic units of genetic pathways and systems biology . We specifically address how to use a continuous phenotype in a manner that is both statistically sound and consistent with the classical approach . We illustrate our method with publicly available expression measurements from Dictyostellium discoideum wild type [20] and deletion mutant strains [21] . This experiment is a classical epistasis analysis that targets the genes of the protein kinase ( PKA ) pathway and measures the gene expression profile of each strain . In the classical epistasis analysis , triplets of deletion mutants combine with a wild type to form a contrast . Each contrast includes two single mutants and a double mutant . Each is described relative to the known wild type phenotype . A hypothetical example of a trait affected by two genes , A and B , can be described as follows , where y is the trait value , μ is the expected value of the wild type , βA and βB are the effects of deleting each gene , and ε is an error term . This adheres strictly to the classical definition , but there is a clear problem; there is no provision if the double mutant does not fall neatly into the same category as one of the single mutants . Gene expression traits fit poorly into the classical framework for this reason . Expression is continuous and intermediate levels are expected . Furthermore , even normalized trait values will inevitably include some measurement error . For these reasons , the double mutant observation is rarely the same as either of the single mutant observations or the wild type . Previous studies have attempted to circumvent this problem by relying on differences between the mutants to determine the most similar mutant pair . However , the assumption that expression is completely masked is poor . To address these issues , we move away from comparing trait values directly . Instead , we evaluate each deletion according to whether it significantly affects the expression of the target and associate unique patterns of significance with models of gene action . We use a linear model to estimate the effect of each deletion . This is a general way to relate all mutants and the wild type without making any assumptions about the nature of the double mutant . We regress the trait value ( e . g . expression ) on indicator variables representing the presence or absence of each wild type allele and an interaction term . The interaction describes effects that are unique to the double mutant . The same example discussed above can be described as follows . Trait value = Wild Type+Effect of deleting A+Effect of deleting B+Interaction+error Various techniques can be used to fit such a linear model . We first fit a full model and then use stepwise backwards selection to drop model terms with coefficients that are not significant at a set level . The resulting reduced model is termed the best-fit model . For any trait , there are eight possible best-fit models . For clarity , we number the reduced models as follows: When the best-fit model has been determined , we estimate parameter values using that model for each trait . Thus , we have a best-fit model and coefficient estimates for each trait . The terms in each best-fit model represent the significant gene and interaction effects acting on that trait . Individual coefficients represent the estimated effect of deleting each gene . Model 7 corresponds to the classical model above when the interaction between the two deletions offsets the effect of one of them , either βI = −βA or βI = −βB . Model 8 describes the case in which the deleted loci have no effect on the trait . A best-fit model describes each gene expression trait . As such , we have dealt with the continuous variable problem . However , by embracing a quantitative genetic model we have lost the appealing feature of the classical experiment: the ability to interpret hierarchical relationships . In the following section we identify sixteen hierarchical relationships and propose that a specific best-fit model supports each . In quantitative genetics , the interaction term in the above model is considered epistasis . However , epistasis in this sense includes both hierarchical and nonhierarchical relationships . Conversely , while Model 7 can clearly be interpreted as hierarchical epistasis with the conditions described above , it does not apply to all possible hierarchies . We considered all combinations of gene order and action within simple ON/OFF models and then predicted the hypothetical effect of deleting genes on each of them ( Figure 1 , Figures S1 , S2 , and S3 ) . There are four points of variation to model for each gene pair relationship . The first is the identity of the upstream gene , i . e . the gene order . Secondly , the upstream gene will turn the downstream gene either on ( enhance ) or off ( repress ) . Thirdly , the downstream gene can enhance or repress the expression of a target gene for which expression is observed . Lastly , we consider that the upstream gene itself will be enhanced or repressed by some initiating factor such as a developmental cue or environmental perturbation . Avery and Wasserman [18] provide a general framework that has been widely used for interpreting epistasis in response to such signals , and note that the effect of a mutation is only observable for a specific signal state . However , knowing the signal state does not give any information about whether the upstream gene is enhanced or repressed in that state . In our models , we focus on the effect on the upstream gene . This model has sixteen possible variants describing hierarchical relationships between two genes and the target gene . The key to our approach is connecting each of the sixteen hierarchical models to one of the eight possible best-fit regression models . If the deletion changes the state of a target gene relative to the wild type in a mutant , then that deletion is predicted to have a significant effect and it will be included in the regression model corresponding to that hierarchical model . Figure 1 gives an example of one possible model , in which A is enhanced by a signal; A is an upstream repressor to B; and B enhances a target gene X . We conclude that the corresponding best-fit regression model will include coefficients for A and an interaction term . Note that if the signal instead represses A , a different best-fit model represents the same relationship between A and B . We applied the same approach to each of the sixteen cases and note several trends . First , the downstream gene's effect upon the target gene X does not influence the corresponding best-fit model . This allows us to reduce the model space to eight hierarchical relationships ( Table 1 ) . This observation is convenient , because expression traits represent all the genes downstream of the deletions . Regardless of the downstream gene's direct effect , some traits will be enhanced while others are repressed . When the upstream gene is a repressor , four distinct regression models represent four unique hierarchical relationships . We can uniquely identify both the gene order and signal effect on the upstream gene . We cannot discern gene order if the upstream gene is an enhancer because the same best-fit model describes both hierarchies . If the upstream gene is merely enhancing the effect of the downstream gene , deleting either gene will affect the trait gene similarly . Six of the eight possible best-fit regression models correspond to the eight hierarchical relationships . It is notable that hierarchies can be indicated even without an interaction effect in the model . We must also consider that there is no hierarchical relationship between A and B , or that they do not affect the target gene ( Table 1 ) . We can distinguish between two types of parallelism . Model 4 , the two-gene additive model with no interaction , represents no epistasis . Model 3 represents buffering epistasis , in which both genes act on the target in the same direction , and the effect of deleting either is not apparent unless both genes are deleted . We refer to this as nonhierarchical epistasis since neither gene is upstream of the other . Deleting a deactivated regulator gene has no effect on the target gene , making it impossible to identify a biological relationship when regulators are deactivated . The remainder of Table 1 represents cases in which one or both genes do not affect the target gene . Expression traits supporting Model 8 ( no significant terms ) may represent target genes that do not lie downstream of A or B , and are uninformative . The result is one-to-many relationships between best-fit regression Models 1 , 2 , and 8 and their corresponding gene expression models . If the upstream gene of a hierarchical pair is turned off , we cannot know whether it is upstream or uninvolved . Typically , expression is measured from thousands of genes simultaneously and we do not expect them all to be informative . Even with clear interpretations for each trait individually , there is a challenge interpreting all traits together . We examine the distribution of all traits . Among informative traits associated with a best-fit model , the majority may represent the underlying biological relationship between the deleted genes . Van Driessche et al . used Dictyostellium discoideum wild type [20] and deletion mutant strains [21] to infer hierarchical epistasis among genes of the protein kinase ( PKA ) pathway . Each strain's gene expression profile was measured using cDNA microarrays with a common reference over 24 hours . These data are well suited for testing our methods for two reasons . First , the epistatic relationships between the deleted genes already have been characterized experimentally . Secondly , the mutant strains are genetically identical at all loci except the few being studied , i . e . there is no variation in their genetic background . The PKA pathway is associated with the developmental aggregation response to nutrient deprivation , which initiated midway through the time course . Data before and after aggregation were considered separately so we can clearly interpret the deletion effects in each signal state . The data represented fold-change on a logarithmic scale , which made the distribution of expression measurements approximately normal; we consider the implications of this in the discussion . We studied 1553 expression traits . The genes we used were measured in both experiments and differentially expressed in the wild type during aggregation [20] . Five deletion strains target genes of the protein kinase A ( PKA ) pathway that is involved in the response to starvation and activates aggregation . This provided three contrasts: pufA/pkaC , pufA/yakA , and regA/pkaR . Although there are ten possible contrasts for these five genes , only these three double mutants were generated , presumably because these are known direct relationships . For each contrast , some traits supported each model ( Figure 2 ) . Additionally , large number of traits showed no deletion effects ( i . e . support Model 8 ) . At a significance threshold of p<0 . 01 , a majority of traits supported Model 8 for every contrast pre-aggregation ( Figure S4 ) and for the regA/pkaR contrast post-aggregation . According to our interpretive models , Model 8 can indicate three possibilities . The first two are hierarchical relationships in which an upstream enhancing gene is turned off during aggregation . The last possibility is that the genes are uninvolved in the expression of the target and the deletions have no effect . Since not all target genes are downstream of the PKA pathway , it is logical that the deletions have no effect on these genes . Similarly , the PKA pathway is invoked during aggregation and it follows that the deletions may affect expression only after aggregation has begun . We assume that the target genes supporting Model 8 are not downstream of the pathway , and that the majority of the remaining target genes reflect the relationship within the pathway . To test this assumption , we looked at the overlap between the expression traits supporting Model 8 for each contrast . We found that all of the expression traits supporting Model 8 for the pufA/yakA contrast also supported Model 8 for the other two contrasts . These traits strongly support the assumption that they are not downstream of the PKA pathway . When we looked at these genes for both the pufA/pkaC and pufA/yakA contrasts , there was strong support for one model over all others post-aggregation . Not only did these models explain more traits post-aggregation , but the models also fit better . On average , the best-fit model explained over half of the expression variation ( R2≥0 . 5 , adjusted for degrees of freedom in the model ) for traits in the pufA/pkaC and pufA/yakA contrasts , and for both contrasts the R2 increased post-aggregation ( t-test with p<0 . 0001 ) . For the pufA/pkaC contrast , Model 2 had the most support of the seven non-null models . Model 2 corresponds to two possible interpretations . The first is that pkaC is the downstream gene , that pufA is a repressor , and that the pufA is turned off in the presence of the aggregation signal . Alternately , we could interpret it to mean that only pkaC has an effect on the downstream targets and that pufA is unrelated . For the pufA/yakA contrast , Model 6 had the most support among non-null models . This model has a one-to-one correspondence to our interpretive models . It asserts that yakA is an upstream repressor of pufA , and that yakA is turned on at aggregation . These conclusions both agree with what has been determined previously about the roles these three genes play during development [22] . YakA represses pufA , which then ceases to repress pkaC . The regA/pkaR was problematic because almost all traits supported Model 8 , the model with no effect terms . For the previous two cases , we assumed that these traits were not downstream of the pathway . Given this assumption , we could have concluded that regA and pkaR were not involved with aggregation . However , the other two contrasts had 435 and 528 traits supporting Model 8 , while regA/pkaR has 1497 . Because of this discrepancy , we suggest that some proportion of these genes support the hierarchical model corresponding to Model 8: that one gene is an enhancer of the other and is deactivated by aggregation . According to previously published results , regA and pkaR work together to repress pkaC pre-aggregation and are in fact deactivated post-aggregation [23] . This is consistent with the potential hierarchical relationship . Because we are modeling nonadditive interactions , the logarithmic scale transformation on these data can potentially alter the results relative to untransformed data [3] , [24] . To test this , we exponentiated the data and repeated our method . Despite dramatic changes to the shape of the data distribution , the resulting distribution of best-fit models agreed with the results presented above . Again , a majority of traits showed no deletion effects ( i . e . support Model 8 ) . Model 2 had the most support for the pufA/pkaC contrast , Model 6 had the most support for the pufA/yakA contrast , and Model 8 had near complete support for the regA/pkaR contrast using the post-aggregation data ( Figure S5 ) . Interestingly , this does not imply that each trait supports the same model regardless of the scale transformation . In fact , only 57% and 47% of traits support the same model with the untransformed data for the pufA/pkaC contrast and pufA/yakA contrast respectively . However , in both these cases the vast majority of changed traits support Model 8 . This result amends our previous interpretation of the traits supporting Model 8; in addition to genes not downstream of the pathway , there may be some proportion of genes for which expression changes due to deletion is not detectable due to issues of scale . Fewer traits supported Model 8 using transformed data , suggesting that these data may be more informative using the logarithmic transformation . Thus , in all three cases our best-fit regression models correspond to a set of interpretative models that includes the true relationship between the genes . Certain regression models have a one-to-many relationship with the interpretive models , but in these cases the number of candidate interpretive models is reduced to a few . Only one interpretation corresponds to Model 6 , which makes the pufA/yakA contrast straightforward to describe . In evaluating pufA/pkaC , Model 2 corresponds to one hierarchical model and one single-gene model . Since the pufA/yakA contrast provides evidence that deleting pufA has an effect , the hierarchical model is a preferable interpretation to the pkaC only model . As we vary the significance threshold for model selection , our results are robust . The best-fit model among models 1–7 was the same for p-value thresholds from 0 . 05 to 0 . 001 ( Figure S6 ) . As the selection criterion becomes stricter we reject more effects as not significant , and more traits support Model 8 . Measuring transcript abundance within a cell will remain a fundamental interest to biologists . Gene expression technologies have become popular over the past decade because of their ability to capture many genes simultaneously . Analyses that traditionally focused on a few genes now must be expanded to consider entire genomes . At this scale , the relationships between genes are of as much interest as the genes' individual effects . Many methods exist to infer gene networks or pathways from expression profiles [25] . Most of these require large datasets and result in large network diagrams that are difficult to interpret . These approaches are useful because they provide a genome scale view of transcription , and they are convenient because they can be applied to data from a variety of easily accessible sources . However , there is a continuing need for experiments that allow us to infer pathways directly . The classical epistasis experiment we recount in our results [21] is one such approach . Because it targets gene pairs directly , we can build pathways a relationship at a time . This local approach results in pathway diagrams that are easily comprehended and biologically relevant . Additionally , it associates genetic variation with expression variation . For these reasons , these types of experiments will be increasingly useful in constructing biological systems diagrams . While there are currently few experiments that measure expression in a genetically variable population , their number is increasing rapidly . Our motivation is to provide a conceptual framework in which these and related experiments can be interpreted . We have addressed the simplest genetically variable data structure for identifying epistasis , in which individuals vary at only two loci , but our ideas can be applied to a range of similar data . Because expression data are continuous by nature , we must address them with quantitative methods . Regression analysis is a standard technique to relate continuous variables . Using a multiple regression model to estimate gene effects and interactions has several advantages . First , it allows us to consider information from all the deletion mutants and the wild type simultaneously . Additionally , it estimates an effect for each allele , allows for variance in allelic effects , and separates these effects from error variance . In a traditional epistasis analysis the double mutant is compared to each single mutant in a rule-based manner , and the two nearest trait values determine epistasis . In contrast to our method , this method does not take advantage of all the information from a given contrast , and it is difficult to distinguish signal from noise . Myriad sophisticated techniques exist for fitting multiple regression models , and these should be employed based on the distributional properties of particular data . We consider individual expression traits rather than an expression profile . A gene expression model represents each trait , but we must infer the correct biological model through the results from the regression step . A corresponding regression model represents each possible gene expression model , but these relationships are not always one-to-one . Hierarchies in which an upstream gene is turned off by a signal are confounded with cases in which the gene has no effect . It makes sense that we cannot observe the effect of a deletion if the gene is already turned off in the wild type . Nonetheless , our framework was consistent with previous characterizations of the pathway in every case . Scale transformations are common in genetics and genomics so that data meet statistical testing assumptions such as normality and homoscadasity [3] . Logarithmic transformations are ubiquitous in the literature for gene expression data such as that presented in our results . However , models with nonadditive interactions are subject to the scale of the data , and transformations can result in support for alternative models . This is a long-standing problem with describing epistasis for complex traits [24] . Often it is difficult to know the most biologically appropriate scale , and the scale is instead often chosen arbitrarily based on the available measurement or statistical convenience . For gene expression traits the scale issue is even more complex . Since there are wide differences in the range of expression variation between genes , it is likely that no one scale will allow detection of the underlying biological interactions for all expression traits . The relationship between scale and epistasis is an area that demands further study , particularly in this era of genetics on biomolecular traits such as gene expression that have not been well studied in this context . When we performed the same analysis on log-transformed and untransformed post-aggregation data , about half the traits supported a different best-fit model , yet the distribution of results led to the same conclusions regarding the underlying relationship between the deleted genes . This suggests our conclusions may be robust to scale effects that would affect single traits because they are based on the distribution of all traits . Those traits that are affected by scale trend toward having no detectable deletion effects with untransformed data . This further confounds the roughly one-third of traits supporting Model 8 , which may also suggest an upstream enhancer or a trait truly unaffected by the deletions . While we do not discount scale effects , we assume most of these traits fit the last category because of the high percentage of these traits , the consistency of traits supporting Model 8 between contrasts , and the logic that deletions should affect only downstream genes . Whichever the case , these concerns make a strong argument for interpreting the distribution of results across expression traits . This contrasts with methods that consider all traits as an expression profile . These assume the profile as a whole supports one underlying pathway [21] . Using our method , it is straightforward to interpret a range of experiments . The alleles being studied do not need to be null alleles , e . g . deletions . The same method could be applied to over-expressed genes , or any polymorphic locus . Additionally , the method can accommodate experiments investigating multiple loci and higher order interactions . Three-way and four-way epistasis models follow from the same principles as the two-way models we present . The regression model is very flexible and easy to extend by adding a parameter for each locus plus interaction terms . Connecting these statistical models to biological models follows the same process we have illustrated . The strengths of our approach are particularly apparent in multi-locus models because we provide a means for estimating effects using the entire population of mutants simultaneously . The number of genotypes increases by a power of two for each additional gene included in the experiment; with a three-locus experiment having eight genotypes . As the number of necessary pair-wise comparisons increases , they will contain more undetected error and become more difficult to interpret . Environmental effects can also be included in the model at the expense of increased complexity in interpretation . We considered observations before aggregation and after aggregation separately in our example for simplicity . By proceeding to add genetic and environmental complexity , it is apparent how the classical epistasis framework connects to the quantitative genetic paradigm . An additional benefit of our method is that it enables comparisons between any population-based expression analyses . Whether study populations consist of deletion mutants , experimentally designed crosses , inbred lines , chromosome substitution strains , or natural populations , each expression trait is the same . For this reason , comparing these results is highly desirable . Estimating the allelic effects and interactions for each expression trait allows direct comparison across a variety of genetic backgrounds . By embracing a common interpretive framework to a range of experiments that use gene expression as a trait , we can integrate results and form clearer insights into the genetic control of systems . We used data originally presented by Van Driessche et al . We use data from Dictyostellium discoideum wild type [20] and eight deletion mutant strains ( pufA− , pkaC− , pufA−pkaC− , yakA− , pufA−yakA− , regA− , pkaR− , regA−pkaR− ) [21] . They measured each strain's gene expression profile over a time course using cDNA microarrays and a common reference that was pooled from all time points . Expression was measured thirteen times over 24 hours and captured the developmental aggregation response to nutrient deprivation , which initiated midway through the time course . We grouped observations before ( hours 0 , 2 , 4 , 6 ) and after ( hours 14 , 16 , 18 , 20 ) aggregation . Expression at these time points is highly correlated ( [Figure 2 in 20] ) and consistent with the regulatory changes previously reported . This data pooling increased the sample size for our regression analysis . Observations during the transitional period ( hours 8 , 10 , and 12 ) were disregarded , as were observations in the late stages of development that were less correlated ( hours 22 and 24 ) . The data represented fold-change on a logarithmic scale . We studied 1553 genes that were measured in both experiments and differentially expressed in the wild type during aggregation [20] . We fit models in the R statistical environment [26] . Stepwise backwards selection entails fitting a fully parameterized model , then eliminating model terms that do not meet a specified significance threshold . The model is refit with the remaining terms until no further terms can be dropped .	Epistasis has long had two slightly different meanings depending on the context in which it is discussed . The classical definition describes an allele at one locus completely masking the effect of an allele at a second locus . Such relationships can be interpreted as hierarchical , and they can be combined to infer genetic pathways . In quantitative genetics , epistasis encompasses a wide range of interactions and can be extended to more than two loci . These two definitions coexist because they are typically applied to different types of study populations and different types of traits . The current trend is to treat gene expression as a trait in a variety of genetic backgrounds . This provides reason to revisit epistasis in this new context . We accommodate the continuous nature of gene expression using ideas from quantitative genetics , but retain the hierarchical interpretation of the classical experiment . These hierarchical relationships are the building blocks of systems diagrams and genetic pathways . This framework can serve as a foundation for future epistasis analyses based on genomic data .	[ "Abstract", "Introduction", "Results", "Discussion", "Materials", "and", "Methods" ]	[ "genetics", "and", "genomics/complex", "traits", "computational", "biology/systems", "biology", "genetics", "and", "genomics/gene", "expression" ]	2008	From Classical Genetics to Quantitative Genetics to Systems Biology: Modeling Epistasis
Dendrites of pyramidal cells exhibit complex morphologies and contain a variety of ionic conductances , which generate non-trivial integrative properties . Basal and proximal apical dendrites have been shown to function as independent computational subunits within a two-layer feedforward processing scheme . The outputs of the subunits are linearly summed and passed through a final non-linearity . It is an open question whether this mathematical abstraction can be applied to apical tuft dendrites as well . Using a detailed compartmental model of CA1 pyramidal neurons and a novel theoretical framework based on iso-response methods , we first show that somatic sub-threshold responses to brief synaptic inputs cannot be described by a two-layer feedforward model . Then , we relax the core assumption of subunit independence and introduce non-linear feedback from the output layer to the subunit inputs . We find that additive feedback alone explains the somatic responses to synaptic inputs to most of the branches in the apical tuft . Individual dendritic branches bidirectionally modulate the thresholds of their input-output curves without significantly changing the gains . In contrast to these findings for precisely timed inputs , we show that neuronal computations based on firing rates can be accurately described by purely feedforward two-layer models . Our findings support the view that dendrites of pyramidal neurons possess non-linear analog processing capabilities that critically depend on the location of synaptic inputs . The iso-response framework proposed in this computational study is highly efficient and could be directly applied to biological neurons . A key objective of theoretical neuroscience is the development of simplified neuron models that incorporate relevant features of neuronal function while abstracting away nonessential complexity . According to the classical point-neuron assumption , synaptic inputs to a neuron sum linearly at a single integrative node , the soma , where the resulting membrane potential is transformed non-linearly to generate the neuronal response [1 , 2] . Although this simplified description might approximate the dynamics of certain neuron types , it is challenged for pyramidal cells that exhibit complex morphologies and spatially modulated distributions of ion channels ( for recent reviews see [3 , 4] ) . These characteristics generate regenerative events , such as Na+ or NMDA ( N-methyl-D-aspartate ) spikes , which are localized in specific branches or subtrees so that the neuron can no longer be described by a single voltage compartment . It has been previously proposed that individual dendritic branches act as independent functional subunits within a two-layer feedforward model [5–7] . The first layer represents the computations of the individual branches , i . e . , the voltage responses to local synaptic inputs , which are linearly summed and passed through a final non-linearity representing the second layer . Experimental and simulation studies consolidated the two-layer model as a suitable abstraction of synaptic integration in basal and proximal apical dendrites [8–11] . Häusser and Mel [12] suggested that even the distal apical tuft may function as a separate two-layer model that feeds into a proximal two-layer model , but a rigorous test of this hypothesis is still missing . In support of it , uncaging experiments have shown that NMDA spikes elicited in the fine branches of tuft dendrites generate voltage transients that are largely confined to the same branch and strongly attenuated across branches [13] . However , NMDA spikes in different branches can sum to initiate an apical calcium spike that propagates back into the distal branches . The impact of this electrical coupling on the core assumption of the two-layer model—subunit independence—remains unclear . Related work [11] has shown that synaptic inputs to nearby branches in tuft dendrites cannot be described by a two-layer model if its integrative node is located at the soma . This observation hints at a partial breakdown of the functional independence of individual branches but might be attributed to an additional non-linearity between the site of the summation of the subunit outputs and the soma—the calcium spike initiation zone . Estimating this non-linearity poses a major problem for tests of the two-layer hypothesis . It requires that the result of the summation of the subunit outputs is known . But this information is rarely available because of the difficulty to record from the fine dendritic branches in the tuft . This obstacle can be bypassed by iso-response methods . For this approach inputs to a system are varied such that a chosen output measure stays constant . As shown by various examples in different neural systems , iso-response methods can readily be integrated into neurophysiological experiments ( see [14] ) . All that is needed is a closed-loop setup with which measurements of the neural output variable can be used to tune the inputs . The stimuli that result in constant outputs define “iso-response curves” that are independent of the specific form of the system’s final non-linearity . An estimation of this non-linearity is not required . Here , we show that the shape of the subunit non-linearities of a two-layer model can be read off from only a few iso-response curves , and that the validity of a two-layer model can be assessed by additional test iso-response curves . We apply this theoretical framework to a detailed multi-compartment model of a CA1 pyramidal cell [6] and investigate the dendritic integration properties of different parts of the apical dendrites including proximal subtrees and the tuft . For this purpose , we analyze sub-threshold somatic membrane-potential responses to brief simultaneous synaptic inputs to pairs of dendritic branches . The two-layer model captures the somatic response to proximal inputs but fails for tuft dendrites . We generalize the two-layer model for the tuft , relax the core assumption of subunit independence , and incorporate additive and multiplicative feedback from a subsequent processing stage . This new model accurately predicts the somatic response to synaptic inputs on proximal apical branches as well as on tuft dendrites . The functional form and range of its non-linear feedback can be readily identified through the iso-response method . We stimulated pairs of branches located in the proximal ( P ) apical dendrites . First , we investigated for which branch pairs the somatic response can be successfully explained by a model that satisfies the classical point-neuron assumption . Here , synaptic inputs are summed linearly at the soma and subsequently transformed non-linearly to generate the neuronal response . We refer to this model as point-neuron model . We found that a large fraction of the proximal apical branches support linear dendritic integration ( Fig 3A ) . In particular , all branches except P13-P16 , P21 and P24 can be collected into one linear functional subunit . However , non-linear dendritic integration occurs in the distal branches of two proximal apical subtrees . More precisely , the point-neuron hypothesis fails only if one of the stimulated branches is part of these subtrees and the other branch is not . Therefore , local synaptic input is first summed linearly within these subtrees and then transformed non-linearly to generate the somatic response . Next , we tested the two-layer model for P-P branch combinations . We find that the model provides an accurate description of dendritic integration for all tested branch pairs with errors below 4% ( Fig 3D ) . The reconstructed subunit characteristics are either linear or supralinear functions ( Fig 3J ) and can be divided into three classes ( S6D and S13 Figs ) according to their input impedance ( S7 Fig ) . First , there are branches close to the trunk ( S6A Fig ) with approximately linear subunit functions . Second , there are branches with weak supralinearities where synaptic inputs trigger somatic spikes before the local depolarizations reach the threshold for dendritic non-linearities . Finally , there are branches with large local depolarizations ( S7A Fig ) that strongly attenuate from the dendrite to the soma ( S7B Fig ) . These branches have supralinear subunit functions with only small linear onsets and rapid saturations as observed in previous simulation studies for distal synaptic input [11 , 15] . The variability of the slopes of these functions reflects the differences in the distances from the stimulation sites to the soma [11] . Supralinear non-linearities occur only for those branches ( P13-P16 ) for which the point-neuron model fails . In an analogous manner , we tested the feedforward two-layer model for oblique branch combinations . We find that the feedforward model accurately predicts the somatic responses to synaptic inputs for all tested branch pairs ( S12 Fig ) . Overall , the feedforward two-layer model is an accurate description of synaptic integration in proximal apical and oblique dendrites . Next , we stimulated pairs of dendritic branches located in the proximal ( P ) apical dendrites and tuft ( T ) dendrites . The point-neuron model fails for nearly all T-T branch pairs ( Fig 3B ) . The few exceptions might be attributed to the specific choices of test iso-response curves and presumably disappear for a different selection of curves . Surprisingly , the majority of the T-T branch pairs is not well described by the two-layer model either , with errors mostly larger than 25% ( Fig 3E ) . An accurate description is only possible within the distal dendrites of one particular subtree ( T22-T27 ) . However , if one branch is located outside of the subtree the model fails . The shapes of the normalized subunit non-linearities for the subtree T22-T27 are shown in Fig 3K ( S14 Fig ) . The point-neuron model fails for most T-P branch combinations as well . Only a few branch pairs consisting of a tuft branch close to the trunk and a linear proximal apical dendrite ( as identified above ) can be modeled by a point neuron . On the other hand , the two-layer model fails for one of the tuft subtrees ( labels starting with T2 ) but mostly predicts the iso-response curves for the other subtree ( labels starting with T3 ) , apart from branch T34 . The reconstructed subunit functions for branch pairs with low errors ( <4% ) resemble those observed for P-P and T-T branch pairs but incorporate additional non-linearities between the outputs of the dendritic branches and the site of their summation ( Fig 3L and S15 Fig ) . Overall , the two-layer model fails to describe dendritic integration in the tuft . We relax the assumption of subunit independence and generalize the purely feedforward model by feedback loops ( Fig 4A ) . The somatic response is now given by the implicit function r = f 3 ( f 1 ( g 1 mult ( r ) · s 1 + g 1 add ( r ) ) + f 2 ( g 2 mult ( r ) · s 1 + g 1 add ( r ) ) ) , where g i add and g i mult denote additive and multiplicative feedback , respectively . As the feedback is determined by the output r ( see Fig 4A ) , it is constant along iso-response curves . Additive feedback shifts the iso-response curves in the stimulus space but does not change their shape . Negative feedback shifts iso-response curves further apart ( Fig 4B ) , whereas positive feedback brings them closer to each other ( Fig 4C ) . Multiplicative feedback rescales the iso-response curves in each dimension of the stimulus space ( Fig 4D ) . For two-layer models with feedback , we quantify the error as the variance of the predicted sum of the subunit non-linearities on the test iso-response curve . As before , we normalize by the corresponding variance of the two training iso-response curves . In contrast to the feedforward model , we now analyze the error of the argument of the final non-linearity f3 and not its output . The final non-linearity can not be estimated from r and the reconstructed subunit non-linearities , because the feedback is unknown for stimuli outside of the training and test iso-response curves . However , both error measures , i . e . , the variance of the sum of the subunit non-linearities on the test iso-response curve and the response variance on the test iso-response curve , are qualitatively similar for the feedforward two-layer model ( S8 Fig ) . The extended two-layer model was applied to T-T and T-P branch pairs ( Fig 5 ) . Additive feedback is sufficient to explain synaptic integration for more than 72% of all branch pairs ( 4% threshold ) . In particular , it describes synaptic integration for all branch pairs within the same subtree ( Fig 5A ) . Additional multiplicative feedback is required to achieve low prediction errors for a few pairs of branches from different subtrees ( Fig 5B ) . For T-P branch pairs additive feedback is sufficient to explain synaptic integration of more than 91% of all branch pairs ( 4% threshold ) . A combination of multiplicative and additive feedback further reduced the prediction error ( Fig 5E ) . In general , purely additive feedback results in lower prediction errors compared to purely multiplicative feedback ( Fig 5C and 5F ) . The additive feedback is almost always positive and up to 10% of the input range ( Fig 5G and 5H ) . This value has to be compared to the average distance between training iso-response curves of about 14% of the stimulus range . Restricting the additive feedback to only positive values has only an insignificant effect on the model error ( Fig 5K ) . Furthermore , the feedback strengths are similar for T-T and T-P branch pairs . The reconstructed multiplicative feedback values peak around 1 and are small ( Fig 5I and 5J ) . A second peak around 1 . 1 for T-T branch pairs indicates that for some pairs stronger multiplicative feedback would further reduce the prediction error . However , we limited the multiplicative feedback values to a range such that the test iso-response curves were long enough to allow a reliable estimation of the model validity . Even within this limited range accurate two-layer models were identified . For a significant fraction of branches the multiplicative feedback values are below one . A summary of the errors of all models is show in Fig 5K . A two-layer model without feedback accurately describes synaptic integration in proximal apical dendrites , but fails for the tuft . Additive feedback alone is sufficient to achieve a low mean prediction error . Restricting the additive feedback to positive values has only a minor effect on the performance . Fig 6 shows cases of branch pairs where dendritic integration can be described by a two-layer model with additive feedback but not without feedback ( Fig 6A–6D ) , and where dendritic integration can be described by a two-layer model with additive and multiplicative feedback but not without multiplicative feedback ( Fig 6E–6H ) . Fig 6B and 6F illustrate the input regime where the feedforward two-layer model without feedback fails . For both two-layer models two sets consisting each of three iso-response curves were used for training so that the reconstructed subunit non-linearities can be compared . For the branch pair where additive feedback is required the reconstructed subunit non-linearities of a two-layer model without feedback are shifted ( Fig 6C ) . For the branch pair where multiplicative feedback is required the reconstructed subunit non-linearities of a two-layer model that implements only additive feedback are rescaled ( Fig 6G ) . We have chosen feedback functions that compensate for both of these transformations . Moreover , the impact of this feedback on the response can be visualized as a shift and a rescaling of iso-response curves . What is the biophysical basis for the feedback in the functional subunits of tuft dendrites ? To answer this question , we repeat the simulation study for a representative subset of branches throughout the dendritic tree ( S9 Fig ) while blocking active ion channels exclusively present in the tuft . These are three voltage-gated calcium currents , an A-type potassium current and a hyperpolarization-activated cation current ( Ih ) . The block of each current improves the performance of the feedforward two-layer model but only the block of voltage-gated calcium currents results in close-to-optimal performance ( Fig 7A ) . This finding indicates that calcium currents disrupt the independence of functional subunits in the tuft by shifting their non-linearities . In this study , we focus on dendritic integration on short time scales that are relevant for spike-timing codes . To test the generality of our findings for other neural coding scenarios we also investigate computations on spike counts within longer time intervals . Here , the input and the response of the pyramidal cell are encoded by the rate of synaptic input and the firing rate of the neuron , respectively . Ten synapses are located with equal spacing on each of two dendritic branches ( for a representative subset of branches ) . All synapses on one branch are stimulated with homogeneous Poisson spike trains with identical rates and a duration of 500 ms . The input rates for the two branches varied individually up to a frequency of 40 Hz . The neuronal response , i . e . the firing rate of the pyramidal cell during stimulation , is averaged over 20 trials . As for computations on short time scales the point neuron model fails to predict the neuronal response to synaptic input to proximal as well as to tuft dendrites ( Fig 7B ) . However , the feedforward two-layer model turns out to be sufficient to describe the firing-rate based non-linear integration . The reconstructed subunit functions for P-P and T-T branch pairs resemble those observed for precisely timed inputs ( Fig 7C and 7D ) . In contrast , the subunit functions for P-T branch pairs ( Fig 7E ) do not incorporate additional non-linearities as found for precisely timed inputs ( Fig 3L ) . Similarly , the iso-response method can be applied to other neural coding scenarios as long as the response variable is a continuous function of the inputs . Additive and multiplicative feedback terms correspond to horizontal shifts and horizontal rescalings of the subunit non-linearities , respectively . Various biophysical mechanisms have been associated with these operations ( for a review see [16] ) . Shifts of dendritic non-linearities have been observed for additional current or conductance inputs in the proximity of a driving input [15 , 17 , 18] . In particular , it was shown for basal dendrites that distal modulatory input within the same branch mainly shifts the subunit non-linearities [15 , 18] . In contrast , proximal modulatory input shifts and changes the slope of the subunit non-linearities . These asymmetric interactions have been linked to the voltage-dependence of NMDA channels and an increase in input resistance with increasing distance between the input and the soma . Our results indicate that a similar modulation occurs across dendritic branches in the tuft . Horizontal shifts of the subunit non-linearities alone are sufficient to explain the response properties of most of the dendritic branches . However , for a small fraction of branches a change in the slope of their subunit non-linearities is required as well to describe the synaptic integration properties . This modulation may be either attributed to a change in the gain of the subunit non-linearity , which changes its maximum value as well , or a horizontal rescaling , which does not change the maximum value . Horizontal rescaling has been observed for modulatory input co-localized with driving input [18] , whereas gain control was reported for proximal modulatory input [15 , 18] . Here , we cannot discriminate between both cases for all branch pairs that implement multiplicative feedback . The subunit non-linearities were not driven to saturation for somatic responses below the action potential threshold , and no conclusion can be drawn about the change of their maximum value . A major difference between the distal tuft and basal dendrites is the presence of hyperpolarization-activated cation currents ( Ih ) and voltage-gated calcium channels . In agreement with experiments [13] we find that Ih currents facilitate subunit independence in the dendritic tuft . Moreover , our simulation study predicts that feedback interactions in tuft dendrites are mediated by calcium currents at the calcium spike initiation zone . Previous work [11] applied the two-layer model to predict the firing rate of a detailed pyramidal cell model in response to homogeneous Poisson input . This feedforward two-layer model correctly predicted the somatic response to a distant pair of apical tuft dendrites but failed for synaptic inputs to nearby branches in the tuft ( see Fig S4 in [11] ) . In contrast , our results show that computations based on firing rates can be well described by a feedforward two-layer model even for tuft dendrites . The failure of the two-layer feed-forward model in [11] to predict the firing-rate responses to synaptic inputs to nearby branches in the tuft can be attributed to either a disruption of the functional subunit independence or a distorted reconstruction of the model parameters . In [11] , the final non-linearity of the two-layer model is approximated by the somatic f-I curve and the individual subunit outputs are assumed to contribute linearly to the total somatic current . If , under these conditions , the dendritic branches operate as independent functional subunits , their non-linearities can be estimated by a least-squares linear regression of the total somatic current on the synaptic inputs . However , for synaptic inputs to nearby distal branches , as shown in Fig S4C of [11] , additional dendritic non-linearities occur between the location of the subunit output summation and the soma . In this case , the somatic current is no more a linear sum of the individual branch outputs and linear regression fails to correctly identify the dendritic non-linearities . As we find no evidence for a violation of the functional independence of dendritic branches in the tuft for computations based on firing rates our results suggest that the reduced performance of the two-layer model in [11] can be attributed to a distorted reconstruction of the final non-linearity . The iso-response method has two advantages when compared to the method proposed in [11] . First , it does not require an estimate of the final non-linearity of the two-layer model that is rarely available for synaptic integration in the tuft because of the difficulty to record from fine dendritic branches . Once the subunit non-linearities of a feedforward two-layer model are identified its final non-linearity can be reconstructed from measurements of the somatic responses to single branch stimulations of varying strength . Second , iso-response curves are optimally efficient in that they are based on the smallest amount of data needed to identify the subunit non-linearities . We have demonstrated that feedback in tuft dendrites impacts synaptic integration on the short time scales relevant for triggering single somatic spikes . The precise timing of spikes relative to hippocampal theta oscillations has been shown to carry information about the location of a rat within its environment [19 , 20] . Moreover , recent work [21] supports the hypothesis that degraded spike timing in the hippocampus results in memory impairments . In contrast , we found no evidence for the functional relevance of feedback for mean somatic responses , i . e . spike counts within longer time intervals . This difference might be attributed to two reasons . First , the brief synaptic input to pairs of dendritic branches was injected simultaneously , whereas the Poisson spike trains used for rate inputs were not synchronized . Simultaneous synaptic stimulation has been shown to enhance the non-linear integrative properties of dendrites [22] . Second , for brief synaptic stimulation we investigated somatic sub-threshold responses in contrast to multiple spikes for rate responses . As proposed in [11] , action potential generation might facilitate subunit independence . Overall , our findings consolidate the feedforward two-layer model as a suitable abstraction of neuronal computations based on firing rates [5–12] but not on precisely timed inputs . Other work has focused on non-linear interactions between tuft dendrites and the soma [23–27] . In contrast , we investigated dendritic integration on a finer spatial scale—within subtrees of the dendritic tuft . Notwithstanding , we observed for all but one proximal-distal branch pair that additive feedback was sufficient to describe the somatic response . In contrast , previous work found a multiplicative coupling between the soma and the tuft [25] , i . e . distal depolarizing current injections increased the gain and reduced the rheobase of the firing rate-current response at the soma . However , the underlying mechanism relied on backpropagating action potentials that haven’t been triggered in our study . We have shown that feedback between the proximal apical dendrites and the tuft occurs also in the sub-threshold regime , and this coupling is additive in most cases . We hypothesize that this additive coupling between tuft dendrites serves as a gating mechanism similar to the proximal-distal coupling . Weakly stimulated dendritic branches in the tuft that would otherwise fail to detect individual input patterns with local membrane potentials sub-threshold to the non-linearities combine the collective evidence resulting in robust pattern recognition . We investigated dendritic integration properties of branches that each receive only a single synaptic input . Hence , the subunit non-linearities are one-dimensional functions of the synaptic input strength . It has been recently shown that pairs of synaptic inputs to the same branch in the basal dendrites are not purely summed but also processed non-linearly in a location dependent manner [15 , 18] . Conceptually , the two-layer model proposed here can be augmented to incorporate this intra-branch integration effects . Moreover , the theoretical framework based on iso-response methods can be applied to identify all non-linearities involved . We focused on spatial integration properties of tuft dendrites , where pairs of synapses are activated simultaneously . A complete description of dendritic integration in pyramidal neurons has to incorporate the experimentally observed temporal coupling between the proximal dendritic compartment , including the basal dendrites , soma and apical obliques , and the distal tuft [23–27] . We expect that the two-layer model can be incorporated into this two-compartment schema . The theoretical framework presented here can be used to identify two-layer models without the necessity of additional information beyond the stimuli and the responses and without extensive datasets . Overall , our findings support the view that the information processing capabilities of pyramidal neurons depend sensitively on the spatial distribution of synaptic inputs . This conclusion has been reached using model neurons . The proposed iso-response techniques could , however , also be integrated into multi-electrode or photo-stimulation paradigms to directly reveal the dendritic computations of biological neurons . The detailed compartmental model [28] was developed by Poirazi et al . [6] in the NEURON simulation environment . The model is well suited for our analysis because it was tested to replicate the regenerative events observed in the dendrites of pyramidal cells . The cell is a reconstructed CA1 pyramidal neuron ( Fig 1A ) and includes various active and passive membrane mechanisms known to be present in CA1 pyramidal cells , such as sodium and potassium currents , A-type potassium currents , m-type potassium currents , a hyperpolarization-activated h-current , voltage-dependent calcium currents , and Ca2+-dependent potassium currents . The densities and distributions of these currents are based on published data . Single synaptic events were triggered simultaneously in pairs of dendritic branches ( Fig 1A and 1B ) . Each synaptic input consisted of an NMDA and an AMPA-type conductance with a ratio of their peak values of 2 . 5 . Synapses were located at the centers of dendritic branches . For purely passive dendrites the NMDA conductances induced highly stereotypical supralinear somatic responses ( S10 Fig ) . The peak conductances were varied between zero and a maximum value . The maximum value was chosen for each synaptic input such that the strongest paired inputs generated a somatic action potential . Maximum values of NMDA conductances varied between 120 and 400 nS . Assuming a maximum conductance of 4 nS for single synapses [8 , 29–31] this corresponds to at most 30-100 synapses per branch—roughly the same input range as studied in related work [11] with at most 20-60 synapses per branch . For the experiments in Fig 7A we blocked L-type and T-type calcium channels with a high threshold for activation , an R-type calcium channel with a medium threshold for activation , an A-type potassium current , and a hyperpolarization-activated cation current ( Ih ) . For computations based on firing rates ( Fig 7B ) the synaptic conductance are chosen such that for maximum paired branch stimulations ( i . e . synaptic Poisson inputs at 40 Hz ) the neuronal response ( i . e . the mean firing rate during a stimulation duration of 500 ms ) is roughly in the same range as the input rates , i . e . with a mean of 40 Hz and a standard deviation of 25 Hz ( For some branches the somatic firing rate saturates below 40 Hz resulting in smaller maximum responses ) . We assume that the gradient of the response r = f3 ( f1 ( s1 ) + f2 ( s2 ) ) in terms of the inputs s1 and s2 exists and is non-zero in the whole synaptic input regime . This is reasonable because the somatic voltage depolarization is a strictly monotonic function of the synaptic input strengths . Under this condition , all stimuli that cause the same response form a one dimensional “iso-response curve” in the two dimensional input space ( for details see [14] ) . Each response value corresponds to a unique iso-response curve ( Fig 1B and 1C ) and each pair of s1 and s2 is an element of an iso-response curve . To estimate iso-response curves a cubic-spline fit of the response function r ( s1 , s2 ) was generated . For a specific response value ri the corresponding iso-response curve was found by solving r ( s1 , s2 ) − ri = 0 using root finding algorithms . The resulting curves were both stored as a function of s1 and as a function of s2 . Fig 2A–2C shows examples of iso-response curves for different combinations of supralinear and sublinear subunit non-linearities . In this simulation study , we measured three iso-response curves L : s 1 ↦ s 2 = L ( s 1 ) H : s 1 ↦ s 2 = H ( s 1 ) T : s 1 ↦ s 2 = T ( s 1 ) . ( 1 ) Two curves , L and H , were chosen at a low and high value of r for the identification of the subunit non-linearities and a third test curve , T , was chosen at a response value right below the threshold for action potential generation . The test iso-response curve was at least half as long as L or H . In the following a relation between iso-response curves and the subunit functions of a conventional two-layer model will be derived . In general , one iso-response curve is not sufficient to identify the subunit non-linearities . However , we will show that two iso-response curves can be used to identify f1 and f2 up to an affine transformation . Furthermore , a third iso-response curve can be used to indicate whether r can be described by a two-layer model within a certain input regime . Assume s1 and s2 = H ( s1 ) . The gradient of r in the direction of the iso-response curve H is zero and ∂ f 1 ( s 1 ) ∂ s 1 + ∂ f 2 ( s 2 ) ∂ s 2 ∂ H ( s 1 ) ∂ s 1 = 0 . ( 2 ) If s1 and s 2 ′ = L ( s 1 ) denote a second point on the iso-response curve L then the ratio of the derivatives of f2 at both points is given by ∂ f 2 ( s 2 ′ ) ∂ s 2 ′ ∂ f 2 ( s 2 ) ∂ s 2 = ∂ H ( s 1 ) ∂ s 1 ∂ L ( s 1 ) ∂ s 1 . ( 3 ) Analogous , for two points with identical s2 but different s1 = H−1 ( s2 ) and s 1 ′ = L - 1 ( s 2 ) follows ∂ f 1 ( s 1 ′ ) ∂ s 1 ′ ∂ f 1 ( s 1 ) ∂ s 1 = ∂ H - 1 ( s 2 ) ∂ s 2 ∂ L - 1 ( s 2 ) ∂ s 2 , ( 4 ) where L−1 and H−1 denote the inverse functions of L and H , respectively . The same relations are valid if the partial derivatives are replaced by finite differences ( Fig 2E ) . Given an odd number of n points that form a stairway { ( s 1 ( 1 ) , s 2 ( 1 ) ) , ( s 1 ( 1 ) , s 2 ( 2 ) ) , ( s 1 ( 2 ) , s 2 ( 2 ) ) , ( s 1 ( 2 ) , s 2 ( 3 ) ) , . . . , ( s 1 ( n / 2 + 1 / 2 ) , s 2 ( n / 2 + 1 / 2 ) ) } , ( 5 ) and lie on the two iso-response curves in alternating order with s 2 ( i ) = H ( s 1 ( i ) ) and s 2 ( j + 1 ) = L ( s 1 ( j ) ) ( 6 ) these relations constrain the subunit functions f1 and f2 at the points s 1 ( i ) and s 2 ( j ) up to an affine transformation that can be absorbed into f3 ( Fig 2F and 2G ) . For arbitrary points that do not form a stairway , the subunit non-linearities can be expanded in basis-functions ( see S1 Methods ) . In this case all iso-response curves of the two-layer model can be predicted within the range of input values used for the identification of the subunit non-linearities . We call two iso-response curves identical if they contain the same points in the input space irrespective of their associated response values . Then , the predicted iso-response curves and the observed iso-response curves are identical if and only if the responses to the stimuli on these iso-response curves can be described by a two-layer model . The “if” part follows from the fact that if the observed responses and the two-layer responses are the same then so are their iso-response curves . The “only if” part follows from the implication that if the observed iso-response curves and the iso-response curves of the two-layer model are identical then there always exists another two-layer model with a final non-linearity such that the resulting predicted responses match the observed ones . Point-neuron models implement a weighted linear summation of their inputs , i . e . r ( s1 , s2 ) = f3 ( w1 s1 + w2 s2 ) . The weights w1 and w2 were set to minimize the mean squared error between the predicted and the observed responses on the two iso-response curves L and H . For a full reconstruction of the point-neuron model and the two-layer model the measurement of f3 is required , e . g . , on the line s1 = s2 . The method was successfully tested on iso-response curves generated by artificial two-layer models ( S1 and S2 Figs ) . To find the subunit non-linearities f1 and f2 of a two-layer model with feedback , we analyze a hypothetical two-layer model without feedback that has the same subunit non-linearities f1 and f2 . The iso-response curve H of this hypothetical two-layer model without feedback can be obtained from an iso-response curve of a two-layer models with feedback , denoted as Hfb , with the transformation H ( s 1 ) = Δ g 2 mult H fb ( s 1 - Δ g 1 add Δ g 1 mult ) + Δ g 2 add Δ g 1 / 2 add = g 1 / 2 add ( r H ) - g 1 / 2 add ( r L ) Δ g 1 / 2 mult = g 1 / 2 mult ( r H ) / g 1 / 2 mult ( r L ) , ( 7 ) where rL and rH denote the response values on the iso-response curves L and H , respectively . This equation can be inserted into Eqs 3 and 4 . We applied a brute-force approach to find the constant optimal values of the feedback functions on the iso-response curves H and T . Feedback values were limited to a range such that the transformed test iso-response curves ( of the hypothetical two-layer model without feedback ) were at least half as long as each of the iso-response curves used for training . The method was successfully tested on iso-response curves generated by artificial two-layer models ( S3 and S4 Figs ) . The error of the optimized two-layer model was tested on the third iso-response curve T . Three error measures were applied . First , for two-layer models without feedback ( Fig 3 ) f3 can be reconstructed and the performance can be computed as the response variance on the test iso-response curve T normalized by the response variance of the training data σ r 2 = 1 K ∑ k = 0 K - 1 ( r ¯ - r k ) 2 ( r H - r L ) 2 . ( 8 ) Here , r ¯ denotes the mean response values of K samples on the test iso-response curve T , and rH and rL denote the response values on the iso-response curves H and L , respectively . Secondly , for two-layer models with feedback ( Fig 5 ) the performance can be similarly assessed for the summed output of the subunit non-linearities m = f1 ( s1 ) + f2 ( s2 ) with σ m 2 = 1 K ∑ k = 0 K - 1 ( m ¯ - m k ) 2 ( m H - m L ) 2 , ( 9 ) where m ¯ denotes the mean value of K samples on the test iso-response curve T , and mH and mL denote the value of m on the two iso-response curves H and L , respectively . Finally , we also calculate for two-layer models without feedback the optimized mean square error between the predictions and the responses on the test iso-response curve T normalized by the response variance of the training data MSE ( r ) = 1 K ∑ k = 0 K - 1 ( r T - r k ) 2 ( r H - r L ) 2 , ( 10 ) where rT denotes the observed response on the test iso-response curve . In general , rT differs from the mean predicted response r ¯ as used for the calculation of σ r 2 . For the feedforward two-layer model , subunit independence can be tested ( Fig 6 ) if the gradient of r is given for four points A , B , C and D in the input space , where A = ( s 1 A , s 2 A ) , such that s 1 A = s 1 B , s 1 C = s 1 D , s 2 B = s 2 C and s 2 A = s 2 D . The angle α between the s1-axis and a gradient-vector is given by tan ( α ( s 1 , s 2 ) ) = d f 2 ( s 2 ) d s 2 d f 1 ( s 1 ) d s 2 . ( 11 ) It follows tan ( α D ) = tan ( α A ) · tan ( α C ) tan ( α B ) . ( 12 ) This relation can be used to predict αD from the three other angles . A non-zero prediction error indicates that the response cannot be described by a two-layer model . In Fig 6B and 6F we measured the gradients on two straight lines through the input space . First , we kept s2 at a constant low value and varied s1 over the whole range ∇r ( s1 , s2 = const ) . Then , we kept s1 at a constant low value and varied s2 over the whole input range ∇r ( s1 = const , s2 ) . From these two straight lines it is possible to predict all gradients of the input space and compare them to the actual measured gradients . Keeping the constant values s1 and s2 at a low level has the benefit that synaptic integration was alway found to be linear and can be described by a two-layer model without feedback . Therefore , it is possible to determine in what input regime feedback effects occur .	Pyramidal neurons are the principal cell type in the cerebral cortex . Revealing how these cells operate is key to understanding the dynamics and computations of cortical circuits . However , it is still a matter of debate how pyramidal neurons transform their synaptic inputs into spike outputs . Recent studies have proposed that individual dendritic branches or subtrees may function as independent computational subunits . Although experimental work consolidated this abstraction for basal and proximal apical dendrites , a rigorous test for tuft dendrites is still missing . By carrying out a computational study we demonstrate that dendritic branches in the tuft do not form independent subunits , however , their integrative properties can be captured by a model that incorporates modulatory feedback between these subunits . This conclusion has been reached using a novel theoretical framework that can be directly integrated into multi-electrode or photo-stimulation paradigms to reveal the dendritic computations of biological neurons .	[ "Abstract", "Introduction", "Results", "Discussion", "Materials", "and", "methods" ]	[ "control", "theory", "action", "potentials", "medicine", "and", "health", "sciences", "engineering", "and", "technology", "nervous", "system", "depolarization", "membrane", "potential", "electrophysiology", "neuroscience", "control", "engineering", "systems", "science", "...	2019	Tuft dendrites of pyramidal neurons operate as feedback-modulated functional subunits
Mutations accumulate during all stages of growth , but only germ line mutations contribute to evolution . While meiosis contributes to evolution by reassortment of parental alleles , we show here that the process itself is inherently mutagenic . We have previously shown that the DNA synthesis associated with repair of a double-strand break is about 1000-fold less accurate than S-phase synthesis . Since the process of meiosis involves many programmed DSBs , we reasoned that this repair might also be mutagenic . Indeed , in the early 1960′s Magni and Von Borstel observed elevated reversion of recessive alleles during meiosis , and found that the revertants were more likely to be associated with a crossover than non-revertants , a process that they called “the meiotic effect . ” Here we use a forward mutation reporter ( CAN1 HIS3 ) placed at either a meiotic recombination coldspot or hotspot near the MAT locus on Chromosome III . We find that the increased mutation rate at CAN1 ( 6 to 21 –fold ) correlates with the underlying recombination rate at the locus . Importantly , we show that the elevated mutation rate is fully dependent upon Spo11 , the protein that introduces the meiosis specific DSBs . To examine associated recombination we selected for random spores with or without a mutation in CAN1 . We find that the mutations isolated this way show an increased association with recombination ( crossovers , loss of crossover interference and/or increased gene conversion tracts ) . Polζ appears to contribute about half of the mutations induced during meiosis , but is not the only source of mutations for the meiotic effect . We see no difference in either the spectrum or distribution of mutations between mitosis and meiosis . The correlation of hotspots with elevated mutagenesis provides a mechanism for organisms to control evolution rates in a gene specific manner . Mutation is an important component of evolution . Organisms need to forge a fine balance between maintaining stasis while allowing enough flexibility so that some members of the population can survive when environmental change occurs . Mitotic DNA replication and repair is a highly accurate process with mutations arising only 3 . 8 or 6 . 4×10−10 per base pair per cell division for URA3 and CAN1 respectively [1] , despite a large burden of continual endogenous and exogenous DNA damage ( estimated to occur at a rate of 103 to 106 lesions per cell per day for most organisms [2] ) . Although mitotic mutations can result in reduced fitness and disease , such as cancer , it is the germ line mutations that contribute to the fitness of future generations and ultimately successful evolution . Our focus here is to determine the rate at which mutations arise as the cells traverse meiosis . An enigma exists between the fitness cost of having a sexual cycle and the near ubiquity of sex among eukaryotes . Asexual organisms are thought to be favored in the short term , but they eventually accumulate too many irreversible deleterious mutations for long-term survival ( Muller's ratchet; [3] ) . It is hypothesized that sexual reproduction improves fitness over the long run via assortment , by providing increased genetic variability , and a mechanism by which deleterious mutations are masked or eliminated [4] . Meiosis differs from mitosis in that diploid cells undergo two consecutive cell divisions to produce germ cells . Meiosis is a highly choreographed process that involves homologous pairing and recombination resulting in the segregation of homologous chromosomes [5] . Recombination occurs during the first meiotic prophase . Meiosis II is similar to a mitotic division where sister chromatid centromeres are segregated from one another . Recombination is strongly induced in the first meiotic prophase by programmed DNA double-strand breaks ( DSBs ) that are introduced by the Spo11 type II topoisomerase [6] . In budding yeast , the number of DSBs is estimated to be ∼160 per cell [7] of which ∼35% result in crossovers [8] , [9] . Meiotic recombination is not uniform across the genome , but rather occurs at either high or low levels , termed hotspots and coldspots respectively . The frequency of meiotic crossovers is positively correlated with the local frequency of Spo11-induced DSBs [10] that , in turn , appear to be influenced by the underlying chromatin context ( [11] , and references cited therein ) . Crossovers themselves are subject to crossover interference , where there are fewer than expected double crossovers near each other [12] . Our laboratory has previously demonstrated that repair of mitotic DSBs are accompanied by 100 to 1000-fold increase in mutations near the site of the break ( Break Repair Induced Mutagenesis -BRIM ) [13]–[15] . High levels of mutation have also been observed to occur during an HO induced mating type switching-like assay [16] , break-induced replication ( BIR ) where mutations are found as much as 36 kb from the initiating break [17] , or associated with fragile genomic sites [18] . Mutagenesis is also elevated during repair after telomere erosion [19] , [20] . A review of mutagenesis associated with DSB repair can be found in [21] . Adaptive mutation is a phenomenon characterized by stress-induced increases in mutation rates ( i . e . starvation ) , and is associated with increased recombination in both bacteria and yeast , and appears to function via a DSB repair pathway [22] , [23] . The Rev3/Rev7 translesion DNA polymerase ( Polζ ) is important for the majority ( 50–75% ) of spontaneous mutations in yeast [24] . We demonstrated that during repair of a mitotically introduced site-specific DSB , Polζ is important for >90% of all base substitution mutations , but only minimally important for the predominating frameshift mutations [13] , [25] . The role of Rev3 in mutagenesis of other DSB induced assays is context dependent ( see [21] for a review ) . In some assays mutagenesis is entirely dependent upon REV3 [20] , while in other assays it has an intermediate effect [17] , [18] , [25] , or is not required [16] . It is not clear what causes Rev3 recruitment to only some DSBs , but one possibility is the length of ssDNA produced during repair . ssDNA is more susceptible to DNA damage than dsDNA , and synthesis on the damaged template may require a translesion polymerase [17] , [20] , [25] . Mutations in Pol ζ do not appear to affect sporulation or viability , although the Rev1 protein ( found in complex with polζ [26] ) has been shown to physically interact with Spo11 [27] . Since recombination during the first meiotic prophase proceeds via DSB repair , we wondered if meiotic recombination was also mutagenic . In the early 1960s Magni and Von Borstel [28] observed an increased level ( 6–20 fold ) of reversion of auxotrophic alleles during yeast meiosis , a process they termed “the meiotic effect” . Subsequently , Magni [29] demonstrated that 71% of the revertants analyzed had an associated crossover while the expected crossover association was 15% . This suggested that the increased mutagenesis might be linked to meiotic recombination . Magni also used the CAN1 gene as a forward mutation reporter [30] . The CAN1 gene encodes the arginine permease and allows cells to take up the toxic arginine analog canavanine [31] . Thus , cells with a wild type allele of CAN1 are sensitive to canavanine , whereas mutations that inactivate the permease render the cell resistant to canavanine . The experiments of Magni and von Borstel , while seminal , had three caveats that we address here . 1 ) For the reversion experiments the nature of the alleles used to score reversion is unknown , hence the required reversion events are also unknown . 2 ) For the experiments with CAN1 the mutation rates in diploids cannot be measured . Based on Magni's previous results he assumed that the diploid mitotic mutation rate was the sum of each of the haploid mutation rates , an assertion that is unlikely because the two parents differed by ∼38 fold in their rates to canavanine resistance ( from 1 . 4×10−9 to 5 . 3×10−8 ) , suggesting that other factors influenced the measured mutation rate . Also we now know that recombination and repair pathways are different in a/α cells than in either a or α cells [32]–[34] . 3 ) Because the CAN1 is located beyond any essential genes on the left arm of chromosome V , it opens a terminal deletion pathway for mutagenesis that may not be a general mechanism [35] . Several additional attempts have been made to confirm these observations but all suffered from similar caveats or insufficient data ( see Discussion ) . In the current study we have revisited the meiotic effect using a diploid with a single CAN1 gene coupled to a HIS3 gene so that by maintaining selection for His+ cells cannot become canavanine resistant simply by loss of heterozygosity ( LOH ) . Importantly , we find that the meiotic effect is entirely dependent upon Spo11 , consistent with the idea that mutations are introduced during DSB repair . The location where we place the CAN1 HIS3 cassette affects the rate of mutation in a manner consistent with frequency of recombination at that locus . We speculate that organisms can control the rate of evolution of different genes by controlling their location relative to meiotic recombination hotspots . We constructed a 3 . 8 kb cassette containing wild type HIS3 and CAN1 such that the two genes are transcribed in opposite directions ( Fig . 1 ) . The normal HIS3 and CAN1 loci were deleted from the parental strains ( see Materials and Methods for exact coordinates of each gene used ) . We inserted the 3 . 8 kb cassette into either of two different locations on chromosome III . Mutations were selected as His+ Canr cells . For each experiment at least 18 colonies were grown to mid-log phase in rich media ( see Materials and Methods for detailed experimental procedures ) . Initially , we measured the mutation rate in the haploids containing the substrate . We then made diploids carrying the HIS3 CAN1 cassette . For the diploid strains , part of each culture was used to determine the mutation rates after mitotic growth while the remainder was transferred to sporulation medium for ∼5 days , and random spores ( disrupted asci ) were plated to determine the change in mutation rate after a single meiotic division . His+ Canr mutants arising during mitotic growth were examined to determine if they were accompanied by crossovers as described in Materials and Methods , however we found very few crossovers among the mitotically arising events in the intervals being scored . His+ Canr mutants existing in each culture prior to sporulation were subtracted from the total after meiosis to allow the measurement of mutations created during meiosis . This allows us to calculate a mutation rate per meiosis . Meiotic recombination was established by tetrad dissection . Because the mutation rate is low , we did not find any Canr mutants in the tetrads we examined . By using random spores , we could plate many more cells . The starting strains were heterozygous for ADE2/ade2 and CYH2/cyh2 ( Table 1 ) allowing us to examine only red ( ade2-1 ) cyhr ( cyh2 ) spores to help eliminate any cells that may have mated after plating . The overall frequency of red ( Ade- ) colonies was no different than the frequency of white ( Ade+ ) colonies . The spores were then examined to determine location of crossovers as described in Materials and Methods . We first examined mutation rates in strains with the substrate inserted in the BUD5 gene 1 . 85 kb centromere proximal to the MATa locus on chromosome III ( Fig . 1C; a schematic of the approximate location on Chromosome III is shown in Fig . 1A ) . The MATα strain harbors a proximal natMX4 marker 7 . 8 kb proximal between the SMN1 and FEN1 genes . We found that the mutation rate to Canr was 2 . 8×10−8 in haploids ( GRY2691 Table 2 ) . Our mutation rate is somewhat ( 5 . 4-fold ) lower than that calculated by Lang and Murray ( 1 . 5×10−7 ) [1] . However , Lang and Murray have also shown that the mutation level of the URA3 gene can vary as much as six-fold dependent upon its location in the chromosome [36] . Thus , it is possible that the area where we are inserting the CAN1 HIS3 cassette shows lower mitotic mutation rates than CAN1 at its native locus . Also , it is possible that strain specific differences influence the mutation rates . Diploid cells had an ∼2 fold elevated mutation rate to 5 . 7×10−8 in an a/α diploid ( GRY3262 ) during mitotic growth . In the diploid the cassette is hemizygous . Since both the haploid and the diploid have a single reporter at the same location , the difference between the mutation rates must be related to cell type and/or ploidy . To determine whether the insertion affected meiotic recombination , we compared meiotic crossovers in strains without and with the HIS3 CAN1 cassette ( Fig . 1B and 1C respectively; Table 3 , GRY3269 and GRY3262 ) . In the strain lacking the reporter cassette ( GRY3269 ) , recombination in the natMX-MAT interval was determined to be 5 . 6% or 2 . 8 cM , resulting in 0 . 29 cM/Kb . The recombination rate is lower than average for chromosome III ( 0 . 48 cM/Kb , http://www . yeastgenome . org/pgMaps/pgMap . shtml ) confirming previous observations that this is a coldspot for recombination [37] , [38] . Insertion of the HIS3 CAN1 cassette further reduced recombination in the natMX-MAT interval to 3 . 2% or 1 . 6 cM , resulting in 0 . 16 cM/Kb ( Table3 , GRY3262 ) . Note that the size of the heterologous insertion is not included in the length calculations for Table 3 , since heterologous regions cannot participate in recombination . Any DNA break initiating within the heterology will either use a sister chromatid , or resect sufficiently to find homology thereby converting away the heterology [39] . The number of spores showing a crossover between MAT and natMX is not statistically significant between the two strains , ( p = 0 . 17 ) . When diploid cells were induced to undergo meiosis the mutation rate was 37×10−8 ( Table 2 , GRY3262 ) , a 6 . 5-fold increase from the mitotic diploid rate ( p = 2×10−8 ) . Our data are in agreement with the early observations from Magni and Von Borstel [28]–[30] , where they observed a 6–20 fold increase in mutation rates after the induction of meiosis . Therefore , we conclude that we see the meiotic effect in our system . To determine whether the increase in mutations that occurred during meiosis was a consequence of meiotic recombination , we constructed diploid strains that were homozygous for spo13 or both spo13 and spo11 . Cells mutated in spo11 are unable to sporulate , however , a concomitant mutation in spo13 , which allows bypass of meiosis I , overcomes the sporulation defect of spo11 mutants , resulting in two diploid spores [40] . Results of this analysis are shown in Table 2 ( Strains GY3273 , GRY3274 and GRY 3275 ) . The mutation rates during mitotic growth are similar for all three strains , although ∼2 fold lower in the spo11/spo11 diploid . Homozygous diploid spo13/spo13 strains show a 5 . 8 -fold increase in the mutation rate after meiosis , similar to the increase in the wildtype strain . In our strain background viability was significantly reduced upon induction of sporulation in the spo13Δ diploids ( to ∼10% ) , requiring an increase in the volume of our starting cultures . This inviability was rescued by a concomitant spo11 mutation as had been previously observed [41] . The spo11 spo13 diploids did not result in an increase in the mutation rate after induction of sporulation . These data provide strong support for the role of recombination and specifically , Spo11 meiotically induced DSBs , in the meiotic effect . Because Polζ is one of the primary polymerases responsible for the majority of both spontaneous and induced mutations in yeast [24] , and is up-regulated during meiosis [27] , [42] we analyzed the role of Rev3 on the meiotic effect . The results are shown in Table 2 ( Strain GRY3276 ) . As expected , Rev3 appears to be responsible for one-half to two-thirds of the spontaneous mitotic events: in haploids there were about twice as many His+ Canr events in the wild type strain ( 2 . 8×10−8 Table 2 , GRY2691 ) as in the rev3 strain ( 0 . 8×10−8 , GRY3265 ) . Similarly , there was a 3-fold difference in the mutation rates during mitotic growth in diploid strains from 5 . 7×10−8 in the wild type , versus 2 . 0×10−8 in the rev3 strain . Induction of meiosis still results in a large increase in the mutation rate . The increase in rev3 strains was 8-fold higher after meiosis as compared to the mitotic mutation rate ( Table 2 , GRY3276 , 2×10−8 versus 16×10−8 ) . Because we observe no differences in the frequency of recombination , sporulation or viability of spores in the rev3 mutant strains , we do not think that Rev3 influences the frequency of Spo11 induced breaks , although this has not been directly tested . Assuming that the efficiency of breaks is not affected between wildtype and rev3 mutants , we consider a better comparison is between the meiotic rates in the wild type strain ( Table 2 , 36×10−8 , GRY3262 ) and the rev3 strain ( Table 2 , 16×10−8 , GRY3276 ) . In this comparison the mutation rate in the rev3 diploid is only about half the expected rate if Rev3 had no role in the meiotic effect . Thus , as in spontaneous mitotic mutations , Polζ appears to be responsible for introducing about half of the meiotic mutations . We saw little difference in recombination between the markers tested in the wild type and rev3 strains by tetrad analysis ( Table 3 , Strain GRY3276 , p>0 . 4 ) . The natMX-MAT interval is 0 . 16 cM/Kb for the wild type versus 0 . 27 cM/kb for the rev3 strain . This observation further supports that Rev3 does not influence the formation of meiotic DSBs per se , but is an important player in introducing mutations during the repair of the breaks when necessary . We sequenced ∼80 independent can1 mutants to determine whether there are any obvious mechanistic differences between mutations generated during mitosis or meiosis . A summary of the sequence analysis is shown in Table 4 , and the data are shown in Supplementary S1 Table and S2 Table . There was very little noticeable difference between mutants generated during mitotic growth and those generated during meiosis . There were slightly more frame-shift mutations in meiosis as compared to mitosis ( p = 0 . 02 ) . There was no noticeable change in the distribution of mutations along the CAN1 gene ( p = 0 . 85 ) . One caveat to our experimental design is that the 3 . 8 kb cassette is hemizygous for CAN1 and could potentially influence both the frequency and types of meiotic events . To determine if the presence of increased homology might influence meiotic recombination rates and mutagenesis in our system , we designed a related cassette to provide homology to the CAN1 ORF on the homologous chromosome . This insertion includes the entire can1 gene with the exception of the promoter and the first 6 codons and increases homology by 2 . 6 kb . LEU2 is substituted for HIS3 ( Fig . 1D ) . Results from this construct are shown in Table 2 ( GRY3263 ) . Again , we found that haploids had a lower mitotic mutation rate than diploids ( 2 . 8×10−8 for GRY2691 versus 8 . 2×10−8 for GRY3263 , Table 2 ) . We analyzed 192 mitotic His+ Canr events for crossovers as described in Materials and Methods , and found no events with a crossover in the natMX-MAT interval . The increased homology resulted in more meiotic crossovers in the natMX-HIS3 interval , consistent with the 2 . 6 kb more homology where crossovers can occur . In unselected tetrads 7/252 ( 2 . 8% ) had crossovers in the hemizygous strain ( Table 3 , GRY3262 ) , and 12/245 ( 4 . 9% ) had crossovers in the strain with can1 homology ( Table 3 , GRY3263; p = 0 . 02 ) . However , the increase in length did not affect overall recombination in the natMX-HIS3 interval ( 0 . 18 cM/kb for the hemizygous strain GRY3262 vs 0 . 24 cM/kb for the strain with can1 homology GRY3263 , Table 3 ) . Likewise , there was no significant difference in crossover frequency between the two strains in the HIS3-MAT interval ( p = 0 . 3 ) . The presence of homology to CAN1 did not eliminate the meiotic effect . Induction of meiosis resulted in a 5 . 9 fold increase in the mutation rate ( Table 2 , GRY3263 , 49×10−8 ) compared with a 6 . 5 fold increase in can1 mutations after meiosis in the hemizygous strain ( Table 2 , GRY3262 , 36×10−8 ) . We conclude that the meiotic effect is independent of the presence of a homolog for the CAN1 ORF . It is true that there remains heterozygosity between HIS3 and LEU2 , and it is possible it influences the types of events seen . However , for the vast majority of cells that have undergone meiosis , the presence of heterozygocity at CAN1 seems to have no effect on the recombination frequency in the area near the insertion of the CAN1 HIS3 cassette ( see below ) . Meiotic recombination varies widely along the chromosome , resulting in coldspots and hotspots that correlate with the level of Spo11 induced breaks [10] . We predicted that since Spo11 is required for meiotic recombination , the rate of mutation induction during meiosis would also be influenced by the relative frequency of Spo11 DSBs . To test this prediction we inserted the CAN1 HIS3 reporter close to a known meiotic hotspot between the BUD23 and ARE1 genes [43] 11 kb distal to MATa ( Fig . 1E , GRY3625 ) . The LEU2 can1 cassette described in the last section , was inserted at the same location on the MATα chromosome to provide homology to the CAN1 ORF ( Fig . 1E-GRY3626 ) . We used a kanMX knockout of YIH1 from the knockout collection [44] as an 11 . 6 kb distal marker for monitoring crossovers . The mutation rates for the strain with the HIS3 CAN1 cassette located between BUD23 and ARE1 are shown in Table 2 ( GRY3630 ) . There was a 3-fold increase in the mutation rate in diploids ( GRY3630 , 8 . 4×10−8 ) as compared to haploids ( GRY3625 , 2 . 5×10−8 ) . Only 1/183 His+ Canr mitotic events from GRY3630 showed evidence of a crossover between HIS3 and yih1::kanMX . To ensure that the insertion of the reporter did not affect the levels of meiotic recombination at the hotspot , we dissected tetrads from the resulting diploid strain , and compared the frequency of crossovers in the MAT-yih1::kanMX interval to that of a strain lacking the insertion . In the absence of the reporter construct , the interval between MAT and yih1::kanMX was 26 . 3 cM ( 1 . 2 cM/Kb GRY3629 , Table 3 ) . Previous meiotic data indicated that the interval between THR4 and MAT was 1 . 2 cM/kb ( http://www . yeastgenome . org/cgi-bin/geneticData/displayTwoPoint ? locus=S000029699 ) . Since THR4 is 7 . 5 kb closer to MAT than YIH1 , it is likely that most of the recombination occurs in the vicinity of the BUD23-ARE1 hotspot . When we inserted the HIS3 CAN1 cassette near the hotspot , recombination between MAT and yih1::kanMX was 24 . 5 cM ( Table 3 , GRY3630 ) resulting in 0 . 98 cM/Kb . Thus , although the insertion did cause a reduction of recombination at the hotspot , recombination was still about twice as frequent as the average for chromosome III ( 0 . 48 cM/Kb ) , and three-fold more frequent than the coldspot insertion between natMX and MAT ( 0 . 33 cM/kb , GRY3263 ) . Meiotic DSBs can be monitored and quantified in strains deficient for sae2 , as these strains are unable to remove the bound Spo11 and initiate resection allowing the DSB to accumulate as unique bands [37] , [45] . In strains lacking the reporter construct ( Fig . 2A . GRY3635 ) 24 . 7% of the DNA accumulated a DSB in the BUD23-ARE1 interval . When the reporter cassette is inserted nearby ( Strain GRY3636 ) , the level of DSBs was is 23 . 6% , consistent with similar levels of meiotic DSBs in the two strains ( Fig . 2B ) . No breaks were detectable near the HIS3 CAN1 cassette located in the coldspot ( Fig . 2C ) . This is consistent with the observations of Pan et al [38] that they observed almost 10 , 000 Spo11 associated oligomers in the 6 kb surrounding the insertion site when it was in the hotspot , but only 384 in the 6 kb surrounding the insertion site at the coldspot . The induction of sporulation resulted in a 3 . 6 fold increase in the mutation rate when the substrate was inserted in the hotspot ( 177×10−8 , GRY3630 , Table 2 ) versus when it was inserted in the coldspot ( 49×10−8 , GRY3263 , Table 2 ) . This correlates well with the differences in meiotic recombination at the two loci ( Table 3 ) either with or without the substrate . When the substrate was inserted near the coldspot ( GRY3262 ) recombination was 0 . 33 kb/cM , versus 0 . 90 cM/kb when the substrate was inserted near the hotspot ( GRY3630 ) , a three-fold difference . Therefore , there is a positive correlation between meiotically induced DSBs and meiotically induced mutations . Although elevated , the frequency of meiotic mutation was too low to determine the crossover ( CO ) association by tetrad analysis . Therefore we examined red ( ade2-1 ) cyhr random spore colonies as described in Materials and Methods . A direct comparison of crossovers between unselected tetrads ( none of which had a mutation in CAN1 ) and the selected His+ Canr random spores was complicated by the fact that from the random spores we cannot distinguish between a gene conversion ( GC ) event versus a double CO of a central marker ( s ) , or a CO versus a GC of an outside marker . Therefore , we also examined random spores from canavanine sensitive ( Cans ) His+ spores . A comparison of the data between tetrads and random spores for the strain with the substrate in the coldspot ( GRY3263 ) is shown in Fig . 3A . The difference between recombination events in tetrads and His+ Cans random spores was not significantly different ( His+ Cans/Tetrads; p = 0 . 5 ) . In contrast the spores that have had a mutation in can1 ( His+ Canr ) were two-to three fold more likely to have had a crossover than either the tetrads , ( Fig . 3A , p = 6 . 7×10−5 ) or the His+ Cans random spores ( Canr/Cans; p = 8 . 3×10−7 ) . This difference was primarily due to an increase in events in the natMX–HIS3 interval ( A ) and apparent double crossovers ( A+B ) . The expected percent of double crossovers is calculated based on total recombinants with a crossover in an interval ( numbers in parentheses ) . There are insufficient tetrads to determine whether there is any interference among the tetrads . However , there appears to be a loss of interference among the His+ Canr random spores , although this could also be due to gene conversions that are counted as crossovers . A similar analysis for the strain with the reporter cassette at the hotspot ( GRY3630 ) is shown in Fig . 3B . The pattern of recombinants seen in His+ Cans total random spores was not significantly different from the pattern of recombinants in tetrads ( His+ Cans/tetrads; p = 0 . 25 ) . However , there was a significant increase ( 10–40 fold ) in crossovers among the His+ Canr spores as compared to either tetrads or the His+ Cans spores ( Canr/Cans; p<1×10−8 ) . Interestingly , the interval showing the greatest increase in crossovers is interval I . which is the furthest from the site of the break ( >14 kb ) . The presence of interference is evident among the tetrads and the random spores , where the observed versus expected ( obs/exp ) ratio is <1 . However there does not appear to be any interference among the His+ Canr random spores where the obs/exp ratio is close to 1 . We conclude that events that have acquired a mutation are ∼3 times more likely to be associated with a crossover than events that did not result in a mutation . Also , these events appear to be associated with crossovers that are quite distant from the initiating DSB . Evolution is driven by the accumulation of mutations that are passed on in the germ line . Survival in evolutionary time scales involves providing sufficient variability so that adaptation can occur with a changing world . Most organisms ensure variability by maintaining a sexual lifestyle despite the cost . Here we explore the concept that the process of meiosis itself may be mutagenic and may also contribute to variability . Support for this hypothesis was first documented in the early 1960′s by Magni and von Borstel [28]–[30] see Introduction ) . Here , we have revisited the meiotic effect with new tools and knowledge in hand , and have attempted to address the caveats present in previous work , as well as provide new data about the mechanism by which meiotic mutations arise . In agreement with the observations of Magni et al [28]–[30] , we find a 4 to 8 -fold increase in the CAN1 mutation rate after the induction of meiosis . Several other groups have attempted to repeat the Magni observerations with little success . Whelan et al [46] also recognized the difficulty of measuring the appropriate diploid rate for can1 mutations , and assumed that the diploid was either equal to the haploid , or twice the haploid rate . They did not subtract the frequency of the mutations generated during mitotic growth from each culture prior to determining the frequency/rate of mutations generated in meiosis . Whereas the events accumulate over several generations during mitosis , all of the meiotic events must accumulate in a single cell division , and therefore will be masked without this adjustment . In a high throughput sequencing approach , Nishant et al [47] found that the mutation rate during mitotic growth was similar to that previously determined . Because of the rarity of mutational events that occur during meiosis , they were only able to estimate that the global meiotic mutation rate in yeast was somewhere between zero to 55-fold higher than the mutation rate during mitotic growth ( and thus well within any observed meiotic effect ) . Finally Qi et al [48] did deep sequencing of a cross between S288C and RMI11-1 ( that diverge by 0 . 5–1% ) and fully sequenced the products of one tetrad . They determined that the limit of their detection was ∼8×10−8/per base per cell division , and thus a 6–20-fold increase of the estimated global rate of mutations is still about 10-fold lower than they could detect . The majority of the events that we sequenced were point mutations in the CAN1 ORF and there appeared to be little difference between mutations made during mitosis or meiosis ( see Supplementary Tables S1 and S2 ) . The presence or absence of homology at the CAN1 ORF did not seem to affect the overall frequency of meiotic recombination , albeit it is impossible to determine whether the presence of any heterozygosity can influence the meiotic mutation rate . We see no significant change in sporulation and/or viability , nor in recombination due to the presence of our heterologies . Without an insert , we cannot measure the meiotic effect in our system . However , the fact that our observations are similar to those of Magni and von Borstel when they measure reversion of a recessive allele [28] suggests that the heterology itself is not inducing the meiotic effect . By comparing the same substrates present in either the hotspot or coldspot , we find that when the substrate is near a coldspot , recombination is 0 . 24 cM/kb and the mutation rate induced by meiosis is increased ∼6 fold . When the substrate is located near a meiotic hotspot recombination is 0 . 9 cM/kb , and the mutation rate is increased 21 fold . Thus , there is a three- to four -fold increase in the level of recombination between the coldspot and the hotspot , and a three- to four-fold increase in the mutation rate . Associated crossovers are 2–3 times more likely to be found among the selected mutant spores than among non-mutants ( Fig . 3A Canr/Cans ) when the substrate is near the coldspot . When the substrate is in the hotspot crossovers are >10 fold more frequent among the mutant spores than non-mutant spores ( Fig . 3B Canr/Cans ) . The largest increase in recombinants is in the interval furthest from the DSB ( interval I ) , where 6 . 8% of the Cans spores have a crossover and 21 . 5% of the Canr spores have a crossover . Meiotic crossovers show interference when the number of double crossovers is less than expected for an interval [12] . It is not clear how interference operates , but the evidence points to very early stages of recombination , possibly at the strand invasion step [49] . The influence of interference on crossovers is quite evident when looking at the strain with the substrate in the hotspot ( GRY3630 , Fig . 3B ) , where the observed/expected ratio for both tetrads and His+ Cans random spores is below one . In contrast , there appears to be a complete loss of interference among the His+ Canr spores , where the observed/expected ratio is actually slightly >1 . The dramatic increase ( 10–40 fold ) in double and triple crossovers is particularly evident when one compares the % observed crossovers between the Cans and Canr mutant spores when the substrate is in the hotspot ( Fig . 3B , Canr/Cans ) . In agreement with these observations is our key finding that the meiotic effect is dependent upon the presence of Spo11 , the protein that introduces meiotic DSBs . Because we cannot distinguish between gene conversion events or double crossovers among the random spores , one possibility is that the events that result in a mutation at CAN1 are unusual in that they are associated with long resection . This is suggested by the fact that at the hotspot , interval I , >14 kb from the site of the DSB ( Fig . 3B ) , has the highest increase in crossovers . Increased resection could have several consequences: a ) increased ssDNA that is more susceptible to DNA damage , b ) increased gene conversion tract lengths that might be confused with crossovers in our random spore analysis , c ) template switching as has been seen in BIR or d ) a loss of crossover interference . The average gene conversion tract in meiosis is ∼1 . 8–2 kb [50] , well below the 14 kb distance of interval I from the break site . We find that about one half of the mutations produced in meiosis are dependent upon Rev3 , a component of the Polζ translesion DNA polymerase . This effect is not very different than that seen for spontaneous mitotic mutations . In contrast , DSB induced mitotic mutations vary significantly in their dependence upon Rev3 , suggesting that context is of key importance for its activity ( see Introduction ) . We saw no evidence for any effect on the frequency , viability , or meiotic recombination in a rev3 mutant as compared with the wildtype , leading us to assume that Rev3 is unlikely to be affecting the rate of Spo11 breakage . If the mutational events do result from longer regions of ssDNA , it is possible that this leads to increased damage , and therefore a potential direct role for Polζ in introducing some of the mutations during lesion bypass synthesis . Mutations occurring long distances ( >8 kb ) from the initiating lesion have been observed in other DSB associated assays [17] , [18] , [20] , suggesting that rare mutational events are associated with exceptional events . The occurrence of multiple template switch events has also been documented during BIR , or when homology is limiting [51] , [52] . These types of events would appear as double or triple crossovers in our random spore analysis . Clearly there is at minimum a loss of crossover interference . Longer resection tracts are most likely associated with delayed repair thereby potentially leading to an uncoupling from the mechanism of crossover interference . If longer resection tracts are associated with mutation , it is possible that a role for Polζ is in copying over DNA damage that might arise during the single stranded phase of repair . However , we cannot distinguish whether the drop in mutations after meiosis seen in the absence of Rev3 is due to repair by an error free mechanism , or whether the cells cannot traverse the lesion and die . Most mutations are thought to be detrimental , and cells have gone to great lengths to keep mutations at a minimum by having multiple repair pathways to deal with the plethora of different lesions that they encounter . So what then is the point of allowing the mutational load to increase , albeit still at a very low level , during meiosis ? We entertain three models . First , perhaps this is part of the compromise organisms make to help maintain variability in the population . The increased error rate may be an unavoidable consequence of the DSB pathway used to initiate meiotic exchange and the advantages of meiosis outweigh the added mutation load . Second , the option to increase mutagenesis during meiosis may have advantages in the sense of increasing the diversity of the germ cell pool . The increased mutational load allows for novel alleles to appear that might have selective advantages . A more provocative third model is that the meiotic effect allows organisms to direct the location of the genes subject to elevated mutagenesis . One of the oddities of meiotic recombination is that there are chromosomal hotspots and coldspots , and the position of these may be highly conserved [50] , [53] . Thus organisms could increase the evolutionary rates of genes by controlling whether they were situated near hot spots of meiotic recombination or protect them from this process by preserving them in cold spots . Indeed , in a survey of yeast genes it was found that essential genes tended to be clustered with each other and are generally cold for meiotic recombination [54] . Since recombination is initiated by DNA double strand breaks , and breaks are usually the recipients of genetic information , it is a conundrum as to how hotspots are maintained . For example , a recent study of several isolated wild strains of Saccharomyces paradoxus indicates that the recombination hotspots are found at similar locations between the evolutionarily separated species S . cerevisiae [53] . It is worth noting that the generated data were exclusive to chromosome III and was obtained by PFGE , therefore at low resolution . A more recent global analysis of DSB sites in yeast suggests that hotspots are located in chromatin-depleted regions that are usually associated with some promoters and active genes [38] . Thus , it is probable that hotspots are conserved because of the underlying structural organization of the genome . Since it appears that meiotic hotspots may be maintained on a global scale , this would allow cells to regulate the position of genes , or more likely the region of meiotic DSB sites so that essential genes are near coldspots , whereas genes where increased variability is desirable are near hot spots . In a high-resolution meiotic mapping experiment in a diploid of two S . cerevisiae haploid strains with 0 . 5% heterology , crossover sites were found to coincide with previously mapped DSB hotspots and with sites of increased variability among yeast strains [50] . On the other hand , Noor [55] found no evidence for increased genomic variability near hotspots between S . cerevisiae and S . paradoxus strains . It is important to note that his conclusions are based on assuming identical hot and cold spots between the two strains , despite the only data suggesting this comes from the low-resolution map of chromosome III by Tsai et al [53] . In contrast , there is an excellent correlation between recombination and sequence divergence in Drosophila [56] . In mammals recombination hotspots are associated with increased SNPs [57] . One major caveat is that recombination is easier to recognize when more SNPs are present . As the positions of meiotic hot spots are determined in more organisms it will be of interest to see whether the proposed correlation of hot spots with gene evolution rates is validated . In summary , we have shown that mutations are increased during meiosis and that these results correlate with increased recombination events and are dependent upon the protein responsible for initiating meiotic recombination . We suggest that these are meiotic events that have gone awry , leading to increased resection , increased DNA damage , and loss of crossover interference . S . cerevisiae cells were grown in YEPD ( Sherman et al . 1986 ) or the appropriate AA-synthetic drop-out media . AA drop-out media is similar to SD media described by Sherman et al . ( 1986 ) except that all amino acids , uracil , adenine , Myo-inisitol are 85 µg/mL , except for leucine , which is at 170 µg/mL , and para-aminobenzoic acid and which is at 17 µg/mL . Drop out plates were only missing the noted amino acid . Canavanine was added at 100 µg/ml and cyclohexamide at 5 µg/ml . To identify red colonies on minimal media the adenine was reduced to 20 µg/ml . Amar spore medium is 2% potassium acetate supplemented with 100 µg/ml adenine and uracil , 50 µg/ml histidine , leucine , lysine , tryptophan , methionine and arginine , 35 µg/ml phenylalanine and 10 µg/ml proline , and is designed to support sporulation without growth of the culture . All strains used in this study are listed in Table 1 . Strain GRY2691 , a MAT a parent ( Table 1 ) was constructed by transformation of GRY1600 with a PvuII fragment from plasmid pMush22 [14] containing CAN1 and HIS3 genes transcribing away from one another ( Fig . 1A ) resulting in an insertion of the 3 . 8 kb cassette 1 . 85 kb proximal to the MAT locus . The sequences of CAN1 present in the cassette are from −148 to +1973 relative to the CAN1 start codon . The strain also harbors a deletion of CAN1 that excludes sequences from −151 to +1995 relative to the ATG , thus there is no homology present between the two loci . The HIS3 insertion includes sequences from −191 to +857 of the HIS3 ORF . The his3-Δ200 mutation extends from −205 to +835 , thus there are only 22 bp of homology on the 3′ end of HIS3 , and no homology on the 5′ end . The MATα parent ( GRY2690 , Table 1 , Fig . 1B ) was constructed by insertion of a natMX cassette from pAG25 [58] between FEN1 and SNM1 on chromosome III and selection on nourseothricin ( clonNAT , Werner BioAgents ) essentially as described for creating the yeast knockout libraries by providing 45 bp of homology on either side of the cassette to the target locus [59] . The kanMX cassette was inserted into GRY2690 by PCR of yih::kanMX from the yeast MATa knockout collection strain ( Open Biosystems ) with an additional ∼250 bp flanking homology and selection on G418 ( Genticin , US Biologicals ) . To construct the promotorless can1 gene ( Fig . 1 C ) we replaced HIS3 , the CAN1 promoter , and the first 6 amino acids of the CAN1 ORF of pMush22 with a LEU2 marker by recombineering [60] with a PCR fragment of LEU2 containing 35 bp of flanking homology to either side of the HIS3 gene . Once the construct was verified and sequenced , it was inserted into strain GRY2690 by one step transplacement [61] . Hotspot constructs were made by PCR of the HIS3 CAN1 and the LEU2 can1 cassettes with 5′ end- tailed primers containing 50 bp homology on either side to a site between ARE1 and BUD23 . The ARE1 and BUD23 genes are transcribed away from one another . The insertion did not delete any base pairs , and was positioned so that it was between −333 of BUD23 and −147 of ARE1 . Strains deleted for SPO11 and SPO13 were constructed by PCR of the appropriate gene disruption from the yeast knockout collection [44] and subsequent transformation into strains GRY2690 and GRY2691 followed by selection on G418 as described [59] . Disruptions were confirmed by PCR , Southern blot and phenotypic analysis where possible . Strains deleted for rev3 were obtained by one-step transplacement of an Xba1 fragment containing a rev3::LEU2 disruption from plasmid pAM56 ( kindly provided by Alan Morrison ) . Transformants were selected on media lacking leucine , and further confirmed by PCR and a reduced level of UV induced papillation to canavanine resistance for strains carrying the HIS3-CAN1 cassette . Strains deleted for sae2 were PCR amplified from a sae2::HygMX mutant strain , which was constructed by marker replacement of the knockout collection [44] , [58] . Tetrads were isolated by patching the various diploid strains on YEPD , then growing up a 5 ml culture in Amar spore media for a minimum of 5 days . Tetrads were treated with zymolyase and dissected onto YEPD . After growth on YEPD , the tetrads were examined for each of the relevant markers by replica plating . Genetic distance was calculated using the Perkins equation ( cM = 100X = ( 100 ( 6N+T ) ) / ( 2 ( P+N+T ) ) [62] . To analyze the rate of the mutations occurring during mitotic cell divisions we used a Luria-Delbruck fluctuation test [63] performed as follows: The relevant haploid strains were mated , and zygotes were isolated by micromanipulation on YEPD . Individual zygote colonies were struck for single cells on YEPD . 9 colonies from each independent zygote colony were inoculated into 5 ml of YEPD . Cells were shaken at 30° overnight and then diluted 1∶50 into 10 ml fresh YEPD and incubated for ∼4–6 hours at 30° to mid-log phase ( 2–4×107 cells/ml ) with shaking ( for the spo13 diploids , we grew 100 ml of culture ) . After washing cultures in sterile water , half of the cells were removed and used to determine the number of His+ cells and the number of His+ Canr mutants from each culture . The remaining diploid cells were then resuspended into 5 ml sporulation media and incubated at 30° with shaking until >90% of the cells had sporulated by microscopic examination ( ∼5 days ) . Mutation rates during mitosis were estimated by the Ma-Sandri-Sarkar Maximum Likelihood Method [64] . This is a recursive algorithm that is the product of the probabilities pr for the experimental results r ( the number of mutants per culture ) . m is the number of mutations per culture , and c is the number of cultures . The proportion of cultures with no mutations is , and cultures with 1 , 2…i mutations are calculated by . The recursive function works as follows: to find the best estimate of m from the fluctuation analysis . The mutation frequencies ( His+ Canr/His+ ) ranged from ∼1×10−5 to 1×10−8 with majority of cultures falling in the 2–5×10−7 range . To determine the 95% confidence intervals we used the following equations and from Foster [65] . For determination of mutation rates during meiosis , we used a random spore analysis as described [66] . Briefly , The ascospores were washed twice in 5 ml water and then resuspended in 5 ml water with 0 . 25 ml 1 mg/ml Zymolyase-100T and 10 µl 2-ME . Cells were incubated overnight at 30°C with gentle shaking . 5 ml 1 . 5% NP40 ( Roche # 11754599001 ) was added along with 2 ml acid washed glass beads and incubated on a roller drum at room temperature for 2 hours with occasional vigorous vortexing . Disruption of spores was monitored microscopically . This procedure lyses all unsporulated cells . Appropriate dilutions were plated onto media to determine the total number of His+ and His+ Canr cells . Determination of the meiotic frequency/rate was calculated by subtracting the mitotic frequency from the meiotic frequency for each culture ( Meiotic His+Canr/His+ per ml minus Mitotic His+Canr/His+ per ml ) . The resulting frequencies were then used to determine a median value reflecting the rate of mutations arising during meiosis , since meiosis involves a single division . For linkage analysis we selected doubly recessive red ( ade2-1 ) Cyhr His+ Canr random spores to maximize analysis of haploid cells versus cells that mated after plating . To identify crossovers among the mitotic mutants we patched ∼200 His+ Canr events onto YPD , phenotypically scored the markers , and replica plated them to sporulation media . After 6–7 days at 30° , spore patches were replica plated to YPD , grown overnight and mated to freshly grown strains GRY633 and GRY634 . Diploids were selected on media lacking both histidine and uracil . Because the HIS3 CAN1 cassette is located near the MATa locus , non-crossovers mate with strain GRY634 ( MATα ) , and give rise to Nats colonies . Crossovers between natMX and HIS3 can1 also only mate with strain GRY634 ( MATα ) but form Natr progeny . Finally , crossovers between HIS3 can1 and MAT mate with strain GRY633 ( MATa ) resulting in Nats colonies . Double crossovers between the same chromatids ( of which none were detected ) would mate with GRY633 and become Natr . No crossovers between HIS3 and MAT were observed among the mitotic diploids analyzed . A similar scheme was used to analyze for mitotic crossover events for the strains with the construct in the hotspot except that the kanMX marker was also scored . After plating for random spores we colony purified 300–600 red His+ Canr Cyhr spore clones from each strain onto media lacking Histidine . These were then retested for Canr , patched and replica plated to test the relevant markers and determine mating type ( by crossing with strains DC14 and DC17 ) . Approximately 50 of the spore colonies from each strain that presented evidence of a crossover were mated to GRY1600 or GRY1601 ( depending on mating type ) and 5–10 tetrads were dissected from each to further confirm linkage . The entire CAN1 gene was PCR amplified from mutant candidates and sequenced on both strands . Three polymorphisms were noted in our CAN1 gene as compared to the published S288c CAN1 sequence . The sequence of the mutations identified in mitosis or meiosis are listed in Supplemental Tables S1 and S2 , respectively . Sequencing was done by the Laboratory of Molecular Technologies Sequencing Facility- SAIC-Frederick , FNLCR . DNA isolation and Southern blots were carried out as described by Sun et al . [67] . For the hotspot analysis DNA was digested with AatII and run on a 0 . 7% agarose gel . For the coldspot analysis DNA was digested with PvuII and run on a 1% agarose gel . 32P-dCTP labeled probes were made with the Agilent Prime it II kit according to manufacturers instructions . Hotspot probe used a PCR fragment spanning coordinates 213283–213848 of Chromosme III . Coldspot probes used a PCR fragment spanning 195735–196774 of Chromosome III . Statistical analyses were calculated by the chi square test for tetrads and random spores: in tetrads we compared PD , NPD and TT between the various strains , for random spore analysis we compared the number of crossovers in each interval for each strain . For distribution analysis along the length of CAN1 the gene was divided into 200 bp windows as described previously [25] . For comparison between mutation rates from mitotic growth and from meiosis we used a student's t test ( http://en . wikipedia . org/wiki/Student's_t-test ) .	Meiosis , the cellular division that gives rise to germ cells , contributes to evolution by reassortment of parental alleles . This process involves recombination initiated by Spo11-induced double-strand breaks early in meiosis . The result is that germ cells from a single meiosis are different from either parent . Here we show that the DNA repair associated with meiotic recombination is inherently mutagenic , providing an additional source of variation that can contribute to evolution . This elevated mutagenesis requires the Spo11 protein , and the rate of mutagenesis correlates positively with the frequency of meiotic double-strand breaks . Furthermore , the mutations that arise show an increased level of associated crossovers , consistent with having been introduced during recombination . We speculate that there is an evolutionary drive to position essential genes in meiotic recombination coldspots for slow evolution , and genes that can afford to evolve more rapidly are placed near meiotic recombination hotspots .	[ "Abstract", "Introduction", "Results", "Discussion", "Materials", "and", "Methods" ]	[ "biology", "and", "life", "sciences" ]	2015	Elevated Mutation Rate during Meiosis in Saccharomyces cerevisiae
A major goal in translational cancer research is to identify biological signatures driving cancer progression and metastasis . A common technique applied in genomics research is to cluster patients using gene expression data from a candidate prognostic gene set , and if the resulting clusters show statistically significant outcome stratification , to associate the gene set with prognosis , suggesting its biological and clinical importance . Recent work has questioned the validity of this approach by showing in several breast cancer data sets that “random” gene sets tend to cluster patients into prognostically variable subgroups . This work suggests that new rigorous statistical methods are needed to identify biologically informative prognostic gene sets . To address this problem , we developed Significance Analysis of Prognostic Signatures ( SAPS ) which integrates standard prognostic tests with a new prognostic significance test based on stratifying patients into prognostic subtypes with random gene sets . SAPS ensures that a significant gene set is not only able to stratify patients into prognostically variable groups , but is also enriched for genes showing strong univariate associations with patient prognosis , and performs significantly better than random gene sets . We use SAPS to perform a large meta-analysis ( the largest completed to date ) of prognostic pathways in breast and ovarian cancer and their molecular subtypes . Our analyses show that only a small subset of the gene sets found statistically significant using standard measures achieve significance by SAPS . We identify new prognostic signatures in breast and ovarian cancer and their corresponding molecular subtypes , and we show that prognostic signatures in ER negative breast cancer are more similar to prognostic signatures in ovarian cancer than to prognostic signatures in ER positive breast cancer . SAPS is a powerful new method for deriving robust prognostic biological signatures from clinically annotated genomic datasets . The identification of pathways that predict prognosis in cancer is important for enhancing our understanding of the biology of cancer progression and for identifying new therapeutic targets . There are three widely-recognized breast cancer molecular subtypes , “luminal” ( ER+/HER2− ) [1] , [2] , [3] , [4] , “HER2-enriched” ( HER2+ ) [5] , [6] and “basal-like” ( ER−/HER2− ) [6] , [7] , [8] , [9] and a considerable body of work has focused on defining prognostic signatures in these [10] , [11] . Several groups have analyzed prognostic biological pathways across breast cancer molecular subtypes [12] , [13] , [14]; a tacit assumption is that if a gene signature is associated with prognosis , it is likely to encode a biological signature driving carcinogenesis . Recent work by Venet et al . has questioned the validity of this assumption by showing that most random gene sets are able to separate breast cancer cases into groups exhibiting significant survival differences [15] . This suggests that it is not valid to infer the biologic significance of a gene set in breast cancer based on its association with breast cancer prognosis and further , that new rigorous statistical methods are needed to identify biologically informative prognostic pathways . To this end , we developed Significance Analysis of Prognostic Signatures ( SAPS ) . The score derived from SAPS summarizes three distinct significance tests related to a candidate gene set's association with patient prognosis . The statistical significance of the SAPSscore is estimated using an empirical permutation-based procedure to estimate the proportion of random gene sets achieving at least as significant a SAPS score as the candidate prognostic gene set . We apply SAPS to a large breast cancer meta-dataset and identify prognostic genes sets in breast cancer overall , as well as within breast cancer molecular subtypes . Only a small subset of gene sets that achieve statistical significance using standard statistical measures achieves significance using SAPS . Further , the gene sets identified by SAPS provide new insight into the mechanisms driving breast cancer development and progression . To assess the generalizability of SAPS , we apply it to a large ovarian cancer meta-dataset and identify significant prognostic gene sets . Lastly , we compare prognostic gene sets in breast and ovarian cancer molecular subtypes , identifying a core set of shared biological signatures driving prognosis in ER+ breast cancer molecular subtypes , a distinct core set of signatures associated with prognosis in ER− breast cancer and ovarian cancer molecular subtypes , and a set of signatures associated with improved prognosis across breast and ovarian cancer . The assumption behind SAPS is that to use a prognostic association to indicate the biological significance of a gene set , a gene set should achieve three distinct and complimentary objectives . First , the gene set should cluster patients into groups that show survival differences . Second , the gene set should perform significantly better than random gene sets at this task , and third , the gene set should be enriched for genes that show strong univariate associations with prognosis . To achieve this end , SAPS computes three p-values ( Ppure , Prandom , and Penrichment ) for a candidate prognostic gene set . These individual P-Values are summarized in the SAPSscore . The statistical significance of the SAPSscore is estimated by permutation testing involving permuting the gene labels ( Figure 1 ) . To compute the Ppure , we stratify patients into two groups by performing k-means clustering ( k = 2 ) of an n×p data matrix , consisting of the n patients in the dataset and the p genes in the candidate prognostic gene set . We then compute a log-rank P-Value to indicate the probability that the two groups of patients show no survival difference ( Figure 1A ) . Next , we assess the probability that a random gene set would perform as well as the candidate gene set in clustering cases into prognostically variable groups . This P-Value is the Prandom . To compute the Prandom , we randomly sample genes to create random gene sets of similar size to the candidate gene set . We randomly sample r gene sets , and for each random gene set we determine a using the procedure described above . The Prandom is the proportion of at least as significant as the true observed Ppure for the candidate gene set ( Figure 1B ) . Third , we compute the Penrichment to indicate if a candidate gene set is enriched for prognostic genes . While the procedure to compute the Ppure uses the label determined by k-means clustering with a candidate gene set as a binary feature to correlate with survival , the procedure to compute the Penrichment uses the univariate prognostic association of genes within a candidate gene to produce a gene set enrichment score to indicate the degree to which a gene set is enriched for genes that show strong univariate associations with survival ( Figure 1C ) . To compute the Penrichment , we first rank all the genes in our meta-dataset according to their concordance index by using the function concordance . index in the survcomp package in R [16] . The concordance index of a gene represents the probability that , for a pair of patients randomly selected in our dataset , the patient whose tumor expresses that gene at a higher level will experience the appearance of distant metastasis or death before the other patient . Based on this genome-wide ranking we perform a pre-ranked GSEA [17] , [18] to identify the candidate gene sets that are significantly enriched in genes with either significantly low or high concordance indices . The GSEA procedure for SAPS has two basic steps . First , an enrichment score is computed to indicate the overrepresentation of a candidate gene set at the top or bottom extremes of the ranked list of concordance indices . This enrichment score is normalized to account for a candidate gene set's size . Second , the statistical significance of the normalized enrichment score is estimated by permuting the genes to generate the Penrichment ( see Refs . [17] , [18] for further description of pre-ranked GSEA procedure ) , which indicates the probability that a similarly sized random gene set would achieve at least as extreme a normalized enrichment score as the candidate gene set ( Figure 1C ) . The SAPSscore for each candidate gene set is then computed as the negative log10 of the maximum of the ( Ppure , Prandom , and Penrichment ) times the direction of the association ( positive or negative ) ( Figure 1D ) . For a given candidate gene set , the SAPSscore specifies the direction of the prognostic association as well as indicates the raw P-Value achieved on all 3 of the ( Ppure prognosis , Prandom prognosis , and Penrichment ) . Since we take the negative log10 of the maximum of the ( Ppure prognosis , Prandom prognosis , and Penrichment ) , the larger the absolute value of the SAPSscore the more significant the prognostic association of all 3 P-Values . The statistical significance of the SAPSscore is determined by permuting genes , generating a null distribution for the SAPSscore and computing the proportion of similarly sized gene sets from the null distribution achieving at least as large an absolute value of the SAPSscore as that observed with the candidate gene set . When multiple candidate gene sets are evaluated , after generating each gene set's raw SAPSP-Value by permutation testing , we account for multiple hypotheses and control the false discovery rate using the method of Benjamini and Hochberg [19] to generate the SAPSq-value ( Figure 1E ) . In our experiments , we have required a minimum absolute value ( SAPSscore ) of greater than 1 . 3 and a maximum SAPSq-value of less than 0 . 05 to consider a gene set prognostically significant . These thresholds ensure that a significant prognostic gene set will have achieved a raw P-Value of less than or equal to 0 . 05 for each of Ppure , Prandom , and Penrichment , and will have achieved an overall SAPSq-Value of less than or equal to 0 . 05 . We chose two model systems to investigate the performance of SAPS . The first is a curated sample of breast cancer datasets previously described in Haibe-Kains et al . [20] . Our analysis focused on nineteen datasets with patient survival information ( total n = 3832 ) ( Table S1 ) . The second dataset was a compendium of twelve ovarian cancer datasets with survival data , as described in Bentink et al . [21] , which includes data from 1735 ovarian cancer patients for whom overall survival data were available ( Table S2 ) . In breast cancer , we used SCMGENE [20] as implemented in the R/Bioconductor genefu package [22] to assign patients to one of four molecular subtypes: ER+/HER2− low proliferation , ER+/HER2− high proliferation , ER−/HER2− and HER2+ . In ovarian cancer , we used the ovcAngiogenic model [21] as implemented in genefu to classify patients as having disease of either angiogenic or non-angiogenic subtype . One challenge in the analysis of large published datasets is the heterogeneity of the platforms used to collect data ( see Table S1 and Table S2 ) . To standardize the data , we used normalized log2 ( intensity ) for single-channel platforms and log2 ( ratio ) in dual-channel platforms . Hybridization probes were mapped to Entrez GeneID as described in Shi et al . [23] using RefSeq and Entrez whenever possible; otherwise mapping was performed using IDconverter ( http://idconverter . bioinfo . cnio . es ) [24] . When multiple probes mapped to the same Entrez GeneID , we used the one with the highest variance in the dataset under study . To allow for simultaneous analysis of datasets from multiple institutions , we tested two data merging protocols . First , we scaled and centered each expression feature across all patients in each dataset ( standard Z scores ) , and we merged the scaled data from the different datasets ( “traditional scaling” ) . In a second scaling procedure , we first assigned each patient in each data set to a breast or ovarian cancer molecular subtype , using the SCMGENE [20] and ovcAngiogenic [21] models , respectively . We then scaled and centered each expression feature separately within a specific molecular subtype within each dataset , so that each expression value was transformed into a Z score indicating the level of expression within patients of a specific molecular subtype within a dataset ( “subtype-specific scaling” ) . After merging datasets , we removed genes with missing data in more than half of the samples and we removed samples that were missing data on more than half of the genes or for which there was no information on distant metastasis free survival ( for breast ) or overall survival ( for ovarian ) . The resulting breast cancer dataset contained 2731 cases with 13091 unique Entrez gene IDs and the ovarian cancer dataset had 1670 cases and 11247 unique Entrez gene IDs for . For each of these reduced data matrices , we estimated missing values using the function knn . impute in the impute package in R [25] . Given that breast cancer is an extremely heterogeneous disease with well-defined disease subtypes , and a primary objective of our work is to identify subtype-specific prognostic pathways in breast cancer , we focus our subsequent analyses on the subtype-specific scaled data . Given that ovarian cancer subtypes are more subtle and less well defined than breast cancer molecular subtypes , we focus our subsequent analyses in ovarian cancer on the traditional scaled data . SAPS scores in breast and ovarian cancer generated from the two different scaling procedures showed moderate to strong correlation across the breast and ovarian cancer molecular subtypes . We downloaded gene sets from the Molecular Signatures Database ( MSigDB ) [17] ( http://www . broadinstitute . org/gsea/msigdb/collections . jsp ) ( “molsigdb . v3 . 0 . entrez . gmt” ) . MSigDB contains 5 major collections ( positional gene sets , curated gene sets , motif gene sets , computational gene sets , and GO gene sets ) comprising of a total of 6769 gene sets . We limited our analysis to gene sets with less than or equal to 250 genes and valid data for genes included in the meta-data sets , resulting in 5320 gene sets in the breast cancer analysis and 5355 in the ovarian cancer analysis . We first applied SAPS to the entire collection of breast cancer cases independent of subtype . Of the 5320 gene sets evaluated , 1510 ( 28% ) achieved a raw P-Value of 0 . 05 by Ppure , 1539 ( 29% ) by Penrichment , 755 ( 14% ) by Prandom , 581 ( 11% ) by all 3 raw P-Values , and 564 ( 11% ) of these are significant at the SAPSq-value of 0 . 05 ( Figure 2 ) . The top-ranked gene sets identified by SAPS and associated with poor prognosis in all breast cancers independent of subtype contained gene sets previously found to be associated with poor prognosis in breast cancer ( Table 1 ) . Thus it is not surprising that these emerged as the most significant , and this result serves as a measure of validation . We note that the list of top gene sets associated with poor breast cancer prognosis identified in our overall analysis includes the gene set VANTVEER_BREAST_CANCER_METASTASIS_DN , which according to the Molecular Signatures Database website is defined as “Genes whose expression is significantly and negatively correlated with poor breast cancer clinical outcome ( defined as developing distant metastases in less than 5 years ) . ” Our analysis suggests that the set of genes is positively correlated with poor breast cancer clinical outcome . Comparison the gene list to the published “poor prognosis” gene list from van't Veer et al . [26] confirms that the gene list is mislabeled in the Molecular Signatures Database and is in fact the set of genes positively associated with metastasis in van't Veer et al . [26] The top-ranking gene sets associated with good prognosis were not originally identified in breast cancers , and represent a range of biological processes . Several were from analyses of hematolymphoid cells , including: genes down-regulated in monocytes isolated from peripheral blood samples of patients with mycosis fungoides compared to those from normal healthy donors , genes associated with the IL-2 receptor beta chain in T cell activation , and genes down-regulated in B2264-19/3 cells ( primary B lymphocytes ) within 60–180 min after activation of LMP1 ( an oncogene encoded by Epstein Barr virus ) . These gene sets suggest that specific subsets of immune system activation are associated with improved breast cancer prognosis , consistent with reports that the presence infiltrating lymphocytes is predictive of outcome in many cancers . We then applied SAPS to the ER+/HER2− high proliferation subtype . Of the 5320 gene sets evaluated , 1503 ( 28% ) achieved a raw P-Value of 0 . 05 by Ppure , 1667 ( 31% ) by Penrichment , 1079 ( 20% ) by Prandom , 675 ( 13% ) by all 3 raw P-Values , and all 675 of these are significant at the SAPSq-value of 0 . 05 . The top-ranking gene sets by SAPSscore are associated with cancer and proliferation . One of the top-ranking gene sets was associated with Ki67 , a well-known prognostic marker in Luminal B breast cancers [27] . Overall , the patterns of significance are highly similar to that seen in breast cancer analyzed independent of subtype ( Figure 3 , Table 2 ) . Next , we used SAPS to analyze the ER+/HER2− low proliferation samples . Of the 5320 gene sets evaluated , 494 ( 9% ) achieved a raw P-Value of 0 . 05 by Ppure , 1113 ( 21% ) by Penrichment , 939 ( 18% ) by Prandom , 303 ( 6% ) by all 3 raw P-Values , and all 303 of these were significant at the SAPSq-value of 0 . 05 . The top-ranking ER+/HER2− low proliferation prognostic gene sets by SAPSscore are also highly enriched for genes involved in proliferation ( Figure 4 , Table 3 ) . Top ranking gene sets associated with good prognosis include those highly expressed in lobular breast carcinoma relative to ductal and inflammation-associated genes up-regulated following infection with human cytomegalovirus . Then , we applied SAPS to the HER2+ subset . Of the 5320 gene sets evaluated , 1247 ( 23% ) achieved a raw P-Value of 0 . 05 by Ppure , 1425 ( 27% ) by Penrichment , 683 ( 13% ) by Prandom , 439 ( 8% ) by all 3 raw P-Values , and 342 ( 6% ) of these are significant at the SAPSq-value of 0 . 05 . Most of the top-ranking prognostic pathways in the HER2+ group by SAPSscore are associated with better prognosis and include several gene sets associated with inflammatory response ( Figure 5 , Table 4 ) . A gene set containing genes down-regulated in multiple myeloma cell lines treated with the hypomethylating agents decitabine and trichostatin A was significantly associated with improved prognosis in HER2+ breast cancer . The top-ranking gene set associated with decreased survival is a hypoxia-associated gene set . Hypoxia is a well-known prognostic factor in breast cancer [28] , [29] , and our analysis suggests it shows a very strong association with survival in the HER2+ breast cancer molecular subtype . Finally , we used SAPS to analyze the poor-prognosis “basal like” subtype which was classified as being ER−/HER2− . Of the 5320 gene sets evaluated , 786 ( 15% ) achieved a raw P-Value of 0 . 05 by Ppure , 1208 ( 23% ) by Penrichment , 304 ( 6% ) by Prandom , 126 ( 2% ) by all 3 raw P-Values , and 25 ( 0 . 5% ) of these are significant at the SAPSq-value of 0 . 05 . Top-ranking gene sets associated with poor survival include genes up-regulated in MCF7 breast cancer cells treated with hypoxia mimetic DMOG , genes down-regulated in MCF7 cells after knockdown of HIF1A and HIF2A , genes regulated by hypoxia based on literature searches , genes up-regulated in response to both hypoxia and overexpression of an active form of HIF1A , and genes down-regulated in fibroblasts with defective XPC ( an important DNA damage response protein ) in response to cisplatin ( Figure 6 , Table 5 ) . This analysis suggests that hypoxia-associated gene sets are key drivers of poor prognosis in HER2+ and ER−/HER2− breast cancer subtypes . Interestingly , cisplatin is an agent with activity in ER−/HER2− breast cancer , and it is has been suggested that ER−/HER2− breast cancers with defective DNA repair may show increased susceptibility to cisplatin [30] . Our analysis for ovarian cancer was similar to that for breast cancer . We began by applying SAPS to the entire collection of ovarian cancer samples independent of subtype . Of the 5355 gene sets evaluated , 1190 ( 22% ) achieved a raw P-Value of 0 . 05 by Ppure , 1391 ( 26% ) by Penrichment , 755 ( 14% ) by Prandom , 497 ( 9% ) by all 3 raw P-Values ( Figure 7 , Table 6 ) , and all 497 of these are significant at the SAPSq-value of 0 . 05 . The top gene sets are involved in stem cell-related pathways and pathways related to epithelial-mesenchymal transition , including genes up-regulated in HMLE cells ( immortalized non-transformed mammary epithelium ) after E-cadhedrin ( CDH1 ) knockdown by RNAi , genes down-regulated in adipose tissue mesenchymal stem cells vs . bone marrow mesenchymal stem cells , genes down-regulated in medullary breast cancer relative to ductal breast cancer , genes down-regulated in basal-like breast cancer cell lines as compared to the mesenchymal-like cell lines , genes up-regulated in metaplastic carcinoma of the breast subclass 2 compared to the medullary carcinoma subclass 1 , and genes down-regulated in invasive ductal carcinoma compared to invasive lobular carcinoma . We then analyzed the angiogenic subtype . Of the 5355 gene sets evaluated , 1153 ( 22% ) achieved a raw P-Value of 0 . 05 by Ppure , 1377 ( 26% ) by Penrichment , 624 ( 12% ) by Prandom , 371 ( 7% ) by all 3 raw P-Values ( Figure 7 , Table 6 ) , and all of these are significant at the SAPSq-value of 0 . 05 . Top-ranking gene sets associated with poor prognosis in the angiogenic subtype include: a set of targets of miR-33 ( associated with poor prognosis ) ( Figure 8 , Table 7 ) . This microRNA has not previously been implicated in ovarian carcinogenesis . Other top hits include several immune response gene sets , which were associated with improved prognosis . Finally , we analyzed the non-angiogenic subtype of ovarian cancer . Of the 5355 gene sets evaluated , 981 ( 18% ) achieved a raw P-Value of 0 . 05 by Ppure , 957 ( 18% ) by Penrichment , 658 ( 12% ) by Prandom , 261 ( 5% ) by all 3 raw P-Values ( Figure 7 , Table 6 ) , and of these , 254 ( 5% ) are significant at the SAPSq-value of 0 . 05 ( Figure 9 , Table 8 ) . The top ranked pathways associated with improved survival are immune-related gene sets and a gene set found to be negatively associated with metastasis in head and neck cancers . To assess similarities and differences in prognostic pathways in both breast and ovarian cancer molecular subtypes , we performed hierarchical clustering of the disease subtypes using SAPSscores . Specifically , we identified the 1300 gene sets with SAPSq-value≤0 . 05 and absolute value ( SAPSscore ) ≥1 . 3 in at least one of the breast and ovarian cancer molecular subtypes . We clustered the gene sets and disease subtypes using hierarchical clustering with complete linkage and distance defined as one minus Spearman rank correlation ( Figure 10 ) . This analysis shows two dominant clusters of disease subtypes , with one cluster containing ER+/HER2− high proliferation and ER+/HER2− low proliferation breast cancer molecular subtypes , and the second cluster containing ovarian cancer molecular subtypes and the ER−/HER2− and HER2+ breast cancer molecular subtypes . SAPSscores for within ER+ breast cancer molecular subtypes , within ER−/HER2− and HER2+ breast cancer molecular subtypes , and within ovarian cancer molecular subtypes show high correlation ( Spearman rho = 0 . 61 , 0 . 68 , and 0 . 51 , respectively , all p<2 . 2×10−16 ) . Interestingly , the SAPSscores for the ER−/HER2− and HER2+ breast cancer subtypes show far greater correlation with the SAPSscores in the ovarian cancer molecular subtypes than with the SAPSscores in ER+ molecular subtypes ( median Spearman rho is 0 . 5 for correlation of ER−/HER2− and HER2+ breast cancer molecular subtypes with ovarian cancer molecular subtypes vs . 0 . 16 for ER− molecular subtypes with ER+ molecular subtypes ( Figure 10 ) . This analysis demonstrates the importance of performing subtype-specific analyses in breast cancer , as breast cancer is an extremely heterogeneous disease and prognostic pathways in ER−/HER2− and HER2+ breast cancer subtypes are far more similar to prognostic pathways in ovarian cancer than with prognostic pathways in ER+ breast cancer subtypes . Recently , the TCGA breast cancer analysis demonstrated that the “basal” subtype of breast cancer ( ER−/HER2− ) showed genomic alterations far more similar to ovarian cancer than to other breast cancer molecular subtypes [31] . Our findings show that ER−/HER2− breast cancers share not only genomic alterations but also prognostic pathways with ovarian cancer . Examining the clusters of gene sets with differential prognostic associations across breast and ovarian cancer molecular subtypes shows three predominant clusters of gene sets . The first cluster is predominantly composed of proliferation-associated gene sets . The second cluster comprised a mixture of EMT-associated gene sets , gene sets associated with angiogenesis , and with developmental processes . The third is comprised predominantly of gene sets associated with inflammation . The proliferation cluster of gene sets is strongly associated with poor prognosis in breast cancer overall and ER+ breast cancer subtypes . This supports prior studies demonstrating that proliferation is the strongest factor associated with prognosis in breast cancer overall [15] and in its ER+ molecular subtypes [6] . Interestingly , the proliferation cluster of gene sets shows little association with survival in ER−/HER2− and HER2+ breast cancer and ovarian cancer and its subtypes , and it is the EMT , hypoxia , angiogenesis , and development-associated cluster of gene sets that are associated with poor prognosis in these diseases/subtypes with these pathways showing little association with poor prognosis in ER+ breast cancer . The cluster of immune-related pathways tends to show association with improved prognosis across breast and ovarian cancer and their subtypes ( Figure 10 ) . A significant body of work has focused on identifying prognostic signatures in breast cancer . Recently , Venet et al . showed that most random signatures are able to stratify patients into groups that show significantly different survival [15] . This work suggests that more sophisticated and statistically rigorous methods are needed to identify biologically informative gene sets based on observed prognostic associations . Here we describe such a statistical and computational framework ( Significance Analysis of Prognostic Signature ( SAPS ) ) to allow robust and biologically informative prognostic gene sets to be identified in disease . The basic premise of SAPS is that in order for a candidate gene set's association with prognosis to be used to imply its biological significance , the gene set must satisfy three conditions . First , the gene set should cluster patients into prognostically variable groups . The p value generated from this analysis is the standard Ppure , which has been frequently used in the literature to indicate a gene set's clinical and biological relevance for a particular disease . A key insight of the SAPS method ( building on the work of Venet et al . [15] ) is that clinical utility and biological relevance of a gene set are two very different properties , necessitating distinct statistical tests . The Ppure assesses the statistical significance of survival differences observed between two groups of patients stratified using a candidate gene set , and thus this test provides insight into the potential clinical utility of a gene set for stratifying patients into prognostically variable groups; however , this statistical test provides no information to compare the prognostic performance of the candidate gene set with randomly generated ( “biologically null” ) gene sets . We believe that it is essential for a candidate prognostic gene set to not only stratify patients into prognostically variable groups , but to do so in a way that is significantly superior to a random gene set of similar size . Therefore , the second condition of the SAPS method is that a gene set must stratify patients significantly more effectively than a random gene set . This analysis produces the Prandom . The Prandom directly compares the prognostic association of a candidate gene set with the prognostic association of “biologically null” random gene sets . Lastly , to avoid selecting a gene set that is linked to prognosis solely by the unsupervised k-means clustering procedure , the SAPS procedure additionally requires a prognostic gene set to be enriched for genes that show strong univariate associations with prognosis . Therefore , the third condition of the SAPS method is that a candidate gene set should achieve a statistically significant Penrichment , which is a measure of the statistical significance of a candidate gene set's enrichment with genes showing strong univariate prognostic associations . Our results in breast and ovarian cancer and their molecular subtypes demonstrate that the Penrichment shows only moderate overall correlation with the Ppure and Prandom ( range Spearman rho = ( 0 . 23–0 . 35 ) , median Spearman rho = 0 . 30 ) ) and there is only moderate overlap between gene sets identified at a raw p value of 0 . 05 by Ppure , Prandom , and Penrichment ( Figures 2A–9A ) . These data suggest that the Penrichment provides useful additional information to the Ppure and Prandom and allows prioritization of gene sets that are enriched for genes showing strong univariate prognostic associations . Summarizing these three distinct statistical tests into a single score is a difficult task as they were each generated using different methods and they test different hypotheses . We chose to use the maximum as the summary function ( as opposed to a median or average , for example ) , as the maximum is a conservative summary measure and it is easily interpretable . It is important to note that the SAPS method provides users with the SAPSscore as well as all 3 component P values ( and the 3 component q-values corrected for multiple hypotheses to control the FDR ) , and therefore the user can choose to use the SAPSscore or to focus on a particular SAPS component , as desired for the specific experimental question being evaluated . Importantly , the SAPS method also performs a permutation-test to estimate the statistical significance of gene set's SAPSscore . To test the utility of SAPS in providing insight into prognostic pathways in cancer , we performed a systematic , comprehensive , and well-powered analysis of prognostic gene signatures in breast and ovarian cancers and their molecular subtypes . This represents the largest meta-analysis of subtype-specific prognostic pathways ever performed in these malignancies . The analysis identified new prognostic gene sets in breast and ovarian cancer molecular subtypes , and demonstrated significant variability in prognostic associations across the diseases and their subtypes . We find that proliferation drives prognosis in ER+ breast cancer , while pathways related to hypoxia , angiogenesis , development , and expression of extracellular matrix-associated proteins drive prognosis in ER−/HER2− and HER2+ breast cancer and ovarian cancer . We see an association of immune-related pathways with improved prognosis across all subtypes of breast and ovarian cancers . Our analysis demonstrates that prognostic pathways in HER2+ and ER−/HER2− breast cancer are far more similar to prognostic pathways in angiogenic and non-angiogenic ovarian cancer than to prognostic pathways in ER+ breast cancer . This finding parallels the recent identification of similar genomic alterations in ovarian cancer and basal-like ( ER−/HER2− ) breast cancer [31] . These results demonstrate the importance of performing subtype-specific analyses to gain insight into the factors driving biology in cancer molecular subtypes . If molecular subtype is not accounted for , prognostic gene sets identified in breast cancer are strongly associated with proliferation [15]; however , when subtype is accounted for , significant and highly distinct pathways ( showing no significant association with proliferation ) are identified as driving prognosis in ER− breast cancer subtypes . Overall , these data show the utility of performing subtype-specific analyses and using SAPS to test the significance of prognostic pathways . Furthermore , our data suggest that ER− breast cancer subtypes and ovarian cancer may share common therapeutic targets , and future work should address this hypothesis . In summary , we believe SAPS will be widely useful for the identification of prognostic and predictive biomarkers from clinically annotated genomic data . The method is not specific to gene expression data and can be directly applied to other genomic data types . In the future , we believe that prior to reporting a prognostic gene set , researchers should be encouraged ( and perhaps required ) to apply the SAPS ( or a related ) method to ensure that their candidate prognostic gene set is significantly enriched for prognostic genes and stratifies patients into prognostic groups significantly better than the stratification obtained by random gene sets . Data-sets were provided as Supplemental Material in Haibe-Kains et al . [20] . Our analysis included 19 datasets with survival data ( total n = 3832 ) ( ) . Data-sets were provided as Supplemental Material in Bentink et al . [21] . Our analysis included 1735 ovarian cancer patients for whom overall survival data were available ( Table S2 ) . For breast cancer , the SCMGENE model [20] was used in the R/Bioconductor genefu package [22] to stratify patients into four molecular subtypes: ER+/HER2− low proliferation , ER+/HER2− high proliferation , ER−/HER2− and HER2+ . In the ovarian datasets we used ovcAngiogenic model [21] as implemented in genefu . For genes with multiple probes , we selected the probe with the highest variance . We tested two procedures for merging of data: subtype-specific scaling , and traditional ( non subtype-specific scaling ) ( as described in “Data-Scaling and Merging” portion of the manuscript ) . We excluded genes and cases with more than 50% of data missing . From these reduced data matrices , we imputed missing values using the impute package in R [32] . These pre-processed meta-data sets are included as Supporting Information in Dataset S1 for both breast and ovarian cancer using subtype-specific and traditional scaling . Gene sets from the Molecular Signatures Database ( MSigDB ) [17] ( http://www . broadinstitute . org/gsea/msigdb/collections . jsp ) ( “molsigdb . v3 . 0 . entrez . gmt” ) . Analyses were limited to gene sets of size greater than 1 and less than or equal to 250 genes . The SAPS procedure is described in “Significance Analysis of Prognostic Signatures ( SAPS ) ” portion of the manuscript . Briefly , for a candidate gene set , SAPS generates 3 component p-values: Ppure , Prandom , and Penrichment . The SAPSscore is the maximum of these values . The Ppure is the standard log-rank p value , computed by performing K-means clustering with a k of 2 and assessing the statistical significance of the survival difference between the 2 resulting clusters , implemented using the survdiff function in the R package survival and extracting the chi-square statistic for a test of equality of the 2 survival curves . To compute the Prandom , we generate a distribution of Ppure from “random” gene sets ( we used 10000 random gene sets for a sequence of 8 gene set sizes ranging from 5 to 250 ) , and we calculate the proportion of random gene sets of a similar size to the candidate gene sets that achieve a Ppure at least as significant as the true Ppure . To compute the Penrichment , we generate “ . rnk” files that include each gene and its concordance index for survival , implemented with the function concordance . index in the survcomp R package . These “ . rnk” files are used in a pre-ranked GSEA analysis implemented with the executable jar file gsea2-2 . 07 ( which is downloadable from: http://www . broadinstitute . org/gsea/downloads . jsp ) . In our analyses , we set a maximum gene set size of 250 and used default GSEA parameters . The SAPSscore for each candidate gene set is then computed as the negative log10 of the maximum of the ( Ppure , Prandom , and Penrichment ) times the direction of the association ( positive or negative ) . The statistical significance of the SAPSscore is determined by permutation-testing . Specifically , in our experiments , we performed 10000 permutations of the gene labels for each of the sequence of 8 of gene set sizes ranging from 5 to 250 . We performed the full SAPS procedure for each of the 80000 permuted gene sets and we generated a null distribution of 10000 SAPSscores for each of the 8 gene set sizes . The SAPSp-value was computed as the proportion of permuted gene sets of a similar size to the candidate gene set that achieved at least as extreme a SAPSscore . The SAPSp-values were then converted to SAPSq-values using the method of Benjamini and Hochberg [19] . Hierarchical clustering was performed on the SAPS scores for breast and ovarian cancer molecular subtypes . Hierarchical clustering was performed with one minus Spearman rank correlation as the distance metric and complete linkage , using the Cluster 3 . 0 package ( http://bonsai . hgc . jp/~mdehoon/software/cluster/ ) . Clustering results were visualized using Java TreeView ( http://jtreeview . sourceforge . net/ ) . The Java TreeView files used to generate the Heatmap in Figure 10 are provided in the Supplementary Information ( “BreastOvary_HC . zip” ) . An R script and R workspaces for running SAPS on the breast and ovarian cancer meta-data sets and generating Scatterplots and Venn Diagrams of the SAPS P-Values ( including all figures from our analyses ) are included in in Dataset S1 ( http://dx . doi . org/10 . 5061/dryad . mk471 ) . The Venn diagrams were generated with the Vennerable package in R .	A major goal in biomedical research is to identify sets of genes ( or “biological signatures” ) associated with patient survival , as these genes could be targeted to aid in diagnosing and treating disease . A major challenge in using prognostic associations to identify biologically informative signatures is that in some diseases , “random” gene sets are associated with prognosis . To address this problem , we developed a new method called “Significance Analysis of Prognostic Signatures” ( or “SAPS” ) for the identification of biologically informative gene sets associated with patient survival . To test the effectiveness of SAPS , we use SAPS to perform a subtype-specific meta-analysis of prognostic signatures in large breast and ovarian cancer meta-data sets . This analysis represents the largest of its kind ever performed . Our analyses show that only a small subset of the gene sets found statistically significant using standard measures achieve significance by SAPS . We identify new prognostic signatures in breast and ovarian cancer and their corresponding molecular subtypes , and we demonstrate a striking similarity between prognostic pathways in ER negative breast cancer and ovarian cancer , suggesting new shared therapeutic targets for these aggressive malignancies . SAPS is a powerful new method for deriving robust prognostic biological pathways from clinically annotated genomic datasets .	[ "Abstract", "Introduction", "Results", "Discussion", "Methods" ]	[ "oncology", "medicine", "breast", "tumors", "mathematics", "gynecological", "tumors", "statistics", "biostatistics", "cancers", "and", "neoplasms", "statistical", "methods" ]	2013	Significance Analysis of Prognostic Signatures
Drugs currently available for leishmaniasis treatment often show parasite resistance , highly toxic side effects and prohibitive costs commonly incompatible with patients from the tropical endemic countries . In this sense , there is an urgent need for new drugs as a treatment solution for this neglected disease . Here we show the development and implementation of an automated high-throughput viability screening assay for the discovery of new drugs against Leishmania . Assay validation was done with Leishmania promastigote forms , including the screening of 4 , 000 compounds with known pharmacological properties . In an attempt to find new compounds with leishmanicidal properties , 26 , 500 structurally diverse chemical compounds were screened . A cut-off of 70% growth inhibition in the primary screening led to the identification of 567 active compounds . Cellular toxicity and selectivity were responsible for the exclusion of 78% of the pre-selected compounds . The activity of the remaining 124 compounds was confirmed against the intramacrophagic amastigote form of the parasite . In vitro microsomal stability and cytochrome P450 ( CYP ) inhibition of the two most active compounds from this screening effort were assessed to obtain preliminary information on their metabolism in the host . The HTS approach employed here resulted in the discovery of two new antileishmanial compounds , bringing promising candidates to the leishmaniasis drug discovery pipeline . Leishmaniasis is a neglected emerging disease without any adequate treatment adapted to the field [1] . The disease can be characterized by skin ulcers ( cutaneous leishmaniasis ) , mucous degeneration , especially from the mouth and internal nose ( mucocutaneous leishmaniasis ) , and visceral organ damage ( visceral leishmaniasis ) , which is lethal if untreated . The different forms of leishmaniasis manifestation depend mainly on the species of parasite but are also related to the host immune system . Official World Health Organization ( WHO ) numbers from the 1990s are still used and estimate 12 million infected people and 350 million at risk living in one of the 88 endemic countries in America , Europe , Africa , the Middle East and Asia [2] . The number of deaths as a consequence of leishmaniasis is higher than 50 , 000 per year , with an incidence of 1 . 5 million annual registered cases of the disfiguring cutaneous leishmaniasis and 0 . 5 million annual registered cases of the potentially fatal visceral leishmaniasis [3] , but these numbers probably underestimate the real burden of the disease [4] , [5] . Leishmaniasis is caused by the kinetoplastid species from the genus Leishmania . Infection takes place when a sandfly vector inoculates Leishmania promastigotes into the mammalian bloodstream; these extracellular flagellated forms of the parasite live in the insect midgut . Once in the bloodstream , parasites are phagocytosed by mononuclear blood cells , especially macrophages , differentiating into the obligatory intracellular amastigote form . Amastigotes proliferate inside the macrophages before inducing the bursting of the host cell and being released into the bloodstream . This process occurs repeatedly , leading to tissue damage [6] . Parasite species and the host immune system determine the clinical status of the disease , ranging from cutaneous ulcers ( cutaneous leishmaniasis ) [7] to visceral organ damage ( visceral leishmaniasis ) [8] , especially of the spleen and the liver . Most of the antileishmanial drugs currently in use for treatment , from the long time established antimonials to the recently introduced miltefosine , have disadvantages , such as patient toxicity , side effects and/or parasite resistance [9] . Lead discovery is currently one of the bottlenecks in the pipeline for novel antileishmanial drugs [10] . High-throughput screening ( HTS ) optimizes the chance of finding lead compounds through the identification of active compounds from a large number of candidates [11]–[12] . We adapted an in vitro fluorometric assay to HTS format using the promastigote form of L . major [13] , one of the causative species of cutaneous leishmaniasis . This was the first reference strain used to sequence the genome of this parasite , completed in 2005 [14] , and genome information can be accessible for future studies , including target identification or mechanism of action determination . To validate the assay in HTS format , we screened a 4 , 000-compound library containing many bioactive compounds with known pharmacological properties , including currently used antileishmanials . Following validation , the assay was applied to the screening of a library containing 26 , 500 structurally diverse chemical compounds . A total of 567 compounds showing a minimum of 70% growth inhibition of the parasite ( L . major ) were identified during the primary screening at 10 µM . Further tests on their cytotoxicity on a human macrophage cell line and specificity filtering were applied , resulting in a list of 124 active compounds . To confirm activity against the intracellular parasites , these 124 compounds were tested in serial dilutions against L . major amastigotes infecting THP-1 differentiated macrophages . Through this process , the two most active compounds with EC50 values lower than 10 µM against L . major were chosen for further characterization . The 124 active compounds were also tested against intramacrophagic L . donovani , one of the causative species of visceral leishmaniasis , to evaluate specificity of the compounds against the parasites causing different clinical manifestations of the disease . To determine the quality of the two most active compounds , we tested their microsomal stability , which would indicate the presence of metabolites , as well as cytochrome P450 ( CYP ) inhibition for drug-drug interaction in multitherapies . The results indicate that the compounds are good candidates for further characterization for leishmaniasis therapy . We discuss the relevance of a developed HTS assay using the promastigote form of the parasite for the discovery of leishmanicidal compounds and the potential of one of the selected hits to become a future lead compound against leishmaniasis . L . major MHOM/IL/81/FRIEDLIN and L . donovani MHOM/ET/67/HU3 promastigotes were cultivated at 28°C in axenic M199 culture medium ( Welgene , S . Korea ) supplemented with 10% heat-inactivated fetal bovine serum ( FBS ) ( Gibco , United States ) and 1% streptomycin/penicillin ( Gibco , United States ) . A total of 4 , 000 small molecules sourced from Sigma , Prestwick and Tocris were all screened at 2–20 µM , 0 . 2–2 µM and 0 . 02–0 . 2 µM . The 26 , 500-compound library screened at 10 µM ( in 1% DMSO ) was sourced from TimTec . This small molecule library contains compounds selected for diversity and drug-like properties as well as small collections of kinase-focused and protease-focused compounds . Ethidium bromide ( EtBr ) ( Sigma E1510 , United States ) , amphotericin B ( Sigma A9528 , United States ) , miltefosine ( A . G . Scientific H-1096 , United States ) and paromomycin sulfate salt ( Sigma P9297 , United States ) were used as reference compounds . After compound addition to the assay plate ( Evotec™ 384-well microplate , Germany ) , 20 , 000 L . major promastigote parasites from an exponential phase culture ( ∼107 parasites/mL ) were diluted in M199 , seeded in 50 µL per well using FlexDrop™ and incubated at 28°C for 28 hours , followed by the addition of 5 mM resazurin sodium salt ( Sigma R7017 , United States ) and further incubation for 20 hours at 28°C . After a 48-hour exposure to compounds , the reference drug ( EtBr ) or control ( 1% DMSO ) , the parasites were fixed with 2% paraformaldehyde ( PFA ) and plates were read in Victor3™ ( Perkin Elmer ) at 530 nm ( excitation ) and 590 nm ( emission ) . This fixation step is not necessary for resazurin readout , but allows flexibility in the automation schedule and increases assay robustness by decreasing metabolic variability between populations across wells and plates . Z-factor , calculated as 1− ( 3×σp+3×σn ) / ( \|μp−μn\| ) , where μp , σp , μn and σn are the means ( μ ) and standard deviations ( σ ) of both the positive ( p ) and negative ( n ) controls [15] , and other parameters , including DRC plates for verification of the reference drug EC50 ( accepted if within the range of 3× higher or lower than a defined value from the literature ) , coefficient of variation not higher than 10% in the controls and edge effect evaluation , were used for screening validation and hits selection . To assess the cytotoxicity of compounds on THP-1 , an acute monocytic leukemia-derived human cell line ( ATCC TIB-202™ ) , a viability assay also using resazurin reduction with minor modifications in concentration and incubation time was performed . Z-factor [15] and other parameters [16] were used for secondary screening validation and hit exclusion . THP-1 cells were cultivated in suspension at a density of 105 to 106 cells/mL in RPMI ( Gibco 1640 , United States ) medium supplemented with 10% heat-inactivated FBS and 1% streptomycin/penicillin at 37°C and 5% CO2 . The differentiation of THP-1 into macrophages was induced by incubation of the cells for 48 hours with phorbol 12-myristate 13-acetate ( PMA ) at 50 ng/ml . For infection , late-stage L . major or L . donovani rich in metacyclic promastigotes was added to differentiated THP-1 cells ( ratio of 50 parasites to 1 host cell ) in Evotec™ 384-well microplates using a FlexDrop™ ( Perkin Elmer ) to dispense 50 µL/well . Compounds were added 24 hours after infection and microplates were incubated at 37°C and 5% CO2 for 4 days . Amphotericin B as the reference compound ( positive control ) at 10 µM as the EC100 and 1% DMSO ( negative control ) were used in all the plates for data normalization ( see bellow ) . After 4 days of incubation , remaining free promastigotes were removed by washing five times with PBS and cells were simultaneously fixed with 2% PFA and the DNA stained with 5 µM of Draq5 ( Biostatus DR50200 , England ) . Microplates were read in an Opera confocal microscope ( Perkin Elmer ) , enabling the determination of the infection ratio of the parasites by image analysis . An algorithm was developed to identify and individualize the macrophages by setting an intensity threshold to discriminate background ( extracellular space ) from foreground ( macrophage cells area ) . The macrophage nuclei and internalized parasites were identified and counted if localized in the foreground previously selected ( Figure S1 ) . With this technique , extracellular parasites were not considered in the calculations , and the infection ratio was determined by the number of infected cells divided by the total number of cells after normalization based on the controls . The average infection ratios from the positive and negative controls were normalized to 0% and 100% infection , and the infection ratio read from each compound activity was proportionally distributed within this range . Z-factor [15] was used for protocol validation and active compound selection acceptance . Compounds at a concentration of 2 µM in 0 . 2% DMSO were incubated with 0 . 5 mg/mL rat ( Sprague-Dawley Rat , BD Gentest ) and human ( human pool donors , BD Gentest ) liver microsomes in potassium phosphate buffer in a reaction started by the addition of NADPH and stopped either immediately or at 5 , 10 , 30 , 60 or 120 minutes for a precise estimate of microsomal stability . The corresponding loss of the parent compound was determined by a quadrupole liquid chromatography-mass spectrometry ( LC-MS ) with diode-array detection ( Agilent 1200 , Agilent Technology ) . The samples were passed through trapping cartridges ( Security Guard Cartridge , Gemini C18 , 4×2 . 0 mm , 3 µm , Phenomenex ) followed by an analytical column ( Gemini C18 , 50×2 . 0 mm , 3 µm , Phenomenex ) . Positive electrospray ionization ( ESI+ ) was employed for this analysis . The mobile phases A ( water with 0 . 1% formic acid ) and B ( acetonitrile with 0 . 1% formic acid ) were used at a flow rate of 0 . 3 mL/min . Gradient elution started with 95% mobile phase A and 5% mobile phase B . Elution was changed to a linear gradient until 50% A and B for 1 min . This condition was held for 0 . 5 min , then increased to 95% B over 0 . 5 min , and held for 1 . 5 min . Then , the elution gradient returned to 95% A and 5% B over 0 . 5 min , was held for the remaining 3 . 5 min . The percentage of the remaining compound was calculated by comparison with the initial quantity at 0 min . Half-life was calculated based on first-order reaction kinetics . Individual fluorescent probe substrates were used with individual rhCYP isozymes and fluorescence detection according to a previously published method [17] . Probe substrates were 7-benzyloxy-4- ( trifluoromethyl ) -coumarin ( BFC ) for CYP3A4 and 3-[2- ( N , N-diethyl-N-methylammonium ) ethyl]-7-methoxy-4-methylcoumarin ( AMMC ) for CYP2D6 in 0 . 5% DMSO . IC50 was determined using an eight-point concentration curve with three-fold serial dilutions . Victor3™ ( Perkin Elmer ) was used for quantification in the fluorescent method ( Perkin Elmer Life and Analytical Sciences ) . The effect of compounds on Leishmania viability was assessed by fluorometric measurement of resazurin reduction [13] . Equivalence between the fluorescence signal from resazurin reduction and the number of parasites was confirmed by the linear correlation between parasites counted by light microscopy and the relative fluorescence unit ( RFU ) value measured ( Fig . 1A ) . Reference compounds for antileishmanial activity ( EtBr , amphotericin B , miltefosine and paromomycin ) were tested ( Fig . 1B ) as a step of the validation process . Values obtained for these compounds in our assay in HTS format were similar to those classically reported for these drugs [18] , [19] . Assay validation was performed on three separate days using 33 microplates ( 384 wells/plate ) containing only controls . Variability between well-to-well , plate-to-plate and day-to-day were measured to confirm assay robustness , resulting in a Z-factor of 0 . 62 ( Fig . 1C ) . After assay validation , we screened a 4 , 000-compound library containing a number of compounds with known pharmacological properties . The screen was performed using three different concentrations with 10× dilution factors for each compound with the highest concentration in the range of 2–20 µM , as compounds in the library did not all have the same molarity . This assay was done in duplicate and results of each concentration assay are represented as individual graphs in Fig . 1D . A 70% proliferation inhibition of the parasites from the lowest compound concentration ( 0 . 2–0 . 02 µM ) was the cut-off used for active compound selection . The list of active chemicals included , but was not limited to , previously reported antileishmanial compounds: anisomycin [20] , pentamidine isethionate [21] , berberine chloride [22] , parthenolide [23] , nitrofural [24] , furazolidone [24] and nifurtimox [25] ( data not shown ) . These results were considered a pharmacological validation of the assay , confirming the ability of the assay to identify antileishmanial compounds . A library containing 26 , 500 structurally diverse chemical compounds was screened at 10 µM against L . major . The positive ( EC100 ) and negative controls ( 1% DMSO ) provided a Z-factor of 0 . 80 ( Fig . 2A ) . The 70% growth inhibition cut-off criterion was used to select the most active compounds . The frequency map distribution based on binned RFUs highlights the distinction of two groups of compounds , in which ∼97% were in the non-active group with RFUs higher than 201 , 600 . Another group contained ∼2% ( 567 compounds ) with RFUs lower than 105 , 500 , representing the active compounds ( Fig . 2B ) . The remaining ∼1% were situated between the other 2 groups and were not sufficiently active for selection . To exclude potentially toxic compounds from the active list , we performed a secondary viability screening using the non-differentiated human macrophage cell line THP-1 . Compounds that interfered with THP-1 viability at a concentration of 10 µM or lower were discarded . In addition , compounds that were found to be active in other in-house , anti-infective HTS campaigns were also excluded to avoid non-specific mechanisms of action . This cytotoxicity determination and Leishmania specificity filter resulted in a list of 124 active compounds ( Fig . 2C ) . Infection of human macrophage cells in vitro with Leishmania has been previously described as a suitable model for screening [26] and was adopted for our purpose . The 124 active compounds selected from the secondary assay were tested in 10-point dose response with 2-fold serial dilutions starting from 20 µM against promastigotes and intracellular amastigotes of L . major and L . donovani infecting macrophages . The average Z-factor calculated per plate based on the reference drug ( amphotericin B ) and carrier ( 1% DMSO ) was 0 . 62 for L . major infection and 0 . 59 for L . donovani infection , and minimum accepted Z-factor for the analysis was 0 . 5 . The amphotericin B EC50 for L . major infecting macrophages was 1 . 06 µM and EC50 for L . donovani infecting macrophages was 0 . 82 µM . From the tested compounds , 5 exhibited activity against L . major intracellular amastigotes and are presented in Table 1 . Activity confirmation was based on elimination or growth inhibition of both promastigotes and intracellular amastigotes ( parasite forms ) of L . major and L . donovani . Differences between promastigote and intracellular amastigote forms regarding compound susceptibility were demonstrated , as shown in Figure S2 . Based on the methods applied , for both L . major and L . donovani all the compounds , except CA272 for L . major , showed more potency against the extracellular form of the parasite ( Figure S2 ) . When comparing species sensitivity we found that some compounds caused different responses in different species . TE122 , for example , was only active against the intracellular L . major but showed no activity against intracellular L . donovani up to 20 µM , as shown in Table 1 and Figure S2 . This different drug sensitivity within Leishmania species is already known [27] and must be considered in future therapy development . Using controls with carrier ( 1% DMSO ) and 10 µM amphotericin B ( EC100 concentration ) ( Fig . 3A ) , the two most active compounds , CH872 and CA272 ( Fig . 3B ) , were selected based on image analysis ( for details , see Figure S1 ) . These compounds showed significant reduction of macrophage infection with L . major after four days incubation ( Fig . 3D , E ) . Moreover , after phenotypic evaluation of the macrophages post-treatment , the compound activity was not due to toxic effects on host cells . Figs . 3C and 3D show , respectively , the infection exposed to a sub-optimal concentration ( 1 nM ) and one example of active concentrations of the compounds ( 0 . 7 µM for CH872 and 10 µM for and CA272 ) according to infection reduction . The dose-response curves against intracellular L . major of both compounds plotted in Fig . 3E resulted in a calculated EC50 of 0 . 3 µM for compound CH872 and 0 . 8 µM for CA272 . Only these compounds showed EC50s against intracellular L . major lower than 1 µM and a ratio CC50/EC50 ( L . major ) greater than 10 and were selected for further characterization . The metabolic stability of the selected compounds in the presence of human and rat liver microsomes was assessed . Compound CH872 was stable in the presence of human liver microsomes , with 99 . 6% of the parent compound remaining after 30 minutes and a long theoretical half-life . However , it was unstable in rat liver microsomes , being completely degraded after 25 minutes , giving a theoretical half-life of 4 . 1 minutes ( Fig . 4 ) . For CA272 , 77 . 8% of the parent compound remained after 30 minutes in the presence of human liver microsomes , with a theoretical half-life of 79 . 4 minutes , whereas 53 . 3% of the parent compound remained in the presence of rat liver microsomes , with a theoretical half-life of 31 . 8 minutes ( Fig . 4 ) . The species difference is most likely due to different enzyme compositions of the liver microsomes . These differences will have to be taken into account if these compounds are to be used with in vivo pharmacokinetic models , which are typically performed in rats . CYP is a large superfamily of enzymes involved in numerous biological processes [28] . Member of the families CYP2 and CYP3 have a role in drug and steroid metabolism . Toxic side effects as well as drug-drug interactions may be predicted by testing the in vitro inhibition of these CYPs by chemical compounds [29] . CYP3A4 and CYP2D6 inhibition assays were performed with both compounds . Whereas CH872 did not inhibit CYP3A4 ( EC50≫10 µM ) or CYP2D6 ( EC50>10 µM ) , CA272 did inhibit CYP3A4 with an EC50 of 3 . 41 µM , but did not inhibit CYP2D6 ( EC50≫10 µM ) , as shown in Table 2 . A high-throughput screening assay for the identification of antileishmanial compounds was developed involving a two-step strategy: primary screening using promastigotes in a 384-well plate format and secondary screening of active compounds using intracellular amastigotes . The primary screening was validated on a robust statistical basis ( Fig . 1C ) and was demonstrated to be able to identify known antileishmanial compounds when a library of known bioactive small molecules was screened as a proof of principle . This assay was then applied to a screen of 26 , 500 structurally diverse small molecules . In this screen , 2 . 1% of the compounds ( 567 ) inhibited parasite growth by at least 70% after 48 hours of compound exposure . From these active compounds , almost 80% were excluded due to cytotoxicity or lack of specificity ( data not shown ) , resulting in 124 compounds that were tested against the amastigote in an infection assay with a human macrophage cell line . Although the clinically relevant stage of the Leishmania parasites is the intracellular amastigote , the extracellular promastigote poses the obvious advantage of being easier and cheaper to handle in the large scale required by HTS . Besides , promastigotes and amastigotes share common metabolic machinery and pathways , and targets against the first form could be relevant against the second one . This screening strategy against promastigotes was applied by St . George et al . to screen 15 , 000 compounds against L . tarentolae [30] and recently by Sharlow et al . to investigate 200 , 000 unique compounds for L . major growth inhibition [31] . As discussed by the authors of the latter study , the use of the promastigote stage for antileishmanial drug discovery may compromise the discovery of macrophage-metabolized prodrugs , such as antimonials . In the present study , antileishmanial activity of the primary selected compounds was further confirmed against intracellular L . donovani amastigotes in a cellular image-based assay in which host cell integrity was taken into account . Although this strategy does not compensate for the possibility of missing prodrugs during the primary screening , it does guarantee that the active compounds are able to cross the macrophage membrane and kill the amastigotes inside the host cells . As expected , most of the compounds were less active or not active against the intracellular amastigote form when compared to the promastigote , as shown in Table 1 and Figure S2 . To be active against the amastigote , a compound must cross two membrane barriers ( cellular membrane of the macrophage and phagolysosome vacuole membrane ) and maintain stability under low pH and in the presence of free radicals in the phagolysosome environment , which increase the attrition rate compared to the promastigote assay . However , in the promastigote extracellular assay , the parasite is directly exposed to the compound . Furthermore , the concentrations at which compounds show activity do not have an observed effect on the macrophage host cell , confirmed by both a cytotoxicity test and image analysis . One of the two most active hit compounds was the hydrazine CA272 , a novel scaffold for antileishmanial compounds . It exhibited good efficacy in L . major infection reduction , although the activity against the intracellular L . donovani amastigote was lower . Other unfavorable properties , such as low metabolic stability against human liver microsomes ( Fig . 4 ) and inhibition of CYP3A4 ( Table 2 ) , might be improved by the optimization of the two phenyl rings , which can be easily modified . Tests against other Leishmania species should also be considered for further studies . Satisfactory activity against intracellular L . major , in addition to low toxicity , indicates a good starting point for a new antileishmanial candidate drug . The most active compound , CH872 , is of interest due to its high in vitro activity and lack of cytotoxicity ( Fig . 3 , Table 1 and Figure S2 ) , along with favorable metabolic stability and CYP inhibition data ( Table 2 ) . Additionally , its core structure , 4-hydroxyquinoline , is a novel scaffold for antileshmanial inhibition . Several modifications of quinolines have been carried out to obtain antileishmanials , such as sitamaquine or quinoline derivatives with side-chains at C4 or at phenyl rings [32] , [33]; however , none of these modifications produced 4-hydroxyquinoline derivatives . Sitamaquine is in clinical trials ( Phase II ) , and no other quinoline derivatives have been approved as antileishmanial drugs . Some issues reported from this class of compounds include kidney toxicity , which can be lethal [34] . Unlike other antileishmanial quinoline derivatives , compound CH872 contains a 4-OH group , which allows this compound to equilibrate with its tautomer CH872A ( Fig . 5 ) . Studies directed toward establishing the structure-activity relationships ( SARs ) and defining the mode of action of CH872 are currently underway . The identification of these two antileishmanial compounds demonstrates that primary screening against the promastigote form can be used to identify novel scaffolds that can serve as the starting point for hit-to-lead programs . Using the experience gained in this effort , we are currently developing and implementing an automated high-throughput drug screening assay in 384-well plate format using amastigotes infecting human macrophages as a primary screening assay . In summary , high-throughput screening of small molecule libraries against promastigotes in a rapid and simple 384-well plate format assay is able to identify novel antileishmanial compounds . Furthermore , these novel compounds have been shown to be active against the intracellular amastigote . As the promastigote screening is much easier and cheaper compared to screening the intracellular amastigote , this strategy could be used to screen libraries of natural compounds commonly available in endemic areas , where resources are often limited . This opens new opportunities for the discovery of future candidates for drug development to be performed locally in endemic countries .	Every year , more than 2 million people worldwide suffer from leishmaniasis , a neglected tropical disease present in 88 countries . The disease is caused by the single-celled protozoan parasite species of the genus Leishmania , which is transmitted to humans by the bite of the sandfly . The disease manifests itself in a broad range of symptoms , and its most virulent form , named visceral leishmaniasis , is lethal if not treated . Most of the few available treatments for leishmaniasis were developed decades ago and are often toxic , sometimes even leading to the patient's death . Furthermore , the parasite is developing resistance to available drugs , making the discovery and development of new antileishmanials an urgent need . To tackle this problem , the authors of this study employed the use of high-throughput technologies to screen a large library of small , synthetic molecules for their ability to interfere with the viability of Leishmania parasites . This study resulted in the discovery of two novel compounds with leishmanicidal properties and promising drug-like properties , bringing new candidates to the leishmaniasis drug discovery pipeline .	[ "Abstract", "Introduction", "Methods", "Results", "Discussion" ]	[ "infectious", "diseases/neglected", "tropical", "diseases" ]	2010	Antileishmanial High-Throughput Drug Screening Reveals Drug Candidates with New Scaffolds
Integrins have emerged as key sensory molecules that translate chemical and physical cues from the extracellular matrix ( ECM ) into biochemical signals that regulate cell behavior . Integrins function by clustering into adhesion plaques , but the molecular mechanisms that drive integrin clustering in response to interaction with the ECM remain unclear . To explore how deformations in the cell-ECM interface influence integrin clustering , we developed a spatial-temporal simulation that integrates the micro-mechanics of the cell , glycocalyx , and ECM with a simple chemical model of integrin activation and ligand interaction . Due to mechanical coupling , we find that integrin-ligand interactions are highly cooperative , and this cooperativity is sufficient to drive integrin clustering even in the absence of cytoskeletal crosslinking or homotypic integrin-integrin interactions . The glycocalyx largely mediates this cooperativity and hence may be a key regulator of integrin function . Remarkably , integrin clustering in the model is naturally responsive to the chemical and physical properties of the ECM , including ligand density , matrix rigidity , and the chemical affinity of ligand for receptor . Consistent with experimental observations , we find that integrin clustering is robust on rigid substrates with high ligand density , but is impaired on substrates that are highly compliant or have low ligand density . We thus demonstrate how integrins themselves could function as sensory molecules that begin sensing matrix properties even before large multi-molecular adhesion complexes are assembled . Cell adhesion to the ECM is mediated by a family of heterodimeric surface receptors called integrins [1] . In addition to their function as mechanical anchors , integrins also participate in signal transduction and thereby regulate important cell behaviors , such as differentiation , motility , survival , and morphogenesis [2] , [3] . To signal , integrins assemble laterally in the membrane and recruit structural and signaling proteins to form a clustered adhesion complex . In addition to their signaling function , assembled adhesion complexes also physically link the cell cytoskeleton to the ECM and transmit traction forces necessary for mechanical cell processes , such as motility and cell shape changes [4]–[7] . Both the physical and chemical properties of the ECM influence integrin adhesion complex assembly [3] , [8]–[15] . The density of matrix ligands and their affinity for integrin receptors determines the number , size and distribution of integrin complexes in the cell membrane [8] , [12] , [15] , [16] . Integrin clustering is especially sensitive to ligand spacing , as nanometer differences in the average spacing between ligands dictates whether or not integrins assemble into large adhesion complexes , such as focal adhesions [8] , [16] . Matrix rigidity also regulates integrin function , as stiff matrices promote the assembly of large integrin complexes ( focal adhesions ) while compliant matrices support the assembly of small point-like integrin structures if any at all [13] , [14] . Since integrin clustering is functionally linked to signal transduction and cell behavior , matrix-regulated adhesion assembly serves as a key sensory process that enables a cell to interrogate and respond to its extracellular environment . Current theory holds that the adhesion complex is embedded with molecular sensors that mediate response to matrix properties . Possibilities include protein switches that undergo tension-dependent conformational changes ( [17]–[20] and reviewed in [21] ) , as well as multivalent adaptor proteins whose incorporation into the adhesion complex are predicted to depend on factors such as matrix ligand density , matrix stiffness , and cell contractility [22]–[25] . Although receiving less attention , the integrin-ligand interaction itself could also be sensitive to the physical and chemical properties of the ECM . When both a receptor and its ligand are tethered , the kinetics and thermodynamics of complex formation depends on the intrinsic chemistry of the interaction; the distance the molecules must stretch to reach each other; and , theoretically , the compliance of the materials the molecules are tethered to [26]–[28] . This suggests that integrins could function as sensors if their aggregation is linked to bond formation with ligand . How could integrin-ligand interaction drive integrin assembly ? One popular hypothesis holds that ligand interaction induces large allosteric changes in integrins that extend to their intracellular domains ( reviewed in [1] , [29] ) . These changes in conformation could facilitate the recruitment of intracellular adaptor or signaling proteins that crosslink and cluster integrins [22] , [30] . Other possibilities , however , likely exist . Following interaction with matrix-immobilized ligands , for example , integrins can assemble into complexes in a matrix-dependent manner prior to the recruitment of intracellular proteins [10] , [31]–[33] . Consistent with this observation , a chemo-mechanical basis for how receptor-ligand interactions can drive receptor clustering independent of intracellular interactions has been described theoretically . When membranes possess two or more receptors of different lengths and chemical affinities , or possess large non-specific repellers ( i . e . large proteoglycans or glycoproteins ) , the receptors tend to phase separate into clustered or ring-like structures upon interaction with a substrate or another membrane [34]–[37] . In essence , receptors aggregate due to a competition between receptor-mediated adhesion and non-specific repulsion that resists adhesion . Integrin clustering could therefore naturally depend on the factors that control adhesion , including ligand chemistry , matrix stiffness , and cell stiffness , and also on the factors that mediate repulsion , such as the physical properties of the glycocalyx . Hence , integrins may be able to respond to matrix properties without the necessity of auxiliary sensor proteins in the adhesion complex . To explore if the glycocalyx can mediate integrin clustering independent of intracellular adaptors and if this clustering is responsive to the chemical and physical parameters that define the ECM , we developed a computational model of integrin-ligand interaction that includes a mechanical description of the cell-ECM interface . The model is based on the simulation algorithm called Adhesive Dynamics [38]–[45] , which was originally devised to study the chemo-mechanics of receptor-mediated cell adhesion under shear flow [46] . Adhesive Dynamics models integrin-ligand bonds as Hookean springs , which allows the distance-dependent kinetic rates of bond formation and rupture to be calculated with a model developed by Bell and co-workers [26] , [47] . In this work , Adhesive Dynamics was expanded to include a lattice spring model ( LSM ) of the cell-ECM interface . The LSM utilizes a defined lattice of nodes with interconnecting springs to calculate the elastic behavior of solid materials [48] , [49] . Integration of the Adhesive Dynamics and LSM algorithms enables integrin dynamics , including force-dependent bond formation and rupture , to be explored in the context of a deformable cell-ECM interface . Using the newly developed computational technique , we evaluate the relationship between integrin clustering , cell and glycocalyx mechanics , and the chemistry and mechanics of the matrix , and in doing so , predict that integrins themselves are responsive to matrix properties . Stress and strain in the interface were calculated with the LSM numerical method . The LSM is a computationally-efficient mesoscopic approach frequently used in fracture mechanics that utilizes a system of regularly spaced nodes and interconnecting harmonic springs to model the mechanical behavior of solids ( reviewed in [49] ) . When the node lattice , arrangement of spring connections , and spring constants are chosen correctly , the large-scale behavior of the LSM directly maps onto linear elasticity theory [48] . The LSM is numerically equivalent to a finite element model that has simple linear elements [48]; however , we employ the LSM methodology over the more commonly used finite element method for two primary reasons . First , the integrin-ligand bonds are described by discrete springs [46] , and these springs can be easily incorporated into the LSM . Second , the LSM avoids computationally expensive remeshing algorithms , which a finite element method would need to call upon each time bond formation or rupture occurred in the interface . To implement the LSM , a node and spring model was constructed for both the membrane/cortex plate and the ECM substrate , as shown in Figure 1B . Nodes were placed on an initially cubic lattice and all nearest {1 0 0} and next-nearest {1 1 0} neighbor nodes were connected by Hookean springs , each having the same spring constant . In response to stress , springs could pivot freely and the nodes could undergo translational movements that minimized the potential energy of the spring network . A system configured in this manner behaves as an isotropic elastic solid that has a fixed Poisson's ratio ν = 1/4 and an adjustable Young's modulus: ( 2 ) where Δx is the LSM lattice node spacing and σ is the Hookean spring constant [57] , [58] . If Δx is small compared to the length scale of interest , the spring system approximates an elastic continuum . The actin cortex and ECM , however , are not continuous on the protein-length scale , which is relevant to integrin-ligand interaction . To better reflect the micro-architecture of cell-ECM interface , we used an LSM lattice spacing of 20 nm , which is on the order of the size of a matrix protein or cytoskeletal filament . Changes in the lattice spacing by an order of magnitude , though , were not expected to alter the qualitative nature of our results if the spring constants were also adjusted to maintain the Young's moduli . In all simulations unless otherwise noted , a 1 . 4 µm×1 . 4 µm area of the cell membrane was simulated . A 40-nm thick membrane/cortex plate and a 400-nm thick ECM substrate spanning this area were constructed using a 70×70×3 and a 70×70×21 node LSM , respectively ( Figure 1B ) . The springs of each LSM were assigned a spring constant ( Table 1 ) that achieved the desired material rigidity ( Equation 2 ) . The harmonic potential ( Equation 1 ) between the membrane and substrate , i . e . the glycocalyx , was added to the model by incorporating additional linear springs between the ECM-substrate and membrane-spring networks . To add the springs , the plate and substrate networks were aligned and each node in the top surface of the substrate network was connected by a Hookean spring to the node directly above it in the bottom surface of the plate network ( Figure 1B ) . The equilibrium spring length of these connections was set equal to the desired thickness of the glycocalyx . The spring constant of the connections , σg , was related to the effective compressibility of the glycocalyx layer ( See Equation 1 ) by the following expression: ( 3 ) Bonds were modeled as Hookean springs and were added to the LSM by connecting the desired node in the top surface of the matrix LSM with a node in the bottom surface of the membrane/cortex LSM . Likewise , a bond was removed by removing the appropriate spring from the model . The deformations in the LSM caused by bond formation were calculated by relaxing the entire spring network to mechanical equilibrium . The potential energy stored in the LSM was given by ( 4 ) where the summation i is over all nodes in the system , the summation j is over all nodes connected to node i , \|rij\| is the distance between node i and j , and σij and lij are the spring constant and equilibrium length of the spring connecting node i and j . The system energy was minimized when the vector sum of forces on each node that can undergo translation was zero , which was achieved by iteratively solving the following system of equations ( 5 ) For relaxation , periodic boundary conditions were applied to the LSM nodes forming the lateral sides of the substrate and membrane/cortex networks . Under this condition , which was implemented to limit finite-size effects , the material strain induced by a stress at one side of the network propagates in a mirror-like fashion on the opposite side of the network . Initially , ligand binding sites and integrin receptors were distributed uniformly and randomly in the cell-ECM interface . Nodes on the top surface of the substrate LSM were selected at random and designated as ligand binding sites until the desired ligand density was achieved ( Figure 1B ) . Since the lattice spacing was 20 nm , the maximum ligand density was 2500 #/µm2 , which is approximately the saturating density for large ECM proteins , such as fibronectin or collagen , absorbed on flat substrates , such as tissue culture plastic or glass slides [59] , [60] . Integrin receptors were placed randomly on the bottom surface of the membrane/cortex plate , but unlike the ECM ligands , the positions of free integrins ( not bound to ligand ) were not limited to sites of LSM nodes . Three integrin states were described in the model that reflect the major conformational states integrins are known to adopt: “inactive” ( low-affinity ) , “active” ( high-affinity ) , and ligand occupied [61] . Although inactive integrins can bind soluble ligands in in vitro binding assays [62] , [63] , when locked in the inactive conformation through molecular engineering and expressed on the cell surface , integrins ( αIIbβ3 and αvβ3 ) do not bind tethered ligands [64] . We thus made the assumption that only active integrins can bind ligand to reflect the relatively low probability of bond formation between matrix-tethered ligands and inactive integrins on the cell surface . Four integrin reactions were therefore modeled in our simulation: activation of inactive integrins , deactivation of active integrins , bond formation between active integrin and ligand , and bond dissociation . In addition , integrin “hop” reactions were included to describe the diffusive movements of unbound integrins . The conversion between active and inactive integrin states was described by simple transition rates , ka and ki , which describe , respectively , the rate of conformational change from the inactive to active and active to inactive states . In the cell , the dynamic equilibrium between active and inactive integrin states depends on a variety of factors , including divalent cations , cell signaling , and intracellular integrin binding partners such as talin . ka and ki in this model can be viewed as phenomenological parameters that take all these influences into account . The distance-dependent rates of bond formation and rupture were calculated according to the equations formulated by Bell and co-workers [26] , [47] . As mentioned , integrin-ligand bonds were modeled as Hookean springs . For such a bond , the reverse reaction rate in the Bell model takes the form of: ( 6 ) where is the unstressed intrinsic dissociation rate , F is the force on the bond , and γ is an empirically measured quantity with units of length describing the bond's sensitivity to force [26] , [28] . The bond force was calculated using Hooke's law based on the equilibrium extension of the bond as determined by Equation 5 . Since the force on each bond could be unique , an individual rupture rate for each bond was calculated . Association rates were calculated for each active integrin and ligand in close proximity . Integrin ligand binding partners that were separated by a lateral cutoff distance greater than 10 nm in the xy-plane were assigned an association rate of exactly zero . For pairs that were within the cutoff , the bond formation rate directly followed from the Boltzmann distribution for affinity [65] and was given by: ( 7 ) where is the unstressed intrinsic association rate and ΔE is the minimum mechanical potential energy change resulting from bond formation [47] . To calculate F and ΔE for a specific pair of binding partners , a bond spring was temporally connected to the desired ligand site , and the system was relaxed to equilibrium . ΔE was then calculated according to Equation 4 and F was determined by Hooke's law . Diffusion of unbound integrins ( inactive and active unbound ) was modeled by hop reactions in the plane of the membrane ( bottom surface of LSM plate ) . As originally proposed by Elf and Ehrenberg [66] , the hops were of discrete length Δℓ and occurred along the four directions defined by the positive and negative x- and y- axes . The rate for a specific integrin to undergo a hop reaction was given by: ( 8 ) where D is the diffusion coefficient for integrins in the membrane ( 10−10 cm2/s ) . The length Δℓ that we used in the simulation is 5 nm , which is on the order of the diameter of the integrin molecule . Periodic boundaries were employed for receptor diffusion to limit finite-size effects . The time evolution of the system was simulated by kinetic Monte Carlo according to the Gillespie algorithm [67] . For a given chemical and mechanical state of the system , the Gillespie algorithm determined the reaction that occurred next and the time that elapsed until that reaction occurred . Reactions were selected through random number sampling of a probability distribution constructed based on the kinetic rates of all possible reactions . The system ultimately was evolved through an iterative process of calculating the reaction rates for the current system state , selecting the next reaction , executing the reaction , updating the rates , and repeating ( Figure 2 ) . To determine the next reaction and the variable time step once the reaction rates were calculated , two random numbers ran1 and ran2 were generated from a uniform probability distribution between 0 and 1 . The next reaction , μ , was selected according to: ( 9 ) where is the rate constant for a particular reaction involving a specific integrin and the summation i is over all possible reactions . The time that elapsed between the last reaction and the newly selected reaction was given by: ( 10 ) After the next reaction and τ were determined , the selected reaction was executed . Either an integrin was moved by Δℓ in a randomly selected direction ( hop reaction ) , the activity state of the integrin was flipped ( activation or deactivation reaction ) , or a bond was incorporated or removed from the LSM at the appropriate ligand site ( bond formation or dissociation ) . The simulation time was then incremented by τ , the spring network was relaxed back to mechanical equilibrium using Equation 5 , and the reaction rates were again calculated . This procedure was repeated until the desired simulation time elapsed ( Figure 2 ) . The mechanical energy minimization defined by Equation 5 was computationally expensive . Two main optimizations were thus implemented to reduce the frequency of calls to the minimization algorithm and increase its efficiency . First , energy minimums were stored to memory upon calculation to avoid repeatedly minimizing the same configuration of integrin bonds . Second , smaller sub-systems of springs and nodes were minimized , as opposed to the entire spring network . Since strain induced by an integrin-ligand bond vanished with sufficient distance from the bond , the total system potential energy minimum could be computed by minimizing a smaller sub-region surrounding the bond . The distance from a bond at which the strain vanished depended on the physical parameters defining the system , and hence sub-system size was optimized for a particular set of matrix , membrane/cortex , glycocalyx , and bond parameters . Typical sub-systems ranged from 400–800 nm in dimension . For minimization with Equation 5 , nodes at the boundary of a sub-system were constrained to their current location to implement the vanishing strain boundary condition . A small number of simulations were executed on rigid matrix substrates that had reaction interfaces spanning a membrane area greater than 1 . 4 µm×1 . 4 µm . Solutions for these larger systems were obtained by approximating the distance-dependent rate of integrin-ligand bond formation . Specifically , the minimum change in system potential energy , ΔE , and the equilibrium force on the integrin ligand bond , F , necessary to compute the bond formation rate were estimated based on the equilibrium separation distance between the unbound integrin and ligand . This approximation avoided the necessity of repeatedly minimizing the system energy to calculate bond formation rates . In order to estimate ΔE and F , the dependence of ΔE and F on equilibrium integrin-ligand separation distance was first determined . To do so , integrin-ligand bonds were randomly and sequentially added to a model cell-ECM interface . For each integrin-ligand bond added , the equilibrium integrin-ligand separation distance before binding and the equilibrium bond force and change in system potential energy after binding were recorded . Plots of F versus initial separation distance and ΔE versus initial separation distance squared were each well-fit by quadratic equations ( Figure S1 ) , at least over the range of physical parameters utilized in this work . Because bonds were added to the system randomly , the relationships did not depend on a specific configuration of bonds . The relationships were dependent , however , on the model's physical parameters , and thus force and energy relationships were determined for each combination of physical parameters examined . During simulation of integrin dynamics , the curve-fits were used to estimate ΔE and F as a function of equilibrium integrin-ligand separation distance to calculate the bond formation rates . For best-estimate system parameters ( See below ) , the average errors in approximating ΔE and F based on curve fits were 6 . 3% and 3 . 5% , respectively , corresponding to an average error in bond formation rate of 3 . 7% according to Equation 7 . Results from simulations of integrin dynamics with estimated ΔE and F were not statistically different from those in which rates were calculated by directly minimizing the system energy ( Figure S2 ) . Table 1 lists the parameters that were used in the simulations unless otherwise noted . The dynamic integrin parameters were based on those reported for fibronectin and the α5β1 integrin . Other possible integrin parameters , however , were also considered to extend the relevance of the model results to other types of integrins and cell surface receptors . For α5β1 , the kinetic rates of the integrin-ligand interaction , the force-dependence of the interaction , the mobility of the integrin in the membrane , and the density of integrin on the cell surface have been reported [62] , [63] , [68]–[70] . The rates of integrin activation and deactivation , however , have not been measured experimentally . Based on experimental reports of the equilibrium distribution of inactive and active integrins on the cell surface [30] , [64] , the free energy of conformational change was approximated to be 2–3 kbT [44] . Considerations of molecular diffusion rates provide an upper limit of ∼1 s for the large structural movement that occurs during activation [71] . We thus used estimates of 0 . 5 s−1 and 5 s−1 for the activation and deactivation rates , respectively , although other possibilities were evaluated . The springs comprising the membrane/cortex plate were assigned a Hookean constant that achieved the experimentally measured flexural rigidity ( bending modulus ) of the actin cortex , 1×10−19 N·m [72] , [73] . The Hookean constant was related to the flexural rigidity , I , of the plate by: ( 11 ) where h is the thickness of the plate . The Hookean constants of the ECM substrate springs were varied according to Equation 1 to achieve elastic moduli in the physiological range of 102–105 Pa [13] , [74] . Since cellular experiments are typically conducted on extremely rigid non-deformable glass or plastic substrates , we also constructed non-deformable substrates in our model by assigning an arbitrarily large spring constant of 1000 pN/nm . This approximates a substrate with a Young's modulus of roughly 0 . 1 GPa ( for comparison , glass or tissue culture plastic is ∼1 GPa [13] ) . The thickness of the glycocalyx is reported to be approximately 40–50 nm [52] and up to 100 nm for certain cell types such as endothelial cells [53] . In this model , a best estimate of 43 nm was used for the glycocalyx spring length , i . e . its thickness , but other values were considered . While the stiffness of the glycocalyx has not yet been measured directly , estimates are available . Agrawal and Radhakrishnan estimated the glycocalyx stiffness by fitting simulations of nano-particle adhesion on the cell surface to analogous experimental data . Based on these results we estimated σg to be 0 . 02 pN/nm [56] . This estimate is in good agreement with purely theoretical estimates calculated by considering the statistical mechanics of chain molecules anchored to a surface [47] . Like the glycocalyx thickness , additional possibilities for σg were explored . To analyze the extent of integrin clustering , a two-dimensional point pattern analysis of the integrin membrane positions projected onto the xy-plane was constructed . The analysis was performed using Ripley's K-function [75] , [76] , which measures the extent to which a point pattern deviates from a random Poisson distribution and is given by: ( 12 ) where the summations i and j are over all integrin point positions , A is the projected area of the membrane , s is the sampling radius , and Wij ( s ) is exactly equal to one if the distance between points i and j is less than s and zero otherwise . Periodic boundaries were utilized in the calculation of Wij ( s ) . To facilitate the interpretation of the statistic , these data were transformed [77] into the following form: ( 13 ) For a point pattern with complete spatial randomness , R ( s ) has an expected value of zero , and if the points are clustered R ( s ) has a positive value . The maximal value of R ( s ) and the radius at which the function is maximal provide a measure of the degree of clustering and the cluster size respectively . To analyze the degree of cooperativity in integrin-ligand binding interactions , Hill plots of the steady state bond fraction versus ligand density were constructed . The plots were fit to a version of the Hill equation that also accounts for the possibility of ligand depletion: ( 14 ) where U is the bond fraction , R is the total integrin receptor density , L is the total ligand density , Kd is the bond dissociation constant and nHill is the Hill coefficient . The model was fit to the Hill plots for both Kd and nHill with non-linear least squares regression . Integrin-ligand binding rates are dependent on the distance the molecules must stretch to reach each other . By inducing mechanical deformations , adhesive bond formation could modify these distances and therefore be cooperative due to mechanical coupling . In order to determine how the cell membrane/cortex deforms during binding , we calculated the equilibrium deformations that were induced by the addition of integrin-ligand bonds into our mechanical model of the cell , glycocalyx , and matrix . A single integrin bond between the cell and a rigid ECM substrate caused a highly localized deformation that extended laterally in the plane of the membrane approximately 150 nm from the bound site ( Figure 3 ) . The placement of additional bonds in the deformed region pulled a significantly larger area of the cell into closer proximity with the matrix substrate ( Figure 3 ) . We thus imagined that new bond formation would be most favorable nearby existing bonds , since the distance between integrins and ligands would be reduced in this area , and that bond formation would become increasingly favorable as additional bonds accumulated together and induced larger deformations . To test if bond formation was indeed cooperative , we ran simulations of integrin dynamics on rigid ECM substrates of varying ligand density and constructed Hill plots of the steady-state bond fraction ( Figure 4A ) . Since the thickness of the glycocalyx determines the initial distance between integrin and ligand partners , the effective thickness of the glycocalyx ( lg−lb ) was also varied in an attempt to manipulate cooperativity . Hill plots were constructed from these simulation results and were fit to a form of the Hill equation which accounts for low ligand density ( Equation 14 ) . The best-fit Hill coefficients were greater than one , indicating cooperative integrin binding , and increased with enhanced glycocalyx thickness ( Figure 4B ) . Cooperative integrin-ligand interactions resulted in a clustered pattern of integrin bonds , as can be seen Figure 4E , which shows integrin positions after 30 minutes of simulation on rigid substrates ( L - 2500 #/µm2 ) . With increasing glycocalyx thickness and hence more cooperative integrin-ligand interactions , integrin clusters became fewer in number , larger in size , and more densely packed with integrins ( Figure 4E ) . To quantify the extent of clustering , we preformed a point-pattern analysis on the steady-state integrin positions by calculating the maximum of the transformed Ripley K-function , R ( s ) . Maximum values greater than zero indicate that the integrins are clustered and the magnitude of the value is related to the degree of integrin clustering . Our point-pattern analysis demonstrated that the degree of clustering increased with enhanced glycocalyx thickness ( Figure 4C ) and was proportional to the level of cooperativity , as indicated by the Hill coefficient ( Figure 4D ) . Kinetically , integrin clusters typically formed within tens of seconds to minutes of simulated time . Figure 5 shows the chemo-mechanical evolution of the integrin system for best-estimate parameters ( Table 1 ) on a rigid matrix ( L - 2 , 500 #/µm2 ) . As Figure 5 demonstrates , new bonds formed rapidly in regions of the cell-ECM interface deformed by prior bonds and formed slowly in regions devoid of bonds . The bonds began to form after approximately a ten second delay , at which point the rate of bond formation accelerated until saturation was reached after approximately 50 seconds ( Figure 6A ) . The statistical measure of integrin clustering , max R ( s ) , exhibited a similar kinetic profile to that of the bond fraction , indicating that clustering was primarily driven by bond formation ( Figure 6B ) . While integrin clustering was primarily driven by the initial binding of integrins to the matrix , integrins continued to condense in the clusters over a much slower time-scale due to bond rearrangements occurring through repeated cycles of bond breakage and reformation ( See white arrows – Figure 5B; See also the slow upward rise in Ripley statistic – Figure 6B ) . Hence , integrin clustering was biphasic and characterized by an initial fast bond formation and clustering step , followed by a slow bond rearrangement and condensing process . Since integrin clustering required both integrin-ligand adhesion and cell-ECM repulsion , we mapped the relationship between integrin-ligand affinity and glycocalyx-mediated cell-ECM repulsion . To do so , we ran simulations in which the glycocalyx-thickness and chemical-affinity parameter space was systematically varied . As shown in Figure 7 , clustering depended strongly on receptor-ligand affinity and the effective thickness of the glycocalyx . For high-affinity interactions but relatively thin glycocalyxes , the majority of integrin receptors were bound but not clustered ( Compare Figure 7A and 7B ) . When the receptor-ligand affinity was relatively low , integrin receptors were neither bound nor clustered . If the glycocalyx was relatively thick compared to the integrin bond length and the receptor-ligand interaction was of sufficient affinity , however , integrins bound ligand and assembled into clusters . Integrin clustering was particularly sensitive to variations in glycocalyx thickness and bond length , as small changes of five to ten nanometers in the effective thickness of the glycocalyx could switch the integrin system from clustered to unclustered or vice versa . While integrins are able to switch between activity states , this property was not essential for integrin clustering . For best estimate parameters , integrins clustered if they were constitutively maintained in the active , ligand-binding conformation or instead were allowed to switch between inactive and active states ( data not shown ) . Receptor density was modified to control the number of active receptors available for binding . Although clusters were smaller and less frequent in number for lower initial densities ( Figure S4 ) , integrins generally clustered when the initial receptor density was high or low . Increased integrin bond stiffness , however , generally enhanced clustering ( Figure 8 and S3 ) . Integrin clustering in our simulations thus was controlled by integrin bond length ( Figure 7 , 8 , and S3 ) , bond stiffness ( Figure 8 and S3 ) , and affinity for ligand ( Figure 7 ) , which all depend or are predicted to depend on integrin activation state . These results suggest a functional link between integrin conformation , glycocalyx properties , and integrin clustering . For physiologically-relevant parameters , integrins clustered even when the total initial receptor density was reduced by a factor of ten ( Figure S4 ) . These results indicate that integrins could still cluster in the presence of soluble ligand , which would effectively reduce the number of available receptors to bind matrix-tethered ligands , if the soluble ligand concentration was non-saturating . Experimentally , small nanometer differences in average ligand-ligand spacing were shown to dictate the strength of cell adhesion and whether or not integrins cluster [8] , [15] , [16] . In our model , we found that the cellular deformations induced by bond formation , which are required for cooperativity , extend laterally only a limited distance ( ∼150 nm ) from the bond in the plane of the membrane ( Figure 3 ) , and that this distance is on the order of the maximum ligand spacing reported to support integrin clustering ( ∼73 nm; [16] ) . Since integrins may not be able to utilize cooperative binding to cluster if ligands are spaced too sparsely , we sought to determine the relationship between integrin clustering and ligand spacing and how this relationship was controlled by the mechanics of the cell and glycocalyx . We first tested how mechanical parameters , including the glycocalyx stiffness , glycocalyx thickness , and membrane/cortical rigidity , affected the lateral width of the cell deformation induced by bond formation . We observed that varying the glycocalyx thickness over a physiological range of possibilities impacted the magnitude of the z-direction height of membrane/cortex deformation above the substrate , but only had a minimal impact on the xy-width of the deformation ( data not shown ) . The ratio of the glycocalyx stiffness ( σg ) to the membrane/cortex stiffness ( σm ) , however , did influence the deformation width , as a decrease in σg/σm was associated with a larger in-plane deformation of the membrane/cortex plate ( Figure 9A ) . We next ran integrin simulations on rigid ECM substrates while varying either glycocalyx thickness or the glycocalyx to membrane/cortex stiffness ratio . For best-estimate σg/σm ( Table 1 ) , a threshold ligand density approximately of 200 #/µm2 was required for integrin clustering regardless of glycocalyx thickness ( Figure 9C ) . This value corresponds to an average intermolecular ligand spacing of 71 nm . Manipulating σg/σm , though , altered the minimal ligand density necessary to support clustering . As suggested by the cell deformations ( Figure 9A ) , enhancing the stiffness ratio shifted the minimal ligand density to higher values ( Figure 9D ) . Glycocalyx stiffness also influenced the characteristics of integrin clusters . Similar to increasing glycocalyx thickness , increasing glycocalyx stiffness resulted in enhanced integrin-binding cooperativity , more extensive integrin clustering , and the formation of more tightly-packed clusters of integrin ( Figure 9B and S4 ) . To test the effect of matrix stiffness on the formation of adhesive bonds and on integrin clustering , we ran dynamic integrin simulations on ECM substrates of varying stiffness . Hill plots of the simulation results were constructed ( Figure 10A ) and fit to the Hill equation accounting for ligand depletion ( Equation 14 ) . We found that an ECM substrate with a Young's modulus of at least 2000 Pa was required to support cooperative integrin binding ( Figure 10B ) . More compliant substrates failed to promote cooperative binding because the highly flexible ligands facilitated fast rates of association between integrin and ligand regardless of position in relation to other bonds . For substrates stiffer than 2000 Pa , the Hill coefficients for integrin binding increased nearly linearly with the logarithm of the substrate stiffness until reaching a plateau at approximately 100 , 000 Pa ( Figure 10B ) . The extent of integrin clustering , max R ( s ) , was correlated with the observed Hill coefficients ( Figure 10C ) , indicating that substrate rigidity controls integrin binding cooperative and clustering . These results suggest one possible mechanism of how integrins could “sense” matrix rigidity . In this work , we built a new model to study integrin adhesion and clustering that couples the chemistry of bond formation with the mechanics of a composite , layered material representing the cell membrane/cortex , glycocalyx , and ECM . The biology incorporated into the model was basic and included only integrin activation/deactivation and association/dissociation reactions . Despite the simplicity of the molecular interactions , when coupled to the mechanics of the system , our model exhibited complex integrin adhesion behaviors that match those reported in the experimental literature . These behaviors can be explained by one simple principle: when deformations in the cell membrane or ECM accompany bond formation , the distance-dependent kinetic rates for other potential integrin-ligand binding interactions are modified . In essence , integrin bonds pull the cell membrane and ECM substrate into closer proximity and new bonds form more readily in these deformed regions . We showed that for realistic model parameters , clustering was sensitive to both the physical and chemical properties of the matrix , suggesting a simple yet efficient mechanism by which integrin adhesions sense matrix properties . Integrin clustering in our model was driven by the interplay between integrin-mediated adhesion and glycocalyx-mediated cell-ECM repulsion . While a relationship between integrin-ligand affinity and integrin clustering has been suggested [10] , [80]–[82] , we now show that the thickness and stiffness of the glycocalyx may regulate this relationship . Indeed , we found that manipulating glycocalyx thickness/stiffness parameters while maintaining the intrinsic integrin-ligand affinity can switch the integrin system from an unclustered state to a clustered state or vice-versa . Similarly , changes in integrin-ligand affinity could also induce a switch in integrin clustering state depending on glycocalyx parameters . Furthermore , changes in integrin length , such as the structural extension that occurs during activation , could change the effective thickness of the glycocalyx to also modulate integrin clustering . In general , high-affinity integrin-ligand interactions in the context of a relatively thick or stiff glycocalyx promoted integrin clustering . A glycocalyx too thick or rigid , however , impeded bond formation and thereby prevented clustering . These results suggest the glycocalyx is a potent regulator of integrin system behavior and signaling . Such a relationship may be extremely important in diseases such as breast cancer , in which 95% of the cells have modified glycocalyx composition or structure and in which integrin clustering is functionally-linked to loss of tissue homeostasis and the development of a malignant phenotype [13] , [83] . Our model provides one explanation for the exquisite sensitivity integrins exhibit in response to variations in matrix-ligand density [8] , [15] , [16] . In cellular experiments on rigid ligand-coated substrates , integrins cluster when the average intermolecular ligand spacing is less than or equal to 58 nm , but not when it is greater than or equal to 73 nm [16] . These results have fueled the notion that cells posses molecular “rulers” that mediate this chemo-sensory process . Our model suggests that the ruler might actually be the cell membrane and associated actin cortex rather than a specific molecule , such as an adhesion plaque protein . In order for integrins to cluster , we found that the average spacing between ligand molecules had to be less than the lateral width of the membrane/cortex deformation induced by an integrin bond . If the deformation was too small relative to the ligand spacing , integrin-ligand binding was not cooperative and integrins did not cluster . For our best-estimate mechanical parameters , the width of cell deformation induced by an integrin bond ( 150 nm ) was on the order of the experimentally-measured ligand spacing at which the unclustered-to-clustered integrin transition occurs experimentally . Moreover , when best-estimate mechanical parameters were utilized in simulations of integrin dynamics , we found that an average intermolecular ligand spacing of 71 nm was necessary to drive integrin clustering in our model , which is in excellent agreement with experimental results . The width of the cell surface deformation was primarily determined by the ratio of the glycocalyx stiffness to membrane/cortex thickness , and hence this ratio controlled the threshold ligand density required for integrin clustering . We thus propose that the integrin adhesion system may be intrinsically sensitive to ligand density and that this sensitivity may be tuned by the mechanical properties of the cell and glycocalyx . We also found that integrin clustering was responsive to matrix stiffness . On progressively more compliant substrates , the rate of integrin-ligand bond formation was increasingly fast due to the enhanced flexibility of the ligand binding site . Consequently bond formation was not cooperative on highly compliant substrates , since new bonds could readily form in the interface regardless of proximity to pre-existing bonds . After evaluating a range of matrix stiffnesses , we determined that integrin clustering in our model requires a substrate with a Young's modulus of at least 2000 Pa , at which point the extent of clustering increases with the logarithm of substrate stiffness until maximum clustering is achieved around 100 , 000 Pa . These results agree well with cellular experiments conducted on ECM-functionalized hydrogels of tunable stiffness , on which integrins assemble into larger and more numerous adhesions on matrices above 1000 Pa [13] , [14] , [84] . Furthermore , cell behaviors correlated with integrin clustering , such as cell spreading , demonstrate an incremental response to increases in matrix stiffness between approximately 1000 Pa to 50 , 000 Pa , which is again in agreement with the sensitivity range for integrin clustering predicted in this work [13] , [84] . While integrin-mediated matrix mechano-sensing has been assumed to require actomyosin contractility to generate matrix probing forces and adhesion plaque proteins to respond to these force ( reviewed in [21] , [85] ) , our model would suggest that integrin themselves can respond to matrix stiffness in one manner independent of myosin or plaque proteins . Experimentally-observed features of integrin clustering , such as its sensitivity to matrix properties , were recapitulated in our model without the incorporation of cytoskeletal adaptor proteins into the model . Indeed , for best-estimate parameters , the kinetic profiles of integrin bond formation and clustering simulated by our model recapitulate the short delay in integrin bond formation observed experimentally when the cell first contacts the ECM , as well as the fast rate of de novo integrin adhesion assembly and clustering observed in cells [10] , [31] , [50] , [78] , [79] . This does not suggest , however , that cytoskeletal interactions are insignificant . Many lines of experimental evidence clearly demonstrate that cytoskeletal interactions regulate the size and signaling activity of integrin adhesion structures ( reviewed in [86] ) . We envision that integrin-cytoskeletal interactions could synergize with the mechanically-coupled integrin-ligand interactions described in this work to drive a more robust integrin clustering response with heightened sensitivity to matrix properties or with additional levels of regulation . Our model , however , does offer an explanation for how integrins can cluster prior to recruiting cytoskeletal adaptor proteins , as has been observed in time lapse studies of adhesion complex assembly [10] , [31] , [50] . Similar to the kinetics of integrin assembly in these time-lapse studies , integrins in our model spontaneously clustered on rigid substrates in tens of seconds to minutes even though cytoskeletal interactions were not included in the model . Provocatively , since clustering was sensitive to matrix properties , our results suggest that integrins may begin to sense matrix properties prior to the assembly of more advanced adhesion structures , such as focal complexes and focal adhesions [86] . It is well-documented that force promotes integrin adhesion complex assembly , which raises the question of whether cytoskeletal forces would influence the myosin-independent integrin clustering described in this work . In our model , integrins cluster because one bond pays a portion of the energy penalty associated with compressing the glycocalyx for the next integrin to complex a nearby ligand site . Myosin-driven forces on integrin bonds could actively pull the cell and ECM into closer spatial proximity , and hence pay this energy penalty [21] . In the context of our integrin clustering model , these force-driven deformations should enhance integrin bond formation and aggregation to possibly achieve states of integrin cluster size or density that would otherwise be unlikely . Similarly , exogenously applied forces to the cell , such as fluid shear forces in the vasculature , could induce deformations in the cell-ECM interface that modify integrin clustering response . Therefore , both endogenous contractile forces and exogenous applied forces could influence integrin distribution through a mechanism similar to that proposed in this work . Many aspects of integrin clustering described by our model are justified experimentally . For example , reports have demonstrated that receptor-ligand interactions are distance-dependent [27] , [28] and that the cell and ECM are in closest proximity at sites containing integrin adhesions [87]–[89] . Perhaps some of the best support of the model is provided by studies with biomimetic lipid vesicles . When lipid vesicles functionalized with adhesion molecules and a repulsive brush border are brought in contact with a complimentary solid surface , receptor-ligand bonds cluster despite the simple chemistry of the vesicle system [90] , [91] . Since the repulsive brush border is required for patterned bond formation , these studies suggest that adhesive bond clustering results from the interplay between adhesion and repulsion , as our model predicts . Several novel predictions stemming from our model , however , must still be validated experimentally . This includes determining if matrix rigidity controls integrin clustering by altering kinetic rates of bond formation , evaluating if cell and glycocalyx stiffness controls the relationship between integrin clustering and ligand density , and determining if the glycocalyx is indeed a potent regulator of integrin function and clustering . Testing these predictions should provide significant insight into how cell adhesions sense and respond to their ECM environment . In conclusion , we showed how the coupling between the chemistry of bond formation and the mechanics of the cell and glycocalyx may drive integrin clustering in a matrix-dependent manner . Our results suggest a mechanism by which integrins function as sensors of matrix rigidity and chemistry .	Critical cell decisions , including whether to live , proliferate , or assemble into tissue structures , are directed by cues from the extracellular matrix , the external protein scaffold that surrounds cells . Integrin receptors on the cell surface bind to the extracellular matrix and cluster into complexes that translate matrix cues into the set of instructions a cell follows . Using a newly developed model of the cell-matrix interface , in this work we detail a simple yet efficient mechanism by which integrins could “sense” important matrix properties , including chemical composition and mechanical stiffness , and cluster appropriately . This mechanism relies on mechanical resistance to integrin-matrix interaction provided by the glycocalyx , the slimy sugar and protein coating on the cell , as well as the stiffness of the matrix and the cell itself . In general , the resistance alters integrin-ligand reaction rates , such that integrin clustering is favored for many physiologically relevant conditions . Interestingly , the mechanical properties of the cell and ECM are altered in many prevalent diseases , such as cancer , and our work suggests how these mechanical perturbations might adversely influence integrin function .	[ "Abstract", "Introduction", "Model", "Results", "Discussion" ]	[ "cell", "biology/cell", "signaling", "biophysics/theory", "and", "simulation", "cell", "biology/extra-cellular", "matrix", "computational", "biology", "cell", "biology/cell", "adhesion" ]	2009	Integrin Clustering Is Driven by Mechanical Resistance from the Glycocalyx and the Substrate
Specification of the dorsal-ventral axis in the highly regulative sea urchin embryo critically relies on the zygotic expression of nodal , but whether maternal factors provide the initial spatial cue to orient this axis is not known . Although redox gradients have been proposed to entrain the dorsal-ventral axis by acting upstream of nodal , manipulating the activity of redox gradients only has modest consequences , suggesting that other factors are responsible for orienting nodal expression and defining the dorsal-ventral axis . Here we uncover the function of Panda , a maternally provided transforming growth factor beta ( TGF-β ) ligand that requires the activin receptor-like kinases ( Alk ) Alk3/6 and Alk1/2 receptors to break the radial symmetry of the embryo and orient the dorsal-ventral axis by restricting nodal expression . We found that the double inhibition of the bone morphogenetic protein ( BMP ) type I receptors Alk3/6 and Alk1/2 causes a phenotype dramatically more severe than the BMP2/4 loss-of-function phenotype , leading to extreme ventralization of the embryo through massive ectopic expression of nodal , suggesting that an unidentified signal acting through BMP type I receptors cooperates with BMP2/4 to restrict nodal expression . We identified this ligand as the product of maternal Panda mRNA . Double inactivation of panda and bmp2/4 led to extreme ventralization , mimicking the phenotype caused by inactivation of the two BMP receptors . Inhibition of maternal panda mRNA translation disrupted the early spatial restriction of nodal , leading to persistent massive ectopic expression of nodal on the dorsal side despite the presence of Lefty . Phylogenetic analysis indicates that Panda is not a prototypical BMP ligand but a member of a subfamily of TGF-β distantly related to Inhibins , Lefty , and TGF-β that includes Maverick from Drosophila and GDF15 from vertebrates . Indeed , overexpression of Panda does not appear to directly or strongly activate phosphoSmad1/5/8 signaling , suggesting that although this TGF-β may require Alk1/2 and/or Alk3/6 to antagonize nodal expression , it may do so by sequestering a factor essential for Nodal signaling , by activating a non-Smad pathway downstream of the type I receptors , or by activating extremely low levels of pSmad1/5/8 . We provide evidence that , although panda mRNA is broadly distributed in the early embryo , local expression of panda mRNA efficiently orients the dorsal-ventral axis and that Panda activity is required locally in the early embryo to specify this axis . Taken together , these findings demonstrate that maternal panda mRNA is both necessary and sufficient to orient the dorsal-ventral axis . These results therefore provide evidence that in the highly regulative sea urchin embryo , the activity of spatially restricted maternal factors regulates patterning along the dorsal-ventral axis . In bilaterians , specification of the dorsal-ventral ( D/V ) axis is a crucial event during embryogenesis to establish the correct body plan . In many species , this process relies on gene products translated from maternal mRNAs deposited in the egg . For example , in Drosophila , specification of the D/V axis of the embryo is initiated by the product of the gurken gene , which is active in the oocyte nucleus during oogenesis and encodes a member of the epidermal growth factor ( EGF ) superfamily that acts as a secreted dorsalizing signal [1–4] . Similarly , in Xenopus and zebrafish , although the D/V axis is not preformed in the unfertilized egg , dorsal determinants are localized to the vegetal pole of the egg [5–8] . Fertilization breaks the radial symmetry of the egg and triggers the asymmetric transport of these determinants from the vegetal pole to the future dorsal side where they activate the canonical Wnt pathway [9 , 10] . While maternal information is clearly important for specification of the D/V axis in a number of species , in contrast , there is very little evidence for the presence of maternal determinants of axis formation in the oocyte of mammals , consistent with the idea that the embryonic axes are specified entirely by cell interactions [11] . Accordingly , it has been argued that the regulative abilities of the first blastomeres of the mouse embryo rule out the possibility that maternal determinants influence axis specification [12] ( reviewed in [13] ) . The sea urchin embryo is well known for its extraordinary developmental plasticity [14] . In a now classical blastomere dissociation experiment , Driesch showed that dissociated blastomeres of the four-cell stage embryo have the potentiality to reestablish a D/V axis [15] . The outcome of this experiment not only demonstrated the impressive regulative ability of the early blastomeres of the sea urchin embryo but also strongly influenced ideas about how D/V patterning is established in this organism . By showing that the D/V axis is very easily respecified , it encouraged the view that there are no determinants for D/V axis formation in echinoderm embryos . On the other hand , egg bisection experiments performed by Hörstadius showed that differences in the fates of presumptive ventral and dorsal regions can be traced back to the egg , consistent with the idea that the oocyte already has a bilateral organization [16] . If there are maternal cues that influence D/V axis formation in this embryo , what could they be ? There is a large body of evidence correlating formation of the D/V axis with the activity of redox gradients and with the asymmetric distribution of mitochondria in the unfertilized sea urchin egg . Classical experiments performed by Child , Pease , and Czihak in the thirties and sixties showed that it is possible to bias the D/V axis by treating embryos with steep gradients of respiratory inhibitors and that the activity of the mitochondrial enzyme cytochrome oxidase can predict the D/V axis as early as the eight-cell stage , with the presumptive ventral side being more oxidizing than the dorsal side [17–20] . This asymmetry of mitochondria activity is the first known manifestation of D/V polarity . Several recent studies by Coffman and colleagues addressed the question of causality between this early asymmetry and the orientation of the D/V axis [21–23] . Although these studies provided evidence that the D/V axis can be entrained by centrifugation , by microinjection of purified mitochondria , or by overexpressing a form of catalase targeted to the mitochondria , the correlations obtained remained modest , and in no case were these perturbations shown to efficiently orient the D/V axis [21–23] . Furthermore , perturbations that were expected to influence D/V axis formation , such as overexpression of a mitochondrially targeted form of superoxide dismutase , which generates the strong oxidizing component H2O2 and that would be predicted to efficiently orient the axis , did not show any effect on the orientation of the D/V axis . Therefore , the redox gradient model of D/V axis formation clearly needs further experimental validation , and the biological significance of the early asymmetry of mitochondria and of redox gradients and their relation to the D/V axis remains largely unclear . At the molecular level , the earliest sign of specification of the D/V axis is the expression of the TGF-β nodal in the presumptive ectoderm at the 32-cell stage . nodal is the first known zygotic gene differentially expressed along the D/V axis , and Nodal signaling orchestrates patterning along the secondary axis first by specifying the ventral ectoderm and second by inducing the expression of BMP2/4 , which acts as a relay to specify the dorsal ectoderm . nodal morphants completely lack D/V polarity , but expression of nodal into one blastomere is capable of completely rescuing D/V polarity in these embryos [24 , 25] . cis-regulatory studies further showed that nodal expression is driven by ubiquitously expressed maternal factors such as the transcription factor SoxB1 and that it requires maternal Wnt/beta catenin signaling as well as signaling by the Vg1/GDF1-related maternal factor Univin [26 , 27] . Intriguingly , nodal is initially expressed very broadly , almost ubiquitously , and then its expression is progressively restricted to a more discrete region of the ectoderm during cleavage . The progressive spatial restriction of nodal expression is thought to rely mostly , if not exclusively , on the ability of Nodal to promote its own expression through an intronic autoregulatory enhancer [26 , 27] and to induce the expression of the long-range Nodal antagonist Lefty . This regulatory mechanism based on the long-range diffusion of the Lefty antagonist fulfills the requirement for a reaction diffusion and is thought to be mainly responsible for amplifying an initial subtle asymmetry , possibly generated by redox gradients , into a robust spatially restricted expression of nodal [23 , 28] . Finally , nodal expression requires the integrity of the p38 pathway [22 , 23 , 26 , 29] . Inhibition of p38 signaling with pharmacological inhibitors abolishes nodal expression . Immunostaining experiments using an anti-phosho p38 and analysis of the spatial distribution of a p38-green fluorescent protein ( GFP ) fusion protein revealed that p38 is first activated ubiquitously and then selectively inactivated on the presumptive dorsal side of the embryo . The signals that regulate p38 in the early embryo are not known , but it has been proposed that redox gradients may be directly responsible for p38 activation [22 , 23 , 26 , 29] . However , direct evidence that p38 mediates the effects of redox gradients is presently lacking , and the transcription factors linking p38 to the machinery that regulates nodal expression are not known . In this paper , we identify a maternal factor that plays a crucial role in D/V axis formation by directing the spatial restriction of nodal expression . First , we discovered that an unidentified TGF-β ligand cooperates with BMP2/4 to restrict nodal expression . We identified this ligand as the product of a maternally expressed TGF-β ligand related to Maverick from Drosophila and GDF15 from vertebrates that we named Panda . Inhibition of maternal panda mRNA translation blocked the early spatial restriction of nodal and caused persistent massive ectopic expression of nodal on the dorsal side despite the presence of Lefty . We further show that while blocking translation of bmp2/4 mRNA alone does not cause ectopic expression of nodal , the double knockdown of panda and bmp2/4 causes an extreme ventralization . We further provide evidence that the panda mRNA is broadly distributed in the early embryo , that local expression of panda mRNA efficiently orients the D/V axis , and that , although panda mRNA is broadly distributed , Panda activity is required locally in the early embryo . Taken together , these findings demonstrate that maternal panda mRNA is required early to restrict the spatial expression of nodal , that it is sufficient to orient the D/V axis when misexpressed , and therefore , that it fulfills the requirements for a maternal factor that specifies the D/V axis . Our results suggest that , although specification of the D/V axis is established by the activity of Nodal in the zygote , maternally provided signaling molecules play crucial roles by antagonizing the activity of Nodal . We showed previously that during D/V patterning in the sea urchin embryo , transduction of the BMP2/4 signals requires the activity of the type-I BMP receptor Alk3/6 , the functional orthologue of Thickveins , which transduces Dpp signals in Drosophila . We noticed , however , that blocking Alk3/6 consistently produced a phenotype much less severe than the BMP2/4 loss-of-function phenotype . For example , while bmp2/4 morphants typically lack a population of immunocytes called pigment cells that requires BMP signaling , alk3/6 morphants always develop with numerous pigments cells ( arrows in Fig 1A ) . This suggested that residual BMP signaling in alk3/6 morphants allows formation of pigment cells and/or that additional BMP type I receptors may contribute to transduction of BMP2/4 signals in the absence of Alk3/6 . Indeed , in addition to alk3/6 , the sea urchin genome contains a second gene encoding a BMP type I receptor named Alk1/2 , which is mostly similar to Alk1 and Alk2 from vertebrates and to Saxophone from Drosophila . Like alk3/6 , alk1/2 is expressed maternally and ubiquitously during the cleavage and blastula stages ( S1 Fig ) . To evaluate the contribution of Alk1/2 in BMP2/4 signaling , we knocked it down with antisense morpholinos . Interestingly , blocking alk1/2 mRNA translation disrupted D/V axis formation and produced a phenotype stronger than that resulting from inhibition of Alk3/6 ( Fig 1A ) . When the alk1/2 morpholino was injected at 1 . 2 mM , most alk1/2 morphants failed to develop their ventral arms and dorsal apex and appeared rounded . Alk1/2 morphants also lacked most pigment cells and developed with an ectopic ciliary band and ectopic spicules on the dorsal side , a phenotype largely identical to the bmp2/4 morphant phenotype . These phenotypes could be suppressed by coinjection of a modified wild-type alk1/2 mRNA immune against the morpholino ( see S1 Fig ) . As shown previously in the case of Alk3/6 and of BMP2/4 , blocking Alk1/2 caused a dramatic expansion of the ciliary band territory at the expense of the dorsal ectoderm , as evidenced by the massive ectopic expression of foxG and onecut on the presumptive dorsal side and the lack of expression of dorsal marker genes such as hox7 ( Fig 1B ) . Unexpectedly , blocking Alk1/2 function , unlike blocking BMP2/4 or Alk3/6 , caused a weak but consistent ventralization , as evidenced by the expression of chordin or foxA that extended to the dorsal side at the gastrula stage ( black arrowheads in Fig 1B ) . Consistent with this ventralization , we found that at blastula stages , embryos injected with high doses of the alk1/2 morpholino displayed a massive ectopic expression of nodal similar to that observed in lefty morphants ( Fig 1C ) . This phenotype , which is not observed in bmp2/4 or alk3/6 morphants , suggests that , in addition to BMP2/4 , Alk1/2 may also be required to transduce an unidentified dorsalizing signal . Finally , consistent with the absence of expression of dorsal marker genes , inhibition of alk1/2 mRNA translation , like inhibition of bmp2/4 or alk3/6 , drastically reduced phospho-Smad1/5/8 signaling in the dorsal ectoderm ( Fig 1D ) . We conclude that Alk1/2 plays a pivotal role in transduction of BMP2/4 in the sea urchin and that the activities of Alk1/2 and Alk3/6 are nonredundant , both being functionally required during D/V patterning to transduce BMP2/4 signals and to activate Smad1/5/8 signaling in the dorsal ectoderm . Furthermore , these results suggest that in addition to BMP2/4 , Alk1/2 may be required for transduction of ( a ) still unidentified signal ( s ) that regulate ( s ) D/V patterning . To further characterize the requirements for Alk1/2 and Alk3/6 in D/V axis patterning , we performed a double knockdown . Our expectations were that the double knockdown of alk3/6 and alk1/2 would produce a phenotype roughly similar to the BMP2/4 loss-of-function phenotype . However , surprisingly , the morphology of the double knockdown embryos was very different from that of the bmp2/4 knockdown . The alk1/2 + alk3/6 morphants were completely radialized and developed with a prominent proboscis in the animal pole region and with an ectopic ciliary band surrounding the vegetal pole region ( Fig 2A , white and black arrowheads , respectively ) . These features are typical of the strongly ventralized phenotype observed in nodal-overexpressing or nickel-treated embryos ( Fig 2A ) . Indeed , molecular analysis revealed that the double inhibition of Alk3/6 and Alk1/2 caused a massive ectopic expression of nodal and of its downstream target genes chordin and foxA in the presumptive ectoderm at the blastula stage , whereas it abolished the expression of the dorsal marker gene hox7 ( Fig 2B and 2C ) . Consistent with the extreme ventralization of the double alk1/2 + alk3/6 morphants , at the gastrula stage , expression of the ciliary band genes foxG and onecut was detected in a belt of cells surrounding the vegetal pole ( black arrowheads in Fig 2B ) , a pattern typically observed in embryos ventralized by nodal overexpression ( Fig 2C ) [30] . Since these results suggest that signaling from these BMP receptors is required to restrict nodal expression , we tested if treatments with recombinant BMP2/4 can antagonize nodal expression . Indeed , treatments with increasing concentrations of recombinant BMP2/4 protein gradually antagonized nodal expression , with low concentrations first causing a typical Nodal loss-of-function phenotype and high concentrations resulting in dorsalization of the ectoderm ( S2 Fig ) [30] . Taken together , these results reveal that specification of the ventral territory is not independent of BMP signaling , as previously thought [25 , 30] . The results suggest instead that , in addition to specifying the dorsal region at the onset of gastrulation , signaling from the two BMP receptors Alk3/6 and Alk1/2 is critically required during or before blastula stages to restrict nodal expression to the ventral side . Importantly , the fact that the bmp2/4 morphant phenotype is considerably weaker than the double alk1/2 + alk3/6 morphant phenotype strongly suggests that an unidentified signal acting through these BMP type I receptors is critically required , in addition to BMP2/4 , for the correct specification of the D/V axis and for the normal restriction of nodal expression . We next attempted to identify the TGF-β ligand acting through Alk3/6 and Alk1/2 that cooperates with BMP2/4 and that restricts the early expression of nodal during D/V axis formation . In other species , BMP ligands of the BMP5/8 and anti-dorsalizing morphogenetic protein ( ADMP ) subfamilies have been shown to cooperate and to act redundantly with BMP2/4 factors during D/V patterning . For example , in Xenopus , while the single knockdown of either ADMP , BMP2 , BMP4 , or BMP7 resulted in partial central nervous system ( CNS ) expansion , the quadruple knockdown of BMP2 , 4 , 7 and ADMP caused full radialization and ubiquitous neural induction [31 , 32] . We therefore tested if in the sea urchin , like in Xenopus , members of the BMP5/8 and ADMP subfamilies of TGF-β ligands cooperate with BMP2/4 during D/V patterning . The simple knockdown of BMP5/8 caused a phenotype much weaker than the phenotype caused by blocking BMP2/4 . Surprisingly , the double knockdown of BMP2/4 and BMP5/8 only slightly increased the severity of the BMP2/4 morphant phenotype ( Fig 3A and S3 Fig ) . Similarly , the double knockdown of BMP2/4 and ADMP did not cause a phenotype dramatically more severe than the BMP2/4 morphant phenotype . Even more surprising , the triple knockdown of BMP2/4 , BMP5/8 , and ADMP did not increase significantly the severity of the BMP2/4 morphant phenotype and did not result in ventralized embryos , suggesting that in the sea urchin , BMP5/8 and ADMP do not act redundantly with BMP2/4 to regulate the spatial restriction of Nodal ( S3 Fig ) . We therefore extended our search for TGF-β ligands that would cooperate with BMP2/4 during D/V patterning to other members of the TGF-β superfamily . In addition to members of the BMP subfamily such as bmp2/4 , bmp5/8 , and admp , the sea urchin genome contains several genes encoding TGF-β ligands structurally related to Activins including TGF-β sensu stricto , Activin as well as SPU_018248 , a less well-characterized gene related to Maverick from Drosophila that we renamed panda ( paracentrotus anti-nodal dorsal activity ) ( see below ) in this study [33] . Blocking Activin or TGF-β sensu stricto did not perturb establishment of the D/V axis , making unlikely the possibility that these factors cooperate with BMP2/4 to restrict nodal expression [34] ( our unpublished results ) . In contrast , blocking translation of the TGF-β Panda strongly affected D/V polarity . While the triple knockdown of bmp2/4 , bmp5/8 , and admp1 did not increase the severity of the bmp2/4 morphant phenotype , in contrast , the double knockdown of bmp2/4 and panda produced a very strong phenotype , indistinguishable from that of the double alk1/2 + alk3/6 morphants . Strikingly , the ventralization induced by the double inactivation of panda and bmp2/4 was so strong that it frequently led to scission of the embryos in two parts by formation of a circular stomodeum and separation of the animal pole-derived proboscis from the vegetal part of the larva that contained the gut ( Fig 3A ) . Indeed , starting at early stages , the double panda + bmp2/4 morphants displayed a massive ectopic expression of nodal , similar to that caused by the double inactivation of Alk1/2 and Alk3/6 ( Fig 3B ) . The extent of this radialization was extremely pronounced , as evidenced by the radial expression of the other ventral markers chordin and foxA and of the ciliary band markers onecut and foxG as well as by the suppression of the dorsal marker hox7 both at the blastula and late gastrula stages . The summary diagram of Fig 3C shows that , while inactivation of bmp2/4 alone does not cause ventralization , in contrast , simultaneous inactivation of both panda and bmp2/4 , like the double knockdown of alk1/2 and alk3/6 , causes unrestricted expression of nodal leading to strong ventralization . These observations strongly support the view that Panda is the elusive factor that , together with BMP2/4 , is required to antagonize Nodal signaling during D/V patterning of the embryo . Taken together , these results also suggest that , in addition to Lefty , the normal restriction of nodal expression during D/V patterning in the sea urchin embryo requires the activities of Panda and BMP2/4 possibly signaling through the two BMP type-I receptors , Alk3/6 and Alk1/2 . In a previous study , we had suggested that the TGF-β encoded by SPU_018248 is related to Maverick sequences from insects and to GDF2 sequences from Crassostrea gigas; however , this analysis failed to identify any deuterostome orthologue of this gene , and the evolutionary origin of this TGF-β remained unclear [33] . To clarify the evolutionary relationships between SPU_018248 and other TGF-β family members and to identify orthologous sequences of this gene in deuterostomes , we performed a novel phylogenetic analysis using a comprehensive set of TGF-β sequences from protostomes , deuterostomes , and cnidarians and including in the analysis the Maverick sequence from Drosophila and the GDF2 sequence from Molluscs as well as a large set of BMP family members from different organisms ( Fig 4 and S4 Fig ) . This analysis confirmed that the sea urchin Panda sequence is phylogenetically related to Drosophila Maverick and "GDF2-like" sequence from Crassostrea . However , it further revealed that Panda and Maverick/GDF15-like factors belong neither to the GDF2/BMP9 family nor to any known subclass of canonical BMP ligands . In addition , this analysis identified GDF15 from vertebrates as well as two genes from hemichordates and cephalochordates ( called myostatin-like ) as additional deuterostome orthologues of Panda ( see also S6 Fig ) . Consistent with these conclusions , Panda , Maverick , and GDF15 share with Inhibins beta chains , TGF-β , and Myostatins a pattern of nine cysteines in the ligand domain , a pattern that is not shared by any prototypical BMP ligand ( see S5 Fig and the alignment provided in the supplementary information ) . Therefore , Panda , Maverick , and GDF15-like sequences define a distinct subclass of TGF-β ligands within a larger branch of the TGF-β superfamily that comprises Inhibins beta chains , Lefty factors , Myostatins , and TGF-β sensu stricto ( see also [35] ) . Previous studies on sea urchin maverick/panda failed to detect expression of this gene by in situ hybridization [33] , while by using an oligonucleotide microarray , a very weak expression was detected in 2 h zygotes and in 72 h pluteus larvae [36] . We reanalyzed the expression of panda by reverse transcription polymerase chain reaction ( RT-PCR ) and in situ hybridization and confirmed that transcripts of this gene are present predominantly in immature ovocytes , unfertilized eggs , and early embryos ( Fig 5A and 5B and S9 Fig ) . Remarkably , a graded distribution of transcripts could be detected in immature ovocytes , in eggs , and during early stages , with one side of the embryo showing a slightly stronger staining than the other , reinforcing the idea that this factor plays an early role in D/V axis formation . Furthermore , double labeling with nodal revealed that the side with the highest concentration of mRNA was the dorsal side , opposite to the side of nodal expression , consistent with the idea that Panda is a factor that cooperates with BMP2/4 to restrict nodal expression ( Fig 5B ) . Finally , starting at the prism stage , panda transcripts accumulated in the ciliary band territory , and strong expression was detected in individual cells within this territory . To further characterize the role of panda during D/V axis formation , we injected two different antisense morpholino oligonucleotides targeting either the translation start site or the 5' UTR region of the transcript . Injecting these two different morpholinos gave rise to similar and remarkable phenotypes ( Fig 6 ) . While at the late gastrula stage , control embryos had started to flatten on the presumptive ventral side and had formed two PMC clusters on each side of the archenteron , the panda morphants were completely radialized , and the PMCs remained arranged into a ring around the archenteron ( Fig 6A ) . Similarly at 48 hpf , when control embryos had developed into elongated pluteus larvae , panda morphants had conserved a radially symmetrical morphology and contained ectopic spicules rudiments ( arrowheads in Fig 6A ) . Surprisingly , at 72 h , these embryos had partially recovered a D/V polarity as indicated by the bending of the archenteron towards the presumptive ventral ectoderm , the opening of the stomodeum , and the preferential elongation of spicules on the presumptive dorsal side ( Fig 6A ) . Indeed , molecular analysis revealed that in most of the embryos ( n > 300 ) , knocking down panda with either the ATG morpholino ( Fig 6B ) or the UTR morpholino ( Fig 6C ) caused a strong ventralization accompanied with massive ectopic expression of nodal and chordin , which were expressed throughout most of the ectoderm at the mesenchyme blastula stage , and a concomitant loss of the dorsal marker gene hox7 . At the late gastrula/prism stage ( 30 hpf ) , panda morphants remained ventralized , as evidenced by the expanded expression of ventral marker genes such as nodal , foxA , and foxG compared to control embryos , occupying about one-half of the embryo . However , consistent with the progressive recovery of D/V polarity observed in live embryos , the expression of hox7 in the dorsal region and of the ciliary band marker onecut at the late gastrula stage indicated that dorsal and ciliary band fates were allocated in panda morphants by the end of gastrulation ( Fig 6B ) . Therefore , although the morphology of panda morphants is radially symmetrical at late gastrula stage , molecular analysis reveals that these embryos are nevertheless patterned along the D/V axis and that radialization is caused by a marked expansion of ventral cell fates . Taken together , these results suggest that Panda function is required early to restrict nodal expression . In the absence of Panda , ventral fates are expanded at the expense of dorsal fates , but this ventralization is most severe during the blastula and gastrula stages , the embryos progressively recovering , to some extent , a D/V polarity after 48 h ( Fig 6D ) . To determine when Panda , Alk1/2 , and Alk3/6 functions are required to restrict nodal expression , we performed a time-course experiment . We compared nodal expression at successive developmental stages , from cleavage to mesenchyme blastula , in control embryos and in embryos injected with either the morpholino oligonucleotide targeting the ATG of panda mRNA or with a morpholino oligonucleotide targeting a splice junction of the panda gene or with a mixture of alk1/2 and alk3/6 morpholino ( Fig 7A–7C ) . Strikingly , in embryos injected with the morpholino targeting the translation start site of panda mRNA , presumed to block both maternal and zygotic panda transcripts , or with a combination of the alk1/2 and alk3/6 morpholinos , nodal expression was never restricted and remained radialized at all stages analyzed ( Fig 7B and 7C ) . In contrast , nodal expression was largely normal in embryos injected with the morpholino targeting the splice junction of panda ( Fig 7B ) , and blocking zygotic panda function did not noticeably perturb development of the embryos ( Fig 7A and S7 Fig ) . RT-PCR analysis indicated that this splice-blocking morpholino reduced the level of the mature panda transcript by more than 90% at the pluteus stage ( S7 Fig ) . This analysis reveals that the function of maternal Panda , but not of zygotic Panda , and the activities of Alk1/2 and Alk3/6 are required very early to restrict nodal expression to the ventral side . The results presented so far indicate that Panda is expressed in a broad D/V gradient and that , like Lefty , Panda is critically required for the correct spatial restriction of nodal to the ventral side during early stages . We then tested if overexpression of Panda , like overexpression of Lefty , efficiently blocks Nodal signaling . Surprisingly , overexpression of panda in the egg did not perturb establishment of the D/V axis , and the panda-overexpressing embryos developed into normal pluteus larvae ( Fig 8A ) . This suggested that unlike Lefty , Panda alone is not capable of suppressing Nodal signaling when overexpressed . We then reasoned that rather than inhibiting Nodal signaling , the function of Panda may instead be to bias early Nodal signaling , perhaps by simply attenuating Nodal signaling on the dorsal side . If this were the case , then local overexpression of panda should efficiently orient the D/V axis . To test if local overexpression of panda is capable of biasing the orientation of the D/V axis , embryos at the two-cell stage were injected into one blastomere with panda mRNA together with a lineage tracer , and at the prism stage , the position of the clone of injected cells was recorded ( Fig 8B ) . Strikingly , in almost 100% of the embryos injected with panda mRNA , the boundaries of the clone precisely coincided with the dorsal part of the embryo . Local overexpression of a constitutively active version of Alk3/6 ( Alk3/6QD ) or Alk1/2 ( ALK1/2Q/D ) mimicked the effects of local overexpression of panda , efficiently orienting the D/V axis in all the injected embryos ( Fig 8B , Table 1 ) . Finally , we tested if overexpression of panda promotes expression of dorsal marker genes . We analyzed the expression of tbx2/3 , the earliest zygotic expressed in all three germ layers in the presumptive dorsal region and that is thought to be induced by low levels of BMP signaling [24 , 25] . Overexpression of panda induced a moderate ectopic expression of tbx2/3 in all three germ layers , suggesting that panda , like bmp2/4 , can activate BMP target genes requiring a low level of BMP signaling ( Fig 8C ) [24 , 25] . We next tested if removing the function of panda from part of the early embryo is also sufficient to orient the D/V axis ( Fig 8B and Table 1 ) . Indeed , injecting the panda morpholino randomly into one blastomere at the two-cell stage significantly biased the orientation of the D/V axis , most embryos ( 77 . 5% ) showing a clone of fluorescently labeled cells in the ventral region . Similarly , injection of alk3/6 morpholino into one blastomere at the two-cell stage efficiently ( 70% ) oriented the D/V axis , supporting the idea that Alk3/6 is involved in the early steps of D/V axis specification . Injection of the alk1/2 morpholino also significantly biased the orientation of the D/V axis , imposing a ventral identity to the clone in about 50% of the injected embryos ( Fig 8B ) . In contrast , injection of the bmp2/4 morpholino into one blastomere did not significantly orient the D/V axis , 37% of the injected embryos displaying a clone of injected cells on the ventral side , further suggesting that bmp2/4 is not involved in the early steps of axis specification ( Table 1 ) . In summary , these results show that while manipulating the levels of BMP2/4 does not appear to have a strong effect on the orientation of the D/V axis , in contrast , up-regulating or down-regulating the levels of Panda or Alk3/6 , and to a lesser extent of Alk1/2 , in part of the early embryo strongly impacts on the orientation of the D/V axis , partially mimicking manipulations of the levels of Nodal signaling . In the course of our functional analysis of panda , we tried to rescue the defects of D/V patterning and the spatial restriction of nodal expression of panda morphants , by injecting a synthetic panda mRNA lacking the sequence targeted by the morpholino . Surprisingly , injection into the egg of a synthetic panda mRNA failed to rescue the severe defects of D/V polarity caused by injection of the panda morpholino ( Fig 9 ) . All the embryos derived from double injection of panda morpholino and panda mRNA at the one-cell stage developed with a phenotype indistinguishable from the panda loss-of-function phenotype . Since Panda is required to restrict nodal expression and since the endogenous panda mRNA is enriched on the presumptive dorsal side , we reasoned that maybe Panda activity had to be provided locally in order to mimic the distribution of endogenous panda mRNA and to rescue D/V polarity of panda morphants . Indeed , while injection of panda mRNA into the egg was inefficient to rescue the D/V axis , injection of panda mRNA into one blastomere completely rescued D/V polarity of embryos previously injected with the panda morpholino , all the embryos developing into perfectly normal pluteus larvae with the dorsal side corresponding to the panda-expressing clone ( Fig 9 ) . This experiment demonstrates that the activity of exogenous Panda has to be spatially restricted to rescue the lack of maternal Panda function , consistent with the idea that the activity of endogenous Panda is spatially restricted in the early embryo . Similarly , injection into one blastomere of alk3/6QD mRNA at low doses that do not dorsalize completely rescued D/V polarity of embryos previously injected with panda morpholino , consistent with previous results showing that local misexpression of an activated form of Alk3/6 is sufficient to antagonize nodal expression and to orient the D/V axis ( Fig 8B ) . The finding that knocking down Panda causes a phenotype similar to that caused by knocking down the two BMP type I receptors Alk1/2 and Alk3/6 , leading to early ectopic expression of nodal , and the fact that local expression of alk3/6QD efficiently rescues D/V polarity in panda morphants indicated that Panda most likely uses Alk1/2 and Alk3/6 to signal . To further address the question of the specificity of the ligands regarding the receptors , we used an assay based on the double knockdown of Panda or BMP2/4 and Alk1/2 or Alk3/6 receptors ( Fig 10 ) . Double inactivation of panda and alk1/2 or of panda and alk3/6 caused a strong ventralization similar to that caused by the double inactivation of panda and bmp2/4 , consistent with the idea that the activities of Alk1/2 and Alk3/6 are both required to transduce BMP2/4 signals ( Fig 10A ) . Similarly , the double inactivation of bmp2/4 and alk1/2 or of bmp2/4 and alk3/6 produced a strong ventralization , suggesting that the activities of Alk1/2 and Alk3/6 are both required to transduce Panda signals , although the phenotype was slightly less severe than that resulting from the double knockdown of panda and bmp2/4 ( Fig 10A and 10B ) . To test directly the hypothesis that Panda requires the BMP type I receptors to orient the D/V axis , we used the axis induction assay . We first injected the alk3/6 morpholino into the egg , and then , at the two-cell stage , we injected panda mRNA into one blastomere . While panda mRNA efficiently oriented the D/V axis when injected alone , previous injection of the alk3/6 morpholino into the egg abolished the ability of panda mRNA to orient the D/V axis , suggesting that the Alk3/6 receptor is required for the activity of Panda ( Fig 10C ) . Taken together , the results presented above strongly suggest that Panda requires the activity of the BMP type I receptors to orient the D/V axis; however , they do not answer the question of what signal transduction pathway is activated by this ligand . Since the axis-inducing activity of Panda requires the BMP type I receptor Alk3/6 and since manipulating the levels of this BMP type I receptor largely mimicked the effects of manipulating the levels of Panda , we expected that Panda , acting through Alk3/6 and Alk1/2 , would activate canonical BMP signaling and Smad1/5/8 phosphorylation . However , intriguingly , previous studies failed to detect phospho-Smad1/5/8 before the hatching blastula stage using western blotting [24 , 25] , suggesting that previous detection methods were not sensitive enough or that Panda may not activate canonical Smad signaling . We therefore attempted to detect endogenous phospho-Smad1/5/8 during the cleavage/early blastula period using an optimized western blotting assay . We were able to detect very strong phosphorylation of endogenous Smad1/5/8 at mesenchyme blastula stages ( Fig 10D ) . However , endogenous phospho-Smad1/5/8 remained below the level of detection during cleavage stages , and overexpression of panda did not detectably increase the level of phosphorylated Smad1/5/8 at early stages . We also tried to detect phospho-Smad1/5/8 during cleavage/early blastula by using a sensitive immunostaining protocol . Fixed embryos were incubated with the anti-phospho-Smad1/5/8 antibody and then with a secondary antibody coupled to alkaline phosphatase ( Fig 10E ) . While overexpression of bmp2/4 or of an activated form of alk3/6 or of alk1/2 induced robust and very strong phosphorylation of Smad1/5/8 starting during cleavage stages , overexpression of panda did not cause any detectable phosphorylation of Smad1/5/8 at these early stages ( Fig 10E ) . However , intriguingly , at mesenchyme blastula , we consistently observed expanded phospho-Smad1/5/8 signals in most embryos overexpressing panda ( Fig 10F ) , consistent with the observed ectopic expression of tbx2/3 ( Fig 8C ) . However , pSmad1/5/8 signaling remained strongly polarized along the D/V axis , consistent with the apparent inability of panda to completely dorsalize embryos . Taken together , these results suggest that panda may not directly activate phosphorylation of Smad1/5/8 and that the expanded pSmad1/5/8 signals may result from Panda antagonizing Nodal signaling and/or promoting BMP2/4 signaling [37] . In conclusion , the results presented in this study show that in addition to Lefty , the spatial restriction of nodal expression critically requires the activity of the maternal TGF-β ligand Panda . Panda is required very early and locally for the spatial restriction of nodal expression and is sufficient to orient the axis when locally overexpressed . Taken together , these properties strongly suggest that Panda may act as a maternal determinant of D/V axis formation in the sea urchin embryo . The current prevailing model postulates that redox gradients generated by mitochondria asymmetrically distributed in the egg regulate the activity of redox-sensitive transcription factors that control the initial asymmetry of nodal expression [21–23 , 26] . However , although very attractive , the hypothesis that mitochondrial redox gradients drive nodal expression is not strongly supported by the extensive experimental work that has addressed this question . Here we provided several lines of evidence demonstrating that the maternally expressed TGF-β ligand Panda acts as an early and central player in the establishment of the D/V axis . First , we showed that the function of Panda is required very early to restrict nodal expression . Second , we showed that panda mRNA is expressed in a broad gradient in the early embryo and that the activity of Panda is spatially restricted . Third , we showed that overexpression of panda promotes the overexpression of tbx2/3 , the earliest zygotic dorsal gene marker . Finally , we showed that local misexpression of panda mRNA or local inhibition of panda very efficiently orients the D/V axis . The broad distribution of Panda mRNA raises the possibility that some other localized factor may be required in the normal embryo for imposing D/V polarity on Panda function . The fact that only local overexpression , and not global overexpression , of Panda rescues the D/V axis of Panda morphants strongly suggests that this is probably not the case , since if another localized factor was playing the role of a maternal determinant , then injection of panda mRNA into the egg would rescue the D/V axis of panda morphants . Therefore , Panda is , to our knowledge , the first signaling factor whose activity is spatially restricted in the embryo and that is both necessary and sufficient to efficiently orient the D/V axis upstream of nodal expression . How can we reconcile the roles of Panda as a maternal signal that orients the D/V axis with the wealth of data correlating redox gradients and the asymmetric distribution of mitochondria with the secondary axis ? One possible mechanism that would reconcile the two bodies of evidence is that formation of the gradient of panda mRNA may be dependent on the activity or the distribution of mitochondria . Alternatively , redox gradients could differentially affect the stability/activity of Panda as shown recently in the case of another TGF-β ligand [38] . Similarly , these findings on Panda could be correlated to the role of p38 in promoting nodal expression during D/V axis formation . As shown by Bradham and colleagues , p38 activity is required for nodal expression , and after a period of ubiquitous activation , it is specifically down-regulated on the presumptive dorsal side [29] . Panda , acting through Alk1/2 and Alk3/6 , may be responsible for this down-regulation of p38 activity on the dorsal side , which may in turn prevent Nodal autoregulation , a hypothesis that we are currently testing . The sea urchin embryo is well known for its remarkable developmental plasticity , the best example of this flexibility being the ability of each blastomere of the four-cell stage to regulate and to develop into smaller but normally patterned pluteus larvae . The outcome of this experiment deeply influenced ideas about how the D/V axis may be specified in this embryo , leading to the commonly accepted view that D/V patterning of the sea urchin embryo relies on cell interactions in the zygote and not on asymmetrically distributed determinants . On the other hand , classical experiments of Horstadius using unfertilized eggs showed that artificially activated meridional halves frequently differentiate as left-right or D/V pairs . These observations led Horstadius to conclude that "there seems to be no doubt as to the existence of a ventral-dorsal axis in the unfertilized sea urchin egg" [14] . The finding that the spatially restricted activity of Panda directs D/V axis formation strongly supports this conclusion . However , the finding that the spatially restricted activity of maternal panda mRNA directs the orientation of the D/V axis may seem at odds with the results of the Driesch experiment . How can we reconcile the fact that the first blastomeres show an equivalent potential to reestablish a secondary axis with the graded activity of a maternal factor controlling formation of the D/V axis in the early embryo ? Following dissociation , each blastomere is expected to inherit a portion of the gradient of activity of Panda . One possibility is therefore that the portion of the gradient of activity of Panda inherited by each blastomere following dissociation is sufficient to reestablish the secondary axis . Indeed , with a reduced gradient of activity of Panda , the reaction-diffusion mechanism between Nodal and Lefty may in some cases be sufficient to amplify an initial asymmetry of the expression of nodal or lefty , leading to the restriction of nodal expression and to reestablishment of the secondary axis . Hörstadius repeated and extended the Driesch experiment by rearing each of the four blastomeres in a separate dish [39] . Interestingly , he noted that in some cases one or two embryos of the quartet differentiated and established a D/V axis faster than the others . This is exactly what would be expected if the blastomeres inherit different portions of the gradient of activity of Panda . The results of Hörstadius are therefore consistent with our finding that there is a maternal gradient of a dorsalizing activity in the early embryo . Our finding that the activity of two type I BMP receptors is essential to restrict nodal expression in the sea urchin embryo adds further support to the previously suggested idea that an antagonism between Nodal and BMP signaling may be an ancestral mechanism to specify the axes [40] . Evidence is accumulating that both in chordates and in echinoderms , a correct balance between BMP signals and Nodal signals is required for normal D/V patterning [10 , 11 , 40–42] . Both in echinoderms and in vertebrates , inactivation of the BMP pathway promotes cell fates controlled by Nodal . The process of specification of the distal visceral endoderm in the mouse embryo offers a striking example of such an antagonism . Formation and positioning of the distal visceral endoderm is regulated by an antagonism between BMP-Smad1 and Activin/Nodal-Smad2 signaling , and activin receptor II ( ACVRII ) has been shown to act as a limiting factor in this process [43 , 44] . Similarly , there is accumulating evidence that during left-right axis specification , the opposing activities of Nodal and BMPs are required for proper patterning along this axis and that BMP signaling is required to spatially restrict nodal expression [45–47] . For example , mouse embryos mutant for smad1 , smad5 , spc4 , or for the gene encoding the BMP type I receptor ACVR1 display bilateral expression of nodal [48–51] . Intriguingly , although this antagonism may be fundamental for axis specification , the underlying mechanism is not well understood , and how the two pathways interact is not known . This antagonism may result from a direct interaction at the level of the signaling components . For example , it has been suggested that a competition at the level of Smad4 may set a threshold on Nodal signaling [52] . Alternatively , this antagonism may result from an interplay at the level of the ligands and secreted antagonists produced downstream of each pathway or at the level of ACVRII , which acts as a common receptor for both pathways . Finally , an antagonism at the level of the transcription factors induced by Nodal or BMP may be responsible for the antagonism between the two signaling pathways . Interestingly , a recent study proposed that the gene encoding the homeobox repressor Hbox12 , a member of the Hbox12/pmar1/micro1 family [53–59] , is expressed early on the dorsal side of the embryo and that it represses nodal expression [60] , raising the possibility that this gene may act downstream of Panda to repress nodal expression . However , preliminary experiments to test this idea did not provide evidence for a link between Panda and Hbox12 ( S8 Fig ) . Although we have detected expanded phospho-Smad signaling following misexpression of panda at the beginning of gastrulation , we failed to detect activation of Smad1/5/8 signaling during the early cleavage period , i . e . , when panda is supposed to work , in embryos overexpressing panda . Therefore , Panda may not activate pSmad1/5/8 signaling directly , and the mechanism by which Panda antagonizes Nodal signaling during early stages remains presently unclear . We can envision several scenarios . Panda may antagonize Nodal signaling by heterodimerizing with Nodal and blocking its function . Such an activity has been reported in the case of BMP7 as well as in the case of Lefty [61 , 62] . Another possibility for the mechanism by which Panda may antagonize Nodal is that Panda may work like the Nodal antagonist Lefty , by sequestering a factor essential for Nodal signaling such as the co-receptor Cripto , ACVRII , or Alk4/5/7 [63] . The finding that blocking locally Nodal or ACVRII fully mimics the effects of overexpressing panda on the orientation of the D/V axis is consistent with this idea . However , these two hypotheses both predict that overexpression of Panda should strongly antagonize Nodal signaling , an activity that is not observed following overexpression of panda . One possibility to explain this result is that Panda may require another factor to act as a strong antagonist of Nodal signaling when overexpressed . Panda may therefore antagonize Nodal signaling in a way similar to that of Inhibin , which disrupts Activin signaling by acting through an intermediary factor and sequesters a factor required for Activin signaling [64] . A major difference between Panda and Inhibin is that while Inhibins have never been shown to require any type I receptor to function , our results indicate that Panda most likely requires the two BMP type I receptors Alk1/2 and Alk3/6 to antagonize Nodal . Therefore , if Panda acts by sequestering a factor required for Nodal signaling , this activity may also require functional Alk1/2 and Alk3/6 , possibly in a complex with these two receptors . Finally , it remains also possible that Panda signals through these type I receptors and activates a noncanonical non-Smad pathway [65] that in turn may antagonize the Nodal pathway . In line with this conclusion , members of the Panda/Maverick/GDF15 subfamily lack a highly conserved leucine residue present in the so-called "wrist" domain of all BMP ligands ( Leu 51 in human BMP2 ) that is critically required for binding of these factors to the type I BMP receptor . This suggests that members of the Panda/Maverick/GDF15 subfamily are low-affinity ligands for the BMP type I receptors or that the interaction between members of this subfamily and the BMP type I receptor may involve residues different from those involved in the interaction between canonical BMP ligands and the BMP type I receptors [66] . Also along these lines , it is intriguing to note that the mechanisms by which vertebrate GDF15 and Drosophila Maverick work remain also largely unknown . During Drosophila development , the maverick gene is broadly expressed during oogenesis and embryogenesis and throughout the larval stages [67] . Its function has long been enigmatic , but recent studies have uncovered a key role for Maverick during synaptogenesis at the neuromuscular junctions [68] . Maverick produced by glial cells was shown to promote expression of Glass bottom boat ( Gbb ) , the fly ortholog of BMP7 , in muscles . Similarly , the function of GDF15 in mice and humans is poorly understood . GDF15 is weakly expressed in most tissues , but its expression is induced in response to tissue injury , notably in the heart following myocardial infarction [69] . Neither Drosophila Maverick nor vertebrate GDF15 have been shown so far to activate any of the signaling pathways normally activated by BMP or Activin type ligands , and the mechanism by which these factors work remains unknown [69–71] . Our results showing that Panda antagonizes nodal expression by acting through type I BMP receptors and that overexpressed Panda induces tbx2/3 without detectably activating Smad1/5/8 signaling points to non-Smad signaling as a potential mechanism for the crosstalk between the Nodal and BMP pathway [37] . However , we cannot completely rule out the possibility that Panda may induce a level of Smad1/5/8 activation below the current limits of detection , a level that would be sufficient to mediate its effects . Finally , a combination of the different mechanisms mentioned above including antagonism between Smad1 and Smad2 , sequestration of rate-limiting components , and antagonism between transcription factors induced downstream of Smads may underlie the antagonism between Nodal and BMP signaling in the sea urchin embryo . Biochemical and functional experiments will therefore be required to dissect the mechanism by which Panda antagonizes Nodal signaling in the sea urchin embryo . An interesting parallel can be drawn between the identification of Panda as a maternal TGF-β ligand acting through BMP receptors that cooperates with the zygotic BMP2/4 and the finding that maternal Univin , a Vg1 related ligand , cooperates with the zygotic Nodal . In the case of Nodal and Univin , it has been suggested that Nodal may heterodimerize with Univin and increase its specific activity [72] . Indeed , while Nodal is a strong ventralizing factor , overexpression of Univin has very modest effects on D/V patterning . Similarly , BMP2/4 has an extremely strong dorsalizing activity , while Panda essentially lacks dorsalizing activity . Heterodimer formation is , however , probably not the mechanism by which Panda and BMP2/4 cooperate , since Panda and BMP2/4 act at different periods during D/V axis formation and the activities of these factors appear to be qualitatively different . Panda is required early , starting at cleavage stages , well before BMP2/4 starts to be expressed , for the spatial restriction of nodal expression , while BMP2/4 is required much later , starting after hatching . Furthermore , while the only known activity of Panda is to limit and orient nodal expression and to induce tbx2/3 , BMP2/4 has a key role in activating a cohort of dorsally expressed transcription factors and signaling molecules . Finally , while BMP2/4 strongly activates phosphorylation and nuclear translocation of Smad1/5/8 , Panda only appears capable of weakly activating pSmad signaling . Therefore , D/V axis specification in the sea urchin embryo requires two phases of signaling from the BMP receptors , but these two phases are temporally and qualitatively different . The first phase of signaling , which covers the period of cleavage up to hatching blastula , is the consequence of maternal Panda signaling through Alk3/6 and Alk1/2 , either through very low canonical Smad signaling or possibly through noncanonical Smad signaling , while the second phase , which starts after hatching and continues late in gastrulation , is the result of zygotically produced BMP2/4 factors binding to the same receptors but activating canonical phospho-Smad signaling . Despite the fact that Panda and Lefty are both expressed early and that both factors are required nonredundantly to restrict nodal expression , the function of Panda is also clearly different from that of Lefty . Panda is capable of orienting the D/V axis when expressed into one blastomere at the two-cell stage , but overexpression of Panda in the egg does not suppress Nodal signaling . Furthermore , Panda is not sufficient to restrict nodal expression in lefty morphants . Therefore , the function of Panda appears to be to break the radial symmetry and to create the asymmetry of nodal expression rather than to maintain the asymmetry of nodal expression . In support of this idea , in the absence of Panda , nodal remains radially expressed up to the beginning of gastrulation . Therefore , although the function of Lefty is normal in these embryos , its activity is not sufficient to restrict nodal expression in the absence of Panda . In other words , without Panda , Lefty is unable to create an asymmetry of nodal expression . The function of Lefty appears therefore important to maintain the asymmetry of nodal expression previously established by Panda rather than to create this asymmetry . We have identified the maternal TGF-β ligand Panda as a novel and central player of the pathway controlling D/V axis formation . Although previous models placed Nodal as the first extracellular signal conveying spatial information for D/V axis formation , we can now place maternal Panda as the earliest known signal orienting the D/V axis upstream of nodal expression . A new model of D/V axis formation in the sea urchin embryo is the following ( Fig 11 ) . During oogenesis , maternal panda mRNA is deposited into the egg , possibly in a graded manner along the D/V axis , and following fertilization , this gradient of mRNA is translated into a shallow gradient of Panda protein . Starting at the 32/60-cell stage , ubiquitously expressed maternal transcription factors and maternal Wnt and Univin signaling activate nodal expression very broadly in the presumptive ectoderm . However , on the presumptive dorsal side , the increased activity of Panda weakly antagonizes nodal expression , introducing a first bias in nodal autoregulation that will initiate the spatial restriction of nodal expression to the presumptive ventral side . Then , starting at the early blastula stage , Nodal signaling induces expression of lefty , and the reaction-diffusion mechanism between Nodal and Lefty further contributes to maintain the spatial restriction of nodal expression . Finally , at the prehatching blastula stage , Nodal induces bmp2/4 and chordin expression . Chordin prevents BMP2/4 signaling on the ventral side while it shuttles BMP2/4 to the opposite dorsal side , where BMP2/4 activates the gene regulatory network responsible for specification of the dorsal side of the embryo . In conclusion , although Nodal remains the pivotal factor that regulates D/V axis formation in the sea urchin embryo , we have shown that the reaction-diffusion mechanism between Nodal and Lefty is not sufficient to break the radial symmetry of the embryo . This process of symmetry breaking is accomplished by a maternal factor , Panda , whose activity is required early and locally in the embryo to restrict the spatial expression of nodal . This work therefore illustrates how in the highly regulative sea urchin embryo , the secondary axis is already "penciled in " by the graded maternal information deposited into the egg in the form of a gradient of panda mRNA . Since nodal plays a key role in specification of the proximal distal axis in mammals and in specification of the secondary and left-right axes in a number of species , this raises the question as to whether members of the Panda/Maverick/GDF15 also provide a blueprint of axial development in these embryos . Adult sea urchins ( Paracentrotus lividus ) were collected in the bay of Villefranche . Embryos were cultured as described in Lepage and Gache ( 1989 , 1990 ) [73 , 74] . For immunostaining and in situ hybridization at early stages , fertilization envelopes were removed by adding 2 mM 3-amino-1 , 2 , 4 triazole 1 min before insemination to prevent hardening of this envelope , followed by filtration through a 75 μm nylon net . Treatments with recombinant BMP2/4 or Nodal proteins were performed by adding the recombinant protein diluted from stocks in 1 mM HCl , in 24-well plates containing about 1 , 000 embryos in 2 ml of artificial sea water [25] . Treatments with NiCl2 were performed by exposing embryos to 0 . 2–0 . 3 mM of chemical . All treatments were carried out from 30 min to 48 h post fertilization . A full-length panda cDNA was obtained by screening a cDNA library with conventional methods and sequencing the corresponding clones . A full-length alk1/2 cDNA was identified from a collection of P . lividus expressed sequence tags ( ESTs ) http://octopus . obs-vlfr . fr/ ) . The complete sequence of this clone was determined . The accessions numbers of panda and alk1/2 mRNA are KF498642 and KF498643 . To make pCS2 Alk1/2-Q225D , the CAG codon encoding Glutamine in position 225 of Alk1/2 was mutated to GAC by oligonucleotide-directed in vitro mutagenesis using the two following oligonucleotides: Alk1/2-Q225D fw: 5ʹ-cgaacagtagcaagagacatcaaccttattcaac -3ʹ Alk1/2-Q225D rev: 5ʹ- gttgaataaggttgatgtctcttgctactgttcg-3ʹ TGF-β sequences from deuterostomes ( vertebrates , cephalochordates , hemichordates , tunicates , and echinoderms ) , from protostomes ( arthropods and molluscs ) , and from cnidarians were recovered from Genebank ( http://www . ncbi . nlm . nih . gov/ ) using well-characterized orthologs of each TGF-β family member from human or mouse . The list of accession numbers of the 162 sequences is provided in the Supplementary Materials ( S1 Text ) . Full-length sequences were aligned using ClustalOmega with default parameters ( http://www . ebi . ac . uk/Tools/msa/clustalo/ ) , and gap optimization and obvious alignment error corrections were made using Bioedit 7 . 0 . 5 . 3 ( http://www . mbio . ncsu . edu/BioEdit/bioedit . html ) . The full complement of TGF-β sequences was recovered and used in the analysis in the case of human , mouse , sea urchin , Saccoglossus , Branchiostoma , and Drosophila . However , only a subset of sequences from Gallus , Xenopus , Danio , Ciona , Crassostrea , Platynereis , Hydra , and Nematostella was included in the analysis . Trees were built either using the maximum likelihood method based on the Whelan and Goldman model [75] or with Mr . Bayes3 . 2 , using the mixed model with two independent runs of 3 million generations [76 , 77] . In the case of the maximum likelihood analysis , the tree was calculated using PhyML [3] with substitution model WAG ( http://atgc . lirmm . fr/phyml/ ) . A consensus tree with a 45% cutoff value was derived from 500 bootstrap analysis using Mega 3 . 1 ( http://www . megasoftware . net/ ) . For the Bayesian analysis , consensus trees and posterior probabilities were calculated once the stationary phase was reached ( the average standard deviation of split frequencies was below 0 . 01 ) . Numbers above branches represent posterior probabilities , calculated from this consensus . The nodal , chordin , foxA , foxG , tbx2/3 , hox7 , and onecut probes have been described previously [24 , 25 , 34] . The panda probe was derived from a full-length cDNA cloned in Bluescript , while the alk1/2 probe was derived from a full-length cDNA cloned in pSport-Sfi . Probes derived from pBluescript vectors were synthesized with T7 RNA polymerase after linearization of the plasmids by NotI , while probes derived from pSport were synthesized with SP6 polymerase after linearization with XmaI . Control and experimental embryos were developed for the same time in the same experiments . Double in situ hybridizations were performed following the procedure of Thisse [78] . Detection of the lineage tracer was performed using an antifluorescein antibody coupled to alkaline phosphatase and using Fastred as substrate . For the time-course analysis of panda expression , total RNA from staged embryos was extracted by the method of Chomczynski and Sacchi [79] and treated with DNase I . cDNA synthesis and PCR were performed using standard procedures using 32–35 cycles of PCR [80] . For the characterization of the splice-blocking morpholino , RNA was extracted at the pluteus stage from batches of 400 embryos injected with increasing doses of the morpholino . Following treatment with DNase-I and phenol-chloroform extraction , RNA samples were reverse transcribed using the QuantiTect reverse transcription kit from Quiagen and following the instructions provided by the manufacturer . The in vivo specificity and efficiency of this morpholino were monitored via semiquantitative RT-PCR using 40 cycles of PCR . PCR primers flanking intron 1 were used to amplify the cDNA products generated in the presence of this splice-blocking oligonucleotide . Both the Phusion DNA polymerase and the kit long-expand PCR from Roche that allows amplification of long DNA fragments were used following the recommendations of the manufacturers . Primer pairs for the panda and mkk3 transcripts were derived from the open reading frames ( respectively 1 , 482 bp and 1 , 020 bp ) : panda-fwd: 5ʹ-GGAAAATGGCTCGACGCACATTCC-3ʹ panda-rev: 5ʹ-TGAGCAGCCGCAACTTTCTACGACCATATC-3ʹ mkk3-fwd: 5ʹ-ATGGCGAGTAAAGGTAAAAAG-3ʹ mkk3-rev: 5ʹ-TTAACTATTCTCCGGATCTCC-3ʹ The antibody we used is an anti-phospho-Smad1/5/8 from Cell Signaling ( Ref 9511 ) raised against a synthetic phosphopeptide corresponding to residues surrounding Ser463/465 contained in the motif SSVS of human Smad5 . Embryos were fixed in paraformaldehyde 4% in microfiltrated sea water ( MFSW ) for 15 min and then briefly permeabilized with methanol . Embryos were rinsed once with Phosphate Buffered Saline Tween ( PBST ) , four times with PBST–bovine serum albumine ( BSA ) 2% , and incubated overnight at +4°C with the primary antibody diluted 1/400 in PBST supplemented with 2% BSA . Embryos were washed six times with PBST-BSA 2% , and then the secondary antibody diluted in PBST-BSA 2% was added to the embryos . In all cases , the antibody was incubated overnight at +4°C . For immunofluorescence , the secondary antibody was washed six times with PBST . Two last rinses were made with PBST-Glycerol 25% and 50% , respectively . Embryos were mounted in a drop of the Citifluor antibleaching mounting medium and then observed under a conventional fluorescence microscope or with a confocal microscope . For Alkaline phosphatase revelation , two rinses were made with PBST following the secondary antibody incubation , and two with Tris Buffered Saline Tween ( TBST ) . Embryos were washed twice with the alkaline phosphatase buffer supplemented with Tween 0 . 1% , and staining was performed with nitro blue tetrazolium ( NBT ) and 5-bromo-4-chloro-3-indolyl phosphate ( BCIP ) as substrates at the final concentration of 50 mM each . In both cases , staining was stopped by four rinses with PBST + EDTA 5 mM and then two rinses with PBST containing glycerol at 25% and 50% . Embryos were mounted and observed with a DIC microscope . Protein samples ( 20 μg/lane ) were separated by SDS-gel electrophoresis and electrophoretically transferred to 0 . 2 μm PVDF filters . After blocking for 2 h with 5% milk in TBST , blots were incubated overnight with the anti-phospho-Smad1/5/8 antibody ( Ref 9511 ) diluted 1/1 , 000 in BSA 5% in TBST . After washing and incubation with the secondary antibody , bound antibodies were revealed by ECL immunodetection using the SuperSignal West Pico Chemiluminescent substrate ( Pierce ) . For overexpression studies , the coding sequence of the genes analyzed was amplified by PCR with a high-fidelity DNA polymerase using oligonucleotides containing restriction sites and cloned into pCS2 . Capped mRNAs were synthesized from NotI-linearized templates using mMessage mMachine kit ( Ambion ) . After synthesis , capped RNAs were purified on Sephadex G50 columns and quantitated by spectrophotometry . RNAs were mixed with Rhodamine Lysine-Fixable Dextran ( RLDX ) ( 10 , 000 MW ) or Fluoresceinated Lysine-Fixable Dextran ( FLDX ) ( 70 , 000 MW ) at 5 mg/ml and injected in the concentration range of 100–2 , 000 μg/ml . Wild-type panda and mutated panda mRNAs were injected at 1 , 000 μg/ml . mRNAs encoding the activated form of alk3/6 and alk1/2 , Alk3/6Q230D ( Alk3/6QD ) , and Alk1/2Q225D ( Alk1/2QD ) [25] were injected at 200 μg/ml . bmp2/4 and nodal mRNAs were injected at 400 μg/ml . To make the Panda and Alk1/2 rescue constructs , oligonucleotides containing nine mismatches in the sequences recognized by the morpholinos were used to amplify the coding sequences . The sequences of these oligonucleotides are as follows: Panda-rescue: 5ʹ-CCCATCGATACCATGGCGAGGCGTACGTTGCAGCGCTTGCAAGGGAGC-3ʹ Alk1/2-rescue: 5ʹ-CCCGGATCCACCATGGCCACCCGTCGTCTTGAGTTTATTTTTATACTTTTGG-3ʹ ( mismatches underlined ) . Morpholinos oligonucleotides were dissolved in sterile water and injected at the one-cell stage together with Tetramethyl Rhodamine Dextran ( 10 , 000 MW ) or Fluorescein dextran ( 70 , 000 MW ) at 5 mg/ml . For each morpholino , a dose-response curve was obtained , and a concentration at which the oligomer did not elicit nonspecific defect was chosen . Approximately 2–4 pl of oligonucleotide solution was used in most of the experiments described here . The sequences for morpholino oligonucleotides used in this study are as follows: Panda-Mo-ATG: 5ʹ-ATCTTTGGAATGTGCGTCGAGCCAT-3ʹ Panda-Mo1-splice: 5ʹ-TACTAATTTGGCGAGCCTACCTGTA-3' Panda-Mo2-splice: 5'-CGGAGGTCCATCTGAACGAAAGAAA-3' Panda-Mo-5' UTR: 5'-TTTCCTCGTGCTTGTAGAAATCTCC-3' Alk3/6-Mo: 5'- TAGTGTTACATCTGTCGCCATATTC-3' Alk1/2-Mo: 5'-TAAATTCTAGTCGTCGCGTCGCCAT-3' BMP5/8-Mo: 5'-CTTGGAGAGAAAATAAGCATATTCC-3' BMP2/4-Mo: 5'-GACCCCAGTTTGAGGTGGTAACCAT-3' ADMP-Mo: 5'-ACACGAAAATAATCTCCATTGTCTT-3' ACVRII-Mo-ATG: 5’- GGATCTTTCCCAGCCATTTCGGATA-3’ The panda , alk3/6 , and alk1/2 morpholinos were used at 1 . 2 mM , except the panda Mo1 splice , which was used at 2 mM . The bmp2/4 and bmp5/8 morpholinos were used at 0 . 3 mM . The acvrII and admp morpholinos were used at 1 . 5 and 0 . 8 mM , respectively . All the injections were repeated multiple times , and for each experiment , >100 embryos were analyzed . Only representative phenotypes present in at least 80% of the injected embryos are presented .	A key event during development of bilaterians is specification of the anterior-posterior and dorsal-ventral axes of the embryo . In some species , such as the fly Drosophila , this process relies on the activity of maternal determinants localized into the egg during oogenesis . However , in other animals , such as mammals or echinoderms , which are renowned for the developmental plasticity of their embryos , there is presently no evidence for maternal determinants controlling axis formation , and how these embryonic axes emerge from radially symmetrical embryos remains unknown . In the sea urchin embryo , specification of the dorsal-ventral axis critically relies on the localized expression of the TGF-β ligand Nodal in the presumptive ventral territory , but what controls the spatially restricted expression of nodal is not known . We discovered that in the sea urchin embryo , the initial restriction of nodal expression is directed by another TGF-β ligand that is expressed maternally , which we named Panda . Panda is both necessary for the early spatial restriction of nodal and sufficient to orient the dorsal-ventral axis when misexpressed locally . Altogether , our findings suggest that Panda may act as a maternal signal that defines the orientation of the dorsal-ventral axis . Thus , an antagonism between Nodal and maternal Panda signaling drives dorsal-ventral axis formation in the sea urchin embryo .	[ "Abstract", "Introduction", "Results", "Discussion", "Materials", "and", "Methods" ]	[]	2015	The Maternal Maverick/GDF15-like TGF-β Ligand Panda Directs Dorsal-Ventral Axis Formation by Restricting Nodal Expression in the Sea Urchin Embryo
Changes in the locations and boundaries of heterochromatin are critical during development , and de novo assembly of silent chromatin in budding yeast is a well-studied model for how new sites of heterochromatin assemble . De novo assembly cannot occur in the G1 phase of the cell cycle and one to two divisions are needed for complete silent chromatin assembly and transcriptional repression . Mutation of DOT1 , the histone H3 lysine 79 ( K79 ) methyltransferase , and SET1 , the histone H3 lysine 4 ( K4 ) methyltransferase , speed de novo assembly . These observations have led to the model that regulated demethylation of histones may be a mechanism for how cells control the establishment of heterochromatin . We find that the abundance of Sir4 , a protein required for the assembly of silent chromatin , decreases dramatically during a G1 arrest and therefore tested if changing the levels of Sir4 would also alter the speed of de novo establishment . Halving the level of Sir4 slows heterochromatin establishment , while increasing Sir4 speeds establishment . yku70Δ and ubp10Δ cells also speed de novo assembly , and like dot1Δ cells have defects in subtelomeric silencing , suggesting that these mutants may indirectly speed de novo establishment by liberating Sir4 from telomeres . Deleting RIF1 and RIF2 , which suppresses the subtelomeric silencing defects in these mutants , rescues the advanced de novo establishment in yku70Δ and ubp10Δ cells , but not in dot1Δ cells , suggesting that YKU70 and UBP10 regulate Sir4 availability by modulating subtelomeric silencing , while DOT1 functions directly to regulate establishment . Our data support a model whereby the demethylation of histone H3 K79 and changes in Sir4 abundance and availability define two rate-limiting steps that regulate de novo assembly of heterochromatin . Heterochromatin , or silent chromatin , is a specialized chromatin structure that plays structural and functional roles on chromosomes . These silent domains are characterized by repression of transcription and recombination , repressive histone modifications , and epigenetic inheritance [1] . The locations and boundaries of heterochromatic loci are dynamic and changing the landscapes of heterochromatin can cause changes in cell identity . In the budding yeast , Saccharomyces cerevisiae , heterochromatin is found at the silent mating loci ( HMLα and HMRa , or the HM loci ) , telomeres and the rDNA gene loci . At the HM loci and telomeres , heterochromatin contains hypoacetylated and demethylated nucleosomes that are bound by the SIR ( Silent Information Regulator ) complex , which consists of three proteins: Sir2 , Sir3 and Sir4 [1 , 2] . Sir2 is the founding member of a conserved family of NAD-dependent protein deacetylases and creates the hypoacetylated domains of nucleosomes within heterochromatin [3–5] . Sir3 and Sir4 are histone-binding proteins that bind with high affinity to deacetylated and demethylated nucleosomes [6–8] . Mutation of any of these three SIR genes abolishes both telomeric and HM silencing [9–14] . Current models for silent chromatin assembly propose that after initial recruitment of the SIR complex to nucleation elements , called silencers , at the HM loci , or to a telomere , iterative rounds of histone deacetylation and SIR complex recruitment lead to the spreading of the SIR complex [15–17] . After SIR complex spreading , recent work has suggested a final maturation step that requires the demethylation of K79 on histone H3 , which is needed to form functional silent chromatin and trigger transcriptional repression [18–20] . The de novo assembly of heterochromatin is best understood in budding yeast where there is a block to assembly in G1 phase , and where assembly requires passage through S phase and dissolution of sister chromatid cohesion [21–23] . The S-phase requirement does not depend on DNA replication , but instead depends on some other event that occurs during S phase [24 , 25] . Current models propose that this cell cycle dependence reflects a cell cycle dependent removal of both histone modifications and the histone H2A variant , Htz1 , which are refractory to heterochromatin assembly , and most cells take two cell cycles to fully silence a new site [19 , 26–28] . Methylation on K4 or K79 of histone H3 , catalyzed by the Set1 and Dot1 methyltransferases respectively , inhibit heterochromatin assembly [29–33] and de novo assembly of heterochromatin occurs faster in dot1Δ or set1Δ mutants , supporting the model that these modifications must be removed before a new site of heterochromatin can assemble [19 , 26 , 34] . To date , however , there is no evidence indicating that methylation in heterochromatic regions is removed at specific points in the cell cycle . Although changes in histone modifications are thought to regulate de novo assembly of silent chromatin , past work proposed that Sir4 may also play a role in regulating establishment . Silencing is sensitive to the dosage of Sir4: cells containing additional SIR4 have improved silencing and heterozygous SIR4/sir4Δ diploids de-repress a weakened HMR locus [35] . These data led to the idea that the abundance of Sir4 may regulate the establishment of silencing , but the assays employed in these studies were not able to differentiate between defects in the establishment , maintenance , or stability of heterochromatin , and there is currently no evidence that Sir4 protein levels fluctuate . We observed that Sir4 levels fall precipitously during a prolonged G1 arrest , and upon release from this arrest , Sir4 protein levels recover after two cell cycles , similar to the time required for de novo establishment of heterochromatin [19 , 21 , 26] . We therefore revisited the question of whether Sir4 dosage regulates de novo establishment using a single cell silencing establishment assay that monitors HMLα repression directly [26] . We find that increasing Sir4 abundance speeds , and decreasing Sir4 abundance slows , the de novo establishment of heterochromatin . ubp10Δ and yku70Δ mutants , like dot1Δ , also speed de novo establishment as well as disrupt telomeric silencing . To investigate the relationship between DOT1 , UBP10 , YKU70 and SIR4 , we examined the speed of de novo establishment in dot1Δ , ubp10Δ and yku70 mutants in conditions that lower Sir4 protein levels or that suppress their telomeric silencing defects . We conclude that UBP10 and YKU70 indirectly influence establishment by liberating Sir4 from telomeres , while DOT1 acts directly at HMLα , and propose a model in which the competition between telomeres and HMLα allows limiting Sir4 levels to regulate de novo establishment . We were interested in whether Sir4 protein levels fluctuate during the cell cycle and observed that during a prolonged G1 arrest induced by the mating pheromone , alpha factor , Sir4 levels fell precipitously ( Fig 1A ) . This decrease is not simply caused by cell cycle arrest , as cells arrested in mitosis with nocodazole maintain Sir4 levels ( Fig 1A ) . Quantification of this experiment shows that Sir4 levels fall four- to five-fold after five hours of growth in alpha factor , but remain unchanged in nocodazole ( Fig 1B ) . When pheromone-arrested cells are released back into the cell cycle , Sir4 protein levels recover after two cell cycles ( Fig 1C ) . We also observed a similar drop in Sir4 protein levels when cells are arrested in stationary phase by prolonged growth in raffinose , a poor carbon source ( S1A Fig ) . Re-feeding with dextrose allowed cells to re-enter the cell cycle and Sir4 levels recover after six hours . This dramatic decrease in Sir4 abundance during a pheromone arrest has not been reported previously , and is paradoxical: Sir4 , and the entire SIR complex , is required to maintain mating type identity , and mating type identity is needed to maintain a pheromone arrest . Sir4 localization by ChIP to a HMLα silencer ( HML-E ) and TELVI-R , however , is not significantly different in cells arrested in G1 by pheromone , in mitosis by nocodazole , or grown asynchronously , confirming that existing sites of heterochromatin are maintained despite falling Sir4 protein levels ( S1B Fig ) . In addition , the average intensity of intra-nuclear Sir4-GFP foci , which have previously been shown to mark clusters of telomeres and HM loci [36 , 37] , are similar between cells arrested in G1 and mitosis ( S1C and S1D Fig ) . Finally , most Sir4 is present on chromatin , and although there is less Sir4 in pheromone arrested cells , the ratio of chromatin to soluble Sir4 is similar in pheromone and nocodazole arrested cells ( S1E Fig ) . Sir4 protein re-accumulates in two cell cycles after release from G1 arrest ( Fig 1C ) . Previous reports have described similar timing for the establishment of de novo silencing [19 , 21 , 22 , 26] , so we considered the possibility that Sir4 abundance may play a role regulating de novo establishment . Previous work has shown that SIR4 is haploinsufficient for silencing at a weakened and modified HMRa ( hmrΔA::ADE2 ) [35] , and we see a similar defect at a telomere proximal URA3 reporter gene ( TELVII-L-URA3 ) in SIR4/sir4Δ heterozygotes ( S2A Fig ) . We wanted to determine whether this haploinsufficiency is caused by a defect in establishment or stability of heterochromatin , so we utilized a single cell silencing establishment assay in which two engineered strains are mated and the assembly of heterochromatin at HMLα is monitored in the resulting zygote and its progeny ( see [26] and Fig 2A for additional details ) . The diploid cells initially behave phenotypically as MATα cells , but upon silencing of HMLα they switch identity and behave as MATa cells . This switch is monitored by the response of the diploids and their progeny to exogenous mating pheromone , which causes cell cycle arrest and polarization in cells that silence HMLα . The number of cell divisions required for silencing establishment is determined by pedigree analysis ( Fig 2A ) . In this assay , approximately 80% of SIR4/SIR4 homozygotes establish silencing at HMLα in two cell cycles ( [26] and Fig 2B ) . sir4Δ/SIR4 heterozygotes , however , establish silencing at HMLα significantly more slowly ( Fig 2B ) . Heterozygous sir4Δ/SIR4 cells contain approximately half the amount of Sir4 as homozygous SIR4/SIR4 cells ( Fig 2C ) and the establishment defect is independent of which strain is deleted for SIR4 ( S2B Fig ) . Once HMLα is silenced ( and the diploid zygotes mate as MATa cells ) , there is no significant difference in the mating efficiency of SIR4/sir4Δ versus SIR4/SIR4 diploids ( Fig 2D ) indicating the defect in sir4Δ/SIR4 cells is specific to silent chromatin establishment and not long-term stability . The pseudo-haploid diploids , however , mate with 3-fold lower efficiency than control haploid MATa cells ( Fig 2D ) , perhaps due to an increase in size and ploidy compared to the control haploid cells . Because a decrease in the levels of Sir4 slows silent chromatin establishment , we wondered if increasing the concentration of Sir4 would speed establishment . Past work has shown that increasing Sir4 levels can improve silencing ( [35] and S3A Fig ) , but has not distinguished whether this improvement is caused by improved stability or changes in the efficiency of establishment . Selection of one , two or four low-copy centromeric plasmids containing SIR4 in zygotes increased the amount of Sir4 protein in these cells ( S3B Fig ) and significantly increased the speed of establishment in the single cell assay ( Fig 3 ) . These changes are not caused by the selection for multiple plasmids , as the presence of two or four empty plasmids does not change the speed of establishment as compared to cells containing no plasmids ( S3C Fig ) . Since the copy number of centromeric plasmids is variable in cells , we tested if other methods of increasing Sir4 would also improve silencing establishment . We created cells that contain SIR4 driven by a galactose inducible promoter ( GAL-SIR4 ) and constitutively express an estrogen receptor-Gal4 DNA binding domain fusion protein ( ER-GAL4-bd ) which allows graded expression of Sir4 using varying estradiol concentrations , rather than induction with galactose [38] ( S3D Fig ) . The speed of silencing establishment in cells with increased Sir4 expression is increased , in a manner similar to that seen with the addition of the SIR4-CEN plasmids ( S3E Fig ) . Further increasing Sir4 levels , however , is detrimental to the establishment of silencing . High copy 2μ-SIR4 plasmids and strong overexpression of SIR4 from a galactose inducible promoter ( using galactose rather than estradiol ) blocks or slows establishment , and causes derepression of both a telomeric URA3 reporter and a weakened and modified HMRa ( hmrΔE::TRP1 ) ( S4 Fig ) . These findings are consistent with past data showing overabundance of Sir4 in cells disrupts silent chromatin by preventing assembly of a complete SIR complex [39–41] . Deletion of DOT1 , the histone H3 K79 methyltransferase , has also been shown to speed the de novo establishment of silent chromatin [19 , 26 , 34] , which led to the proposal that the removal of histone H3 K79 methylation at silent loci may be a regulated step in de novo assembly . The effect of increasing Sir4 levels is similar to deleting DOT1 , so we wondered if changes in Sir4 abundance acted upstream , downstream or independently of DOT1 function ( Fig 4A ) . To test these three models we monitored silencing establishment in dot1Δ SIR4/dot1Δ sir4Δ diploids ( Fig 4B ) . If Dot1 functions downstream of Sir4 abundance ( Fig 4A , Model 1 ) then dot1Δ SIR4/dot1Δ sir4Δ diploids should be indistinguishable from dot1Δ/dot1Δ diploids . These diploids , however , are defective in the establishment of silent chromatin compared to SIR4/SIR4 cells , although to a lesser extent than SIR4/sir4Δ cells . This result clearly rules out Model 1 , in which Sir4 abundance functions upstream of Dot1 , but this result alone cannot distinguish between the two remaining models ( Fig 4A , Models 2 and 3 ) . In order to distinguish between the two remaining models ( Fig 4A ) we first considered how Dot1 , a histone modifying enzyme , might act upstream of Sir4 abundance . Although the positive effect of dot1Δ on silencing establishment has been proposed to reflect changes in chromatin state at the HM loci [19 , 26] , dot1Δ cells are also defective for subtelomeric silencing ( [32 , 33] and S5A Fig ) , and this phenotype might indirectly affect the HM loci . Past studies have shown that telomeres and the HM loci compete for silencing proteins [42 , 43] , and that loss of telomeric silencing can have indirect effects on other phenotypes due to re-localization of silencing proteins [44 , 45] . Mutation of UBP10 , a ubiquitin protease that targets K123 on histone H2B , allowed us to examine these two possible functions of Dot1 in silencing establishment . Unlike dot1Δ cells , deletion of UBP10 increases Dot1-dependent K79 and Set1-dependent K4 methylation on histone H3 by increasing K123 ubiquitination on histone H2B ( [46 , 47] and S6 Fig ) . However , similar to dot1Δ cells , ubp10Δ cells exhibit subtelomeric silencing defects detectable in strains containing a telomeric URA3 reporter ( [48] and S5A Fig ) . Interestingly , we also find that ubp10Δ/ubp10Δ cells , like dot1Δ/dot1Δ cells , speed the rate of establishment ( Fig 5A ) . To further test if loss of subtelomeric silencing could cause an earlier establishment phenotype at HMLα , we deleted YKU70 , a component of the Ku complex that is required for non-homologous end joining , normal telomere length maintenance and telomeric silencing [49 , 50] , but is not required for silencing of the native HM loci ( Fig 5B and 5C and [51–53] ) . Like dot1Δ/dot1Δ cells , yku70Δ/yku70Δ cells show a dramatic increase in the speed of silencing establishment ( Fig 5A ) . Deletion of YKU70 , UBP10 or DOT1 are specific to silencing establishment because the respective homozygous mutants have no effect on the long-term stability of HMLα silencing in pseudo-MATa diploids ( Fig 5B ) or HMRa silencing in MATα haploids ( Fig 5C ) , similar to the SIR4/sir4Δ heterozygote ( Fig 2D ) . Deletion of UBP10 or YKU70 also has no significant effect on K79 methylation of histone H3 at HMLα relative to wild type cells ( [46 , 47] and S6 Fig ) , demonstrating their effect on de novo silencing establishment is not due to inhibition of Dot1 . Like dot1Δ/dot1Δ cells , removing one copy of SIR4 in ubp10Δ/ubp10Δ or yku70Δ/yku70Δ cells slows the speed of establishment compared to SIR4/SIR4 cells ( S7A and S7B Fig ) , which resembles the phenotype of a SIR4/sir4Δ heterozygote . This supports the hypothesis that all three of these proteins may speed de novo establishment indirectly by derepressing subtelomeric silencing . Double mutant diploids that combine dot1Δ , ubp10Δ and yku70Δ were also tested for epistatic effects using the single cell establishment assay . None of the pairwise deletions further increased the speed of establishment compared to the more penetrant single mutant within the pair ( S7C–S7E Fig ) . Deletion of the telomere binding proteins RIF1 and RIF2 suppress the subtelomeric silencing defect in yku70Δ cells ( [54] and S5B Fig ) . This suppression allowed us to test if the advanced de novo establishment phenotype of yku70Δ cells at HMLα depends on their subtelomeric silencing defect . We find that rif1Δ/rif1Δ and rif1Δ rif2Δ/rif1Δ rif2Δ mutants also suppress the earlier silencing establishment phenotype of yku70Δ/yku70Δ ( Fig 6A and S7E Fig ) , supporting a model whereby effects on telomeric silencing in yku70Δ cells indirectly modulate the speed of silencing establishment at HMLα and act upstream of Sir4 availability ( Fig 4A , Model 2 ) . We also find that deletion of RIF1 and RIF2 suppress the telomeric silencing defects of dot1Δ and ubp10Δ cells ( S5B Fig ) , allowing us to rigorously test whether there is a correlation between the speed of silencing establishment at HMLα and the strength of subtelomeric silencing . As in yku70Δ/yku70Δ cells , rif1Δ rif2Δ/rif1Δ rif2Δ cells also suppress the earlier silencing establishment phenotype of ubp10Δ/ubp10Δ cells ( Fig 6B ) . This suppression , however , is not seen in dot1Δ/dot1Δ cells ( Fig 6C ) , despite the complete rescue of subtelomeric silencing in rif1Δ rif2Δ dot1Δ cells ( S5B Fig ) , suggesting two distinct pathways operate to regulate de novo establishment of heterochromatin . rif1Δ rif2Δ cells have extremely long telomeres which recruit large amounts of the telomere binding protein Rap1 [43 , 55 , 56] . Rap1 , which also binds at the silencer elements of HMR and HML , binds to both Sir4 and Sir3 , and is required for the nucleation of silent chromatin [57–60] . The increased recruitment of Rap1 to telomeres in rif1Δ rif2Δ cells has been speculated to cause the loss of silencing at weakened hmr silencers ( [43 , 55] and S5B Fig ) , suggesting that the suppression of the subtelomeric silencing defects of dot1Δ , ubp10Δ and yku70Δ cells is also caused by increased recruitment of Sir4 and Sir3 to telomeres . Consistent with this model and our hypothesis that the release of subtelomeric Sir4 speeds de novo establishment in yku70Δ/yku70Δ and ubp10Δ/ubp10Δ cells , we find that rif1Δ rif2Δ/rif1Δ rif2Δ cells have the opposite phenotype and slow the de novo establishment of silencing in a manner similar to that of SIR4/sir4Δ cells ( Fig 6D ) . Like SIR4/sir4Δ cells ( Fig 2D ) , rif1Δ and rif1Δ rif2Δ MATα haploids have similar mating efficiency as wild type cells ( Fig 6E ) , indicating that once silencing is established at HMLα , these mutants have no defects in maintaining the silent state . We observed that Sir4 protein levels fall during a prolonged arrest in G1 and re-accumulate over two cell cycles after release from this arrest ( Fig 1 ) . This is the first report of cell cycle-dependent changes in Sir protein abundance , and prompted us to directly test if Sir4 abundance modulates the de novo assembly of silent chromatin . Using a single cell establishment assay we have shown that decreasing Sir4 dosage in a SIR4/sir4Δ heterozygote slows de novo assembly at HMLα ( Fig 2 ) , and increasing Sir4 dosage speeds assembly ( Fig 3 ) . Importantly , the phenotype of SIR4/sir4Δ cells is specific to establishment , as a quantitative mating assay shows that there is no defect in the long-term stability of silencing at HMLα ( Fig 2B ) . These results suggest that in the single-cell pedigree assay the accumulation of Sir4 is a rate-limiting step in silencing establishment . The original experiments that suggested Sir4 dosage might regulate establishment used an ADE2 reporter placed at a weakened and modified HMRa locus ( hmrΔA-ADE2 ) and monitored repression and/or activation of ADE2 by red/white colony sectoring [35] . Although this assay was proposed to monitor establishment of silent chromatin , changes in sectoring may also monitor the maintenance or stability of heterochromatin . We tested if this sectoring assay could be used as a simpler alternative to the single-cell pedigree assay , and found that dot1Δ and yku70Δ cells , which speed establishment in the pedigree assay ( Fig 5A ) , derepress ADE2 ( S5A Fig ) . Thus , in the context of a weakened silencer at HMR , these mutants negatively impact silencing . We conclude that although experiments using hmrΔA::ADE2 strains identified Sir4 as a dose-dependent regulator of silencing [35] , this assay may not always reflect changes in de novo establishment . Past studies have shown that removal of histone H3 K4 and K79 methylation , and of Htz1-containing nucleosomes , which are all refractory to heterochromatin assembly , is rate-limiting for de novo silencing establishment [19 , 26–28 , 34] . We therefore wondered whether Dot1-dependent histone H3 K79 methylation acted in the same pathway as SIR4 , or in an independent pathway . The epistasis analysis between dot1Δ/dot1Δ and SIR4/sir4Δ cells clearly demonstrates that DOT1 inhibition of establishment is not downstream of the effects of changes in Sir4 abundance ( Fig 4A and 4B , Model 1 ) . Our finding that yku70Δ/yku70Δ and ubp10Δ/ubp10Δ cells also speed establishment ( Fig 5A ) suggested a model whereby the loss of subtelomeric heterochromatin common to the dot1Δ , yku70Δ and ubp10Δ mutants [32 , 33 , 41 , 49] may indirectly cause faster establishment of silencing at HMLα . Loss of one copy of SIR4 in all three mutants causes an intermediate phenotype between SIR4/SIR4 and SIR4/sir4Δ ( Fig 4B and S7A and S7B Fig ) , so these experiments cannot differentiate between a model that these genes act upstream or independently of Sir4 abundance ( Fig 4A , Models 2 and 3 ) . An intermediate phenotype might be expected if these genes acted independently of Sir4 abundance . Similarly , because SIR4/sir4Δ is not a null mutant and still contains Sir4 , some suppression of the SIR4/sir4Δ phenotype might be expected if these genes acted upstream of Sir4 abundance , and when mutated liberated Sir4 from telomeres . Our analysis of how deletion of RIF1 and RIF2 interacts with these three mutants significantly clarified our conclusions and suggests that DOT1 and Sir4 abundance function in independent pathways to regulate de novo establishment . Although rif1Δ rif2Δ cells suppress the subtelomeric silencing defects of yku70Δ , ubp10Δ and dot1Δ cells ( S5B Fig and [54] ) , the earlier establishment phenotype is only rescued in rif1Δ rif2Δ yku70Δ/rif1Δ rif2Δ yku70Δ and rif1Δ rif2Δ ubp10Δ/rif1Δ rif2Δ ubp10Δ cells ( Fig 6A and 6B ) . In contrast , rif1Δ rif2Δ dot1Δ/rif1Δ rif2Δ dot1Δ cells , like dot1Δ/dot1Δ cells , establish heterochromatin faster than wild type cells . These differences suggest that DOT1 functions in a distinct pathway from YKU70 and UBP10 ( Fig 7 ) . Our data argue that either increased Sir4 or the absence of histone H3 K79 methylation speeds de novo establishment of heterochromatin . These two events could regulate a common step in heterochromatin assembly , or function independently of each other . Because we see similar phenotypes in cells with increased Sir4 as with mutation of DOT1 and YKU70 , and we don’t observe additive effects in dot1Δ yku70Δ/dot1Δ yku70Δ or dot1Δ ubp10Δ/dot1Δ ubp10Δ mutants ( S7C–S7E Fig ) , we favour a model in which a common step in heterochromatin assembly is regulated by both Sir4 and histone H3 demethylation ( Fig 7 ) . Past work has shown that recruitment of Sir4 to silencers and telomeres occurs independently of Sir2 and Sir3 [15–17 , 61] and this first step in the nucleation of silent chromatin may be slowed during periods of limiting Sir4 protein ( Fig 7A , regulated nucleation ) . Dot1-dependent histone methylation directly antagonizes Sir3 binding to nucleosomes [62 , 63] , so the absence of H3 K79 methylation in dot1Δ cells may also regulate heterochromatin nucleation by improving Sir3 recruitment to the HML silencer . In this model , coincident recruitment of Sir3 and Sir4 , and their binding to one another , would be a slow step required for efficient nucleation ( Fig 7A and [64] ) . Overexpression of Sir3 does not speed establishment [26] , which is consistent with a model that Sir3 recruitment is determined by the extent of demethylation , not the concentration of Sir3 . The extent of histone H3 K79 methylation at HML-E is similar ( and low ) in the presence or absence of SIR4 ( S6 Fig ) , suggesting that histone H3 K79 demethylation at HML-E may be regulated independently of heterochromatin assembly , and that this function of demethylation is unlikely to be a regulated step in de novo establishment . Alternatively , because histone H3 K79 methylation often correlates with active transcription [65–67] and competes with Sir3 and Sir4 for binding to nucleosomes [31 , 68 , 69] , Dot1 may control Sir3 spreading , which could also function as a rate limiting step in de novo assembly ( Fig 7B , regulated completion of assembly ) . In this model , low levels of Sir4 would be sufficient for the spreading of Sir proteins , but additional Sir4 would be necessary at a later step , arguing that the occupancy of Sir4 would change during transcriptional repression of this region ( Fig 7B ) . The recruitment of the Sir4/Sir2 complex could be linked to the demethylation of histone H3 K79 , as previous work has shown that Dot1 competes with Sir3 and Sir4/Sir2 for binding to the N-terminal tail of histone H4 [31 , 68 , 69] . Supporting this model , several studies have suggested histone H3 K79 demethylation functions at a late step in silent chromatin maturation [18–20 , 70] and one showed that de novo establishment at a modified HMRa occurred after SIR complex recruitment and spreading [20] . This work investigated de novo establishment at HMRa , which occurs more slowly than at HMLα [71 , 72] , and establishment was triggered by the tethering of several Sir1 proteins , which recruit both Sir3 and Sir4 [73 , 74] . Both aspects of this experimental design , therefore , may alter the requirements for Sir4 and histone H3 K79 demethylation . Although histone H3 K79 demethylation speeds establishment at strong silencers , in the context of mutant silencers ( hmrΔE or hmrΔA ) , TELVII-L , or the deletion of the recruitment protein Sir1 , the absence of this methylation causes defects in silencing ( S5 Fig and [26 , 31–33 , 41 , 75] ) . We speculate that these defects are likely caused by the inability of the weak silencers to compete effectively with the recruitment of Sir3 non-specifically to other genomic loci [32 , 33 , 70] . Past work showing that telomeres act as reservoirs of Sir proteins [42 , 43 , 76–82] suggest a model in which the loss of subtelomeric silencing in yku70Δ and ubp10Δ cells liberates Sir4 from telomeres and increases its availability for de novo establishment at HMLα ( Fig 7 ) . In such a model the competition between telomeres and potential new sites of silencing would allow changes in Sir4 abundance to regulate the speed at which these new sites of heterochromatin can assemble . rif1Δ and rif1Δ rif2Δ cells have been shown to antagonize silencing at weakened HM loci , likely due to increased recruitment of the SIR complex to telomeres , and like sir4Δ/SIR4 cells , rif1Δ rif2Δ/rif1Δ rif2Δ cells slow de novo establishment ( Fig 6D ) . These mutants provide independent support of our hypothesis that telomere sequestration of Sir4 regulates de novo establishment . The phenotypes of rif1Δ rif2Δ/rif1Δ rif2Δ cells are milder than sir4Δ/SIR4 cells , which may explain why yku70Δ sir4Δ/yku70Δ SIR4 cells establish silencing slower than yku70Δ rif1Δ rif2Δ/yku70Δ rif1Δ rif2Δ cells . We observe similar differences in the phenotype of dot1Δ/dot1Δ cells when combined with sir4Δ/SIR4 or rif1Δ rif2Δ/rif1Δ rif2Δ , but in this situation we hypothesize that less Sir4 is needed for robust establishment at HMLα even though telomeric silencing is strengthened in rif1Δ rif2Δ/rif1Δ rif2Δ cells . Competition between heterochromatic sites has also been described in fission yeast in which sequestration of Swi6 at telomeres regulates the efficiency of assembly of heterochromatin at other sites , and release of Swi6 from telomeres or increased Swi6 expression allows bypass of RNAi-dependent assembly of pericentric heterochromatin [83] . This work suggested that a function of telomeric heterochromatin is to buffer cells from changing levels of heterochromatin factors and prevent inappropriate assembly of potentially harmful heterochromatin . Our work supports this model , but also suggests that competition between telomeric heterochromatin could function to generate phenotypic diversity within a population . Sir-dependent regulation of subtelomeric genes has been shown to influence cell adhesion , cell wall remodeling and stress resistance in budding yeast , expression of cell surface antigens in Plasmodium falciparum , and colony morphologies in Candida albicans [84–89] . Some of these changes are induced by environmental factors , but variation in the expression of a heterochromatin protein , like Sir4 , may also influence the competition between different subtelomeric regions to generate more subtle changes in cell identity . Our investigation into the role of Sir4 in de novo establishment began with the surprising observation that Sir4 abundance falls precipitously after prolonged G1 arrest ( Fig 1 ) . The re-synthesis of Sir4 after release from this arrest takes two cell cycles , which correlates with the time needed for cells to establish a new site of heterochromatin [19 , 21 , 26] . Although we are unable to monitor Sir4 protein levels during mating in individual cells , we hypothesize that the prolonged exposure to pheromone during mating also causes a drop in Sir4 protein abundance . After mating , SIR4/SIR4 zygotes would then require two cell cycles to re-synthesize Sir4 and re-establish silencing of HMLα . The slow kinetics of Sir4 degradation may explain why this drop in abundance has not been observed previously , and suggests that the decrease in Sir4 protein levels may be induced by pheromone , and not cell cycle arrest . SIR4 transcription is unchanged during pheromone treatment [90] , thus the drop in protein level is likely caused by regulated translation or degradation . Recent work has defined a pheromone-induced pathway that slows cell growth by causing inhibition of ribosome synthesis and translation [91] . The strength of this response depended on the concentration of pheromone and the extent of polarization , thus we speculate that the slow drop in Sir4 over five hours of arrest ( Fig 1 ) may indicate that Sir4 is regulated by the same pathway . Our finding that the abundance and availability of Sir4 can regulate de novo heterochromatin establishment suggests that changes in Sir4 abundance may be a cell cycle regulated step in de novo establishment [21 , 22 , 24 , 25] . However , although we have shown that modulating Sir4 abundance changes the speed of establishment in single cells ( Figs 1 and 2 ) , we have not been able to test if preventing the loss of Sir4 in pheromone arrested cells will allow for immediate re-establishment during the arrest . Identifying the mechanism that regulates Sir4 abundance will allow us to test this directly . Chromatin fractionation revealed that both the soluble and chromatin-bound fraction of Sir4 drops during G1 arrest ( S1E Fig ) . An explanation for this behavior is that the majority of chromatin-bound Sir4 binds to non-specific chromosomal sites and that this fraction rapidly equilibrates with soluble Sir4 , such that bound and unbound bulk levels of the protein drop equally during G1 arrest . Sir4 bound at specific chromosomal sites ( S1B Fig ) , however , may not exchange as rapidly , allowing maintenance of silencing despite low levels of Sir4 . Although the level of Sir4 protein during a pheromone arrest may be insufficient to establish new sites of silent chromatin , past work has shown that existing subtelomeric silent chromatin are more stable during a pheromone arrest [92] . The stability of existing sites would imply that two factors are at play: low levels of Sir4 prevent new sites of heterochromatin from forming , and structural or post-translational changes in Sir4 ( or another silencing protein ) prevent disassembly of existing sites . Recent work has shown that very low levels of Sir3 , like Sir4 , are also sufficient to maintain a silenced state [93] . This strategy may be useful for cells arrested in G1 , where maintenance of cell identity is critical as cells make developmental choices including the decision to mate , to initiate the meiotic program , to enter the cell cycle , or to enter quiescence . In addition , low levels of Sir4 have also been shown to improve silencing within the rDNA [82] , providing a mechanism for cells to maintain rDNA integrity during persistent G1 arrest . Similar developmental decisions are made in vertebrate cells in G1 , and similar mechanisms may be used to protect cell identity . This study was performed in strict accordance with standards for animal care and use outlined in the Canadian Council on Animal Care Standards . The University of Ottawa is a registered research facility under the Province of Ontario's Animals for Research Act . The protocol was approved by the University of Ottawa Animal Care Committee ( Permit Number: BMI-113 ) . All surgery was performed under sodium pentobarbital anesthesia , and every effort was made to minimize suffering . Supporting information S2 Table lists the strains used in this work . All strains , except the mating testers ( ADR3081 and ADR3082; gifts from Fred Winston , Harvard Medical School ) are derivatives of the W303 strain background ( W303-1a; Rodney Rothstein , Columbia University , New York , NY ) . Strains used for the single cell mating assay ( JRY8828 and JRY8829 ) were a gift from Erin Osborne and Jasper Rine ( UC , Berkeley , CA ) . All deletions and replacements were confirmed by immunoblotting , phenotype or PCR . The sequences of all primers used in this study are available upon request . The bacterial strains DH5α and Rosetta ( EMD-Millipore ) were used for amplification of DNA . All deletions were made using cassettes amplified from pAR747 ( C . albicans URA3 ) , pFA6a-His3MX6 ( S . pombe his5+ ) , pFA6a-kanMX6 [94] , pAG29 ( natMX4 ) [95] and pYM22 ( K . lactis TRP1 ) [96] . pAR747 was constructed by cloning the CaURA3 gene from pKT176 [97] as a BglII/XmaI fragment into pAG29 ( patMX4 ) [95] cut with BglII/XmaI . The TELVII-L::URA3 and adh4::URA3 strains were made by using pTEL::URA3 and pADH4::URA3 , respectively ( kindly provided by Dan Gottschling , Fred Hutchison Cancer Research Center ) to create ADR2828 and ADR2830 , respectively . hmrΔE::TRP1 [35] strains were constructed by crossing derivatives of CCFY100 ( kindly provided by Kurt W . Runge , The Lerner Research Institute , Cleveland , OH ) [98] to create ADR4062 which was subsequently used to create other hmrΔE::TRP1 strains . hmrΔA::ADE2 [35 , 99] strains ( kindly provided by David Shore ( University of Geneva , Switzerland ) via Marc Gartenberg ( Rutgers , NJ ) ) were constructed by deleting DOT1 , UBP10 and YKU70 in GCY317 . The SIR4-eGFP strain was created using pKT127 [97] . his3-11 , 15::pGAL-SIR4-HIS3 strains were created by integrating pAR655 cut with BsiWI into the appropriate strains . pAR655 was constructed by amplifying the entire SIR4 ORF by PCR and cloning it as a ClaI/EcoR1 fragment into pAR121 . pAR121 is pRS303 [100] with the GAL1-10 promoter cloned between the KpnI/XhoI sites . leu2-3 , 112::pMRP7-GAL4-ER-VP16-LEU2 strains were created by integrating pAR941 cut with XcmI . pAR941 was created by cloning the pGEV cassette as a FspI fragment from pGEV-HIS3 [38] into pRS305 [100] . SIR4-CEN plasmids , pAR646 and pAR722 were created by cloning the XhoI/EcoR1 fragment that contains SIR4 from pAR465 into pRS313 ( HIS3 ) and pRS316 ( URA3 ) [100] respectively . pAR465 is a modified version of LSD343 ( pRS314-SIR4; kindly provided by David Shore ) which includes 216 nt downstream of the STOP codon followed by a XhoI site and a silent MluI site has been introduced at nt2850 ( changing AAGAGT to ACGCGT ) . pAR450 contains the same SIR4 fragment as pAR465 ( without the MluI site ) cloned into pRS313 [100] . pRS313 and pRS316 [100] were used as empty CEN plasmids in pedigree experiments . The SIR4-2μ plasmid was created by cloning the XhoI/EcoR1 fragment that contains SIR4 from pAR646 into pRS423 ( HIS3 ) [100] . pRS423 was used as the empty 2μ plasmid in pedigree experiments . The single cell establishment pedigree assay was performed as described in Osborne et al . [26] with minor variations . Cells were grown overnight on plates of either YEP + 2% Dextrose , YEP + 2% Raffinose or synthetic selective media at 30°C , with the exception of ku70Δ dot1Δ ( ADR5920 and ADR5921 ) and ku70Δ dot4Δ ( ADR5944 and ADR5945 ) cells which were grown at 25°C overnight due to a slight temperature sensitivity . A small number of cells of each experimental strain were resuspended in YEP and each spotted onto a YEP +2% dextrose plate along with a thick streak of MATα ( ADR22 ) cells . Individual cells from the two experimental strains were micro-manipulated ( Nikon Eclipse 50i , 20X S Plan Fluor , NA0 . 45 , with TDM50 micromanipulator ) next to one another and allowed to mate over one to two hours . Zygotes were micro-manipulated adjacent to the streak of MATα cells and allowed to grow at 30°C . As the zygote ( and subsequent daughters ) divided , daughters were separated and moved in order to observe and score the behavior of the pedigree . Cells were monitored every one hour by microscopy and tracked for three divisions or until all cells in the pedigree had arrested and shmooed . Similar to Osborne et al . [26] , we did not observe plate-specific or day-specific effects on the pedigree patterns , and we therefore pooled data from several plates to compile the distributions for each genotype tested . Statistical differences between pedigree distributions were determined by likelihood ratio test as described [26] . Each distribution was compared pair-wise , and the complete dataset can be found in Supporting Information S1 Table . Cell cycle arrests were performed with 10μg/mL nocodazole ( Sigma-Aldrich ) , 1μg/mL α-factor ( Biosynthesis ) or 0 . 2M hydroxyurea ( HU; Sigma ) at 25°C . To image Sir4 foci , SIR4-GFP and wild type ( ADR4006 ) control cells were grown overnight in YEP + 2% dextrose to log phase , fixed in 4% paraformaldehyde for 10 minutes , washed , sonicated and resuspended in 100mM KPO4 containing 1 . 2M sorbitol . Samples were imaged using a Nikon TI microscope ( Nikon ) with a Nikon Plan Apo 60X 1 . 4 NA objective and FITC filter set ( Chroma ) at room temperature with a Photometrics CoolsnapHQ2 camera ( Photometrics ) . 13 fluorescent images using no ND filters and an exposure time of 2s were obtained separated by 0 . 5μm along the Z-axis . A single brightfield image was obtained at the central plane with an exposure time of 200ms . Example images were prepared using ImageJ software . Imaging was done in mixed populations of nocodazole ( mitotic ) and α-factor ( G1 ) treated cells; the two cell types were determined by cell morphology . The same linear look-up-table was used for each example image . Fluorescence quantification was done using NIS-Elements software . For each cell three fluorescent foci were analyzed . A total of eight cells of each genotype were obtained in two separate experiments . Mean fluorescence in a 0 . 25um2 circular ROI was obtained for each focus . For each focus , mean fluorescence from a similar sized background ROI obtained from the same cell and focal plane was subtracted . Non-focus fluorescence measurements were obtained as described above but all images were obtained at the same Z-plane . Three measurements were obtained in each of eight cells from two separate experiments . For serial dilution assays , cells were grown for two days in YEP + 2% dextrose or synthetic selective media + 2% dextrose at 30°C , and cultures were spotted on the indicated media in 10-fold serial dilutions using a multi-prong applicator ( Dan-Kar ) , and grown at 30°C for two to three days . Quantitative mating was performed as described previously [61] . Tested strains and tester strains were placed on YEP + 2% glucose plates and grown overnight at 30°C . On the following morning , the cells were scraped from the plates , and heavily inoculated into YEP + 2% dextrose liquid medium and grown for two to four hours at 25°C to an OD600 of one to two for the tested strains and to an OD600 of four to six for the tester strains . Quantities of 106 , 105 , 104 , 103 , and 102 cells of the tested strain ( in 100μl ) were mixed with 107 to 5 X 107 cells of the tester strain ( in 300μl ) . The mixture was plated directly on synthetic minimal + 2% dextrose plates and grown at 25°C for 3 days , and then colony counting was performed to determine the mating efficiency . A quantity of 102 cells of the tested strain was also plated on synthetic complete + 2% dextrose plates to accurately determine the number of viable cells that were mated in the experiment . Within each experiment , matings were done in duplicate , and at least three independent mating experiments were performed . Mating efficiencies are typically between 50 and 100% for wild type cells ( ADR21 and ADR22 ) . The matings in Figs 2D , 5B and 5C and 6E were performed by different individuals ( S . P . , C . D . and A . D . R . ) and accounts for the differences in efficiencies between identical ( JRY8828 X JRY8829 ) and similar ( ADR21 and ADR22 ) strains . hmrΔA::ADE2 cells were grown at 30°C overnight in liquid YEP + 2% dextrose media , 200–500 cells plated on YEP + 2% dextrose plates and grown for 3 days at 30°C . The plates were then left at 4 degrees to allow the red color to develop . Cells were scored for their ability to completely silence/express the ADE2 locus ( red or white ) or switch between silenced and unsilenced states ( sectored colonies ) . These methods have been described previously [61 , 101] . Yeast extracts for Western blotting were made by bead beating ( multitube bead beater; Biospec ) frozen cell pellets in 1X sample buffer ( 2% SDS , 80mM Tris-HCl pH6 . 8 , 10% glycerol , 10mM EDTA , bromophenol blue , 5% 2-mercaptoethanol ) . Samples were normalized by cell concentration before harvesting . Standard methods were used for polyacrylamide gel electrophoresis and protein transfer to nitrocellulose ( Pall ) . Typically , samples were run on 12 . 5% polyacrylamide gels ( 120:1 acrylamide:bis , respectively , with no added SDS ) . Blots were stained with Ponceau S to confirm transfer and equal loading of samples , and then were blocked for 30 min in blocking buffer ( 4% nonfat dried milk ( Carnation ) in TBST ( 20mM Tris-HCl ( pH 7 . 5 ) , 150mM NaCl , 0 . 1% Tween 20 ) . All antibodies were incubated overnight at 4°C or for 2 h at 25°C . After washing with TBST , the blots were incubated with horseradish peroxidase-conjugated anti-rabbit antibodies ( Bio-Rad ) at a 1:5 , 000 dilution in blocking buffer for 30 min at 25°C , washed again , incubated with Western Lightning reagents ( Perkin-Elmer ) and then exposed to X-Omat film ( Kodak ) or imaged on a GE Image Quant LAS4010 ( GE ) imaging system . Densitometry of bands was performed using ImageQuant TL ( GE ) software . Affinity-purified rabbit polyclonal anti-Sir2 [102] and anti-Sir4 antibodies were used at a dilution of 1:2 , 500 , and anti-Sir3 and anti-histone H4 ( Millipore; 04–858 ) were used at a dilution of 1:1000 , in antibody storage buffer ( autoclaved 4% nonfat dried milk , TBST , 5% glycerol , 0 . 02% NaN3 ) . Anti-Cdk1 antibody [101] was used at a dilution of 1:1000 in BSA storage buffer ( sterile 2% BSA , TBST , 0 . 02% NaN3 ) . Anti-Sir3 and anti-Sir4 antibodies were generated as follows . His6-Sir3 ( aa522-978 ) and GST-Sir4 ( aa1165-1358 ) was expressed in bacteria from pAR1007 and pAR411 , respectively ( kindly provided by Danesh Moazed , Harvard Medical School , Boston , MA ) . 1mg of each fusion protein was injected into rabbits every 4 weeks for 8 to 16 weeks ( uOttawa animal facility ) . Rabbit serum was harvested , clarified by centrifugation and loaded on Affigel-10 ( Bio-rad ) columns coupled to purified His6-Sir3 ( aa522-978 ) and malE-Sir4 ( aa1165-1358 ) , respectively . malE-Sir4 ( aa1165-1358 ) was expressed from pAR653 in which Sir4 ( aa1165-1358 ) was cloned as BamH1/Sal1 fragments into pMAL-c2 ( NEB ) . Antibody was eluted from Affigel columns with either 100mM triethylamine pH 11 . 5 or 100mM glycine pH 2 . 3 . The triethylamine and glycine elutions were neutralized , dialyzed in PBS + 50% glycerol and stored at -80°C . Chromatin fractionation was performed as described previously [103] . 0 . 5 X 109 cells at ~2 X 107 cells/ml were harvested and sodium azide was added to 0 . 1% . Cells were washed once in pre-spheroplasting buffer ( 100mM PIPES ( pH 9 . 4 ) , 10mM DTT ) and then incubated for ten minutes at room temperature in 5 ml of the same buffer . Cells were collected by centrifugation and resuspended in 1 ml of spheroplasting buffer ( 50mM KH2PO4/ K2HPO4 ( pH 7 . 5 ) , 0 . 6M Sorbitol , 10mM DTT ) containing 50μl of 2mg/ml Lyticase ( Sigma ) and incubated at room temperature with end-over-end mixing for 10–20 minutes . Spheroplasting was considered complete when the OD600 of a 1:100 dilution of the cell suspension ( in water ) dropped to <10% of the value before digestion . Spheroplasts were washed with 1 ml of ice-chilled wash buffer ( 100mM KCl , 50mM HEPES-KOH ( pH 7 . 5 ) , 2 . 5mM MgCl2 , and 0 . 4M Sorbitol ) , pelleted at 4000 rpm for 1 min in a microcentrifuge at 4°C , and resuspended in 500μl of extraction buffer ( 100mM KCl , 50mM HEPES-KOH ( pH 7 . 5 ) , 2 . 5mM MgCl2 , 50mM NaF , 1mM NaVO3 , and added fresh: 1mM PMSF , 1mM benzamidine HCl , and 10μg/ml of leupeptin , pepstatin , bestatin and chymostatin ) . Spheroplasts were lysed by adding Triton X-100 to 0 . 25% and incubating on ice for 5 min with gentle mixing . A portion of the total lysate ( T ) was removed and mixed with 2X sample buffer , and the remaining lysate was spun at 12 , 000 rpm for 10 min at 4°C . The supernatant ( S ) was removed and a portion mixed with 2X sample buffer . The crude chromatin pellet ( C ) was washed with extraction buffer containing 0 . 25% Triton X-100 , and spun again at 10 , 000 rpm for 5 min at 4°C . The pellet was resuspended in the wash buffer and then mixed with 2X sample buffer . Fractions were heated at 65°C for ten minutes and equal cell equivalent volumes were run on polyacrylamide gels ( 12 . 5% and 20% ) and processed for western blotting with the indicated antibodies . Chromatin immunoprecipitation was performed as described previously {Rudner:2005fd} . Sir4 was precipitated with 1μl of affinity purified rabbit polyclonal anti-Sir4 antibody , histone H3 was precipitated with 1μl of rabbit anti-histone H3 antibody ( Millipore; 04–928 ) and histone H3 methylated on K79 was precipitated with 1μl of rabbit anti histone H3-pan-methyl K79 antibody ( AbCam; ab28940 ) . PCR was performed with 0 . 1mCi [α-32P] dCTP ( 3 , 000 Ci/mmol; Perkin Elmer ) , reactions were resolved on 6% acrylamide ( 30:1 acrylamide:bis ) -Tris-borate-EDTA gels , and quantified by phosphorimaging on a Typhoon Trio phosphorimager and ImageQuant software ( GE Healthcare ) . Relative fold enrichment values for each strain were calculated as follows: [silent locus ( immunoprecipitate ) /ACT1 ( immunoprecipitate ) ]/[silent locus ( input ) /ACT1 ( input ) ] . The average and standard error for three independent experiments were determined . For clarity , these values were scaled so that the average values of sir4Δ in the Sir4 ChIP was set to one; the average values were scaled so that the dot1Δ values in the histone H3 ChIP was set to one .	Heterochromatin , characterized by the repression of transcription , is a specialized chromatin structure that plays both structural and functional roles on chromosomes . Heterochromatic domains are dynamic , switching between active and inactive states , and this property is used by cells during developmental switches and may generate phenotypic diversity . We have shown that competition between different heterochromatic domains for limiting amounts of a heterochromatin protein , Sir4 , plays a critical role in the switch from an active to an inactive state . Previous work has suggested that this switch is regulated by turnover of histone modifications in these regions and our data suggests that modulating Sir4 abundance acts in parallel to these changes to influence the rate of de novo assembly . This work supports a model in which competition between different chromosomal domains is exploited by cells to regulate cell identity .	[ "Abstract", "Introduction", "Results", "Discussion", "Materials", "and", "Methods" ]	[]	2015	Competition between Heterochromatic Loci Allows the Abundance of the Silencing Protein, Sir4, to Regulate de novo Assembly of Heterochromatin
Hospital-acquired pneumonia is associated with high rates of morbidity and mortality , and dissemination to the bloodstream is a recognized risk factor for particularly poor outcomes . Yet the mechanism by which bacteria in the lungs gain access to the bloodstream remains poorly understood . In this study , we used a mouse model of Pseudomonas aeruginosa pneumonia to examine this mechanism . P . aeruginosa uses a type III secretion system to deliver effector proteins such as ExoS directly into the cytosol of eukaryotic cells . ExoS , a bi-functional GTPase activating protein ( GAP ) and ADP-ribosyltransferase ( ADPRT ) , inhibits phagocytosis during pneumonia but has also been linked to a higher incidence of dissemination to the bloodstream . We used a novel imaging methodology to identify ExoS intoxicated cells during pneumonia and found that ExoS is injected into not only leukocytes but also epithelial cells . Phagocytic cells , primarily neutrophils , were targeted for injection with ExoS early during infection , but type I pneumocytes became increasingly injected at later time points . Interestingly , injection of these pneumocytes did not occur randomly but rather in discrete regions , which we designate ““fields of cell injection” ( FOCI ) . These FOCI increased in size as the infection progressed and contained dead type I pneumocytes . Both of these phenotypes were attenuated in infections caused by bacteria secreting ADPRT-deficient ExoS , indicating that FOCI growth and type I pneumocyte death were dependent on the ADPRT activity of ExoS . During the course of infection , increased FOCI size was associated with enhanced disruption of the pulmonary-vascular barrier and increased bacterial dissemination into the blood , both of which were also dependent on the ADPRT activity of ExoS . We conclude that the ADPRT activity of ExoS acts upon type I pneumocytes to disrupt the pulmonary-vascular barrier during P . aeruginosa pneumonia , leading to bacterial dissemination . Hospital-acquired pneumonia ( HAP ) is a severe form of nosocomial infection associated with attributable mortality rates of approximately 30% [1] . Dissemination of bacteria from the lung to the bloodstream is a particularly poor prognostic sign in HAP [2] . Yet the mechanisms by which bacteria disrupt the pulmonary-vascular barrier to reach the bloodstream remain largely unexplored . P . aeruginosa is the cause of approximately 15–20% of HAP cases [3–5] , and P . aeruginosa bacteremic pneumonia is associated with mortality rates considerably higher than non-bacteremic pneumonia [6 , 7] . Among the many virulence determinants of P . aeruginosa is a type III secretion system that injects toxic effector proteins directly into the cytosol of host cells [8] . P . aeruginosa secretes four known effector proteins by this pathway: ExoS , ExoT , ExoU , and ExoY . A functional type III secretion system has been associated with worse clinical outcomes and higher mortality rates in patients with P . aeruginosa pneumonia [9 , 10] . Approximately 70% of clinical strains contain the gene encoding ExoS [11] , underscoring the importance of understanding its contribution to P . aeruginosa pathogenesis . ExoS is a bi-functional effector protein , with GTPase activating protein ( GAP ) and ADP-ribosyltransferase ( ADPRT ) activities . The GAP and ADPRT domains of ExoS have been implicated in cell rounding , apoptosis , bleb-niche formation , and anti-internalization phenotypes using in vitro assays [12–16] . Likewise , ExoS secretion has been associated with more severe disease in several animal models , including a mouse model of pneumonia [17 , 18] . In both pneumonia and burn models , secretion of ExoS led to higher rates of dissemination from the site of infection , which may have contributed to the worse outcomes [18 , 19] . However , the mechanism by which ExoS causes dissemination remains unclear . There are few studies examining the effects of ExoS on pulmonary alveolar epithelial cells due in part to difficulties in isolating and culturing these cell types from mouse lungs . Lung alveoli consist of two types of epithelial cells , type I pneumocytes and type II pneumocytes , both of which are important for defense against bacterial pathogens . Type I pneumocytes are flat , elongated cells that are responsible for gas exchange within the lungs [20] . Along with the endothelial cells in lung capillaries and the basement membrane components , type I pneumocytes comprise the pulmonary-vascular barrier that prevents bacterial pathogens from entering the blood . Type II pneumocytes are smaller cuboidal cells that produce and secrete surfactant [20] . Surfactant decreases surface tension to facilitate lung expansion , interacts with resident macrophages to enhance phagocytosis of pathogens , and creates a physical barrier for entrapping bacteria [21] . Lung epithelial cells are also a major source of cytokines and chemokines that mediate rapid recruitment of immune cells into the lungs during infection . We hypothesized that interactions between ExoS and alveolar epithelial cells might contribute to the ability of P . aeruginosa to disseminate from the lungs to the bloodstream during pneumonia . In the present study , we used a novel in situ approach to examine the role of ExoS in acute pneumonia . We showed that early during pneumonia only phagocytic cells were appreciably injected with ExoS . However at later time points , clusters of type I pneumocytes were injected , and these discrete foci of injected cells increased in size as the infection progressed . The growth of these foci was associated with increased disruption of the pulmonary-vascular barrier and with increased bacterial dissemination into the bloodstream . We conclude that type I pneumocyte intoxication with ExoS causes breakdown of the pulmonary-vascular barrier , allowing bacterial dissemination during pneumonia . We previously showed that ExoS caused a heightened inflammatory response in the lungs , with the majority of infiltrating cells being neutrophils [22] . Furthermore , ExoS inhibited phagocytosis by these neutrophils [18 , 22] . Despite the ability of P . aeruginosa to inject a variety of cell types in vitro , we hypothesized that phagocytic cells would comprise the majority of intoxicated cells in vivo during early infection . We analyzed ExoS leukocyte injection using a mouse model of pneumonia and a β-lactamase reporter assay . In this assay , a fluorogenic β-lactam substrate , CCF2-AM , is applied to infected cells to detect translocation of ExoS fused with a β-lactam tag [23 , 24] . After diffusion into host cells , intact CCF2-AM exhibits green fluorescence . However , in the presence of injected ExoS containing a β-lactamase tag , CCF2-AM is cleaved , resulting in blue fluorescence . Cells can then be categorized as injected or not injected based upon their blue:green fluorescence ratio . Infections were performed using a variant of the clinical isolate PA99 , which naturally secretes ExoU , ExoT , and ExoS [18] . We wished to study the contribution of ExoS to disease progression in the absence of confounding effects caused by ExoU and ExoT . To accomplish this , we used a previously generated strain of PA99 with disruptions in exoU , exoT , and exoS ( designated PA99null [18] ) . PA99null was complemented at a neutral chromosomal site with a single copy of an exoS allele encoding ExoS with a C-terminal β-lactamase tag . This exoS allele is under the control of its endogenous promoter . We designated this strain PA99Sbla . We infected mice with PA99Sbla and recovered leukocytes from minced whole lungs at various time points post-infection to analyze the injected cell types . As expected , neutrophils were the predominant white blood cell type injected with ExoS ( Fig 1 ) . A substantial proportion of neutrophils were injected as early as 6 hr post-infection , and this proportion increased as the infection progressed . By 24 hr post-infection , approximately 15% of recovered neutrophils were injected with ExoS . It is plausible that this is an underestimation of the actual number of neutrophils injected with ExoS since at least 100 β-lactamase molecules are required per cell for detection of injection [23] . These results demonstrate that neutrophils are the leukocyte type most frequently injected with ExoS during early pneumonia . While abundant numbers of leukocytes can be recovered from minced whole lungs , epithelial cells such as type I and type II pneumocytes remain entrapped in large aggregates and cannot be analyzed . To determine whether ExoS intoxication of these cells might also occur , we adapted the CCF2-AM/ β-lactamase reporter assay for use on lung tissue sections . Unlike previous studies , this in situ analysis allowed us to study toxin injection in the context of intact lung tissue architecture [25 , 26] . For these experiments , whole lungs were excised from mice with pneumonia and incubated with CCF2-AM while intact to allow for β-lactamase cleavage of the fluorogenic substrate . The lungs were then fixed , frozen , and sectioned , and individual sections were stained with cell specific markers . We then used the TissueFAXS system ( see Materials and Methods ) to image cross-sections of entire lobes of the lungs ( S1A Fig ) . Next , TissueQuest software was used to identify and count injected cells based upon their blue:green fluorescence ratio ( S1B Fig ) . Blue:green fluorescence thresholds for detection of injected cells were determined by first using lung sections from mice infected with a strain secreting ExoS lacking a β-lactamase tag to set background fluorescence levels . Type I pneumocytes , type II pneumocytes , and phagocytic cells ( neutrophils and monocytes ) were identified and counted based upon staining with antibodies recognizing caveolin-1 , pSP-C , or Gr1 , respectively ( S1C–S1E Fig ) . A detailed description of the algorithm used to identify injected cells of each cell type is provided in S2 Fig and the Materials and Methods section . Mice were infected with PA99Sbla and sacrificed at various times post-infection . Lungs were removed and analyzed for injected cells . At 12 hr post-infection , phagocytes were the most frequently injected cell type within lung tissue sections ( Fig 2A ) , confirming our flow cytometry injection results ( Fig 1 ) . Phagocytes continued to comprise a large proportion of injected cells at 18 hr and 23 hr post-infection . Interestingly , relatively few type I pneumocytes were injected at 12 hr post-infection , but this cell type became increasingly injected at 18 and 23 hr post-infection ( Fig 2A ) . In contrast , type II pneumocytes represented a very small proportion of the total ExoS-injected cells even 23 hr after infection . This may be because type II pneumocytes are of relatively low abundance within the lungs ( Fig 2B ) or because they less frequently make contact with P . aeruginosa bacteria relative to type I pneumocytes and phagocytic cells . Alternatively , P . aeruginosa type III secretion may preferentially inject phagocytic cells and type I pneumocytes relative to type II pneumocytes . These results indicate that phagocytes are injected early during the course of P . aeruginosa pneumonia , that type I pneumocytes become increasingly injected at later time points , and that type II pneumocytes are not appreciably injected even at 23 hr after infection . In addition to identifying and counting injected cells , the TissueFAXS imaging system has the capability of marking these cells within the tissue sections of entire lung lobes . In this manner , we were able to examine the spatial distribution of ExoS-injected cells ( S3A Fig ) . We noticed that injected cells were not uniformly distributed throughout the lung sections but rather often occurred in distinct regions ( Fig 3 and S3B–S3D Fig ) . Examination of lungs infected for 12 , 18 , and 23 hr indicated that these regions of injected cells increased in size as the pneumonia progressed ( Fig 4A and 4B ) . Furthermore , the concentration of injected cells within these regions increased over time ( S4 Fig ) . Injection appeared to start at a few distinct sites within the lungs . These regions of injection subsequently expanded outward and coalesced as the infection progressed . We will refer to these regions of injected cells as “fields of cell injection” ( FOCI ) . We next examined the cell types that were injected within FOCI . We first examined lung sections taken from mice at a relatively early stage of infection ( 12 hr ) , when small FOCI were just becoming apparent . Even at this early time point , injected phagocytes were present throughout large regions of the lungs ( Fig 5C and 5D ) . These injected phagocytes caused a diffuse background of light blue throughout much of the lung . As expected , these regions corresponded to the distribution of bacteria in the lungs ( Fig 5A and 5B ) . Within the diffuse background of injected phagocytes , small FOCI were apparent as areas of more intense blue fluorescence . These FOCI contained injected type I pneumocytes ( Fig 5E and 5F and S4A Fig ) , as well as numerous bacteria ( S5 Fig ) . At later time points , more and more of the type I pneumocytes within FOCI became injected , and the regions of injected type I pneumocytes increased in size ( S4 Fig ) . Together with our time-course results ( Fig 2A ) , these observations indicate that bacteria were widely distributed in the lungs by 12 hr post-infection , and that these bacteria were competent to inject phagocytes . Within this background of injected phagocytes , small FOCI consisting of injected type I pneumocytes formed and subsequently increased in size and intensity . To address whether FOCI formation was unique to strain PA99 , we investigated type I pneumocyte injection using two additional strains . BL12 is a P . aeruginosa clinical isolate that was cultured from the blood of a patient with bacteremia , and PAK is a commonly used P . aeruginosa laboratory isolate . Unlike PA99 , both BL12 and PAK naturally secrete ExoS and ExoT , and in this respect are typical of the P . aeruginosa isolates most commonly cultured from patients [11] . We introduced an exoS allele encoding enzymatically inactive ExoS ( R146A and E379A/E381A substitutions resulting in loss of GAP and ADPRT activities , respectively ) tagged with β-lactamase into a neutral site in the BL12 and PAK chromosomes ( see Materials and Methods section ) . The resulting strains were designated BL12+S ( R146A/E379A/E381A ) bla and PAK+S ( R146A/E379A/E381A ) bla . In this way , we could identify injected cells using the TissueFAXS imaging system without affecting the normal progression of the infection [27 , 28] . The lungs were incubated with CCF2-AM , sectioned , stained for the type I pneumocyte marker , and analyzed using the TissueFAXS imaging system . BL12 caused a very severe pneumonia in mice , causing them to succumb prior to 23 hr , so results were only available for 12 or 18 hr . With this isolate , type I pneumocytes were injected earlier ( at 12 hr ) and to higher levels ( >80% of injected cells at 23 hr ) than with PA99Sbla ( Fig 6A ) . As a result , FOCI were readily observable even at the 12 hr time point ( Fig 6B ) . PAK likewise injected approximately 80% of type I pneumocytes by 23 hr ( Fig 7A ) , and FOCI were observed at this time point ( Fig 7B ) . These findings indicate that FOCI formation is not unique to strain PA99Sbla and suggest that P . aeruginosa strains differ in the rate and extent to which they form FOCI . As mentioned , most clinical isolates of P . aeruginosa secrete both ExoS and ExoT . Since these effector proteins are quite similar ( 76% amino acid identity [29] ) , we examined whether ExoT was sufficient to cause the formation of FOCI . Technical difficulties prevented us from doing this in the PA99 background , so we used strain PAK . The exoS allele encoding enzymatically inactive ExoS tagged with β-lactamase was inserted into a neutral chromosomal site of strain PAKΔexoS . This strain , designated PAKΔS+ S ( R146A/E379A/E381A ) bla , secreted ExoT but no enzymatically active ExoS ( S9 Fig ) . It was compared to PAK+S ( R146A/E379A/E381A ) bla , which secreted both ExoS and ExoT . Mice were infected for 12 or 23 hr , at which times they were sacrificed and their lungs removed . Lungs were incubated with CCF2-AM , sectioned , stained for the type I pneumocyte marker , and analyzed using the TissueFAXS imaging system . In the absence of enzymatically active ExoS , PAK secreting ExoT injected relatively few type I pneumocytes ( Fig 7A ) . Type I pneumocytes that were injected were relatively dispersed , and did not form readily apparent FOCI ( Fig 7C ) . These findings indicate that ExoT by itself is not sufficient to cause robust formation of FOCI . This may be because ExoT has different enzymatic substrates than ExoS [30] or because bacteria secreting ExoS persist in higher numbers in the lungs than bacteria secreting only ExoT [18 , 31] . ExoS is capable of causing cell death in cell culture systems [12 , 14 , 15 , 32] . We therefore characterized the viability of cells within FOCI . Using CCF2-AM in combination with a stain that identifies dead cells based upon membrane permeability , we found that many of the cells within FOCI were nonviable ( Fig 8A ) . The majority of these cells had morphologies consistent with type I pneumocytes ( Fig 8A ) . We confirmed the identity of these dead cells by staining lung sections with cell specific markers for type I pneumocytes ( caveolin-1 ) and phagocytes ( Gr1 ) in addition to the cell viability stain and performing confocal microscopy . The majority of non-viable cells within FOCI were indeed type I pneumocytes ( Fig 9A ) , although some dead cells stained positive for a phagocytic cell marker ( Fig 9C ) . These findings indicate that FOCI caused by ExoS injection contain large numbers of dead type I pneumocytes . We next investigated which of the enzymatic activities of ExoS were responsible for the formation of FOCI . We used strains of P . aeruginosa secreting ExoS with amino acid substitutions at residues critical for GAP activity ( R146A ) or ADPRT activity ( E379A/E381A ) or both GAP and ADPRT activities ( R146A/E379A/E381A ) , and fused these variants with C-terminal β-lactamase tags . These strains were designated PA99S ( R146A ) bla , PA99S ( E379A/E381A ) bla , and PA99S ( R146A/E379A/E381A ) bla , respectively . Because the loss of ExoS enzymatic activity may result in more rapid bacterial clearance [18 , 22] , we determined a dose for each mutant strain that yielded bacterial numbers in the lungs at 23 hr post-infection similar to those observed with wild type bacteria ( S6 Fig ) . Mice were infected with the corresponding CFU of each mutant strain . At 23 hr post-infection , mice were sacrificed . The lungs were incubated with CCF2-AM , sectioned , stained for the type I pneumocyte marker , and analyzed using the TissueFAXS imaging system as described above . Lungs infected with the ADPRT-deficient strains PA99S ( E379A/E381A ) bla or PA99S ( R146A/E379A/E381A ) bla showed a trend towards smaller FOCI within the lungs ( Fig 4B and 4C ) , suggesting that FOCI development and expansion was delayed during infection caused by these strains . As the infections progressed , enlarging FOCI associated with wild-type ExoS coalesced to encompass much of the lung lobe ( Fig 4A ) , whereas those associated with ADPRT- ExoS remained small and distinct ( Fig 4C ) . These results indicate that the ADPRT activity of ExoS is necessary for the rapid expansion of FOCI during pneumonia . Corresponding experiments were performed to examine the death of cells within FOCI . The use of a dead-cell stain demonstrated a higher proportion of death among injected cells in lungs infected with bacteria secreting fully active ExoS or GAP-deficient ExoS compared to bacteria secreting ADPRT-deficient or ADPRT/GAP-deficient variants of ExoS ( Fig 8B ) . Visual inspection indicated that many of these dead cells were localized to FOCI ( S5 Fig ) . Interestingly , although a substantial proportion of cells injected with GAP-deficient ExoS were killed , this proportion did not reach that observed with wild-type ExoS ( Fig 8B ) , suggesting that ExoS GAP activity may also play a role in cell death . We next examined the identity of these dead cells . As stated previously , wild-type ExoS was primarily associated with the death of type I pneumocytes within FOCI , although some dead phagocytes were also noted ( Figs 8A and 9 ) . In contrast , relatively few dead cells were observed in FOCI associated with ADPRT-deficient ExoS-secreting bacteria , and those dead cells that were present had a rounded morphology , consistent with immune cells ( Fig 8C ) . Cell type markers identified the majority of these cells as phagocytes ( Fig 9D ) , while relatively few had markers of type I pneumocytes ( Fig 9B ) . Together , these results indicate that the ADPRT domain of ExoS is responsible for the death of type I pneumocytes within FOCI during pneumonia . It is conceivable that the ADPRT-negative variants of ExoS caused less type I pneumocyte cell death because they were preferentially injected into other cell types , not because they were less toxic . To examine this possibility , we measured the proportion of injected cells that were type I pneumocytes during pneumonia caused by bacteria secreting wild-type ExoS or the ExoS catalytic variants . At 23 hr post-infection , we did not observe a smaller proportion of type I pneumocytes injected with enzymatic variants of ExoS compared to wild-type ExoS ( S7 Fig ) . In fact , the opposite was true . A higher proportion of type I pneumocytes was injected with enzymatically inactive ExoS , ADPRT-deficient ExoS , and GAP-deficient ExoS than was injected with fully active ExoS . This may have accounted for the overall increased blue fluorescence from sections of lungs injected with ADPRT-deficient ExoS compared to wild-type ExoS ( Fig 8C ) and is consistent with the enzymatic activities of ExoS rendering injected cells non-viable over time . The death and subsequent disintegration of cells injected with ADPRT+ ExoS would be expected to result in fewer intact injected cells in lung sections relative to cells injected with ADPRT- ExoS . Bacterial dissemination from the lungs into the bloodstream is normally prevented by type I pneumocytes , which form the pulmonary-vascular barrier . Since ExoS is associated with bacterial dissemination during infection [17 , 18 , 33] , we hypothesized that the formation of FOCI may be linked to disruption of this barrier and subsequent bacterial spread to the bloodstream . We therefore measured the integrity of the pulmonary-vascular barrier during pneumonia . Mice were infected with P . aeruginosa strains , and FITC-albumin was administered into their lungs by nasal aspiration . Leakage of the FITC-albumin into the bloodstream was then measured . As a control to determine what proportion of the disruption of the pulmonary-vascular barrier was due to ExoS and not other bacterial products or the overall inflammatory response , we used an isogenic non-secreting strain of P . aeruginosa designated PA99null . PA99null has a functional type III secretion system and is identical to PA99S except that it contains a deletion in the exoS gene . Since PA99null is rapidly cleared from the lungs [18 , 22 , 34] , we adjusted the inoculum of this bacterium so that equivalent CFU of PA99null and PA99Sbla were present in the lungs at 23 hr post-infection ( S6 Fig ) . At 12 hr post-infection , only background levels of FITC-albumin were detected in the blood of mice infected with either PA99Sbla or PA99null ( Fig 10A ) . However , at 23 hr post-infection , greater amounts of FITC-albumin were detected in the blood of PA99Sbla-infected mice compared to PA99null-infected mice , indicating that ExoS contributed to disruption of the pulmonary-vascular barrier during infection . We next investigated whether one or both of the enzymatic activities of ExoS was responsible for the pulmonary-vascular leakage . We repeated the FITC-albumin experiments using the strains secreting ExoS enzymatic variants . After controlling for bacterial CFU differences by adjusting the dose of each enzymatic mutant ( S6 Fig ) , we found that an intact ADPRT domain of ExoS was associated for the heightened leakage of FITC-albumin into the blood ( Fig 10B ) . These results indicate that the ADPRT activity of ExoS contributes to disruption of the pulmonary-vascular barrier . Since our previous results indicated that ExoS disrupted the pulmonary-vascular barrier , we next wanted to examine whether this was linked to bacterial dissemination . We enumerated bacterial CFU in the blood of PA99Sbla-infected mice with pneumonia over time . As the pneumonia progressed , significantly higher densities of bacteria were detected in the blood of infected mice ( Fig 11A ) . Likewise , bacteria were detected in a higher proportion of infected mice ( Fig 11B ) . A functional ADPRT domain of ExoS was required for high levels of dissemination ( Fig 11A and 11B ) . We also measured bacterial dissemination by quantifying bacterial numbers in the liver . Mice were infected with PA99Sbla or PA99S ( E379A/E381A ) bla , and bacterial counts in the liver were enumerated at 23 hr post-infection . Higher bacterial CFU in the liver were observed following infection with PA99Sbla compared to ADPRT-deficient PA99S ( E379A/E381A ) bla ( Fig 11C ) . These results are in agreement with previous findings [18] and show that the ADPRT activity of ExoS contributes to bacterial dissemination to the blood and liver . However , bacteria were still detected in the blood of some mice infected with the ADPRT-deficient strain . It is plausible that other factors or activities allow dissemination to occur , albeit in a delayed manner . Dissemination of bacteria from the lungs to the bloodstream bodes poorly for patients with pneumonia . Despite its clinical importance , the mechanisms by which bacteria escape the pulmonary compartment to spread systemically remain unclear . In the present study , we have explored this problem by focusing on P . aeruginosa and one of its major virulence determinants , ExoS . We used a novel imaging methodology to study the temporal and spatial aspects of ExoS injection into epithelial and immune cells in the context of a mouse model of pneumonia . We found that phagocytic cells , primarily neutrophils , were injected early during infection but that type I pneumocytes were injected at later time points . Injected type I pneumocytes were not observed throughout the lungs but were clustered into discrete regions ( designated FOCI ) that expanded as the pneumonia progressed . FOCI contained large numbers of dead cells , suggesting that they represent regions where the pulmonary-vascular barrier is compromised . Consistent with this hypothesis , pulmonary leakage of FITC-albumin and bacterial dissemination to the bloodstream both increased with expansion of FOCI . FOCI formation , pulmonary leakage , and bacterial dissemination all depended on the ADPRT activity of ExoS . We conclude that the ADPRT activity of ExoS facilitates spread of P . aeruginosa bacteria to the bloodstream following focal injection of ExoS into type I pneumocytes and subsequent disruption of the pulmonary-vascular barrier . Injection of epithelial cells by ExoS had been previously demonstrated in vitro but not in vivo , primarily because these cells are difficult to collect by bronchoalveolar lavage or from excised lungs . In addition , the CCF2-AM injection reporter system requires the use of living cells and therefore cannot be readily applied to fixed tissue sections . For these reasons , current methodologies allow only limited assessment of the spatial distribution of injection in the lung . We used two innovations to overcome these difficulties . First , we employed the TissueFAXS imaging system , which allows analysis of an entire lung cross-section and , similar to cell sorters , quantifies the number of cells in the cross-section with specified staining characteristics . Second , we applied the CCF2-AM stain to intact lungs prior to fixation and sectioning . These modifications allowed us to discern spatial relationships between injected cells of various types , including type I pneumocytes . The combination of the TissueFAXS approach and the β-lactamase reporter assay allowed us to identify regions with large numbers of injected type I pneumocytes , which we have designated FOCI . This method may have utility in the study of other pulmonary pathogens as well as to other organ systems . The factors dictating which regions of the lungs evolve into FOCI remain unclear . In the mouse model of pneumonia , bacteria spread rapidly throughout the lungs and are already found in the alveoli and distal airways at 3 hr post-infection [35] . However , bacteria are likely not uniformly distributed and may be more abundant in some regions of the lungs than others . Although our analysis did not quantify bacterial numbers , we did qualitatively observe higher concentrations of bacteria inside of FOCI than outside of FOCI ( S5 Fig ) . Therefore FOCI may form in regions of the lung that receive an unusually large inoculum of bacteria . Alternatively , the increased bacterial numbers may be the result and not the consequence of FOCI . FOCI may foster higher bacterial numbers because their dead cells release nutrients that fuel increased bacterial replication or because the tissue damage in these regions impairs an effective immune response . Interestingly , FOCI often contained small clusters in which injected type I pneumocytes were directly adjacent to one another ( Fig 5F and S4 Fig ) . It may be that injection of a single type I pneumocyte initiates a process whereby neighboring type I pneumocytes become prone to subsequent injection as the result of a breach in the polarized nature of the monolayer . Such a positive-feedback process would then result in rapidly enlarging clusters of injected cells . Interestingly , discreet plaque-like areas of injected cells have been described in several cell culture systems using primary human , mouse airway epithelial cells , or canine kidney epithelial cells [36–39] . Our findings suggest that FOCI formation may be an in vivo correlate of this in vitro phenomenon . Although ExoS has both GAP and ADPRT activities , we found that the ADPRT activity was primarily responsible for robust FOCI formation and bacterial dissemination . Others have similarly noted that the GAP activity of ExoS plays at best a minor role in the disruption of endothelial and epithelial cell monolayers in vitro [39 , 40] . Following infection with a mutant secreting an ADPRT-deficient form of ExoS , type I pneumocytes were still injected , but the resulting FOCI remained small . In addition , relatively few dead type I pneumocytes were present in these FOCI . Thus rapid expansion of FOCI requires the ADPRT activity of ExoS , although other mechanisms may suffice for the formation of FOCI . It remains unclear which of the phenotypes associated with the ExoS ADPRT activity ( e . g . cell rounding , apoptosis , bleb-niche formation , or anti-internalization [12–16 , 41] ) is responsible for the formation of FOCI . Whereas large numbers of neutrophils were already injected with ExoS by 6 hr post-infection , relatively few type I pneumocytes were injected with this effector protein even at 12 hr post-infection . Several explanations may account for the delayed injection of type I pneumocytes relative to neutrophils . Neutrophils and other phagocytic cells may be preferentially targeted for rapid injection by the P . aeruginosa type III secretion system , as is the case for the closely related Yersinia pseudotuberculosis type III secretion system [42 , 43] . The active chemotaxis of phagocytic cells towards bacteria may allow them to more rapidly achieve direct contact with bacteria , a prerequisite for type III secretion . Alternatively , bacteria may be prevented from contacting type I pneumocytes early during infection by the protective layer of surfactant that coats the alveoli and which may form a physical barrier that the type III secretion needle cannot penetrate . As the infection progresses , this protective layer may be compromised by P . aeruginosa products such as elastase B or host inflammatory factors such as cathepsin G , neutrophil elastase , and proteinase-3 , all of which degrade surfactant protein SP-A [44 , 45] . Perhaps the most likely explanation is that the polarized nature of type I pneumocytes establishes an apical barrier that is relatively resistant to P . aeruginosa attachment and type III injection [46 , 47] . We focused on the role of ExoS in facilitating bacterial dissemination to the bloodstream , but disruption of the pulmonary-vascular barrier may also adversely affect disease outcomes in other ways . In vitro studies indicate that ExoS causes loss of focal adhesions and retraction of endothelial cells [40 , 48] . Although we did not examine endothelial cells in our study , disruption of the epithelial barrier has been associated with corresponding disruption of the endothelial barrier in pneumonia caused by ExoS+ P . aeruginosa [17] . Such effects on endothelial cells could further contribute to the disruption of the overall pulmonary-vascular barrier and lead to prolonged alveolar edema [49–51] , which may potentiate severe pneumonia [49 , 52] . Of note , P . aeruginosa pneumonia in humans is accompanied by a high incidence of acute lung injury , the clinical correlate of destruction of the alveolar epithelia and enhanced permeability of the pulmonary-vascular barrier [49–51] . Thus ExoS injection into pulmonary epithelial cells may play an important role in the progression of pneumonia . We propose the following model to explain the role ExoS in bacterial dissemination during pneumonia . Bacteria enter the lungs and cause a rapid recruitment of inflammatory cells , primarily neutrophils . Early during infection , neutrophils are the predominant cell type injected with ExoS , which inhibits phagocytosis and prevents bacterial clearance [12 , 22] . As the infection progresses , small , discrete foci of type I pneumocytes become injected . The ADPRT activity of ExoS targets host cell substrates essential for maintenance of type I pneumocyte tight junctions and viability . This in turn leads to expansion of the foci and compromise of the pulmonary-vascular barrier . As a result , bacteria disseminate from the lungs into the bloodstream . The net effects are the worse clinical outcomes associated with ExoS+ P . aeruginosa infections . This work highlights one example of how a pulmonary pathogen utilizes a effector protein to promote dissemination to the bloodstream . Additional studies are necessary to determine whether other respiratory pathogens use similar mechanisms to escape from the lungs during pneumonia . The P . aeruginosa clinical isolate PA99 secretes ExoU , ExoS , and ExoT [11 , 53] . This strain was chosen for the present study because it is well characterized in the context of the mouse model of pneumonia and because it allows for the comparison of the effects of ExoS to those of ExoU in an isogenic background [18 , 22 , 31 , 34 , 54–56] . PA99S was previously derived from PA99 by disruption of the exoU and exoT genes [11] . PA99null , which possess an intact type III secretion apparatus but does not secrete known effector proteins , was previously generated by disrupting all three effector genes ( exoU , exoS , and exoT ) [18] . Also previously generated was PA99ΔpscJ , a strain identical to PA99 except that it has a disrupted pscJ gene , which renders its type III secretion apparatus defective [18] . A strain complemented for ExoS secretion , designated PA99null+S , was previously generated by integrating the exoS gene with its promoter into a neutral site in the chromosome using plasmid mini-CTXexoS [18] . PAK is a laboratory strain of P . aeruginosa that secretes ExoS and ExoT [57]; PAKΔS secretes only ExoT due to disruption of the exoS gene [15] . BL12 is a clinical isolate cultured from the blood of a patient at Northwestern Memorial Hospital , Chicago , IL . It secretes ExoS and ExoT ( Fig 9A ) . Further details on strains and plasmids used in this study can be found in S1 Table . Bacterial strains were streaked from frozen cultures onto Vogel-Bonner minimal ( VBM ) agar [58] . Overnight cultures were grown in 5 ml MINS medium [59] at 37°C . Cultures were diluted into fresh medium the next day and re-grown to exponential phase prior to infections . The stop codon of the exoS gene in plasmid mini-CTXexoS [18] was removed by site-directed mutagenesis with primers ( 5’- GGCCTTGATCTGGCCGGACCGGTCGTAAA-3’ ) and ( 5’- CGTCTTTCTTTTACGACCGGTCCGGCCAGAT-3’ ) to generate mini-CTXexoSΔstop . The DNA fragment encoding TEM-1 β-lactamase was amplified from pBR322 by PCR using primers with engineered AgeI sites as previously described [54] . This fragment was ligated into mini-CTXexoSΔstop to generate mini-CTXexoSbla , which encodes for a translational fusion of ExoS and the TEM-1 β-lactamase ( “ExoS-Bla” ) . This plasmid was then transformed into E . coli S17 . 1 and conjugated into the PA99null chromosome via integrase-mediated recombination at the attB site as previously described [60] to generate PA99Sbla . mini-CTXexoSbla was also introduced into the chromosome of PA99ΔpscJ [18] to create PA99ΔpscJ+Sbla , a strain identical to PA99Sbla except that it has a disrupted type III secretion apparatus . Proper secretion , translocation , and/or cytotoxic activity of tagged ExoS-Bla were confirmed by immunoblot analyses , cell death assays , CCF2-AM translocation assays , and/or bacterial persistence studies using the mouse model of pneumonia ( S8 and S9 Figs ) . To generate enzymatically inactive ExoS variants with β-lactamase tags , site-directed mutagenesis was performed on mini-CTXexoS-bla using primer sets as previously described [18] . In this way , we generated plasmids mini-CTXexoS ( R146A ) bla , mini-CTXexoS ( E379A/E381A ) bla , and mini-CTXexoS ( R146A/E379A/E381A ) bla [27 , 28] . These plasmids were then transformed into E . coli S17 . 1 as described above to generate PA99S ( R146A ) bla , which is GAP- , PA99S ( E379A/E381A ) bla , which is ADPRT- , and PA99S ( R146A/E379A/E381A ) bla , which is both GAP- and ADPRT- . mini-CTXexoS ( R146A/E379A/E381A ) bla was complemented into PAK , PAKΔS , and BL12 to study ExoS injection in the context of these strain backgrounds . In this way , we generated PAK+S ( R146A/E379A/E381A ) bla , PAKΔS+ S ( R146A/E379A/E381A ) bla , and BL12+S ( R146A/E379A/E381A ) bla . Proper secretion and translocation of the ExoS-bla fusion protein by these strains was confirmed using immunoblot analysis and a fluorescence cell assay ( S9 Fig ) . The mouse model of acute pneumonia described by Comolli et al . [61] was used for all animal experiments . Briefly , 6- to 8- week-old female BALB/c mice were anesthetized by intraperitoneal injection of a mixture of ketamine ( 75 mg/kg ) and xylazine ( 5 mg/kg ) . Mice were intranasally inoculated with 4 . 6–9 . 2 x 106 CFU PA99Sbla or PA99S ( R146A ) bla , or 1 . 8 x 107 CFU PA99S ( E379A/E381A ) bla , PA99S ( R146A/E379A/E381A ) bla , or PA99null in phosphate buffered saline ( PBS ) . These doses led to equivalent CFU of bacteria in the lungs at 23 hr post-infection ( S6 Fig ) . Inocula were confirmed by plating serial dilutions on VBM agar for enumeration . At appropriate times post-infection , the mice were anesthetized and sacrificed by cervical dislocation . For dissemination experiments , organs were excised and homogenized in PBS . Viable bacteria were enumerated by plating serial dilutions on VBM agar . For quantifying bacterial CFUs within blood , cardiac punctures were performed immediately post-mortem and the obtained blood placed into tubes containing 5 USP units of heparin . Bacteria were enumerated by plating 100 μl aliquots on VBM agar . Animals were purchased from Harlan Laboratories , Inc . ( Indianapolis , IN ) and housed in the containment ward of the Center of Comparative Medicine at Northwestern University . All experiments were approved by the Northwestern University Animal Care and Use Committee . To remove circulating blood , the vasculatures of lungs were flushed post-mortem by injection of 2 ml PBS into the right ventricle of the heart . Lungs were excised , pressed through 40-μm filters ( BD Falcon , Becton , Dickinson and Company , Franklin Lakes , NJ ) , and rinsed repeatedly with PBS . The recovered cells were pelleted by centrifugation at 500 x g for 5 min at 4°C . Red blood cells were lysed by adding 5 ml of cold , sterile H2O and gently shaking for 30 sec . Five milliliters of 2X normal saline ( 1 . 8% NaCl ) was added to stop lysis . The remaining cells were pelleted and resuspended in 1 ml PBS . Viable cells were quantified using a hemocytometer by counting cells that excluded trypan blue . A total of 2 x 105 cells were pelleted and resuspended in 100 μl PBS containing 1X CCF2-AM loading solution ( Invitrogen ) . Cells were incubated in the dark for 1 hr at room temperature . Cells were then pelleted and incubated with 10% rat serum ( 1:4 final dilution ) ( Sigma-Aldrich , St . Louis , MO ) and anti-CD16/32 ( 1:4 final dilution ) in fluorescence-activated cell sorting ( FACS ) buffer ( 1% BSA , 1 mM EDTA in PBS ) for 15 min on ice . To identify individual cell types , cells were incubated in the dark with cell-discriminatory antibodies at appropriate dilutions ( see below ) in 125 μl FACS buffer for 30 min . Cells were pelleted , fixed in 3 . 7% paraformaldehyde in PBS for 2 min , and washed twice with FACS buffer . Final cell suspensions were filtered through 80-μm nylon mesh ( Small Parts Inc . , Miami Lakes , FL ) into 12 x 75 mm round-bottom tubes ( BD Falcon ) . Final antibody dilutions were as follows: anti-CD16/32 , 1:50; anti-CD45 , 1:1 , 500 , anti-Gr1 , 1:1 , 500; anti-F4/80 , 1:50; anti-CD4 , 1:500; anti-CD8 , 1:500; anti-CD19 , 1:500; anti-CD49 , 1:500; anti-CD11b , 1:2 , 500; anti-CD11c , 1:500; and isotype controls , 1:100 each . All antibodies were purchased from eBioscience ( San Diego , CA ) . Analysis was performed on the BD FACSCanto II flow cytometer ( Becton , Dickinson and Company ) and analyzed using FlowJo version 8 . 8 . 6 software ( Tree Star , Inc . , Ashland , OR ) . Immune cells were gated as follows: CD11b+Gr1hi , neutrophils; CD11b+Gr1int , recruited monocytes; Gr1-F4/80+ , resident macrophages; CD11cintGr1- , dendritic cells; CD4+ , helper T cells; CD8+ , cytotoxic T cells; CD19+ , B cells; and CD49+ , NK cells . The total number of viable inflammatory cells per mouse lung was determined by equating the number of viable inflammatory cells measured on the flow cytometer ( using forward and side scatter properties ) to the number of trypan blue-negative cells counted on the hemocytometer . Lungs were excised and instilled with 800 μl of 6X CCF2-AM loading solution ( Life Technologies , Carlsbad , CA ) and then placed in a vial of 3 ml 6X CCF2-AM loading solution for 1 hr at room temperature in the dark . For cell viability experiments , LIVE/DEAD Fixable Red Dead Cell stain ( Life Technologies ) was added to the CCF2-AM mixture at a final dilution of 1:500 . The lungs were then moved to 10 ml of 4% paraformaldehyde and incubated overnight at room temperature . Lungs were cryoprotected by incubation in 15% sucrose solution for 8 hr and 30% sucrose solution overnight . Lungs were frozen using Clear Frozen Section Compound ( VWR , Radnor , PA ) in a dry ice/isopentane bath . Sections ( 6 μm thickness ) were cut by the Mouse Histology and Phenotyping Core of the Robert H . Lurie Comprehensive Cancer Center at Northwestern University . Slides were stored at -80°C prior to additional staining . For visualization of cell types , slides were acclimated to room temperature and blocked by incubation in 10% mouse serum , 1% BSA in Tris-buffered saline ( TBS ) for 2 hr . Slides were then incubated overnight at 37°C in a humidified chamber with 1:1 , 000 dilutions of primary antibodies in 1% BSA in TBS . Primary antibodies to caveolin-1 ( Abcam , Cambridge , England ) , pSP-C ( Abcam ) , or Gr1 ( Abcam ) were used to detect type I pneumocytes , type II pneumocytes , and phagocytes ( neutrophils and monocytes ) , respectively . ( Note that Gr1- resident macrophages comprise a very small fraction of phagocytic cells in the lungs at 12 hr and later during infection [22] . ) For detection of type I and type II pneumocytes , slides were washed twice with 0 . 05% Triton X-100 in TBS for 5 min and then incubated with AlexaFluor 555-conjugated goat anti-rabbit secondary antibody ( Molecular Probes , Eugene , OR ) diluted 1:1 , 000 in 1% BSA in TBS for 1 hr at 37°C . No secondary antibody was necessary for the detection of phagocytes , as the primary Gr1 antibody was conjugated to Cy5 . For visualization of bacteria , slides were blocked as described above and then incubated with 1:1 , 000 diluted primary antibodies to clinical isolate PA99 [56] for 2 hr at 37°C . Slides were washed twice with 0 . 05% Triton X-100 in TBS for 5 min and then incubated with 1:1 , 000 diluted AlexaFluor 555-conjugated goat anti-rabbit secondary antibody for 1 hr at 37°C . Slides were washed twice for 5 min with 0 . 05% Triton X-100 in TBS , air-dried , and mounted using ProLong Gold antifade mounting media ( Molecular Probes ) . Images were acquired at 200X using the TissueFAXS imaging system ( TissueGnostics , Vienna , Austria ) located in the Cell Imaging Facility at Northwestern University . Analysis was performed using the TissueQuest 4 . 1 software ( TissueGnostics USA Ltd . , Tarzana , CA ) . This technique employs a cellular reconstruction algorithm called microscopy-based multicolor tissue cytometry [62 , 63] . In this algorithm , high and low fluorescent intensity thresholds are used to filter each pixel . If a pixel falls within these thresholds , it is considered valid , and it is grouped with all directly adjacent valid pixels to form an object . The object is enlarged by continued assessment and addition of adjacent pixels until no further adjacent pixels are found that meet the criteria for validity . At that point , the resulting object is designated a cell [62] . Pixels are evaluated across an entire lung cross-section . In this way , all cells in a lung cross-section are counted . At least four mice were analyzed per time point , with two lobes chosen at random from each mouse . Due to limitations of the fluorescence spectral overlap of CCF2-AM , only one primary antibody could be imaged along with CCF2-AM per slide . Three adjacent 6 μm lung sections per mouse lung were therefore used to analyze for injected type I pneumocytes , type II pneumocytes , and phagocytic cells . Lung sections infected with PA99null+S ( no β-lactamase tag ) were used to determine baseline blue:green fluorescence ratio thresholds for each cell type ( ≥ 1 . 3 for type I pneumocytes and phagocytic cells; ≥1 . 1 for type II pneumocytes ) , since all cells in these sections lack blue fluorescence from cleaved CCF2-AM . Injected cells from lung sections infected with bacteria secreting β-lactamase-tagged ExoS were identified as those cells with a blue:green fluorescence ratio exceeding these baseline thresholds . In this way , the total population of ExoS-injected cells in a tissue cross-section of an entire lung lobe was determined . These cells were then subdivided into populations of ExoS-injected cells staining positive for each individual cell type marker . A schematic overview of this process is provided in S2 Fig . TissueQuest software was used to analyze the spatial distribution of ExoS-injected cells within lung sections . The average area of FOCI in a lung section was calculated by manually drawing a border around each region of the tissue section that contained a high density of injected cells , calculating the area of each FOCI in the lung section , summing all these areas , and dividing the total area by the total number of FOCI in the lung section . To quantify the number of dead cells , the percentages of injected and uninjected cells that stained positive with the LIVE/DEAD Fixable Red Dead Cell stain were determined . TissueQuest software was used to count the total number of ExoS-injected and non-injected cells per lung lobe . Then the number of those cells also positive for the DEAD stain was determined to calculate the final percentages of dead cells . To determine which cell types were dead , slides were stained with antibodies that recognized cell discriminatory markers as described above . Imaging was performed on a Zeiss UV LSM510 confocal microscope in the Nikon Cell Imaging Facility at Northwestern University . Mice were anesthetized and intranasally administered 50 μl FITC-conjugated albumin ( Sigma ) at a concentration of 50 mg/ml in 0 . 9% saline . Two hours after the FITC-albumin administration , mice were anesthetized and sacrificed by cervical dislocation . Blood was collected by cardiac puncture , and BALF was collected by twice instilling and withdrawing 1 ml PBS . The collected blood was centrifuged at 4000 x g for 10 min to separate plasma from red blood cells . FITC fluorescence of the BALF and plasma were measured using a SpectraMax M3 fluorescence plate reader ( Molecular Devices , Sunnyvale , CA ) . A standard curve of FITC-albumin fluorescence vs . concentration was generated and used to determine the FITC-albumin concentration in each sample . The ratio of FITC-albumin concentration in the blood to that in the BALF was calculated . Statistical analysis was performed using GraphPad Prism 6 ( GraphPad Software , Inc . , La Jolla , CA ) . An analysis of variance ( ANOVA ) followed by Bonferroni correction for multiple comparisons was performed for injection , foci measurement , cell viability , and pulmonary-vascular leakage experiments . The Kruskal-Wallis comparison of medians followed by Dunn’s test for multiple comparisons was performed for blood CFU experiments . Fisher’s exact test was performed on frequencies of dissemination . A p-value of 0 . 05 was used as a threshold for significance .	Dissemination to the bloodstream is a poor prognostic sign in patients with hospital-acquired pneumonia , yet the mechanism by which this occurs is poorly understood . To begin to address this issue , we have used a mouse model of P . aeruginosa pneumonia to study the mechanism by which the type-III-secreted effector protein ExoS enhances bacterial dissemination . We show that intoxication of type I pneumocytes by ExoS leads to cell death and disruption of the pulmonary-vascular barrier , allowing bacterial dissemination into the bloodstream . These effects required the ADP-ribosyltransferase activity of ExoS , as strains secreting an ExoS variant lacking this activity demonstrated reduced type I pneumocytes death and pulmonary-vascular breakdown . This study indicates that inhibitors of the ADP-ribosyltransferase activity of ExoS could serve as novel therapeutics for the prevention of bacteremic pneumonia .	[ "Abstract", "Introduction", "Results", "Discussion", "Materials", "and", "Methods" ]	[]	2015	The Role of ExoS in Dissemination of Pseudomonas aeruginosa during Pneumonia
Toxoplasma gondii is an obligate intracellular parasite for which the discharge of apical organelles named rhoptries is a key event in host cell invasion . Among rhoptry proteins , ROP2 , which is the prototype of a large protein family , is translocated in the parasitophorous vacuole membrane during invasion . The ROP2 family members are related to protein-kinases , but only some of them are predicted to be catalytically active , and none of the latter has been characterized so far . We show here that ROP18 , a member of the ROP2 family , is located in the rhoptries and re-localises at the parasitophorous vacuole membrane during invasion . We demonstrate that a recombinant ROP18 catalytic domain ( amino acids 243–539 ) possesses a protein-kinase activity and phosphorylate parasitic substrates , especially a 70-kDa protein of tachyzoites . Furthermore , we show that overexpression of ROP18 in transgenic parasites causes a dramatic increase in intra-vacuolar parasite multiplication rate , which is correlated with kinase activity . Therefore , we demonstrate , to our knowledge for the first time , that rhoptries can discharge active protein-kinases upon host cell invasion , which can exert a long-lasting effect on intracellular parasite development and virulence . Toxoplasma gondii is an obligate intracellular parasite belonging to the protozoan phylum Apicomplexa , which includes a large number of human and animal parasites responsible for diseases such as malaria , toxoplasmosis , coccidiosis , and cryptosporidiosis . As for all other members of the phylum , host cell invasion by T . gondii involves specialized apical organelles of the invasive stage , namely micronemes and rhoptries , which discharge their contents successively [1 , 2] . The exocytosis of micronemal proteins is associated with gliding and attachment to the host cell [3–6] . Then , a complex of microneme and rhoptry neck proteins forms a moving junction with the host cell plasma membrane that propels the parasite within the developing parasitophorous vacuole [7 , 8] . Subsequently , proteins of the bulb of the rhoptries ( ROP proteins ) become associated with the parasitophorous vacuole membrane ( PVM ) that forms from host plasma membrane and rhoptry components during invasion [9] . Among rhoptry proteins is a series of related proteins , the ROP2 family [10–12] , named after the ROP2 protein , which is translocated into the PVM during invasion [13] . The N-terminal ( Nt ) domain of ROP2 has been shown to interact with the mitochondrial import machinery and to mediate the association of host mitochondria to the PVM [14] . Targeted depletion of ROP2 using a ribozyme-modified antisense RNA strategy results in disruption of rhoptry biogenesis and affects cytokinesis , association of host cell mitochondria with the PVM , host cell invasion , and virulence in mice [15] . Several other members of the family have been characterized more recently , and they are also targeted to the PVM upon invasion [16–18] . The importance of ROP2 and the fact that the parasite is synthesizing simultaneously several ROP2 homolog proteins suggest that these proteins serve crucial functions; yet , the apparent indispensability of ROP2 suggests that they may not complement one another and may have distinct functions . We have recently shown that the ROP2 family could be expanded to at least 12 members , some of which show a full set of features compatible with protein-kinase activity , whereas ROP2 and its closest relatives have lost some of these features [12] . This raises the question of the role played by these proteins . Indeed , parasitic kinase ( s ) acting on host cell inhibitor of nuclear factor κB ( IκB ) have been suggested to be present at the PVM level [19] . Thus , T . gondii could be capable of manipulating the host cell machinery using its own kinases to favour its survival and development . Recently , many investigations have focused on searching protein-kinases in unicellular parasites , based on the fact that the vast phylogenetic distance between the organisms and their vertebrate hosts may have generated divergences in the properties of their protein-kinases that could be exploited for specific inhibition of the parasite enzymes [20–23] . This has prompted us to study the new members of the ROP2 family predicted to possess a fully functional protein-kinase domain . We report here the cloning and characterization of ROP18 , a novel ROP2-related rhoptry protein that is translocated to the PVM during invasion . We show that ROP18 is a protein-kinase that can phosphorylate a tachyzoite substrate of 70 kDa . We therefore demonstrate that rhoptries can discharge active kinases at the parasite–host cell interface upon cell invasion . In addition , overexpression of ROP18 in tachyzoites led to a dramatic stimulating effect on intracellular parasite multiplication , strongly suggesting that this protein plays a role in the control of parasite proliferation and may therefore be involved in T . gondii virulence . The open reading frame ( ORF ) included in the expressed sequence tag ( EST ) Cluster 100121072 ( APIDBest , http://www . apidb . org/apidb ) corresponding to the ROP18 protein has been amplified from T . gondii genomic DNA and sequenced [12] . The ROP18 protein–deduced primary sequence aligns with the ROP2 sequence with 25% identity ( Figure 1 ) . ROP18 is more closely related to ROP5 ( 28% ) than to the other ROP2 prototypes , such as ROP2 , 4 , 7 , and 8 . As other ROP2 family proteins , ROP18 contains an Nt peptide signal sequence , with a predicted cleavage site between residues 28 and 29 ( the second Met of the ORF has been considered as the start codon by homology to ROP2 , and used as position 1 ) or between residues 32 and 33 , according to SignalP . Almost all rhoptry proteins described so far in T . gondii , including members of the ROP2 family proteins ( ROP2 , 4 , and 7 ) , are synthesized as pro-proteins that are subjected to proteolytic cleavage during trafficking to rhoptries removing the Nt pro-region [10 , 17 , 24] . The exact site of cleavage for the ROP2 family proteins is unknown , but the sequence ( SWLE ) present at the end of the pro-domain of ROP2 , ROP4 , and ROP8 has been suspected to be the cleavage site by the maturase TgSUB2 ( the proposed consensus being SΦXE , where Φ represents bulky hydrophobic residues and X is any amino acid [25] ) . ROP5 lacks this sequence and is not processed [18] . ROP18 contains at the same location the sequence SLLE , and may therefore be cleaved after amino acid 82 . Following the predicted cleavage site , several arginine-rich stretches are observed in ROP2 family proteins , including ROP2 , 4 , 5 , 7 , and 8 [12] . Two such arginine-rich segments are clearly recognized in ROP18 sequence at positions 101–113 and 129–142 , and a third one at 152–163 is more degenerate . However , the precise role of these basic and amphipathic segments remains unknown , although they may serve to anchor proteins onto membrane surfaces . These stretches are followed by a linker region ( residues 173–233 ) whose function and structure are unknown . A putative serine/threonine protein-kinase domain in the C-terminal ( Ct ) half of the ROP18 sequence was identified by PSI-BLAST search [26] with significant e-value ( below e−7 ) . However , this domain comprises a hydrophobic stretch conserved in other ROP2 family proteins that has been previously considered as a transmembrane segment . The corresponding segment in ROP18 is weakly predicted by TopPred as a putative transmembrane segment ( 443 and 463 ) with a score of 0 . 614 . Our recent sequence analysis of all ROP2 family sequences has led us to rule out the transmembrane prediction for this conserved segment [12] . We rather predicted that the Ct region ( 234–539 in ROP2 ) adopts a protein-kinase fold . In order to confirm this hypothesis , molecular modelling was performed on ROP18 . Comparative modelling was initiated using fold-recognition through the meta-server @TOME [27] . Significant scores of fold-compatibility were obtained with various serine/threonine kinases ( see results at http://www . infobiosud . cnrs . fr/bioserver/ROP/suppl . html ) despite a low overall sequence identity ( ∼20% over the whole Ct domain ) . Molecular modelling of this domain ( Figure 2 ) further demonstrated the conservation of the protein-kinase fold , especially all the residues critical for the domain stability and the protein-kinase activity [28] . The hydrophobic segment appears to be completely buried inside the helical domain of the protein core and to bear residues essential for protein stability ( including D450 , W452 , and G455 ) . Indeed , it corresponds to the Hanks motif “DxxxxG” numbered as IX . Among the other conserved motifs , those involved in catalysis , regulation , and peptide recognition were further scrutinized using the theoretical models . Motifs I , II , VIb , VII , and VIII as defined by Hanks [28] were clearly detected . The catalytic lysine ( K266 in ROP18 ) and aspartate ( D394 ) residues were present . The region 427 to 435 perfectly matched the Hanks motif VIII of serine/threonine kinases . ROP18 is unique among ROP2-like proteins in having this peptide-binding motif perfectly conserved . The sub-sequence GTP ( 427-GTP-429 in ROP18 ) is expected to recognize serine or threonine residues to be phosphorylated . This functional prediction was confirmed by the sequence of the motif VIb ( 392-HTDIKPAN-399 in ROP18 ) [29] . This motif bears a consensus sub-sequence , “KpeN , ” that is specific to serine/threonine kinases ( versus aarN in tyrosine kinases ) . The absence of an arginine at the second residue position of this motif in ROP18 ( bearing a threonine T393 instead ) suggests that ROP18 does not need phosphorylation of its activation loop to become active . These in silico predictions were further supported by experimental data gained on another member of the ROP2 protein family sharing the same predicted structural features [12] . Indeed , refolded recombinant ROP18 was rather unstable and could not be obtained in sufficient amounts , whereas previous work had shown that ROP2 could be obtained directly as a stable soluble recombinant protein [30] . Dynamic light scattering ( DLS ) spectra of both recombinant proteins confirmed their size similarity in solution ( Figure S5 ) . The existence of a soluble and compact domain was evaluated by small angle X-ray scattering ( SAXS ) experiments performed on a recombinant ROP2 ( 196–561 ) construct at high protein concentration ( up to 18 mg/ml ) . SAXS data on ROP2 were recorded to a maximum resolution of s = 4 . 63 nm−1 ( Figure S2 ) . Low resolution data showed only minor signs of aggregations , and the Guinier plot followed a straight line between s*Rg limits of 0 . 87 and 1 . 25 ( Rg being the radius of gyration ) . The Rg calculated from the Guinier plot ( 3 . 90 nm ) was in very good agreement with the one obtained by GNOM ( 3 . 89 nm ) . GNOM analysis also indicated the maximum particle diameter to be 13 . 5 nm . By comparison with the I0 value of a BSA standard , a molecular weight of 37 kDa was estimated for recombinant ROP2 , in reasonable agreement with a calculated molecular weight of 42 kDa for a monomer . Ab initio shape calculations yielded two-lobed 40 × 40 × 65 Å ellipsoidal structures with an ∼75 Å tail , a form which is already apparent from the distance distribution ( Figure S3 ) . The ellipsoidal structure compares very well in shape and size with a typical protein-kinase domain ( Figure S4 ) . SAXS and homology modelling data suggest that , going from Ct to Nt , the 40 N-terminal residues additional to the protein-kinase domain first run over one side of the C-terminal lobe ( up to residue 40–35 ) , and then form an unstructured and solvent-exposed extension . Trp39 might be important for pinning the Nt to the C-terminal lobe . The scattering curve calculated from a homology model of ROP2 fits the experimental SAXS data well , given the only modest sequence homology ( Figure S2 ) . Altogether , SAXS data corroborate that recombinant ROP2 consists of a protein-kinase domain with an unstructured 40 residue–N-terminal tail . The size observed for freshly refolded ROP18 by DLS ( see below ) and the significant sequence similarity shared by ROP18 and ROP2 suggests that these two proteins ( as well as other ROP2-like proteins ) possess the same structural organization containing a folded protein-kinase domain . Similarly , ultraviolet circular dichroism ( UV-CD ) measurement ( wavelength 195–260 nm; unpublished data ) suggested that both ROP18 and ROP2 proteins are mainly composed of alpha-helices in agreement with their predicted fold . In conclusion , ROP18 is predicted to be composed of an N-terminal domain of unknown structure , while the large C-terminal domain would be folded as a soluble and functional serine/threonine protein-kinase . To investigate the expression and subcellular localization of ROP18 , we raised a polyclonal serum against a recombinant ROP18 protein . On immunoblots of tachyzoites , anti-ROP18 antibodies reacted with a single protein with an apparent molecular mass of 55 kDa ( Figure 3 ) . The specificity of the antibodies was assessed by Western blotting transgenic parasites expressing a Ty-tagged version of ROP18 . When ROP18-Ty tachyzoites were analyzed , two bands at 56 and 60 kDa were found with anti-Ty monoclonal antibody ( mAb ) ( Figure 3 ) . The same bands were detected when probing with anti-ROP18 , together with the 55-kDa band , confirming the antibody specificity . The mobility shift observed for the 56-kDa band is consistent with the addition of a Ty-1 epitope . The 60-kDa band was interpreted as unprocessed ROP18-Ty protein ( see below ) . When analyzed by immunofluorescence assay ( IFA ) on T . gondii–infected human foreskin fibroblasts ( HFFs ) , ROP18 was found at the apical end of RHΔhx ( HX ) tachyzoites , co-localizing exactly with ROP1 ( Figure 4 , HX ) . In transgenic parasites expressing ROP18-Ty , the anti-Ty antibodies labelled the rhoptries ( Figure 4 , R18Ty ) , and , in some cases , vesicles located between the rhoptries and the nucleus ( not shown ) . To determine whether ROP18 is processed during trafficking to rhoptries , we studied the biosynthesis of ROP18 by pulse-chase metabolic labelling with [35S] methionine . For this analysis , the transfected strain ROP18-Ty was used . After a 20-min pulse , one major labelled protein of 60 kDa and a minor of 56 kDa were immunoprecipitated by mAb anti-Ty ( Figure 5 ) . The 56 kDa was strongly enriched when a 1-h chase was performed , suggesting that ROP18 is processed in the biosynthetic pathway . When compared with ROP2-ROP4 immunoprecipitated on mAb T3 4A7 , the mature 56-kDa form of ROP18-Ty appeared slightly earlier than mature ROP2-ROP4 . Whether this is due to the expression of ROP18-Ty under a tubulin promoter rather than under its native one , or to different kinetics of trafficking in the pathway , remains to be established . The persistence of unprocessed ROP18-Ty observed on Western blots ( Figure 3 ) could also be explained by some untimely synthesis due to the tubulin promoter that may lead to accumulation in vesicles located between the rhoptries and the nucleus , or trafficking to compartments that do not contain the processing enzymes when rhoptries are not being produced . This demonstration of proteolytic processing , together with the presence of the SLLE motif at the expected cleavage site in the sequence , tends to reinforce the hypothesis of TgSUB2 being the processing enzyme [25] , as the only unprocessed member of the family known so far is ROP5 , which lacks this motif . We then investigated the fate of ROP18 during HFF invasion . Rhoptries are discharged during the invasion process [2 , 9 , 13] , and their contents associate with the nascent vacuole membrane . When invasion is interrupted with cytochalasin-D ( Cyt-D ) , rhoptry-derived vesicles named evacuoles accumulate in the host cell cytoplasm [31] . When Cyt-D–arrested parasites were labelled with anti-ROP18 , we found that ROP18 was associated with evacuoles ( Figure 6 , evac ) . ROP18 was also associated with the PVM of invaders and of recently invaded parasites , co-localizing with ROP1 ( Figure 6 , inv1 and inv2 ) . The tropism of ROP18 for the PVM was also confirmed upon infection of BHK21 cells that transiently expressed ROP18 ( Figure 6 , BHK ) . The nucleotide coding sequence corresponding to mature ROP18 ( ROP18ΔPro , amino acids 83–539 , by deletion of the peptide signal and putative propeptide ) was cloned in frame with a sequence coding for the V5 epitope , in the mammalian expression vector pTracer-A , which allows co-expression of the sequence of interest and of the green fluorescent protein ( GFP ) . GFP is expressed in the cytoplasm and its intrinsic fluorescence allows direct visualization of transfected cells . In these cells expressing ROP18ΔPro , anti-V5 antibodies produced a punctuate labelling homogeneously distributed in the cytosol ( Figure 6 , BHK , upper row ) . When these cells were infected with tachyzoites 4 h after transfection and fixed 16 h after infection , the anti-V5 labelling was found prominently around the PVM , with some extension in the parasitophorous vacuole that may correspond to the PVM-derived intra-vacuolar network ( Figure 6 , BHK , lower row; Figure S1 ) . In contrast , the distribution of GFP was unchanged . Control PVM in non-transfected cells ( GFP-negative cells ) were not labelled ( not shown ) . These results indicate that ROP18 possesses a strong affinity for the PVM . Collectively , these results demonstrate that ROP18 is a rhoptry protein secreted during the invasion process that associates with the PVM-surrounding intracellular parasites . Since ROP18 showed a full set of features compatible with protein-kinase activity , we investigated the predicted catalytic properties of this protein . We therefore expressed the catalytic domain with a Ct His-tag in Escherichia coli . The recombinant protein was found in bacterial inclusion bodies in the various conditions tested so far ( unpublished data ) . We therefore used a denaturation-refolding procedure to purify the recombinant protein . Refolding was monitored by light scattering ( Figure 7A ) and was confirmed by tryptophan fluorescence ( unpublished data ) . The refolded protein was incubated with either heat-inactivated parasite or HFF lysate and assayed for kinase activity ( Figure 7B ) . A major phosphorylated band of 70 kDa and a minor one of 68 kDa were detected in parasite ( Figure 7B , lane 2 ) , but not in HFF lysate ( Figure 7B , lane 4 ) . Autophosphorylation by the refolded kinase was not observed ( unpublished data ) . To confirm that the observed phosphorylation was due to the catalytic activity of ROP18 , we generated a recombinant catalytic domain with a mutation of aspartic acid D394 ( domain VIb , according to Hanks [28] ) , required for the activity , to an alanine . We expressed and refolded the mutated catalytic domain as done for the native catalytic domain and compared the activity of equal amounts of both proteins . The kinase assay with the mutated protein on parasite extracts did not lead to any significant labelling ( Figure 7B , lanes 1 and 3 ) , clearly demonstrating that mutation of the ROP18 catalytic aspartic acid D394 to an alanine leads to the loss of kinase activity . Cultivation of the transfected ROP18-Ty strain routinely showed an earlier release of parasites compared with that of wild-type . We therefore investigated whether this was due to higher invasion rate or faster intracellular multiplication . To make sure that a position effect of the transformation was not involved , we duplicated the experiments with parasite clones isolated from two independent ROP18-Ty transfections . Evaluating the invasion rate of the various parasites did not show any significant difference between wild-type and transfected clones ( not shown ) . In contrast , when counting the number of parasites per vacuole at 16 h after infection , we observed a significant increase in parasite proliferation in the ROP18-Ty–transfected parasites , compared with that of wild-type tachyzoites . The results of five experiments led to a mean reproduction rate of 2 . 47 ± 0 . 39 parasites per vacuole at 16 h after infection for wild-type , whereas the ROP18-Ty showed a rate of 4 . 07 ± 0 . 34 ( Figure S6 ) . A statistical analysis of these data using Student's t-test led to a p-value of 0 . 001 . In order to know whether this property was related to the enzymatic activity of ROP18 , we created transfectants expressing a D394A-mutated ROP-18Ty . The ROP18-TyD394A localization was verified by IFA . As expected , the protein was present in the rhoptries and secreted during the invasion process . By counting the number of parasites per vacuoles in three experiments , we showed that the multiplication rate of ROP18-TyD394A–transfected parasites at 16 h ( 2 . 40 ± 0 . 45 ) was not significantly different from wild-type ( Figure S6 ) . A graph of one typical experiment showing the relative distribution of the number of parasites per vacuole at 16 h post-infection is shown in Figure 8A . The level of expression of ROP18-Ty and ROP18-TyD394A proteins was verified to be equivalent by quantification on Western blots of equal amounts of parasites ( Figure 8B ) . In order to evaluate whether the presence of an overexpressor in a cell could influence the reproduction rate of a co-infecting wild-type parasite , we performed a co-infection experiment with HX- and ROP18-Ty–transfected parasites . The two types of vacuoles were differentiated by IFA on the Ty-tag expression . This experiment showed that there was no influence of co-infection , with both parasite types behaving as in the single infections experiments , regardless of the presence of the other type in the same cell ( Figure 8C ) . Therefore , we obtained evidence that overexpression of an active ROP18 kinase leads to an increase in the rate of parasite replication , but that this effect is restricted to the vacuole containing these parasites . We have identified an active kinase stored in T . gondii tachyzoites rhoptries , which is secreted in the host cell during invasion and is involved in the intracellular proliferation of the parasite . The simultaneous expression by the parasite of such a family of closely related proteins is still poorly understood . All of these proteins possess a region sharing significant homologies with the canonical kinase domain as described by Hanks [12 , 28] . Some of them , like ROP4 , 5 , 7 , and 8 , lack the glycine loop and the catalytic aspartic acid required for activity . Others , like ROP2 , lack the glycine loop , although having kept the catalytic loop , they may still interact with a substrate without being able to phosphorylate . Remarkably , several other members of the family possess the complete set of features needed for kinase activity , which led us to investigate them further . We focused our attention on ROP18 . The characteristic features of the family , such as rhoptry location , ORF size , hydrophobic segment near Ct , and arginine-rich stretches near the Nt [12] , are well conserved in ROP18 . We could express and refold as an active protein-kinase its complete Ct domain ( 243–539 ) . We showed that it was indeed capable of phosphorylating parasite proteins . These findings demonstrate directly the presence of an active kinase in T . gondii rhoptries . In T . gondii lysate , a major 70-kDa protein and a minor one of 68 kDa are phosphorylated; these sizes do not correspond to any parasitic protein characterized so far . Moreover , we do not know whether the negative result obtained with HFF lysates corresponds to a total absence of activity on host cell substrates , or to a defect in the experimental procedure impairing the activity . We and others have shown that several members of the ROP2 family are translocated to the parasitophorous membrane upon invasion [13 , 16–18] . ROP18 follows the same route . In addition , we show that , when expressed in the cytoplasm of the host cell , ROP18 also homes to the PVM , suggesting a specific interaction with this membrane . In this location , ROP18 could modify other PVM proteins ( such as another rhoptry protein or a dense granule protein ) or signal and/or control host cell functions . What phosphorylation ( s ) occurs is yet to be identified , but two related events have already been described , namely the phosphorylation of ROP4 on several serine/threonine residues after translocation in the PVM [16] , and the phosphorylation of host IκB that correlates with the activation of NF-κB , which is required for the inhibition of apoptosis [19] . As several other ROP proteins are also putative rhoptry kinases [12 , 32] , the parasite is likely to modulate several host cell function or PVM properties soon after entry . The harnessing of the host cell by T . gondii was demonstrated by Blader et al . [33] , who showed a wide range of changes in host cell transcription pattern after parasite invasion . Our observation that overexpression of ROP18 increases parasite proliferation rate , with this property being strictly linked to the protein-kinase catalytic activity , fits perfectly with these data . Such a shortening of the parasite cell cycle triggered by overexpression of a parasite protein has not been described so far . It tends to suggest that the mutants are metabolically more efficient either by activating the cell metabolism for their benefit , or by getting their supply from the host cell more efficiently . A modification of the PVM would fit with the second possibility , which is consistent with our observation that the effect of overexpression does not extend to other vacuoles in the same cell . In addition , as the length of the cell cycle differs between T . gondii strains , and as more virulent strains have higher multiplication rates , a direct connection could exist between expression of ROP18 and virulence . Such a correlation has actually been independently observed by genetic mapping of T . gondii virulence [34 , 35] . In conclusion , we have shown here that ROP18 is a protein-kinase belonging to the ROP2 family of rhoptry proteins . To our knowledge , we have provided the first direct demonstration of the presence of an active kinase in T . gondii rhoptries; in addition , we have shown a direct effect of the expression of this protein on the proliferation rate of the parasite , suggesting a possible role in virulence , which expands the part played by the ROP2 family proteins in the biology of T . gondii . All parasites were maintained by serial passage in HFFs grown in Dulbecco's modified Eagle medium ( DMEM ) ( GibcoBRL , http://www . invitrogen . com ) supplemented with 10% fetal calf serum ( FCS ) and 2 mM glutamine . Tachyzoites of the RH strain of T . gondii [36] and of HX deleted for hypoxanthine guanine phosphoribosyl transferase [37] were used throughout the study . BHK-21 ( baby hamster kidney ) cells ( ATCC CCL-10 ) were grown in BHK-21 medium ( GibcoBRL ) supplemented with 5% FCS , 2 mM tryptose , 100 U/ml penicillin , and 100 μg/ml streptomycin . Invasion and intracellular tachyzoite multiplication rates were measured on HFFs plated on 12-mm coverslips in 24-well plates and fixed 16 h after infection by equal numbers of freshly released tachyzoites . In some experiments , uninvaded parasites were washed 30 min after contact with the cells to avoid possible bias due to differences in kinetics of invasion . Coverslips were fixed and stained with eosine-methylene blue ( RAL 555 ) and then mounted permanently ( Pertex; Microm Microtech France , http://www . microm . fr ) . Fields were randomly selected , and the number of vacuoles per field and the number of parasites per vacuole were counted using a 40× objective in ten fields per coverslip , with three coverslips per assay . Five independent experiments were performed . Data were analyzed using Student's t-test . A p-value less than 0 . 05 was regarded as significant . In one experiment , coverslips were co-infected simultaneously by both wild-type and transfected parasites , and counts were performed after anti-TY IFA ( see below ) to differentiate between both types and compare their respective proliferation in single and double infections . Antibodies used in this study included mAb anti-Ty-1 tag [38] , mAb T3 4A7 specific for ROP2 , 3 , and 4 [10] , anti-recombinant ROP1 and ROP2 rabbit sera ( J . F . Dubremetz and O . Mercereau-Puijalon , unpublished data ) , and a rabbit anti-ROP18 obtained by rabbit immunization ( see below ) . Preliminary genomic and/or cDNA sequence data was accessed via ToxoDB ( http://www . toxodb . org ) and/or the Toxoplasma gondii Genome Project ( http://www . tigr . org/tdb/t_gondii ) . The ROP18 cloning was based on the EST cluster ( 100121072 ) found in ToxoDB APIDBest ( http://www . apidb . org/apidb ) . The ROP18 gene was PCR-amplified from genomic DNA with primers HH32 ( 5′-GTGATGTTTTCGGTACAGCGGCCA-3′ ) and HH33 ( 5′-CTTTTATTCTGTGTGGAGATGTTC-3′ ) , and subcloned into a PCR blunt II Topo vector ( Invitrogen , http://www . invitrogen . com ) , to generate pROP18 . The plasmid pROP18-Ty was designed to express a Ct Ty-tagged ROP18 protein in RH tachyzoites . It was constructed by inserting the coding sequence of ROP18 under the control of the tubulin promoter ( TUB ) . The ROP18 gene coding sequence was PCR-amplified from pROP18 with forward primer HH51 ( 5′-ATG CAATTGATGTTT TCGGTACAGCGGCCA-3′; MfeI site underlined ) and reverse primer HH50 ( 5′-TGCATGCATGTTCTGTGTGGAGATGTTCCTG-3′; NsiI site underlined ) , and subcloned as an MfeI/NsiI fragment into pTUB8mycGFPPftailTY , which was generously given by D . Soldati . Plasmid pET-ROP18 was designed to express in E . coli a Ct His-tagged recombinant protein corresponding to the predicted kinase catalytic domain of ROP18 . The DNA sequence coding for amino acids 243–539 was amplified by PCR from the pROP18 plasmid using forward primer HH40 ( 5′-GGGTTTCATATGACTACCGGTGAAACCCGG-3′; NdeI site underlined ) and reverse primer HH41 ( 5′-AAATATGCGGCCGCTTCTGTGTGGAGATGTTC-3′; NotI site underlined ) and cloning into NdeI and NotI sites of pET-24a vector ( Novagen , http://www . emdbiosciences . com/html/NVG/home . html ) to generate pET-ROP18 . Plasmid pTracer-ROP18ΔPro was designed to express the mature ROP18 protein spanning amino acids 83 to 539 , such as the one stored in T . gondii rhoptries . It was constructed by PCR amplification of sequence from pROP18 plasmid using forward primer ML193 ( 5′-GCGGCCGCATGGAAAGGGCTCAACACCGGGTA-3′; NotI site underlined ) and reverse primer ML194 ( 5′-TCTAGATTCTGTGTGGAGATGTTCCTG-3′; XbaI site underlined ) and cloning into NotI and XbaI sites of pTracer-A 5 ( Invitrogen ) . All constructs were verified by sequencing . PSI-BLAST program [26] was applied with standard parameters to search for homologous proteins in the Swiss-Prot Translated EMBL ( SPTrEMBL ) and National Center for Biotechnology Information ( NCBI ) non-redundant sequence databases . Fold-compatibility for the full-length and truncated sequences of ROP18 was searched and evaluated as previously described for other ROP2-like proteins [12] . Domain organisation was refined using fold-recognition results ( see http://www . infobiosud . cnrs . fr/bioserver/ROP/suppl . html ) . Sequence–structure alignments , including ROP18 and its paralogs and distinct protein-kinases ( see alignment in [12] ) , were manually refined with the help of the program ViTO [39] . Improved three-dimensional models were built for using MODELLER 7 . 0 with the loop optimization procedure . SAXS data were collected from beamline X33 at Deutsches Elektronen-Synchrotron ( DESY ) , European Molecular Biology Laboratory ( EMBL ) Hamburg . Data were collected at 10 °C , using a wavelength of λ = 1 . 5 Å . ROP2 was overexpressed and purified as previously described [30] . ROP2 was used at a concentration of 12 . 3 mg/ml in 0 . 2 M KPO4 ( pH 8 ) . Prior to data recording , the samples were extensively centrifuged to eliminate aggregates , and supplemented with 2 mM DTT . Diffusion spectra for buffer only were taken before and after the protein sample , averaged , and subtracted from the protein scattering curve . Data analysis and ab initio shape calculations were performed using PRIMUS , GNOM , GASBOR , and DAMMIN [40] . The His-ROP18 recombinant protein was expressed in Plys E . coli ( Stratagene , http://www . stratagene . com ) that had been induced at 37 °C for 2 h with 1 mM isopropyl-β-D-thiogalactopyranoside . The bacterial pellets were resuspended in lysis buffer ( Tris-HCl 50 mM [pH 7 . 5] , NaCl 50 mM , EDTA 0 . 1 mM , and Complex mixture protease inhibitor tablet [Roche Applied Science , http://www . roche-applied-science . com] ) and cells were broken using a French press ( Thermo Spectronic; Thermo Scientific , http://www . thermo . com ) operated at 20 , 000 p . s . i . , then centrifuged at 12000g for 15 min; the pellet was washed with buffer ( Tris-HCl 50 mM [pH 7 . 5] , NaCl 50 mM , 0 . 1% Triton , and EDTA 0 . 1 mM ) . The protein was extracted from bacterial inclusion bodies by denaturation-refolding . Denaturation was performed in 6 M guanidinium chloride followed by ultracentrifugation of 100 , 000g for 30 min at 4 °C . The protein solution was then brought to 4 M guanidinium chloride . Refolding was obtained by a 10-fold dilution of the supernatant in the refolding buffer ( Tris-HCl 50 mM [pH 8 . 3] and CsCl 100 mM ) . Proper refolding was assessed by DLS using a Zetasizer NanoZS ( Malvern , http://www . malvern . com ) . The refolded His-ROP18 fusion protein was purified on Ni-NTA resin ( Qiagen , http://www . qiagen . com ) and eluted at 100 mM imidazole . The His-ROP18 D394A recombinant protein ( see below ) was expressed and refolded by the same procedure as the His-ROP18 recombinant protein . Specific antibodies were obtained by subcutaneous immunizations of a rabbit with 1 mg of the recombinant protein separated on sodium dodecyl sulfate polyacrylamide gel electrophoresis ( SDS-PAGE ) and transferred to nitrocellulose , which was then crushed in Freund's complete adjuvant and injected . Two further injections were performed at 3-wk intervals in Freund's uncomplete adjuvant . The rabbit was then bled and the specific antibodies were affinity purified on the recombinant protein electrophoresed and Western blotted on nitrocellulose . QuickChange Site-Directed Mutagenesis system ( Stratagene ) was used to introduce point mutation in the catalytic domain of ROP18 gene . The reaction was performed according to the manufacturer's instructions . Plasmid pET-ROP18 was used as template for construction of plasmid pET-ROP18-D394A where the triplet encoding the catalytic aspartic acid D394 has been changed to encode an alanine . The mutated catalytic domain sequence was obtained using primers HH54 ( 5′-ATTGTGCATACGGCTATCAAACCGGCG-3′ ) and HH59 ( 5′-CGCCGGTTTGATAGCCGTATGCACAAT-3′ ) . The same mutation was also introduced in plasmid pROP18-Ty using the same primers to generate plasmid pROP18-Ty D394A , further identified as pROP18-TyMUT . The presence of the expected mutations was verified by sequencing . Transgenic parasites expressing ROP18-Ty or ROP18-TyMUT were obtained by electroporation of either 30 μg of pROP18-Ty or pROP18-TyMUT into 107 HX tachyzoites as described previously [41] . After overnight growth , transfectants were selected with 25 μg/ml mycophenolic acid and 50 μg/ml xanthine , and cloned by limiting dilution under drug selection . Two independent transformation experiments were performed . Transfections were carried out using Lipofectamine Reagent ( GibcoBRL ) as instructed by the manufacturer with 3 × 105 BHK-21 cells grown on coverslips for 24 h in 6-well plates . After four h with Lipofectamine , cells were washed and incubated for four additional hours with complete BHK-21 medium . Then , the wells were infected with one million parasites for 16 h prior to fixation and immunofluorescence analysis . For IFAs on intracellular parasites , HFFs were seeded on coverslips and infected with tachyzoites 24 h before fixation . Infected cells were fixed with 4% paraformaldehyde in phosphate buffered saline ( PBS ) for 30 min at room temperature , washed and permeabilized with 0 . 1% Triton X-100 in PBS for 10 min , and blocked with 10% FCS in PBS for 10 min . Coverslips were subsequently washed in PBS , then incubated with primary antibodies for 30 min at room temperature . Dilutions were 1:200 for mAb anti-V5 , 1:100 for mAb anti-Ty and mAb T3 4A7 , 1:500 for rabbit anti-ROP1 , and 1:10 for rabbit anti-ROP18 affinity-purified antibodies . Coverslips were washed in PBS and then incubated with affinity-purified goat anti-mouse immunoglobulin G ( IgG ) conjugated to FITC ( Sigma , http://www . sigmaaldrich . com ) and with goat anti-rabbit IgG conjugated to RITC ( Jackson ImmunoResearch , http://www . jacksonimmuno . com ) at 1:500 for 30 min . Finally , coverslips were washed and mounted onto microscope slides using Immunomount ( Calbiochem , http://www . emdbiosciences . com/html/CBC/home . html ) . IFA of invading parasites were done as described previously [8] . Briefly , after 2 min of invasion at 37 °C , coverslips were fixed with 4% paraformaldehyde in PBS and infected cells were permeabilized with 0 . 05% saponin in PBS . IFA was performed as described above . Cyt-D treatment was by incubating with 1 μM of the drug before and during invasion as described previously [18] . All observations were performed on a Leica DMRA2 microscope ( Leica Microsystems , http://www . leica-microsystems . com ) equipped for epifluorescence; images were recorded with a CoolSNAP CCD camera ( Photometrics , http://www . photomet . com ) driven by Metaview ( Universal Imaging , http://www . moleculardevices . com ) and processed using Adobe Photoshop 7 . 0 ( Adobe Systems , http://www . adobe . com ) . Freshly released tachyzoites were boiled in SDS-PAGE sample buffer and separated on 10% polyacrylamide gels according to Laemmli [42] . Mr markers ( Bio-Rad , http://www . bio-rad . com ) were used for calibration . Proteins were transferred to nitrocellulose membranes ( Protran; Schleicher & Schuell ) at 0 . 8 mA/cm2 for 90 min by semi-dry transfer . The nitrocellulose strips were saturated for 1 h in 5% non-fat dry milk in 15 mM Tris-HCl ( pH 8 ) , 150 mM NaCl , and 0 . 05% Tween 20 ( TNT ) . They were then incubated with mAbs ( mouse ascitic fluids ) or with polyclonal rabbit antibodies diluted 1:500 in TNT for 1 h . After washing , the strips were incubated with alkaline phosphatase–conjugated anti-mouse diluted 1:1000 in TNT and stained with BCIP-NBT . Infected monolayers were solubilized in lysis buffer ( Tris-HCl 50 mM [pH 8 . 3]/ NaCl 150 mM/ EDTA 4 mM/ PMSF 1 mM/1% Nonidet 40 -NP40- ) for 1 h at 4 °C . The lysate was centrifuged 1 h at 16 , 000g , and the supernatant was collected for immunosorption . The immunosorbents were prepared by incubating 20 μl of ascitic fluid with 20 μl of Protein G-Sepharose for 1 h in 1 ml of PBS; they were then incubated with radiolabelled lysate at 4° C for 2 h under gentle agitation , washed four times with a buffer containing 1M NaCl and 0 . 5% NP40 in 50 mM Tris-HCl ( pH 8 . 3 ) and then in 5mM Tris-HCl ( pH 6 . 8 ) . Elution was then performed during 5 min at 95 °C with electrophoresis sample buffer . After SDS-PAGE , the gel was impregnated with Amplify ( Amersham , http://www . amershambiosciences . com ) , dried , and exposed to Biomax film ( Kodak , http://www . kodak . com ) at −80°C . Heavily infected HFF monolayers were incubated in methionine and cysteine-free DMEM ( Invitrogen ) containing 1% dialyzed FCS for 30 min at 37 °C in a 5% CO2 incubator prior to the addition of 50 μCi/ml [35S] methionine/cysteine ( 700 Ci/mM; MP Biomedicals , http://www . mpbio . com ) . The infected monolayers were then labeled for 30 min , rinsed with complete DMEM containing 10% FCS , and either arrested or incubated in this medium for 2 h chase prior to immunoprecipitation as described above . The assays were performed in a standard reaction buffer ( 30 μl ) containing 25 mM Tris-HCl ( pH 7 . 5 ) , 15 mM MgCl2 , 2 mM MnCl2 , 15 μM ATP , and 20 μCi of [γ-32P]ATP ( 4500 Ci/mM; MP Biomedicals and Qbiogene , http://www . qbiogene . com ) and the lysate of 107 parasites or of 105 HFFs that had been heated at 56 °C for 30 min to inactivate endogenous kinases . The reactions were initiated by addition of 10 μg each of the recombinant protein-kinase or of the recombinant mutated protein-kinase . The reaction proceeded for 30 min at 30 °C and was stopped by adding gel loading buffer and heating immediately at 95 °C for 5 min . Proteins were analyzed by electrophoresis on 12% SDS-polyacrylamide gel . The gels were dried and submitted to autoradiography . The GenBank ( http://www . ncbi . nlm . nih . gov/Genbank ) accession numbers for the sequences discussed in this paper are ROP2 ( CAA85377 ) , ROP4 ( CAA96467 ) , ROP5 ( DQ116423 ) , ROP7 ( AM056071 ) , ROP18 ( AM075204 ) , TgSUB2 ( AF420596 ) .	Apicomplexa are unicellular eukaryotes that cause a number of diseases , including malaria . Most of them are obligate intracellular parasites , developing in a parasitophorous vacuole ( PV ) within their host cell . PV formation during invasion is associated with the exocytosis of parasite secretory organelles named rhoptries , whose role is unknown . Toxoplasma gondii is a model Apicomplexa responsible for toxoplasmosis , a fatal congenital or opportunistic infection in humans and animals . We have studied a novel rhoptry protein dubbed ROP18 , which is translocated to the PV membrane upon invasion . ROP18 belongs to a family of rhoptry proteins that share homologies with serine-threonine kinases , but those described so far lack residues critical for enzyme activity . We show that ROP18 possesses all the features needed to be active , and we experimentally demonstrate this activity , which phosphorylates at least one parasite protein . We show that overexpression of ROP18 causes a dramatic increase in parasite multiplication rate that is correlated with kinase activity , and likely dependent on a PV membrane modification . We therefore demonstrate that rhoptries can discharge active protein-kinases upon invasion , which can exert a long-lasting effect on intracellular parasite development and virulence .	[ "Abstract", "Introduction", "Results", "Discussion", "Materials", "and", "Methods", "Supporting", "Information" ]	[ "eukaryotes", "cell", "biology", "microbiology" ]	2007	ROP18 Is a Rhoptry Kinase Controlling the Intracellular Proliferation of Toxoplasma gondii
Environmental and anthropogenic changes are expected to promote emergence and spread of pathogens worldwide . Since 2013 , human urogenital schistosomiasis is established in Corsica island ( France ) . Schistosomiasis is a parasitic disease affecting both humans and animals . The parasite involved in the Corsican outbreak is a hybrid form between Schistosoma haematobium , a human parasite , and Schistosoma bovis , a livestock parasite . S . bovis has been detected in Corsican livestock few decades ago raising the questions whether hybridization occurred in Corsica and if animals could behave as a reservoir for the recently established parasite lineage . The latter hypothesis has huge epidemiological outcomes since the emergence of a zoonotic lineage of schistosomes would be considerably harder to control and eradicate the disease locally and definitively needs to be verified . In this study we combined a sero-epidemiological survey on ruminants and a rodent trapping campaign to check whether schistosomes could shift on vertebrate hosts other than humans . A total of 3 , 519 domesticated animals ( 1 , 147 cattle; 671 goats and 1 , 701 sheep ) from 160 farms established in 14 municipalities were sampled . From these 3 , 519 screened animals , 17 were found to be serologically positive but were ultimately considered as false positive after complementary analyses . Additionally , our 7-day extensive rodent trapping ( i . e . 1 , 949 traps placed ) resulted in the capture of a total of 34 rats ( Rattus rattus ) and 4 mice ( Mus musculus ) . Despite the low number of rodents captured , molecular diagnostic tests showed that two of them have been found to be infected by schistosomes . Given the low abundance of rodents and the low parasitic prevalence and intensity among rodents , it is unlikely that neither rats nor ruminants play a significant role in the maintenance of schistosomiasis outbreak in Corsica . Finally , the most likely hypothesis is that local people initially infected in 2013 re-contaminated the river during subsequent summers , however we cannot definitively rule out the possibility of an animal species acting as reservoir host . The increasing movements of humans and animals and the ongoing climate changes at the global scale are expected to promote the emergence or the spread of tropical infectious diseases in temperate areas [1] . In this context , in summer 2013 more than 100 cases of human urogenital schistosomiasis were contracted in a very confined locality in the south of Europe ( Corsica , France ) [2 , 3] . In summer 2014 the incriminated river ( i . e . the Cavu ) was closed to recreational activities , but during the two next summers , 2015 and 2016 , new local contamination events occurred [4 , 5] . Molecular analyses performed on parasites from patients infected in 2013 and 2016 revealed that the parasite was the same unique strain [4] . In this scenario , it has been hypothesised that vertebrate hosts , either human , animal or both , would had been the original source and ongoing reservoirs for this parasite strain , maintaining the parasite transmission in this area [4] Schistosomes are trematode parasites affecting either humans , animals or both , according to the species concerned in tropical and sub-tropical countries [6] . Schistosoma haematobium is responsible for human urogenital schistosomiasis , it is widely distributed through sub-Saharan Africa , Egypt , Sudan and the Arabian Peninsula and it is estimated to affect more than 110 million people [7] . Schistosoma bovis is a schistosome species infecting livestock ( cattle , sheep , goats , pigs , equines , and dromedaries ) , wild ruminants , and rodents [8] . This parasite species is widely distributed throughout Africa , the Middle East and to a lower extent in the Mediterranean islands and Spain [9] . It is estimated that at least 165 million cattle are infected with schistosomes worldwide causing serious socio-economic damages [10] . Considering the One Health approach , it has recently been proposed to better quantify these economic losses and eventually to treat the schistosomes infected livestock [11] . Surprisingly , the parasite that emerged in Corsica was not a pure S . haematobium parasite but a hybrid between S . haematobium and S . bovis [12] . It has previously been showed that this hybrid was imported from Senegal [2] . S . haematobium x S . bovis hybrids have also been reported in Niger , Mali and Benin [13] . Previous studies have found these hybrid parasites in human and rodent hosts [8] . Only two studies have investigated the presence of such hybrid parasites in domestic cattle but did not find any [14 , 15] . However , as the own authors recognised , they only searched the blood vessels of the intestinal tract ( infection site of pure S . bovis ) and not the vessels of the urinary tract ( infection site of pure S . haematobium ) , where the hybrids may have passed unnoticed [14 , 15] . The fact that S . bovis was historically known to be present in the Corsican island [16 , 17] makes the epidemiological situation much more complex and suggests the likely presence of local reservoir hosts and the potential of local hybridization events . The last study on the presence of S . bovis dates from 1962 [18] and since that time , S . bovis has not been investigated . Identifying infected animals in slaughterhouses relies on the detection of worms of approximately 1 . 5 cm in the cattle mesenteric system . Moreover , clinical manifestations in the S . bovis infected animals are poorly documented and the disease is mainly sub-clinical and chronic . As a consequence , the presence of S . bovis in the island is enigmatic and need to be investigated [19] . Another concern is the potential presence of rodent reservoir hosts . It is well known that schistosome parasites have affinity for rodent hosts , as evidenced by the fact that most species of schistosomes can be maintained under laboratory conditions using several rodent species as vertebrate hosts ( e . g . mouse , hamster , rat , gerbil or guinea pig ) . There is experimental evidence that S . bovis can infect the Nile rat ( Arvicanthis niloticus [20] ) and a wide range of wild rodents has been found to be naturally infected by this parasite [21]: Arvicanthis niloticus [8 , 20] , Mastomys natalensis [22] , Praomys albipes , Rattus rattus , Mastomys coucha and Lophuromys flavopunctatus [23] . However , rodents are poorly compatible for S . haematobium and only Mastomys coucha and Arvicanthis niloticus were successfully experimentally infected so far [23] . To our knowledge there is no evidence of natural S . haematobium rodent infection . Interestingly however , one S . haematobium x S . bovis hybrid parasite were recently detected in Mastomys huberti in Senegal [8] . The close genetic similarities between parasites established in Corsica and those present in Senegal hence raise the question whether rodents present in Corsica could constitute potential reservoir hosts locally . Our aim was to identify if animals ( livestock and/or wild rodents ) can be reservoir host of zoonotic schistosomes in Corsica . This study included both a large-scale sero-epidemiological survey on ruminants and a rodent trapping campaign in the vicinity of the Cavu where the schistosomiasis transmission is still persisting . We have also experimentally tested the ability of schistosome hybrids from Corsica to infect Rattus norvegicus laboratory hosts . ELISA tests were performed to detect potential anti-schistosome antibodies in the sera of animals using a tegumental extract ( TG ) of S . bovis adult worms as coating antigen . This S . bovis TG extract was obtained as previously described [25] . Multi-well polystyrene plates ( Corning , Ref 3369 ) were coated overnight at 4°C with 0 . 5 μg/well of TG extract in 100 μl of carbonate buffer , pH 9 . 6 . The following day , the plates were first washed three times with 0 . 05% Tween-20 in PBS ( TPBS ) and then blocked with 200 μl/well of 1% bovine serum albumin ( BSA ) in PBS for 1 h at 37°C . After a series of three additional washes , the sera were placed into duplicated wells ( 100 μl/well of a 1/100 dilution in TPBS ) and incubated for 1 hour at 37°C . After a last washing step , 100 μl/well of a 1/16 , 000 dilution of peroxidase-labelled anti-bovine IgG ( Sigma , A5295 ) or a 1/6 , 000 dilution of peroxidase-labelled anti-Sheep/Goat IgG ( AbD Serotec , STAR88P ) was added and the plates were incubated for 1 hour at 37°C . Finally , the plates were incubated for 10 minutes at room temperature with 100 μl/well of substrate solution ( 0 . 6 mg/ml of orto-phenylenediamine and 0 . 4 μl/ml of H2O2 in citrate buffer pH 5 . 0 ) . The reaction was stopped with 100 μl/well of 3N sulfuric acid , and the optical densities ( OD ) at 492 nm were read with the Multiskan GO spectrophotometer ( Thermo Scientific ) . In each plate , 3 positive and 3 negative control sera were included . The negative and positive control sera used for the analysis of sheep and goats were obtained from naive sheep and from sheep experimentally infected with S . bovis . For the analysis of the bovine sera , positive and negative control sera were collected from cattle during an epidemiological study performed in the province of Salamanca ( Spain ) [26] . Any positive and/or doubtful sera potentially detected were reanalyzed independently following the same protocol . Additionally , bearing in mind that the trematode Fasciola hepatica is also present in Corsica [27] , the potential reactivity of the TG extract with antibodies against F . hepatica infection was checked . To do so , serum samples from six lambs experimentally infected with 100 F . hepatica metacercariae obtained in a previous work [28] were analyzed by ELISA against the excreted/secreted antigen of F . hepatica ( E/S Fh ) and the TG extract of S . bovis . The E/S Fh antigen was prepared as previously described [29] . To analyze the results of the serology and to compare serological patterns between samples , each optical density ( OD ) was transformed into an Elisa Index ( EI ) by applying the following formula: OD of each serum / mean OD of the negative control sera included in each plate . Receiver–Operator Characteristic ( ROC ) curve were built for the TG extract and used to establish the cut-off value of EI used to discriminate between positive and negative animals . A ROC curve is obtained by calculating the sensitivity and specificity of a test or an antigen at every possible cut-off point , and plotting sensitivity against specificity [30] . The current ROC analysis was performed using 50 well-defined sera from sheep infected experimentally with S . bovis and 164 sera from naive sheep ( without schistosomes ) . After that , the cut-off selected was the EI value that gave the highest diagnostic performance for the TG extract , which was calculated as the sum of the sensitivity and specificity divided by two . Then , this value was used to establish the positive/negative state of all the animal sera analysed . This ROC analysis was performed using the SPSS v17 software package . For ELISA positive animals , coprological examination was assessed by recovering faecal sample directly from the rectum . A faecal sample of approximately 100 grams was diluted in saline solution and passed through a series of stainless steel sieves ( 420 μm , 250 μm , and 45 μm ) . The eggs should be retained in the last sieve . The trapping campaign was done along the Cavu between the 29th of May and the 9th of June 2018 . The distance between the river and the traps was generally less than 50 meters and never greater than 150 meters . We have focused our survey in the vicinity of the human transmission sites identified in 2013 and 2015 [2 , 4] . The traps ( 27x9x9 cm; Caussade , France ) were deposited along the forest in steep locations and also near human activities . Our protocol considered the nocturnal activity of the targeted species . Thus the traps were baited every late afternoon , and inspected in subsequent early morning . A total of 1 , 949 traps were placed during 10 trapping sessions . Nontarget species such as shrews ( Crocidura suaveolens ) , hedgehogs ( Erinaceus europaeus ) or weasels ( Mustela nivalis ) were immediately released at the point of capture . Trapped small rodents ( mouse and rats ) were returned alive to the laboratory for necropsy . Animal were euthanized with lethal injection of 1 mg per kg body weight of a sodium pentobarbital solution ( Dolethal , Vetoquinol , Lure , France ) and then perfused using an hepatic perfusion technique [31] . The resulting collected blood was filtered to recover potential adult worms . Animal internal organs ( i . e . liver , mesenteric vessels , peri-vesical area ) were examined to identify possible traces of schistosome infection . Some suspicious zones such as liver cysts and abnormal ( hemozoin-like ) pigments were biopsied and the fragments of tissues were stored in ethanol 95% for subsequent genetic diagnostics . DNA extraction from worms and biopsies were performed using E . Z . N . A . tissue DNA extraction kit ( Omega Bio-tek , USA ) following the manufacturer's instruction . Extracted DNA from worms and biopsies were then amplified by PCR using Sh73/DraI assay [32] . This PCR diagnostic tool is very sensitive and specific to the S . haematobium parasite group and the resulting genetic profiles allow distinguishing between S . bovis and S . haematobium [32] . The PCR target a 121 bp repeated sequence of the schistosome genome . According to the PCR efficiency some extra bands can be observed representing multiples of 121 bp ( e . g . 242 ) . For worms only , the complete internal transcribed spacer ( ITS ) and a partial region of the mitochondrial cytochrome oxidase subunit I ( cox1 ) gene were amplified by PCR and sequenced by a subcontractor ( Genoscreen , Lille , France ) . Sequences were first visualised and manually corrected before being compared to a set of reference sequences from the National Center for Biotechnology Information ( NCBI ) database including the sequences corresponding to the parasites from the 2013 and 2015 outbreaks [2 , 4] . Detailed PCR protocols are available in S1 Text . The parasite that emerged in Corsica is routinely maintained in the lab using hamsters as vertebrate hosts and a sympatric lineage of Bulinus truncatus as intermediate host [33] . Three Wistar rats ( Rattus norvegicus ) were exposed to three different doses of cercariae: 400 , 600 and 800 following standard procedure [31 , 34] . Rats were returned to their facilities and reared under standard experimental conditions for 105 day to give time for the potential infecting parasites to develop in their hosts . After this developing period , the rats were euthanized with lethal injection of 0 . 3 ml of a sodium pentobarbital solution ( Dolethal , Vetoquinol , Lure , France ) and perfused using hepatic perfusion technique [31] . The blood from each rat was filtered and visually inspected for the presence of adult worms . Similarly to wild animals , organs were also thoroughly examined to identify possible traces of schistosome infection . All experiments were carried out according to national ethical standards ( NOR: AGRG1238753A ) . The French Ministry of Agriculture ( Ministère chargé de l’Agriculture ) , and the French Ministry for Higher Education , Research and Technology ( Ministère de l’Education Nationale de la Recherche et de la Technologie ) approved the experiments carried out for this study and provided permit A66040 for animal experimentation . The investigator possesses the official certificate for animal experimentation ( Decret n˚ 87–848 du 19 Octobre 1987 ) . The sera of a total of 3 , 519 animals from 160 farms localized in 14 municipalities were sampled . This sampling effort represents 15% ( n = 1 , 147 ) , 48% ( n = 671 ) and 22% ( n = 1 , 701 ) of the total cattle , goats and sheep reared in this area respectively ( Fig 1 ) . The resulting sera were stored at -20°C and sent to the laboratory of parasitology at IRNASA ( CSIC ) in Salamanca ( Spain ) for ELISA tests . ROC analysis allowed us to set up a positivity/negativity cut-off between EI 2 . 5 and 2 . 8 , with a diagnostic performance of 96 . 7% ( 97% sensitivity and 99 . 4% specificity ) . Since farmed animals are expected to be in contact with a greater number of pathogens and parasites compared to the experimental animals infected under laboratory conditions , we conservatively established a cut-off slightly higher ( namely , EI 3 ) . Accordingly , EI values comprised between 2 . 5 and 3 were considered unclear results and the associated sera were re-analysed . More generally , we applied the following criterion: animals with serum samples displaying EI < 2 . 5 were considered as negative , those with sera that displayed EI values between 2 . 5 and 3 were considered as doubtful , and finally animals with a detected EI value > 3 were considered as infected by Schistosomes . Table 1 shows the overall results of the ELISA analyses of 3 , 519 sera and Fig 2 represents the frequency of EI values obtained for each host species ( i . e . goat , cattle , sheep ) . These results indicate that 99% of the sera analysed were negative , 0 . 52% remained doubtful after a double-analysis and 0 . 48% were found to be positive with EI values higher than 3 . In order to explain these positive reactions , we first analyzed the cross-reaction to F . hepatica , which is an autochthonous trematode in Corsica [27] . The sera from sheep infected with F . hepatica showed increasingly detectable levels of antibodies to F . hepatica from the second week of infection onwards . However , none of these sera reacted at all with the TG antigen , clearly indicating that a possible concurrent infection by F . hepatica would not interfere in the S . bovis-ELISA diagnostic test . To confirm the diagnosed positive cases , additional blood samples from some animals found to be seropositive were collected and analyzed independently . From the 17 animals ( 12 cows– 1 sheep– 4 goats ) detected positives during the first serological screening some animals was declared dead by the breeders ( 1 sheep– 1 goat– 1 cow ) and for 8 animals ( 3 goats and 5 cows ) , the breeders refused a second serological sampling . Only 6 cows were serologically tested a second time . These cows were also diagnosed based on coprological examinations . This additional series of diagnostic tests revealed that these six cows were in fact negative based on both serological and coprological ( i . e . absence of schistosome eggs ) examinations . The Table 2 represents the GPS coordinates of the different sites and the number of traps by site . Out of the 1 , 949 traps placed in the field , a total of 34 rats ( Rattus rattus ) , 4 mice ( Mus musculus ) , 2 shrews ( Crocidura suaveolens ) , 3 hedgehogs ( Erinaceus europaeus ) and 2 weasels ( Mustela nivalis ) were trapped . No trace of schistosome was found in the mice ( Mus musculus ) . Among the 34 rats , one individual ( called R1 ) was infected by a unique male schistosome . This rat also displayed granuloma on the liver ( that could indicate the past or present occurrence of schistosome eggs in the liver ) and black pigments on the mesenteric system . These pigments were similar to the hemozoin pigment specifically found in mature female schistosomes . Both granuloma and a piece of the mesenteric vein harbouring the black pigment were biopsied . A second rat ( R2 ) did not display adult worms after the perfusion but some black pigmentation on the mesenteric system and on the hepatic portal system similar to those found in the rat R1 . Biopsies were performed on these two zones . Additionally , samples from the liver , the mesenteric vessels and the bladder from a third rat ( R3 ) where no trace was observed were also biopsied as negative controls . Fig 3 shows PCR profiles after the Sh73/DraI PCR diagnostic test performed on the biopsies from the two suspicious rats ( R1 , R2 ) and the negative non-infected rat R3 . No band amplification was observed on the biopsies from the non-infected rat R3 ( well 1 , 2 , 3 ) . Conversely , a PCR amplification was obtained for the adult worm and all biopsies collected on the rat R1 ( well 4: worm; well 5: biopsy of mesenters; well 6: biopsy of the granuloma ) and all amplifications displayed a migration pattern specific to pure S . haematobium ( well 12 ) . In respect with the rat R2 , we find less intense but positive PCR signal ( i . e . presence of the 121 bp diagnostic band ) on the mesenteric biopsy ( well 8 ) . No amplification was obtained on the biopsy performed on the hepatic portal system of the same rat ( well 9 ) . The ITS sequence ( Genbank accession number: MK797748 ) from the single worm from the rat R1 was assigned to S . haematobium and the cox1 sequence was assigned to S . bovis , a pattern commonly found in hybrids from the lineage established in Corsica . The cox1 sequence ( Genbank accession number: MK797748 ) obtained from this worm is strictly identical to the Sb2 haplotype found during the 2013 and 2015 outbreaks [2 , 4] . Among the three Wistar rats experimentally exposed to the schistosome strain that emerged in Corsica only two were slightly infected . No worms were recovered from the rat exposed to 800 cercariae and no trace of schistosomes was observed in the organs . The rat exposed to 600 cercariae displayed one schistosome pair , two single males and a few granulomas on the liver . Finally , the rat exposed to 400 cercariae displayed a single male worm and no granuloma in the liver . No pigment trace was observed on the three animals . In 2013 , the emergence of schistosomiasis in Corsica was clearly unexpected for both the scientific community and public health authorities [35] . Even more surprising was the persistence of the outbreak with evidences of new local infections during summer 2015 and 2016 [4 , 36] . Because free larval stages cannot resist in the environment , the persistence of the parasite has to be attributed to the maintenance inside intermediate and/or definitive hosts . Considering the longevity of infected snail intermediate hosts in the lab ( 170 days at best [37] ) and if we suppose that these snails can resist over one winter season this cannot explain the 2015 and 2016 cases . The existence of local vertebrate reservoir hosts either human and/or animal is the more likely explanation . The diagnostic of schistosomiasis in humans or animals is far from being trivial . For humans , several diagnostic commercial kits exist based on the ELISA , Indirect Haemagglutination ( IHA ) or Western Blot methods [38] . During the outbreak in Corsica , the French authorities for human health recommended the concurrent use of these two methods ( IHA and ELISA ) , followed by a Western Blot analysis when the two former tests were discordant . For animals , no such commercial kits exist to detect bovine schistosomiasis . Infections of ruminants by S . bovis can be assessed either directly , i . e . through the detection of parasite eggs in faeces or adult worms in the mesenteric veins in sacrificed animals , or indirectly by detecting antibodies specific to S . bovis in animals’ blood serum . Coprological diagnosis , which is the most frequently used by veterinarians , has high specificity but is very laborious and offers low sensitivity [39] . The detection of specific antibodies to S . bovis antigens in the animal host is more sensitive than coprology and furthermore , the serodiagnostic methods ( as ELISA ) can be applied to large-scale epidemiological studies [40] . In this study , we performed a wide serological survey in domestic ruminants originating from the geographic area where human infections by schistosomes were first reported in south Corsica . We used the S . bovis TG extract as antigen , which has proved to be very efficient and reliable for the detection of antibodies of S . bovis in sheep [41] . This method allowed us to detect a total of 17 positive animals . However , for several reasons we believe these positively diagnosed animals were in fact false positives . First , cows found to be serologically positive after the first set of analyses were found to be negative after a second batch ELISA analyses and a sub-sample of these animals were found to be negative using coprological test . Second , S . bovis usually exhibits a focalized distribution and infected animals are expected to cluster into few particular farms [42] . Contrary to this expectation , the 17 positively diagnosed animals originated from 10 different farms sometimes geographically distant from one another . Third , despite our first technical validation , glycan antigens can be responsible for many cross-reactions between parasite infections [26 , 43] . We thus believe that the positive diagnostic of these 17 sera result from unspecific reactions to the glycoproteins that contain the TG extract [41] . In this regard , the potential influence of the hybrid status of schistosomes on the efficiency of serological diagnostic is not known . The sensitivity of serological commercial tests using S . mansoni antigens is lower for S . haematobium infection than for S . mansoni infection [44] . S . bovis and S . haematobium are closely related species and S . bovis antigen has proved to be useful for the detection of human infections by S . mansoni , S . intercalatum and also S . haematobium [45–47] . However , the sensitivity of serological tests needs to be evaluated in the case of hybrid parasite infections in human . Concerning rodents , both our experimental ( i . e . on R . norvegicus ) and field ( on R . rattus ) approaches indicate that rats are partially permissive to the hybrid schistosome strain that emerged in Corsica . Unfortunately , we did not experimentally test the compatibility of the schistosome strain naturally present in Corsica to black rats ( R . rattus ) since no laboratory strain of R . rattus was available . Importantly , both the black rats [48–50] and the brown rats [50] are known to be natural alternative reservoir hosts of the parasite S . japonicum . In the case of S . mansoni the situation is more complex . The two rodent host species are partially permissive to experimental infections . In particular , adult worms develop in the brown rat but the eggs produced are non fertile eggs and these eggs are not excreted , thus this species is not considered as a suitable vertebrate host for parasite transmission [51] . Conversely , the black rat is known to be a suitable host for S . mansoni transmission [52] . We lack information concerning the suitability of rats for either S . haematobium or S . bovis . Experimental infections showed that both rat species can be infected but are not very permissive to S . haematobium parasite [53] . To our knowledge only one study showed that R . rattus is permissive to S . bovis [23] . The ability of hybrid schistosomes to increase its host range and thus for S . haematobium hybrids to become a zoonotic parasite is a real concern . To date , no livestock animals was found to be infected by S . haematobium x S . bovis hybrid parasites [14] and only few rodents were found to be infected in Senegal [8] . In the case of Corsica , very few animals were caught ( 2 . 3% of trapping success ) hence indicating a small abundance of rats in the vicinity of the identified transmission sites . Considering both the low parasite infectivity in experimental infection and the field observation showing low parasite prevalence and intensity in rodents ( only two rats infected with very few worms and no trace of reproduction ) , we suppose that even if rats are permissive to infection their role in the maintenance and transmission of schistosomiasis locally is expected to be negligible . In summary , our survey strongly suggests that bovine schistosomiasis is absent from the south of Corsica , and livestock or wild rodents may not play a role as reservoir host for human schistosomiasis . The host spectrums of schistosomes are at least partially known in their original distribution area in Africa , Asia and south America , but the invasion of a new ecosystem by a hybrid parasite strain may offer new host possibilities . As an example , the mouflon ( Ovis gmelini musimon ) is endemic from Corsica , and nothing is currently known about the ability of S . bovis to infect such Ovidae . The fact that S . bovis is known to infect sheep and the presence of a mouflon population in the Cavu valley could support this last hypothesis . Moreover , the hybrid status of the parasite makes the situation much more complex because the host spectrum of hybrid schistosomes remains enigmatic . Some positive patients emitting schistosome eggs could serve as reservoir for schistosome infection in Corsica . This hypothesis is likely because up to 66% of the patients that have been found to be infected so far were asymptotic [54] , the diagnosis is difficult to establish and false negative are possible [54–56] , and the median time for first symptom apparition is 30 weeks [54] . To conclude , even if the most likely hypothesis is that local people initially infected in 2013 re-contaminated the river during subsequent summers , we cannot definitively rule out the possibility of an animal species acting as reservoir host .	There is an increasing interest on the effect of global changes on the transmission of infectious diseases . Both environmental and anthropogenic changes are expected to promote outbreaks and spread of pathogens . In particular , tropical infectious diseases are expected to move towards more temperate latitudes . Until 2013 , urogenital schistosomiasis was restricted to tropical and sub-tropical areas . In summer 2013 , a schistosomiasis outbreak has emerged in Corsica ( France ) with more than 100 cases . Corsica is a French Mediterranean island , which is very popular for tourists from throughout Europe due to the natural beauty of the environment . Surprisingly , in summer 2015 and 2016 , the contamination has resumed , and schistosomiasis has been classified in the list of French notifiable infectious disease . In this context it has been hypothesised that reservoir vertebrate hosts , either human and/or animal are at the origin of the maintenance of the local transmission . This paper shows that ruminants ( cow , sheep and goats ) should not play a role of reservoir host but we found that rodents living in the vicinity of the transmission sites have been infected by the parasite . Considering the low abundance of rodents and the low parasitic prevalence/intensity among rodents , it is unlikely that rats play a significant role in the maintenance of schistosomiasis outbreak in Corsica and that other animals or human could maintain the parasite locally .	[ "Abstract", "Introduction", "Methods", "Results", "Discussion" ]	[ "schistosoma", "invertebrates", "livestock", "medicine", "and", "health", "sciences", "ruminants", "helminths", "vertebrates", "parasitic", "diseases", "animals", "mammals", "animal", "models", "model", "organisms", "experimental", "organism", "systems", "research", "and"...	2019	Epidemiological surveillance of schistosomiasis outbreak in Corsica (France): Are animal reservoir hosts implicated in local transmission?
Oscillations are omnipresent in neural population signals , like multi-unit recordings , EEG/MEG , and the local field potential . They have been linked to the population firing rate of neurons , with individual neurons firing in a close-to-irregular fashion at low rates . Using a combination of mean-field and linear response theory we predict the spectra generated in a layered microcircuit model of V1 , composed of leaky integrate-and-fire neurons and based on connectivity compiled from anatomical and electrophysiological studies . The model exhibits low- and high-γ oscillations visible in all populations . Since locally generated frequencies are imposed onto other populations , the origin of the oscillations cannot be deduced from the spectra . We develop an universally applicable systematic approach that identifies the anatomical circuits underlying the generation of oscillations in a given network . Based on a theoretical reduction of the dynamics , we derive a sensitivity measure resulting in a frequency-dependent connectivity map that reveals connections crucial for the peak amplitude and frequency of the observed oscillations and identifies the minimal circuit generating a given frequency . The low-γ peak turns out to be generated in a sub-circuit located in layer 2/3 and 4 , while the high-γ peak emerges from the inter-neurons in layer 4 . Connections within and onto layer 5 are found to regulate slow rate fluctuations . We further demonstrate how small perturbations of the crucial connections have significant impact on the population spectra , while the impairment of other connections leaves the dynamics on the population level unaltered . The study uncovers connections where mechanisms controlling the spectra of the cortical microcircuit are most effective . Understanding the origin and properties of oscillations [see 1 , for a review] is of particular interest due to their controversially discussed functional roles , such as binding of neurons into percepts and selective routing of information [reviewed in 2 , esp . part VI] . Specific frequencies have been localized in different layers and linked to top-down and bottom-up processes [3 , 4] . Oscillations in population signals correlate with multi-unit spiking activity [5] , predominantly at high frequencies [6 , 7] , while firing probabilities relate to the phase of low frequency oscillations [8] . Coherent oscillations at the population level can arise from clock-like firing cells [9 , 10] and more robustly [11] from irregularly firing neurons synchronizing weakly [12 , 13] . Neurons in vivo tend to fire irregularly [14] and population oscillations resemble filtered noise rather than clock-like activity [15 , 16] . Balanced random networks of leaky integrate-and-fire neurons in the asynchronous irregular ( AI ) regime can sustain such weakly synchronized oscillatory states [17] and reproduce the stochastic duration and power spectra of γ oscillations [18 , 19] . Focusing on the network aspect , rather than on intrinsic cell properties , the PING and ING mechanisms have been suggested to underlie the generation of low- and high-γ frequencies [20 , reviewed in 21] . Inter-neuron γ ( ING ) consists of a self-coupled inhibitory population producing an oscillation frequency primarily determined by the time course of the inhibitory postsynaptic potential ( IPSP ) , the dynamical state of the neurons [20 , 22 , 10 , 23] and the delays [24] , constraining the generated frequency to the high-γ ( > 70 Hz ) range . Lower-γ frequencies ( 30–70 Hz ) arise from the interplay of pyramidal- and inter-neurons ( PING ) with the frequency determined by the dynamical state of the neurons and the connection parameters [25 , 26 , 27 , 28] . Early network models combining ING and PING motifs [29] , as well as self-coupled excitatory populations ( studied later in [30] ) , enabled the phenomenological study of γ oscillations . The two mechanisms were originally formulated for the fully synchronized regime and the analytical treatment of weakly synchronizing networks is restricted to at most two populations [17 , 31] , neglecting the variety of dynamical states of neuronal populations embedded in a larger circuitry . Modeling studies considering neurons of various level of detail assess the link between network structure and induced oscillations [32] . Experiments find specific frequencies at different depths of the layered cortex [33] , while these locations in the tissue are characterized by distinct connectivity patterns [34] . Pronounced slow oscillations ( < 1 Hz ) are found in deeper layers , such as layer 5 [35 , 36] , and hypotheses regarding their origin range from intrinsic cell mechanisms [37] to network phenomena [38 , 39] . In contrast , fast oscillations in the γ and high-γ range are primarily observed in the upper layers [40 , 41] and show different phase relationships than the γ oscillations in the lower layers [42] . To the best of our knowledge , theoretical descriptions of coexisting oscillations requiring complicated network structures , as well as a method identifying these structures in a given circuit have not yet been established . The present work sheds light on the influence of sub-circuits integrated in larger networks and the properties of individual connections relevant for the emergence of specific oscillations . The multi-layered spiking cortical network model used throughout this study was introduced by Potjans and Diesmann [34] . The model is composed of four layers ( L2/3 , L4 , L5 and L6 ) , each layer containing an excitatory and an inhibitory population of neurons ( Fig 1A ) . The number of neurons in each population , as well as the number of connections between and within populations are extracted from experimental data sets [for a full list of references see Table 1 of ref . 34] . Combining the data yields the 8 × 8-dimensional indegree matrix K ( Fig 1C ) , where the element Kij describes the number of connections from population j to population i . Given the total number of connections between populations , the pre- and postsynaptic neurons of the individual connections are drawn randomly . Each population receives additional Poisson spike trains resembling the activity of other brain regions . Potjans and Diesmann show by simulations that the population firing rates generated within the model reproduce those observed in experiments [43 , 44] . The neurons are modeled by leaky integrate-and-fire ( LIF ) dynamics with exponentially decaying synaptic currents: τ m d V k i ( t ) d t = - V k i ( t ) + R I k i ( t ) τ s d I k i ( t ) d t = - I k i ( t ) + τ s ∑ l = 1 N ∑ j = 1 M l w k i , l j ∑ n δ ( t - t l j n - d k i , l j ) + τ s ∑ j = 1 M e x t w k i , j ∑ n δ ( t - t j n ) . ( 1 ) Here Vki ( t ) describes the membrane potential of the i-th neuron in the k-th population and Iki ( t ) the incoming synaptic current to this neuron . R denotes the resistance of the membrane and τm the membrane time constant , τs the synaptic time constant , w and d the weight and delay associated to the incoming events , and t l j n the time of the n-th spike of neuron j in population l . The number of populations is given by N and the number of neurons in the l-th population is denoted by Ml . In addition to the spikes from within the network , each neuron receives spikes from Mext external sources representing the input of other brain regions by spike times t j n drawn from Poisson processes with rates specified in [34] . Simulating the microcircuit model , we first reproduce the dynamics observed in [34] and additionally investigate the correlation structure of the system . After simulating the circuit for T = 10 s with a time resolution of 0 . 01 ms , we observe averaged population specific firing rates between 0 . 9 Hz and 8 . 6 Hz , which reflect tendencies of population specific firing rates in experimental data [34] . The average coefficients of variation ( CV ) of the neurons are around 0 . 55 for the populations with low firing rates ( 2/3E and 6E ) and around 0 . 8 for the other populations , characterizing the spike trains of individual neurons as irregular . The irregular nature of the spike trains is underlined by the raster plot ( Fig 1B ) showing all spike times in a 50 ms segment . The vertical stripes visible in the spike times of some populations suggest a certain degree of synchrony in the activity of the neurons on the population level . However , this regularity is barely exhibited on the single neuron level , which participate only in a fraction of the cycles ( red dots in Fig 1B ) . A comparison of the auto-correlations of the individual neurons with the auto-correlations of the population shows that the population spectrum is dominated by the cross-correlations and the contribution of the individual auto-correlations is negligible in agreement with Tetzlaff et al . [45] . The description of fluctuations in spiking networks deployed in this study proceeds in two steps , as summarized in Fig 2 . In the first step , we use mean-field theory for spiking neurons [46] to determine the stationary state of the network , i . e . the time-independent averaged firing rates of each population . In particular , the original high-dimensional system , composed of the two dynamical variables in Eq ( 1 ) for each of the N neurons , is reduced by means of diffusion approximation . Here we exploit the fact that each neuron receives a large number of inputs , with each incoming impulse eliciting only a small deviation of the membrane potential , which allows us to treat inputs linearly . Assuming additionally that correlations between the populations are only displayed by fluctuations around a stationary state but that the stationary rates of the populations can be obtained by neglecting these correlations , enables us to describe the total synaptic input to population i as Gaussian white noise characterized by mean and variance [46 , 13] μ i = τ m w ∑ j ∈ E K i j r ¯ j - g ∑ j ∈ I K i j r ¯ j + K ext , i r ext σ i 2 = τ m w 2 ∑ j ∈ E K i j r ¯ j + g 2 ∑ j ∈ I K i j r ¯ j + K ext , i r ext . ( 2 ) Here w denotes the synaptic weight , Kij the indegrees from population j to population i , g the ratio between excitation and inhibition and r ¯ j the yet to be determined stationary firing rate of the j-th population . The mean and variance of the input current to a population is referred to as its working point . In this approximation , the stationary firing rates are a function of the working point , yielding the self-consistency equation r ¯ i = ν i ( μ i , σ i ) = ν i ( μ i ( r ¯ ) , σ i ( r ¯ ) ) ( 3 ) where the vector r ¯ = ( r ¯ 1 , . . . , r ¯ N ) denotes the stationary rates of the N populations and the function ν ( μ , σ ) is derived from the stationary solution of the Fokker-Planck equation for the probability distribution of the membrane potentials ( see equation 4 . 33 in [47] ) . Fig 1D shows that this approximation suffices to predict the rates in the microcircuit model . In the second step , we analyze how small fluctuations around this stationary state propagate within the network and how their dynamics can be mapped to that of a linear rate model [48 , 49] . To this end we employ linear response theory , applied to the leaky integrate-and-fire model , using an extension of the work of [13] and [12] to colored synaptic noise [50] . We here summarize the main results of the mapping of the dynamics of the fluctuations , referring the reader to [49] for the detailed derivations . The reduction allows for a self-consistent dynamical description of the fluctuations of the population rates . The output rate ( left-hand side ) relates to the input ( right-hand side ) via R ( ω ) = M ˜ d ( ω ) Y ( ω ) , with Y ( ω ) = R ( ω ) + X ( ω ) . ( 4 ) Here , R ( ω ) denotes the eight dimensional rate vector in Fourier space . The realization of the instantaneous rate as a spike train is approximated by Poisson statistics , giving rise to the noise term X ( ω ) . The input to the populations is weighted by the connectivity and filtered by the transfer function of the populations ( summarized in the effective connectivity matrix M ˜ d ( ω ) ) . The formulation of the rate dynamics yields predictions for the population rate spectra ( for further details see “Fluctuation dynamics” ) . Fig 1E shows that the low-γ peak around 64 Hz , visible in the spectra of all populations , is well predicted by this theory . As suggested by the regularly occurring vertical stripes in Fig 1B , we observe a high-frequency peak in all populations varying from 235 Hz to 303 Hz , which is most prominent in layer 4 . It can be shown that in the context of the low-γ peak , the network is in the asynchronous irregular ( AI ) regime [12] , where the linear response theory suffices to describe the noise fluctuations . However , on the time scale of the high-frequency peak the network verges on the border of the synchronous irregular state ( SI ) , resulting in deviations of the theoretical prediction from the observed oscillations . The frequency of the fast oscillations depends strongly and inversely on the synaptic time constants , which are small in the present model ( τsyn = 0 . 5 ms ) . Therefore the frequency would be considerably lower for longer time constants ( see “The high-γ peak” ) . The high-frequency peak will therefore , in the following , be referred to as the high-γ peak . Given the density of connections in the circuit ( Fig 1C ) , the similarity of the spectra hints at the oscillation being generated in a sub-circuit of the microcircuit and subsequently imposed onto all populations . This prevents the identification of the sub-circuitry generating the oscillation on the basis of the spectra . Thus the analytical tools developed in our study up to this point enable the prediction of the population firing rate spectra Eq ( 18 ) , but do not allow for the inspection of the underlying circuits determining the characteristics of the spectra . Rate profiles have been observed to vary across cortical layers [44] , with inhibitory neurons displaying higher rates than excitatory neurons [for a review see 34] . As a result of the population specific rates , each population processes the afferent time-dependent activity with its specific temporal filter , called transfer function in systems theory [51] . Here , we summarize how peaks at different frequencies in the spectra can be associated with the activity of coexisting dynamical modes , which can be found by a linear basis transformation of the rate fluctuations , as described in “Dynamical modes” . Intuitively , a mode describes the tendency of a set of neuronal populations to co-fluctuate with a fixed relationship between their amplitudes and relative phases . Subsequently we discuss how population specific transfer functions result in the commingling of modes across frequencies and therefore hinder the analytical tractability of the anatomical origins of the oscillations . Considering the linearity of the relation between input to the populations and the resulting output rate Eq ( 4 ) ( for details see Eq ( 14 ) and the derivation in “Fluctuation dynamics” ) , the influence of the connectivity on the rate dynamics and thus the shape of the spectrum appears to be straightforwardly investigated by applying tools from linear algebra to the effective connectivity matrix , which is defined by the element-wise product ( Hadamard product ) of the anatomical connectivity Eq ( 13 ) , determined by connection weights , indegrees and the population specific transfer functions M ˜ d , i j ( ω ) = M i j A H d , i j ( ω ) . Different modes of the circuit are found by eigenvalue decomposition of the effective connectivity matrix M ˜ d , i j ( ω ) and are , in the following , therefore referred to as eigenmodes . Their dynamical behavior is linked to the frequency dependence of the corresponding eigenvalues as shown in Fig 3B and discussed in “Dynamical modes” . Indeed , we observe that the term p ( ω ) = \|1/ ( 1 − λi ( ω ) ) \| exhibits a peak at around 60 Hz with a shape reminiscent of the peaks in the theoretically predicted spectra as well as in the spectra observed in simulations ( Fig 1 ) . This similarity is not surprising , since the inverse of the distance of the eigenvalue to one contributes to the spectrum of all populations Eq ( 22 ) . At higher frequencies the eigenvalues of four modes produce a maximum with the dominant mode being largest at 275 Hz , corresponding to the high-γ peak in Fig 1E . All modes but one exhibit small terms p ( ω ) for low frequencies . The involvement of the mode that corresponds to the large values of p ( ω ) at low frequencies in the generation of slow rate fluctuations is discussed in the following sections . At peak frequencies the dynamics of the circuit can be approximated by the dynamics of the dominant mode Eq ( 31 ) . Results for the spectra in the excitatory populations of layer 2/3 and 4 are shown in Fig 3C . The reduced circuitry suffices to approximate the spectrum around the low-γ peak , but for lower and in particular higher frequencies the absence of contributions of the remaining modes becomes apparent . Since the dynamics in the vicinity of a peak are well approximated by the dominant mode , the question arises as to how much information about the minimal anatomical circuit producing the same oscillation is contained in the mode representation . As discussed in “Dynamical modes” and illustrated in Fig 3A , a frequency independent projection from the dynamics of the full circuit to the dynamical modes is only attainable if all populations have the same firing rate and transfer function . In this case each dynamical mode can be traced back to one particular set of anatomical connections which does not influence the dynamics of the other modes . Heterogeneous rates and response properties of the populations result in dynamical modes composed of different anatomical connections at different frequencies . Thus the mapping between a set of anatomical connections and a dynamical mode is limited to one particular frequency , while the same set of connections might influence other modes at other frequencies . The representation of the activity modes determining the characteristics of the spectra is hence frequency dependent . Therefore the eigenmode cannot straightforwardly be mapped to the relevant anatomical connection as in the case of networks with populations in homogeneous dynamical states . In this section we provide an intuitive understanding of how the dynamics of the eigenvalues determine the spectra as well as how the eigenvalues originate from the connectivity of the individual layers and are shaped by the connections between the layers . Readers primarily interested in the final method used to detect the origin of the oscillations may skip this section . The dynamics of the eigenmodes are mainly described by the eigenvalues . The frequency dependence of the complex-valued eigenvalues , termed “trajectories” in the following , is visualized in parametric plots ( Fig 4A ) . These plots , also known as Nyquist plots [51 , Chapter 11] , allow for the simultaneous investigation of the real and imaginary part of the eigenvalue . Fig 4A shows the trajectories of the eight eigenvalues of the microcircuit in the complex plane up to 400 Hz . The movement of a trajectory is reminiscent of a spiral starting at the eigenvalue of the effective connectivity matrix at zero frequency and spiraling clockwise towards zero with increasing frequency . Fig 3B together with the final expression for the spectrum Eq ( 22 ) shows that the amplitude of the spectrum increases the closer an eigenvalue approaches the value one and decreases the further it moves away . The spectrum diverges if the eigenvalue assumes one , which will therefore be referred to as the critical value one . The implications of this critical value for the stability of the circuit are discussed in more detail in “Stability of the dynamical modes” . All trajectories eventually converge to zero , reflecting that the modes cannot follow very high frequencies . Hence the term p ( ω ) = 1/\| ( 1 − λi ( ω ) ) \| , which contributes considerably to the shape of the spectrum Eq ( 22 ) , converges to one for high frequencies . The eigenvalue trajectories λi ( ω ) in the microcircuit are typically continuous and spin clockwise for all frequencies . Hence , if a trajectory reaches the sector of positive real parts it will , in the general case , for at least one frequency ωp assume a closer distance to one than in the large frequency limit . At this frequency the term p ( ωp ) assumes larger values than one , yielding a peak in the spectrum . The eigenvalues can be interpreted as the transfer function of the modes ( Fig 3A ) , which offers an intuition for the impact of synaptic parameters the spectra , as outlined in “Eigenvalue trajectories are transfer functions of the dynamical modes” . From the shape of the transfer functions corresponding to the populations Eq ( 15 ) we deduce that small delays slow down the spinning of the trajectories . The eigenvalue therefore passes by one at a high frequency resulting in rapid oscillations . Long delays accelerate the trajectories yielding slow oscillations . However , the longer the delay the larger the radius of the trajectories . Once the eigenvalue assumes a real part larger than one at the frequency where it passes closest to the critical value one the mode produces activity in the SI regime . The transition from the regime where the real part of the eigenvalue is smaller than one to the regime where the real part is larger than one is characterized by a change in stability . Below one , the dynamics relaxes back to the stationary rates in an oscillatory fashion determined by the frequency at which the distance to one is minimal with a damping related to this minimal distance . Nevertheless , since the noise propagating in the system repeatedly excites these oscillatory decaying modes , the oscillations are visible in the population rate spectra . Altering the parameters of the circuit such that the eigenvalue assumes the value one for some frequency ω , the system ( defined by Eq ( 4 ) in the Fourier and by Eq ( 11 ) in the time domain ) transits from damped to amplified oscillatory modes ( as discussed in more detail in “Stability of the dynamical modes” ) . Growing modes are restrained by the non-linearities of the neurons . Hence , the point at which one eigenvalue assumes the value one characterizes the passage of the corresponding mode dynamics from oscillatory decaying rate perturbations to the onset of sustained oscillations . If the eigenvalue assumes real values larger than one , the equivalent linear rate system is unstable and displays an irresistibly growing oscillatory mode . The system of LIF model neurons , however , provides stabilizing non-linearities ( due to the reset after firing and the fact that the rates of the neurons cannot become negative ) which prevent the dynamics from exploding . Since the presented theoretical framework does not account for non-linearities , it is limited to predicting the tendencies of the spectra in the latter case as visible in the high frequency peak in Fig 1E . The radius of the trajectory is compressed by widely distributed delays Eq ( 15 ) , allowing for the production of slower oscillations caused by long delays without destabilization of the dynamics . In order to analyze the dynamics originating in the individual layers we calculate the eigenvalue trajectories of the isolated layers . The input of other layers is provided by means of Poisson spike trains . This approach holds the dynamic state of the populations constant while neglecting the correlations induced by the input from other layers . Hence the collective dynamics emerging locally in each layer can be analyzed . The eigenvalue trajectories corresponding to the isolated layers are displayed in Fig 4B . Since the connectivity within the layers is more pronounced than the connectivity between the layers , we can deduce the origin of some of the eigenvalue trajectories by comparing their characteristics with the characteristics of the trajectories in the original circuit ( Fig 4A and 4B ) . In isolation , layer 2/3 produces an eigenvalue which passes closest to one at 87 Hz . Since the distance to one is large , the eigenvalue trajectory produces only a small peak in the spectrum of layer 2/3 ( Fig 4C ) . Layer 4 in isolation does not generate a low-γ peak ( Fig 4C ) . Connecting the layers , the eigenvalue trajectories of layers 2/3 and 4 mix and produce a trajectory with a positive imaginary offset and a sufficiently small real-valued starting point to pass close by one at a relatively low frequency ( 60 Hz ) , resulting in a peak in the spectra of all populations . In the high frequency range we observe four trajectories originating in the four layers passing close by one . The course of the trajectories is only mildly impacted when integrated into the full circuit . Therefore we predict a high frequency peak visible in the populations , even in the isolated layers . However , considering the spectra in layer 2/3 and 4 we observe that the high frequency peak in layer 4 matches the peak observed in the full circuit , whereas the peak produced in layer 2/3 is of smaller frequency and amplitude . Therefore we can already conclude that the high-γ peak originates in layer 4 and is propagated to the other layers when embedded in the full circuit . In addition to the origin of the oscillation , the eigenvalue trajectories shed light on the stability of the circuit and the associated oscillations . Following the classification of Brunel [54] , the dynamics of a mode transits from the AI to the SI regime via a Hopf bifurcation if there is a frequency at which the corresponding eigenvalue equals one . The dynamics is in the AI regime , i . e . the system possesses a stable fixed point of the rates , if the closest encounter of the corresponding trajectory with the critical value one is located on the left of the latter Fig 4C ( see also “Stability of the dynamical modes” ) . Here , temporal structure in the firing rates arises from oscillatory , but decaying perturbations , which are continuously excited by the noise generated within the system . The damping factor of decaying perturbations grows with the distance of the eigenvalue from one , while the amplitude of the peak in the spectrum diminishes . In addition , Brunel et al . [13] show that networks of LIF-model neurons close to a bifurcation point are stabilized by the non-linearity of the neuron dynamics . Hence , perturbations decaying with low-γ frequency are strongly damped . The assumption of independence on the level of individual neurons , which underlies the mean-field approximation in the first step of our framework , is therefore justified . This is reflected by the match of the theoretical prediction of the low-γ peak with the peak observed in simulations of the microcircuit ( Fig 1 ) . Fig 3C shows that one of the eigenvalues associated to the high-γ peak lies to the right of one ( corresponding to the SI regime in [52] ) , which , in a linear system , would yield growing oscillatory perturbations . However , in networks of LIF-model neurons these oscillations are tamed by the non-linearity of the neurons . Accordingly , for the high-γ peak , the theoretical framework only suffices to predict the tendencies of the peak in the spectrum ( Fig 1 ) . In summary , employing a combination of mean-field and linear response theory we consider the dynamical contributions of the individual layers . In an iterative fashion we narrowed down the origin of the high-γ peak to layer 4 . In addition we find indication for the low-γ peak being shaped in layers 2/3 and 4 . Fig 3C and 3D shows the spectra of distinct layers of a network that can in isolation be described as band-pass filters . However , the connections between the layers contribute strongly to the generated oscillation . This is reflected in the fact that the filter properties of the layers change once inter-layer connections are considered showing that the network needs to be considered in its entirety to predict the spectra . Hence this iterative approach provides insight regarding the structures shaping the oscillations , but is potentially time-consuming especially when circuits with large numbers of populations are considered . Here we set out to develop a systematic approach identifying the connections involved in the generation of the frequency peaks . So far we identified the eigenmode responsible for the peak generation by considering its proximity to the critical value one . Since the distance of the eigenvalue at peak frequency to one scales the amplitude of the peak in the power spectrum , we can define important anatomical connections as connections the eigenvalue is particularly sensitive to . In the following , the eigenvalue evaluated at peak frequency is referred to as the critical eigenvalue . Mathematically sensitivity is assessed by introducing a small perturbation to the indegree matrix at the connection from the l-th to the k-th population K ^ i j ( α k l ) = 1 + α k l δ k i δ l j K i j . ( 5 ) Thus , depending on the sign of the perturbation , the kl-th element of the indegree matrix is decreased or increased by the fraction αkl . Before we continue the formal perturbation analysis let us briefly look at the interpretation of such a perturbation in the context of our theoretical framework . Our aim is to analyze the contribution of the connection from population j to population i on the fluctuations of activity expressed by the spectra . The linear response theory treats fluctuations of network activity around the stationary state up to linear order . The stationary state itself ( determining the firing rates ) as well as the transfer functions of the populations are in this approximation therefore not effected by the activity fluctuations . We hence study the affect of the connections while conserving their embedment in the full circuit . This separation of the contribution of connections to the correlations from their contribution to the stationary state can be realized in direct simulations by counteracting the perturbation in the number of synapses within the circuit by an adjustment of the external input to the populations . In this way the stationary properties remain fixed since the mean and variance of the input to the neurons are unaltered . However , the correlation structure generally changes since the connections within the circuit , which induce correlations due to the specificity of the connectivity , are substituted by external connections providing uncorrelated input . The perturbed effective connectivity matrix is obtained by inserting the new indegrees into Eqs ( 13 ) and ( 12 ) M ^ i j ( α k l ) = 1 + α k l δ k i δ l j M ˜ i j . ( 6 ) We define the sensitivity measure Zkl ( ω ) as the derivative of the critical eigenvalue of the perturbed system at frequency ω with respect to the perturbation [53] Z k l ≔ ∂ λ ^ c ( α k l ) ∂ α k l \| α k l = 0 = v ^ c T ( α k l ) ∂ M ^ ( α k l ) ∂ α k l u ^ c ( α k l ) v ^ c T ( α k l ) u ^ c ( α k l ) \| α k l = 0 = v c , k M ˜ k l u c , l v c T u c , ( 7 ) where M ˜ k l is the kl-th element of the effective connectivity matrix and v c T , uc are its left and right eigenvectors corresponding to the critical mode . For brevity , the frequency dependence of the matrix and the eigenvectors is omitted . The elements of the matrix Z ( ω ) describe the direction and amplitude of the shift of the critical eigenvalue after perturbing the indegrees of the corresponding connections . The frequency dependence of the perturbed eigenvalue can be linearly approximated by λ ^ ( α k l , ω ) ≃ λ ( ω ) + Z k l ( ω ) α k l , ( 8 ) which describes the displacement of the eigenvalue to linear order after perturbing the kl-th element of the indegree matrix . Hence the sensitivity measure evaluated at peak frequency exhibits large entries for connections having a strong influence on the position of the critical eigenvalue . Fig 5A shows the real and imaginary part of the sensitivity measure Z evaluated at 64 Hz . The influence of the individual elements on the eigenvalues can be visualized in the complex plane ( Fig 5B ) . Given the inverse proportionality of the peak height to the distance of the eigenvalue to one Eq ( 22 ) , a perturbation in a connection causing a shift of the eigenvalue towards or away from one results in an increased or decreased peak amplitude in the spectrum . If the perturbation causes a shift of the trajectory purely in the direction of one , the trajectory will pass by one at approximately the same frequency leaving the position of the peak in the spectrum unaltered . This direction is labeled by the vector k in Fig 5B k = ( 1 - ℜ ( λ c ) , ℑ ( λ c ) ) / ( 1 - ℜ ( λ c ) ) 2 + ℑ ( λ c ) 2 . A perturbation resulting in a shift of the critical eigenvalue along the perpendicular direction k⊥ k ⊥ = ( - ℑ ( λ c ) , 1 - ℜ ( λ c ) ) / ( 1 - ℜ ( λ c ) ) 2 + ℑ ( λ c ) 2 alters the trajectory such that it passes closest to one at a lower or higher frequency while conserving the height of the peak . This suggests a basis transformation of the complex sensitivity measure to the coordinate system spanned by the two vectors k and k⊥: Z i j amp = ℜ ( Z i j ) , ℑ ( Z i j ) k T , Z i j freq = ℜ ( Z i j ) , ℑ ( Z i j ) k ⊥ T . ( 9 ) The resulting matrices Zamp ( ω ) = Zk ( ω ) and Z freq ( ω ) = Z k ⊥ ( ω ) ( shown in Fig 5C ) determine the impact of the connections on amplitude and frequency of the peak . The sensitivity measure ( Fig 5C ) exhibits large entries in the sub-circuit composed of layer 2/3 and 4 . The finding of layer 2/3 and 4 being involved in the generation of the 64 Hz oscillation is in agreement with insights gained from the eigenvalue trajectories in the previous section . We observe that the amplitude of the peak is mostly determined by connections between layers 2/3 and 4 , as well as inhibitory connections within layer 4 . The connections dominating the amplitude of the peak originate mostly in populations 4I , 4E , and 2/3E . The frequency , on the other hand , is shaped by the connections within the layers , with connections within layer 4 having larger impacts than connections in layer 2/3 . The connection from 2/3E to 4I is the only connection from layer 2/3 to layer 4 contributing to the amplitude of the peak . Therefore this connection closes the dynamic loop between layer 2/3 and layer 4 . Its role in the generation of the oscillation is discussed in the following sections . Other connections contributing to the amplitude of the peak originate and terminate in 5E . From Fig 4 we identify four modes that potentially contribute to the generation of the high-γ peak . Evaluating the sensitivity measure for these modes at their respective peak frequencies reveals each mode being shaped by the self-coupling of one inhibitory population ( Fig 6 ) . This mechanism has been termed ING [20] and the peak frequency of the modes and the resulting peak in the spectrum depends on the delay of the synapses , the refractory time and the decay time of the IPSPs [17] . Inserting the time constants used in the microcircuit model into Eq ( 15 ) of [17] , predicts an oscillation frequency of 288 Hz corresponding to the high-γ oscillations observed in the simulations . The rapidity of the high-γ oscillation in the model is hence explained by the choice of small time constants for the IPSCs ( τsyn = 0 . 5 ms ) . Larger synaptic time constants yield an ING peak of lower frequency , for example 127 Hz , 97 Hz and 80 Hz for time constants of 2 ms , 3 ms and 4 ms , respectively . In the original microcircuit model the synaptic time constants are chosen to be small and equal for all neurons to investigate the contributions of the connectivity to the emergent dynamics . The formalism developed in [54] and [50] delivers good predictions for the stationary firing rate and transfer function of the populations for synaptic time constants in the range of a few milliseconds and therefore provides the basis of a successful application of the mean-field and linear response theory . For larger time constants the analytically predicted transfer function can still serve as an approximation to predict the tendencies of the population rate spectra . When the synaptic time constant exceeds the membrane time constant ( τsyn > τm ) an adiabatic approximation [55] is applicable . The dominant mode determining the high-γ oscillation of the full circuit originates in the self-coupling of 4I ( Fig 6A ) . The sign of the entries in Zamp and Zfreq reveals that an increase in the high-γ oscillation , by alterations of the connectivity , goes along with an increase in the oscillation frequency and vice versa . Adjustments of the I-I-loop within layer 4 has an opposite effect than alterations of the I-I-loops within other layers . It turns out that the eigenvalue corresponding to the dominant mode has a real part which is slightly larger than one . Weakening the connections in the 4I-4I-loop stabilizes the circuit dynamics . Once the trajectory is shifted past one , the sensitivity measures takes the opposite sign and predicts decreased high-γ oscillations when connections from 4I to 4I are removed . This shift of the eigenvalue from real parts larger than one to real parts smaller than one describes the transition of network dynamics from the SI to the AI regime [introduced in 12] . Interestingly , in the stabilized circuit ( see “Stability of the dynamical modes” ) , the alterations of connections from 4I to 4I has opposing effects on the amplitude and frequency of the low-γ ( Fig 5C ) and the high-γ oscillation . Since the sensitivity measure analyzes the eigenvalues of the effective connectivity matrix , it sheds light on the static properties of the circuit when evaluated at zero frequency ( Fig 7 ) . The eigenvalue with the largest real part determines the stability of the circuit . At the same time , the measure evaluated at zero frequency reveals the connections shaping low frequency fluctuations . These two statements describe the same phenomenon , since a circuit near an instability exhibits slowly decaying modes when perturbed in the direction of the eigenmode corresponding to the eigenvalue with the largest real part . Technically , there is a peak at zero frequency , but in practice the power in a wide range of low frequencies is elevated , as visible in the spectra in Fig 8B and in the corresponding traces of instantaneous firing rates in Fig 8D . The largest entries of the sensitivity measure evaluated at zero frequency correspond to connections within layer 5 . This finding is in agreement with experimental literature [38 , 39] , where the onset of slow fluctuations was observed to be initiated in layer 5 , as recorded in the sensory-motor areas of the mouse and area V1 of the cat . In contrast to the low- and high-γ oscillations , the slow oscillations are independent of the delay and time course of the neuronal responses . Thus , the amplitude of the slow fluctuations depends solely on the anatomical connections of the circuit and the slope of the f-I curve of the neurons . The indegree matrix ( Fig 1C ) shows that the number of connections from 5E to 5I is low compared to other indegrees in layer 5 . The reduced excitatory input to 5I results in lower rates of the inhibitory neurons relaying less inhibition back to 5E . The comparably stronger E-E-loop is driven towards a rate instability , exhibiting slowly decaying modes of the population rate , which appear as low frequency components in the spectrum . In agreement with the previous considerations , the slow oscillations become stronger if the self-coupling of the populations in layer 5 is strengthened and weaker if the cross-coupling is increased Fig 7 . Further relevant connections are located within layer 4 and starting in population 2/3E and 4E projecting onto layer 5 . The measure predicts that strengthening connections from 2/3E to 5E reduces slow oscillations . We now exploit the sensitivity measure to predict changes in the spectra in different frequency ranges when individual connections are altered . The predictions are validated by simulations of the microcircuit with perturbed indegrees . According to the sensitivity measure shown in Fig 5C , increasing the self-coupling of population 4I should lower the amplitude of the low-γ peak and decrease the frequency . Simulations confirming these predictions are shown in Fig 8A . Since we are interested in the contribution of the connection to fluctuations of the activity , we fix the dynamical state of the populations by simultaneously decreasing the external input to 4I . The left panel demonstrates the shift of the eigenvalue trajectory when altering the connectivity . Since the connection from 4I to 4I strongly influences the dynamics at 64 Hz ( Fig 5 ) , while having small impact on the low frequency spectrum ( Fig 7 ) , the spectrum produced by the altered circuitry deviates from the original one only at frequencies around the low-γ peak . Simulating the microcircuit for increased self-coupling of population 4I confirms the theoretical predictions . Note that reducing the number of synapses from one connection in the microcircuit by as little as 10% can cause an attenuation of the peak amplitude to 7% of its original value and a frequency shift of 11 Hz . The reduction of the oscillation is also visible in the spiking activity . Simultaneously inspecting all spike times of the neurons in population 4E ( Fig 8C top ) , we observe three population burst for the circuit with the original connectivity . The populations bursts become less prominent ( Fig 8C , middle and bottom ) when the number of connections from 4I to 4I is increased , an observation which is in agreement with the predicted population rate spectra shown in Fig 8A . Given that layer 4 is the input layer , we show here that the spectrum exhibited by the circuit is highly sensitive to variations within layer 4 , which could originate either from within the circuit or from external drive . Starting from the hypothesis that slow rate fluctuations are controlled within layer 5 , suggested by the sensitivity measure ( Fig 7 ) , we perturb the indegree from 5E to 5I . The right panel in Fig 8B shows the expected increase of the peak at low frequencies in the spectrum of 5E for fewer connections from 5E to 5I . The predictions match the simulation results . The low frequency oscillations are reflected as slow rate fluctuations in the instantaneous firing rates ( Fig 8D ) . While the stationary firing rate of population 5E is almost not affected by perturbation of the connectivity ( compare the positions of the dashed lines in the three panels of Fig 8D ) , the amplitude of the rate fluctuations increases visibly . The enhanced peak amplitude of the spectrum is explained by the onset of the corresponding eigenvalue trajectory being shifted towards one ( left panel , Fig 8B ) . In agreement with the prediction of the sensitivity measure , the spectrum for frequencies above 20 Hz is unaffected by alterations of the connectivity in layer 5 . Thus we conclude that layer 5 is capable of locally eliciting slow rate fluctuations while leaving the properties of the full circuit at high frequencies unimpaired . The preceding sections investigate how the sensitivity measure predicts the influence of individual connections on the spectrum . Next we apply these insights to uncover the minimal circuitry generating the 64 Hz oscillation . The sub-circuit is obtained by starting from an unconnected circuit , i . e . missing input from other populations is compensated by Poisson spike trains with the same mean and variance . In this setup the populations display the same stationary firing rates as in the original network , but the correlations on the population level are negligible , resulting in a flat population rate spectrum . The empty connectivity matrix is then successively filled with the connections that have largest entries in Zamp ( 64 Hz ) and Zfreq ( 64 Hz ) , while ensuring stability of the resulting system ( instabilities can arise when adding an excitatory connection without an inhibitory counterpart ) . We continue this procedure in decreasing order of sensitivity until the peak frequency of the original spectrum is approximately restored . It turns out that the five largest entries of Zamp and the eight largest entries of Zfreq suffice to reproduce at least 95% of the peak frequency and 83% of the logarithmic peak amplitude in all populations contributing to the circuit . Fig 8A visualizes the resulting connectivity along with simulation results of the reduced circuit and the analytical prediction of the spectra for the original and the reduced circuit . Confirming our previous conclusions , the minimal circuit is located in a sub-system composed of layers 2/3 and 4 . The blocks along the diagonal in Fig 9A show that all connections within layers 2/3 and 4 contribute to the minimal circuit . Additional connections start in the populations of layer 4 and terminate in population 2/3E . The loop is closed by the projection from population 2/3E to the inhibitory population in layer 4 , revealing the special role of this connection in ensuring the recurrence of the oscillation-generating circuit . Testing this hypothesis , we simulate the circuit with the original connectivity , leaving out the connection from 2/3E to 4I . As predicted the peak vanishes entirely ( Fig 9B ) . In summary , these considerations show that given the dynamical state of the populations in the microcircuit , the circuit depicted in Fig 9A is shaping the spectrum around 64 Hz . However , the spectrum generated by the sub-circuit in isolation , i . e . without the substitution of the input of other populations by Poisson spike trains , would potentially be different . Here the advantage of the two-step reduction in the derivation of the theoretical framework becomes apparent . Performing the diffusion approximation the firing rates and response properties of the populations are established and can be verified by experimental data . The analysis of the dynamical contributions of the individual connections or sub-circuits can then be conducted after having fixed these quantities . The sensitivity measure reveals that the peak in the low-γ range is generated by a sub-circuit consisting of layer 2/3 , layer 4 and the connections from layer 4 to 2/3E and from 2/3E to 4I . This finding is in agreement with experimental literature locating γ oscillations in the upper layers . Furthermore , we identify the feedback connection from 2/3E to 4I and the feed-forward connections from layer 4 to layer 2/3 as crucial for the amplitude of the peak . The oscillation generated by the cooperation of the two upper layers is of lower frequency than the oscillation produced by the layers in isolation . A hint on layers 2/3 and 4 teaming up to generate a low frequency γ peak has been found in Ainsworth et al . [58] . The frequency of the peak is predominantly determined by connections within the input layer 4 . This implies that excitation of the column will be reflected in a frequency shift of the γ peak , which results from an alteration of the dynamical state of the populations and therefore of the effective connectivity . The variability of the generated frequency caused by inputs to layer 4 has been demonstrated experimentally [18 , 59] . The collective oscillations could also be shaped by alterations of the synaptic efficacies between layers 2/3 and 4 ( e . g . by short term plasticity ) . Further experimental studies need to probe the influence of perturbations in weight and number of synapses on the amplitude and frequency of γ peaks in the population rate spectra . The sensitivity measure can be utilized to verify the parameters used in the model and to reveal shortcomings of the theoretical description , which potentially arise from the assumptions of simplified neuron-models and negligible auto-correlations . High-γ peaks are found to be generated in the I-I-loops of each layer , with the loop in layer 4 dominating the spectra . This mechanism , termed ING , has been analyzed previously [20] and experimentally located in upper layers . In the microcircuit , the second largest contribution arises from the I-I-coupling in layer 6; we hence propose to target this layer experimentally to test this hypothesis . Connections determining slow rate fluctuations and the stability of the circuit are identified by the sensitivity measure at zero frequency . The measure shows that connections within layer 5 as well as the connections from population 2/3E and 4E to layer 5 are crucial . We conclude that there are too few connections from 5E to 5I to counteract the rate fluctuations which accumulate due to the amplification within the strong 5E-5E loop . Our findings are in good agreement with experimental results demonstrating the initiation of slow frequency oscillations in layer 5 , as well as the stronger amplification of low frequency oscillations in response to a stimulus applied to layer 5 than to a stimulation of layer 2/3 [38] . Given the dynamical state of 5E , the circuit is stabilized when removing connections from 2/3E to 5E , resulting in a decrease of slow rate fluctuations . In contrast , an impairment of the connections from 4E to 5E has the effect of strengthening the self-amplification of fluctuations and therefore strengthens slow oscillations . With the emerging optogenetic toolbox it may be possible to experimentally test these two predictions in the future . Our analysis suggests a refinement of the parameters of the microcircuit model , which are so far deduced from direct measurements of anatomical and physiological connectivity alone [34] . Experimental studies show that the amplitude of γ oscillations depends on the stimulus strength [60] , suggesting that the current microcircuit model captures the cortical tissue in a semi-stimulated regime . Lowering the external input to the excitatory neurons in layer 4 decreases the low-γ power in the idle state , which in addition sensitizes population 4E to evoke γ oscillations when stimulated . Synaptic delays do not influence the stationary state of the network , characterized by the time-averaged firing rates of all populations , but crucially shape the fluctuations around this stationary set point . We provide an intuitive understanding of the influence of delays on oscillations with parametric plots of the eigenvalues of the activity modes determining the spectra of the circuit . Small delays cause fast oscillations , while long delays support slow ones . Larger delays move the network towards the regime of sustained oscillations , which is counteracted by heterogeneity in the delays . The frequency of the oscillation is highly sensitive to the delays , but the static properties of the circuit , which depend on the dynamic state of the neurons and the anatomical connectivity , determine whether a network displays fast or extremely slow oscillations . The newly derived sensitivity measure determines crucial connections for the frequency and amplitude of population rate oscillations . Since its applicability is not constrained to the analysis of indegrees , it permits a systematic investigation of complicated networks with respect to parameters such as the synaptic delay , connection weight , or excitation-inhibition balance . In these pages we exemplified its use by the analysis of a particular model , but it can in principle be utilized to identify dynamically relevant circuits embedded in any high-dimensional network . Our work thus extends existing methods analyzing single- or two-population network models to more intricate structures . The significance of the identified connections is validated by demonstrating how small changes in the number of synapses can have a large impact on the spectra of all populations . The formalism requires the neurons to work in a regime where the activity fluctuations of the inputs are summed linearly on the considered time scale . Simulations of networks of LIF-model neurons confirm the validity of the linear approximation . Experimental evidence supports the existence of cortical networks operating in this regime [61 , 62 , 63 , 64] . Since the sensitivity measure can be applied to any network whose dynamics can be approximated by a linear rate model , the applicability goes beyond circuits composed of LIF-model neurons . For example , responses of modified IF models have been shown to approximate neural responses in vivo [65] . Several studies treat the stationary and dynamical properties of these models in the linear regime ( see [66] for EIF and [67] for QIF ) . Grabska-Barwinska et al . [68] emphasize that theoretical predictions for networks composed of QIF neurons in the asynchronous regime , by trend , also hold in networks operating in a more synchronized regime , in which individual neurons are exposed to larger input fluctuations . For neuron models with conductance-based synapses a reduction to effective current based synapses exists [69 , 70] and therefore enables the usage of the theoretical framework developed in [13 , 12] . Furthermore , networks of current and conductance based model neurons have been pointed out to be qualitatively comparable ( see section 3 . 5 . 3 . in [65] ) . Alternatively the measure can be fed with experimentally obtained firing rates and transfer functions [71 , 61 , 72] of neuronal populations to analyze the underlying circuits generating the oscillations . The proposed method also finds application in systems where the non-linearities affect the dynamics on a slower time scale than the considered oscillation . Such non-linearities can be taken into account by reevaluating the measure for different mean-inputs corresponding to different phases of the slow input fluctuations . Employing the measure in the described iterative fashion results in a phase-dependent identification of relevant connections for the generation of the fast rhythm and thus sheds light on the anatomical origin of phase-amplitude coupling [reviewed in 21] . The method can also be exploited in reverse to engineer circuits with a desired oscillatory behavior in a top-down fashion . The results presented here lead to clear interpretations of experimental data on network activity and to new hypotheses . It should be noted , however , that the model of the microcircuit represents an early draft and was purposefully designed by its authors as a minimal model with respect to the number of populations and the heterogeneity in the neuronal dynamics . Therefore , failure in the reproduction of certain phenomena found in nature or in the confirmation of a hypothesis should not be attributed to the mathematical method developed here , but to shortcomings of the investigated model . The method is applicable to any update of the original model as structural data and single neuron properties become more refined , given that the assumptions underlying the mean-field and linear response theory are still met . One potential extension is the subdivision of the inhibitory neurons into multiple populations representing different types of inter-neurons , with connection probabilities that yield specific connectivity motifs , as recently reported in [73] . The sensitivity measure uncovers the contribution of the laminar structure to the population rate spectra and produces predictions which can be tested experimentally by comparing the spectra generated by different species or brain areas with distinct laminar structures . In summary the current work introduces a method which elucidates the relation between anatomy and dynamical observables of layered cortical networks . Even though a specific model is used to exemplify the method and to derive concrete predictions , the novel method provides a general framework for the systematic integration of the anatomical and physiological data progressively becoming available into ever more consistent models of cortical circuitry . While analyzing the oscillatory properties of the microcircuit model in this work it turned out that the model with its original parameters [as specified in Table 5 of 34] is in a dynamical regime very close to the onset of sustained population oscillations , resulting in spectra with distinct frequency peaks . We stabilized the circuit by removing 15% of the connections from 4I to 4E and increasing the standard deviation of the delay distribution of all connections to 1 ms . To keep the rates fixed we compensate for the lack of inhibitory input to 4E by removing 19% of the external excitatory input . All simulations were carried out using the simulation software NEST [74] . The source code describing the cortical microcircuit is included in the examples within the release package of NEST as of version 2 . 4 . The population rate spectra from simulations are computed by extracting the spike trains of all neurons in one population from a simulation of 10 s duration and subsequently applying the Fast Fourier Transform ( FFT ) algorithm using a binning of 1 ms . In addition we provide spectra obtained by averaging over 500 ms windows . We here use the term “mean-field theory” for the first step of our analysis , i . e . the equation determining the time-averaged activity characterized by the firing rates of the neurons . This notion , to our knowledge , has its origin in the literature on disordered systems [75 , 76 , 77] and entered the neuroscience literature by the works of Amit et al . [46] for spiking model neurons , Sompolinsky et al . [78] for non-linear rate models and van Vreeswijk et al . [79] for binary model neurons . Note that these theories include synaptic fluctuations . In contrast , mean-field theory in its original meaning is applied to systems without disorder , where it follows from the lowest order saddle point approximation in the local order parameter ( see e . g . [80] , Chapter 4 . 3 , Ferromagnetic transition for classical spins ) , which neglects fluctuations altogether . In the second step of our analysis , we employ linear response theory to characterize the dynamical properties of the populations by a transfer function [13 , 81 , 50] . On the basis of this ingredient , we utilize the finding that a linear rate model with output noise [49] captures the dynamics of circuits composed of LIF-model neurons in the asynchronous irregular regime . The term “rate model” is used in its general sense , as a set of coupled stochastic differential or convolution equations of time-dependent signals . Therefore the observed population-averaged spiking activity yi ( t ) of the i-th population can be interpreted as the fluctuating time density of spike emission ri ( t ) of the neurons with an additive noise component xi ( t ) obeying y i ( t ) = r i ( t ) + x i ( t ) , ⟨ x i ( t ) ⟩ = 0 ⟨ x i ( s ) x j ( t ) ⟩ = δ i j δ ( s - t ) r ¯ i M i , i , j ∈ E , I . ( 10 ) The white noise effectively describes the fluctuations caused by the spiking realization of the point process . Here r ¯ i denotes the average rate of the population with size Mi and the last line shows that the noise produced by different populations is uncorrelated . Correlations between the populations are induced by the connectivity of the network of populations . The rate yi ( t ) describes a signal which fluctuates around the offset ri ( t ) . The amplitude of these fluctuations is infinite in the precisely defined sense of a white noise [82] . The necessity for this additive white noise arises from demanding the equivalence between the original spiking signal and its stochastic counterpart yi ( t ) Eq ( 10 ) on the level of their pairwise statistics: the noise for a rate signal corresponding to a single spike train has to be chosen such that the autocorrelations of the two signals agree . In this case , the white noise generates a Dirac δ peak weighted by the firing rate . The additional factor 1/Mi in Eq ( 10 ) arises from the uncorrelated superposition of Mi such signals [for the formal derivation cf . 49 , esp . Section 4] . Even though the white noise formally has infinite variance , all observable quantities , namely averages over short time intervals , exhibit a finite variance corresponding to that of a Poisson process . In other words , binning a sufficiently long time series y ( t ) with bin size Δt , the variables y ˜ ( t ) = 1 Δ t ∫ t t + Δ t y ( t ′ ) d t ′ ( describing the observed fluctuating spike density in each bin ) are characterized by a distribution with mean r ( t ) and variance r ¯ M 1 Δ t . In this work the validity of the linear approximation is tested by simulations of networks of LIF-model neurons , expressing a non-linearity by their hard threshold on the membrane potential . The description suffices since the network-generated noisy activity effectively linearizes the response of the neurons . This is a fundamental property of non-linear systems subject to noisy inputs , often studied in the context of stochastic resonance in biology [83 , 84 , 85] and reviewed in [86] . In signal processing , the impulse response characterizes the output of a system after the application of a short external input [51] . The time fluctuations of the population rates are obtained by integration over the history of all incoming impulses convolved by the impulse response Hij ( t ) r i ( t ) = ∫ - ∞ t ∑ j = 1 N M i j A H i j ( t - s ) r j ( s - d i j ) + x j ( s - d i j ) d s , ( 11 ) where dij denotes the delay of the connection from population j to i . The impulse response Hij ( t ) of a population of LIF-neurons is obtained by applying linear response theory to the corresponding Fokker-Planck Eq ( 13 ) . We here use the recently derived extension incorporating exponentially decaying synaptic currents [50 , Eq ( 30 ) ] . The effective connectivity matrix M ( t ) with elements M i j ( t ) = M i j A H i j ( t ) ( 12 ) summarizes the rate response of population i to an impulse sent from population j . This matrix has two contributions . The first part , termed the anatomical connectivity M i j A , determines the size of the incoming input . The anatomical connectivity matrix is element-wise composed of the indegree matrix K and the weight matrix W M i j A = K i j W i j , W i j = J E if j ∈ E J I if j ∈ I . ( 13 ) Here Kij describes the number of incoming connections from population j to population i and Wij their respective weight . The second part describes the time course of the rate response Hij ( t ) . The substitution s → s + d when integrating Eq ( 11 ) permits the absorption of the time delay into the effective connectivity matrix Md ( t ) = M ( t − d ) . Transforming Eq ( 11 ) into Fourier space yields R ( ω ) = M ˜ d ( ω ) ( R ( ω ) + X ( ω ) ) ⇒ R ( ω ) = ( M ˜ d - 1 ( ω ) - I ) - 1 X ( ω ) ( 14 ) with M ˜ d , i j ( ω ) = M ˜ i j ( ω ) e - i ω d i j . Since the delays are Gaussian distributed we need to average over all possible realizations of the delays . This averaging can formally be done by weighting the contributions involving the delays with the probability density function f ( y ) describing the delay distribution e - i ω d i j → ∫ - ∞ ∞ e - i ω y f ( y ) d y . Here the probability function is given by a renormalized Gaussian distribution truncated at zero ( since the delays are positive ) , yielding the effective connectivity M ˜ d , i j ( ω ) = M ˜ i j ( ω ) 2 π σ d i j 1 - Φ - d i j σ d i j ∫ 0 ∞ d y e - i ω y e - ( y - d i j ) 2 2 σ d i j 2 , = 1 - Φ - d i j + i ω σ d i j 2 σ d i j 1 - Φ - d i j σ d i j M ˜ i j ( ω ) e - i ω d i j e - σ d i j 2 ω 2 2 ( 15 ) with σ d i j being the standard deviation of the delay from population j to population i , dij the average delay , and Φ ( x ) = 1 2 1 + erf x 2 , with the error function erf ( x ) . The integration can be performed for any probability density function . Therefore the formalism generalizes to models incorporating delay heterogeneities with other statistics than a Gaussian distribution . The activity composed of the output rate and the additional noise is thus given by Y ( ω ) = R ( ω ) + X ( ω ) = P ( ω ) X ( ω ) , ( 16 ) where we define P ( ω ) = ( I - M ˜ d ( ω ) ) - 1 as the propagator determining how the noise is mapped via the network onto the observable activity Y . The cross-correlations between the activities are given by C ( ω ) = ⟨ Y ( ω ) Y T ( - ω ) ⟩ = P ( ω ) D P T ( - ω ) , ( 17 ) where D = 〈X ( ω ) XT ( − ω ) 〉 is the diagonal matrix of correlations between the effective noise sources X , which represent the spiking realization of the neuronal signals . Due to the initial independence of the neurons Eq ( 10 ) , the correlation matrix has diagonal form with the elements defined by the average firing rate of the neurons and the population size ( D i i = r ¯ i / M i ) . The stationary firing rates of LIF model neurons supplied with colored noise is derived in Fourcaud et al . [47] . The spectrum of the i-th population can be directly read off the diagonal of the cross-correlation C i i ( ω ) = ⟨ Y ( ω ) Y T ( - ω ) ⟩ i i . ( 18 ) In Fourier space the effective connectivity matrix is a function of frequency ω . For every frequency the matrix can be decomposed , resulting in N = 8 eigenvalues with the corresponding left and right eigenvectors M ˜ ( ω ) u i ( ω ) = λ i ( ω ) u i ( ω ) v i T ( ω ) M ˜ ( ω ) = λ i ( ω ) v i T ( ω ) . ( 19 ) The eigenvectors are normalized such that the product of the left and right eigenvector equals one . The propagator shares its eigenvectors with the effective connectivity matrix and the eigenvalues are given by P ( ω ) u i ( ω ) = 1 1 - λ i ( ω ) u i ( ω ) . ( 20 ) The noise can be expressed in the new basis as X ( ω ) = ∑ i α i ( ω ) u i ( ω ) , α i ( ω ) = v i T ( ω ) X ( ω ) . ( 21 ) Hence the cross-correlations in the new basis take the form C ( ω ) = ∑ i , j = 1 N α i ( ω ) α j * ( ω ) ( 1 - λ i ( ω ) ) ( 1 - λ j * ( ω ) ) ︸ ≕ β i j ( ω ) u i ( ω ) u j * T ( ω ) ︸ ≕ T i j ( ω ) . ( 22 ) Here Tij ( ω ) is the matrix given by the outer product of the eigenvectors of the i-th and j-th mode evaluated at frequency ω , where we employed u i ( - ω ) = u i * ( ω ) . This relation holds since the impulse response Hi ( t ) entering the effective connectivity matrix is real valued in the time domain . When one eigenvalue approaches unity at frequency ω0 ( λc ( ω0 ) ≈ 1 ) , the spectrum at this frequency is dominated by the contribution of the critical mode c and we can approximate the spectrum visible in the k-th population by C k k ( ω 0 ) ≈ \| α c ( ω 0 ) 1 - λ c ( ω 0 ) \| 2 u c , k ( ω 0 ) u c , k * ( ω 0 ) = β c c ( ω 0 ) T c c , k ( ω 0 ) . ( 23 ) In a simplified circuit with all populations having the same transfer function H ( ω ) the eigenvalue decomposition of the effective connectivity matrix reads M ˜ ( ω ) = H ( ω ) ∑ i = 1 N λ i A u i A v i A , T . ( 24 ) Here λ i A is the i-th eigenvalue of the anatomical connectivity matrix and u i A and v i A are the associated right and left eigenvectors , respectively . The propagator matrix Eq ( 20 ) , mapping the noise of the system to the rate , is determined by the effective connectivity matrix and thus has the same eigenvectors and the eigenvalues 1 / ( 1 - H ( ω ) λ i A ) . Mapping the rate vector R ( ω ) into the coordinate system spanned by the right and left eigenvectors of the anatomical connectivity matrix ( u i A , v i A , T ) , the rates of the initial populations R i ( ω ) = e i T R ( ω ) ( where e i is the unit vector being one at position i and zero everywhere else ) are converted to the dynamic modes v i A , T R ( ω ) . Fig 3A shows a scheme of the coordinate transformation . The activity of the i-th mode is fed back solely to itself with the connection weight λ i A v i A , T u i A and filtered by the transfer function H ( ω ) . By expressing the ongoing spiking activity propagating through the system Eq ( 21 ) as a linear combination of the eigenmodes , the total activity is described by the sum of the activity of decoupled modes . The diagonal elements of the cross-correlation matrix describing the spectrum of the populations can be expressed in the new basis C k k ( ω ) = ∑ i , j = 1 N α i ( ω ) α j * ( ω ) ( 1 - H ( ω ) λ i A ) ( 1 - H * ( ω ) λ j A * ) ︸ ≕ β i j A ( ω ) u i , k A u j , k A * ︸ ≕ T i j , k A , k ∈ 1 , . . , N , ( 25 ) with α i ( ω ) = v i A , T X ( ω ) being the projection of the noise into the new coordinate system . The contribution of one mode dominates if H ( ω 0 ) λ c A ≈ 1 and we can approximate the spectrum at ω0 with C k k ( ω 0 ) ≈ \| α c ( ω 0 ) 1 - H ( ω 0 ) λ c A \| 2 u c , k A u c , k A * = β c c A ( ω 0 ) T c c , k A . ( 26 ) This section devises a method to break a circuit down into smaller independent circuits each describing distinct characteristics of the spectrum by means of eigenvalue decomposition of the effective connectivity matrix . The activity R k ( ω ) = e k T R ( ω ) of population k is given by R k ( ω ) = ∑ l = 1 N M ˜ d , k l ( ω ) ( R l ( ω ) + X l ( ω ) ) = H k ( ω ) ∑ l = 1 N M k l A ( R l ( ω ) + X l ( ω ) ) . ( 27 ) and illustrated in the top of Fig 3A . We now consider a simplified circuit where all populations have the same transfer functions . Here , the anatomical and dynamical part of the effective connectivity can be treated separately M ˜ i j ( ω ) = H ( ω ) M i j A . ( 28 ) The anatomical part M i j A can be split into eight modes using eigenvalue decomposition Eq ( 24 ) yielding the activity of one eigenmode R ˜ k ( ω ) = v k A , T R ( ω ) R ˜ k ( ω ) = v k A , T ∑ i = 1 N λ i A H ( ω ) u i A v i A , T ( R ( ω ) + X ( ω ) ) = λ k A H ( ω ) v k A , T ( R ( ω ) + X ( ω ) ) = λ k A H ( ω ) ( R ˜ k ( ω ) + X ˜ k ( ω ) ) , as visualized in the bottom left of Fig 3A . Since the eigenvectors of the effective connectivity matrix for homogeneous transfer functions are frequency independent , the mapping to the mode activity R ˜ k ( ω ) is also constant across frequencies . The modes can be considered as decoupled circuits , whose activity is fed back to itself and can be treated in isolation . I . e . an adjustment of the connectivity of one mode does not influence the activity of another mode . The sum of activities in the circuit , however , is independent of the representation R ( ω ) = ∑ i = 1 N R i ( ω ) e i = ∑ i = 1 N R ˜ i ( ω ) u i A . The spectrum produced by the original circuit is given by the sum of the spectra generated by all possible mode pairs Eq ( 25 ) C k k ( ω ) = ∑ i , j = 1 N β i j A ( ω ) T i j , k A , k ∈ 1 , . . , N . ( 29 ) Here the spectrum visible in population k receives contributions from all mode pairs i and j . The prefactors βAij ( ω ) are common to the spectrum of all populations and thus determine the global frequency dependence of the spectra . The visibility of the global characteristics of the spectrum in the spectra of the individual populations is determined by the frequency independent factor T i j , k A . The prefactor βAij ( ω ) is large if one of the eigenvalues of the effective connectivity matrix comes close to one at a particular frequency ω0 , resulting in a peak of the spectrum . Therefore , at peak frequency ω0 the contribution of the critical mode ( i = j = c ) constitutes the dominant part of the spectrum and we can approximate the spectrum of the circuit by the spectrum of the critical mode Eq ( 26 ) C k k ( ω 0 ) ≈ β c c A ( ω 0 ) T c c , k A . ( 30 ) The anatomical sub-circuit responsible for the peak can now directly be deduced from the definition of T c c , k A as the outer product of the eigenvectors of the critical mode . Removing the correlations induced by these connections ( i . e . substituting the input provided by these connections with white noise ) from the anatomical connectivity matrix M A → M A - λ c A u c A v c A , T removes the contributions of the critical mode ( in particular the peak in the spectrum ) , but leaves contributions of the remaining modes to the spectrum unaltered . The assumption of identical transfer functions of the populations entering the previous argument requires equal dynamic states of all populations . This in turn results in all populations displaying the same firing rates , which disagrees with experimental findings . We therefore need to take population specific transfer functions into account , resulting in frequency dependent eigenvectors and eigenvalues and hence frequency dependent representations of the dynamical modes . The activity of one mode is now given by R ˜ k ( ω ) = v k T ( ω ) R ( ω ) R ˜ k ( ω ) = λ k ( ω ) ( R ˜ k ( ω ) + X ˜ k ( ω ) ) as illustrated in the bottom right of Fig 3A . In this case not only the prefactor but also the outer product of the eigenvectors is frequency dependent C k k ( ω 0 ) ≈ β c c ( ω 0 ) T c c , k ( ω 0 ) . ( 31 ) The peak in the spectrum and hence the dynamics of the critical eigenmode at ω0 could be removed by the adjustment M ( ω ) → M ( ω ) - λ c ( ω ) u c ( ω ) v c T ( ω ) . However , due to the frequency dependence of the mode representation , the same set of anatomical connections relevant for this mode at ω0 might also take part in the generation of the dynamics of another mode at a different frequency . Therefore removing the dynamical contribution of the anatomical connections contributing to one mode at one frequency will remove this particular oscillation , but may also impair other modes . The rates observed in population k are given by R k pop ( ω ) = e k T R ( ω ) = ∑ j = 1 N M ˜ d , k j ( ω ) ( R j pop ( ω ) + X j pop ( ω ) ) ( 32 ) with X j pop ( ω ) = e j T X ( ω ) representing the noise projected into the direction of the j-th population . The rate of one dynamical mode is given by R k mode ( ω ) = v k T ( ω ) R ( ω ) = λ k ( ω ) ( R k mode ( ω ) + X k mode ( ω ) ) ( 33 ) with X j mode ( ω ) = v k T ( ω ) X ( ω ) representing the noise projected into the direction of the k-th mode . From the latter expression we conclude , that the eigenvalue trajectory acts as the transfer function of the dynamical modes . Considering Fig 4A and 4B we observe that the shape of the eigenvalue trajectory can roughly be approximated by a low pass filter with additional factors accounting for the mean delay and the delay distribution λ k ( ω ) ≈ λ k ( 0 ) 1 + i ω τ k eff e - i ω d k eff e - σ d k eff 2 ω 2 2 . The time constant τ k eff , the delay d k eff , as well as the variance of the delays σ d k eff 2 are effective parameters which , in multi-dimensional networks , are analytically intractable functions of the network parameters . Their definitions , however , suffice to gain an intuition for changes in synaptic parameters . The effective time constant determines the convergence speed of the trajectory towards zero . Hence larger effective time constants , potentially arising from large synaptic time constants , prevent the transmission of large frequencies . Larger effective delays accelerate the spinning of the trajectory , resulting in resonances at smaller frequencies , but also support rate instabilities in terms of Hopf bifurcations ( see “Stability of the dynamical modes” ) Larger variances of the effective delays compress the eigenvalue trajectory , resulting in smaller peaks in the spectrum . This effect of desynchronization of population dynamics by heterogeneity in the connection parameter has been demonstrated previously [87] . To analyze the stability of the circuit , we consider the convolution equation , that describes the rates in a self-consistent manner , without noise r i ( t ) = ∫ - ∞ t ∑ j = 1 N M i j A H i j ( t - s ) r j ( s - d i j ) d s = ∑ j = 1 N M i j A H i j * r j ( ∘ - d i j ) . ( 34 ) The variable describing the rate of each population can be replaced by its Laplace back-transformation r i ( t ) = 1 2 π i ∫ - i ∞ i ∞ e z t R i ( - i z ) d z ( 35 ) for complex z . Here Ri ( ω ) is the Fourier transform of ri evaluated at the complex Laplace frequency z = iω . Since convolutions simplify to multiplication in Laplace space we get 1 2 π i ∫ - i ∞ i ∞ d z R i ( - i z ) - ∑ j H i j ( - i z ) e - z d i j M i j A ︸ = M ˜ i j ( - i z ) R j ( - i z ) e z t = 0 ⇒ 1 - M ˜ ( - i z ) R ( - i z ) = 0 . ( 36 ) This condition is fulfilled if either R ( −iz ) is an eigenvector R ^ ( - i z ) of M ˜ ( - i z ) with eigenvalue λ ( − iz ) = 1 or R ( − iz ) equals zero . The integration in Eq ( 35 ) can hence be rewritten as a sum over all solutions z′ ∈ Z′ for which M ˜ ( - i z ′ ) R ^ ( - i z ′ ) = R ^ ( - i z ′ ) r i ( t ) = 1 2 π i ∫ - i ∞ i ∞ e z t R i ( - i z ) d z = 1 2 π i ∫ - i ∞ i ∞ e z t ∑ z ′ ∈ Z ′ α z ′ R ^ i ( - i z ′ ) δ ( z - z ′ ) d z = 1 2 π i ∑ z ′ ∈ Z ′ α z ′ R ^ i ( - i z ′ ) e z ′ t , where αz′ are as yet undetermined constants that could be determined when tackling the inhomogenous problem . Expressed in Fourier domain we have z = iω so that ez′t turns into eiω′t and R ^ i ( - i z ) into R ^ i ( ω ) r i ( t ) = 1 2 π i ∑ ω ′ ∈ Ω ′ α i ω ′ R ^ i ( ω ′ ) e i ω ′ t = 1 2 π i ∑ ω ′ ∈ Ω ′ α i ω ′ R ^ i ( ω ′ ) e ( - ℑ ( ω ′ ) + i ℜ ( ω ′ ) ) t . ( 37 ) The transfer function h ( t ) and therefore the effective connectivity as well as the activity ri ( t ) are real valued functions in time domain . With the complex component in Fourier domain originating from the argument ω′ , we conclude that M ˜ ( - ω ′ ) = M ˜ * ( ω ′ ) has the eigenvector R ^ ( - ω ′ ) = R ^ * ( ω ′ ) if ω ′ ∈ R . Thus Ω′ contains pairs of values ω + ′ = ℜ ( ω ′ ) + i ℑ ( ω ′ ) and ω - ′ = - ℜ ( ω ′ ) + i ℑ ( ω ′ ) for each ω′ . Eq ( 37 ) can hence be written as: r i ( t ) = 1 π ∑ ω ′ ∈ Ω + ′ ℜ α i ω ′ R ^ i ( ω ′ ) e i ℜ ( ω ′ ) t e - ℑ ( ω ′ ) t . ( 38 ) with Ω + ′ containing all ω + ′ ∈ Ω ′ . The equation above reveals that unstable modes exist if there is a solution with ℑ ( ω′ ) < 0 . In the context of the eigenvalue trajectories ( Fig 4A ) in the microcircuit one needs to investigate the solutions for the eigenvalues for complex frequencies ω . It turns out that all eigenvalue trajectories spiraling around the right side of the critical value one exhibit an unstable solution ( λ ( ω′ ) = 1 for ℑ ( ω′ ) < 0 ) ( see inset in Fig 4A ) and vice versa . If the course of an eigenvalue trajectory is altered by changes in parameters ( for example indegrees ) such that it passes the value one for a real valued frequency ( λ ( ω′ ) = 1 with ℑ ( ω′ ) = 0 ) , the system would undergo a Hopf bifurcation . This can also be shown by mapping Eq ( 36 ) to the system discussed in [Eq . 2 . 8 in 88] .	Recordings of brain activity show multiple coexisting oscillations . The generation of these oscillations has so far only been investigated in generic one- and two-population networks , neglecting their embedment into larger systems . We introduce a method that determines the mechanisms and sub-circuits generating oscillations in structured spiking networks . Analyzing a multi-layered model of the cortical microcircuit , we trace back characteristic oscillations to experimentally observed connectivity patterns . The approach exposes the influence of individual connections on frequency and amplitude of these oscillations and therefore reveals locations , where biological mechanisms controlling oscillations and experimental manipulations have the largest impact . The new analytical tool replaces parameter scans in computationally expensive models , guides circuit design , and can be employed to validate connectivity data .	[ "Abstract", "Introduction", "Results", "Discussion", "Methods" ]	[ "action", "potentials", "medicine", "and", "health", "sciences", "neural", "networks", "engineering", "and", "technology", "electrical", "circuits", "population", "dynamics", "membrane", "potential", "electrophysiology", "neuroscience", "mathematics", "algebra", "population...	2016	Identifying Anatomical Origins of Coexisting Oscillations in the Cortical Microcircuit
Snake envenoming is a significant public health problem in underdeveloped and developing countries . In sub-Saharan Africa , it is estimated that 90 , 000–400 , 000 envenomations occur each year , resulting in 3 , 500–32 , 000 deaths . Envenomings are caused by snakes from the Viperidae ( Bitis spp . and Echis spp . ) and Elapidae ( Naja spp . and Dendroaspis spp . ) families . The African continent has been suffering from a severe antivenom crisis and current antivenom production is only sufficient to treat 25% of snakebite cases . Our aim is to develop high-quality antivenoms against the main snake species found in Mozambique . Adult horses primed with the indicated venoms were divided into 5 groups ( B . arietans; B . nasicornis + B . rhinoceros; N . melanoleuca; N . mossambica; N . annulifera + D . polylepis + D . angusticeps ) and reimmunized two times for antivenom production . Blood was collected , and plasma was separated and subjected to antibody purification using caprylic acid . Plasmas and antivenoms were subject to titration , affinity determination , cross-recognition assays and in vivo venom lethality neutralization . A commercial anti-Crotalic antivenom was used for comparison . The purified antivenoms exhibited high titers against B . arietans , B . nasicornis and B . rhinoceros ( 5 . 18 x 106 , 3 . 60 x 106 and 3 . 50 x 106 U-E/mL , respectively ) and N . melanoleuca , N . mossambica and N . annulifera ( 7 . 41 x 106 , 3 . 07 x 106 and 2 . 60 x 106 U-E/mL , respectively ) , but lower titers against the D . angusticeps and D . polylepis ( 1 . 87 x 106 and 1 . 67 x 106 U-E/mL ) . All the groups , except anti-N . melanoleuca , showed significant differences from the anti-Crotalic antivenom ( 7 . 55 x 106 U-E/mL ) . The affinity index of all the groups was high , ranging from 31% to 45% . Cross-recognition assays showed the recognition of proteins with similar molecular weight in the venoms and may indicate the possibility of paraspecific neutralization . The three monospecific antivenoms were able to provide in vivo protection . Our results indicate that the anti-Bitis and anti-Naja antivenoms developed would be useful for treating snakebite envenomations in Mozambique , although their effectiveness should to be increased . We propose instead the development of monospecific antivenoms , which would serve as the basis for two polyvalent antivenoms , the anti-Bitis and anti-Elapidae . Polyvalent antivenoms represent an increase in treatment quality , as they have a wider range of application and are easier to distribute and administer to snake envenoming victims . Snake envenoming prevention and treatment has been a worldwide effort in underdeveloped and developing countries in Africa , Asia and South America . In sub-Saharan Africa , it is estimated that 90 , 000–400 , 000 envenomations occur per year , resulting in 3 , 500–32 , 000 fatalities [1] . Amputations and complications caused by snake envenoming affect 3% and 5 . 5% of envenomation victims , respectively [2] . As accidents with snake occur mostly with young men in rural areas , they can result in the removal of these men from the workforce , causing great personal and economic financial losses . Snake envenoming is caused mainly by snakes from the Viperidae ( Echis spp . and Bitis spp . ) and Elapidae ( Naja spp . and Dendroaspis spp . ) families . Viperidae snakes have a venom rich in metalloproteinases ( SVMP ) that can cause hemorrhagic effects and coagulatory-inducing disturbances [3] . Echis occelatus is responsible for most accidents [4] , while Bitis arietans has the widest territorial distribution [5] . Elapidae snake venoms have a more pronounced neurotoxic action , targeting neuromuscular junctions , and accidents can evolve to respiratory failure [6] . Spitting cobra bites ( Naja spp . ) are regarded as the most medically important due to their lethality [7] . The most effective treatment against snakebite envenoming is the administration of specific antivenom . Antivenom was introduced in Africa in 1950; there were three major producers–Behringwerke A . G . ( Germany ) , Sanofi-Pasteur ( France ) and the indigenous South African Institute for Medical Research ( SAIMR ) [8] . After the 1980’s , the European companies ceased or greatly reduced their production due to the high cost of antibody production , and SAIMR struggled financially . The present production of antivenom ( 200 , 000 ampules/year ) meets less than 25% of the African continent’s demand for snakebite treatment [9] . In an effort to solve the problem , African authorities began importing antivenoms from India and Asia . These antivenoms are not specific against African snakes and this treatment has little efficacy , causing the population to be distrustful and look for alternatives , such as traditional healing routes [10] . Even with a new wave of antivenoms being researched [11 , 12 , 13] , there is still much to be done towards fighting snakebite envenomation in sub-Saharan Africa . In this study , we concentrate on the development of antivenoms against eight snake species found in Mozambique: Bitis arietans , B . nasicornis , B . rhinoceros , Naja melanoleuca , N . mossambica , N . annulifera , Dendroaspis angusticeps and D . polylepis . Our results are promising , and provide the basis for polyvalent antivenoms that could prove viable for widespread use . Tris buffer ( Tris HCl , 25 mM; pH 7 . 4 ) , complete MMT80 ( Marcol Montanide ISA 50 , 2 mL; sodium chloride 0 . 15 M , 5 mL; Tween 80 , 1 mL; lyophilized BCG , 1 mg ) , incomplete MMT80 ( Marcol Montanide ISA 50 , 2 mL; sodium chloride 0 . 15 M , 5 mL; Tween 80 , 1 mL ) , solution A for SDS buffer ( Tris , 6 . 25 mM; SDS , 6 . 94 mM; pH 6 . 8 ) , SDS buffer for non-reducing conditions ( solution A , 8 . 5 mL; glycerol , 1 mL; bromophenol blue 1% , 2 mL ) , PBS buffer ( potassium chloride , 2 . 6 mM; monobasic potassium phosphate , 1 . 5 mM; sodium chloride , 76 mM; disodium phosphate , 8 . 2 mM; pH 7 . 2–7 . 4 ) , AP buffer ( Tris HCl , 100 mM; sodium chloride , 100 mM; magnesium chloride , 5 mM; pH 9 . 5 ) , NBT solution ( NBT , 50 mg; dimethylphormamide , 700 μL; H2O , 300 μL ) , BCIP solution ( BCIP , 50 mg; dimethylphormamide , 1 mL ) , developing solution for Western blot ( AP buffer , 5 mL; NBT solution , 33 μL; BCIP solution , 16 . 5 μL ) , citrate buffer ( citric acid , 0 . 1 M; monobasic sodium phosphate , 0 . 2 M; pH 5 . 0 ) , OPD solution ( OPD , 20 mg; citric acid , 1 mL ) and substrate buffer for ELISA ( citrate buffer , 5 mL; OPD solution , 100 μL; H2O2 30 volumes , 5 μL ) were used . Except for the NBT/BCIP , obtained from Molecular Probes ( USA ) , the reagents used were obtained from Sigma-Aldrich ( USA ) . The protein concentrations of the venoms and plasma/antivenoms were assessed by the bicinchoninic acid method [14] using a Pierce BCA Protein Assay Kit ( Rockford , IL ) . B . arietans , B . nasicornis , B . rhinoceros , N . melanoleuca , N . mossambica , D . angusticeps , D . polylepis and N . annulifera venoms were supplied by Venom Supplies Pty Ltd ( 59 Murray Street , Tanunda , Australia ) and stored at Laboratório de Venenos , Instituto Butantan . Each venom batch was made from sample mixtures of several snake specimens and lyophilized . Adult horses ( 400–450 kg ) were used to produce the anti-venoms , and they were divided into 5 groups: anti-B . arietans , n = 12; anti-B nasicornis + B . rhinoceros , n = 12; anti-N . melanoleuca , n = 12; anti-N . mossambica , n = 6; anti-D . angusticeps + D . polylepis + N . annulifera , n = 9 . The animals were primed ( 4 injections , 15 days apart ) and maintained in a special animal house at the São Joaquim Farm , Instituto Butantan , São Paulo , Brazil . Before immunization , the animals were vaccinated against common equine infectious diseases . All animals used in this study were maintained and treated under strict ethical conditions in accordance with the “International Animal Welfare recommendations” [15] and the “Committee Members , International Society on Toxinology” [16] . This project was approved by the Ethics Committee of Animal Usage in Research ( Protocol No: 1137/13 ) of the Instituto Butantan . The horses received the following mixtures: anti-B . arietans ( n = 12 ) , 3 . 5 mg/animal of crude B . arietans venom; anti-B . nasicornis + B . rhinoceros ( n = 12 ) , 3 . 5 mg/animal of crude B . nasicornis and B . rhinoceros venom mixture ( 1:1 ) ; anti-N . melanoleuca ( n = 12 ) , 3 . 5 mg/animal of crude N . melanoleuca venom; anti-N . mossambica ( n = 6 ) , 3 . 5 mg/animal of crude N . mossambica venom; anti-D . angusticeps + D . polylepis + N . annulifera ( n = 9 ) , 3 . 5 mg/animal of crude D . angusticeps , D . polylepis and N . annulifera venom mixture ( 1:1:1 ) . The subcutaneous injections were performed 15 days apart at four different points in the dorsal region of each animal . The animals were primed ( 4 inoculations in 6 mL of complete or incomplete MMT80 adjuvant ) , and later re-immunized for this experiment ( 2 inoculations in 6 mL of PBS ) . Fifteen days after each inoculation post-priming , blood was collected ( 8 mL/animal ) in vials with anticoagulant solution ( heparin by venipuncture of the jugular vein ) . The plasma was separated by centrifugation ( 1 , 500 rpm for 15 min at 4°C ) and stored at -20°C . Horse blood was collected and the plasma was separated as described above . Four equine plasma samples ( Batch No: #143 , #158 , #223 and #356 ) from horses immunized with C . d . terrificus venom , according to WHO guidelines [17] , were used to produce anti-Crotalic serum . These samples were provided by “Divisão de Desenvolvimento Tecnológico e Produção–Seção de Processamento de Plasmas Hiperimunes , Instituto Butantan” , and they were used as a standard for comparison . The obtained plasma samples were pooled within their respective groups and immunoglobulins were purified using caprylic acid . Equine antibodies were purified from plasma ( 2nd immunization samples ) following the procedure described by Dos Santos et al . [18] . The plasma samples were pooled within their respective groups , heated at 56°C for 15 min to achieve complement inactivation and centrifuged at 900 g for 10 min . The pH of the samples was adjusted to 5 . 0 through the addition of 0 . 1 N acetic acid . Caprylic acid was added slowly with vigorous shaking to a final concentration of 8 . 7% , and then the samples were left to shake for 30 min at room-temperature . After a second centrifugation step ( 11 , 000 g for 15 min ) , the supernatants were collected , and their pH were adjusted to between 7 . 0 and 7 . 5 through the addition of 0 . 1 N NaOH . The samples were filtered through a Sterile Millex 0 . 45 μm ( Milliford Corporation , Billeric , MA ) . Dialysis was performed against PBS buffer using a Centricon 50 kDa Centrifugal Device ( Milliford Corporation , Billeric , MA ) 4 , 000 g for 10 min , three times . The samples were then diluted to 30 mg/mL and stored at -20°C . Polystyrene , high-affinity ELISA plates ( 96 wells ) were coated with 1 . 0 μg/well of crude B . arietans , B . nasicornis , B . rhinoceros , N . melanoleuca , N . mossambica , D . angusticeps , D . polylepis or N . annulifera venoms in 100 μL of PBS buffer and kept overnight at 4°C . In some assays , the plates were coated with 1 . 0 μg/well of crude C . d . terrificus venom for standard antivenom quantification . The plates were blocked for 2 h at 37°C with 200 μL/well of PBS plus 5% BSA . The plates were washed with 200 μL/well of PBS . Serial dilutions of horse plasma ( 1:4 , 000 to 1:512 , 000 ) or IgG antivenoms ( 1:2 , 000 to 1 , 024 , 000 ) in PBS plus 0 . 1% BSA were prepared , and 100 μL/well of each dilution was added to their respective antigens . The plates were then incubated at 37°C for 1 h , and washed three times with the wash buffer ( PBS plus 0 . 1% BSA and 0 . 05% Tween-20 , 200 μL/well ) . Rabbit peroxidase-conjugated anti-horse IgG ( whole molecule ) ( Sigma Aldrich , St . Louis , MO ) diluted ( 1:20 , 000 ) in PBS plus 0 . 1% BSA ( 100 μL/well ) was added to the plates . The plates were incubated for 1 h at 37°C . After three washes with the wash buffer , 50 μL/well of the substrate buffer was added , and the plates were incubated at room temperature for 15 min . The reaction was terminated by the addition of 50 μL/well of 4 N sulfuric acid . The absorbance at 492 nm was recorded using an ELISA plate reader ( Labsystems Multiskan Ex , Thermo Fisher Scientific Inc . , Walthan , MA ) . IgG from horses collected before immunization was used as a negative control ( fixed dilution of 1:2 , 000 ) . The antivenom dilution with an optical density of 0 . 2 was used to calculate the U-ELISA per milliliter of the undiluted antivenom solution . One U-ELISA was defined as the smallest dilution of antibody that presented an OD of 0 . 2 under the conditions used in the ELISA assay , as described previously [19] . The value was then multiplied by 10 to convert it to milliliters . The affinities of the horse plasmas ( 2nd immunization samples ) or IgG antivenoms were measured using ELISA , following the methodology described above , with the inclusion of a potassium thiocyanate ( KSCN ) elution step [20 , 21 , 22] . After the serum incubation step , dilutions of KSCN ( 0 . 0 to 5 . 0 M , in intervals of 1 . 00 M ) in distilled H2O were added to the wells and incubated for 30 min at room temperature . The remaining bound antibodies were detected with rabbit peroxidase-conjugated anti-horse IgG ( whole molecule ) ( Sigma Aldrich , St . Louis , MO ) diluted ( 1:20 , 000 ) in PBS plus 0 . 1% BSA ( 100 μL/well ) . After three washes with the wash buffer , 50 μL/well of substrate buffer was added , and the plates were incubated at room temperature for 15 min . The reaction was terminated with 50 μL/well of 4 N sulfuric acid . The absorbance at 492 nm was recorded using an ELISA plate reader ( Labsystems Multiskan Ex , Thermo Fisher Scientific Inc . , Waltham , MA ) . The results are expressed as follows: affinity index ( AI ) = % bound antibodies at KSCN 5 M . Western blot analysis was carried out according to the method previously described by Towbin et al . [23] . Crude B . arietans , B . nasicornis , B . rhinoceros , N . melanoleuca , N . mossambica , D . angusticeps , D . polylepis and N . annulifera venoms ( 10 μg ) were treated with non-reducing SDS-PAGE sample buffer and resolved in 12 . 5% polyacrylamide gel . Gels were either submitted to silver staining or electroblotted onto nitrocellulose membranes , according the method described by Laemmli [24] . These membranes were blocked with PBS buffer containing 5% BSA at 37°C for 2 h , washed with PBS , and treated with the antivenom diluted to 1:20 , 000 in PBS plus 0 . 1% BSA for 1 h at room temperature on a horizontal shaker . Each membrane was treated with only one antivenom . After being washed three times with PBS plus 0 . 05% Tween-20 , the membranes were incubated with rabbit anti-horse IgG conjugated to alkaline phosphatase ( whole molecule ) diluted to 1:7 , 500 in PBS plus 0 . 1% BSA . Then , the membranes were incubated for 1 h at room temperature on a horizontal shaker . The membranes were washed three times with PBS plus 0 . 05% Tween-20 and placed in developing solution for Western blotting . The reaction was terminated by washing with distilled water . Male Swiss mice , 18–20 g , were used in protocols to determine the lethality ( LD50 ) of the venoms and the neutralizing potency ( ED50 ) of the antivenoms . For LD50 determination , different venom quantities were prepared in 0 . 85% NaCl solution and intraperitoneally injected ( 500 μL ) in the mice . Four mice were used per venom dose . Deaths were recorded after 48 h , and LD50 was estimated by probits analysis [25] . For ED50 determination , a fixed amount of 3 LD50 of snake venom and various dilutions of the respective purified antivenoms ( 1:5 , 1:10 , 1:20 , and 1:40 ) were incubated for 30 min at 37°C . Venom samples incubated only with 0 . 85% NaCl solution were used as controls . After incubation , 500 μL aliquots of the mixtures were intraperitoneally injected in the mice . Four mice were used per dilution . The death/survival ratio was recorded at 3 h , 24 h and 48 h after the injection . ED50 was estimated by probits analysis . Data is expressed as Specific Activity ( μg of venom neutralized by 1mg of antibodies ) . The data was analyzed using one-way ANOVA , followed by the Dunnett’s Multiple Comparison Test ( standard antivenom as comparison ) , or two-way ANOVA , followed by the Bonferroni Post-Test ( standard antivenom as comparison ) . Differences were considered to be significant if P < 0 . 05 . Analysis was performed using GraphPad Prism 5 for Windows ( GraphPad Software , San Diego , USA ) . The titration of the plasmas against the respective venoms was performed using ELISA . The plates were sensitized with 1 μg/well of antigen , and the plasma dilutions ranged from 1:4 , 000 to 1:512 , 000 . Titration curves from different immunizations and groups showed similar kinetics ( Fig 1A ) . When titers were compared , differences between different immunizations were not found ( Fig 1B ) . The groups anti-B arietans and anti-B . nasicornis + B . rhinoceros showed similar results , with titers ranging from 3 . 47 x 106 to 4 . 62 x 106 U-ELISA/mL . The anti-N . mossambica and anti-N . melanoleuca groups displayed the highest titers , 4 . 55 x 106 to 5 . 12 x 106 U-ELISA/mL . Significant differences were not found between those groups and the standard anti-Crotalic antivenom ( 4 . 21 x 106 U-ELISA/mL ) . The group anti-D . angusticeps + D . polylepis + N . annulifera showed higher titers against N . annulifera venom ( from 2 . 67 x 106 up to 3 . 93 x 106 U-ELISA/mL ) , and lower titers against the Dendroaspis venoms , between 1 . 46 x 106 and 2 . 23 x 106 U-ELISA/mL . This group was also the only one to show a statistically significant difference when compared to the standard anti-Crotalic antivenom . Affinity determination was performed by ELISA with the inclusion of an elution step with KSCN , a chaotropic agent , in concentrations ranging from 0 M to 5 M . Only plasma samples from the 2nd immunization were used , and the dilution was fixed at 1:20 , 000 . The affinity curves obtained for all groups presented similar kinetic behaviors ( Fig 2A ) . The affinity index determination showed a similar percentage of bound antibodies between the experimental groups with the affinity index ranging from 40% to 49% against most of the venoms tested ( Fig 2B ) . The highest affinity was obtained with the anti-N . melanoleuca plasma ( 60% ) . No statistical significant difference was found between the experimental groups and the standard anti-Crotalic antivenom ( 44% ) . Plasmas samples ( 2nd immunization ) were used for the IgG antibody purification with caprylic acid . The titration of the resulting IgG antivenoms was performed with ELISA , as described previously , with antibody dilutions ranging from 1:2 , 000 to 1:1 , 024 , 000 . The titration curves again displayed similar kinetics , with a more pronounced decline towards greater antibody dilutions ( Fig 3A ) . We observed a small increase in titers after purification in the anti-B . arietans ( 5 . 18 x 106 U-ELISA/mL ) , anti-B . nasicornis + B . rhinoceros ( 3 . 50 x 106 and 3 . 60 x 106 U-ELISA/mL , respectively ) and an small decrease in titers in anti-D . angusticeps + D . polylepis + N . annulifera ( 1 . 87 x 106 , 1 . 67 x 106 and 2 . 60 x 106 U-ELISA/mL , respectively ) groups . The anti-N . mossambica group had a high decrease in titers ( 3 . 07 x 106 U-ELISA/mL ) , while the anti-N . melanoleuca group showed a big increase; it had the highest titers ( 7 . 41 x 106 U-ELISA/mL ) . We found statistically significant differences between all the groups , with the exception of anti-N . melanoleuca , when compared to the purified standard anti-Crotalic antivenom ( 7 . 55 x 106 U-ELISA/mL ) ( Fig 3B ) . The affinity curves displayed similar kinetics both between groups and with pre- and post-purification samples ( Fig 4A ) . The affinity indices decreased compared to the pre-purification samples , ranging between 31% and 37% against most venoms . The anti-N . melanoleuca group showed the highest results ( 45% bound antibodies at KSCN 5 M ) . This group was also the only one to show a statistically significant difference when compared to the standard anti-Crotalic antivenom ( 33% ) ( Fig 4B ) . The venoms ( 10 μg ) were treated with non-reducing SDS-PAGE resolved in 12 . 5% polyacrylamide gel . Samples were either stained with silver ( Fig 5A ) or electroblotted onto nitrocellulose membranes for Western blot ( Fig 5B–5F ) recognition with the purified antibodies ( 2nd immunization ) , diluted 1:20 , 000 . The venom from B . arietans , B . nasicornis and B . rhinoceros showed a similar pattern , with protein bands close to 15 kDa , related to PLA2 , 20–60 kDa , related to SVSPs , and 60–120 kDa , related to SVMPs [26 , 27] . The N . mossambica , N . melanoleuca and N . annulifera venoms were also similar , with protein bands close to 20 , 25 and 60 kDa , related to SVMPs . The N . melanoleuca and N . annulifera venoms also displayed protein bands close to 15 kDa , related to PLA2 , and two high weight proteins bands , 85 kDa ( related to SVMPs ) and 120 kDa ( CVF ) [28] . The venom from D . angusticeps and D . polylepis showed a few protein bands , close to 25 , 40 , 60 and 85 kDa , related to SVMPs [29] . The anti-B . arietans ( Fig 5B ) and anti-B . nasicornis + B . rhinoceros antivenoms ( Fig 5C ) exhibited bands close to 20 , 25 , 60 , 85 and 120 kDa , present with the different Bitis venoms , and bands close to 60 kDa with the other venoms . The anti-B . nasicornis + B . rhinoceros antivenom was also able to resolve bands close to 15 kDa on the Bitis venoms . The anti-N . melanoleuca ( Fig 5D ) and anti-N . mossambica antivenoms ( Fig 5E ) showed bands close to 20–25 , 60 and 85 kDa on most venoms tested , a band close to 10 kDa on the N . melanoleuca and N . mossambica venoms and a band over 120 kDa on the N . melanoleuca and N . annulifera venoms . The anti-D . angusticeps + D . polylepis + N . annulifera antivenom ( Fig 5F ) recognized the same bands as the monospecific Naja antivenoms , plus a band between 10 and 15 kDa on the Dendroaspis venoms ( BPTI-like toxins ) . Protein bands with 25 , 60 and 85 kDa appeared on all the venoms tested and were cross-recognized by all experimental antivenoms . Cross-recognition titration was performed using ELISA ( Fig 6 ) against all venoms , using the purified antibodies ( 2nd immunization ) with dilutions ranging from 1:2 , 000 to 1:1 , 024 , 000 . The anti-B . arietans ( Fig 6A ) and anti-B . nasicornis + B . rhinoceros antivenoms ( Fig 6B ) yielded high titers against their respective venoms ( 4 . 42 x 106 U-ELISA/mL to 7 . 62 x 106 U-ELISA/mL ) and lower titers against the other Bitis venoms ( 1 . 70 x 106 U-ELISA/mL to 2 . 03 x 106 U-ELISA/mL ) . The titration was negligible against the other venoms tested . The anti-N . melanoleuca antivenom ( Fig 6C ) again showed the highest titers against its respective venom ( 7 . 89 x 106 U-ELISA/mL ) , and high titers against the other Naja venoms ( 3 . 14 x 106 U-ELISA/mL to 5 . 45 x 106 U-ELISA/mL ) . Titers against Dendroaspis venoms were low ( < 1 . 60 x 106 U-ELISA/mL ) , and the titration against the Bitis venom was negligible . The anti-N . mossambica antivenom ( Fig 6D ) yielded high titers against all three Naja venoms ( 3 . 35 x 106 U-ELISA/mL to 4 . 08 x 106 U-ELISA/mL ) , and negligible titers against the venoms from other groups . The anti-D . angusticeps + D . polylepis + N . annulifera antivenom ( Fig 6E ) resulted in medium titers against Naja ( 1 . 05 x 106 U-ELISA/mL to 2 . 49 x 106 U-ELISA/mL ) and the Dendroaspis venoms tested ( 1 . 30 x 106 U-ELISA/mL to 1 . 99 x 106 U-ELISA/mL ) , and it showed negligible titers against the Bitis venoms . In vivo protection ( ED50 ) was determined by the injection of 3 LD50 of the venom , incubate with different concentrations of the respective antivenom ( 1:5 , 1:10 , 1:20 , 1:40 ) . The three monospecific antivenoms were effective in neutralizing venom lethality ( Table 1 ) . Anti-B . arietans showed the highest specific activity , with 77 μg of venom neutralized by 1 mg of antivenom , and it was followed by anti-N . mossambica ( 16 μg venom/mg of antivenom ) and anti-N . melanoleuca ( 11 μg of venom/mg of antivenom ) . Anti-B . nasicornis + B . rhinoceros provided protection against B . nasicornis venom ( 54 μg of venom/mg of antivenom ) , however it was not able to protect against B . rhinoceros lethal activity . The polyspecific anti-D . angusticeps + D . polylepis + N . annulifera was not able to neutralize the lethality of any of the three venoms , and in vivo protection was not achieved in this group . The antivenoms produced in this study , specific against medically important African snakes , showed high titers , affinity , ability to cross-recognize venoms in in vitro essays and were capable of in vivo protection . The proposed polyvalent anti-Bitis and anti-Elapidae antivenoms would be effective in the treatment of snake envenoming in Mozambique and could be further developed for continental use . The availability and distribution of these antivenoms to the population in the rural areas would represent an important step in treating snake envenomation in Africa .	Snake envenoming has long been a serious public health problem in underdeveloped and developing countries . In sub-Saharan Africa , the number of bites can be as high as 400 , 000 and result in up to 32 , 000 deaths . As accidents involving snakes are most common among young men in rural areas , complications and amputations can mean the loss of livelihood . Specific antivenoms are the most effective treatment available , but the only African antivenom producer can only supply 25% of the continent’s demand . In this study , we have developed antivenoms against the most important snakes commonly found in Mozambique: Bitis spp . ( puff adders ) , Naja spp . ( cobras ) and Dendroaspis spp . ( mambas ) . The experimental antivenoms were made by immunizing horses with the specific venoms , then collecting and processing their plasma to purify the antibodies . The experimental antivenoms were compared to the commercially available anti-Crotalic ( rattlesnake ) antivenom . The antivenoms produced had high titers , showed affinity for the specific venoms , were able to cross-recognized similar venoms and provide in vivo protection . The data in this study indicates that the antivenoms would be effective in treating B . arietans , B . nasicornis , N . melanoleuca and N . mossambica envenomations . We propose the development of monospecific antibodies as a strategy to increase antivenom quality , and as the basis for the production of two polyspecific antivenoms , anti-Bitis and anti-Elapidae .	[ "Abstract", "Introduction", "Methods", "Results", "Discussion" ]	[]	2016	Development of Equine IgG Antivenoms against Major Snake Groups in Mozambique
Hierarchical processing is pervasive in the brain , but its computational significance for learning under uncertainty is disputed . On the one hand , hierarchical models provide an optimal framework and are becoming increasingly popular to study cognition . On the other hand , non-hierarchical ( flat ) models remain influential and can learn efficiently , even in uncertain and changing environments . Here , we show that previously proposed hallmarks of hierarchical learning , which relied on reports of learned quantities or choices in simple experiments , are insufficient to categorically distinguish hierarchical from flat models . Instead , we present a novel test which leverages a more complex task , whose hierarchical structure allows generalization between different statistics tracked in parallel . We use reports of confidence to quantitatively and qualitatively arbitrate between the two accounts of learning . Our results support the hierarchical learning framework , and demonstrate how confidence can be a useful metric in learning theory . In real-world environments , learning is made difficult by at least two types of uncertainty [1] . First , there is inherent uncertainty in many real-world processes . For instance , the arrival of your daily commute may not be perfectly predictable but subject to occasional delays . Faced with such random fluctuations , learners should integrate as many observations as possible in order to obtain a stable , accurate estimate of the statistics of interest ( e . g . the probability of delay ) [2 , 3] . Second , there is the higher-order uncertainty related to sudden changes in those very statistics ( change points ) . For instance , engineering works may increase the probability of delay for an extended period . When faced with a change point , learners should discount older observations and rely on recent ones instead , in order to flexibly update their estimate [2 , 4 , 5] . Confronted with both forms of uncertainty , the optimal learning strategy is to track not only the statistic of interest but also the higher-order probability of change points . This enables learners to render their estimate stable when the environment is stable ( i . e . between change points ) and flexible when the environment changes [2 , 4 , 6–9] . Importantly , this approach assumes that learners use a hierarchical generative model of their environment . Such a model comprises multiple levels , of which lower levels depend on higher ones: current observations ( level 1 ) are generated according to statistics of observations ( level 2 ) which themselves may undergo change points ( level 3 ) . The hierarchical approach is widely used to study learning in both health [2 , 10] and disease [11 , 12] . However , efficient learning in dynamic environments is also possible without tracking higher-order factors such as the likelihood of individual change points [3 , 13–16] , and a large body of work indeed uses such a solution to model behavioral and brain responses [17–19] . Computationally , this approach is very different as it assumes that learners do not take higher-level factors ( e . g . change points ) into account , and hence use a non-hierarchical or flat model of the world . The possibility that the brain uses internal hierarchical models of the world is an active area of research in cognitive science [20] , and has important consequences for neurobiology , since hierarchical models [21 , 22] and non-hierarchical ones [3 , 18] require different neural architectures . In learning theory however , internal hierarchical models pose somewhat of a conundrum , being simultaneously assumed critical by some frameworks for learning under uncertainty [2 , 4 , 8] and unnecessary by others [3 , 16–18 , 23] . One possible explanation for this conundrum is that the brain might resort to different learning algorithms in different situations . Here , we explore another explanation ( compatible with the former ) : in many situations , flat approximations to hierarchical solutions are so efficient that both accounts become difficult to distinguish in practice . Indeed , previous studies using quantitative model comparison reported conflicting results: some authors found that learning was best explained by hierarchical models [10 , 11 , 24 , 25] while others found that flat models best explained their results [17 , 18 , 26] . Here , our goal is to provide an experimental learning situation in which learners demonstrably use a hierarchical model . We provide a task and a simple analysis that reliably tests whether learners use a hierarchical model of the world . Our test relies not just on comparing model fits , but also on detecting a qualitative signature or hallmark that is uniquely characteristic of an internal hierarchical model . By tracking the higher-order probability of change-points , learners using a hierarchical model can adjust their weighting of prior knowledge and new observations to the changeability of the environment . A highly influential result suggesting that human learners might apply this strategy demonstrated that the apparent learning rate—technically the ratio between the update size and the prediction error at any given observation—was modulated by change points in humans [2 , 5] . This was argued a hallmark of hierarchical processing , since increasing the learning rate after change points would only be expected from a hierarchical learner ( tracking both the statistics of observations and changes in those statistics ) and not from a flat learner ( tracking only the statistics ) . However , we show a counter-example: a flat learning model whose parameters are kept fixed and nevertheless shows systematic modulations of the apparent learning rate without actually tracking the higher-level likelihood of change points ( Fig 1 , Methods and Supplementary Results 2 in S1 File ) . Although the modulations are smaller in the flat model than in the hierarchical one , they are qualitatively identical , demonstrating that such modulations are not uniquely characteristic of hierarchical models . This counter-example suggests that the mere presence of apparent learning rate modulations is not sufficiently specific , and that a new qualitative test must tap into a different property in order to reveal a truly unique hallmark of hierarchical learning . When developing a new test , we first asked what quantity or metric the test should target . In earlier studies on learning under uncertainty , subjects tracked changing statistics such as the probability of reward or of a stimulus [2 , 10 , 17 , 25 , 27 , 28] , or the mean of a physical quantity like the location or magnitude of reward [5 , 7] . Learning was then probed either from choices guided by the learned statistics [2 , 10 , 17 , 25 , 26] or using explicit reports of those statistics [5 , 7 , 8 , 27] . Both choices and explicit reports are first-order metrics , as they only reflect the estimated statistics themselves . However , since a first-order metric only describes the level of observations , and since all models aim at providing a good description of observations , a first-order metric may be seldom unique to a single model . By contrast , second-order metrics , such as the learner’s confidence about her estimates , also describe the learner’s inference and may convey additional information to discriminate models , not contained in the first-order estimate [29] . For illustration , we simulated a hierarchical model and a flat model ( same models as in Fig 1 ) . The latter is minimally different from the classic delta-rule , but extended so as to provide confidence levels ( see Methods and Supplementary Results 1 in S1 File ) . We simulated a typical learning problem , where participants estimate a reward probability that unpredictably changes discontinuously over time . Over a large range of possible task parameters , the probability estimates of the optimal hierarchical model and a near-optimal flat model were highly correlated ( Pearson ρ>0 . 9 ) whereas their confidence levels were much less correlated ( see Fig 2 and Methods ) . The same conclusion holds when simulating a more complex learning task that we present below ( See panels A-B in S1 Fig ) . Note that while we used confidence , results should be similar when using other , related metrics ( see Discussion ) . Importantly , these results only reflect average correlations and do not speak to unique hallmarks or signatures . Rather , they simply show that even when first-order estimates of two models are nearly indistinguishable , their second-order metrics can be much less correlated , thus offering an additional source of information not often considered in learning theory . A hierarchical model is defined by its levels , the variables in each level and the dependencies between those variables . However , these aspects are often confounded in previous experiments because their task structure was quite simple . In a typical experimental situation [2 , 5 , 7 , 10 , 17 , 25 , 27 , 28] the observations received by the subject ( level 1 ) are controlled by only one statistic ( level 2 ) , such as the probability of reward , whose value abruptly changes at change points ( level 3 ) . This task structure is hierarchical , but so simple that it takes the form of a linear chain of dependencies: at each level , a single variable depends on a single variable of the next level ( see panel A in S2 Fig ) . With such a structure , the notions of hierarchy and changeability are confounded . This enables a learner that can cope with changeability , for instance by using leaky ( and perhaps adjustable ) integration [3 , 18] , to learn effectively without needing an internal , hierarchical model of the task structure . By contrast , using a more complex task structure gives rise to forms of inference and learning that are possible only if learners rely on an internal hierarchical model . Here , we build on a previously used task [8 , 24] in which two changing statistics are governed by the same change points . This breaks the pure linear structure of previous tasks by introducing a branch which renders its hierarchical nature more prominent , see panel B in S2 Fig . Crucially , in this situation , only a hierarchical model can leverage the coupling of multiple statistics arising from a common higher-order factor and generalize appropriately between the estimated statistics . Probing this generalization will be our test for hierarchy , as we will detail below . In the task ( see Fig 3A ) , participants observed long sequences of two stimuli ( A and B ) , the occurrence of which was governed by two transition probabilities which subjects had to learn: p ( At\|At-1 ) and p ( Bt\|Bt-1 ) . The value of each probability was independent , but at unpredictable moments ( change points ) both simultaneously changed . Subjects were informed about this generative process in advance . They passively observed the stimuli and were asked to report both their estimate of the transition probability leading to the next stimulus , and their confidence in this estimate . Before testing for an internal hierarchical model , we first wanted to verify whether subjects had performed the task well , in the sense that their responses were consistent with the hierarchical , Bayes-optimal solution . As a benchmark , we used the optimal model for this task; this model was not fitted onto subjects’ data , but set so as to optimally solve the task by ‘inverting’ its hierarchical generative structure using Bayes’ rule ( see methods ) . As displayed in Fig 3B and 3C , linear regressions show an agreement between participants’ probability estimates and optimal probability estimates ( defined as the mean of the posterior; regressions are computed at the subject-level: β = 0 . 66±0 . 06 s . e . m . , t22 = 11 . 13 , p = 1 . 7 10−10 ) , and subjective confidence reports and optimal confidence ( defined as the log precision of the posterior; β = 0 . 10±0 . 03 s . e . m . , t22 = 3 . 06 , p = 0 . 0058 ) ; for further checks of robustness , see Supplementary Results 3 in S1 File . In order to quantify the variance explained , we also computed the Pearson correlations corresponding to those regressions: ρ = 0 . 56±0 . 04 s . e . m . for probability estimates and ρ = 0 . 19±0 . 05 s . e . m . for confidence . Despite being somewhat noisier , confidence reports also showed many properties of optimal inference ( see Supplementary Results 4 in S1 File ) . Since we propose that subjects’ confidence reports can convey useful information over and above their first-order estimates , the next thing we verified was that confidence reports indeed conveyed information that was not already conveyed implicitly by the first-order estimates . We tested this in our data by regressing out the ( theoretically expected ) covariance between subjects’ confidence reports and several metrics derived from first-order estimates ( see Supplementary Results 4 in S1 File ) ; the residuals of this regression still co-varied with optimal confidence ( β = 0 . 028±0 . 012 , t22 = 2 . 3 , p = 0 . 029 ) . This result was replicated by repeating the analysis on another dataset [8]: β = 0 . 023±0 . 010 , t17 = 2 . 2 , p = 0 . 0436; and also in the control experiment detailed below: β = 0 . 015±0 . 006 , t20 = 2 . 3 , p = 0 . 034 . These results indicate that subjective confidence and probability reports are not redundant , and thus that confidence is worth investigating . Having verified that confidence and probability reports closely followed estimates of an optimal hierarchical model , and that both metrics were not redundant , we then tested whether subjects’ reports , overall , could not be better explained by a different , computationally less sophisticated model: the flat model introduced above ( Fig 1 , also see Methods and Supplementary Results 1 in S1 File ) , that approximates the full Bayesian model extremely well . Both models have the same number of free parameters , so model comparison ( at least using standard comparison methods like BIC or AIC ) boils down to comparing the goodness-of-fit . We first took the parameters that provide the best estimate of the true generative probabilities . The goodness-of-fit , assessed as mean square error ( MSE ) between subjects’ and models’ estimates , was better for the hierarchical model than for the flat model ( paired difference of MSE , hierarchical minus flat: -0 . 0051±0 . 0014 s . e . m . , t22 = -3 . 7 , p = 0 . 0013 ) . Note that subjects’ estimates of volatility , a key task parameter here , usually deviate from the optimum and show a large variability [5 , 30] , which could bias our conclusion . We therefore fitted the model parameters per subject , and we found that the difference in fit was even more significant ( -0 . 0077±0 . 0019 s . e . m . , t22 = -3 . 97 , p = 6 . 5 10−4 ) . This result replicates a previous finding [24] . We then repeated the comparison for confidence levels . When model parameters were set to best estimate the true transition probabilities , the hierarchical model showed a trend toward a significantly lower MSE compared to the flat model ( paired difference of MSE , hierarchical minus flat: -0 . 0017±0 . 0010 s . e . m . , t22 = -1 . 8 , p = 0 . 084 ) . When model parameters were fitted onto each subjects’ confidence reports , this difference was significant ( -0 . 0027±0 . 0012 s . e . m . , t22 = -2 . 36 , p = 0 . 028 ) . In sum , these results show that participants successfully performed the task and that the hierarchical model was quantitatively superior to the flat model in explaining subjects’ probability estimates and confidence ratings . This leaves us with the last and perhaps most important question: did subjects also show a qualitative signature that could only be explained by a hierarchical model ? Identifying the qualitative signature proposed here was possible because our task involves two transition probabilities , P ( A\|A ) and P ( B\|B ) , whose changes were coupled , occurring simultaneously . In this context , a flat learner only estimates the value of each transition probability , while a hierarchical model also estimates the probability of a global change point . Faced with a global change point , the hierarchical learner then reacts optimally and makes its prior knowledge more malleable by becoming uncertain about both P ( A\|A ) and P ( B\|B ) . Importantly , using this mechanism , an internal hierarchical model should allow for generalization: if a change point is suspected after observing just one type of transition ( e . g . AAAAAAA , when P ( A\|A ) was estimated to be low ) a hierarchical learner would also become uncertain about the other quantity , P ( B\|B ) , despite having acquired no direct evidence on this transition ( Fig 4A ) . Critically , this form of indirect inference is unique to hierarchical models and thus offers a powerful test of hierarchical theories of learning . To test for this generalization effect , we focused on streaks of repetitions , and distinguished between streaks that seem unlikely in context and may signal a change point ( suspicious streaks ) and streaks that do not ( non-suspicious streaks ) . Stimulus sequences were carefully selected to contain enough suspicious and non-suspicious streaks and to control for confounds such as streak duration ( see Methods ) . Questions were inserted just before and after the streak , so that subjects reported their estimate of ( and confidence in ) the other , non-repeating transition ( Fig 4A ) . Exact theoretical predictions for both models are found in Fig 4B . In the hierarchical model , confidence decreases strongly after suspicious , but much less after non-suspicious streaks . In the flat model , however , there is no such difference . Strikingly , subjective reports followed the hierarchical account ( see Fig 4C ) : confidence decreased strongly after suspicious ( -0 . 12±0 . 04 s . e . m , t22 = -3 . 2 , p = 0 . 004 ) but not after non-suspicious streaks ( -0 . 02±0 . 03 s . e . m . , t22 = -0 . 7 , p = 0 . 51 ) , and this interaction was significant ( paired difference , 0 . 10±0 . 03 s . e . m . , t22 = 3 . 7 , p = 0 . 001 ) . We now rule out a series of potential confounding explanations . First , one concern is that the analysis above uses models optimized to estimate the true transition probabilities of the task . However , our conclusions remain unaffected if we use models fitted onto each subject ( panels C , E in S3 Fig ) . Another concern is that our analyses assume subjects were tracking transition probabilities , while they may in fact have been tracking another ( heuristic ) quantity , perhaps using a flat model . Detailed analysis revealed that subjects did in fact track transition probabilities ( see Supplementary Results 4 in S1 File ) and that no heuristic flat model could explain the selective decrease of confidence ( panels B , D , F in S3 Fig ) . Finally , we also considered models that were technically hierarchical but that erroneously assume that the two transition probabilities have independent ( rather than identical ) change points ( see panel C in S2 Fig ) . These models did not show the critical effect of streak type ( panels A , C , E in S3 Fig ) , indicating that our test is diagnostic of the ability to transfer knowledge between dependent variables . This transfer is not afforded by all hierarchical models , but only those which entertain the correct , relevant dependencies . One may also wonder whether the effect reported in Fig 4 for confidence is also found in another variable . S4 Fig shows that probability estimates ( the ones about which confidence is reported and shown in Fig 4 ) are not affected by streak types neither in subjects ( paired difference between streak types , -0 . 01±- 0 . 02 s . e . m . , t22 = -0 . 5 , p = 0 . 59 ) nor in the hierarchical model ( -0 . 01±0 . 01 s . e . m . , t22 = -1 . 4 , p = 0 . 17 ) . A more subtle effect is that , when a change point is suspected , generalization should reset the estimate of the unobserved transition probability , which should thus get closer to the prior value 0 . 5 . However , this effect is less straightforward , because the estimated transition probability may already be close to 0 . 5 before the streak . Indeed , even in the hierarchical , Bayes-optimal model , streak type had only a weak effect ( paired difference , 0 . 02±0 . 01 sem , t22 = 2 . 9 , p = 0 . 008 ) , compared to the effect on confidence ( Fig 4 , t22 = 11 . 7 , p = 6 . 9 10−11 ) . The expectedly weaker effect of streak type on the distance to the prior was not detected in participants ( -0 . 0036 +/- 0 . 01 s . e . m . , t22 = -0 . 3 , p = 0 . 76 ) . We also tested reaction times since they often co-vary with confidence . Here , when the optimal confidence was lower , subjects took longer to respond to the prediction question ( slope of reaction times vs . optimal confidence: -0 . 57±0 . 19 s . e . m . , t22 = -3 . 07 , p = 0 . 005 ) , but not to the confidence question ( slope: 0 . 04±0 . 08 s . e . m . , t22 = 0 . 48 , p = 0 . 64 ) . However , this significant correlation is reducible to first-order estimates . We repeated the analysis previously reported for the subjects’ confidence but now with their reaction times: after regressing out several metrics derived from first-order estimates , the residuals of this regression no longer co-varied with optimal confidence ( β = -0 . 103±0 . 124 s . e . m . , t22 = -0 . 8 , p = 0 . 41 ) ; and standardized regression coefficients indeed differed between the two regressions ( paired difference of standardized βs: -0 . 233±0 . 080 s . e . m . , t22 = -2 . 9 , p = 0 . 008 ) . In addition , there was no effect of streak type on ( raw ) reaction times both for the probability estimate and reports ( paired difference between streak types , both p>0 . 27 ) . A final alternative explanation for the effect shown in Fig 4 is that suspicious streaks were more surprising and that subjects may become generally uncertain after surprising events . In this case , the effect would not reflect hierarchical inference but simply general surprise . We therefore performed a control experiment , in a different group of subjects , in which both probabilities changed independently ( Panel C in S2 Fig ) : here , suspicious streaks were equally surprising but no longer signaled a global change point ( Fig 5A ) . Indeed , generalization of a decrease in confidence was no longer observed for the hierarchical model with the correct task structure or in subjects ( paired difference between suspicious and non-suspicious streaks: 0 . 03±0 . 02 s . e . m . , t20 = 1 . 5 , p = 0 . 15 ) , see Fig 5B . This absence of an effect in the control task is significantly different from the effect found in the main task ( difference of paired differences , two-sample t-test , t42 = -2 . 03 , p = 0 . 048 ) . This difference is not due poor performance in the control experiment ( see Fig 5C ) : linear regression between the optimal hierarchical model for uncoupled change ( the optimal model for this task ) and subjects showed a tight agreement for both predictions ( β = 0 . 61±0 . 06 s . e . m . , t20 = 10 . 27 , p = 2 10−9 ) and confidence ( β = 0 . 08±0 . 01 s . e . m . , t20 = 6 . 83 , p = 1 . 2 10−6 ) , as in the main task ( see Fig 3B and 3C ) . The difference between the two tasks suggests an effect of higher-level factors ( coupled vs . uncoupled change points ) and thus further support the hierarchical model . Choices and first-order reports are often used in behavioral science , but other metrics like a subject’s confidence and reaction times also proved useful to study cognition , see [31] . A feature that distinguishes our task from previous work on adaptive learning is the use of explicit confidence ratings to dissociate between flat and hierarchical models . Since the flat models considered here are known to provide very accurate first-order approximations to hierarchical ( optimal ) models [3 , 15 , 32] , we reasoned that second-order estimates might prove useful as an additional source of information . Our simulations showed that even when first-order metrics are nearly indistinguishable , confidence was much less correlated between models . Although this conclusion holds both in a standard task in which one non-stationary statistic is learned ( Fig 2 ) and in our more complex task ( S1 Fig ) , we acknowledge that this does not guarantee that confidence is generally more diagnostic than first-order metrics . Rather , we take our experiment as an example case showing that second-order metrics can be worth studying in learning theory . Importantly , while we believe confidence can be useful to discriminate hierarchical and flat models , we do not want to claim confidence is necessary ( i . e . that it is impossible with other metrics ) . In particular , both the low correlations ( Fig 2 ) and the critical dissociation between models ( Fig 4 ) should in principle also be detectable using other , related metrics . Reaction times are one obvious candidate , as they are often an implicit measure of a subject’s confidence [33–36] . In our study , we found a correlation between reaction times and confidence , however , unlike for confidence reports , this correlation was reducible to first-order aspects such as the predictability of the next stimulus , and we found no effect of streak type that serves as our hallmark . Note that in our study , in contrast to accumulation [36] or waiting-time [35] paradigms , there is no general , principled reason for reaction times to co-vary with confidence . On the contrary , motor effects may even artifactually corrupt the relationship between confidence and reaction times since in our task reporting a more extreme estimate ( typically associated with higher confidence ) requires to move the response cursor further . This may explain why reaction times showed only a partial correlation with confidence and eventually did not show the hallmark of hierarchical inference . Another interesting metric is the apparent learning rate . Previous studies reported modulations of the apparent learning rate by change points [2 , 5] . The optimal , hierarchical model indeed shows such modulations because its updates are confidence-weighted [4 , 24]: for a given prediction error , its updates are larger when confidence about prior estimates is lower , which is typically the case when a change point is suspected . However , we found that in simple experiments that require to monitor only the frequency of a stimulus or a reward , a flat model could exhibit similar modulations , which are therefore not diagnostic of hierarchical inference . In more complex experiments like the one here , the apparent learning rate could nevertheless show our signature of hierarchical inference . Theoretical analysis supports this hypothesis ( see Supplementary Results 5 in S1 File ) but we cannot assess it in our data , since this requires a trial-by-trial measure of the apparent learning rate , and thus trial-by-trial ( not occasional ) reports of first-order estimates . A trial-by-trial , model-free measure of the apparent learning rate is neither accessible if subjects make choices at each trial . In such studies [2 , 37] , the authors could only use choices to compute an apparent learning rate in a sliding window of trials but this analysis lacks trial-by-trial resolution . In our task , investigating the apparent learning rate would require subjects to report their probability estimates after each trial , and hence to constantly interrupt the stimulus stream . This would probably interfere with the participants’ ability to integrate consecutive observations , which is critical for tracking transition probabilities , and therefore seems difficult to implement in practice . Furthermore , if an effect of streak type were observed on the apparent learning rate , it would probably be mediated by the subject’s confidence [4 , 24] , in that case one may prefer to probe confidence directly . We acknowledge that there are drawbacks of using confidence as the metric of interest . Although our simulations suggested that in theory confidence might discriminate flat and hierarchical models more reliably , in practice we found that the model fit between participants and the hierarchical , Bayes-optimal model was more precise for probability estimates than for confidence ratings ( see Figs 3B , 3C and 5C ) . This noisy character of confidence measurements was also reported previously [8 , 38–40] and may hinder the use of confidence as a metric to discriminate between models . This problem may be even worse when using an indirect indicator of confidence , such as reaction times . We also acknowledge the possibility that asking subjects for explicit confidence reports about their inferences may have promoted the use of a sophisticated learning algorithm of a hierarchical nature . By contrast , simpler tasks in which subjects only make choices without being asked to reflect upon the reliability of their estimates , as we do here with confidence reports , could favor the use of simpler , possibly non-hierarchical learning algorithms . We cannot explore further this cautionary note in our current design because confidence reports are precisely key to detect the use of a hierarchical model here . Quantitative model comparison is a widely used method for contrasting competing models . Following this approach , multiple models are fit onto the subjects’ data , and the model that achieves the best fit with respect to its complexity is deemed most likely [41 , 42] . This approach is attractive because it is generally applicable and it provides a common metric ( e . g . goodness-of-fit , Bayes-Factor , exceedance probability ) to compare different models . However , a limitation of quantitative model comparison is that it is not always clear what underlying factors are contributing to differences in fitness , and whether these factors are indeed most relevant to the question at hand . Moreover , quantitative model comparison only allows for relative conclusions , such as one model being better than other tested models , and is thus inherently restricted in scope to the models that are tested . Here , we used two models . The choice for the hierarchical model was straightforward: it is the optimal solution and therefore served as a benchmark . The choice for the flat model was motivated by both its resemblance to the classic , widely used delta rule with fixed learning rate [15 , 19] , also see Supplementary Results 1 in S1 File; and because it is known to approximates the hierarchical model extremely well [3] . This renders it a worthy competitor , although not representative of all flat models . A complementary approach is qualitative model comparison: analyzing specific , critical trials for the presence of qualitative signatures or hallmarks that uniquely identify or exclude one type of model . This approach is not only more transparent , but also enables more general conclusions , such as the falsification of one type of model [43] . Here , generality was granted by the fact that our test taps into a unique property of hierarchical models with an appropriate representation of the task structure . This argument derives from a priori principles , but we verified it by trying several parametrizations , priors and hypotheses of our models ( S2 Fig ) , which supported that a model entertaining the correct dependencies can appropriately generalize . Importantly , we do not claim that the full Bayesian solution is the only model with this property . Other hierarchical models can be envisaged , such as two leaky accumulators ( one per transition probability ) coupled by a third mechanism that globally increases their leak whenever a change point is suspected in either accumulator . Note that here , the quantitative and qualitative approaches were indeed complementary: quantitative model comparison provided evidence in favor of a hierarchical account , and the qualitative approach tested for unique hallmarks and thereby falsified a strictly flat account . Last , we acknowledge that we falsify the flat account only in a strict sense: we rule out that humans only make use of a flat algorithm in our task . Our results cannot rule out the existence of flat learning algorithms in general . Several learning algorithms may co-exist and the brain may switch between strategies depending on context . Instead , the results show that humans are capable of hierarchical learning and that they use it even in a task that does not critically requires it , as shown by the near-optimal performance of the flat model . The human capacity of hierarchical learning is compatible with what others have suggested [4 , 6 , 10 , 21] and we provide a task for studying it . Our results falsify a strictly flat account of learning under uncertainty , therefore they also falsify the use of solely leaky integration by neural networks to solve adaptive learning tasks . Leaky integration is often deemed both biologically plausible and computationally efficient [3 , 18 , 19 , 37] . A sophisticated version of the leaky integration with metaplastic synapses allows partial modulation of the apparent learning of the network , without tracking change points or volatility [18] . Others have suggested that computational noise itself could enable a flat inference to automatically adapt to volatility [16 , 44] . Those approximate solutions dismiss the need to compute higher-level factors like volatility , they are thus appealing due to their simplicity; however , we believe that such solutions cannot explain the generalization afforded by hierarchical inference that we showed here . As such , at least for explicitly hierarchical tasks like the one studied here , such models have to be complemented to include higher-level factors . One previously proposed bio-inspired model seems compatible with our result [6] . This model comprises two modules: one for learning and the other for detecting change points , or “unexpected surprise” [1] . When a change point is detected , a reset signal is sent to the learning module . Converging evidence indicates that noradrenaline could play such a role [45–48] . A global reset signal could promote learning for the two transition probabilities that are maintained in parallel in our task , thereby allowing the reset of both when only one arouses the suspicion of a change point . Such a hypothesis nevertheless needs to be refined in order to account for the fact that the two statistics can also be reset independently from one another , as in the control task . Our task structure is more complex than many previous experiments which required to monitor only one generative statistic [2 , 5 , 7 , 10 , 27 , 49] . This may hamper translating our results to other experiments , but it has a certain ecological appeal since in real-life situations , multiple regularities are often embedded in a single context . We believe that more complex tasks are well-suited to distinguish complex computations and approximations thereof . Both are likely to be equivalent in simpler situations , whereas in highly structured environments with multiple interdependent levels [20 , 50] , an effective learning algorithm can hardly obliviate the hierarchical nature of the problem to solve . Note that while we believe that hierarchically structured tasks are theoretically and practically better suited to test for hierarchical information processing , we do not claim that such tests are impossible in simpler tasks . An interesting and difficult problem that we leave unaddressed here is how subjects may discover the task structure [20 , 51 , 52] . In our task , the optimal hierarchical model is able to correctly identify the current task structure ( coupled vs . uncoupled change points ) , but only with moderate certainty even after observing the entire experiment presented to one subject ( log-likelihood ratios range from 2 to 5 depending on subjects ) . Therefore , in principle , subjects who are not endowed with optimal computing power cannot identify reliably the correct structure from observations alone . We speculate that in real-life situations , some cues or priors inform subjects about the relevant dependencies in their environment; if true , then our experiment in which subjects were instructed about the correct task structure may have some ecological validity . Interestingly , while the importance of hierarchical inference remains controversial in the learning literature [4 , 5 , 7 , 10 , 13 , 14 , 16–18 , 26 , 27] , it seems more clearly established in the domain of decision making and action planning [50 , 53–57] . For instance , it was suggested that the functional organization of cognitive control is nested: low level cues trigger particular actions , depending on a stimulus-response association which is itself selected depending on a particular context [58] . In this view , negative outcomes may indicate that the ( higher-level ) context has changed and thus that a new rule now applies . This inference even seems to be confidence-weighted in humans: the suspicion of a change in context is all the stronger that subjects were confident that their action should have yielded a positive outcome under the previous context [59] . Those two studies feature an important aspect of hierarchy: a succession of ( higher-level ) task contexts separated by change points governs the ( lower-level ) stimuli . Our task also leverages another feature of hierarchy: it allows generalization and transfer of knowledge . A rule learned in a particular context can be applied in other contexts , for instance see [60 , 61] . Our results go beyond mere transfer: they show that the brain can update a statistic in the absence of direct evidence thanks to higher-level dependencies . In sum , we showed that previously proposed hallmarks are insufficient to distinguish hierarchical from non-hierarchical models of learning under uncertainty , and provided a new way to test between the two . Our results provide support to the hierarchical framework . Moreover , our work demonstrates the importance of using more complex task structures to test for hierarchy explicitly , and the usefulness of confidence as a source of information in learning theory . We believe our test can be applied beyond human learning to animal and computational models like neural networks , for which it may not be clear whether they make inferences that are hierarchical or not . As such , our test will be of interest to experimentalists and theoreticians alike . Participants were recruited by public advertisement . They gave a written informed consent prior to participating and received 20 euros for volunteering in the experiment . The study was approved by the local Ethics Committee ( CPP n°08–021 Ile de France VII ) . 26 participants ( 17 females , mean age 23 . 8 , s . e . m . : 0 . 49 ) performed the main task and 21 other participants performed the control task ( 11 females , mean age 23 . 0 , s . e . m . : 0 . 59 ) . We excluded participants who showed poor learning performance , which we quantified as the Pearson ρ coefficient between their probability estimates and the hierarchical , Bayes-optimal estimates . We used a threshold corresponding to 5% of the ( lowest ) values measured in this task ( ρ<0 . 18 , from a total of 105 participants in this study and others ) This excluded 3 subjects from the main task , and none from the control task . Including those subjects does not change our main conclusion: regression of subject vs . optimal probability ( resp . confidence ) : p = 3 . 3 10–9 , ( resp . p = 0 . 007 ) ; quantitative model comparison , with values fitted onto subject’s probability estimates ( resp . confidence reports ) supports the hierarchical model: p = 0 . 0003 ( res . p = 0 . 050 ) ; effect of streak type of pre-post change in confidence: p = 0 . 047 . The task was run using Octave ( Version 3 . 4 . 2 ) and PsychToolBox ( Version 3 . 0 . 11 ) . Each participant completed a total of 5 blocks: 1 training block and 4 experimental blocks ( 2 auditory , 2 visual ) . Auditory and visual blocks alternated , with the modality of the first block randomised across participants . In each block , we presented binary sequences of 380 stimuli ( 1520 total ) denoted A and B , which were either visual symbols or sounds and were perceived without ambiguity . Sequences were generated according to the same principles as in previous studies [8 , 24] . A and B were randomly drawn based on two hidden transition probabilities which subjects had to learn . These probabilities were stable only for a limited time . The length of stable periods was randomly sampled from a geometric distribution with average length of 75 stimuli , truncated at 300 stimuli to avoid overly long stable periods . Critically , and contrary to other studies [2] the volatility was thus fixed ( at 1/75 ) . Transition probabilities were sampled independently and uniformly between 0 . 1–0 . 9 , with the constraint that , for at least one of the two probabilities , the change in odds ratio ( p/1-p ) between consecutive stable periods was at least fourfold , thus guaranteeing that the change was effective . Across sequences and subjects , the actually used generative values indeed covered the transition probability matrix 0 . 1–0 . 9 uniformly , without any correlation ( Pearson ρ = −0 . 0009 , p = 0 . 98 ) . Occasionally , the sequence was interrupted and subjects had to estimate the probability that the next stimulus would be either an A or a B and report their confidence in that estimate . Questions were located quasi-randomly , semi-periodically once each 15 stimuli on average ( 100 in total ) . Of the 100 questions , 68 questions were randomly placed; the remaining 32 questions were intentionally located just before and after 16 selected streaks ( 8 suspicious , 8 non-suspicious ) and functioned as pre/post-questions to evaluate the effect of these streaks ( see Fig 4 ) . For details on the definition and selection of suspicious/non-suspicious streaks , see below . To familiarize participants with the task they were carefully instructed and performed one training block of 380 stimuli ( or ~12 minutes ) . To make sure they were fully aware of the volatile nature of the generative process , participants had to report when they detected changes in the hidden regularities . In the experimental blocks , reporting change points was omitted , but participants knew the underlying generative process was the same . The control task was very similar to the main one , with only two differences . ( 1 ) When a change occurred , it impacted only one of the two transition probabilities ( randomly chosen ) . ( 2 ) During the training block , when subjects were required to report when they detected change points , they also reported which of the two transition probabilities had changed . Each randomly generated sequence was evaluated computationally and carefully selected to ensure that each subject encountered enough target moments during which the models make qualitatively different predictions , and that all sequences were balanced in terms of potential confounds such as streak duration and location . To this end , 4 random sequences of 380 stimuli long ( each corresponding to one block ) were analyzed computationally with the hierarchical and flat learning models , yielding 4 simulated ‘blocks’ . The sequences , and associated trial-by-trial transition probability estimates from both models , were concatenated to form a single experimental sequence ( of 1520 stimuli ) . This experimental sequence was then submitted to several selection criteria . First , we assessed whether the sequence contained at least 8 suspicious and 8 non-suspicious ‘streaks’ . Consecutive repetitions were defined as ‘streaks’ if they consisted of at least 7 or more stimuli , and started after the 15th stimulus of a block . Streaks were classified as ‘suspicious’ if they aroused the suspicion of a change in the hierarchical , Bayes optimal model . Computationally , this was defined via the confidence in the probability of the observed repetition decreasing on average during the streak . To ensure the effect would be observable , only sequences in which the suspicious streaks led to a sizeable decrease in theoretical confidence levels were selected . To control for factors that may potentially confound decreases in confidence , only sequences in which the average duration of suspicious and non-suspicious streaks was approximately identical , and in which there was at least one streak of each type in each block , were selected . In addition , subjects were not informed about the distinction between suspicious and non-suspicious streaks or that between random questions and pre-post questions that targeted the critical moments before and after streaks . Interviews performed after the experiment ruled out that subjects understood the goal of the experiment , as no subject had noticed that a sizable fraction ( ~30% ) of questions purposefully targeted streaks . The models used in this study are implemented in a Matlab toolbox available on GitHub and described in a previous publication [32] . The model termed “hierarchical” and “flat” here correspond respectively to the hidden Markov model ( HMM ) and the leaky integrator model in the toolbox . Here , we summarize the essential aspects of those models . The hierarchical and flat models ( M ) are both ideal observer models , in the sense that they both use Bayes rule to infer what the posterior distribution of the statistic they estimate , θt , should be , under a given set of assumptions . However , only for the hierarchical model , the set of assumptions actually corresponds to the generative process , rendering its estimates statistically optimal for the task . Nevertheless , both use Bayes rule to estimate the posterior distribution of θt , based on a prior on this statistic and the likelihood provided by previous observations , y1:t ( here , a sequence of As and Bs ) : p ( θt\|y1:t , M ) ∝p ( y1:t\|θt , M ) p ( θt , M ) ( Eq 1 ) Subscripts denote the observation number within a sequence . In the main text , the models estimate the transition probabilities between successive stimuli , so that θ is a vector with two elements: θ = [p ( A\|A ) , p ( B\|B ) ] . Note that those two probabilities suffice to describe all transitions , since the others can be derived as p ( B\|A ) = 1-p ( A\|A ) and p ( A\|B ) = 1-p ( B\|B ) . In S3 Fig , we also consider variants in which the model estimates another statistic , the frequency of stimuli: θ = p ( A ) . Note that p ( B ) is simply 1-p ( A ) . The estimation of θ depends on the assumption of the ideal observer model ( M ) . The flat model considers that θ is fixed , and evaluates its value based on a leaky count of observations . The internal representation of this model therefore has only one level: θ , the statistic of observations . When the true generative statistic is in fact changing over time , the leakiness of the model enables it to constantly adapt its estimate of the statistic and therefore to cope with changes . If the leakiness is tuned to the rate of change , the estimate can approach optimality ( see panel A in S1 Fig ) . By contrast , the hierarchical model entertains the assumption that θ can abruptly change at any moment . The internal representation of the model therefore has several levels beyond observations: a level characterizing the statistic of observations at a given moment ( θt ) and a level describing the probability that of a change in θ occurs ( pc ) . Conceivably , there could be higher-order levels describing changes in pc itself [2]; however this sophistication is unnecessary here and we consider that pc is fixed . The flat and the hierarchical models have one free parameter each , respectively ω ( the leakiness ) and pc ( the prior probability of change point ) . Unless stated otherwise , the analysis reported in the main text used the parameters that best fit the true probabilities used to generate the sequences of observations presented to subjects . More precisely , for each sequence of observations , we computed the probability of each new observation given the previous ones , as estimated by the models using Eq 5 and we compared it to the true generative probability . We adjusted the free parameters ω and pc with grid-search to minimize the sum of squared errors ( SSE ) over all the sequences used for all subjects . The resulting values are ω = 20 . 3 and pc = 0 . 014 ( indeed close to the generative value 1/75 ) . We also fitted the parameters to the responses of each subject ( S3 Fig ) . For probability estimates , the above grid-search procedure was repeated after replacing generative values with the subject’s estimates of probabilities at the moment of questions . For confidence reports , we used a similar procedure; note however that subjects used a bounded qualitative slider to report confidence whereas the model confidence is numeric and unbounded , so that there is not a direct mapping between the two . Therefore , the SSE was computed with the residuals of a linear regression between subject’s confidence and the model’s confidence . Here are the fitted values we obtained , reported as median ( 25–75 percentile ) : for the hierarchical model pc = 0 . 0360 ( 0 . 0043–0 . 3728 ) when fitted onto probability estimates and pc = 0 . 0134 ( 0 . 0006–0 . 0526 ) when fitted onto confidence reports: for the flat model , ω = 34 . 4 ( 25 . 9–100 . 5 ) and ω = 20 . 8 ( 8 . 2–50 . 2 ) , respectively . All linear regressions between dependent variables ( e . g . probability estimates , confidence ratings ) and explanatory variables ( optimal estimates of probabilities and confidence , surprise , prediction error , entropy ) included a constant and were estimated at the subject level . The significance of regression coefficients was estimated at the group level with t-tests . For multiple regressions , explanatory variables were z-scored so that regression coefficients can be compared between the variables of a given regression . Unless stated otherwise , all t-tests are two-tailed .	Learning and predicting in every-day life is made difficult by the fact that our world is both uncertain ( e . g . will it rain tonight ? ) and changing ( e . g . climate change shakes up weather ) . When a change occurs , what has been learned must be revised: learning should therefore be flexible . One possibility that ensures flexibility is to constantly forget about the remote past and to rely on recent observations . This solution is computationally cheap but effective , and is at the core of many popular learning algorithms . Another possibility is to monitor the occurrence of changes themselves , and revise what has been learned accordingly . This solution requires a hierarchical representation , in which some factors like changes modify other aspects of learning . This solution is computational more complicated but it allows more sophisticated inferences . Here , we provide a direct way to test experimentally whether or not learners use a hierarchical learning strategy . Our results show that humans revise their beliefs and the confidence they hold in their beliefs in a way that is only compatible with hierarchical inference . Our results contribute to the characterization of the putative algorithms our brain may use to learn , and the neural network models that may implement these algorithms .	[ "Abstract", "Introduction", "Results", "Discussion", "Materials", "and", "methods" ]	[]	2019	Confidence resets reveal hierarchical adaptive learning in humans
High throughput measurement of gene expression at single-cell resolution , combined with systematic perturbation of environmental or cellular variables , provides information that can be used to generate novel insight into the properties of gene regulatory networks by linking cellular responses to external parameters . In dynamical systems theory , this information is the subject of bifurcation analysis , which establishes how system-level behaviour changes as a function of parameter values within a given deterministic mathematical model . Since cellular networks are inherently noisy , we generalize the traditional bifurcation diagram of deterministic systems theory to stochastic dynamical systems . We demonstrate how statistical methods for density estimation , in particular , mixture density and conditional mixture density estimators , can be employed to establish empirical bifurcation diagrams describing the bistable genetic switch network controlling galactose utilization in yeast Saccharomyces cerevisiae . These approaches allow us to make novel qualitative and quantitative observations about the switching behavior of the galactose network , and provide a framework that might be useful to extract information needed for the development of quantitative network models . One of the primary goals of systems biology is to uncover the dynamics of cellular networks . Sometimes , this has meant collecting time-series data and applying tools for time-series analysis such as Fourier methods to identify periodically expressed genes [1]–[3] or temporal clustering to identify different dynamic “modes” [4]–[6] . In other cases , it has meant the construction of explicit state-based dynamical models , based either on qualitative expectations of system behavior [7]–[10] or based more directly on quantitative experimental data [11] , [12] . Another common goal has been to characterize the steady-state behavior of the network , which is of particular interest if the system exhibits multistability [13]–[15] . In these cases , the steady-states , along with their basins of attraction , have been likened to distinct cell types [16]–[18] , and thus define the repertoire of “behaviors” available to the cell . Mathematically , the analysis of steady states falls into the domain of bifurcation theory , which addresses the existence , number and stability of fixed points or limit cycles/attractors of dynamical systems and how these change as a function of system parameters or inputs [19] . Usually , this analysis is performed on deterministic mathematical models such as differential equations or difference equations . Here , we are concerned with the experimental and computational quantification of bifurcation-like behavior in stochastic genetic switches . There is considerable evidence that signalling networks in a population of genetically-identical cells exhibit large cell-to-cell variability in their output , despite operating in a homogeneous external environment ( see e . g . , [20] , [21] ) . In some cases , inherent fluctuations in the internal state of the cells leads to distinguishable subpopulations , even when cells are genetically identical and experience a homogenous environment . For example , a ubiquitous network motif is the bistable genetic switch , with output variability distributed about high and low states dependent upon the level of an external input signal [13] , [22]–[25] . Accurate estimates of bifurcation structure from noisy experimental can provide important qualitative , and in some cases quantitative , information about system behavior , guide model development and parameter estimation efforts , or help to discriminate among competing hypotheses regarding network architectures . For example , recent work has demonstated that the statistics of the fluctuations about the steady-states provides significant constraints on kinetic parameter estimation [26] . Two ingredients are necessary for empirical analysis of the bifurcation behavior of a cellular network . One is single-cell measurements of one or more cellular variables , such as gene expression . Technologies such as microarrays , SAGE or quantitative mass spectrometry , which operate on collections of cells or whole tissues , obscure potential heterogeneity in the sample . They do not discriminate , for example , between a 100% increase in expression of a gene , and a 200% increase in its expression in 50% of the cells . With technologies such as fluorescent cell imaging and flow cytometry , however , the state of each cell can be ascertained . As a result , one can determine whether the cell population is homogenous or if it comprises a set of subpopulations—each undergoing different dynamical behaviors corresponding to different growth strategies , differentiation endpoints , etc . The other necessary ingredient is a method for experimental manipulation of some system parameter ( s ) or environmental condition ( s ) , in order to study how subpopulations change under varying conditions . This may mean changing the concentration of ligands or nutrients in the cellular environment or artificially manipulating the activity of regulatory factors inside individual cells . For example , Ozbudak et al . [13] recently used single-cell fluorescence microscopy to establish an empirical map of the two-dimensional bifurcation diagram for the lactose utilization network in Eschericia coli as a function of the systematic variation of two environmental parameters . Moreover , targeted disruption of feedback loops within the galactose utilization network of Saccharomyces cerevisiae has provided key insights into the control of cell-cell variability in gene expression and mechanisms underlying the stochastic switching between distinct epigenetic expression states [25] , [27] . Increased use of these techniques demands the establishment of methods for analyzing the generated data in a statistically robust and computationally efficient manner . The organization of this paper is as follows . First , we discuss traditional bifurcation analysis in greater detail , introducing in particular saddle-node bifurcations , a type of bifurcation widely associated with the dynamics of gene regulatory switches . We then describe the necessity of generalizing the notion of bifurcation behavior to account for the inherent noise ( stochasticity ) in cellular networks . Next , we present the data that motivated our study—single-cell flow cytometry data measuring activity in the yeast galactose utilization network over a range of extracellular galactose concentrations . We then report on two broad approaches to analyzing this data and extracting estimates of bifurcation structure , namely , mixture density modeling and conditional mixture density modeling . We evaluate the relative strengths of these approaches , and describe a number of novel qualitative and quantitative observations about switching in the galactose network . Bifurcation analysis is a branch of dynamical systems theory concerned with steady-state or asymptotic behaviors of a dynamical system [19] . Typically , bifurcation analysis is applied to a deterministic dynamical model , such as a system of difference equations or differential equations . To give a concrete example inspired by the data presented and analyzed later in this paper , imagine a situation where a single gene is activated by an input signal , representing , for example , the activity of transcription factor protein . Let denote the gene's protein product . Suppose that the gene is an auto-activator: the protein product acts as a transcription factor to upregulate its own expression . Following standard modeling approaches ( e . g . [28] ) we describe the time-varying behaviour of the protein abundance by the differential equation ( 1 ) where the parameter corresponds to a basal level of protein production , is the maximal additional production attributable to regulation , and characterize the effects of the activators , and indicates the rate of protein degradation or dilution due to cell growth . Figure 1A is a bifurcation diagram for this system , showing the steady state values of as a function of the input , which in this context is called the bifurcation parameter . Intuitively , if levels of are low , then little is produced and the system reaches a steady state at a low level of . Conversely , if is highly abundant , then a great deal of is produced , leading to a high steady state . Most interestingly , when lies in and intermediate range , three steady states coexist . Intermediate levels of and a large initial amount of will stimulate sufficient production to maintain at a high concentration . However , if initially the level of is low , production is not maintained , and the system reaches a low steady state . There is also a third , unstable steady state between the low and high steady states . The values of at which the number of steady states changes , i . e . , the turns of the ‘S’-shaped curve in Figure 1 , are called bifurcation points and correspond in a deterministic system to the critical values of where a small change in this parameter may cause the system to transition between states of low and high levels of . In contrast with deterministic models , real cellular networks can be significantly noisy , with system variables fluctuating over time for a variety of reasons , including , for example , fluctuations in biochemical reaction rates , random partitioning of cellular content at cell division , and variation in cell size and cell age ( see e . g . , [21] ) . Thus , if one were to observe multiple instances of a bistable system—say , a culture of genetically identical cells experiencing a homogeneous medium—one would not expect the experimental measurements to agree with the predictions of a deterministic model , even after the culture has attained a steady behaviour [29] . Noticeably , stochastic fluctuations will constantly push individual cells on excursions away from a stable expression state , causing a broadening of the population distribution around this state . The mean of the population distribution will reflect the steady state expression only when these excursions are symmetric , and the mode of the distribution , which corresponds to the state where the system on average spends most time , may be the better surrogate of deterministic steady states in a stochastic dynamical system ( e . g . , [30] ) . In a bistable system , fluctuations can induce stochastic transitions between the two expression states such that some cells are expressing at low level while others express at high levels . The result is the emergence of a bimodal population distribution and subpopulations with distinct expression characteristics . Figure 1B depicts what the steady state distribution for might look like as a function of , assuming the stochastic system would show a lognormal distribution for about the deterministic steady states . ( For a graph of real data from that galactose network , see Figure 2 . ) In this case , the time-invariant steady state distribution of the system is reached when the probability that a cell will switch from the low to the high expression state is the same as that associated with a transition from the high to the low expression state . The time it takes for the system relax to steady state , which is set by the kinetic rate parameters and the level of noise in the system , can range from the order of seconds to several tens of cell generations [31] . It is also noted that very rapid transitions between expression states may result , at the population level , in a persistent subpopulation that is not associated with a steady state in the deterministic model , and that noise , under certain conditions , may shift the location of bifurcation points or induce new bifurcations ( see e . g . [32] ) . How can we capture the bifurcation behavior of a stochastic dynamical system ? Suppose that represents the bifurcation parameter ( e . g . , an externally controlled parameter or variable ) , and represents an observed variable of the system , such as the protein abundance . Suppose that for any value of , and under a specified set of experimental conditions , we observe a population of cells with values of following some distribution . We propose that the stochastic bifurcation structure of the system should specify four pieces of information as a function of the parameter : This is not a formal definition of stochastic bifurcation structure; these are principles , which might be formalized in a number of different ways . For example , as mentioned above , the modes of the steady state distribution of a stochastic dynamical system have previously been proposed as analogs to the steady states of a deterministic model . Thus , one might use the modes of the distribution to determine the number and location of subpopulations , satisfying the first two parts of the definition above . In particular , one could use bimodality as a defining feature of bistability in a stochastic switching system and associate bifurcation points with parameter values where the population distributions change from unimodal to bimodal . In many cases , this may work well , although below we will show some reason to question the use of modes as defining of the number of subpopulations . If one can assign every cell to a subpopulation , then the variance of within each subpopulation and the relative sizes of the subpopulations provide natural answers to the third and fourth parts of the definition above . As with the locations of the modes , these features of the stochastic bifurcation structure may be related to properties of a deterministic model . For example , the degree of variation around a mode , or the fraction of time the system spends near the mode , are related to the degree of stability of the state in the deterministic model [32] . Below , we use the formalism of mixture models to instantiate these four principles of stochastic bifurcation structure . First , however , we present our experimental data on the galactose network . Our thoughts on stochastic bifurcation structure and methods to estimate it were motivated , indeed necessitated , by data we collected on activity in the galactose utilization network in S . cerevisiae . The network includes genes for the import and metabolism of galactose as well as various regulatory genes [33] , [34] , and is known to behave as a bistable switching network . For a range of external galactose concentrations , cells stochastically switch between induced and non-induced states [25] . To assay this behavior , a standard laboratory strain was augmented with a gene encoding a fluorescent protein under the control of the promoter region normally regulating the transcription of the endogenous Gal10 gene ( see Materials and Methods ) . Gal10 is a general indicator of activation of the network , hence the fluorescent reporter should be expressed when and only when the native network is itself active . Cells were cultured for 22 hours in 17 different constant concentrations of galactose from two different initial conditions—pregrowth in the absence of galactose to establish a non-induced initial state or pregrowth in the presence of galactose at high concentration to establish an initial state where all cells are induced . The activity of the network in individual cells was quantified by flow cytometry to measure the intensity of the fluorescence emitted by the expressed reporter gene . Four biological replicates were made of every experiment . The collected data comprises counts of how many cells were detected in each of 1024 fluorescence channels , which are logarithmically related to real fluorescence intensity and have a dynamical range of four orders of magnitude ( i . e . , channel 1024 represents 10 , 000 times the intensity of channel 1 ) . Figure 2 displays the data , which is broadly consistent with previous experiments [25] . At low galactose levels , all cells show low network activity . At higher galactose concentrations , a highly active subpopulation emerges , and at yet higher levels , the highly active subpopulation dominates and the low-activity subpopulation disappears . While these overall trends in the data are visually clear , the challenges in analyzing the data quantitatively include robustly determining the locations and sizes of the subpopulations , especially when one is much smaller than the other , dealing with cells not clearly attributable to any one subpopulation , and separating cell-to-cell variability from replicate-to-replicate variability . Ideally , these should be done in a statistically robust , computationally simple , and objective manner . We have defined a notion of stochastic bifurcation structure suitable for studying the behavior of stochastic genetic switches , and we have generated an extensive map of the response of the canonical bistable yeast galactose utilization network to variation in external galactose concentrations . While the data broadly conforms to our expectations for stochastic switching between low and high expression states within the network , several additional properties are noteworthy . The establishment of a “high” expressing subpopulation occurs rather abruptly and fairly consistently at a concentration of approximately 0 . 003% galactose , although this state is initially overlapping the low expressing subpopulation . By contrast , the low subpopulation fades away more gradually at higher concentrations , while maintaining clear separation from the high subpopulation . Activity within the high subpopulation , in terms of fluorescent intensity , increases substantially as a function of galactose concentration—by approximately 300% over the range of concentrations tested . Activity within the low subpopulation is fairly constant , and is , in most cases , indistinguishable from that of cells not expressing the reporter gene ( data not shown ) , though there may be a mild increase in expression as the galactose concentration increases . Hence , the response of the network to varying conditions appears to combine a boolean-type “binary” switch between “on” and “off” expression states with a continuous “graded” modulation of activity within the “on” state . From a methodological point of view , we proposed that mixture density estimation and conditional mixture density estimation are ideally suited to extracting stochastic bifurcation structure from real , noisy data . Our tests of two different mixture fitting methods and one conditional mixture fitting method suggested that , in most respects , the methods are equally accurate in fitting the data . It is possible that the conditional mixture model was less accurate . Visually , it appears to overestimate the location of the high subpopulation at smaller galactose concentrations , and underestimate it at higher concentrations ( see Figures 3A or 4A , B ) . However , this is due simply to the affine form assumed for the dependence of subpopulation location on concentration . Alternative forms could readily be chosen to allow greater flexibility in fitting the data . Regardless , the overall level of disagreement between methods appeared smaller than the variability between different biological replicates . One potentially important distinction between the two mixture modeling approaches , standard EM and mode estimation followed by EM , is that the former is able to identify a “high” subpopulation at lower galactose concentrations than the second approach . This is because , at the lowest galactose concentrations , the “high” subpopulation is very broad and partially merged with the typical low subpopulation—in some cases , to such a degree that the overall distribution is still unimodal ( see Figure 5 ) . The standard EM method , because it requires multiple runs to avoid the problems of local minima and for cross-validation , is considerable slower than either of the other methods . Still , all methods run orders of magnitude faster than the data collection takes , so this is a minor concern . Conditional mixture models have several additional advantages compared to fitting the data at each galactose level separately: they use fewer total parameters , and are thus less likely to overfit the data , and they explicitly represent and make predictions for the bifurcation structure at all values of the bifurcation parameter—not only the values tested experimentally . This approach worked well on our data . The drawback of this approach is that it requires choosing functional forms to represent the dependence of mixture probabilities and mixture component parameters on the bifurcation parameter . In this case , a proper means of representing mixture probabilities only became clear after doing the individual fits . In early conditional mixture model fits , we assumed the mixture probabilities were independent of galactose concentration . This had the unfortunate side affect that the high component would start to “capture” cells from the low subpopulation at low galactose levels , dragging down the whole mean curve for the high subpopulation until it intersected and overlapped with the low subpopulation . The form we chose for the mixture probabilities avoids this problem by definitively assigning cells to the low component at all galactose levels below some threshold . This illustrates that the strength of using few parameters and explicitly generalizing across bifurcation parameter values also implies a danger of poor performance if an inappropriate representation is chosen . While this is a truism in the statistics and machine learning communities , it is all the more important to keep in mind in systems biology where there is a greater focus on interpreting models , as opposed to , say , being concerned only about prediction accuracy . Despite our focus on mixture modeling , one can imagine other approaches for estimating stochastic bifurcation structure . For example , clustering methods such as K-means or self-organizing maps could readily be applied in much the same way as we applied mixture density estimation . Nonparametric density estimation techniques might also be applied , although it would take extra effort to extract subpopulations from a nonparametric density estimate . Investigating such alternative approaches is an important topic for future research . Part of our contribution is in specifying four types of information that should be included in a stochastic bifurcation analysis: the number of distinct subpopulations , the fraction of cells they contain , the level of expression and the variance within each subpopulation . Our notion of stochastic bifurcation structure is considerably different from ideas employed in stochastic bifurcation theory , which addresses the behavior of explicitly stochastic dynamical models , such as stochastic differential equations [36] . The primary concern of stochastic bifurcation theory is the number and stability of different steady state distributions of a model . In gene regulatory networks , it is not unreasonable to assume that any given cell could eventually , through random fluctuations , reach the same state as any other cell [23] , [25] . Such a system is said to be “communicating” , and under fairly general conditions , has only a single steady state distribution for each bifurcation parameter value [37] . By the gross standard of stochastic bifurcation theory , such a system does not show any bifurcations at all . We , by contrast , have attempted to paint a finer-grained picture of the dependence of a stochastic dynamical system on an experimentally manipulated parameter . This picture is largely consistent with the expectation that fluctuations within the context of a deterministic network model constantly push individual cells on excursions away from a stable expression state , and induce stochastic transitions between the two expression states to generate bimodal population distributions [29] . Indeed , our focus identifying subpopulations is closely related to the idea in Kepler and Elston [29] of defining bifurcations via the number of critical points in the steady state distribution . However , our approach is much different; whereas they start with first-principles stochastic chemical descriptions of simple gene regulatory models , we start with empirical measurements of a complex gene regulatory system . Stochastic bifurcation structure may provide useful information for the development of quantitative regulatory network models , however this remains to be investigated . The exact relationship between stochastic observables and model features is not yet clearly established . For example , models of gene regulatory networks are usually derived from molecular interactions within individual cells and rarely consider effects due to population dynamics . The gradual fading of the low-expressing subpopulation observed in our experiments could be due the stochastic dynamics of the regulatory network itself , or it could be due to a reduced growth rate of the low-expressing cells . Additionally , while we took steps to present the cells in each culture with homogenous extracellular conditions ( see Materials and Methods ) , it is likely that there was some variability in the conditions experienced by different cells or by the same cell over time . Depending on the magnitude of this effect , it too might need to be estimated , if possible , and separated from intrinsic cell-to-cell variability if one wants accurate estimates of cellular network parameters . Careful quantitative estimation of stochastic bifurcation structure facilitates comparison between different experimental conditions or genetic backgrounds . For example , the yeast strain studied by Acar et al . ( W303 ) is much less sensitive to galactose and displays an almost 10-fold shift of the bimodal region ( to concentrations between approximately 0 . 02% and 0 . 3% ) compared to our strain ( an equivalent of BY4743; see also discussion in Bennett et al . [38] ) . Thus , even subtle differences in DNA-encoded parameters may have significant impact on the stochastic bifurcation structure of a given gene regulatory network . It should be possible to link DNA sequence information to quantitative properties of gene regulatory networks . This may require the development of several methodological , in addition to experimental , approaches that can extract consistent information about stochastic bifurcation structures . For example , it would be necessary to compare different , empirically-measured stochastic bifurcation structures associated with different genotypes to determine whether there is a statistically significant difference between them and , if so , identify the origin of the difference using a dynamical systems theory or other type of modelling framework . In addition , such methods could be useful to investigate how gene regulatory networks have evolved , to infer regulatory relationships between genes , or refine our knowledge of them , based on stochastic bifurcation behavior in experiments involving systematic genetic perturbations , such as gene deletions , gene knockdown or overexpression experiments . The experiments use a diploid Saccharomyces cerevisiae strain expressing a single copy of yeast-enhanced green fluorescent protein ( yEFPG ) from the native promoter of the GAL10 gene ( ) . The diploid was obtained by mating two haploid strains , a Mat a strain ( yHP101 ) derived from BY4741 ( Mat a , ; ; ; , Open Biosystems ) by PCR-mediated replacement of the open reading frame of the Ade2 gene by a Leu2 expression cassette , and a Mat strain ( yHP201 ) derived from BY4742 ( Mat ; ; ; ; , Open Biosystems ) by PCR-mediated gene replacement of Ade2 by a DNA fragment carrying the reporter cassette and an expression cassette conferring histidine auxotrophy . Following PCR validation of the appropriate gene replacements , the diploid strain , designated yHP301 ( Mat , ; ; ; ; ; ) was stored at in rich media ( YPD ) containing 20 g/L Yeast Bacto-Peptone ( Wisent ) , 10 g/L yeast extract ( Wisent ) 20 g/L glucose ( Sigma-Aldrich ) and 1% w/vol adenine ( Sigma-Aldrich ) supplemented with 15% w/vol glycerol ( Sigma-Aldrich ) . Prior to quantification , yHP301 was streaked onto synthetic dropout medium ( Wisent , Inc . ) agar plates without leucine and histidine supplemented with 2% w/vol glucose and 1% w/vol adenine . Individual colonies were used to inoculate 3 mL rich media ( YPR ) containing 20 g/L Yeast Bacto-Peptone , 10 g/L yeast extract , 1% w/vol adenine and 2% w/vol raffinose ( Wisent ) or YPR media supplemented with 2% w/vol galactose ( Becton , Dickenson ) . Following growth for 24 hours at and continuous shaking ( 250rpm ) , twenty-one aliquots of each culture were transferred to a deep well block and washed twice with YPR media supplemented with varying amounts of galactose ( final concentrations 0 . 0 , 0 . 0015 , 0 . 0022 , 0 . 0033 , 0 . 0038 , 0 . 0043 , 0 . 0050 , 0 . 0057 , 0 . 0066 , 0 . 0076 , 0 . 0087 , 0 . 0100 , 0 . 0115 , 0 . 0132 , 0 . 0174 , 0 . 020 , 0 . 080 , 0 . 20 , 0 . 50 , 2 . 0%w/vol ) . Following the wash , cells were resuspended in of the appropriate media and optical density ( OD ) quantified with a Perkin Elmer Victor3V plate reader using cultures . A fraction of the remaining volume was subsequently used to inoculate fresh media containing the appropriate amount of galactose to an OD of , and grown in a 96 deep well block for 22 hours at and 250rpm prior to analysis . Reporter gene expression was quantified in individual cells using a Beckman-Coulter FC500 flow cytometer . A total of 60 , 000 events were collected for each condition and filtered using custom-written software script using a fixed elliptical forward/side-scatter autogate capturing approximately 50% of the events in each sample . The fluorescence intensity ( 488nm excitation , 510–550nm emission ) associated with these events was used to generate representative expression distributions for each sample condition . A total of four replicates were obtained , for each final galactose concentration and both pre-growth conditions . Mixture density estimation using EM used 100 runs in an effort to avoid problems with stopping at solutions that were only locally optimal . Each of the 100 runs began from different random initial parameters . The means of each Gaussian component were chosen uniformly between the lowest and highest data point . Standard deviations were initialized to 50—roughly the level observed at single-subpopulation galactose concentrations—and initial mixture probabilities for the Gaussians were set to where is the number of Gaussians . ( Recall that a fixed 0 . 02-weighted uniform density is also part of the mixture ) . The exception to this rule was the mode-estimation-plus-EM approach , for which means were initialized to the mode estimates , and we used a single run of EM . The parameter updates during the M-step were as described , e . g . , in Bishop [35] . If the variance of a Gaussian shrank below , the component was eliminated , because such a Gaussian is focussed on a single fluorescence channel , and does not represent a true subpopulation . The EM fitting employed cross-validation to determine the proper number of Gaussian components to have in the mixture for each replicate and at each galactose level . After fitting a model with Gaussian components , we tested whether an Gaussian model would be significantly better by performing 10-fold cross-validation . In each fold , 90% of the data was used to fit an Gaussian model , which was scored by the mean ( across data points ) log likelihood of the remaining 10% of the data . We calculated the mean and standard deviation ( across folds ) of the Gaussian model scores . If the mean was standard deviations greater than the score of the Gaussian model , we accepted the increase to Gaussians , and performed the process again . We chose , as we found this was sufficiently stringent to prevent splitting of what were clearly single subpopulations ( e . g . , at zero galactose concentration ) . Mode estimation for the mode-estimation-plus-EM approach began by smoothing the data by taking a running average over a window of size 71 channels . Call this . First and second derivatives , and , were estimated by computing centered finite differences , with the same width of 71 channels . A mode in the density was detected at channel point if the first derivative crossed from positive to negative ( i . e . , and ) and if . Ordinarily , one might threshold only the second derivative . However , small bumps in the data series are characterized by both smaller first and second derivatives in the vicinity of a mode , and combining them in this way lead to more robust and balanced detection of peaks of all sizes in preliminary tests . The choices of a 71-width averaging window and the −0 . 0002 threshold were based on pilot testing on a separate , but related , set of flow cytometry data . For fitting the conditional mixture density models , we used only a single run of EM , as further runs did not improve accuracy . Updates are standard , as given in Bishop [35] . Low and high subpopulation means were initialized to have means of 200 and 700 respectively ( independent of galactose level ) , standard deviations were initialized to 50 , and mixture probabilities to 0 . 49 . All code is written in MATLAB . Code and raw data are available upon request , as well as on TJP's website: http://www . perkinslab . ca	Decades ago , Waddington , and later Kauffman , likened the dynamics of a differentiating cell to a marble rolling downhill on bumpy terrain—the epigenetic landscape . In this metaphor , the valleys of the landscape represent the paths that cells can follow towards a stable cell type , and the fate of the cell is determined by the constant modulation of the epigenetic landscape by internal and external signals . With new technologies for measuring single-cell gene expression , it is increasingly feasible to map out these valleys and how external variables influence cellular responses . Moreover , it is possible to quantify population level effects , such as what fraction of a population of cells arrives at one valley or another , and variability at the cellular level , such as how individual cells bounce around within , and possibly between , valleys due to the stochasticity of cellular biochemistry . In this paper , we discuss which characteristics of the epigenetic landscape can readily be extracted from single-cell gene expression data , and describe computational methods for doing so .	[ "Abstract", "Introduction", "Results", "Discussion", "Materials", "and", "Methods" ]	[ "molecular", "biology/bioinformatics", "computational", "biology/systems", "biology" ]	2010	Estimating the Stochastic Bifurcation Structure of Cellular Networks
Annual mass treatment with ivermectin and albendazole is used to treat lymphatic filariasis in many African countries , including Tanzania . In areas where both diseases occur , it is unclear whether HIV co-infection reduces treatment success . In a general population study in Southwest Tanzania , individuals were tested for HIV and circulating filarial antigen , an indicator of Wuchereria bancrofti adult worm burden , before the first and after 2 consecutive rounds of anti-filarial mass drug administration . Testing of 2104 individuals aged 0–94 years before anti-filarial treatment revealed a prevalence of 24 . 8% for lymphatic filariasis and an HIV-prevalence of 8 . 9% . Lymphatic filariasis was rare in children , but prevalence increased in individuals above 10 years , whereas a strong increase in HIV was only seen above 18 years of age . The prevalence of lymphatic filariasis in adults above 18 years was 42 . 6% and 41 . 7% ( p = 0 . 834 ) in HIV-negatives and–positives , respectively . Similarly , the HIV prevalence in the lymphatic filariasis infected ( 16 . 6% ) and uninfected adult population ( 17 . 1% ) was nearly the same . Of the above 2104 individuals 798 were re-tested after 2 rounds of antifilarial treatment . A significant reduction in the prevalence of circulating filarial antigen from 21 . 6% to 19 . 7% was found after treatment ( relative drop of 8 . 8% , McNemar´s exact p = 0 . 036 ) . Furthermore , the post-treatment reduction of CFA positivity was ( non-significantly ) larger in HIV-positives than in HIV-negatives ( univariable linear regression p = 0 . 154 ) . In an area with a high prevalence for both diseases , no difference was found between HIV-infected and uninfected individuals regarding the initial prevalence of lymphatic filariasis . A moderate but significant reduction in lymphatic filariasis prevalence and worm burden was demonstrated after two rounds of treatment with albendazole and ivermectin . Treatment effects were more pronounced in the HIV co-infected subgroup , indicating that the effectiveness of antifilarial treatment was not reduced by concomitant HIV-infection . Studies with longer follow-up time could validate the observed differences in treatment effectiveness . Lymphatic Filariasis ( LF ) is a mosquito-borne disease caused either by Wuchereria bancrofti which is distributed throughout the tropics , or Brugia malayi and Brugia timori , both limited to Southeast-Asia . It is estimated that 120 million people world-wide are infected with one of these pathogens , and 1 billion are at risk to acquire LF during their lifetime [1] . Before larger treatment programmes started , LF was present in most of the 21 regions of Tanzania with up to 63 . 8% of individuals testing positive for circulating filarial antigen , a marker for LF infection [2] . Since the year 2000 the “Global Alliance to Eliminate Lymphatic Filariasis” uses annual mass drug administration ( MDA ) , with the aim to control and ultimately eliminate the disease [3] . The campaign of the Tanzanian National Lymphatic Filariasis Elimination Programme ( NLEFP ) commenced in 2001 in the coastal regions of Tanzania . In the Mbeya district in Southwest-Tanzania the treatment programme started in October 2009 with the annual distribution of albendazole ( 400mg ) and ivermectin ( 150–200μg/kg ) . Ivermectin is considered to be mainly microfilaricidal [4] , for albendazole an effect on the release of intrauterine antigen components of the adult worm was described [5] . Some studies report on the treatment effectiveness of the combination of albendazole and ivermectin after 12-month: in Ghana a significant reduction in circulating filarial antigen ( CFA ) levels but no measurable reduction of CFA prevalence was described in 370 individuals receiving both drugs [6 , 7] . A longitudinal study from Northern Tanzania showed only small reductions of CFA positivity after two annual drug distributions ( from 53 . 3% to 51 . 4% ) , but a significant drop to 44 . 9% and 19 . 6% after four and seven years of treatment , respectively [8] . In South Western Tanzania , both LF and HIV are public health concerns . The HIV prevalence in the country has been documented in several national surveys [1 , 9 , 10] . The third population based Tanzanian HIV/AIDS and Malaria Indicator Survey in 20011/2012 ( THMIS ) revealed a country-wide HIV prevalence of 5 . 1% in Tanzanian adults between the age of 15 and 49 years , and a prevalence of 9 . 0% for this age-group in Mbeya Region [10] . Large scale distribution of antiretroviral ( ART ) drugs was initiated in Tanzania in 2005 . At the time of our study , ART was not widely available in Southwest Tanzania . [10–12] . Local differences in initial prevalence , coverage of treatment programs , co-infection with other pathogens , etc . can all affect treatment success , thus careful surveillance of the programs is necessary to control the infection . [1 , 13–17] . Only few manuscripts focus specifically on the possible interaction of HIV with LF and most of these use cross-sectional data [18–21] . Only one recently published study investigates the treatment effectiveness of MDA drugs in HIV/LF co-infected individuals [22] , but focusses on changes in CD4 and HIV viral load after antifilarial treatment in selected HIV-positive individuals . No study concentrated on the antifilarial treatment effectiveness of MDA drugs in HIV/LF co-infected individuals . Our study assesses LF prevalence in the Mbeya Region , before and after the governmental eradication program reached the area and examines the potential impact of HIV co-infection on LF treatment . Data were collected during the SOLF cohort-study ( Surveillance of Lymphatic Filariasis , http://www . mmrp . org/projects/basic-research/solf . html ) in the Kyela district/Mbeya region in Southwest Tanzania which was conducted at the National Institute for Medical Research ( NIMR ) —Mbeya Medical Research Centre ( MMRC ) between 2009 and 2011 . The study was embedded into the population based EMINI ( Evaluation and Monitoring of the Impact of New Interventions , http://www . mmrp . org/projects/cohort-studies/emini . html ) cohort study , which was carried out in 9 selected communities in the Mbeya region ( Fig 1 ) from 2006 to 2011 . More than 170 , 000 inhabitants from ~42 , 000 households of these communities were registered and 10% of households randomly selected to participate in the study . No additional households entered the surveillance , but some new participants entered through birth or marriage into included household . The SOLF study was approved by the Mbeya Medical Research and Ethics Committee and the Tanzanian National Institute for Medical Research—Medical Research Coordinating Committee as an amendment to the EMINI cohort study . Prior to enrolment , each EMINI participant had provided written informed consent regarding study participation . Parents consented for their children below 18 years of age . In addition , children above the age of 12 years signed their own assent form . Data and samples from participants in the Kyela site of the EMINI study were collected annually from 2007 until 2009 . During the last two surveys ( 2010 and 2011 ) only half of the study households were visited in each year . During each visit , which took place between 8 am and 2 pm , blood , urine and stool samples were collected from each participant . Samples from 2 , 165 participants from March 2009 were used to estimate the prevalence of LF directly before the government treatment program commenced in Kyela in October 2009 . In March 2011 , 18 month after the first and 6 month after the second delivery of antifilarial treatment , samples from 1 , 010 participants were used to evaluate treatment impact . From each study participant , 2 . 7 ml of blood was collected during morning hours in EDTA tubes and immediately stored at 4°C . Cells and plasma were separated within 24 hours and subsequently stored at -80°C . All laboratory tests were performed at NIMR-MMRC , Mbeya Tanzania . HIV testing was performed using the SD-Bioline HIV-1/2 3 . 0 ( Standard Diagnostics , Kyonggi-do , South Korea ) rapid diagnostic test ( RDT ) . Negative RDT results from one survey , followed by another negative RDT result in a subsequent survey , were regarded as confirmed negative and not further tested . All positive results were confirmed using an ELISA HIV test ( Enzygnost Anti HIV 1/2 Plus , DADE-Behring , Marburg , Germany ) , and tested by Western blot ( MPD HIV Blot 2 . 2 , MP Biomedicals , Geneva , Switzerland ) if discordant . For all HIV incident cases , the negative result of the previous year , as well as the new positive results was confirmed by the testing algorithm described above . For children below the age of two years , HIV testing was done by PCR . Further details are described elsewhere [23] . Because confidential disclosure of the HIV-status could not be ensured during household visits , we did not inform participants about their HIV status . Instead they were offered voluntary counseling and testing by an independent team , which was travelling with our study team , who provided referral to the local care and treatment center , to everyone who was tested positive . A commercially available ELISA ( TropBio Og4C3 serum ELISA , Townsville , Australia ) was used to detect circulating filarial antigen ( CFA ) using 100 μl of the collected sera . The Og4C3 antibody detects Wuchereria bancrofti antigen with high specificity ( 98 . 5% ) and no known cross-reaction to Onchocerca volvulus , Brugia spp . , Mansonella , Dracunculus medinensis , Ascaris lumbricoides or Strongyloides stercoralis [24] . Sensitivity varies between 73% [25] and 100% [26] , but was found 97 . 9% in individuals carrying microfilariae [24] . CFA is secreted by fully developed W . bancrofti adults and can be found at similar levels during day and night . Antigen levels thus reflect the W . bancrofti worm burden . The measurement of CFA with the Trop Bio ELISA is semi-quantitative; seven control tubes with standardized amounts of antigen are supplied and allow an estimation of the filarial antigen levels in the analysed plasma according to the measured optical density ( OD ) . LF test results were considered negative , indeterminate or positive if the OD was <0 . 2 , ≥0 . 2 and ≤0 . 3 , or >0 . 3 respectively . Statistical analyses were performed using Stata statistics software ( version 14; Stata Corp . , College Station , TX ) . Pearson´s chi-squared test was used to compare binominal outcomes between groups and to compare CFA positivity before and after treatment in all participants . McNemar´s exact test for paired data was used to compare CFA positivity before and after treatment in those individuals who participated in both surveys . The non-parametric Wilcoxon rank sum test was used to compare selected baseline characteristics of continuous variables , since none of these was normally distributed . In order to examine the association of LF infection with HIV status and other potentially important covariates we performed uni- and multi-variable log link binomial regression analyses with robust variance estimates . In March 2009 , before the first national MDA commenced , valid CFA results were obtained from 2 , 104 individuals ( Table 1 ) . Indeterminate results were found for the 61 of the tested 2 , 165 samples ( 2 . 8% ) . Their median age was 16 . 6 years ( range 0–94 , IQR: 8 . 8 to 34 ) , and 51 . 0% were female . Only 4 ( 1 . 6% ) of the 245 children below the age of 5 years were CFA-positive; LF prevalence started to rise in participants above 10 years and was 42 . 3% in adults above 18 years of age ( Fig 2 ) . When including all age groups , 24 . 8% of the study population were CFA-positive with a trend to higher prevalence in males ( 26 . 5% ) than in females ( 23 . 1% , chi-squared p = 0 . 074 ) . In the adult population above 18 years the difference in CFA-positivity between males ( 47 . 3% ) and females ( 38 . 0% ) was significant ( chi-squared p = 0 . 003 ) . In March 2011 , 18 months after the first MDA and six months after the second , ~50% of the initially included households were revisited for interviews and blood sample collection . Some scheduled participants were not found in 2011 , and some new individuals had entered the visited households ( see study population and design ) . In addition to an analysis where the data of all participants form each Survey ( = open cohort ) are evaluated , which reflects more a cross sectional design , a second analysis included only the 798 individuals who actively participated in both years of the surveillance longitudinally ( = closed cohort ) . The numbers of participants is shown in Table 1 . Of the 974 valid test results in 2011 , 19 . 7% were CFA-positive , leading to a calculated prevalence reduction of 5 . 1% ( 24 . 8% vs . 19 . 7% , chi-squared p = 0 . 002 ) when including all subjects who participated in at least one survey ( Table 1 , open cohort ) . In the analysis of samples from 798 individuals who actively participated in both surveys ( Table 1 , closed cohort ) , a lower prevalence reduction ( 21 . 6 to 19 . 7% , McNemar´s exact p = 0 . 036 ) was measured ( Fig 3 ) . At baseline the overall HIV prevalence in our study cohort was 8 . 9% , with a prevalence of only 2 . 1% in children and adolescents below the age of 18 years , and a prevalence of 16 . 9% in individuals ≥18 years of age ( Fig 3 ) . HIV-infection was more prevalent in female ( 10 . 7% ) , compared to male participants ( 7 . 1% , chi-squared p = 0 . 003 ) . Sixty-eight of the 968 adult individuals ( 7 . 0% ) were infected with both pathogens and among the whole group of 2 , 104 individuals 69 co-infections ( = 3 . 3% ) were observed . The initial univariable analysis of the potential association of HIV with LF infection showed a higher prevalence of LF in HIV-positive ( 36 . 9% ) ; compared to HIV-negative individuals ( 23 . 6% ) ( RR = 1 . 56 , 95% CI = 1 . 26 to 1 . 94 , p<0 . 001 ) . But we already demonstrated that HIV and LF are both less common in children than in adults , which confounds this association . To further study the pattern of co-infection we analysed CFA positivity in HIV infected and uninfected individuals stratified by age ( Fig 4 ) ; in adults ( > = 18 years ) only; and in log-link binomial multivariable regression adjusted for age and gender . None of these analyses showed a significant association of LF infection with HIV , neither within the single age strata nor overall in the multivariable regression model where the influence of age and gender were confirmed , but where the adjusted RR for HIV was only 1 . 04 ( Table 2 ) . When only analysing data from adults above 18 , the CFA prevalence was 42 . 6% in the HIV-negative and 41 . 7% in the HIV-positive subgroup ( univariable log-link regression RR = 0 . 98 , 95% CI = 0 . 80 to 1 . 20; p = 0 . 84 ) . In order to compare antifilarial treatment success in the HIV-negative and positive subgroups we again performed two analysis: one for all tested individuals who participated in at least one survey ( open cohort using chi-squared testing ) , and one only for the individuals who participated in both surveys before and after treatment ( closed cohort , using McNemar´s exact test ) . For the open cohort a CFA prevalence reduction from 23 . 6% to 18 . 9% ( chi-squared p = 0 . 015 , relative drop = 19 . 7% ) was found in HIV-negative participants , and from 36 . 9% to 27 . 5% ( chi-squared p = 0 . 023 , relative drop = 25 . 4% ) in HIV-positives . For the closed cohort we observed a drop in CFA positivity from 20 . 9% to 19 . 4% ( McNemar´s exact p = 0 . 117 , relative drop = 7 . 3% ) in 723 HIV-negatives and from 27 . 5% to 21 . 7% ( McNemar´s exact p = 0 . 125 , relative drop = 21 . 1% ) in 69 HIV-positive participants . The reason for this pronounced difference ( 7 . 3% vs . 21 . 1% ) is a higher incidence of CFA positivity in the HIV negative participants where 15 ( 2 . 6% ) of the 572 initially CFA negative participants turned CFA-positive , whereas none of the 50 HIV-positive participants who were initially CFA-negative turned CFA-positive ( chi-squared p = 0 . 246 ) . The proportion of initially CFA positives who turned CFA negative was very similar in HIV-negative ( 26/151 = 17 . 2% ) and HIV-positive participants ( 4/19 = 21 . 1% , chi-squared p = 0 . 679 ) . When combining this information about change in LF status in one outcome variable ( -1 = turned CFA negative; 0 = no change in CFA status; 1 = turned CFA positive ) univariable linear regression modelling resulted in a coefficient β for the HIV infected subgroup of -0 . 043 ( 95%CI = -0 . 102 to 0 . 016 , p = 0 . 154 ) . Analysing the prevalence reduction in the closed cohort for adults > = 18 years only , a drop from 42 . 7% to 40 . 3% ( McNemar´s exact p = 0 . 248 , relative drop = 5 . 6% ) was noted for HIV-negatives , and from 32 . 7% to 25 . 5% ( McNemar´s exact p = 0 . 125 , relative drop = 22 . 0% ) in the HIV-positive subgroup . Summarizing our results , we found more pronounced drops in prevalence among the HIV positive subgroup , compared with the HIV negative , no matter , whether all participants or only adults are analysed and also with both possible ways of evaluating the data ( open cohort or closed cohort ) . The measurement of CFA with the Trop Bio ELISA is semi-quantitative; with the OD of the plasma samples reflecting the participant’s worm burden . Our findings for CFA intensities parallel those for CFA prevalence: geometric mean intensities before treatment were relatively similar between HIV-positives ( 157 units ) and HIV-negatives ( 179 units , Wilcoxon rank sum p = 0 . 34 ) , which is also true for the relative reduction of geometric mean intensity after treatment , which was 26% and 30% respectively ( Wilcoxon rank sum p = 0 . 50 ) In an area with high prevalence of and no previous treatment against LF we investigated the potential association of HIV and LF infection . When adjusting for age we found similar CFA prevalence and intensities in HIV-positive and negative participants . After two rounds of treatment a significant reduction in CFA prevalence and intensity was demonstrated , which was more pronounced in the HIV-positive compared to HIV-negative participants . Hence , HIV co-infection does not seem to negatively affect antifilarial treatment .	Parasite infections and HIV show large geographical overlap in sub-Saharan Africa and could hence potentially interact in co-infected individuals . In a general-population study conducted in Southwest Tanzania , we found high prevalence of both , lymphatic filariasis and HIV , with 42 . 5% of the adult population infected with Wuchereria bancrofti and 16 . 8% infected with HIV . Seven percent of the adults were infected with both pathogens . When adjusting for age , there was no statistically significant difference in initial prevalence or worm burden between HIV-positive and negative participants . For 798 individuals test results for both diseases were available in 2009 , before and in 2011 , after 2 rounds of treatment against lymphatic filariasis . Between 2009 and 2011 , a significant drop of prevalence and worm burden in infected individuals were observed , which was more pronounced in the HIV co-infected subgroup . Hence , HIV co-infection does not seem to negatively affect lymphatic filariasis treatment programmes .	[ "Abstract", "Introduction", "Methods", "Results", "Discussion" ]	[ "invertebrates", "hiv", "infections", "medicine", "and", "health", "sciences", "pathology", "and", "laboratory", "medicine", "pathogens", "tropical", "diseases", "microbiology", "geographical", "locations", "parasitic", "diseases", "animals", "retroviruses", "viruses", "i...	2016	Prevalence of Lymphatic Filariasis and Treatment Effectiveness of Albendazole/ Ivermectin in Individuals with HIV Co-infection in Southwest-Tanzania
Eukaryotic positive-strand RNA [ ( + ) RNA] viruses are intracellular obligate parasites replicate using the membrane-bound replicase complexes that contain multiple viral and host components . To replicate , ( + ) RNA viruses exploit host resources and modify host metabolism and membrane organization . Phospholipase D ( PLD ) is a phosphatidylcholine- and phosphatidylethanolamine-hydrolyzing enzyme that catalyzes the production of phosphatidic acid ( PA ) , a lipid second messenger that modulates diverse intracellular signaling in various organisms . PA is normally present in small amounts ( less than 1% of total phospholipids ) , but rapidly and transiently accumulates in lipid bilayers in response to different environmental cues such as biotic and abiotic stresses in plants . However , the precise functions of PLD and PA remain unknown . Here , we report the roles of PLD and PA in genomic RNA replication of a plant ( + ) RNA virus , Red clover necrotic mosaic virus ( RCNMV ) . We found that RCNMV RNA replication complexes formed in Nicotiana benthamiana contained PLDα and PLDβ . Gene-silencing and pharmacological inhibition approaches showed that PLDs and PLDs-derived PA are required for viral RNA replication . Consistent with this , exogenous application of PA enhanced viral RNA replication in plant cells and plant-derived cell-free extracts . We also found that a viral auxiliary replication protein bound to PA in vitro , and that the amount of PA increased in RCNMV-infected plant leaves . Together , our findings suggest that RCNMV hijacks host PA-producing enzymes to replicate . Positive-strand RNA [ ( + ) RNA] viruses are the most abundant plant viruses , and include many viruses economically important in agriculture . ( + ) RNA plant viruses have a limited coding capacity . To replicate and achieve successful infection in their hosts , they need to use host proteins , membranes , lipids , and metabolites . All characterized eukaryotic ( + ) RNA viruses replicate their genomes using viral replication complexes ( VRCs ) , which contain multiple viral and host components on intracellular membranes [1–6] . A growing number of studies have suggested that plant viruses have evolved ways to hijack plant host factors and reprogram host cell metabolism for their successful infection [6 , 7] . Conversely , plants have evolved the ability to recognize viruses through specific interaction with viral proteins or viral double-stranded RNA intermediates for restricting virus infection [8 , 9] . Viruses must circumvent or suppress such surveillance systems and host defense mechanisms . Thus , viruses must be evolved to achieve a good balance between hijacking/reprogramming host factors for efficient viral replication and avoiding the danger of stimulating antiviral defense responses . Red clover necrotic mosaic virus ( RCNMV ) is a ( + ) RNA plant virus and a member of the genus Dianthovirus in the family Tombusviridae . The genome of RCNMV consists of RNA1 and RNA2 . RNA1 encodes a p27 auxiliary replication protein , p88pol RNA-dependent RNA polymerase ( RdRp ) , and a coat protein [10] . RNA2 encodes a movement protein that is required for viral cell-to-cell movement [10 , 11] . p27 , p88pol , and host proteins form a 480-kDa replicase complex , which is a key player in the viral RNA replication [12] . p27 and p88pol colocalize at the endoplasmic reticulum ( ER ) [13 , 14] , where RCNMV replication takes place [15] . Our previous studies showed that RCNMV uses host heat shock proteins ( HSPs ) , HSP70 and HSP90 [16] , and ADP-ribosylation factor 1 ( Arf1 ) [15] for the formation of the 480-kDa replicase complex and p27-induced ER membrane alterations . Arf1 is a small GTPase that regulates COPI vesicle formation . Sar1 , another small GTPase that regulates COPII vesicle-mediated trafficking , and Arf1 are recruited from their original subcellular locations to RCNMV replication sites via p27 , and p27 interferes with host membrane trafficking pathway in plant cells [15 , 17] . Mammalian and yeast Arf1 recruits and/or stimulates its effector proteins , including a coatomer , phosphatidylinositol 4 kinase III β ( PI4KIIIβ ) , and phospholipase D ( PLD ) [18] . Arf1 can activate mammalian PLD1 and PLD2 directly . PLD hydrolyses structural phospholipids such as phosphatidylcholine ( PC ) and phosphatidylethanolamine ( PE ) to produce phosphatidic acid ( PA ) and remaining headgroups . PA production resulting from Arf1-mediated PLD activation has been proposed to be associated with vesicle formation [19] . The 12 different PLD isoforms encoded in the Arabidopsis thaliana genome are classified into six groups ( α , β , γ , δ , ε , and ζ ) based on sequence similarity and in vitro activity [20] . PLDζ1 and ζ2 have N-terminal phox homology ( PX ) and pleckstrin homology ( PH ) domains and share high sequence similarities to two PX/PH-PLDs in mammals . The remaining PLDs contain the Ca2+-dependent phospholipid-binding C2 domain and are unique to plants . PA is normally present in small amounts ( less than 1% of total phospholipids ) , but rapidly and transiently accumulates in lipid bilayers in response to different abiotic stresses such as dehydration , salt , and osmotic stress [20–22] . PA also accumulates in response to several microbe-associated molecular patterns ( MAMPs ) in plant cells and positively regulates salicylic acid ( SA ) -mediated defense signaling [23–27] . Moreover , effector proteins of bacterial and fungal pathogens , such as Cladosporium fulvum Avr4 and Pseudomonas syringae AvrRpm1 and AvrRpt2 , trigger PA accumulation in their host cells , and multiple PLD isoforms contribute to AvrRpm1-triggered resistance in Arabidopsis thaliana [28–30] . PLDδ plays a positive role in MAMPs-triggered cell wall based immunity and nonhost resistance against Blumeria graminis f . sp . hordei [31] . Moreover , overexpression of rice diacylglycerol ( DAG ) kinase , which catalyzes the conversion of DAG to PA , enhances resistance against tobacco mosaic virus and Phytophthora parasitica infections in tobacco plants [32] . In accordance with this , direct application of PA to leaves has been shown to induce the expression of pathogenesis-related ( PR ) genes and cell death [28 , 33] . These findings indicate that PA is a positive regulator in plant defense against pathogens . In contrast , PLDβ1 acts like a negative regulator of the generation of reactive oxygen species ( ROS ) , the expression of PR genes , and plant defenses against biotrophic pathogens in rice and Arabidopsis [34–36] . In this study , using two-step affinity purification and liquid chromatography–tandem mass spectrometry ( LC/MS/MS ) analysis , we identified Nicotiana benthamiana PLDα and PLDβ as interaction partners of RCNMV replication protein , p88pol . Gene-silencing and pharmacological inhibition approaches show that PLDs-derived PA plays a positive role in viral RNA replication . Consistent with this role , direct application of PA to plant cells or plant-derived cell-free extracts enhanced RCNMV RNA replication and negative-strand RNA synthesis , respectively . We found that p27 auxiliary replication protein interacted with PA in vitro and that the accumulation of PA increased in RCNMV-infected plant leaves . Together , our findings suggest that RCNMV hijacks host PA-producing enzymes to achieve successful RNA replication . To test whether NbPLDα and NbPLDβ are required for infection of host plants with RCNMV , endogenous transcript levels of NbPLDα or NbPLDβ were downregulated using Tobacco rattle virus ( TRV ) -based virus-induced gene silencing ( VIGS ) in N . benthamiana plants . A TRV vector harboring a partial fragment of NbPLDα ( TRV:NbPLDα ) or NbPLDβ ( TRV:NbPLDβ ) was expressed via Agrobacterium-mediated expression . An empty TRV vector ( TRV:00 ) was expressed as a control . Newly developed leaves were inoculated with RCNMV RNA1 and RNA2 at 18 dai . Two days after RCNMV inoculation , three inoculated leaves from three different plants were harvested and mixed , and total RNA was extracted . No morphological defects such as chlorotic and stunted phenotypes were observed at this stage ( S4 Fig ) . Semiquantitative reverse-transcription-PCR ( RT-PCR ) or quantitative RT-PCR ( RT-qPCR ) analyses confirmed the specific reduction of NbPLDα or NbPLDβ mRNAs in plants infiltrated with TRV:NbPLDα or TRV:NbPLDβ , respectively ( Figs 3 and S5 ) . Northern blot analyses using ribonucleotide probes specifically recognizing RCNMV RNA1 or RNA2 showed that the accumulation of RCNMV RNAs was dramatically reduced in both NbPLDα- and NbPLDβ-knockdown plants compared with control plants ( Fig 3 ) . It is known that , in PLDβ1 knockdown transgenic rice plants , the generation of ROS and the expression of defense-related genes are induced even in the absence of pathogen infection [35] . Therefore , it is possible that the poor viral infection in NbPLDβ-knokedown plants was due to activated defense responses . To address this possibility , we tested the effects of NbPLDα- or NbPLDβ-knockdown by TRV-mediated VIGS on the expression of defense-related genes by RT-qPCR analysis . The defense-related genes analyzed here were SA-signaling marker genes ( PR-1 , PR-2 , and PR-5 ) [39 , 40] , jasmonic acid ( JA ) -signaling marker genes ( LOX1 , PR-4 , and PDF1 . 2 . ) [39 , 41] , ROS-detoxification enzymes ( APX , GST , and SOD ) [39] , MAMP-triggered immunity marker genes ( CYP71D20 and ACRE132 ) [42] , and mitogen-activated protein kinases ( WIPK and SIPK ) [42] . The expression levels of these defense-related genes were not significantly increased in NbPLDα- or NbPLDβ-knockdown plants compared with those in TRV control plants ( S5 Fig ) , excluding the possibility that the reduced viral infection in NbPLDα- or NbPLDβ-knockdown plants are due to activated defense responses in these plants . Some genes including PR-1 , PR-2 , and CYP71D20 genes were even repressed in NbPLDα- or NbPLDβ-knockdown plants ( S5 Fig ) , consistent with the positive roles of PLDs and PLD-derived PA on plant defense signaling [24–30] . Altogether , these results suggest that both NbPLDα and NbPLDβ play a positive role in RCNMV infection . To test the possible contribution of PLD-derived PA to viral RNA replication , we exploited the transphosphatidylation activity of PLDs , which uses primary alcohols as substrates to form an artificial phosphatidyl alcohol . The preferential formation of this compound impairs PA production [43] . Thus , we tested the effect of n-butanol that inhibits PA production by PLDs on RCNMV RNA replication . N . benthamiana protoplasts were inoculated with RCNMV RNA1 and RNA2 and incubated with n-butanol or tert-butanol , an alcohol with no inhibitory effect on PA production , and viral RNA accumulation was determined by northern blot analysis . Increasing n-butanol concentration caused a progressive reduction of viral RNA accumulation ( Fig 4A ) . By contrast , viral RNA accumulation was only moderately reduced in protoplasts treated with tert-butanol compared with the water control . Note that n-butanol did not affect the accumulation of rRNA ( Fig 4A ) . The inhibitory effect of n-butanol on RCNMV replication was also observed in tobacco BY-2 protoplasts ( S6 Fig ) . We also tested the effect of n-butanol and tert-butanol on defense-related gene expressions in N . benthamiana protoplasts . n-butanol did not increase the expression of defense-related genes ( S7 Fig ) . This result corresponds well with the finding that n-butanol has no effects on the basal transcription of the PR-1 gene in Arabidopsis seedlings [27] . In contrast , tert-butanol treatment caused the induction of CYP71D20 , ACRE132 , and WIPK genes ( S7 Fig ) . This may explain weak negative effect of tert-butanol on RCNMV replication ( Figs 4A and S6 ) . Altogether , these results suggest that PLD-derived PA plays a positive role in viral RNA replication . To verify further the importance of PA in viral RNA replication , commercially available soy-derived PA was supplied to RCNMV-inoculated N . benthamiana protoplasts and viral RNA accumulation was determined by northern blot analysis . Exogenously added PA enhanced the accumulation of RNA1 in a dose-dependent manner ( up to 6-fold increase by 5 μM PA ) , whereas the effect of PA on the accumulation of RNA2 was negligible ( Fig 4B ) . Neither exogenously supplied PC nor PE significantly affected the accumulation of viral RNA in N . benthamiana protoplasts ( S8 Fig ) . These results suggest that PA plays an important role in RCNMV RNA replication , and that the requirement for PA may differ between RNA1 and RNA2 . Replication of RNA2 depends entirely on RNA1 simply because RNA2 uses replication proteins supplied from RNA1 . The negative impact of n-butanol on the accumulation of RNA2 ( Fig 4A ) may be the indirect action of this primary alcohol through inhibition of RNA1 replication . Therefore , it is possible that RNA2 replication does not require any PLD-derived PA . To investigate this possibility , we inoculated N . benthamiana protoplasts with RNA2 and plasmids expressing p27 and p88pol , as suppliers of the replication proteins . The accumulation of RNA2 was decreased by n-butanol in a dose-dependent manner ( Fig 4C ) . Note that in this experiment , the accumulation of p27 replication protein was not significantly changed ( Fig 4C ) . These results indicate that PLD-derived PA was also required for the replication of RNA2 as in the case of RNA1 . However , the replication of RNA2 was not enhanced by exogenously supplied PA ( Fig 4D ) , suggesting that the threshold of PA requirement for RNA2 replication is lower than that for RNA1 . The differential requirement for PA in the replication of RNA1 and RNA2 is discussed later . Next , we investigated the effect of n-butanol on RNA replication of Brome mosaic virus ( BMV ) , another plant ( + ) RNA virus , which is unrelated to RCNMV . Increasing n-butanol concentration caused progressive reduction in the accumulation of BMV RNA ( Fig 5A ) . Co-IP experiments showed interactions between BMV replication proteins and NbPLDβ-FLAG ( Fig 5B and 5C ) . These results suggest that PLD-derived PA is also important for BMV RNA replication . However , exogenously added PA did not affect the accumulation of BMV RNA ( Fig 5D ) as similarly seen for RCNMV RNA2 . PA acts as a second messenger in signal transduction during multiple biotic and abiotic stress responses and plays multiple roles including that for transcriptional reprogramming [20–22] . To investigate whether PA has a direct role in viral RNA replication , we took advantage of BYL , a nucleus-depleted ( therefore , the effects of PA on transcriptional reprogramming are negligible ) in vitro translation/replication system that has been used successfully to recapitulate the negative-strand RNA synthesis of RCNMV [12 , 44–49] . Addition of PA into BYL stimulated the accumulation of newly synthesized negative-strand RNA ( Fig 6A ) , and moderately enhanced the accumulation of the 480-kDa replicase complex ( Fig 6B ) , suggesting that PA stimulates the activity and/or assembly of the viral replicase complex in a direct manner . The stimulation effects of PA on the accumulation of the 480-kDa replicase complex was less obvious than that on the accumulation of newly-synthesized viral RNA at the highest concentration of PA used in this experiment ( 50 μM ) . This result may also support a direct role for PA in the enhancement of RdRP activity rather than the formation of the replicase complex . Because PA directly stimulated the viral negative-strand RNA synthesis in BYL , we hypothesized that p27 , the multifunctional RCNMV replication protein , has an affinity for PA . To investigate this possibility , we conducted a lipid overlay assay using bacterially expressed , purified C-terminally FLAG-tagged p27 ( p27-FLAG ) [49] . Purified p27-FLAG protein was incubated with phospholipid-spotted nitrocellulose membranes , and the interaction between p27-FLAG and phospholipids was detected using anti-FLAG antibody . p27 gave strong PA binding signals on the blot ( Fig 7A and 7B ) . p27 also exhibited weak binding to phosphatidylinositol-4-phosphate ( PI4P ) , phosphatidylinositol ( 3 , 5 ) -bisphosphate , phosphatidylinositol ( 4 , 5 ) -bisphosphate , and phosphatidylinositol ( 3 , 4 , 5 ) -trisphosphate , but negligible binding to other lipids , including PC and PE ( Fig 7A and 7B ) . These results were consistent with the findings that neither exogenously supplied PC nor PE promoted RCNMV replication in N . benthamiana protoplasts ( S8 Fig ) . N- or C-terminal halves of p27 fragments did not give PA binding signals ( Fig 7C and 7D ) , suggesting that overall protein conformation may be important for p27–PA binding . Next we investigated whether RCNMV infection affects the amount of PA in plant leaves . N . benthamiana leaves were inoculated with RCNMV via agroinfiltration . At 2 dai , lipids were extracted and the amount of PA was analyzed by thin layer chromatography ( TLC ) . Compared with control plant leaves infiltrated with Agrobacterium harboring an empty vector , the signal intensity of a lipid spot that showed a migration similar to that of soy-PA was increased in RCNMV-infected plant leaves ( about 3-fold higher ) ( Fig 8A and 8B ) . To identify the lipid species of the spot , the same samples were again subjected to TLC , and the spot was scratched out from Coomassie Brilliant Blue R-250 stained TLC plates and subjected to LC/MS analysis . The lipid was identified as PA ( Fig 8C–8E ) . We concluded that RCNMV infection upregulated PA accumulation in plants . It is known that expressions of PLD genes were induced by pathogen infection [50] . Therefore , the enhanced accumulation of PA in RCNMV-infected plants could be due to elevated accumulation of PLD through induction of PLD gene expression . To examine this possibility , we investigated whether mRNA levels of NbPLDα and NbPLDβ were upregulated in RCNMV-infected plants by RT-qPCR analysis . The accumulation levels of NbPLDα and NbPLDβ transcripts in RCNMV-infected plants were about 1 . 2- and 1 . 9-fold higher , respectively , than those in control plants , although the increase in NbPLDα transcripts was insignificant by the Student’s t-test ( S9 Fig ) . We compared the accumulation of endogenous PA in NbPLDα or NbPLDβ knockdown plants with that in TRV control plants by TLC analysis . As expected from their predicted function , NbPLDα- or NbPLDβ-knockdown plants showed reduced accumulation of PA compared with TRV control plants ( S10 Fig ) , suggesting that these PLDs contribute to PA production in N . benthamiana . These results further supported the idea that RCNMV-induced PLDs-derived PA plays an important role in RCNMV replication . A growing number of studies have suggested that PLD and PLD-derived PA play vital roles in environmental responses in plants [19–22 , 50] . The properties of PA ( i . e . , normally present in small amounts , and rapidly and transiently accumulates in response to various environmental cues ) seem to be suitable for its function in biotic and abiotic responses in which plants need to rapidly accommodate their surrounding environments . Indeed , PA has been shown to accumulate in response to several MAMPs and pathogen effector proteins , or SA that is a key hormone involved in plant resistance against biotrophic pathogens [23–29] . PA induces PR gene expression and cell death [28 , 33 , 34] , and has been proposed to act as an important component in resistance to biotrophic pathogens such as tobacco mosaic virus and Phytophthora parasitica . Plants have diverse numbers of PLD isoforms and they appear to have distinct but somewhat overlapping functions in cellular responses [50] . In Arabidopsis , it is proposed that multiple PLD isoforms cooperatively contribute to AvrRpm1-triggered resistance [30] . In rice , PLDβ1 acts like a negative regulator of defense signaling because PLDβ1-knockdown rice plants exhibit constitutive ROS production , expression of PR genes , and enhanced resistance against pathogens [35] . In this study , we showed that PLD and PA are essential for and play a key role in RCNMV replication . Poor infection of RCNMV to NbPLDα- or NbPLDβ-knockdown plants was not due to constitutively activated defense responses , indicating that both NbPLDα and NbPLDβ act as essential host factors in RCNMV replication . The findings reveal novel aspects of PLD and PA in their roles during biotic stress responses in plants . The replication of Tomato bushy stunt virus is enhanced by deletion of the PAH1 gene encoding a PA phosphatase , which converts PA into DAG in yeast [51] . Moreover , ectopic expression of Arabidopsis PA phosphatase , Pah2 in N . benthamiana results in the inhibition of tombusviruses and RCNMV infection [51] . These findings suggest that PA is positively involved in the life cycles of these viruses . However , whether viruses manipulate PA production for viral replication has been unknown . In the current study , we showed that RCNMV replication proteins interacted with PA-producing enzymes , NbPLDα and NbPLDβ . RCNMV infection induced a high accumulation of PA in plant tissues , suggesting that RCNMV alters cellular lipid metabolism to establish a suitable environment for viral replication . It is known that transcription of PLD genes is induced by pathogen infection [50] . RCNMV infection increased the accumulation of NbPLDα and NbPLDβ transcripts ( S9 Fig ) . Currently , whether the enhancement of PA accumulation observed in RCNMV-infected plants is due to the upregulation of PLD gene expression remains unknown . PLDs may be activated through direct or indirect interaction with RCNMV replication proteins . Although we failed to show the interaction between p88pol and NbPLDα or NbPLDβ in vivo because p88pol accumulates below detection limits in the absence of viral RNA replication in N . benthamiana [12] , p88pol interacted with PLDs , at least with NbPLDα in a co-IP assay in BYL . The interaction of p88pol with PLDs seems to make sense for viral replication strategy because it brings them to the sites of replication . This strategy could increase PA only at VRCs and not at other cellular membranes where PA might affect cellular metabolism or activate the SA-mediated defense responses that is detrimental for successful viral infection . This is critical for the viral life cycle because PA interacts with p27 auxiliary replication protein ( Fig 7 ) and enhances viral replication through upregulating the activity and/or assembly of the 480-kDa replicase complex ( Fig 6 ) . To our knowledge , this is the first demonstration of a functional role for PA in ( + ) RNA virus replication . It is likely that PA is involved in RNA replication of many ( + ) RNA viruses . Indeed , the replication proteins , 1a and 2apol of BMV , a virus unrelated to RCNMV , also interacted with NbPLDβ , and BMV RNA replication was sensitive to n-butanol , which is an inhibitor of PLDs-derived PA production ( Fig 5 ) . However , exogenously added PA did not affect the accumulation of BMV RNA ( Fig 5 ) , implying that the threshold of PA requirement for BMV RNA replication seems to be lower than that for RCNMV . The differential PA requirement may be explained by very weak affinity of BMV replication proteins with NbPLDα , which is more active in PA production than NbPLDβ ( S10 Fig ) . Moreover , Dengue virus induces the accumulation of several lipids in infected mosquito cells , including PA [52] . It has been reported that Coxsackievirus B3 and mouse hepatitis coronavirus replication is insensitive to n-butanol [53 , 54] . However , because PA can also be formed through the combined action of phospholipase C and DAG kinase [55] , it is unknown whether replication of these viruses depends on PA or not . How does PA affect viral RNA replication ? PA binding to proteins modulates the catalytic activity of target proteins , tethers proteins to the membranes , and promotes the formation and/or stability of protein complexes [55] . We found that exogenous PA enhanced the accumulation of newly synthesized viral RNA and the formation of 480-kDa replicase complexes in BYL in vitro translation/replication systems , and that p27 had affinity for PA in vitro ( Figs 6 and 7 ) . Therefore , PA could promote viral replicase activity and/or assembly directly . The 480-kDa replicase complexes contain p27 oligomer . Therefore , it is possible that PA-binding to p27 in the replicase complexes assists the conformational change of p27 that is suitable for RNA synthesis . Alternatively , PA could serve as an assembly platform for host PA-binding proteins . PA binds to various proteins , including transcription factors , kinases , phosphatases , enzymes involved in central metabolism , and proteins involved in vesicular trafficking and cytoskeletal rearrangements [20 , 21] . Several known cellular PA-binding proteins were also identified in our LC/MS/MS analysis ( S1 Table ) . These include NADPH oxidase [56] , glyceraldehyde-3-phosphate dehydrogenase ( GAPDH ) [57 , 58] , and SNF1-related kinase [58] . RCNMV-induced accumulation of high PA levels may facilitate the recruitment of these PA-binding proteins to viral replication sites . One of the candidate proteins is GAPDH-A . Although GAPDH-A is not required for RCNMV replication , it recruits RCNMV movement protein to viral replication sites and plays an important role in virus cell-to-cell movement [59] . Therefore , PA may also function in bridging viral replication and cell-to-cell movement . RCNMV induces large ER aggregates in infected cells , which are thought to be viral RNA replication factories [13–16] . PA may play a direct role in the formation of RNA replication factories . In support of this hypothesis , deletion of pah1 in yeast causes the expanded ER membranes and leads to enhanced TBSV replication on these membranes , although TBSV replicates normally at peroxisomes [51] . In addition , it is also possible that PA might affect unknown cellular factors that are involved in viral RNA replication . Further studies are needed to elucidate the molecular functions of PA in viral RNA replication . Our results suggest that RNA1 and RNA2 have differential PA requirements in RNA replication . How does this difference contribute to RCNMV infection ? RNA1 requires a larger amount of PA for maximum efficiency of its replication than that required by RNA2 . This may reflect differences in the translation and replication mechanisms between RNA1 and RNA2 . Translation of MP from RNA2 couples with RNA replication: only progeny RNA2 generated de novo through the RNA replication pathway could function as mRNA [60] . Poor enhancement of RNA2 replication by PA may be beneficial for switching from replication to translation . By contrast , because RNA1 has a cap-independent translation enhancer that is an effective recruiter of translation factors [61] , newly synthesized RNA1 serves as a template for further translation of the replication proteins that would activate PA production in infected cells . Moreover , RNA1 also serves as a template for transcription of subgenomic RNA , which encodes CP . Therefore , PA-mediated enhancement of RNA1 replication may be suitable for the production of CP subgenomic RNA during the late stage of viral infection . High PA requirement for maximum RNA1 replication may explain the use of both NbPLDα and NbPLDβ as essential host factors . A growing number of studies have suggested that multiple lipid species affect virus replication and that ( + ) RNA viruses employ a multifaceted strategy to rewire host machinery involved in lipid transport and synthesis [62] . HCV infection stimulates the production of cellular PC , PE , and PI4P and HCV co-opts PI4KIIIα for replication [63 , 64] . In addition , HCV infection stimulates the accumulation of cellular sphingomyelin , which binds to and activates HCV NS5B polymerase [65 , 66] . Enterovirus utilizes PI4KIIIβ for RNA replication and viral RdRP 3Dpol binds to PI4P [67] . Enteroviruses also upregulate cellular uptake of fatty acids , which are channeled toward highly upregulated PC synthesis in infected cells [68] . The elevated PC has been proposed to serve as a building block for the formations of the viral replication factory [68 , 69] . Recently , it has been shown that PE and PC stimulate TBSV RdRP activity in vitro [70] . RCNMV p27 showed affinity , not only for PA but also for other phospholipids such as PI4P in vitro . However , our LC/MS/MS analysis failed to detect any PI4P-producing enzymes , such as PI4K . Combined lipidomics , proteomics , and transcriptome analysis will be helpful for a comprehensive understanding of lipid species involved in viral RNA replication . Plasmids given the prefix ‘‘pBIC” were used for Agrobacterium infiltration , ‘‘pUC” , ‘‘pRC” and ‘‘pR” were used for in vitro transcription , ‘‘pCold” was used for protein expression in Escherichia coli . pUCR1 [71] and pRC2_G [72] are full-length cDNA clones of RNA1 and RNA2 of the RCNMV Australian strain , respectively . pB1TP3 , pB2TP5 , and pB3TP8 are full-length cDNA clones of RNA1 , RNA2 , and RNA3 of the BMV M1 strain , respectively [73] ( generous gift from Paul Ahlquist ) . The constructs described previously used in this study include pBICp27 [71] , pBICp88 [71] , pBICR2 [71] , pBICR1R2 [71] , pBICp19 [71] , pCOLDIp27-FLAG [49] , pCOLDIp27N-FLAG [49] , and pCOLDIp27C-FLAG [49] . pUC118 was purchased from Takara Bio Inc . ( Shiga , Japan ) . Escherichia coli DH5α was used for the construction of all plasmids . All plasmids constructed in this study were verified by sequencing . RNA extraction from Nicotiana benthamiana leaves was performed using an RNeasy Plant Mini Kit ( Qiagen , Hilden , Germany ) . Reverse-transcription-PCR ( RT-PCR ) was catalyzed by Superscript III reverse transcriptase ( Invitrogen ) using oligo ( dT ) [16] . Primers to amplify coding sequences of NbPLDα or NbPLDβ were designed based on the N . benthamiana RNA seq data ( Transcriptome version 5: http://sydney . edu . au/science/molecular_bioscience/sites/benthamiana/ ) [74] . N . benthamiana plants were grown on commercial soil ( Tsuchi-Taro , Sumirin-Nosan-Kogyo Co . Ltd . ) at 25 ± 2°C and 16 h illumination per day . RCNMV RNA1 and RNA2 were transcribed from SmaI-linearized pUCR1 and pRC2_G , respectively , using T7 RNA polymerase ( TaKaRa Bio , Inc ) . BMV RNAs were transcribed from EcoRI-linearized pB plasmids using T7 RNA polymerase and capped with a ScriptCapm7G capping system ( Epicentre Biotechnology ) . Capped mRNAs were transcribed from NotI-linearized pBYL plasmids using T7 or SP6 RNA polymerase ( TaKaRa Bio , Inc ) and capped with a ScriptCapm7G capping system ( Epicentre Biotechnology ) . All transcripts were purified with a Sephadex G-50 fine column ( Amersham Pharmacia Biotech ) . RNA concentration was determined spectrophotometrically , and its integrity was verified by agarose gel electrophoresis . Four-week-old N . benthamiana plants were agroinfiltrated as described previously [15] . At 2 days postinfiltration ( dpi ) , total proteins were extracted from 6 g of leaves in 10 ml of buffer A ( 50 mM HEPES , 150 mM NaCl , 0 . 1% 2-mercaptoethanol , 0 . 5% Triton X-100 , 5% glycerol , pH 7 . 5 ) containing 30 mM imidazole and a cocktail of protease inhibitors ( Roche ) . Following the removal of cell debris by filtering the mixture through cheesecloth , the extract was centrifuged at 21 , 000g at 4°C for 10 min and the supernatant was mixed with Ni-NTA beads ( 400 μl ) ( Qiagen , Hilden , Germany ) and incubated for 1 h at 4°C with gentle mixing . The beads were washed three times with 1 ml of buffer A containing 100 mM imidazole . The bound proteins were eluted with 1 ml of buffer A containing 500 mM imidazole . The eluted proteins were mixed with 50 μl of ANTI-FLAG M2-Agarose Affinity Gel ( Sigma-Aldrich ) and incubated for overnight at 4°C with gentle mixing . The beads were washed 3 times with 1 ml of buffer A . The bound proteins were eluted with 300 μl of buffer A containing 150 μg/ml 3 × FLAG peptides ( Sigma-Aldrich ) . The eluted proteins were concentrated by acetone precipitation and dissolved in 1 × NuPAGE sample buffer ( Invitrogen ) . The purified proteins were separated by sodium dodecyl sulfate ( SDS ) -PAGE ( NuPAGE 3%–12% bis-Tris gel: Invitrogen ) and visualized by silver staining ( Wako , Osaka , Japan ) . Proteins in excised gel pieces were subjected to digestion with trypsin , LC–MS/MS analysis , and MASCOT searching as described previously [12] . Appropriate combinations of silencing vectors were expressed via Agrobacterium infiltration in 3- to 4-week-old N . benthamiana plants as described previously [16] . At 18 dpi , the leaves located above the infiltrated leaves were inoculated with in vitro transcribed RNA1 and RNA2 ( 500 ng each ) . At 2 days after inoculation , three inoculated leaves from three different plants infected with the same inoculum were pooled , and total RNA was extracted using RNA extraction reagent ( Invitrogen ) , treated with RQ1 RNase-free DNase ( Promega , Madison , WI ) , purified by phenol–chloroform and chloroform extractions , and precipitated with ethanol . Viral RNAs were detected by northern blotting , as described previously [16] . The mRNA levels of NbPLDα and NbPLDß were examined by RT-PCR using primer pairs 5′ -TATCAAGGTAGAGGAGATAGGTGC-3′ and 5′-TACATCATCTCCATCGTTCTCCTC-3′ , and 5′-GAAGGCTTCAAAGCGCCATG-3′ and 5′-CTTAGGCAAGGGACATCAGC-3′ , respectively . As a control to show the equal amounts of cDNA templates in each reaction mixture , the ribulose 1 , 5-biphosphate carboxylase small subunit gene ( RbcS ) , a gene that is constitutively expressed , was amplified by RT-PCR as described previously [16] . N . benthamiana protoplasts were prepared according to Kaido et al . ( 2014 ) [59] . N . benthamiana protoplasts were inoculated with RCNMV RNA1 ( 1 . 5 μg ) and RNA2 ( 0 . 5 μg ) and incubated with various concentrations of n-butanol ( Sigma-Aldrich ) , tert-butanol ( Sigma-Aldrich ) , phosphatidic acid ( PA ) ( Soy-derived; Avanti Polar Lipid ) , phosphatidyl choline ( PC ) ( Soy-derived; Avanti Polar Lipid ) or phosphatidyl ethanolamine ( PE ) ( Soy-derived; Avanti Polar Lipid ) at 20°C for 18 h . Phospholipids were dissolved in dimethylsulfoxide . Total RNA was extracted and subjected to northern blotting , as described previously [15] . Each experiment was repeated at least three times using different batches of protoplasts . The preparation of BYL was as described previously [38 , 46] . The BYL translation/replication assay was performed essentially as described previously [46] . Briefly , 300 ng of RNA1 was added to 30 μL of BYL translation/replication mixture in the presence of various concentrations of PA . The mixture was incubated at 17°C for 240 min . Aliquots of the reaction mixture were subjected to northern and immunoblotting analyses , as described previously [45 , 46 , 48] . Four-week-old N . benthamiana plants were agroinfiltrated as described previously [15] . At 4 days postinfiltration ( dpi ) , total proteins were extracted from 0 . 33 g of leaves in 1 ml of buffer A containing a cocktail of protease inhibitors ( Roche ) . Following the removal of cell debris by centrifugation at 21 , 000g at 4°C for 10 min , the supernatant was mixed with GFP-Trap agarose beads ( 10 μl ) ( ChromoTek ) and incubated for 1 h at 4°C with gentle mixing . The beads were washed 3 times with 1 ml of buffer A . The bound proteins were eluted by addition of 1 × SDS gel loading buffer , followed by incubation for 3 min at 95°C . Protein samples were subjected to SDS-PAGE , followed by immunoblotting with appropriate antibodies . FLAG- or HA-tagged proteins were expressed in BYL by adding an in vitro transcript . After incubation at 25°C for 120 min , a 10-μl bed volume of anti-HA Affinity Matrix ( Roche ) or ANTI-FLAG M2-Agarose Affinity Gel ( Sigma-Aldrich ) was added to the BYL and further incubated for 60 min with gentle mixing . The resin was washed three times with 200 μl of TR buffer [38] supplemented with 150 mM NaCl and 0 . 5% Triton X-100 . The bound proteins were eluted by addition of 1 × SDS gel loading buffer , followed by incubation for 3 min at 95°C . Protein samples were subjected to SDS-PAGE , followed by immunoblotting with appropriate antibodies . Appropriate combinations of fluorescent protein-fused proteins were expressed in N . benthamiana leaves by Agrobacterium infiltration . Fluorescence of GFP and mCherry was visualized with confocal microscopy at 4 dai as described previously [59] . Protein expression in E . coli BL21 ( DE3 ) and subsequent purification were done as described previously [15] . The concentration of purified protein was measured using a Coomassie Protein Assay Kit ( Thermo Fisher Scientific , Waltham , MA ) . The purified protein was subjected to SDS-PAGE and visualized with Coomassie brilliant blue R-250 to check its purity . Lipid overlay assays were conducted as recommended by the manufacture’s protocol . Briefly , the membrane ( PIP Strips or Membrane Lipid Arrays; Echelon Bioscience Inc ) was incubated in 3% fatty acid free BSA ( Sigma-Aldrich ) in a mixture of phosphate-buffered saline and 0 . 1% Tween 20 ( PBST ) for 1 h at room temperature ( RT ) and then incubated in the same solution containing 500 ng of purified recombinant protein for 1 h at RT . After washing three times with PBST , the membrane was incubated with a mouse anti-FLAG antibody ( 1:10000; Sigma-Aldrich ) for 1 h at RT , followed by three washes with PBST . An anti-mouse IgG conjugated with horseradish peroxidase ( 1:10000; KPL ) was used as a secondary antibody . Binding of proteins to phospholipids was visualized by incubation with a chemiluminescent substrate . Four-week-old N . benthamiana plants were inoculated with RCNMV via agroinfiltration . At 2 dai , 0 . 33 g of infiltrated leaves were ground in liquid nitrogen and extracted in 900 μl of water . Total lipids were extracted by adding 3 ml CHCl3/CH3OH ( 2:1 , v/v ) to each sample . The samples were vortexed and centrifuged at 1690g , at 4°C for 10 min . The organic phase was recovered and dried under nitrogen gas stream . Lipids were dissolved in 100 μl CHCl3/CH3OH ( 2:1 , v/v ) . Subsequently , 5 μl of the samples were analyzed on TLC plates ( Merck , Germany ) . The chromatography was performed using CHCl3/CH3OH/formic acid/acetic acid ( 12:6:0 . 6:0 . 4 , v/v ) . Plates were air-dried , soaked in 10% CuSO4 , and charred at 180°C for 10 min to visualize lipids . To identify lipid species , the air-dried TLC plates were stained for an hour in a 0 . 03% Coomassie Brilliant Blue R-250 solution containing 20% of methanol and 0 . 5% of acetic acid . Destaining of the plates was performed with 20% methanol containing 0 . 5% acetic acid for 5 min . After drying the plates for a few minutes , the blue bands of interest were scratched out and transferred to glass tubes . The scratched silica gels were mixed with chloroform/methanol ( 3/7 , v/v ) , followed by the centrifugation at 1 , 690g , at 4°C for 10 min . The supernatants were subjected to the LC-MS analysis using LCMS-IT-TOF mass spectrometer ( Shimadzu , Kyoto , Japan ) . A TSK gel ODS-100Z column ( 2 . 0 × 150 mm , 5 μm , Tosoh , Tokyo , Japan ) was eluted isocratically with acetonitrile/methanol/2-propanol/water ( 6/131/110/3 , by volume ) containing 19 . 6 mM of ammonium formate and 0 . 2% of formic acid at a flow rate of 0 . 2 mL/min . The MS was performed using an electrospray ionization interface operated in negative ion mode , under the following conditions: CDL temperature , 200°C; block heater temperature , 200°C; nebulizing gas ( N2 ) flow , 1 . 5 L/min . The MS data were acquired in the range of m/z 600 to 1 , 000 using 10 msec ion accumulation time . The MS2 data were acquired in the range of m/z 125 to 500 , using 50 msec ion accumulation time and automatic precursor ion selection in the range of m/z 650 to 750 . CID parameters were follows: energy , 50%; collision gas ( argon ) 50% . Total RNA extracted from N . benthamina leaves or protoplasts were subjected to reverse transcription using PrimeScript RT reagent Kit ( Takara ) using oligo-dT and random primers according to manufacturer’s protocol . Real-time PCR was carried out using SYBR Premix Ex Taq ( RR420A , Takara ) . Primers used for real-time PCR analysis were listed in S2 Table . Quantitative analysis of each mRNA was performed using a Thermal cycler Dice Real Time System TP800 ( Takara ) . NbPLDα and NbPLDβ were registered through DDBJ and accession numbers LC033851 and LC033852 , respectively , were given on March 11 2015 .	All characterized eukaryotic positive-strand RNA [ ( + ) RNA] viruses replicate their genomes using the viral replication complexes ( VRCs ) , which contain multiple viral and host components , on intracellular membranes . Phospholipids are major constituents of cellular membranes; however , the function ( s ) of phospholipids in genome replication of ( + ) RNA viruses remains largely unknown . Here , we show that Red clover necrotic mosaic virus ( RCNMV ) , a plant ( + ) RNA virus , induces a high accumulation of phosphatidic acid ( PA ) in infected plant leaves . PA-producing enzymes , phospholipase Dα ( PLDα ) and PLDβ , are associated with RCNMV VRCs . PA interacts with the viral replication protein and enhances the viral replication by upregulating the activity/assembly of the VRCs in vitro . In summary , RCNMV alters cellular lipid metabolism via PLD to establish a suitable environment for viral replication .	[ "Abstract", "Introduction", "Results", "Discussion", "Materials", "and", "Methods" ]	[]	2015	Phosphatidic Acid Produced by Phospholipase D Promotes RNA Replication of a Plant RNA Virus
During early meiotic prophase , homologous chromosomes are connected along their entire lengths by a proteinaceous tripartite structure known as the synaptonemal complex ( SC ) . Although the components that comprise the SC are predominantly studied in this canonical ribbon-like structure , they can also polymerize into repeated structures known as polycomplexes . We find that in Drosophila oocytes , the ability of SC components to assemble into canonical tripartite SC requires the E3 ubiquitin ligase Seven in absentia ( Sina ) . In sina mutant oocytes , SC components assemble into large rod-like polycomplexes instead of proper SC . Thus , the wild-type Sina protein inhibits the polymerization of SC components , including those of the lateral element , into polycomplexes . These polycomplexes persist into meiotic stages when canonical SC has been disassembled , indicating that Sina also plays a role in controlling SC disassembly . Polycomplexes induced by loss-of-function sina mutations associate with centromeres , sites of double-strand breaks , and cohesins . Perhaps as a consequence of these associations , centromere clustering is defective and crossing over is reduced . These results suggest that while features of the polycomplexes can be recognized as SC by other components of the meiotic nucleus , polycomplexes nonetheless fail to execute core functions of canonical SC . The faithful segregation of chromosomes away from their homologs at the first meiotic division is the physical basis for Mendelian inheritance . In most organisms , chromosome segregation is achieved by recombination between paired homologs , a process that results in chiasmata and in the production of gametes bearing recombined chromosomes . Both the maintenance of homolog pairing and crossing over depend on the production of double-strand breaks ( DSBs ) in the context of the synaptonemal complex ( SC ) . Meiotic DSBs are catalyzed by SPO11 homologs , and a subset of these DSBs are then converted into crossovers only in the presence of the SC [1 , 2] . By maturing into chiasmata , these crossovers serve to physically interlock homologs at metaphase I and thus ensure segregation . In Drosophila melanogaster females , as well as many other organisms , the SC assembles along each pair of homologous chromosomes during early pachytene [3 , 4 , 5] . Proper SC formation is required not only for chromosome synapsis , but also for the maturation of DSBs into crossovers [6 , 7 , 8 , 9] . The SC also plays a pivotal role in the clustering of centromeres in Drosophila females [10 , 11] . As analyzed by electron microscopy ( EM ) , the SC has a tripartite zipper-like structure that is highly conserved across many organisms [5 , 12 , 13 , 14] . Electron dense lines , referred to as the lateral elements ( LEs ) , associate along each of the homologous chromosomes . Lateral element components include the cohesin and cohesin-related proteins that connect the chromatin to the rest of the SC ( Fig 1A ) . The structure between the LEs is called the central region , and the electron-dense line that runs down the center of the central region is referred to as the central element . While the amino acid sequences of SC proteins evolve rapidly between even closely related species [15] , the basic tripartite structure of the SC is maintained from yeast to humans , indicating the SC’s structure is crucial for its function in meiosis [5] . A schematic of the current structural model of the Drosophila SC is shown in Fig 1A . The transverse filament protein Crossover Suppressor on 3 of Gowen ( C ( 3 ) G ) spans the distance between LEs [6] . C ( 3 ) G is thought to form homodimers with their C-terminal ends at the LE and their N-terminal ends at the central element [6 , 17] . The Corona ( Cona ) and Corolla proteins appear to help stabilize and/or assemble C ( 3 ) G since the loss of any of the three proteins leads to a loss of SC assembly [7 , 8] . SC components are assembled and disassembled in a highly controlled fashion ( Fig 1B ) . However , the process by which they are loaded and unloaded from the chromosomes in the Drosophila ovary is poorly understood . Recent work in other organisms has demonstrated the importance of post-translational modifications in regulating the timing and pattern of SC assembly and disassembly [5] . These modifications , including phosphorylation , acetylation and sumoylation can be on the SC components themselves as well as on regulators of SC assembly [5 , 18 , 19 , 20 , 21] . When assembly of central region components along the chromosomes is disrupted ( for example , when LE proteins are absent or when excess free SC components are present ) , aberrant SC-like structures called polycomplexes can form [22 , 23 , 24 , 25 , 26 , 27] . Polycomplexes have been observed in many organisms , and by EM they appear to be repeating layers of SC [12 , 22 , 23 , 26 , 28] . Immunofluorescence studies of these structures in yeast , worms and flies have demonstrated that the polycomplexes contain central region proteins [12 , 21 , 22 , 24 , 29 , 30 , 31 , 32] . Polycomplex-like structures have been observed in cultured mammalian somatic cells upon expression of only the major transverse filament protein , SYCP1 [33] . Polycomplexes have been observed both in the cytoplasm and within the nucleus , and many of the better-characterized examples of polycomplexes are not associated with chromatin [22 , 23 , 24] . The mechanisms that ensure SC proteins assemble between homologous chromosomes and thus block self-assembly into polycomplexes are not well understood in many organisms [23 , 27] . We show here that mutations in the gene encoding the E3 ubiquitin ligase seven in absentia ( sina ) result in the aberrant formation of numerous and large SC polycomplexes in D . melanogaster females , demonstrating that Sina is required to promote normal SC assembly ( perhaps by blocking PC formation ) . These polycomplexes can persist into meiotic stages beyond those at which SC has been fully disassembled in wild-type oocytes , suggesting that normal disassembly of the SC also requires a functional Sina protein . While these polycomplexes can associate with chromatin at both centromeres and at sites of DSBs , they fail to maintain centromere clustering or promote wild-type levels of crossover formation . Intriguingly , lateral element and cohesin proteins appear to be recruited to many of the sina mutant-induced polycomplexes . Previous work in the Drosophila eye as well as numerous studies of homologs in other organisms indicate Sina and its homologs function as E3 ubiquitin ligases that target proteins for degradation [34 , 35 , 36 , 37 , 38] . Therefore , it seems likely that Sina regulates the destruction of a protein crucial for directing proper SC assembly and maintenance along the arms of the chromosomes and that loss of functional Sina protein leads to uncontrolled assembly of SC components into polycomplex structures . In somatic tissues , Sina acts as an E3 ubiquitin ligase that targets specific proteins for degradation [34 , 35 , 36 , 37] . For example , in the Drosophila eye , Sina has been shown to be required for degradation of the transcription factor Tramtrack [36] . Null alleles of sina are reported to cause numerous phenotypes , including lethargy , short lifespan , eyes with missing R7 photoreceptors , and male and female sterility [39] . However , a role for sina in meiosis has not been reported . A forward genetic screen in our laboratory for mutants with elevated levels of X chromosome meiotic nondisjunction produced a mutant , termed A4 , which exhibited elevated meiotic nondisjunction . Deficiency mapping and sequencing identified A4 as a new allele of sina , hereafter called sinaA4 . The sinaA4 mutation changes the highly conserved amino acid at position 270 from an alanine to a valine in the predicted Sina protein ( Fig 1C ) . Unlike null alleles of sina , which are female-sterile [39] , sinaA4 females ( as well as sinaA4/sinaDf females ) are semi-fertile , demonstrating that sinaA4 is a hypomorphic mutation ( S1A Fig ) . Additionally , both sinaA4 and sinaA4/sinaDf flies survive the 10 days necessary for genetic assays while sina null flies are lethargic and short-lived [39] . The increase in fertility and hardiness of sinaA4 females compared to sina null females makes this a more tractable background in which to study sina’s meiotic roles . In nondisjunction assays , sinaA4 mutant females carrying normal-sequence X chromosomes displayed 13 . 4% X and 6 . 9% 4th chromosome nondisjunction , compared to 0 . 4% X and 0 . 2% 4th chromosome nondisjunction in control flies ( Table 1 ) . Since sinaA4 is a hypomorphic allele and semi-fertile in combination with sinaDf ( S1A Fig ) , we next analyzed the amount of chromosome nondisjunction in sinaA4/sinaDf females ( Table 1 ) . sinaA4/sinaDf females showed a more severe phenotype , with X and 4th chromosome nondisjunction levels increasing to 47 . 4% and 28 . 2% , respectively . This level of X chromosome nondisjunction indicates the X chromosomes are segregating nearly at random . An N-terminal FLAG-tagged overexpression wild-type sina construct ( FLAGsinaWT ) rescued the X chromosome nondisjunction phenotype , as well as the defect in egg hatch rate from sinaA4/sinaDf females , confirming that sina is indeed our gene of interest ( S1A and S1B Fig ) . To elucidate the cause of the chromosome nondisjunction in sinaA4 and sinaA4/sinaDf mutants , we examined the early steps of meiosis . Because similar high levels of chromosome nondisjunction are also observed in females mutant for SC components [6 , 7 , 8 , 9] , we wondered whether SC formation was defective in sina mutants . In wild-type Drosophila females , central region components of the SC are first observed to load near the centromeres during the premeiotic mitotic divisions that produce the 16-cell interconnected cyst [41] . As the cyst enters meiosis in early pachytene ( region 2A ) of the germarium , the central region proteins rapidly load along the chromosome arms in up to four cells of the 16-cell cyst ( Fig 1B ) , appearing as curved , ribbon-like tracks . As the cyst matures and moves through the germarium , the SC disassembles from three of the nuclei , leaving only the pro-oocyte nucleus with full-length SC by mid-pachytene ( region 3 ) ( reviewed in [3] ) . Full-length SC is maintained until approximately stage 5 , when the SC along the chromosome arms begins to progressively disassemble . At approximately stage 7–8 , SC components remain only at the centromeres . Deconvolution immunofluorescence microscopy using antibodies recognizing the central region proteins C ( 3 ) G and Corolla in sinaA4 and sinaA4/sinaDf ovaries revealed the presence of aberrant SC ( Fig 2 ) . In wild type , the SC forms as curved tracks between the chromosomes in early pachytene ( region 2A ) and the tracks of SC are present in the oocyte nucleus in mid-pachytene ( region 3 ) ( Fig 2A and 2B and S2A Fig ) . In sinaA4 ovaries , the SC began to assemble with relatively normal timing , exhibiting tracks of SC in multiple nuclei in early pachytene ( region 2A ) ( Fig 2C and S2B Fig ) . Full length SC tracks were observed in 46 . 5% ( 20/43 ) of sinaA4 nuclei in early pachytene ( region 2A ) . As the cysts progressed through the germarium , the SC lost its curved , track-like pattern and SC components began to form narrow rod-like structures ( Table 2 , Fig 2D and S2 Fig ) . In early pachytene ( region 2A ) 18 . 6% ( 8/43 ) of the nuclei displayed a combination of tracks and rod-like structures and 34 . 9% ( 15/43 ) of the nuclei contained only these rod-like structures . By early/mid-pachytene ( region 2B ) and mid-pachytene ( region 3 ) , 72 . 2% ( 13/18 ) and 87 . 5% ( 14/16 ) of nuclei respectively , had no clear SC tracks visible among the rod-like structures , and the remaining nuclei had partial pieces of track-like SC and polycomplexes ( Fig 2D and S2B Fig ) . As the cysts budded from the germaria into the vitellarium at mid-prophase ( stages 2–9 ) , sinaA4 oocyte nuclei retained the abnormal SC structures in 85 . 7% ( 12/14 ) of nuclei with 5 nuclei also containing fragments of track-like SC . In the remaining two nuclei only fragmented SC tracks were observed . Consistent with the stronger meiotic nondisjunction phenotype , sinaA4/sinaDf females showed an exacerbated SC phenotype ( Fig 2E and 2F ) , including fewer SC tracks and earlier polycomplex formation . We observed very small rod-like SC structures in the premeiotic region 1 where only foci of SC components are observed in wild type ( S2D and S2E Fig ) . By early pachytene ( region 2A ) , the SC had already assembled into aberrant structures in sinaA4/sinaDf ovaries with 100% ( 29/29 ) of nuclei containing polycomplexes ( Fig 2E ) . In early pachytene ( region 2A ) 58 . 6% ( 17/ 29 ) of the nuclei had SC tracks of various lengths along with the polycomplexes . The number of nuclei with only polycomplexes then increased by mid-pachytene ( region 3 ) ( Fig 2F ) , but a least one small SC track-like structure could still be observed with the polycomplexes in 31 . 8% ( 7/22 ) of early/mid-pachytene ( region 2B ) , 23 . 1% ( 3/13 ) mid-pachytene ( region 3 ) , and 20 . 0% ( 5/25 ) mid-prophase ( stages 2–9 ) nuclei . It is important to note that sinaA4 homozygotes had a less severe defect than sinaA4/sinaDf females , with the presence of nuclei with long curved SC tracks in early pachytene ( region 2A ) and less severe chromosome nondisjunction ( Table 1 ) , which supports our earlier assessment that the sinaA4 mutation is a hypomorphic allele . For reasons more fully described below , we have concluded that these aberrant SC structures are best classified as polycomplexes . In early pachytene ( region 2A ) sinaA4 ovaries had an average of 7 . 7 polycomplexes per nucleus , but the number ranged from 1–15 polycomplexes ( Table 2 ) . A similar average number of polycomplexes could be observed in sinaA4 ovaries through mid-pachytene ( region 3 ) , but then the average number decreased to 5 . 6 polycomplexes in mid-prophase ( stage 2–9 ) oocytes ( Table 2 ) . The sinaA4/sinaDf germaria displayed a lower average number of polycomplexes compared to sinaA4 homozygotes with only 3 . 8 polycomplexes per nucleus in early pachytene ( region 2A ) ( Table 2 ) with a range of 1–8 polycomplexes per nucleus . However , the average number of polycomplexes per nucleus in sinaA4/sinaDf germaria increased to 5 . 0 in early/mid-pachytene ( region 2B ) and remained similar through mid-prophase ( stages 2–9 ) ( Table 2 ) . While sinaA4/sinaDf nuclei had a lower average number of polycomplexes than sinaA4 homozygotes , the polycomplexes displayed more variability in width and had a greater average width than were seen in sinaA4 homozygotes ( Table 3 , Fig 3 , S3 Fig ) . For example , at early pachytene ( region 2A ) sinaA4 had an average polycomplex width of 0 . 39 μm versus 0 . 69 μm for sinaA4/sinaDf polycomplexes . The polycomplexes in sinaA4 oocytes varied greatly in length , ranging from 0 . 20–4 . 28 μm but showed a relatively narrow range of widths with most polycomplexes less than 1 micron ( Table 3 , Fig 3 , S3 Fig ) . Fig 2D illustrates this uniformity of width but not length of the polycomplexes , most of which were similar in appearance to rods . In sinaA4/sinaDf ovaries the average width of the polycomplexes was greater than was observed in sinaA4 homozygotes at all stages ( Table 3 , Fig 3 , S3 Fig ) . A subset of the polycomplexes showed a similar width of around 0 . 5 microns , but polycomplexes that were much wider could also be observed ( Fig 3 , S3 Fig ) . The average length of the polycomplexes in sinaA4/sinaDf nuclei remained similar throughout the stages , with a length of 2 . 17 μm at early pachytene , but as was observed for polycomplexes in sinaA4 homozygotes there was a large degree of variability . The maximum polycomplex length increased to 5 . 49 μm in sinaA4/sinaDf ovaries versus 4 . 28 μm in sinaA4ovaries ( Table 3 , Fig 3 , S3 Fig ) . Rather than predominantly rod-like structures some of the larger polycomplexes tapered at one or both ends in sinaA4/sinaDf nuclei , such as shown in Fig 2E and 2F . We find the relatively small variation in width , but not length of the polycomplexes in sinaA4 nuclei intriguing . Fig 3 shows the that width of the majority of the polycomplexes are 0 . 5 μm or a narrower while the length ranges over several microns . In sinaA4/sinaDf ovaries similar polycomplex widths predominate for polycomplexes under 2 μm in length , but as the polycomplexes increase in length the variability in width increases ( Fig 3 and S3 Fig ) . As the sinaA4 allele is a hypomorph , the mutated Sina protein may retain enough function to constrain the polycomplex width in sinaA4 homozygotes; but reducing the mutated protein dose by half in sinaA4/sinaDf nuclei may loosen this constraint . Polycomplexes were also observed in females carrying the sina3 null allele in trans to either sinaDf or sinaA4 [42] ( S4A–S4C Fig ) , as well as in females bearing the sinaP21 allele in trans to sinaDf [43 , 44] ( S4D Fig ) . High chromosome nondisjunction was also observed for sinaA4/sina3 females ( S1 Table ) with 52 . 4% X and 34 . 7% 4th chromosome nondisjunction . As the sina3 allele only affects the sina coding sequence [39] ( Fig 1C ) , the presence of polycomplexes in sinaA4/sina3 ovaries ( S4B and S4C Fig ) further supports our conclusion that the sina A4 mutation is responsible for the polycomplex phenotype in sina A4 homozygotes . We noted earlier that overexpression of a FLAGsinaWT construct rescued the chromosome nondisjunction of sinaA4/sinaDf females , and overexpression of the same construct rescued the aberrant SC phenotypes ( S1C Fig ) . Thus , polycomplex formation appears to be a phenotype common to multiple sina mutants and the formation of these polycomplexes is correlated with defects in chromosome segregation . We used structured illumination microscopy ( SIM ) to examine the organization of the abnormal SC structures in greater detail . Using SIM , the LEs of a normal SC can be resolved as two parallel tracks along the chromosome arms with an antibody recognizing the C-terminus of C ( 3 ) G . Similarly , the CE can be identified with an antibody recognizing Corolla , which localizes between these two parallel tracks [8 , 17] . This wild-type , tripartite SC pattern was observed in sinaA4/+ heterozygotes , as expected given that the sinaA4 mutation is recessive ( Fig 4A ) . Examination of sinaA4 homozygotes by SIM revealed stretches of tripartite SC in early pachytene ( region 2A ) ( Fig 4B ) , indicating that sinaA4 can assemble some tracks of visually normal-looking SC . We next examined the polycomplexes in both sinaA4 homozygotes and sinaA4/sinaDf females using the C-terminal C ( 3 ) G and Corolla antibodies . SIM revealed alternating layers of Corolla and C ( 3 ) G C-terminus along the length of the polycomplex structures ( Fig 4C and 4D ) . In both genotypes the C ( 3 ) G and Corolla proteins appeared to localize only on the outside surface of the larger structures ( inset , Fig 3C ) . Morphologically , the structures looked like rods , cones , or complex shapes ( Fig 4C and 4D ) . The presence of repeated alternating layers of SC proteins has been shown to be a characteristic of polycomplexes [27] . This repeating pattern supports our earlier assessment that the aberrant SC structures in sina mutants are polycomplexes . Polycomplexes have been observed in Drosophila oocytes as well as in other organisms [12 , 22 , 23 , 24 , 26 , 28 , 45 , 46 , 47] . However , the polycomplexes found in sina mutants are more numerous and tend to be longer than more recently characterized examples in Drosophila [12 , 45] . To further characterize the architecture of the polycomplexes , sinaA4/sinaDf ovaries were examined using electron microscopy ( EM ) . EM revealed a clear alternating pattern of electron-dense lines that resembled the LE and central element lines observed by EM of wild-type SC ( Fig 4E and 4F ) [12] . Immuno-EM with antibodies to Corolla and the C-terminus of C ( 3 ) G revealed that Corolla , which localized to the denser lines resembling the CE in wild-type SC , alternated out-of-phase with the C-terminus of C ( 3 ) G , which localized to the less-dense lines resembling LEs in wild-type SC ( Fig 4F ) . A similar repeating pattern has been previously observed by EM in Drosophila [12] . An EM image of what appears to be a slice through two polycomplexes supports what was observed by SIM for the larger polycomplexes , that the larger polycomplexes have the appearance of a hollow middle ( Fig 4G ) . Similar views of previously characterized polycomplexes observed by EM in mutants that only express a version of C ( 3 ) G lacking the C-terminus also appeared to be hollow in the middle [45] . Since the central region proteins Corolla , C ( 3 ) G , and Cona are mutually dependent on each other for wild-type SC formation [7 , 8] , we next examined whether Cona was present in the sina polycomplexes . Both an antibody recognizing Cona and a Venus-tagged Cona overexpression construct localized to the polycomplexes in sinaA4 and sinaA4/sinaDf ovaries ( S5 Fig ) . By SIM , the Venus-tagged Cona incorporated robustly into the large polycomplexes in sinaA4 ovaries and was located between the layers formed by the C-terminal C ( 3 ) G antibody ( S5E Fig ) , similar to Corolla staining ( Fig 4C ) . Taken together , these results support the conclusion that mutations in sina lead to the formation of numerous , large polycomplexes during prophase in Drosophila females . Moreover , the repeating array of SC proteins illustrates that the sina polycomplexes have an organized structure and are not merely amorphous aggregates of SC proteins . While polycomplexes have been observed previously in Drosophila oocytes and other organisms , their protein composition has been extensively examined in only a few organisms [21 , 22 , 25 , 29 , 30 , 31] . In a number of these investigations , the polycomplexes do not appear to be associated with DNA [22 , 23 , 24 , 29] . Additionally , examples of polycomplexes lacking LE proteins have been identified . For example , in Caenorhabditis elegans , cohesin and axial/ LE proteins fail to localize to polycomplexes that result from a mutation in the LE protein htp-3 [22 , 29] . Polycomplex-like structures have even been observed in mammalian cell lines when only the primary transverse filament protein was overexpressed [33] . To examine the polycomplexes in sinaA4 and sinaA4/sinaDf mutants for the association of chromatin and LE proteins , we used a spread protocol in which soluble proteins are removed and only those proteins that are chromatin-associated remain bound to the slide . As was observed for the SC in wild-type nuclei ( Fig 5A ) , we found that polycomplexes could be readily identified within nuclei in chromosome spreads of sinaA4 and sinaA4/sinaDf ovaries ( Fig 5B and 5C ) . While this protocol would fail to reveal the presence of a subpopulation of non-chromatin-associated polycomplexes , these studies provide evidence that at least some of the sina polycomplexes are attached to chromatin . The polycomplexes in Fig 5B and 5C are entirely within the DAPI-stained regions suggesting that the polycomplexes in these nuclei are attached to the chromatin at multiple points along the polycomplexes . As described in Methods , we used a cocktail of antibodies to both Structural maintenance of chromosomes 1 and 3 ( Smc1/3 ) to examine the localization of these core cohesin components to the sina polycomplexes ( Fig 5B and 5C ) . In wild type , Smc1/3 can be observed as tracks similar to the C-terminus of C ( 3 ) G ( Fig 5A ) as has been reported previously [48] . In chromosome spreads of sinaA4 and sinaA4/sinaDf mutants , the Smc1/3 antibodies localized to the sina-induced polycomplexes ( Fig 5B and 5C ) . To the best of our knowledge , the presence of core cohesin proteins has not been reported for polycomplexes previously characterized in Drosophila , although a recent example of polycomplexes in C . elegans does appear to contain SMC3 [31] . We also examined the kleisin-like LE protein C ( 2 ) M ( Crossover Suppressor on 2 of Manheim ) with respect to sina-induced polycomplexes by using an N-terminal HA-tagged C ( 2 ) M overexpression construct ( 3XHAc ( 2 ) M ) [9] . In whole-mount preparations this construct localized to many , but not all , of the SC polycomplexes in sinaA4/sinaDf oocytes ( Fig 5D ) . By SIM , 3XHAC ( 2 ) M clearly localized in two tracks near the C-terminus of C ( 3 ) G in wild-type nuclei ( Fig 5E ) and in a striped pattern near the C-terminus of C ( 3 ) G for sinaA4/sinaDf polycomplexes ( Fig 5F ) . SIM also showed a failure of 3XHAC ( 2 ) M to localize to a subset of sina-induced polycomplexes ( Fig 5G ) . Upon further examination of images of sinaA4/sinaDf germaria , we found that the thin polycomplexes formed in the mitotic nuclei in region 1 consistently appeared to lack 3XHAC ( 2 ) M ( Fig 5D ) . This was not surprising , because C ( 2 ) M is not normally expressed in region 1 of the germarium [9] . The few polycomplexes lacking 3XHAC ( 2 ) M in early pachytene ( region 2A ) were of a similar size to the polycomplexes formed in region 1 . We speculate that the polycomplexes formed in the absence of C ( 2 ) M in region 1 may be unable to later recruit C ( 2 ) M as the nuclei enter meiosis in early pachytene ( region 2A ) . Our current studies cannot differentiate whether the polycomplexes that form in the premeiotic divisions are a different structure that prevents the localization of C ( 2 ) M or whether C ( 2 ) M can only load when polycomplexes are formed de novo . Answering this question will require further investigation . In wild-type Drosophila ovaries , up to four nuclei of the 16-cell interconnected cyst initiate SC assembly along the chromosome arms in early pachytene ( region 2A ) . By mid-pachytene ( region 3 ) all but one of the nuclei disassemble their SC ( Fig 1B ) . The single remaining SC-positive nucleus becomes the oocyte , while the other formerly SC-positive nuclei assume a nurse cell fate . In sinaA4/sinaDf germaria , all the nuclei that normally initiate SC assembly appear to form polycomplexes , while in sinaA4 homozygotes , polycomplexes can be observed in both nuclei in the early/mid- pachytene ( region 2B ) cysts that still contain SC proteins ( S2B and S2C Fig ) . Once formed , polycomplexes remained in mid-pachytene ( region 3 ) in a subset of sina mutant nuclei that would have normally disassembled their SC to become nurse cells . To determine if the persistence of nuclei with sina-induced polycomplexes was the result of a failure to specify the oocyte , we examined Orb localization in sina mutants . Orb is a protein required for the determination of the oocyte nucleus and concentrates in the cytoplasm of the specified oocyte [49 , 50] . In wild-type , as well as sinaA4 and sinaA4/sinaDf mutant ovaries , there was only a single nucleus accumulating Orb protein in mid-pachytene ( region 3 ) ( S2A–S2C Fig ) . This indicates that oocyte specification occurred normally and that the persistence of polycomplexes in the additional nuclei is likely caused by a disassembly delay ( S2B and S2C Fig ) . SC is normally present only in the oocyte nucleus in mid-prophase in wild type ( Fig 6A ) . Polycomplexes were observed to persist in nurse cells in mid-prophase ( stages 2–9 ) in sinaA4/sinaDf oocytes , including in nurse cell nuclei that had begun their endoreduplication cycles as based on DAPI ( Fig 6B and 6C ) . There were typically only one or two persisting polycomplexes per nurse cell nucleus , while multiple polycomplexes could be observed in the four nuclei building polycomplexes in early pachytene ( Table 2 ) . This demonstrates that only a subset of the polycomplexes resisted timely disassembly in the nurse cell nuclei of sina mutants . More intriguingly , a subset of the polycomplexes within sinaA4/sinaDf oocyte nuclei were also resistant to disassembly . While SC progressively disassembles along the chromosome arms in stages 5–7 in wild type , polycomplexes could still be observed in sinaA4/sinaDf oocytes that appeared to be in late prophase ( stages 10–12 ) . Later-stage polycomplexes may not be associated along their entire lengths with the chromatin , since examples could be identified that appeared to associate at one end in the chromatin while the other end of the polycomplex extended past the DAPI staining ( Fig 6D ) . Most surprisingly , while central region components are absent during prometaphase I in wild-type oocytes ( Fig 6E ) , polycomplex-like structures could be observed after spindle assembly in sinaA4/sinaDf oocytes ( Fig 6F ) . These polycomplexes were still present in 81 . 6% of sinaA4/sinaDf prometaphase I/ metaphase I oocytes ( n = 98 ) . They were either associated with the meiosis I spindle or pushed free into the cytoplasm rather than near the chromosomes . In Drosophila females , the chromosomes promote meiosis I spindle assembly , so it is possible the growing microtubules dislodge any remaining polycomplexes near the chromosomes . Polycomplexes were not observed after spindle assembly in sinaA4 oocytes ( n = 46 ) , and no SC structures were observed in wild-type after spindle assembly ( n = 37 ) ( Fig 6E ) . The ability of the polycomplexes to persist after spindle assembly in sinaA4/sinaDf demonstrates that at least a subset of the polycomplexes resist timely disassembly . While polycomplexes have been reported after spindle assembly in other organisms , this is the first observation of polycomplexes present during prometaphase I of Drosophila oocytes [23 , 27] . The central region components of the SC are first loaded at the centromeres in the pre-meiotic dividing nuclei , and the centromeres are the last place the SC components are disassembled at mid/late prophase [10 , 11 , 41] . We examined the localization of centromeres with respect to the polycomplexes in comparison to wild-type nuclei ( S6A Fig ) . We used an antibody recognizing the centromere-specific histone Centromere identifier ( Cid ) [51] to mark centromeres , a Corolla antibody to identify polycomplexes , and an antibody against lamin to clearly demarcate nuclear bounds and assist with scoring individual nuclei ( S6 Fig ) . The use of the lamin antibody demonstrated the polycomplexes were confined to the nucleus ( S6 Fig ) . Small polycomplexes were observed in the pre-meiotic nuclei in region 1 that were associated with one or more centromeres in sinaA4/sinaDf nuclei , similar to the foci of SC components associated with centromeres in region 1 of wild type ( S6B and S6C Fig ) [41] . In early pachytene ( region 2A ) and throughout early/mid-prophase , centromeres appeared associated near one or more of the polycomplexes in sinaA4/sinaDf ovaries ( S6D and S6E Fig ) . While only a subset of the sina polycomplexes were near centromeres , nearly all centromeres were observed near a polycomplex in pachytene . In the hypomorphic sinaA4 mutant , we observed Cid foci associated with the tracks of SC in early pachytene ( region 2A ) in a manner similar to wild type ( S6A Fig ) , but centromeres associated with the polycomplexes once polycomplex formation was initiated ( S6F and S6G Fig ) . Examples where the centromeres appeared to be organized along or around the polycomplexes in sina mutants were observed ( S6 Fig ) . This close association of the centromeres to polycomplexes is more apparent when examining centromeres using stimulated emission depletion ( STED ) microscopy ( S7 Fig ) . In the nurse cells that maintained polycomplexes into mid-prophase , the remaining polycomplexes were observed to also remain associated with Cid foci ( Fig 6B and 6C ) . The SC has been shown to be required for the process of centromere clustering in Drosophila oocytes [10 , 11] . In wild-type Drosophila oocytes , the centromeres from the four sets of chromosomes form an average of approximately two clusters ( Fig 7 and S6A Fig ) [10 , 11] . We examined whether the process of centromere clustering was affected in sina mutant ovaries . While centromeres can associate with the sina polycomplexes , sina mutant ovaries failed to promote and maintain centromere clustering . Consistent with previously published results [10 , 11] , in wild-type nuclei an average of 1 . 5 Cid foci were observed in early pachytene ( region 2A ) and a similar level of clustering was maintained into mid-prophase ( Fig 7 ) . In early pachytene ( region 2A ) in sinaA4 homozygotes , where some nuclei have assembled full or partial SC tracks , there was an average of 2 . 4 Cid foci ( Fig 7 ) . However , as the cyst progresses through pachytene , the increase in polycomplex formation was correlated with a worsening of centromere clustering , averaging 3 . 4 Cid foci at early/mid-pachytene ( region 2B ) . Subsequently , centromere clustering improved at mid-prophase to 2 . 4 Cid foci . The analysis of centromere cluster number in sina mutants was complicated by the varying patterns of Cid localization observed . In some oocytes , Cid appeared as elongated smears wrapping along or around the polycomplexes , thus requiring the nuclei with the more aberrant Cid localization patterns to be excluded from quantification ( S6E Fig ) . In sinaA4/sinaDf ovaries centromere clustering was more severely disrupted . In early pachytene ( region 2A ) , sinaA4/sinaDf nuclei had an average of 4 . 2 Cid foci , with the number of Cid foci varying from one to seven ( Fig 7 and S6D Fig ) . Centromere clustering remained defective in sinaA4/sinaDf ovaries throughout pachytene ( Fig 7 ) . An average of four Cid foci usually indicates that pairing of homologous centromeres is occurring but clustering of non-homologous chromosomes is disrupted ( as Drosophila has four pairs of homologous chromosomes ) . While the average number was four Cid foci , there were examples of nuclei with five or more Cid foci in sinaA4/sinaDf ovaries , indicating that in some nuclei the pairing of homologous centromeres can also be disrupted in sinaA4/sinaDf ovaries ( Fig 7 ) . As we showed above , centromeres are able to associate near the sina-induced polycomplexes , but disruption of sina prevents centromere clustering when polycomplexes form . We cannot rule out that the process of centromere clustering is directly affected by the absence of sina function . However , we prefer a model in which that it is the presence of the polycomplexes that disrupts the ability of the centromeres to cluster . The weaker defect in centromere clustering observed in early pachytene ( region 2A ) in sinaA4 homozygotes ( before full polycomplex formation ) further supports the hypothesis that polycomplexes disrupt centromere clustering and , again , that sinaA4 is a hypomorphic mutation . The centromeres or centromere-associated proteins may have an affinity for some feature of the polycomplexes , and/or centromeres may act as an initiation point for polycomplex formation . The polycomplexes that persisted into later stages in both the pro-oocyte and nurse cells frequently were associated with centromeres ( Fig 6B and 6C ) . Potentially , the centromeres are involved in the mechanism preventing the disassembly of a subset of polycomplexes , similar to the delayed disassembly of SC components near the centromeres in wild type . Null mutations in central region components of the SC cause a reduction in the number of meiotic DSBs , abolish crossover formation , and subsequently cause high levels of chromosome nondisjunction [6 , 7 , 8] . This is likely due to the failure of central region mutants to assemble SC . In sina mutant germaria , however , SC components primarily assemble into aberrant polycomplexes . We wanted to determine if the sina-induced SC polycomplexes were able to function in the formation of crossovers , and therefore we analyzed both the presence of meiotically induced DSBs and the repair of those DSBs into crossover products in sina mutant females . We identified DSBs using an antibody that recognizes γH2AV , the histone modification made in response to DSB formation . In wild-type oocytes , γH2AV foci are first observed in early pachytene ( region 2A ) of the germarium , following the formation of full-length SC ( Fig 8A ) [52] . Foci persist into early/mid-pachytene ( region 2B ) , and steadily decrease in number as cysts progress into mid-pachytene . Very few γH2AV foci are observed in mid-pachytene ( region 3 ) , indicating repair of DSBs into crossovers and noncrossovers has been initiated . In both sinaA4 and sinaA4/sinaDf ovaries , γH2AV foci could be observed at a similar frequency as wild type in early pachytene ( Fig 8B–8D , Table 4 ) , demonstrating that the induction of meiotic DSBs is not disrupted in sina mutants . In addition , the kinetics of DSB repair in sina mutants was not significantly different than wild-type ( Table 4 ) . Taken together these results suggest that DSB formation and repair are not significantly disrupted in the absence of Sina function . In our analysis , we observed that many of the γH2AV foci appeared to be in close proximity to polycomplexes when present in sina mutants ( Fig 8B–8D ) . Therefore , we analyzed the position of the γH2AV foci in relation to the SC for both wild-type and sina mutants . We found in wild-type the γH2AV foci were associated with tracks of SC 100% of the time in both early pachytene ( 52 γH2AV scored from 5 nuclei ) ( Fig 8A ) and early/ mid-pachytene nuclei ( 28 γH2AV scored from 10 nuclei ) . The vast majority of γH2AV foci were also found associated with SC in both sinaA4 and sinaA4/sinaDf from early pachytene throughout early/mid-pachytene . In sinaA4 we found that 100% of the γH2AV foci were associated with SC in early pachytene nuclei containing polycomplexes ( 92 . 8% associated with polycomplexes , 7 . 2% associated with track-like SC , 70 γH2AV scored from 6 nuclei ) . Similar association was observed in early/mid-pachytene ( 85 . 3% associated with polycomplexes , 14 . 7% associated with track-like SC , 34 γH2AV foci scored from 10 sinaA4 nuclei ) . In early pachytene sinaA4/sinaDf nuclei the γH2AV foci were 89 . 0% associated with polycomplexes and 9 . 6% associated with tracks or foci of SC ( 73 γH2AV foci scored from 7 nuclei ) . In early/mid-pachytene sinaA4/sinaDf nuclei the γH2AV foci were 90 . 0% associated with polycomplexes and 3 . 3% percent associated with a focus of SC ( 30 γH2AV foci scored from 14 nuclei ) . Examples of the γH2AV signal appearing to form patches along or around the polycomplexes could be observed in some sina mutant nuclei , making it difficult to assign a location and number of γH2AV foci in these nuclei . These patches of signal may represent association of multiple closely spaced DSBs , but we could not rule out the localization pattern representing a spread of the γH2AV signal from one DSB . While sites of DSBs localize near sina polycomplexes the DSBs may not be interacting directly with the central region components of the sina polyocomplex . Proteins , such as the DSB initiation and repair machinery could be mediating the localization pattern of γH2AV foci in regions near the polyocomplexes . Since DSBs occur at wild type frequency in sina mutant oocytes and can associate with the SC tracks and polycomplexes , we next asked if the DSBs could be converted to crossovers in these mutants . Meiotic recombination along the entire length of the X chromosome was assayed in sinaA4 and sinaA4/sinaDf females and compared to wild type . For sinaA4 homozygotes , the total recombination along the X chromosome was reduced ( to ~20% of wild type ) , particularly for the sc–cv and cv–v intervals , which are the intervals most distal to the centromere ( Table 5 ) . In sinaA4/sinaDf females , recombination was further reduced ( to ~11% of wild type ) but was not abolished . Although DSBs can be recruited to the sina polycomplexes , the polycomplexes may not allow for the efficient maturation of DSBs into crossovers between homologous chromosomes . The failure to form crossovers likely contributes to the high X chromosome nondisjunction of sinaA4/sinaDf females . The higher level of recombination in sinaA4 homozygotes correlates with both the lower levels of X chromosome nondisjunction and the increase in track-like SC observed in early pachytene ( region 2A ) in this mutant background . Our observation that crossover formation is decreased in sinaA4/sinaDf females is consistent with the analysis of prometaphase I/metaphase I oocytes . Chiasmata are required to hold homologous chromosomes at the metaphase I plate . If oocytes fail to make a crossover on at least one pair of homologs , the chromosomes will not remain at the metaphase I plate . The chromosomes will instead move extremely far apart , in some cases nucleating multiple spindles , or will appear to enter anaphase I [53 , 54] . When examining prometaphase I/metaphase I sinaA4/sinaDf oocytes , 69 . 4% ( N = 98 ) of oocytes showed recombination-defective chromosome configurations ( Fig 6F ) that were absent in control oocytes ( N = 37 ) ( Fig 6E ) . Only 28 . 3% ( N = 46 ) of sinaA4 homozygotes displayed recombination-defective chromosome configurations , consistent with the lower chromosome nondisjunction observed for this mutant background . Track-like SC is also more prevalent in early pachytene in sinaA4 germaria when DSBs are induced . It is possible that the DSBs that are matured into crossovers are those associated with track-like SC in sinaA4 and the partial tracks in sinaA4/ sinaDf . Taken together these data suggest that the sina polycomplexes cause a defect with respect to crossover formation , despite the localization of DSBs near the polycomplexes . Since the LE protein C ( 2 ) M localizes to many of the sina polycomplexes , we examined the consequences of loss of c ( 2 ) M on polycomplex formation . We generated females doubly homozygous for a null mutation in c ( 2 ) M ( c ( 2 ) MEP2115 ) and sinaA4 , as well as c ( 2 ) MEP2115; sinaA4/sinaDf females ( Fig 9 and S8 Fig ) . In c ( 2 ) MEP2115 germaria , the SC begins to load as small patches near the centromeres and multiple additional locations , but fails to elongate the central region components into full tracks of SC ( Fig 9A and S8D Fig ) [9] . Surprisingly , in both double mutant backgrounds , the number of polycomplexes was reduced , but not eliminated , compared to each sina mutant alone ( Table 2 and Fig 9B and 9C and S8 Fig , see Fig 2 for sina mutants alone ) . In c ( 2 ) MEP2115; sinaA4/sinaDf nuclei , there was an average of 1 . 3 polycomplexes in early pachytene ( region 2A ) compared to 3 . 8 polycomplexes in sinaA4/sinaDf alone ( Fig 9C , Table 2 ) . This difference in average polycomplex number was even greater in subsequent stages ( Table 2 ) . The phenotype in c ( 2 ) MEP2115; sinaA4 females was more complex . In early pachytene ( region 2A ) , c ( 2 ) MEP2115 mutants display numerous puncta of SC staining , while sinaA4 mutants display mostly track-like SC in the earliest cysts and then form many thin polycomplexes . In c ( 2 ) MEP2115; sinaA4 germaria , only puncta of SC could be observed in 6/37 early pachytene ( region 2A ) nuclei . Polycomplexes were present in the remaining early pachytene nuclei of c ( 2 ) MEP2115; sinaA4 females with an average of 1 . 5 polycomplexes compared to 7 . 7 polycomplexes in sinaA4 single mutants ( Table 2 ) . This difference in polycomplex number between the two genotypes continued in later stages ( Table 2 ) . Intriguingly , the 1–2 polycomplexes in both c ( 2 ) MEP2115; sinaA4 and c ( 2 ) MEP2115; sinaA4/sinaDf nuclei were frequently associated near centromeres , with association defined as Cid signal touching the SC signal [41] ( Fig 9B and 9C ) . At mid-pachytene in c ( 2 ) MEP2115; sinaA4 ( region 3 ) 90% of polycomplexes were associated with Cid ( 10 polycomplexes scored from 8 nuclei ) and in c ( 2 ) MEP2115; sinaA4/sinaDf nuclei 100% of polycomplexes ( 21 polycomplexes scored from 18 nuclei ) were associated with Cid . As with the single sina mutants , the polycomplexes in c ( 2 ) MEP2115; sinaA4/sinaDf showed a large variation in length and width ( Fig 3 and S3 Fig , Table 3 ) , with less variability in polycomplex width in c ( 2 ) MEP2115; sinaA4 oocytes in early pachytene . However , we did observe in c ( 2 ) MEP2115; sinaA4 oocytes the polycomplex width became more variable at later stages ( Fig 3 and S3 Fig , Table 3 ) . These results indicate that while C ( 2 ) M localizes to many of the polycomplexes in sina mutants , it is not required for the formation of all polycomplexes . The propensity for nuclei with 1–2 polycomplexes instead of multiple polycomplexes in sinaA4 mutants indicates that C ( 2 ) M facilitates the initiation of polycomplexes , perhaps similar to how C ( 2 ) M is required for the elongation of central region components into long tracks along the chromosome arms in wild-type oocytes [9] . This idea is further supported by the observation that the 1–2 polycomplexes in each nucleus in the double mutants were often centromere associated . The loading of the central region components to the centromeres requires the Sunn , Solo and Ord protein complex in wild-type oocytes [10 , 48 , 55] . This same complex could be responsible for initiating polycomplexes near the centromeres and in the premeiotic region of the ovary , while initiation of additional polycomplexes at pachytene might be facilitated by a C ( 2 ) M-based protein complex . The different protein requirements of the polycomplexes in sina mutants indicates that polycomplex formation may not be uniform . Sina family members in multiple organisms have been shown to function as E3 ubiquitin ligases to target proteins for degradation . In the Drosophila eye , Sina , in combination with its cofactor Phyllopod , targets the transcriptional repressor Tramtrack for degradation to allow for R7 photoreceptor specification [36] . Closely related homologs of Sina in vertebrates have been implicated in the degradation of numerous proteins [34 , 35 , 37 , 38] , supporting the idea that Sina might act during meiosis to mediate the degradation of a regulator of SC assembly in the oocyte . Indeed , we can imagine two variants on this hypothesis: 1 ) Sina-mediated ubiquitination functions primarily to stabilize those SC components that are properly assembled into SC between the homologs or 2 ) Sina-mediated ubiquitination acts to destabilize polycomplexes thus directing SC components to proper SC formation . That fact that in sinaA4/sinaDf mutant oocytes we observed the formation of rod-like PC , even before canonical SC is formed supports the proposal that the primary role of wild-type Sina is to block PC formation . Much of our ongoing studies focuses on identifying Sina targets , and specifically to determine whether Sina acts to induce the modification of SC structural components or rather regulators of SC assembly . If Sina directly degraded a SC component one would expect the phenotype of sina mutants to be nuclei with normal SC with additional polycomplexes due to the excess SC component self-assembling . sina mutant nuclei can be identified lacking normal SC and displaying only polycomplexes supporting the idea that Sina degrades a protein that regulates the choice between normal and polycomplex SC assembly . Recent studies in other model systems have illustrated the roles of post-translational modifications , including phosphorylation and sumoylation , in controlling the assembly and disassembly of SC components along the chromosomes [5 , 18 , 19 , 20 , 21] . For example , expression in yeast of a version of the central region component Ecm11 that cannot be sumolyated results in the assembly of polycomplexes [21] . Knock-down of a proteasome component led to polycomplex formation in C . elegans [56] . In addition , ubiquitination has been associated with the regulation of meiotic recombination [56 , 57] . Mutation of mouse siah1a , a sina homolog , leads to defects at metaphase/ anaphase I in males [58] . While prophase appeared normal in the mutant male mice , this result suggests that ubiquitination , and specifically by Sina homologs , may play roles in regulating meiosis across species . The C-terminal half of Sina shows high sequence homology with its mammalian Siah1 and Siah2 homologs ( see [16] for full description of the homology among the Sina family of proteins ) . This region of Sina homologs appears to mediate the binding of Sina to many of its targets and cofactors and to facilitate homo- and heterodimerization of Siah family members [59] . The sinaA4 mutation described here resides in a C-terminal beta-sheet that mediates dimerization of human Siah1 in crystallography studies [59 , 60] and is absolutely conserved with the mammalian Siah1 and Siah2 proteins . The sinaA4 mutation is an A to V amino acid change that behaves as a hypomorphic mutation . As an E3 ubiquitin ligase , Sina homologs interact with cell-specific cofactors to recognize targets for degradation [36 , 61 , 62] . The sinaA4 mutation may impact the structure of Drosophila Sina and hamper its ability to recognize a specific meiotic cofactor/target ( s ) . This hypothesis is supported by the observation that sinaA4 mutant flies survive longer in our meiotic assays than null sina mutant flies , are partially fertile , and do not display obvious external mitotic defects [39 , 63] . This suggests the A4 mutation disrupts Sina’s ability to target a protein ( s ) for degradation in the germline more severely than mitotic targets . Alternatively , the sinaA4 mutation may be disrupting the stability of the Sina protein , particularly in the female germline . Varying the dosage of wild-type Sina product clearly affects the onset of polycomplex formation . As noted above , in the hypomorphic sinaA4 homozygotes , SC components attempt to assemble into track-like SC in early pachytene but then disassemble from the tracks and switch to polycomplex formation as the cyst progresses through the germarium . However , polycomplexes are seen earlier in sinaA4/sinaDf mutant ovaries with small polycomplexes being observed in pre-mitotic nuclei in region 1 ( at this stage in wild-type oocytes , central region components are normally only observed as foci associated with the centromeres ) . We interpret these data to indicate that polycomplex assembly needs to be prevented ( presumably by Sina ) even before SC assembly , and likely at multiple stages in meiosis . Perhaps in homozygotes , the sinaA4 mutant protein retains sufficient function until early pachytene to allow for SC assembly along the euchromatin , but ultimately fails to degrade enough of the target to prevent SC components from disassembling along the chromosome arms and initiating polycomplex formation . Moreover , the delayed appearance of polycomplexes in sinaA4 homozygotes indicates that promotion of SC components into canonical SC may be an active process that needs to be continued throughout pachytene . Sina may be needed throughout pachytene to actively shift assembly of SC components away from self-assembling into polycomplexes . It is unclear why in sinaA4/sinaDf ovaries a subset of polycomplexes failed to be disassembled in late prophase and appeared to be maintained into prometaphase I ( Fig 6 ) . This persistence of polycomplexes until prometaphase I has been noted in other organisms [23 , 27] . In wild-type Drosophila females , the SC disassembles along the chromosome arms earlier than it does from the centromeres . This difference in the timing of SC disassembly indicates the oocyte must have a mechanism to differentiate between SC at different locations . A similar mechanism may mark some sina-induced polycomplexes as ready for disassembly in mid-prophase while others are marked to resist degradation at the time the SC is disassembled at the euchromatin in wild-type oocytes . The differential disassembly of sina polycomplexes may provide insight into the regulation of wild-type SC assembly and disassembly . As is seen for the rare spontaneously occurring polycomplexes previously observed by EM in Drosophila [12 , 28] , EM of the polycomplexes in sinaA4/sinaDf mutants shows repeating layers of lateral and central elements . Both Corolla and Cona ( the two known central element proteins in Drosophila ) localize to the sina polycomplexes . Based on immuno-EM ( Corolla ) and SIM ( Cona and Corolla ) , the proteins alternate with the C-terminus of C ( 3 ) G , which in wild type localizes to both LEs of the SC [8 , 12 , 17] . This arrangement suggests that the polycomplexes are composed of repeating layers of wild-type tripartite SC . How one “layer” of SC interacts with the next to form polycomplexes has been a question pondered for many years [23 , 27] . In sina mutants , the LE protein C ( 2 ) M localizes to most , but not all , of the polycomplexes in a pattern similar to the C-terminus of C ( 3 ) G ( Fig 4F ) ; this indicates that like in wild type , C ( 2 ) M localizes near the C-terminus of C ( 3 ) G . Since early sina polycomplexes lack C ( 2 ) M , this protein may not be required to connect the layers of the polycomplex but rather may play a role in initiating and promoting additional polycomplex formation . This idea is supported by the observation that polycomplex formation is reduced , but not abolished in c ( 2 ) M; sina double mutants ( Fig 9 and S8 Fig ) . More intriguing is the localization of the core cohesin proteins Smc1/3 to sina polycomplexes . In C . elegans , SMC3 and meiotic kleisin proteins were recently observed to localize to polycomplexes in akir-1; ima-2 double mutants , which likely have defects in the nuclear import and loading of cohesin proteins [31] . Conversely , core cohesin proteins in C . elegans were not observed to localize to polycomplexes caused by loss of meiotic kleisin proteins or the LE protein HTP-3 [22 , 25 , 29] . These examples , as well as the sina polycomplexes , illustrate that polycomplexes display variations between and within species despite exhibiting structural similarities . Normally , the proteins C ( 2 ) M and the cohesins Smc1/3 are localized within the LE along the chromosome axes . In Drosophila , C ( 3 ) G spans the distance between the LEs , and by EM the C-terminus of C ( 3 ) G localizes at the LE [12] . One explanation for the localization of LE proteins to the polycomplexes is that the LE proteins are attracted to the C-terminal ends of C ( 3 ) G present in the sina polycomplexes . In the absence of competing central region components ( as is the case for null mutants for central region components ) , the LE components remain associated with the chromosomes [7 , 9 , 48] . Alternatively , Sina may degrade a regulator that controls the localization of LE proteins in the homolog axes . Failure to degrade this regulator in sina mutants might drive the loss of LE proteins from the chromosome axes . Decreased LE proteins along the chromosome axis could prevent stable association of the central region along the homologs to promote polycomplex formation . Indeed , females mutant for both c ( 2 ) M and ord , which would disrupt both proposed meiotic cohesin complexes , display polycomplex formation indicating that loss of multiple lateral element protein complexes can cause polycomplex formation [11 , 55] . The possibility that Sina affects the localization of LE proteins would also be consistent with mechanisms of polycomplex formation in C . elegans in which loss of cohesin and LE proteins leads to polycomplex formation [22 , 25 , 29] . The presence of LE proteins could provide an explanation for the chromatin association of sina polycomplexes . While many previously studied polycomplexes do not show chromatin association [22 , 23 , 24] , there is at least one reported example of DNA-associated polycomplexes . In a yeast mutant for the meiosis-specific transcription factor ndt80 , DNA was shown in close association with the lines similar in appearance to the LEs of the polycomplexes as observed by EM [64] . Conversely , the polycomplexes in ord; c ( 2 ) M mutants , which would disrupt both proposed cohesin complexes , were seen in the cytoplasm indicating these polycomplexes are not chromatin associated [11 , 55] . Both observations are concordant with the idea that the LE proteins that associate with the sina polycomplexes may be mediating the binding of sina polycomplexes to chromatin . While the sina polycomplexes are abnormal SC structures , there may well be aspects of their structure that resemble wild-type SC for both centromeres and DSBs ( or centromere and DSB associated proteins ) to localize near the polycomplexes . In wild-type oocytes , central region components first load to the centromeres and they are required for centromere clustering [8 , 10 , 41] . In yeast the transverse filament protein Zip1 also localizes early in meiosis to centromeres [65] . It is unclear how the SC interacts with the centromeres and how it is preferentially loaded to centromeres first in the mitotic divisions in wild-type flies . The centromeres or proteins that bind the centromeres may recognize some protein or substructure common to both wild-type SC and polycomplexes . As is the case for centromeres , the recombination machinery may be able to recognize the polycomplexes as SC to affect the localization of DSBs . In Drosophila females the Vilya and Narya proteins , which are required for DSB formation and repair , initially localize to the central region of the SC , as well as to DSBs [66 , 67] . It is possible these and other DSB associated proteins are mediating the localization of DSBs near sina polycomplexes . However , the abnormal centromere clustering and decreased recombination in sina mutants indicates both processes require wild-type SC structure for completion . Decreased recombination is also associated with polycomplex formation in several C . elegans mutants [31 , 32] . The sina polycomplexes provide a tool for dissecting which aspects of SC structure are required for different SC functions . The observation of polycomplexes in numerous organisms illustrates that SC components have the intrinsic ability to polymerize into polycomplexes [13 , 22 , 23 , 24 , 25 , 27] . It would be beneficial to develop mechanisms to prevent polycomplex formation , which could interfere with the proper assembly of SC components to synapse the homologous chromosomes . Preventing polycomplex formation would also be crucial at times when SC components are expressed but chromosomes are not fully synapsed , such as during the final mitotic divisions of the 16-cell cyst in the Drosophila ovary when SC components are restricted to assembly only at the centromeres . The phenotypes of sina mutants in the germline suggest that Sina plays a pivotal role in inhibiting SC components from assembling into polycomplexes during these periods ( Fig 10 ) . As an E3 ubiquitin ligase , this role on SC formation is likely indirect . Sina likely targets for degradation a protein that promotes the assembly of SC components ( both central region and LE proteins ) into polycomplexes or inhibits normal assembly between the homologs . The localization of LE proteins to the sina polycomplexes could allow DNA association , including centromeres , with the polycomplexes . When both the LE protein C ( 2 ) M and Sina are absent , polycomplex formation is reduced , indicating that C ( 2 ) M further promotes polycomplex formation in sina mutants . Since Sina proteins normally target other proteins for degradation , what is the likely target protein being degraded ? One possibility is a protein that modifies the chromosome axis to prevent the loading and/or maintenance of SC components along the chromosomes . A second possibility is that Sina targets a protein that post-translationally modifies SC components to promote self-assembly into polycomplexes . In the premeiotic divisions of the 16-cell cyst , central region components are present and load to the centromeres but must be prevented from loading along the chromosome arms . Potentially , Sina could be degrading a meiotic regulator that helps to promote this pre-meiotic state of SC assembly . While overexpression of the FLAGsinaWT construct rescued the meiotic phenotypes of sinaA4/sinaDF females ( S1 Fig ) , the resulting FLAGSinaWT protein recognized by an anti-FLAG antibody did not consistently localize to a cellular structure to suggest the potential target of Sina ( S9A–S9C Fig ) . In addition , an anti-Sina antibody did not recognize any specific cellular structure in wild-type germaria ( S9D Fig ) . This lack of localization of Sina to the SC supports the idea that Sina’s target is a protein that can modify SC assembly , rather than a SC component itself . Identifying this Sina target is an area of active research , as it will provide insight into determining what controls the decision of when and where to assemble full-length SC . The gene seven in absentia ( sina , CG9949 , FBgn0003410 ) is located on the 3rd chromosome . The sinaA4 mutation is an A-to-V mutation at amino acid 270 that was isolated in an ethylmethanesulfonate-induced forward genetic screen . The mutagenized chromosome was recombined to pick up the visible markers sr e ca to remove linked mutations on 3R resulting from the mutagenesis ( see Flybase for descriptions ) . sinaA4 refers to the genotype sinaA4 sr e ca . The sina3 mutation is a deletion in the C-terminal half of the Sina protein that leads to an early stop at amino acid 212 and is presumed to be a null allele of sina [39] . The molecular lesion in the sinaP21 allele has not been reported [43 , 44] . The deficiency sinaSH ( named sinaDf here for simplicity ) is a small deficiency removing parts of the sina and sinaH genes as described [68] . Wild type was y w; spapol except for experiments with genetic overexpression constructs and recombination . Overexpression constructs include pUASp: 3XHA-c ( 2 ) M ( gift from Kim McKim ) [9] , pUASP-Cona-venus [7] and FLAGsinaWT ( described below ) . Overexpression constructs were driven in the germline by Pnos-Gal4::VP16 [69] located on the X chromosome . Imaging of early stages in whole-mount ovaries was carried out as described in [67] with two changes: female flies were yeasted and dissected 1–2 days post eclosion , and ovaries were dissected in PBS plus 0 . 1% TWEEN-20 ( PBST ) . For imaging of late-stage oocytes , ovaries were dissected in 1X Robb’s buffer ( 55 mM sodium acetate , 8 mM potassium acetate , 20 mM sucrose , 0 . 44 mM magnesium chloride , 0 . 01 mM calcium chloride , 20 mM HEPES ) with 1% bovine serum albumin ( BSA ) . Ovaries were fixed in 16% EM-grade formaldehyde ( Ted Pella , Redding , CA ) mixed 1:1 with a solution of 200 mM potassium cacodylate , 200 mM sucrose , 80 mM sodium acetate , and 20 mM EGTA for 5 min . Ovaries were washed 3 times in PBS plus 0 . 1% Triton X-100 ( PBSTx ) for at least 15 min each and then manually dechorionated between two frosted slides . Ovaries were blocked in PBSTx with 5% normal goat serum for at least 1 hr and then primary antibodies were left overnight in fresh blocking solution at 4°C . After 3 washes in PBSTx , secondary antibodies were applied in blocking solution for 4 hr with DAPI added during the last 15 min . After 3 washes in PBSTx , the samples were mounted in Prolog Gold ( ThermoFisher Scientific ) . Chromosome spreads were carried out as described in [48] with the few following minor changes . After later stages were removed in the hypo-buffer ( 30 mM Tris pH 8 . 0 , 50 mM sucrose , 17 mM trisodium citrate dihydrate , 5 mM EDTA , 0 . 5 mM DTT , and 1X protease inhibitor cocktail ( Sigma-Aldrich ) ) , the ovary tips were transferred to 30 μl of 100 mM sucrose for fine mincing . The entire 30 μl of solution was spread along a single angled slide that had been dipped into fix ( 25 mL of a 1% paraformaldehyde solution with 350 μl Triton X-100 ) for 15 sec . For immunolocalization , the rehydrated slides were blocked in 5% goat serum , 2% BSA , 0 . 1% Triton X-100 , 0 . 01% sodium azide in PBS under parafilm for 1 hr at room temperature in a humid chamber or overnight at 4°C . Primary antibodies were applied for 3 hr at room temperature or overnight at 4°C . Primary antibodies used for immunofluorescence include mouse anti-C ( 3 ) G 1A8-1G2 , 5G4–1F1 , and 1G5–2F7 ( 1:500 each ) ( 1A8-1G2 was used for standard IF and immuno-EM , and a cocktail of all three C ( 3 ) G antibodies was used for STED ) [12] , affinity-purified rabbit anti-Corolla ( animal 210 ) ( 1:2000 ) [8] , guinea pig anti-Cona ( 1:500 ) [7] , rabbit anti-Cid ( 1:2000 ) ( gift of Dr . Gregory Rogers ) , rat anti-Cid ( 1:2000 ) [70] , mouse anti-Orb antibodies 4H8 and 6H4 ( 1:20 each ) ( Developmental Studies Hybridoma Bank , Iowa ) , mouse anti-γH2AV ( 1:1000 ) [71] , high-affinity rat anti-HA clone 3F10 ( 1:100 ) ( Roche , Indianapolis , IN ) , rabbit anti-GFP ( 1:500 ) ( AB6556 , AbCam Inc . ) , mouse anti-lamin Dm0 ADL84 . 12 ( 1:100 ) ( Developmental Studies Hybridoma Bank , Iowa ) , rat α-tubulin ( 1:250 ) ( Serotec ) , mouse anti-FLAG ( 1:500 ) ( F3165 , Sigma-Aldrich ) , mouse anti-Siah2 ( 1:10 ) ( 24E6H3 , Novus Biologicals ) , rat anti-Smc1 ( 1:250 ) ( described below ) , and guinea pig anti-Smc3 ( 1:250 ) ( described below ) . Secondary goat anti-mouse , rabbit , guinea pig or rat Alexa-488 , Alexa-555 and Alexa-647 IgG H&L chain conjugated antibodies ( Molecular Probes , ThermoFisher Scientific ) were used at 1:500 for early whole-mount samples and 1:400 for chromosome spreads and late stage-samples that used the DeltaVision and OMX microscopes . For STED , goat anti-mouse Abberior STAR 635 ( Sigma Aldrich ) and goat anti-rabbit Alexa-594 IgG H&L chain conjugated antibodies ( Molecular Probes , ThermoFischer Scientific ) were used at 1:500 . For the production of Smc1 and Smc3 antibodies , the following peptides were made by KLH using Biosynthesis: MTEEDDDVAQRVATAPVRKP-cys ( Smc1 ) and CVTREEAKVFVEDDSTHA ( Smc3 ) . Peptides were then injected into a rat ( Smc1 ) or guinea pig ( Smc3 ) by Cocalico Biologicals . A Deltavision Elite system from GE Healthcare equipped with an Olympus IX70 inverted microscope and a high-resolution CCD camera was used for most images . An Applied Precision OMX Blaze microscope ( Issaquah , WA , USA ) furnished with a PCO Edge sCMOS camera was used for standard super-resolution images . SoftWoRx v . 6 . 1 or 7 . 0 . 0 ( Applied Precision/GE Healthcare ) was used to deconvolve images ( Deltavision and OMX ) and reconstruct ( OMX ) images . SoftWoRx v . 6 . 1 or Imaris software 8 . 3 . 1 ( Bitplane , Zurich , Switzerland ) was used for image analysis . Brightness and contrast were adjusted minimally to visualize signals during figure preparation . STED images were acquired with a Leica SP8 Gated STED microscope with a 100x , 1 . 4 N . A . objective . Abberior STAR 635 labelled probe was imaged with a pulsed white light ( 80 MHz ) tuned to 635 nm; Alexa Fluor 594 labelled probe was imaged with the same white laser tuned at 594 nm . Both probes used a pulsed STED 775-nm laser as the depletion laser . All images were acquired in 2D mode with 80%–90% of maximum depletion laser power to maximize lateral resolution , and each image was averaged 8 times in line average mode . The emission photons were collected with an internal Leica HyD hybrid detector with a time gate between 1–6 ns . Raw STED images were deconvolved with the STED module in the Huygens professional deconvolution software ( version 14 . 10; Scientific Volume Imaging ) . Deconvolution was processed with theoretical estimated point spread function and background from raw data , and the signal to noise was set in the range of 15–20 depending on the signal intensity . Other settings used were the system default . Whole-mount immuno-EM samples were prepared as described in [67] . Primary antibodies were mouse anti-C ( 3 ) G ( 1:500 ) and rabbit anti-Corolla ( 1:2000 ) . Secondary antibodies used were anti-rabbit Alexa-488 and anti-mouse ultra-small gold ( Electron Microscopy Sciences , Hatfield , PA ) . For on-section immunogold labeling , ultrathin sections of 50–70 nm in thickness were cut with a Diamond knife ( DiATOME , PA ) , and mounted on Formvar-coated Nickle grids ( EMS ) . The samples were incubated 1 hr at RT with the primary antibodies . After washing in PBS , samples were incubated for 1 hr with goat anti-mouse IgG conjugated to 6-nm gold particles and goat anti-rabbit IgG conjugated to 12-nm gold particles ( Jackson ImmunoResearch Laboratories , PA ) . All sections were post-stained with uranyl acetate and lead citrate , observed , and imaged under a FEI transmission electron microscope . For the quantification of polycomplexes in sinaA4 homozygotes and sinaA4/sinaDf ovaries images were scored that contained lamin antibody staining to demarcate the nuclei boundaries . Only nuclei with low optical distortion in z and minimal overlap with other SC positive nuclei were chosen for analysis . The number of polycomplexes ( labeled with an antibody against Corolla ) was scored by going through the entire nucleus defined by the lamin staining . Foci , puncta and track-like staining of central region components was excluded from the total polycomplex averages . For measurements of polycomplex size a subset of polycomplexes within the scored nuclei were analyzed that contained clearly defined ends for length measurements , showed low optical distortion , and were not too complex in shape ( structures that differed in both length and width at multiple points were excluded for size measurements ) . For simple cylinder shapes the length and width were recorded . For cones , tapered ellipses , and other structures that changed in width , the width of the polycomplexes was taken at the widest point . For double mutants with the c ( 2 ) MEP2115 mutation nuclei were scored without the use of the lamin antibody . The decreased density of polycomplexes allowed for accurate scoring of polyocomplex number and size using only the C ( 3 ) G or Corolla antibodies and DAPI . For analysis of γH2AV association , nuclei were examined by going through the z-stacks for each nuclei . A γH2AV focus was considered associated with a track of SC or a polycomplex if the γH2AV signal touched the Corolla signal or was within 0 . 2 of the nearest SC track or polycomplex . Virgin females 1–3 days post eclosion of the indicated genotype were mated with wild type ( y w/y+ Y; spapol ) males and allowed to acclimate in cages with grape plates supplemented with wet yeast paste for 1–2 days . Females were allowed to lay eggs on fresh grape plates with yeast paste for periods of 2–4 hr . The number of hatched and unhatched eggs was scored approximately 48 hr later . P-values were calculated using Fisher’s exact test in combination with a False Discovery Rate ( FDR ) correction for multiple testing . To measure the rate of both X and 4th chromosome nondisjunction , virgin females of the indicated genotype were mated to multiple X^Y , In ( 1 ) EN , v f B; C ( 4 ) RM , ci eyR males . Assays of wild type ( y w; spapol ) were done one female per vial , but due to the lower fertility of the sina mutants , two ( y; sinaA4; spapol ) or three ( y w/y; sinaA4/sinaDf; spapol ) virgin females per vial were assayed to improve progeny viability . Calculations to determine the percentage of X and 4th chromosome nondisjunction were performed as previously described [72 , 73] . To calculate adjusted progeny in Table 1 , the number of X chromosome exceptional progeny , as well as the number of X+4th chromosome exceptional progeny , were doubled to account for inviable classes of progeny . This total was added to the total number of progeny with normal chromosome segregation and the number of 4th chromosome exceptional progeny to yield the final adjusted total of progeny . To measure only X chromosome nondisjunction in S1B Fig and meiotic recombination along the X chromosome , virgin females ( one virgin per vial for wild type , two for sinaA4 and three for sinaA4/sinaDf due to decreased fertility ) were mated to y sc cv v f/BSY males . For a description of the calculations of adjusted total progeny and X nondisjunction , see [66] . For recombination of the X chromosome in Table 5 , genotypes were y sc cv v f y+/y w ( wild type ) , y sc cv v f y+/y; sinaA4 , and y sc cv v f y+/y w; sinaA4/sinaDf . For recombination , only female progeny resulting from normal X chromosome segregation were scored for all markers . The FLAG-tagged and UASp-driven wild-type sina rescue transgene was built by amplifying the sina coding region from cDNA using PCR while adding a 5’ NotI and a 3’XbaI restriction site . The fragment was subcloned into the pUASp-attB-3XFLAG vector [74] after digestion with NotI and XbaI . The construct was sequence verified and sent to Best Gene Inc . for injection using ɸC31 site-specific integration into the attP40 line . For the alignment shown in Fig 1C the following sequences were obtained from UniProt: D . melanogaster Sina ( P21461 ) , Homo Sapiens Siah1 ( Q8IUQ4 ) , Mus musculus Siah1 ( P61092 ) , Danio rerio Siah1 ( Q72VG6 ) , Xenopus laevis Siah1 ( Q6GQJ5 ) and Hydra vulgaris Siah1 ( T2MA38 ) . The protein domains are depicted as described in [16] .	Mistakes that occur during meiotic chromosome segregation can lead to fetal death or various disorders in offspring , hence proper chromosome segregation is crucial . To ensure optimal execution of this process , crossing over must occur between homologous chromosomes . Crossovers form only in the presence of a large proteinaceous structure called the synaptonemal complex ( SC ) , which forms between homologs during early meiosis . How SC components assemble in a controlled manner into a normal ribbon-like structure between homologs is poorly understood . In female fruit flies , mutations in seven in absentia ( sina ) cause SC components to self-assemble into large structures called polycomplexes , rather than into normal SC . Sina is a member of a family of E3 ubiquitin ligases , which are present in a number of organisms and target other proteins for degradation . We propose that there are proteins that promote the assembly of SC components into polycomplexes , and that Sina is required to degrade such proteins , allowing the normal assembly of SC components between homologous chromosomes .	[ "Abstract", "Introduction", "Results", "Discussion", "Methods" ]	[ "invertebrates", "plant", "anatomy", "medicine", "and", "health", "sciences", "homologous", "chromosomes", "reproductive", "system", "chromosome", "structure", "and", "function", "centromeres", "animals", "germ", "cells", "animal", "models", "oocytes", "drosophila", "me...	2019	The E3 ubiquitin ligase Sina regulates the assembly and disassembly of the synaptonemal complex in Drosophila females
Courtship is a widespread behavior in which one gender conveys to the other a series of cues about their species identity , gender , and suitability as mates . In many species , females decode these male displays and either accept or reject them . Despite the fact that courtship has been investigated for a long time , the genes and circuits that allow females to generate these mutually exclusive responses remain largely unknown . Here , we provide evidence that the Krüppel-like transcription factor datilógrafo ( dati ) is required for proper locomotion and courtship acceptance in adult Drosophila females . dati mutant females are completely unable to decode male courtship and almost invariably reject males . Molecular analyses reveal that dati is broadly expressed in the brain and its specific removal in excitatory cholinergic neurons recapitulates the female courtship behavioral phenotype but not the locomotor deficits , indicating that these are two separable functions . Clonal analyses in female brains identified three discrete foci where dati is required to generate acceptance . These include neurons around the antennal lobe , the lateral horn , and the posterior superior lateral protocerebrum . Together , these results show that dati is required to organize and maintain a relatively simple excitatory circuit in the brain that allows females to either accept or reject courting males . Animals are capable of a staggering array of complex behaviors and many of them rely on innate abilities to compare different scenarios and generate specific and appropriate responses . For instance , most animals can determine with ease whether the best option is to confront or retreat from a predator or opponent . Risk assessment and similar mutually exclusive behaviors are likely to rely on neural circuits that collect information , remove irrelevant and noisy information , and quickly determine a course of action . Courtship rituals are ancient forms of communication that allow animals to identify and rank potential mates in the midst of a noisy and usually complex environment . Thus , it is not surprising that courtships usually deploy a series of displays that involve bright colors , unusual sounds , and rhythmicities . The recipients of these displays , which in many species are females , evaluate their quality and generate the mutually exclusive behaviors of accepting or rejecting courtship . One of the most fascinating aspects of the ability to generate courtship and respond with a decision is the fact that both behaviors are largely genetically encoded; that is , animals are capable of executing them perfectly with minimal practice and no instruction every generation . Pioneering work has established clear associations between individual male courtship behaviors with specific genes and alleles in Drosophila [1] , and even led to the mapping of foci in the central nervous system required to generate discrete behaviors [2]–[5] . However , little is known about how females interpret and integrate aspects of the male's displays and decide if and when to accept male courtship [6] . This is a longstanding question of significance not only to our understanding of the molecular mechanisms of reproductive behavior but also to any comprehensive understanding of how neural circuits generate mutually exclusive decisions . In Drosophila , males show their interest in females by making wing displays , singing a courtship song , dispersing airborne and contact pheromones , and physically contacting them [7]–[9] . In response to these cues , receptive females slow their movement and allow the male to proceed , to finally posture themselves to allow the male to mount them for copulation . In contrast , a disinterested or unreceptive female will engage in a number of rejection behaviors , such as fleeing , kicking the male , extruding her ovipositor , and raising or curling her abdomen [7] . Early studies have shown that no single sensory modality alone determines acceptance or rejection in mature females . Instead , the likelihood of acceptance or rejection relies on different sensory modalities that individually contribute to the final behavioral output [10]–[13] . Genes and alleles that either enhance or inhibit female receptivity have been isolated [14]–[17] . Mutations in these genes provide a unique opportunity to determine the genetic contribution to cell organization and physiological responses required to generate female mate choice [18] . In addition to mutations , somatic mosaics have been employed to determine the regions of the brain underlying female behavior [18] , [19] . Nevertheless , critical information about the neural circuitry involved in female decision-making behavior and the genes that pattern these circuits is still sorely lacking . Here , we describe dati , a neural-specific transcription factor that is required for female courtship acceptance and locomotion , and use it to begin probing the nature of the circuit by which females integrate the signals they receive from courting males to reach the correct behavioral output . Flies were reared at 25°C or 22°C on standard cornmeal–molasses–yeast media ( http://flystocks . bio . indiana . edu/Fly_Work/media-recipes/molassesfood . htm ) . The descriptions of fly stocks used and mutations therein can be found at ( http://flybase . bio . indiana . edu/ ) , unless otherwise stated . The mutant l ( 4 ) 102CDd2 [20] was molecularly mapped in this study . The single breakpoints of the chromosomal deletions Df ( 4 ) C1-7A , Df ( 4 ) B6-4A , and Df ( 4 ) B6-2D and the compound chromosome C ( 4 ) DRA-1 were described previously [21] , [22] . The mutations KG02689 ( dati1 ) and KG01667 ( dati2 ) were also described previously [22] , [23] and correspond to single P element insertions in the second intron and 300 bp upstream of the first exon of CG2052 , respectively . datiF11 . 4 was generated by the excision of the P-element KG02689 inserted in dati1 and corresponds to a precise excision as revealed by direct sequencing of the region that flanks the insertion . UAS-datiF32D is a leaky UAS transgene containing the cDNA of dati ( GH06573 , CG2052-PA ) that rescues the embryonic lethality of l ( 4 ) 102CDd2 up to the first instar in the absence of a driver and its insertion was determined to be on the second chromosome in the neurally expressed gene Mmp2 . The D . melanogaster stocks used were as follows: wild-type Canton-S ( CS ) , w1118 , y w , y w; Df ( 4 ) B6-4A/ciD spapol , y w; Df ( 4 ) B6-2D/ciD spapol , y w; Df ( 4 ) C1-7A/C ( 4 ) DRA-1 , y w; Df ( 4 ) BH/C ( 4 ) DRA-1 , y w; C ( 4 ) DRA-1/dati1 , y w; datiF11 . 4 , y w; SM/Cha-Gal4 UAS-GFP; C ( 4 ) DRA-1/dati1 , y w hs-FLP; SM/FRT42D Actin-Gal4; ciD spa/dati1 , y w hs-FLP; SM/FYGal80T; spapol [23] , w; UAS-mCherry . NLS , w; dati RNAi/dati RNAi [FBst0472372] , w UAS-dcr-2; dati RNAi , y w hs-FLP; SM/FRT42D Actin-Gal4; C ( 4 ) DRA-1/dati1 , elav-Gal4 [FBst0008760] , w; TM3 , Sb/repo-Gal4 [FBst0007415] , w; CyO/Cha-Gal4 UAS-GFP [FBst0006793] , w; ple-Gal4 [FBst0008848] , w; Ddc-Gal4 [FBst0007009] , w; Gad-Gal4 ( gift of T . Sakai ) [17] , Ubi-GFP . NLS [FBst0005626] , Ubi-RFP . NLS [FBst0035496] , and FYT/GAF [23] . Experimental crosses were raised at 25°C , unless otherwise specified . Both males and females used for mating experiments were collected as pupae and aged 3–6 d posteclosion before mating tests . All mating tests were performed at 22°C , between 1 and 4 pm EST . Mating tests were performed in small arenas made by superimposing two sliding sheets of transparent polycarbonate containing 24 wells each ( 2 . 54 cm in diameter and 1 . 27 cm depth ) [24] . Each well was divided in half by a thin removable sheet of plastic . Canton-S males and experimental females 3 to 6 d old were loaded into opposing sides of each chamber without anesthesia with a manual aspirator . Once all wells were loaded , the thin plastic sheet was removed and all pair matings began simultaneously . The chamber was lit from below by an Artograph LED LightPad ( Artograph Inc . ) , and mating behavior was recorded using a Sanyo FH1-A ( Sanyo Inc . ) camcorder for 1 h . For each experimental group , we calculated the courtship acceptance rate , defined as the number of pairs that successfully copulated in the 1 h observation period divided by the number of pairs observed . The average Courtship Index ( CI ) was calculated for each experimental group . CI is defined as the fraction of time a male spent courting in a given observation period [25] . Male courtship for each pairing was observed for 10 min , starting at the onset of courtship . CIs for each pair mating in an experimental group were then aggregated into an average CI . Sample sizes are shown in the corresponding figure in results . Females 3 to 5 d old of experimental and control genotypes were pair-mated to Canton-S males in the mating chamber described above and video recorded for 1 h . For each pair mating , female behavior was analyzed for 10 min from the onset of courtship or until mating occurred . For this time period , every time a male initiated a step of courtship , the female reaction to courtship was recorded . The following six discrete rejection behaviors were quantified: fleeing , kicking , extruding ovipositor , jumping , flicking wings , and standing still . For each female , a Behavioral Index ( BI ) was obtained by calculating the frequency of each behavior displayed over the frequency of all behaviors , and these indices were averaged for each genotype . Locomotor behavior was analyzed using an adaptation of the negative geotaxis assay [26] . Five to eight flies of 3 to 5 d old were placed in a 15 mL Falcon tube without anesthesia and allowed to acclimate for 5 min . After this period of acclimation , the Falcon tube was inverted and rapped sharply against a fly transfer pad three times to knock flies to the bottom of the tube . The tube was then placed in front of the camcorder and flies were allowed to climb the walls . The heights reached by each of the flies after 5 s was assessed from the camcorder footage . Over 30 flies were analyzed for each experimental group . For each Falcon tube of flies , this assay was repeated for a total of five trials , spaced 30 s apart , and the heights of all flies from each trial averaged together . Antibody staining for brains was performed according to standard protocols [27] . Primary antibodies used were chicken anti-GFP ( 1∶1 , 000 , Invitrogen ) , rabbit anti-RFP ( 1∶1 , 000 , Invitrogen ) , mouse anti-nc82 ( 1∶1 , 000 , DSHB ) , Guinea pig anti-Dati ( 1∶1 , 000 , gift from T . Isshiki [35] ) , rat anti-Elav ( 1∶1 , 000 , DSHB ) , and mouse anti-FasII ( 1∶100 , DSHB ) . Secondary antibodies used were donkey anti-mouse 647 ( 1∶500 ) , goat anti-rat 555 ( 1∶500 , Invitrogen ) , donkey anti-guinea pig 647 ( 1∶500 , Invitrogen ) , donkey anti-chicken 488 ( 1∶500 , DyLight ) , and donkey anti-rabbit 555 ( 1∶500 , Invitrogen ) . All samples were mounted in SlowFade ( Invitrogen ) and scanned on a Zeiss LSM 700 confocal microscope . Images generated from Z-stacks taken at 1 or 2 µm intervals are displayed as maximum intensity projections using Zeiss Zen 2009 or as orthogonal projections/surface projections using Image J . Automated cell counting was performed on confocal slices using Fiji software [28] . Briefly , a two-channel stack stained for dati ( green ) and Cha-Gal4 UAS-RFP . NLS ( red ) was converted to RGB , and the yellow overlap was segmented with white color using “Threshold Color” function . The blue channel containing the segmented nuclear overlaps was extracted and the noise removed by filtering the stack with the function “Despeckle . ” Three-dimensional segmentation counts were generated by the plugin “3D Object Counter” [29] . Due to the large size of posterior brain stacks , they were stitched together using the plugin “Pairwise Stitching” before segmentation [30] . Clonal analyses were performed using the FYT ( FLP-recombinase recognition target site-yellow+-Translocation ) system previously described [23] . After clone induction , third instar larvae containing GFP+ clones were handpicked , placed in a single vial , and allowed to develop up to adults 3 to 6 d old . Single females carrying clones were tested with single Canton-S male in the courtship arenas described above and video recorded for 1 h . After this time the number of couples that mated was recorded and the Courtship Indices determined . In the next day , the females that rejected males were retested with new Canton-S males for rejection , and only those that passed in the double rejection test were analyzed further [19] . Females that accepted and rejected males were referenced to specific wells and had their brains dissected . Each clone was located in a grid that divides the brain in 40 anterior and 40 posterior sectors . Because each brain may vary slightly in size or in the way it is mounted , the grid was manually stretched to find the best fit for each sample . In total , 491 clones from 83 brains were analyzed . In these experiments , we used a T-Maze [31] with 2 µl of benzaldehyde in one of the ends . Individual flies were loaded into the elevator of the apparatus and immediately lowered to the level containing the two ends with and without odor . After 10 s , the number of flies that moved away from the aversive odorant or towards it was recorded . Brains of different genotypes were dissected and stained for the mushroom body marker FasII and imaged as z-stacks at 1 µm intervals . Selected z-stacks containing the gamma lobe were manually segmented using the Fiji plugin “Segmentation Editor . ” Measurement of γ lobe morphological defects was done in Fiji . A similar procedure was done to segment the antennal lobe , except that the limits of the segmented structures were defined by the expression of GFP in the pattern of CHA . All statistical analyses were performed using MiniTab 16 . 1 . 0 ( Minitab Inc . ) . For all comparisons of courtship acceptance rate between control and experimental groups , a 2-Proportion Test was performed , and Fisher's exact p test value was used for the determination of significance level between two groups , unless otherwise indicated . CI data were arcsine transformed prior to statistical analysis as previously described [32] and analyzed by one-way ANOVA . The difference in climbing ability in locomotor tests was analyzed by one-way ANOVA . Behavioral indices ( Figure S2 ) were analyzed by Mann–Whitney U Test . All other tests are two sample t tests unless otherwise noted . We previously generated a series of molecularly mapped terminal deletions on the fourth chromosome that define relatively small genomic intervals that can be used to map mutations [22] , [33] . These deletions were then used to map a collection of mutants available for this chromosome , to later test for locomotion and other behavioral abnormalities . One of them was l ( 4 ) 102CDd2 , an unmapped embryonic lethal mutation isolated nearly 50 years ago by Ben Hochman [20] . While mapping l ( 4 ) 102CDd2 we found that 5%–8% of the heterozygotes between this mutation and two deletions ( Df ( 4 ) B6-2D and Df ( 4 ) B6-4A ) escaped the lethality of l ( 4 ) 102CDd2 and exhibited a phenotype of uncoordinated movements , which becomes stronger with age ( compare Movies S2 and S3 ) . Due to the tapping of the forelegs of these genotypes , we named the mutation datilógrafo ( dati ) [34] , which means typist in Portuguese . Subsequent analyses revealed that mutations in dati also render females completely unable to accept male courtship , as will be shown later . We located molecularly the mutation in l ( 4 ) 102CDd2 , which corresponds to a deletion that disrupts dpr7 and CG2052 plus eight other genes in between , and renamed it as deficiency on the fourth chromosome of Ben Hochman [Df ( 4 ) BH] ( Figure 1A and Text S1 ) . Because two single fourth chromosome P-element insertions localized at the breakpoint of these deletions in the CG2052 gene ( KG02689 and KG01667 ) exhibited the same phenotypes as homozygotes or heterozygotes for Df ( 4 ) BH , we focused our analyses on the insertion KG02689 ( dati1 ) , the strongest of these two alleles . dati encodes a zinc finger transcription factor closely related to rotund ( rn ) and squeeze ( sqz ) with homologs in several species , including humans ( Figure 1B ) . Consistent with its reported requirement in specifying late born neurons during embryogenesis [35] , dati is specifically expressed in the central nervous system in embryos ( Figure S1A ) . In larval stages , dati is expressed in the brain and ventral nerve cord ( Figure S1B ) but not in other larval tissues ( e . g . , wing , leg , eye , and antennal discs; unpublished data ) . In adults , dati is broadly expressed in the brain ( Figure S1C ) . dati1 mutants usually stand still for long periods of time , but when courted by males , they can flee at considerable speed . In addition , when cornered by a courting male , they engage in a series of rejection behaviors that include kicking and curling their abdomen ( Movies S1 , S2 , S3 ) [7] , [36] . To investigate how the behavior of dati1 females departs from the wild type , we quantified six discrete behaviors normally displayed by wild-type females in response to male courtship ( i . e . , fleeing , kicking , extruding ovopositor , jumping , flicking wings , and standing still ) . dati females display all of the aforementioned behaviors but spend more time kicking and less time standing still than the wild type ( Figure S2 ) . To further quantify the abnormal mating behavior of dati1 mutant females , we compared their mating success with that of wild-type Canton-S , y w , and dati precise excision revertant females . From these data it becomes evident that the behavior of dati1 is significantly different from the wild-type Canton-S , y w and the revertant datiF11 . 4 females , which exhibit normal acceptance rates ( Figure 2A ) . To test whether the deficit in matings was exclusively due to the female rejection , we assessed the sex appeal of dati1 homozygous females using the CI ( Figure 2B ) [25] , [37] . These experiments reveal that males respond to dati1 females normally , with courtship indices indistinguishable between all four groups . The rejection of dati mutants was tested over a longer time by measuring the frequency of females that produced progeny with a wild-type male in 6 d . The difference between the two groups is not significant ( dati1 6 d = 2/14 versus dati1 1 h = 0/32 , p = 0 . 08 ) , but both are significantly different than Canton-S ( dati1 6 d versus Canton-S 6 d = 29/30 , p<0 . 0001 and Figure 2A ) . This result is consistent with the fact that females that fail to accept males within 30 min are unlikely to mate afterwards [38] . To determine in which tissues dati is required for normal courtship behavior and locomotion , we knocked down its expression using RNAi and UAS-dcr-2 to enhance the knockdown . The knockdown of dati with the ubiquitous Actin-Gal4 [23] at 25°C resulted in few adult individuals that died shortly after eclosion with extreme locomotor abnormalities ( unpublished data ) . To obtain a less severe phenotype more similar to dati1 homozygotes , the UAS-dcr-2 construct was removed from the genotype and the flies were reared at 18°C . Under these conditions , females expressing the dati RNAi from Actin-Gal4 showed defects in acceptance and locomotion ( Figure 2C , D; unpublished data ) . Similarly , the knockdown of dati with elav-Gal4 caused rejection and locomotor defects ( Figure 2C ) . elav is a bona fide postmitotic marker , except for a transient embryonic expression in glial cells and neuroblasts in thoracic and abdominal segments [39] . However , we show that the knockdown of dati in glial cells using repo-Gal4 produced no effect ( Figure 2 ) , indicating that the courtship behavioral phenotypes are not generated in these cells . In addition , we later provide evidence that the behavioral effects of dati knockdown with elav-Gal4 are not associated with neuroblasts of the embryonic ventral nerve cord . Because our previous results suggested that dati might be required in some capacity in neurons , we next asked whether a specific neuronal population could phenocopy the mating deficit observed . The fly brain employs several neurotransmitters including dopamine , acetylcholine , GABA , glutamate , serotonin , histamine , octopamine , and tyramine [40]–[45] . To begin an unbiased search for specific neuronal populations , we first knocked down the expression of dati by RNAi using four Gal4 drivers of genes involved in the synthesis of different neurotransmitters ( Dopa decarboxylase , pale , Choline Acetyltransferase , and Glutamic acid decarboxylase 1 ) ( Figure 2E , F ) to later test other neuronal types if necessary . Out of the four drivers tested , Choline Acetyltransferase Gal4 ( Cha-Gal4 ) produced a strong and significant reduction in courtship acceptance ( Figure 2E , F ) . Thus , the inability of dati females to accept males affects a particular neuronal type . Interestingly , the removal of dati in cholinergic neurons does not impair locomotion as can be observed from “negative geotaxis” escape response tests [26] , [46] . In these tests , dati1 homozygous females normally achieve a much lower mean height 5 s after being knocked to the ground compared to wild-type Canton-S females ( Figure 2G ) . Revertants also have a significantly better climbing ability than dati homozygotes ( Figure 2G ) . However , their climbing ability was not completely restored to the levels of y w , indicating that although most of the climbing deficits can be ascribed to the mutation in dati , other genes in the genetic background contribute to the locomotor deficits observed . In contrast , the climbing abilities of dati RNAi knockdowns with the Cha-Gal4 driver were not different from wild-type Canton-S ( Figure 2G ) , indicating that the male rejection behavior of dati1 mutants is separable from the locomotor deficits . The results above revealed that the acceptance deficits of the dati1 mutant are generated in cholinergic neurons . Because the mushroom bodies in Drosophila express CHA and have been implicated in memory formation , learning , and olfactory processing , we initially tested whether this neuropile was abnormal in dati1 mutants [47] , [48] . The alpha and beta lobes of dati1 mutants appear indistinguishable from the wild-type mushroom bodies , but the gamma lobes are malformed with a generally withered appearance ( Figure 3A , B , D ) and have significantly different curvature ( Figure 3E ) . To determine if the gamma lobe defects could be responsible for the behavioral rejection , we asked whether the knockdown of dati expression in CHA+ cells could recapitulate the morphological defect in the gamma lobe and behavioral phenotypes observed . These experiments revealed that although Cha-Gal4 UAS-dati-RNAi females reject males , the gamma lobe is not affected ( Figure 3C , E ) . Together these experiments allowed us to conclude that although the loss of dati disrupts the gamma lobe neuropile , the focus of dati-mediated courtship acceptance lies elsewhere in the brain . To narrow the region where dati is required for female acceptance , we asked whether DATI- and CHA-positive neurons corresponded to a smaller subset than CHA neurons . DATI is broadly expressed in a complex pattern that involves a few thousand neurons . Automated cell counts indicate that there are around 2 , 400 neurons of the central brain that express dati , which corresponds roughly to 6 . 6% of all neurons of the fly's central brain ( Figure S3 ) [49] . The overlap between CHA- and DATI-positive neurons is much smaller , comprising 345±55 . 3 ( mean ± s . d ) neurons of the anterior central brain ( N = 5 ) , and 1 , 049±134 neurons of the posterior central brain ( N = 8 ) . Based on these cell counts , DATI- and CHA-positive cells ( i . e . , cells that cause rejection with RNAi ) correspond to a modest 4% of the total neurons in the central brain . Besides reducing the complexity of the neural circuit required for acceptance , these experiments revealed that dati is not required to determine cholinergic cell identity . Instead , dati appears to specify a subtype of neuronal identity that is presumably shared by neurons that express different neurotransmitters . To determine the brain regions that mediate acceptance , we performed a clonal analysis using a new genetic tool we developed that allows for the systematic and efficient generation of somatic clones of fourth chromosome mutants , named the FYT system ( Figure 4 ) [23] . In these experiments , we randomly removed dati in different positions in the brain , tested whether females accepted or rejected males , and located the position of each clone within a grid that divides the brain in 80 sectors ( Figure 5A–D ) . By compiling a collection of 491 clones in the brain of females that either produce acceptance or double rejection ( i . e . , rejection in 2 consecutive days ) , it becomes clear that some regions in the brain produce significant deficits in acceptance while others do not . In the anterior brain , a single statistically significant region was identified in anterior sector B2 ( AntB2 , p = 0 . 029 , Figure 5A ) . In the posterior brain , two regions stood out as highly significant ( PosA3 , p = 0 . 004 and PosC4 , p≤0 . 001 ) ( Figure 5B ) . The anterior region AntB2 encompasses the first focus identified for female acceptance behavior using gynandromorphs [19] and also a region populated by extensively characterized local neurons ( LNs ) that express Sex lethal [50] , [51] . The posterior region PosA3 is located in the posterior superior lateral protocerebrum ( pslpr ) immediately above the lateral horn . In contrast , PosC4 spans over the ventral part of the lateral horn , the edge of the posterior inferior lateral protocerebrum ( pilpr ) , and posterior lateral protocerbrum ( plpr ) ( Figure 5D ) . Together , these results show that dati is required in discrete neurons along a known olfactory path [52] , [53] , which involves second-order olfactory neurons and also third-order neurons located around the lateral horn . Interestingly , the ventral lateral horn has been recently identified as the region that processes pheromones [54] , [55] . In contrast , PosA3 appears to be a novel focus implicated in female receptivity . To narrow down the position of the neurons in each sector , we analyzed the neurons that express CHA and DATI within these regions . In the anterior brain , within the region AntB2 , we can discern 13 . 8±2 neurons per hemi-antennal lobe ( N = 16 , Figure 6A and C ) . The posterior brain regions PosA3 and PosC4 that produced the most significant acceptance deficits also have very few DATI CHA neurons . Indeed , in these two regions we can identify 16 . 82±2 . 4 neurons that are positive for DATI and CHA ( N = 15 , Figure 6B and D ) . In the PosA3 sector ( pslpr ) , we found 3 . 64±1 . 18 neurons ( Figure 6D ) , and in the ventral lateral horn and posterior inferior lateral protocebrum ( pilpr ) , there are 13 . 17±2 . 52 DATI CHA-positive neurons ( N = 15 ) . These results suggest that a strikingly small number of DATI CHA neurons are essential for female acceptance . Because we had observed that the removal of dati in olfactory neurons in the region AntB2 impairs female acceptance and dati is required in the specification of late born neurons [35] , we expected that the mutant might fail to specify a neuronal subtype DATI CHA . To begin addressing this issue , we compared the GFP expression patterns of Cha-Gal4 in wild-type and dati1 homozygotes ( Figure 7A–G ) . These experiments revealed severe abnormalities in the cholinergic tracts of the antennal lobes ( Figure 7B–G ) . A closer examination reveals that the population of dorsal lateral neurons in the region AntB2 are either reduced or transformed to cholinergic neuronal types with a distinct morphology than those normally found in this region ( Figure 7C and F ) . These transformations within antennal lobe neurons affect several glomeruli , which include DA1 , the target of the male pheromone cis-vaccenyl acetate ( cVA ) ( Figure 7D–H ) [56] . The experiments above revealed that the loss of dati disrupts olfactory glomeruli . To test whether these disruptions lead to olfactory deficits , we assayed the performance of dati mutant females in a T-Maze in which flies are tested for moving away or towards an aversive odor . In this test , only 3% of the Canton-S flies ( 1 out of 30 ) moved towards the aversive odor compared to 32% of the dati mutant females ( 11 out of 34 ) , indicating that olfactory behavior is indeed impaired in dati mutants ( wild type versus dati , p = 0 . 001 ) . In the lateral horn , the loss of dati leads to a reduction of approximately 10% of the lateral horn neuropile area ( Figure 8C and D; dati , N = 8; WT , N = 10 ) and the cholinergic projections from the antennal lobe towards the lateral horn are also affected ( Figure S4 ) . Like in the antennal lobe , we note the presence of larger neurons in the lateral horn of dati mutants , which are not present in the wild type ( Figure 8A and B ) . Furthermore , there are more CHA-positive cells around the lateral horn , suggesting that in the absence of dati some neuronal precursors can proliferate to later assume a cholinergic fate or , alternatively , that in the absence of dati some cells assume a cholinergic fate ( Figure 8A–D ) . Together , these results show that dati is required in postmitotic neurons as well as in the precursors of these cells . From the previous experiments , we found evidence that dati specifies a subpopulation of cholinergic neurons that project into the antennal olfactory glomeruli . Olfactory neurons in the antennal lobe descend from few neuroblast lineages that generate remarkably different neurons within and across lineages [50] , and it has been suggested that morphologically different neurons are dedicated to specific neurocomputations [57] . This heterogeneity has been traditionally investigated in great detail in clones of single or few neurons using Gal4 drivers that reveal discrete neuronal populations [58]–[60] . However , we are often confronted with the opposite problem , which is to estimate whether a selected neuronal population makes simple , complex , or both simple and complex connections when a discrete Gal4 driver for these neurons is not available . This distinction is important to determine whether dati intrinsically modifies cell shapes or other aspects of neuronal physiology [61] . To that end , we developed a simple system of nuclear bar coding that distinguishes different DATI CHA neurons by color . Nuclear Bar Coding ( NBC ) consists of labeling nuclei of neurons with small or large volumes with different colors by expressing a localized nuclear RFP ( mCherry . NLS ) and GFP-S65T ( nuclear and cytoplasmic ) under the control of a Gal4 driver ( in this case Cha-Gal4 ) . Cells expressing the two fluorescent proteins from the same promoter are expected to be produced and degraded at comparable rates and result in nuclei with an overlay of two colors ( Figure S5 ) [62] , [63] . Assuming that these two proteins are not subject to a different regulation , the overlay of two colors should vary depending on the cellular volume . In cells with long or more intricate processes , GFP-S65T should be expected to fill up the cellular processes and shift the overlay of the two signals in the cell bodies towards that of the localized nuclear fluorescence ( i . e . , red color from RFP ) . Evidence for this shift was obtained in comparisons between cells with short and long cell processes ( Figure 9A–C ) . Conversely , when both GFP and RFP are targeted to the nucleus , the shifts of nuclear bar coding are abolished ( Figure S5 ) . If , to this simple bar coding , we add a third color that detects DATI-positive cells ( Figure 9D ) , then we can globally assess whether dati cholinergic neurons have simple or more complex projections . NBC allowed us to easily identify the descending neurons ( Figure 9A ) , as well as long projection neurons located immediately above the antennal lobe , known as anterior–dorsal projection neurons ( adPNs; Figure 9C , D ) , and LNs imbedded in antennal lateral neurons ( Figure 9C–E ) . In addition , the NBC method reveals that the DATI CHA neurons within the region AntB2 make both short and long connections ( Figure 9C–E ) . Thus , we conclude that dati does not specify only one type of cell shape , like other transcription factors that specify particular neurons [61] . Here we described DATI , a zinc finger transcription factor related to the Drosophila Rotund and Squeeze and the vertebrate ZNF384 , one of the three genes known to be involved in acute lymphoblastic leukemia ( ALL ) [64] , [65] . A survey of the sequences related to dati suggests that it descends from a Krüppel/rotund prototype present in cnidarians ( e . g . , Nematostella , gb\|ABAV01025004 . 1\| ) . Later this prototype evolved to become the rotund-like found in nematodes ( e . g . , C . elegans , Lin29 ) and mollusks ( e . g . , M . galloprovincialis , gb\|GAEN01018610 . 1\| ) and was inherited by both vertebrates and invertebrates . Due to its similarity with Lin29 , dati was previously referred to as Dmel/Lin29 . However , orthology tests show that the ortholog of the C . elegans Lin29 is rotund , not dati . The first true ortholog of dati is found in marine arthropods ( e . g . , Daphnia pulex , Dpdati , gb\|ACJG01001740 . 1\| ) , which appeared in the Cambrian some 540 Mya [66] . Like its vertebrate homolog , dati is expressed in the nervous system and required for stem cell development [35] , [67]–[69] . During embryogenesis , dati is one of the last genes to be activated in a serial activation of transcription factors that determines the identity of specific neuronal lineages in the ventral nerve cord [35] . The present study shows that dati is later required to specify regions of the central brain required for appropriate female acceptance . dati mutant flies are moderately uncoordinated and almost invariably reject male courtship ( Figure S1C and Figure 2 ) . This rejection is so intense and persistent that it does not seem to be due to the mere loss of single sensory modalities , which inhibit but do not abolish acceptance [36] . Because of this strong rejection , we expected that dati might impair either more than one path required to generate acceptance in the brain or an area in which sensory information converges . In addition , we also tested if the locomotor and decision-making defects were associated or separable . The mapping of foci by clonal analyses revealed individuals with clones that exhibited rejection but not locomotor defects ( unpublished data ) . Conversely , we also found individuals with locomotor defects that were perfectly capable of accepting courtship and mating properly ( unpublished data ) . Further evidence that locomotion and female behavior are separable was obtained in the experiments in which dati was knocked down in neurons that express different neurotransmitters ( Figure 2 ) . In this case , we found that none of the four drivers used ( Ddc , Gad , ple , and Cha-Gal4 ) produced locomotor defects like those observed using either a ubiquitous driver or the neuronal driver elav-Gal4 , but the removal of dati in CHA neurons resulted in strong female behavior deficits . Thus , we conclude that the locomotor defects and female acceptance map to different brain regions and distinct cells that express specific neurotransmitters . Our results suggest dati has two roles in the nervous system—one developmental and another constitutive—both affecting female behavior . The over/underproliferation of cholinergic neurons in dati homozygotes suggests a requirement in neuronal precursors , which is consistent with the previous study that showed dati is transiently expressed in developing ganglion mother cells [35] . However , there is a requirement in neurons , as the courtship behavioral phenotype is recapitulated when dati is removed in postmitotic neurons . Further evidence for this requirement in adult neurons is the fact that dati is indeed expressed in neurons well into adulthood , and in fact , we identified a small group of neurons that only initiates expression of dati in adult neurons ( unpublished data ) . Together these results suggest that dati may be required to maintain a neuronal identity . Because not all dati-positive neurons are cholinergic , and vice versa , it is unlikely that its primary role would be to determine the expression of this neurotransmitter . The Nuclear Bar Coding analysis suggests that dati does not evidently define any specific cell morphology either . We speculate that dati specifies a type of neuronal identity that allows neurons to respond to neurotransmitters that other cholinergic neurons without dati cannot . In this scenario , it is easy to see that removing dati from mature neurons would deprive them from the appropriate receptor ( s ) needed to receive input from their synaptic partners , and consequently silence female receptivity . Future tests should resolve whether dati indeed regulates channels/receptors to generate courtship acceptance . Different mutants and experimental approaches , including gynanders , spinster mosaics , mapping of cVA processing neurons , and the use of dati mosaics , here have identified some common and other distinct foci for female decision making . For instance , the first focus AntB2 that we identified maps to Sp11 , the first brain region identified for female acceptance using mosaic gynandromorphs [19] . AntB2 also maps within the Spin-D site identified by mosaics of spinster [18] , a gene also required for female behavior . In addition , the two other highly significant regions , PosC4 and PosA3 , flank the lateral horn , and we note that the focus PosC4 co-maps with regions previously implicated in pheromonal processing in the female brain [52] , [54] . Notably , the lateral horn may have a larger role in sensory integration , as it receives projections from centers that process visual and mechanosensory information [52] . Thus , the picture that emerges from previous work and the present study suggests that female decision making in Drosophila is modulated by a core circuit involving the antennal lobe and the lateral horn . However , we note that there are regions with ratios of acceptance and rejection that intuitively may appear to be relevant but that failed to reach statistical significance . In particular , there are three regions in the anterior brain ( AntB3 , AntB4 , and AntD3 ) and seven regions in the posterior brain ( PosB3 , PosB4 , PosC1 , PosC2 , PosC3 , PosD2 , and PosD3 ) . We believe that these regions are unlikely foci for female receptivity , as our sample had resolution to identify the great significance of a relatively small focus like PosA3 . Also , a similar study that analyzed a larger sample of Spinster foci for female receptivity also found brain regions that did not reach statistical significance but had ratios that could be intuitively interpreted as almost significant like ours . Like us , these authors disregarded these data as significant [18] . Besides providing the locations where courtship acceptance decisions are generated in the brain and the type of neurotransmitter involved , our results also reveal a significant neural mechanism at play . The DATI-CHA neurons mapped in the antennal lobe correspond to a subset of extensively studied cholinergic population known as the excitatory dorsal lateral Projection Neurons ( ePNs ) and excitatory lateral neurons ( eLNs ) [70]–[75] . The central role of excitatory cholinergic neurons revealed by our study and the localization of a region where sensory information is integrated constitute a nearly perfect cellular and molecular representation of the “Summation Hypothesis , ” elaborated by Manning and others several decades ago based on behavioral inference [38] , [76] , [77] . This hypothesis states that acceptance of courtship involves the convergence of multiple excitatory stimulations provided by different sensory modalities until the stimulation reaches a critical threshold point that generates acceptance [76] . Most importantly , the Summation Hypothesis predicts that the two opposite female responses ( i . e . , rejection or acceptance ) are not the result of opposing neural activities ( e . g . , excitation and inhibition ) but rather the result of two different levels of excitation . Until now , there was no molecular and cellular evidence in support of this prediction . In this regard , our results are in agreement with this prediction , as the absence or presence of DATI in an excitatory circuit generates either complete rejection or overwhelming acceptance , respectively . Corroborating our results , recent findings show that pheromone processing is not subject to the inhibitory mechanisms that apply to the processing of other odors [78] . Taken altogether , our results suggest that few dozen excitatory neurons converging in as few as three brain foci make the core components to generate a mating decision in Drosophila . Given that dati-related genes are present in a wide variety of organisms , it is likely that their common ancestor had the same or a similar mechanism of female acceptance .	Males of the fruit fly Drosophila melanogaster generate a series of courtship displays that convey visual , auditory , and olfactory information that females must decode in order to accept or reject mating . Despite the central role of female decision in sexual selection , relatively little is known about how genes and neural circuits generate this behavior . Here we show that the transcription factor datilografo ( dati ) is required to organize and maintain the neural circuitry required for acceptance in the central brain . Strikingly , dati is required in an excitatory circuit involving few neurons that express acetylcholine as their neurotransmitter and are located in the olfactory lobe , the first entry point for odor processing in the brain . In addition , dati is required in two other brain centers: a region where olfaction and presumably other senses are integrated and a novel region . Together these results show that a complex behavior can be generated by very few excitatory neurons , suggesting that the sharp cutoffs between acceptance and rejection may involve different thresholds of stimulation as postulated decades ago .	[ "Abstract", "Introduction", "Materials", "and", "Methods", "Results", "Discussion" ]	[ "invertebrates", "animal", "genetics", "rna", "interference", "nervous", "system", "neuroscience", "animals", "gene", "function", "animal", "models", "drosophila", "melanogaster", "model", "organisms", "epigenetics", "drosophila", "research", "and", "analysis", "methods",...	2014	The KRÜPPEL-Like Transcription Factor DATILÓGRAFO Is Required in Specific Cholinergic Neurons for Sexual Receptivity in Drosophila Females
Efficient and highly organized regulation of transcription is fundamental to an organism’s ability to survive , proliferate , and quickly respond to its environment . Therefore , precise mapping of transcriptional units and understanding their regulation is crucial to determining how pathogenic bacteria cause disease and how they may be inhibited . In this study , we map the transcriptional landscape of the bacterial pathogen Streptococcus pneumoniae TIGR4 by applying a combination of high-throughput RNA-sequencing techniques . We successfully map 1864 high confidence transcription termination sites ( TTSs ) , 790 high confidence transcription start sites ( TSSs ) ( 742 primary , and 48 secondary ) , and 1360 low confidence TSSs ( 74 secondary and 1286 primary ) to yield a total of 2150 TSSs . Furthermore , our study reveals a complex transcriptome wherein environment-respondent alternate transcriptional units are observed within operons stemming from internal TSSs and TTSs . Additionally , we identify many putative cis-regulatory RNA elements and riboswitches within 5’-untranslated regions ( 5’-UTR ) . By integrating TSSs and TTSs with independently collected RNA-Seq datasets from a variety of conditions , we establish the response of these regulators to changes in growth conditions and validate several of them . Furthermore , to demonstrate the importance of ribo-regulation by 5’-UTR elements for in vivo virulence , we show that the pyrR regulatory element is essential for survival , successful colonization and infection in mice suggesting that such RNA elements are potential drug targets . Importantly , we show that our approach of combining high-throughput sequencing with in vivo experiments can reconstruct a global understanding of regulation , but also pave the way for discovery of compounds that target ( ribo- ) regulators to mitigate virulence and antibiotic resistance . The transcriptional architecture of bacterial genomes is far more complex than originally proposed . The classical model of an operon describes a group of genes under the control of a regulatory protein where transcription results in a polycistronic mRNA with a single transcription start site ( TSS ) and a single transcription terminator site ( TTS ) [1] . However , many individual examples have established that the same operon may encode alternative transcriptional units under varying environmental conditions [2 , 3] . Furthermore , advancements in sequencing technology that enable highly accurate mapping of TSSs and TTSs on a genome-wide level have demonstrated that the number of TSSs and TTSs can significantly exceed the number of operons [4] . Thus it seems likely that the bacterial transcriptional landscape , or the genome-wide map of all possible transcriptional units , is shaped by an operon architecture that encodes many TSSs and TTSs within single operons , thus significantly increasing complexity with the objective of enabling diverse transcriptional outcomes [5 , 6] . To achieve a complex landscape of alternative transcriptional units , transcriptional regulation occurs on multiple levels . In addition to the many protein activators and repressors that control transcription initiation , there are also many non-coding RNAs ( ncRNAs ) , including both small ncRNAs ( sRNAs ) and highly structured portions of mRNAs that play essential roles as regulatory elements controlling metabolism , stress-responses , and virulence [7–9] . Trans–acting small RNAs ( sRNAs ) , which are found in intergenic regions [10] or derived from 3’ UTRs [11] , allow selective degradation or translation of specific mRNAs [10] . Cis–acting mRNA structures , such as riboswitches , which interact with small molecules including metal ions , and protein ligands , and other regulatory sequences that are found in the long 3’ UTR of mRNAs , affect expression of their respective genes by regulating transcription attenuation or translation inhibition [12 , 13] . RNA regulation has been shown to play a key role in shaping the transcriptional landscape of a wide range of pathogenic bacteria including Staphylococcus aureus , Listeria monocytogenes , Helicobacter pylori , and strains of Streptococci [6 , 14–22] . Several RNA regulators have been validated and associated with pathogenicity and virulence [23 , 24] , and could be used as highly specific druggable targets [25 , 26] , however , only a select set of regulators have been targeted to date [27 , 28] . Streptococcus pneumoniae is a major causative agent of otitis media , meningitis , pneumonia , and bacteremia . It causes 1 . 2 million cases of drug-resistant infection in the US annually and results in ~1 million deaths per year worldwide [29–31] . While high-resolution transcriptional mapping data are available for other Streptococcus species , these studies have shown limited experimental validation [20] , or have focused primarily on the role of sRNAs in virulence [32] . Additionally , previous studies of the S . pneumoniae transcriptome have demonstrated the presence of ncRNA regulators and assessed their roles in infection and competence , however , these studies also largely focused on sRNAs [16 , 18 , 33] . Thus high-resolution validated maps of the genome-wide transcriptional landscape for different S . pneumoniae strains would be incredibly valuable . Here a comprehensive characterization of the S . pneumoniae TIGR4 transcriptional landscape is created using single and paired-end RNA-Seq [34] , 5’ end-Seq [22] , and term-seq ( 3’-end sequencing ) [35] during early-log and mid-log growth in the presence and absence of the antibiotic vancomycin . We obtain a global transcript coverage map , identifying all TSSs , and all TTSs , which highlights a highly complex S . pneumoniae transcriptional landscape including many operons with multiple TSSs and TTSs . Furthermore , we demonstrate how TSS and TTS mapping under one set of conditions can be leveraged to analyze independently obtained RNA-Seq data collected under a variety of conditions , and we experimentally validate this approach with several cis-acting RNA regulators . Finally , we demonstrate that the functionality provided by the RNA cis–regulator pyrR is critical for S . pneumoniae in vivo in a mouse infection model . Importantly , our work demonstrates how a variety of high-throughput sequencing efforts can be combined to map out a comprehensive transcriptional landscape for a bacterial pathogen as well as identify potentially druggable ncRNA targets . To characterize the transcriptional landscape of Streptococcus pneumoniae TIGR4 ( T4 ) , we first determined transcript boundaries by mapping transcription start ( TSSs ) and termination sites ( TTSs ) from 5’ and 3’ end sequencing reads obtained from 3 different conditions: early-log and mid-log growth phase in the absence and presence of vancomycin ( Fig 1A and 1B ) . Since the TSS predictions ( both whether a TSS exists and the position identified ) correspond well among all three conditions for the vast majority of genes ( S1 Fig ) , reads from both mid-log phase growth conditions were pooled for downstream analyses . To describe a comprehensive transcriptional landscape for T4 , the 2341 annotated genes were supplemented by annotating the homologs of known structured non-coding RNAs ( ncRNAs ) in T4 identified using Infernal ( an RNA specific homology search tool [36] ) , and previously described small RNAs ( sRNAs ) [16 , 18] resulting in a total of 2557 features in T4 . Given this feature set , a total of 742 primary , and 48 secondary , high confidence TSSs were identified , with a processed coverage greater than 100 and a processed/unprocessed ratio greater than 4 , where the primary TSS has the highest processed/unprocessed ratio . A total of 1286 low confidence primary TSSs were called that have coverage greater than 2 and a processed/unprocessed ratio greater than 1 . In addition , 74 low confidence secondary TSSs that have coverage greater than 100 , but a lower processed/unprocessed ratio than the primary TSS were called to yield 2150 total TSSs . From the pooled 3’-end sequencing reads , 1864 TTSs were identified that have a minimum coverage of 10 , and are enriched at least two-fold over the background ( Fig 2 , S1 and S2 Tables ) . Of the 742 high confidence TSSs , 625 showed the presence of the GRTATAAT motif at the -10 position and 215 had the MTTGAMAA motif at the -35 position ( S1 Fig ) . In order to assess how the high confidence TSSs and TTSs contribute to transcriptional complexity , we reconstructed T4 transcripts from paired-end RNA-Seq data . StringTie [37 , 38] detected 343 single gene operons and grouped 1857 genes into 388 multi-gene operons ( S3 Table , and Fig 3 ) . Based on the reconstructed transcripts , we classified operons into five categories based on the number of internal primary high confidence TSSs and TTSs ( Fig 2 ) . Simple operons , transcriptional units with a single gene TSS , and TTS are predominantly identified in the S . pneumoniae genome at 47% of all operons ( Fig 3A ) . Traditional operons with multiple genes and a single TSS and TTS make up 10% of the operons , while multiTSS operons make up 1% , and multiTTS operons make up 16% , of all transcriptional units respectively . However , complex operons ( most of which consist of two genes ) with multiple TSSs and TTSs are the second largest category comprising 26% of all operons ( Fig 3A and 3B ) . Most complex operons are defined by a secondary internal TSS and TTS , however there are several significantly more complex examples where the operon contains multiple TSSs and TTSs ( Fig 3B ) , indicating an intricate system of possible transcripts . Of note , transcript prediction from paired-end RNA-Seq using StringTie agrees partially with operon predictions run using only single-end RNA-Seq data [39] , or established operon databases that use genome context [40] . Such discrepancies between strategies for operon calling have been observed previously ( e . g . Escherichia coli MG1655 ) [4 , 41] ) . For example , StringTie groups genes SP_0001-SP_0014 into a single transcript while Rockhopper splits them into three operons . However , it is evident from the paired-end RNA-Seq coverage map ( see S2 Fig ) that despite the fluctuations in coverage between the coding regions , there is still large enough expression in the intergenic regions for StringTie to group these genes together . This example demonstrates the role of internal transcriptional features in creating alternative transcriptional units . Of note , operon predictions based on either method can only be validated by directed , low-throughput approaches . However , despite the differences arising from the methodologies used , the complexity in operon structure we deduce remains nearly the same except for the portion of multiTTS operons ( see S2 Fig ) . The TSS , TTS , our supplemented annotations , and reconstructed transcripts can be easily visualized using files available at https://github . com/nikhilram/T4pipeline . Since our data revealed many operons with complex structure , we sought to corroborate specific examples using additional data sources . One complex operon we identified , which is also present in existing databases of operon structure [5 , 42] , consists of 9 genes ( SP_1018-SP_1026 ) encoding thymidine kinase , GNAT family N-acetyltransferase , peptide chain release factor 1 , peptide chain release factor N ( 5 ) -glutamine methyltransferase , threonylcarbomyl-AMP synthase , N-acetyltransferase , serine hydroxymethyltransferase , nucleoid-associated protein , and Pneumococcal vaccine antigen A respectively , with two primary high confidence and six other internal TSSs and eight TTSs ( Fig 4A ) . In addition , this operon displays unequal and complex gene expression patterns when independently collected RNA-Seq data from diverse media conditions is mapped to the transcript . In poor growth medium ( MCDM ) the operon can be split into two parts based on expression observed from RNA-Seq coverage maps , where the last five genes in the operon ( SP_1022–1026 ) are expressed higher than the first four genes ( SP_1018–1021 ) , while in rich medium ( SDMM ) the read depth across the operon is similar . While none of the individual genes show significant differential expression between the two conditions ( >2 fold change ) [43 , 44] , there is a drastic difference between the two conditions in the average coverage per gene amongst the first 4 versus the last 5 genes . In rich medium , SP_1022–1026 show an average coverage of ~1 . 9x more than SP_1018–1021 , while in the poor medium the average coverage of SP_1022–1026 is ~3 . 6x that of SP_1018–1021 . This observation corroborates the role of the internal regulatory mechanisms in creating alternate transcriptional units to maintain differences in gene expression between different growth conditions , especially since we identify a high confidence TSS upstream of SP_1022 . Additional validation of our data and analysis approach derives largely from existing low-throughput experiments . The mal regulon is a multiple operon system under the control of a single protein , malR ( SP_2112 ) , which downregulates regulon expression at the malM ( SP_2107 ) promoter [45] . Our data shows that the mal regulon includes operons belonging to two different categories , a multiTTS operon ( malMP/SP_2016–2107 ) and a complex operon ( malXCD/SP_2108–2112 ) . From the RNA-Seq coverage maps it is clear that the two operons can be differentially controlled and expressed in rich vs poor media ( Fig 4B ) . Again , although the genes are not significantly differentially expressed in the two conditions [43 , 44] , the average coverage for each gene is greater in the poor medium in comparison to that in the rich medium . Furthermore , the TSS and TTS identified by our analysis reveal features that have been previously described in lower throughput assays [24] . Despite experimental analyses grouping SP_2108–2110 and SP_2111–2112 into two different operons , it is evident from the paired-end RNA-Seq coverage map as well as RNA-Seq of T4 grown in MCDM that transcription continues past the TTS of SP_2110 into SP_2111 . In SDMM , however , transcription seems to mimic what has been experimentally shown . Thus , although our data may highlight many examples of complex transcriptional architecture , these examples are verifiable through the incorporation of additional RNA-Seq data , and where applicable are consistent with low-throughput studies done in the past . To identify RNA regulators that act through premature transcription termination , we compiled 3’ end sequencing reads upstream of translational start sites , allowing a minimum 5’-untranslated region ( UTR ) length of 70 bases . We detected 162 such early TTS sites that represent novel high and low confidence putative regulatory elements , of which 141 were completely intergenic ( represented in blue in the 5th band of Fig 2 and S4 Table ) . Intriguingly , these putative regulatory elements are not limited to the 5’-UTR of protein coding genes . Of these 141 novel putative regulatory elements , 129 are upstream of protein coding genes while the rest are in the 5’-UTR of annotated ncRNAs . A ncRNA under the regulation of another ncRNA has been shown in Enterococcus faecalis [46] and Listeria monocytogenes [47] but this is the first putative example of this mechanism in Streptococcus pneumoniae TIGR4 . By screening these regulatory elements against 380 published S . pneumoniae strains [48] , covering a large part of the pan-genome , we found that 80 candidates ( ~57% ) were conserved across all genomes , 107 ( 76% ) were identified in at least 350 genomes , 12 ( ~9% ) candidates were identified in fewer than half of the genomes ( S3A Fig ) , while the least distributed candidate , identified upstream of a hypothetical protein , was found only in 52 genomes analyzed . Interestingly , 140 ( 99% ) candidates were found as single copies within a genome , while copies of the putative candidate upstream of the large subunit of the rRNA were found preceding each of the 3 rRNA copies in each genome . Evolutionary distance of each candidate cluster was estimated using MEGA-CC [49] , which reveals that each cluster is made of highly similar , if not identical , sequences ( S3B Fig ) . We reasoned that since we mapped RNA-regulators by means of 5’end and 3’end sequencing , we would be able to associate these regulators with specific growth conditions using environment dependent RNA-Seq data . To confirm the biological relevance of an RNA regulator and associate it with a specific condition one would expect to see a change in the 5’ UTR coverage relative to the accompanying gene . For instance , if a regulator forms an early terminator the RNA-Seq coverage in the 5’UTR is relatively high , while the coverage in the controlled gene would be much lower . Alternatively , if the environment relieves the formation of the early terminator the coverage across the 5’UTR and gene would become less skewed . To determine the applicability of this assumption we leveraged independently collected RNA-Seq data sampled under different nutrient conditions including , rich and poor media [44] , and nutrient depletion conditions where a single nutrient was removed from the environment . RNA-Seq data were mapped to each putative regulatory region and coverage was calculated and averaged across the length of the 5’ UTR regulatory element and the downstream gene . From our list of candidate 5’-UTR regulatory elements , 36 showed more than two fold change in read-through between rich and poor media , with the majority showing an increase in readthrough in poor media . Furthermore , as more RNA-Seq data from other conditions become available , this list is likely to grow to include more putative RNA regulators that show differential read-through under alternative conditions . To associate a regulator to a specific growth condition , we validated homologs of two previously identified RNA cis-regulators; the thiamine pyrophosphate ( TPP ) ( preceding SP_0716 ) and flavin mononucleotide ( FMN ) ( preceding SP_0178 ) riboswitches , using qRT-PCR ( Fig 5A and 5C ) , confirming their media-specific changes in transcription . In many bacteria the TPP riboswitch binds thiamine pyrophosphate and regulates thiamine biosynthesis and transport [50] . Similarly , the FMN riboswitch regulates biosynthesis and transport of riboflavin by binding to FMN [51] . While we validated that these riboswitches respond to poor media by increasing expression of their respective genes ( Fig 5A and 5C ) we suspected that this was due to depletion of each specific ligand in the poor media . Indeed , when poor media is supplemented either with thiamine or riboflavin , expression of the TPP or FMN controlled gene ( SP_0716 and SP_0178 respectively ) decreases by more than 3-fold ( Fig 5B and 5D ) , suggesting that the observed differences between rich and poor media can be directly attributed to the activity of these riboswitches . In addition to previously well-known RNA regulators , a putative regulator preceding ribosomal protein L11 methyltransferase ( SP1782 ) that showed a decrease in read-through and expression in poor media was validated ( Fig 5E ) . While the magnitude of the expression change apparent from qPCR is not as strong as that calculated from the RNA-Seq coverage , there are significant changes in expression that match the direction of change . Since many ribosomal proteins are regulated through the action of cis-regulatory RNAs [52] , this seems like a strong candidate for further study . However , due to the considerable challenges associated with identifying the exact ligand for this putative regulator , a definitive biological function has yet to be assigned and is currently under investigation . In an attempt to validate the feasibility of directly associating RNA-regulators with a highly specific change in the environment we performed RNA-Seq in the presence and absence of uracil . One specific regulatory element that is sensitive to uracil is the pyrR RNA element , which in many bacteria regulates de novo pyrimidine nucleotide biosynthesis through a transcription attenuation mechanism mediated by the PyrR regulatory protein [53 , 54] . In the presence of the co-regulator UMP , PyrR binds to the 5’ UTR of the pyr mRNA transcript ( the pyrR RNA element ) and disrupts the anti-terminator stem-loop thereby promoting the formation of a factor-independent transcription terminator resulting in reduced expression of downstream genes [53] ( Fig 6A ) . In contrast , the co-regulator 5-phosphoribosyl-1-pyrophosphate ( PRPP ) antagonizes the action of UMP on termination by binding to the PyrR protein when UMP concentration is low [55] ( Fig 6A ) . For S . pneumoniae , our data confirms that pyrR RNA elements are present in the 5’ UTR of two pyr operons ( SP_1278–1276; SP_0701–0702 ) , and the uracil transporter ( SP_1286 ) . Furthermore , in response to the absence of uracil the coverage across the two genes directly adjacent to the regulators ( SP_1278 and SP_0701 ) and over the entire two operons increases drastically ( Fig 6B ) , demonstrating that the regulatory elements effectively turn the genes/operons on , which we confirmed by qRT-PCR ( Fig 6C ) . Thus , while term-seq can be used to map novel regulatory RNA candidates on a genome-wide scale , RNA-Seq data can be leveraged , even in retrospect , to identify environmental conditions the regulator responds to . To further investigate the importance of the pyrR regulatory RNA element in growth , three different mutants were constructed that variably affect the 5’ RNA secondary structure ( Fig 6D ) : 1 ) mutation M1 interferes with the binding of PyrR to the pyr mRNA; 2 ) mutation M2 renders the regulatory element in an “always on” state by destabilizing the rho-independent terminator stem-loop structure that is formed in the presence of UMP; 3 ) M3 locks the terminator and creates an “always off” state ( Fig 6D ) . Wild type and mutant strains were cultured in the presence or absence of uracil and the effect of the mutations on expression of SP_1278 were assessed with qRT-PCR ( Fig 6E ) and further confirmed by β-galactosidase reporter assays ( S4 Fig ) . As expected , expression in the wild type decreased ( 9 . 5-fold ) in the presence of uracil confirming the repressive effect of exogenous pyrimidine ( Fig 6E ) [56] . M1 , which should be insensitive to the presence of PyrR and its co-regulator UMP ( Fig 6D ) is indeed unresponsive to the presence of uracil ( Fig 6E ) . M2 triggers constitutive expression of the pyr operon ( Fig 6E ) and M3 has a ~5-fold reduction in expression compared to the wild type regardless of the presence of uracil ( Fig 6E ) . Previously we showed that the pyrimidine synthesis pathway in S . pneumoniae is partially regulated by a two-component system ( SP_2192–2193 ) and that genes in this pathway are important for growth [57] . To determine the importance of a functional pyrR regulatory RNA element in growth , we performed growth experiments with mutants M1 , M2 and M3 in the absence and presence of uracil . These data suggest that a functional pyrR does not appear to be absolutely necessary . For instance , while M1 may have a slight growth defect when cultured in the absence of uracil , M2 has no growth defect in the presence or absence of uracil ( Fig 7A and 7B ) . Although both mutations result in constitutive expression of the pyr operon , mutation M1 leads to higher expression ( Fig 6E , S4 Fig ) indicating that overexpression of the pyr genes may result in accumulation of end products that are detrimental to the cell . Alternatively , the M2 pyrR RNA element can still bind excess UMP-bound PyrR ( as its PyrR binding domain is intact ) thus reducing the effective concentration of UMP in the cell and thereby potential accumulation associated side-effects . Importantly , M3 has a severe growth defect compared to wild type in the absence of uracil ( Fig 7A ) ( p<0 . 002 ) , which can be partially rescued upon addition of uracil ( Fig 7B ) . This suggests that while a constitutive off-state is detrimental for the bacterium in the absence of uracil a constitutive on-state can be overcome , indicating that efficient transcriptional control may not be essential . To determine whether we can manipulate the manner in which the pyrR RNA element affects growth , we determined growth in the presence of 5-Fluoroorotic acid ( 5-FOA ) , a pyrimidine analog . 5-FOA is converted into 5-Fluorouracil ( 5-FU ) a potent inhibitor of thymidylate synthetase , whose activity is essential for DNA replication and repair [58] . Additionally , 5-FU competes with UMP for interacting with the PyrR protein [59] . 5-FU can thus work as a decoy , signaling that UMP is present in the cell; triggering the formation of a terminator and reducing expression of the pyr operon . The wild type strain displayed a severe growth defect in the presence of 5-FOA ( Fig 7C & 7D ) ( p<0 . 05 ) , while M1 ( which should not interact with PyrR and should thus be largely insensitive to the presence of 5-FOA ) displayed a much smaller growth defect ( Fig 7C & 7D ) . In addition , M2 , which constitutively over expresses the pyr operon , is also less sensitive to 5-FOA then wild type ( Fig 7C & 7D ) ( p<0 . 05 ) . Thus , the mutations we introduced into the pyrR RNA element affect the secondary structure in the manner that we intended , and can have far reaching regulatory and fitness effects . Importantly , it shows that a drug targeted against the secondary structure can directly manipulate and severely hamper growth . A remaining key question is the importance of RNA regulatory elements in colonization and the induction of disease . Somewhat surprisingly our in vitro growth curves suggest that constitutive expression of the pyr operon ( M1 ) and constitutive overexpression ( M2 ) is not substantially detrimental to growth , indicating that efficient regulation is not critical . To assess the effect of loss of regulation on bacterial fitness in vivo , the pyrR RNA element mutants were tested in 1x1 competition assays ( mutant vs . wild type ) in a mouse infection model ( Fig 8 ) . While fitness for all three mutants is similar to the unmodified strain in vitro in the presence of uracil , M1 and M3 are unable to colonize and survive in the mouse nasopharynx , or infect and survive in the lung and transition and survive in the blood ( Fig 8A & 8C ) . M2 has less of a defect in vivo , but still has a significantly diminished ability to infect and survive in the lung ( Fig 8B ) . These results indicate that efficient regulation of the pyr operon in vivo is critical for growth and survival of S . pneumoniae within the host . While we had previously shown that genes in the pyr operon are important in vivo [57] , the regulatory findings in this project take our understanding a step further and , importantly , in combination with the findings that 5-FOA can efficiently interact with the RNA regulatory element , suggests that it is feasible to modulate in vivo fitness and thereby virulence by targeting such regulatory elements . With the advent of deep sequencing technologies , our understanding of prokaryotic transcriptional dynamics is rapidly advancing [60] and underlining that bacterial transcriptomes are not as simple as previously thought . Analysis of the S . pneumoniae TIGR4 transcriptome using three different sequencing techniques ( RNA-Seq , term-seq , and 5’end-Seq ) has led to a comprehensive mapping of its transcriptional landscape . We successfully map 742 primary and 48 secondary high confidence TSSs and 74 low confidence secondary TSSs ( 2150 total ) and 1864 TTSs . These results along with a recent study that annotates the S . pneumoniae D39 genome [61] , deeply enriches the understanding of transcriptional regulation in this important pathogen . Besides identifying the transcript boundaries , we also uncovered a complex operon structure , which has also been found in E . coli [4] . Importantly , such complexity likely allows for environment-dependent modulation of gene expression producing variable transcripts in response to varying conditions , which we illustrated here through analyses of a 9-gene complex operon and the mal regulon ( Fig 4 ) . Additionally , similar environment-dependent versatile operon behavior has been observed in E . coli [4] and to a lesser extent in Mycoplasma pneumoniae [62] . This means that our understanding is shifting dramatically and it is thus becoming clear that operons in bacteria should be seen as adaptable structures that can significantly increase the regulatory capacity of the transcriptome by responding to environmental changes in a highly specific manner . Furthermore , this technology can potentially be extended to in vivo growth environments to determine how the bacterium modulates its gene expression in the context of the host immune system . A major challenge with respect to such procedures is the quantity and quality of RNA to generate TSSs , TTSs and RNA-Seq coverage data . However , such limitations could be overcome by obtaining only RNA-Seq coverage data and leveraging these data , as we have done in this study , to identify RNA regulators specific to in vivo environments . Another central aspect of our approach is the identification of putative 5’-UTR structured regulatory elements . Riboswitches and other untranslated regulatory elements ( binding sites for small regulatory RNAs ) are important bacterial RNA elements that are thought to regulate up to 2% of bacterial genes [7 , 8] . However , the discovery of new regulators is difficult when relying solely on computational methodology and sequence conservation [63] . Here we show that through term-seq [35] it is possible to identify such RNA elements on a genome-wide scale and by combining it with RNA-Seq performed in different conditions transcriptional phenotypes can be directly linked to the RNA element . This strategy thus makes it possible to screen for regulatory RNA elements in retrospect by making use of already existing or newly generated RNA-Seq data . Importantly , besides the ability to re-construct an organism’s intricate transcriptional landscape we show that there is also a direct application of our multi-sequencing approach , namely the ability to inhibit operons and/or pathways with specific chemicals or drugs that target the RNA regulatory element . We show that this is possible for the pyrR RNA element , a regulatory element that is important for pneumococcal growth and virulence , which means that this regulatory element could be a potential antimicrobial drug target . This idea is further strengthened by the fact that S . pneumoniae displays a growth defect in the presence of 5-FOA , which directly relates to misregulation of pyrR RNA confirming its drug-able potential . We believe that the presented multi-omics sequencing strategy brings a global understanding of regulation in S . pneumoniae significantly closer , and because the approach is easily transferable to other species , it will enable species-wide comparisons for conservation of operon structure and regulatory elements . In addition , such detailed regulatory understanding creates new regulatory control tools for synthetic biology purposes . Moreover , the combination with in vivo experiments shows that it is a realistic goal to design or select specific compounds that target ribo-regulators in order to mitigate virulence or antibiotic resistance . For RNA-Seq , term-seq and 5’end-Seq library preparation , Streptococcus pneumoniae TIGR4 ( T4 ) was cultured in rich media ( SDMM ) to mid-log phase ( OD600 = 0 . 4 ) . Cultures were diluted to an OD600 of 0 . 05 in fresh media , grown for one doubling ( T0 ) . Cultures were then grown in the presence or absence of 0 . 24 μg/ml vancomycin . At 0 min ( T0 ) and after 30 min of growth ( T30 ) 10ml culture was harvested by means of centrifugation ( 4000 rpm , 7 min at 4°C ) followed by flash freezing in a dry ice ethanol bath and storage at -80°C until RNA extraction . Sample collection was performed in four biological replicates and total RNA was isolated using an RNeasy Mini kit ( Qiagen ) . For qRT-PCR analyses , T4 was cultured in SDMM to mid-log phase ( OD600 = 0 . 4 ) and after centrifugation cultures were washed with 1X PBS and diluted to an OD600 of 0 . 003 in appropriate media . Cultures were harvested at mid-log followed by RNA extraction as described above . 5’end-Seq libraries were generated by dividing the total RNA into 5’ polyphosphate treated ( processed ) and untreated ( unprocessed ) samples that were subsequently processed and sequenced according to protocols described in [22] and [64] with few modifications . See supplemental methods for a detailed protocol ( S1 Methods ) and RNA adapters used ( S5 Table ) . RNA-Seq libraries were generated by using the RNAtag-Seq protocol [34 , 44] . Briefly , 400 ng RNA was fragmented in FastAP buffer , DNase-treated with Turbo DNase , 5’-dephosphorylated using FastAP . Barcoded RNA adapters were then ligated to the 3’ terminus , samples from different conditions were pooled and ribosomal RNA was depleted using the Ribo-zero rRNA removal kit . Illumina cDNA sequencing libraries were generated by first-strand cDNA synthesis , 3’ linker ligation and PCR with Illumina index primers for 17 cycles . The final concentration and size distribution were determined with the Qubit dsDNA HS Assay kit and the dsDNA HS D1000 Tapestation kit , respectively . RNA was extracted in triplicate from cells grown to mid-logarithmic phase ( OD620 = 0 . 4 ) using the QiaShredder and RNeasy Kit ( Qiagen ) by following the kit instructions with the exception of the lysis step , which was performed by membrane disruption using zirconia/silica beads . After extraction , RNA was treated with DNA Free DNA removal kit ( Thermo ) to remove DNA . RNA-seq library was generated via ScriptSeq vs RNA-seq library kit ( epicenter ) using 1 ug RNA . The kit instructions were followed using AMPure beads ( Agencourt ) for cDNA and library purification . Each replicate was individually indexed using Failsafe PCR enzyme ( Epicentre ) . Completed libraries were analyzed for insert size distribution on a 2100 BioAnalyzer High Sensitivity kit ( Agilent Technologies , Santa Clara , California ) . Libraries were quantified using the Quant-iT PicoGreen ds DNA assay ( Life Technologies , ) . One hundred cycle paired end sequencing was performed on an Illumina HiSeq 4000 . term-seq libraries were generated as previously described [35] with few modifications . 2 μg total RNA was depleted of genomic DNA using Turbo DNase , 5’ dephosphorylated , ligated to barcoded RNA adapters at the 3’ terminus and fragmented in fragmentation buffer . Barcoded and fragmented RNA from different conditions were pooled and ribosomal RNA was depleted using Ribo-zero . cDNA libraries were generated by first strand cDNA synthesis and RNA template was degraded as mentioned in the 5’end-Seq library preparation . Second 3’ linker was ligated and PCR amplified with Illumina index primers for 17 cycles . All four library preparations ( RNAtag-Seq , term-seq , 5’end-Seq processed and 5’end-Seq unprocessed ) were pooled according to the method of preparation and sequenced at high depth ( 8 . 5 million reads/sample ) on an Illumina NextSeq500 . The S . pneumoniae TIGR4 ( NC_003028 . 3 ) genome annotation was supplemented with all known structured ncRNAs and sRNAs . Homologs of known ncRNA families deposited in Rfam [65 , 66] were identified in the genome by the cmsearch function of Infernal 1 . 1 [36] . sRNAs previously identified in S . pneumoniae [16 , 18] were identified in T4 using BLAST [67] . Coordinates for the hits from the above searches are included in the supplemented annotation . The single-end sequencing reads from the 5’ end-Seq sequencing , 3’ end sequencing ( term-seq ) , and RNAtag-Seq were processed and mapped to the supplemented S . pneumoniae TIGR4 ( NC_003028 . 3 ) genome using the in-house developed Aerobio pipeline . Aerobio runs the processing and mapping in two phases . Phase 0 employs bcl2fastq to convert BCL to fastq files , quality control and de-multiplexing and compilation of the reads based on the sample conditions . Phase 1 maps the de-multiplexed reads against the genome , under default parameters , using Bowtie2 [68] and streams the uniquely mapped reads to SAMtools [69] to generate sorted and indexed BAM files for each sample . Perl code from [35] was adapted to map the strand specific single nucleotide coverage of the 5’ positions of the 5’ end and 3’ end sequencing reads respectively estimated using BEDTools [70] . With this nucleotide level coverage data calculated from the 5’ end-Seq , regions up to 500 nucleotides upstream of the translational start sites described in the annotated TIGR4 genome ( NC_003028 . 3 ) were scanned for mapped reads with a minimum coverage of 2 and a processed/unprocessed ratio of 1 as in [35] . Predicted primary TSSs were called as high confidence if the position had a minimum processed coverage greater than 100 and a processed/unprocessed ratio of 4 . Secondary TSSs are greater than 10 nucleotides up or downstream of predicted primary TSSs . When multiple putative TSSs were identified in a 5’ UTR , the higher confidence , or the highest processed/unprocessed ratio was assigned as the primary TSS for the downstream gene . Regions up to 100 nucleotides upstream of the predicted high confidence TSSs were split into three datasets—TSS-20; TSS-20-50; and TSS-100 , and searched for enriched promoter motifs and transcription factor binding sites listed in [71] using MEME [72] . Similar to the identification of the TSSs , TTSs were identified by scanning up to 150 nucleotides downstream of the translational stop site for mapped 3’ end reads with a minimum coverage of 10 in the pooled dataset . The position with the highest coverage was considered the most likely TTS for a gene . Predicted TTSs were called as high confidence if the coverage of the site was at least double the average coverage in the 150 nucleotide region . 40 nucleotides upstream of the predicted TTS were scanned for a stable stem-loop structure using RNAFold [73] , and the number of uridines in the 8 nucleotides immediately upstream calculated . The modified perl scripts , sample input files , and the track files for visualization are available at https://github . com/nikhilram/T4pipeline . Paired-end reads were mapped to the T4 genome using TopHat2 [74] . The BAM file of the mapped reads were quality filtered and used to predict transcript boundaries and group genes into operons using a stringent run of StringTie [37 , 38] . Specifically , 0 locus gap separation was allowed and a minimum junction coverage of 10 was required . Predicted operons were compared with the genome based predictions listed in the Database of Prokaryotic Operons [5 , 42 , 75] , and complexity in the operon structure was characterized by surveying the number of high confidence primary internal TSSs and enriched TTSs similar to [4] . Once the TSSs were identified , 5’-UTR regions with a length of at least 70 nucleotides were scanned for mapped 3’ end sequencing reads with a minimum coverage of 2 to identify putative early terminators . 5’-UTR regions with a predicted early TTS were binned as candidate regulatory elements . The nucleotide sequence for each candidate element was obtained and folded using RNAFold [73] . Secondary structures and free energy values were compiled for each candidate . The response of candidate regulatory elements to different media conditions were assessed by calculating the RNA-Seq coverage in both the regulatory element and the regulated gene . Read-through was calculated for each of the candidates as described previously [35] . Briefly , read-through is the ratio ( denoted in percentage ) of the average coverage across the gene to that of the 5’-UTR identified here . The greater the read-through , the higher the expression of the gene with respect to the 5’-UTR . That is , if the regulator reduced the expression of the gene , read-through would be small . If the regulator turned on gene expression in response to certain conditions , the read-through would be large . A local BLAST [67] database was generated with the genomes of 30 S . pneumoniae strains available in Refseq 77 [76] and 350 strains from [48] . Each of the candidate regulators identified in the genome of TIGR4 was BLASTed against this database , and hits in the other genomes were extracted and aligned using MAFFT version 7 [77] . The degree of conservation across the 380 genomes was determined by surveying each candidate cluster post filtering to remove sequences that were less than 70% in length of the query and with e-values greater than 1x10-4 . RNA was isolated from cultures using the Qiagen RNeasy kit ( Qiagen ) . DNase treated RNA was used to generate cDNA with iScript reverse transcriptase supermix for RT-qPCR ( BioRad ) . Quantitative PCR was performed using a Bio-Rad MyiQ . Each sample was normalized against the 50S ribosomal gene , SP_2204 and were measured in biological replicate and technical triplicates . No-reverse transcriptase and no-template controls were included for all samples . Primers used in this study are listed in S6 Table . To generate a pyrR RNA mutant strain , wild-type pyrR RNA element region and ~ 1 kb regions of homology flanking on either side of the RNA element was PCR amplified from S . pneumoniae TIGR4 genomic DNA ( GenBank accession number NC_003028 . 3 ) . Mutations to the pyrR RNA element were obtained either by site-directed mutagenesis or by PCR assembly using appropriate primers ( S6 Table ) . A chloramphenicol resistance cassette was PCR amplified from pAC1000 . The amplified products were assembled such that the chloramphenicol resistance cassette was inserted immediately upstream of the promoter in the opposite orientation . The assembled products were transformed into S . pneumoniae TIGR4 as previously described [78] and transformants were screened for resistance to chloramphenicol and mutation verified by Sanger sequencing . Wild type and pyrR RNA mutants of T4 were grown for 2 hours and diluted to an OD600 of 0 . 015 in fresh media , with varying concentrations of uracil and/or 5-FOA . Growth assays were performed in 96-well plates for 16 hours by taking OD600 measurements every half hour using a Tecan Infiniti Pro plate reader ( Tecan ) . Growth assays were performed no less than three times . 1 x 1 competition experiments were performed with pyrR RNA mutants ( M1 to M3 ) that were competed against the wild-type strain after which bacterial fitness was calculated as previously described [57] with a few modifications . Lung removal and homogenization ( in 10 ml 1X PBS ) , blood collection ( 100 μl ) and nasopharnyx lavage ( with 1 ml 1X PBS ) were perfromed on all animals 24 hours post infection , with the exception of pyrR M3 , which due to the large fitness defect were harvested at 6 hours post infection . Experiments involving animals were performed in accordance with guidelines of IRB/IACUC and approval of the Boston College Animal Facility .	The canonical relationship between a bacterial operon and the mRNA transcript produced from the operon has become significantly more complex as numerous regulatory mechanisms that impact the stability , translational efficiency , and early termination rates for mRNA transcripts have been described . With the rise of antibiotic resistance , these mechanisms offer new potential targets for antibiotic development . In this study we used a combination of high-throughput sequencing technologies to assess genome-wide transcription start and stop sites , as well as determine condition specific global transcription patterns in the human pathogen Streptococcus pneumoniae . We find that the majority of multi-gene operons have alternative start and stop sites enabling condition specific regulation of genes within the same operon . Furthermore , we identified many putative RNA regulators that are widespread in the S . pneumoniae pan-genome . Finally , we show that separately collected RNA-Seq data enables identification of conditional triggers for regulatory RNAs , and experimentally demonstrate that our approach may be used to identify drug-able RNA targets by establishing that pyrR RNA functionality is critical for successful S . pneumoniae mouse colonization and infection . Thus , our study not only uses genome-wide high-throughput approaches to identify putative RNA regulators , but also establishes the importance of such regulators in S . pneumoniae virulence .	[ "Abstract", "Introduction", "Results", "and", "discussion", "Methods" ]	[ "sequencing", "techniques", "medicine", "and", "health", "sciences", "pathology", "and", "laboratory", "medicine", "pneumococcus", "nucleobases", "gene", "regulation", "pathogens", "microbiology", "operons", "nucleotides", "dna", "transcription", "regulator", "genes", "ur...	2018	The Transcriptional landscape of Streptococcus pneumoniae TIGR4 reveals a complex operon architecture and abundant riboregulation critical for growth and virulence
Altered metabolism in cancer cells has been viewed as a passive response required for a malignant transformation . However , this view has changed through the recently described metabolic oncogenic factors: mutated isocitrate dehydrogenases ( IDH ) , succinate dehydrogenase ( SDH ) , and fumarate hydratase ( FH ) that produce oncometabolites that competitively inhibit epigenetic regulation . In this study , we demonstrate in silico predictions of oncometabolites that have the potential to dysregulate epigenetic controls in nine types of cancer by incorporating massive scale genetic mutation information ( collected from more than 1 , 700 cancer genomes ) , expression profiling data , and deploying Recon 2 to reconstruct context-specific genome-scale metabolic models . Our analysis predicted 15 compounds and 24 substructures of potential oncometabolites that could result from the loss-of-function and gain-of-function mutations of metabolic enzymes , respectively . These results suggest a substantial potential for discovering unidentified oncometabolites in various forms of cancers . Otto Warburg observed that cancer cells convert most of their consumed glucose into lactate , despite the presence of sufficient oxygen [1] . This metabolic state , called “aerobic glycolysis” of cancer cells , has been viewed as a passive response required for a malignant transformation [2] , and was originally hypothesized to be a necessary adaptation offsetting dysfunctional mitochondria . In contrast to this initial hypothesis , later studies have found that most tumor mitochondria are not defective in their ability to carry out oxidative phosphorylation [3]–[5] . The notion of passive cancer metabolism is being challenged by recent studies . It was shown that altered metabolism can by itself be a driver for oncogenic [6]–[10] . Recently characterized isocitrate dehydrogenase ( IDH1 , IDH2 ) mutations have established a new paradigm in oncogenesis in that the heterozygous point mutations confer a new metabolic enzymatic activity that produce an oncometabolite ( e . g . 2-hydroxyglutarate ( 2-HG ) , from α-ketoglutarate ) ( Figure 1A; left ) . Surprisingly , 2-HG shows a 100-fold increased concentration in glioma and acute myeloid leukemia's ( AML ) patients with IDH1 or IDH2 missense mutations . This increased concentration of 2-HG competitively inhibits α-ketoglutarate ( α-KG ) binding to histone demethylases , thus blocking differentiation of cells [6] , [7] . In parallel to IDH , loss-of-function mutations on succinate dehydrogenase ( SDHA , SDHB , SDHC , and SDHD ) and fumarate hydratase ( FH ) cause the accumulation of succinate and fumarate , respectively , which also acts a competitive inhibitor of α-KG-dependent oxygenases that regulate hypoxia-inducible factor ( HIF ) oncogenic pathway ( Figure 1A; middle , right ) [9]–[12] . Curiously , although IDH1 and IDH2 mutations are clearly powerful drivers of low grade glioma and AML , they seem to be rare or absent in other tumor types . This observation highlights the importance of the specific cellular context in understanding metabolic perturbations in cancer cells [6] , [13] . Metabolism represents a complex network of biochemical reactions and it may be hard to decipher phenotypic consequences based on single reaction alterations . Annotated genomes and biochemical legacy data , however , have enabled the construction of genome-scale models ( GEMs ) of metabolism [14] , [15] that have been successfully used to compute many observed metabolic states and properties [16]–[19] . A GEM is a formal and mathematical representation of reconstructed metabolism as a genome-scale network [18] , [19] , which consist of collections of metabolic reactions , their stoichiometry , the enzymes and the genes that encode them . GEMs offer a novel mechanistic link between genetic parameters and computed metabolic states . Two versions of the human metabolic reconstruction are available [14] , [20] . With the rapid development of high-throughput experimental methods , recent integrative studies using disparate omics data types have deciphered characteristics of cancer metabolism using in silico GEMs [21]–[23] . Shlomi et al . dissected underlying principles of elevated glycolysis through the simulation of biomass production rates using GEMs [21] . Folger et al . identified drug targets for cancers based on synthetic lethal gene pair analysis using a generic GEM of cancer [22] . Further , the GEM analysis was applied to hereditary leiomyomatosis and renal cell cancer ( HLRCC ) to unravel the survival mechanism that enables the HLRCC cells to operate the mitochondrial electron transport chain despite mutations on FH [23] . The GEM of human metabolism has thus already shown its utility for the analysis and understanding of cancer metabolism . In this study , we predict putative oncometabolites by incorporating genetic mutation information on a massive scale collected from more than 1 , 700 cancer genomes into context-specific GEMs of metabolism for nine cancer types . We reconstructed context-specific cancer and matching metabolic models from corresponding normal tissue for nine cancer types using gene expression profiles of primary cancer cells and site-matched normal cells . By integrating the exome mutation data source with the reconstructed GEM , we predict potential oncometabolites that could show altered concentration in cancer cells due to the loss- or gain-of-function mutations on enzymes ( Figure 1B ) . We collected sets of exome mutation and gene expression data from The Cancer Genome Atlas ( TCGA , http://cancergenome . nih . gov/ ) , Cancer Cell Line Encyclopedia ( CCLE ) [24] , and NCBI Gene Expression Omnibus ( http://www . ncbi . nlm . nih . gov/geo/ ) ( Table 1 ) . We selected three cancer types ( breast , kidney , and squamous cell carcinoma lung cancer ) from the TCGA project with accompanying publically available genetic mutation and gene expression data sets . In addition , we collected mutation information for six types of cancers studied in the CCLE project ( gastric , leukemia , liver , adenocarcinoma lung , ovarian , and pancreas ) . As the CCLE project does not provide gene expression for matched normal samples , we separately collected expression data of cancer and site-matching normal tissues from the NCBI GEO database by matching cancer histological types to the CCLE mutational data . Finally , nine unique histology types of cancer with mutation and gene expression data measured in cancer and site-matching normal were used in the present study ( Table 1 ) . To identify potential oncometabolites that originated from genomic variations in metabolic enzymes , we determined the total number of mutations in enzymatic genes across nine cancer types . The metabolic genes were identified from the global human metabolic network Recon 2 [14] . For every gene in Recon 2 , the number of coding region genetic variants including classes of missense , nonsense , frame shift , in frame indel , silent , and splice site mutations were counted . Between 5 to 20 metabolic genes per sample were mutated in cancer cells ( Figure 2A ) . Also , when we tallied the total mutation count per cancer type , the mutational frequency became more pronounced in each type of cancer ( Figure 2B ) . In the CCLE data sets ( gastric , leukemia , liver , ovarian , and pancreas ) , the initial mutation calling was made within a set of targeted metabolic genes . Thus we expect that the real number of mutated metabolic genes would be higher than the current numbers . Second , we confirmed that the missense mutation , which is the consensus type of mutation observed in IDH , was the most dominant mutation class in metabolic genes ( Figure 2C ) . This analysis was also conducted on additional TCGA mutation data sets , which showed qualitatively similar results ( Table S1 in Text S1 , Figure S1 in Text S1 ) . In order to predict oncometabolites that could originate from mutations , we selected metabolic genes that are recurrently found to be mutated in more than 5% of samples . Transporters were removed from our analysis , as they usually do not represent canonical metabolic transformations . As shown in the examples of IDH and FH , mutations could change the enzyme activity such that it gains a new function ( gain-of-function ) or loses its original function ( loss-of-function ) . In this study , we divided the potential oncometabolites into two categories: ( i ) native oncometabolites that could change concentration due to the loss-of-function ( LoF ) mutations , and ( ii ) promiscuous oncometabolites that could change concentration due to the gain-of-function ( GoF ) mutations . Therefore , for this oncometabolites analysis , we classified mutations into two classes . First , we adopted a definition for LoF mutations to be correlated with loss of function: nonsense ( stop codon introducing ) , splice site indels&SNP ( splice site disrupting mutations ) , and frame shift indels ( disrupting reading frame ) [25] [26] . Second , we focused on the possibility of missense mutations as GoF mutations , which is the consensus type of mutation that is observed in IDH oncometabolite studies . Finally , the recurrently mutated genes are categorized into three classes: ( i ) recurrently mutated genes from a type of GoF mutation , ( ii ) recurrently mutated genes from types of LoF mutations ( nonsense , frame shift indels , splite site indels&SNP ) [26] , and ( iii ) recurrently mutated genes from both of GoF and LoF mutations . As a result , we found 96 enzyme encoding metabolic genes that were recurrently mutated in the nine types of cancer ( Figure 3A , File S1 ) . From the 96 recurrently mutated metabolic genes , we next selected genes that could have significant functional impacts on their catalytic activities due to the mutations . First , isoenzymes were filtered out because an unexpected mutational malfunction of one isoenzyme could be substituted by functions of other isoenzymes having duplicated catalytic activities . Second , mutated genes having smaller functional impact were removed . The impact of a mutation in the recurrently mutated genes was assessed by using the functional impact score ( FIS ) [27] which demonstrated a consistent higher accuracy in a recent systematic assessment [28] . A FIS is derived from multiple sequence alignments of amino acid sequence homologs , thus , the score is based on the evolutionary conservation of a mutated residue in a protein family . By applying the aforementioned two criteria , 20 metabolic genes were selected for detailed analysis . These 20 genes are recurrently mutated in samples ( ≥ 5% ) , and are expected to have significant functional impact from their genomic sequence mutations ( Figure 3B , Figure S2 in Text S1 , File S1 ) . Notably , we observed mutation recurrence in LoF SDHB that is one of the previously identified oncometabolite-producing enzymatic genes ( Table S2 , S3 in Text S1 ) . In order to predict oncometabolites originating from LoF mutations on enzymes , we simulated flux changes of reactions in a cancer cell using GEMs . For the reconstruction of cancer specific and matched normal metabolic networks , gene expression data sets of cancer and matched normal were used . Here , the specific characteristics of cancer and normal metabolic models were represented by the rerouted network structure based upon the presence or absence of an enzymatic reaction in the intracellular environment of cancer and normal cells . This presence or absence of reactions was determined by the present/absent ( P/A ) calls of enzymatic genes ( see Materials and Methods for details ) . With the result of P/A calls of enzyme-encoding genes in nine cancer and normal gene expression data sets , we evaluated the relevance of P/A call results to the cancer specific metabolism . Here , we first assessed gene-wise P/A alterations in cancer vs . normal across nine cancer types . In this analysis , we confirmed that the P/A alterations in cancer vs . normal of enzymatic genes were not consistent across different cancer types ( Figure S3A in Text S1 ) . This inconsistency is in accordance with previous finding that the expression differences of individual genes vary from cancer to cancer [29] . However , when the alterations were evaluated at the level of functional pathways , several pathways related to common malignancy features showed significant P/A alterations ( Figure S3B in Text S1 ) . A pathway is considered to be significantly altered if the pathway is found to have a significantly higher number of genes with P/A alterations than the corresponding number found in a random set of genes ( see Text S1 for details ) . Specifically , reactions transporting ( exporting , secreting ) substances inside and outside of the cell were frequently altered across nine types of cancer . Furthermore , we confirmed that these altered pathways had significant over representation of mutations ( Figures S3C-S3E in Text S1 ) . We reconstructed nine cancer and normal matching models using P/A gene expression calls the Gene Inactivity Moderated by Metabolism and Expression ( GIMME ) algorithm ( see Materials and Methods for details , Table 2 , Figures S4-S6 and Tables S4-S6 in Text S1 ) . The accuracy of the cancer and normal matched metabolic models were evaluated based on: ( i ) how well the structure of the reconstructed network represents gene expression data , and ( ii ) how well the simulated fluxes predict metabolic states of cancer and normal cells . Since the characteristics of the reconstructed model are mainly determined by the result of gene expression P/A calls , we first evaluated how well the network structure of the model represents the gene expression P/A calls . Since a gene can be associated with multiple reactions , or vice versa , the correlation between P/A calls and the network structure was calculated in two steps . The final correlation was determined by correlations of two vectors of Pearson's correlation coefficient ( PCC ) calculated from pairwise correlations of presence/absence of gene expression and pairwise correlations of presence/absence of reactions between cancer types ( Figure 4A ) . Correlation coefficients varied from 0 . 90 to 0 . 98 , and the PCC values were significantly higher than random PCC values ( Figure S7 in Text S1 ) . Thus , the reaction content of the cancer and normal models represent the gene expression data sets well . Second , we tested the accuracy of the simulated flux states of the reconstructed models by evaluating whether the flux states correctly predict metabolic states of cancer and normal cells . We hypothesized that if the flux through a reaction was predicted to increase ( or decrease ) its magnitude in cancer compared to normal , then the expression level and/or abundance of the associated active enzyme in cancer will increase ( or decrease ) in order to meet the change in flux . With this hypothesis , we compared the changing pattern ( increase or decrease ) between flux and gene expression for the reactions with a significant flux change . Flux changes were calculated using the Markov chain Monte Carlo ( MCMC ) ( see Materials and Methods for details ) . Figure 4B shows the accuracy of the predictions for the significantly changed fluxes ( P-value < 0 . 001 , fold-change ≥ 2 ) . The results are fairly accurate . Among the nine cancer models , seven cancer types , except the lung ( SCC ) cancer and leukemia , showed significantly better accuracy than random tests ( P-value < 0 . 05 ) . Furthermore , our results were qualitatively robust with variations in the P-value and fold-change thresholds ( Figure S8 in Text S1 ) In addition , we evaluated the accuracy of predicting essential genes . The accuracy of reconstructed models was evaluated by the enrichment of in vivo essential genes among in silico essential genes computed from Flux Balance Analysis ( FBA ) ( see Materials and Methods for details ) . Across the nine cancer types , we found that about 3–30% of genes contributed to cell growth , about 2–14% of genes were essential for the cell growth ( Figure 4C ) . Further , in several cancer types ( breast , kidney , lung ( SCC ) , and liver ) , we found that genes predicted to have higher contributions to the biomass formation are more enriched in the in vivo essential genes ( Figure 4D ) . The analyses presented here confirm that the simulated flux through the reconstructed models effectively predicts in vivo cellular metabolic activity of cancer and normal cells . From the substrates and products of the nine LoF mutant enzymes ( Figure 3B ) , we chose the metabolites that significantly change the flux state in cancer cells with LoF mutation activity relative to the corresponding normal tissue . These metabolites were identified as potential oncometabolites associated with LoF mutations . To predict LoF oncometabolites , we first modified a reconstructed cancer model into mutated enzyme deficient models . For example , as shown in Figure 3B , FASN , ACACB , and CAD genes were found to be recurrently mutated in leukemia . For each gene , we built a corresponding leukemia model with a deficiency in that gene ( e . g . . FANS-deficient , ACACB-deficient , and CAD-deficient leukemia models ) . We built a total of 13 mutated enzyme deficient models for nine genes across nine cancer types that have feasible flux solution states with the deficient function of the mutated enzyme ( Figure S9 in Text S1 ) . Once the deficient models were built , we simulated flux changes that result from the enzyme deficiency by employing an MCMC sampling method to the enzyme deficient models and matching normal ( no cancer with non-enzyme deficient ) models . Substrates or products of reactions with significantly changed flux for deficient models as compared to normal models were chosen as potential oncometabolites . Finally , 15 unique metabolites catalyzed by the mutated enzymes and surrounded by significantly changing flux were predicted as context-specific LoF oncometabolites ( Figure 5A ) . Notably , previously known oncometabolites , succinate and fumarate , were predicted as potential oncometabolites due to the LoF of SDBH in gastric cancer ( Figure 5A , Table S7 in Text S1 ) [9]–[12] . As described in the example of IDH , mutated enzymes could possibly confer new catalytic activities by simultaneously changing the native reaction mechanism and/or catalyzing different substrates . In this study , we used a chemoinformatics approach to predict promiscuous catalytic activities of enzymes resulting from GoF mutations . This chemoinformatics approach has been shown to be useful in predicting enzymes promiscuity [30] , [31] , assigning Enzyme Commission ( EC ) number [32] , [33] , and analyzing reaction databases [34] . To predict candidate oncometabolites resulting from promiscuous activities of mutated enzymes , a systems framework was developed to determine new enzyme functionalities due to mutated enzymes . First , each substrate and product present in the mutated enzyme reactions were compare against the Human Metabolome Database ( HMDB ) [35] . In order to decrease computational efforts and select a manageable set of metabolites for biochemical reaction operators ( BROs ) simulation , Tanimoto coefficient cut off values were determined for each mutated enzyme reactions . Then , synthetic reactions were constructed by applying generic BROs to previously HMDB selected metabolites ( see Materials and Methods for details ) . Then , simulated synthetic reactions were compared against the reactions associated with the mutated enzymatic genes . To do this , Tanimoto coefficients of substrates and reactions were calculated between pairs of reactions . In order to emulate the GoF enzyme behavior , different Tanimoto coefficient cut-off values were imposed on each reaction according to the IDH GoF mutation case . Pairs of reactions with Tanimoto similarity scores of less than or equal to a specific cut-off were saved and identified as possible gain of function pairs ( see Figure S10 in Text S1for details ) . Finally , from these reaction pairs , structural features of oncometabolites were predicted . Among the 17 genes with recurrent GoF mutations , seven genes could have catalytic activity in cancer ( see Figure S11 in Text S1 for details ) . The final GoF mutated enzymatic gene list for this analysis contains seven unique genes . Among the reactions associated with the seven mutants , reactions catalyzing large molecules that could yield more than 30 , 000 promiscuous activities due to the compounds complexity were excluded from the analysis , and finally 33 reaction associations were identified from the enzymatic list . Notably , cofactors were not taken into account for the compound list generation . A summary of mutated enzymes associated reactions and their predicted promiscuity catalytic activities is given in Table S8 in Text S1 . For example , for the CAD gene which catalyzing L-glutamine , 2866 potential promiscuity catalytic reactions associated with 170 substrates and 1644 products were predicted in CAD mutant lung ( SCC ) cancer and gastric cancer ( Table S8 in Text S1 , Figure 5B , all detailed information of predicted promiscuity reactions is shown in the File S2 ) . In most cases promiscuous substrates show similarities with the native substrate [36] , [37] . We conducted compound similarity analysis in order to demonstrate dominant substructures of promiscuous substrates and products ( Table S8 in Text S1 ) . The compound structure shown in the Table S8 in Text S1 is the dominantly observed substructure of promiscuous substrates and products . Notable , several promiscuous substrates and products did not have dominant substrates . Finally , we identified 24 promiscuous compound substructures as features of GoF oncometabolites ( Figure 5B ) . In this study , we predicted potential oncometabolites in nine types of cancer by analyzing the massive scale genetic variants integrated with cancer and normal GEMs . We first predicted potential oncometabolites that could result from LoF mutations by simulating flux changes in the metabolic network due to the LoF of mutated enzymes . Second , we predicted oncometabolites that could result from GoF mutations by inferring promiscuous catalytic activities of enzymes resulting from their GoF mutations . Setting aside the generally accepted LoF criterion ( nonsense , splice-variants and frame-shift ) , we rarely have information about the mutational directionality of a given missense mutation whether it functions as GoF or LoF . Specifically in the oncometabolite analysis , enzymes can be disrupted in both ways . Whereas an LoF of an enzyme gives a predictable malfunctions ( e . g . an accumulation of a target product ) , a GoF mutation is much more unpredictable and is the very place in which an in silico analysis should be targeted . As in the example of IDH , IDH gained an unexpected catalytic activity that is result from missense mutations . Therefore , in this study , we focused on the possible metabolic perturbation of a GoF mutation in an enzyme can raise . Note that , assuming a mutation to be GoF is different from asserting the mutation is GoF . Using the aforementioned criteria , we predicted 15 oncometabolites resulting from the LoF mutations , and 24 substructures of oncometabolites resulting from the GoF mutations . These predictions can be used as a guide to examine select mutant enzymes and generation of oncometabolites . Notably , in our reconstructed models , several cancer and/or normal models do not uptake glucose or secrete CO2 in their optimal flux states ( optimal solution ) . In order to prevent this problem , we allowed small amounts of uptakes to important vitamins exchange reactions ( Table S4 in Text S1 ) . Also , we constrained models to uptake glucose , oxygen and secrete CO2 , biomass . With the updates , now most of models produce presumably reasonable uptake and secrete flux states ( see flux variation ranges in the Table S5 , S6 in Text S1 ) . Although still several models do not secrete any CO2 in the optimal flux states , this problem is not a critical problem in our study since the representative flux states in our results were determined from the sampling points within the suboptimal solution space ( 90% of optimality solution space ) that is the flux variation ranges shown in the Table S5 and S6 in Text S1 . Altered energy metabolism in cancer cells is accepted as a hallmark of cancer [2] , [5] . With the discovery of oncometabolites such as 2-HG , succinate , and fumarate that originate from mutations in key enzymes , alteration of metabolism is now considered to be a strong oncogenic factor . With the existence of oncometabolites established , there is clearly a great interest in determining if there are additional metabolites with oncogenic potential . Large-scale data sets available for a variety of cancers and genome-scale models of metabolism can be used to predict the existence of oncometabolites , and can contribute to therapies and biomarkers in cancer . Oncometabolites could be used to develop therapies and identify biomarkers associated with cancer . Recent studies showed that inhibition of mutant IDH1 delays growth of glioma cells and induces cellular differentiation in leukemia , which shows possibility of a potential application as a therapy for cancer [38] , [39] . Also , it has been reported that a breast cancer subtype with elevated level of 2-HG was associated with reduced survival . This study indicated that high levels of 2-HG may be a useful biomarker for breast cancer diagnosis and prognosis [40] . Thus , predicting existence of potential oncometabolites would be beneficial in cancer therapies and biomarker identification . In order to reconstruct cancer type specific metabolic models and their paired normal models , we used data sets of gene expression experiments collected from TCGA consortium and GEO Database which are composed of expression profiles of primary cancer cells and site-matched normal cells ( Table 1 ) . In order to maximize platform consistency , we focused on using Affymetrix and Illumina Hiseq RNA-seq platforms . The present/absent ( P/A ) calls of the Affymetrix gene expression were made using the ‘mas5calls’ function in the ‘affy’ package ( ver . 1 . 28 . 0 ) implemented in R ( ver . 2 . 15 . 0 ) , and a threshold of 10 read maps was used to define detection of the P/A calls of RNA-seq data at the gene level [41] . In our study , genes that are expressed in more than 99% of the total number of samples in a data set are finally determined as present ( expressed ) genes . The P/A call results of gene expression are then incorporated to the reactional space using Gene Inactivity Moderated by Metabolism and Expression ( GIMME ) algorithm implemented in COBRA Toolbox v2 . 0 [42] , [43] . For the medium condition , a standard RPMI-1640 condition was used in all simulations [21] , [22] . Although , same medium uptake rates were applied to different cancer and normal models , this is not likely to cause serious artifacts since the proposed method is based on differences of network structure of cancer vs . normal models as determined by gene expression P/A call results . As cancer cells are known to maximize their proliferation rate , the biomass formation was maximized in all cancer models [21] , [22] ( Formula 1 ) . However , normal cells may have different objectives depending on the growth signals [3] , thus we designed a new objective function which can cover the different cellular objectives in both proliferative and quiescent normal cells ( Formula 2 ) . Because normal cells have different , or multiple , physiological objectives ( i . e . , biomass or ATP formation ) , we constructed an objective function that was a linear combination of these physiological functions ( Formula 2 ) . Each reaction flux in the objective function was scaled by its maximum achievable flux in the given growth condition . The contribution of each component may be weighted with an additional coefficient ( a and b ) with a value of 0 denoting no contribution to the objective . Here , coefficient a and b are set to be 1 and 1 , respectively ( equal contribution of biomass and ATP ) . The distribution of feasible fluxes in the models used in this study was determined using Markov chain Monte Carlo ( MCMC ) sampling [44] , and was implemented with the COBRA Toolbox v2 . 0 [42] . Specifically , the objective function was provided a lower bound of 90% of the optimal growth rate as computed by flux balance analysis [45] . Thus , the sampled flux distributions represented sub-optimal flux-distributions , but still simulated fluxes relevant to cell growth and maintenance . MCMC sampling was used to obtain thousands of feasible flux distributions using the artificially centered hit-and-run algorithm with slight modifications , as described previously [46] , [47] . As a result , a mixed fraction of approximately 0 . 50 was obtained , suggesting that the space of all possible flux distributions is nearly uniformly sampled . For each reaction , a distribution of feasible steady-state flux values was acquired from the uniformly sampled points , subject to the network topology and model constraints . Similar measures were taken for all other models in this work . To simulate changes in reaction flux occurring in a shift between cancer and normal , the sampled fluxes for each reaction were compared between cancer and normal as follows . First , reactions that carried no flux in both conditions or that were involved in loops [48] were removed and not used in further analysis . Next , flux magnitudes were normalized between each pair of media conditions . To do this , the flux value of each sample point was divided by the sum of all flux magnitudes for the sample point ( Formula 3 ) . ( 3 ) Once the flux values were normalized , the changes of fluxes between two conditions were determined as previously described [47] . Briefly , differential reaction activity was determined by assuming that a reaction is differentially activated if the distributions of feasible flux states ( obtained from MCMC sampling ) under two different conditions do not significantly overlap . For each metabolic reaction , a P-value was obtained by computing the probability of finding a flux value for a reaction in one condition that is equal to or more extreme than a given flux value in the second condition . The significance of P-values was adjusted for multiple hypotheses ( FDR ≤ 0 . 01 ) . The approach presented here extends the pre-published method called Flux Space Shift analysis ( FSS ) [49] . FSS utilizes MCMC sampling of the metabolic solution space to compute the distribution of all possible steady-state fluxes an enzymatic reaction can carry in a cell in a given growth condition . For each reaction , a P-value is computed from the distributions of possible fluxes for the reaction in a cancer cell and its matched normal cell . This P-value represents the probability of choosing a flux value from a reaction in a cancer cell that is also within the distribution for that reaction in a normal cell . The P-values are then corrected for multiple hypotheses , and the list of reactions that show significantly different fluxes for cancer vs . normal is returned , along with the direction of the change in magnitude ( up or down ) . All significantly changed fluxes are then decomposed into a list of genes that help to catalyze the reactions using the gene-protein-reaction associations in the model . Through this , one can obtain lists of genes that are expected to be up-regulated or down-regulated . Here , in order to minimize the ambiguity , genes that are associated both with reactions that increase and other reactions that decrease were removed , and reactions catalyzed by more than one isoenzyme were also filtered out from the analysis since not all isozymes are necessarily to change their gene expression level . Once the lists of genes that are expected to be up- or down-regulated are gathered , the predictions were compared to the actual gene expression changes . The accuracy of the prediction was demonstrated by the ratio of the number of correctly predicted genes divided by the number of total predictions ( Formula 4 ) . The significance of the accuracy is demonstrated by a comparison with the background distribution using 10 , 000 random accuracy tests . ( 4 ) ( TP: True positive , TN: True negative , FP: False positive , FN: False negative ) Accuracy of reconstructed cancer specific models was assessed using experimentally determined essential genes . We collected a list of cancer essential genes from [50] . Ultimately , 14 cancer cell essential metabolic genes which were reported as common essential genes across 13 cancer cell line were used . We performed in silico single gene deletion tests on each model , and genes whose deletion effect reduces the maximum objective reaction by more than 10% were determined to be in silico essential genes . Finally , as the total number of predicted in silico essential genes varies according to the models , the accuracy of the models were evaluated with P-values from the hypergeometric enrichment tests of experimental essential genes against the in silico essential genes . For synthetic reaction construction , we first defined a set of 374 irreversible generic biochemical reaction operators ( BROs ) that has been used in previous studies , mostly for prediction purposes for metabolic engineering [51]–[53] , enzyme promiscuity analysis [31] , [54] , and xenobiotics degradation [55] . These BROs can represent approximately 75% of enzymes present in the KEGG Database and 72% of BRENDA EC numbers . Essentially , a BRO is constructed based on the smallest substructure representing the structural changes of substrates and products in a specific reaction . Each BRO is related to specific cofactors and a third-level EC number for further reaction reconstruction and identification . For BRO representation , we used SMIRKS [56] , a language used for describing generic reactions ( transformations ) by using SMARTS [56] representation of the reaction's substructures . A SMARTS pattern may include not only a specification of reaction center , but also a specification of a local structure that must occur or is necessarily absent based on our best understanding of the relevant biochemistry [57] . After BRO construction , we generate all possible reactions that may occur and every compound that may be produced given the previous selected list of human metabolites . Then , specific cofactors were assigned , and for filtering purposes mass balance was performed . For each mutated enzyme , Maximum Common Substructure ( MCS ) analysis was used in order to identify the most common chemical patter from all predicted products and substrates . The idea is to identify from a range of chemical structures , the largest substructure common in all of them . For all calculations , with regard to handling compounds , building reaction , substrate fingerprint generation , reaction fingerprint generation , TC dissimilarity calculations , BROs simulations , and MCS analysis we used MATLAB linked with ChemAxon's package libraries , specifically Marvin , JChem Base , Standardizer and Reactor [ ( ChemAxon , Budapest , Hungary , www . chemaxon . com ) ] .	Cancer and metabolism have been considered to be associated for a long period since Otto Warburg observed that tumor cells consume glucose and convert most of it to lactate , despite the presence of oxygen . However , the role of the Warburg effect in oncogenesis had been under doubt because no solid links between genetic variations in metabolic genes and cancer had been observed until recent days . When with the development of sequencing technologies , researchers found mutations in IDH1 , IDH2 ( isocitrate dehydrogenase ) in medium-grade glioma and acute leukemia . As these mutated metabolic genes initiate unexpected enzymatic reactions , cancer cells show altered concentration of particular metabolites , here called “oncometabolites” . The oncometabolites regulate the epigenetic controls of cell differentiations . In this study , we predict potential oncometabolites that might originate from loss or gain-of-function mutations in nine types of cancer from massive scale cancer mutation data with a systems biology approach .	[ "Abstract", "Introduction", "Results", "Discussion", "Materials", "and", "Methods" ]	[ "systems", "biology", "biology", "and", "life", "sciences", "computational", "biology" ]	2014	A Systems Approach to Predict Oncometabolites via Context-Specific Genome-Scale Metabolic Networks
African trypanosomiasis is a severe parasitic disease that affects both humans and livestock . Several different species may cause animal trypanosomosis and although Trypanosoma vivax ( sub-genus Duttonella ) is currently responsible for the vast majority of debilitating cases causing great economic hardship in West Africa and South America , little is known about its biology and interaction with its hosts . Relatively speaking , T . vivax has been more than neglected despite an urgent need to develop efficient control strategies . Some pioneering rodent models were developed to circumvent the difficulties of working with livestock , but disappointedly were for the most part discontinued decades ago . To gain more insight into the biology of T . vivax , its interactions with the host and consequently its pathogenesis , we have developed a number of reproducible murine models using a parasite isolate that is infectious for rodents . Firstly , we analyzed the parasitical characteristics of the infection using inbred and outbred mouse strains to compare the impact of host genetic background on the infection and on survival rates . Hematological studies showed that the infection gave rise to severe anemia , and histopathological investigations in various organs showed multifocal inflammatory infiltrates associated with extramedullary hematopoiesis in the liver , and cerebral edema . The models developed are consistent with field observations and pave the way for subsequent in-depth studies into the pathogenesis of T . vivax - trypanosomosis . African trypanosomiasis , one of the most neglected diseases , consists of a number of important human and animal pathologies caused by parasitic protists of the order Kinetoplastida . Human African Trypanosomiasis ( HAT ) , or sleeping sickness , and animal trypanosomosis , or Nagana , are vector-borne diseases , that are primarily cyclically transmitted by tsetse flies . HAT is a major public health problem in 35 sub-Saharan countries . The related animal challenge , caused by several species , i . e . Trypanosoma vivax , Trypanosoma congolense and to a lesser extent to Trypanosoma brucei brucei causes about 3 million deaths annually in cattle and has a marked impact on African agriculture , causing annual livestock production losses of about US$ 1 . 2 billion . T . vivax accounts for up to half of total Trypanosoma prevalence in West Africa where it is considered the major pathogen for livestock and small ruminants [1] , [2] , [3] . Outside tsetse endemic areas , West African T . vivax isolates were introduced long ago into South American countries where it represents a real threat since it can be efficiently transmitted across vertebrate hosts by mechanical means and by various biting flies and tabanids [4] , [5] , [6] . The severity of the disease depends on parasite strain , endemicity and host species , but the key steps in the T . vivax - host interactions are still largely unknown . Several pieces of evidence point to the importance of host genetic factors in determining individual susceptibility and/or resistance to this infection [3] , [7] , [8] , [9] , [10] , [11] . Trypanotolerance is defined as the ability demonstrated by cattle of different genetic backgrounds to control trypanosomosis [12] , [13] . It has previously been reported that increased bovine resistance to trypanosomosis is associated with more control over parasitemia and related anemia , two of the main pathogenic effects of trypanosome infections [14] , [15] . However , dissimilar courses of the infection may be due to genetic polymorphism and to the virulence of the parasite isolates , thus leading to moderate , progressive and/or lethal pathologies and therefore affecting mortality rates [5] , [6] , [7] . It is widely accepted that if trypanosomosis is to be successfully treated in the field , a number of parameters must be taken into account , including the seasonal trypanosome prevalence and vector abundance , the severity of the disease , the magnitude of the anemia , the stock nutritional state and the prescription of an appropriate trypanocidal drug [6] , [16] , [17] , [18] . However , the antigenic complexity of trypanosomes , their ability to expose a variety of genetically-controlled surface coat antigens ( VSG ) , and the diversity of the immune responses presented by unrelated hosts [19] , [20] , [21] , call for the discovery of new parasite genetic markers and more in-depth knowledge of host trypanotolerance mechanisms . Several early studies were conducted in more affordable mouse or rat experimental models of infection in attempts to throw light on trypanotolerance , antigenic variation , the pathogenesis of intravascular coagulation , and T . vivax immunobiology and dynamics [7] , [11] , [19] , [22] , [23] , [24] , [25] . However , these studies used a variety of more or less virulent isolates from cattle , goats , sheep , horses and donkeys to explore the ability of T . vivax stocks to infect several intact or immunosuppressed mouse strains . Although these studies had a huge impact on research into T . vivax , the diversity of the results they yielded and the difficulties encountered in establishing axenic parasite cultures or reliable in vivo infections that entirely resemble natural infections [26] , constrained the work performed with these models . In consequence , more than 20 years ago , while biological investigations into VSG and the identification of serodemas were usual for more than a few trypanosomes of the Trypanozoon subgenus , studies on T . ( Dutonella ) vivax VSG molecules and structure of the coat were just been encouraged [26] . Research into T . vivax then focused on characterizing parasite surface proteins or comparing genetic diversity of Western to Eastern African parasite stocks and more recently on analyzing population clonality , [10] , [27] , [28] , [29] , [30] , [31] , [32] , [33] , but somehow neglected the further development of suitable rodent models . Now , and in an attempt to circumvent the major constraints inherent to studying T . vivax/host interactions in the field and data inconsistencies arising from the difficulties encountered in the past , we have developed in vivo murine models of trypanosomosis using a T . vivax isolate known to maintain infectivity to rodents [23] . Here we show that this T . vivax isolate retains its original characteristics after several years of cryopreservation . The parasites can grow , multiply and be transmitted in vivo following predictable kinetics in the peripheral blood of different mouse strains selected for their susceptibility or resistance to different parasite inocula . Sustained and reproducible infections are obtained that successfully mimic the dynamics of the parasitological , histological and pathological features of the infection and closely resembling those observed for cattle trypanosomosis in the field . We have thus developed reliable mouse research models that can be used to elucidate the immunopathological mechanisms involved in T . vivax infection and associated disease . It is worth noting that T . vivax was recently shown to express a functional gene involved in the non specific polyclonal activation of host B cells and that this gene is absent in more widely studied T . brucei and T . congolense [34] . Furthermore , the work presented here is expected to be a useful and complementary tool for the further studies of T . vivax immunobiology and will thus provide valuable information about trypanotolerance , Trypanosoma evasion strategies from host immune system , and immunopathogenesis . Trypanosoma ( Dutonella ) vivax stabilates ( STIB 731-A ) , cryopreserved on September 25 , 1996 after 9 passages in mice , were kindly provided by R . Brun ( Swiss Tropical Institute , Basel , Switzerland ) . STIB 731-A stabilates were originally prepared in November 1982 using bloodstream forms of IL 1392 T . vivax stock obtained from the blood of goat #M918 , at ILRAD ( ILRI ) , Nairobi , Kenya . This West African IL 1392 goat stock was derived from the Zaria Y486 Nigerian isolate of a naturally infected Zebu steer maintained by 62 serial passages in mice [23] . VSG ILDat 1 . 2 ( ILRAD Duttonella antigen type 1 . 2 ) specific primers were deduced from the VSG ILDat 1 . 2 full length sequence ( TvY486_0004810 variant surface glycoprotein putative , 1215 bp ) obtained from the GeneDB of the Zaria Y486 Trypanosoma vivax nuclear genome ( Sanger Institute Pathogen Sequencing Unit ( PSU ) , http://www . sanger . ac . uk/Projects/T_vivax/ ) : VSG-1 . 2F ( 5′ AATTTTGGTGAGTGTCGGTGT 3' ) and VSG-1 . 2R ( 5' ATTTCCTCCACCACGTAGCTC 3' ) . T . vivax- specific forward and reverse ribosomal promoter primers were also deduced from the Zaria Y486 chromosome 3: TvrDNAF ( 5' CTGATTTCGCCACTGCTATTATTTGC 3' ) and TvrDNAR ( 5' CGCTTCACTTGATGATCGTTTCG 3' ) , respectively . Parasites were maintained by weekly passages in mice and new stabilates were appropriately and regularly frozen in polysoma buffer/glycerol , as previously described [11] . Blood smears were prepared from infected mouse blood , air dried , fixed in methanol for 5 minutes and further stained with 5% Giemsa for 20 minutes . Seven to 10-week-old male BALB/c ( H2d ) , C57BL/6 ( H2b ) or Swiss outbred ( CD-1 , RJOrl:SWISS ) mice ( Janvier , France ) were used in all the studies . Mice were injected intraperitoneally with 101–105 bloodstream forms of T . vivax obtained at the peak of parasitemia ( day 8 post infection ) . For parasite enumeration , five microliters of blood were harvested individually from the tail vein and appropriately diluted in buffered saline when necessary . Blood parasite counts were established under a light microscope and expressed as number of parasites per milliliter of blood . All animal work was conducted in accordance with relevant national and international guidelines ( see here below ) . All mice were housed in our animal care facilities in compliance with European animal welfare regulations . The Institut Pasteur is member of the Committee #1 of the Comité Régional d'Ethique pour l'Expérimentation Animale ( CREEA ) , Ile de France . The Animal housing conditions and protocols used in the present work were previously approved by the “Direction des Transports et de la Protection du Public , Sous-Direction de la Protection Sanitaire et de l'Environnement , Police Sanitaire des Animaux” under the number B 75-15-28 accordingly to the Ethics Chart of animal experimentation which includes appropriate procedures to minimize pain and animal suffering . PM has permission to perform experiments on vertebrate animals #75-846 issued by the Department of Veterinary Services of Paris , DDSV and is responsible for all the experiments and protocols carried out personally or under her direction in the framework of laws and regulations relating to the protection of animals . 50 µl of retro-orbital blood were recovered onto 0 . 5 M EDTA . Samples were analyzed in a Scil Vet abc ( Scil , Strasbourg , France ) using pre-established and normalized parameters for the different mouse strains . Peripheral reticulocytes were counted as described [35] , modified by S . Bagot ( personal communication ) . Briefly , 5 µl of blood were fixed in 1 ml of 0 . 25% glutaraldehyde in PBS pH 7 . 4 and further stained with 1 µM Hoechst 33258/Thiazole orange 0 . 1 µg/ml in PBS pH 7 . 4 for 1 h at 37°C . Twenty days after infection , mice were anesthetized with an i . p . injection of 0 . 1 ml per 10 g mouse body weight of a solution containing 1 mg/ml xylazine ( Rompun 2% , Bayer , Leverkusen , Germany ) and 10 mg/ml ketamine ( Imalgène 1000 , Merial , Lyon , France ) , and then sacrificed by cervical dislocation . After a complete post-mortem examination , the spleen , liver , kidneys , lung , heart and specimens of the central nervous system were removed and immediately fixed in 10% neutral-buffered formalin . Tissue samples from these organs were embedded in paraffin; five-micrometer sections were cut and stained with hematoxylin and eosin ( HE ) . All the experiments were performed two or three times using at least 5 mice per experimental group and per time point . Mice were analyzed individually and the differences between the groups used in this study were tested for statistical significance using Student's test or the Log-rank ( Mantel-Cox ) test whenever appropriate ( Prism software , Graph Pad , San Diego , CA ) . Data are expressed as arithmetic means are presented as arithmetic means +/− the standard deviation ( SD ) of the means . The IL 1392 West African stock of T . vivax is derived from the Nigerian isolate Zaria Y486 [23] which is infective for rodents and can be cyclically and/or mechanically transmitted [36] , [37] . Rodent-infective derived clones of Y486 T . vivax , notably the IL 1392 , have already been shown to express VSG ILDat 1 . 2 ( ILRAD Duttonella antigen type 1 . 2 ) [19] , [28] , [38] , [39] in a relatively stable fashion . This VSG can be readily recognized by its specific 20 amino acid N-terminal sequence ( ANNFAETDMEGVCTGALTLR ) [30] , [31] . As can be seen in Figure 1A , VSG-1 . 2 forward and VSG-1 . 2 reverse oligonucleotide primers were deduced from the full length ILDat 1 . 2 gene sequence and flanking this 20 amino acid specific sequence . PCR reactions were then used to ascertain the identity of the initial IL 1392 T . vivax stabilate used in the present work , as previously shown [30] , [31] . VSG-1 . 2F and VSG-1 . 2R primers amplified a 148 bp fragment of genomic DNA in IL 1392 bloodstream forms , and when sequenced , this showed more than 99% similarity with ILDat 1 . 2 and only 2 point mutations as compared to Zaria Y486 ( See Fig . 1B ) , confirming the presence of ILDat 1 . 2 VSG from the rodent infective West African T . vivax 1392 isolate , not reactive with Eastern T . vivax strains or with DNA from T . brucei and T . congolense [27] , [28] , [40] It is also worth noting that a PCR reaction comprising TvDNAF ( forward ) and TvDNAR ( reverse ) oligonucleotides amplified a 1 , 8 kb DNA product whose sequence was flanked by the two primers and presented 95% homology to the highly specie-specific ribosomal promoter of the Y486 T . vivax reference strain ( not shown ) . Furthermore , we recently showed that the IL 1392 T . vivax genome possesses a functional proline racemase gene ( TvPRAC ) that is absent in other trypanosomatid genomes [34] . Altogether these results established and confirmed the molecular identity of the IL 1392 T . vivax parasites used in the present work . IL 1392 T . vivax bloodstream forms readily infected all mouse strains tested and were regularly maintained hereafter in the laboratory without losing infectivity through weekly passages in 7- to 8-week-old outbred Swiss ( outbred ) mice ( CD-1 , RJOrl:SWISS ) ( Janvier , France ) by intra-peritoneal ( i . p . ) injection of 103 parasite forms . As can be seen in Figures 1C and 1D , the parasites showed a predominantly slender morphology , an anterior free flagellum and a narrow posterior end containing a large sub terminal kinetoplast , similar to stained trypanosomes from cattle , as previously described [10] , [23] . Figure 1E shows large numbers of T . vivax in blood , at the peak of parasitemia in outbred mice . Despite the high degree of gene synteny observed in kinetoplastids , genes coding for essential proteins associated with key metabolic reactions are not necessarily ubiquitous among members of the order . Accordingly , T . vivax but not Trypanosoma brucei , Trypanosoma congolense nor Leishmania spp possesses a TvPRAC enzyme responsible for the interconversion of L- and D-proline enantiomers [34] . This enzyme , earlier described in Trypanosoma cruzi parasites , was shown to be essential for parasite metabolism and triggers non-specific polyclonal B cell responses in the host thus contributing to mechanisms of parasite escape from the host immune system [41] , [42] . Taking into account the fact that T . vivax multiplies extracellularly in the host bloodstream , unlike the intracellular and extracellular T . cruzi , it is conceivable that TvPRAC may also play a role in triggering non specific polyclonal B cell responses , contributing to antibody diversity , host immunosuppression , parasite evasion and persistence . In an attempt to further address these questions , we decided to develop a reliable and consistent mouse model and for this purpose studied several parasitological , hematological and immunological parameters of the infection using a parasite stock of defined antigenic identity . The experimental murine infection was initially studied using 7 to 8-week-old intact BALB/c inbred mice infected with different inocula ( 101 to 105 ) of ILRAD 1392 T . vivax bloostream forms . The initial results showed that appearance of parasitemia was highly dependent upon the number of parasites injected as parasites could be detected as early as two days post-injection when a high inoculum ( 105 ) was used , while three to six days were necessary when lower parasite numbers were injected ( 101 to 104 ) ( not shown ) . Death seemed to correlate with parasite load since average time to death was also dependent on the number of parasites in the inoculum ( not shown ) . Thus , the more elevated the parasite inoculum , the lower the survival rate , corroborating published data based on the original parasite isolate [43] . In order to compare the impact of host genetic background on the establishment of infection , we then conducted studies using BALB/c ( H2d ) , another inbred mouse strain ( C57BL/6 ) which bears a different haplotype ( H2b ) , and an outbred mouse stock , all infected with an inoculum consisting of 102 bloodstream forms of T . vivax . BALB/c mice showed a rapid and pronounced increase in parasitemia that reached 4 . 108 parasites/ml , as recorded by daily monitoring ( Fig . 2A ) . As compared to BALB/c and C57BL/6 , detectable parasitemia ( 4–6 days post infection - d . p . i . - , ≥104 parasites/ml ) was slightly delayed following infection of the outbred mice ( Figures 2B and 2C ) . While parasitemia in all three mouse strains reached maximum levels 6 to 8 days post-infection , survival rates were significantly higher in the C57BL/6 and outbred mice than in the BALB/c mice which died in the first week of infection ( Figure 2D ) . Moreover , while outbred mice showed a parasitemia plateau after 10 days of infection , recurrent parasitemia peaks were observed in the C57BL/6 mice over the same period , as also observed by De Gee et al . [7] and Mahan & Black [19] , indicating that the C57BL/6 mice were partially controlling the parasite load . Since BALB/c mice proved to be highly susceptible to the infection , we continued our studies using only C57BL/6 and outbred mice as these were able to endure the infection over a longer period of time . Microscopic examination of the peripheral blood of infected animals indicated an apparent loss of red blood cells , concomitant with high levels of parasitemia . To monitor this phenomenon , peripheral blood samples taken from individual mice were analyzed throughout the infection and subjected to hematological analysis . Complete blood counts showed similar and severe changes in both mouse strains ( Figure 3 ) . Firstly , hemoglobin concentrations were significantly decreased in both C57BL/6 ( from 13 . 3±0 . 1 to 5 . 7±1 . 0 g/dl ) and outbred mice ( 14 . 7±0 . 2 to 7 . 2±0 . 6 g/dl ) . This decrease was associated with a fall in the red blood cell counts ( from 8 . 5±0 . 1 to 3 . 7±0 . 8 106 cells/mm3 and from 9 . 0±0 . 1 to 4 . 3±0 . 4 106 cells/mm3 in the C57BL/6 and outbred mice , respectively ) and in hematocrit values ( from 42 . 7±0 . 5 to 20 . 6±3 . 6% and from 49 . 6±0 . 9 to 25 . 4±2 . 4% in the C57BL/6 and outbred mice , respectively ) ( Fig . 3A , 3B and 3C ) . Taken together , these alterations indicated that the infection gave rise to severe anemia as reported for natural cases of bovine trypanosomosis caused by T . vivax [6] , [44] . An evaluation made to measure immature red cell production in the blood showed a transient 5-fold increase in the number of reticulocytes 14 days post-infection ( data not shown ) . Although only 20% of the injected animals were still alive 20 days post-infection , these results suggest that , at the time of death , the mice were suffering from regenerative , normocytic and normochromic anemia . Also , severe thrombocytopenia reported as an universal complication in the course of trypanosome infections [45] was observed in both the outbred and C57BL/6 mice , stemming from a dramatic fall in the platelet count as early as seven to ten days post-infection ( Figure 3D ) . In addition , and except during the initial increase in white blood cell counts seen in the first days of the infection , a leucopenia was observed as the infection progressed and was more pronounced in the C57BL/6 mice ( Figure 3E ) . The number of circulating lymphocytes fell significantly during the second week of infection , more precisely at around 20 d . p . i for the outbred mice and was accompanied by an increase in neutrophils and monocytes , as previously described with another Y-486 derived strain of T . vivax and albino mice [22] ( see accompanying paper ) . The extent of the tissue damage caused by T . vivax infection was assessed by means of anatomic pathological and histopathological examinations . Here , we chose to use the highly reproducible outbred model as it gave lower inter-individual differences within the mouse groups and sustained and elevated parasitemia . Outbred mice were infected with 1×102 parasites and a general anatomic pathological assessment of disturbances was conducted 20 days post-infection . At necropsy , gross lesions were observed only in the spleen and liver; the other organs were macroscopically normal . The spleens were uniformly enlarged , thereby characterizing marked splenomegaly , but did not show any congestion . They were firm and a little blood oozed from the cut surfaces . Randomly scattered white or red foci , ranging from 1 to 5 mm in diameter were observed on the capsules and the cut surfaces . Livers showed discrete , pale red or sometimes white foci that were sharply delineated from the adjacent parenchyma . These foci varied in size from 0 . 5 to 2 mm . The tissues of outbred mice infected with T . vivax were also subjected to histopathological analysis 20 days post-infection . Lymphoid and non-lymphoid organs showed significant lesions ( Figure 4 ) . The spleens of infected animals ( Figures 4A to 4D ) , showed diffuse lesions , more diffuse at the periphery of the organ , involving both the red and white pulps ( Figure 4A ) . Lesions in the red pulp were characterized by necrosis with replacement of the normal tissue by an acidophilic and amorphous to fibrillar material containing cell debris , fibrin , extravasated erythrocytes and trypanosomes that were often clustered together in hemorrhagic foci ( Figure 4B and inset arrowhead ) . Extramedullar hematopoiesis foci were reduced in number ( Figure 4C ) . The white pulp showed disorganized lymphoid structure associated with marked infiltration by activated macrophages displaying vesiculous , euchromatic and nucleolated nuclei and abundant acidophilic cytoplasm . Infiltration by several lymphocytes and plasma cells was also noted ( Figure 4D ) . Numerous plasma cells called ‘Mott cells’ , characterized by their round shape and a polar cytoplasm containing stored immunoglobulins ( Russel bodies ) , were observed in both the red and white pulps ( Figure 4D , arrow ) . Collectively , these lesions were characteristic of diffuse , sub acute , necrotizing and hemorrhagic splenitis , associated with intralesional trypanosomes . A multifocal to coalescing lesion , primarily centered on the portal tracts but also involving centrilobular veins , was observed in the liver ( Figure 4E ) . As can be seen , the lesion was characterized by marked infiltration of plasma cells , lymphocytes and macrophages ( Figure 4F ) . Trypanosomes were frequently observed in the vascular spaces , i . e . sinusoids and the portal and terminal hepatic veins . Many necrotic foci were also observed , randomly distributed in the liver parenchyma and associated with hemorrhages and trypanosomes ( Figure 4F , inset arrowhead ) . A very high density of extramedullar hematopoiesis foci was noted in the liver sinusoids ( Figure 4G ) . Collectively , these lesions were characteristic of multifocal to coalescing necrotizing and hemorrhagic hepatitis , associated with extramedullary hematopoiesis and intralesional trypanosomes . The infection also induced bilateral , multifocal and sub acute tubulointerstitial nephritis , as can be seen in Figures 4H to 4J . The kidneys showed bilateral multifocal lesions , mostly involving the renal cortex and characterized by interstitial perivascular and periglomerular infiltration of plasma cells , lymphocytes and macrophages ( Figures 4H and 4I ) . Very few neutrophils were seen in this lesion . Some randomly distributed tubular epithelial cells were noted , showing acidophilic cytoplasm and a condensed hyper basophilic ( pycnotic ) and/or fragmented nucleus ( necrotic cells ) . Trypanosomes were observed in the blood vessels , mostly in the arcuate arteries at the corticomedullary junction ( Figure 4J , star and 4J , inset arrowhead ) . The histopathological investigation of the central nervous system also discovered multifocal lesions centered primarily on small and medium-sized veins , and more severe in the cerebellum ( Figure 4K ) . Evidence was often noted of dilatation and filling of the blood vessel lumen by erythrocytes , proteins and numerous trypanosomes . Vasogenic edema was observed , characterized by the presence of an amorphous and unstained material accumulated in perivascular spaces . Some angular and shrunken neurons close to these lesional blood vessels contained acidophilic cytoplasm and a condensed hyper basophilic nucleus , characteristic of ischemic necrosis ( Figure 4L ) . Trypanosomes were also seen in the meningeal blood vessels ( Figure 4M , star and inset , arrowhead ) . Interestingly , and although observed in only a few animals , moderate lymphoid hyperplasia was noted in the lymph nodes , apparently associated with intravascular or intrasinusal trypanosomes , as previously described in ruminants [46] ( not shown ) . In addition , histopathological examination of the heart , revealed the presence of numerous parasites in the ventricular cavities as well as in the blood vessels located in the periphery of the myocardium in some mice . Some of these lesions were accompanied by an infiltration of mononuclear cells ( multifocal myocarditis ) , i . e . plasmocytes , lymphocytes and macrophages . These findings are suggestive of a myocardial commitment induced by the infection and are consistent with the congestive heart failure that has previously been reported for cattle trypanosomosis [26] , [47] , [48] . While Human African Trypanosomiasis ( HAT ) has drawn the attention of many research groups over the last three decades , lesser consideration has been given to animal trypanosomosis ( Nagana ) despite its considerable impact on the development and fertility of livestock and the economical hardship it causes in several countries . Most studies , both in the distant past and more recently , have concentrated on analyzing the genetic factors involved in tolerance to trypanosomosis , or on describing the general deregulation of the immune response as expressed by a few individuals in different cattle species in the field [13] , [49] , [50] , [51] , [52] , [53] , [54] , [55] , [56] . For instance , T . brucei , which is of little clinical importance in livestock , has generally constituted the parasite of choice in experimentally and genetically controlled studies [57] , [58] , [59] , [60] . But trypanosomosis , which is overwhelmingly the most prevalent cattle illness in Africa and South America , is mainly caused by T . vivax and T . congolense . The hallmark of their pathogenesis in the field is severe anemia accompanied by a general immunosuppressive condition [6] , [21] . Since various types of tissue damage have been described for ruminants , horses , sheep and goats , distinct strategies are used to explain host resistance ( “tolerance” ? ) or susceptibility to trypanosomes and/or the etiology of the lesions and tissue damage observed [20] , [25] , [61] , [62] , [63] . To overcome these difficulties , several experimental models were developed in rats and mice ( see [21] for a review ) . Athough these studies showed that the mouse was potentially an important tool in understanding the pathogenesis of trypanosomosis and most particularly the immunobiology of host-parasite interactions , most subsequent studies focused on trypanosomes of the subgenus Trypanozoon ( i . e . T . brucei ) leaving T . ( Duttonella ) vivax infections poorly characterized . Thus , despite the fact that remarkable progress was made , the array of features shown by diverse stocks of T . vivax isolates has not painted a clear picture of the factors that could be central to the development of appropriate immuno ( chemo ) therapies [15] , [64] . To better investigate the relationship between trypanosomosis and genetically-controlled rodents , we therefore undertook to develop new mouse models of Trypanosoma ( Dutonnella ) vivax infection . This parasite not only differs from other trypanosomes belonging to the Trypanozoon subgenus ( i . e . T . brucei and T . equiperdum , but not T . evansi ) with regard to its transmission and tissue distribution in the host , but also is generally recognized as possessing diverse isolates which may or not express the ability to infect laboratory rodents [10] , [23] , [36] , [65] . For instance , East African isolates are known to induce mild infections and hemorhagic syndromes in cattle and only some stocks are adapted to rodents . Conversely , West African T . vivax isolates , obtained at different stages of natural infection , are responsible for the majority of trypanosomosis cases in cattle and other ruminants and may express mild , intermediate or high virulence to mice [7] , [65] . The work presented here used a well characterized West African T . vivax isolate which is infective to rodents ( IL 1392 ) [19] , [23] , [28] , [31] , [38] , [39] and describes in detail the parasitological , hematological and histopathological parameters of the infection in different inbred and outbred mouse strains . IL 1392 was chosen for its stable expression of VSG ILDat 1 . 2 , characteristic of rodent-adapted West African T . vivax isolates that together with closely related South American stocks pertain phylogenetically to the same clade [66] . Initially , our studies showed that the IL 1392 isolate , retained its infective characteristics and mouse infection profile after a long cryopreservation period ( see Material and Methods ) [11] , [22] . In addition , the experimental mouse models used in this work proved easy to handle on infection and reflected the general characteristic features observed in livestock , namely the remodeling of secondary lymphoid organs , cardinal severe anemia , genetically-related differences in resistance to the parasite ( but not to death ) and the development of multifocal tissue hemorrhages , necrosis and consequent systemic pathologies . Briefly , the work confirmed widespread observations that BALB/c mice are highly susceptible to T . vivax infection with the shortest survival time as compared to C57BL/6 or outbred mice . Furthermore , the BALB/c but also C57BL/6 and outbred experimental mouse models showed an early exponential increase in parasitemia , closely resembling acute trypanosomosis in the field [46] . Regardless of the fact that C57BL/6 and outbred mice proved to be more “tolerant” than BALB/c to T . vivax , limiting the pathological consequences of the infection and delaying mortality , all the mice presented common pathognomonic signs of the disease , such as anemia , cachexia , thrombocytopenia associated with high parasitemia and a tendency to leucopenia in the terminal stages [22] , [45] , [67] . As frequently observed in cattle and goats infected by T . vivax , the histopathological analysis of mouse tissues committed by the infection showed accumulation of trypanosomes nuclei and debris in the blood vessels of mouse spleen , liver and brain . Some extravascular foci associated with intense inflammatory and degenerative tissue disorders and cerebral edema were also observed [5] , [22] , [68] . It is noteworthy that both the red and white pulps of the spleen showed severe necrosis , with germinal centers depleted of lymphocytes , which could explain the lymphopenia observed at late stages in the infection . The marked splenitis , hepatitis and central nervous system involvement with vasogenic oedema and ischemia , reflect to what extent the model is of value in studies of T . vivax pathogenesis . The characteristic erythrocytopenia experienced at the peripheral level is suggestive of a decreased erythropoiesis . However , we cannot rule out possible extravascular haemolysis , or more specific apoptosis due to an autoimmune process triggered by the infection . A detailed study of bone-marrow cell populations before and during the infection will better address these questions ( see accompanying paper ) . The present systematic analysis described here using different mouse strains infected with a well characterized West African T . vivax isolate showed that these models are reliable and constitute new experimental tools for the study of trypanosomosis . The studies conducted aimed to develop models that could be used in the future to gain further insight into the genetic mechanisms involved in drug resistance and the discovery of new drug targets for purposes of parasite control . However , further studies using Eastern or other Western T . vivax stocks and the present mouse strains are encouraged to better approach the influence of genetic parasite divergences on disease outcome . These models of infection will be useful in the in vivo testing of new chemicals against T . vivax trypanosomosis since tests in the definitive hosts are prohibitively expensive . Finally , the mouse models given herein provide a description of the parasitological , hematological and histopathological features of T . vivax infection and pave the way to a more in-depth understanding of the immune responses involved in disease tolerance and susceptibility to T . vivax .	While most research efforts have focused on T . b . brucei trypanosomosis , infections caused by T . vivax and T . congolense which predominate in livestock and small ruminants have been subject to little study . In order to circumvent the major constraints inherent to studying T . vivax/host interactions in the field , we developed in vivo murine models of T . vivax trypanosomosis . We show here that the mouse experimental model reproduce most features of the infection in cattle . More than reflecting only the main parasitological parameters of the animal infection , the mouse model can be used to elucidate the immunopathological mechanisms involved in parasite evasion and persistence , and the tissue damage seen during infection and disease . Studies planned for the future will allow us to further investigate T . vivax–induced immunopathology in an experimental context for which all the necessary tools are now available .	[ "Abstract", "Introduction", "Materials", "and", "Methods", "Results", "Discussion" ]	[ "immunology/leukocyte", "development", "infectious", "diseases/neglected", "tropical", "diseases", "immunology/immune", "response", "immunology/genetics", "of", "the", "immune", "system", "infectious", "diseases/protozoal", "infections" ]	2010	Trypanosoma vivax Infections: Pushing Ahead with Mouse Models for the Study of Nagana. I. Parasitological, Hematological and Pathological Parameters
Bacteria encounter sub-inhibitory concentrations of antibiotics in various niches , where these low doses play a key role for antibiotic resistance selection . However , the physiological effects of these sub-lethal concentrations and their observed connection to the cellular mechanisms generating genetic diversification are still poorly understood . It is known that , unlike for the model bacterium Escherichia coli , sub-minimal inhibitory concentrations ( sub-MIC ) of aminoglycosides ( AGs ) induce the SOS response in Vibrio cholerae . SOS is induced upon DNA damage , and since AGs do not directly target DNA , we addressed two issues in this study: how sub-MIC AGs induce SOS in V . cholerae and why they do not do so in E . coli . We found that when bacteria are grown with tobramycin at a concentration 100-fold below the MIC , intracellular reactive oxygen species strongly increase in V . cholerae but not in E . coli . Using flow cytometry and gfp fusions with the SOS regulated promoter of intIA , we followed AG-dependent SOS induction . Testing the different mutation repair pathways , we found that over-expression of the base excision repair ( BER ) pathway protein MutY relieved this SOS induction in V . cholerae , suggesting a role for oxidized guanine in AG-mediated indirect DNA damage . As a corollary , we established that a BER pathway deficient E . coli strain induces SOS in response to sub-MIC AGs . We finally demonstrate that the RpoS general stress regulator prevents oxidative stress-mediated DNA damage formation in E . coli . We further show that AG-mediated SOS induction is conserved among the distantly related Gram negative pathogens Klebsiella pneumoniae and Photorhabdus luminescens , suggesting that E . coli is more of an exception than a paradigm for the physiological response to antibiotics sub-MIC . Vibrio cholerae is a severe human pathogen growing on crustacean shells or planktonically in the aquatic environment . V . cholerae couples environmental stress and adaptation through SOS response dependent gene expression and mutagenesis , induced , for instance , during ssDNA uptake [1]–[2] or after antibiotic treatment [3] . The SOS stress response occurs through the induction of an entire regulon controlled by the LexA repressor [4] . In the presence of abnormal levels of single stranded DNA ( ssDNA ) in the cell , RecA , the pivotal protein of homologous recombination , forms a nucleofilament on ssDNA which catalyses the self-cleavage of the LexA repressor . Inactivation of LexA releases the transcription of recombination and repair genes belonging to the SOS regulon . ssDNA levels increase in the cell during horizontal gene transfer but also when replication is blocked in the presence of any DNA damaging agent such as antibiotics that target DNA ( like fluoroquinolones or mitomycin C ) or UV irradiation . We recently reported the effects on the SOS response induction of sub-Minimal Inhibitory Concentrations ( sub-MIC ) of antibiotics from different families in V . cholerae [3] . Sub-MIC of antibiotics may be found in several environments [5] , in particular in the mammalian hosts of pathogenic and commensal bacteria , where they can play a very important role for the selection of resistant bacteria [6] . Moreover , a large part of the ingested antibiotics is rejected unchanged in the environment [7]–8 . Unlike above-MIC , the biological effects of sub-MIC of antibiotics have not been studied in detail . Transcriptome and proteome analyses have shown that many antibiotics exhibit contrasting properties when tested at low compared to high concentrations: it has been noted that sub-MIC of antibiotics induce several changes in expression profiles of a wide range of genes unrelated to the target function [9] . Sub-MIC of antibiotics induce phenotypic changes , including an increase in mutagenesis [3] . Strikingly , we observed that sub-MIC of aminoglycosides ( referred to as AGs throughout this manuscript ) , chloramphenicol , rifampicin and tetracycline induce SOS in V . cholerae [3] . These antibiotics do not directly target DNA synthesis or the DNA molecule and they do not induce SOS in E . coli [3] . The fact that they induce SOS in V . cholerae suggests a role for intermediate factors that cause stress and lead to DNA damage in this bacterium . A recent study demonstrated how beta-lactams , fluoroquinolones ( FQs ) and AGs stimulate production of reactive oxygen species ( ROS ) in bacteria [10] . ROS can damage DNA and proteins , and induce mutagenesis , increasing as such the odds of resistance conferring mutations , ultimately leading to multiple resistances . DNA damage can indeed be caused by ROS [11]: DNA is the target of hydroxyl radicals ( OH− ) and the Fenton reaction is the major source of OH− formation [12] in the presence of iron ions [13]–[14] . Iron can localize along the phosphodiester backbone of nucleic acids and OH− attacks DNA sugar and bases and ultimately causes double strand breaks , which are repaired by the RecBCD homologous recombination pathway [15] . Another type of DNA damage caused by oxidative attack is the incorporation of oxidized guanine residues ( 7 , 8-Dihydro-8-oxo-guanine or 8-oxo-G ) . E . coli and V . cholerae possess a defense system against 8-oxo-G which involves MutT , MutY and MutM [16] . MutT hydrolyses 8-oxo-G in the nucleotide pool , whereas MutY and MutM limit incorporation of 8-oxo-G and mismatch formation [17] . Here too , incomplete action of this base excision repair system may lead to double strand DNA breaks that are cytotoxic if unrepaired [18] . On the other hand , it is known that mistranslated and misfolded proteins are more susceptible to oxidation [19] . Moreover , mistranslational corruption of proteins may lead to replication fork collapse and induction of SOS [20] . It has been shown in Deinococcus radiodurans that cell death by radiation is not caused by direct DNA damage but primarily by oxidative damage on proteins , which eventually results in the loss of DNA repair [21] . Proteome protection against ROS thus seems to be at least as important as DNA protection for keeping cell integrity in times of oxidation . Knowing that AGs target protein translation , it was appealing to determine if sub-MIC of these antibiotics induce ROS formation in V . cholerae and to try to characterize mechanisms involved in the induction of SOS . Finally , oxidative stress is known to induce the RpoS regulon [22]–[23] . RpoS is the stationary phase sigma factor , which is induced in exponential phase in response to stress [24]–[26] . Genes expressed following RpoS regulon induction , namely catalases ( KatE , KatG ) and iron chelators ( Dps ) , protect cells from ROS related DNA damage [27] , e . g . double strand DNA breaks caused by hydroxyl radicals generated through the Fenton reaction [12]–[13] , [28] . It has been shown in Vibrio vulnificus ( a human pathogen ) that RpoS is essential in exponential growth phase to overcome H2O2 related oxidative stress [29] . Moreover , it was observed that V . vulnificus is more sensitive to H2O2 than E . coli and that KatG catalase expression is more strongly reduced in a ΔrpoS mutant in V . vulnificus compared to E . coli . These observations pointed to an effect of RpoS in the differences on the response to H2O2 between these two species , and as V . cholerae is more phylogenetically related to V . vulnificus than to E . coli , one would expect more conservation for the RpoS response between the two Vibrio species . RpoS levels are regulated at several stages in E . coli: ( i ) transcription [30] ( ii ) mRNA stability [31] and ( iii ) protein stability . Indeed , the RpoS protein is targeted to the ClpXP protease for degradation by the adaptor protein RssB [32]–[34] . ΔclpP mutants accumulate RpoS [34] . Anti-adaptors IraD , IraM , IraP are induced following stress ( respectively: oxidative stress , magnesium and phosphate starvation ) [32] and prevent this targeting by RssB , thus stabilizing the RpoS protein [33] . A ΔrssB mutant relieves H2O2 sensitivity of an E . coli ΔiraD mutant in an RpoS dependent fashion . Interestingly , the IraD anti-adaptor , which is induced during oxidative stress and DNA damage is conserved only in E . coli and Salmonella and is absent from V . cholerae [24]–[25] , whereas RssB is present in V . cholerae , suggesting that E . coli RpoS is more effectively protected from degradation during stress than V . cholerae RpoS . Moreover , among Gram negative pathogens , E . coli and Salmonella seem to be the only species that carry these anti-adaptors , suggesting that other pathogens may behave like V . cholerae in terms of RpoS degradation . Here , we test and show that sub-MIC of the AG tobramycin induces oxidative stress in V . cholerae . Moreover , we show that this SOS induction is mostly due to hydroxyl radical formation and 8-oxo-G incorporation in DNA using GFP as a reporter of SOS thanks to fusions of gfp with the SOS-dependent intIA promoter constructed for our previous studies [1] , [3] . Finally , we provide evidence for a role of RpoS in the protection of E . coli and V . cholerae cells against sub-MIC tobramycin induced stress and propose that most of the induction is linked to the rapid degradation of the RpoS protein in V . cholerae . We further show that the SOS induction by AGs is conserved among distantly related Gram negative pathogens . In order to understand how sub-MIC of antibiotics such as AGs that do not directly target DNA ( AGs ) impact the bacterial cell and induce a stress response in V . cholerae , we aimed at determining if SOS induction by sub-MIC takes place following oxidative stress . First we addressed whether there is a difference in ROS formation between E . coli and V . cholerae . We chose to use low doses of tobramycin ( 100 fold below the MIC ) in both bacteria because this antibiotic was used in previous studies [3] . These concentrations ( 0 . 1 µg/ml for E . coli and 0 . 01 µg/ml for V . cholerae ) were previously shown to induce SOS in V . cholerae and not in E . coli [3] . We performed growth in LB medium with and without tobramycin in the presence of dihydrorhodamine 123 ( DHR ) , a chemical agent that becomes fluorescent upon oxidation into rhodamine 123 in the presence of intracellular ROS [35] . DHR oxidation actually reports the presence of H2O2 and intracellular peroxidases [36] , and peroxidases are induced in response to oxidative stress and catalyze the formation of ROS . Fluorescence is thus interpreted as peroxidase induction and generation of free oxygen radicals [36] . We measured the intracellular generation of free radicals/peroxidase induction through rhodamine fluorescence at the middle and end of exponential phase ( OD600 0 . 5 and 0 . 8 ) . Our results showed that , when these bacteria were grown in the presence of tobramycin at 1/100 MIC , peroxidase induction ( i . e . oxidative stress ) took place in V . cholerae but not in E . coli ( Figure 1 ) . Ciprofloxacin ( a fluoroquinolone ) was used as a control known to induce ROS formation in E . coli [10] . Ciprofloxacin also induced SOS in both bacteria at sub-MIC ( at a concentration of 1/100 of the MIC ) [3] . As expected , sub-MIC of ciprofloxacin induced DHR oxidation in E . coli and in V . cholerae . These results point to a significant activation of the oxidative stress response and ROS formation in V . cholerae in response to sub-MIC tobramycin treatment . In order to study the nature of DNA damage caused by sub-MIC tobramycin , we alternatively inactivated the RecFOR gap repair and RecBCD double strand break repair pathways by deleting recF and recB respectively in the wild type V . cholerae strain ( Figure 2 ) . We used GFP fused to the intIA promoter as the SOS reporter ( plasmid p4640 for V . cholerae ) [1] , [3] . Mitomycin C ( MMC ) cross-links the two DNA strands , leading to double strand break formation , which induces SOS . MMC was thus used as an SOS inducer and tested as a positive control . The basal GFP fluorescence on LB was at the same level for all strains shown in Figure 2 . The RecFOR pathway is involved in the induction of SOS by allowing RecA nucleo-filament formation on single strand DNA lesions , while the RecBCD pathway recruits RecA on double strand DNA breaks [15] . SOS induction following tobramycin treatment decreased from 3 fold for wild type to 1 . 2 fold for recB and 1 . 6 for recF mutants ( Figure 2 ) . This suggests that both double strand breaks and single strand lesions are formed on DNA , with a more dramatic effect on double strand DNA breaks . The recB mutant also grew more slowly in tobramycin than the recF mutant ( data not shown ) . The presence of double strand ( ds ) lesions is compatible with the fact that hydroxyl radicals ( OH− ) target DNA and cause ds breaks . The Fenton reaction is the major source of OH− radical formation in the presence of Fe2+ ions [12]–[13] . If OH− radicals are responsible for SOS induction , then iron is also expected to be essential for inducing SOS . We used 2 , 2′-dipyridyl ( DP ) , an iron chelator that prevents the Fenton reaction , to test whether iron depletion affects SOS induction by tobramycin . DP did not change the basal level of fluorescence in LB . In the presence of tobramycin , the SOS induction was decreased from 3 to 1 . 3 fold upon addition of DP ( Figure 2 ) . Single strand and double strand lesions can occur during mismatch repair . Mismatches are formed when a DNA base is incorporated in a non Watson-Crick manner . This can happen during oxidative stress , which leads to the presence of higher levels of oxidized guanine residues ( 8-oxo-G ) . During replication , when the DNA polymerase encounters an 8-oxo-G residue , it pairs it to an adenine instead of a cytosine . The repair of such mismatches involves MutY [37] . MutT is also involved in the response to 8-oxo-G by decreasing its concentration in the nucleotide pool . We over-expressed MutY and MutT in V . cholerae from a plasmid . The empty plasmid showed higher SOS induction by tobramycin than the wild type strain without any plasmid . This is also reproducible with other plasmids ( not shown ) , and can be explained by the fact that such high copy plasmids introduce more DNA to replicate and thus more potential for DNA damage and repair . When we over-expressed MutY or MutT , we found that they strongly decreased the SOS induction , from 6 . 2 fold for the wild type carrying the empty vector plasmid to 1 . 4 and 1 . 2 fold respectively , confirming that sub-MIC tobramycin treatment led to incorporation of 8-oxo-G residues ( Figure 2 ) . Moreover , when the base excision repair pathway for 8-oxo-G was impaired in E . coli , ( using the E . coli ΔmutT ΔmutM ΔmutY strain ) , sub-MIC tobramycin and kanamycin ( another AG ) treatments resulted in SOS response induction , whereas no SOS induction was observed in BER proficient E . coli , further implicating the presence of incorporated 8-oxo-G residues as responsible for SOS induction ( Figure 3 ) . Altogether , this first set of results shows that V . cholerae is more prone to react to oxidative stress than E . coli , forming reactive oxygen species at low doses of antibiotics , which can explain SOS induction by sub-MICs in this species . In order to understand and clarify the origin of these differences between E . coli and V . cholerae , we decided to address the role of the RpoS regulon on sub-MIC tobramycin-induced SOS . It is well established that the RpoS-dependent general stress response is triggered in response to oxidative stress . Several genes expressed after RpoS regulon induction , namely catalases ( KatE , KatG ) and iron chelators ( Dps ) , protect cells from ROS related DNA damage [12]–[13] , [28] . It has been shown in Vibrio vulnificus that RpoS is essential in exponential growth phase to overcome H2O2 related oxidative stress [29] . We constructed a V . cholerae strain deleted for rpoS . We performed viability tests and found that the mutant strain did not grow in 2 mM H2O2 ( data not shown ) , confirming that RpoS is essential for V . cholerae growth during oxidative stress . RpoS is targeted to the ClpXP complex for degradation by the RssB protein . In E . coli , IraD binds to and titrates RssB during oxidative stress induction so that less free RssB is present in the cell to bind RpoS . IraD thus protects RpoS from degradation . As mentioned in the introduction , IraD is absent from the genome of a majority of bacterial species , whereas RssB is conserved . In order to address whether the absence of IraD plays a role in the induction of SOS by sub-MIC tobramycin in other species , we decided to test two other unrelated pathogens , Klebsiella pneumoniae and Photorhabdus luminescens . In order to follow SOS induction in these species , we transformed them , as we did for E . coli , with the plasmid p9092 carrying the Pint-gfp fusion . We found that sub-MIC tobramycin treatment induced SOS in both species , as it did for V . cholerae ( Figure 4 ) . As the iraD gene is absent from the V . cholerae genome , we hypothesized that higher levels of RpoS are present in the cell during sub-MIC tobramycin treatment in E . coli , than in V . cholerae , allowing E . coli to cope more easily with oxidative stress and avoid eventual DNA damage . In order to test this hypothesis , we first deleted iraD in E . coli . [32] . We used the Pint-gfp fusion ( GFP fused to the intIA promoter ) as a reporter of SOS induction ( p9092 for E . coli ) [2] . sfiA is another gene regulated by the SOS response and commonly used for SOS measurement assays ( as in [1] ) . GFP fused to the sfiA promoter was also tested and gave the same induction profiles as the intIA promoter ( not shown ) . We chose to carry on with the intIA promoter in order to have the same reporter promoter in E . coli as in V . cholerae . IraD protects RpoS from degradation . Deleting iraD in E . coli MG1655 resulted in SOS induction following sub-MIC tobramycin treatment ( Figure 5A ) . RssB targets RpoS to degradation . Conversely , we found that RssB over-expression strongly induced SOS in these conditions ( Figure 5A ) . We then expressed the E . coli anti-adaptor IraD in V . cholerae ( Figure 5B ) and found that in this context sub-MIC tobramycin dependent SOS induction was decreased in V . cholerae , whereas SOS induction by MMC was not affected ( 4 fold induction , not shown on the graph ) . Over-expression of RpoS also relieved SOS induction following sub-MIC tobramycin treatment . Conversely , when rpoS was deleted or when RssB was over-expressed in V . cholerae , SOS was significantly induced with or without tobramycin . The higher basal levels when RpoS is low ( deletion or high level of RssB ) could be a clue to impaired replication in these strains in LB . For instance , when sub-units of DNA polymerase are impaired ( such as in dnaEts or dnaNts mutants at semi-permissive temperature ) , such SOS induction can also be detected [38] . Over-expression of IraD or RssB in a V . cholerae ΔrpoS strain had no effect on SOS , whereas RpoS over-expression complemented efficiently the V . cholerae ΔrpoS strain , confirming that the effect observed in the wild type strain is dependent on RpoS ( Figure 5B ) . Western blotting and detection with an RpoS-specific antibody showed that in the presence of tobramycin , the amount of RpoS increased ( or RpoS degradation decreased ) in wild type E . coli and not in wild type V . cholerae during exponential phase ( Figure 6 ) . Moreover , these increased RpoS amounts were not observed in an E . coli ΔiraD strain . Altogether , these data show that protection of RpoS levels is sufficient to relieve sub-MIC AG induced SOS response in V . cholerae . We then addressed whether cellular ROS levels were modified in E . coli and V . cholerae mutants . We performed the DHR assay on rpoS and iraD mutants . As the growth of these strains was impaired in tobramycin , instead of measuring fluorescence at OD600 nm 0 . 5 after TOB 0 . 01 µg/ml treatment as we did in Figure 1 , we either measured fluorescence at the beginning of exponential phase ( OD 0 . 2 ) for the same concentration of TOB ( 0 . 01 µg/ml , Figure 5C ) , or at the same OD ( 0 . 5 ) for a decreased TOB concentration ( 0 . 001 µg/ml , Figure 5D ) . In both cases , we observed an increase of ROS in E . coli and V . cholerae when rpoS or iraD ( for E . coli ) were deleted , showing that decreased RpoS levels lead to ROS formation after TOB treatment in both bacteria . Finally , it has been observed that addition of polyamines ( e . g . putrescine , spermidine ) induces RpoS transcription [39] and reduces intracellular ROS production and DNA fragmentation [40]–[41] . Polyamines thus have antioxidant properties and a protective effect on DNA . We tested the effect of spermidine on tobramycin dependent SOS induction in V . cholerae wild type or Δrpos . We observed that spermidine decreases SOS induction after tobramycin treatment in an RpoS dependent way ( Figure 5B ) . These results support the idea that RpoS is involved in SOS induction by oxidative stress after AG treatment in V . cholerae , and suggest that the RpoS protein level is less stable in V . cholerae . We show here that V . cholerae is subject to oxidative stress in response to AG ( here tobramycin ) at concentrations 100 times below the MIC . These concentrations of AGs seem to lead to the formation of ROS and ultimately to DNA damage through double strand breaks and 8-oxo-G incorporation into DNA . E . coli on the other hand , has a stronger resistance to this kind of feeble stress triggered by AGs . We found that a difference between V . cholerae and E . coli is that the RpoS sigma factor is protected from degradation by anti-adaptor proteins in E . coli upon oxidative stress induction , whereas it seems to be more easily degraded in V . cholerae . Indeed , we showed that the protection of E . coli RpoS from degradation by the anti-adaptor IraD is important for the protection of the cells against oxidative stress that is triggered in the presence of tobramycin . IraD is part of a genomic island [42] . We looked for other genomes that possess iraD orthologs in KEGG ( http://www . genome . jp/kegg/ ) and MicroScope ( http://www . cns . fr/agc/microscope/home/index . php ) databases containing respectively 2020 and 872 bacterial genomes , including commensal bacteria ( like Bacillus subtilis , Streptococcus agalactiae ) and pathogens such as Acinetobacter baumannii , Proteus mirabilis , Escherichia coli , Klebsiella pneumoniae , Listeria monocytogenes , Pseudomonas aeruginosa , Streptococcus pneumoniae , Vibrio cholerae , Yersinia pestis and others . In KEGG , iraD orthologs were found in 69 genomes ( 49 E . coli and 5 Shigella with 70 to 100% protein identity; 15 Salmonella with 40 to 50% identity ) . In MicroScope , a total of 49 genomes appear to possess iraD orthologs where 44 are E . coli , 4 Shigella and 2 Salmonella . No sequenced bacterial organism other than E . coli , Shigella and Salmonella species possess iraD , which suggests that E . coli is more of an exception than a paradigm for the physiological response to antibiotics sub-MIC . We cannot rule out the possibility that other anti-adaptors exist in V . cholerae that are not homologous to IraD . Indeed , the three known E . coli anti-adaptors IraD , IraP ( responding to phosphate starvation ) and IraM ( responding to magnesium starvation ) do not resemble each other , although they all interact with RssB to protect RpoS from degradation . RpoS levels are also regulated at transcriptional level by rprA and dsrA small RNAs that bind the rpoS mRNA , and protect it from degradation in E . coli . Interestingly , like the IraD protein , rprA and dsrA small RNAs are also absent from V . cholerae [31] . Figure 7 represents the model we propose for the induction of SOS after AG treatment: the presence of sub-MIC AG leads to an increase of reactive oxygen species in the bacterial cell , causing oxidative stress . We propose that RpoS levels are insufficient to cope with oxidative stress caused by sub-MIC tobramycin in V . cholerae . A more steady presence of RpoS , as in E . coli , can lead to more efficient protection from oxidative stress and avoidance of DNA damage . The protective effect of RpoS can be through proteome protection , namely by decreasing the synthesis of iron rich proteins [43] and through genome protection by limiting available free iron and OH− formation [43]–[44] . Mistranslated or misfolded proteins have indeed been shown to be more susceptible to oxidation [19] . Protein oxidation is thus not only a function of ROS availability but also of the levels of aberrant proteins . AGs target the ribosome and cause mistranslation , which could lead to oxidation of mistranslated proteins , and thus an increase in oxidized proteins and eventually DNA damage due to impaired replication and repair . Interestingly , stabilization of a single oxidative stress sensitive protein was shown to be sufficient to enhance oxidative stress resistance of the whole organism in V . cholerae [45] , confirming the weight of protein oxidation on V . cholerae's ability to cope with stress and RpoS could be a key factor in this process . Moreover , it was shown in V . cholerae that cell envelope damage ( caused by chemical treatment , genetic alterations , physical damage ) leads to internal oxidative stress , formation of oxygen radicals , changes in iron physiology ( increased iron storage ) [46] and induces RpoS [46]–[48] . The presence of AGs could be responsible for such envelope stress in V . cholerae . RpoS regulates various factors that allow protection from oxidative stress , such as antioxidant agents , but also iron availability sensors like Fur , which prevents iron acquisition when the free iron level is high [43]–[44] . Indeed , excess iron reacts with ROS formed as a natural consequence of aerobic metabolism , and generates hydroxyl radicals through the Fenton reaction . Iron availability is necessary for the toxicity of antibiotics [49] . Another protein induced by RpoS is Dps , a DNA binding iron chelator involved in the protection against killing by bactericidal antibiotics ( through hydroxyl radicals ) upon oxidative stress during exponential phase [49] . Dps binds and stores iron , which attenuates OH− formation . It is tempting to state here that the effect of RpoS is synergistic with SOS for genome protection , in that RpoS prevents OH− attack on DNA whereas SOS repairs it . Our results using DP are consistent with an effect of iron in the process of DNA damage by sub-MIC tobramycin . So why has V . cholerae selected for an RpoS system that is less efficient than that of E . coli ? This could be explained by the need for V . cholerae to quickly up- and down-regulate the RpoS regulon during infection . Indeed , in contrast to its role in stress response , the effects of RpoS on pathogenesis and virulence are highly variable and depend on bacterial species and their niches [50] . RpoS is conserved within alpha , beta and gamma proteobacteria but the RpoS regulon composition is subject to modification between species [51] . In Salmonella , RpoS is essential for virulence . In E . coli , RpoS induces virulence genes , but a ΔrpoS mutant can outcompete wild type cells in the mouse colon . A tradeoff has been shown between self preservation and nutritional competence in E . coli infected patients , determined by levels of RpoS that naturally occur in different E . coli cells from the same infecting strain [52]–[53] . In Vibrio anguillarum , a fish pathogen , virulence is also up-regulated by RpoS [54] . Conversely , and illustrating the high variability of the RpoS regulon between even closely related species , in V . cholerae RpoS represses virulence genes ( a ΔrpoS strains produces more cholera toxin ) , and is not required for intestinal survival , even though RpoS contributes to overcoming host-specific stresses in the gut [55] . Nevertheless , RpoS plays an important role in the pathogenicity of V . cholerae because it is required for detachment from the epithelial cells after infection and release of bacteria in the environment ( the “mucosa escape response” ) [56] . This state is characterized by high RpoS levels , RpoS dependent up-regulation of chemotaxis , motility , flagella and reduced virulence gene expression . V . cholerae thus has to down-regulate RpoS in order to be virulent in the gut and up-regulate RpoS in order to spread in the environment and survive . This is not the case for E . coli where RpoS induces virulence and has no known role in its escape and spread to the environment . The absence of proteins protecting RpoS from degradation in V . cholerae could increase the efficiency of the switch between low RpoS ( cholera ) and high RpoS ( spread ) . Considering this RpoS regulon variability , it was proposed that horizontally transferred genes , which may enhance host adaptation , integrate into the RpoS regulon [51] , [57] . In Gram-negative pathogens , acquired multiple antibiotic resistance is in many cases associated with mobile integrons . This natural genetic engineering system is composed of a gene coding an integrase belonging to the site-specific recombinase family and a primary recombination site where gene cassettes can be integrated . The existence of sedentary integron platforms gathering up to 200 gene cassettes in the chromosomes of environmental proteobacteria has been demonstrated [58]–[61] . This is the case for V . cholerae , which carries a superintegron gathering more than 170 cassettes [52] . These cassettes can code for antibiotic resistance and other adaptive genes , and their integration by the integrase allows expression from a constitutive promoter upstream of the insertion site [62]–[63] . Moreover , integrase expression is controlled by the SOS response , suggesting that the evolutionary success of these elements largely accounts for the coupling of stress and cassette array remodeling [64]–[65] . We show here that sub-MIC AGs induce SOS and thus integron integrase expression in V . cholerae . This AG driven induction , which is apparently common in gamma proteobacteria that lack IraD , likely explains how the numerous cassettes coding for resistance to all AG ( 43 in 2009 [66] ) characterized in mobile integrons were recruited . This implies that the use of these antibiotics may promote cassette rearrangements and expression of integron-borne resistance genes to all families of antibiotics , including the ones that do not induce SOS in E . coli . Bacteria face different growth conditions in different environments . V . cholerae has two different niches: the aquatic environment where it grows as a biofilm on crustacean shells where nutrients are scarce , and the host gut , a rich environment where virulence is induced . In both locales , V . cholerae may face varying antibiotic concentrations , due to rejections in the environment or antibiotic treatments of the host . Other pathogens are also known to encounter sub-MIC of antibiotics , such as Pseudomonas aeruginosa , which causes chronic lung infections where antibiotics subsist as gradients . Strikingly , it was shown that metronidazole treatment of a patient infected with P . aeroginosa induced β-lactamase and ceftazidime resistance through SOS-mediated integron rearrangements [67] . It is tempting to make a parallel here with our work with AGs , as neither AGs nor metronidazole cause direct DNA damage , though they are both able to induce the SOS response . It is thus important to better understand how different ecological niches and different life styles modulate evolution of stress responses , which have a major impact on the evolution of genome plasticity and antibiotic resistance , and understanding the molecular mechanisms that drive the emergence of drug resistance can facilitate the design of more effective treatments .	Bacteria can remodel their genome in order to adapt to new environments . This genomic plasticity can be mediated by horizontal gene transfer , genome rearrangements , and mutations . One important factor is the bacterial SOS response , induced upon DNA damaging stress conditions . SOS response induction leads to increased mutation frequencies and genome rearrangements . Antibiotics such as fluoroquinolones cause direct DNA damage and induce the SOS response . Other antibiotics such as aminoglycosides ( AGs ) do not target DNA and do not induce SOS in E . coli . However , it was recently shown that low concentrations of AGs , which do not inhibit bacterial growth , do induce SOS in V . cholerae . We show here that AG-dependent SOS induction is caused by oxidative stress in V . cholerae and that E . coli is well protected against this kind of oxidative insult thanks to efficient oxidative stress response through the RpoS general stress response sigma factor . We propose that the model bacterium E . coli is not a paradigm in this case , as other pathogens such as Klebsiella pneumoniae and Photorhabdus luminescens also induce SOS in response to sub-MIC AGs .	[ "Abstract", "Introduction", "Results", "Discussion" ]	[ "flow", "cytometry", "microbial", "mutation", "cellular", "stress", "responses", "genetic", "mutation", "gene", "regulation", "microbiology", "mutational", "hypotheses", "dna", "transcription", "gene", "function", "escherichia", "coli", "emerging", "infectious", "diseases...	2013	RpoS Plays a Central Role in the SOS Induction by Sub-Lethal Aminoglycoside Concentrations in Vibrio cholerae
Paracoccidioides brasiliensis is a dimorphic fungus that is the causative agent of paracoccidioidomycosis , the most important prevalent systemic mycosis in Latin America . Recently , the existence of three genetically isolated groups in P . brasiliensis was demonstrated , enabling comparative studies of molecular evolution among P . brasiliensis lineages . Thirty-two gene sequences coding for putative virulence factors were analyzed to determine whether they were under positive selection . Our maximum likelihood–based approach yielded evidence for selection in 12 genes that are involved in different cellular processes . An in-depth analysis of four of these genes showed them to be either antigenic or involved in pathogenesis . Here , we present evidence indicating that several replacement mutations in gp43 are under positive balancing selection . The other three genes ( fks , cdc42 and p27 ) show very little variation among the P . brasiliensis lineages and appear to be under positive directional selection . Our results are consistent with the more general observations that selective constraints are variable across the genome , and that even in the genes under positive selection , only a few sites are altered . We present our results within an evolutionary framework that may be applicable for studying adaptation and pathogenesis in P . brasiliensis and other pathogenic fungi . The neutral theory of evolution states that most evolutionary change at the molecular level is caused by the fixation of neutral alleles through random genetic drift [1] . Nonetheless , it is the impact of natural selection on genomic evolution that is of interest if we wish to understand patterns of adaptive evolution by distinguishing between selectively neutral and non-neutral evolutionary change , and relate this change to the biology and history of the organism . The arms race between hosts and their pathogens is a particularly useful system for relating potentially non-neutral evolutionary change to the biology and history of the organisms [2] , [3] because of the role natural selection plays in maintaining or fixing different alleles in both host and pathogen populations [4] . Human-fungal interactions provide a privileged system to study the impact of natural selection on the genome of fungal pathogens . Paracoccidoides brasiliensis is the etiological agent of paracoccidioidomycosis ( PCM ) , a human systemic mycosis of importance in Latin America [5] . It is endemic to an area extending from Mexico to Argentina , and infects an estimated 10 million people [6] . Recently , the existence of genetically distinct evolutionary lineages within P . brasiliensis was demonstrated through analysis of DNA sequence data for multiple genes [7] , [8] . These groups are currently designated S1 ( species 1 ) , PS2 ( phylogenetic species 2 ) , PS3 ( phylogenetic species 3 ) and Pb01 . Additional support for these lineages comes from variation in virulence and expression levels of antigenic proteins previously found between P . brasiliensis isolates which are now known to belong to S1 and PS2 groups [9] . The recent publication of genomic sequences in the form of expressed sequence tag ( EST ) databases for several isolates of the different genetic groups of P . brasiliensis [10] , [11] , [12] and the closely-related species Histoplasma capsulatum ( Ajellomyces capsulatum ) ( unpublished results ) presents an opportunity to investigate the role that natural selection may have played in shaping the molecular evolution of the P . brasiliensis genome . Comparative studies between the P . brasiliensis genetic groups and H . capsulatum can be useful to understand host-pathogen evolution , especially in the genes encoding pathogenesis-related proteins which are likely to evolve in response to selective pressure from the host's immune system . Detecting natural selection at the molecular level requires statistical tests that distinguish the genomic signature of selection from that of neutral mutation and genetic drift alone . Positive selection is inferred when ω [13] ( the ratio of non-synonymous ( dN ) to synonymous ( dS ) mutations between species ) exceeds 1 . Positive directional selection occurs when successive amino acid changes make a protein better adapted in a particular biological context , and as a result the changes will tend to be fixed in future lineages . Positive diversifying selection occurs when multiple phenotypes in a population are favored , resulting in an overall increase of the genetic diversity within the species [14] , [15] . Several likelihood methods have also been developed to detect deviations from neutral expectation . Under an infinite-sites model , the level of DNA polymorphism within a species is proportional to the amount of divergence at that locus among closely related species [16] . Deviations from this model form the basis for various tests of natural selection , such as the HKA test [17] , and the M-K test [18] . Moreover , likelihood methods that allow ω to vary among the branches in a phylogeny , as well as between codons , have been proposed [19] , [20] , [21] , [22] , [23] . Using such methods , several genes involved in defense systems and immunity , as well as toxic protein genes , have been shown to be under diversifying or positive directional selection [24] , [25] , [26] , [27] . In this study , we sought to understand the molecular evolution of candidate genes associated with P . brasilensis fungal pathogenesis , which are hypothesized as being under positive selection due to their role in the host-pathogen immune system interaction . Thirty-two putative virulence factors described in previous studies [9] , [10] , [11] , [12] , [28] were selected from two available EST databases [10] , [11] . In addition , we randomly selected 32 putative housekeeping genes without known antigenic or virulence properties to be used as controls . Orthologous sequences from P . brasiliensis and H . capsulatum were tested for positive selection by means of the Nei and Gojobori method [29] , which calculates the average ratio across all amino acid sites . For those genes that showed some evidence of positive selection we obtained sequences from the three lineages of P . brasilensis and used maximum likelihood methods to identify amino acid residues on which positive selection has acted [30] . Our results suggest that positive selection has indeed played an important role in the molecular evolution of virulence factors of P . brasiliensis . The P . brasiliensis strains used in this study were described previously [7] . The sample included individuals from four biotypes recognized for P . brasiliensis: Pb01 ( n = 1 ) , S1 ( n = 46 ) , PS2 ( n = 6 ) and PS3 ( n = 23 ) and was representative of six endemic areas for paracoccidiodomycosis . We used sequences from GenBank under accession numbers DQ003724 to DQ003788 as well as new sequences obtained by methods previously described [7] . Briefly , total DNA was extracted from the yeast culture with protocols using glass beads [31] or maceration of frozen cells [32] . PCR primers and conditions were as previously reported [7] . The new sequences were deposited in GenBank under accession numbers EU283774 to EU283809 . Molecular genetic tools are still not fully developed for P . brasiliensis , hindering studies that seek to molecularly define genetic factors involved in P . brasiliensis pathogenesis . For the dimorphic fungi , a virulence factor has been functionally defined as a gene product that has an effect on the survival and growth of the organism in its mammalian host but is not essential for growth of the parasitic phase in vitro [33] . Nevertheless , the study of virulence genes sensu Rappleye and Goldman in isolation [33] does not provide full picture of their evolution , because the molecular basis of virulence involves complex networks that comprise many classes of genes . We focused on all the genes proposed to have an impact on the virulence of P . brasiliensis . Table S1 lists the genes that , following genomic analysis in P . brasiliensis , were considered as potential virulence factors and , as such , candidates for this survey [10] , [11] , [12] . For a gene to be included in this analysis , it had to fulfill three conditions: ( i ) to have been reported as a putative virulence factor in the previous literature [9] , [10] , [11] , [12] , [28] , ( ii ) to be present in the three analyzed databases ( two ESTs databases from P . brasiliensis and the genome of H . capsulatum ) , and ( iii ) have been demonstrated to be a virulence factor or be an ortholog of a proven virulence factor and have a high homology with it ( <1E-10 ) . Fifty percent ( 32 genes ) of the 64 initial candidates fulfilled our requirements and were analyzed to detect positive selection . To validate our results , we selected a smaller subset of genes that had demonstrated to be under positive selection pressures and for which population datasets were available . The only genes that fulfilled these characteristics were gp43 , p27 , fks and cdc42 . In this set of sequences we searched for evidence of positive selection using the CODEML program of the PAML package ( version 4 ) [22] , [30] by using several likelihood-based tests . For each test , equilibrium codon frequencies were estimated from the average nucleotide frequencies at each codon position , amino acid distances were assumed to be equal , and the transition/transversion ratio ( κ ) was estimated from the data . For all other parameters , we used the default settings described by Yang and Bielawski [30] . Given the observed intraspecific variability , the lack of homoplasy found in individual gene trees , and the phylogenetically recognized groups , we assumed linkage between colinear sites ( i . e . , there was no recombination within each data set ) . To determine which model best fit the data , likelihood ratio tests ( LRTs ) were performed by comparing the differences in log-likelihood values ( LRT = −2lnL ) between two models using a χ2 distribution , with the number of degrees of freedom equal to the difference in the number of parameters between the models . We used six models implemented in PAML [13] , [22] , [30] to test for the presence of sites under positive selection ( ω>1 ) . The one-ratio model ( M0 ) assumes one ω for all sites . The neutral model ( M1 ) assumes two classes of sites in the protein: the conserved sites at which ω = 0 , and the neutral sites that are defined by ω = 1 . The beta model ( M7 ) uses a β distribution of ω over sites: β ( p , q ) , which , depending on parameters p and q , can take various shapes in the 0 to 1 interval . The other three models allow sites with ω>1 and can be considered as tests of positive selection . The selection model ( M2 ) has an additional class of sites compared to the neutral model , in which ω is a free parameter and , as such , can change among residues . The discrete model ( M3 ) uses a distribution with three site classes , with the proportions ( p0 , p1 , and p2 ) and the ω ratios ( ω0 , ω1 , and ω2 ) estimated from the data . The beta and ω model ( M8 ) added an extra class of sites to the beta model , estimating the proportion of ω from the data . We used LRTs to make 3 comparisons: to find out whether positive selection has played a role in the molecular evolution of these genes the one-ratio model ( M0 ) was compared with the discrete model ( M3 ) and the neutral model ( M1 ) was compared with the selection model ( M2 ) . A third comparison ( the beta model ( M7 ) vs . the beta and ω model , M8 ) [30] was used to identified particular sites in the genes that were likely to have evolved under positive selection by using the Bayesian Empirical Bayes ( BEB ) analysis previously proposed by Yang [13] . Bayes' theorem was used to estimate the posterior probability that a given site came from the class of positively selected sites [13] , [30] , [43] . In order to predict potential antigenic determinants for HLA recognition , we used the program SYPFETHI [44] . To determine whether any of the studied loci presented coalescence times within the P . brasiliensis clade ( which were older than any other loci ) we calculated the Time to the Most Recent Common Ancestor ( TMRCA ) . TMRCAs for S1 and PS2 were estimated based on genetic variation at the eight nuclear loci using the program IM [45] . Estimates of TMRCA do not directly estimate the date of divergence; they provide the timing of coalescence of alleles within a taxon . TMRCA estimates can post- or pre-date the speciation event , and thus can indicate whether the polymorphism in any given gene is older or more recent than the polymorphism in the other genes . Thirty-two putative virulence factors fulfilled the requirements for inclusion in this analysis . All the virulence factors showed to be single-copy genes ( data not shown , available upon request ) . To be considered as being under positive selection , these genes had to exhibit a dN/dS ratio larger than 1 and a p-value for the Z-test below 0 . 05 . Table 1 shows the dN/dS ratios for the putative virulence factors and their p-values as determined by using the Z test . According to these criteria , 12 genes were determined to be under positive selection . The dN/dS ratio is correlated to the strength of selection , where values >1 indicate positive selection , and larger values indicate stronger selection . Thirty-two housekeeping genes were randomly selected from the P . brasiliensis available sequences by using a PERL script and their dN/dS ( and associated Z values ) were calculated and were used as source of comparison . None of these genes showed evidence of being under positive selection in the P . brasiliensis branches , as illustrated in Table 2 . A possible explanation for the high proportion of genes under positive selection is that the high proportion of virulence factors showing significantly higher dN/dS are partly artifacts caused by the methods used to estimate the number of non-synonymous and synonymous mutations [46] . Such an explanation would require saturation to occur faster in synonymous than in non-synonymous sites , i . e . , the number of non-synonymous nucleotide differences should be a concave function of the number of synonymous nucleotide differences [41] . We plotted the number of non-synonymous nucleotide differences between the two groups of P . brasiliensis and their common ancestor , against the number of synonymous nucleotide differences ( Figure 2 ) . No differences were found between the linear and the quadratic models , neither for virulence factors ( LRT = 2 . 134 , p = 0 . 144 ) , nor the housekeeping genes ( LRT = 0 . 112 , p = 0 . 7378 ) , nor for the pooled data ( LRT = 1 . 631; p = 0 . 2015 ) indicating that the lineal model is more appropriate to explain the relationship between dN and dS . Therefore , mutational saturation is not responsible for the elevated dN/dS ratios observed in the virulence factors . Similar comparisons were performed including H . capsulatum: one virulence factor ( ags1 ) and housekeeping gene ( Gp_dh_N ) were found to be under positive selection in the branch that leads towards H . capsulatum ( data not shown ) . Another possibility is that sequencing errors had inflated dN values . Such errors could artificially increase the significance level of the dN/dS test because they would tend to elevate the number of non-synonymous mutations . However , sequencing errors should also elevate the proportion synonymous mutations and missense mutations . If sequencing errors had , indeed , increased dN , then a large proportion of points in Figure 2 should be located in the upper-left region of the plane . Because no such pattern is observed in Figure 2 , we consider this explanation unlikely . Detection of positive selection by several computing packages program is “reliable” but “conservative” [19] , [30] , [47] when few sequences are used . Increased accuracy and power are most easily gained with more sequences [19] , [30] . Therefore , to further validate our methods and distinguish between directional and diversifying selection , we selected a subset of genes . We choose from among the 12 genes that showed both evidence for positive selection and had more than 25 sequences of P . brasiliensis in GenBank , then reapplying population genetics analysis to these genes . From the 12 genes listed in Table 1 , four were selected to be analyzed more in-depth: gp43 , p27 , cdc42 and fks . For the gp43 case , the M-K test yielded no significant results between H . capsulatum and P . brasiliensis ( Fischer's exact test , P = 0 . 40 , Table 3 ) . M-K tests were significant for p27 , cdc42 and fks ( p27: P = 0 . 043594; cdc42: P = 0 . 000993; fks: P = 0 . 000017; Table 3 ) when H . capsulatum was used as an outgroup . The TMRCAs for S1 and PS2 were estimated based on genetic variation at the gp43 locus and seven other nuclear loci . The results showed that the TRMCA for the gp43 alleles is longer than for any other gene in P . brasiliensis ( Table 6 ) , indicating that the polymorphism in gp43 is significantly older than the polymorphism in the other genes ( Signed rank test; P<0 . 01 ) . This constitutes evidence for balancing selection [49] , [50] . Additional evidence for the balancing selection hypothesis in gp43 comes from the haplotype network previously described for this gene , in which several high frequency haplotypes are separated by long branches [7] . Conversely , the TRMCAs for cdc42 , p27 and fks were significantly lower than the other genes as is expected if a gene is under positive directional selection . Comparisons of DNA sequence differences within and between closely related species can give insights into the temporal scales of molecular evolutionary processes , and into selective pressures on different type of loci . In this study , evidence of different types of positive selection acting on the putative virulence factors was obtained from analysis of the ratio between non-synonymous and synonymous substitution rates in coding regions . A comparison of these virulence factors with housekeeping genes in P . brasiliensis showed that a higher proportion of virulence genes evolve under positive selection ( 37 . 5% vs . 0% ) , suggesting that at least some of these genes have an adaptive role . Substantial heterogeneity in the mode of evolution was found both among and within the genes investigated in this study . As predicted from previous studies of evolution of virulence factors in other organisms , the 12 putative virulence factors genes identified as having evolved under positive selection have a wide variety of functions ( Table 1 , Table S1 and Text S1 ) [27] . This analysis of positive selection using genomic data identified a set of genes that together with data derived from genetic , expression and biochemical essays , provides some insights into the evolution of P . brasiliensis virulence . Some of these genes are involved in the escape from immune recognition ( tsa1 , sod1 ) . However , this is just one aspect of the ability of a pathogen to successfully invade and colonize its host , and other genes have proven to be important in pathogenesis , such as the case of heat shock genes that are connected to virulence [32]–[34] . Previous studies have suggested that although virulence factors sensu Rappleye and Goldman [33] are key factors in pathogenesis , their study as isolated entities does not provide a holistic picture of the evolutionary dynamics of virulence . The results of this study , and others , support the notion that many essential genes participate in complex networks that comprise the molecular basis of virulence , and that their history is shaped by natural selection . For most of the genes found to be under positive selection ( 10 out of 12 ) , biochemical and physiological characteristics are known . Only two genes ( p27 and gp43 ) have unknown functions . All the others were classified in four different categories of genes according to their functions: metabolic related genes ( fas2 , his1 ) , cell wall related genes ( fks , mnn5 , ags1 ) , heat shock proteins , detoxification related genes ( tsa1 , sod1 , hsp88 ) and signal transduction genes ( cdc42 , cst20 ) . A detailed biochemical description and information related to these genes is presented in the Text S1 . p27 , cdc42 and fks are genes that are depauperate in genetic variation , as is expected for regions in which advantageous amino acid replacements have been fixed by positive selection . Judging by the significant results of the M-K tests , positive selection has played an important role in the history of these three genes and the depletion of genetic variation within P . brasiliensis ( at these three loci ) is a consequence of positive selection . The M-K test was not significant for gp43 . This test has proven to be robust because the sites in which synonymous and non-synonymous mutations occur are interspersed , so that they would be similarly affected by genetic drift and changes in geography [20] , [45] . In gp43 , the M-K test was not able to detect positive selected within the P . brasiliensis lineage due to the excess of non-synonymous substitutions within and across species . The persistence of non-synonymous intra- and trans-specific gp43 polymorphisms within and between lineages of the P . brasiliensis complex suggests they have been maintained by historical or contemporary selection [51] . Several recent studies have used the power of modern molecular selection analyses to design experiments based on the molecular evolutionary hypothesis [20] . An example of the importance of this kind of study is that immunization with gp43 epitopes from one isolate would not be expected to be effective against allthe species complex due to the high level of polymorphism in gp43 . This has profound implications for the development of a gp43 vaccine and immunotherapy [52] . It is likely that the evolution of putative virulence factors of P . brasiliensis has been driven by the interaction between the pathogen and its extracellular environment . However , it remains unclear whether the positive pressure was derived from the environment when the fungus is in its free-living stages , or from the host's immune system . Determining the function and biochemical roles of the proteins encoded by the genes found to be under positive selection in P . brasiliensis should shed light on the corresponding selective pressures . Molecular evolutionary analysis should facilitate the identification of biologically important genes through the comparison of nucleotide sequences . Although the methods for positive selection used here are not perfect [23] , the identification of positively selected proteins offers a good approach for understanding human pathogenic fungi , in which transformation or production of mutants is difficult ( McEwen , personal communication ) . Positive selection in virulence factors might have different outcomes , including: adaptation of a species to optimize the process of infection , to escape host immune response , inhabit different environmental niches , and also lead to functional diversification of members of multi-gene families . We hope that identifying and cataloging these loci for this and other groups of fungi will provide others with an evolutionary framework for pursuing directed mutation experiments on the specific functional significance of these genes .	The fungus Paracoccidioides brasiliensis is the causative agent of paracoccidioidomycosis , a severe pulmonary mycosis that is endemic to Latin America , where an estimated 10 million people are infected with the fungus . Despite the importance of this disease , we know little about the ecological and evolutionary history of this fungus . Here , we present a survey of genetic variation in putative virulence genes in P . brasiliensis in what constitutes the first systematic approach to understand the molecular evolution of the fungus . We used a population genetics approach to determine the role has natural selection played in the coding genes for proteins involved in pathogenesis . We found that nonsynonymous mutations are more common in genes that code for virulence factors than in housekeeping genes . Our results suggest that positive selection has played an important role in the evolution of virulence factors of P . brasiliensis and is therefore an important factor in the host–pathogen dynamics . Our results also have implications for the possible development of a vaccine against paracoccidioidomycosis , since gp43—the main vaccine candidate—has a high level of polymorphism maintained by natural selection .	[ "Abstract", "Introduction", "Materials", "and", "Methods", "Results", "Discussion" ]	[ "evolutionary", "biology", "evolutionary", "biology/evolutionary", "and", "comparative", "genetics" ]	2008	Evidence for Positive Selection in Putative Virulence Factors within the Paracoccidioides brasiliensis Species Complex
Bacterial sepsis is a major killer in hospitalized patients . Coagulase-negative staphylococci ( CNS ) with the leading species Staphylococcus epidermidis are the most frequent causes of nosocomial sepsis , with most infectious isolates being methicillin-resistant . However , which bacterial factors underlie the pathogenesis of CNS sepsis is unknown . While it has been commonly believed that invariant structures on the surface of CNS trigger sepsis by causing an over-reaction of the immune system , we show here that sepsis caused by methicillin-resistant S . epidermidis is to a large extent mediated by the methicillin resistance island-encoded peptide toxin , PSM-mec . PSM-mec contributed to bacterial survival in whole human blood and resistance to neutrophil-mediated killing , and caused significantly increased mortality and cytokine expression in a mouse sepsis model . Furthermore , we show that the PSM-mec peptide itself , rather than the regulatory RNA in which its gene is embedded , is responsible for the observed virulence phenotype . This finding is of particular importance given the contrasting roles of the psm-mec locus that have been reported in S . aureus strains , inasmuch as our findings suggest that the psm-mec locus may exert effects in the background of S . aureus strains that differ from its original role in the CNS environment due to originally “unintended” interferences . Notably , while toxins have never been clearly implied in CNS infections , our tissue culture and mouse infection model data indicate that an important type of infection caused by the predominant CNS species is mediated to a large extent by a toxin . These findings suggest that CNS infections may be amenable to virulence-targeted drug development approaches . Bacterial sepsis is a frequent cause of death in hospitalized patients . Coagulase-negative staphylococci ( CNS ) are the leading cause of nosocomial sepsis , especially in neonates [1–3] . CNS sepsis most often originates from the infection of indwelling medical devices , such as in catheter-related bloodstream infections ( CRBSIs ) or central line-associated blood stream infections ( CLABSIs ) [4] . Most prominent among CNS infections are those due to the skin commensal Staphylococcus epidermidis [5] . However , the bacterial factors contributing to the development of sepsis , in particular in CNS , are poorly understood . Given that toxins have long been assumed to be widely absent from CNS [6] , sepsis caused by S . epidermidis and other CNS , similar to other Gram-positive bacteria , has so far been believed to be due predominantly to an overwhelming immune reaction directed against invariable , pro-inflammatory cell surface molecules , such as teichoic acids and lipopeptides [7] . Recently , the notion that CNS do not commonly produce toxins had to be revised with the discovery of the pro-inflammatory and cytolytic phenol-soluble modulin ( PSM ) staphylococcal toxin family [8] . However , due to the difficulties associated with genetic manipulation of S . epidermidis and other CNS , the roles of PSMs in CNS infections , including most notably sepsis , have hitherto remained unexplored . Most S . epidermidis blood infections are caused by methicillin-resistant strains ( MRSE ) , with methicillin resistance rates even exceeding those found among S . aureus [9] . Methicillin resistance is encoded on so-called staphylococcal chromosome cassette ( SCC ) mec mobile genetic elements , which are believed to have originated from CNS , from where they were transferred to S . aureus [10] . While other PSMs are core-genome encoded [8] , one PSM toxin , called PSM-mec , is encoded within SCCmec elements of subtypes II , III , and VIII [11 , 12] . The psm-mec gene is embedded in a short regulatory ( sr ) RNA , which in S . aureus has been reported to down-regulate the production of other PSMs and thereby decrease virulence [13 , 14] . While this effect has been claimed to generally explain lower virulence of hospital-associated as compared to community-associated MRSA strains [13] , it is quite moderate and extensively strain-dependent [11 , 13] . Recently , the psm-mec locus has been introduced on a plasmid into some CNS that naturally lack psm-mec , and was reported to trigger gene regulatory changes [15]; but the roles that the PSM-mec peptide or the psm-mec srRNA naturally play in CNS including S . epidermidis are unknown . Here we analyzed the role of the psm-mec locus in S . epidermidis sepsis by using tissue culture and animal infection models . Our findings show for the first time that a toxin can have a strong impact on CNS sepsis , setting the stage for anti-virulence strategies directed against this frequent and deadly infection . To analyze the impact of the psm-mec locus on S . epidermidis sepsis , we produced isogenic psm-mec deletion mutants ( Δpsm-mec ) in two MRSE strains , a clinical isolate ( SE620 ) and the genome-sequenced strain RP62A . PSM-mec production in these strains is representative of clinical PSM-mec-positive MRSE ( S1 Fig ) , which we determined in a clinical S . epidermidis strain collection from Norway to occur in ~ 2/3 ( 59/91 ) of the ~ 50% ( 91/180 ) methicillin-resistant S . epidermidis . We also introduced a point mutation in the start codon of the psm-mec gene in the genome of strain SE620 to differentiate between effects mediated by the PSM-mec peptide versus those due to the psm-mec srRNA ( psm-mec* ) . Notably , the stability of the psm-mec RNA was not significantly altered by introduction of the 1-basepair start codon mutation ( S2 Fig ) . We first analyzed those mutants in a murine sepsis model . Mortality was significantly reduced in the Δpsm-mec mutants of both strains ( Fig 1A and 1B ) . There was no significant difference between the Δpsm-mec mutant and the psm-mec* start codon mutant ( Fig 1A ) . Furthermore , CFU in the blood and the kidneys were strongly reduced in the Δpsm-mec mutants of both strains and the psm-mec* start codon mutant ( Fig 1C–1F ) . These results demonstrate a strong contribution of the PSM-mec toxin to bacteremia and mortality due to S . epidermidis sepsis , while the psm-mec srRNA did not show any impact . We showed previously that synthetic PSM-mec peptide is strongly pro-inflammatory and has moderate to strong cytolytic capacity [12] . To analyze the contribution that the psm-mec locus has to pro-inflammatory and cytolytic capacity in the S . epidermidis background , we measured cytokine concentrations during experimental murine sepsis and determined cytolytic capacity of the bacterial strains toward human neutrophils in vitro . Cytokine concentrations during sepsis are the result of a systemic reaction due to several immune cell types , and are thus best determined in vivo , while cytolytic capacity can be most accurately measured in vitro . The PSM-mec peptide , but not the psm-mec srRNA , had a strong and significant impact on the production of cytokines during murine sepsis ( Fig 2 ) . At 12 h after infection , the mouse IL-8 homologue CXCL1 was significantly reduced when mice were infected with the Δpsm-mec or psm-mec* start codon mutant of strain SE620 , to about half the concentration measured in mice infected with the wild-type strain ( Fig 2A ) . Concentrations of IL-1β and TNF-α were even more strongly reduced to levels not significantly different from those measured in mock ( PBS ) infected animals ( Fig 2B and 2C ) . In the RP62A background , the phenotypes were similar , with differences being more pronounced during earlier stages of the infection ( measured at 2 versus 12 h ) ( Fig 3 ) . These results showed that the cytokine storm that commonly accompanies bacterial sepsis is strongly dependent on the PSM-mec toxin in S . epidermidis . In addition to being pro-inflammatory , the PSM-mec toxin has pronounced cytolytic capacity [12] . Cytolysis by PSMs is believed to be most important for infection when bacteria are engulfed in the phagosome of neutrophils and other phagocytes [16 , 17] . Survival of bacteria when incubated with human neutrophils and survival in whole human blood was significantly higher with the S . epidermidis wild-type strain than with Δpsm-mec or psm-mec* start codon mutants , as was killing of neutrophils when incubated with whole bacteria ( Fig 4 ) , emphasizing the role of the PSM-mec toxin in evasion of neutrophil killing and resistance to the strong bactericidal capacities of immune defense mechanisms in human blood . Together , these results indicate that the both the pro-inflammatory and cytolytic capacities of the PSM-mec peptide contribute to the development of S . epidermidis sepsis . In S . aureus , the psm-mec locus has also been implicated in biofilm-forming capacity , although effects were generally minor and highly strain-dependent [12] . Similar to S . aureus , biofilm formation in S . epidermidis was affected only slightly by the psm-mec locus , and as this was seen only in one strain , similarly strain-dependent ( Fig 5 ) . In that strain , SE620 , the effect was due to the PSM-mec peptide , not the psm-mec srRNA . These findings indicate that during indwelling medical device-associated blood stream infections by S . epidermidis , the impact of PSM-mec generally is by contributing to the development of sepsis , as we have shown here , rather than by promoting biofilm formation on the device itself . Our results showed that the psm-mec srRNA is not involved with sepsis or other relevant virulence phenotypes in S . epidermidis . As a previous study suggested that the psm-mec srRNA leads to gene regulatory changes in S . epidermidis [15] , based on the introduction of a psm-mec expressing plasmid into S . epidermidis , we also directly investigated whether the psm-mec locus has a gene regulatory impact in S . epidermidis . The most important regulatory effect of the psm-mec locus in S . aureus , by which the sometimes negative impact of the psm-mec locus on virulence in S . aureus was explained , has been reported to consist in the alteration of the expression of other , core genome-encoded PSMs [18] . PSM expression was altered only to a very low extent in the psm-mec-negative as compared to the wild-type S . epidermidis strains , with changes only significant for some PSMs and never exceeding a factor of ~ 1 . 5 ( Fig 6 ) . This demonstrates that there is only a very minor effect of the psm-mec srRNA on PSM expression when analyzed directly in the S . epidermidis background . Furthermore , we analyzed genome-wide gene expression in the psm-mec mutants of both strains by microarray analysis ( Tables 1 and 2 ) . For microarray analysis , strains were grown to the maximum of PSM-mec expression as determined by qRT-PCR ( 10 h ) ( Fig 7 ) . While we observed gene regulatory changes that were due to the psm-mec srRNA , they mostly comprised metabolic ( e . g . , riboflavin and purin/pyrimidine synthesis ) rather than virulence genes , and were inconsistent between the two strains . Notably , the results of the previously claimed impact of the psm-mec locus on virulence would be negative [13 , 15 , 18] , contrasting the positive effect we observed in the mouse sepsis model . Such a gene regulatory mechanism can thus be ruled out as underlying psm-mec-mediated development of S . epidermidis sepsis . Our results may explain the highly inconsistent phenotypes that have been attributed to psm-mec in S . aureus [11–13] , inasmuch as the psm-mec locus may exert effects in the background of S . aureus strains that differ from its original role in the CNS environment . One such possibility that remains to be investigated is that the highly expressed psm-mec mRNA interferes with other DNA or RNA sequences in S . aureus . Furthermore , the psm-mec srRNA barely exceeds the limits of the psm-mec gene [14] , which contrasts the only other case of an srRNA with an embedded peptide toxin in staphylococci , namely the well-described regulatory RNAIII of the staphylococcal accessory gene regulator ( Agr ) system . RNAIII significantly exceeds the boundaries of the embedded PSM peptide gene , hld [19] . Together , these observations suggest that the psm-mec srRNA does not serve a well-defined general purpose in virulence gene regulation . In conclusion , our study reveals that sepsis due to MRSE is mediated to a large extent by the PSM-mec peptide toxin , representing the first example of a toxin being made responsible for the development of CNS sepsis . Our study was largely based on the investigation of isogenic psm-mec mutants in clinical strains of S . epidermidis , using tissue culture and animal infection models . Future clinical work is needed to assess whether PSM-mec and/or other toxins contribute to sepsis in humans . Importantly , our results suggest that CNS sepsis may be amenable to virulence-targeted therapeutic approaches , such as those targeting the quorum-sensing system Agr [20] , which strictly regulates PSM expression [21] , or monoclonal antibody-based therapy directed against the toxin . Strain RP62A is a genome-sequenced clinical MRSE isolate [22] . Strain SE620 is an MRSE clinical isolate from Norway [23] . Isogenic Δpsm-mec deletion mutants and the psm-mec* start codon mutant were produced with the constructs previously used for S . aureus [12 , 14] , using a strategy with the allelic exchange vector pKOR1 [24] . The psm-mec locus and adjacent DNA do not differ between S . aureus and S . epidermidis [25] . For construction of the psm-mec* mutant , the start codon mutation was created by introducing a ClaI restriction site ( introducing ATC instead of the ATG start codon ) using primer PSMEClarev GAGGGTATGCATATCGATTTCACTGGTGTTATTACAAGC and primer PSMECladir ( reverse complement of PSMEClarev ) . Two PCR fragments were amplified using those primers and primers psmEatt1 and psmEatt2 , respectively [12] , cut with ClaI , ligated , and cloned into pKOR1 . The resulting plasmid was used for allelic replacement as described [24] . Growth patterns of the mutants were indistinguishable from those of the wild-type ( S3 Fig ) . Strains were grown in tryptic soy broth ( TSB ) , unless otherwise noted . Female , 6–10 weeks old , C57BL/6NCRl ( Charles River ) mice were used . The mice were injected via the tail vein with 5 x 108 CFU in 100 μl phosphate-buffered saline ( PBS ) of the indicated bacterial strains grown to mid-exponential growth phase and monitored for disease development every 8 h for up to 120 h . This dosis was determined to be minimally necessary to achieve mortality and production of inflammatory cytokines ( S4 Fig ) . Animals were euthanized immediately if showing signs of respiratory distress , mobility loss , or inability to eat and drink . Cytokine concentrations were measured at 2 and/or 12 h , as indicated , using commercially available ELISA kits ( IL-1β , TNF-α , BD BioSciences; CXCL1 , R&D Systems ) . For survival in whole blood experiments , about 108 bacteria in 100 μl Dulbecco’s PBS from mid-exponential growth phase were added to 500 μl heparinized human blood and mixtures were incubated for 6 h . Aliquots were taken at 2-h intervals , and CFU were determined by plating and incubating plates overnight at 37°C . For neutrophil interaction experiments , neutrophils were isolated from the venous blood of human volunteers as described [26] . Bacteria from mid-exponential growth phase were mixed with neutrophils at an MOI ( bacteria/neutrophils ) of 10:1 . Bacteria/neutrophil mixtures were incubated at 37°C , 5% CO2 , 90% humidity for 6 h . At 2-h intervals , 50 μl of Triton X-100 was added to the 200-μl bacteria/neutrophil suspensions , aliquots were plated , and plates incubated at 37°C overnight for CFU counting . Alternatively , the rate of neutrophil lysis promoted by the bacteria was determined after 4–h incubation using a lactate dehydrogenase ( LDH ) assay at an MOI of 100:1 . Biofilm formation was assessed in a semi-quantitative 96-well microtiter plate assay as previously described [27] , using TSB + 0 . 5% glucose . Relative PSM concentrations in culture filtrates were determined as described using reversed-phase high-pressure liquid chromatography/electrospray mass spectrometry ( RP-HPLC/ESI-MS ) [28] . Quantitative RT-PCR was performed as previously described [29] with the following oligonucleotides: psm-mecF , TGCATATGGATTTCACTGGTGTTA , psm-mecR , CGTTGAATATTTCCTCTGTTTTTTAGTTG , psm-mec probe , ATTTAATCAAGACTTGCATTCAG . Expression was measured relative to that of 16S RNA . Cultures were grown to the maximum of psm-mec expression as determined by qRT-PCR ( 10 h ) . Total RNA and cDNA were prepared as described [30] . Biotinylated S . aureus cDNA was hybridized to custom Affymetrix GeneChips ( RMLChip 3 ) with 100% coverage of chromosomal genes from strains S . epidermidis RP62A and scanned according to standard GeneChip protocols ( Affymetrix ) . Each experiment was replicated 3 times . Affymetrix GeneChip Operating Software was used to perform the preliminary analysis of the custom GeneChips at the probe-set level . Subsequent data analysis was performed as described [30] . The complete set of microarray data was deposited in NCBIs Gene Expression Omnibus ( GEO , http://www . ncbi . nlm . nih . gov/geo/ ) and is accessible through GEO Series accession number GSE85265 . To determine psm-mec mRNA stability in strains S . epidermidis SE620 and the psm-mec* start codon mutant , bacteria were cultured at 37°C for 10 hours . At time t = 0 min , rifampicin ( 50 mg/ml stock in DMSO ) was added to the cultures to a final concentration of 100 μg/ml . One-ml aliquots were taken and immediately centrifuged at 4°C to pellet cells , which were then frozen at -70°C . Remaining cultures were further incubated at 37°C with shaking; one-ml aliquots were taken at the indicated times and RNA was subsequently isolated from all cell pellets as described [29] . Samples were analyzed by qRT-PCR using primers psm-mecR and psm-mecF with a SuperScript III Platinum SYBR Green One-Step qRT-PCR kit ( Invitrogen ) according to the manufacturer’s instructions . Expression was measured relative to that of 16S RNA . Statistical analysis was performed using GraphPad Prism Version 6 . 0 . Comparisons were by 1-way or 2-way ANOVA for comparisons of three and more , and by unpaired t-tests for comparisons of 2 groups . Error bars show ±SEM . The animal protocol ( LB1E ) was reviewed and approved by the Animal Care and Use Committee at the NIAID , NIH , according to the animal welfare act of the United States ( 7 U . S . C . 2131 et . seq . ) . All mouse experiments were performed at the animal care facility of the NIAID , Building 50 , in accordance with approved guidelines . All animals were euthanized by CO2 at the end of the studies . Human neutrophils were isolated from blood obtained under approved protocols at the NIH Blood Bank or with a protocol ( 633/2012BO2 ) approved by the Institutional Review Board for Human Subjects , NIAID , NIH . All subjects were adult and gave informed written consent .	Coagulase-negative staphylococci ( CNS ) are the leading cause of sepsis in hospitalized patients , causing a significant number of deaths . This situation is further worsened by a limitation of therapeutic options due to the fact that most CNS infectious isolates are resistant to methicillin . CNS sepsis has been assumed to be due to on over-reacting immune response triggered by invariant bacterial surface structures . By using tissue culture and animal infection model-based evidence , we here show that in contrast to that notion , the PSM-mec toxin produced by methicillin-resistant strains of the leading CNS species Staphylococcus epidermidis has a strong impact on the severity of sepsis and its outcome . This is the first report to link a toxin to the pathogenesis of the most frequent bacterial cause of sepsis . Notably , these findings pave the way for anti-virulence strategies against this widespread and deadly type of infection .	[ "Abstract", "Introduction", "Results", "and", "Discussion", "Methods" ]	[ "blood", "cells", "medicine", "and", "health", "sciences", "toxins", "body", "fluids", "pathology", "and", "laboratory", "medicine", "immune", "cells", "gene", "regulation", "pathogens", "immunology", "microbiology", "staphylococcus", "aureus", "toxicology", "toxic", ...	2017	Toxin Mediates Sepsis Caused by Methicillin-Resistant Staphylococcus epidermidis
Major depression is a common and severe psychiatric disorder with a highly polygenic genetic architecture . Genome-wide association studies have successfully identified multiple independent genetic loci that harbour variants associated with major depression , but the exact causal genes and biological mechanisms are largely unknown . Tissue-specific network approaches may identify molecular mechanisms underlying major depression and provide a biological substrate for integrative analyses . We provide a framework for the identification of individual risk genes and gene co-expression networks using genome-wide association summary statistics and gene expression information across multiple human brain tissues and whole blood . We developed a novel gene-based method called eMAGMA that leverages tissue-specific eQTL information to identify 99 biologically plausible risk genes associated with major depression , of which 58 are novel . Among these novel associations is Complement Factor 4A ( C4A ) , recently implicated in schizophrenia through its role in synaptic pruning during postnatal development . Major depression risk genes were enriched in gene co-expression modules in multiple brain tissues and the implicated gene modules contained genes involved in synaptic signalling , neuronal development , and cell transport pathways . Modules enriched with major depression signals were strongly preserved across brain tissues , but were weakly preserved in whole blood , highlighting the importance of using disease-relevant tissues in genetic studies of psychiatric traits . We identified tissue-specific genes and gene co-expression networks associated with major depression . Our novel analytical framework can be used to gain fundamental insights into the functioning of the nervous system in major depression and other brain-related traits . Major Depression is a highly disabling mental health disorder that accounts for a sizable proportion of the global burden of disease . The global lifetime prevalence of major depression is around 12% ( 17% of women and 9% of men ) [1] , and ranks as the fourth most disabling disorder in Australia in terms of years lived with disability [2] . Major Depression has a complex molecular background , driven in part by a highly polygenic mode of inheritance . A recent genome-wide association study ( GWAS ) meta-analysis of 135 , 458 major depression cases and 344 , 901 controls identified 44 loci associated with the disorder [3] . A meta-analysis of this study with two other GWAS [4 , 5] ( 246 , 363 cases and 561 , 190 controls ) identified 102 independent variants associated with major depression [6] , 87 of which were replicated in an independent sample of 1 , 509 , 153 individuals . Detailed functional studies showed these loci harboured common ( minor allele frequency , MAF > 0 . 01 ) single nucleotide polymorphisms ( SNPs ) that regulate the expression of multiple genes in brain tissue with putative roles in central nervous system development and synaptic plasticity . Furthermore , large scale gene expression studies have identified altered immune pathways in whole blood [7 , 8] . These results suggest disease-associated SNPs modify major depression susceptibility by altering the expression of their target genes in a tissue-specific manner . Genes regulate the activity of one-another in large co-expression networks . Therefore , SNPs may not only affect the activity of a single target gene , but the activity of multiple biologically related genes within the same co-expression network to influence the manifestation of a phenotype . The integration of GWAS SNP genotype data with gene co-expression networks across multiple tissues may be used to elucidate biological pathways and processes underlying highly polygenic complex disorders such as major depression . Genome-wide gene expression data has been successfully integrated with SNP genotype data to prioritise risk genes and reveal possible mechanisms underlying susceptibility to a range of psychiatric disorders [9–11] . However , the collection of phenotype , SNP genotype , and gene expression data measured from the same individuals is impeded by cost and tissue availability , and identifying causal variants can be difficult due to linkage disequilibrium ( LD ) and confounding from environmental and technical factors . Recent approaches address these limitations by integrating GWAS summary statistics with independent gene expression data provided by large international consortia , such as the multi-tissue Genotype-Tissue Expression ( GTEx ) project [12–14] . The most recent release of the GTEx project ( version 7 ) contains SNP genotype data linked to gene expression across 53 tissues from 714 donors , including 13 brain tissues from 216 donors . This represents a valuable resource with which to study gene expression and its relationship with genetic variation , known as expression quantitative trait loci ( eQTL ) mapping [15] . Recent genetic studies have leveraged GTEx data in gene-based analyses to prioritise individual risk genes whose expression is associated with major depression [16 , 17] . While these analyses identified individual risk genes for major depression , they provide little insight into the molecular context within which the risk genes operate . We propose the use of GTEx data to build co-expression networks consisting of highly correlated genes in multiple tissues . The gene network modules provide a detailed map of gene co-expression in a given tissue , and provide a biological substrate to test the enrichment of major depression GWAS signals . Enriched gene modules can be characterised using gene pathway analysis , and provide a valuable resource for the integration of additional molecular data . This approach may characterise the broader molecular context of risk genes in major depression and thereby facilitate the identification of gene pathways for diagnostic , prognostic , and therapeutic intervention . We built gene co-expression networks using RNA-Seq data from 13 brain tissues and whole blood in GTEx ( v7 ) . In total , 464 tissue samples ( including 216 brain samples ) and 17793 protein-coding genes were used to build the co-expression networks , although the number of samples ( range: 80–369 ) and genes ( range: 14834–16892 ) differed by tissue ( Table 1 ) . The number of gene co-expression modules within each gene network ranged from 11 modules in brain cortex to 24 modules in amygdala , and the number of genes within a module ranged from as few as 30 ( 0 . 18% of network genes , amygdala ) to 9144 ( 55% of network genes , anterior cingulate cortex ) . We used gene pathway analysis to characterise biological processes in each co-expression module ( S1 Table ) . Co-expression networks were largely enriched for a single type of biological process ( e . g . transcriptional regulation or immune response ) . These data suggest the network gene co-expression modules represent biologically homogeneous units . To assign major depression risk SNPs to genes , we applied two gene-based strategies: first , proximity-based gene mapping with MAGMA , which assigns SNPs to the nearest gene within a genomic window; and second , eQTL gene mapping using eMAGMA , which uses tissue-specific SNP-gene associations from GTEx to assign SNPs to genes based on their association with gene expression . To further prioritise gene-level results , we performed a transcriptome-wide association study using S-PrediXcan . Both tissue-specific and P value thresholds for each gene-based method , calculated using Bonferroni correction for the number of associations , are shown in Table 1 . We identified 137 unique mapped depression-associated genes with MAGMA ( S2 Table ) , 217 significant tissue-specific gene associations with eMAGMA ( representing 99 unique mapped genes ) ( S3 Table ) , and 86 tissue-specific gene associations with S-PrediXcan ( S4 Table ) . A total of 41 genes were implicated by both MAGMA and eMAGMA mapping strategies in at least one tissue ( Fig 1; S5 Table ) . Among significant eMAGMA associations , 35 ( 16% ) also had a significant S-PrediXcan association in the same tissue ( S6 Table ) , and 16 associations were significant across all three gene-based methods ( Table 2 ) . A biological pathway analysis of gene lists produced by each method identified a single overlapping pathway—“Butyrophilin ( BTN ) family interactions”—across all three methods ( S7 Table ) . Taken together , these results point to potential functional links for the GWAS-associated variants and give higher credibility to genes with convergent evidence of association from multiple methods . We tested for the enrichment of MAGMA ( S2 Table; N = 137 ) and eMAGMA associations ( S3 Table; N unique = 99 ) in gene co-expression modules from the brain and whole blood . Gene modules in four brain tissues ( amygdala , cerebellar hemisphere , frontal cortex , and nucleus accumbens ) were enriched with MAGMA association signals , while one module in hypothalamus and one module in putamen were enriched with eMAGMA associations ( Table 3 ) . Gene modules enriched with MAGMA remained significant after removal of genes in the MHC region ( S8 Table ) , however modules enriched with eMAGMA associations were no longer significant after empirical multiple testing correction ( S9 Table ) . No enrichment of gene-based association signals was observed for modules identified in whole blood , despite the larger sample size ( and hence increase power ) compared to brain tissues . We plotted the overlap in gene modules enriched with gene-based major depression associations ( S2 Fig ) . A total of 217 genes overlapped across four modules enriched with MAGMA associations , suggesting similar biological processes may underlie the modular enrichments . Indeed , Pathway analysis of 217 genes overlapping four modules enriched with MAGMA gene-based associations revealed chemical synaptic transmission ( GO:0007268; P = 1 . 24 × 10−14 ) and the neuronal system ( R-HSA-112316; P = 6 . 62 × 10−10 ) pathways ( S10 Table ) . In gene pathway analyses of the major depression enriched modules , we found enrichment of neuronal and synaptic signalling pathways in amygdala , frontal cortex , nucleus accumbens , putamen ( e . g . trans-synaptic signalling in frontal cortex , P = 2 . 81 × 10−24 ) , as well as membrane trafficking related pathways in cerebellar hemisphere ( e . g . Membrane Trafficking , P = 2 . 19 × 10−13 ) and vascular-related pathways in hypothalamus ( e . g . blood vessel morphogenesis , P = 5 . 67 × 10−15 ) ( Fig 2 , S11 Table ) . Our network-based approach allows the discovery of major depression associated gene modules as well as the preservation ( or reproducibility ) of those associated modules across tissues . We assessed the preservation of gene co-expression modules across brain tissues and whole blood using the WGCNA modulePreservation algorithm , highlighting the preservation of modules enriched with major depression GWAS signals . Strong modular preservation ( Z score > 10 ) was observed across all brain regions , while weak to moderate preservation was observed in whole blood ( Z score < 10 ) . Major depression modules enriched with synaptic signalling pathways ( Modules M1 [Amygdala] , M3 [Frontal cortex] , M4 [Nucleus accumbens] , and M6 [Putamen] ) showed particularly strong preservation across brain tissues , while module M2 ( cerebellar hemisphere ) , enriched with cellular localisation and transport pathways , and module M5 ( Hypothalamus ) , enriched with vascular related pathways , showed relatively weak preservation ( Fig 3 ) . Our network-based approach identified novel gene candidates and gene co-expression networks enriched with both major depression GWAS signals and biological pathways related to synaptic signalling and neuronal development . The implicated modules were strongly preserved across brain tissues , with weaker preservation observed in whole blood . Our results suggest the study of gene co-expression networks may improve our understanding of the complex molecular systems governing the susceptibility to major depression and other neuropsychiatric disorders . More specifically , by describing the correlation structure of major depression risk genes with their nearest neighbours , we provide a large molecular substrate for detailed functional analyses than offered by traditional gene list-based approaches . The study of gene networks reduced the dimensionality of genome-wide gene expression data across multiple brain tissues and whole blood without the loss of important biological information , and thereby alleviated the multiple testing burden associated with traditional single gene-based methods . A similar network-based approach has been applied to gene expression data for other brain-related disorders , including post-traumatic stress syndrome [18] , schizophrenia [19] , and psychosis [20] . However , these studies typically included a small number of individuals ( fewer than 100 ) from a single brain region and are therefore limited in their statistical power and generalisability across different tissues . Our approach used a total of 216 individuals with a tissue sample from at least one of 13 brain regions , and 464 individuals with the inclusion of whole blood , thereby improving the resolution and robustness of gene networks . Our network approach identified between 11 ( Cortex ) and 24 ( Amygdala ) mutually exclusive modules within tissues , and ranged in size from 30 to 9144 genes . Each module was enriched with distinct and highly significant biological pathways ( e . g . immune signalling ) , suggesting our approach generated robust modules of functionally related genes . To identify genes and gene-sets associated with major depression , we assigned disease-associated SNPs to their nearest gene using both proximity and tissue-specific eQTL information . We first used MAGMA , a proximity-based approach that assigns SNPs to their nearest gene . This approach appropriately corrects for correlated SNPs ( i . e . linkage disequilibrium [LD] ) , and also adjusts for correlated gene expression in gene-set analysis and multiple-testing correction . However , SNPs are simply assigned to their nearest gene based on an arbitrary genomic window . It is well known that such proximity-based approaches often miss the functional SNP-gene association [21] . Therefore , we created eMAGMA , which modifies the annotation stage of the MAGMA pipeline by mapping SNPs to genes based on tissue-specific eQTL information in GTEx . We found some overlap ( N = 16 ) in risk genes between these two methods and those identified by S-PrediXcan , despite each method using different strategies for mapping SNPs to genes ( i . e . proximity versus eQTL information ) . This should not be surprising given the eQTL-based approaches ( eMAGMA and S-PrediXcan ) use cis-eQTLs which have been precomputed in a +/- 1 MB cis window around the transcription start site of a given gene . As such , in some instances , the most proximal gene ( identified by MAGMA ) will also be an eGene—that is , a gene whose expression is significantly associated with one or more SNPs in cis—identified by eMAGMA and S-PrediXcan . Our eMAGMA approach identified novel and biologically meaningful candidate risk gene associations for major depression across multiple tissues . Of 99 significant eMAGMA genes ( representing 217 unique gene-tissue associations ) , 58 were not identified by ( proximity-based ) MAGMA . Noteworthy among these associations is Complement Factor 4A ( C4A ) , recently implicated in the development of schizophrenia through its role as a mediator of synaptic pruning during postnatal development [22] . C4A was significant in 12 of 14 investigated tissues , including whole blood , and was one of 24 significant eMAGMA genes located on chromosome 6p21—a region with complex LD structure that flanks the centromeric end of the major histocompatibility complex . Future work using simulated GWAS data will be required to compare the performance ( e . g . true positive rate for the association with disease ) of our eMAGMA approach against other gene-based approaches . Further work to test whether C4A is involved in a shared mechanism between major depression and schizophrenia may also be performed using ( for example ) a joint GWAS—or phenome-wide association study—to identify loci that harbour one or more SNPs with pleiotropic effects on the two disorders . We tested for the enrichment of candidate risk genes in tissue-specific network co-expression modules , while adjusting for correlated gene expression , gene size , and gene density . We identified six co-expression modules across six individual brain tissues , four of which were involved in synaptic signalling and neuronal development pathways ( Amygdala , Putamen , Frontal cortex , and hypothalamus ) . These results align with recent pathway analyses of genetic associations in major depression , which identified genes and gene-sets involved in synaptic transmission and neuronal mechanisms , among other pathway groupings [3 , 5] . Furthermore , structural changes in frontal cortex have been identified in a recent meta-analysis of brain magnetic resonance imagining findings in adult major depression cases [23] , highlighting the central role of frontal cortex in major depression aetiology . It is important to note that co-expression networks across all ( N = 13 ) brain tissues contained a gene co-expression module associated with synaptic and neuronal pathways , but only four were enriched with major depression association risk genes . This suggests risk genes underlying major depression susceptibility manifest their effect in specific brain regions , consistent with tissue-specific gene expression [24] and highlighting the importance of studying multiple tissues in integrated studies of complex traits such as major depression . We assessed the preservation ( or reproducibility ) of connectivity patterns ( i . e . correlations ) between genes across multiple brain tissues and whole blood . This approach may determine whether the connectivity between genes within a network module associated with major depression differs both across brain tissues and between brain and whole blood , and may therefore identify ( peripheral ) surrogate tissues for molecular studies of major depression . We observed strong preservation of network modules across all brain regions , but not whole blood , suggesting blood-based molecular studies of major depression may fail to capture important disease-related processes in brain . Our findings therefore support the use of brain tissues from large international consortia , such as the GTEx study or the CommonMind consortium , for the characterisation of genetic association signals , despite reduced sample sizes compared to blood-based datasets and the potential for technical biases associated with the use of post-mortem samples . The results of this study should be interpreted in view of the following limitations . Our analyses rely on the stability of gene networks both within and between tissues . The relatively small sample sizes of brain tissues , which ranged from 80 in substantia nigra to 154 in cerebellum , may render the gene networks susceptible to spurious gene modules . A number of methods to assess the stability of clustered data are available [25] , but are too computationally intensive for high dimensional gene expression data . We therefore applied a permutation procedure , where each gene was randomised across individuals for the tissue with the smallest sample size , and clustered the data for the presence of modules ( methods ) . The permuted data did not yield a single co-expression module , suggesting our observed major depression module enrichments were not built upon spurious gene correlations . Nevertheless , our approach requires validation using independent expression data ( for example , from the CommonMind Consortium; www . synapse . org/cmc ) and the latest GWAS data for major depression , which was recently made available on some 800 , 000 individuals [6] . These analyses may further characterise co-expression modules in major depression , and identify molecular targets for follow-up functional studies . Current results support a common variant genetic architecture of major depression , where variants with relatively high frequency ( e . g . minor allele frequency > 0 . 01 ) in the general population , but low penetrance , are the major contributors to genetic susceptibility to the disorder . Therefore , as sample sizes grow larger , thousands of lead SNPs associated with major depression are likely to be identified , as shown in the latest GWAS meta-analysis of major depression [6] . With these impending data , new methods for the interpretation of genetic signals for major depression and other common complex disorders will be required . Our network-based approach provides a gene expression substrate across multiple human tissues for the integration and characterisation of GWAS signals . By exploiting the connectivity between genes , this approach will allow the identification of perturbations in the activity of a system rather than individual genes . Furthermore , network-based methods may identify regulatory hubs whose perturbation may have wider consequences for major depression and other ( co-morbid ) psychiatric and/or neurological disorders by virtue of their interaction with other genes . Finally , gene our co-expression networks can be integrated with epigenetic DNA methylation and chromatin interaction data from ( for example ) psychENCODE [26] and the impending release enhanced GTEx [27] to annotate and further prioritise risk SNPs and genes associated with major depression . These “multi-omic” analyses may identify regulatory elements involved in brain development and disease risk , and will be critical for understanding the properties of biological systems underlying complex disorders such as major depression . An overview of our analytical pipeline is shown in S1 Fig . Fully processed , filtered and normalised gene expression data for 13 brain tissues and whole blood ( Table 1 ) were downloaded from the Genotype-Tissue Expression project portal ( version 7 ) ( http://www . gtexportal . org ) ( Table 1 ) . Only genes with ten or more donors with expression estimates > 0 . 1 Reads Per Kilobase of transcript ( RPKM ) and an aligned read count of six or more within each tissue were considered significantly expressed . Within each tissue , the distribution of RPKMs in each sample was quantile-transformed using the average empirical distribution observed across all samples . Expression measurements for each gene in each tissue were subsequently transformed to the quantiles of the standard normal distribution . Detailed methods , including a description of population cohorts , quality control of raw SNP genotype data , and association analyses for the major depression GWAS is described elsewhere [3] . The major depression GWAS included a mega-analysis of 29 samples ( PGC29 ) ( 16 , 823 major depression cases and 25 , 632 controls ) of European ancestry and additional analyses of six independent European ancestry cohorts ( 118 , 635 cases and 319 , 269 controls ) . Cases in the PGC29 cohort satisfied diagnostic criteria ( DSM-IV , ICD-9 , or ICD-10 ) for lifetime major depression . Cases in the expanded cohort were collated using a variety of methods: Generation Scotland employed direct interviews; iPSYCH ( Denmark ) used national treatment registers; deCODE ( Iceland ) used national treatment registers and direct interviews; GERA used Kaiser-Permanente ( health insurance ) treatment records ( CA , US ) ; UK Biobank combined self-reported major depression symptoms and/or treatment for major depression by a medical professional; and 23andMe used self-report of treatment for major depression by a medical professional . Controls in PGC29 were screened for the absence of major depression . A combination of polygenic scoring and linkage disequilibrium score regression showed strong genetic homogeneity between the PGC29 and additional cohorts and between samples within each cohort . SNPs and insertion-deletion polymorphisms were imputed using the 1000 Genomes Project multi-ancestry reference panel [28] . Logistic regression association tests were conducted for imputed marker dosages with principal components covariates to control for population stratification . Ancestry was evaluated using principal components analysis applied to directly genotyped SNPs . Summary statistics for 10 , 468 , 942 autosomal SNPs were made available by the PGC and were utilized in our study . Gene co-expression modules were individually constructed for 13 brain tissues and whole blood using the weighted gene co-expression network analysis ( WGCNA ) package in R [29] . An unsigned pairwise correlation matrix—using Pearson’s product moment correlation coefficient—was calculated . An appropriate “soft-thresholding” value , which emphasizes strong gene-gene correlations at the expense of weak correlations , was selected for each tissue by plotting the strength of correlation against a series ( range 2 to 20 ) of soft threshold powers . The correlation matrix was subsequently transformed into an adjacency matrix , where nodes correspond to genes and edges to the connection strength between genes . Each adjacency matrix was normalised using a topological overlap function . Hierarchical clustering was performed using average linkage , with one minus the topological overlap matrix as the distance measure . The hierarchical cluster tree was cut into gene modules using the dynamic tree cut algorithm [30] , with a minimum module size of 30 genes . We amalgamated modules if the correlation between their eigengenes—defined as the first principal component of their genes’ expression values—was greater or equal to 0 . 8 . The stability of gene co-expression modules was assessed using a permutation procedure , where the expression values for each gene in substantia nigra—the brain tissue with the smallest sample size ( n = 88 ) —were randomly permuted , in a step-wise manner , 1000 times across individuals . This ensured each gene retained the same expression values , but the inherent correlations across individuals was removed . A WGCNA analysis was performed on the 1000 permuted gene expression datasets to identify gene modules , which were subsequently compared to the observed modules . We identified and prioritised risk genes for major depression using three approaches . First , we performed gene-level analyses using MAGMA v1 . 06 [31] . This approach assigns SNPs to their nearest gene using a pre-defined genomic window ( here a 35 kb upstream or 10 kb downstream of a gene body ) and computes a gene-based statistic based on the sum of the assigned SNP–log ( 10 ) P values while accounting for the correlation ( i . e . linkage disequilibrium ) between nearby SNPs . Second , we modified the MAGMA approach by integrating eQTL information from the GTEx project . That is , for a given interrogated tissue , we assigned SNPs to target genes based on significant ( FDR<0 . 05 ) SNP-gene associations in GTEx . This approach , which we will refer to as “eMAGMA” , is a tissue-specific , eQTL-informed method for assigning SNPs to genes . Gene-based statistics were subsequently computed using the sum of the assigned SNP–log ( 10 ) P values , in a similar manner to proximity-based MAGMA . Third , we used S-PrediXcan to integrate eQTL information from GTEx with major depression GWAS summary statistics to identify genes whose genetically predicted expression levels are associated with major depression . For S-PrediXcan , we used expression weights for 13 brain tissues and whole blood generated from GTEx ( v7 ) [32] , and LD information from the 1000 Genomes Project Phase 3 [33] . These data were processed with beta values and standard errors from the GWAS of major depression [3] to estimate the expression-GWAS association statistic . For each gene-level approach , we corrected for multiple testing using Bonferroni correction . For MAGMA , we corrected for the total number of genes tested ( i . e . 0 . 05/18 , 041 = 2 . 77×10−6 ) . For the multi-tissue eMAGMA and S-PrediXcan , we applied two correction thresholds ( Table 1 ) : a “liberal” threshold , which corrected for the number of tests within each tissue ( i . e . ignoring the number of tissues tested ) , and a “conservative” threshold , which corrected for the total number of tests performed ( i . e . all tests across all tissues ) . To identify gene co-expression modules enriched with major depression risk genes , we performed gene-set analysis of both ( proximity ) MAGMA and eMAGMA gene-level results in tissue-specific gene co-expression modules using the gene-sets analysis function in MAGMA v1 . 06 . The competitive analysis tests whether the genes in a gene-set ( i . e . gene co-expression module ) are more highly associated with major depression risk genes than other genes while accounting for gene size and gene density . We applied an adaptive permutation procedure [31] ( N = 10 , 000 permutations ) to obtain P values corrected for multiple testing . The 1000 Genomes European reference panel ( Phase 3 ) was used to account for Linkage Disequilibrium ( LD ) between SNPs . For each tissue and gene-based enrichment method , a quantile-quantile plot of observed versus expected P values was generated to assess inflation in the test statistic . Gene-set enrichment analyses were re-performed after excluding genes in the MHC region . Gene expression modules enriched with major depression GWAS association signals were assessed for biological pathways and processes using g:Profiler ( https://biit . cs . ut . ee/gprofiler/ ) [34] . Ensembl gene identifiers within major depression gene modules were used as input; we tested for the over-representation of module genes in Gene Ontology ( GO ) biological processes , as well as KEGG[35] and Reactome[36] gene pathways . The g:Profiler algorithm uses a Fisher’s one-tailed test for gene pathway enrichment; the smaller the P value , the lower the probability a gene belongs to both a co-expression module and a biological term or pathway purely by chance . Multiple testing correction was done using g:SCS; this approach accounts for the correlated structure of GO terms and biological pathways , and corresponds for an experiment-wide threshold of α = 0 . 05 . To examine the tissue-specificity of biological pathways , we assessed the preservation ( i . e . replication ) of network modules across GTEx tissues using the “modulePreservation” R function implemented in WGCNA [37] . Briefly , the module preservation approach takes as input “reference” and “test” network modules and calculates statistics for three preservation classes: i ) density-based statistics , which assess the similarity of gene-gene connectivity patterns between a reference network module and a test network module; ii ) separability-based statistics , which examine whether test network modules remain distinct in reference network modules; and iii ) connectivity-based statistics , which are based on the similarity of connectivity patterns between genes in the reference and test networks . For simplicity , we report two density and connectivity composite statistics: “Zsummary” and “medianRank” . A Zsummary value greater than 10 suggests there is strong evidence a module is preserved between the reference and test network modules , while a value between 2 and 10 indicates weak to moderate preservation and a value less than 2 indicates no preservation . The median rank statistic ranks the observed preservation statistics; modules with lower median rank tend to exhibit strong preservation than modules with higher median rank .	Although genome-wide association studies have identified genetic risk variants associated with major depression , our understanding of the mechanisms through which they influence disease susceptibility remains largely unknown . Genetic risk variants are highly enriched in non-coding regions of the genome and affect gene expression . Genes are co-expressed and regulate the activity of one another to form highly organized networks . In this study , we generate tissue-specific gene co-expression networks , each containing groups of functionally related genes or “modules” , to delineate gene co-expression and thereby facilitate the identification of gene processes in major depression . We developed and applied a novel research methodology ( called “eMAGMA” ) which integrates genetic and transcriptomic information in a tissue-specific analysis to identify risk genes and test for their enrichment in gene co-expression modules . Using this novel approach , we identified gene modules in multiple tissues that are both enriched with major depression genetic association signals and biologically meaningful pathways . We also show the disease implicated gene modules are strongly preserved across brain regions , but not in whole blood , suggesting unique patterns of gene co-expression within the two tissue types . Our novel analytical framework provides important insights into the functional genetics major depression and can be applied to other neuropsychiatric disorders .	[ "Abstract", "Introduction", "Results", "Discussion", "Methods" ]	[ "genome-wide", "association", "studies", "medicine", "and", "health", "sciences", "genetic", "networks", "volcanology", "magma", "body", "fluids", "neural", "networks", "brain", "neuroscience", "network", "analysis", "genome", "analysis", "geology", "mood", "disorders",...	2019	A gene co-expression network-based analysis of multiple brain tissues reveals novel genes and molecular pathways underlying major depression
In October 2012 , the Haitian Ministry of Health and the US CDC were notified of 25 recent dengue cases , confirmed by rapid diagnostic tests ( RDTs ) , among non-governmental organization ( NGO ) workers . We conducted a serosurvey among NGO workers in Léogane and Port-au-Prince to determine the extent of and risk factors for dengue virus infection . Of the total 776 staff from targeted NGOs in Léogane and Port-au-Prince , 173 ( 22%; 52 expatriates and 121 Haitians ) participated . Anti-dengue virus ( DENV ) IgM antibody was detected in 8 ( 15% ) expatriates and 9 ( 7% ) Haitians , and DENV non-structural protein 1 in one expatriate . Anti-DENV IgG antibody was detected in 162 ( 94% ) participants ( 79% of expatriates; 100% of Haitians ) , and confirmed by microneutralization testing as DENV-specific in 17/34 ( 50% ) expatriates and 42/42 ( 100% ) Haitians . Of 254 pupae collected from 68 containers , 65% were Aedes aegypti; 27% were Ae . albopictus . Few NGO workers reported undertaking mosquito-avoidance action . Our findings underscore the risk of dengue in expatriate workers in Haiti and Haitians themselves . Dengue is the most common mosquito-borne viral disease in the world , and resulted in an estimated 390 million infections and 96 million symptomatic cases throughout the tropics and subtropics in 2010 [1] , [2] . Over the last decade , the incidence and the severity of dengue have increased in the Americas , including the Caribbean [3] , [4] , where the four dengue virus-types ( DENV-1–4 ) that cause dengue and the mosquitoes ( i . e . , Aedes aegypti and Ae . albopictus ) that transmit DENV are endemic [1] , [5]–[7] . The risk of acquiring dengue can be greatly reduced by following key mosquito avoidance activities , such as applying mosquito repellent multiple times a day and wearing long sleeves , pants or permethrin-treated clothing [8] , [9] . Despite an absence of routine systematic surveillance data , dengue is likely endemic in Haiti , as it is in the Dominican Republic , which shares the island of Hispaniola with Haiti . Both Ae . aegypti and Ae . albopictus have been detected in Haiti , as have all four dengue virus-types [10] , [11] . A 2007 study in Port-au-Prince showed that 65% of children <5 years of age had evidence of prior infection with a DENV [12] , and a two-year prospective study in an outpatient clinic in Léogane found that 2% of patients presenting with undifferentiated fever tested positive for DENV infection by a rapid diagnostic test ( RDT ) [13] . Similarly , of 885 patients with acute febrile illness who were admitted to four hospitals in Haiti during 2012–2013 , 4% tested positive for DENV infection by RDT [14] . Although dengue has been documented in US military personnel and expatriate relief workers in Haiti in the past two decades [15]–[19] , visitors often do not regularly employ mosquito avoidance practices . In a survey conducted among American missionaries returning from Haiti in 2010 , only 24% reported using mosquito repellent multiple times a day [15] , and in a 1997 study of US military personnel in Haiti only 18% of febrile patients reported always using mosquito repellant [20] . In October 2012 , the International Federation of Red Cross and Red Crescent ( IFRC ) and Red Cross-Haiti alerted the Haitian Ministry of Public Health and Sanitation ( French acronym: MSPP ) and the US Centers for Disease Control and Prevention ( CDC ) of 25 recent RDT-positive dengue cases among Haitian and expatriate staff of non-governmental organizations ( NGOs ) based mostly in Port-au-Prince and Léogane . Seven ( 28% ) of the 25 cases were evacuated from Haiti for advanced medical care . To estimate the incidence of recent and previous DENV infection and identify demographic and behavioral risk factors for infection , we conducted a serologic survey among and administered a questionnaire to Haitian and expatriate NGO workers in Léogane and Port-au-Prince . Additionally , to better understand entomologic risk factors for human infection , we carried out an entomologic investigation around work sites and workers' residences . Of 776 NGO workers ( 106 expatriates and 670 Haitians ) in Léogane and Port-au-Prince , 181 ( 23% ) participated in the investigation , including 52 expatriates and 129 Haitians . Of those , 173 ( 96% ) provided a blood specimen for diagnostic testing . The majority of participants were male ( 76% ) and Haitian ( 71% ) , and the median age was 33 years ( Table 1 ) . Most participants worked in administrative or office duties , construction , or community or field work . Less than a quarter ( 21% ) of expatriates reported being born in a dengue-endemic country . Nearly all expatriates ( 94% ) but less than a quarter ( 23% ) of Haitians reported ever living in ( >1 month ) or traveling to ( >1 week ) a dengue-endemic country other than Haiti in their lifetime . Nearly all expatriates ( 96% ) , but less than half ( 39% ) of Haitians , reported ever hearing of dengue . Expatriates reported greater knowledge of DENV transmission ( 89% vs . 29% ) and dengue prevention ( 96% vs . 13% ) compared with Haitians . While 6% of expatriates reported a previous dengue diagnosis , no Haitians reported ever being diagnosed with dengue . Overall , 89% and 23% of expatriate staff reported receiving a yellow fever or Japanese encephalitis vaccination , respectively . In contrast , among Haitian staff , only 4% and 0% reported receiving a yellow fever vaccination or Japanese encephalitis vaccination , respectively . The majority ( 87% ) of expatriates and half ( 47% ) of Haitians reported using mosquito repellent , but less than half ( 44% ) of expatriates and only a small proportion ( 9% ) of Haitians reported using mosquito repellent multiple times a day ( Table 2 ) . While most expatriates and Haitians reported using a bed net , only a small percentage of expatriates and Haitians ( 10% and 2% , respectively ) reported using permethrin-treated clothing . Of the 52 expatriate workers , most ( 87% ) said they had made a travel consultation prior to their current trip to Haiti . Of the 45 workers who reported making a travel consultation , approximately half ( 47% ) went to a travel medicine clinic , 71% received mosquito-avoidance information during their consultation , and 39% received information about dengue . Compared with their Haitian colleagues , more expatriates reported having screens on their windows or doors , and air conditioning at their sleep site . Expatriates also reported more standing water and trash near their work site , off-hours ‘hang-out’ places , and sleep site as compared to Haitians ( Table 2 ) . DENV nucleic acid was not detected in any of the 173 NGO workers who provided blood specimens for dengue diagnostic testing . Both NS1 and anti-DENV IgM antibody were detected in one asymptomatic expatriate . Anti-DENV IgM antibody was detected in 17 ( 10% ) NGO workers ( 8 [15%] expatriates and 9 [7%] Haitians ) ( Table 3 ) . Of the 17 participants with evidence of current and/or recent DENV infection , six ( 35% ) participants ( five expatriates and one Haitian ) reported being ill in the past 90 days , five ( 29% ) reported missing at least one day of work , and three ( 18% ) were hospitalized and subsequently required medical evacuation to the Dominican Republic . Of these three evacuated participants , two had dengue with warning signs: one had menorrhagia and the other had a pleural effusion . Of 173 specimens tested , 161 ( 93% ) had detectable anti-DENV IgG antibody , including 41 ( 79% ) expatriates and all 121 Haitians . Prior DENV infection was confirmed by microneutralization assay in 17 ( 50% ) of the 34 IgG-positive/IgM-negative specimens from expatriates , and in all 42 randomly selected specimens from the 121 IgG-positive Haitians ( 95% confidence interval [CI]: 94 . 5%–100% ) ( Table 4 ) . Participants who reported working near “open water sources” had greater odds of having had a current and/or recent DENV infection ( odds ratio [OR] = 3 . 6 , 95% CI = 1 . 3–10 . 1; Table 3 ) . Participants who reported using mosquito repellent multiple times a day ( OR = 3 . 5 , 95% CI = 1 . 22–10 . 04 ) , having very good knowledge of infectious disease in Haiti ( OR = 3 . 6 , 95% CI 1 . 16–10 . 98 ) , and knowing how to prevent mosquito bites ( OR = 6 . 2 , 95% CI 1 . 92–19 . 72 ) had greater odds of having had a current and/or recent DENV infection . No other risk factors were found to be statistically significant ( Table 5 ) . One hundred premises were surveyed , including 8 NGO work sites and 28 adjacent buildings , 8 NGO residences and 27 adjacent buildings , and 29 Haitian employee residences . In total , 2 , 664 containers were inspected for immature mosquitoes . Of these containers , 756 ( 28% ) contained water , of which 198 ( 26% ) contained immature mosquitoes . We collected 254 pupae from 68 water-holding containers; Ae . aegypti was the most abundant mosquito species identified ( 65% ) , followed by Ae . albopictus ( 27% ) . The remaining 8% of mosquitoes identified were Ae . mediovittatus , Culex species , or were unidentifiable . Vector indices were similar between NGO work sites and residences . All mosquito abundance indices were elevated . For all premises combined , the Premise index was 61% , the Container index was 26% , and the Breteau index was 198 ( Table 6 ) . Pupae were found in 46% of tires , 29% of cans , 28% of water drums , 21% of cisterns , and 20% plastic containers that held water . In our investigation of NGO workers in Léogane and Port-au-Prince , Haiti , we found that a substantial proportion ( 15% of expatriates and 7% of Haitians ) had recently been infected with a DENV . Six of the infected workers reported being ill , and three required evacuation from Haiti for medical care . This rate of recent DENV infection is similar to findings from two previous studies in Haiti that reported rates of infection as high as 25% in expatriates and 29% in military personnel ( 12 , 15 ) . These findings demonstrate the risk of dengue for visitors to and residents of Haiti , and also illustrate the potential economic consequences of dengue through missed work days , hospitalization , and medical evacuation [27] , [28] . While there is no vaccine to prevent dengue , people at risk for DENV infection , such as the NGO workers in our investigation , can reduce their chance of getting infected through a number of preventive measures like applying mosquito repellent multiple times a day and wearing permethrin-treated clothing [8] , [9] . The NGO workers in our investigation variably employed these preventive measures: less than half of expatriates reported using mosquito repellent multiple times a day , and only 10% of expatriates used permethrin-treated clothing . The majority of expatriates in our investigation had a pre-travel health consultation . This rate is higher than previous reports of pre-travel health consultations among US citizens who traveled to countries with elevated public health risks [29] , [30] . However , in our investigation , less than half of expatriates received information about dengue during their pre-travel consultation . Improving pre-deployment education of expatriate NGO workers could increase the likelihood that they will employ preventive measures once they are in the field . All Haitian NGO workers had evidence of prior DENV infection , providing further evidence of dengue endemicity in Haiti . While our investigation and previous studies [11] , [12] , [14] , [31] collectively provide strong evidence for dengue endemicity in Haiti , questions remain about the clinical course of dengue among Haitians . Some studies have hypothesized that Haitians and persons of African descent are less likely to experience severe dengue [11] , [32] . In fact , in our investigation , some Haitian NGO staff said that dengue was not a health threat to Haitians and therefore declined to participate . In our investigation , only one Haitian ( 11% ) with a recent DENV infection reported dengue-like symptoms , and no Haitians reported symptoms of severe dengue . However , hospital and clinic-based surveillance conducted over the last two years in Haiti has shown that dengue is associated with both clinic visits and hospital admissions among Haitians [13] , [14] . Broader surveillance should be undertaken in Haiti to better understand the burden and clinical course of dengue in Haitians . Among all the sites we inspected in Léogane , the Premise and Breteau indices were 61% and 198 , respectively , reflecting an increased risk for DENV transmission [2] , [33] . These findings , according to WHO guidelines , indicate a need to prioritize vector control [26] . The density of immature vectors found in Léogane was greater than what was reported in a survey conducted in Port-au-Prince in May 2011 [34] . In our risk factor analysis , we found that NGO staff who worked near open water had an increased risk of DENV infection . Although this question did not clearly define open water with examples , anecdotally respondents interpreted this question to mean open containers filled with water . Other studies , including a recent study conducted in Saudi Arabia [35] , have identified proximity to standing water as a risk factor for DENV infection . Efforts should be made by NGOs and individuals to eliminate mosquito-breeding habitats by systematically reducing standing water in containers around worksites and residences . While vector control has had mixed results with regard to decreasing DENV infections [36] , it is still effective at reducing DENV-transmitting mosquitoes by eliminating container habitats [37] , [38] . However , dengue risk perceptions need to be addressed within these communities to make these efforts sustainable [39] , [40] . We found some unexpected results in our risk factor analysis . The use of mosquito repellent multiple times a day was associated with DENV infection . This finding is most likely due to sampling bias; four ( 24% ) of the recently infected participants in this investigation had received a diagnosis of dengue in the three months prior to our investigation and subsequently received education about dengue . Their responses to these questions likely reflected a change in risk perceptions and an awareness of dengue that they acquired after receiving their diagnosis [41] , [42] . Because the questionnaire did not distinguish whether information acquired after receiving a diagnosis of dengue had an effect on knowledge or practice , our findings related to this risk factor are likely spurious . Avoidance of mosquito bites by use of mosquito repellent is a widely supported measure for dengue prevention [8] , [9] , [43] . Our investigation was subject to several limitations . Because we used a convenience sample and less than half of all NGO workers at the Léogane-based NGOs and IFRC in Port-au-Prince participated , our results may not be representative of all workers at these NGOs . Also , we were not able to systematically evaluate whether there were demographic differences between participants and non-participants . While anecdotally some NGO workers declined to participate because of doubts about the relevance of dengue in Haiti , others were unavailable at the time of the survey . In addition , the investigation was conducted only among NGO workers in the two cities , and therefore our results may be not generalizable to other parts of the expatriate and Haitian population . Our analysis of risk factors for DENV infection was limited by a relatively small sample size . Finally , we were not able to link the results from the serosurvey with our findings from the entomologic investigation . Our findings underscore the risk of dengue in expatriate workers in Haiti . Expatriate NGO staff should be briefed on dengue risk and prevention measures prior to their arrival in Haiti , and NGOs should systematically employ vector control measures at their work sites and residences to reduce mosquito populations . We found evidence of acute dengue virus infections in Haitians and we found a high rate of previous infection among Haitians . Surveillance and research should be undertaken to better understand clinical dengue in Haitians .	Dengue is the most common mosquito-borne viral disease in the world , and caused an estimated 390 million infections and 96 million cases in the tropics and subtropics in 2010 . Over the last decade , the number of cases of dengue and the severity of dengue virus infections have increased in the Americas , including the Caribbean , yet little is still known about dengue in Haiti . Following an outbreak of dengue in mostly expatriate NGO workers , the investigators of this study took blood samples from expatriate and Haitian NGO workers living in two cities in Haiti and tested them for evidence of current , recent , and past dengue virus infection . They also investigated the amount and kinds of mosquitoes at homes and work sites . The study found recent infections among some Haitians and expatriates and widespread past infections among all Haitians and most expatriates . It also found that many people were not doing basic things to avoid mosquito bites , like applying mosquito repellent multiple times a day and wearing long sleeves or pants . These findings highlight the likely endemicity of dengue virus in Haiti , and the need to improve knowledge and awareness of dengue prevention among expatriates visiting Haiti and local Haitians .	[ "Abstract", "Introduction", "Results", "Discussion" ]	[ "invertebrates", "medicine", "and", "health", "sciences", "viral", "transmission", "and", "infection", "infectious", "disease", "epidemiology", "immunology", "vector-borne", "diseases", "microbiology", "tropical", "diseases", "geographical", "locations", "animals", "epidemi...	2014	Dengue Virus Infections among Haitian and Expatriate Non-governmental Organization Workers — Léogane and Port-au-Prince, Haiti, 2012
Health planners use forecasts of key metrics associated with influenza-like illness ( ILI ) ; near-term weekly incidence , week of season onset , week of peak , and intensity of peak . Here , we describe our participation in a weekly prospective ILI forecasting challenge for the United States for the 2016-17 season and subsequent evaluation of our performance . We implemented a metapopulation model framework with 32 model variants . Variants differed from each other in their assumptions about: the force-of-infection ( FOI ) ; use of uninformative priors; the use of discounted historical data for not-yet-observed time points; and the treatment of regions as either independent or coupled . Individual model variants were chosen subjectively as the basis for our weekly forecasts; however , a subset of coupled models were only available part way through the season . Most frequently , during the 2016-17 season , we chose; FOI variants with both school vacations and humidity terms; uninformative priors; the inclusion of discounted historical data for not-yet-observed time points; and coupled regions ( when available ) . Our near-term weekly forecasts substantially over-estimated incidence early in the season when coupled models were not available . However , our forecast accuracy improved in absolute terms and relative to other teams once coupled solutions were available . In retrospective analysis , we found that the 2016-17 season was not typical: on average , coupled models performed better when fit without historically augmented data . Also , we tested a simple ensemble model for the 2016-17 season and found that it underperformed our subjective choice for all forecast targets . In this study , we were able to improve accuracy during a prospective forecasting exercise by coupling dynamics between regions . Although reduction of forecast subjectivity should be a long-term goal , some degree of human intervention is likely to improve forecast accuracy in the medium-term in parallel with the systematic consideration of more sophisticated ensemble approaches . Infectious pathogens with short generation times pose public health challenges because they generate substantial near-term uncertainty in the risk of disease . This uncertainty is most acute and shared globally during the initial stages of emergence of novel human pathogens such as SARS [1] , pandemic influenza [2] , or the Zika virus [3] . However , at national and sub-national levels , uncertainty arises frequently for epidemic pathogens such as seasonal influenza , dengue , RSV and rotavirus; causing problems both for health planners and at-risk individuals who may consider changing their behavior to mitigate their risk during peak periods . Seasonal influenza affects populations in all global regions and is forecast annually in temperate populations , either implicitly or explicitly [4] . Peak demand for both outpatient and inpatient care is driven by peak incidence of influenza in many years [5] . Therefore , the efficient provision of elective procedures and other non-seasonal health care can be improved by accurate forecasts of seasonal influenza . Implicitly , most temperate health systems use knowledge of historical scenarios with which to plan for their influenza season . The current situation is then assessed against the deviation from the historical averages and worst-cases as observed in their own surveillance system . The United States Centers for Disease Control and Prevention ( CDC ) has sought to formalize regional and national forecasts by introducing an annual competition [6] . Each week , participating teams submit weekly estimates of incidence for the next four weeks , season onset , and timing and intensity of the peak . Methods used by teams include purely statistical models , [7–9] mechanistic models [10 , 11] machine learning and hybrid approaches [12–14] . Expert-opinion surveys have also been used and performed well . Some teams augment their forecasts of the official ILI data with the use of potentially faster or less-noisy datasets such as Google Flu Trends [15] . Here we describe our mechanistic-model-supported participation in the 2016-17 CDC influenza forecasting challenge , as an example of a disease forecasting process . We emphasize a subjective human component of this process and also describe a retrospective evaluation of the models for the previous six seasons . All the models described are implemented in the R package Dynamics of Interacting Community Epidemics ( DICE , https://github . com/predsci/DICE ) . The CDC Outpatient Influenza-like illness Surveillance Network ( ILINet ) Health and Human Services ( HHS ) region and national data were downloaded from the CDC-hosted web application FluView [16] and used to create a historic database of ILI cases . S1 Fig shows which states are grouped into each HHS region , along with the population of each region . Because we require an absolute number of cases per week , the CDC ILINet data is converted from percent ILI cases per patient to ILI cases . We estimate the absolute number of weekly ILI cases by dividing the weighted percent of ILI cases in the CDC data by 100 and multiplying it by the total weekly number of patients . We assume two outpatient visits per person per year so that the total weekly number of patients is estimated as: ( total regional population ) x ( 2 outpatient visits per person per year ) x ( 1 year/52 weeks ) . The estimate of two outpatient visits per year is based on two studies . In 2006 Schappert and Burt [17] studied the National Ambulatory Medical Care Surveys ( NAMCS ) and the National Hospital Ambulatory Medical Care Surveys ( NHAMCS ) and calculated an ambulatory rate of 3 . 8 visits per capita-year . The 2011 NAMCS [18] and NHAMCS surveys [19] estimated ambulatory visit rates of 3 . 32 visits per capita per year for physician’s offices , 0 . 43 for hospital emergency departments and 0 . 33 hospital outpatient departments . We sum these rates to get an outpatient visit per capita-year of 4 . 08 . We further estimate from the surveys that only half of these outpatient clinics are sites that report to ILINet , and hence we rounded to our two outpatient visits per year estimate . Specific humidity ( SH ) is measured in units of kg per kg and is defined as the ratio of water vapor mass to total moist air mass . Two other measurements of humidity are absolute humidity and relative humidity . SH is included in DICE as a potential modifier of transmissibility for this time period and uses Phase-2 of the North American Land Data Assimilation System ( NLDAS-2 ) database provided by NASA [20–22] . The NLDAS-2 database provides hourly specific humidity ( measured 2-meters above the ground ) for the continental US at a spatial grid of 0 . 125° which we average to daily and weekly values . The weekly data is then spatially-averaged for the states and CDC regions . School vacation schedules were collected for the 2014-15 and 2015-16 academic years for every state . For each state , a school district was identified to represent each of the three largest cities . Vacation schedules were then collected directly from the district websites . These three school vacation schedules were first processed to a weekly schedule with a value of 0 indicating class was in session all five weekdays and a 1 indicating five vacation days . Next , the representative state schedule was produced by averaging the three weekly district schedules . Region schedules are obtained by a population-weighted average of the state schedules . Similarly , the national schedule is generated by a population-weighted average of the regions . For the 2016-17 season we determine start and end times as well as spring and fall breaks from the previous year’s schedules . Thanksgiving and winter vacation timing was taken from the calendar where the winter break is assumed to be the last two calendar weeks of the year . Based on the proportion of schools closed and number of days closed , p ( t ) is assigned a value in the range [0 , 1] . For example in week ti , if all schools are closed for the entire week then we define the proportion of open schools p ( ti ) = 1 . However , if all schools have Monday and Tuesday off ( missing 2 of 5 days ) , then p ( ti ) = 0 . 4 . Similarly , if 3 of 10 schools have spring break ( entire week off ) , but the other 7 schools have a full week of class then p ( ti ) = 0 . 3 . If all schools have a full week of class then p ( ti ) = 0 . The DICE package has been designed to implement meta-population epidemic modeling on an arbitrary spatial scale with or without coupling between the regions . Our model for coupling between spatial regions follows ref [23] . We assume a system of coupled S-I-R equations ( susceptible-infectious-recovered ) for each spatial region . In this scenario , the rate at which a susceptible person in region j becomes infectious ( that is transitions to the I compartment in region j ) depends on: ( 1 ) the risk of infection from those in the same region j , ( 2 ) the risk of infection from infected people from region i who traveled to region j , and ( 3 ) the risk of infection encountered when traveling from region j to region i . To account for the three mechanisms of transmission , ref [23] defined the force of infection , or the average rate that susceptible individuals in region i become infected per time step as: λ i ( t ) = ∑ j = 1 D β j ( t ) m i j ∑ l = 1 D m l j I l ∑ p = 1 D m p j N p ( 1 ) where D is the total number of regions . In our case , unlike reference [23] , the transmissibility is not the same for all regions and it is allowed to depend on time: βj ( t ) . Given this force of infection we can write the coupled S-I-R equations for each region as: d S j d t =- λ j ( t ) S j , ( 2 ) d I j d t =λ j ( t ) S j - I j T g , ( 3 ) d R j d t =I j T g . ( 4 ) Eqs ( 2–4 ) are the coupled version of the familiar S-I-R equations , where Tg is the recovery rate ( assumed to be 2-3 days in the case of influenza ) . The mobility matrix , mij , of Eq 1 describes the mixing between regions . Thus , element i , j is the probability for an individual from region i , given that the individual made a contact , that that contact was with an individual from region j . As shown below , the sum over each row in the mobility matrix is one and in the limit of no mobility between regions the mobility matrix mij is the identity matrix so that λ i ( t ) = β i ( t ) I i N i and we recover the familiar ( uncoupled ) S-I-R equations: d S j d t =- β j ( t ) S j I j N j , ( 5 ) d I j d t =β j ( t ) S j I j N j - I j T g , ( 6 ) d R j d t =I j T g . ( 7 ) The level of interaction between spatial regions is determined by the mobility matrix and its interaction kernel , κ ( rij ) : m i j = N j κ ( r i j ) 1 ∑ k N k κ ( r i k ) . ( 8 ) This kernel is expected to depend on the geographic distance between the regions ( rij ) , and following Mills and Riley [23] we use a variation of the off-set power function for it: κ ( r i j ) = 1 1 + ( r i j / s d ) γ ( 9 ) where sd is a saturation distance in km and the power γ determines the amount of mixing between the regions: as γ decreases there is more mixing while as γ increases , mixing is reduced . In the limit that γ → ∞ there is no mixing between regions and we recover the uncoupled SIR Eqs ( 5–7 ) . The DICE package is designed to allow the estimation of these two parameters ( γ and sd ) , but they can also be set to fixed values . The S-I-R equations model the total population , but the data are the number of weekly observed cases or incidence rate for each spatial region ( I j R ) . The weekly incidence rate is calculated from the continuous S-I-R model by discretizing the rate-of-infection term λj ( t ) Sj ( or β j ( t ) S j I j N j in the uncoupled case ) : I j R ( t i ) = B j + p j C ∫ t i - 1 - Δ t t i - Δ t λ j ( t ) S j ( t ) d t , ( 10 ) scaling by percent clinical p j C , and adding a baseline Bj . The term p j C is the proportion of infectious individuals that present themselves to a clinic with ILI symptoms and Bj is a constant that estimates non-S-I-R or false-ILI cases . The integral runs over one week determining the number of model cases for week ti . Δt approximates the time delay from when an individual becomes infectious to when they visit a sentinel provider for ILI symptoms and is set to 0 . 5 weeks based on prior calibration [24 , 25] . Eq 10 describes how DICE relates its internal , continuous S-I-R model to the discrete ILI data . In the next section we describe the procedure used for fitting this property ( by optimizing the parameters: βj , sd , γ , Bj , and P j C ) to an ILI profile . To allow for different models for the force of infection/contact rate , we write this term in the most general way as a product of a basic force of infection , R j 0 , multiplied by three time dependent terms: β j ( t ) = R j 0 T g · F 1 ( t ) · F 2 ( t ) · F 3 ( t ) ( 11 ) The first time dependent term , F1 ( t ) , allows for a dependence of the transmission rate on specific-humidity , the second ( F2 ( t ) ) on the school vacation schedule , and the third ( F3 ( t ) ) allows the user to model an arbitrary behavior modification that can drive the transmission rate up or down for a limited period of time . For the purpose of the CDC challenge we only considered models involving either F1 ( t ) , F2 ( t ) , both , or none ( i . e . , the contact rate does not depend on time ) , and the functional form of these terms is discussed in S1 and S2 Texts . The DICE fitting procedure determines the joint posterior distribution for the model parameters using a Metropolis-Hastings Markov Chain Monte Carlo ( MCMC ) procedure . [26] We describe the procedure starting with the simpler uncoupled case . In the uncoupled scenario transmission can only happen within each HHS region , since there is no interaction between different spatial regions . The uncoupled regions are run sequentially and posterior distributions for the model parameters and forecasts are obtained . For each region , we simulated three MCMC chains each with 107 steps and a burn time of 2 × 106 steps . The smallest effective sample size that we report for any parameter is greater than 100 . After sampling from the individual posterior densities of each region , we calculate our national forecast as the weighted sum of the regional profiles with the weights given by the relative populations of the regions . The national curve was also fitted directly ( without any regional information ) using all the models and priors , but these direct results were only used at the end of the season when estimating the performance of each of our procedures . In the coupled scenario , the MCMC procedure uses Eqs ( 2–4 ) along with Eq ( 10 ) to simultaneously generate candidate profiles for the coupled ten HHS regions . The log-likelihood of the ten regional profiles is calculated and combined with the proper relative weights to generate a national log-likelihood which is minimized . It is important to note that in the coupled scenario we only optimize the national log-likelihood , and not the individual region-level likelihoods , but the parameters we optimize are still mostly region specific ( only sd and γ are not ) . We also tried fitting the coupled model to the regional log-likelihoods , however the results of the fits were not as accurate as the ones obtained when the national likelihood is optimized ( see Discussion ) . Both the coupled and uncoupled scenarios begin with the entire population of each region , minus the initial seed of infections which were fitted , in the susceptible state . We also fitted the onset time of each local epidemic and the proportion of infections that were cases . We also considered fitting an initially removed fraction R ( 0 ) , but found that pc and R ( 0 ) were very strongly correlated if both were fitted parameters . We fit the data using four model variants for the force of infection: ( i ) The force of infection as a function of specific humidity only ( H ) , ( ii ) school vacation only ( V ) , ( iii ) both ( HV ) or , ( iv ) none ( F ) . In all four variants we fit R j 0 and fixed Tg at a value of 3 days . The allowed range for R j 0 is between 1 and 3 with typical values being in 1 . 1 − 1 . 3 . We compare the performance of the model variants to: ( 1 ) Each other; ( 2 ) An ensemble model which is the equally weighted average of all the model variants; and ( 3 ) A historic NULL model , calculated as the weekly average of the past ten seasons ( excluding the 2009 pandemic season ) . We selected different FOI variants during different weeks . At the regional level , although we selected the most flexible humidity and school vacation assumptions ( HV ) more often ( 47 . 9% 134/280 ) than the alternatives ( Fig 1 ) , we did select humidity-only ( H , 47/280 16 . 8% ) , school-vacations-only ( V , 62/280 22 . 1% ) and fixed transmissibility ( F , 37/280 13 . 2% ) models on a number of occasions . For the national model , we only used the aggregated forecast of the regional models on 9/28 ( 32 . 1% ) occasions . ( The CDC challenge lasts 28 weeks and there are 10 HHS regions , hence the 28 and 280 in the denominators ) . For assumptions about the inferential prior , before the epidemic peaked ( EW 06 for the nation ) we rarely chose an uninformed prior ( UP ) . Most often we chose the heated data augmentation option ( HDA ) . Early in the season ( EWs 43-49 ) , when our options did not include the coupled procedure , we chose the informed prior ( IP ) or heated informed prior ( HIP ) most often . Once the season had peaked the uninformed prior was selected often both for the national profile and individual regions , we continued to select the data augmentation ( DA ) and HDA options for the nation well after the season has peaked ( EWs 12-14 , 16 and 18 ) . For assumptions about coupling , once the coupled procedure was available , it was often selected for both the national profile and most regions ( 1 , right panel ) , with the exception of regions 1 and 8 . We found that the coupled procedure used regions 8 ( and to a lesser extent 1 ) as a way to reduce the error to the national fit , at the cost of producing poor fits to these regions , hence their coupled results were rarely selected for submission . The aggregate option for the national selection was only selected at EWs 5 and 6 , the weeks prior to the peak and the peak week itself . For these two weeks our errors for both season targets and 1 − 4 week forecasts were large ( see Discussion below ) . At both the national and regional levels , the accuracy of the weekly %-ILI forecasts decreased as the lead time increased . The %ILI 1-4 week forecast and observed data for the national data and the three largest ( by population ) HHS regions ( S1 Fig ) : 4 , 5 , and 9 is shown in Fig 2 . Early in the season , up to and including EW01 , the national curve is nearly identical to the historic national curve , whereas our mechanistic forecasts consistently overestimated incidence . After EW01 , our national predictions improve significantly , while the historic curve no longer follows the 2016-17 national profile . Similarly , for the three largest HHS regions , the historic curve is similar to the 2016-17 profile until EW01 , at which point they start to deviate and our forecasts become more accurate . Averaged over the entire season , our selected national forecast does better than the historic NULL model only for the 1-week prediction window ( top left panel in Fig 2 ) . However , for the largest HHS region ( region 4 , top right panel ) we perform better for all four prediction time horizons , for region 9 ( bottom left panel ) for the first three , and for region 5 ( bottom right panel ) for the first two . Although our forecasts gave potentially useful information over and above the NULL model for the timing of the peak week ( Fig 3 ) and for the amplitude of peak intensity , the peak week of EW06 was the same as the historical mean . Between EW50 ( eight weeks before the season peaks ) and EW04 ( two weeks before the season peak ) our forecast correctly predicted to within ±1 week of the observed peak week ( EW06 ) . One week before the season peaks , and at the peak week ( EW05 and EW06 ) , our model forecast has an error of two weeks . Forecasts based on the mechanistic model performed better than the historic NULL model for the peak intensity ( Fig 3A/3B/3C/3D ) . Two weeks before the peak week ( and three weeks early in the season ) we started predicting the correct peak intensity of 5 . 1% ( to within ±0 . 5% ) . The mean and median historic values are significantly lower ( 4 . 4% and 4 . 1% respectively ) and outside the ±0 . 5% range . Our apparent forecast performance for intensity appears to drop off at the end of the season . However , this is an artifact of the forecasting work flow . Once the peak had clearly passed , the final model was selected for reasons other than the peak intensity and the already-observed peak intensity was submitted . Selected forecasts based on the mechanistic model did not accurately predict onset . Both the mean of onset ( EW51 ) and median ( EW50 ) historic values were within a week of the observed 2016-17 onset week ( EW 50 ) . However , our model was unable to properly predict the onset until it happened . As with peak values , once onset had been observed in the data , we used the observed value in our formal submission , which was not reflected in onset values from the chosen model . Similar models were correlated with each other in their forecasts of 1- to 4-week ahead ILI , but with decreasing strength as forecasts were made for longer time horizons . S5 Fig in the SI shows the Pearson correlation between the 32 models calculated for the 1- , 2- , 3- , and 4- week forward forecast using all 28 weeks of the challenge . The high correlation shows how closely related to each other many of the models are . But this Fig also shows that this correlation decreases when the forecast horizon increases and that there is a spread in the predictions ( manifested by negative values of the Pearson correlation ) . Once the challenge was over , we examined retrospectively the performance of all mechanistic model variants over the course of all seasons in the historical database and separately for the 2016-17 season . To assess the quality of all the near-term forecasts ( 1 − 4 weeks ) from the different models and assumptions about priors , we show in Fig 4 their weekly CDC score ( see Methods ) , for the 4 different forecast lead times ( 1- , 2- , 3- and 4- weeks ahead ) , and for the prediction of the National %ILI intensity . The models are arranged based on their CDC score ( averaged across all weeks , numbers on the right y-axis ) from best ( top ) to worst ( bottom ) . Coupled models were more accurate than non-coupled models for the 2016-17 season and for historical seasons for all 4 lead times . The model labeled ‘ensemble’ is the average of the 32 model variants . For the 1- , 2- , and 3-weeks ahead forecasts the subjectively selected model does substantially better than the ensemble model and for the 4-weeks ahead forecast they score practically the same . The performance of mechanistic models was comparable to that of the historical average NULL model at the beginning and end of the season . However , in the middle of the season , when there is greater variation in the historical data , the performance of the best mechanistic model variants was substantially better than that of the historical average model . For predicting ILI incidence for the 2016-17 season , which followed similar trend to the historical average , coupled models that used data augmentation were more accurate than coupled models that did not use data augmentation . However , on average for historical seasons , coupled models that did not use augmented data were more accurate than those that did . Also , on average for historical seasons , coupled models that included humidity were more accurate than those that did not ( see dark banding in upper portion of charts on the right hand side of Fig 4 ) . We examined the performance of the different model variants for individual regions for the near-term forecasting of %-ILI ( S3 and S4 Figs ) . Again , the coupled models with uninformative prior outperformed other model variants . Although for some regions the improvement in forecast score for the uninformative prior variants over other coupled variants was less pronounced ( regions 1 and 7 ) , these models never appeared to be inferior to the other variants . In a similar way , we examined the forecast accuracy of different mechanistic model variants in forecasting season-level targets: onset , peak time and peak intensity for the 2016-17 season and on average across all seasons ( Fig 5 ) . Here too the models are arranged based on their overall performance from best to worse ( top to bottom ) . Again , for all three targets during 2016-17 , the coupled uninformative model variant was at least as good as other coupled options and better than the non-coupled variants . For all three season targets , the selected model performed better ( and in the case of peak week significantly better ) than the ensemble model . We note that in the latter part of the season , after the single observed onset and peak had passed , results from a single season do not contain much information about model performance . However , the performance of the coupled uninformative prior model was on average better than other model variants across the historical data and different epidemic weeks for all three targets , other than one exception . From EW01 onwards for peak intensity , uncoupled heated augmented prior variants performed better than did coupled uninformative prior model variants . We were able to compare our performance over the course of the season to the performance of the other teams using the public website that supports the challenge ( www . cdc . gov/flusight ) . Averaged across all weeks of forecast and all forecast targets , we were ranked 13 out of 29 teams . For 1- , 2- , 3- and 4-weeks ahead forecasts we were ranked 6 , 11 , 9 and 16 respectively; again , out of 29 teams . We were ranked 14 for the timing of onset , 5 for the timing of the peak and 14 for intensity of the peak . Probing beyond the overall rankings , our performance was similar to the other better-performing teams in the challenge . Also , our performance improved substantially as measured by both in absolute terms and relative to other teams across the season ( S6 and S7 Figs ) . In this study , we have described our participation in a prospective forecasting challenge . Although we drew on results from a large set of mechanistic models , our single forecast for each metric was made after choosing between available model results for that metric in that week and was therefore somewhat subjective . We performed poorly at the start of the competition when our mechanistic models consistently over-estimated incidence . However , during the middle phase of the season , our models produced less biased estimates and consistently outperformed non-mechanistic models based on the average of historical data . A robust testing of model variants using historical data suggests that spatially coupled models are systematically better than historical NULL models during the middle of the season and are not significantly worse even at the start of the season . We evaluated a simple ensemble and showed that the subjective model choice was better . However , the ranking of individual models suggests that an ensemble of coupled models may outperform our subjective choice . We are considering exactly this experiment for the upcoming season . This study is slightly different from some prior studies of influenza forecasting [29] in that it describes and assesses a subjective choice between multiple mechanistic models as the basis of a prospective forecast , rather than describing the performance of a single model or single ensemble of models used for an entirely objective forecast . Although this could be viewed as a limitation of our work , because individual subjective decisions cannot be reproduced , we suggest that the explicit description of a partially subjective process is a strength . In weather forecasting , there is a long history of evaluating the accuracy of entirely objective forecasts versus partially subjective forecasts [30 , 31] . Broadly , for each different forecast target and each forecast lead-time , there has been a gradual progression over time such that objective forecasts become more accurate than subjective forecasts . We note also that although we describe the subjective process as it was conducted , we also provide a thorough retrospective assessment of the predictive performance of each model variant . We may refine our ensemble approach for future iterations of the competition . It seems clear that the coupled models produce more accurate forecasts than the uncoupled models for most targets , so we would consider an ensemble only of the coupled model variants . We will also consider weighted ensembles of models and attempt to find optimal weights by studying all prior years . Also , data were often updated after being reported and we did not include an explicit reporting model in our inferential framework ( also sometimes referred to as a backfill model ) . Rather , we used knowledge of past adjustments to data during our discussions and eventual subjective choice of models . We aim to include a formal reporting model in future versions of our framework . By reflecting on our choices and their performance , we can evaluate the importance of a number of different model features . Our coupled model variants performed much better than uncoupled variants consistently across the 2016-17 season , for different targets and when evaluated using the historical data . This prospective study supports recent retrospective results suggesting that influenza forecasts can be more accurate if they explicitly represent spatial structure [32 , 33] . Given that the model structure we used to represent space was relatively coarse [23] , further work is warranted to test how forecast accuracy at finer spatial scales can be improved by models that include iteratively finer spatial resolution . In submitting forecasts based on uninformed mechanistic priors using an uncoupled model at the start of the season , we failed to learn lessons that have been present in the influenza forecasting literature for some time [29] . Historical variance is low during the start of the season and the growth pattern is not exponential . Therefore , it would be reasonable to forecast early exponential growth only in the most exceptional of circumstance , such as during the early weeks of a pandemic . Model solutions that are anchored to the historical average in some way , such as by the use of augmented data for not-yet-seen time points , are likely to perform better . Also , forecasting competitions may want to weight performance differentially across time , with greater weight given to forecasts during periods where there is a higher variance in incidence . Models that included humidity forcing performed better on average in our analysis of all historical data than equivalent models that did not include those terms , especially for the forecasting of ILI 1- to 4-weeks ahead [34] . However , we did not see similar support for the inclusion of school vacation terms improving accuracy , which has been suggested in a retrospective forecasting study at smaller spatial scales ( by this group ) [35] . The lack of support for school vacations in the present study could indicate that the prior work was under-powered or that averaging of school vacation effects across large spatial scales—both in the data and the model—degrades its contribution to forecast accuracy . Also , we chose to present our accuracy of predicting peak height and timing relative to actual epidemiological week so that it was consistent with our presentation of accuracy of other forecast targets . However , it may be more appropriate in some circumstances to present accuracy of targets associated with the peak relative to the eventual peak [10] . We found the experience of participating in a prospective forecasting challenge to be different to that of a retrospective modeling study . The feedback in model accuracy was much faster and the need for statistically robust measures of model likelihood or parsimony less obvious . We encourage the use of forecasting challenges for other infectious disease systems as a focus for better understanding of underlying dynamics in addition to the generation of actionable public health information .	It is estimated that there are between 3 and 5 million worldwide annual seasonal cases of severe influenza illness , and between 290 000 and 650 000 respiratory deaths . Influenza-like illness ( ILI ) describes a set of symptoms and is a practical way for health-care workers to easily estimate likely influenza cases . The Centers for Disease Control and Prevention ( CDC ) collects and disseminates ILI information , and has , for the last several years , run a forecasting challenge ( the CDC Flu Challenge ) for modelers to predict near-term weekly incidence , week of season onset , week of peak , and intensity of peak . We have developed a modeling framework that accounts for a range of mechanisms thought to be important for influenza transmission , such as climatic conditions , school vacations , and coupling between different regions . In this study we describe our forecast procedure for the 2016-17 season and highlight which features of our models resulted in better or worse forecasts . Most notably , we found that when the dynamics of different regions are coupled together , the forecast accuracy improves . We also found that the most accurate forecasts required some level of forecaster interaction , that is , the procedure could not be completely automated without a reduction in accuracy .	[ "Abstract", "Introduction", "Methods", "Results", "Discussion" ]	[ "medicine", "and", "health", "sciences", "pathology", "and", "laboratory", "medicine", "statistics", "atmospheric", "science", "influenza", "pathogens", "biological", "locomotion", "health", "care", "seasons", "mathematics", "forecasting", "patients", "humidity", "researc...	2019	Forecasting national and regional influenza-like illness for the USA
Sleep is an ancient animal behavior that is regulated similarly in species ranging from flies to humans . Various genes that regulate sleep have been identified in invertebrates , but whether the functions of these genes are conserved in mammals remains poorly explored . Drosophila insomniac ( inc ) mutants exhibit severely shortened and fragmented sleep . Inc protein physically associates with the Cullin-3 ( Cul3 ) ubiquitin ligase , and neuronal depletion of Inc or Cul3 strongly curtails sleep , suggesting that Inc is a Cul3 adaptor that directs the ubiquitination of neuronal substrates that impact sleep . Three proteins similar to Inc exist in vertebrates—KCTD2 , KCTD5 , and KCTD17—but are uncharacterized within the nervous system and their functional conservation with Inc has not been addressed . Here we show that Inc and its mouse orthologs exhibit striking biochemical and functional interchangeability within Cul3 complexes . Remarkably , KCTD2 and KCTD5 restore sleep to inc mutants , indicating that they can substitute for Inc in vivo and engage its neuronal targets relevant to sleep . Inc and its orthologs localize similarly within fly and mammalian neurons and can traffic to synapses , suggesting that their substrates may include synaptic proteins . Consistent with such a mechanism , inc mutants exhibit defects in synaptic structure and physiology , indicating that Inc is essential for both sleep and synaptic function . Our findings reveal that molecular functions of Inc are conserved through ~600 million years of evolution and support the hypothesis that Inc and its orthologs participate in an evolutionarily conserved ubiquitination pathway that links synaptic function and sleep regulation . Sleep is an evolutionarily ancient behavior present in vertebrates and invertebrates [1] . The similar characteristics of sleep states across animal phylogeny suggest that both the functions of sleep and the regulation of sleep may have a common evolutionary basis [2] . In diverse animals including mammals and insects , sleep is regulated similarly by circadian and homeostatic mechanisms [3] . The circadian regulation of sleep is better understood at a molecular level , and numerous studies have revealed that the underlying genes and intracellular pathways are largely conserved from flies to humans [4–10] . In contrast , the molecular mechanisms underlying the non-circadian regulation of sleep—including those governing sleep duration , consolidation , and homeostasis—remain less well defined , and their evolutionary conservation is largely unexplored . Mutations of the Drosophila insomniac ( inc ) gene [11] severely curtail the duration and consolidation of sleep but do not alter its circadian regulation [11 , 12] . inc encodes a protein of the Bric-à-brac , Tramtrack , and Broad / Pox virus zinc finger ( BTB/POZ ) superfamily [13] , which includes adaptors for the Cullin-3 ( Cul3 ) E3 ubiquitin ligase complex [14–17] . Cul3 adaptors have a modular structure , in which the BTB domain binds Cul3 and a second distal domain recruits substrates to the Cul3 complex for ubiquitination [14–17] . The BTB domain also mediates adaptor self-association , enabling the oligomerization of Cul3 complexes and the efficient recruitment and ubiquitination of substrates [18 , 19] . Biochemical and genetic evidence supports the hypothesis that Inc is a Cul3 adaptor [11 , 12] . Inc and Cul3 physically interact in cultured cells [11 , 12] and Inc is able to self-associate [12] . In vivo , neuronal RNAi against inc or Cul3 strongly reduces sleep [11 , 12] , and reduction in the levels of Nedd8 , a protein whose conjugation to Cullins is essential for their activity , also decreases sleep [11] . While Inc is thus likely to function as a Cul3 adaptor within neurons to promote sleep , the neuronal mechanisms through which Inc influences sleep are unknown . Three proteins similar to Inc—KCTD2 , KCTD5 , and KCTD17—are present in vertebrates [11 , 20 , 21] , but their functions in the nervous system are uncharacterized and their functional conservation with Inc has not been addressed . KCTD5 can self-associate and bind Cul3 [20] , suggesting that it may serve as a Cul3 adaptor , yet no substrates have been identified among its interacting partners [22 , 23] . One KCTD17 isoform has been shown to function as a Cul3 adaptor for trichoplein , a regulator of primary cilia [24] . However , trichoplein-binding sequences are not present in Inc , KCTD2 , KCTD5 , or other KCTD17 isoforms , and trichoplein is not conserved in Drosophila . Thus , it remains unclear whether Inc and its vertebrate homologs have conserved molecular functions , particularly within neurons and cellular pathways relevant to sleep . Here , we assess the functional conservation of Inc and its mammalian orthologs and elucidate a neuronal mechanism through which they may impact sleep . Inc and each of its orthologs bind Cul3 and self-associate , supporting a universal role for these proteins as Cul3 adaptors . Inc and its orthologs furthermore exhibit biochemical interchangeability within Inc-Inc and Inc-Cul3 complexes , indicating that the oligomeric architecture of Inc-Cul3 complexes is highly conserved . Strikingly , KCTD2 and KCTD5 can functionally substitute for Inc in vivo and restore sleep to inc mutants , indicating that these Inc orthologs readily engage the molecular targets through which Inc impacts sleep . Our studies furthermore reveal that Inc and its orthologs localize similarly within fly and mammalian neurons and are able to traffic to synapses . Finally , we show that inc mutants exhibit defects in synaptic structure and physiology , indicating that inc is essential for both sleep and synaptic function . Our findings demonstrate that molecular functions of Inc are conserved from flies to mammals , and support the hypothesis that Inc and its orthologs direct the ubiquitination of conserved neuronal proteins that link sleep regulation and synaptic function . Inc functions in neurons to impact sleep [11 , 12] . To assess whether Inc orthologs are expressed in the mammalian nervous system , we performed RT-PCR on whole adult mouse brain RNA using primers specific for KCTD2 , KCTD5 , and KCTD17 . All three genes are expressed in the brain ( Fig 1A ) , and in situ hybridizations reveal expression within cortex , thalamus , striatum , pons , and cerebellum among other brain regions ( S1 Fig ) . Cloning of RT-PCR products revealed single transcripts for KCTD2 and KCTD5 , encoding proteins of 263 and 234 residues respectively , and two alternatively spliced transcripts for KCTD17 , encoding proteins of 225 and 220 residues with distinct C-termini ( Fig 1B and S2A Fig ) . These KCTD17 isoforms , KCTD17 . 2 and KCTD17 . 3 , have not been characterized previously and lack residues of a longer KCTD17 isoform required to bind trichoplein ( S2B Fig ) [24] . Thus , KCTD17 . 2 and KCTD17 . 3 are likely to have different molecular partners . KCTD17 . 2 and KCTD17 . 3 behaved indistinguishably in our experiments except where noted below . Inc and its mouse orthologs share ~60% sequence identity , and have variable N-termini followed by a BTB domain , an intervening linker , and conserved C-termini ( S2A Fig ) [11 , 20 , 21] . We next assessed the expression of KCTD2 , KCTD5 , and KCTD17 proteins , using a polyclonal anti-KCTD5 antibody that cross-reacts with mouse Inc orthologs and Drosophila Inc ( S3A Fig ) . This antibody detected a strongly reactive species of ~26 kD and additional species of ~28 to 29 kD in extracts from mouse and rat brain , cultured rat cortical neurons , and human 293T cells ( Figs 1C and 5D , and S3A Fig ) , consistent with the range of molecular weights predicted for KCTD2 , KCTD5 , and KCTD17 ( Fig 1B ) . The size of these immunoreactive species and their biochemical properties , described further below , indicate that they correspond to one or more isoforms of KCTD2 , KCTD5 , and KCTD17 . The expression of Inc orthologs in the mammalian brain and Inc in the fly brain [11] , together with the similarity of their primary sequences , suggests that Inc defines a protein family that may have conserved functions in the nervous system . Inc and KCTD5 are able to bind Cul3 and to self-associate [11 , 12 , 20] , key attributes of BTB adaptors [14–17] . To determine whether these attributes are universal to Inc orthologs and their isoforms expressed in the nervous system , we first examined the physical interactions of these proteins with mouse Cul3 . Co-immunoprecipitations revealed that KCTD2 , KCTD5 , and KCTD17 are able to associate with mouse Cul3 ( Fig 2A ) , with KCTD5 exhibiting stronger or more stable interactions than KCTD2 and KCTD17 . Thus , the ability to bind Cul3 is a conserved property of Inc and its mouse orthologs . Epitope-tagged Cul3 also co-immunoprecipitated the endogenous ~26 kD species detected by anti-KCTD5 antibody , confirming that this species represents KCTD5 or another Inc ortholog ( S3B Fig ) . To assess the extent to which the Inc-Cul3 interface is evolutionarily conserved , we tested whether Drosophila Inc is able to associate with mouse Cul3 in a cross-species manner . We observed that fly Inc and mouse Cul3 interact ( Fig 2A ) , indicating that Inc readily assembles into mammalian Cul3 complexes . Conversely , we tested whether mouse KCTD2 , KCTD5 , and KCTD17 can associate with fly Cul3 in Drosophila S2 cells , and observed that each Inc ortholog associated with fly Cul3 in a manner indistinguishable from Inc ( Fig 2B ) . The interchangeable biochemical associations of Inc family members and Cul3 indicate that the Inc-Cul3 interface is functionally conserved from flies to mammals . Self-association is a critical property of BTB adaptors that enables the oligomerization of Cul3 complexes and that stimulates substrate ubiquitination [18 , 19] . To test whether Inc orthologs self-associate in a manner similar to Inc [12] , we co-expressed FLAG- and Myc-tagged forms of each Inc ortholog in mammalian cells and performed co-immunoprecipitations . KCTD2 , KCTD5 , and KCTD17 each homomultimerized strongly ( Fig 3A ) . Thus , homo-oligomerization is a shared property of Inc family proteins . The presence of three Inc orthologs in mammals and their likely co-expression in brain regions such as thalamus and cortex ( S1 Fig ) led us to test whether these proteins can also heteromultimerize . We observed robust heteromeric associations between all pairwise combinations of Inc orthologs ( Fig 3A ) , a property that may enable functional redundancy in vivo or the assembly of functionally distinct complexes . To further probe the multimeric self-associations of Inc family members , we tested whether KCTD2 , KCTD5 , and KCTD17 can heteromultimerize with Drosophila Inc . We observed that each Inc ortholog associates readily with Inc in both mammalian and Drosophila cells ( Fig 3A and 3B ) . Thus , the multimerization interface of Inc family members is highly conserved through evolution . Together with the interchangeable associations of Inc family members and Cul3 ( Fig 2 ) , these findings strongly suggest a conserved oligomeric architecture for complexes containing Cul3 and Inc family members . The biochemical interchangeability of Inc and its mouse orthologs prompted us to test whether Inc function , including its presumed ability to serve as a Cul3 adaptor in vivo and ubiquitinate specific neuronal proteins relevant to sleep , is conserved between flies and mammals . We therefore generated UAS transgenes expressing Myc-tagged forms of Inc , KCTD2 , KCTD5 , KCTD17 . 2 , and KCTD17 . 3 , integrated each at the attP2 site [25] , and backcrossed these lines to generate an isogenic allelic series . Expression of these transgenes panneuronally with elavc155-Gal4 yielded similar levels of expression for Inc , KCTD2 , and KCTD5 , weak expression of KCTD17 . 2 , and low levels of KCTD17 . 3 expression near the threshold of detection ( Fig 4A ) . Next , we assessed the behavioral consequences of expressing mouse Inc orthologs in vivo . Animals expressing Inc orthologs under the control of elavc155-Gal4 slept largely indistinguishably from control animals expressing Inc or from those lacking a UAS transgene , as indicated by analysis of sleep duration , daytime and nighttime sleep , sleep bout length , and sleep bout number ( Fig 4C and S4A–S4D Fig ) . Thus , neuronal expression of Inc and its mouse orthologs does not elicit significant dominant negative effects or otherwise inhibit endogenous inc function , similar to ubiquitous or neuronal expression of untagged Inc [11] . To assess whether mouse Inc orthologs can functionally substitute for Drosophila Inc , we measured their ability to rescue the sleep defects of inc1 null mutants [11] . Sleep is a behavior sensitive to genetic background [26 , 27] , environment [28] , and other influences , and thus the ability of Inc orthologs to confer behavioral rescue represents a stringent test of inc function . Inc and its orthologs expressed under inc-Gal4 control in inc1 animals accumulated with relative levels similar to those in elav-Gal4 animals ( Fig 4A and 4B ) . Expression of Myc-Inc under inc-Gal4 control fully rescued sleep in inc mutants to wild-type levels , indicating that Myc-Inc recapitulates the function of endogenous Inc protein ( Fig 4D ) . Strikingly , expression of mouse KCTD2 and KCTD5 rescued most of the sleep deficits of inc mutants , including those in total sleep duration , daytime sleep , sleep bout length , and sleep bout number ( Fig 4D and 4E , and S4E–S4H Fig ) . Nighttime sleep was rescued partially by KCTD2 but not by KCTD5 ( S4E Fig ) . The ability of KCTD2 and KCTD5 to rescue inc phenotypes indicates that these proteins not only recapitulate Inc-Inc and Inc-Cul3 interactions ( Figs 2 and 3 ) , but also retain other critical aspects of Inc function including its presumed ability to engage and ubiquitinate neuronal target proteins in vivo . In contrast , KCTD17 . 2 and KCTD17 . 3 failed to rescue inc phenotypes ( Fig 4D and S4E–S4H Fig ) . The restoration of sleep by KCTD2 and KCTD5 but not by isoforms of KCTD17 contrasts with the apparent biochemical interchangeability of these proteins with respect to Inc-Inc and Inc-Cul3 associations ( Figs 2 and 3 ) . The inability of KCTD17 . 2 and KCTD17 . 3 to rescue inc sleep defects may reflect the lower abundance of these proteins in transgenic flies ( Fig 4A and 4B ) or differences in the intrinsic activities of these KCTD17 isoforms , including their ability to engage Inc targets relevant to sleep . Inc is required in neurons for normal sleep-wake cycles [11 , 12] , indicating that it impacts aspects of neuronal function that are essential for sleep . The subcellular localization of Cul3 adaptors reflects their biological functions and variously includes the cytoplasm [29] , nuclear foci [30] , cytoskeletal structures [31 , 32] , and perisynaptic puncta in neurons [33] . The subcellular distribution of Inc is unknown , as available antisera do not efficiently detect endogenous Inc [11] , and the localization of Inc orthologs in neurons is similarly uncharacterized . Anti-KCTD5 antibody did not efficiently detect endogenously expressed Inc orthologs in immunohistochemical staining . We therefore examined the subcellular localization of Inc family members fused to epitope tags and expressed in cultured cells , primary neurons , and in flies in vivo . In S2 cells , Myc-tagged Inc was localized to the cytoplasm and excluded from the nucleus ( S5A Fig ) , and an Inc-GFP fusion was distributed similarly in both live cells and after fixation ( S5A Fig ) . Inc orthologs , Inc-GFP , and mouse Cul3 were similarly distributed in the cytoplasm and excluded from the nucleus in cultured mammalian cells ( S5B–S5D Fig ) . We next assessed the localization of tagged Inc orthologs in primary neurons cultured from cortex , a region of the brain in which they are expressed in vivo ( Fig 1C and S1 Fig ) . In cortical neurons , KCTD2 , KCTD5 , and KCTD17 were excluded from the nucleus and localized to the cytosol and to dendritic and axonal projections ( Fig 5A ) . Within dendrites , these proteins trafficked to spines as indicated by co-staining against PSD95 , a component of the postsynaptic density ( Fig 5B ) . In axons , Inc orthologs were present at varicosities costaining with synapsin , a vesicle-associated protein that marks presynaptic termini ( Fig 5C ) . Inc localized similarly to its mammalian orthologs including at pre- and postsynaptic structures ( Fig 5A–5C ) , suggesting that intrinsic determinants governing the localization of Inc and its orthologs in neurons may be functionally conserved . While the expression of tagged Inc orthologs may not fully recapitulate the abundance or distribution of endogenous proteins , these data suggest that KCTD2 , KCTD5 , and KCTD17 can localize to various sites within neurons , including the cytosol , projections , and synapses . To further assess whether the localization of Inc orthologs at synapses in transfected neurons reflects the distribution of corresponding endogenous proteins , we fractionated cortex to isolate synaptosomes and subjected them to biochemical analysis . Enrichment of presynaptic and postsynaptic proteins in these preparations was confirmed by blotting for synapsin and PSD95 respectively ( Fig 5D ) . Probing with anti-KCTD5 antibody revealed that Inc orthologs were present in synaptosome fractions ( Fig 5D ) . Similarly , endogenous Cul3 was present in synaptic fractions in its native and higher-molecular weight neddylated form , indicating that active Nedd8-conjugated Cul3 complexes are present at synapses ( Fig 5D ) . Thus , Inc orthologs and Cul3 are present endogenously at mammalian synapses in vivo , suggesting that they form functional ubiquitin ligase complexes at synapses and that their substrates may include synaptic proteins . To determine whether the localization of Inc in vivo is similar to that of its mouse orthologs , we examined flies expressing Myc-Inc and 3×FLAG-Inc , forms of Inc that rescue the sleep defects of inc mutants ( Fig 4 and S6 Fig ) and which are thus likely to recapitulate attributes of endogenous Inc including its subcellular distribution . In the adult brain , expression of 3×FLAG-Inc under the control of inc-Gal4 yielded strong anti-FLAG signal in cell bodies and in projections including those of the mushroom bodies , ellipsoid body , and fan-shaped body ( Fig 6A ) . To assess the subcellular localization of Inc in adult neurons more clearly , we first identified sparse neuronal populations likely to express inc natively , as indicated by their expression of the inc-Gal4 driver that fully rescues inc mutants ( Fig 4D ) . Animals bearing inc-Gal4 and a nuclear localized GFP reporter ( UAS-nls-GFP ) exhibited GFP signal in corazonin positive ( CRZ+ ) neurons in the dorsal brain [34] and in pigment dispersing factor-expressing ( PDF+ ) circadian pacemaker neurons [35] ( Fig 6B and S7A Fig ) . We then utilized Gal4 drivers specific for these neuronal subpopulations to express Myc-Inc and assess its localization . In both populations , Myc-Inc was largely excluded from the nucleus and present in neuronal cell bodies and arborizations ( Fig 6C and S7B , S7B’ and S7D Fig ) . To assess the nature of projections containing Inc , we compared the pattern of Inc localization to that of the pre- and postsynaptic markers synaptotagmin-GFP ( Syt-eGFP ) [36] and DenMark [37] expressed in the same neuronal populations . In corazonin neurons , whose presynaptic and dendritic compartments are well separated , Myc-Inc trafficked both to medial dendritic structures and to lateral puncta located in the same regions as presynaptic termini of these neurons ( Fig 6C and 6D–6D” ) . In PDF+ neurons , Myc-Inc was abundant in cell bodies and was detectable in dorsal , ventral , and contralateral projections ( S7B , S7B’ and S7C–S7C” Fig ) [38] . Thus , the localization of Myc-Inc in vivo suggests that Inc may act within the cytosol of neurons as well as their pre- and postsynaptic compartments . We also assessed Myc-Inc localization in the third instar larval brain and at the larval neuromuscular junction ( NMJ ) , the latter which permits higher resolution analysis of synaptic termini [39] . In animals bearing inc-Gal4 and UAS-Myc-Inc , we observed Inc signal in motor neuron cell bodies and their axonal projections innervating the NMJ ( Fig 6E ) . As in the adult brain ( Fig 6A ) [11] , Inc expression was present in a subset of neurons , as evident in a fraction of Myc- and HRP-positive projections emanating from the ventral ganglion . At the NMJ itself , Inc was enriched at synaptic boutons and was more prominently expressed within type Is boutons ( Fig 6F–6F”’ and S8A Fig ) . The presence of Inc in motor neurons , their axonal projections , and at boutons circumscribed by postsynaptic Discs Large ( Dlg ) signal indicates that Inc signal in these preparations is presynaptic ( Fig 6F ) . Muscle nuclei exhibited weaker Inc signal ( Fig 6F”’ ) , suggesting , along with comparisons of female and male larvae ( S8B Fig ) , that inc may also be expressed postsynaptically at the NMJ . Taken together with findings that Inc orthologs are present at mammalian synapses and have functions conserved with those of Inc ( Figs 4 and 5 ) , these data suggest that Inc family members may have evolutionarily conserved functions at synapses . A prominent hypothesis invokes synaptic homeostasis as a key function of sleep [40] , and findings from vertebrates and flies support the notion that sleep modulates synaptic structure [e . g . 41–45] . Neuronal inc activity is essential for normal sleep [11 , 12] , but whether inc impacts synaptic function is not known . To test whether the distribution of Inc at synaptic termini reflects a synaptic function , we assessed the anatomical and physiological properties of inc mutants at the NMJ . inc1 and inc2 null mutants both exhibited significantly increased bouton number with respect to wild-type animals ( Fig 7A and 7B ) , indicating that Inc is essential for regulation of synaptic growth or plasticity . To assess whether these anatomical defects are associated with altered synaptic transmission , we recorded postsynaptically from muscle in control and transheterozygous inc1/inc2 animals . While the amplitude of spontaneous miniature postsynaptic potentials was not significantly altered in inc mutants , their frequency was reduced ( Fig 7C–7E ) . The amplitude of evoked postsynaptic potentials triggered by presynaptic stimulation was significantly reduced in inc1/ inc2 mutants , and quantal content was similarly decreased ( Fig 7F–7H ) . The attenuation of evoked potentials and increased bouton number in inc mutants suggest that a compensatory increase in synaptic growth may arise in response to defects in synaptic transmission , though this increase does not compensate for the decreased strength of inc synapses . These data indicate that inc is vital for normal synaptic structure and physiology and suggest , together with the ability of Inc and its orthologs to localize to synapses , that Inc family members may direct the ubiquitination of proteins critical for synaptic function . The presence of sleep states in diverse animals has been suggested to reflect a common purpose for sleep and the conservation of underlying regulatory mechanisms [46] . Here we have shown that attributes of the Insomniac protein likely to underlie its impact on sleep in Drosophila—its ability to function as a multimeric Cul3 adaptor and engage neuronal targets that impact sleep—are functionally conserved in its mammalian orthologs . Our comparative analysis of Inc family members in vertebrate and invertebrate neurons furthermore reveals that these proteins can traffic to synapses and that Inc itself is essential for normal synaptic structure and excitability . These findings support the hypothesis that Inc family proteins serve as Cul3 adaptors and direct the ubiquitination of conserved neuronal substrates that impact sleep and synaptic function . The ability of KCTD2 and KCTD5 to substitute for Inc in the context of sleep is both surprising and notable given the complexity of sleep-wake behavior and the likely functions of these proteins as Cul3 adaptors . Adaptors are multivalent proteins that self-associate , bind Cul3 , and recruit substrates , and these interactions are further regulated by additional post-translational mechanisms [47] . Our findings indicate that KCTD2 and KCTD5 readily substitute for Inc within oligomeric Inc-Cul3 complexes , and strongly suggest that these proteins recapitulate other aspects of Inc function in vivo including the ability to engage neuronal targets that impact sleep . The simplest explanation for why KCTD2 and KCTD5 have retained the apparent ability to engage Inc targets despite the evolutionary divergence of Drosophila and mammals is that orthologs of Inc targets are themselves conserved in mammals . This inference draws support from manipulations of Drosophila Roadkill/HIB and its mammalian ortholog SPOP , Cul3 adaptors of the MATH-BTB family that regulate the conserved Hedgehog signaling pathway [48] . While the ability of SPOP to substitute for HIB has not been assessed by rescue at an organismal level , clonal analysis in Drosophila indicates that ectopically expressed mouse SPOP can degrade the endogenous HIB substrate Cubitus Interruptus ( Ci ) , and conversely , that HIB can degrade mammalian Gli proteins that are the conserved orthologs of Ci and substrates of SPOP [48] . By analogy , Inc targets that impact sleep are likely to have orthologs in vertebrates that are recruited by KCTD2 and KCTD5 to Cul3 complexes . While our manipulations do not resolve whether KCTD17 can substitute for Inc in vivo , the ability of KCTD17 to assemble with fly Inc and Cul3 suggests that functional divergence among mouse Inc orthologs may arise outside of the BTB domain , and in particular may reflect properties of their C-termini including the ability to recruit substrates . The finding that Inc can transit to synapses and is required for normal synaptic function is intriguing in light of hypotheses that invoke synaptic homeostasis as a key function of sleep [40] . While ubiquitin-dependent mechanisms contribute to synaptic function and plasticity [49–52] and sleep is known to influence synaptic remodeling in both vertebrates and invertebrates [41–45] , molecular links between ubiquitination , synapses , and sleep remain poorly explored . Other studies in flies have indicated that regulation of RNA metabolism may similarly couple synaptic function and the control of sleep [53 , 54] . Alterations in the activity of the Fragile X mental retardation protein ( FMR ) , a regulator of mRNA translation , cause defects in the elaboration of neuronal projections and the formation of synapses as well as changes in sleep duration and consolidation [53 , 55 , 56] . Loss of Adar , a deaminase that edits RNA , leads to increased sleep through altered glutamatergic synaptic function [54] . Like Inc , these proteins are conserved in mammals , suggesting that further studies in flies may provide insights into diverse mechanisms by which sleep influences synaptic function and conversely , how changes in synapses may impact the regulation of sleep . Our findings at a model synapse suggest that the impact of Inc on synaptic function may be intimately linked to its influence on sleep but do not yet resolve important aspects of such a mechanism . The synaptic phenotypes of inc mutants—increased synaptic growth , decreased evoked neurotransmitter release , and modest effects on spontaneous neurotransmission—are qualitatively distinct from those of other short sleeping mutants . Shaker ( Sh ) and Hyperkinetic ( Hk ) mutations decrease sleep in adults [26 , 57] but increase both excitability and synaptic growth at the NMJ [58–60] , suggesting that synaptic functions of Inc may affect sleep by a mechanism different than broad neuronal hyperexcitability . While a parsimonious model is that Inc directs the ubiquitination of a target critical for synaptic transmission both at the larval NMJ and in neuronal populations that promote sleep , this hypothesis awaits the elucidation of Inc targets , definition of the temporal requirements of Inc activity , and further mapping of the neuronal populations through which Inc impacts sleep [11 , 12 , 61] . Finally , determining the localization of endogenous Inc within neurons is essential to distinguish possible presynaptic and postsynaptic functions of Inc and whether Inc engages local synaptic proteins or extrasynaptic targets that ultimately influence synaptic function . A clear implication of our findings is that neuronal targets and synaptic functions of Inc may be conserved in other animals . While the impact of Inc orthologs on sleep in vertebrates is as yet unknown , findings from C . elegans support the notion that conserved molecular functions of Inc and Cul3 may underlie similar behavioral outputs in diverse organisms . INSO-1/C52B11 . 2 , the only C . elegans ortholog of Inc , interacts with Cul3 [14] , and RNAi against Cul3 and INSO-1 reduces the duration of lethargus , a quiescent sleep-like state , suggesting that effects of Cul3- and Inc-dependent ubiquitination on sleep may be evolutionarily conserved [62] . The functions of Inc orthologs and Cul3 in the mammalian nervous system await additional characterization , but emerging data suggest functions relevant to neuronal physiology and disease . Human mutations at the KCTD2/ATP5H locus are associated with Alzheimer’s disease [63] , and mutations of KCTD17 with myoclonic dystonia [64] . Cul3 lesions have been associated in several studies with autism spectrum disorders [65–67] and comorbid sleep disturbances [67] . More generally , autism spectrum disorders are commonly associated with sleep deficits [68] and are thought to arise in many cases from altered synaptic function [69] , but molecular links to sleep remain fragmentary . Studies of Inc family members and their conserved functions in neurons are likely to broaden our understanding of how ubiquitination pathways may link synaptic function to the regulation of sleep and other behaviors . Total RNA was isolated with TRIZOL ( ThermoFisher ) from a single brain hemisphere of a mixed C57BL/6 background adult mouse . 5 μg total RNA was annealed to random hexamer primers and reverse transcribed with Thermoscript ( ThermoFisher ) according to the manufacturer’s protocol . KCTD2 , KCTD5 , and KCTD17 transcripts were amplified using primer pairs oNS286 and oNS287 , oNS288 and oNS289 , and oNS290 and oNS291 , respectively . For in situ hybridization , DNA templates bearing a terminal SP6 promoter for in vitro transcription were generated by PCR amplification of C57BL/6 mouse genomic DNA , using primer pairs oNS1204 and oNS1205 for KCTD2 , oNS1207 and oNS1208 for KCTD5 , and oNS1213 and oNS1214 for KCTD17 . Riboprobes were transcribed with SP6 polymerase and DIG-11-UTP or Fluorescein-12-UTP ( Roche ) . In situ hybridization was performed as described [70] , amplifying Fluorescein- and DIG-labeled probes with Fluorescein-tyramide and Cy5-tyramide ( Perkin Elmer ) respectively . Vectors for Drosophila transgenesis were as follows: pUASTattB-Myc-Inc ( pNS346 ) encodes a N-terminal 1×Myc epitope ( MEQKLISEEDLAS ) fused to Inc , and was generated by three piece ligation of BglII-XhoI digested pUASTattB [71] , a BglII-NheI 1×Myc fragment generated by phosphorylating and annealing oligonucleotides oNS283 and oNS284 , and an NheI-XhoI inc fragment liberated from the PCR amplification product of pUAST-Inc ( pNS272 ) [11] template and primers oNS277 and oNS285 . pUASTattB-Myc-KCTD2 ( pNS347 ) was generated similarly to pNS346 , substituting a NheI-XhoI KCTD2 fragment liberated from the PCR amplification product of mouse brain cDNA and primers oNS286 and oNS287 . Amplified KCTD2 sequences are identical to those within GenBank accession NM_183285 . 3 . pUASTattB-Myc-KCTD5 ( pNS348 ) was generated similarly to pNS346 , substituting a NheI-XhoI KCTD5 fragment liberated from the PCR amplification product of mouse brain cDNA and primers oNS288 and oNS289 . Amplified KCTD5 sequences are identical to those within GenBank accession NM_027008 . 2 . pUASTattB-Myc-KCTD17 . 3 ( pNS349 ) and pUASTattB-Myc-KCTD17 . 2 ( pNS350 ) were generated similarly to pNS346 , substituting NheI-XhoI KCTD17 fragments liberated from the PCR amplification products of mouse brain cDNA and primers oNS290 and oNS291 . The smaller and larger NheI-XhoI fragments respectively corresponding to KCTD17 . 3 and KCTD17 . 2 were gel purified and ligated separately . Amplified KCTD17 . 3 and KCTD17 . 2 sequences are identical to those within GenBank accession NM_001289673 . 1 and NM_001289672 . 1 respectively . pUAST-Inc-HA ( pNS273 ) encodes Inc fused to a C-terminal 1×HA epitope ( GSYPYDVPDYA ) and was generated by three piece ligation of pUAST BglII-XhoI , a BglII-EcoRI inc fragment liberated from pNS272 [11] , and an EcoRI-XhoI HA fragment generated by phosphorylating and annealing oligonucleotides oNS191 and oNS192 . pUASTattB-3×FLAG-Inc ( pNS404 ) encodes a N-terminal 3×FLAG epitope ( MDYKDDDDKGSDYKDDDDKGSDYKDDDDKAS ) fused to Inc and was generated by three piece ligation of EcoRI-XhoI digested pUASTattB , an EcoRI-NheI 3×FLAG fragment liberated from the PCR amplification product of pNS311 template and primers ACF and oNS241 , and a NheI-XhoI inc fragment liberated from pNS351 . Vectors for expression in S2 cells were as follows: pAc5 . 1–3×Myc-Inc ( pNS351 ) encodes a N-terminal 3×Myc epitope ( MEQKLISEEDLGSEQKLISEEDLGSEQKLISEEDLAS ) fused to Inc in a derivative of pAc5 . 1/V5-HisA ( ThermoFisher ) , and was generated by ligating NheI-XhoI digested pNS309 [11] to a NheI-XhoI inc fragment prepared as for pNS346 . pAc5 . 1–3×Myc-KCTD2 ( pNS352 ) was generated similarly to pNS351 , substituting a NheI-XhoI KCTD2 fragment prepared as for pNS347 . pAc5 . 1–3×Myc-KCTD5 ( pNS353 ) was generated similarly to pNS351 , substituting a NheI-XhoI KCTD5 fragment prepared as for pNS348 . pAc5 . 1–3×Myc-KCTD17 . 3 ( pNS391 ) was generated by ligating NheI-XhoI digested pNS309 to a NheI-XhoI KCTD17 . 3 fragment liberated from the PCR amplification product of pNS354 template and primers oNS290 and oNS612 . pNS354 was generated by ligating NheI-XhoI digested pNS309 and the NheI-XhoI KCTD17 . 3 fragment prepared as for pNS349 . pAc5 . 1–3×Myc-KCTD17 . 2 ( pNS392 ) was generated by ligating NheI-XhoI digested pNS309 to a NheI-XhoI KCTD17 . 2 fragment liberated from the PCR amplification product of pNS355 template and primers oNS290 and oNS955 . pNS355 was generated by ligating NheI-XhoI digested pNS309 and the NheI-XhoI KCTD17 . 2 fragment prepared as for pNS350 . pAc5 . 1–3×HA-Inc ( pNS402 ) was generated by ligating NheI-XhoI digested pNS310 [11] and a NheI-XhoI inc fragment liberated from pNS351 . pAc5 . 1–3×HA-mCul3 ( pNS367 ) encodes a N-terminal 3×HA epitope ( MYPYDVPDYAGSYPYDVPDYAGSYPYDVPDYAAS ) fused to mouse Cul3 , and was generated by three piece ligation of NheI-NotI digested pNS310 [11] , a 0 . 3 kb NheI-HindIII 5’ mCul3 fragment liberated from the PCR amplification product of pCMV-SPORT6-mCul3 template ( ThermoFisher , GenBank accession BC027304 ) and primers oNS313 and oNS314 , and a 2 . 2 kb HindIII-NotI 3’ mCul3 fragment generated by digesting pCMV-SPORT6-mCul3 with AvaI , blunting with T4 DNA polymerase , ligating a NotI linker , and digesting with HindIII-NotI . pAc5 . 1–3×FLAG-Cul3 ( pNS403 ) encodes a N-terminal 3×FLAG epitope fused to Drosophila Cul3 and was generated by ligating NheI-NotI digested pNS311 and a NheI-NotI Cul3 fragment liberated from pNS314 [11] . pNS311 contains a N-terminal 3×FLAG tag and was generated from pNS298 , a derivative of pAc5 . 1/V5-HisA that contains a C-terminal 3×FLAG tag . To construct pNS298 , oligonucleotides oNS234 and oNS235 were phosphorylated , annealed , and cloned into XhoI-XbaI digested pAc5 . 1/V5-HisA . To construct pNS311 , the EcoRI-NotI fragment liberated from the PCR amplification product of pNS298 template and primers oNS240 and oNS241 was ligated to EcoRI-NotI digested pAc5 . 1/V5-HisA . pAc5 . 1-Inc-GFP ( pNS275 ) encodes Inc fused at its C-terminus to GFP and was generated by three piece ligation of EcoRI-NotI digested pAc5 . 1/V5-HisA , an EcoRI-BamHI inc fragment liberated from pNS273 , and a BamHI-NotI EGFP fragment liberated from pEGFP-N3 ( Clontech ) . Mammalian expression vectors were as follows: pcDNA3 . 1–3×Myc-Inc ( pNS358 ) was generated by ligating NheI-XhoI digested pcDNA3 . 1 ( + ) ( ThermoFisher ) with a SpeI-XhoI 3×Myc-Inc fragment liberated from pNS351 . pcDNA3 . 1–3×Myc-KCTD2 ( pNS359 ) was generated similarly to pNS358 , substituting a SpeI-XhoI 3×Myc-KCTD2 fragment liberated from pNS352 . pcDNA3 . 1–3×Myc-KCTD5 ( pNS360 ) was generated by ligating EcoRI-XhoI digested pcDNA3 . 1 ( + ) with a EcoRI-XhoI 3×Myc-KCTD5 fragment liberated from pNS353 . pcDNA3 . 1–3×Myc-KCTD17 . 3 ( pNS393 ) was generated by three piece ligation of EcoRI-XhoI digested pcDNA3 . 1 ( + ) , an EcoRI-NheI 3×Myc fragment liberated from the PCR amplification product of pNS351 template and primers ACF and oNS318 , and a NheI-XhoI KCTD17 . 3 fragment liberated from the PCR amplification product of pNS354 template and primers oNS290 and oNS612 . pcDNA3 . 1–3×Myc-KCTD17 . 2 ( pNS394 ) was generated similarly to pNS393 , substituting a NheI-XhoI KCTD17 . 2 fragment liberated from the PCR amplification product of pNS355 template and primers oNS290 and oNS955 . pcDNA3 . 1–3×HA-Cul3 ( pNS365 ) was generated by ligating EcoRI-NotI digested pcDNA3 . 1 ( + ) with the EcoRI-NotI 3×HA-Cul3 fragment from pNS314 [11] . pcDNA3 . 1–3×HA-mCul3 ( pNS369 ) was generated by ligating KpnI-NotI digested pcDNA3 . 1 ( + ) with the KpnI-NotI 3×HA-mCul3 fragment liberated from pNS367 . pcDNA3 . 1–3×FLAG-Inc ( pNS395 ) encodes a N-terminal 3×FLAG epitope fused to Inc , and was generated by three piece ligation of EcoRI-XhoI digested pcDNA3 . 1 ( + ) , an EcoRI-NheI 3×FLAG fragment liberated from the PCR amplification product of pNS386 template and primers ACF and oNS318 , and the NheI-XhoI inc fragment liberated from pNS351 . pNS386 was generated by ligating NheI-XhoI digested pNS311 and an NheI-XhoI inc fragment liberated from pNS351 . pcDNA3 . 1–3×FLAG-KCTD2 ( pNS396 ) was generated similarly to pNS395 , substituting an NheI-XhoI KCTD2 fragment liberated from pNS352 . pCDNA3 . 1–3×FLAG-KCTD5 ( pNS397 ) was generated similarly to pNS395 , substituting an NheI-XhoI KCTD5 fragment liberated from pNS353 . pcDNA3 . 1–3×FLAG-KCTD17 . 2 ( pNS398 ) was generated similarly to pNS395 , substituting the NheI-XhoI fragment liberated from the PCR amplification product of pNS355 template and primers oNS290 and oNS955 . pCDNA3 . 1–3×FLAG-KCTD17 . 3 ( pNS399 ) was generated similarly to pNS395 , substituting the NheI-XhoI fragment liberated from the PCR amplification product of pNS354 template and primers oNS290 and oNS612 . pEGFPN3-Inc ( pNS279 ) encodes Inc fused at its C-terminus to GFP and was generated by ligating EcoRI-BamHI digested pEGFP-N3 ( Clontech ) and an EcoRI-BamHI inc fragment liberated from pNS273 . Oligonucleotides used in this work are as follows: oNS119 5’-GTCCGCGCGATTCCCTTGCTTGC-3’ , oNS191 5’-AATTTTGGGAATTGGATCCTACCCCTACGATGTGCCCGATTACGCCTAAC-3’ , oNS192 5’-TCGAGTTAGGCGTAATCGGGCACATCGTAGGGGTAGGATCCAATTCCCAA-3’ , oNS198 5’-ACTGGGATCCATCCGCCTGTGTGGCTGGGACGG-3’ , oNS234 5’-TCGAGGCTAGCGACTACAAGGATGATGACGATAAGGGCTCCGATTACAAGGACGACGATGATAAGGGATCCGATTACAAGGATGATGACGACAAGTGAT-3’ , oNS235 5’-CTAGATCACTTGTCGTCATCATCCTTGTAATCGGATCCCTTATCATCGTCGTCCTTGTAATCGGAGCCCTTATCGTCATCATCCTTGTAGTCGCTAGCC-3’ , oNS240 5’-ACTGGAATTCCGCGGCAACATGGACTACAAGGATGATGACGATAAGGGC-3’ , oNS241 5’-ACTGGCGGCCGCTCCTAGGGTGCTAGCCTTGTCGTCATCATCCTTGTAATCGGAT-3’ , oNS254 5’-GTGCGCCAAGTGTCTGAAGAACAACTGG-3’ , oNS255 5’-GATGAGCTGCCGAGTCAATCGATACAGTC-3’ , oNS277 5’-ACGTGCTAGCATGAGCACGGTGTTCATAAACTCGC-3’ , oNS283 5’-GATCTCAACATGGAGCAGAAGCTGATCAGCGAGGAGGATCTGG-3’ , oNS284 5’-CTAGCCAGATCCTCCTCGCTGATCAGCTTCTGCTCCATGTTGA-3’ , oNS285 5’-ACGTGCTAGCTCGAGGGGTTGTGTGTGAATATATAGCGCGA-3’ , oNS286 5’-ACGTGCTAGCATGGCGGAGCTGCAGCTGG-3’ , oNS287 5’-ACGTGCTAGCTCGAGCCGCTTACATTCGAGAGCCTCTCTCC-3’ , oNS288 5’-ACGTGCTAGCATGGCGGAGAATCACTGCGAGCTG-3’ , oNS289 5’-ACGTGCTAGCTCGAGCCTCACATCCTTGAGCCCCGTTC-3’ , oNS290 5’-ACGTGCTAGCATGCAGACAACGCGGCCGGCG-3’ , oNS291 5’-ACGTGCTAGCTCGAGCCCAAGGCAGGAGTGAGTCTCAGC-3’ , oNS313 5’-ACGTGCTAGCATGTCGAATCTGAGCAAAGGCACGGG-3’ , oNS314 5’-GCCGAAGATGATCCCTAATACACCCATACCG-3’ , oNS318 5’-ACGTCTCGAGTTACTGCGTCACGTTGTAGAACTC-3’ , oNS612 5’-ACGTCTCGAGTCACATCCGGGTGCCTCTGGCTT-3’ , oNS955 5’-ACGTCTCGAGTCACTGCAAGCTCAGGCTTGGGTCTG-3’ , oNS1204 5’-TAATACGACTCACTATAGGGGAAGGCAAGAGAGCAATCGGC-3’ , oNS1205 5’-GCGATTTAGGTGACACTATAGAAGAAAAGGCTGCAGAAGCAGTTAC-3’ , oNS1207 5’-TAATACGACTCACTATAGGGGGCTCAAGGATGTGAGGAATGCTGAG-3’ , oNS1208 5’-GCGATTTAGGTGACACTATAGAAGCAGCCTCTATCCCAGGCACAAC-3’ , oNS1213 5’-TAATACGACTCACTATAGGGTTACAAGCCAGAGGCACCCGGA-3’ , oNS1214 5’-GCGATTTAGGTGACACTATAGAAGCAGCTCAACCCGTTACACCTGTC-3’ , ACF 5’-GACACAAAGCCGCTCCATCAG-3’ , attP2-5’ 5’-CACTGGCACTAGAACAAAAGCTTTGGCG-3’ . 293T cells were cultured in DMEM containing 10% FBS , penicillin , and streptomycin , and transfected with Lipofectamine 2000 ( ThermoFisher ) according to the manufacturers protocol . S2 cells were cultured in S2 media containing 10% FBS , penicillin , and streptomycin , and were transfected with Effectene ( Qiagen ) as described previously [11] . For both 293T and S2 cells , transfections were performed in 6 well or 12 well plates for ~24h until liposome-containing media was replaced with fresh culture media . Cells or coverslips were harvested for lysis or immunohistochemistry 36-48h after transfections were initiated . For transfections involving more than one plasmid , an equal amount of each was used . Rat cortical neurons were cultured on poly-D-lysine coated coverslips and transfected with calcium phosphate at 7 days in vitro ( DIV ) as described previously [72] . For co-immunoprecipitations from 293T cells , samples were lysed with ice-cold RIPA buffer ( 50 mM Tris-Cl pH 7 . 6 , 150 mM NaCl , 50 mM NaF , 2 mM EDTA , 0 . 5% sodium deoxycholate , 1% NP40 , 0 . 1% SDS ) containing protease inhibitor ( Sigma , P8340 ) . For co-immunoprecipitation of Inc family members from S2 cells , samples were lysed with ice-cold NP40 buffer ( 50 mM Tris pH 7 . 6 , 150mM NaCl , 0 . 5% NP40 ) containing protease inhibitors or RIPA buffer as above; for S2 cell co-immunoprecipitations involving Cul3 , ice-cold NP40 buffer was used . Protein extracts were quantitated in duplicate ( BioRad , 5000111 ) and 160–400 μg ( 293T ) or 800–1000 μg ( S2 ) was immunoprecipitated with 20 μl ( 50% slurry ) of anti-FLAG ( Sigma , F2426 ) or anti-HA ( Sigma , E6779 ) affinity gel for 1 hr nutating at 4°C . Samples were then washed 4×5 min at 4°C with lysis buffer , denatured in SDS sample buffer , separated on Tris SDS-PAGE gels , and transferred to nitrocellulose . Membranes were blocked for 1 hr at room temperature or 4°C overnight in LI-COR Odyssey buffer ( LI-COR , 927–40000 ) or 1% casein in PBS . Membranes were subsequently incubated in blocking buffer containing 0 . 1% Tween 20 and the appropriate primary antibodies: rabbit anti-Myc ( 1:2 , 000 , Sigma , C3956 ) , mouse anti-FLAG ( 1:2 , 000 , Sigma , F1804 ) , rat anti-HA ( 1:1 , 000–1:2 , 000 , Roche , 11867431001 ) , rabbit anti-Cul3 ( 1:1 , 000 , Bethyl Laboratories , A301-109A ) , and rabbit anti-KCTD5 ( 1:2 , 000 , Proteintech , 15553-1-AP ) . After washing 4×5 min in a solution containing 150 mM NaCl , 10mM Tris pH 7 . 6 , and 0 . 1% Tween 20 ( TBST ) , membranes were incubated in the dark for 30 min at room temperature with appropriate secondary antibodies , all diluted 1:30 , 000 in blocking buffer containing 0 . 1% Tween 20 and 0 . 01% SDS: Alexa 680 donkey anti-rabbit ( Life Technologies , A10043 ) , Alexa 790 anti-mouse ( Life Technologies , A11371 ) , and Alexa 790 anti-rat ( Jackson ImmunoResearch , 712-655-153 ) . Membranes were then washed 4×5 min in TBST , 1×5 min in TBS , and imaged on a Li-Cor Odyssey CLx instrument . Fly protein extracts were prepared from whole animals or from sieved heads by manual pestle homogenization in ice-cold NP40 lysis buffer supplemented with protease inhibitors . 50 μg was separated on Tris-SDS-PAGE gels and blotted as described above . Primary antibodies were mouse anti-Myc ( 1:1 , 000 , BioXCell , BE0238; or 1:1 , 000 , Cell Signaling Technology , 2276 ) , rabbit anti-tubulin ( 1:30 , 000 , VWR , 89364–004 ) , or mouse anti-actin ( 1:1 , 000 , Developmental Studies Hybridoma Bank ( DSHB ) , JLA20 ) . Secondary antibodies were Alexa 680 donkey anti-rabbit , Alexa 680 anti-mouse ( Life Technologies , A10038 ) , and Alexa 790 anti-rabbit ( Life Technologies , A11374 ) . Mouse and rat brain extracts not used for synaptosome preparations were prepared by homogenizing brains in ice-cold NP40 lysis buffer supplemented with protease inhibitors . Extracts were separated with SDS-PAGE and blotted as described above . Synaptosomes were prepared from rat brain essentially as described [73] , and probed with rabbit anti-Cul3 ( 1:1 , 000 ) , mouse anti-Actin ( 1:1 , 000 ) , rabbit anti-KCTD5 ( 1:2 , 000 ) , guinea-pig anti-synapsin ( 1:1 , 000 , Synaptic Systems , 106 004 ) , and mouse anti-PSD95 ( 1:1 , 000 , Neuromab , 75–028 ) . Alexa 680 and Alexa 790 secondary antibodies were used as described above . Rat cortical neurons transfected at 7 DIV were processed at 13 DIV for immunohistochemistry . Samples were fixed for 15 min with ice-cold 4% paraformaldehyde in Dulbecco’s PBS containing 4 mM EGTA and 4% sucrose , and subsequently permeabilized for 10 min in PBS containing 0 . 5% normal donkey serum ( Lampire Biological , 7332100 ) and 0 . 1% Triton X-100 . Samples were then blocked at room temperature for 30 min in PBS containing 7 . 5% normal donkey serum and 0 . 05% Triton X-100 , incubated overnight at 4°C in primary antibody cocktail prepared in PBS containing 5% normal donkey serum and 0 . 05% Triton X-100 , and washed 3×5 min in PBS at room temperature . Secondary antibody cocktails were prepared similarly and incubated with samples at room temperature for 30–40 min in the dark . Samples were then washed 3×5 min in PBS at room temperature and mounted on microscope slides in Vectashield ( Vector Labs , H-1000 ) . Primary antibodies were rabbit anti-Myc ( 1:200 ) , mouse anti-PSD95 ( 1:1 , 000 ) , and guinea pig anti-synapsin ( 1:1 , 000 ) . Secondary antibodies , all diluted at 1:1 , 000 , were Alexa 488 donkey anti-rabbit ( Life Technologies , A21206 ) , Alexa 488 donkey anti-guinea pig ( Jackson ImmunoResearch , 706-545-148 ) , Alexa 568 donkey anti-mouse ( Life Technologies , A10037 ) , and Alexa 568 donkey anti-rabbit ( Life Technologies , A10042 ) . Immunohistochemistry for 293T cells was performed similarly; cells were plated on poly-L-lysine treated coverslips and cultured and transfected as described above . 0 . 4 μg/ml DAPI was included in the penultimate wash prior to mounting coverslips on microscope slides . For immunohistochemistry of adult fly brains , whole animals were fixed with 4% paraformaldehyde in PBS containing 0 . 2% Triton X-100 ( PBST ) for 3 hr at 4°C , and subsequently washed 3×15 min at room temperature with PBST . Brains were dissected in ice-cold PBST , incubated for 30 min at room temperature in blocking solution containing PBST and 5% normal donkey serum , and incubated overnight at 4°C in primary antibody cocktail diluted in blocking solution . After 3×15 min washes in PBST at room temperature , samples were incubated in secondary antibody cocktail in blocking solution for 1–3 days at 4°C , washed 3×15 min at room temperature with PBST , and mounted on microscope slides in Vectashield . Primary antibodies were mouse anti-FLAG ( 1:100 ) , rabbit anti-GFP ( 1:3 , 000; Life Technologies , A11122 ) , mouse anti-GFP ( 1:1 , 000 , DSHB , G1 ) , mouse anti-PDF ( 1:1 , 000 , DSHB , C7 ) , rabbit anti-CRZ ( 1:1 , 000 , [74] ) , rabbit anti-DsRed ( 1:1 , 000 , Clontech , 632496 ) , mouse anti-Myc ( 1:100 , BioXCell , BE0238 ) , and rabbit anti-Myc ( 1:500 , Sigma , C3956 ) . Secondary antibodies , all used at 1:1 , 000 , were Alexa 488 donkey anti-mouse ( Life Technologies , A21202 ) , Alexa 488 donkey anti-rabbit , Alexa 568 donkey anti-mouse , and Alexa 568 donkey anti-rabbit . For immunohistochemistry of larval brains and neuromuscular junctions , wandering third instar larvae were dissected in PBS and pinned to 35mm Sylgard-coated petri dishes . Where experiments required larvae of a specific sex , gonads were identified prior to dissection by visual inspection of animals under PBS immersion as described [75] . Larval filets were fixed for 30 min at room temperature in 4% paraformaldehyde in PBS , and subsequently rinsed twice and washed 3×20 min in PBST . Samples were blocked in 5% normal donkey serum in PBST at room temperature for 30 min and incubated overnight at 4°C in primary antibody cocktail diluted in blocking solution . Samples were then washed 3×20 min in PBST at room temperature and incubated overnight at 4°C with secondary antibody cocktail in blocking solution , washed , and mounted in Vectashield . Primary antibodies were rabbit anti-myc ( 1:500 ) , mouse anti-Dlg ( 1:1 , 000 , DSHB , 4F3 ) , and Alexa 647 goat anti-HRP ( 1:200 , Jackson ImmunoResearch , 123-605-021 ) . Secondary antibodies , used at 1:1000 , were Alexa 488 donkey anti-rabbit and Alexa 568 donkey anti-mouse . Neuromuscular junctions were imaged with a confocal microscope and z-stacks were captured using 40× or 63× oil objectives at 512×512 resolution . Boutons were counted offline using a manual tally counter while manipulating z-stacks in 3-dimensional space using Zen software ( Zeiss ) ; each axon branch was counted separately to avoid undercounting or duplicate counts and counts were performed three times to ensure consistency . All bouton counting was performed in a double-blind manner with codes revealed after the entire experiment was scored . For live imaging of Inc-GFP in S2 cells , cells were cultured on poly-L-lysine coated coverslips and transfected with Effectene as described above . 48 hours post-transfection , coverslips were inverted onto a drop of PBS on microscope slides and imaged immediately ( within 5 minutes ) on a confocal microscope . For imaging of fixed Inc-GFP signal in S2 cells , coverslips were washed twice with PBS , fixed for 25 min in PBS containing 4% paraformaldehyde , washed twice in PBS , and inverted onto drops of Vectashield containing DAPI ( Vector ) on microscope slides . All imaging was performed on Zeiss LSM510 or LSM800 confocal microscopes . elavc155-Gal4 [76] and inc1 , inc2 , and inc-Gal4 [11] have been described previously . Unless noted otherwise , all experiments were performed with the X-linked inc-Gal4 transgene . The inc1 inc-Gal4 stock was generated by meiotic recombination between isogenic inc1 and inc-Gal4 chromosomes and verified with duplex PCR using primers oNS254 and oNS255 for Gal4 and oNS119 and oNS198 for inc1 . pUASTattB-based vectors generated in this study were integrated at attP2 [25] with phiC31 recombinase ( BestGene ) ; integration was verified by PCR using primer attP2-5’ paired with oNS277 , oNS286 , oNS288 , or oNS290 for Inc , KCTD2 , KCTD5 , and KCTD17 respectively . All transgenes were backcrossed eight generations to Bloomington stock 5905 , an isogenic w1118 stock described elsewhere as iso31 [77] . Crosses were set with five virgin females and three males on cornmeal , agar , and molasses food . One to four day old male flies eclosing from LD-entrained cultures raised at 25°C were loaded in glass tubes containing cornmeal , agar , and molasses food . Animals were monitored for 5–7 days at 25°C in LD cycles using DAM2 monitors ( Trikinetics ) . The first 36–48 hours of data was discarded to permit animals to acclimate to glass tubes , and an integral number of days of data ( 3–5 ) were analyzed using custom Matlab software as described previously [11] . Locomotor data were collected in 1 min bins , and sleep was defined by inactivity for 5 minutes or more; a given minute was assigned as sleep if an animal was inactive for that minute and the preceding four minutes . Dead animals were excluded from analysis by a combination of automated filtering and visual inspection of locomotor traces . Electrophysiological recordings were performed from abdominal segment 3 muscle 6/7 of third instar larvae as previously described [78] . One-way ANOVA and Tukey post-hoc tests were used for comparisons of total sleep , daytime sleep , nighttime sleep , sleep bout number , and bouton number . Nonparametric Kruskal-Wallis tests and Dunn’s post hoc tests were used for comparisons of sleep bout length . Unpaired two-sided Student’s t-tests were used for comparisons of all electrophysiological parameters . Alignments were performed with Clustal Omega 1 . 2 . 1 and BOXSHADE . GenBank accession numbers for transcript variants referred to in S1 and S2 Figs are: KCTD2 , NM_183285; KCTD5 , NM_027008; KCTD17 . 1 , NM_001289671; KCTD17 . 2 , NM_001289672; KCTD17 . 3 , NM_001289673; KCTD17 . 4 , NR_110357; KCTD17 . v1 , XM_006521460; KCTD17 . v2 , XM_011245739; KCTD17 . v3 , XM_011245740; KCTD17 . v4 , XM_011245741; KCTD17 . v5 , XM_006521461; KCTD17 . v6 , XM_006521462; KCTD17 . v7 , XM_011245742; KCTD17 . v8 , XM_006521465; hKCTD17 . 2 , NM_024681 .	Sleep is ubiquitous among animals and is regulated in a similar manner across phylogeny , but whether conserved molecular mechanisms govern sleep is poorly defined . The Insomniac protein is vital for sleep in Drosophila and is a putative adaptor for the Cul3 ubiquitin ligase . We show that two mammalian orthologs of Insomniac can restore sleep to flies lacking Insomniac , indicating that the molecular functions of these proteins are conserved through evolution . Our comparative analysis reveals that Insomniac and its mammalian orthologs can localize to neuronal synapses and that Insomniac impacts synaptic structure and physiology . Our findings suggest that Insomniac and its mammalian orthologs are components of an evolutionarily conserved ubiquitination pathway that links synaptic function and the regulation of sleep .	[ "Abstract", "Introduction", "Results", "Discussion", "Methods" ]	[ "invertebrates", "medicine", "and", "health", "sciences", "asses", "nervous", "system", "sleep", "vertebrates", "electrophysiology", "neuroscience", "animals", "mammals", "sleep", "disorders", "animal", "models", "physiological", "processes", "drosophila", "melanogaster", ...	2017	Conserved properties of Drosophila Insomniac link sleep regulation and synaptic function
P-glycoprotein ( P-gp ) is an ATP-dependent transport protein that is selectively expressed at entry points of xenobiotics where , acting as an efflux pump , it prevents their entering sensitive organs . The protein also plays a key role in the absorption and blood-brain barrier penetration of many drugs , while its overexpression in cancer cells has been linked to multidrug resistance in tumors . The recent publication of the mouse P-gp crystal structure revealed a large and hydrophobic binding cavity with no clearly defined sub-sites that supports an “induced-fit” ligand binding model . We employed flexible receptor docking to develop a new prediction algorithm for P-gp binding specificity . We tested the ability of this method to differentiate between binders and nonbinders of P-gp using consistently measured experimental data from P-gp efflux and calcein-inhibition assays . We also subjected the model to a blind test on a series of peptidic cysteine protease inhibitors , confirming the ability to predict compounds more likely to be P-gp substrates . Finally , we used the method to predict cellular metabolites that may be P-gp substrates . Overall , our results suggest that many P-gp substrates bind deeper in the cavity than the cyclic peptide in the crystal structure and that specificity in P-gp is better understood in terms of physicochemical properties of the ligands ( and the binding site ) , rather than being defined by specific sub-sites . P-glycoprotein ( P-gp ) is an ATP-dependent transport protein that is selectively expressed at entry points of xenobiotics in tissues such as the intestinal epithelium , capillary brain endothelium , and kidney proximal tubules among others [1] . Acting as an efflux pump , it prevents exogenous substances from entering sensitive organs and , as such , plays a key role in the absorption and blood-brain barrier penetration of many drugs , affecting their distribution and elimination [2] , [3] . Moreover , overexpression of this protein , also known as MDR1 , has been linked to multidrug resistance ( MDR ) in cancer tumor cells where higher levels of the protein result in increased efflux of chemotherapeutic compounds [4] . Finally , there is also accumulating evidence that P-gp , in addition to its role in drug transport , may transport endogenous molecules such as signaling lipids , and play a role in tumor biology and cancer progression [5] . A major hurdle in the drug discovery process [6] , [7] , P-gp has inspired the development of several assays aimed at identifying its substrates [8] , [9] . One widely used assay , the monolayer efflux ratio ( ER ) assay , measures transport rates of molecules in different directions across a single layer of specialized cells . The ratio or difference of the two rates , basal-to-apical and apical-to-basal , is used to identify P-gp substrates . Another commonly used assay , aimed at identifying P-gp inhibitors as well as substrates , is the calcein-AM ( CAM ) inhibition assay , in which accumulation of the fluorescent calcein molecule inside the cells indicates an interaction between P-gp and the molecule being tested . Despite being widely used , both assays have limitations [10] . For example , the monolayer efflux assay may fail to identify P-gp substrates with high passive permeability ( >300 nm/s ) because efflux by P-gp can be masked by the high diffusion rate of the compounds through the membrane . There is also no standard value of the efflux ratio used to distinguish substrates from nonsubstrates , with cutoff values from 1 . 5 to 3 being used [11] , [12] , [13] , [14] . Because the CAM assay is based on the competitive inhibition of calcein transport by compounds that interact with P-gp , the assay may not detect P-gp substrates with low passive membrane diffusion rates that reach the P-gp binding site at a much slower rate than the fluorescent compound . Both assays are also expensive and time consuming , and results in different labs can vary significantly . For example , midazolam has been identified as a nonsubstrate [13] , an inhibitor [8] , [11] , a substrate [15] and an inducer [16] in different studies . Doxorubicin , resistance to which has been linked to P-gp overexpression , is another example [17] . The drug has been cited repeatedly as a classical P-gp substrate [18] , [19]; however , it has also been classified as a nonsubstrate by several in vitro studies [8] , [13] . To complement experimental assays , several in silico methods have been developed to predict P-gp binding . Among them , pharmacophore models based on anywhere from two [20] to nine [10] features to an ensemble of 100 pharmacophores [21] have been generated . Other approaches have included quantitative structure-activity relationship ( QSAR ) models and machine-learning algorithms , some of them incorporating up to 70 descriptors [22] , [23] , [24] . Even though several of the methods report sensitivity of 80% or higher , the extraordinary chemical diversity of the P-gp substrates , reflected by the large numbers of pharmacophores and descriptors , have frustrated efforts to make sense of the chemical data . A multitude of theories about the number , sizes , and locations of the binding sites has further complicated the issue [25] , [26] , [27] , [28] . The recent publication of the mouse P-gp crystal structure [29] ( 87% identical amino acid sequence to human P-gp ) presents an opportunity to develop new prediction methods that take advantage of the receptor structural information to not only identify molecules that bind to P-gp but also to guide chemical optimization , e . g . , to attempt to modify interactions with the protein . Among the three published structures , two were crystallized with stereoisomers of a cyclic inhibitor , QZ59 , that defines the drug-binding cavity . All structures are in an inward-facing conformation that is believed to be relevant for initial substrate recognition . Located in the transmembrane region , the large and hydrophobic drug-binding pocket is lined with various aromatic side chains and has no clearly defined sub-sites . Taking into account the well-known ability of the protein to accommodate ligands of various shapes and sizes , such an arrangement strongly supports the “induced fit” ligand binding model proposed by Loo et al . [30] . This , combined with the relatively low resolution of the structure , suggests that it is essential to treat the binding site as flexible while modeling binding site interactions , which we demonstrate explicitly in our results here . In this study , we employ a flexible receptor docking method [31] , together with scoring methods that include the Glide XP scoring function [32] and a molecular mechanics scoring function with generalized Born implicit solvent ( MM-GB/SA ) [33] , [34] , to develop a new prediction algorithm for P-gp binding specificity . We benchmark the method in several ways , including a blind test on a series of peptidic cysteine protease inhibitors , confirming the ability to predict compounds more likely to be P-gp substrates . We also apply this approach to evaluate the ability of P-gp to discriminate endogenous vs . exogenous compounds , and to predict that several endogenous metabolites may be P-gp substrates . Overall , our results suggest that specificity in P-gp is better understood in terms of physicochemical properties of the ligands ( and the binding site ) , rather than being defined by specific sub-sites . We also suggest that many P-gp substrates bind deeper in the cavity than the cyclic peptide in the crystal structure . All molecular docking calculations were performed using the mouse P-glycoprotein crystal structure ( Protein Data bank [PDB] code 3G60 ) . BLAST [35] sequence alignment with human P-gp revealed 87% overall sequence identity and ∼100% identity within the binding cavity with the exception of mSer725/hAla729 directly facing the binding cavity . The docking calculations were performed using Glide ( version 5 . 6 ) [36] with the OPLS2005 force field [37] , [38] . The receptor structure was prepared and minimized within the Protein Preparation Wizard . For rigid docking , a rigid receptor grid defined by a 10×10×10 Å inner box was generated . The docking site was either defined by the centroid of the co-crystallized ligand , QZ59-RRR , with a center at ( 19 . 1 , 52 . 3 , −0 . 3 ) Å or defined higher than the original ligand with the center at ( 19 . 0 , 46 . 0 , −6 . 0 ) Å ( same as in induced fit docking , see below ) . All ligands were docked in both standard precision ( SP ) Glide and extra precision ( XP ) Glide modes ( Figure S1 ) . Flexible receptor docking was performed using a multi-stage induced fit docking protocol ( IFD ) [31] as implemented in Schrödinger Suite 2010 . Briefly , in the first stage , the van der Waals radii of protein and ligand are scaled by a factor of 0 . 5 and ligands are docked into the receptor using the default Glide SP mode . Next , Prime is used to predict and optimize selected protein side chains ( details below ) . Finally , the poses are scored and filtered , after which ligands are redocked using Glide XP mode and scored . Final scoring in this work was implemented using the extra precision ( XP ) Glide [32] scoring function and an MM-GB/SA [33] , [34]rescoring function . The specific protocol was developed and refined using well-known P-gp substrates and inhibitors from Table S1 . Since no binding modes are known for any compounds ( with the exception of QZ59 ) , optimal parameters were selected based on binding scores and the ability to distinguish binders from non-binders ( as described below ) . The parameters we varied included the inner box coordinates , van der Waals radii scaling , the number of poses saved , and the number and identity of ‘trimmed’ ( mutated temporarily to Ala ) residues in the initial docking stage . Specifically , we chose to delete the side chains of Phe71 , Phe332 , and Phe728 in the 1st docking stage . These three residues are located in the center of the cavity , and were most responsible for preventing potent inhibitors from achieving good scores , by binding deeper in the cavity . In the primary IFD round , a 10×10×10 Å inner box with coordinates ( 19 . 0 , 47 . 0 , −6 . 0 ) Å was used , which is centered deeper in the binding cavity than the cyclic peptide in the crystal structure , and roughly centered on the docked poses of the initial test set . At this stage , all residues lining the cavity were optimized by Prime [39] ( Table S2 ) , whereas in the following IFD round , only residues within 5 Å of each ligand were minimized . The number of poses saved during the initial docking was set to 100 . For all subsequent docking calculations inner box coordinates ( 19 . 0 , 46 . 0 , −6 . 0 ) Å were used . For each ligand , up to 20 top poses were saved and scored with the Glide XP function and MM-GB/SA . Ligand coordinates were obtained from the DrugBank [40] and PubChem Compound ( http://pubchem . ncbi . nlm . nih . gov/ ) databases and processed using the Ligprep 2 . 4 module . The parameters were assigned based on the OPLS2005 force field . For the QZ59-RRR ligand , selenium atoms were replaced with sulfur atoms . For molecules with stereocenters , only the known active forms were docked . For drugs used as racemic mixtures , both stereoisomers were investigated . The isomer with the more favorable docking score was used in the data analysis . Ionization states were assigned by Epik , and groups with pKa between 5 and 9 were treated as neutral while those outside the range were treated as charged . Initial testing of our approach was conducted with two datasets . One of them was comprised of 24 well-known P-gp binders from Table 1 of the review article by Hennessy et al . [41] and 102 endogenous molecules selected from the KEGG database [42] to represent different classes of biological compounds . The rationale for this first test was that most endogenous molecules would not be effluxed by P-gp , providing insight into how P-gp discriminates between endogenous and exogenous molecules . The second dataset was based on the Doan et al . study of FDA approved drugs [11] that generated consistent experimental data from the monolayer efflux ratio as well as the calcein-inhibition assays . We used the intersection of the results in the two assays to define sets of compounds that were clear P-gp substrates ( i . e . , positive in both assays ) or non-substrates . We did not consider the compounds that were positive in only one of the two assays . The complete list of compounds and their scores are provided in Tables S1 and Tables S3 , S4 , and S5 . Reagents and solvents were purchased from Aldrich Chemical , Alfa Aesar , Chem Impex international or TCI America and used as received . Reactions were carried out under an argon atmosphere in oven-dried glassware using anhydrous solvents from commercial suppliers . Air and/or moisture sensitive reagents were transferred via syringe or cannula and were introduced into reaction vessels through rubber septa . Solvent removal was accomplished with a rotary evaporator at ∼10–50 Torr . Automated column chromatography was carried out using a Biotage SP1 system and silica gel cartridges from Biotage or Silicycle . Analytical TLC plates from EM Science ( Silica Gel 60 F254 ) were employed for TLC analyses . 1H NMR spectra were recorded on a Varian INOVA-400 400 MHz spectrometer . Analogs 1 [43] , [44] , 3 , 6 , and 9 were synthesized in one step from commercially available N-benzyloxycarbonyl ( Cbz ) protected amino acids according to the following general procedure . A solution of the N-benzyloxycarbonyl protected L-amino acid ( 0 . 33 mmol ) in 2 mL of DMF was treated with aminoacetonitrile bisulfate ( 0 . 37 mmol , 1 . 1 equiv . ) , 1-hydroxybenzotriazole ( 0 . 33 mmol , 1 . 0 equiv ) , N- ( 3-dimethylaminopropyl ) -N′-ethylcarbodiimide hydrochloride ( 0 . 67 mmol , 2 . 0 equiv . ) , and N , N-diisopropylethylamine ( 2 . 0 mmol , 6 . 0 equiv . ) . The reaction was stirred at room temperature and monitored until judged complete by TLC or HPLC . The reaction mixture was then poured into ethyl acetate and the resulting organic solution washed in succession with aqueous 1 N HCl ( for non-basic analogs only ) , 50% aqueous NaHCO3 , saturated aqueous NaCl , and then dried ( MgSO4 ) , filtered , and concentrated . The crude product thus obtained was purified using automated silica gel flash chromatography ( Biotage SP1 , ethyl acetate-hexane ) to afford the desired products . Analogs 2 , 4 , 5 , 7 , and 10 were synthesized in two steps from N- ( benzyloxycarbonyl ) -L-serine lactone [45] according to the following procedure . A solution of N- ( benzyloxycarbonyl ) -L-serine lactone ( 0 . 45 mmol ) in 2 mL of acetonitrile was added dropwise to a solution of the relevant amine or N-trimethylsilylamine ( 1–10 equivalents depending on the amine , see below ) in ∼3 mL of acetonitrile . The reaction was monitored at room temperature or in some cases heated at 50°C , depending on the reactivity of the amine ( see below ) . When the reaction was judged complete by TLC or HPLC , the reaction mixture was concentrated and the desired amino acid separated from undesired amide side product in one of the following ways . For the amino acid leading to 2 , the crude product was partitioned between ethyl acetate and water and the water phase ( containing the desired product ) was then lyophilized . For the amino acids leading to 4 and 5 , purification by automated silica gel flash chromatography ( Biotage SP1 , methanol-dichloromethane ) afforded the desired amino acids . For intermediate amino acids leading to 7 and 10 , the crude residue was partitioned between dichloromethane and 1 N aqueous NaOH , followed after phase separation by acidification of the aqueous phase with 1 N HCl to effect precipitation of the amino acid , which was collected on a filter , washed with cold water , and dried . The procedures described above provided the desired amino acids in sufficient purity for use in the subsequent coupling reaction with aminoacetonitrile , which was carried out according to the general coupling protocol described for analogs 1 , 3 , 6 , and 9 above . Analog 8 was prepared in three steps by reaction of indoline with N- ( benzyloxycarbonyl ) -L-serine lactone as described above , using automated silica gel flash chromatography ( Biotage SP1 , methanol-dichloromethane ) to isolate the desired amino acid . The amino acid intermediate was coupled to aminoacetonitrile according to the general procedure and finally , the resulting indoline product was oxidized to the desired indole 8 by reaction with 1 . 05 equivalents of 2 , 3-dichloro-5 , 6-dicyano-1 , 4-benzoquinone ( DDQ ) in dichloromethane for 30 minutes . The final product was purified by automated silica gel flash chromatography ( Biotage SP1 , ethyl acetate-hexane ) . Additional details and NMR data are provided in Supplementary Methods ( Text S1 ) . Permeability measurements were performed by Wuxi Apptec , using the following procedures . MDCK-MDR1 cells ( obtained from Piet Borst at the Netherlands Cancer Institute ) were seeded onto polyethylene membranes ( PET ) in 96-well BD insert systems at 2×105 cells/cm2 for 4–6 days to obtain confluent cell monolayer formation . Test compounds were diluted with the transport buffer ( HBSS , pH 7 . 4 ) from a 10 mM stock solution to a concentration of 2 µM and applied to the apical ( A ) or basolateral ( B ) side of the cell monolayer . Permeation of the test compounds from the A to B direction or B to A direction was determined in triplicate over a 150-minute incubation at 37°C and 5% CO2 with a relative humidity of 95% . In addition , the efflux ratio of each compound was also determined . Test and reference compounds were quantified by LC-MS/MS analysis based on the peak area ratio of analyte/IS . The apparent permeability coefficient Papp ( cm/s ) was calculated using the equation:where dCr/dt is the cumulative concentration of compound in the receiver chamber as a function of time ( µM/s ) ; Vr is the solution volume in the receiver chamber ( 0 . 075 mL on the apical side , 0 . 25 mL on the basolateral side ) ; A is the surface area for the transport , i . e . 0 . 084 cm2 for the area of the monolayer; C0 is the initial concentration in the donor chamber ( µM ) . The efflux ratio ( ER ) was calculated using the equation: Percent recovery was calculated using the equation:Where Vd is the volume in the donor chambers ( 0 . 075 mL on the apical side , 0 . 25 mL on the basolateral side ) ; Cd and Cr are the final concentrations of transport compound in donor and receiver chambers , respectively . Cc is the compound concentration in the cell lysate solution ( µM ) , and Vc is the volume of insert well ( 0 . 075 mL in this assay ) . Permeability determinations were performed in triplicate and are reported as mean values . The mean total recovery was greater than 90% for all compounds tested ( compounds 1–10 ) . We initially developed the docking strategy using a number of well-known P-gp substrates and inhibitors . Specifically , we used a set of 24 known binders ( Hennessy et al . [41] , Table 1 ) that included , among others , HIV protease inhibitors , anthracyclines , vinca alkaloids , and taxanes ( Table S1 ) . Initial docking using a rigid receptor and docking box coordinates centered on the co-crystallized ligand generated poses that largely overlapped with the coordinates of the cyclic peptide in the crystal structure , with most of the compounds showing extensive exposure to solvent at the base of the cavity . By contrast , the flexible-receptor docking protocol resulted in ligands receiving much more favorable docking scores ( Figure 1 ) as well as ligand poses in which the ligands bound much ‘deeper’ in the cavity ( Figure 2 ) , with little solvent exposure . Only side chains in the binding site were treated as flexible ( Table S2 ) , and the conformational changes that allowed the ligands to bind more deeply in the cavity were modest . The most important conformational changes were those of the side chains of Phe71 , Phe332 , and Phe728 , which are located in the center of the cavity . As shown in Figure 2A , the rotamer changes in these side chains result in a more open cavity than that in the initial crystal structure . Representative top poses and Glide XP binding scores of some of the compounds from the final IFD round are shown in Figures 2B and 2C . Additional scores are provided in , and are represented as a histogram in Figure 1 , highlighting the much more favorable docking scores achieved with flexible-receptor docking . Below , we also demonstrate that the flexible receptor approach greatly improves the ability to discriminate binders from non-binders . In all of this work , we used two different scoring functions to rank compounds , Glide XP and a molecular mechanics based scoring function ( MM-GB/SA ) , in addition to the default Glide SP scoring function . The results using Glide XP and MM-GB/SA scoring were , on average , remarkably similar , considering the very different functional form and methods of parameterization . Both scoring functions performed much better than Glide SP in distinguishing binders from non-binders . ( On the other hand , the Glide XP and MM-GB/SA scoring functions did not always identify the same poses as top-ranked; see , e . g . , doxorubicin in Figure S2 . Given the size and flexibility of the binding cavity , it is likely that for some , if not all molecules , several binding modes are possible . ) For simplicity , we present mainly the results using Glide XP here , and present the remaining results using MM-GB/SA in supplementary tables and figures , in part because the results with Glide XP are slightly better in some cases . This is not altogether surprising because the molecular mechanics scoring function has been useful primarily in ranking compounds that are chemically similar , e . g . , congeneric series , and the series of compounds we use in most of the tests here are quite diverse . However , the similarity of the results using the very different scoring functions is striking , and we use the results with MM-GB/SA scoring to investigate , for example , the role of desolvation in binding by P-gp . Also shown in Figure 2D are the rigid- and flexible-receptor poses of a well-known ( non-drug ) P-gp substrate , rhodamine B , the binding mode of which has been partially elucidated by experimental data obtained using its Cys-linked analog . The flexible-receptor pose selected by Glide XP is qualitatively consistent with the experimental data in that the molecule is reasonably close to residues facing the binding cavity and shown to interact with the ligand [26] , [46] . By comparison , the pose obtained by rigid docking appears less consistent with the experimental data and also has a much less favorable docking score ( −5 . 4 kcal/mol , vs . −15 . 3 kcal/mol for the flexible receptor pose ) . As an additional control , we docked QZ59-RRR back into the crystal structure using both rigid and flexible docking protocols . The results are illustrated in Figure 3 . Rigid docking reproduced the binding mode of QZ59 , but the molecule was ‘flipped’ in comparison to its position in the crystal structure . Induced fit docking produced a similar pose with the molecule slightly shifted upwards from the original position . ( The shift was seen regardless of whether the docking box was centered on the original ligand position or shifted deeper into the cavity ) . The ligand still maintained contact with Phe724 and Val978 deemed to be important for drug binding ( as discussed in Aller paper ) , as well as with the majority of the hydrophobic residues indicated to be within 4–5 Å of the crystal pose ( Table S6 ) . The flexible docking score in this case , though not particularly high , is also more favorable than that from rigid docking . Based on the IC50 value reported for QZ59-RRR inhibition of verapamil-stimulated ATPase activity ( 4 . 8±2 . 6 µM ) , the ligand is a rather weak inhibitor , which could partially explain the weak binding score . In addition , as discussed in Methods , it was not possible to dock the compound containing selenium atoms , and these were replaced with sulfur . We next docked a set of 102 common metabolites , comprising representatives of four major classes of biological molecules including carbohydrates , amino acids , fatty acids , and nucleic acids ( Table S3 ) . We reasoned that most of these metabolites would be non-binders based on the notion that their efflux would be inefficient to cell function ( as would inhibition of P-gp by metabolites ) . As shown in Figure 1 , most of the metabolites did in fact have much less favorable docking scores than the drugs discussed above . However , a small fraction of the metabolites received docking scores similar to those of the drugs . For example , of the 26 drugs in Table S1 known to interact with P-gp , 23 had Glide XP scores of −12 kcal/mol or better , and 15 had scores <−14 kcal/mol . By contrast , 18 of the 102 metabolites received scores more favorable than −12 kcal/mol , and only 8 had scores <−14 kcal/mol . Among these metabolites with favorable scores , we subsequently identified four ( thyroxin , vitamin D3 , progesterone , and cholesterol ) that have been reported to interact with P-gp [47] , [48] , [49] , [50] and reassigned them to the binders set ( which had little affect on the ROC-type curve ) . We also searched for literature data on approximately 20 randomly selected metabolites with less favorable docking scores and were unable to find any evidence of these being P-gp substrates . We also could not identify any direct evidence for other top-scoring metabolites , such as riboflavin , retinol , and leukotriene C4 , interacting with P-gp , but it is possible that some of these metabolites are currently unrecognized substrates ( or inhibitors ) . In fact , P-gp has been suggested to export naturally derived toxins in healthy cells [51] as well as to play a role in transport of cancer-signaling lipids [5] . Investigation of a more extensive set of biologically relevant molecules is currently under way . In the absence of any direct evidence for the other metabolites , we consider them non-binders , and we quantify the ability to discriminate the known binders ( 26 drugs+4 metabolites ) and presumed non-binders ( 98 metabolites ) using an ROC-type curve in Figure 4A . Clearly , the flexible-receptor protocol results in much better discrimination between these two sets of compounds ( area under the curve , AUC = 0 . 93 ) than the rigid receptor docking , either with the docking box centered on the co-crystallized QZ59 ligand ( AUC = 0 . 78 , Figure 4A ) , or with the docking box shifted deeper into the cavity as in the flexible docking results ( AUC = 0 . 83 , Figure S1 ) . The results using flexible-receptor docking and the MM-GB/SA scoring function are very similar in this case , AUC = 0 . 93 ( Figure S3 ) . Next , we tested the ability to qualitatively reproduce results of in vitro assays regularly used to evaluate P-gp binding . For that purpose we selected the Doan et al . [11] dataset of FDA approved drugs that included results of the monolayer efflux and CAM inhibition assays . Based on the assay results , we defined P-gp binders ( N = 13 ) as molecules positive for both assays ( ER>1 . 5 and >10% CAM inhibition ) , and nonbinders ( N = 34 ) as compounds negative for both assays ( ER<1 . 5 and <10% CAM inhibition ) . The ROC curves obtained with Glide XP and the default treatment of ionization ( see Methods ) are shown in Figure 4B . The induced fit approach ( AUC = 0 . 90 ) again outperformed rigid docking ( AUC = 0 . 71 ) , although there is clearly some overlap in the distribution of scores between the binders and non-binders in this set . Some of this overlap is due to the somewhat arbitrary criteria used for distinguishing the sets of compounds , as discussed below . The results using the MM-GB/SA ( AUC = 0 . 81 ) scoring function were somewhat worse than using Glide XP ( Figure S3 ) , using the default treatment of ionization states . However , when all compounds were docked as neutral species , regardless of their pKa values , the MM-GB/SA scoring function performed much better ( AUC = 0 . 92 , Figure S4 ) ; the results using the Glide XP scoring function were similar ( AUC = 0 . 92 ) . It is not surprising that the MM-GB/SA scoring function is more sensitive to the treatment of protonation states , since charged compounds have large , unfavorable desolvation energies . It is not completely clear why treating all compounds as neutral results in better discrimination , although we note in this regard that one of the prevalent theories in the field is that P-gp substrates enter the binding cavity from the membrane ( where they are assumed to be electrically neutral ) rather than directly from the aqueous environment of the cytoplasm [4] . It is currently challenging to measure binding affinities to P-gp , and the results of standard assays are generally interpreted qualitatively ( i . e . , is it a substrate or not ) , as we have done here . However , the ratio or difference of the rates of permeability in the two directions ( basal-to-apical , PBA , and apical-to-basal , PAB ) provides a quantitative measure of how ‘strong’ a substrate a given compound is . Although there is no reason to expect the computed docking scores to necessarily correlate well with these metrics , there is , in fact , a reasonable correlation with both the difference in the rates ( Pactive = PBA−PAB ) and the more commonly used log of the efflux ratio ( Figure 5A ) . ( Results using the MM-GB/SA scoring function are qualitatively similar , and again the results using this scoring function are better when all molecules are treated as neutral ( Figures S5 and S6 ) . One advantage of representing the data this way is that it avoids somewhat arbitrary criteria used for classifying the compounds as substrates and non-substrates . The plots illustrate the two different , if somewhat overlapping ranges of the binding scores for P-gp binders and nonbinders . When the same data was plotted versus rigid docking scores , the two classes were undistinguishable ( Figure 5B ) . When interpreting these plots , one must keep in mind that efflux ratio values are a result of a complex interplay between the binding affinities of the compounds and kinetic aspects of the P-gp efflux , and may be influenced by rate of passive membrane permeability , and we do not claim that it should be possible to quantitatively predict the efflux ratio based on docking calculations alone . The P-gp binding site is highly hydrophobic and the ligand desolvation energy may have a significant effect on the ligand binding . To investigate this point , we computed free energy of transferring compounds from water to a low dielectric solvent ( chloroform ) using an approach described previously for predicting passive membrane permeability [52] , [53] . The plot of the P-gp binding scores vs . free energy of desolvation showed no correlation between the two ( Figure S7 ) , indicating that binding scores reflect specific interactions with P-gp and are not dominated solely by the polarity/hydrophobicity ( quantified using the solvation free energy ) of the compounds . Some of these specific interactions are illustrated in Figure S2 and include pi-stacking , cation-pi interactions and hydrogen bonding . These results also suggest that the physicochemical properties of the ligands that define their passive membrane permeability are different from the physicochemical determinants that define their interactions with P-gp . This finding in turn suggests that it might be possible to reduce a ligand's P-gp binding using chemical modifications without dramatically reducing its membrane permeability . We note that the computations we perform here attempt to predict the ( path-independent ) thermodynamics of transferring a ligand from water to the P-gp binding site , and thus our computations do not provide any direct information about whether the ligand enters the binding site through the membrane , or directly from the cytoplasm . Finally , we have performed a first ‘blind’ test of the method , using a series of peptidic cysteine protease inhibitors bearing natural and unnatural amino acid residues . These compounds were originally designed to test hypotheses concerning passive membrane permeability , and those results will be reported elsewhere . However , we also tested the compounds in a cell-monolayer assay ( performed by WuXi AppTec ) , specifically using P-gp transfected MDCK cells . The results are summarized in Table 1 and Figure 6 , where we again plot the predicted docking scores for binding to P-gp versus measures of the asymmetry of the permeability across the monolayer . The compound that showed the strongest evidence for P-gp mediated efflux , compound 6 , had an efflux ratio of 9 with moderate passive membrane permeability . Encouragingly , this compound had the most favorable Glide XP score ( −14 . 6 ) among the series , comparable to many of the P-gp substrates in the benchmarking results discussed above . Similarly , the compound with the least favorable docking score ( −9 . 9 , compound 2 ) had a much lower efflux ratio ( 1 . 6 ) , and the compound with the lowest measured efflux ratio ( 1 . 2 , compound 1 ) had the second lowest docking score ( −10 . 9 ) . The results are not perfect . Compound 9 has relatively polarized efflux ( ER = 5 . 8 ) , but has a docking score that would classify it only as a ‘possible’ binder ( −12 . 3 ) . However , we note that this compound has 10× lower passive membrane permeability than any other member of the series , PAB = 5×10−7 cm/s , making the determination of P-gp mediated efflux more uncertain . In general , the correlation between the docking scores and the experimental results is ‘noisier’ than in the benchmarking study using the Doan et al . data set . However , we note that the range of experimental efflux ratios , as well as the range of docking scores , is narrower in this series of peptidic compounds , which is expected due to the compounds being much more chemically similar . As such , it is gratifying to be able to confirm an ability to predict compounds more/less likely to be P-gp substrates , even in the more challenging case of a more chemically homogeneous series , albeit not with ‘quantitative’ accuracy . In this regard , we again emphasize that , even with perfect ability to predict binding affinities to P-gp ( which we certainly do not claim ) , there is no reason to expect any simple relationship between the docking scores and experimental measures of P-gp mediated efflux . Thus the docking results , as well as the experimental results , should be interpreted in qualitative terms . As general guidelines based on the work presented here , we view a Glide XP docking score of approximately −14 or lower , using the flexible docking protocol , to be a predictor of P-gp interaction , while scores of −12 or greater predict non-interaction . Intermediate scores of approximately −12 to −14 are less conclusive , but many compounds in this range show evidence of being relatively ‘weak’ substrates . In summary , we have developed an in silico method suitable for predicting compounds that are more likely or less likely to interact with P-gp . We tested the ability of this method to differentiate between binders and nonbinders of P-gp using consistently measured experimental data from P-gp efflux and calcein-inhibition assays . Treating the P-gp binding cavity as flexible is critical for obtaining good results . We suspect that this observation reflects intrinsic flexibility of the binding site , but may also be related to the relatively low resolution of the crystal structures; most of the side chains orientations are not well defined by the electron density . Encouragingly , a first ‘blind’ test of the flexible-receptor approach on a series of peptidic protease inhibitors provided additional evidence for the predictive ability of the method . Overall , our results suggest that specificity in P-gp is better understood in terms of physicochemical properties of the ligands ( and the binding site ) , rather than being defined by specific sub-sites . We also suggest that many P-gp substrates bind deeper in the cavity than the cyclic peptide in the crystal structure . Finally , we also make testable predictions concerning metabolites that may be P-gp substrates . We have not explored whether we can distinguish substrates and inhibitors of P-gp . Experimentally distinguishing these two classes is not simple , and available evidence suggests that substrates can competitively inhibit efflux of other compounds to varying extents [19] . From the standpoint of the computations , it is clear that both substrates and inhibitors must bind to P-gp , presumably in the ‘inward’ configuration represented in the crystal structures . We speculate that the mode of binding ( i . e . , where in the binding site the compounds bind ) might relate to these complicated considerations . However , the biophysical basis of coupling between ligand binding and ATP hydrolysis ( enhancement and inhibition ) remain poorly characterized , making further progress difficult at this time .	With many drugs failing in the preclinical stages of drug discovery due to undesirable ADMETox ( absorption , distribution , metabolism , excretion and toxicity ) properties , improvement of these properties early on in the process , alongside the optimization of the compound activity , is emerging as a new focus in the pharmaceutical field . One of the key players affecting pharmacokinetic profiles of many clinically relevant compounds is an active efflux transporter , P-glycoprotein . Expressed predominantly at various physiological barriers , it can influence drug absorption ( intestinal epithelium , colon ) , drug elimination ( kidney proximal tubules ) and drug penetration of the blood-brain barrier ( endothelial brain cells ) . Moreover , its increased expression in cancer cells has been linked to resistance to multiple drugs in tumors . In this study we describe a computational approach that allows prediction of which compounds are more likely to interact with P-gp . We have tested the ability of this method to differentiate between binders and nonbinders of P-gp by using consistently measured in vitro experimental data . We also implemented a blind test on a series of peptidic cysteine protease inhibitors with encouraging outcome . Overall , our results suggest that this method provides a qualitative , quick , and inexpensive way of evaluating potential drug efflux problem at the early stages of drug development .	[ "Abstract", "Introduction", "Methods", "Results", "Discussion" ]	[ "biomacromolecule-ligand", "interactions", "biochemistry", "biology", "computational", "biology", "drug", "discovery" ]	2011	Predicting Binding to P-Glycoprotein by Flexible Receptor Docking
Ticks salivate while feeding on their hosts . Saliva helps blood feeding through host anti-hemostatic and immunomodulatory components . Previous transcriptomic and proteomic studies revealed the complexity of tick saliva , comprising hundreds of polypeptides grouped in several multi-genic families such as lipocalins , Kunitz-domain containing peptides , metalloproteases , basic tail secreted proteins , and several other families uniquely found in ticks . These studies also revealed that the composition of saliva changes with time; expression of transcripts from the same family wax and wane as a function of feeding time . Here , we examined whether host immune factors could influence sialome switching by comparing sialomes of ticks fed naturally on a rabbit , to ticks artificially fed on defibrinated blood depleted of immune components . Previous studies were based on transcriptomes derived from pools of several individuals . To get an insight into the uniqueness of tick sialomes , we performed transcriptomic analyses of single salivary glands dissected from individual adult female I . ricinus ticks . Multivariate analysis identified 1 , 279 contigs differentially expressed as a function of time and/or feeding mode . Cluster analysis of these contigs revealed nine clusters of differentially expressed genes , four of which appeared consistently across several replicates , but five clusters were idiosyncratic , pointing to the uniqueness of sialomes in individual ticks . The disclosure of tick quantum sialomes reveals the unique salivary composition produced by individual ticks as they switch their sialomes throughout the blood meal , a possible mechanism of immune evasion . Ticks remain attached to their hosts for several days or weeks while increasing their weight several hundred fold . The tick acquires its meal from a skin feeding cavity containing a haemorrhagic pool that continually supplies blood to the ectoparasite . This feeding pool is maintained by tick saliva that is intermittently secreted into the host while alternating sucking and spitting throughout the meal [1] . Tick saliva contains a complex potion of pharmacologically active compounds including non-peptidic components such as vasoactive and immunosuppressive prostaglandins and adenosine , as well as peptidic kratagonists ( agonist binding proteins active against serotonin , histamine , inflammatory leukotrienes and cytokines ) , anti-clotting , anti-platelet , anti-complement and immunosuppressive peptides; metalloproteases can digest fibrin and inhibit angiogenesis [2 , 3] . Hosts can mount immune responses against ticks and non-natural hosts can mount very effective anti-tick immunity [4] . Perhaps for this reason , tick salivary proteins evolve at a fast pace to avoid immune recognition . Many salivary components of Ixodes ticks are members of expanded multigene families , such as lipocalin , Kunitz , cystatin , basic tail , ixostatin , antigen-5 , and metalloprotease families ( for information on these families , see [2] ) . Interestingly , many members of these families are expressed only at a particular time during feeding [5–7] . In other words , the mechanism of paralogue switching may cause the host antibody-mediated response to fail because the antibody response takes several days , but by the time the antibody is produced , its target is gone and a new paralog is in place . The sialome switching mechanism could theoretically be driven by a physiological feeding “clock” . Alternatively , the switch could be triggered by a host response , such as innate and/or acquired immunity or a tick “stressor” signal , or both . The information on tick sialomes ( from the Greek sialo = saliva ) has been so far assembled from conventional Sanger or next generation ( 454 or Illumina ) transcriptomes , made from pools of dozens of salivary glands . This approach is informative for obtaining the general sialomic repertoire that a given population of a species of tick can produce but it is not informative about which components of this repertoire are being recruited by individual members of that population . In the present work , we made libraries from single adult female I . ricinus ticks that were fed for 24 , 48 , or 72 h , artificially , using membrane feeding , or naturally on a rabbit . The artificial feeding treatment was an attempt to investigate whether the lack of host inflammatory and innate immunity responses would suppress sialome changes . Moreover , this comparative approach led to the identification of several transcripts that are specifically up-regulated by the host immune response at an early stage of feeding , and therefore may have potential as candidates for effective anti-tick interventions . We identified an additional 434 novel polypeptide sequences within these 1 , 279 contigs , most from secreted products at 72 h of feeding , a time point not previously explored with I . ricinus sialo-transcriptomes [7 , 8] . Nearly 2 , 000 novel sequences were identified and deposited in public databases . All animal experiments were carried out in accordance with the Animal Protection Law of the Czech Republic No . 246/1992 Sb , ethics approval No . 095/2012 and protocols approved by the responsible committee of the Institute of Parasitology , Biology Centre of the Czech Academy of Sciences . Male and female adult Ixodes ricinus ticks were collected by flagging in a forest near the town České Budějovice in the Czech Republic . Laboratory rabbits reared in the animal facility of the Institute of Parasitology were used for adult tick feeding . Females were allowed to oviposit . Larvae and derived nymphs were fed on guinea pigs . Adult ticks from a single egg batch were used for this work ( Fig 1 ) . Resulting adult female siblings were then fed either naturally on a rabbit or fed a reconstituted rabbit blood exploiting an artificial membrane feeding system ( see below ) . Rabbit blood was manually defibrinated using an autoclaved fork . The whole blood was then separated by centrifugation 2500 x g , 15 min , 4°C . Resulting serum was then transferred into a fresh tube and the red blood cells were washed 3 times in sterile PBS to remove remaining white cells ( as evidenced by Giemsa staining ) . Collected serum was heat inactivated at 56°C for 30 min in 15-ml tubes using a water bath . The heat inactivated ( HI ) serum was then sterile filtered using a 0 . 22 μm filter ( Merck Millipore ) . The efficiency of heat inactivation was verified as described [9] . Briefly , Escherichia coli ( TOP10 , Invitrogen ) was grown to OD = 1 in Lysogeny Broth ( LB ) , then diluted one thousand times to 106CFU/ml , and 50 μl of the culture was mixed with 50 μl of: a ) sterile PBS , b ) gentamicin ( 10 μg/ml ) , c ) active serum , d ) heat inactivated serum . The cultures were then incubated at 225 rpm , 37°C for 3 h and then plated out on LB plates . To reconstitute the whole blood , the HI serum was then added back to washed red blood cells . Unaltered hematocrit levels were verified by SDS-PAGE ( 20μg protein per lane ) and UV/VIS spectrophotometry ( 2μl drop of 100 times diluted blood meal ) , as previously described [10] ( S1 Fig ) . Artificial membrane feeding of I . ricinus females was performed according to Kröber and Guerin [11] using stationary feeding units as described previously [10 , 12] . The feeding unit was lined with a thin ( 80‒120 μm ) silicone membrane , treated with a bovine hair extract prepared in dichloromethane ( 0 . 5 mg of low volatile lipids ) , and UV-sterilised . Fifteen I . ricinus females were put in the assembled feeding unit and allowed to attach . After 8 hours , all unattached females were removed and an equal number of males to attached females were added into the feeding unit . Reconstituted rabbit blood was served without supplementation and was regularly exchanged in 8h intervals . At least three females of comparable sizes were then forcibly removed from the membrane after 24 , 48 , and 72 hours of feeding . Salivary glands were dissected from a single tick in biological triplicates for both treatments–rabbit-fed ( R ) , membrane-fed ( M ) , and three time points of feeding ( 24 h , 48 h and 72 h ) , giving in total 18 individual pairs of salivary glands ( Fig 1 ) . Ticks were dissected on a double sticky tape in a drop of DEPC-treated PBS . The purity of dissected SG was checked by microscope and all contaminating tissues , mainly trachea , were removed . The salivary glands were then homogenized in RA1 buffer ( Machery Nagel ) using a syringe and 29G needle . Total RNA was extracted using a Nucleospin RNA kit ( Machery Nagel ) . The quality of the RNA samples was confirmed by lab-on-chip analysis using the 2100 Bioanalyzer ( Agilent Technologies , Inc . Santa Clara , CA , USA ) . Total RNA samples ( labelled according to the feeding mode and time as M24_1 , 2 , 3; R24_1 , 2 , 3; M48_1 , 2 , 3; R48_1 , 2 , 3; M72_1 , 2 , 3; R72_1 , 2 , 3 ) were submitted to the North Carolina State Genomic Sciences Laboratory ( Raleigh , NC , USA ) for Illumina RNA library construction and sequencing . Purification of messenger RNA ( mRNA ) was performed using the oligo-dT beads provided in the NEBNext Poly ( A ) mRNA Magnetic Isolation Module ( New England Biolabs , USA ) . Complementary DNA ( cDNA ) libraries for Illumina sequencing were constructed using the NEBNext Ultra Directional RNA Library Prep Kit ( NEB ) and NEBNext Mulitplex Oligos for Illumina ( NEB ) using the manufacturer-specified protocol . Briefly , the mRNA was chemically fragmented and primed with random oligonucleotides for first strand cDNA synthesis . Second strand cDNA synthesis was then carried out with dUTPs to preserve strand orientation information . The double-stranded cDNA was then purified , end repaired and “a-tailed” for adaptor ligation . Following ligation , the samples were selected for a final library size ( adapters included ) of 400–550 bp using sequential AMPure XP bead isolation ( Beckman Coulter , USA ) . Library enrichment was performed and specific indexes for each of the 18 samples were added during the protocol-specified PCR amplification . The amplified library fragments were purified and checked for quality and final concentration using an Agilent 2100 Bioanalyzer with a High Sensitivity DNA chip ( Agilent Technologies , USA ) . The final quantified libraries were pooled in equimolar amounts for sequencing on four lanes of an Illumina HiSeq 2500 DNA sequencer , utilizing a 150 bp single end sequencing flow cell with a HiSeq Reagent Kit v4 ( Illumina , USA ) . Flow cell cluster generation for the HiSeq2500 was performed using an automated cBot system ( Illumina , USA ) . The software package Real Time Analysis ( RTA ) , version 1 . 18 . 64 , was used to generate raw bcl , or base call files , which were then de-multiplexed by sample into fastq files for data submission using bcl2fastq2 software version v2 . 16 . 0 . The raw fastq files were deposited in the Sequence Read Archives ( SRA ) of the National Center Biotechnology Information ( NCBI ) under accession SRP071001 of bioproject PRJNA312361 and biosample SAMN04497582 . Assembly of all reads was done as described previously using the assemblers Abyss and Soapdenovo-Trans with every kmer ending in 1 and 5 ( -k program switch ) from 21 to 95 [13–17] . Resulting contigs were re-assembled by a pipeline of blastn and cap3 assembler [18] as described earlier [19] . Coding sequences were extracted based on blastx [20] results deriving from several database matches , including a subset of the non-redundant protein database of the NCBI containing tick and other invertebrate sequences , as well as the Swissprot and Gene Ontology ( GO ) databases . Open reading frames larger than 150 nt were also extracted if they had signal peptides indicative of secretion , as evaluated by version 3 . 0 of the SignalP program [21] . Reads from the four libraries were mapped back into the CDS by blastn with a word size of 25 and allowing one gap . Reads were mapped up to a maximum of five different CDS if the blast scores were the same for all matches . The program edgeR was used in ancova mode to detect statistically significant differentially expressed genes ( DEG ) according to treatment or time variables and displayed as Volcano plots [22] . EdgeR inputted the read matrix for genes having at least one library expressing a RPKM [23] ( fragments per thousand nucleotides per million reads ) equal or larger than 10 . For heat map display [24] of the CDS temporal expression , Z scores of the RPKM values were used . Heatmaps were produced with the programs gplots and heatmap . 2 using R [25] . Differential gene expression clustering was done with the program Expander version 7 . 1 [26] , using as input , RPKM data and the click algorithm . More details of the input are available in the results section , which also contains hyperlinks to several databases , as explained previously [19 , 27] . Deduced coding sequences and their translations were deposited to the Transcriptome Shotgun Assembly database DDBJ/EMBL/GenBank under the accessions GEGO01000001-GEGO01007692 cDNA preparations were made from 0 . 1 μg of total RNA using the Transcriptor High-Fidelity cDNA Synthesis Kit ( Roche Diagnostics , Germany ) . The cDNA served as templates for subsequent quantitative expression analyses by RT-qPCR . Samples were analysed by a LightCycler 480 ( Roche ) using Fast Start Universal SYBR Green Master Kit ( Roche ) . Relative expressions were calculated by the ΔΔCt method . The expression profiles were normalised to I . ricinus elongation factor 1α ( ef-1α ) . Primers used for validation are listed in S1 Table . The feasibility of the recently introduced technique for artificial feeding of hard ticks [11] allowed us to investigate the as yet insoluble problem of identification of biologically active components from tick saliva that specifically react to the host’s immune/inflammatory response . In order to address the question , how do ticks avoid active host defense mechanisms , we compared the sialomes of individual ticks fed at different times on a natural host to the sialomes of ticks fed artificially on a blood meal deprived of active host immunity components . The deprived natural immunity was produced by serum heat inactivation and washing out of the remaining white cells from manually defibrinated blood ( Fig 1 and S1 Fig ) . Adult I . ricinus females that were used in this study were derived from a single egg batch to reduce genetic variation . Our experimental design consisted of two feeding modes: ( a ) natural feeding on a laboratory rabbit and ( b ) artificial feeding of heat-inactivated rabbit blood . Individual pairs of salivary glands were dissected at three time points ( 24 , 48 and 72 h ) with 3 replicates for each time-point , thus needing the construction of 18 libraries ( S2 Table ) . The “de novo” assembly of these 18 tick transcriptomes enabled the identification of 1 , 907 novel protein sequences , 406 of which were classified as of a secreted nature . We additionally extended 115 sequences that were at least 95% identical to publicly available proteins , but 5% or longer in length . Previous I . ricinus sialo-transcriptomes were assembled from over half a billion Illumina and pyrosequencing reads [5 , 7 , 8] , but they did not include 72h samples from adult ticks , from whence the majority of these new transcripts were derived , emphasizing the still unknown dimension of the full sialome repertoire of I . ricinus . These 18 libraries were loaded in each of three Illumina HiSeq 2500 lanes ( single ended protocol ) , obtaining from 18 . 5 to 27 million reads for each library , averaging 149‒150 nt in length with a median and L50 size of 151 after removal of contaminating primers ( S2 Table ) . The ~435 million reads were assembled together with the previously assembled salivary and midgut transcriptome of I . ricinus [5 , 12] , from which 40 , 490 coding sequences ( CDS ) were extracted . All deducted coding sequences and their reads are available for browsing as a hyperlinked Excel spreadsheet ( further referred to as Source data 1 ) at http://exon . niaid . nih . gov/transcriptome/Ixric-18/S1-web . xlsx . After mapping the reads of the 18 libraries to these CDS and calculating the RPKM for each , 20 , 773 contigs contained at least one library yielding a RPKM equal or larger than 10 ( Source data 1 ) . To identify differentially expressed genes ( DEG ) , we submitted the read matrix of these 20 , 773 contigs to an analysis of covariance ( ANCOVA ) test using the edgeR package . A total of 1 , 279 contigs were identified as being differentially expressed with a false discovery rate ( FDR ) < 0 . 05 . The results of the ANCOVA can be found in columns FW‒GG of the Source data 1 spreadsheet , and DEG’s for feeding treatment and times of feeding can be identified by sorting the columns for the respective fold change . Volcano plots ( Fig 2 ) show the large degree of variation found , with ranges of expression being over 210 . These uncommonly large rates have been detected previously in tick salivary transcriptomes as a function of time [5 , 6] . Selected up-regulated transcripts from 48 h to 24 h and from 72 h to 48 h of naturally-fed ticks are listed in S2 and S3 Tables , respectively . Most abundant transcripts across replicated libraries of each time-point of naturally-fed ticks are listed in S5 , S6 , and S7 Tables . We hypothesised that the absence of active host immunity would lead to a stable sialome , without much time change in the M as opposed to the R mode of feeding . Indeed the ANCOVA revealed the mode significance , suggesting this hypothesis to be true . However , the M group also showed time-dependent differential expression . To obtain further insight into the sialo-transcriptome DEG’s , we submitted the RPKM data for all 1 , 279 contigs identified as DEGs by edgeR for cluster analysis using the click algorithm of the Expander package [26 , 28] . Following Z normalization of the data and using default parameters , nine clusters were identified , leaving out 21 singletons . Column X of the Source data 1 spreadsheet contains the cluster membership number of the contigs . Sorting on this column can retrieve the contigs for each cluster . Of the nine clusters , four were clearly related to time and/or type of feeding ( Fig 3A ) . Cluster 1 has 541 genes over-represented in both M and R types at 72 h , representing a time dependent/feeding mode-independent group of genes . Cluster 3 with 177 genes is M 24h over-represented; clusters 5 and 8 are both over-represented in rabbit feeding at 24 h / 48h and 48 h / 72 h , respectively . Clusters 5 and 8 , specifically , may represent groups of genes responsive to host immune/inflammatory interaction . The over-representation was not always observed in all three replicates . Fig 4 depicts the heat maps of these clusters . For each contig , the sample producing the larger RPKM value was recorded regarding type of feeding ( M or R ) , time of feeding ( 24 , 48 , or 72 h ) , and replicate number ( 1 , 2 , or 3 ) . While four of the nine clusters were associated with time or mode of feeding of a particular set of ticks , five of the nine clusters had a single tick responsible for the cluster of DEG’s ( Fig 3B ) . Heat maps of four of the idiosyncratic clusters shown in Fig 5 display their unique gene expression pattern . The majority of these idiosyncratic DEGs were of the secreted class , and included classic salivary members of the Kunitz , lipocalin , basic tail , and 8 . 9 kDa . For each contig , the sample producing the larger RPKM value was recorded regarding type of feeding ( M or R ) , time of feeding ( 24 , 48 , or 72 h ) , and replicate number ( 1 , 2 , or 3 ) . Since five of nine clusters of DEGs were ascribed to single ticks ( Fig 3B ) and since the majority of the encoded proteins in these clusters are secreted proteins , these results indicate that five of the 18 ticks were secreting a unique sialome repertoire . This may indicate that sialomes switch more frequently than the 24 h intervals that were used in this study . These idiosyncratic sialomes could represent , for example , “typical” average sialomes of 36 h or 60 h , not sampled in this study . Even though there is a growing body of data that suggests tight time-dependent regulation of gene expression in salivary glands of feeding ticks , virtually nothing is known about the actual mechanisms that orchestrate the individual sialome switches . We have identified multiple transcripts encoding histone-modifying enzymes that were suggested to manage changes in sialome diversity [5] . Histone acetylases of GNAT ( Ir-271878 ) and MYST ( Ir-260682 , Ir-242196 , Ir-256281 ) families were moderately expressed ( RPKM values range spans between 4‒33 ) across all libraries . Histone deacetylases ( Ir-240628 , Ir-261552 , Ir-963 , Ir-262316 ) were expressed with RPKM values ranging between 2‒37 across all libraries . We have also identified two transcripts encoding bromodomain proteins ( Ir-248264 and Ir-250997 ) , which are expressed with RPKM values ranging between 4‒16 across all libraries . Surprisingly , we have identified around forty contigs encoding lysine methyltransferases , but only a single arginine methyltransferase ( Ir-262971 ) and single histone demethylase ( Ir-239028 ) . All of these transcripts , however , displayed very stable expression levels with very low inter-individual variability , and it is therefore difficult to conclude that these enzymes would orchestrate sialome switching . Tick salivary glands also express numerous copies of general transcription factor-encoding transcripts . Since approximately half of the assembled transcripts encode secretory proteins , not only timed orchestration of gene transcription , but also appropriate protein synthesis and folding must be organised . A transcript ( Ir-258585 ) encoding X-box binding protein 1 ( XBP-1 ) was found as the most abundant transcription factor across our libraries . This gene encodes a homologue of Drosophila melanogaster “bZIP”-containing transcription factor , which participates in the unfolded protein response , an evolutionarily conserved signalling pathway activated by an overload of misfolded proteins in the endoplasmic reticulum [29] . This transcription factor was implicated in the biology of the fly salivary glands , where the protein was suggested to stimulate the folding capacities of the ER in cells committed to intense secretory activities [30] . This transcription factor might , therefore , switch on a battery of proteins , as a stress response , to facilitate correct folding of thousands of proteins in tick salivary glands . Feeding treatment and time both promoted significant differentially expressed transcripts . However , the mode of feeding yielded a narrower range of differential expression and a smaller number of DEGs in comparison to time variables . We have identified individual transcripts with statistically significantly up-regulated levels in libraries of artificially fed ticks in comparison to naturally fed ticks . The transcripts are listed in Table 1 for the 24 h time-point . To support our DEG analysis , we ran RT-qPCR analyses on cDNA templates from an independent tick cohort . We selected 24 h libraries , as this time interval seems to be the most relevant for potential targeting of tick-borne pathogen transmission . We have confirmed elevated levels of most identified DEGs in libraries of naturally fed ticks ( Fig 6 ) . These transcripts encode metalloproteases ( Ir-249265 , Ir-226907 , Ir-237695 ) , 18 . 3 kDa basic tail superfamily proteins ( Ir-SigP-242556 , Ir-SigP-258570 ) , anti-complement proteins ( Ir-SigP-241930; Ir-261824 , Ir-SigP-239926 ) , and secreted protein precursor ( Ir-1315 ) . To rebut the argument that membrane fed ticks underwent slower feeding progression and , therefore , did not reach the peak levels of expression , we also verified transcript levels at 48 h and 72 h time-points . As the RPKM values of these transcripts in membrane-fed ticks did not increase , even at later time-points , we concluded that expression of the verified DEGs was indeed induced by active components of the host blood . DEGs upregulated in naturally fed ticks at 48 h and 72 h are listed in S8 and S9 Tables , respectively . Our results show that three out of sixteen contigs encoding members of the 18 . 3 kDa subfamily of basic tail superfamily ( use alphabetical sorting in column AB of the Source data 1 spreadsheet ) are clearly responsive to active host immunity compared to membrane feeding ( Table 1 ) . On the other hand , contig Ir-SigP-255938 was substantially expressed in early M ( 24 ) -fed compared to R ( 24 ) -fed ticks and the other 18 . 3kDa members displayed quite inconsistent inter-individual variability ( Source data 1 ) . Some of the other 54 basic-tail proteins encoding transcripts ( tagged as BTSP in column AB of the Source data 1 spreadsheet ) were highly expressed independent of the feeding mode during all feeding intervals and belong to the most abundant transcripts during the 48 and 72 hours post attachment ( S6 and S7 Tables , respectively ) . These results accord well with previously published results on expression of BTSP superfamily members in I . ricinus females [7] and collectively suggest an important and specific role of these molecules at the tick-host interface , which is still not well understood and definitely worthy of further investigation . One of the largest multigene families found in our transcriptome project comprises 173 contigs annotated as secreted metalloproteases ( Column AB of Source data 1 spreadsheet ) , five of those belonging to the most host-responsive transcripts at the early stage of feeding ( Table 1 ) . These enzymes are believed to function as anti-hemostatic and anti-angiogenic factors [31 , 32] and their expression in I . ricinus salivary glands was reported to be specifically dependent on the stage and feeding status [5] . Of special interest are three transcripts encoding host-responsive I . ricinus anti-complement proteins at the early stage of feeding ( Table 1 ) . Anti-complement activity of tick saliva was discovered three decades ago in I . scapularis ( formerly I . dammini ) [33] and one molecule responsible for this activity , tagged as Isac ( for I . scapularis salivary anti-complement ) was purified and cloned [34] . The sialome analysis of I . scapularis later revealed the existence of a multigene family encoding Isac paralogs in this species [32] . In our transcriptome ( Source data 1 ) we found 10 contigs encoding proteins homologous or closely related to the previously characterized Isacs from I . ricinus , namely IRAC I and IRAC II [35] , IxAC B1‒5 [36] , or other putative Isac anti-complement proteins identified more recently in the I . ricinus salivary gland transcriptome [7] . IRAC I and II as well as IxAC B1‒B5 proteins were reported to inhibit the alternative pathway of mammalian complement activation via binding to properdin that prevents the formation of the C3/FactorB convertase complex [36] . The sequential expression of individual Isac functional homologs with divergent primary structures ( antigenic variability ) illustrates how ticks are capable of avoiding the host antigen-specific immune response in the course of feeding [37] . Since the vertebrate complement system apparently plays a decisive role in susceptibility or resistance to infection by Borrelia burgdorferi sensu lato [38–40] , an effective targeting of anti-complement activities in tick saliva holds promise as a possible strategy to prevent the transmission of Lyme disease . The tick salivary gland is a complex tissue that facilitates fluid absorption and secretion when off- or on-host , respectively [41] . As ticks feed , the tissue undergoes structural changes to facilitate secretion of bioactive components of saliva [42] . Previous studies revealed that feeding progression is linked with a timed regulation of gene expression in several tick species [5 , 43 , 44] . These studies clearly demonstrated that mRNA transcript repertoires differ substantially among groups of pooled salivary glands dissected at different time-points of feeding . In this study , we evaluated salivary gland transcriptomes from an “individualised” point of view and described single tick sialome responsiveness to active host immune components at the initial time-points of feeding . The list of all assembled contigs and their respective RPKM values are available through a hyper-linked spreadsheet: http://exon . niaid . nih . gov/transcriptome/Ixric-18/S1-web . xlsx , and can be blasted using a TSA blast , BioProject number 312361 . The disclosure of tick quantum sialomes reveals the unique salivary composition of individual ticks as they switch their sialomes throughout the blood meal . The idiosyncratic nature of several quantum sialomes suggests sialomes switch within less than 24 h , the sampling time of this study . These switches have been proposed to exist as a mechanism of immune evasion [5–7 , 32] but we now can start to appreciate their diversity and individual tempos . The rapid changes in transcript repertoires beg the following questions to be addressed because interference with the mechanism of sialome switching may become an interesting target for tick and tick-borne pathogen control: What is the mechanism underlying sialome switching ? Does it involve classical transcription factor activation/suppression activated by a signal transduction cascade ? Is it subject to epigenetic regulation ? Moreover , we demonstrated that artificial membrane feeding of hard ticks enables the identification of salivary gland transcriptome responses to dietary stimuli . Comparing sialomes at given time-points , but differing in the mode of blood-meal status ( immunologically-passive blood meal in artificial feeding versus immunologically-active blood meal in natural feeding ) , we identified several transcripts that were expressed only upon feeding on active blood-meal , i . e . naturally-fed on a host ( Fig 2 , Table 1 ) . Anti-complement proteins , metalloproteases , the 18 . 3 kDa basic-tail protein superfamily , and a secreted protein precursor were substantially up-regulated in salivary glands of naturally-fed ticks at 24 h , a time-point essential for tick attachment and pathogen transmission . These data thus reveal a novel list of potential vaccine candidates that are inducibly expressed and might assist in evading host immunity and/or facilitate pathogen transmission .	In this work , we confirm previous reports that the repertoire of tick salivary gland transcripts changes as a function of time , but in addition , we now identify transcripts that change their levels according to the mode of feeding of ticks . Implementation of membrane feeding allowed us to feed ticks on an immune-deficient diet and identify transcripts that are subject to immunity-stimulated expression . Such identification may help to prioritise selection of salivary gland transcripts for further investigation . One novelty of this work was creating cDNA libraries from a single pair of salivary glands , which helped to gain insight into sialomic diversity at the single tick level . We observed that ticks express a battery of genes in defined clusters as feeding progresses ( over tested replicates ) , but also individual ticks were found to express idiosyncratic clusters of genes . Such a biological phenomenon may imply novel tick mechanisms for evading host-mediated recognition of tick antigens .	[ "Abstract", "Introduction", "Methods", "Results", "and", "discussion" ]	[ "medicine", "and", "health", "sciences", "body", "fluids", "enzymes", "gene", "regulation", "regulatory", "proteins", "metalloproteases", "enzymology", "vertebrates", "rabbits", "dna-binding", "proteins", "saliva", "animals", "mammals", "animal", "models", "experimental",...	2018	Sialome diversity of ticks revealed by RNAseq of single tick salivary glands
Synthesis of ribosomal RNA by RNA polymerase I ( RNA pol I ) is an elemental biological process and is key for cellular homeostasis . In a forward genetic screen in C . elegans designed to identify DNA damage-response factors , we isolated a point mutation of RNA pol I , rpoa-2 ( op259 ) , that leads to altered rRNA synthesis and a concomitant resistance to ionizing radiation ( IR ) -induced germ cell apoptosis . This weak apoptotic IR response could be phenocopied when interfering with other factors of ribosome synthesis . Surprisingly , despite their resistance to DNA damage , rpoa-2 ( op259 ) mutants present a normal CEP-1/p53 response to IR and increased basal CEP-1 activity under normal growth conditions . In parallel , rpoa-2 ( op259 ) leads to reduced Ras/MAPK pathway activity , which is required for germ cell progression and physiological germ cell death . Ras/MAPK gain-of-function conditions could rescue the IR response defect in rpoa-2 ( op259 ) , pointing to a function for Ras/MAPK in modulating DNA damage-induced apoptosis downstream of CEP-1 . Our data demonstrate that a single point mutation in an RNA pol I subunit can interfere with multiple key signalling pathways . Ribosome synthesis and growth-factor signalling are perturbed in many cancer cells; such an interplay between basic cellular processes and signalling might be critical for how tumours evolve or respond to treatment . Multicellular organisms use genetically determined cell death mechanisms – most prominently apoptosis – to ensure the timely and innocuous removal of superfluous , damaged , or potentially harmful cells . Apoptosis is tightly regulated and an integral part of the delicate balance between cell proliferation and cell loss , which is essential for the formation and maintenance of tissues and organs . Disturbances leading to excessive or diminished apoptosis contribute to devastating human diseases with increasing prevalence worldwide , such as neurodegeneration , immunological disorders , and cancer . The integrity of our genome is continuously challenged by endogenous processes ( replication errors , reactive oxygen species , by-products of cellular metabolism ) and by exogenous genotoxic agents – naturally occurring or iatrogenic [e . g . , 1] . Depending on the cell type , cell cycle stage , metabolic state and probably also tissue context , excessive DNA damage can be a strong stimulus for cells to undergo apoptosis . Detailed knowledge of the DNA damage response network and its failures is required for our understanding of tumor formation and progression; and also for effective and safe tumor treatment , as many of the current treatments produce DNA damage to induce replicative arrest or death of rapidly proliferating tumor cells . Nucleoli have been increasingly acknowledged as pivot systems for homeostatic regulation and stress responses [e . g . ] , [ 2]–[4] . Hypertrophic and irregularly shaped nucleoli were reported to be characteristic of malignant cells already at the end of the 19th century , and have gained a high prognostic value for several human neoplasias [5] . However , large nucleoli and increased ribosome biogenesis are common signatures of proliferating cells given the increased demand for protein synthesis . It has therefore been challenging to explore whether nucleolar hypertrophy in cancer cells is a mere expression of rapid proliferation , or whether the enlarged nucleoli could play a causative role in tumor development [6] . Synthesis of ribosomal RNA accounts for up to 75% of total transcriptional activity in yeast [7] and at least 50% of the synthetic effort of rapidly proliferating eukaryotic cells are expended on ribosome production [8] . Transcription of rRNA by RNA polymerase I ( RNA pol I ) is the rate-limiting step for ribosome synthesis [e . g . , 9] and a major target of many signalling pathways of cellular growth and proliferation , e . g . , Ras/ERK [10] , Myc [e . g . , 11] , mTOR [12] , pRb [13] , and p53 [14] . In turn , physiological or pathological changes in rRNA transcription and processing , or in nucleolar integrity affect cellular fate decisions through altered translation , but also through more direct regulatory signalling [6] , [15] , [16] . Various disturbances in early steps of rRNA synthesis have strong pro-apoptotic effects [17]–[19] . Somewhat contrasting , partial depletion of ribosomal proteins in zebrafish proved to be cancerogenic [20] . Also in humans , tumor predisposing diseases could be linked to mutations in rRNA processing and ribosome assembly factors , such as Dyskeratosis congenita ( to pseudouridine synthase DKC1 ) or Diamond-Blackfan anaemia ( to various ribosomal proteins ) [e . g . ] , [ 16 , 21] . The germ line of Caenorhabditis elegans has proven a versatile model to dissect the classical DNA damage responses [22] in the context of an optically transparent , living organism [23] . Proliferation arrest of mitotic germ stem cells and apoptosis of meiotic germ cells occur in two spatially separate areas and can be morphologically followed by Nomarski differential interference contrast ( DIC ) microscopy . Under standard laboratory conditions , germ cells at the exit from late meiotic pachytene stage alternatively progress into large oocytes or undergo apoptosis [24] . This developmental , ‘physiological’ apoptosis affects an estimated half of all oocyte precursors in an apparently stochastic manner , possibly to supply the rapidly growing oocytes with cytoplasmic constituents [reviewed in [25]] . Two regulatory pathways of global importance for cellular growth and proliferation have been associated with physiological cell death: the Rb complex [26] , [27] and Ras/MAP kinase signalling [28] , [29] . In contrast to physiological apoptosis , DNA damage-induced apoptosis depends on CEP-1 ( a functional C . elegans homolog of mammalian p53 [30] , [31] ) , which transcriptionally upregulates the two pro-apoptotic BH3-only proteins EGL-1 and CED-13 [32] . In the classical model ( schematic in Fig . S13A ) , EGL-1 activates the core apoptotic machinery by binding to CED-9/Bcl-2 , therewith releasing CED-4/Apaf-1 [33] , which in turn oligomerises and serves as a platform for activation of the effector caspase CED-3 [reviewed in [34]] . Many of the factors involved in the maintenance of genome integrity are evolutionarily conserved in C . elegans . In this work , we characterise the mutation rpoa-2 ( op259 ) – isolated in an unbiased forward genetic screen – which affects the second largest subunit of RNA polymerase I ( RNA pol I ) and thus an essential and highly conserved gene . Taking advantage of this viable allele , we identified causal links between ribosome synthesis and cellular checkpoints in the context of a whole organism . We show that rpoa-2 ( op259 ) leads to stimulation of stress signalling and notably to chronic activation of the pro-apoptotic CEP-1/p53 pathway . It simultaneously interferes with Ras/MAPK signalling , possibly by increasing LIP-1 phosphatase activity , and thereby modulates apoptosis downstream of CEP-1 , likely at the level of CED-9 . Only the combination of these two effects explains the apoptotic phenotype of the RNA pol I mutant under normal conditions and in response to irradiation . We searched for additional regulators of cellular DNA damage responses with a forward genetic ( EMS mutagenesis ) screen ( see Methods ) . Of 2'000 haploid genomes tested , one candidate mutation , op259 , led to reduced apoptosis of meiotic germ cells following ionising radiation ( IR ) ( Fig . 1A ) . In op259 , the basal levels of germ cell corpses were lower than wild type , and IR-induced apoptosis was strongly reduced ( Fig . 1B ) . op259 mutants also showed a reduced apoptotic response to ultraviolet light ( UV-C ) ( Fig . S1A ) , speaking for a more global impairment of damage-induced apoptosis . Genetic mapping and molecular characterisation of op259 mutants ( see Methods ) led to the identification of a C→T transition in the second exon of F14B4 . 3 ( Fig . S2A ) , resulting in a Proline to Serine change ( P70S ) . F14B4 . 3 codes for the second-largest subunit of eukaryotic RNA polymerase I; we hence named the gene rpoa-2 . Several lines of evidence suggest that this missense mutation is the cause of the apoptotic defect in op259 mutants . First , we tried to phenocopy rpoa-2 ( op259 ) by knockdown of F14B4 . 3 using RNAi by feeding [35] , [36] . Whereas worms grown on control RNAi conditions showed a strong increase of the cell corpse number upon irradiation , F14B4 . 3 ( RNAi ) -treated animals had only few germ cell corpses ( Fig . S1C ) . Second , transgenic expression of wild-type RPOA-2 could mostly rescue the apoptotic phenotype of rpoa-2 ( op259 ) ( Fig . S1B ) . Third , rpoa-2 ( op259 ) failed to complement the independent deletion allele rpoa-2 ( ok1970 ) available from a public source [37] . Because this allele is homozygous lethal , we generated transheterozygous rpoa-2 ( ok1970/op259 ) animals by crossing rpoa-2 ( op259 ) males with rpoa-2 ( ok1970 ) /hT2 hermaphrodites . rpoa-2 ( ok1970/op259 ) were viable and shared the apoptotic phenotype of rpoa-2 ( op259 ) homozygotes ( Text S1 and Fig . S1E ) . Taken together , our observations confirm that rpoa-2 ( op259 ) is the cause of the observed apoptotic defects . The core subunits of the DNA-dependent RNA polymerases ( Table S1 ) are evolutionarily highly conserved [38] , particularly the catalytic ( largest and second-largest ) subunits [39] . Alignment of the RPOA-2 protein sequence with its homologs reveals that the P70S mutation affects a highly conserved residue at the beginning of a well-conserved stretch of amino acids both when comparing diverse orthologous eukaryotic RNA pol I β-subunits ( Fig . S2B ) as well as the three paralogous C . elegans β-subunits of RNA pol I , II , and III ( Fig . S2C ) . In the crystal structure of the RNA pol II core complex from yeast ( [40] , PDB entry 1I50 ) , the proline corresponding to the residue mutated in RPOA-2 maps to a region that lies distant to the catalytic centre of the polymerase , towards an outer surface of the core complex ( not shown ) , suggesting that the synthesis of RNA is preserved and possibly explaining why the op259 mutants , unlike the ok1970 null animals , are viable . Consistent with the expected function of RPOA-2 , we found that a YFP::RPOA-2 transgene under the control of its endogenous promoter is ubiquitously expressed , and localises mainly to the nucleolus ( Fig . S3 and Fig . S8B ) . To test for potential effects of the op259 mutation on RPOA-2 expression and localisation , we compared YFP-tagged mutant and wild-type protein . In spite of variable expression levels between different transgenic strains , YFP::RPOA-2 ( P70S ) consistently showed stronger relative cytoplasmic fluorescence than YFP::RPOA-2 ( wt ) ( Fig . S3 ) . In conclusion , we found that op259 is a viable hypomorphic mutation in the essential rpoa-2 gene and thus offers a valuable means to investigate disturbances in the process of ribosome synthesis in a living system . Many C . elegans DNA damage response mutants show not only defective apoptosis but also defects in DNA repair and cell cycle arrest of proliferating germ cells [23] , [41] . In wild-type worms , DNA damage induces cell cycle arrest of mitotic germ cells until the damage has been repaired [42] . Nuclear growth and cytoplasmic expansion meanwhile persist , resulting in a visible enlargement of the germ cell nuclei [43] . Chromatin staining of isolated gonads from control animals showed a uniform pool of nuclei in the mitotic compartment , which was interspersed with larger nuclei shortly after irradiation; at 6 hours , almost all nuclei were considerably enlarged ( Fig . 1C ) . The nuclei in rpoa-2 ( op259 ) gonads showed a similar response ( while the nucleoli grew only little ) , confirming that germ cells in the mutant do arrest following IR . To assess DNA repair in rpoa-2 ( op259 ) germ lines , we used the fluorescent reporter RAD-54::YFP expressed from the transgene opIs257 [44] . RAD-54 is involved in homologous recombination at sites of DNA double strand breaks ( DSB ) . In addition to its role in meiotic recombination , RAD-54 is also recruited to repair foci when DNA has been damaged exogenously [44] , [45] . We found that meiotic cells in rpoa-2 ( op259 ) animals had similar numbers of RAD-54::YFP foci both under control conditions ( 0 . 3±0 . 1 vs . 0 . 4±0 . 1 ( 95% CI ) foci per nucleus ) , and following IR ( 3 . 7±0 . 3 vs . 3 . 4±0 . 4 foci at 3 hours ) . In the mitotic region , rpoa-2 ( op259 ) and wild-type worms alike had very few foci under control conditions; and upon irradiation , the number strongly increased in both strains ( Fig . 1D ) . Twenty-four hours after IR , we noticed a strong association of RAD-54::YFP foci with enlarged cells ( Fig . 1D ) , i . e . , cells arrested in the cell cycle , likely due to unrepaired DNA damage . The numbers of RAD-54::YFP foci per large mitotic cell were 6 . 4±1 . 8 ( 95% CI ) for wild-type and 6 . 7±1 . 0 for rpoa-2 ( op259 ) gonads . Taken together , our observations indicate that rpoa-2 ( op259 ) mutants have no gross defect in cell cycle arrest response or DSB-repair . rpoa-2 ( op259 ) mutants develop and reproduce less rapidly than wild-type worms . A full generation cycle at 20°C took at least 24 hours longer in rpoa-2 ( op259 ) than in the wild type ( Fig . S4A ) . Most significantly delayed was the progression from the L4 stage to the appearance of the first eggs on the plate . Further , the egg-laying rate of rpoa-2 ( op259 ) animals in young adulthood was reduced in comparison to wild type ( Fig . S4B ) ; however , the total number of progeny laid per animal was less divergent ( 179±32 vs . 200±38 , n = 6 ) . rpoa-2 ( op259 ) mutants had no clear signs of oocyte retention or stacking of embryos that would explain a lower output; we thus considered that germ cell proliferation or maturation might be slowed down and assessed germ line dynamics in early adulthood ( Fig . S5A and Text S1 ) . The total number of germ cells indeed increased at a slower rate in rpoa-2 ( op259 ) mutants ( Fig . S5B ) . However , in contrast to many mutants of germ line proliferation characterised so far [e . g . ] , [ 46] , [47 , chapters in 48] , the numerical proportions of the distinct germ cell stages – mitotic , transition or mid-late pachytene cells – did not significantly deviate from the wild-type pattern; at 24 hours after onset of egg laying , wild type and rpoa-2 ( op259 ) gonads showed very similar germ cell populations ( 174 vs . 201 mitotic , 516 vs . 511 meiotic pachytene cells ) ( Fig . S5C ) . Thus , rpoa-2 ( op259 ) mutants have a reduced germ cell production rate , not so much due to an altered mitotic cell pool but most likely due to slower cell cycles . A lower rate of cells exiting late meiotic pachytene might thus contribute to the lower number of apoptotic cells in rpoa-2 ( op259 ) mutants . However , reduced proliferation alone is unlikely to explain the lower apoptosis rate and weak IR-response in rpoa-2 ( op259 ) gonads ( see further results ) . rpoa-2 ( op259 ) mutants were smaller than wild type at the L4 stage . The mutants however caught up during adulthood and exceeded the wild-type length by 10% at the third day of adulthood . Older animals also contained significantly more intestinal and interstitial lipids , indicative of metabolic alterations . Knockdown of certain translation initiation factors had been shown to considerably prolong adult lifespan in C . elegans [e . g . ] , [ 49 , 50] . In contrast , rpoa-2 ( op259 ) animals had a slightly shortened median survival ( Fig . S4C ) . Whereas rpoa-2 ( op259 ) worms were fertile at 15°C or 20°C , animals raised at 25°C were sterile . Most rpoa-2 ( op259 ) worms had not switched from spermatogenesis to oogenesis by 18 to 24 hours after L4 as wild-type worms would [51] , and more than half of all gonads revealed the presence of small cells resembling pre-diakinetic germ cells between the most proximal oocyte and the uterus ( Fig . S6A ) . At later stages , these cell clusters expanded to large masses of innumerable cells ( Fig . S6B ) . DAPI staining of these tumours revealed a chromatin pattern typical for mitotic germ cells ( Fig . S6B ) . A similar proximal proliferation ( Pro ) phenotype had previously been described by Hubbard and colleagues [52] . Interestingly , the conditions that provoked this Pro phenotype were mutations or RNAi knockdowns of genes involved in rRNA processing [53] . Formation of the proximal tumours in these conditions involves events in early germ line development ( as many germ cell proliferation phenotypes do [e . g . ] , [ 54 , 55] ) and is likely a result of spatiotemporal mismatch between early germ cells and the signalling environment [56] . Similarly , we found in temperature shift experiments that the critical developmental stage that determines tumor formation and fertility in rpoa-2 ( op259 ) mutants was the L3 larval stage ( not shown ) . We observed a further distinct germ line differentiation defect in a small fraction of rpoa-2 ( op259 ) mutant gonads: the ectopic presence of large , apparently mature oocytes in the distal arm , followed proximally by apparently normal early pachytene nuclei ( Fig . S7A ) . The penetrance of this defect , which is described in further detail in the Supplementary Results section ( Text S1 ) as Gogo phenotype , was clearly enhanced by irradiation of mutant animals at adult stage . In contrast to the Pro phenotype , which arose independently of cep-1 and ced-3 ( not shown ) , the Gogo phenotype could be suppressed by loss of cep-1 function ( Fig . S7C ) . In summary , rpoa-2 ( op259 ) mutants have a delayed germ line development and a reduced germ cell proliferation rate , and show at least two distinct germ cell maturation disorders under restrictive conditions . One of these , the proximal tumor phenotype , is shared with other mutants of ribosome synthesis . Considering that rpoa-2 codes for the second-largest subunit of RNA pol I , we wished to determine the effect of the rpoa-2 ( op259 ) mutation on ribosomal RNA synthesis . Germ cell nucleoli in C . elegans make up a significant fraction of the nuclear volume and can be readily observed by DIC microscopy . Nucleolar volume was reduced by more than 50% in rpoa-2 ( op259 ) mutant gonads ( Fig . 2B ) . Additionally , the nucleoli in rpoa-2 ( op259 ) mutants often exhibited an enlargement of a substructure visible by DIC ( Text S1 and Fig . 2A ) . Similar changes could also be observed in somatic nucleoli ( Fig . S8A ) . Eukaryotic nucleoli form around the tandem rDNA repeat units , which are transcribed by RNA pol I as one precursor rRNA ( pre-rRNA ) that is subsequently processed to generate the 18S , 5 . 8S and 28S rRNA [e . g . , 57] . The 18S rRNA together with 33 canonical ribosomal proteins ( in yeast ) forms the 40S subunit ( SSU ) ; the 26S ( 28S in mammals ) , 5 . 8S and the RNA pol III-transcribed 5S rRNAs combine with 46 ribosomal proteins to the 60S subunit ( LSU ) . Mature ribosomal RNA is thought to constitute about 60% of total cellular RNA in yeast [58] . To quantify such highly abundant RNA by qRT-PCR , we adopted the competimer strategy to the worm ( see Methods ) . We designed competimer primer sets for amplicons of the three RNA pol I transcribed rRNAs ( Fig . S11A and Table S4 ) . On average , the relative levels of 18S , 5 . 8S or 26S in rpoa-2 ( op259 ) were only about 70% of wild type ( Fig . 2C ) . For the short-lived precursors ( pre-rRNA ) , we used primers that annealed to the transcribed spacers ets , its1 or its2 , and would thus amplify the cDNAs of transcripts that had not yet been processed in the respective regions , but not the fully mature rRNAs . Overall , pre-rRNA levels were only slightly reduced in rpoa-2 ( op259 ) mutants ( Fig . 2C ) . Measurements of in vivo rRNA synthesis activity using 5-fluorouridine ( 5-FU ) incorporation also suggested that rRNA transcription is grossly normal in rpoa-2 ( op259 ) mutants ( Text S1 and Fig . S9B ) . Eukaryotic pre-rRNA is cleaved at specific sites in a well-defined sequence of exo- and endonucleolytic events [e . g . , 59] , and more than 100 nucleotides are modified in human rRNA , to yield the mature ribosomal RNAs . Numerous non-ribosomal factors and small nucleolar RNPs are required for this processing and for the assembly and nuclear export of ribosomal subunits [e . g . ] , [ 60 , 61] . Defects of certain early steps of eukaryotic rRNA processing can lead to accumulation of pre-rRNA or to processing by alternative routes , eventually resulting in a characteristic pattern of processing intermediates that can be separated by electrophoresis . For instance , pro-1 mutants in C . elegans differently process rRNA at the internal transcribed spacer its2 , similar to yeast mutants of its homolog Ipi3 [53] . We examined rpoa-2 ( op259 ) mutant worms for aberrant rRNA intermediates using digoxigenin ( DIG ) -labelled antisense RNA probes to various regions of the polycistronic pre-rRNA ( ets1 , its1 , its2; and 18S , 5 . 8S , 26S rRNAs ) ( Text S1 and Fig . S11A ) . We did not detect significant alterations in the levels of early pre-rRNA species or shorter processing intermediates that retained transcribed spacer sequences ( not shown ) . However , we noticed a distinct band between the outstanding 26S and 18S rRNA bands already when separating total RNA in denaturing agarose gels and staining with EtBr ( Fig . 2E ) . This product was more prominent in rpoa-2 ( op259 ) samples than in wild-type extracts . We found that this molecule corresponded to a truncated version of the 26S rRNA ( 26S-short ) , with a processed 3′ end without signs of polyadenylation ( Text S1 and Fig . S11D ) . We assessed the quantitative difference of the 26S-short band between wild type and various mutants from northern blots with probes complementary to the 5′ terminus . The intensity relative to the 26S rRNA was on average 1 . 4-fold higher in rpoa-2 ( op259 ) than in wild type ( Fig . 2F ) . Truncated versions of ribosomal RNA had been described in other species; in mammalian cells , truncated 28S rRNA were observed in the context of apoptosis or of viral infection . We did , however , not find a dependence of the 26S-short band on apoptosis execution ( tested with ced-3 ( lf ) ) or on the multi-exonuclease exosome ( tested with crn-3 ( lf ) ) ( Fig . 2F; for details and references , see Text S1 ) . We further looked at the two rRNA processing mutants pro-2 ( na27 ) ( Noc2 in yeast ) and pro-3 ( ar226 ) ( Sda1 ) , which share the Pro phenotype with rpoa-2 ( op259 ) . Similarly to rpoa-2 ( op259 ) , the 26S-short to 26S rRNA ratio was about 1 . 4-fold that of wild type for both mutants . The accumulation might thus result from defects in certain steps of rRNA processing , and it supports an effect of rpoa-2 ( op259 ) on ribosome synthesis beyond transcription of ribosomal RNA . Quantitative or qualitative changes in transcription and processing of ribosomal RNA likely alter the composition and the activity of ribosomes , and therefore might influence translation . Interfering with translation regulation mechanisms often leads to alterations in protein expression that are confined to a subset of factors rather than being global [e . g . ] , [ 62 , 63] . Expression of some genes or groups of genes might thus be differentially affected in rpoa-2 ( op259 ) – even with a background of largely normal translation . To determine the effect of the rpoa-2 ( op259 ) mutation on the proteome level , we performed mass spectrometric analysis , applying a label-free quantitation technique on whole worm protein extracts . We could identify a total of 342 proteins in wild-type samples and 343 in rpoa-2 ( op259 ) ; of these , 328 proteins were present in both samples and could be compared quantitatively . Seventy proteins were ribosomal; thus , most of the total 80 ribosomal proteins were represented . Strikingly , in the mutant , most ribosomal proteins were reduced to about 50–70% of wild type when normalised by their fraction of the total in each sample . The average of the normalised ratios between rpoa-2 ( op259 ) and wild type of individual ribosomal proteins was 0 . 66 ( 0 . 69 for SSU and 0 . 64 for LSU ) ( Fig . 2D ) . Collectively , ribosomal proteins formed 21 . 8% of total protein in wild type , whereas in rpoa-2 ( op259 ) they represented a mere 14 . 2% . Altogether , we have found that rpoa-2 ( op259 ) mutants have no significant decrease of ribosomal RNA transcription , but a clear reduction of mature rRNA levels and a similar reduction in the abundance of ribosomal proteins . Cells in rpoa-2 ( op259 ) animals might therefore have a smaller pool of mature ribosomal subunits and thus a somewhat reduced protein synthesis capacity . In addition , we have characterised an unconventional RNA of relatively high abundance , which has the nucleotide sequence of a 3′-terminally truncated , non-polyadenylated 26S rRNA , and which has higher relative abundance in rpoa-2 ( op259 ) and in mutants of rRNA processing . Based on the data above , we considered three alternatives how the rpoa-2 ( op259 ) mutation might lead to disturbed cell death . First , the apoptotic defects in rpoa-2 ( op259 ) animals might be a result of globally reduced translation; DNA damage-induced apoptosis is induced through CEP-1-dependent transcriptional upregulation of the BH3 domain protein EGL-1 [30] , [64] and therefore clearly requires new protein synthesis . Second , alterations in ribosome biogenesis and the nucleolus might affect germ cell apoptosis through mechanisms that do not necessarily involve protein synthesis . Third , the role of RPOA-2 in apoptosis regulation might be an independent , specific characteristic of this second largest polymerase subunit or even of the mutated site , and might not be directly linked with the overall performance of RNA pol I as a transcription apparatus . To distinguish between these possibilities , we tested the effect of knocking down the expression of other RNA pol I core subunits or associated factors , as well as of other factors involved in ribosome biogenesis or protein synthesis ( Table S2 and Table S3 ) . We chose the L3/L4 stage for initiation of RNAi by feeding to minimize detrimental effects on germ line development . Knockdown of RNA polymerase subunits frequently resulted in fertility and growth defects , but did , with the exception of F14B4 . 3 ( rpoa-2 ) and F36A4 . 7 ( ama-1 ) not lead to defective DNA damage-induced apoptosis ( Table S2 ) . A possible explanation for the apoptosis defect in ama-1 ( which codes for the largest RNA pol II subunit ) is that the reduction of mRNA transcription might become rate-limiting for the rapid transcriptional upregulation of pro-apoptotic BH3-only factors that is required for DNA-damage induced apoptosis or critical for the expression levels of other pro-apoptotic factors . rRNA processing- and ribosomal assembly factors are critical for the maturation of the small 40S and the large 60S ribosomal subunits . There is a high number of factors involved specifically in the synthesis of either of the two subunits as well as factors needed for both [61] . Interestingly , several worm mutants of rRNA processing factors had been isolated from one genetic screen , based on their conditional proximal proliferation ( Pro ) phenotype described above [52] , [53] . We thus looked for apoptotic defects in the viable Pro mutants pro-2 ( na27 ) and pro-3 ( ar226 ) . pro-2 ( na27 ) animals indeed failed to exhibit increased apoptosis upon IR at the permissive 20°C , similarly to rpoa-2 ( op259 ) ( Fig . 1E ) . This strengthens the notion that rpoa-2 ( op259 ) and pro-2 ( na27 ) have very similar phenotypes and that they likely result in similar molecular defects . We addressed the question of whether failure in properly processing rRNA might more generally lead to changes in apoptosis . RNAi knockdown of nucleolar proteins , of rRNA processing factors and of representative ribosomal proteins indeed frequently blocked IR-induced apoptosis . However , knockdown of several of these factors entailed strong germ line defects that became visible soon after apoptosis scoring , so that an interpretation is difficult in these cases ( Text S1 and Table S3 ) . Nevertheless , these observations support the notion that proper rRNA processing and ribosome synthesis are critical for IR-induced apoptosis of germ cells . Finally , we looked at the effect of interfering with translation . eIF4E has been recognised as a central factor in translation initiation and is frequently altered in various proliferative diseases [65] . C . elegans has five homologs of eIF4E , coded for by the genes ife-1 to ife-5 [66] . The five isoforms show specificity for translation of distinct sets of mRNAs [e . g . ] , [ 67 , 68] . We tested the major isoforms for somatic translation , ife-2 , and for germ line translation , ife-1 . The ife-1 loss-of-function alleles ok1978 ( Fig . 1F ) and bn127 ( data not shown ) reduced apoptosis in response to IR , whereas ife-2 ( lf ) animals were not obviously defective ( Fig . 1F ) . We also used the bacterial toxin cycloheximide ( CHX ) to pharmacologically block translation ( see Text S1 ) . At 500 µg/ml , most of the IR response was abolished , whereas baseline apoptosis remained unaffected ( Fig . S12A ) . However , this dose also led to massive impairment of animal health and growth . Lower doses that only marginally impaired development had little to no effect on apoptosis ( see Text S1 ) . In summary , we found that interfering with different steps of ribosome synthesis can lead to inhibition of germ cell apoptosis . Chemically interfering with protein translation also inhibits germ cell apoptosis , but only under conditions that are so drastic that they result in widespread defects . By contrast , rpoa-2 ( op259 ) animals appear largely normal and healthy and with a largely normal proteome profile ( excepting ribosomal proteins ) , suggesting that the rpoa-2 ( op259 ) mutation is able to reduce apoptotic irradiation response without drastically hampering global translation . Even though we can herewith not exclude that the apoptotic phenotype is due to rate-limiting factors of apoptosis pathways whose expression is critically sensitive to particular alterations of the translation apparatus , the disturbed IR response is likely a more direct effect of abnormal ribosome biogenesis . The data above suggest that rpoa-2 ( op259 ) might selectively interfere with DNA damage-induced apoptotic signalling . CEP-1 – a functional homolog of mammalian p53 family proteins [30] , [69] – is a key player in this pathway , mediating apoptosis mainly by transcriptionally activating the two pro-apoptotic BH3-only proteins EGL-1 and CED-13 [30] , [64] . We evaluated the possibility that insufficient activation of CEP-1 is responsible for the weak IR response in rpoa-2 ( op259 ) by measuring transcript levels of EGL-1 and CED-13 by qRT-PCR . To our surprise , we found that both EGL-1 and CED-13 were efficiently induced in irradiated rpoa-2 ( op259 ) mutants ( Fig . 3A ) . Moreover , the levels in non-irradiated rpoa-2 ( op259 ) control animals were much higher than in the wild type . Increased basal levels of EGL-1 and CED-13 were CEP-1-dependent , as they were abrogated in cep-1 ( gk138 ) rpoa-2 ( op259 ) double mutants ( Fig . 3A ) . These observations suggest that rpoa-2 ( op259 ) mutant animals suffer from a chronic activation of CEP-1/p53 , while at the same time being surprisingly resistant to this activation . Consistent with the hypothesis that rpoa-2 ( op259 ) mutants are resistant to CEP-1-dependent apoptosis , we found that rpoa-2 ( op259 ) could suppress the strongly increased baseline levels of germ cell apoptosis in the dsDNA break repair mutant rad-51 ( lg8701 ) ( Fig . S13C ) as well as the hypersensitivity to irradiation-induced apoptosis in in the Abl kinase mutant abl-1 ( ok171 ) ( Fig . S13B ) , both of which involve CEP-1 [70] . Intriguingly , cep-1 ( gk138 ) rpoa-2 ( op259 ) double mutants had extremely few corpses , far less than either cep-1 ( gk138 ) or rpoa-2 ( op259 ) alone . We used the engulfment defective background ced-6 ( n1813 ) for better numerical ‘resolution’ of germ cell corpses around the low baseline levels and could confirm these findings ( Fig . 3B and Fig . 3C ) . Thus , loss of cep-1 function not only suppresses the remaining moderate levels of IR-induced death in rpoa-2 ( op259 ) gonads , but also eliminates baseline cell death . This result is surprising in so far that most of the basal ‘physiological’ apoptosis observed in adult gonads normally is CEP-1-independent [30] . A hypothesis consistent with all the above information would be that rpoa-2 ( op259 ) mutants show a reduced sensitivity not only to DNA damage-induced apoptosis , but also to physiological germ cell death cues . The basal apoptosis observed in non-irradiated rpoa-2 ( op259 ) could then be explained by the increased CEP-1 activity observed in these mutants . Considering this broader defect in germ cell apoptosis downstream of CEP-1-induced EGL-1 transcription , the rpoa-2 ( op259 ) mutation ought to influence cell death at the level or downstream of the core apoptotic machinery . In experiments that we describe in detail in the Supplementary Results ( see Text S1 ) we observed that rpoa-2 ( op259 ) had no influence on the transcript levels of the core factors CED-4 and CED-3 and no obvious effect on the CED-4 expression pattern in pachytene stage germ cells ( Fig . S14C ) . By contrast , rpoa-2 ( op259 ) showed a strong genetic interaction with ced-9 ( RNAi ) or with the hypomorphic ced-9 ( n1653 ) mutation ( Fig . S15C and Fig . S15A ) . The combination rpoa-2 ( op259 ) ; ced-9 ( n1653 ) led to highly excessive apoptosis , which could be suppressed by egl-1 ( lf ) ( Fig . S15B ) . These findings suggest that rpoa-2 ( op259 ) impacts the apoptotic machinery at the level of CED-9 . MAPK signalling pathways are central regulators of cell proliferation and growth . In the C . elegans gonad , Ras/MAPK is required for ‘physiological’ germ cell death and has recently also been shown to modulate DNA damage-induced apoptosis [71] . In addition to the lack of physiological apoptosis , rpoa-2 ( op259 ) mutants showed a shift of the appearance of the first large oocyte to the proximal arm of the gonad , another germ line feature reminiscent of reduced Ras/MAP kinase activity . This suggested that the loss of CEP-1-independent baseline germ cell death in rpoa-2 ( op259 ) could be due to reduced Ras/MAPK pathway activity . Hypothetically , such a reduction might also contribute to the weak irradiation response in rpoa-2 ( op259 ) . To measure the levels of activated MPK-1 in situ , we performed immunofluorescence analysis on dissected gonads . MAPK-YT , an antibody to di-phosphorylated ERK with cross-reactivity to activated MPK-1 [72] detects high ppMPK-1 levels in the most proximal oocytes and distinct accumulation in the region of late pachytene cells , whereas the distal gonad is generally devoid of signal [73] . We co-stained dissected gonads for ppMPK-1; for total MPK-1; and for dsDNA . For wild-type worms , we observed the signal pattern described above ( Fig . 4A ) . By contrast , in rpoa-2 ( op259 ) gonads , fluorescence intensity of activated MPK-1 was clearly reduced in the late pachytene region , where germ cells start maturing into oocytes or become apoptotic ( Fig . 4A and Fig . 4B ) . These findings support that the low levels of germ cell apoptosis in rpoa-2 ( op259 ) mutants could be due to reduced Ras/MAPK activity . The EGFR/Ras/ERK signalling cascade has been studied in much detail in the context of vulval development in C . elegans [e . g . , 74] . Briefly ( Fig . 5B ) , binding of the LIN-3 ( EGF ) signal to LET-23 ( EGFR ) and activation of the receptor tyrosine kinase activity leads to activation of the Ras GTPase homolog LET-60 , and via a kinase cascade , to phosphorylation of the ERK homolog MPK-1 . The phosphatase LIP-1 antagonises MAPK activity through dephosphorylation of MPK-1 . LIP-1 is also an important regulator of germ cell proliferation [75] , oocyte maturation [76] , and germ cell apoptosis [77] . We genetically tested whether an increase in Ras/MAPK pathway activity would compensate for the abolished ‘physiological’ cell death in rpoa-2 ( op259 ) and what the effect would be on irradiation-induced germ cell apoptosis . We generated double mutants between rpoa-2 ( op259 ) and the let-60 gain-of-function alleles let-60 ( n1046gf ) and let-60 ( ga89 ) , which both lead to constitutive moderate activation of Ras GTPase activity [77] . In both strains , germ line anatomy was largely normal at 20°C; yet , the cell corpse number upon irradiation rose significantly higher than in wild type ( Fig . 5A ) . Thus , both let-60 ( gf ) alleles can at least partially compensate for the apoptotic defect of rpoa-2 ( op259 ) . The germ lines of lip-1 ( zh15 ) loss-of-function mutants had slightly increased baseline apoptosis and were strongly hypersensitive to IR ( Fig . 5A ) , consistent with a recent report [71] . rpoa-2 ( op259 ) ; lip-1 ( zh15 ) double mutant worms also had higher than wild-type levels of germ cell death at baseline and were equally hypersensitive to IR as lip-1 ( zh15 ) animals , indicating that lip-1 ( zh15 ) is fully epistatic to rpoa-2 ( op259 ) in terms of IR-induced germ cell apoptosis . To evaluate the possibility that the rpoa-2 ( op259 ) mutation affected LIP-1 , we performed anti-LIP-1 immunostaining [75] on extruded gonads of adult animals ( see Text S1 ) : the wild-type pattern of membrane-association puncta was lost in the late pachytene region of rpoa-2 ( op259 ) , leaving a relatively higher cytoplasmic signal ( Fig . S17B ) . We also looked at the somatic expression pattern of a Plip-1::GFP transcriptional reporter [78] in late larval stages , and found a strongly increased signal in hypodermal cells ( Fig . S17A ) . The genetic and expression data support that rpoa-2 ( op259 ) might reduce MAPK activity by activating the phosphatase LIP-1 . The observation of antagonistic effects of rpoa-2 ( op259 ) and let-60 ( gf ) on germ cell apoptosis levels would fit a model where rpoa-2 ( op259 ) stimulates phosphatase activity and counteracts phosphorylation of MAPK resulting from constitutive MAPK-kinase activation by let-60 ( gf ) . We and others previously described mutations that lead to increased baseline levels of germ cell death [79] . gla-3 mutants , which show high levels of CEP-1-independent apoptosis , have increased Ras/MAPK activity both in vulval development and in the germ line [80] . We tested whether reduced gla-3 would also render germ cells hypersensitive to IR . Due to very close linkage of gla-3 , cep-1 , and rpoa-2 , we used gla-3 ( RNAi ) instead of a mutation . Germ cell death was increased to about 25 corpses per gonad in gla-3 ( RNAi ) treated worms; upon irradiation , this number rose massively , to 50 corpses on average ( Fig . S16A ) . gla-3 ( RNAi ) therefore does not only increase baseline apoptosis , but it potently enhances irradiation response . In rpoa-2 ( op259 ) mutants , gla-3 ( RNAi ) increased the corpse number to approximately wild-type levels . Moreover , gla-3 ( RNAi ) significantly restored the IR response in rpoa-2 ( op259 ) mutants ( Fig . S16A ) , similarly to let-60 ( gf ) . We also tested whether lip-1 ( zh15 ) , let-60 ( n1046gf ) or gla-3 ( RNAi ) would restore CEP-1-independent – i . e . ‘physiological’ – germ cell apoptosis in rpoa-2 ( op259 ) . Indeed , all these conditions re-established baseline apoptosis in cep-1 ( gk138 ) rpoa-2 ( op259 ) at levels that were similar to cep-1 ( gk138 ) ( Fig . S16B ) . Remarkably , the conditions leading to activation of Ras/MAPK increased the cell corpse number in rpoa-2 ( op259 ) without a significant effect on the slow ( germ line ) development of this mutant . Taken together , these observations suggest that rpoa-2 ( op259 ) animals suffer from reduced Ras/MAPK pathway activity in the germ line , which can account for both the reduced physiological germ cell death and the reduced sensitivity to DNA damage-induced apoptosis in this mutant . To determine whether the negative effect of rpoa-2 ( op259 ) on MAPK signalling extends beyond the germ line , we looked at vulval development , where MPK-1 controls the cell fates of vulval precursor cells . let-60 ( gf ) hyperactivates MPK-1 and leads to extra inductions of vulval precursor cells and to multiple vulvae ( Muv phenotype ) in the adult animal [81] . The excessive vulval inductions caused by let-60 ( gf ) were almost completely suppressed in rpoa-2 ( op259 ) ; let-60 ( n1046gf ) double mutants ( Fig . 5D ) . To exclude the possibility that rpoa-2 ( op259 ) acts downstream of mpk-1 in this model , we tested epistasis with the MPK-1 target LIN-1 . In its non-phosphorylated state , LIN-1 acts as a negative regulator of genes that regulate vulva formation [82]; it is inhibited through phosphorylation by MPK-1 . Loss of lin-1 function thus mimics Ras/MAPK overactivity downstream of mpk-1 . We found that all rpoa-2 ( op259 ) ; lin-1 ( n304 ) animals had the maximal number of four extra vulvae , like lin-1 ( n304 ) single mutants ( 50 out of 50 animals in both strains ) , indicating that the effect of the rpoa-2 ( op259 ) mutation is at the level or upstream of mpk-1 . Finally , we tested mutants of MPK-1 and two other MAP kinase pathways in C . elegans , PMK-1/p38 and JNK-1/Jnk , for apoptotic IR response . The mpk-1 ( ga111 ) allele is temperature-sensitive; at 25°C , many gonads show the mpk-1 loss-of-function pachytene-exit defect and brood size is strongly reduced , while at 20°C , germ cell progression into oocytes still occurs and viable progeny are produced . Even at this permissive temperature , mpk-1 ( ga111 ) animals had lower than wild-type levels of ‘physiological’ germ cell death and showed a reduced response to IR ( Fig . 5C ) . We also observed an apoptotic defect with RNAi knockdown of mpk-1 ( not shown ) . These results confirm the importance of MPK-1 in DNA damage-induced germ cell death . The MAP kinase p38 family homolog PMK-1 is involved in the response to danger signals , such as infectious agents or excess transition metals [83] , [84]; lack of functional PMK-1 sensitizes C . elegans to toxic effects of pathogens [85] . We tested whether this pathway might also play a role in irradiation-induced apoptosis . Indeed , pmk-1 ( km25 ) animals had reduced numbers of apoptotic germ cell corpses at baseline or following irradiation ( Fig . 5C ) . In contrast , animals with the jnk-1 ( gk7 ) loss-of-function mutation had normal , if not slightly increased levels of germ cell corpses following IR ( Fig . 5C ) . As described above , MAPK signalling has a strong impact on irradiation-induced apoptosis . mpk-1 ( rf ) and rpoa-2 ( op259 ) mutants had only weak IR response; conversely , elevating MAPK activation had a potentiating effect on apoptosis levels , most significantly following IR . We wondered whether the role of MAPK pathway activity was solely to facilitate apoptosis , independently of the additional pro-apoptotic cues , or whether the pathway might actually be activated by irradiation responses , and examined the effect of IR three hours after treatment . The ppMPK-1 signal in the late pachytene region was indeed stronger in IR treated wild-type worms than in non-irradiated controls ( Fig . 4A ) , as also recently reported [71] . By contrast , rpoa-2 ( op259 ) mutants did not significantly respond to irradiation , and ppMPK-1 signal intensity remained lower than in non-irradiated wild-type worms ( Fig . 4B ) . Taken together , our observations support the hypothesis that rpoa-2 ( op259 ) animals have low activation of Ras/MAPK , which likely is responsible for the reduced germ cell apoptosis observed in these animals . Ras/MAPK activity is a prerequisite for germ cell death and seems to sensitise cells for further pro-apoptotic signals . Possibly , the early increase in Ras/MAPK pathway activity following IR additionally boosts germ cell death in response to this genotoxic and cytotoxic treatment . The rpoa-2 ( op259 ) mutation affects a highly conserved residue in one of the two main catalytic subunits of the RNA polymerase . With various assays , we detected no strong reduction of early pre-rRNA transcripts , but a significant drop of mature ribosomal RNAs , and an according reduction of ribosomal proteins to approximately 70% of wild type . Thus , rpoa-2 ( op259 ) mutant cells likely have a reduced pool of mature ribosomal subunits , either as a cumulative result of steadily subnormal rRNA transcription , or due to a post-transcriptional effect of rpoa-2 ( op259 ) on rRNA maturation and ribosome assembly . Supporting the latter possibility , transcription of rRNA has been shown to be interdependent and co-regulated with rRNA processing and pre-ribosome assembly , especially of the SSU [87] , [88] . Intriguingly , a mutation in yeast Rpa135 , the RPOA-2 homolog , also shows clear evidence for such a link [89] . Northern blot analyses of rpoa-2 ( op259 ) worm RNA extracts revealed an increase of an RNA band which we were able to characterise as a truncated , non-polyadenylated version of the 26S rRNA . Increased 26S-short levels were also present in the rRNA processing mutant pro-2 ( na27 ) , which shares various phenotypic aspects with rpoa-2 ( op259 ) , as well as in mutants of wdr-46 ( UTP7 in yeast , involved in small subunit assembly ) [90] . Even though the relevance of this molecule remains speculative ( see Text S1 ) , its relative increase in rpoa-2 ( op259 ) indicates altered rRNA processing or turnover and thus an effect of the rpoa-2 ( op259 ) mutation beyond transcription . Given the localisation of the P70S substitution on the surface of the RPOA-2 protein , it is conceivable that this mutation disrupts the interaction between the RNA pol I core complex and associated proteins required for proper assembly or disassembly of processing factors , e . g . , of the SSU processosome . rpoa-2 ( op259 ) mutants show increased CEP-1/p53 activation in the absence of exogenous DNA damage . Studies over the last two decades clearly place the nucleolus and ribosome biogenesis at the crossroads of cellular metabolism , cell cycle regulation , growth control , cellular stress responses , aging , and cell death [e . g . , 91] . Nucleolar disruption has been recognised as a major hallmark of cellular stress; it can result from reduced rRNA synthesis [e . g . , 18] and is possibly even recruited as a mechanistical step towards cellular fate in response to stress like DNA damage [2] , [15] . A key function is attributed to the regulation of p53 by nucleolar integrity or nucleolar disintegration . Mdm2 , the major ubiquitin ligase and negative regulator of p53 , is controlled by the nucleolar tumor suppressor protein ARF ( p19ARF ) . ARF is released upon nucleolar breakdown , eventually leading to p53 stabilisation [e . g . , 4] . The Mdm2–p53 interaction is also target of ribosomal proteins , e . g . , RPL11 or RPL26 , that are released from the nucleolus following different types of cellular stress [92] . rpoa-2 ( op259 ) mutant germ cells have smaller than wild-type nucleoli and enlarged nucleolar substructures , indicating altered nucleolar physiology . Unfortunately , the proposed molecular link between nucleolar integrity and p53 stability cannot be easily transferred from mammals to the worm . Whereas p53 is functionally – and less tightly by sequence – conserved in CEP-1 [30] , [31] , no homolog of Mdm2 has been found in C . elegans ( it is conceivable that sequence conservation is very low and therefore the homolog has not been identified , or that alternative ubiquitin ligases regulate CEP-1 stability ) , and there is no obvious C . elegans homolog of ARF . Nevertheless , loss of nucleolar proteins was found to increase resistance to pathogenic bacteria , via a mechanism that involves CEP-1 [93] . Thus , whether by the yet to be identified homologs of ARF and Mdm2 or by alternative mechanisms , it is well possible that the nucleolus is involved in stress responses in C . elegans and that CEP-1/p53 activation in rpoa-2 ( op259 ) is a result of disturbed nucleolar function . It remains to be determined whether the CEP-1::GFP-positive substructures that we observed in germ cell nucleoli [see Text S1; similar structures have also been reported in mammalian cells] are relevant for CEP-1 regulation . Contrasting with the expected pro-apoptotic effect of increased CEP-1 signalling , germ cell apoptosis levels were not increased in rpoa-2 ( op259 ) . We show that there is an additional effect of rpoa-2 ( op259 ) in this in vivo system – reduced Ras/MAPK activity – that eventually blocks the pro-apoptotic drive from increased CEP-1/p53 activity , leading to lower than wild-type levels of germ cell death both under normal growth conditions and following DNA damage ( Fig . 6A ) . How might disturbances in ribosome production lead to reduced Ras/MAPK signalling ? rpoa-2 ( op259 ) counteracts Ras/MAPK activation not only in the germ line , but also e . g . , during vulval development . It is possible that disturbed ribosome production , beyond the specific nucleolar signalling described above , leads to a stress response , which might include a reduction in Ras/MAPK signalling . Consistent with this hypothesis , a recent study demonstrated upregulation of pathogen- and xenobiotic-associated defences in worms that had been treated with drugs or RNAi to reduce ribosome synthesis and other essential cellular processes [94] . That study and work in many species have shown that cellular stress leads to changes in the activity of MAPK superfamily members and in their cross-talks [e . g . ] , [ 95 , 96] , including possibly a reduction in MPK-1/ERK activity . In our quantitative proteome comparison of rpoa-2 ( op259 ) and wild type adults , we have found a tendency of the group of proteins annotated with ‘stress response’ towards higher relative abundance in the mutant ( Fig . S10 ) . Additionally , we have tested four reporters of stress response genes in the rpoa-2 ( op259 ) mutant background . Three of them showed increased expression in rpoa-2 ( op259 ) ( Fig . S18 ) . Remarkably , the formerly mentioned study found that upregulation of defence genes was raised globally even if only specific tissues like intestine or hypodermis were targeted with RNAi [94] . Such cross-tissue effects are likely to be in play in rpoa-2 ( op259 ) as well; altered MAPK activity and germ cell apoptosis need not be a germ line autonomous effect of rpoa-2 ( op259 ) . Indeed , we found that tissue specific knockdown of rpoa-2 specifically in the soma reduced germ cell apoptosis as much as knockdown specifically in the germ line ( Fig . S1C ) , suggesting a more global mechanism for the transmission from altered ribosome synthesis to cell death regulation . It is also conceivable that impaired ribosome synthesis signals starvation-like conditions . Nutritional signalling – particularly the insulin/IGF pathway – is critical for the soma-germ line interaction during germ line development and also for germ cell apoptosis [86] , [97] , [98] . Additionally , starvation of larval stage worms was shown to counteract the effects of let-60 ( gf ) in vulval development [99] . rpoa-2 ( op259 ) mutants have slower larval growth than wild-type worms and show increased lipid accumulation in adults , indicating significant metabolic alterations . The rpoa-2 ( op259 ) mutation leads to reduced MPK-1 activity in the gonad and antagonises the effects of LET-60/Ras overactivity on IR-induced germ cell death and on cell fate determination in somatic development . Intriguingly , a very similar phenotypic pattern has recently been reported for the insulin signalling mutants daf-2 and pdk-1 [86] . This congruence suggests that an alteration in insulin signalling by rpoa-2 ( op259 ) , issued in the germ line and/or the soma , could be the link to reduced MAPK activity . Conversely , since mammalian target of rapamycin ( mTOR ) coordinates ribosome synthesis in response to metabolic state [12] , insulin pathway mutants could lead to altered ribosome production and thereby affect MAPK activity and apoptosis . Our genetic analyses of rpoa-2 ( op259 ) confirm the recently discovered functional relevance for the Ras/MAPK pathway in regulating IR-induced apoptosis [71] and additionally indicate that Ras/MAPK activity modulates germ cell death at a genetic level downstream of CEP-1/p53 , likely at the level of CED-9/Bcl-2 . Several other mutants beyond rpoa-2 ( op259 ) are known to show a reduced sensitivity to IR despite functional CEP-1 activation: e . g . , the pRb homolog lin-35 [26] , the Sirtuin homolog sir-2 . 1 [100] , the ceramide synthesis mutant lagr-1 [101] , or the ubiquitin ligase ARF-BP1 homolog eel-1 [102] . Whether these mutations also lead to reduced MPK-1 signalling remains to be determined . Supporting the model that MAP kinases play a role in stress-induced germ cell apoptosis in parallel to , or downstream of , CEP-1/p53 activation , apoptotic response to excessive arsenite [103] or copper [84] were found to depend on various MAP kinase ( ERK/JNK/p38 ) cascades but not on CEP-1 . MAPK pathways are also required for germ cell apoptosis induced by osmotic , oxidative , or heat shock stress [104] . Finally , pathogen-driven germ cell death requires the p38 MAP kinase PMK-1 but not CEP-1 [85] . Altogether , these findings support a model in which CEP-1 and MPK-1 collaboratively establish the level of IR-induced germ cell apoptosis ( Fig . 6B ) . Ras/MAPK activity , beyond facilitating ‘physiological’ germ cell death , is critical for how sensitive cells are towards further pro-apoptotic signals . Thus , MAP kinases might act as “master modulators” of germ cell apoptosis in C . elegans . Our analysis of rpoa-2 ( op259 ) mutants demonstrates that a single point mutation affecting a basic cellular process can readily influence multiple key signalling networks , and that the combinatorial effect of these disturbances can lead to quite complex phenotypes . The rpoa-2 ( op259 ) mutation on the one hand leads to CEP-1/p53 activation , on the other hand reduces MAPK pathway activity ( Fig . 6A ) . Both effects influence germ cell apoptosis , but in opposite directions; reduced Ras/MAPK activity in rpoa-2 ( op259 ) mutants abolishes ‘physiological’ germ cell death and significantly impairs CEP-1-induced apoptosis; this explains the weak IR-response as well as the reduced baseline germ cell death levels in rpoa-2 ( op259 ) . At the same time , we find activation of cytoplasmic stress response factors; we can guess from the literature but have no direct evidence from our studies that the activation of these specific stress response pathways leads to the observed reduced levels of activated MAPK . Nevertheless , our observations support the notion that moderate disturbance of an elemental process can lead to very specific changes in signalling rather than an amalgam of generally disrupted cellular functioning . A similar conclusion was recently reached by Melo and Ruvkun , who showed that interfering with basic cellular processes can lead to the specific activation of pathogen-associated response pathways in C . elegans [94] . The altered signalling becomes particularly evident when tissues are additionally challenged through exogenous stimuli . Conditional germ line phenotypes in rpoa-2 ( op259 ) and in other mutants of basic cellular processes are indicative of the delicate new balance in these systems regarding cell proliferation , differentiation , and death . As the signalling pathways and cellular processes described in this work are all highly conserved through evolution , similar mechanisms might also operate in humans . This has particularly implications for the field of cancer genomics , as it suggests that mutations in genes participating in core cellular processes might , through their modulatory effects on cancer signalling pathways , significantly contribute to the development of at least some types of cancer , and influence how these respond to therapeutic treatment . Worms were cultured at 20°C ( unless indicated otherwise ) on NGM agar plates seeded with OP50 strain E . coli , according to standard procedures . Synchronisation was reached by bleaching gravid adults . The moment when the first few eggs had been laid by a synchronous worm population was defined as the 0 hour reference time point for adulthood ( 12 hours post L4 stage in wild-type worms at standard conditions ) . Genotypes used in this study ( information available on www . wormbase . org ) : N2 wild type , rpoa-2 ( op259 ) ( this study ) ; rpoa-2 ( ok1970 ) /hT2 , pro-2 ( na27 ) , pro-3 ( ar226 ) , ife-1 ( ok1978 ) , ife-1 ( bn127 ) , ife-2 ( ok306 ) , ncl-1 ( e1865 ) , hus-1 ( op241 ) , cep-1 ( gk138 ) , ced-6 ( n1813 ) , ced-3 ( n717 ) , rad-51 ( lg8701 ) , abl-1 ( ok171 ) , let-60 ( n1046 ) , let-60 ( ga89 ) , lip-1 ( zh15 ) , gap-1 ( ga133 ) , lin-1 ( n304 ) , mpk-1 ( ga111 ) , pmk-1 ( km25 ) , jnk-1 ( gk7 ) , crn-3 ( ok2269 ) , rrf-1 ( pk1417 ) , mut-7 ( pk204 ) , unc-119 ( ed3 ) . opIs257[Prad-54::rad-54::yfp; unc-119 ( + ) ] , opIs219[Pced-4::ced-4::gfp; unc-119 ( + ) ] , opIs198[Pcep-1::cep-1::gfp; unc-119 ( + ) ] , zhIs4[Plip-1::gfp; unc-119 ( + ) ] , dvIs19[Pgst-4::GFP::NLS] , frIs7[Pnlp-29::GFP + Pcol-12::DsRed] , muIs84[Psod-3::GFP + rol-6] , zcIs13[Phsp-6::GFP] , opIs110[Plim-7::act-5::yfp; unc-119 ( + ) ]; opIs372[Prpoa-2::yfp::rpoa-2 ( + ) ; unc-119 ( + ) ] , opIs375[Prpoa-2::yfp::rpoa-2 ( op259 ) ; unc-119 ( + ) ] , opIs413[Phus-1::yfp::rpoa-2 ( op259 ) ; unc-119 ( + ) ] , opEx1431[Phus-1::yfp::rpoa-2 ( + ) ; unc-119 ( + ) ] , opEx1416[Prpoa-2rpoa-2 ( + ) ; unc-119 ( + ) ] . For the [rpoa-2] transgenes , the genomic sequence of rpoa-2 was fused with optionally an YFP tag and the endogenous promoter- and 3′UTR sequences in the Lazyboy cloning system ( primer sequences , see Table S4 ) . Plasmid vectors were introduced into the C . elegans germ line by ballistic transformation [105] of rpoa-2 ( op259 ) ; unc-119 ( ed3 ) animals . The Pro phenotype and temperature sensitive sterility at 25°C could be rescued with [rpoa-2 ( + ) ] transgenes ( Fig . S6C ) ; no rescue was attained , though , with those transgenes that did not show germ line expression . Gene expression knockdown was performed with RNAi by feeding as described [35] , [36] . Briefly , bacteria expressing double-stranded RNA ( plasmids were sequenced to confirm the correct target genes ) were seeded on NGM agarose plates supplemented with ampicillin and 3 mM IPTG for efficient induction . Worms were synchronised by bleaching; depending on the stage when RNAi had to be initiated , freshly hatched L1 larvae were transferred to the RNAi plates directly , or they were grown on OP50 seeded plates until reaching the appropriate stage and then washed and transferred . As controls , an empty vector RNAi clone and an unc-22 clone were always included . The forward genetic screen was performed as a F1 clonal screen . Because the screen was fairly laborious , we limited our search to 2000 haploid genomes , which is significantly below saturation level . Adult wild-type worms were mutagenized using the chemical alkylating agent ethyl methane sulfonate ( EMS , 1 . 25 mM ) , 1000 F1 progeny were transferred singly onto new plates , and their offspring were exposed to 60 Gy of X-rays at young adulthood; multiple of these F2 animals were examined for DNA damage response defects using DIC microscopy . One candidate mutation of this screen , op259 , was mapped with a combination of three-factor and two-factor mapping techniques to the right arm of chromosome I . Fine mapping and gene sequencing in the candidate region revealed a C→T transition in the second exon of F14B4 . 3 . The mutant allele was 6× backcrossed into the wild-type genetic background prior to further characterisation . Apoptotic germ cell corpses – visible as refractile discs – were scored using Nomarski optics ( DIC ) . Irradiation treatments were performed at the reference time point of adulthood as described above . The number of apoptotic corpses in the late meiotic pachytene region of one of the gonads ( anterior or posterior ) was counted in 20 worms per condition , either in a time course , or selectively at 24 hours post treatment , when the irradiation-induced corpse number was tending towards a plateau and the germ lines were still without strong signs of aging . For ionizing irradiation ( IR ) , well-fed worms on agar plates were exposed to X-rays in an Isovolt irradiation device for 18 . 5 min , corresponding to a dose of 60 Gy; for UV-irradiation , worms were treated with short pulses of UV-C in a Stratalinker ( 254 nm ) to reach the indicated energy doses . Gonads were extruded from adult worms and stained with Hoechst at the indicated time points after irradiation treatment , and images were acquired immediately; the fluorescent images depict slightly more than a simple axial cross-section through the gonads , as these were flattened by the preparation . The total number of mitotic cells and the fraction of enlarged cells were counted according to the chromatin-staining pattern . RAD-54::YFP foci were counted using the ImageJ [106] local intensity peak detection tool with a defined threshold . The six vulval precursor cells ( VPCs ) are instructed to adopt a certain fate mainly through signalling from the anchor cell ( EGFR/Ras/MAPK ) and by lateral crosstalk ( Notch ) . Aberrant vulval cell inductions can result in extra vulvae ( Muv ) or a missing vulva ( Vul ) in adult animals . Cellular inductions can be accurately assessed with DIC microscopy of L4 larvae [81] . The vulval induction index represents the average number of VPCs that have adopted a 1° or 2° vulval fate ( in wild type , VI index is 3 . 0 ) . Adult worms were collected once the majority of the population had started laying eggs but no progeny had hatched yet ( 18 hours post L4 larval stage for wild type; 24 hours later for rpoa-2 ( op259 ) mutants ) . Gravid adults were washed from plates , cleared from bacteria and external progeny by settling in M9 buffer 3×5 min , and frozen in liquid nitrogen . Total RNA was extracted using the Nucleospin II kit ( Machery and Nagel ) , and cDNA synthesis performed on 400 ng DNA-free RNA by Superscript III ( Invitrogen ) . qPCR reactions were performed in triplicates , and the mean measured levels for transcripts of interest were normalised to the mean of a set of internal controls , including ( in part ) PGK-1 , RPL-29 , RPS-4 and transcripts shown to behave relatively stably between many conditions [107] , namely CDC-42 , PMP-3 , or Y45F10D . 4 . Proteins were extracted from well-synchronised adult worms ( collected as for RNA extraction , with 2 additional washes in ddH2O on ice ) using glass beads and Urea/Thiourea ( 7 M/1 M ) buffer and supplemented with 2% CHAPS and 75 mM DTT . Equal amounts of protein were digested with trypsin ( 100 mM Tris pH 8 . 5 , 1 mM CaCl2 ) , and the resulting peptides were – without further fractionation – purified on an MCX cartridge; the high sample volume after elution ( 5% formic acid/methanol ) was reduced by evaporation , and the concentrates were desalted with C18 ZipTips . Peptide concentrations in the final samples were determined to ensure similar loading of the mass spectrometry column . Peptides were measured in triplicate LTQ-FT runs of each non-labelled sample . Estimates of individual protein abundance in the original samples were derived from spectral counting: based on the relative abundance of proteotypic peptides for the proteins identified by a ‘Mascot’ search , a quantitative value was assigned to each protein with the Scaffold 3 software [109] . Antibodies used for immunodetection were: MAPK-YT ( Sigma M9692 , 1∶100 ) raised against di-phosphorylated ERK and binding specifically to the homologous ppMPK-1 in C . elegans as well [72]; anti-total ERK ( 1∶200 ) , cross-reacting with MPK-1; a protein-independent control antibody to dsDNA ( Abcam HYB331-01; 1∶500 ) for normalisation . Secondary antibodies: Alexa Fluor goat anti-mouse IgG1 ( MAPK-YT ) , goat anti-rabbit IgG ( ERK ) , goat anti-mouse IgG2a ( anti-dsDNA ) , all 1∶500 . Blocking buffer: 10% goat serum in antibody buffer according to [110] . In our staining protocol , gonads were extruded by dissecting adult hermaphrodites 3 hours after irradiation; samples were fixed in 3% PFA for 30 min at 4°C , freeze cracked , fixed in 100% methanol for 10 min at −20°C , stained ( blocking for 1 hour at RT; 1° antibodies in blocking buffer o/n at 4°C; 2° antibodies in blocking buffer for 1 hour at RT; >3 washes with PBS/Tween 0 . 1% between all steps ) and mounted on poly-lysine coated slides . To test for specificity of the MPK-1 antibodies , we applied the immunofluorescence protocol to mpk-1 ( RNAi ) treated worms . Loss of MPK-1 function was not very severe since most animals showed only minute phenotypes at the stage of analysis . Accordingly , ppMPK-1 and total MPK-1 signals were reduced but not completely absent . Fluorescence pictures were acquired using constant exposure settings and analysed with ImageJ .	Maintenance of genome integrity and tissue homeostasis is critically important for organisms to survive . For adequate responses to DNA damage or other types of stress , cells have evolved intricate surveillance mechanisms such as repair , cell cycle arrest and apoptosis . Disruption of these response pathways leads to an enhanced probability of cancer . The nematode C . elegans has proven a valuable model to study the genetic basis of apoptosis as an adaptive response to genotoxic or cytotoxic insults . Here , we have used C . elegans in a non-biased genetic screen to identify further regulators of DNA damage-induced apoptosis and find that ribosome synthesis factors critically determine cell death levels . We show that the newly isolated mutant rpoa-2 ( op259 ) , which has a ribosomal RNA synthesis defect , is resistant to apoptosis as a result of concurrent alterations in p53 and Ras/MAP kinase signalling . Changes in ribosome synthesis and altered Ras/MAPK activity can be observed in many tumor types , and perturbed p53 activity is among the most frequently detected abnormities in human cancer . The interaction between ribosome synthesis and major signalling pathways is potentially of high relevance for tumor biology .	[ "Abstract", "Introduction", "Results", "Discussion", "Methods" ]	[]	2013	Ribosome Synthesis and MAPK Activity Modulate Ionizing Radiation-Induced Germ Cell Apoptosis in Caenorhabditis elegans
Cytokinesis occurs through the coordinated action of several biochemically-mediated stresses acting on the cytoskeleton . Here , we develop a computational model of cellular mechanics , and using a large number of experimentally measured biophysical parameters , we simulate cell division under a number of different scenarios . We demonstrate that traction-mediated protrusive forces or contractile forces due to myosin II are sufficient to initiate furrow ingression . Furthermore , we show that passive forces due to the cell's cortical tension and surface curvature allow the furrow to complete ingression . We compare quantitatively the furrow thinning trajectories obtained from simulation with those observed experimentally in both wild-type and myosin II null Dictyostelium cells . Our simulations highlight the relative contributions of different biomechanical subsystems to cell shape progression during cell division . Cytokinesis , the separation of a mother cell into two daughter cells , is a highly stereotypical cell shape change . During most mitotic events , cytokinesis requires the careful orchestration of many cellular systems to ensure that the cell separates the genomic material into two genetically equivalent daughter cells [1] , [2] . However , the core process can be altered to produce asymmetric cell division events in which the daughter cells differ dramatically in size and/or cell differentiation fate [3] , [4] , [5] . For cytokinesis , myosin II is a key but non-essential mechanoenzyme that converts the energy of ATP hydrolysis into mechanical work [6] . Myosin II works on the actin network to alter the cell's mechanical properties in complex ways . By pulling on the filaments , myosin II can slide the polymers . This activity is the core of the traditional contractile ring model in which myosin II slides filaments , contracting the ring in a manner analogous to the contracting muscle sarcomere [7] . However , the actin polymers are held together by various actin crosslinking proteins , each with its own unique kinetic characteristics , force-sensitivity , and concentration . Thus , myosin II pulls on anchored actin filaments , leading to an effective tension due to the stalling of the myosin II motor in the isometric state [8] , [9] . As a result , myosin II is not rate-limiting for furrow ingression , and previous analyses have indicated that the furrow ingresses some 30–50-fold more slowly than predicted from the myosin II unloaded actin filament sliding velocity [10] . Ultimately , appreciating how the cell integrates three properties – biochemistry , mechanics and morphology – is the crux of understanding all cell shape changes . Because cytokinesis proceeds through genetic strain-specific geometries and characteristic dynamics , it is particularly well suited for studying how cell shape changes arise from biochemical mechanisms . This view has led to the concept that cytokinesis requires the function of the entire cortex and cytoplasm and is governed by two basic modules , global and equatorial actin-associated proteins [9] . Myosin II is found throughout the cortex but in a roughly two-fold concentration gradient between the equatorial and polar cortical domains [11] . The myosin II-mediated force generation is only one of several major mechanical systems of the cell . Two other systems include polar protrusive forces and the viscoelasticity of the cytoskeleton [8] , [12] . Another major mechanical component is derived from the cell's surface cortical tension and surface curvature , which leads to fluid pressure differentials that make cytokinesis in particular , and cell shape change in general , hydrodynamic in character . These pressure differentials lead to net flows of cytoplasm away from regions of high surface curvature to regions of lower curvature , allowing the furrow to ingress with dynamics that are controlled by the fluid dynamical and mechanical features of the cell [10] . Here , we present a computational model that demonstrates how the cell's major mechanical subsystems are integrated to drive and control cytokinesis . In particular , the model considers these separate mechanical subsystems , and explains the dynamical features of wild type and mutant cytokinesis events . Most significantly , the model demonstrates that these biomechanical systems are sufficient to explain cytokinesis . We next sought to determine whether our model cells could undergo traction-mediated cytofission , a process whereby multinucleated cells can divide during interphase [15] . We incorporated adhesion into the model taking advantage of recent measurements of the traction experienced by motile Dictyostelium cells ( Fig . 1C ) [16] . Starting from a spherical cell , we applied protrusive forces in directions 180° apart ( Fig . 1D ) . Though this assumption represents a geometrical simplification that allows us to take advantage of cylindrical symmetry , the amount of force is proportional to the cross-sectional area of the cell ( initially a circle ) and is representative of the protrusive force experienced by a cell that makes a hemispherical contact with the substrate . This force led to relatively slow cell elongation and initially , concomitant slow furrow ingression ( Fig . 2B; Video S1 ) . However , as the furrow narrowed , the cortical tension combined with an increase in local curvature to amplify the local stress . This , in turn , accelerated the rate of furrow ingression , increasing the local curvature further . This positive feedback loop caused a drastic pinching of the furrow , leading to daughter cell separation ( Fig . 2B , C ) . It must be noted that the mean curvature depends on the 3-D nature of the local geometry which involves both axial and radial components . The former is decreasing as the furrow ingresses , but the latter increases greatly during constriction . Experimentally , it is documented that separate molecular mechanisms are needed to promote the scission of the bridge joining the two daughter cells [17] , [18] . Furthermore , measurements of the furrow ingression dynamics show the existence of a bridge-dwelling step that is quantitatively separable from the mechanical stresses that drive furrow ingression [10] . For these reasons , we did not attempt to simulate the final bridge severing and stopped the simulations at this point . The rapid rate at which curvature-induced differences in cortical tension enabled furrow ingression in the previous simulation led us to posit whether spatial differences in the material properties of the cell could initiate ingression and eventually give rise to sufficient forces leading to cell division . Using micropipette aspiration , we previously measured the effective cortical tension under several contrasting conditions , including interphase vs . mitotic , WT vs . myoII null , and furrow vs . polar regions and demonstrated that the furrow exhibits a 20–30% higher effective cortical tension relative to the poles [8] , [12] . We incorporated this heterogeneity into the model and simulated cytokinesis in non-adherent ( Fig . 3A ) and adherent conditions ( Fig . 3B; Fig . S5; Video S2 ) . In both cases , heterogeneity in effective cortical tension and the resultant difference in Laplace-like pressures cause furrow ingression . In non-adherent cells , however , furrow ingression stops shortly after commencing and is not sufficient to cause further ingression or cell division . By increasing the difference in effective cortical tension , we were able to achieve cell division , but this required non-physiological differences ( 3–10 fold ) in effective cortical tension between pole and equator ( not shown ) . On the other hand , the addition of transient adhesive and protrusive forces led to successful cell division ( Fig . 3B ) . These forces appear to be required to induce a sufficient change in morphology ( specifically , curvature ) from which cortical tension can complete furrow ingression . It is well documented that Dictyostelium cells lacking functional myosin II cannot divide in suspension , but successfully divide when placed on an adhesive surface [19]; similar observations have been made of mammalian cell culture cells [20] ( Fig . 3C ) . Though this division is similar to those observed in WT cells , there are some significant differences . The furrow ingression dynamics ( quantified as the time-dependent change in the relative furrow diameter ) display biphasic behavior , in which a slow phase of ingression is followed by a rapid one [10] . We found strong agreement between the furrow-thinning dynamics predicted by our simulation and those measured experimentally in myoII null cells ( Fig . 3D; Video S3 ) . Plotting the curvature at furrow and poles during division , it is clear that the second rapid phase of furrow ingression can be attributed to the large increase in force that comes from an increase in mean curvature at the furrow ( Fig . 3E ) as the radial component of curvature begins to dominate . There are some noticeable differences in the shapes of the simulated cells when compared to the myoII null cells ( Fig . 3C , D ) . In real cells , protrusions are more “stochastic” causing ruffling at the poles . In our model , protrusive stresses are applied uniformly across the boundary and lead to a rounded shape . The treatment of adhesions is also likely to cause some of these differences . In our model , adhesion is modeled as a homogeneous friction , whereas in cells it is more likely to be localized , and this will affect the shape [21] . Furthermore , in myoII null cells , cortexillin I is not as focused in the cleavage furrow as in wild-type cells [22] , [23] , which could broaden the zone of increased elasticity Having established that material heterogeneities cannot initiate division but can provide the required force to finish it , we next considered the effect of a myosin II contractile force in our simulations . To this end , we determined the location of myosin II motors from fluorescent images of GFP-myosin II ( Fig . S1 ) and distributed a contractile force temporally and spatially based on the measured distribution of myosin II motors in the cortex ( Methods ) . Incorporating this contractile force in simulations of non-adherent cells led to successful division ( Fig . 4A; Video S4 ) . This demonstrates that a cell in suspension can initiate division by substituting the initial ingression provided by adhesion and protrusion on surfaces by myosin II constriction at the furrow . We also observed division in simulations of adherent cells ( Fig . 4B; Video S5 ) . Interestingly , cells that are adherent but do not apply protrusive forces did not divide successfully in simulation ( Fig . S2 ) . This suggests that the primary advantage of the adherent surface is that it enables cells to apply protrusive forces . Without these , adhesion acts to resist the myosin II forces and prevent sufficient cellular deformation that would otherwise enable cell division to proceed successfully . Defective cytokinesis on adherent surfaces has been documented in several Dictyostelium strains that have aberrant actin polymerization . In cells lacking coronin , an actin binding protein , attachment to the surface does not facilitate cell division [24] . Similarly , cells lacking AbiA , a component of the SCAR complex , exhibit deficient cytokinesis in adherent conditions [25] . Beyond the cell's ability to divide in non-adherent conditions , these simulations show some further differences from those of myoII null cells . The initial rate of furrow ingression in these simulations is faster than observed in the simulations devoid of myosin II contractile force . This is expected as the initial deformation now includes the cooperative interaction of two force generating subsystems . Differences are also seen in the shape of the daughter cells , as these simulations give rise to rounder cells than cells from simulations that lack myosin II contractile forces . These observations are in agreement with experimentally measured differences between WT and myoII null cells ( Fig . 4B vs . 3B ) [10] . Comparing the simulated furrow-thinning trajectory to that measured experimentally in WT cells did reveal some important differences ( Figs . 3 , 4 ) . The furrows in our simulations exhibit the same sharp drop in radius that is seen in our models of myoII null cells , which can be attributed to the large rise in pressure due to the increase in curvature . This sharp drop-off , which is not seen experimentally , leads to faster division than in real cells . To account for this difference we considered the possible role that strain-stiffening may have on furrow ingression . Strain-stiffening is a non-linear effect whereby materials harden when deformed sufficiently; this has been observed in several biopolymers [26] . Hallmarks of strain-stiffening can be seen in other aspects of Dictyostelium cellular and cytokinesis mechanics in a myosin II-dependent manner . For example , in response to pressure jumps from micropipette aspiration , cells missing myosin II show non-linear effects that are absent in WT cells , suggesting that myosin II pre-stresses the network , leading to strain-stiffening [12] . We incorporated a phenomenological description of strain-stiffening into our model ( Methods ) and simulated the system . As expected , the initial rate of furrow ingression was unaffected . However , as the furrow diameter became small enough to cause strain-stiffening , the furrow ingressed more slowly , matching the rates observed experimentally ( Fig . 4C–E; Videos S6 and S7 ) . While strain stiffening slows down the cytokinetic progression of WT strains , we have not observed this slowdown in experiments of myoII null cells . This suggests that myosin II is a fundamental component that provides this stiffening effect , an observation that is consistent with our measured material properties of myoII null cells [8] , [12] . Using this full model we considered the effect that changing the material properties of the cell have on the furrow ingression dynamics . For example , we varied the parameter controlling elasticity ( K in Fig . 1B , according to Equation 11 ) and simulated furrow ingression ( Fig . S3 ) . Increasing the elasticity constant by 40% led to a slower , more linear initial ingression ( cross-over time increased from 370 to 420 s ) , as well as slower division overall ( 415 to 495 s ) . In contrast , decreasing the elastic constant 30% shortened the cross-over time ( 370 to 350 s ) as well as the total trajectory ( 415 to 380 s ) . The simulated trajectories of the model with reduced elasticity are reminiscent of experiments of cells lacking globally-distributed proteins , such as RacE and dynacortin , that have a strong effect on the viscoelastic moduli and act to slow furrow ingression [10] . Finally , the model allows us to sort out an additional point about cytokinesis furrow ingression dynamics . In particular , it is often thought that myoII null cells divide by simply crawling apart . However , our simulations indicate key differences in mitotic cell division for both WT ( Fig . 4B , C ) and myoII null cells ( Fig . 3B ) and interphase traction-mediated cytofission ( Fig . 2B ) . By plotting the pole-to-pole distance as a function of time ( Fig . 4F ) , it can be seen that interphase cells drive fission solely by crawling apart . This leads to significant pole separation as well as long and thin morphologies ( Fig . 2B ) . In contrast , mitotic cells that have spatial heterogeneity in their mechanical properties initiate division through protrusion , but divide quite differently , with pole-to-pole distances that are similar to WT cells . Computational modeling presents an opportunity to dissect the different subsystems that contribute to force generation and subsequent cell shape changes during cytokinesis . Using an experimentally validated viscoelastic model of a Dictyostelium cell , and relevant measured data on adhesion , protrusion and myosin II-generated contractile forces , we successfully simulated cell division in several distinct virtual strains . We show that cytokinesis can be divided into three distinct phases: 1 , initial furrow ingression; 2 , Laplace-like pressure dominated , and 3 , bridge-dwelling phase [10] , [27] . Initial furrow ingression can be achieved in multiple ways using separate subsystems . Adherent cells can pull themselves apart by applying protrusive forces in two opposite directions . Alternatively , in the absence of adhesion , the initial ingression can come from the contractile forces provided by myosin II [28] . We note that alone , both of these subsystems require certain special conditions to complete division; either traction to apply protrusive forces ( Fig . 2B ) or the absence of resistance from adhesion ( Fig . S2 ) . Both our simulations and previous experimental evidence show that Dictyostelium cells can initiate cytokinesis using either of these two force producing processes . In other cell types which are less adherent , it is possible that myosin II-driven ingression may play a more important role during this first phase of ingression . While these subsystems are important to start cytokinesis , the major shape change occurs during phase 2 when the bulk of the force is provided by passive Laplace-like pressure differences that result from induced changes in mean curvature ( Fig . 5 ) . Our results demonstrate that either adhesion in combination with protrusive forces or myosin II are sufficient to drive the cell to phase 2 to allow the Laplace-like pressures to take over . Our results are also consistent with experiments of Dictyostelium cells flattened by agar overlay where full myosin II mechanochemistry is required to overcome the added mechanical stress from the compression by the sheet of agar [29] . The combination of Laplace pressures and myosin II-generated forces are large enough to make the cell divide faster than what is observed experimentally , suggesting the presence of another component that acts to slow down cell division . Several possibilities exist for this resistive force , including an axial compression acting on the ends of the furrow to counteract the effects of Laplace-like pressures and/or elastic relaxation [10] . More recent observations indicate that the slowdown depends on the lever-arm length of myosin II [30] . Wild type and a longer lever-arm mutant myosin II ( 2×ELC ) lead to furrow-thinning trajectories that are WT-like . In contrast , a short lever-arm mutant deleted for both light chain binding sites ( ΔBLCBS ) shows myoII null-like furrow-thinning trajectory though it accumulates at the cleavage furrow , demonstrating that it is not the presence of myosin II bipolar thick filaments alone that are responsible for the slower WT furrow ingression dynamics . Rather , the lever-arm length dependency suggests that it is the stalling of myosin II in the isometric state that is responsible for the slower ingression dynamics . This locking of the myosin II motor on the actin filaments then leads to an increase in myosin II-mediated crosslinking and tension and consequently an increase in the furrow stiffness ( i . e . strain-stiffening ) . While it is difficult to directly quantify the level of this increase or the time-scales over which the strain-stiffening is prominent , our simulations do suggest that non-linear strain-stiffening properties of the cortex may account for the slowdown of furrow ingression . In actuality , all three , compressive stress , elastic relaxation and strain-stiffening , are likely to contribute to varying degrees to the slowdown . Though most conceptions of cytokinesis contractility have focused almost exclusively on the contractile ring [7] , our simulations demonstrate that cell division is the result of multiple force-generating subsystems , acting on the cellular mechanical network . This explanation is particularly compelling because our model , using only experimentally measured parameters , accurately reproduces WT and mutant cell division events . While it is often considered that cytokinesis is regulated spatiotemporally by linear biochemical pathways ( such as by small GTPases and kinases ) , another level of control is equally important . For example , myosin II not only generates contractility but also controls the cortical tension , elastic modulus , and strain-stiffening . Thus , myosin II regulation affects both a force-generating subsystem and the mechanical network on which the force acts , highlighting the complex nature of the system . The level set method takes an Eulerian approach , tracking a moving boundary ( denoted Γ ( t ) ) on a static Cartesian grid deformed by a continuum stress field across the simulation domain [13] . In our simulations , we take a two-dimensional domain and assume cylindrical symmetry about the division axis ( Fig . 1A ) . The level set formalism defines a potential function φ ( x , t ) for which the boundary is the zero-level set: Γ ( t ) = {x∈R2 \| φ ( x , t ) = 0} . In our simulations , we initialize the potential function with the signed distance function , whose magnitude equals the shortest distance from a point x∈R2 to the curve Γ ( t ) and whose sign is positive if the point is outside the cell and negative otherwise . In practice , as the potential function evolves over time , it can become quite steep or flat , leading to numerical errors . These can be minimized by re-initializing the potential function periodically using the equation ( 1 ) where S ( φ ( x , 0 ) ) is taken as +1 inside the cell , −1 outside the cell and zero on the cell membrane . The potential function evolves according to the Hamilton-Jacobi equation ( 2 ) The vector v ( x , t ) is the velocity of the level set moving in the outward normal direction which , in our simulations , describes the cell's membrane protrusion and retraction velocities . These are driven by a combination of active and passive stresses acting on a mechanical model of the cell , to be described next . Previously we developed a mechanical description of a cell in the level set framework and fitted a viscoelastic model topology with parameters obtained from measurements of cells deformed using micropipette aspiration [14] . The model assumes that the cell deformation obeys , where v is the velocity defined above , and xm is the displacement of the membrane ( Fig . 1A , B ) . The total membrane displacement is the sum of the displacements of the cortex ( xcor ) and cytoplasm ( xcyt ) . To describe how stresses affect these , we use a Voigt model , which consists of the parallel connection of elastic ( K ) and viscous ( D ) elements , to represent the cortex connecting the cell membrane and the cytoplasm ( Fig . 1B ) . The viscous component describes the association and dissociation dynamics of actin cross-linkers . The cytoplasm is modeled by a purely viscous element ( B ) placed in series with the Voigt element . In our simulations , we use stress rather than force to drive the cellular deformations thus accounting for the extra µm2 found in the parameters in our model . The model assumes that these displacements occur normal to the cell surface . This neglects bending effects , which are relevant at much smaller length-scales than those we consider in modeling cytokinesis [27] , [31] . In the simulations , the total stress ( σtot ) is applied at the cell boundary , according to: , where xcor and xcyt represent the positions of the cortex and cytoplasm , respectively . Using the membrane displacement , xm = xcor+xcyt , we can rewrite the system of equations as ( 3 ) We thus obtain the membrane velocity solving first for xcor and then for . This value is entered into the Hamilton-Jacobi Equation ( Eqn . 2 ) . The total net stress ( σtot ) is computed for the simulation domain as the vector sum of all stresses acting on the cell . This includes stress contributions from active components , adhesion ( σadh ) , protrusion ( σpro ) , and myosin-based contraction ( σmyo ) , as well as passive components due to surface tension ( σten ) and volume regulation ( σvol ) . Thus ( 4 ) These individual components are now described in detail . Our model of adhesion uses a continuum stress field to counteract cellular deformations [32] and is based on defining an adhesion map , as previously described [33] . Though our simulations assume that the cell has cylindrical symmetry , for the purposes of computing adhesion and protrusion , we instead consider the cross-sectional area in the ( z , r ) plane , which more closely corresponds to the contact area between cell and substrate . We compute area densities in both r and z directions , normalized to the total cell cross-sectional area: ( 5 ) Here 1 ( z , r ) is the indicator function that equals one when the point ( z , r ) is inside the cell and zero otherwise , and the summations are done over all simulation points in either the z- or r-direction ( Fig . 1C ) . These densities describe the fraction of the cell-substrate contact area that lie in the respective strips either in the z- or r-directions . We multiply these two densities and scale by the maximum adhesion stress ( σadh-max ) to generate a spatial adhesion map: ( 6 ) This adhesion is applied spatially as a resistive stress element that counteracts the net effect of the other stresses . To evaluate this new model we simulated the cellular response to a series of pulses ( Fig . S4 ) and compared this response to that of the nominal model . Simulations that incorporate adhesion show a delayed initial response and these cells also take longer to reach steady state . We incorporate protrusive forces based on several assumptions ( Fig . 1D ) . First , protrusion acts at both ends of the cell to drive the cell apart . Thus , the protrusive stress acts away from the z = 0 line . Second , the local protrusive forces depend on the contact area ( as calculated by Dr ( z ) above ) and increase as you move away from the cleavage furrow ( scaled by a linear function l ( z ) with values of zero at the center of the division axis ( z = 0 ) and one at the poles ) . Finally , the protrusive force decreases over time as the cell is dividing . We incorporate this by including an exponential function indexed by the furrow diameter ( wf ( t ) , defined as the diameter of the cell at the midpoint along the z-axis ) . Together , these assumptions lead to a protrusion stress whose magnitude is given by ( 7 ) where σpro-max is maximum stress applied ( Table 2 ) . Though the stress is assumed to act along the z-axis ( Fig . 1D ) , only the component normal to the surface is used in the simulations . The model used here is phenomenological , but captures the net movement of the membrane away from the division plane . Other approaches , which look at finer scale effects for modeling protrusion , have been considered in the literature [34] , [35] . An active contractile force from the work of myosin II against the cytoskeleton is present in wild type cells . This force acts tangentially to the cortex , thereby constricting the cell and , because we assume cylindrical symmetry , this reduces the circumference ( Fig . 1E ) . This has the net effect of reducing the furrow diameter ( with a stress reduced by a factor of 2π to account for conversion from circumference to radius ) . Thus , to incorporate this into our model , we assume that the contractile stress acts radially inward . The magnitude of the local force depends on two things , the maximum stress generated by myosin II and the local distribution of myosin II . To compute the maximum stress we note that if we assume 3 . 4 µM total cellular concentration of myosin II monomers ( each monomer is composed of two heavy chains , two essential light chains and two regulatory light chains ) [11] , [36] , then a mitotic cell with a radius of 5 µm contains 1×106 myosin monomers ( 2×106 heads ) . Given the Dictyostelium myosin II unloaded duty ratio ( 0 . 6% ) and the force generated by the power stroke of the myosin ( 3 pN ) , the maximum total force that can be generated from myosin II is ∼40 nN , assuming no load-dependent shifts in the duty ratio . Because only ∼20% of the myosin II is found in the assembled bipolar thick filament state , most of which resides in the cortex [11] , [37] , the resulting maximal force is 10 nN . This number is used to compute the total maximum stress by dividing by the cellular area ( 4πR2 ) . To apportion this stress spatially , we imaged myoII::GFP-myoII cells ( mhcA ( HS1 ) :: pBIG:GFP-myosin II; pDRH:RFP-tubulin ) undergoing cytokinesis as previously described [38] . From this movie , the GFP-myosin II fluorescent intensities were extracted to quantify myosin density . Cell images were aligned by their centroids and along the division axis . For each image , edge detection was performed to identify cell periphery . Using this edge , the GFP-myosin II intensity was computed for 5 pixels ( 1 µm ) inwardly normal from the boundary , a region likely to contain cortical myosin . An average of these intensities was assigned as the local myosin density at that boundary point . The cell shape was averaged across both its axes of symmetry along with the GFP-myosin II distributions to construct a symmetric myosin profile along the division axis . This profile was smoothed using a cubic smoothing spline . For each image in the time series , a one-dimensional profile was constructed , indexed to the position along the division axis and the measured furrow diameter ( Fig . S1 ) . The resultant map ( myo ( r , z ) ) describes the distribution of myosin as a function of radius and is used to generate a stress: ( 8 ) where n is the outward normal unit vector . Local differences in mean curvature and surface tension give rise to spatially heterogeneous stresses on the cell . The stress differential across the boundary , described by the Young-Laplace relationship , is given by σten = γ ( z ) κmean ( z ) n , where γ ( z ) describes the local cortical tension , κmean is the mean curvature and n is a normal unit vector . The mean curvature , κmean , is the arithmetic mean of two principal curvatures ( κmean = ½ ( κ2D+κP ) ) [39] . The first is computed using a Lagrangian formulation based on the cellular boundary: κ2D ( x , y ) = ( x′y″−y′ x″ ) / ( x′2+y′2 ) 3/2 where the point ( x , y ) ∈Γ . The primes denote spatial derivatives along the boundary and are approximated by the center weighted difference between two points [13] . The computation of the second principal curvature takes advantage of the cell's cylindrical symmetry: κP = Nr ( r ) /r , where Nr ( r ) is the normal in the radial direction at a given point , and r is the radius of the cell at that location [39] . For interphase cells , we assume that cortical tension is homogeneous around the cell with a nominal value of 1 nN/µm [10] . For mitotic myoII null cells , we assume a spatially heterogeneous γ with values of 0 . 5 and 1 . 0 nN/µm at the pole and furrow , respectively [8] , [12] . We interpolate these values using a Gaussian profile: ( 9 ) where R0 is the initial radius of the cell and z is the horizontal position between the pole and furrow . In wild type cells , the cortical tension at the pole and furrow are 1 and 1 . 8 nN/µm , respectively [10] . In these simulations , we interpolate between these two values according to the measured myosin II concentration ( described below ) . This profile is used as a means of marking intracellular changes in the material properties of the cell during division , not necessarily implying that surface tension comes from myosin . We considered other schemes for spatially varying the cortical tension , but all gave similar results . For example , simulations of cells lacking myosin contractility were run varying cortical tension using a Gaussian distribution . Additionally , we performed simulations using both the myosin density profile and a normal distribution to simulate the surface tension profile but found little difference between the two . We assume that the cellular volume remains constant [14] . To enforce this constraint we implement a stress ( 10 ) where n is the outward normal . The cell's volume is evaluated by assuming the cell is radially symmetric: Vactual = ∫cell lengthπr ( z ) dz . Large values of Kvol keep the cell volume relatively constant , but can lead to small oscillations as the stress overshoots the required target . In our simulations , we set Kvol = 0 . 1 nN/µm5 , which was sufficiently high to ensure that both volume changes were small but maintained the stability of the simulations , though some oscillations ( as seen in the furrow measurements in Fig . 3E ) do appear . We assume that the elastic component of the cell undergoes strain stiffening . Though no precise model for strain stiffening is currently available , in Dictyostelium cells , we have previously observed the effect of nonlinearities in cellular responses to deformations of varying size . These differences depend on the presence of myosin II , likely due to stalling of the myosin II motors [12] , [30] . Hence , we posit a plausible phenomenological model of strain stiffening that includes the effect of both the strain ( by incorporating the change in the furrow diameter ) and the local myosin II-density . The increased elasticity at point x is given by ( 11 ) where K0 is the nominal elasticity ( Table 3 ) , myo ( z , r ) is the myosin density profile ( described above ) and wf ( t ) is the furrow diameter . The resulting temporally and spatially varying map of elasticity is then applied to the material model . The simulations are based on the Level Set Toolbox [40] and are coded in Matlab ( Mathworks , Natick , MA ) . The code is extended to implement the local level set algorithm [41] , a modification of the level set method that decreases the computational complexity by solving quantities only near the boundary . Simulations were implemented on a dynamic grid of fixed height ( 12 µm ) and varying width ( 12–24 µm ) , with density of 20 points/µm and 5-ms time steps . Simulation takes approximately 2 hours for every minute of cell division on a desk top PC . Strains used to determine experimental furrow thinning trajectories are the myoII null ( mhcA ( HS1 ) :: pLD1A15SN; pDRH:GFP-tubulin ) and the rescued myoII null as WT ( mhcA ( HS1 ) :: pBIG:GFP-myosin II; pDRH:GFP-tubulin ) [8] , [12] . Time-lapse DIC images of were taken at 2-s intervals with a 40× ( N . A . 1 . 3 ) objective with 1 . 6× optivar [8] , [12] . To determine the relative furrow diameter , we find the furrow diameter ( wf ( t ) , the diameter of the cell at the midpoint along the z-axis ) and the furrow length ( Lf ( t ) , the distance between the points of inflection in the furrow region ) . The point when the two are equal is the cross-over time , tx and this marks the cross-over distance ( Dx = wf ( tx ) = Lf ( tx ) ) . We define the relative furrow diameter as the ratio wf ( tx ) /Dx . Rescaled time is defined by shifting time so that tx = 0 . Furrow diameter and length at each time point were measured using ImageJ ( http://rsbweb . nih . gov/ij/ ) .	Cytokinesis , the physical separation of a mother cell into two daughter cells , requires force to deform the cell . Though there is ample evidence in many systems that myosin II provides some of this force , it is also well known that some cell types can divide in the absence of myosin II . To elucidate the mechanisms by which cells control furrow ingression , we developed a computational model of cellular dynamics during cytokinesis in the social amoeba , Dictyostelium discoideum . We took advantage of a large number of experimentally measured parameters and well-characterized furrow ingression dynamics for a number of different strains . Our simulations demonstrate that there are distinct phases of cytokinesis . Myosin II plays a role providing the stress that initiates furrow ingression . In its absence , however , this force can be supplied by a combination of adhesion and protrusion-mediated stresses . Thereafter , Laplace-like pressures take over and provide stresses that enable the cell to divide . Overall , we show how various mechanical parameters quantitatively impact furrow ingression kinetics , accounting for the cytokinesis dynamics of wild type and mutant cell-lines .	[ "Abstract", "Introduction", "Results", "Discussion", "Methods" ]	[ "physics", "computer", "science", "computer", "modeling", "biophysic", "al", "simulations", "biology", "cell", "mechanics", "computational", "biology", "biophysics", "biomechanics" ]	2012	Deconvolution of the Cellular Force-Generating Subsystems that Govern Cytokinesis Furrow Ingression
In the interest of identification of new kinase-targeting chemotypes for target and pathway analysis and drug discovery in Trypanosomal brucei , a high-throughput screen of 42 , 444 focused inhibitors from the GlaxoSmithKline screening collection was performed against parasite cell cultures and counter-screened against human hepatocarcinoma ( HepG2 ) cells . In this way , we have identified 797 sub-micromolar inhibitors of T . brucei growth that are at least 100-fold selective over HepG2 cells . Importantly , 242 of these hit compounds acted rapidly in inhibiting cellular growth , 137 showed rapid cidality . A variety of in silico and in vitro physicochemical and drug metabolism properties were assessed , and human kinase selectivity data were obtained , and , based on these data , we prioritized three compounds for pharmacokinetic assessment and demonstrated parasitological cure of a murine bloodstream infection of T . brucei rhodesiense with one of these compounds ( NEU-1053 ) . This work represents a successful implementation of a unique industrial-academic collaboration model aimed at identification of high quality inhibitors that will provide the parasitology community with chemical matter that can be utilized to develop kinase-targeting tool compounds . Furthermore these results are expected to provide rich starting points for discovery of kinase-targeting tool compounds for T . brucei , and new HAT therapeutics discovery programs . Human African trypanosomiasis ( HAT ) is a parasitic infection that affects 10 , 000 patients annually [1] . Current therapies for HAT have significant issues of toxicity , inconvenient dosing regimens , and emerging resistance . While there are new therapeutic options currently being evaluated in the drug development pipeline , such as SCYX–7158 [2] , nifurtimox-eflornithine combination therapy [3] , and fexinidazole [4 , 5] , there remains a need for back-up approaches to new drugs for this disease . In recent years , in response to repeated calls for new drugs from the World Health Organization and clear guidelines for new drug specifications for HAT [6] , drug discovery efforts have increased worldwide for this otherwise neglected disease . The acute lack of financial incentive for undertaking the costly drug discovery process is increasingly being addressed by a combination of public research funding , as well as philanthropic and industrial contributions . The OpenLab Foundation was established in 2010 as a means to provide financial support to drug discovery efforts performed at GlaxoSmithKline ( GSK ) in collaboration with investigators outside the company who have identified new approaches to fighting tuberculosis [7] , and tropical diseases such as HAT , malaria [8 , 9] , Chagas disease , and leishmaniasis . Indeed , in recent years large sets of screening data for compounds tested against the pathogens causing these diseases have emerged from this unique combination of industrial , philanthropic , and non-industrial collaborators [7 , 8] . One powerful approach to discovery of new drugs for HAT has been directed at repurposing established knowledge about classes of molecular targets that the pathogen holds in common with humans , recognizing that the huge body of historical knowledge around homologous targets could be quickly redirected to inhibiting pathogen growth [10] . These target families include phosphodiesterases [11] , histone deacetylases [12] , and kinases [13 , 14 , 15 , 16 , 17] . T . brucei expresses 176 kinases [18 , 19] , an observation that has driven our efforts to identify classes of kinase inhibitors that can be useful for discovery of new parasite growth inhibitors . For example , we recently reported that the human PI3K and mTOR inhibitor NVP-BEZ235 , a Phase III clinical candidate for solid tumors , is highly potent against trypanosomes in culture and in vivo [14] . Other PI3K and mTOR inhibitors were also shown to be potent lead compounds for killing trypanosomatid parasites . Recognizing that we had only tested a very small fraction of the kinase inhibitor chemotype space , and that the kinome of T . brucei , Leishmania , and T . cruzi have been analyzed and found to have moderate similarity to human kinases [18] , we hypothesized that testing a much wider set of human kinase inhibitor chemotypes should allow identification of a wide range of starting points for HAT therapeutics . Over the last few decades , drug discovery efforts have focused on discovery of inhibitors for specific molecular targets ( enzymes , receptors ) , though it is now becoming apparent that combining whole-cell activity data with molecular mechanism of action information is , in fact , the most productive path forward to new discovery [20] . Further , since little detail is known about the functions of the T . brucei kinome , the discovery of putative kinase-targeting inhibitors that potently inhibit cell growth represent an opportunity not only for new leads for HAT , but also new tool compounds for elucidation of kinase function in the pathogen . With this in mind , we now report the assessment of 42 , 444 kinase-targeted compounds in a high-throughput , cell-based T . brucei growth assay , followed by evaluation of cellular selectivity , characterization of the rate and reversibility of action , coupled with a range of predictive and experimental determinations of drug-like properties that can inform prioritization for future drug discovery efforts for HAT . To our knowledge , this represents the largest kinase-targeted HTS against T . brucei , resulting in discovery of 797 validated T . brucei growth inhibitors , grouped into 59 clusters ( plus 53 singleton compounds ) , intended to prompt new studies of mechanism of action and further pursuit for drug optimization for HAT . We further describe the prioritization of 46 clusters based on potency , rate-of-action , cidality , and predicted central nervous system ( CNS ) exposure . The progression through the HTS process and follow-up experiments is shown in Figure 1 . A kinase inhibitor subset was selected to be screened in the growth assay using three approaches . First , compounds from the GSK compound screening collection that possess>70% Tanimoto similarity to the previously-reported active PI3K/mTOR inhibitor chemotypes [14] were selected ( 2 , 979 compounds ) . Second , we included the GSK Published Kinase Inhibitor Set ( 367 compounds ) [26] . Third , we included the subset of 39 , 098 kinase inhibitor compounds from the wider screening collection to create a full 42 , 444 member kinase-targeted library . A high-throughput screen ( HTS ) was developed using a T . b . brucei ( Lister 427 strain ) whole-cell assay based on a widely-validated resazurin viability test [27] in HTS format . During assay development , final experimental conditions were assessed in terms of DMSO and compound concentration , inoculum density and resazurin-dependent fluorescence , and validated with standard trypanocidal drugs and kinase inhibitors [14 , 28] . The selected compounds from the GSK collection were tested at 4 µM concentration in a single point assay to test their growth inhibition in a log-phase culture of T . b . brucei using the optimized assay conditions . The HTS was performed in 384-well format , with a Z′ robust value mean of 0 . 78 , a mean signal-to-background> 5 and a throughput of ∼17 , 600 compounds per day [29] . Representative HTS performance metrics are available in Figure 2 , and additional details are described in the Supporting Information . Approximately 15% of the original set ( 6368 out of 42 , 444 ) displayed more than 50% growth inhibition at 4 µM drug concentration in the HTS close to the robust statistical cut-off ( 3 SD ) . These compounds were subjected to a confirmatory single concentration assay , and tested in duplicate sets , leading to 4 , 574 compounds to be advanced to T . brucei and HepG2 cell dose-response assays . We selected a relatively short incubation time for the HepG2 cell assay ( 24-48 hours ) to allow identification of compounds with acute host cell toxicity . Taking these assays together , we found 797 compounds to have EC50 values ≤1 µM against T . brucei cells , with at least 100-fold selectivity over HepG2 cells . Ligand efficiency ( LE ) is a characterization of a compound's efficiency on a per-heavy-atom basis . Calculated by 1 . 37*pEC50 divided by the number of heavy atoms , a value of LE ≥0 . 3 is typically accepted as a good starting point for optimization efforts [30] . Although ligand efficiencies come from a thermodynamic analysis of target-ligand interactions , we use them here as a score or descriptor to prioritize compounds based on their balance of activity vs molecular size and lipophilicity . LE is most commonly utilized in analysis of target-based assay results , but , as demonstrated in a recent antimalarial program , it can be utilized to normalize potency for molecular size in cell assays as well [31] . Of the 797 hit compounds , 538 ( or 68% ) show an LE>0 . 3 ( Figure 3A ) . The cLogP and molecular weight of the hits are plotted in Figure 3B , color coded for compounds that are predicted to permeate the CNS based on a predictive model ( vide infra ) . Lipophilic ligand efficiency ( LLE , pEC50-cLogP ) has recently been shown to be reliable and meaningful metric of inhibitor quality [32 , 33] , with a targeted LLE≥4 [34] . Of the hit compounds , 200 ( 25% ) show an LLE value of at least 4 , and 242 compounds ( 30% of the hits ) have a LLE value between 3–4 . Computed properties of the hit compounds , as a comparison to the overall screening set are shown in Figure 4 and are summarized in Table 1 . The distributions of the potent and specific compounds are narrower , compared to the overall screening set , with mean values slightly increased when compared to the whole HTS set . Noting that new HAT therapeutics must be centrally-acting , we then computationally characterized the hit compounds for predicted central nervous system ( CNS ) activity . We applied the method recently disclosed by Wager et al . that utilizes commonly computed properties ( cLogP , cLogD , molecular weight , topological polar surface area ( TPSA ) , hydrogen bond donors , and pKa ) to predict each compound's likelihood of CNS activity [35] . In this model , preferred ranges for each of the properties above are identified , compounds are scored based on compliance with each of those property ranges , and these properties scores are summed to obtain the CNS multiparameter optimization ( MPO ) score . Using this method , compounds with a CNS MPO score of 4 or higher is predicted to be centrally acting . ( We note that one difference between our MPO calculation and that performed by Wager et al . is the method for computation of pKa . We utilized the pKa computation from ChemAxon , whereas previous work utilized that available from ACD/Labs ) . Figure 3B shows a plot of cLogP versus molecular weight of the 797 potent and selective compounds; 329 compounds ( 41% of the total ) have a MPO score ≥4 ( green ) , and are predicted to have a high propensity for CNS activity . In order to identify the most effective anti-trypanosomal agent , we sought compounds that would rapidly ( ≤18 hours ) and irreversibly kill T . brucei cells . While the initial assays above were carried out at 72 h using resazurin as a redox indicator of cellular viability and density , we performed a different cell viability dose-response assay at shortened compound-treatment periods ( 6 , 12 , 18 and 24 h ) [2] . For these experiments we utilized ATP content as an indicator of living cells , which achieves a lower level of detection than the resazurin assay [36] . This allowed us to use lower cell densities to quantify cell death as signal decay . We note that the different readout had no effect upon compound potency at 72 hours . Since cell density remains almost invariable for 6 hours of incubation , the biological activity of compounds near this time point predominantly reflects induction of cell death . However , at incubation times>24 hours , when cell density has significantly increased , the action of compounds almost exclusively reflects growth inhibition . Thus , assessment of cell density at intermediate incubation times ( i . e . 12 and 18 hours ) can allow us to discern between compounds that act via cell killing or via growth inhibition . Compounds that do not quickly kill the parasite or arrest its growth will appear inactive at 6 hours in the ATP content assay and active at 72 hours in the resazurin-based one . Conversely , compounds with a rapid action will already exhibit their effects at short ( 6 hour ) incubation times , with increasing potency with longer incubation times . With this analysis in mind , the pEC50 values for each compound were plotted as a function of time , generating rate-of-action curves , which were shape-clustered into 11 clusters by using a K-means algorithm with Euclidean distances ( Figure 5 ) . The average shape of the curves in each cluster ( the black curves in Figure 5 ) was delineated by fitting a third degree polynomial to all the points in each cluster , allowing classification of the curve clusters , such that different clusters display different rate of action behavior . For instance , slow-acting compounds ( pEC50≥6 only after 72 hours ) are contained mainly in clusters 1 , 4 and 7 . Fast-acting compounds are contained mainly in clusters 8–11 , where pEC50≥6 was achieved as early as 12 hours of incubation time . The remaining clusters show intermediate behaviors , with a moderate but consistent increase in potency with increasing incubation time . Thus , 242 compounds showed a pEC50≥6 by 18 hours of incubation , were classified “rapidly acting” compounds and are shown as dark blue curves in Figure 5 . We selected these rapidly-acting molecules for progression into compound washout studies to determine their cidal/static properties , and also tested 46 slower-acting compounds that possessed computed properties consistent with good CNS penetration ( CNS MPO score≥4 ) . Dose-response assays were performed as above , but , after 18 hours of incubation , drugs were washed from the cell cultures . The resazurin cell viability assay was performed 72 h after drug removal to determine the extent to which the cells were able to resume replication and growth . We defined the minimum trypanocidal concentration ( MTC , EC99 ) as the compound concentration that abolishes 99% of growth recovery when compared with DMSO controls . We define “cidal” compounds to be those with pEC99>6 after 18 hours of incubation . By this definition , 56% ( or 137 ) of these rapidly acting compounds are cidal . Within this set , 24 compounds achieved pEC99 values>7 . Further analysis also revealed that 96% of the slow acting compounds that were included in these cidality assays were static in behavior . Stated differently , fast acting rate-of-action clusters ( clusters 8–11 ) contain compounds that are predominantly cidal , while other clusters are predominantly static ( Figure 6 ) . At this point we performed a series of computations that would help further prioritize these compounds in combination with the CNS MPO scores described above . We decided that a multi-factorial prioritization of screening hits was warranted with a focus on specific properties that we decided would be most valuable for selection of the most promising chemotypes for future hit-to-lead optimization . We developed a Composite Score that consisted of potency , ligand efficiency , lipophilic efficiency , rate of action , and the cidal/static nature of the inhibitor . These criteria were selected on the basis of the desired properties for HAT therapeutic candidates ( prioritizing potent , fast-acting , cidal , and predicted brain-penetrant compounds ) . The relative scoring was designed with this in mind , and based on our collective drug discovery experience . Under this scheme , the maximal score was 16 , calculated as shown in Table 2 . For example , compounds with pEC50≥8 scored 3 points , those between 7–8 , 2 points , and between 6–7 , 1 point . Compounds that were rapidly acting ( pEC50>6 at 18 hours ) received 2 points , and those whose pEC99 of 6 or greater received 2 points . We used a similar approach for scoring predicted CNS activity ( MPO Scores ) , and compound efficiencies ( LE , LLE ) . Points were totaled for each compound , providing the Composite Score . Table S5 in the Supporting Information shows the distribution of the HTS hits in each of the Composite Scoring bins . Compounds were visually grouped on the basis of common substructures , which resulted in the delineation of 59 clusters . Of these clusters , 46 clusters had at least one molecule with a composite score of at least 6 . Figures 7 and 8 shows the best-scoring representatives of each of these 46 clusters . Structures of the 13 top-scoring singleton compounds are shown in Figure 9 , and a summary of their measured and computed data is tabulated in the Supporting Information ( Table S2 ) . The Supporting Information ( Table S1 ) tabulates average values of key molecular properties for each compound cluster , along with metrics regarding the percentage of each cluster that is fast killing and/or cidal . This allows easy sorting of the compound clusters for prioritization of hit-to-lead medicinal chemistry follow-up . We selected nine of the high-potency cluster representatives shown in Figures 7 and 8 for assessment against 15 human kinases in order to ascertain information about potential selectivity liabilities . These kinases were chosen for general selectivity screening owing to their broad representation of the human kinome . The results are shown in the Supporting Information ( Table S3 ) . In addition to the computed chemical properties that are often predictive of ADME properties , we mined the GSK database for ADME properties for the hit compounds , and generated this data for the most promising hits when it was not available . We tabulate this data , where available , for the 46 cluster representatives highlighted in the Supporting Information , Table S4 . We selected three of the most attractive hit compounds to be assessed in mouse pharmacokinetic experiments , with an eye towards selecting at least one to test in an in vivo model of HAT . We selected NEU-1200 ( cluster 32 , composite score 13 ) , NEU-1207 ( cluster 34 , composite score 13 ) , and the top-scoring singleton NEU-1053 ( composite score 12 ) . The results of these experiments are shown in Figure 10A , and the pharmacokinetic parameters obtained are tabulated in the Supporting Information ( Tables S7 ) . Of these three , NEU-1053 was selected for assessment in the mouse bloodstream model of HAT on the basis of its combination of high potency ( T . b . brucei pEC50 = 9 . 17; T . b . rhodesiense pEC50 = 9 . 6; T . b . gambiense pEC50 = 9 . 7 , see Table S6 ) , rapid , cidal activity and excellent blood exposure following intraperitoneal ( ip ) dosing ( Figure 10A ) . The total blood exposure was well in excess of the EC99 observed for the compound for>24 hours , though we also observed a high plasma protein binding ( >99% ) in human plasma . This led us to elect to use a higher dose for the efficacy experiments . We focused an in vivo efficacy experiment on NEU-1053 on the basis of its excellent PK results and its very high in vitro potency . We found this compound to be well-tolerated in mice at i . p . doses of 20 mg/kg . In the bloodstream infection experiments ( that mimic Stage 1 HAT in humans ) , female NMRI mice were infected with T . b . rhodesiense ( EATRO3 ETat1 . 2 TREU164 ) on day 0 , and after three days , the infected animals were treated with 10 mg/kg doses of NEU-1053 twice a day for 4 days ( days 3–6 ) . Following a four day hiatus , drug was administered for another four days . On day 5 , none of the mice treated with NEU-1053 showed any detectable parasitemia ( Detection limit: 500 parasites/mL of blood ) , while the control group showed parasitemia of 15-940 million parasites/mL of blood ( Supplementary Information Table S9 ) . Control mice succumbed to parasitemia on days 9-13 ( 80% on days 9–10 ) , while the treated group maintained undetectable parasitemia , except one of them , which had a peak of parasitemia on day 11 ( during the second round of treatment ) . The rapid clearance of the parasitemia just 24–48 hours after the first treatment suggests NEU-1053 has a potent trypanocidal activity in vivo that is consistent with the rate of action and cidality assays described above . An identical in vivo experiment was performed using the highly virulent strain of T . b . brucei ( Lister 427 ) . In this case , after day 3 , none of the mice treated showed any detectable parasitemia , while the control group showed parasitemia of 520–1100 million parasites/mL of blood ( Supplementary Information Table S8 ) . All the control mice succumbed to parasitemia on day 5 ( the third day of the treatment ) . In contrast , the treated group maintained undetectable parasitemia . After the four day treatment hiatus , parasitemia was detected in 2 out of 4 mice . This second round of treatment resulted again in a reduction of the parasitemia to undetected levels for both positive mice . We continued to follow parasitemia after the second four-day regimen . One of the previously infected mice died on day 21 , with 1 . 2×108 parasites apparent in the blood on the day prior . However , the three other mice have been deemed to be cured , since 90 days after NEU-1053 treatment the mice are alive and no parasitemia was detected . In sum , while all the control mice succumbed to the infection , all but one of the drug-treated mice showed parasitological cure for 90 days following the drug treatment . Supplementary data tables are included in the Supporting Information as noted throughout the text , annotated with NEU registry numbers to enable online searching within the publically available data set on www . collaborativedrug . com , and in the ChEMBL database .	Human African trypanosomiasis , or sleeping sickness , affects 10 , 000 patients annually , yet current drugs for this disease are poor , with high toxicity and inconvenient dosing requirements . Trypanosoma brucei , the parasite that causes sleeping sickness , is sensitive to a class of compounds called kinase inhibitors , and our project was aimed at identifying kinase-targeting compounds that rapidly and irreversibly inhibit parasite growth . This was accomplished by high-throughput screening of over 42 , 000 compounds , which resulted in identification of 797 potent inhibitors of parasite growth that are non-toxic to human cells . These inhibitors were studied for the speed of their effects and reversibility of growth inhibition , and were grouped on the basis of chemical structure similarity . One compound was shown to cure mice from a bloodstream of infection of T . brucei . These compounds can now be utilized by the research community as starting points for new drug discovery , and also as tool compounds for understanding the function of kinases in T . brucei .	[ "Abstract", "Introduction", "Results", "Discussion" ]	[ "medicine", "and", "health", "sciences", "trypanosoma", "medicinal", "chemistry", "protozoans", "biology", "and", "life", "sciences", "pharmacology", "drug", "screening", "trypanosoma", "brucei", "gambiense", "physical", "sciences", "chemistry", "parasitic", "protozoans",...	2014	Identification and Characterization of Hundreds of Potent and Selective Inhibitors of Trypanosoma brucei Growth from a Kinase-Targeted Library Screening Campaign
Clonorchis sinensis is a carcinogenic human liver fluke , prolonged infection which provokes chronic inflammation , epithelial hyperplasia , periductal fibrosis , and even cholangiocarcinoma ( CCA ) . These effects are driven by direct physical damage caused by the worms , as well as chemical irritation from their excretory-secretory products ( ESPs ) in the bile duct and surrounding liver tissues . We investigated the C . sinensis ESP-mediated malignant features of CCA cells ( HuCCT1 ) in a three-dimensional microfluidic culture model that mimics an in vitro tumor microenvironment . This system consisted of a type I collagen extracellular matrix , applied ESPs , GFP-labeled HuCCT1 cells and quiescent biliary ductal plates formed by normal cholangiocytes ( H69 cells ) . HuCCT1 cells were attracted by a gradient of ESPs in a concentration-dependent manner and migrated in the direction of the ESPs . Meanwhile , single cell invasion by HuCCT1 cells increased independently of the direction of the ESP gradient . ESP treatment resulted in elevated secretion of interleukin-6 ( IL-6 ) and transforming growth factor-beta1 ( TGF-β1 ) by H69 cells and a cadherin switch ( decrease in E-cadherin/increase in N-cadherin expression ) in HuCCT1 cells , indicating an increase in epithelial-mesenchymal transition-like changes by HuCCT1 cells . Our findings suggest that C . sinensis ESPs promote the progression of CCA in a tumor microenvironment via the interaction between normal cholangiocytes and CCA cells . These observations broaden our understanding of the progression of CCA caused by liver fluke infection and suggest a new approach for the development of chemotherapeutic for this infectious cancer . Cholangiocarcinoma ( CCA ) is an aggressive malignancy of the bile duct epithelia associated with local invasiveness and a high rate of metastases . It is the second most common primary hepatic tumor after hepatocellular carcinoma , which is considered to be a highly lethal cancer with a poor prognosis due to the difficulty in accurate early diagnosis [1] . There are several established risk factors for CCA , including primary cholangitis , biliary cysts and hepatolithiasis [2] . Another critical factor is infection with the liver flukes Opisthorchis viverrini and Clonorchis sinensis , resulting in the highest incidences of CCA being in Southeast Asian countries [3] . The proposed mechanisms of liver fluke-associated cholangiocarcinogenesis include mechanical damage to bile duct epithelia resulting from the feeding activities of the worms , infection-related inflammation , and pathological effects from their excretory-secretory products ( ESPs ) , consisting of a complex mixture of proteins and other metabolites ) [4] . These coordinated actions provoke epithelial desquamation , adenomatous hyperplasia , goblet cell metaplasia , periductal fibrosis , and granuloma formation , all contributing to the production of a conducive tumor microenvironment . Eventually , malignant cholangiocytes undergo uncontrolled proliferation that leads to the initiation and progression of CCA [5] . Like other parasitic helminths , liver flukes release ESPs continuously during infection , in this case into bile ducts and surrounding liver tissues . These substances play pivotal roles in host–parasite interactions [6] . Exposure of human CCA cells and normal biliary epithelial cells to liver fluke ESPs results in diverse pathophysiological responses , including proliferation and inflammation [7 , 8] . Additionally , profiling of differential cancer-related microRNAs ( miRNAs ) expression has revealed that the miRNAs involved in cell proliferation and the prevention of tumor suppression are dysregulated in both CCA cells and normal cholangiocytes exposed to C . sinensis ESPs [9] . These results suggest that there are ESP-responsive pathologic signal cascades that are common to both cancerous and non-cancerous bile duct epithelial cells . Another aspect of carcinogenic transformation is the tissue microenvironment , consisting of the extracellular matrix ( ECM ) and surrounding cells and is a crucial factor in the regulation of cancer cell motility and malignancy [10] . The diverse responses of tumor cells , cholangiocytes , and immune cells in the CCA microenvironment cooperatively affect cancer progression , including invasion , and/or metastasis [11] . Chronic inflammation of the bile duct due to the presence of liver flukes is closely associated also with the development of CCA , because it causes biliary epithelial cells to produce various cytokines and growth factors including interleukin-6 , -8 ( IL-6 , -8 ) , transforming growth factor-β ( TGF-β ) , tumor necrosis factor-α ( TNF-α ) , platelet-derived growth factor and epithelial growth factor [12] . Exposure to cytokines and growth factors induces their endogenous production by CCA cells through a crosstalk loop , enhancing malignant features such as invasion , metastasis , chemoresistance and epithelial-mesenchymal transition ( EMT ) [13] . Cytokines driven by chronic inflammation contribute to the pathogenesis of CCA and should be collectively considered in studies on tumor microenvironment . We have established a three-dimensional ( 3D ) cell culture assay previously that contains a gradient of C . sinensis ESPs in the ECM and mimics the complex CCA microenvironment . In this previous study , CCA cells ( HuCCT1 ) were morphologically altered to form aggregates in response to C . sinensis ESPs , and these CCA cells could only invade the type I collagen ( COL1 ) hydrogel scaffold in response to ESP gradient treatment . This response was accompanied with an elevation of focal adhesion protein expression and the secretion of matrix metalloproteinase ( MMP ) isoforms [14] , suggesting that C . sinensis ESPs may promote CCA progression . Additionally , this study revealed the chemoattractant effect of C . sinensis ESP gradients for CCA cells and to expand this work , we explored the more complicated tumor microenvironment subjected to ESPs from C . sinensis . In the present study , we developed an in vitro clonorchiasis-associated tumor microenvironment model that consisted of the following factors: ( 1 ) a 3D culture system of normal cholangiocytes using a microfluidic device as 3D quiescent biliary ductal plates on ECM; ( 2 ) physiological co-culture of CCA cells with normal cholangiocytes coupled to the directional application of C . sinensis ESPs to reconstitute a 3D CCA microenvironment; and ( 3 ) visualization and assessment of the interactions between tumor cells and their microenvironments to assess how the malignant progression of CCA corresponds with carcinogenic liver fluke infestation ( Fig 1 ) . To reconstitute the microenvironment of a normal bile duct on an ECM , H69 cells were cultured three dimensionally on a COL1 hydrogel within a microfluidic device . The cells formed an epithelial layer and sprouted 3-dimensionally into the hydrogel one day after seeding ( Fig 2A ) . The sprouts formed 3D tube-like structures resembling newly-developed small bile ducts ( Fig 2A and 2B ) . This morphological change can be referred to as cholangiogenesis , and hepatic neoductule formation from an existing biliary ductal plate [15] . The sprouting was suppressed in this study to form quiescent mature biliary ductal plates by varying the composition of the culture medium , namely complete , fetal bovine serum-free ( FBS ( - ) ) , and FBS-free/epidermal growth factor-depleted ( FBS ( - ) /EGF ( - ) ) . In complete culture medium , H69 cells dynamically sprouted and expanded into the COL1 hydrogel and the boundary between the biliary ductal plate and the COL1 hydrogel ( Fig 2C , Day 4 ) moved far from the initial cell seeding point ( Fig 2C , Day 1 ) . In the absence of FBS and EGF , cholangiogenesis decreased dramatically ( Fig 2D ) . Additionally , the H69 cells in FBS-free/EGF-depleted medium were in G0 phase ( Fig 2E ) and expressed a basolateral polarity marker ( Integrin α6 ) along the region of the COL1 hydrogel scaffold that was in contact with the cell layer ( Fig 2F ) [16] . Therefore , we designated this cluster of H69 cells as representing a quiescent 3D biliary ductal plate . The mechanical properties of the COL1 hydrogel were modulated by altering the initial pH or concentration to identify other factors that could suppress the H69 cell sprouting . When the pH of the COL1 solution prior to gelation was basic ( pH 11 ) , the resulting hydrogel was stiffer than one gelled at pH 7 . 4 and one produced with a high concentration ( 2 . 5 mg/mL ) [17] . H69 cells on stiffer COL1 hydrogel showed more numerous sprouts with larger surface areas ( Fig 3A and 3B ) . H69 cells cultured on normal COL1 hydrogels ( 2 . 0 mg/mL and pH 7 . 4 ) and in FBS ( - ) /EGF ( - ) medium formed a quiescent biliary ductal plate . To mimic C . sinensis infestation , ductal plates were treated with ESPs ( 4 μg/mL ) by either application to the channel containing the H69 cells ( direct application ) or to the other channel of the microfluidic device ( gradient application ) . After gradient application , ESPs diffused through the COL1 hydrogel and toward the basal side of the biliary duct plate , forming a complex concentration profile . Computational simulation results showed that after 24 hours ESP concentration reached 3 μg/mL at the apical side of the biliary ductal plate ( 2~2 . 5 μg/mL at basal ) upon direct application , and 1 . 5~2 μg/mL at the basal side of the biliary ductal plate ( 1 μg/mL at apical ) upon gradient application ( Fig 4A ) . Based on the observation that the H69 cells produced three stratified layers and each layer is 10 μm in thick , ~40% of the local ESP concentration difference was applied to a single HuCCT1 cell entered the biliary ductal plate under gradient application . The 3D biliary ductal plate was stably maintained and remained healthy after either type of ESP treatment and neither treatment induced cholangiogenesis ( Fig 4B and 4C ) . HuCCT1 CCA cells labeled with GFP were seeded onto the apical side of the 3D quiescent biliary ductal plate formed by H69 cells under the culture condition as defined above . The HuCCT1 cells were then exposed to ESPs ( direct or gradient ) for 3 days . After the ESP treatment , the HuCCT1 cells actively invaded the biliary ductal plate and reached the COL1 hydrogel ( Fig 5A ) . After gradient ESP application , 1 . 71-fold and 1 . 85-fold more HuCCT1 cells invaded the biliary ductal plate compared to non-treated control , or those treated with direct application , respectively ( Fig 5B and 5D ) . Interestingly , the number of individualized HuCCT1 cells in the biliary duct layer and COL1 hydrogel were similar , both after gradient and direct treatment ( Fig 5C and 5D ) . It has been reported that elevated plasma levels of IL-6 and TGF-β1 are correlated with histophathological changes in the livers of C . sinensis-infected mice [18 , 19] . Moreover , interaction of these cytokines appears to be assoiciated with an increased malignancy of CCA cells [13] . These findings prompted us to examine whether IL-6 and TGF-β1 were involved in invasion and migration of HuCCT1 cells in our system . First , we measured IL-6 and TGF-β1 levels in the culture supernatants of ESP-treated H69 cells using ELISA , and found that secretion of both IL-6 and TGF-β1 was significantly elevated at 12 hours post-ESP treatment , compared to the non-treated control ( Fig 6A ) . Elevated secretion of IL-6 was maintained at 24 hours and increased further increased by 48 hours , while the TGF-β1 secretion level increased in a time-dependent manner . To assess the crosstalk of IL-6 and TGF-β1 from H69 cells with co-cultured HuCCT1 cells , the induction of IL-6 and TGF-β1 in ESP-treated H69 cells was attenuated by means of small interfering ( si ) RNA transfection . The culture supernatants from each of four groups of 48 hour-ESP-treated H69 cells ( transfected with siRNAs of scrambled oligonucleotide or with siRNAs for IL-6 , TGF-β1 or both ) were substituted for 24 hour-ESP-treated medium in HuCCT1 cell cultures . Then , these HuCCT1 cells were incubated further for 48 hours and their culture supernatants were analyzed using ELISA . The levels of both ESP-induced IL-6 and TGF-β1 secretion by HuCCT1 cells , as well as H69 cells , were significantly decreased in the respective siRNA transfectants , when compared with those of untransfected and scrambled siRNA-transfected . cells ( Fig 6B ) . Moreover , a greater reduction in the secretion of these cytokines was observed when using the supernatant from cells treated with siRNA for both IL-6 and TGF-β1 siRNA than when using ones from cells treated with either siRNA alone , suggesting that an IL-6/TGF-β1 autocrine/paracrine signaling network may be in effect between non-cancerous and cancerous co-cultured cells . Next , we examined ESP-mediated changes in E- and N-cadherin expression in HuCCT1 and H69 cells , which are , respectively , epithelial and mesenchymal markers , that are regarded as functionally significant factors in cancer progression . Decreased amounts of immunoreactive E-cadherin were detected in HuCCT1 cells following 24 hours post-ESP treatment , and this decreased further at 48 hours . Increased expression of N-cadherin was obvious at 24 hours post-ESP treatment and maintained up to 48 hours ( Fig 7A ) . However , in H69 cells , the expression of E-cadherin was significantly elevated at 24 hours and increased further subsequently , while there was no substantial change in N-cadherin expression during the same period of ESP treatment ( Fig 7B ) . This suggests that ESPs may contribute to facilitating EMT-like changes only in HuCCT1 cells , leading to the promotion of migration/invasion . Finally , we evaluated the involvement of IL-6 and TGF-β1 in the cadherin switching of HuCCT1 cells treated with culture supernatants from siRNA-treated cells described above . Silencing of IL-6 and TGF-β1 markedly attenuated the reduction of E-cadherin and the elevation of N-cadherin expression induced by ESPs . The levels of E- and N-cadherin expression in double silencing supernatant-treated HuCCT1 cells were almost the same as those of the non-treated control ( Fig 7C ) , indicating that the IL-6 and TGF-β1 expression induced by the ESPs contributed to EMT progression in HuCCT1 cells . The microfluidic model of a CCA tumor microenvironment used in this study consisted of a quiescent 3D biliary ductal plate formed by H69 cells ( cholangiocytes ) on a COL1 ECM that has been stimulated by C . sinensis ESPs . HuCCT1 cells ( CCA cells ) responded to microenvironmental factors actively by proliferating , migrating and invading the 3D biliary ductal plate and passing into the neighboring ECM . HuCCT1 cells exhibited different cellular behaviors when co-cultured on the biliary duct layer , compared to when they were cultured on ECM alone , as described previously [14] . As a CCA tumor microenvironment factor , characteristics of normal cholangiocytes were carefully investigated , and compared with previous reports [20] . Ishida Y et al . reported ductular morphogenesis and functional polarization of human biliary epithelial cells when embedded three dimensionally in a COLI hydrogel [21] . Tanimizu N et al . also reported the development of a 3D tubular-like structure during the differentiation of mouse liver progenitor cells [16 , 22] . However , these traditional dish-based culture platforms only generated 3D tube-like structures whose apical-basal polarities differed from those observed in vivo , and which were unsuitable for co-culturing with CCA cells to monitor tumor malignancy changes upon invasion/migration in a tumor milieu . In the microfluidic 3D culture platform described here , H69 cells formed a cholangiocyte layer and sprouted into the COL1 hydrogel . This mimicked an asymmetrical ductal structure at the parenchymal layer on the portal vein side and a primitive ductal structure during the early stage of biliary tubulogenesis during cholangiogenesis [15] . H69 cholangiocytes lining small bile ducts are layer-forming biliary epithelial cells with a potential proliferative capacity , but under normal conditions are quiescent or in the G0 state of the cell cycle [23] . The mechanical and biochemical properties of the ECM and culture medium within the microenvironment of the biliary epithelium were characterized and shown to be conducive to the formation of a stable 3D biliary ductal plate and primitive ductal structure , which are crucial steps in cholangiogenesis . The reconstituted biliary ductal plate on the ECM formed a 3D CCA tumor microenvironment , which was then seeded with CCA tumor cells and treated with C . sinensis ESPs . Many publications have described the co-culture of tumor cells with stromal cells ( mainly fibroblasts ) and reported upregulated tumor cell malignancy , however , only a few of these studied CCA cells [24] . One study co-cultured various CCA cells ( HuCCT1 and MEC ) with hepatic stellate cells as a CCA stroma and reported increased invasion and proliferation by CCA cells [25] . To our knowledge , this is the first attempt to construct a co-culture system that facilitates the direct contact of normal and CCA tumor cells from a single type of tissue; both cells were cultured separately and the combined to produce the pathophysiological effect . Therefore our work describes an advanced method for orchestrating complex CCA microenvironments , especially in 3D . The first components of the CCA microenvironment are growth factors , which are present in the culture medium and are candidates for promoting the proliferation , differentiation and migration of cholangiocytes . FBS should be preferentially excluded to arrest the cells in a quiescent state in order to assess the direct effects of C . sinensis ESPs on CCA malignancy . Although the precise effect of EGF on normal cholangiocytes remains to be elucidated , key roles of EGF in biliary duct development and cholangiocytes differentiation including cholangiogenesis and neoductule formation from an existing biliary ductal plate , have been reported [16 , 26] . Additionally , signaling via EGF and its receptor ( EGFR ) facilitates the progression of hepato-cholangiocellular cancer [27 , 28] . Thus , we excluded EGF from our 3D co-culture model system because the complex and diverse roles of EGF might mask direct ESP-dependent effects on the CCA microenvironment . The second component in the microenvironment was the COL1 ECM . The mechanical properties of the COL1 hydrogel , such as fibril diameter and stiffness , can be altered by controlling the collagen concentrations or adjusting the pH in of the collagen solution prior to gel casting . High pH reduces the diameter of COL1 nanofibers after gelation and this increases the stiffness of the COL1 hydrogel dominantly; the linear modulus of a COL1 hydrogel produced from pH 11 and 2 . 0 mg/mL is 2 . 7-fold and 3 . 1-fold higher than those from pH 7 . 4 and 2 . 5 mg/mL and from pH 7 . 4 and 2 . 0 mg/mL , being approximately 53 kPa , 20 kPa and 17 kPa , respectively [17] . It has been reported that hepatobiliary cells express diverse cellular behaviors with respect to proliferation , differentiation , adhesion , and migration under stiff ECM conditions , with close association in development , homeostasis and disease progression [29 , 30] . The morphological changes in H69 cells on stiff COL1 probably reflect the highly-activated proliferation of cholangiocytes ( ductular reaction ) in liver fibrosis , and “atypical” proliferation of cholangiocytes commonly seen in patients with prolonged cholestatic liver diseases , such as primary sclerosing cholangitis or primary biliary cirrhosis [31 , 32] . C . sinensis ESPs were the third component of the CCA microenvironment considered in this study , and ESP stimuli induced 3D morphological changes in biliary ductal plate . Some H69 cells in the 3D biliary ductal plate grown on a COL1 hydrogel interacted with it by sprouting; however , the majority maintained the layer structure during the entire experimental periods , independent of direct or gradient ESP application ( Fig 4B and 4C ) . While no obvious changes in N-cadherin expression in H69 cells were observed , the expression of E-cadherin in H69 cells was increased gradually in a time-dependent manner during the experiments ( Fig 7B ) , implying that the ESPs may cause H69 cells to exhibit more epithelial characteristics . However , we do not rule out the possibility that more intense stimulation , such as with a higher dose of ESPs and/or longer exposure times , could produce EMT-like effects in H69 cells . We determined that ESPs are implicated in the acquisition of CCA malignant characteristics; increased invasion and migration . Single cell invasion by HuCCT1 cells was similarly increased by both direct and gradient ESP application , while migration increased significantly upon gradient application only . The concentration profile produced by computational simulation explained these differential effects on HuCCT1 cells; the average concentration of ESPs over the entire area of the 3D biliary ductal plate was estimated at over 1 . 5 μg/mL and the H69 cells forming the biliary ductal plate were exposed to high concentrations of ESPs ( over 800 ng/mL ) , sufficient to induce significantly increased levels of IL-6 and TGF-β1 ( Fig 4A , red dotted line ) . In contrast , the concentration of ESPs over the channel where HuCCT1 cells were seeded was considerably higher after direct application versus gradient application; yet , the magnitude of the effect on both migration and invasion was smaller ( Fig 5 ) . These results suggest multiple pathological effects of ESPs in the CCA microenvironment , such as the stimulation of normal tissues near the CCA and the chemoattraction of CCA cells . It has been reported previously that C . sinensis ESP-triggered CCA cell migration/invasion is mediated by ERK1/2-NF-κB-MMP-9 and integrin β4-FAK/Src pathways , suggesting that ESPs may function as detrimental modulators of the aggressive progression of liver fluke-associated CCA [33 , 34] . In the present study , the morphological features of HuCCT1 cells exposed to ESPs in 3D co-culture with H69 cells differed from those of HuCCT1 cells cultured alone . Co-cultured HuCCT1 cells exhibited increased motility , as represented by single cell invasion , while HuCCT1 cells exhibited aggregation in the 3D culture [14] . This implies that the interaction between HuCCT1 and H69 cells contributes to a change in HuCCT1 cell phenotype . Cytokines generated by various types of cells within the tumor microenvironment play pro- or anti-tumorigenic roles , depending on the balance of different immune mediators and the stage of tumor development [35] . During liver fluke infection , chronically-inflamed epithelia are under constant stimulation to participate in the inflammatory response by continuous secretion of chemokines and cytokines . This creates a vulnerable microenvironment that may promote malignant transformation and even cholagiocarcinogenesis . IL-6 is considered a proinflammatory cytokine that has typically pro-tumorigenic effects during infection . Liver cell lines , including H69 cells , preferentially take up O . viverrini ESPs by endocytosis , resulting in proliferation and increased secretion of IL-6 [7] . Elevated plasma concentrations of IL-6 are associated with a significant dose-dependent increase in the risk of opisthorchiasis-associated advanced periductal fibrosis and CCA [36] . The TGF-β-mediated signaling pathway is involved in all stages of liver disease progression from initial inflammation-related liver injury to cirrhosis and hepatocellular carcinoma [37] . A crude antigen from C . sinensis differentiates macrophage RAW cells into dendritic-like cells and upregulates ERK-dependent secretion of TGF-β , which modulates the host’s immune responses [38] . C . sinensis infection activates TGF-β1/Smad signaling promoting fibrosis in the livers of infected mice [19] . Additionally , it has been reported that the E/N-cadherin switch via TGF-β-induced EMT is correlated with cancer progression of CCA cells and the survival of patients with extrahepatic CCA [39 , 40] . Consistent with these studies , we observed that the decreased E-cadherin and increased N-cadherin expression in ESP-exposed HuCCT1 cells ( Fig 7A ) was associated with increased secretion of IL-6 and TGF-β1 by H69 cells ( Fig 6A ) as well as by HuCCT1 cells , as reported previously [41] . The cytokine mediated-interaction between H69 and HuCCT1 cells was evaluated by means of siRNAs , which the levels of IL-6 and TGF-β1 secretion were suppressed in the culture supernatants of siRNA-IL-6 and -TGF-β1 H69 transfectants ( Fig 6B ) . The suppression of these cytokines was correlated with an impairment of the change in E-/N-cadherin expression in HuCCT1 cells triggered by the ESPs ( Fig 7C ) . This suggests that local accumulation of these cytokines , as the result of constitutive and dysregulated secretion of both cell types , promotes a more aggressive pathogenic process in the tumor milieu . Therefore , it is tempting to speculate that ESPs facilitate a positive feedback loop of elevated inflammatory cytokine secretion in both non-cancerous and cancerous cells , triggering an E/N-cadherin switch in HuCCT1 cells that subsequently increased invasion and/or migration mediated by the EMT . We will conduct future studies to explore this possibility . In conclusion , HuCCT1 cells exhibited elevated single cell invasion after both direct and gradient ESP application , with increased migration occurring only after gradient treatment ( ESPs applied to the basal side ) . These changes were caused by coordinated interactions between normal cholangiocytes , CCA cells and C . sinensis ESPs , which resulted in increased secretion of IL-6 and TGF-β1 and a cadherin switch in ESP-exposed cells . Therefore , the combined effects of these detrimental stimulations in both cancerous and non-cancerous bile duct epithelial cells during C . sinensis infection may facilitate a more aggressive phenotype of CCA cells , such as invasion/migration , resulting in more malignant characteristics of the CCA tumor . Our findings broaden our understanding of the molecular mechanism underlying the progression of CCA caused by liver fluke infection . These observations provide a new basis for the development of chemotherapeutic strategies to control liver fluke-associated CCA metastasis and thereby help to reduce its high mortality rate in the endemic areas . Cell culture medium components were purchased from Life Technologies ( Grand Island , NY ) , unless otherwise indicated . Polyclonal antibodies against the following proteins were purchased from the indicated sources: Ki-67 and integrin α6 ( Abcam , Cambridge , UK ) ; E-cadherin ( BD Biosciences , San Jose , CA ) ; N-cadherin ( Santa Cruz Biotechnology , Santa Cruz , CA ) ; glyceraldehyde-3-phosphate dehydrogenase ( GAPDH; AbFrontier Co . , Seoul , Korea ) . Horseradish peroxidase ( HRP ) -conjugated secondary antibodies were obtained from Jackson ImmunoResearch Laboratory ( West Grove , PA ) . All other chemicals were obtained from Sigma-Aldrich ( St . Louis , MO ) . Human HuCCT1 cholangiocarcinoma cells ( originally established by Miyagiwa et al . in 1989 [42] ) was maintained in RPMI 1640 medium supplemented with 1% ( v/v ) penicillin/streptomycin and 10% FBS . Human H69 cholangiocyte cells , which are SV40-transformed bile duct epithelial cells derived from non-cancerous human liver [43] , were kindly provided by Dr . Dae Ghon Kim of the Department of Internal Medicine , Chonbuk National University Medical School , Jeonju , Korea . H69 cells were grown in DMEM/F12 ( 3:1 ) containing 10% FBS , 100 U/mL penicillin , 100 μg/ml streptomycin , 5 μg/ml of insulin , 5 μg/ml of transferrin , 2 . 0 ×10−9 M triiodothyronine , 1 . 8 × 10−4 M adenine , 5 . 5 × 10−6 M epinephrine , 1 . 1 × 10−6 M hydrocortisone , and 1 . 6 × 10−6 M EGF . Both cell types were cultured at 37°C in a humidified atmosphere containing 5% CO2 . Clonal cell lines that stably expressed a green fluorescent protein ( GFP ) were generated by transfection of HuCCT1 cells . Briefly , HuCCT1 cells were grown to ~70% confluence and were transfected using Lipofectamine 2000 ( Invitrogen , Calsbad , CA ) and a pGFP-C1 vector ( Clontech Laboratories , Inc . , Palo Alto , CA ) for 24 hours . To generate stable lines , the cells were cultured for 3 weeks in a complete medium containing 1 mg/ml G 418 disulfate salt ( Sigma-Aldrich ) that was changed every 2~3 days . Colonies with uniform GFP fluorescence were screened and two clonal cell lines with approximately similar levels of GFP overexpression were chosen for further experiments . Adult C . sinensis specimens for the preparation of ESPs were obtained from infected , sacrificed New Zealand albino rabbits to collect adult worms . Animal care and experimental procedures were performed in strict accordance with the national guidelines outlined by the Korean Laboratory Animal Act ( No . KCDC-122-14-2A ) of the Korean Centers for Disease Control and Prevention ( KCDC ) . The KCDC-Institutional Animal Care and Use Committee ( KCDC-IACUC ) /ethics committee reviewed and approved the ESPs preparation protocols ( approval identification number; KCDC-003-11 ) . The ESPs from C . sinensis adult worms were prepared as described previously [41] . Briefly , adult worms were recovered from the bile ducts of male New Zealand albino rabbits ( 12 weeks old ) orally infected with ~500 metacercariae 12 weeks earlier . Worms were washed several times with cold phosphate-buffered saline ( PBS ) to remove any host contaminants . Five fresh worms were cultured in 1 mL of prewarmed PBS containing a mixture of antibiotics and protease inhibitors ( Sigma-Aldrich ) for 3 hours at 37 °C in a 5% CO2 environment . Then the culture fluid was pooled , centrifuged , concentrated with a Centriprep YM-10 ( Merck Millipore , Billerica , MA ) membrane concentrator , and filtered through a sterile 0 . 2-μm syringe membrane . After measuring the ESP protein concentration , the aliquots were stored at −80°C until use . The microfluidic device was prepared as described previously [14] . Briefly , the microfluidic devices were produced by curing polydimethylsiloxane ( PDMS , Silgard 184 , Dow Chemical , Midland , MI ) overnight on a micro-structure-patterned wafer at 80°C . The device was punched to produce ports for the hydrogel and cell suspension injections . After sterilization , the device and s glass coverslip ( 24 × 24 mm; Paul Marienfeld , Germany ) were permanently bonded to each other and the surfaces of microchannels in the device were coated with poly-D-lysine by treatment 1 mg/mL solution . The devices were stored under a sterile condition until use . The gel region of microfluidic device was filled with an unpolymerized COL1 solution ( 2 . 0 mg/mL , pH 7 . 4 ) and then placed in a 37°C humidified chamber to polymerize the hydrogel . EGF-depleted H69 medium containing 1% FBS was injected into the medium channels to prevent shrinkage of the COL1 hydrogel , and the devices were stored at 37°C in a 5% CO2 incubator until cell seeding . H69 cells ( 5 × 105 cells ) suspended in conditional medium ( FBS-free , EGF-depleted ) were loaded into one medium port . After filling a medium channel by the cells in the suspension by hydrostatic flow , the device was positioned vertically for 2 hours at 37°C in a 5% CO2 incubator to allow the cells to attach to the COL1 hydrogel wall by gravity . One day after seeding with H69 cells , HuCCT1-GFP cells suspended in conditional medium at 10 × 105 cells/mL were seeded into the cell channel in a manner identical to the H69 cells . ESPs were diluted in conditional medium to a concentration of 4 μg/mL and then added either to the cell channel ( direct application ) or to the medium channel ( gradient application ) . The medium was replaced every day with fresh conditional medium supplemented with ESPs ( Fig 1B ) . H69 cells cultured in a microfluidic device were washed twice with PBS and fixed with a 4% paraformaldehyde solution for 30 minutes . A 0 . 1% Triton X-100 solution was treated to permeabilize the cell membranes for 10 minutes . The cells were incubated with 1% bovine serum albumin and primary antibodies against Ki67 or Integrin α6 ( 1:1000 dilution ) , followed by Alexa Fluor 488 secondary antibody ( 1:1000 dilution; Invitrogen ) . After staining with 4' , 6-Diamidino-2-Phenylindole ( DAPI , 1:1000 dilution , Invitrogen ) , and rhodamine phalloidin ( to stain F-actin , 1:200 dilution , Invitrogen ) , the cells were examined by a confocal laser-scanning microscope ( LSM700; Carl Zeiss , Jena , Germany ) and by fluorescent microscope ( Axio Observer Z1; Carl Zeiss , Jena , Germany ) . We used the siRNAs ( Ambion Silencer Select ) of IL-6 , TGF- β1 , and scrambled oligonucleotide as a negative control from Thermo Fisher Scientific ( Waltham , MA ) . H69 or HuCCT1 cells were seeded on 24-well culture plate and transiently transfected with either each or both target-specific siRNAs using Lipofectamine RNAiMAX ( Invitrogen ) according with the manufacturer’s protocols . Each siRNA transfection was performed in quadruplicate . After 24 hours , the transfection mixture on the cells was replaced with fresh culture medium . At 60 hour after transfection , H69 cells were depleted of FBS gradually , followed by incubation in conditional medium supplemented with 800 ng/mL ESPs for 48 hours . The culture supernatants from H69 cells were collected and clarified by brief centrifugation . Then , the 24 hour-ESP ( 800 ng/mL ) -treated medium of HuCCT1 cells was replaced by these supernatants . The supernatants and cells were harvested after 48 hours and used for both ELISA and immunoblot analyses . HuCCT1 or H69 cells treated with 800 ng/mL ESPs , for the indicated times , were washed with ice-cold PBS and then lysed with a RIPA buffer containing a complete protease inhibitor cocktail ( Sigma-Aldrich ) . Thirty μg of total soluble protein was separated by SDS-PAGE and electrophoretically transferred to a nitrocellulose membrane ( Millipore , Bedford , MA ) . Membranes were probed with primary antibodies against E-cadherin ( 1:3000 dilution ) or N-cadherin ( 1:1000 ) . After incubation with host-specific secondary antibodies , the immunoreaction was detected with a West-Q-chemiluminescent substrate kit ( GenDEPOT , Barker , TX ) and quantitated by densitometric scanning of the X-ray film with a Fluor-S Multimager ( Bio-rad , Hercules , CA ) . The blots were normalized for protein loading by washing in Blot-Fresh Western Blot Stripping Reagent ( SignaGen Laboratories , Gaithersburg , MD ) and re-probing with a polyclonal antibody against GAPDH ( 1:5000 dilution ) . Immunoreactive TGF-β1 and IL-6 in non-treated and ESP-treated ( 800 ng/mL ) culture supernatants of H69 cells were quantitated in duplicate using commercially available ELISA kits ( Enzo Life Sciences Inc . , Ann Arbor , MI ) , according to the manufacturer’s instructions . The levels of each cytokine were determined by measuring absorbance at 450 nm and by comparing the absorbance values against a standard curve obtained using a four-parameter logistic curve fit . COMSOL Multiphysics 5 . 0 ( COMSOL , Sweden ) software was used to simulate the concentration profile of ESPs . We assumed the molecular weight of ESPs to be 40 kDa , which is the molecular weight of cathepsin B , major component of ESPs [44] . The diffusion coefficients of ESPs in various milieus were estimated 8×10−11 m2/s in to be cell growth medium , 5 . 8×10−11 m2/s in a collagen hydrogel and 1 . 57×10−12 m2/s in a cell layer , based on previous studies on the VEGF transport [45] , because the molecular weight of cathepsin B is similar to VEGF ( 42 kDa ) .	The oriental liver fluke , Clonorchis sinensis , is a biological carcinogen of humans and is the cause of death of infectious cancer patients in China and Korea . Its chronic infection promotes cholangiocarcinogenesis due to direct contact of host tissues with the worms and their excretory-secretory products ( ESPs ) ; however , the specific mechanisms underlying this pathology remain unclear . To assess its contribution to the progression of cholangiocarcinoma ( CCA ) , we developed a 3-dimensional ( 3D ) in vitro culture model that consists of CCA cells ( HuCCT1 ) in direct contact with normal cholangiocytes ( H69 ) , which are subsequently exposed to C . sinensis ESPs; therefore , this model represents a C . sinensis-associated CCA microenvironment . Co-cultured HuCCT1 cells exhibited increased motility in response to C . sinensis ESPs , suggesting that this model may recapitulate some aspects of tumor microenvironment complexity . Proinflammatory cytokines such as IL-6 and TGF-β1 secreted by H69 cells exhibited a crosstalk effect regarding the epithelial-mesenchymal transition of HuCCT1 cells , thus , promoting an increase in the metastatic characteristics of CCA cells . Our findings enable an understanding of the mechanisms underlying the etiology of C . sinensis-associated CCA , and , therefore , this approach will contribute to the development of new strategies for the reduction of its high mortality rate .	[ "Abstract", "Introduction", "Results", "Discussion", "Materials", "and", "methods" ]	[ "biliary", "system", "invertebrates", "amorphous", "solids", "medicine", "and", "health", "sciences", "innate", "immune", "system", "liver", "immune", "physiology", "cytokines", "engineering", "and", "technology", "helminths", "gene", "regulation", "immunology", "carcin...	2019	Clonorchis sinensis excretory-secretory products increase malignant characteristics of cholangiocarcinoma cells in three-dimensional co-culture with biliary ductal plates
Dendrodendritic synaptic interactions between olfactory bulb mitral and granule cells represent a key neuronal mechanism of odor discrimination . Dendritic release of gamma-aminobutyric acid ( GABA ) from granule cells contributes to stimulus-dependent , rapid , and accurate odor discrimination , yet the physiological mechanisms governing this release and its behavioral relevance are unknown . Here , we show that granule cells express the voltage-gated sodium channel α-subunit NaV1 . 2 in clusters distributed throughout the cell surface including dendritic spines . Deletion of NaV1 . 2 in granule cells abolished spiking and GABA release as well as inhibition of synaptically connected mitral cells ( MCs ) . As a consequence , mice required more time to discriminate highly similar odorant mixtures , while odor discrimination learning remained unaffected . In conclusion , we show that expression of NaV1 . 2 in granule cells is crucial for physiological dendritic GABA release and rapid discrimination of similar odorants with high accuracy . Hence , our data indicate that neurotransmitter-releasing dendritic spines function just like axon terminals . Inhibitory interactions in the olfactory bulb ( OB ) play a crucial role for spatiotemporal processing of olfactory information [1–5] . The inhibitory network forming the mammalian OB circuitry [6–9] is dominated by a reciprocal dendrodendritic synapse established between mitral cells ( MCs ) and granule cells ( GCs ) [10–15] . This synapse has been shown to define olfactory discrimination time in a stimulus-dependent manner [16] . The structural basis for this behavioral function is represented by two adjacent and inversely oriented synaptic contacts formed between an MC lateral dendrite and a GC spine: an asymmetric contact mediating glutamatergic excitation of the GC and a symmetric contact mediating GABAergic inhibition of the MC [11] . Intense research has addressed the physiological properties of this synapse [17–22] , but important aspects of fast and synchronous neurotransmitter release from GC spines and its molecular underpinnings remain poorly understood . A key issue concerns the mechanisms translating the glutamatergic excitatory postsynaptic potential ( EPSP ) into gamma-aminobutyric acid ( GABA ) release from the GC spine . In axonal nerve terminals , fast and synchronous neurotransmitter release is triggered by a highly localized cytoplasmatic Ca2+ nanodomain generated by voltage-dependent calcium channels ( VDCCs ) in response to an action potential invading the presynaptic terminal [23–27] . Mainly , the Ca2+ tail current during the repolarization phase of the action potential boosts Ca2+ entry and thereby drives neurotransmitter release [25] . This axonal mechanism of neurotransmitter release constitutes a hallmark of synaptic transmission , because it ensures fast and synchronous release [25 , 26] . Dendritic release is thought to function differently . Previous work at the dendrodendritic synapse suggested that N-methyl-D-aspartate receptor ( NMDAR ) -mediated Ca2+ inflow into the GC spine directly triggers GABA release [17–19 , 28] . However , such a dendritic release mechanism would require Ca2+ to diffuse over a large distance ( approximately 1 μm ) from the glutamatergic postsynaptic density to the GABAergic active zone [11] , implying a large Ca2+ domain directly driving neurotransmitter release . Yet such large Ca2+ domains are unlikely to exist in GCs , because the high intrinsic buffer capacity of GCs [29] limits Ca2+ diffusion , and EGTA does not affect dendrodendritic inhibition [20] . Alternatively , the glutamatergic EPSP may be sufficiently strong to directly activate VDCCs located next to GABAergic vesicles [20] . Specifically , the Na+ inward current provided by α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptors ( AMPARs ) may either directly trigger a local Ca2+ spike [20] or relieve Mg2+ block of NMDARs and thereby contribute to the local Ca2+ signal causing GABA release [19] . Other conductances , such as transient receptor potential-canonical ( TRPC ) channels [30] or T-type VDCCs [21] , could amplify this process . Nevertheless , none of these mechanisms can provide a sufficiently steep voltage gradient to boost Ca2+ entry via the tail current mechanism . Only one report pointed toward a role of voltage-gated sodium channels ( VGSCs ) in GABA release from GCs [19]: tetrodotoxin ( TTX ) application decreased dendrodendritic inhibition when applying short depolarizing pulses to MCs , while application of long depolarizing pulses increased it . Because this study relied on applying depolarizing square pulses ( 3 and 50 ms ) that have no similarity with the time course and shape of an action potential , and furthermore used 0 mM Mg2+ and TTX in the external solution , these results describe a functional state that cannot be considered physiological . Hence , the contribution of VGSCs to physiological GABA release from GCs remains unknown . Although dendritic expression of the VGSC NaV1 . 6 α-subunit has been demonstrated in hippocampal neurons [31] , the molecular identity of dendritic VGSCs and , in particular , their contribution to dendritic transmitter release remain poorly described . Furthermore , the network and behavioral consequences of dendritic VGSC function remain unknown . To address these issues , we evaluated the expression of VGSCs ( reviewed in [32] ) in mouse OB GCs and targeted the function of VGSCs in GC dendritic processing . Using 3D-immunohistochemistry [33] , we found that GCs exclusively express the NaV1 . 2 α-subunit strongly clustered across the entire GC surface , including the spine heads . We specifically knocked down NaV1 . 2 expression in GCs using viral short hairpin RNA ( shRNA ) delivery . Na+-current and GC spiking were abolished in knockdown GCs . Dendrodendritic inhibition of MCs was strongly attenuated . Discrimination of highly similar odorants required more time for accurate discrimination , yet discrimination of dissimilar odors and odor discrimination learning remained unaffected in NaV1 . 2 knockdown mice . These results establish that VGSCs play a pivotal role in synchronous GABA release under physiological conditions at the dendrodendritic synapse that is required for fast odor discrimination . To clarify the role of VGSCs in GC function , we first analyzed the expression of VGSC α-subunits in the mouse OB ( Fig 1 ) . Reverse transcriptase PCR ( rT-PCR ) showed that the SCN1A , SCN2A , SCN3A , and SCN8A mRNAs were abundant in the OB , while only negligible amounts of SCN5A and SCN11A mRNA could be detected ( Fig 1A ) . Because SCN5A ( NaV1 . 5 ) is expressed in cardiac myocytes [34] and in limbic regions [35] , and SCN11A ( NaV1 . 9 ) is expressed only in dorsal root ganglia ( DRG ) neurons [36] , we focused our subsequent protein analysis on a subset of four VGSC α-subunits known to be widely expressed in the brain [32] . Western blot analysis revealed the expression of five α-subunits in the mouse OB ( Fig 1B ) , four of which ( NaV1 . 1 , NaV1 . 2 , NaV1 . 3 , and NaV1 . 6 ) correspond to the genes identified by rT-PCR ( SCN1A , SCN2A , SCN3A , and SCN8A , respectively ) . Additionally , the VGSC α-subunit NaV1 . 7 was prominently expressed in the OB despite the absence of its mRNA ( SCN9A ) . Furthermore , we performed immunohistochemistry using subunit-specific antibodies to assess the distribution of the VGSC α-subunits in the mouse OB ( Fig 1C ) . We used the auxiliary β-subunit NaV2 . 1 to normalize the signal intensity of the VGSC α-subunits , because it is expressed ubiquitously with the α-subunits [32] . Indeed , we have observed that the signal intensity of the β-subunit NaV2 . 1 is homogeneous across the OB layers ( Fig 1D; ANOVA , F = 0 . 57 , p = 0 . 64 ) . Our region of interest ( ROI ) -based analysis of the signal intensity of the VGSC α-subunits ( Fig 1E ) suggests that NaV1 . 2 is most abundantly expressed in the OB , in particular in the granule cell layer ( GCL ) and in the external plexiform layer ( EPL ) . NaV1 . 1 is strongly expressed in the mitral cell layer ( MCL ) and olfactory nerve layer ( ONL ) , while NaV1 . 3 mostly occurs in the ONL . NaV1 . 6 is strongly expressed in the ONL and shows a weak and punctate expression in the GCL . The NaV1 . 7 subunit is strongly expressed in the ONL , consistent with its expression in olfactory sensory neurons [37] and the absence of its mRNA in our OB sample ( see Fig 1A ) . The monoclonal antibodies against NaV1 . 1 and NaV1 . 2 target epitopes on the N-terminus of these α-subunits that share a high sequence similarity . We used human embryonic kidney 293 ( HEK293 ) cells transfected with a plasmid expressing the NaV1 . 1 and NaV1 . 2 epitope sequences to confirm the specificity of these antibodies ( S1 Fig ) . In conclusion , we have identified the distribution of VGSC α-subunits in the mouse OB , while it remains unclear which subunits are expressed in GCs . To identify the VGSC α-subunits expressed in mature GCs , 3D-immunohistochemistry [33] of sparsely prelabeled GCs was performed . To achieve this , the GCL of 3-week-old mice was stereotaxically injected [16] with recombinant adeno-associated virus 2/1 ( rAAV2/1 ) encoding membrane-bound green fluorescent protein ( mGFP ) [38] . After 3 weeks of expression , tissue containing prelabeled GCs was immunostained with antibodies against the VGSC α-subunit of interest . Only cells with clear GC morphology ( soma diameter of <10 μm , dendrite projecting into the EPL [39] ) were considered for further analysis . The immunosignal residing inside the mGFP-delimited GC was excised , thereby delineating the expression pattern of α-subunits in individual GCs . Strikingly , only the NaV1 . 2 α-subunit was expressed consistently in GCs at detectable levels ( Fig 2 and S2 Fig ) . The other subunits found to be present in the GCL ( NaV1 . 1 , NaV1 . 3 , and NaV1 . 6 ) are likely expressed in axons of MCs and cell types other than GCs [40 , 41] . In some instances , we observed a random partial overlap between NaV1 . x and mGFP ( S2 Fig ) owing to technical limitations of 3D-immunohistochemistry [33] . To further corroborate our finding , we used phrixotoxin-3 , a selective antagonist of NaV1 . 2 at low nanomolar concentrations [42] , to block Na+ currents in GCs ( S3A–S3C Fig ) . This experiment showed a 75% ± 10% ( n = 4 ) reduction of the Na+ current at −30 mV and further reduction to 96% ± 0 . 8% after adding 1 μM TTX . The incomplete block by phrixotoxin-3 was due to the low concentration that had to be used to selectively block NaV1 . 2 ( 1 nM , IC50 is 0 . 6 nM [42] ) . In contrast to GCs , phrixotoxin-3 caused only a minor reduction of Na+ currents elicited in MCs ( S3D–S3F Fig ) . Hence , phrixotoxin-3 block of the Na+ current is consistent with the expression of NaV1 . 2 reported by immunohistochemistry . NaV1 . 2 was found in a strongly clustered arrangement in the cell body ( Fig 2A ) , dendritic shaft ( Fig 2B ) , and in the gemmules ( Fig 2C ) of the GCs . Close-up 3D reconstructions of the gemmules revealed clusters of NaV1 . 2 immunoreactivity within the spines ( Fig 2D and 2E ) . To quantify the distribution of NaV1 . 2 clusters in GCs , we determined the ratio of the NaV1 . 2 cluster area relative to the area delineated by mGFP expression ( Fig 2H ) . This analysis revealed the highest density of NaV1 . 2 clusters within the spine heads and dendrites . At GC somas , the cluster density was significantly lower ( soma: 7% ± 1 . 3%; GCL dendrites: 32 . 2% ± 6 . 7%; EPL dendrites: 31 . 1% ± 6 . 5%; 43% ± 3 . 7% ANOVA , F = 9 . 101 , p = 0 . 001 ) . This distribution pattern predicts a dominant function of NaV1 . 2 channels in the dendrites and particularly within gemmules of GCs . While NaV1 . 2 has been described as an axonal subunit [32] , this expression pattern is remarkable for axonless GCs , suggesting that the NaV1 . 2 α-subunit might trigger synchronous GABA release from GC spine heads . To abolish NaV1 . 2-dependent Na+ currents in GCs , we designed shRNA molecules to specifically target SCN2A ( see Materials and methods , S4 Fig ) . Four different shRNAs bicistronically expressing enhanced green fluorescent protein ( eGFP ) were cloned in rAAV2/1 vectors and stereotaxically delivered to the GCL . Upon 5 weeks of expression , OB acute slices were prepared to assess Na+ currents in eGFP-positive GCs expressing the respective shRNA ( S4 Fig ) . shRNA#14 showed the most potent reduction of the Na+ current in GCs ( 90% ± 3% , n = 8 ) and was therefore chosen for all subsequent experiments . The specificity of shRNA#14 for NaV1 . 2 was further examined in silico ( S5 Fig ) . The Na+ current was reliably reduced in comparison to control cells and cells infected with a mismatch shRNA ( shRNAmm ) ( Fig 3A and 3B ) , suggesting a specific knockdown of the NaV1 . 2 α-subunit in the mouse OB . The observed reduction of the Na+ current amplitude after knockdown abrogated action potential firing in GCs upon somatic current injections ( Fig 3C ) . GCs expressing shRNA#14 failed to fire action potentials , while control cells or cells expressing the shRNAmm fired a single action potential upon 3-ms current injections ( Fig 3C , left set of traces ) or fired tonically upon longer current injections ( Fig 3C , right set of traces ) . These experiments demonstrate that a selective knockdown of the NaV1 . 2 α-subunit abolishes action potential firing in GCs . To further corroborate shRNA-mediated NaV1 . 2 knockdown , we examined the morphology of transduced GCs and quantified NaV1 . 2 expression . The general morphological features of GCs were unchanged upon shRNA expression ( Fig 3D ) , and we observed a reduction of the NaV1 . 2 staining ( Fig 3E ) upon viral transduction of the GCL with rAAVs carrying shRNA#14 , in comparison to control infection ( mGFP ) and shRNAmm14 ( Fig 3F ) . For a quantitative comparison , we calculated fluorescence ratios relative to the mean fluorescence of the control ( control: 1 . 00 ± 0 . 019 , shRNAmm: 1 . 01 ± 0 . 02 , shRNA#14: 0 . 44 ± 0 . 02 , ANOVA , F = 265 . 2 , p < 0 . 001 , see Materials and methods ) . While we observed no intensity ratio differences between the control and the shRNAmm ( t = 0 . 42 , p > 0 . 05 ) , the intensity ratio was strongly reduced when comparing shRNA#14 expression with control ( t = 19 . 73 , p < 0 . 001 ) and shRNAmm ( t = 20 . 16; p < 0 . 001 ) . In summary , we demonstrated the effectiveness of shRNA#14 in reducing NaV1 . 2 subtype expression in GCs , leading to a strong reduction of Na+ currents and abrogating action potential firing in GCs . Before analyzing the physiological and behavioral consequences of NaV1 . 2 knockdown in GCs , we investigated the spread of rAAV1/2 infection for widespread expression of shRNA#14 within the OB ( S6A Fig ) . We observed that virtually all green fluorescent protein ( GFP ) -positive cells were confined to the GCL , indicating that the stereotaxic delivery of rAAV2/1 particles into the GCL spares the MCL , EPL , and glomerular layer ( GL ) , as previously demonstrated [16] . Furthermore , rAAV transduction did not spread beyond the OB , thereby constituting a highly local genetic perturbation , a prerequisite to causally link molecular function to odor discrimination behavior ( S6B Fig ) [16] , [38] . While this approach is not inherently GC specific , more than 99% of the cells expressing GFP were GCs as judged by the diameter of the soma and location of the dendrite ( S6C–S6E Fig ) [39 , 40] . Expression of the shRNA#14 ( kd ) and the shRNAmm control were tracked by coexpression of eGFP and compared to eGFP expression alone ( control ) . Additionally , to quantify the number of infected cells in the GCL , the neuronal marker NeuN was used to identify the nuclei of neurons populating the OB , while 4 , 6-diamidino-2-phenylindole ( DAPI ) was used to label nuclei of all cell types in the OB sections ( Fig 3G; S6F Fig ) . The number of DAPI-positive nuclei did not vary among the conditions tested ( control: 529 . 7 ± 22 . 07 , n = 4; mm: 486 . 5 ± 34 . 40 , n = 6; kd: 520 . 0 ± 30 . 82 , n = 6; ANOVA , p = 0 . 58 ) , indicating that neither cell death nor glial proliferation occurred ( S6G Fig ) . This is consistent with the unaltered ratio of NeuN/DAPI cells in the GCL ( S6H Fig; control: 73 . 92% ± 2 . 25%; mm: 80 . 43% ± 2 . 20%; kd: 76 . 16% ± 3 . 84%; ANOVA , F = 1 . 33 , p = 0 . 31 ) . To assess the number of neurons infected in the GCL , we counted the nuclei that were positive for both NeuN and eGFP ( Fig 3H ) . The fraction of infected neurons was not significantly different ( control: 47 . 57% ± 0 . 93%; mm: 54 . 92% ± 1 . 73%; kd: 55 . 45% ± 2 . 35%; ANOVA , p = 0 . 12 ) , suggesting that each viral preparation transduced approximately 50% of the GCs ( Fig 3H ) . While the GCL is dominated by GCs , other types of neurons are present as well , although at much lower numbers [40 , 41] . Because viral infection might also transduce a small number of these neurons ( S6D and S6E Fig ) , we refer to mice infected with rAAVs carrying the shRNA#14 as NaV1 . 2ΔGCL mice and mice infected with rAAVs carrying the shRNAmm as mmΔGCL . Herewith , we show that our acute genetic perturbation approach is effective in selectively infecting the GCL , leaving other OB cell layers unaffected . To test if a lack of action potential firing in GCs would affect dendrodendritic inhibition of MCs by GCs , we performed whole-cell current clamp recordings from MCs in acute OB slices at near physiological temperature ( 34 ± 1 °C ) . Somatic current injections in MCs reliably generated action potentials ( Table 1 , Fig 4A and 4B ) . In the control and shRNA conditions , similar amounts of current were required to trigger action potentials with indistinguishable properties ( Table 1 , ANOVA , F = 0 . 80 , p = 0 . 45 ) . The action potential was followed by a rapid and pronounced hyperpolarization that increased with the number and frequency of action potentials elicited ( Fig 4A and 4B ) . To further define the nature of this hyperpolarization , we blocked dendrodendritic transmission with ionotropic glutamate receptor ( 6-cyano-7- nitroquinoxaline-2 , 3-dione [CNQX] and 2-amino-5-phosphonopentanoic acid [APV] ) or GABAA receptor ( gabazine ) antagonists ( S7 Fig ) . In both cases , a remaining hyperpolarization of 8%–11% was observed that could be attributed to MC-intrinsic conductances such as the afterhyperpolarization ( S8 Fig ) . Therefore , we conclude that our recording conditions give rise to a fast hyperpolarization that predominantly reflects the synchronous recurrent inhibitory postsynaptic potential ( rIPSP ) . In control ( n = 10 ) and mmΔGCL cells ( n = 12 ) , action potential firing produced hyperpolarizations ( Fig 4B and 4C ) with indistinguishable amplitudes ( control: 6 . 35 ± 0 . 62 mV; mmΔGCL: 5 . 816 ± 0 . 47 mV; t = 0 . 71 , p > 0 . 05 ) . However , in NaV1 . 2ΔGCL cells , the hyperpolarization amplitudes were strongly decreased ( 2 . 87 ± 0 . 46 mV [n = 13]; control versus NaV1 . 2ΔGCL t = 4 . 73 , mmΔGCL versus NaV1 . 2ΔGCL t = 4 . 21 , p < 0 . 001 in both conditions ) . Moreover , we observed the same effect for trains of 5 and 20 action potentials ( Fig 4C; 5 action potentials—control: 6 . 68 ± 1 . 01 mV , mmΔGCL: 6 . 18 ± 0 . 47 mV , NaV1 . 2ΔGCL: 1 . 97 ± 0 . 39 mV; control versus mmΔGCL t = 0 . 56 p > 0 . 05 , control versus NaV1 . 2ΔGCL t = 5 . 28 p < 0 . 001 , mmΔGCL versus NaV1 . 2ΔGCL t = 4 . 95 p < 0 . 001; 20 action potentials—control: 8 . 15 ± 0 . 92 mV , mmΔGCL: 7 . 41 ± 0 . 81 mV , NaV1 . 2ΔGCL: 2 . 50 ± 0 . 41 mV; control versus mmΔGCL t = 0 . 74 p > 0 . 05 , control versus NaV1 . 2ΔGCL t = 5 . 53 p < 0 . 001 , mmΔGCL versus NaV1 . 2ΔGCL t = 5 . 04 p < 0 . 001 ) . Hence , given that 90% of the measured hyperpolarization is contributed by the rIPSP ( see above ) , knockdown of NaV1 . 2 in GCs highly significantly reduces dendrodendritic inhibition . While the action of ionotropic glutamate receptor blockers and GABAA receptors appears stronger than the effect of NaV1 . 2 knockdown ( S8 Fig ) , this is expected because NaV1 . 2 is only knocked down in approximately 55% of the GCs ( see Fig 3H and Discussion ) . To simulate odor-induced spike trains [4 , 43] , we stimulated MCs for different durations , with trains of action potentials at 100 Hz . As described above , in control cells , the hyperpolarization amplitude increased with the number of action potentials triggered in the MC ( Fig 4A and 4C ) , suggesting that brief bursts of action potentials facilitated GABA release from GCs . Upon NaV1 . 2 α-subunit knockdown in the GCL , the frequency-dependent facilitation of the hyperpolarization amplitude was abolished ( Fig 4C ) . Furthermore , action potentials in MCs remained unaffected ( Table 1 ) , supporting the notion that rAAV infection is restricted to the GCL . In conclusion , these results demonstrate that NaV1 . 2 VGSC α-subunits are required for synchronous release of GABA to produce dendrodendritic inhibition of MCs . The go/no-go operant conditioning paradigm [16 , 44] was used to address the role of GC NaV1 . 2 subunits in odor discrimination behavior ( Fig 5 ) . Three groups of mice—wild-type controls , mmΔGCL controls , and NaV1 . 2ΔGCL knockdown mice—were pretrained to discriminate the odorants cineol and eugenol ( cineol versus eugenol [CvE] ) for task habituation . Subsequently , the test odorants amyl acetate and ethyl butyrate ( amyl acetate versus ethyl butyrate [AAvEB] ) and their binary 6:4 mixtures ( 6v4/4v6 ) were tested . The three groups of mice learned equally well to discriminate the tested odorants with accuracies exceeding 90% ( Fig 5A and 5B ) . For the simple odors , mice had similar discrimination times in both rounds of testing ( Fig 5C; first round AAvEB—control: 295 . 2 ± 10 . 81 ms , mmΔGCL: 281 . 50 ± 16 . 78 ms , NaV1 . 2ΔGCL: 310 . 60 ± 14 . 55 ms , ANOVA , F = 1 . 04 , p = 0 . 37; second round AAvEB—control: 291 . 7 ± 7 . 032 ms , mmΔGCL: 284 . 5 ± 10 . 07 ms , NaV1 . 2ΔGCL: 316 . 10 ± 10 . 07 ms , ANOVA , F = 3 . 11 , p = 0 . 07 ) . However , NaV1 . 2ΔGCL mice showed a significant increase of the discrimination time for highly similar odorants ( Fig 5C ) , in comparison to control and mmΔGCL mice , in both rounds of testing ( first round 6/4v6/4—control: 333 . 40 ± 16 . 86 ms , mmΔGCL: 330 ± 12 . 16 ms , NaV1 . 2ΔGCL: 418 . 20 ± 27 . 65 ms , ANOVA , F = 6 . 23 , p = 0 . 008; control versus mmΔGCL t = 0 . 11 p > 0 . 05; control versus NaV1 . 2ΔGCL t = 3 . 00 p < 0 . 05; mmΔGCL versus NaV1 . 2ΔGCL t = 3 . 11 p < 0 . 005; second round 6/4v6/4—control: 301 . 9 ± 11 . 33 ms , mmΔGCL: 299 . 40 ± 20 . 64 ms , NaV1 . 2ΔGCL: 364 . 50 ± 20 . 64 ms , ANOVA , F = 5 . 58 , p = 0 . 01; control versus mmΔGCL t = 0 . 11 p > 0 . 05; control versus NaV1 . 2ΔGCL t = 2 . 84 p < 0 . 05; mmΔGCL versus NaV1 . 2ΔGCL t = 2 . 95 p < 0 . 05 ) . The intertrial interval , taken as a motivational indicator , was unchanged , suggesting constant motivational conditions ( Fig 5D ) . Our behavioral data demonstrate that lack of MC inhibition caused by NaV1 . 2 knockdown in GCs leads to a stimulus-dependent prolongation of odor discrimination time , but it does not change discrimination accuracy or learning . Our expression analysis shows that GCs exclusively express the sodium channel α-subunit NaV1 . 2 , a subunit predominantly expressed in axons [32] . At first glance , it may sound surprising that an axonless neuron expresses an axonal VGSC subunit in its dendrites . However , given that GCs use dendritic spines both to receive synaptic input and to generate output , the presence of the NaV1 . 2 subunit may imply that neurotransmitter release from spines utilizes the same mechanisms established in axon terminals . Another VGSC subunit , NaV1 . 6 , has been found in proximal and distal dendrites of hippocampal CA1 neurons [31] supporting a role in mediating backpropagating action potentials [45] . Lorincz and colleagues found NaV1 . 2 only within axons and presynaptic terminals but not in the dendritic domain . Hence , while dendritic expression of certain VGSC subunits appears to be a general phenomenon , we postulate that NaV1 . 2 occurs only in dendrites capable of neurotransmitter release . NaV1 . 2 is expressed in small clusters over the entire extent of a GC , including the heads of GC spines ( e . g . , Fig 2E ) . Cluster formation of NaV1 . 2 may be required to achieve a sufficiently large local current density for sodium channel activation and action potential initiation , similar to the situation found at nodes of Ranvier or the axon initial segment . VGSC clusters in the dendritic shaft may propagate the action potential along the dendritic tree of the GC , thereby generating a global action potential . Clusters situated in the GC spines may mediate a locally restricted action potential , as discussed in a separate section below . We suggest that under physiological conditions , the glutamatergic EPSP ( Fig 6A and 6B ) leads to a local depolarization that reaches the threshold of NaV1 . 2 activation ( Fig 6C ) , which then drives the initiation of an action potential in the spine . Subsequently , activated VDCCs , presumed to be localized in the GABAergic active zone , produce a nanodomain with high Ca2+ concentration during the repolarization phase of the action potential ( Fig 6D ) , when the driving force for Ca2+ is high , similar to the mechanisms known to exist in axon terminals [25 , 26 , 46] . In GC spines , this mechanism allows for fast and synchronous release of GABA and temporally well-resolved inhibition of MCs ( Fig 4 ) . Knockdown of NaV1 . 2 in GCs activity-dependently reduced the hyperpolarization amplitude in MCs by 60%–75% ( Fig 4B , S8 Fig ) . Considering that 10% of the control hyperpolarization is mediated by the afterhyperpolarization or other intrinsic MC conductances such as Ih , an rIPSP component of approximately 15%–30% remained ( see S8 Fig ) . However , if GABA release depends on a local all-or-none spine action potential , expression of the shRNA should result in a complete block of the rIPSP . This apparent discrepancy may be due to the incomplete transduction of GCs connected to the stimulated MC with rAAVs expressing shRNA and the incomplete knockdown of NaV1 . 2 ( Fig 3 ) . Thus , a mixed population of unperturbed GCs and GCs with NaV1 . 2 knockdown defines the strength of recurrent inhibition found in MC recordings presented here . Alternatively , after NaV1 . 2 knockdown , T-type Ca2+ channels may contribute to GABA release from GC spines and thereby generate some of the remaining MC hyperpolarization . The contribution of VGSC α-subunits other than NaV1 . 2 appears unlikely , given that none of these could be detected in GCs by immunohistochemistry ( S2 Fig ) . Hence , the strong reduction of the rIPSP amplitude indicates that an all-or-none action potential is required for fast and synchronous GABA release from GC spines . While we demonstrated a clustered distribution of NaV1 . 2 within GCs , with higher densities in the spine heads and dendrites compared to the soma , the NaV1 . 2 knockdown is likely to affect clusters in all locations of the GC . Hence , despite the highest density of clusters occurring in the spine heads , it remains difficult to know at which site action potentials get initiated: spine head or dendrite . Assuming the former raises the question whether this action potential remains a local event or propagates into the GC dendrite , thereby causing a global GC action potential . Recent work suggested that the high impedance of GC spine necks may insulate GC spines to function as “mini-neurons” or single processing units [47] . Such isolated autonomous compartments would mainly support local recurrent MC inhibition and may employ local plasticity mechanisms but would not allow dendritic integration of multiple glutamatergic inputs from different MCs . Yet the spine neck has little effect on EPSP propagation in the forward direction [48] . Such an impedance mismatch circuit allows coincidence detection within the dendrite and , in combination with dendritic inhibition , controls action potential generation in the dendritic tree . Hence , the design of the GC dendrite provides a solution to both ensure efficient local neurotransmitter release , plasticity , and dendritic computations of multiple synchronously active MC inputs . This NaV1 . 2-dependent model of reciprocal synaptic communication seemingly contradicts previous work claiming that dendrodendritic GABA release is governed by NMDAR-mediated currents that activate P/Q-type VDCCs [17 , 28] or that Ca2+ entry by NMDARs suffices to drive GABA release [18 , 19] , such that VGSCs are not required for GABA release at the dendrodendritic synapse [17] . These studies used Mg2+-free solutions containing TTX combined with very long MC current injections . Such unphysiological conditions will yield an unnaturally strong release of glutamate from MC dendrites , followed by a sustained Ca2+ inflow through NMDARs into the GC spine heads that will suffice to trigger slow , asynchronous GABA release . In contrast , synchronous GABA release physiologically triggered by MC action potentials occurs on the time scale of a few milliseconds ( see Fig 4B ) . Thus , a comparison of the studies mentioned above with our results needs to consider that two different modes of transmitter release are affected—asynchronous and synchronous release , respectively—each operating on different release sensors [49] . Electron microscopy has revealed a segregated organization of the postsynaptic density and the active zone harboring GABA-filled synaptic vesicles [11 , 13] , suggesting that NMDAR-mediated Ca2+ entry is not spatially coupled to the release machinery . However , maximal release rates only occur at high local Ca2+ concentrations [50 , 51] requiring a close spatial relationship ( a few nm ) between the VDCCs and the release machinery [26] . Thus , synaptic NMDAR activation will not suffice to generate a Ca2+ transient sufficient to release GABA from GC spines , but the unphysiological conditions described above may result in a Ca2+ domain reaching from the glutamatergic postsynaptic density to the GABAergic active zone , where it could trigger the release of GABA . Consistent with nanodomains mediating GABA release , EGTA perfused into GCs did not affect dendrodendritic inhibition [20] . An MC action potential triggers recurrent inhibition with a delay of a few milliseconds ( Fig 4 ) . NMDARs activate rather slowly , half-maximal currents are reached after approximately 3 ms , and the peak current is reached after approximately 10 ms . Therefore , NMDAR-mediated Ca2+ [52] inflow into the GC will only occur after the peak of the MC recurrent hyperpolarization is already declining , consistent with our notion that synchronous GABA release from GCs cannot be mediated by NMDARs under physiological conditions . What then are NMDARs needed for at the dendrodendritic synapse ? Our previous work [16] has shown that deletion of NMDARs in GCs slows the EPSP time course , reduces MC inhibition , and slows odor discrimination time , albeit to a much smaller extent than reported here for deletion of NaV1 . 2 . These results suggest that NMDARs are not essential for synchronous GABA release from GC spines under physiological conditions . Alternatively , NMDARs may enhance the local depolarization after AMPAR-mediated relief of Mg2+ block and may contribute a Ca2+ signal that could modulate the release machinery via high-affinity Ca2+ sensors or exert actions via other Ca2+-dependent pathways . This Ca2+ signal could also be detected by high-affinity release sensors and directly translated into long-lasting asynchronous release [49] , consistent with the slow time course of the IPSC reported by previous studies [18 , 19] . Due to the long-lasting activation ( hundreds of ms ) of NMDARs caused by a single synaptic glutamate transient [53] , repetitive glutamatergic input may recruit these mechanisms even more strongly and may transiently increase GABA release ( see Fig 4A ) . Alternatively , NMDARs may establish synaptic plasticity mechanisms and contribute to local signals that do not reach VGSCs’ activation threshold . GCs have been shown to fire action potentials in response to odorants [2 , 4 , 54 , 55] . Moreover , awake behaving animals show much stronger GC activity with low temporal structure compared to anesthetized animals [55–57] , implying that GCs modulate MC activity through extended lateral interactions independent of the respiratory cycle [55] . These observations contradict the claim of the GC spine being a “mini-neuron” [47] , for which consequently the GC would appear rather silent , with dendritic spines working independently . We propose that EPSPs generated in several spines within close spatial proximity may depolarize the dendrites to firing threshold , causing the GC to fire a global action potential . In this context , T-type Ca2+ channels may play an important role in amplifying small glutamate receptor–mediated depolarizations [21] in gemmules and thereby generate a Ca2+ signal in the neighboring compartments of the dendritic tree . Our model of GC function allows establishing rapid and synchronous local recurrent inhibition [17 , 58] in parallel to global dendritic computations , resulting in lateral inhibition [18 , 21 , 22] . Yet these events cannot occur segregated or in a gradual fashion , as previously proposed [17 , 22 , 58 , 59] , because once the depolarization reaches action potential threshold at the dendrite , a global action potential will be elicited . GC inhibition can shunt action potential propagation and consequently isolate dendritic compartments to compute stimuli from different sources . Indeed , GC dendrites can respond differently to odors than the soma [57] because of GC inhibitory mechanisms that might play a pivotal role to compartmentalize GC responses to a given stimulus . At the spine level , A-type voltage-gated potassium channels ( VGKCs [20] ) will dampen the developing interspike depolarization to temporally space successive action potentials more widely [60] . The A-type current may lead to the low spiking rate observed in GCs [55 , 61 , 62] . Altogether , we suggest that GCs use VGSCs to drive GABA release in a fast and synchronous manner at the GC spine and that generation of dendritic spikes at multiple dendritic domains allows for coincidence detection and dendritic integration . The viral approach used in this study is not inherently selective for GCs . The parameters of stereotaxic injection can effectively limit viral spread and transduction to the GCL of the OB [16 , 38] . However , in addition to GCs , the GCL also contains deep short axon ( dSA ) cells , a heterogeneous population of cells classified based on their soma size , location , and morphology ( reviewed in [39] ) . More than 99% of the cells analyzed in our study had dendrites of typical morphology reaching into the EPL and soma sizes of less than 10 μm ( S6 Fig ) . On the functional level , knockdown of Na channels in dSA cells would cause a loss of dSA activity and would , based on their inhibitory nature and their synaptic connections with GCs [39 , 41] , yield GCs more excitable . This would in turn increase MC inhibition ( see [38] ) . However , we found the opposite , indicating that dSA cells were not perturbed in a physiologically relevant manner . GCL-specific deletion of NaV1 . 2 channels resulted in a stimulus-dependent slowing of the time needed for highly accurate odor discrimination , not interfering with odor discrimination learning and accuracy ( Fig 5 ) . These results recapitulate our previous observations after GCL-specific deletion of the NMDA-type glutamate receptor subunit 1 ( GluN1 ) [16] . Nevertheless , NaV1 . 2 deletion had a much stronger effect on the rIPSP amplitude , indicating that AMPARs remaining in the absence of NMDARs are sufficient to maintain dendrodendritic inhibition at a level stronger than after NaV1 . 2 knockdown . Interestingly , both GluN1 deletion and NaV1 . 2 knockdown resulted in a stimulus-dependent phenotype: only discrimination of highly similar binary mixtures , but not of dissimilar stimuli , was affected . In conclusion , under physiological conditions , Nav1 . 2 activation is crucial for inhibitory interactions of GCs and MCs , underlying an important molecular mechanism for the OB to enhance discrimination of highly similar activity patterns . What is the mechanistic link between altered MC inhibition and odor discrimination behavior ? This and our previous work [16 , 38] suggest a correlation between the strength of dendrodendritic inhibition and discrimination time: stronger inhibition will accelerate odor discrimination and vice versa . On the cellular level , the simplest mechanism causing a shift in discrimination time could be the time required to build up a hypothetical level of recurrent MC inhibition sufficient to discriminate similar stimuli [16] . However , understanding how network connectivity , neuronal ensemble formation , decision-making , and coupling of olfactory and motor areas give rise to odor discrimination behavior remains a challenge to be addressed . Mice used in this study were handled in agreement with the European FELASA guidelines , and all procedures were approved by the national authorities , Regierungspraesidium Karlsruhe , Germany , under the approved protocol number 35–9185 . 81/G-100/09 . All animal care procedures were conducted in agreement with the European Directive 2010/63 , at the Heidelberg University , under the supervision of the Interfakultaere Biomedizinische Forschungseinrichtung . In all experiments , mice were housed in standard cages , in a constant day–night cycle ( 12 hours–12 hours ) and in a temperature- ( 22 ± 2 °C ) and humidity- ( 60% ± 4% ) controlled environment . Individuals used for behavior were kept in an inverted light cycle , and the experiments were performed in the night period . All mice ( strain C57Bl6 ) were purchased from Charles River . To assess VGSC expression in the main OB , RNA was extracted from whole brain , whole OB , heart , and muscle using TRIzol ( Ambion , cat . # 15596–018 ) . For each extract , a cDNA library was created by inverse transcription using SuperScript II kit ( Invitrogen , cat . # 18064–014 ) . PCRs were made using primers previously published [63 , 64] for the different subunits . Primer specificity was tested using whole brain , heart , and muscle cDNA . All known mRNAs could be detected by rT-PCR . Due to the high molecular weight of the VGSCs subunits , western blots were done following the protocol of Fairbainks [65 , 66] . Antibodies against NaV1 . 1 ( AB_2238842 ) , NaV1 . 2 ( AB_2184197 ) , NaV1 . 6 ( AB_2184197 ) , and NaV1 . 7 ( AB_2184355 ) were purchased from Neuromab ( Antibodies , United States of America ) . The antibody against the NaV1 . 3 ( pab0279-P ) subunit was purchased from Covalab ( France ) . These antibodies were also used in the immunohistochemistry procedures . The design of shRNA molecules was performed using the InvivoGen algorithm ( www . sirnawizard . com ) . To restrict the number of candidate molecules , the following constrains were used: ( 1 ) sequences containing TTATT were discarded ( known to induce immune response ) , ( 2 ) sense and antisense oligos should contain 21 nts length each , ( 3 ) C-G amount <70% . The sequence of the shRNA loop was TAATATTAT . Specificity was controlled by determining the E value with BLAST ( blast . ncbi . nlm . nih . gov ) . Four molecules ( sense sequences: shRNA#5: GAA AGC AAT CTC TCG GTT CAG; shRNA#14: GTT GGA AGA CCC TAC ATC AAG; shRNA#22: GAT GGA AAC GGG ACG ACC AGT; shRNA#44: GTG GAC CTC CCG ATT GTG ACC ) were selected and tested for their efficacy in reducing Na+ currents in GCs ( S4 Fig ) . For the most effective shRNA , a shRNAmm was produced by changing each third base of the original oligo ( mmhRNA#14: GTA GGG AGT CCG TAG ATG AAC ) . To avoid possible interferences with other cellular mRNAs , the specificity of the mismatch molecule was predicted using BLAST . Mice , 21 days old , were stereotaxically injected using a stereotax ( myNeurolab , USA ) . The rAVV chimeric vectors ( 1:1 ratio of AAV1 and AAV2 capsid proteins ) carrying rAAV-specific expression cassettes ( pAM plasmid ) were injected in the GCL of the OB , as previously described [16] . For 3D-immunohistochemistry , mGFP expression under the chicken beta actin ( CAG ) promoter was used . For in vitro whole-cell electrophysiology and behavioral experiments , control plasmids expressed eGFP under the CAG promoter . Expression of shRNAs was driven by the U6 promoter with bicistronic expression of eGFP under the control of the CAG promoter . This approach combines viral-mediated labeling of cells with immunohistochemistry [33] . Fixed tissue sections containing neurons labeled with mGFP for precise detection of even thin processes ( typically not visible with soluble cytoplasmic eGFP expression ) were treated with the antibody of interest . Dual-color confocal image stacks were acquired in the serial scanning mode . Using ImageJ or AMIRA software , the 3D morphology template of the cell , as delineated by mGFP , was used to excise the immunohistochemistry signals residing within the cell of interest . To reconstruct full GC cell bodies or EPL dendrites , 50–100 consecutive confocal image frames were used . The result was rendered in 3D and shows the distribution of the immunosignal within the labeled cell . Practically , 21-day-old wild-type mice were stereotaxically injected with rAAV-mGFP , and after 3–4 weeks of expression , mice were transcardially perfused using 4% paraformaldehyde ( PFA ) . The brains were removed , and free-floating 50-μm-thick slices of the OB were prepared using a vibratome . Afterwards , the slices were incubated for 45 minutes at room temperature in vehicle buffer ( 10% normal goat serum , 1% bovine serum albumin fraction V , 0 . 3% Triton X-100 , in 1x PBS , pH = 7 . 4 ) with 0 . 1% cold fish gelatin ( blocking solution ) . The antibody against the NaV1 . 2 subunit was carried in vehicle ( 1:1 , 000 dilution ) and incubated overnight at 4 °C . After the third wash in vehicle ( 10 minutes each wash step ) , the alexa dye conjugated secondary antibody ( 1:1 , 000 in vehicle; life technologies cat . #: A-21244 ) was incubated for 90 minutes at room temperature . The slices were washed three times in 1x PBS , pH = 7 . 4 , and mounted on coverslips using Moviol . All incubation steps were performed under gentle agitation using a horizontal rocking shaker ( neoLab , Germany ) . Images were acquired in a confocal microscope ( Leica SP5 , Leica , Germany ) using a 63× glycerol immersion objective ( NA = 1 . 3 ) . OB horizontal 300-μm-thick slices were prepared from mice , 6–10 weeks old , using a vibratome ( Leica ) while submersed in ice-cold oxygenated slicing solution ( in mM ) : 125 NaCl , 2 . 5 KCl , 25 NaHCO3 , 1 . 25 NaH2PO4 , 3 myo-Inositol , 2 Na-pyruvate , 0 . 4 ascorbic acid , 0 . 1 CaCl2 , 3 MgCl2 , 25 glucose . Slices were transferred and incubated for about 30 minutes in a 37 °C warm oxygenated bath solution ( in mM ) : 125 NaCl , 2 . 5 KCl , 25 NaHCO3 , 1 . 25 NaH2PO4 , 2 CaCl2 , 1 MgCl2 , 25 glucose . This bath solution was continuously aerated with carbogen and used for all recordings at a temperature of 33–35 °C . Whole-cell recordings were established using an EPC9 amplifier ( HEKA , Lambrecht , Germany ) . To evaluate Na+ currents in GCs , voltage-clamp recordings were performed using pipettes with resistances ranging 3 . 5–4 . 5 MΩ , filled with ( in mM ) 130 CsCl , 4 TEA-Cl , 10 Na2-phosphocreatine , 10 HEPES , 5 EGTA , 4 Mg-ATP , 0 . 3 Na-GTP , 2 Ascorbate , pH = 7 . 2 with CsOH . TEA ( 10 mM ) was added to the bath solution . Current-clamp recordings were established from MCs to access recurrent inhibition . To induce a single action potential , a 3-ms current pulse was used . To induce 100-Hz stimulation , 3-ms current pulses interleaved with 8 ms of no current injection were used . Pipettes had resistances of 3–4 MΩ , filled with 135 K-gluconate , 10 HEPES , 10 Na2-phosphocreatine , 4 MgATP , 4 KCl , 0 . 3 Na-GTP , pH = 7 . 2 adjusted with KOH . All recordings were performed using a micro-salt bridge in the electrode attached to the pipette holder [67] . To access cell morphology , the dye alexa 594 hydrazide ( 10 μM; Molecular Probes , cat . # A10438 ) was added routinely in the intracellular solutions in all experiments . For action potential recordings in GCs , the same solutions were used . Current injections of 1-ms duration and 100-ms duration were used to analyze action potential firing in GCs . For the behavioral experiments , animals were kept separated in macrolon type II cages in a temperature- and humidity-controlled environment operated under a reverse light cycle ( 12 hours–12 hours ) . Experiments were conducted in a dark room during the night period directly adjacent to the animal room . The behavioral protocol was described previously [16 , 44] . The behavioral pretraining started 4–6 weeks after the surgery and 2–3 days before the animals were kept under water restriction ( by periods no longer than 12 hours ) . The weight of the animals was strictly monitored and kept >85% of the initial body weight . The pretraining took 3–5 days , and the behavioral training took usually no longer than 8 weeks . The control of the eight-channel semiautomated olfactometers [68] ( Knosys , Washington ) and data acquisition were carried out with custom programmed software ( S1 File ) written in Igor Pro 6 ( Wavemetrics ) . Odor presentation tasks began after all animals finished the pretraining successfully . All odors were diluted to 1% in mineral oil and further air-diluted at a 1:20 ratio in the olfactometers . Odors were presented in a fully randomized way , with no more than 5 consecutive trials having the same stimulus . Bias toward any of the odors presented was avoided by counterbalancing between animals . The odor pair CvE was used for task habituation; usually , 400 trials sufficed to achieve an accuracy >80% . Test odors ( AAvEB and 6v4/4v6 ) were used to determine the reaction times . To study the localization and distribution of the NaV1 . 2 subunit in GCs , the ImageJ software was used for data visualization . Electrophysiology and behavior data analyses ( S1 File ) were performed using custom software written in Igor Pro6 ( Wavemetrics ) . All data are presented as mean ± SEM . ANOVA refers to one-way ANOVA , and t-values were derived from the Bonferroni multiple comparison test , except when otherwise denoted . Statistical analyses were performed in Prism 5 ( GraphPad Software ) .	In axonal nerve terminals , neurotransmitter release is triggered by a localized Ca2+ nanodomain generated by voltage-gated calcium channels in response to an action potential , which in turn is mediated by voltage-gated sodium channels . Dendritic neurotransmitter release has been thought to work differently , mainly depending on Ca2+ entering directly through N-methyl-D-aspartate ( NMDA ) receptors , a subtype of ligand-gated ion channel . To further investigate how dendritic neurotransmitter is released , we studied granule cells in the olfactory bulb of mice , which establish inhibitory dendrodendritic synapses with mitral cells . We show that granule cells express voltage-gated sodium channels predominantly localized in dendrites and spines . Down-regulation of these channels precludes action potential firing in granule cells and strongly reduces mitral cell inhibition . Behaviorally , these mice require more time to discriminate highly similar odorants at maximal accuracy . Therefore , the inhibition of mitral cells relies on neurotransmitter released from the dendrites of granule cells by a mechanism that resembles axonal neurotransmitter release much more than previously thought .	[ "Abstract", "Introduction", "Results", "Discussion", "Materials", "and", "methods" ]	[ "action", "potentials", "medicine", "and", "health", "sciences", "neurochemistry", "membrane", "potential", "green", "fluorescent", "protein", "electrophysiology", "neuroscience", "histochemistry", "and", "cytochemistry", "techniques", "luminescent", "proteins", "cognition", ...	2018	Axonal sodium channel NaV1.2 drives granule cell dendritic GABA release and rapid odor discrimination
Porcine Reproductive and Respiratory Syndrome ( PRRS ) is a panzootic infectious disease of pigs , causing major economic losses to the world-wide pig industry . PRRS manifests differently in pigs of all ages but primarily causes late-term abortions and stillbirths in sows and respiratory disease in piglets . The causative agent of the disease is the positive-strand RNA PRRS virus ( PRRSV ) . PRRSV has a narrow host cell tropism , limited to cells of the monocyte/macrophage lineage . CD163 has been described as a fusion receptor for PRRSV , whereby the scavenger receptor cysteine-rich domain 5 ( SRCR5 ) region was shown to be an interaction site for the virus in vitro . CD163 is expressed at high levels on the surface of macrophages , particularly in the respiratory system . Here we describe the application of CRISPR/Cas9 to pig zygotes , resulting in the generation of pigs with a deletion of Exon 7 of the CD163 gene , encoding SRCR5 . Deletion of SRCR5 showed no adverse effects in pigs maintained under standard husbandry conditions with normal growth rates and complete blood counts observed . Pulmonary alveolar macrophages ( PAMs ) and peripheral blood monocytes ( PBMCs ) were isolated from the animals and assessed in vitro . Both PAMs and macrophages obtained from PBMCs by CSF1 stimulation ( PMMs ) show the characteristic differentiation and cell surface marker expression of macrophages of the respective origin . Expression and correct folding of the SRCR5 deletion CD163 on the surface of macrophages and biological activity of the protein as hemoglobin-haptoglobin scavenger was confirmed . Challenge of both PAMs and PMMs with PRRSV genotype 1 , subtypes 1 , 2 , and 3 and PMMs with PRRSV genotype 2 showed complete resistance to viral infections assessed by replication . Confocal microscopy revealed the absence of replication structures in the SRCR5 CD163 deletion macrophages , indicating an inhibition of infection prior to gene expression , i . e . at entry/fusion or unpacking stages . Porcine reproductive and respiratory syndrome ( PRRS ) is one of the most economically important infectious diseases affecting pigs worldwide . The “mystery swine disease” was first observed almost simultaneously in North America and in Europe in the late 1980s [1 , 2] . The causative agent of PRRS was identified to be a virus later named PRRS virus ( PRRSV ) . Infected pigs may present with symptoms involving inappetence , fever , lethargy , and respiratory distress . However , the most devastating effects of PRRSV infection are observed in young piglets and pregnant sows . In pregnant sows an infection with PRRSV can cause a partial displacement of the placenta , leading to full abortions or to death and mummification of fetuses in utero [3] . Late-term abortions occur in up to 30% of infected sows with litters containing up to 100% stillborn piglets . Live-born piglets from an antenatal infection are often weak and display severe respiratory symptoms , with up to 80% of them dying on a weekly basis pre-weaning [4 , 5] . Young piglets infected with PRRSV often display diarrhea and severe respiratory distress caused by lesions in the lung . In pre-weaned piglets the infection may be transmitted via the mammary gland secretions of an infected sow [6] . At this age the infection has a fatal outcome in up to 80% of animals . After weaning mortality rates reduce , but continued economic losses due to reduced daily gain and feed efficiency are often observed [4 , 7 , 8] . Due to reduction or loss of pregnancies , death in young piglets , and decreased growth rates in all PRRSV infected pigs it is estimated that more than $650m are lost annually to pork producers in the United States alone [9 , 10] . PRRSV is an enveloped , plus-strand RNA virus belonging to the Arteriviridae family in the order Nidovirales [11 , 12] . The PRRSV genome ( ~15 kb ) encodes at least 12 non-structural and seven structural proteins . The viral RNA is packaged by the nucleocapsid protein N , which is surrounded by the lipoprotein envelope , containing the non-glycosylated membrane proteins M and E , as well as four glycosylated glycoproteins GP2 , GP3 , GP4 , and GP5 , whereby GP2 , 3 , and 4 form a complex [13–17] . PRRSV has a very narrow host range , infecting only specific subsets of porcine macrophages [18–20] . It is unknown yet how widespread PRRSV infections are within the superfamily of the Suoidea . Whereby European wild boars have been shown to act as a reservoir for PRRSV [21] , little is known about infection in African suids , such as bushpigs and warthogs . In vitro virus replication is supported by the African Green Monkey cell line MARC-145 . Entry of PRRSV into macrophages has been shown to occur via pH-dependent , receptor mediated endocytosis [22 , 23] . Various attachment factors and receptors have been indicated to be involved in the PRRSV entry process ( reviewed in [24] ) . Heparan sulphate was identified early as an attachment factor of the virus [25–27] . In vitro infection of pulmonary alveolar macrophages ( PAMs ) but not MARC-145 cells was shown to be inhibited by an antibody targeting CD169 ( sialoadhesin ) , a lectin expressed on the surface of macrophages [28] . Overexpression of CD169 in previously non-permissive PK-15 cells showed internalization but not productive replication of PRRSV [29] . Finally , an in vivo challenge of genetically modified pigs in which the CD169 gene had been knocked out revealed no increased resistance to PRRSV infection , suggesting that CD169 is an attachment factor that is not essential for PRRSV infection [30] . Even though cell surface protein expression is a major determinant of PRRSV binding and internalization , there appears to be a redundancy amongst cell surface attachment factors , with the potential for additional , as yet unidentified receptors , being involved [31] . The scavenger receptor CD163 , also known as haptoglobin scavenger receptor or p155 , is expressed on specific subtypes of macrophages and has been identified as a fusion receptor for PRRSV . The extracellular portion of CD163 forms a pearl-on-a-string structure of nine scavenger receptor cysteine-rich ( SRCR ) domains and is anchored by a single transmembrane segment and a short cytoplasmic domain [32] . CD163 has a variety of biological functions , including mediating systemic inflammation and the removal of hemoglobin from blood plasma ( reviewed in [33 , 34] ) . Overexpression of CD163 renders non-susceptible cells permissive to PRRSV infection [35] , whereby it was found that CD163 does not mediate internalization but is crucial for fusion [36] . The transmembrane anchoring and an interaction with the SRCR domain 5 ( SRCR5 ) of CD163 were found to be essential for successful infection with PRRSV [34 , 35] . Recent in vivo experiments with CD163 knock-out pigs confirmed that CD163 is required for PRRSV infection [37] . However , as CD163 has important biological functions the complete knockout could have a negative physiological impact on the animal , particularly with respect to inflammation and/or infection by other pathogens . Interestingly , whereas all the other eight SRCR domains have been shown to be involved in different biological functions , no specific role has been associated with SRCR5 , other than in PRRSV infection [34] . Therefore , this study aimed to generate pigs with a defined CD163 SRCR5 deletion and to assess the susceptibility of macrophages from these pigs to PRRSV infection . The CD163 gene is not correctly represented in the current pig reference genome sequence ( Sscrofa10 . 2 ) [38] . Through targeted sequencing we established a detailed model of the porcine CD163 locus ( S1 File ) . Briefly , CD163 is encoded by 16 exons with exons 2–13 predicted to encode the SRCR domains of the protein [39] . Interestingly , SRCR5 is predicted to be encoded by one single exon , namely exon 7 ( Fig 1A ) . Thus , an editing strategy was developed to excise exon 7 using the CRISPR/Cas9 genome editing system [40 , 41] . A combination of two guide RNAs , one located in the intron 5’ to exon 7 and one in the short intron between exons 7 and 8 was predicted to generate a deletion of exon 7 , whilst allowing appropriate splicing of the remaining exons . Due to the short length of the intron between exons 7 and 8 ( 97 bp ) only one suitably unique targeting sequence ( crRNA ) with a corresponding protospacer adjacent motif was identified . Three candidate crRNA sequences were selected in the immediate upstream area of exon 7 . All four sequences were assessed in vitro for cutting efficiency by transfection of porcine kidney PK15 cells with a plasmid based on px458 [42] encoding the complete single guide sequence ( sgRNA ) , driven by the hU6 promoter , and a CAG promoter driving NLS-Cas9-2A-GFP . Transfected cells were isolated by fluorescence activated cell sorting ( FACS ) for GFP and cutting efficiency at the target site was assessed using a Cel1 surveyor assay . Three out of four guides were shown to direct cutting of DNA as anticipated ( 2 upstream and one downstream of exon 7 ) . Following double transfection assay and subsequent PCR analysis it was found that only the combination of guides SL26 and SL28 generated the desired exon 7 deletion in the CD163 gene ( Fig 1B ) . Based on these results the guide combination of sgSL26 and sgSL28 was used for in vivo experiments . sgRNAs SL26 and SL28 were microinjected together with mRNA encoding the Cas9 nuclease into the cytosol of zygotes ( Fig 1C ) . Editing efficiency was assessed in a small number of injected zygotes by in vitro culture to the blastocyst stage , genomic DNA extraction , whole genome amplification and PCR amplification across exon 7 . The analysis revealed that two out of 17 blastocysts contained a deletion of the intended size and Sanger sequencing confirmed the deletion of exon 7 . Edited blastocyst B2 showed a clean deletion and subsequent re-ligation at the cutting sites of sgSL26 and sgSL28 , whilst edited blastocyst B14 showed that in addition to the intended deletion there was also a random insertion of 25 nucleotides at the target site ( S1 Fig ) . None of the full length PCR products showed nucleotide mismatches at either cutting site in a T7 endonuclease assay . The overall editing rate in the blastocysts was 11 . 7% . To generate live pigs , 24–39 zygotes injected with sgSL26 , sgSL28 , and Cas9 mRNA were transferred into the oviduct of recipient gilts . A total of 32 live piglets were born and genotyping of ear and tail biopsies revealed that four of the piglets had an exon 7 deletion , corresponding to 12 . 5% of the total . In addition to the intended deletion of exon 7 , three out of the four animals showed insertions of new DNA at the target site probably as a consequence of non-homologous end joining ( NHEJ ) repair . Pig 347 showed a 2 bp truncation at the sgSL26 cutting site and a 66 bp insertion between the cutting sites , pig 346 showed a deletion of 304 bp after the cutting site of sgSL26 , and pig 310 showed a short 9 bp insertion at the cutting sites ( S2 Fig ) . Pig 345 was found to have a precise deletion of exon 7 without insertion or deletion of random nucleotides at the cut sites ( S2B & S2D Fig ) . Interestingly PCR amplification indicated that pigs 310 , 345 , and 347 were all mosaic for an editing event , with pig 310 having a low frequency heterozygous ( one allele edited ) compared to unedited cells , whilst pigs 345 and 347 have both homozygous ( both alleles edited ) and heterozygous cell types ( S2A & S2C Fig ) . To generate fully homozygous and heterozygous pigs , 310 was mated with 345 . This mating yielded a litter of 6 heterozygous , 2 biallelic/homozygous CD163 SRCR5 deletion ( ΔSRCR5 ) , and 4 wild type CD163 piglets ( S3 Fig ) . Sequencing of the animals revealed all the heterozygotes to have inherited their edited allele from 345 . Pig 629 was found to be biallelic for the exon 7 deletion with one allele carrying the genotype of 345 and the other allele the one from 310 . Interestingly 630 was found to be homozygous for the edited allele with the 9 bp insert between the cutting sites of sgSL26 and sgSL28 as found in the 310 founder / parent ( Fig 2 ) . We conclude that this homozygous state has arisen from a gene conversion event in the zygote . Animals 627 , 628 , 629 , 630 , 633 , and 634 were selected for further analysis , representing the various genotypes ( wild type , heterozygous , and biallelic/homozygous ) and genders . Growth rates of both ΔSRCR5 and heterozygous animals were comparable to wild type animals ( Fig 2 ) . Blood samples were taken from all six animals at 10 weeks of age and analyzed by a full blood count conducted by the diagnostics laboratory at the Royal ( Dick ) School of Veterinary Sciences , University of Edinburgh . The blood counts of all animals were within reference values ( S1 table ) . Size , stature and other morphological features of ΔSRCR5 and heterozygous pigs were comparable to their wild type siblings ( Fig 3A ) . At 8 weeks of age , pulmonary alveolar macrophages ( PAMs ) were isolated from all six animals by bronchoalveolar lavage ( BAL ) . DNA was extracted from the PAMs and analyzed by PCR and Sanger sequencing . The PAM genotype confirmed the results obtained from the ear biopsies; 628 and 633 were wild type , 627 and 633 heterozygous , and 629 and 630 ΔSRCR5 , respectively . Sequencing of PCR products confirmed that all editing events had resulted in complete deletion of exon 7 . Whilst pigs 627 and 633 had a clean deletion of exon 7 with precise religation at the sgSL26 and sgSL28 cutting sites in one allele , 629 had one allele with a clean deletion and one allele with a 9 bp insertion between the sites , and pig 630 had both alleles with the 9 bp insertion ( Fig 3B ) . RNA was extracted from the PAMs , converted into cDNA using oligo ( dT ) primed reverse transcription , amplified by PCR and analyzed by Sanger sequencing . PCR products spanning exons 4 to 9 showed the expected 315 bp deletion in both heterozygous and ΔSRCR5 animals ( Fig 3C ) . A third fragment situated between the full length and exon 7 deletion band in 627 and 634 was confirmed to be a hybrid of the full length and the exon 7 deletion fragment . This shows that deletion of exon 7 has not disrupted the use of the correct splice acceptor site of exon 8 . Expression of CD163 protein was assessed by western blot of PAM lysate . The wild type pigs 628 and 633 expressed the full length protein with a predicted size of 120 kDa but is described to run at roughly 150 kDa [43] , likely due to glycosylation , whereby a protein band at roughly 100 kDa may indicate the expression of another isoform , which could correspond to the described human isoform CRA_a or CRA_b ( GenBank references EAW88664 . 1 and EAW88666 . 1 ) . Heterozygous animals 627 and 634 express both the full-length and the ΔSRCR5 protein ( Fig 3D ) . The band of the full-length protein is clearly stronger , indicating either higher expression of the full-length gene or increased binding of the full-length protein by the polyclonal CD163 antibody used in this study . To further examine this , gene expression was quantified by RT-qPCR on RNA extracted from PAMs and normalized to β-actin expression , demonstrating no significant difference in total CD163 mRNA expression between wild type , heterozygous and ΔSRCR5 animals ( Fig 3E ) . To assess the differentiation potential of monocytes into CD163-expressing macrophages we isolated peripheral blood monocytes ( PBMCs ) from whole blood and then differentiated them into macrophages by CSF1-induction for seven days . Expression of macrophage specific markers was assessed by immunofluorescence labelling and FACS analysis . CD14 and CD16 levels are clear indicators of the differentiation of peripheral blood monocytes with levels of both increasing significantly upon differentiation [44 , 45] . CD14/CD16 staining of the PMMs from the ΔSRCR5 , heterozygous , and wild type animals were all within the previously observed and documented levels [46] , with difference being observed between the various genotypes ( Fig 4A ) . CD172a , also known as SIRP α , is expressed at high levels on both monocytes and macrophages [45] and was expressed at high levels in cells from all animals . CD169 , described as an attachment factor for PRRSV [29] , is not expressed in monocytes but is highly expressed in tissue macrophages [47] and was expressed at expected levels in cells from our animals ( Fig 4B ) . An additional differentiation marker found to be expressed on PMMs is SWC9 , also known as CD203a , as well as the putative PRRSV attachment factor CD151 [48 , 49] . Expression of SWC9 highlighted the full differentiation of the PMMs . CD151 expression together with the previously shown CD169 expression demonstrated that both putative PRRSV attachment factors or receptors are still expressed on macrophages from ΔSRCR5 animals ( Fig 4C ) . As in humans , expression of CD163 in pigs is restricted to monocytes and macrophages . CD163 is expressed at high levels in tissue macrophages , but at low levels in blood monocytes and in bone marrow-derived macrophages [50] ( porcine macrophage markers are reviewed in [51] ) . Both the wild type and the ΔSRCR5 CD163 were recognized on the surface of the PAMs ( Fig 4D ) . This indicates that the ΔSRCR5 version of CD163 is likely to be properly folded as the clone 2A10/11 antibody only recognizes the protein in a non-reduced , native conformation . The medians of CD163 fluorescence intensity of pigs 628 , 633 , 627 , 634 , 629 , 630 were 23 . 3 , 16 . 7 , 18 . 3 , 16 . 5 , 18 . 8 , and 17 . 2 , respectively , with the isotype control medians ranging from 1 . 88–3 . 79 . Overall , PBMCs isolated from all animals , independent of their genotype were shown to be fully differentiated into PMMs upon CSF1 induction . They all expressed macrophage-specific surface markers , including CD169 , CD151 , and CD163 , which have putative functions in PRRSV entry . To confirm the results from the in vitro differentiation PAMs were isolated by BAL and characterized for the expression of macrophage-specific surface proteins CD14 , CD16 , CD169 , CD172a , and CD163 as described above . CD14/CD16 staining of the PAMs from the ΔSRCR5 , heterozygous , and wild type animals were all within the previously observed and documented levels [46] ( S4A Fig ) . Also CD169 and CD172a were within expected levels , confirming full differentiations ( S4B Fig ) . The medians of CD163 fluorescence intensity of pigs 628 , 633 , 627 , 634 , 629 , 630 were 35 . 9 , 22 . 7 , 26 . 4 , 24 . 4 , 17 . 9 , and 26 . 7 , respectively , with isotype control medians ranging from 2 . 13–3 . 84 ( S4C Fig ) . This indicates slightly higher expression levels of CD163 on PAMs compared to PMMs . Overall , PAMs isolated from all animals , independent of their genotype were shown to be fully differentiated and to express macrophage-specific surface markers , including CD169 and CD163 , which have implicated functions in PRRSV entry . PRRSV has two different genotypes with distinct geographic distribution , with genotype 1 being found primarily in Europe and Asia and genotype 2 in the Americas and Asia . The two genotypes show differences in both antigenicity and severity of pathology and show an evolutionary divergence of >15% on a whole genome scale and ∼40% on the nucleotide level between them ( reviewed in [52] ) . Genotype 1 can be further divided into three subtypes , based on the ORF7 sequence and geographical distribution , whereby subtype 1 is pan-European whilst subtypes 2 and 3 are currently limited to Eastern Europe [53] . Here we tested all genotype 1 subtypes of PRRSV , represented by subtype 1 strain H2 ( PRRSV H2 ) [54] , subtype 2 strain DAI ( PRRSV DAI ) [55] , and subtype 3 strain SU1-Bel ( PRRSV SU1-Bel ) [56] , originally isolated from the UK , Lithuania , and Belarus , respectively . PAMs were infected at an MOI = 1 in a single-round infection . 19 hours post inoculation ( hpi ) the cells were harvested and stained with a FITC-labelled antibody against PRRSV-N protein . Infection levels were assessed by FACS analysis . All three virus subtypes resulted in infection levels of 40–60% in wild type and heterozygous animals , with more than 98% of infected cells being classified as CD163 positive . A slightly higher , statistically significant infection was observed in heterozygous animals infected with PRRSV H2 and DAI . The reason for this is unclear , but may reflect either altered CD163 protein expression profile in heterozygous animals or other , as yet unidentified , genetic properties . By contrast , cells from both ΔSRCR5 animals ( 629 and 630 ) were found to be highly resistant to infection in this assay ( Fig 5A–5C ) . A second assay was performed to assess whether virus could replicate in PAMs then infect neighboring cells in a multiple-round infection time course . Cells were inoculated at MOI = 0 . 1 and supernatant samples collected at indicated time points . Viral RNA was extracted from the supernatants and analyzed by RT-qPCR . For PRRSV H2 and SU1-Bel specific probes and primers against ORF7 were employed . To quantify PRRSV DAI vRNA specific primers against ORF5 and BRYT green dye binding were used due to the limited genome information available on this strain . All wild type and heterozygous animals replicated the virus to similar levels . Virus levels started to rise by 12 hpi and increased exponentially up to 36 hpi when they plateaued . PRRSV SU1-Bel levels reached their plateau at 48 hpi . The quantification limit of the RT-qPCR corresponded to a CT value of 35 , which corresponded to 1E4 TCID50/ml for PRRSV H2 , 1E3 TCID50/ml for PRRSV DAI , and 5E3 for PRRSV SU1-Bel . vRNA levels in supernatants from ΔSRCR5 PAMs in this multiple round infection did not increase above the quantification limit ( Fig 5D–5F ) . In order to assess whether infectious virions were produced a TCID50 assay was conducted on supernatant collected at 48 hpi , when all three subtypes had reached a plateau . Serial dilutions were started at a 1:10 dilution , corresponding to a detection limit of 63 TCID50/ml . Virus produced from PAMs of wild type or heterozygous origin was infectious and levels measured were comparable to those calculated for the vRNA extractions . By contrast , homozygous ΔSRCR5 PAMs did not support virus production at the detection limit of this assay ( Fig 5G–5J ) . In summary , PAMs from ΔSRCR5 animals could not be infected by PRRSV genotype I at a high MOI nor did they replicate the virus over a 72 h time course . To explore the possibility that PMMs could be a suitable alternative to monitor PRRSV infection and investigate whether ΔSRCR5 PMMs , like PAMs , are resistant to PRRSV infection we tested infectivity with all three genotype 1 subtypes of PRRSV , represented by the strains described above . PMMs were infected and assessed as described for PAMs above in both single-round infections ( S5A–S5C Fig ) and multiple-round infections ( S5D–S5F Fig ) . The results obtained from PMMs confirmed the ones obtained in PAMs as no replication of PRRSV was observed in cells from ΔSRCR5 animals . Interestingly , PMMs replicated all viruses to higher levels than PAMs , suggesting that PMMs are not only suitable but may in fact be a superior model for in vitro infection studies with PRRSV . As there could be a genetic variation of CD163 within the Suidae superfamily we performed an in vitro control experiment to assess the susceptibility of warthog ( Phacocherus africanus ) PMMs to PRRSV infection . Interestingly , warthog PMMs were found to be as susceptible to infection with all PRRSV genotype 1 subtypes as the pig PMMs . They all replicated the virus at a similar rate and to comparable titers ( S6 Fig ) . This also shows that the virus poses a threat to African pig breeding countries . To assess the infectability of the ΔSRCR5 macrophages with the Asian/American genotype 2 of the virus we selected two strains associated with high virulence , pathology , morbidity and mortality; strain VR-2385 , isolated in Iowa in 1992 [57] and MN184 , isolated in Minnesota in 2001 [58] , ( both kindly provided by Prof . Tanja Opriessnig ) . PMMs from the different CD163 genotypes were subjected to a multiple-round infection . Therefore , cells were inoculated at MOI = 0 . 1 and supernatant samples collected throughout the progression of infection at 6 , 24 , 32 , 48 , and 72 hpi . Viral RNA was extracted from the supernatants and analyzed by RT-qPCR . Virus levels for both strains started to rise after the 6 hpi time point and increased exponentially up to 32 hpi when they plateaued ( Fig 6A & 6B ) . The virus amplification in the male homozygous and heterozygous macrophages appears to reach higher levels for strain VR-2385 , whereby no difference could be observed for strain MN184 . The quantification limit of the RT-qPCR for VR-2385 was found to be at a CT value of 32 , which corresponded to 3E2 TCID50/ml , for MN184 the quantification limit was at CT value of 36 , corresponding to 25 TCID50/ml . vRNA levels in supernatants from ΔSRCR5 PMMs in this multiple round infection did not increase above the quantification limit . No replication of PRRSV was observed in ΔSRCR5 animals . In the porcine kidney cell line PK-15 , lacking CD163 expression , transfected with the PRRSV attachment factor CD169 the virus was found to be internalized but not to undergo uncoating [36] . This indicates that CD163 , in a close interplay with attachment/internalization factors , plays a major role in the fusion of PRRSV . To assess whether the infection process in ΔSRCR5 macrophages is arrested prior to replication we inoculated PAM cells with all three PRRSV genotype 1 subtypes , represented by the strains described above , at MOI = 2 . The inoculum was removed 3 hpi and infection allowed to continue up to 22 hpi . Cells were fixed and stained for the replication-transcription complexes ( RTC ) formed by PRRSV upon replication initiation . PRRSV nsp2 protein , involved in the formation of double membrane vesicles ( reviewed in [59] ) was chosen as a representative marker for the RTC . The cells were permeabilized and stained for the presence of PRRSV nsp2 . We found that macrophages from both the wild type and the heterozygous animals infected with PRRSV formed RTCs , independent of the subtype . However , in the macrophages from ΔSRCR5 animals no RTC formation was observed . Representative for all strains the SU1-Bel infection is shown in Fig 7 , a figure of all infections may be found in S7 Fig This underlines the involvement of CD163 in the entry and uncoating process of PRRSV infection . It also supports the deletion of SRCR5 as an effective method to abrogate PRRSV infection before the virus or viral proteins are amplified . In addition to its contribution to PRRSV susceptibility , CD163 has been described to have a variety of important biological functions . CD163 is an erythroblast binding factor , enhancing the survival , proliferation and differentiation of immature erythroblasts , through association with SRCR domain 2 and CD163-expressing macrophages also clear senescent and malformed erythroblasts . SRCR domain 3 plays a crucial role as a hemoglobin ( Hb ) -haptoglobin ( Hp ) scavenger receptor . Free Hb is oxidative and toxic; once complexed with Hp it is cleared through binding to SRCR3 on the surface of macrophages and subsequent endocytosis . This prevents oxidative damage , maintains homeostasis , and aids the recycling of iron . Recently , CD163 was also shown to interact with HMGB1-haptoglobin complexes and regulate the inflammatory response in a heme-oxygenase 1 ( HO-1 ) dependent manner [60] . CD163-expressing macrophages were also found to be involved in the clearance of a cytokine named TNF-like weak inducer of apoptosis ( TWEAK ) , with all SRCRs apart from SRCR5 being involved in this process [61] . Soluble CD163 can be found at a high concentration in blood plasma but its function in this niche is still partially unknown ( reviewed in [34 , 62] ) . However , a recent publication by Akahori et al . showed the TWEAK interaction of CD163 to be involved in ischaemic injury tissue regeneration [63] . Maintaining these biological functions is likely to be crucial to the production of healthy , genetically edited animals . Interestingly , none of the biological functions assigned to CD163 have yet been linked to SRCR5 . In order to confirm whether ΔSRCR5 macrophages were still able to take up Hb-Hp complexes we performed a variety of in vitro experiments . Hb-Hp complex uptake in PMMs in vitro has been investigated extensively in the past , with PMMs able to take up both Hb and Hb-Hp complexes in a CD163-dependent manner and the inducible form of heme oxygenase , HO-1 , being upregulated upon Hb-Hp uptake [64 , 65] . PBMCs were differentiated into PMMs by CSF1-induction for seven days , following which PMMs were incubated in the presence of Hb-Hp for 24 h to stimulate HO-1 upregulation . The HO-1 mRNA upregulation , assessed by RT-qPCR , increased 2- to 6-fold in the PMMs from all animals ( Fig 8A ) with no significant difference between the different genotypes . To assess HO-1 levels by western blotting PMMs were incubated in the presence of Hb-Hp for 24 h , lysed using reducing Laemmli sample buffer , and proteins separated by SDS-PAGE . The levels of HO-1 were assessed using a monoclonal antibody against the protein , with a monoclonal antibody against calmodulin as a loading control . HO-1 protein expression was found to be upregulated in all animals , independent of CD163 genotype ( Fig 8B ) . To evaluate the uptake of Hb-Hp directly Hb was labelled with Alexa Fluor 488 ( AF488 ) . PMMs were incubated with HbAF488-Hp for 30 min and followed by FACS analysis . Independent of the CD163 genotype , HbAF488-Hp was taken up efficiently by the PMMs with medians of green fluorescence being 329 , 305 , 329 , 366 , 340 , and 405 for animals 628 , 633 , 627 , 634 , 629 , and 630 , respectively , whilst the background mock-treated cell medians ranged from 2 . 41–4 . 74 ( Fig 8C ) . The uptake of HbAF488-Hp into the PMMs was confirmed by confocal microscopy . In a further experiment PMMs were incubated with HbAF488-Hp for 30 min , followed by fixation and staining for CD163 . The HbAF488-Hp was found in distinct spots , presumably endosomes , with no obvious co-localization with CD163 ( Representative animals wild type 628 and ΔSRCR5 630 shown in Fig 8D , all animals shown in S8 Fig ) . This lack of co-localization is not surprising as the majority of HbAF488-Hp complexes observed were likely already located in late endosomes and lysosomes . Overall , this data demonstrates that macrophages from ΔSRCR5 animals retain the ability to perform their role as hemoglobin-haptoglobin scavengers . The results of this study show that live pigs carrying a CD163 SRCR5 deletion are healthy and maintain the main biological functions of the protein , whilst the deletion renders target cells of PRRSV resistant to infection with the virus . By using two sgRNAs flanking exon 7 of CD163 in CRISPR/Cas9 editing in zygotes we achieved excision of this exon from the genome of pigs yielding a CD163 ΔSRCR5 genotype . The expression of the truncated gene was confirmed by PCR of cDNA , RT-qPCR and western blotting against CD163 . Macrophages isolated from the lungs of wild type CD163 , heterozygous and ΔSRCR5 animals showed full differentiation and expression of macrophage surface markers characteristic of macrophages isolated from the pulmonary alveolar areas . Assessing infection of PAMs from the different genotype animals in both high dose , single-round infections and low dose , multiple-round infections showed PAMs from ΔSRCR5 pigs to be resistant to infection in vitro . The differentiation ability of cells of the monocytes/macrophages lineage from genetically edited CD163 animals was further confirmed by isolation and differentiation of PBMCs . PMMs from ΔSRCR5 pigs were also shown to be resistant to PRRSV infection . PMMs have a crucial biological role , serving as scavengers for Hb-Hp complexes in the blood . Using uptake experiments of fluorescently labelled Hb-Hp complexes as well as gene upregulation assays to monitor the increase of HO-1 upon Hb-Hp stimulation we confirmed that this important biological function is maintained in macrophages isolated from ΔSRCR5 animals . Using CRISPR/Cas9 editing in zygotes we generated live pigs with exon 7 CD163 deletions . Editing efficiency was highly variable , dependent on day of the procedure/surgery , in both in vitro cultivated blastocysts as well as born animals . However , it needs to be considered that overall numbers are low . The reagents used on the various surgery days were the same and both insemination and surgery times were kept consistent . However , there are many elements in the genome editing process that rely on highly skilled personnel and technical reproducibility . Recent developments in nucleic acid delivery methods for genome editing in zygotes may offer possible solutions to standardize the genome editing process . Various groups recently reported successful genome editing by in vitro electroporation of CRISPR/Cas9 regents into zygotes isolated from mice and rats without the necessity to remove the zona pellucida [66–68] . Using electroporation to deliver genome editing reagents in vivo Takahasi et al . showed high success with this method in mouse embryos after 1 . 6 days of gestation [69] . Use of in vitro electroporation could standardize the injection process and reduce the requirement for highly trained personnel . As an alternative , in vivo electroporation would remove both the requirement for donor animals and the long handling process of zygotes prior to re-implantation , however this procedure has currently only been developed for mice and may prove difficult to adapt to the porcine system ( reviewed in [70] ) . Three out of four of the founder animals were found to be edited in a mosaic pattern , although again caution in over interpretation is to be advised due to the low numbers involved . In animal 310 the mosaicism seems to result from a delayed activity of the CRISPR/Cas9 complex , resulting in an edit of one allele in a single cell at the 4- or 8-cell stage . In animals 345 and 347 an initial editing event appears to occur in one allele at the 1-cell stage and a second editing event , modifying the second allele in one of the cells at the 2-cell stage , resulting in homozygous/heterozygous mosaic animals . Mosaicism has been observed in various studies employing injection of genome editors into porcine zygotes [71–73] . Asymmetric spreading of introduced mRNA seems unlikely following results of Sato et al . , who performed in vitro EGFP mRNA injections using parthenogenetically activated porcine oocytes , whereby a relatively homogenous fluorescence pattern could be observed [73] . Rather , mosaicism seems to result from Cas9 protein/sgRNA complexes remaining active throughout several cell divisions or delayed mRNA expression possibly triggered by cell division . The former theory is supported by the genotype of 345 and 347 , which very likely have developed from an initial editing step in one allele at the one cell stage and editing of the second allele in one of the 2-cell or 4-cell stage cells . To generate more biallelic animals by direct injection of zygotes , a more active reagent set is required . Recent studies indicate that injection of Cas9/sgRNA ribonucleoproteins ( RNPs ) is more efficient than mRNA injection . Also , RNP injection can be combined with in vitro electroporation [74] . The mating of the F0 generation animals 310 and 345 resulted in wild type , heterozygous and biallelic edited animals . This showed that despite mosaicism both animals are germline heterozygous . None of the offspring showed any adverse effect from the genome editing under standard husbandry conditions . Interestingly , the genotype of one of the animals , 630 , was consistent with a gene conversion event at the edited CD163 locus . Based on the mechanism of interallelic gene conversion we assume that a homologous recombination occurred in this animal between one allele showing the edited genotype of 345 and the other allele the edited genotype of 310 . The gene conversion appears to have occurred at the zygote stage , rendering 630 homozygous for the genotype of 310 ( reviewed in [75] ) . PRRSV shows a very narrow host cell tropism , only infecting specific porcine macrophage subsets . Isolating these cells from the F1 generation offspring of our genetically edited animals and their wild type siblings we showed that removal of the CD163 SRCR5 domain results in complete resistance of the macrophages towards PRRSV infection . We further demonstrated that macrophages from ΔSRCR5 animals are not only resistant to infection with all European subtypes of genotype 1 but also highly pathogenic and highly virulent strains of the Asian/American genotype 2 . This shows that a targeted removal of SRCR5 is sufficient to achieve complete resistance to PRRSV infection in vitro . PRRSV attachment factors CD151 and CD169 are still expressed on ΔSRCR5 macrophages underlining that these proteins are not sufficient for PRRSV infection . PRRSV infection on macrophages from the ΔSRCR5 animals was halted before the formation of replication transcription complexes proving CD163 to be involved in the entry or uncoating stage of the PRRSV replication cycle . The ΔSRCR5 macrophages will provide a new tool to study this process in detail in a true-to-life system . The ΔSRCR5 animals have several advantages over previously described genome edited animals resistant to PRRSV infection . Whitworth et al . generated animals with a premature stop codon in exon 3 of the CD163 gene , resulting in an ablation of CD163 expression [37] . In another recent publication the group replaced the SRCR5 of the porcine CD163 with the human CD163-L1 ( hCD163-L1 ) domain SRCR8 , utilizing the strategy employed by van Gorp et al . in their in vitro studies of the PRRSV-CD163 interaction [76 , 77] . However , when replacing the SRCR5 with hCD163-L1 SRCR8 the resulting pigs were found to be susceptible to PRRSV genotype 2 infection . In contrast to both approaches we have demonstrated that specific application of genome editing tools in vivo can be used to efficiently generate PRRSV genotype 1 and genotype 2 resistant animals with precise deletion of exon 7 of CD163 , and that these animals retain expression of the remainder of the CD163 protein on the surface of specific differentiated macrophages in a native confirmation . We further showed that the macrophages from these ΔSRCR5 animals retain full differentiation potential , both in PAMs as well as PBMCs stimulated to differentiate by CSF-1 addition , and that macrophages from edited animals retain the ability to perform crucial biological functions associated with CD163 expression , such as hemoglobin-haptoglobin uptake . Furthermore , SRCR5 animals are not transgenic in contrast to hCD163-L1 SRCR8 replacement animals . The hCD163-L1 SRCR8 animals or any variation thereof , which would render the animals resistant to PRRSV genotype 2 as well as genotype 1 , bear another risk , which is the adaptation of the virus to the new amino acid sequence of the replacement SRCR domain , thereby turning into a potential human pathogen due to the human sequence used . A removal of SRCR5 has the advantage of removing the virus’s target interaction sequence all together , thereby making it more difficult for the virus to adapt . Overall , this study demonstrates that it is possible to utilize a targeted genome editing approach to render livestock resistant to viral infection , whilst retaining biological function of the targeted gene . Introduction of CD163 SRCR5 deletion animals in pig breeding could significantly reduce the economic losses associated with PRRSV infection . Primary pulmonary alveolar macrophages ( PAMs ) for the propagation of PRRSV genotype 1 , subtype 1 strain H2 ( PRRSV H2 ) [54] , subtype 2 strain DAI ( PRRSV DAI ) [55] , and subtype 3 strain SU1-Bel ( PRRSV SU1-Bel ) [56] were harvested from wild type surplus research animals aged 6–9 weeks as previously described [46] . Briefly , animals were euthanized according to a schedule I method . Lungs were removed and transferred on ice to a sterile environment . PAMs were extracted from lungs by washing the lungs twice with warm PBS , massaging to release macrophages . Cells were collected by centrifugation for 10 min at 400 x g . When necessary red cells were removed using red cell lysis buffer ( 10 mM KHCO3 , 155 mM NH4Cl , 0 . 1 mM EDTA , pH 8 . 0 ) for 5 min before washing again with PBS . Cells were collected by centrifugation as before and frozen in 90% FBS ( HI , GE Healthcare ) , 10% DMSO ( Sigma ) . Cells were frozen gradually at 1°C/min in a -80°C freezer before being transferred to -150°C . Genotype 2 PRRSV strains VR-2385 , isolated in Iowa in 1992 [57] and MN184 , isolated in Minnesota in 2001 [58] , ( both kindly provided by Prof . Tanja Opriessnig [78] ) were assessed for infectivity on PAM cells prior to use in infection experiments . PAMs from the animals 627 , 628 , 629 , 630 , 633 , and 634 were collected at 8 weeks of age . For this the piglets were sedated using a Ketamine/Azaperone pre-medication mix and anaesthetized with Ketamine/Midazolam . Anesthesia throughout the procedure was maintained using Sevoflurane . PAMs were collected by bronchoalveolar lavage ( BAL ) through an intubation with an air flow access . Three lung segments were flushed in each animal using 2 x 20 ml PBS . Fluid recovery was between 60–80% . Cells were collected by centrifugation for 10min at 400g from the BAL fluid and frozen as above . Peripheral blood monocytes ( PBMCs ) were isolated as described previously [46] . Briefly , blood was collected using EDTA coated vacuum tubes from the jugular vein of the piglets at 10 weeks of age . Blood was centrifuged at 1200 x g for 15 min and buffy coat transferred to PBS . Lymphoprep ( Axis-Shield ) was overlaid with an equal volume of buffy coat/PBS and centrifuged for 45 min at 400 x g . The mononuclear cell fraction was washed with PBS , cells collected and frozen as described above . PAM cells were cultivated in RPMI-1640 , Glutamax ( Invitrogen ) , 10% FBS ( HI , GE Healthcare ) , 100IU/ml penicillin and 100μg/ml streptomycin ( Invitrogen ) ( cRPMI ) . PBMCs were cultivated in cRPMI supplemented with rhCSF1 ( 1×104 units/ml; a gift from Chiron ) for 6 days prior to infection . PK15 cells were cultured in DMEM supplemented with Glutamax ( Invitrogen ) , 10% FBS ( HI , GE Healthcare ) , 100IU/ml penicillin and 100μg/ml streptomycin ( Invitrogen ) . Three potential guide RNA sequences were selected in the 200 bp of intron 6 and one in the 97 bp long intron 7 . Oligomers ( Invitrogen ) were ordered , hybridized as previously described [79] then ligated into the BbsI sites of plasmid pSL66 ( a derivative of px458 with modifications to the sgRNA scaffold as described by [42] ) . The generated plasmids contain a hU6 promoter driving expression of the guide RNA sequence and a CBA promoter driving Cas9-2A-GFP with an SV40 nuclear localization signal ( NLS ) at the N-terminus and a nucleoplasmin NLS at the C-terminus of Cas9 . Cutting efficiency of each guide was assessed by transfection of the plasmids into pig PK15 cells using a Neon transfection system ( Invitrogen ) set at 1400 mV with 2 pulses of 20 mS . 48 hours post-transfection GFP positive cells were collected using a FACS Aria III cell sorter ( Becton Dickinson ) and cultured for a further 4 days prior to preparation of genomic DNA ( DNeasy Blood & Tissues Kit , Qiagen ) . PCR across the target sites was with oSL46 ( ACCTTGATGATTGCGCTCTT ) and oSL47 ( TGTCCCAGTGAGAGTTGCAG ) using AccuPrime Taq DNA polymerase HiFi ( Life Technologies ) to produce a product of 940 bp . A CelI assay ( Transgenomic; Surveyor Mutation Detection Kit ) was performed as previously described [80] . Co-transfection of PK15 cells with pairs of plasmids encoding guides flanking exon 7 were carried out as described above with the exception that cells were harvested at 40 hours post-transfection without enrichment for GFP expression . In this instance a truncated PCR product was observed in addition to the 940 bp fragment , indicating deletion of exon 7 . Based on both single and double cutting efficiencies guide RNAs SL26 ( GAATCGGCTAAGCCCACTGT ) , located 121 bp upstream of exon 7 , and SL28 ( CCCATGCCATGAAGAGGGTA ) , located 30 bp downstream of exon 7 were selected for in vivo experiments . A DNA oligomer fragment containing the entire guide RNA scaffold and a T7 promoter was generated by PCR from the respective plasmid template as follows; a forward primer containing the T7 promoter followed by the first 18 bp of the respective guide RNA and the reverse primers oSL6 ( AAAAGCACCGACTCGGTGCC ) were used in combination with the Phusion polymerase ( NEB ) . DNA fragments were purified on a 1 . 5% agarose gel using the MinElute Gel Extraction Kit ( Qiagen ) according to the manufacturer’s instructions . DNA eluate was further treated with 200 μg/ml Proteinase K ( Qiagen ) in 10 mM Tris-HCl pH 8 . 0 , 0 . 5% SDS for 30 min at 50°C followed by phenol/chloroform extraction . Guide RNAs were generated from the resultant DNA fragment using the MEGAshortscript Kit ( Thermo Fisher ) according to the manufacturer’s instructions . RNA was purified using phenol/chloroform extraction followed by ethanol precipitation and resuspended in EmbryoMax Injection Buffer ( Millipore ) . Purity and concentration of the RNA was assessed using an RNA Screen Tape ( Agilent ) on an Agilent TapeStation according to the manufacturer’s instructions . Embryos were produced from Large-White gilts as described previously [80] . Briefly , gilts were superovulated using a regumate/PMSG/Chorulon regime between day 11 and 15 following estrus . Following heat , the donor gilts were inseminated twice in a 6 hour interval . Zygotes were surgically recovered from mated donors into NCSU-23 HEPES base medium , then subjected to a single 2–5 pl cytoplasmic injection with an injection mix containing 50 ng/μl of each guide ( SL26 and SL28 ) and 100 ng/μl Cas9 mRNA ( PNA Bio or Tri-Link ) in EmbryoMax Injection buffer ( Millipore ) . Recipient females were treated identically to donor gilts but remained unmated . During surgery , the reproductive tract was exposed and 24–39 zygotes were transferred into the oviduct of recipients using a 3 . 5 French gauge tomcat catheter . Litter sizes ranged from 5–12 piglets . Uninjected control zygotes and injected surplus zygotes are cultivated in NCSU-23 HEPES base medium , supplemented with cysteine and BSA at 38 . 5°C for 5–7 days . Blastocysts were harvested at day 7 post cultivation and the genome amplified using the REPLI-g Mini Kit ( Qiagen ) , according to the manufacturer’s instructions . Genotyping was performed as described below . Genomic DNA was extracted from ear biopsy or tail clippings taken from piglets at 2 days postpartum using the DNeasy Blood and Tissue Kit ( Qiagen ) . The region spanning intron 6 to exon 8 was amplified using primers oSL46 ( ACCTTGATGATTGCGCTCTT ) and oSL47 ( TGTCCCAGTGAGAGTTGCAG ) , generating a 904 bp product from the intact allele and a 454 bp product if complete deletion of exon 7 had occurred . PCR products were analyzed by separation on a 1% agarose gel and subsequent Sanger sequencing of all truncated fragments . Fragments corresponding to the wild type length were further analyzed by T7 endonuclease I ( NEB ) digestion according to the manufacturer’s instructions . RNA was isolated from 1E6 PAM cells , isolated by BAL as described above , using the RNeasy Mini Kit ( Qiagen ) , according to the manufacturer’s instructions , including an on-column DNase digestion . First-strand cDNA was synthesized using an Oligo-dT primer in combination with SuperScript II reverse transcriptase ( Invitrogen ) , according to the manufacturer’s instructions . The cDNA was used to assess the RNA phenotype across exons 4 to 9 using primers P0083 ( ATGGATCTGATTTAGAGATGAGGC ) and P0084 ( CTATGCAGGCAACACCATTTTCT ) , resulting in a PCR product of 1686 bp length for the intact allele and 1371 bp following precise deletion of exon 7 . PCR products were analyzed by separation on a 1% agarose gel and subsequent Sanger sequencing of deletion fragments . 4E5 PAM cells isolated by BAL were collected by centrifugation at 300 rcf for 10 min . The pellet was resuspended in Laemmli sample buffer containing 100 mM DTT , boiled for 10 min at 95°C and subjected to electrophoresis on 7 . 5% acrylamide ( Bio-Rad ) gels . After transfer to a nitrocellulose membrane ( Amersham ) , the presence of cellular proteins was probed with antibodies against CD163 ( rabbit pAb , abcam , ab87099 ) at 1 μg/ml , and β-actin ( HRP-tagged , mouse mAb , Sigma , A3854 ) at 1:2000 . For CD163 the blot was subsequently incubated with HRP-labelled rabbit anti-mouse antibody ( DAKO , P0260 ) at 1:5000 . Binding of HRP-labelled antibodies was visualized using the Pierce ECL Western Blotting Substrate ( Thermo Fisher ) , according to the manufacturer’s instructions . RNA was isolated from 1E6 PAMs using the RNeasy Mini Kit ( Qiagen ) , according to the manufacturer’s instructions , including an on-column DNase digestion . RNA levels were measured using the GoTaq 1-Step RT-qPCR system ( Promega ) according to the manufacturers' instructions on a LightCycler 480 ( Roche ) . mRNA levels of CD163 were quantified using primers P0074 ( CATGGACACGAGTCTGCTCT ) and P0075 ( GCTGCCTCCACCTTTAAGTC ) and reference mRNA levels of β-actin using primers P0081 ( CCCTGGAGAAGAGCTACGAG ) and P0082 ( AAGGTAGTTTCGTGGATGCC ) . PAMs were seeded one day prior to analysis . PBMCs were seeded seven days prior to analysis and differentiated by CSF1 stimulation to yield PBMC-derived macrophages ( PMMs ) . Cells were harvested by scraping with a rubber policeman and fixed in 4% formaldehyde/PBS for 15 min at room temperature . Cells were incubated with blocking solution ( PBS , 3% BSA ) for 45 min before staining with antibodies . Cells were stained with antibodies targeting either mouse anti pig CD14 ( AbD Serotec , MGA1273F , 1:50 ) and mouse anti pig CD16 ( AbD Serotec , MCA2311PE , 1:200 ) , mouse anti pig CD169 ( AbD Serotec , MCA2316F , 1:50 ) and mouse anti pig CD172a ( SoutherBiotech , 4525–09 , 1:400 ) , mouse anti human CD151 ( AbD Serotec , MCA1856PE , 1:50 ) and mouse anti pig SWC9 ( CD203a ) ( AbD Serotec , MCA1973F , 1:50 ) , mouse anti pig CD163 ( AbD Serotec , MCA2311PE , 1:50 ) , or mouse IgG1 or an IgG2b negative control ( AbD Serotec , MCA928PE , MCA691F , or Sigma , F6397; same concentration as primary Ab ) . The cells were washed three times with PBS and resuspended in FACS buffer ( 2% FBS , 0 . 05M EDTA , 0 . 2% NaN3 in PBS ) . Gene expression determined by antibody labelling was assessed by FACS analysis on a FACS Calibur ( Becton Dickinson ) using FlowJo software . PAMs were seeded one day prior to infection . PBMCs were seeded seven days prior to infection and differentiated by CSF1 stimulation to yield PBMC-derived macrophages PMMs . Cells were inoculated at MOI = 1 of the respective virus strain ( PRRSV H2 , DAI , or SU1-Bel ) in cRPMI for 3 h at 37°C . The inoculum was replaced by warm cRPMI . At 19 hpi cells were detached by using a cell scraper . Cells were fixed in 4% Formaldehyde ( Sigma-Aldrich ) in PBS ( Gibco ) for 15 min at RT , washed with PBS , and subsequently permeabilized in PBS containing 0 . 1% Triton-X-100 ( Alfa Aesar ) for 10 min . Cells were incubated with antibody against PRRSV-N ( SDOW17-F , RTI , KSL0607 , 1:200 ) and CD163 ( AbD Serotec , MCA2311PE , 1:50 ) or mouse IgG1 negative controls , as described above , in 3% BSA in PBS . The cells were washed three times with PBS and resuspended in FACS buffer . Infection levels , determined by antibody labelling , were assessed by FACS analysis on a FACS Calibur ( Benson Dickson ) using FlowJo software . PAMs were seeded one day prior to infection . PBMCs were seeded seven days prior to infection and differentiated by CSF1 stimulation to yield PMMs . Cells were inoculated at MOI = 0 . 1 with the respective virus strain ( PRRSV H2 , DAI , or SU1-Bel ) in cRPMI for 3 h at 37°C . Inoculum was removed , cells washed 1x with PBS , and infection continued . At the indicated times post inoculation samples were harvested to be assessed . All samples were frozen and processed once all samples from a time course had been collected . Viral RNA ( vRNA ) was extracted from the supernatant samples using the QIAmp Viral RNA Mini Kit according to the manufacturer’s instructions . The viral RNA levels were quantified by RT-qPCR using the GoTaq Probe 1-Step RT-qPCR system ( Promega ) for PRRSV H2 and SU1-Bel and the GoTaq 1-Step RT-qPCR system ( Promega ) for PRRSV DAI , VR-2385 , and MN184 , according to the manufacturer’s instructions . For this the following primers and probes were used: H2 fwd ( GATGACRTCCGGCAYC ) , H2 rev ( CAGTTCCTGCGCCTTGAT ) , H2 probe ( 6-FAM-TGCAATCGATCCAGACGGCTT-TAMRA ) , ( optimal H2 primer/probe sequences obtained from JP Frossard , AHVLA ) , SU1-Bel fwd ( TCTTTGTTTGCAATCGATCC ) , SU1-Bel rev ( GGCGCACTGTATGACTGACT ) , SU1-Bel probe ( 6-FAM-CCGGAACTGCGCTTTCA-TAMRA ) , DAI fwd ( GGATACTATCACGGGCGGTA ) , DAI rev ( GGCACGCCATACAATTCTTA ) , VR-2385 fwd ( CTGGGTAAGATCATCGCTCA ) , VR-2385 rev ( CAGTCGCTAGAGGGAAATGG ) , MN184 fwd ( CTCTCGCGACTGAAGATGAC ) , MN184 rev ( GCCTTGGTTAAAGGCAGTCT ) . RNA levels were measured on a LightCycler 480 ( Roche ) using a standard curve generated from vRNA isolates of high titer stocks . Infectivity of the virus produced was assessed using a TCID50 assay of selected time points on PAMs isolated from wild type surplus research animals . PBMCs were seeded seven days prior to analysis and differentiated by CSF1 stimulation to yield PMMs . Hemoglobin ( Hb , Sigma-Aldrich , A0 , H0267 ) and Haptoglobin ( Hp , Sigma Aldrich , Phenotype 2–2 , H9762 ) were mixed in a 1:1 wt/wt ratio in PBS for 15 min on a vertical roller before experimentation . PMMs were incubated with 100 μg/ml Hb-Hp in cRPMI for 24 h at 37°C . Cells were harvested by scraping with a rubber policeman . RNA was isolated from 1E6 cells using the RNeasy Mini Kit ( Qiagen ) , according to the manufacturer’s instructions , including an on-column DNase digestion . RNA levels were measured using the GoTaq 1-Step RT-qPCR system ( Promega ) according to the manufacturers' instructions on a LightCycler 480 ( Roche ) . mRNA levels of heme oxygenase 1 ( HO-1 ) were quantified using primers P0239 ( TACATGGGTGACCTGTCTGG ) and P0240 ( ACAGCTGCTTGAACTTGGTG ) and reference mRNA levels of β-actin using primers P0081 and P0082 . For analysis of protein levels of HO-1 cells were collected by centrifugation at 300 rcf for 10 min . The pellet was resuspended in Laemmli sample buffer containing 100 mM DTT , boiled for 10 min at 95°C and subjected to electrophoresis on 12% acrylamide ( Bio-Rad ) gels . After transfer to a nitrocellulose membrane ( Amersham ) , the presence of cellular proteins was probed with antibodies against HO-1 ( mouse mAb , abcam , ab13248 , 1:250 ) , and calmodulin ( rabbit mAb , abcam , ab45689 , 1:1000 ) . The blot was subsequently incubated with HRP-labelled goat anti-rabbit antibody ( DAKO , PI-1000 ) at 1:5000 . Binding of HRP-labelled antibodies was visualized using the Pierce ECL Western Blotting Substrate ( Thermo Fisher ) , according to the manufacturer’s instructions . PBMCs were seeded seven days prior to analysis and differentiated by CSF1 stimulation to yield PMMs . For fluorescence microscopy , cells were seeded on glass cover slips . Hemoglobin ( Sigma-Aldrich , A0 , H0267 ) was labeled with Alexa Fluor 488 ( AF-488 ) using a protein labelling kit ( Molecular Probes ) according to the manufacturer’s instructions . HbAF488 and Hp were mixed in a 1:1 wt/wt ratio in PBS for 15 min on a vertical roller before experimentation . PMMs were incubated with 10 μg/ml HbAF488-Hp in cRPMI for 30 min at 37°C . For quantification by FACS the cells were collected with a rubber policeman and washed three times with Ca2+/Mg2+-free PBS to remove surface bound HbAF488-Hp as described previously [65] . Cells were fixed in 4% ( wt/v ) formaldehyde ( Sigma-Aldrich ) in PBS ( Gibco ) for 15 min at RT , washed with PBS , and subsequently permeabilized in PBS containing 0 . 1% Triton-X-100 ( Alfa Aesar ) for 10 min . Cells were stained with mouse anti pig CD163 antibody ( AbD Serotec , MCA2311PE , 1:50 ) as described above then washed three times with PBS and resuspended in FACS buffer . Gene expression determined by antibody labelling was assessed by analysis on a FACS Calibur ( Becton Dickinson ) using FlowJo software . For immunofluorescence imaging cells were washed three times with Ca2+/Mg2+-free PBS and fixed in 4% formaldehyde ( Sigma-Aldrich ) in PBS ( Gibco ) for 15 min at RT , washed with PBS , then permeabilized in PBS containing 0 . 1% Triton-X-100 ( Alfa Aesar ) for 10min . Cells were washed with PBS and incubated with antibody against CD163 ( rabbit pAb , abcam , ab87099 , 5 μg/ml ) in blocking buffer ( PBS , 3% FBS ) for 1 h , washed , and incubated with secondary goat anti-rabbit AF594 antibody ( A11037 , 1:100 ) , AF647 phalloidin ( A22287 , 1:100 ) , and DAPI ( 1:10 , 000; all Life Technologies ) . The samples were analyzed using a confocal laser-scanning microscope ( Zeiss LSM-710 ) . PAMs were seeded onto coverslips one day prior to infection . Cells were inoculated at MOI = 2 of the respective virus strain ( PRRSV H2 , DAI , or SU1-Bel ) in cRPMI for 3 h at 37°C . The inoculum was replaced by warm cRPMI . At 19 hpi cells were fixed in 4% formaldehyde ( Sigma-Aldrich ) in PBS ( Gibco ) for 15 min at RT , washed with PBS , and permeabilized as described above . Cells were washed with PBS and incubated with antibody against PRRSV nsp2 ( A gift from Ying Fang , Kansas State University , [81] , 1:400 ) in blocking buffer for 1 h , washed , and incubated with secondary goat anti-mouse AF488 antibody ( A11029 , 1:100 ) , AF568 phalloidin ( A12380 , 1:100 ) , and DAPI ( 1:10 , 000; all Life Technologies ) . The samples were analyzed using a confocal laser-scanning microscope ( Zeiss LSM-710 ) .	Porcine Reproductive and Respiratory Syndrome is an endemic infectious disease of pigs , manifesting differently in pigs of different ages but primarily causing late-term abortions and stillbirths in sows and respiratory disease in piglets . The causative agent of the disease is the positive-strand RNA PRRS virus ( PRRSV ) . PRRSV only infects a specific subset of cells of the innate immune system of the monocyte/macrophage lineage . Previous research found that the virus needs a specific receptor , CD163 , in order to make its own membrane fuse with the host cell membrane in an uptake vesicle to release the viral genetic information into the cytosol and achieve a successful infection . CD163 has a pearl-on-a-string structure , whereby the “pearl”/ domain number 5 was found to interact with the virus and allow it to infect a cell . Here we describe how we generated pigs lacking the CD163 subdomain 5 using so-called CRISPR/Cas9 gene editing in zygotes . The pigs were healthy under normal husbandry conditions and other biological functions conducted by the CD163 were found to be intact . We isolated a variety of monocyte and macrophage cells from these pigs and found them to be completely resistant to PRRSV infection .	[ "Abstract", "Introduction", "Results", "Discussion", "Materials", "and", "methods" ]	[ "blood", "cells", "livestock", "medicine", "and", "health", "sciences", "immune", "cells", "body", "fluids", "immunology", "rna", "extraction", "variant", "genotypes", "vertebrates", "animals", "mammals", "genetic", "mapping", "germ", "cells", "zygotes", "molecular", ...	2017	Precision engineering for PRRSV resistance in pigs: Macrophages from genome edited pigs lacking CD163 SRCR5 domain are fully resistant to both PRRSV genotypes while maintaining biological function
Trypanosoma cruzi enters host cells by subverting the mechanism of cell membrane repair . In this process , the parasite induces small injuries in the host cell membrane leading to calcium entry and lysosomal exocytosis , which are followed by compensatory endocytosis events that drive parasites into host cells . We have previously shown that absence of both LAMP-1 and 2 , major components of lysosomal membranes , decreases invasion of T . cruzi into host cells , but the mechanism by which they interfere with parasite invasion has not been described . Here we investigated the role of these proteins in parasitophorous vacuole morphology , host cell lysosomal exocytosis , and membrane repair ability . First , we showed that cells lacking only LAMP-2 present the same invasion phenotype as LAMP1/2-/- cells , indicating that LAMP-2 is an important player during T . cruzi invasion process . Second , neither vacuole morphology nor lysosomal exocytosis was altered in LAMP-2 lacking cells ( LAMP2-/- and LAMP1/2-/- cells ) . We then investigated the ability of LAMP-2 deficient cells to perform compensatory endocytosis upon lysosomal secretion , the mechanism by which cells repair their membrane and T . cruzi ultimately enters cells . We observed that these cells perform less endocytosis upon injury when compared to WT cells . This was a consequence of impaired cholesterol traffic in cells lacking LAMP-2 and its influence in the distribution of caveolin-1 at the cell plasma membrane , which is crucial for plasma membrane repair . The results presented here show the major role of LAMP-2 in caveolin traffic and membrane repair and consequently in T . cruzi invasion . Trypanosoma cruzi is the causative agent of Chagas disease . This parasite is naturally transmitted through the feces of an infected vector , a triatomine bug , but transmission may also occur through contaminated food , blood transfusion , placenta or organ transplantation [1] . Therefore , although originally endemic to Latin America , where the vector is widespread , Chagas disease is now found in non-endemic countries , especially in the southern part of the United States and Europe due to human migration [2–5] . Chagas disease is a serious and debilitating illness with a variable clinical course , ranging from asymptomatic to very serious cardiac and/or gastrointestinal disease [6] . Available treatment is not efficient , especially considering the chronic phase of the infection [7] . In order to survive and replicate in the vertebrate host , T . cruzi needs to interact with and invade host cells . Therefore the comprehension of the mechanisms involved in these processes is extremely important for the development of more efficient treatment and disease control . The parasite is able to invade a wide variety of cell lines , professional and non-professional phagocytic cells . In order to gain entry into host cells , T . cruzi subverts the mechanism by which cells repair small injuries in their plasma membrane . These small membrane tears lead to extracellular calcium influx into the cell cytoplasm . Increase in intracellular calcium induces lysosome recruitment and fusion at the site of injury , which leads to the release of an enzyme called acid sphingomyelinase . This enzyme cleaves sphingomyelin , generating ceramide , which induces a compensatory endocytosis event that carries the damaged site to the cell interior , resealing the plasma membrane . In the case of T . cruzi , intracellular calcium increase may occur through a series of molecules released by the parasite or located at its surface , which will trigger host signaling events that promote a rise in intracellular calcium [8–12] . In parallel , T . cruzi is also able to induce microinjuries at the host cell membrane , leading to calcium influx from the extracellular media into the cells [13] . In both cases , calcium leads to lysosomal exocytosis at the site of parasite attachment/injury [14 , 15] , followed by compensatory endocytosis events that pull T . cruzi into host cells [13] . To the newly formed parasitophorous vacuole , more lysosomes fuse until the entire vacuole is covered with lysosomal membrane markers . Lysosomal association and fusion during T . cruzi host cell invasion is essential for the formation of a viable parasitophorous vacuole , without which parasites could exit host cells [15] . Lysosomes could be contributing to anchor and drag parasite into host cells . In fact , we have previously shown that cells lacking Lysosomal Associated Membrane Proteins 1 and 2 ( LAMP1/2-/- cells ) , the major integral membrane proteins of lysosomes , are less susceptible to T . cruzi infection [16] . However , the mechanism by which these proteins interfere with invasion is still unclear . LAMPs are highly glycosylated proteins , rich in sialic acid , and estimated to cover approximately 80% of the lysosome luminal surface [17 , 18] . Due to its high sialic acid content and the previously reported role of host cell sialic acid in T . cruzi invasion process [19–22] , it has been suggested that LAMP’s sialic acid moieties could be contributing to the invasion phenotype in LAMP knock out cells [16] . However , despite being heavily sialylated and structurally very similar , LAMP-1 and 2 have only around 37% similarity in amino acid sequence [23 , 24] and differ in function , as revealed by the generation of LAMP1 and 2 single knock out mice [18 , 25–27] . Therefore we were interested in evaluating the real role of LAMP during T . cruzi invasion of host cells . In order to evaluate the mechanism by which LAMP proteins participate in T . cruzi host cell invasion , we decided to investigate the role of these proteins in known important events involved in this process: parasitophorous vacuole morphology , host cell lysosomal exocytosis , and membrane repair ability . For this , we used LAMP1/2-/- cells in comparison to WT cells . In parallel , we performed the same assays using LAMP2 single knock out fibroblast ( LAMP2-/- ) , since the lack of LAMP-2 caused a more severe alteration in cells and mice [28 , 29] . Anti–mouse LAMP-1 mAb ( 1D4B ) , as well as anti-mouse LAMP-2 mAb ( ABL-93 ) , were obtained from the Developmental Studies Hybridoma Bank . Anti-T . cruzi polyclonal antibodies were obtained from serum of rabbits immunized with T . cruzi trypomastigotes as described previously [30 , 31] . Secondary antibodies , anti-rat IgG-Alexa fluor 488 , anti-mouse-Alexa fluor 488 and anti-rabbit IgG-Alexa fluor 546 were obtained from Thermo Fischer Scientific . Mouse fibroblasts cell lines , derived from wild type ( WT ) , LAMP1 and 2 ( LAMP1/2-/- ) or LAMP2 ( LAMP2-/- ) knock out C57BL6 mice , were obtained from a collection of cell lines from Dr . Paul Saftig’s laboratory ( Biochemisches Institut / Christian-Albrechts-Universität Kiel , Germany ) , which were previously generated by spontaneous immortalization of primary fibroblasts in culture around passages 10–20 [18 , 29 , 32] . The cells were maintained in high-glucose DMEM ( Thermo Fischer Scientific ) supplemented with 10% fetal bovine serum ( FBS ) , 1% penicillin/streptomycin ( 100U/mL and 100μg/mL , respectively ) and 1% glutamine ( DMEM 10% ) . Tissue culture trypomastigotes ( TCTs ) from the T . cruzi Y strain were obtained from the supernatant of infected LLC-MK2 monolayers and purified as described by Andrews et al . ( 1987 ) [30] . For invasion assays , 4x104 cells ( WT , LAMP1/2-/- and LAMP2-/- fibroblasts ) in high glucose DMEM 10% were plated in each well of a 24-well plate , containing 13mm round glass coverslips . Cells were plated 24 h before the experiment and incubated at 37°C and 5% CO2 . Cells were then exposed to T . cruzi TCTs Y strain for 20 min at 37°C at a multiplicity of infection ( MOI ) of 100 . After parasite exposure , the monolayers were washed 4 times with phosphate buffered saline containing Ca2+ and Mg2+ ( PBS+/+ ) , in order to remove the non internalized parasites , and fixed in paraformaldehyde 4% overnight . After fixation , cells were processed for immunofluorescence . After fixation , coverslips with attached cells were washed three times in PBS , incubated for 20 min with PBS containing 2% BSA and processed for an inside/outside immunofluorescence invasion assay as described previously [30] . Briefly , extracellular parasites were immunostained with rabbit anti-T . cruzi polyclonal antibodies in a 1:500 dilution in PBS/BSA for 1h at room temperature , washed and labeled with Alexa Fluor-546 conjugated anti-rabbit IgG antibody ( Thermo Fischer Scientific ) in a proportion of 1:500 in PBS/BSA for 45min . After the inside/outside immunofluorescence staining , host cell lysosomes were also immunostained using a 1:50 dilution of either rat anti-mouse LAMP-1 hybridoma supernatant ( 1D4B ) or rat anti-mouse LAMP-2 hybridoma supernatant ( Abl 93 ) in PBS/BSA/saponin and the appropriate fluorescent labeled secondary antibody anti-rat IgG-Alexa fluor 488 , as described previously [15 , 30] . After that , the DNA of host cells and parasites was stained for 1 min with DAPI ( 4' , 6-Diamidino-2-Phenylindole , Dihydrochloride—Sigma ) , 0 , 1μM in PBS , mounted , and examined on an Olympus BX51 microscope equipped with a Q color 3 camera controlled by the ImagePro Express Software ( Olympus ) . WT , LAMP1/2-/- and LAMP2-/- fibroblasts were infected with T . cruzi TCTs at a MOI of 100 for 20 min , washed in PBS+/+ and then fixed with 2 . 5% Glutaraldehyde in 0 . 1M PHEM buffer ( 5mM MgCl2 . 6H2O , 70mM KCl , EGTA 10mM , HEPES 20mM , PIPES 60mM ) , pH7 . 2 . After fixation cells were gently scraped off with a rubber policeman , harvested by centrifugation and incubated in 3% low melting agarose . The hardened agarose with the sample was cut into small pieces and washed in PHEM buffer . All samples were post-fixed with 2% osmium tetroxide containing 1 . 5% potassium ferrocyanide in PHEM buffer for 1h at room temperature . The cells were dehydrated using increasing concentrations of graded acetone before embedding the pellet in Araldite . Ultrathin sections in 200 mash copper grids were stained with lead citrate and analyzed on Tecnai G2-12 –SpiritBiotwin FEI electron microscope . Quantification of vacuole volume density was performed by the ratio between the total area of the vacuole and the parasite area , using the Image J software . Values closer to 1 indicate a tight vacuole , while values greater than 1 indicate looser vacuoles . To evaluate the lysosomal exocytosis , a β-hexosaminidase secretion assay was performed according to previous work [31] . Briefly , WT , LAMP1/2-/- and LAMP2-/- cells were treated with Ionomycin ( Calbiochem ) 5 and 10 μM for 10 min at 37°C . After treatment , cell extracellular media was collected and cells were lysed with Triton x-100 ( Sigma ) 1% in PBS . Extracellular media and cell lysates were incubated with 50μL of β-hexosaminidase substrate , 6mM 4 methylumbelliferyl-N-acetyl-B-D-glucosaminide ( Sigma-Aldrich ) , dissolved in Na-citrate-PO2 buffer ( pH4 . 5 ) . After 15–20 min of incubation at 37° the reactions were stopped by adding 100μL of stop solution ( 2M Na2CO3-H2O , 1 . 1M glycine ) . 100μL of this solution was used for reading at 365nm of excitation and 450 nm of emission in a spectrofluorimeter ( Synergy 2 , Biotek in the Center of Flow Cytometry and Fluorimetry , Department of Biochemistry and Immunology , ICB-UFMG ) . Compensatory endocytosis after injury was measured using a scrape wound assay followed by trypan blue fluorescence quenching , as previously described [33] . Briefly , WT cells were grown in 10cm plates , washed with HBSS at 4°C containing or not containing 1 . 8mM Ca2+ and labeled on the plasma membrane with 1μg/mL wheat germ agglutinin ( WGA ) –Alexa Fluor 488 ( Invitrogen ) for 1 minute at 4°C , followed by two more washes in HBSS . Cells were then wounded by scraping in the presence or absence of Ca2+ and incubated for 2 minutes at 37°C with 0 . 2% trypan blue before washing and FACS analysis . Due to WGA-Alexa Fluor 488 susceptibility to quenching by the membrane impermeable trypan blue , Alexa Fluor 488 fluorescence measurement will correspond to the amount of membrane endocytosed after trypan blue exposure . WT , LAMP1/2 and LAMP2 knockout cells were plated on 10 cm round dishes 24 hours prior to experiments at 2 x 106 cells/dish . Cell monolayers were washed 3 X with PBS and incubated at 4o C with 2 mL of cold HBSS media containing Ca2+ . To induce PM damage , exocytosis of lysosomes and PM repair each cell type was scrapped at 4° C , gently re-suspended with a serological pipette and incubated at 37° C for 5 minutes to induce lysosomal secretion and plasma membrane repair . As controls , non-scraped cells were incubated for 5 minutes at 37° C and the media were collected to assess constitutive exocytosis . The suspensions containing the exocytosed content from lysosomes were transferred to ice and centrifuged for 30 minutes at 10 , 000 g in order to remove possible detached cells . The supernatants containing the enzymes released from lysosomes were concentrated 20 times using 3 kDa Amicon Ultra filter units and analyzed by SDS-PAGE and Western Blot . Immunoblot assay was performed using rabbit polyclonal anti-ASM ( Abcam ab83354 ) . Protein samples were prepared with reducing sample buffer , boiled , separated on 10% SDS-PAGE and blotted onto nitrocellulose membranes ( Bio-Rad ) using the Trans-Blot transfer system ( Bio-Rad ) for 1 hour and 30 minutes at 100 V . The membrane was blocked for 1 h with 5% milk solution . After incubation with the primary antibody and peroxidase conjugated secondary antibody , detection was performed using Luminata Forte Western HRP Substrate ( Millipore ) and a Fuji LAS-4000 Imaging System with Image Reader LAS-4000 software ( Fuji ) . WT , LAMP1/2-/- and LAMP2-/- cells were cultured in 35mm culture dishes ( 3x105 cells/dish ) at 37°C in 5% CO2 in DMEM 10% . After 24 hours , cell monolayers , treated or not with MβCD , were washed at 4°C with Ca2+-free PBS followed by two more washes in Ca2+-free PBS . Cells were then incubated with HBSS containing Ca2+ at 4°C , wounded by scrapping and incubated for 5 minutes at 37°C , to allow plasma membrane repair . Afterwards , cells were incubated at 4°C for 5 minutes with Propidium Iodide ( PI ) ( 10μg/mL in HBSS ) . For hitting control , cells were incubated with Ca2+-free HBSS at 4°C , wounded by scraping in the presence of PI and incubated for 5 minutes at 37°C . Alternatively , cell monolayers were injured by incubation with DMEM 10% containing PI ( 10μg/mL ) in the presence or absence of T . cruzi TCTs Y strain for 30 min at 37°C at a multiplicity of infection ( MOI ) of 100 . After flow cytometry ( FACS Scan; Becton Dickinson ) , the data were analyzed using FlowJo v10 . 1 software ( Tree Star , Inc . ) . WT , LAMP1/2-/- and LAMP2-/- cells were plated on 24 well plates containing 13 mm round coverslips at a density of 5x104 cells/well 24 hours before the assay . Cells were then fixed with paraformaldehyde ( PFA ) 4% for 1 hour at 4°C . After fixation , coverslips with attached cells were washed three times in PBS+/+ , permeabilized with Triton X-100 0 . 1% ( Sigma-Aldrich ) and incubated for 30 min with PBS+/+ containing 1% BSA ( PBS/BSA 1% ) . For labeling polymerized actin , cells were incubated with Phalloidin-Alexa Fluor 546 ( Thermo Fischer Scientific ) using a dilution 1:40 in PBS/BSA 1% at room temperature , followed by three additional washes . Labeled coverslips were mounted on glass slides and examined on a Zeiss Axio Imager . Z2 ( ApoTome . 2 structured illumination system ) microscope . Plasma membranes ( PM ) from WT , LAMP1/2-/- and LAMP2-/- cells were separated from nuclei and large granule fraction through differential centrifugation and subsequently isolated from endoplasmatic reticulum using equilibrium density ultracentrifugation as previously described [34] , with few modifications . After isolation , PM subfraction was saponified with 25 ml ethanolic sodium hydroxide ( 1M ) under heating . The unsaponifiable phase ( enriched in cholesterol ) was extracted with ethyl acetate and dried out overnight . The unsaponifiable matter was treated with N , O-Bis ( trimethylsilyl ) trifluoroacetamide ( BSTFA ) to obtain trimethyl-silyl ( TMS ) derivatives [35] . Gas chromatography ( GC ) analyses were performed in a gas chromatograph HP7820A ( Agilent ) equipped with a flame ionization detector . An HP5 column ( Agilent; 30 m x 0 . 32 mm x 0 . 25um ) was used following a ramp protocol starting from 250°C at a rate of 10°C/min for 5 min , with an injector ( split 1:30 ) at 300°C and a detector at 300°C . Hydrogen was used as the carrier gas ( 3 ml/min ) ; and an injection volume of 3μl . Peak identification and cholesterol concentration were obtained comparing peak areas with a standard curve using known concentration of cholesterol ( Sigma-Aldrich ) derivatized under same conditions as samples . The final content of cholesterol was normalized by total protein concentration taken from particle free supernatant obtained during PM subfraction isolation and protein concentration was determined according to Lowry et al . using bovine serum albumin as standard [36] . For Western blot detection of total caveolin-1 content , WT , LAMP1/2-/- and LAMP2-/- cells were plated the day before , scraped and lysed with RIPA buffer ( 25mM Tris-HCl pH 7 . 6 , 150mM NaCl , 1% NP-40 , 1% sodium deoxycholate , 0 . 1% SDS ) . Protein content was measured , samples were prepared with reducing sample buffer , boiled and 50 μg of protein were loaded in each lane on a 12% SDS-PAGE . After transfer ( see above–same as for acid sphingomyelinase ( ASM ) detection ) the membrane was incubated with anti-Caveolin 1 ( BD–cat . # 610406 ) diluted 1:2000 . Detection and imaging were performed as described for ASM detection . For immunofluorescence labeling , WT , LAMP1/2-/- and LAMP2-/- cells were plated onto round glass coverslips the day before , fixed in 4% paraformaldehyde for 10 min and washed 3 x in PBS , followed by incubation for 45 min at room temperature in PBS containing 2% BSA and 0 . 05% saponin ( PBS/BSA/saponin ) . Cells were then incubated for 2 h with anti-Caveolin 1 ( BD–cat . # 610406 ) antibody diluted 1:500 in PBS/BSA/saponin , washed , followed by 1 h incubation with secondary anti-mouse antibodies conjugated with Alexa Fluor 488 ( Thermo Fischer Scientific ) diluted 1:500 in PBS/BSA/saponin . Coverslips were mounted using ProLong1Gold antifade reagent ( Thermo Fischer Scientific ) and imaged using a Zeiss Axio Imager . Z2 ( ApoTome . 2 structured illumination system ) microscope with an Axiocam 503 monochrome camera controlled by the Zen Blue Software ( Zeiss ) . Alternatively , cells were imaged in a Zeiss Axio Imager . Z2 Microscope to create a 3D reconstruction , which has been made following capture of 16 optical sections with an approximate 0 . 93μm interval , using the 63x oil objective . The 3D imaging stack has been reconstructed using Zen Blue Software . We first decided to test the influence of lysosomal integral membrane protein 2 , LAMP-2 , in T . cruzi cell infection , when compared to LAMP1/2-/- deficient or WT control fibroblasts . It had been shown before that LAMP-1 deficiency had little effect on cell morphology , metabolism or viability and led to overexpression of LAMP-2 in cells , suggesting that LAMP-2 might compensate for LAMP-1 deficiency [25] . On the other hand , LAMP-2 deficiency in mice led to a series of lysosomal defects , which together compromised mice viability [27 , 28] . For this , we performed cell invasion assays using fibroblast cell lines generated from a LAMP2 knock out mouse ( LAMP2-/- fibroblasts ) and compared to the invasion rates obtained from WT and LAMP1/2 double knock out mouse ( LAMP1/2-/- fibroblasts ) [29] . The phenotypes of the cell lines were confirmed through immunolabeling with both anti LAMP-1 and LAMP-2 antibodies ( S1 Fig ) . As expected from previous data from our group [16] , lack of both LAMP-1 and 2 led to a reduction in the invasion rate of T . cruzi TCTs when compared to WT fibroblasts ( Fig 1A–1C ) . The number of internalized parasites per 100 counted cells ( Fig 1A ) , as well as the percentage of infected cells ( Fig 1B ) in LAMP1/2-/- fibroblast cultures are about 3 times lower when compared to their WT counterparts . The same was observed for LAMP2-/- fibroblasts , revealing that absence of LAMP-2 alone was sufficient to reduce T . cruzi invasion rates to the same level observed for LAMP1/2-/- fibroblasts and indicating a primary role for this protein in the invasion process ( Fig 1A–1C ) . Interaction of T . cruzi with its vacuolar membrane has been shown to interfere with parasite invasion ability [22] . Therefore , we decided to test whether , in the absence of LAMP proteins , parasite interaction with its vacuole was compromised , contributing to the decreased cell invasion observed in LAMP knock out cells . For this , we evaluated parasitophorous vacuole morphology in the different cell lines ( WT , LAMP1/2-/- and LAMP2-/- fibroblasts ) through Transmission Electron Microscopy ( TEM ) . TEM images of parasitophorous vacuoles containing recently internalized trypomastigotes , in the different cell lines ( Fig 2A–2C ) , were processed using the software Image J and the ratio between the parasitophorous vacuole and parasite areas were calculated ( Fig 2D ) . As expected , membranes of parasite and vacuole were tightly apposed to each other in WT cells , showing almost no intravacuolar space between parasite plasma membrane and parasitophorous vacuolar membrane ( Fig 2A ) . The same was observed for LAMP1/2-/- or LAMP2-/- fibroblasts ( Fig 2B and 2C , respectively ) . Quantification of the ratio between vacuolar area and parasite area confirmed that presence or absence of LAMP did not interfere with vacuolar morphology ( Fig 2D ) . The ratio between parasitophorous vacuole and parasite areas for all three cell lines were very close to one , showing that the two membranes , parasite and vacuolar , were intimately associated even in the absence of LAMP . As it is well known , T . cruzi invasion depends on lysosomal secretion induced by calcium signaling events [14 , 37] , followed by compensatory endocytosis , which drives parasites into host cells [13] . To test whether impairment in T . cruzi entry was due to deficiency in lysosomal exocytosis in cells lacking LAMP-2 , we performed a lysosomal exocytosis assay using WT , LAMP1/2-/- and LAMP2-/- fibroblasts stimulated with Ionomycin , a calcium ionophore . Cells were exposed to Ionomycin in two different concentrations , 5 and 10μM and lysosomal exocytic events were measured by assaying β-hexosaminidase activity in cell culture supernatants . Supernatant of non-treated cells showed very little amounts of enzyme activity , as expected . On the other hand , treatment with Ionomycin did trigger lysosomal exocytosis in all cell lines , as it is demonstrated by the increase in β-hexosaminidase activity values in the cell supernatant upon treatment ( Fig 3A ) . Treatment with 5 or 10μM of the drug led to the same exocytic values in all cell lines . Additionally and most important , no difference in lysosomal content release was observed among the distinct cell lines either before or after cell stimulation with the drug , indicating that absence of LAMP does not affect host cell lysosomal exocytic ability and could not contribute to the invasion phenotype observed for LAMP-2 deficient cells . Once neither exocytosis nor vacuole morphology were compromised by LAMP deficiency , we decided to test whether compensatory endocytosis induced by these exocytic events , another important step of T . cruzi induced entry process [13] , was affected . Compensatory endocytosis was measured by FACS analysis after scrape wounding using a trypan blue quantitative quenching assay [33 , 38] . For this , WT , LAMP1/2-/- and LAMP2-/- fibroblasts were treated with WGA-Alexa Fluor 488 , upon which cell membranes were fluorescently labeled , submitted to scrape wounding in the presence or absence of calcium and allowed to recover from injury . In this process , scrape wounding of plasma membrane will induce lysosome secretion , which will trigger compensatory endocytosis carrying the damaged WGA-Alexa Fluor 488 labeled membrane to cell interior . Cell external fluorescence was then quenched by trypan blue , to keep only fluorescence from internalized membranes , and cells were read using FACS . In the absence of calcium , endocytosis events were minimal , while in the presence of calcium endocytosis reached its maximum , enhancing intracellular WGA- Alexa Fluor 488 fluorescence . As observed in Fig 3B , a significant increase in intracellular WGA- Alexa Fluor 488 fluorescence is observed when WT cells were submitted to scrape wounding in the presence of calcium . However , no increase in intracellular WGA- Alexa Fluor 488 fluorescence was observed when LAMP1/2-/- and LAMP2-/- fibroblasts were submitted to the same procedure , indicating that LAMP-2 plays a critical role in the process of membrane repair by interfering with compensatory endocytosis after lysosome secretion ( Fig 3B ) . Since the endocytic process triggered by lysosomal secretion during membrane injury events is dependent on lysosomal acid sphingomyelinase ( ASM ) action on the extracellular leaflet of PM , we decided to evaluate whether ASM was present in the lysosomal secreted content during PM repair in WT , LAMP1/2-/- and LAMP2-/- cells . As shown in Fig 3C , the enzyme was normally exocytosed from lysosomes in all cell lines during resealing of mechanical wounds provoked by scraping . Thus , the defect in endocytosis observed in LAMP-2 deficient cells was not due to lack of ASM secretion and it is most likely due to events downstream of lysosomal secretion . If endocytosis induced during plasma membrane repair was compromised in LAMP-2 deficient cells we should also observe a defect in these cells' ability to repair their plasma membrane . To test whether cells lacking LAMP were really deficient in repairing injured membranes , an event that requires compensatory endocytosis , we performed a plasma membrane repair assay using the membrane impermeable fluorophore propidium iodide ( PI ) . For this , cells were exposed to PI during or after scraping . In the first experiment , cells were exposed to PI during scraping to measure the efficiency in mechanical wounding of plasma membrane . Therefore , PI labeled cells provided the total number of cells injured by scraping from plates ( Fig 4A ) . Scraping in the presence of PI resulted in 94 . 7% of WT cells injured by scraping , while only 5 . 3% were not injured in this process . A similar amount of injured/non-injured cells was found for LAMP1/2-/- and LAMP2-/- ( 95 . 7% / 4 . 3% and 96 . 5% / 3 . 5% , respectively ) . In the second experiment , cells were exposed to PI after scraping in order to measure cells' ability in repairing injured membrane . In this case , PI labeled cells provided the total number of cells unable to recover from membrane injury . On the other hand , cells that excluded the fluorophore included the cells that were never injured and the ones that had been injured , but were able to repair their membranes ( Fig 4B ) . In order to calculate the percentage of cells that recovered from injury by membrane repair , for each cell type , we subtracted the percentage of not injured cells given by the first experiment ( Fig 4A , PI- ) from the total percentage of viable cells given by the second experiment ( Fig 4B , PI- ) . As shown in Fig 4A and 4B , 5 . 3% of cells were negative for PI when exposed to the fluorophore during scraping , while 36 . 8% of cells were negative for PI when exposed to the fluorophore after scraping , indicating that 31 . 5% of cells recovered from injury by plasma membrane repair . On the other hand , only 7 . 7% ( 12% - 4 . 3% ) of LAMP1/2-/- cells and 19% ( 22 . 5% - 3 . 5% ) of LAMP2-/- cells were able to recover from membrane injury . This result confirms the importance of LAMP-2 for the membrane repair process . In order to confirm that the decreased invasion in cells lacking LAMP-2 was due to their deficiency in compensatory endocytosis and not an inability of these cells to be injured by T . cruzi , we have also measured membrane injury , using the parasite as the source of membrane tear . For this , cells were exposed to T . cruzi trypomastigotes for 30 minutes in the presence of PI . Cultures not exposed to the parasite , but incubated with PI for the same amount of time , were used as controls in order to measure membrane injuries that may occur even in the absence of the parasite . As shown in Fig 4C , upon parasite exposure , it was possible to observe an increase in the number PI positive events for all cell lines ( WT , LAMP1/2-/- and LAMP2-/- ) when compared to control condition ( cell cultures without parasite exposure ) . WT cells exposed to T . cruzi showed about 7 . 3% more PI positive cells than its respective control , followed by LAMP1/2-/- with 9 . 9% and LAMP2-/- with 15 . 2% of PI positive cells ( Fig 4D ) . LAMP absence , especially LAMP-2 , had been previously shown to induce cholesterol accumulation in lysosomes [26 , 29] . The latter could compromise the levels of cholesterol delivered to the cell plasma membrane and consequently interfere with caveolin-1 distribution at the cell surface , which is important for the compensatory endocytosis process triggered during membrane repair [39] . Since plasma membrane cholesterol sequestration using MβCD has been shown to induce actin stress fiber formation [31] , we first decided to evaluate the actin cytoskeleton organization in WT , LAMP2-/- and LAMP1/2-/- using phalloidin staining of actin filaments . Analysis of the actin organization revealed that cells lacking LAMP-2 ( LAMP2-/- and LAMP1/2-/- ) showed a differential organization of actin stress fibers in their cytoplasm , especially at the cell periphery , strongly suggesting that cholesterol content at the plasma membrane level was compromised in these cells ( Fig 5A ) . In order to prove that the cholesterol content at the plasma membrane of cells lacking LAMP-2 was in fact lower when compared to WT cells , we prepared lipid extracts from cell plasma membrane and measured the cholesterol content . Cells lacking LAMP-2 showed a reduction of 70% in the levels of cholesterol at the cell plasma membrane when compared to WT cells ( Fig 5B ) . We also labeled caveolin-1 using an anti-cav1 antibody and a secondary conjugated with Alexa-Fluor-488 . WT cells show caveolin staining in cell interior , seen as small dots , as well as a very strong labeling at the cell surface ( Fig 6A , S3 Fig and S1 Video ) . On the other hand , in cells lacking both LAMP-1 and 2 no surface labeling is observed , only the doted labeling in cell interior ( Fig 6A , S3 Fig and S1 Video ) . LAMP2-/- cells show a profile similar to LAMP1/2-/- cells , although in the first it is possible to see some labeling at the cell surface in few cells ( Fig 6A , S3 Fig and S1 Video ) . In order to show that the amounts of caveolin produced by these cells were the same and only the distribution was different , we also evaluated by Western Blot the total amount of caveolin-1 in protein cell extracts . As it can be seen in Fig 6B and 6C , WT , LAMP1/2-/- and LAMP2-/- cells have the same amount of Caveolin-1 . In order to gain entry into host cells , T . cruzi stimulates them by interacting with their proteins and /or producing small injuries in their plasma membrane [10 , 12 , 40–42] . These events lead to the increase in intracellular calcium , which will in turn trigger lysosome exocytosis [43–45] . The latter is followed by a compensatory endocytic event that carries the parasite into the host cell [13] . Lysosomes have also been shown to be important for parasite retention inside cells [15] . Therefore these organelles have a pivotal role during T . cruzi invasion . Lysosomal proteins LAMP-1 and 2 have been shown before to interfere with parasite invasion , since LAMP1/2 knock out cells led to very low levels of cell infection . LAMPs are not only the most abundant , but also highly glycosylated proteins , rich in sialic acid , and estimated to cover about 80% of the luminal surface of this organelle [17 , 18 , 46 , 47] . Since sialic acid had been shown to be important for T . cruzi entry into host cells , it had been proposed that those residues could contribute to the observed LAMP knock out phenotype [16 , 21 , 48] . However , cells lacking only LAMP-2 were able to reproduce the invasion defect produced by abrogation of both LAMP proteins , strongly suggesting that LAMP-2 alone was responsible for the LAMP1/2-/- invasion phenotype and that sialic acid moieties , specifically from LAMPs , were most likely not important for this process . This was reinforced by the fact that parasitophorous vacuole morphology was not altered in LAMP2-/- or LAMP1/2-/- cells as compared to WT cells . It had been shown before by Lopez and co-workers ( 2002 ) that cells lacking sialic acid are less susceptible to infection and that this phenotype was a result of the formation of a looser vacuole , where membranes of parasite and vacuole were not tightly apposed [22] . Altogether these data reinforced that intrinsic characteristics of LAMP-2 , other than its sialic acid residues were important for T . cruzi invasion of host cells . In order to investigate how LAMP-2 was involved with T . cruzi host cell invasion , we evaluated the different steps involved with parasite internalization . First we investigated the ability of these LAMP2 knock out cells to perform lysosomal exocytosis . It had been shown before that lysosome fusion with phagosomes was somewhat disturbed in LAMP1/2-/- cells [49] , indicating that lysosome mobility could be affected upon loss of LAMP . We showed that this was not the case , since upon stimuli these LAMP knock out cells were able to induce lysosome exocytic events . These data corroborate previous work from our group , which had shown that no parasite loss was observed in LAMP1/2-/- cells [16] , as would be expected when lysosomal fusion is blocked [15] . On the other hand we showed that compensatory endocytosis triggered by lysosomal exocytosis is compromised in cells lacking LAMP-2 ( LAMP2-/- and LAMP1/2-/- ) . These compensatory endocytic events are extremely important during membrane repair , since they are responsible for removing the injured membrane and promoting membrane resealing , without which cells would die [33 , 38] . In fact , we showed here that LAMP-2 deficiency leads to more death of scraped injured cells . Although the latter could be explained by a compensatory endocytosis defect , it could also be a consequence of the fact that these cells are more prone to injury by scraping when compared to WT cells . The higher the number of injuries in one cell could lower its chance of membrane repair and recovery . In fact a larger number of cells presenting higher PI labeling values , was observed for cells lacking LAMP-2 . The same was observed when we used the parasite as the source of injury . This susceptibility to membrane injury is probably linked to the fact that cells deficient in LAMP-2 retain cholesterol in lysosomes [29 , 50] . LAMP-2 has been shown to bind cholesterol and help in its traffic to the plasma membrane [51] , leading to less cholesterol at the cell surface , as demonstrated here . The decrease in the levels of cholesterol at the cell plasma membrane leads to actin cytoskeleton reorganization and cell stiffening [31 , 52] , which could make cells more prone to mechanical injury . We then tested whether cells pre-treated with MβCD , a drug able to decrease cholesterol from cell plasma membrane and induce actin cytoskeleton rearrangement would also lead to increased cell injury and death . In the conditions tested here , even though MβCD treated cells were more prone to injury by scraping , they were almost as efficient as non-treated cells in recovering from injury ( see S2 Fig ) . Therefore , the inability of LAMP-2 deficient cells in recovering from injury should really be due to their inability to endocytose the injured membrane and not due to the fact that they are excessively injured . We further investigated why compensatory endocytosis events were compromised in cells lacking LAMP-2 . Compensatory endocytosis triggered by lysosomal exocytic events are dependent on the secretion of Acid Sphingomyelinase ( ASM ) , an enzyme that cleaves sphingomyelin into ceramide inducing its coalescence and membrane internalization [33] . The levels and secretion of ASM were not altered in LAMP-2 deficient cells . This corroborates previous data by Eskelinen and coworkers ( 2004 ) , which showed that although these LAMP-2 deficient fibroblasts present an accumulation of autophagic vacuoles no deficiency in protein degradation was observed , indicating that lysosomal enzyme content seemed to be unaffected by LAMP-1 or 2 deficiency [29] . Additionally , contrary to other cellular models of cholesterol storage defects , such as NPC ( Niemann-Pick type C ) patient cells , it has been demonstrated that LAMP1/2-/- cells do not show significant differences in the levels of sphingomyelin , ceramide and gangliosides at the cell surface when compared to WT cells . Therefore , not only ASM but also its substrate were available to trigger endocytosis in cells lacking LAMP-2 [50] . Thus subsequent events had to be responsible for the compensatory endocytosis defect . LAMP1/2-/- fibroblasts had also been shown to have a marked defect in the maturation of autophagosomes and phagosomes to degradative autolysosomes and phagolysosomes , indicating an alteration in lysosome mobility in these cells [29 , 49 , 53] . In the work by Huynh and coworkers , late endosomes/lysosomes , as well as phagosomes , show reduced ability to move on microtubules towards the cell center in LAMP1/2-/- cells , most likely due to impairment in the interaction between them . This reduced mobility could also be accounting for the reduction in the observed endocytic events . However , it has also been shown that in LAMP1/2-/- cells the phagosomes acquired Rab5 and accumulated phosphatidylinositol 3-phosphate normally , suggesting that the first steps of the endocytic pathways had not been altered when LAMPs are not present [49] . Moreover Schneede and co-workers have described that the maturation defect of autophagosomes and phagosomes to degradative autolysosomes and phagolysosomes in LAMP1/2-/- cells was not observed in MEFs lacking only LAMP-1 or 2 . Therefore mobility was not likely to be responsible for the reduced compensatory events observed for LAMP-2 deficient cells . Corrote and coworkers described that the endocytosis of PM wounds is dependent on caveolin-1 and caveolar structures , this being a major mechanism used by cells to reseal plasma membrane injuries [54] . Interestingly , our results showed that the distribution of caveolin-1 at the cell surface is seriously compromised in cells lacking LAMP-2 . Since caveolin-1 associates with cholesterol to form caveoale , cholesterol decrease in plasma membrane could be leading to dispersion of caveolin-1 as previously suggested [55] . Additionally , it has been shown that during caveolin traffic to the plasma membrane it accumulates in the medial Golgi , where it associates with cholesterol in order to be relocated to cell plasma membrane [56] . The altered intracellular traffic of cholesterol could be holding caveolin inside the cell and preventing its relocation to cell surface . In fact , we showed that the amount of caveolin produced in LAMP-2 deficient cells is the same as the one from WT cell . However , in LAMP-2 deficient cells caveolin is found only inside cells , apparently in small vesicles , most likely lysosomes as shown before [55] . Therefore , the lack of caveolin-1 at the plasma membrane could be responsible for the defect in compensatory endocytosis phenotype of LAMP2 knock out cells . This would also consequently account for less parasite internalization , especially considering that we have shown that T . cruzi was able to promote membrane injuries in all three cell lines , but failed to efficiently invade cells lacking LAMP-2 . The results shown here not only demonstrate the importance and mechanism by which LAMP-2 interferes with T . cruzi host cell entry but also indicate a major role for this protein in regulating plasma membrane repair . Consequently , it may help to understand the mechanisms involved with other diseases , related not only to LAMP-2 deficiency , such as Danon disease , but also with some lysosomal storage maladies or any genetic disorder in which impairment in plasma-membrane repair is observed .	Trypanosoma cruzi is the etiological agent of Chagas disease , a very debilitating illness that has no efficient treatment to date . Better knowledge of the mechanisms involved with host cell infection is important to change this scenario . T . cruzi enters host cells by subverting the mechanism by which cells repair small injuries in their plasma membrane . In this process , parasites interact with host cells causing membrane injuries . These injuries lead to secretion of lysosomal content to the extracellular media , which in turn causes the internalization of plasma membrane wounds via endocytosis . During this endocytic process the parasite is internalized by the host cell . We have previously shown that absence of two proteins located at the lysosomal membrane , LAMP-1 and 2 , severely interferes with T . cruzi entry into host cells . In the present manuscript we show that absence of LAMP-2 alone is enough to compromise T . cruzi entry into host cells and it does so by compromising the endocytic events that follow lysosome secretion , which in turn is responsible for driving T . cruzi into the host cell interior .	[ "Abstract", "Introduction", "Materials", "and", "methods", "Results", "Discussion" ]	[ "lysosomes", "medicine", "and", "health", "sciences", "viral", "transmission", "and", "infection", "cell", "processes", "microbiology", "parasitic", "diseases", "parasitic", "protozoans", "fibroblasts", "protozoans", "connective", "tissue", "cells", "cellular", "structure...	2017	LAMP-2 absence interferes with plasma membrane repair and decreases T. cruzi host cell invasion
Schistosomiasis is one of the most prevalent parasitic diseases , affecting millions of people in developing countries . Amongst the human-infective species , Schistosoma mansoni is also the most commonly used in the laboratory and here we present the systematic improvement of its draft genome . We used Sanger capillary and deep-coverage Illumina sequencing from clonal worms to upgrade the highly fragmented draft 380 Mb genome to one with only 885 scaffolds and more than 81% of the bases organised into chromosomes . We have also used transcriptome sequencing ( RNA-seq ) from four time points in the parasite's life cycle to refine gene predictions and profile their expression . More than 45% of predicted genes have been extensively modified and the total number has been reduced from 11 , 807 to 10 , 852 . Using the new version of the genome , we identified trans-splicing events occurring in at least 11% of genes and identified clear cases where it is used to resolve polycistronic transcripts . We have produced a high-resolution map of temporal changes in expression for 9 , 535 genes , covering an unprecedented dynamic range for this organism . All of these data have been consolidated into a searchable format within the GeneDB ( www . genedb . org ) and SchistoDB ( www . schistodb . net ) databases . With further transcriptional profiling and genome sequencing increasingly accessible , the upgraded genome will form a fundamental dataset to underpin further advances in schistosome research . Schistosoma spp . are platyhelminth ( flatworm ) parasites responsible for schistosomiasis , a tropical disease endemic in sub-tropical regions of Africa , Brazil , Central America , regions of China and Southeast Asia , which causes serious morbidity , mortality and economic loss . An estimated 779 million people are at risk of infection and more than 200 million are infected [1] . The paired adult males and females of S . mansoni reside in the hepatic portal vasculature , each female depositing 200–300 eggs per day near the intestinal wall . These eggs either pass into the gut lumen to be voided in the faeces and continue the life cycle or pass up the mesenteric veins and lodge in the liver , where they can cause serious pathology including granulomatous inflammation response and fibrosis . On contact with fresh water , free-living motile miracidia hatch from the eggs to infect aquatic snails ( Biomphalaria spp . ) , where parasites undergo two rounds of asexual multiplication and are released as infective cercariae into water . Cercariae infect the human host , by penetrating unbroken skin , and transform into schistosomula . After several days the parasites exit the cutaneous tissue via blood ( or lymphatic ) vessels and travel first to the lungs and onward into the systemic vasculature . They may make multiple circuits before arriving in the hepatic portal system; only then do they start to feed on blood , mature and pair up , the whole process taking approximately five weeks [2] . Two Schistosoma draft genomes ( S . mansoni and S . japonicum ) were recently published [3] , [4] and represent the only described genomes amongst parasitic flatworms to date . Their assemblies were generated by conventional capillary sequencing but are highly fragmented ( S . mansoni , 19 , 022 scaffolds; S . japonicum , 25 , 048 scaffolds ) and severely compromise gene prediction , as well as comparative and functional genomics analyses . The transcriptome has similarly only been partially characterised by large-scale expressed sequence tag ( EST ) sequencing and low-resolution cDNA-based microarrays . Second-generation sequencing technologies provide new opportunities to characterise both genomes and transcriptomes in depth . In addition to whole genome de novo sequencing [5] , [6] and genome improvement [7] , massively parallel cDNA sequencing ( RNA-seq ) can identify transcriptionally active regions at base-pair resolution [8]–[11] and accurately define the exon coordinates of genes [12] . In addition , the quantitative nature and high dynamic range of RNA-seq allows gene expression to be scrutinised [11] , [13] , [14] in a more sensitive and accurate way than other previous high-throughput methods [15] , [16] . In this study we systematically improved the draft genome of S . mansoni , using a combination of traditional Sanger capillary sequencing , second generation DNA sequencing from clonal parasites and reanalysis of existing genetic markers [17] . This allowed us to assemble 81% of the genome sequence into chromosomes . We have also used RNA-seq data from several life-cycle stages to refine the structures of 45% of existing genes as well as to identify new genes and alternatively spliced transcripts . In addition to cis splicing , our data highlight extensive trans-splicing and provide clear evidence that the latter can be used to resolve polycistronic transcripts . With RNA-seq we profiled the parasite's transcriptome during its transformation from the free-living , human-infectious cercariae to the early stages of infection and in the mature adult . As the infective form transforms into a mammalian-adapted parasite , the relative abundance of transcripts shifts markedly during a 24-hour period , from those involved in glycolysis , translation and transcription to those required for complex developmental and signalling pathways . The improved sequence and new transcriptome data are available to the community in a user-friendly and easy to query format via both the GeneDB ( www . genedb . org ) and SchistoDB ( www . schistodb . net ) databases . These data demonstrate that revisiting a previously published draft genome , to upgrade its quality , is an option that should not just be reserved for model organisms . S . mansoni clonal DNA was obtained from single miracidium infections of Biomphalaria snails . Male and female adults ( NMRI strain , Puerto Rican origin ) were obtained from infected C57Bl/6 mice . DNA extraction was performed and sequencing libraries were prepared as previously described [18] . Eight and lanes were sequenced for the male samples and one lane for the female sample , both as 108-base paired reads . For RNA-seq samples , total RNA samples were obtained from cercariae , 3 hours and 24 hours post-infection schistosomula , and 7-week old mixed sex adult worms . Schistosomula samples were obtained using mechanical transformation [19] . RNA-seq libraries were prepared using a modified version of the protocol described in [8] and sequenced as 76-base paired reads . All samples were sequenced using the Illumina Genome Analyzer IIx platform . Raw sequence data were submitted to public data repositories; DNA reads were submitted to ENA http://www . ebi . ac . uk/ena/ under accession number ERP000385 and RNA-seq reads were submitted to ArrayExpress http://www . ebi . ac . uk/arrayexpress/ under accession number E-MTAB-451 ) . The Arachne assembler ( version 3 . 2 , [20] ) was used to produce a new assembly using the existing capillary reads from the previously published draft assembly [3] , supplemented with an additional ∼90 , 000 fosmid and BAC end sequences . FISH-mapped BACs [3] were also end-sequenced generating 438 reads that were incorporated into the assembly . Illumina reads were used to close gaps with the IMAGE pipeline [7] . The sequences of 243 published linkage markers [17] of S . mansoni were retrieved and used as anchors within the assembly by incorporating them as faux capillary reads . Scaffolds containing these reads were ordered , orientated and merged into chromosomes . Except where indicated , all analyses reported in the present study refer to a frozen dataset taken at this stage of the assembly process ( S . mansoni genome v5 . 0 , available at http://www . sanger . ac . uk/resources/downloads/helminths/schistosoma-mansoni . html ) . All comparisons were made against the previously published draft genome ( v4 . 0 ) . As part of the active finishing process , we randomly checked ∼20% ( 2 , 062 ) of the gaps automatically closed by IMAGE and found 90% of these could be verified by visual inspection . Contigs containing telomeric repeat sequences ( TTAGGG ) [21] were extended by oligo-walking pUC clones until a unique sequence was identified . Where the unique sequence was linked to a known marker , the telomere could be placed onto a chromosome . All manual improvement changes were included in a subsequent snapshot of the data ( v6 . 0 ) . To transfer the existing annotation to the latest reference we used RATT [22] ( with the old assembly split into four parts and using options –q and –r ) to define regions with synteny between both assemblies and transform the annotation coordinates onto the new assembly . The annotated genome sequence was submitted to EMBL http://www . ebi . ac . uk/embl/ under the accession numbers HE601624 to HE601631 ( nuclear chromosomes ) ; HE601612 ( mitochondrial genome ) ; and CABG01000001 to CABG01000876 ( unassigned scaffolds ) . Each lane of RNA-seq reads was independently aligned to the genome using TopHat [23] and the resulting alignment files used as the input for the gene finder Cufflinks [12] . Transcript fragments with less than 10× average read depth coverage and fewer than 50 codons were excluded from subsequent analyses . JIGSAW [24] was used to combine existing models with Cufflinks' transcript fragments . The final set of gene models can be accessed through GeneDB http://www . genedb . org/Homepage/Smansoni and SchistoDB http://www . schistodb . net . RNA-seq read pairs that contained the splice leader ( SL ) sequence [25] were used to find trans-splicing sites; where a gene was found within 500 bases from a trans-splice site its transcript was tagged as putative trans-spliced . By looking for genes whose 3′ end was located within 2 , 000 bp upstream of a putative trans-spliced acceptor site , putative polycistronic units were identified . RT-qPCR was performed to validate both trans-spliced and polycistronic transcripts . RNA-seq reads were aligned to the reference genome using SSAHA2 [26] . A minimum mapping score 10 was applied to filter aligned reads . Reads per gene and RPKMs ( reads per Kilobase per million mapped reads [8] ) were calculated using only coding regions coordinates . We also estimated the background signal for non-coding regions ( RPKM<2 ) . Total reads per gene were used to identify differentially expressed genes ( using only genes with >background RPKM ) in pair wise comparisons ( adjusted p-value<0 . 01 – adjusted for multiple testing [27] ) using the edgeR package [28] implemented in the Bioconductor R-package [29] . Gene Ontology ( GO ) term enrichment analysis was performed with TopGO [30] , also implemented in R [31] . The procedures involving animals in the UK were carried out in accordance with the UK Animals ( Scientific Procedures ) Act 1986 , and authorised on personal and project licences issued by the UK Home Office . The study protocol was approved by the Biology Department Ethical Review Committee at the University of York . The procedures involving animals in the US were carried out in strict accordance with the Animal Welfare Act ( Public Law 91–579 ) and the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health ( OLAW/NIH , 2002 ) . The protocol was approved by the University of Texas Health Science Center Institutional Animal Care and Use Committee ( IACUC , Protocol Number: 08039x ) . Using the existing Sanger-sequencing data from the published draft genome [3] , supplemented with an additional ∼90 , 000 fosmid and BAC ends , we produced an improved version of the S . mansoni genome de novo using the Arachne assembler [20] . With only 885 scaffolds , the new assembly contains less than one-twentieth of the original number of scaffolds ( Table 1 ) . Half of the 364 . 5 Mb genome is represented in scaffolds greater than 2 Mb and 90% are over 0 . 5 Mb . Ordering and orientating scaffolds based on 243 available linkage markers [17] and end-sequences from FISH-mapped BACs [3] further improved the continuity of the genome . The largest scaffold of 10 Mb contains 8 microsatellite markers from Chromosome 6 and no ambiguities , i . e . , the order of the contigs in this scaffold is the same as the order of the markers in the linkage group . Chromosome 1 represents the largest placed chromosome of 79 . 6 Mb with 41 . 8 Mb of the sequences ordered and concatenated as a single scaffold . There were only 6 microsatellite ambiguities in the whole assembly and these were corrected by targeted manual finishing . We then used genomic DNA from a clonal adult male population ( see Methods , Text S1 and Figure 1A ) to reduce the number of gaps within scaffolds and generally improve the assembly . Using the Genome Analyzer IIx platform , we generated 11 Gb of 108-base paired reads , approximately 60-fold genome coverage . IMAGE [7] was then used to iteratively extend contigs into gaps by performing local assemblies of the Illumina reads ( Figure 1B ) . After 33 iterations with a range of k-mer sizes , IMAGE closed a total of 11 , 158 gaps ( 53 . 4% of all possible gaps ) . The closed gaps had an average length of 315 bp with the largest gap being 6 . 5 kb ( Figure S1 ) . The statistics of the improved new assembly are shown in Table 1 . Compared with the previous draft assembly , the number of contigs was reduced from 50 , 292 to 9 , 203 and the N50 was increased from 16 to 78 kb . Because the linkage markers were associated with much larger scaffolds , we were able to allocate an additional 84 Mb of consensus sequence data into individual chromosomes , bringing the total to 81% ( Figure S2 ) . The improvement is best illustrated in chromosome 6 , which consists of the largest and 5 smaller scaffolds in the new assembly , but corresponds to 1 , 537 scaffolds from the old assembly . Illumina reads from clonal worms , mapped to both assemblies , were also used to assess assembly improvement . Table 1 shows that the mapping statistics were broadly similar in both assemblies . However , in terms of absolute numbers , more reads mapped to the new assembly despite the total genome length having been reduced by ∼10 Mb . Further , an increased number of read-pairs mapped in their correct orientation , within a distance predicted by the sequencing library fragment size , indicating fewer mis-assemblies . Following assembly , the genome was further improved by manual finishing . In particular , 305 , 465 Sanger reads ( comprising repetitive sequences that were previously excluded by assembly software ) were manually incorporated , three more scaffolds were ordered into chromosome sequences , and 17 new contigs were assembled to further extend the ends of chromosomes . For example , by closing 33 gaps , one end of chromosome 6 has been extended by 1 . 4 Mb and now includes its telomeric tract . The S . mansoni genome has one pair of sex chromosomes . Females are the heterogametic sex with both Z and W chromosomes and males are homogametic with a ZZ pair . We found Z and W assembled together into 34 scaffolds , which could be ordered and orientated based on 51 previously reported genetic linkage markers [17] and comprised a total of 59 Mb . We used differences in coverage of reads mapped from male and female DNA , to identify both Z and W specific regions ( Text S1 ) . Approximately 30% of the Z/W chromosome was Z-specific ( Figure S3 ) and contained 23 Z-specific genetic markers [17] . Amongst the unplaced sequences that lack genetic markers , were an additional 69 Z-specific scaffolds ( >100 kb ) and a further 114 unplaced scaffolds ( ∼1 . 1 Mb ) that were W-specific . Repeats comprise 90% of the latter , and include previously identified female-specific repeat [32] as well as 0 . 1 Mb of previously uncharacterised female-specific sequences . These scaffolds usually have female reads mapped many fold higher than the average coverage of the assembly , for example scaffold 1570 has 26 times higher coverage than the average , suggesting that the heterochromatin portion of the W chromosome have been collapsed into these scaffolds . Based on the differences between the genome-wide assembly coverage and the coverage of these scaffolds , we estimate these heterochromatin portions of the W chromosome to comprise ∼3 . 3 Mb collapsed into the 1 Mb of consensus . Interestingly , the W-specific scaffolds appear to contain no coding genes whereas the Z-specific portion of Z/W sequence contains 782 genes , ∼95% of which exist as single-copies within the assembly . Amongst the unassembled reads there were 5 , 647 that originated from mitochondrial DNA . An independent assembly of these reads using CAP3 [33] generated a single contig of 21 kb ( to which 15 scaffolds from the previous genome assembly could be aligned ) . The first 14 kb of the contig was 99 . 9% identical to the published coding portion of the S . mansoni mitochondrial genome [34] . Based on restriction fragment analysis , a long non-coding region that is repetitive and highly variable between individuals has previously been partially characterised [35] . In our data , the additional 9 kb non-coding portion of the mitochondria genome is now complete and comprises known 62 bp repeats [35] , plus additional 558 bp repeats and long tracts of low complexity sequence . We obtained total RNA from four time points of the life cycle of S . mansoni: 1 ) free-living mammalian-infectious cercariae , mechanically transformed schistosomula at 2 ) three hours and 3 ) twenty hours post infection , and 4 ) seven-week old mixed-sex adults recovered from hamster host . The 183 million 76-base RNA-seq read pairs were mapped to the new reference genome using SSAHA2 alignment tool [26] . An average 70% of the RNA-seq reads generated in each sequenced library aligned as proper pairs to the genome ( Table 2 ) , an improvement over the previous version of the genome . Less than 6% of reads mapped to the mitochondrial genome in each sample; the lowest ( 0 . 5% ) corresponding to the schistosomula stages . The majority ( 91% ) of the 11 , 799 gene models from the previous version of the genome could unambiguously be transferred onto the new assembly . Splitting gene models from the previous assembly increased the gene count by 307; however , the coalescence of genes previously located on multiple different scaffolds caused some redundancy ( an example is shown in Figure 2 ) , removal of which reduced the number of transferred genes to 10 , 123 . Of the 1 , 065 genes that could not be transferred to the new assembly , at least 83% were presumed to represent incorrect annotations due to a lack of sequence similarity and their short lengths , 1- or 2-exon structures ( Figure S4 ) or a lack of start or stop codons . RNA-seq data has been used to refine and improve gene model predictions in various organisms [10] , [36] , [37] . In the first draft of the S . mansoni genome , gene models were generated using a combination of ab initio gene predictions and EST evidence [38] , with only a few hundred manually curated genes . To systematically upgrade the quality of annotations , we aligned pooled RNA-seq reads using TopHat [23] , which allows gaps in the read-to-reference alignment at putative splice sites . Using the upgraded genome sequence 30% more RNA-seq reads with putative splice junctions aligned , highlighting putative new genes or structural refinements that could be made to existing genes . Cufflinks [12] was used to aid the refinement of gene structures by creating transcript “fragments” with sharply defined exon boundaries [23] . Using transcript fragments with at least 10 reads coverage at each base we found 78% of previous gene models had evidence of transcriptional activity within the sampled life cycle stages . Of these models , 3 , 604 ( 45% ) were modified to include new exons derived from RNA-seq data , hence generating alternative gene predictions ( Table 3 ) . Using the transcript data as a guide , 236 genes were merged and 26 split into two or more gene models . To assess the accuracy of gene models , we calculated two metrics: the proportion of intron-exon junctions found in previous models that matched to the same intron-exon junction in a transcript fragment , and the proportion of the coding sequence in previous models that overlapped with the transcript fragments . Figure 3A is a heatmap showing these two metrics; existing models are clustered around top right of the plot , which indicates that RNA-seq evidence-based transcript fragments are similar to the existing models . Sixteen percent of gene models were perfectly reproduced by the transcript fragments ( Figure 3B ) , while 90% of gene models with transcriptional evidence have at least 70% of the coding region covered by the transcript fragments . In the new dataset , only 53% of gene models have at least 70% of their exon boundaries preserved . There are two main reasons for this low specificity in predicting exon boundaries . First , Cufflinks was unable to successfully predict the small introns typically observed in the 5′ end of many S . mansoni genes ( Figure 3C and [3] ) . Consistently , when the first four exons of the old gene models were excluded , we found that transcript fragments could perfectly predict 90% of exon boundaries . Second , sequencing errors in the previous assembly resulted in introns being falsely incorporated into gene models during prediction to compensate for apparent frameshifts . These “intron” sequences are no longer necessary to preserve the reading frame and were identified as part of exons by Cufflinks in the new assembly ( Figure 3D ) . For the two reasons above , we used JIGSAW [39] to combine existing models with those produced from RNA-seq data , resulting in 1 , 264 exon coordinates being changed . We identified 1 , 370 transcripts corresponding to putative full length coding sequences but which did not overlap with existing gene models . To check whether they indeed represented novel genes , we first screened them against known repeats and transposable elements . The 36 previously published transposable element sequences in S . mansoni matched 866 of the transcribed fragments , the longest of which ( 5 , 061 bp ) was 99% identical to the coding portion of the LTR retrotransposon Saci-1 [40] . Of the remaining 504 complete transcript fragments we found sequence similarity for 231 in the NCBI nr protein database , mostly to other genes already annotated in S . mansoni ( presumably representing gene duplications or members of multi-gene families ) or S . japonicum . However , seven out of the remaining 273 full-length transcript fragments did show at least one conserved domain: a putative Tpx-1/SCP related allergen , a rhodopsin-like GPCR domain , a DNA-protein interaction domain , a epidermal growth factor-like ( EGF-like ) domain , and a polypeptide encoding a fascicline-like domain ( FAS1 ) domain ) , and two transcripts with ArsR transcriptional regulator sequences . The new transcript fragments were on average shorter ( 261 bp ) and exhibited unusual codon usage ( Wilcoxon rank sum test , p<0 . 01 , Figure S5 ) compared with a typical schistosome gene . Although we cannot rule out at this stage that the small set of atypical genes are non-coding RNA species , they are included in the total number of putative protein coding genes , which stands at 10 , 852 . Both cis and trans-splicing are used to produce mature transcripts in S . mansoni . By filtering for RNA-seq reads containing the spliced leader ( SL ) sequence [25] , strongly supported trans-splicing events could be mapped on a genome-wide scale and highlighted 1 , 178 ( ∼11% ) genes ( an example is shown in Figure 4A ) , a figure in close agreement with a previous prediction [41] . For validation , we randomly chose ten putative trans-spliced gene models and could verify the existence of their trans-spliced transcripts by RT-PCR ( Figure 4B , Table S1 ) . In many cases , mapping information suggests a second trans-splicing acceptor site , usually within 20–50 bases from the primary acceptor site , indicating that alternative splicing operates at the trans as well as cis levels . Using Gene Ontology enrichment [30] , we could find no particular functions or processes enriched within the trans-spliced genes , agreeing with the previous report [41] . Polycistronic transcripts originate from a single promoter but are later processed to generate two or more individual mRNAs . This type of transcriptional regulation is characteristic of trypanosomatids [42] and is present in C . elegans [43] and other organisms [44] . It has been suggested [45] that the S . mansoni Ubiquinol-cytochrome-c-reductase ( UbCRBP ) and phosphopyruvate hydratase ( Smp_024120 and Smp_024110 respectively ) genes might be transcribed as a polycistronic unit and that trans-splicing of the phosphopyruvate hydratase might resolve the polycistron into individual transcripts . In our study we provide strong evidence that this is indeed the case . One of the characteristics of polycistronic transcripts is a short intergenic distance ( <200 bp ) between individual “monocistrons” . We identified a total of 46 trans-splicing acceptor sites that fall between gene models that have a maximum intergenic distance of 200 bp , and 115 cases ( Figure 4C , Table S2 ) where the intergenic regions expands up to 2 kb ( maximum reported for C . elegans ) . We validated four of these polycistrons by RT-PCR ( Figure 4D , Table S1 ) and Sanger sequencing ( data not shown ) . Unlike C . elegans , which uses a second spliced leader ( SL2 ) to resolve polycistrons [43] , S . mansoni seems to use the same SL for both polycistronic- and non-polycistronic- trans-spliced transcripts . The role of polycistrons in schistosome gene expression remains to be determined but no pattern could be discerned between the ascribed functions of genes within each polycistron . In order to profile the transcriptional landscape of the parasite establishing in the mammalian host , the RNA-seq data from four key time points in the parasite's life cycle were analysed independently . Consistent with RNA-seq experiments elsewhere [16] , we found good reproducibility between biological replicates , indicated by high correlation coefficients ( average Pearson correlation of log RPKM values , across five pairs of biological replicates , was 0 . 95; Figure S6 ) . A total of 9 , 535 ( 88% ) genes were expressed ( above an empirically determined background RPKM cut-off of 2 – Text S1 and Figure S7 ) in at least one surveyed time point and the remaining 12% were regarded as genes with expression too low to be detected or expressed during life stages not surveyed in this study ( e . g . intra-molluscan stages ) and therefore were excluded from further analysis . Of the excluded genes , 65% are annotated as hypothetical proteins ( higher than the genome-wide figure of 44% ) . To gain better insight into the resolution of the RNA-seq approach in S . mansoni , we compared our results with a few example genes that have been described to undergo pronounce changes in their expression along the parasite's life cycle: an 8 kDa calcium binding protein , associated with tegument remodelling during cercariae transformation into schistosomula [46] , [47]; a heat shock protein 70 ( HSP70 ) , active in schistosomula after penetration through mammalian host skin [48]–[50]; and the tegument antigen Sm22 . 6 [51] , associated with resistance to re-infection in adult patients of endemic areas [52] . Our RNA-seq results broadly agree ( Figure 5 ) with relative gene expression measurements obtained through other approaches . We also investigated how well the RNA-seq data correlate with previous microarray studies [53] , [54] . Comparing normalised intensity values of the array features against the RNA-seq read depth for each microarray probe location in the genome ( Figure S8 ) suggests that these data broadly correlate ( Pearson's correlation of the log values 0 . 67 ) . A total of 2 , 194 genes had detectable expression in at least one stage but not another and were therefore differentially expressed . We also used a pair-wise approach to analyse genes differentially expressed between the following life cycle stages: cercariae vs . 3-hour schistosomula , 3-hour schistosomula vs 24-hour schistosomula , and 24-hour schistosomula vs . adult . A total of 3 , 396 non-redundant transcripts ( excluding alternative spliced forms ) were differentially expressed ( adjusted p-value<0 . 01 ) within the three pair wise comparisons ( Table 4 and Table S3 ) . An example showing differential expression between cercariae and 3-hour post-infection schistosomula is presented in Figure 6 . To obtain a broad overview of the biological changes occurring at the gene expression level , we used Gene Ontology term enrichment to identify annotated functions and processes that were overrepresented in genes that were statistically ( adjusted p-value<0 . 01 ) up- or down- regulated . Aerobic energy metabolism pathways were down regulated in schistosomules compared to cercariae and antioxidant enzymes were overrepresented in transcripts from adults . Three-hour post-infection schistosomula showed enrichment of transcripts involved in transcriptional regulation , G-protein coupled receptor ( GPCR ) and Wnt signalling pathways , cell adhesion and a considerable number of genes involved in potassium/sodium transport ( Table S4 ) . Most of the categories enriched at 3 hours post transformation persist through to 24 hours ( e . g . GPCR signalling pathways ) . A total of 165 proteins are found to be associated with GPCR signalling pathways ( annotated via GO ) . Of these , 30 and 18 were up regulated in 3 and 24 hours post-infection schistosomula , respectively , compared with cercariae . In order to investigate major processes occurring individually in each life cycle stage , we studied genes with expression above the 95 percentile in cercariae , 24-hour schistosomula and adults ( Figure 7 ) . Across the life cycle stages studied , some core cellular processes are consistently highly expressed , including glycolytic enzymes and protein translation but other broad changes are also apparent . Free-living cercariae utilise internal glycogen stores; accordingly genes involved in glycolysis and the tricarboxylic acid cycle ( TCA ) are highly expressed . After penetrating the skin and transforming into obligate endoparasites , the schistosomula switch to anaerobic metabolism [55] , [56] before aerobic metabolism partly resumes in the adult . These events are also reflected in the transcriptome . At the schistosomulum stage there is a switch to high expression of L-lactate dehydrogenase , while TCA cycle transcription markedly decreases . As noted above , the cercariae and adult samples have relatively high contributions from the mitochondrial transcriptome ( Figure S9 ) reflecting the high energy-demands of these two stages . Other genes highly expressed in the schistosomula are involved in protein re-folding and chaperone function: 5 heat shock proteins ( Smp_008545 , Smp_035200 , Smp_062420 , Smp_072330 , HSP70/Smp_106930 ) are among the top 50 most expressed genes at this stage and may reflect a response to the rapid temperature rise between fresh-water ( ∼28°C ) , in which the cercariae are found , and the warmer mammalian host ( ∼37°C ) . Within the host , schistosomes are exposed to potentially damaging reactive oxygen species produced during metabolism . Consistent with previous work [57] we found that antioxidant enzymes - particularly the peroxiredoxins ( Prx1 , Smp_059480 and Prx2 , Smp_158110 ) - are highly expressed in adults , 24 hours after transformation and for Prx1 , as early as 3 hours after transformation . Our results highlight the advantages of RNA-seq transcriptome profiling , especially its ability to dramatically improve the gene annotation alongside accurately recording changes in gene expression . In 2009 a draft genome of S . mansoni was published and provided a major resource for gene discovery and data mining . Our motivation for this study was to take S . mansoni's genome to the next level , to systematically upgrade its draft sequence so that gene structures can be more accurately predicted and the genomic context of genes can be better explored . Although systematic manual finishing has occurred for some parasite genomes , it is not an economically viable option for most non-model organisms . The genome of S . mansoni is approximately 10 times larger that the genomes of protozoan parasites and is set in the context of a field that attracts less funding . Although additional “traditional” targeted , long-range capillary sequence was introduced , more than 40 , 000 gaps were closed simply by re-sequencing at deep coverage , from a low-polymorphic population of adult worms . Further substantial changes were made from re-evaluating existing genetic marker information . As a result , the genome is measurably more accurate and its continuity has been transformed; 81% of the data is now assembled into chromosomes . We have also upgraded the annotation using deep coverage RNA-seq . Compared with the 2009 draft genome , the net change in the gene content is that there are now ∼900 fewer genes . However , 500 genes are new and more than 1600 low confidence or erroneous predictions have been removed . Across the genome , more than one third of genes now have new sequences . The value of the genome resource will therefore be tangibly improved: data mining approaches to identify genes will be more sensitive and trawling through kilobases of sequence for missing exons will be come less common . Our results also highlight the major benefit of using RNA-seq for transcriptome profiling - its ability to dramatically improve the gene annotation , whilst accurately recording changes in gene expression . We see major expected changes , for example , the well described metabolic switch on host penetration , plus some previously overlooked ones , such as a battery of receptors up regulated at the onset of infection in the mammalian host . Our data also define with high resolution some of the important building blocks of the schistosome transcriptome – long transcripts , cis and trans-splicing , and for the first time , clear evidence of the trans-splicing being used to resolve polycistrons . By increasing the quality of the genome , we have increased the utility of our RNA-seq data and taken it well beyond the levels attainable by previous microarray approaches . Although only a broad view of gene expression changes are presented herein , the resolution of our analyses reflects the functional annotation that has been previously ascribed . The true value of these data will arise from their use within the context of genome databases such as GeneDB and SchistoDB to query the behaviour of specific genes or groups of genes . The quality of a genome directly influences the uses to which it can be put and with many more , low-cost , draft-genome sequencing projects underway , the requirement for higher quality reference material , is increasing . Chain et al . 2009 recently defined several levels or standards for genome assemblies [58] . In the present study , we have taken an existing draft genome and demonstrated that in relatively modest period of time it can be upgraded to annotation-directed grade using second generation sequencing technology without the need for extensive manual finishing . The much improved genome assembly and gene structures , along with the expression data , are available at GeneDB and SchistoDB and will be an excellent resource not only for the helminth research community but also for in depth comparative genomics studies across metazoa .	Schistosomiasis is a disease caused by parasitic blood flukes of the genus Schistosoma . Human-infective species are prevalent in developing countries , where they represent a major disease burden as well as an impediment to socioeconomic development . In addition to its clinical relevance , Schistosoma mansoni is the species most widely used for laboratory experimentation . In 2009 , the first draft of the S . mansoni and S . japonicum genomes were published . Both genome sequences represented a great step forward for schistosome research , but their highly fragmented nature compromised the quality of potential downstream analyses . In this study , we have substantially improved both the genome and the transcriptome resources for S . mansoni . We collated existing data and added deep DNA sequence data from clonal worms and RNA sequence data from four key time points in the life cycle of the parasite . We were able to identify transcribed regions to single-base resolution and have profiled gene expression from the free-living larvae to the early human parasitic stage . We uncovered extensive use of single transcripts from multiple genes , which the organism subsequently resolves by trans-splicing . All data from this study comprise a major new release of the genome , which is publicly and easily accessible .	[ "Abstract", "Introduction", "Materials", "and", "Methods", "Results", "Discussion" ]	[ "biology", "genomics", "genetics", "and", "genomics" ]	2012	A Systematically Improved High Quality Genome and Transcriptome of the Human Blood Fluke Schistosoma mansoni
Mass vaccinations are a main strategy in the deployment of oral cholera vaccines . Campaigns avoid giving vaccine to pregnant women because of the absence of safety data of the killed whole-cell oral cholera ( rBS-WC ) vaccine . Balancing this concern is the known higher risk of cholera and of complications of pregnancy should cholera occur in these women , as well as the lack of expected adverse events from a killed oral bacterial vaccine . From January to February 2009 , a mass rBS-WC vaccination campaign of persons over two years of age was conducted in an urban and a rural area ( population 51 , 151 ) in Zanzibar . Pregnant women were advised not to participate in the campaign . More than nine months after the last dose of the vaccine was administered , we visited all women between 15 and 50 years of age living in the study area . The outcome of pregnancies that were inadvertently exposed to at least one oral cholera vaccine dose and those that were not exposed was evaluated . 13 , 736 ( 94% ) of the target women in the study site were interviewed . 1 , 151 ( 79% ) of the 1 , 453 deliveries in 2009 occurred during the period when foetal exposure to the vaccine could have occurred . 955 ( 83% ) out of these 1 , 151 mothers had not been vaccinated; the remaining 196 ( 17% ) mothers had received at least one dose of the oral cholera vaccine . There were no statistically significant differences in the odds ratios for birth outcomes among the exposed and unexposed pregnancies . We found no statistically significant evidence of a harmful effect of gestational exposure to the rBS-WC vaccine . These findings , along with the absence of a rational basis for expecting a risk from this killed oral bacterial vaccine , are reassuring but the study had insufficient power to detect infrequent events . ClinicalTrials . gov NCT00709410 The recombinant cholera toxin B subunit , killed whole-cell oral cholera ( rBS-WC , Dukoral ) vaccine , has been found to be safe and protective in a range of settings over the last 30 years [1] , [2] , [3] . This vaccine is mainly used by tourists visiting endemic areas [4] where the control of cholera has traditionally been based on safe water supply , sanitation and health education [5] . A more affordable oral cholera vaccine which could be used more widely in endemic settings has recently been developed , licensed , and prequalified for purchase by UN agencies [6] . This second generation killed oral cholera vaccine ( Shanchol ) is composed of a different set of V . cholerae strains than the rBS-WC vaccine , includes not only O1 but also an O139 strain , does not include the recombinant B subunit ( rBS ) , therefore does not require buffer for administration , and has afforded 66% protection during a 3 year trial in Kolkata , India [7] . In early 2010 , the Strategic Advisory group of the World Health Organization ( WHO ) recommended that oral cholera vaccines be used preventively as well as reactively in the management of cholera outbreaks [8] . Since cholera tends to affect all age groups in endemic settings and during outbreaks , mass vaccination is considered an important vaccine deployment strategy . To achieve maximum impact of mass cholera vaccination , it is crucial to immunize the highest possible percentage of the population at risk . This includes women in the reproductive age group , defined here as being between 15 and 50 years old . In endemic and epidemic settings , women are at high risk for cholera and other diarrhoeal diseases , not least because mothers tend to be exposed to infectious children [9] . Without prompt rehydration , cholera during pregnancy can result in abortions , premature childbirth and maternal death [10] , [11] . There are good reasons for women in the reproductive age group in endemic areas to participate in interventions that prevent cholera . Excluding potentially pregnant women from mass vaccination campaigns is logistically and ethically challenging . But administering oral cholera vaccines to this highly vulnerable population causes a dilemma since the safety of the vaccine during pregnancy has not been documented . There are several reasons why it is thought that oral cholera vaccines are unlikely to have a harmful effect on foetal development . First , the bacteria in the rBS-WC vaccine are killed and do not replicate . Second , the vaccine antigens act locally on the gastro-intestinal mucosa is not absorbed and does not enter the maternal or foetal circulation . Third rBS-WC vaccines don't trigger systemic reactions ( e . g . fever ) linked to abortions early in pregnancy . However , no actual safety studies of the rBS-WC vaccine in pregnancy have been carried out [12] . The uncertainty regarding the use of the vaccine during pregnancy has resulted in differing recommendations . The recommendations from the WHO state the following . “The primary targets for cholera vaccination in many endemic areas are preschool-aged and school-aged children . Other groups that are especially vulnerable to severe disease and for which the vaccines are not contraindicated may also be targeted , such as pregnant women and HIV-infected individuals . ” [13] . The package insert of Dukoral , states: “The effect of DUKORAL [Oral , Inactivated Travellers' Diarrhoea and Cholera Vaccine] on embryo-foetal development has not been assessed and animal studies on reproductive toxicity have not been conducted . No specific clinical studies have been performed to address this issue . The vaccine is therefore not recommended for use in pregnancy . However , DUKORAL is an inactivated vaccine that does not replicate . DUKORAL is also given orally and acts locally in the intestine . Therefore , in theory , DUKORAL should not pose any risk to the human foetus . Administration of DUKORAL to pregnant women may be considered after careful evaluation ofthe benefits and risks . ” The package insert of the second generation vaccine ( Shanchol ) uses similarly guarded language: “The vaccine is not recommended for use in pregnancy . However , Shanchol is a killed vaccine that does not replicate , is given orally and acts locally in the intestine . Therefore , in theory , Shanchol should not pose any risk to the human foetus . Administration of Shanchol to pregnant women may be considered after careful evaluation of the benefits and risks in case of a medical emergency or an epidemic . ” A mass oral cholera vaccination was conducted in Zanzibar in 2009 . Pregnant women were advised not to participate in the campaign . To assess whether any pregnant women had inadvertently received the vaccine , and to investigate birth outcomes , we visited all women residing in the study area and in the reproductive age group more than nine months after the last dose of the vaccine had been administered . The objective of the study was to determine whether there was any difference between the outcomes of pregnancies exposed and not exposed to the oral rBS-WC cholera vaccine . The study was conducted according to the principles expressed in the Declaration of Helsinki . Individual verbal consent was obtained from each respondent after the purpose of the study was explained . The Institutional Review Board of the Government of Zanzibar ( ZAMREC ) , of the International Vaccine Institute , Seoul , Korea , and the Research Ethics Review Committee of the World Health Organization , Geneva , Switzerland approved this project . The informed consent process was done in several phases . Community informed consent was obtained through meetings with the local leaders ( She has ) . A multistage community outreach campaign was conducted to disseminate information about the planned study activities . During the census , individual verbal informed consent was obtained prior to the interview of each household head or his or her representative . During the mass vaccination , individual verbal informed consent was obtained from each participant or from his or her guardian , if they were less than 18 years of age . In addition , verbal assent from children 12 to 17 years of age was obtained . The participants received information regarding the vaccine , including advice for children less than 2 years of age and pregnant women not to receive the vaccine . There was no screening for pregnancy prior to vaccine administration . The interview of pregnant women was closely linked with the census , for which oral consent was provided . Like the census interview , the interview of pregnant women posed minimal risks and oral consent was deemed appropriate . Provision of oral consent by each participant was documented in a logbook . The use of oral consent was approved by the ethics review boards . After the surveillance was completed the three ethics review boards were informed about the conduct and the findings of the birth surveillance . The archipelago of Zanzibar lies about 50 kilometres east of mainland Tanzania and consists of two main islands , Unguja and Pemba , as well as smaller islets . Zanzibar had a population of about 1 . 1 million in 2009 . In Unguja , we included the shehias of Chumbuni , Karakana , and Mtopepo , which are informal , urbanized areas extending from the capital , Zanzibar City also known as Stonetown . These shehias arose without the corresponding development of adequate water and sanitation facilities . In Pemba , we included the shehias of Mwambe , Kengeja , and Shamiani , located in the mainly rural southeast of the island . Each dose of the rBS-WC cholera vaccine ( Dukoral ™ , SBL Vaccine AB , Sweden ) consists of ca . 1×1011 vibrios [12]: The full dose of vaccine was mixed with 75 or 150 ml of buffer solution for participants aged from two to six years and over six years , respectively . A formal census was conducted from November to December 2008 , collecting demographic and socio-economic information . Verbal informed consent was obtained from the head of each household prior to the interviews . The number of household members , ownership of various capital goods and household building materials were recorded . Data was directly entered into handheld computers , also known as personal digital assistants ( PDA ) [15] . A unique identification number was assigned to each resident in the study sites . After the census was completed , household identification cards were distributed in early January 2009 . At the time of card distribution , all healthy , non-pregnant residents of the study sites who were two years of age and older were invited to participate in the mass vaccination campaign . Study residents were requested to bring their household identification cards when coming to a vaccination outpost to facilitate identification . In August 2009 , a second census was conducted in the study sites to update the study population database . The mass vaccination campaign was implemented by the Expanded Program on Immunization of the Zanzibar Ministry of Health and Social Welfare with WHO technical support . The first round of immunizations was conducted from January 11 to 26 , 2009 , the second round from February 7 to 16 , 2009 . The vaccine vial was shaken , opened and its contents poured into a cup with buffer solution and stirred . The participants drank the mixture under direct observation and completeness of ingestion was recorded . During the first round , a card was issued to each vaccine recipient to record the subject's name , age , address , household head , date of vaccination , and completeness of ingestion of the dose . At the time of dosing , this information was also recorded in a PDA-based vaccination registry . Only those who had received a first dose ( as documented in the vaccination card or the PDA registry ) were given a second dose of the vaccine . The birth surveillance was conducted more than 9 months after the mass vaccination campaign was completed , between January 15 and February 15 , 2010 . A list of all women between 15 and 50 years of age at the time of the vaccination campaign and living in the study area was prepared based on the study population database . Following training in study procedures fieldworkers visited the listed women and asked whether they had been pregnant in 2009 . Women who had been pregnant were asked about the following: the date of delivery , duration and outcome of the pregnancy based on their last menstrual period , number of deliveries , age of the last child born before this delivery , antenatal clinic attendance during this pregnancy and person who attended the delivery . Birth outcomes were described as miscarriage or live births . We further defined a miscarriage as either a spontaneous abortion or a stillbirth . A spontaneous abortion was defined as a termination of a pregnancy within 20 weeks of conception . A stillbirth was defined as a foetus born after 20 weeks of gestation without a pulse . Live births that died later during infancy were described as infant deaths . For live births , the disposition of the baby was recorded . During the visit the field worker asked whether the baby is free from recurring illness , without gross malformations , and is feeding , urinating , defecating , crying , sleeping and growing normally . For the purpose of this surveillance , a recurrent illness was defined as an illness lasting more than two weeks or occurring twice or more often [16] . Only illnesses requiring the attention of medical staff were included . A gross malformation was defined as a physical defect present in a baby at birth . It includes any abnormality visible on a naked baby ( e . g . cleft lip or palate , Down syndrome , spina bifida , limb defects , etc . ) . Whether feeding , urinating , defecating , crying , sleeping and growing was within the normal range was recorded according to the mother's definition . If the field worker considered the infant as sick or abnormal , the infant was seen by a paediatrician . The paediatrician completed a standardized history , physical examination and assessment and provided treatment or referral according to national guidelines . The field workers and paediatricians were blinded regarding the vaccination status of the mother . The information collected during birth surveillance was linked to the population census and vaccination databases . Receipt of the cholera vaccine during the mass immunization program was ascertained based on the vaccination database . Linkage to the vaccination registry was made blinded to pregnancy outcome . Baseline data on socio-behavioural , economic , and environmental variables were obtained from the census database . To calculate the date of conception , we subtracted the duration of the pregnancy ( as defined by the mother in weeks based on the last menstrual period ) from the date of delivery . The pregnancy was considered exposed to the vaccine if the period from conception to delivery included the dates when the woman received at least one vaccine dose . Additionally , because it is difficult to know the exact date of conception , we included pregnancies with calculated conception dates within two weeks before ingestion of the first vaccine dose as potentially exposed . A pregnancy was considered unexposed if the period from two weeks before the calculated conception date to the date of delivery did not include receipt of any oral cholera vaccine dose . We compared the frequency of adverse birth outcomes between exposed and unexposed pregnancies . The number of miscarriages , live infants and infant deaths ( birth outcomes ) among the exposed and unexposed pregnancies were initially compared using chi-square or Fisher's exact test , as appropriate . Characteristics of women who had exposed and unexposed pregnancies were compared using chi-square and Student's t-test for binary/categorical and continuous variables , respectively . In the assessment of the risk for negative outcomes ( miscarriage and infant sickness , abnormality or death ) , a stepwise elimination method was used to select variables most closely associated with exposure and non-exposure and to fit them into a logistic regression model . All p values and 95% confidence intervals were interpreted in a two-tailed fashion . Statistical significance was designated as a p value less than 0 . 05 . Stata/SE 8 ( Stata Corporation , Texas , USA ) was used for statistical analysis . The population census enumerated 14 , 564 women between 15 and 50 years of age residing in the study sites . During the birth surveillance , 13 , 736 ( 94% ) of this population were located and interviewed . Women who participated had a significantly different health care utilization pattern , tended to be from a lower socio-economic background as suggested by the possession of fewer capital items ( mobile phone , bicycle etc . ) , came from larger households and tended to be less well educated ( Table S1 ) . Out of the interviewed women , 1 , 453 ( 11% ) had a delivery in 2009; and 1 , 151 ( 79% ) of these deliveries occurred during the period where the foetus could have been exposed to the vaccine . The large majority 955 ( 83% ) out of these 1 , 151 mothers had not been vaccinated; the remaining 196 ( 17% ) mothers had received at least one dose of the oral cholera vaccine ( 82 received 1 dose , 114 received 2 doses ) . The flow of the pregnant women is shown in Figure 1 . We compared the outcomes of pregnancies exposed and unexposed to the cholera vaccine ( Table 1 ) . There was no statistically significant difference in the number of miscarriages among the exposed compared to the unexposed pregnancies [10/196 ( 5% ) vs . 27/955 ( 3% ) , adjusted odds ratio ( AOR ) 1 . 62 ( 95% confidence interval ( 95% CI 0 . 76 to 3 . 43 ) ] . Similarly , there was no statistically significant difference in the number of infant deaths among the exposed compared to the unexposed non-miscarriage pregnancies [3/186 ( 2% ) vs . 13/928 ( 1% ) , AOR 1 . 46 ( 95% CI 0 . 41 to 5 . 29 ) ] . The frequency of infant illness and abnormalities among the live infants verified by a paediatrician was 8/183 ( 4% ) among the exposed versus 46/915 ( 5% ) among the unexposed ( AOR 0 . 79 , 95th CI 0 . 36 to 1 . 75 ) . Logistic regression models , adjusted for variation in background characteristics , found no significant difference in the frequency of miscarriages , sickness or abnormality , and infant deaths between the exposed and unexposed pregnancies . We assessed the timing of the exposure to the cholera vaccine in relation to the gestational period of the ten miscarriages ( Figure 2 ) . Vaccine exposure occurred during the first trimester in three , during the second trimester in four , and during the third trimester in three pregnancies . We compared individual and household characteristics of the mothers who had exposed and unexposed pregnancies ( Table 2 ) . Pregnant women who participated in the mass vaccination campaign differed in several aspects from pregnant women who didn't participate in the vaccinations . The women who received the vaccine were significantly older , had had more deliveries , attended antenatal care less frequently , had more frequently lived in the same household during the past 5 years and lived in a larger household with lower socio-economic status as suggested by the ownership of capital items and household construction materials . This study found no significant increase in adverse events involving the foetus or newborn among pregnant women who inadvertently received killed oral cholera vaccine . Because the sample size was small , our findings cannot rule out the possibility that rBS-WC vaccine could cause adverse events during pregnancy , but the study provides reassurance that such events are not common . The findings from this study support the current recommendation that killed oral cholera vaccine is not contraindicated during pregnancy , but the decision to administer the vaccine should depend on the epidemiological context and after weighing the potential benefits and risks [12] . Randomized , controlled studies of the rBS-WC vaccine in pregnant women are ethically not justifiable , but future , larger mass vaccinations may allow further evaluation of birth outcomes after inadvertent exposure of pregnant women to the vaccine .	Pregnant women are more vulnerable to complications of cholera than other people . It would be helpful to include pregnant women in vaccination campaigns against cholera but pregnant women and their unborn children are highly vulnerable to the potential adverse effects of biological products such as vaccines . The safety of oral cholera vaccines in pregnant women has up to now not been evaluated . During a large mass cholera vaccination campaign in Zanzibar in 2009 , women were advised not to participate if they thought they may be pregnant . The large majority ( 955 or 83% ) of women residing in the study area who were to be pregnant during the 9 months following the vaccinations did not participate in the campaign . The remaining 196 ( 17% ) women received the vaccine . A comparison between vaccine exposed and unexposed pregnancies did not reveal any significant differences in outcome between the two groups . The small number of miscarriages , infant deaths and ill infants was similarly distributed between the two groups . These findings are reassuring but continued monitoring of this vaccine when given during pregnancy is recommended .	[ "Abstract", "Introduction", "Methods", "Results", "Discussion" ]	[ "medicine" ]	2012	Safety of the Recombinant Cholera Toxin B Subunit, Killed Whole-Cell (rBS-WC) Oral Cholera Vaccine in Pregnancy
Since the discovery of tumour initiating cells ( TICs ) in solid tumours , studies focussing on their role in cancer initiation and progression have abounded . The biological interrogation of these cells continues to yield volumes of information on their pro-tumourigenic behaviour , but actionable generalised conclusions have been scarce . Further , new information suggesting a dependence of tumour composition and growth on the microenvironment has yet to be studied theoretically . To address this point , we created a hybrid , discrete/continuous computational cellular automaton model of a generalised stem-cell driven tissue with a simple microenvironment . Using the model we explored the phenotypic traits inherent to the tumour initiating cells and the effect of the microenvironment on tissue growth . We identify the regions in phenotype parameter space where TICs are able to cause a disruption in homeostasis , leading to tissue overgrowth and tumour maintenance . As our parameters and model are non-specific , they could apply to any tissue TIC and do not assume specific genetic mutations . Targeting these phenotypic traits could represent a generalizable therapeutic strategy across cancer types . Further , we find that the microenvironmental variable does not strongly affect the outcomes , suggesting a need for direct feedback from the microenvironment onto stem-cell behaviour in future modelling endeavours . Heterogeneity among cancer cells within the same patient contributes to tumour growth and evolution . A subpopulation of tumour cells , called Tumour Initiating cells ( TICs ) , or cancer stem cells , has recently been shown to be highly tumourigenic in xenograft models and have some properties of normal stem cells . Evidence continues to emerge that TICs can drive tumour growth and recurrence in many cancers , including , but not limited to , brain [1] , breast [2] and colon [3] . These tumour types can be broadly classed as hierarchical as they have been posited to have a hierarchical organisation similar but not identical to non-neoplastic stem-cell ( SC ) driven tissues . In these hierarchical tumors , TICs can differentiate to produce non-TIC cancer cells or self-renew to promote tumor maintenance . As TICs have been demonstrated to be resistant to a wide variety of therapies including radiation and chemotherapy , the TIC hypothesis has important implications for patient treatments [4] . Specifically , the effect of current strategies on the tumor cell hierarchy should be defined , and TIC specific therapies are likely to provide strong benefit for cancer patients . In a simplified view of the tumour cell hierarchy , TICs can divide symmetrically or asymmetrically to produce two TIC daughters or a TIC daughter and a more differentiated progeny [5] , [6] , respectively . More differentiated TIC progeny which still have the capability of cell division and are similar to transient amplifying cells ( TACs ) in the standard stem-cell model and are capable of several rounds of their own symmetric division before the amplified population then differentiates into terminally differentiated cells ( TDs ) which are incapable of further division . This mode of division and differentiation , which we will call the Hierarchical Model ( HM ) is schematized in Figure 1 . In the HM , there are a number of cellular behaviours that govern the system . In this study , we choose to study three: the rate of symmetric versus asymmetric division of the stem cells ( ) , the number of rounds of amplification that transient amplifying cell can undergo before terminal differentiation ( ) , and the relative lifespan of a terminally differentiated cell ( ) . While it is a simplification of reality to study only these three parameters and leave out others ( for example: differing proliferation rates for the different cell types [7] or the differing metabolic demands of stem vs . non-stem cells [8] ) rigorous quantification of these parameters has been extremely difficult to pin down experimentally and so the majority of the work to describe them has been in silico . Most germane to the loss of homeostasis is the work by Enderling et al . [9] which showed the changes to the size of a mutated tissue ( tumour ) as they varied the number of rounds of amplification of TACs . Other recent work attempting to quantify the rate of symmetric to asymmetric division in putative glioma stem cells was presented by Lathia et al . [10] , who showed that this rate can change depending on the presence or absence of growth factors , suggesting yet another method by which a tissue can lose or maintain homeostasis: in reaction to microenvironmental change . A critical limitation of in vivo lineage tracing performed to date is an inability to determine the impact of microenvironmental heterogeneity on TIC symmetric division . While the HM appears to be quite straight forward , there is growing evidence of complexity to be further incorporated into the model . There are likely to be differences in the extent of TIC maintenance or the ability of tumour cells to move toward a TIC state . TICs appear to reside in distinct niches suggesting there may be differences in the biology of these cells , but defining differences in TICs is limited by cell isolation and tumour initiation methods . Prospective isolation of TICs relies on surface markers , including CD133 , CD151 and CD24 which can be transient in nature [11] , due to modulation by the tumour microenvironment [12] or methods of isolation [13] . Characterisation of these sorted cells then requires functional assays including in vitro and in vivo limiting dilution assays [14] . As the importance of TICs becomes more evident as it pertains to aspects of tumour progression like heterogeneity [15] , treatment resistance [16] , [17] , recurrence [18] and metastasis [19] , the need for generalizable therapeutic strategies based on conserved motifs in these cells grows . We therefore aim to understand how the phenotypic traits discussed earlier ( asymmetric division rate , allowed rounds of transient amplification and lifespan of terminally differentiated cells ) and microenvironmental changes ( modelled as differences in oxygen supply ) effect resultant tissue growth characteristics . To this end , we present a minimal spatial , hybrid-discrete/continuous mathematical model of a hierarchical SC-driven tissue architecture which we have used to explore the intrinsic , phenotypic , factors involved in the growth of TIC-driven tumours . We consider parameters that involve the rates of division of the cells involved in the hierarchical cascade as well as micro-environmental factors including space and competition between cell types for oxygen . We present results suggesting that there are discrete regimes in the intrinsic cellular parameter space which allow for disparate growth characteristics of the resulting tumours , specifically: TICs which form tumours that are unsustainable , TICs that are capable of forming only small colonies ( spheres ) , and TICs that are capable of forming fully invasive tumours in silico , just as we see diversity in biological experiments ( Figure 2 ) . A systematic parameter exploration of the three key parameters relating to vascularisation of the domain , symmetric vs . asymmetric division ( ) and progenitor division potential ( ) was performed . We also explored the parameter determining the lifespan of differentiated cells ( ) and found that the only impact of longer lifespans is an increase in the amount of time before the simulations reach a steady state , but does not change the qualitative nature of the results . These results are summarised in Figure 3 . Each of the three panels represents the results for a different degree of vascularisation ( 0 . 01 , 0 . 05 and 0 . 1 ) . A density of vascularisation of 0 . 05 would mean 12 , 500 oxygen sources in the domain . To determine the diffusion coefficient , we used the estimate of approximately 70 micrometers of effective oxygenation [20] . Each plot shows the total tissue size after 50 , 000 time steps as we change the proliferative potential of progenitor cells . Each of the lines shows a different rate of symmetric vs . asymmetric divisions . These results show that all these three parameters have a critical range where homeostasis is disrupted ( tumourigenesis ) . Figure 4 shows examples of the typical results produced by this model . Although the proliferation rates of all the cells remain the same , due to space constraints and the differences in , the population of TICs does not grow at the same rate as the non-stem population . Figure 4A shows an example of an unviable tissue ( parameters: , , and day ) where the vascularisation does not support the potential tissue size of that TIC , resulting in an area of hypoxia affecting the region that contains the TIC . That leads to the death of the stem cell and , eventually , the rest of the cells in the tissue . Figure 4B shows a case of slightly increased symmetric division , resulting in a dynamic homeostasis where cell birth and death is balanced so that tissue size remains relatively constant - which could represent the enigmatic dormant phase [9] . Finally , figure 4C shows an example where the system never achieves true homeostasis . In this case is slightly higher when compared with the previous example , suggesting a critical value at which overgrowth occurs . Over time , the number of TICs increases , allowing for the ‘tumour phenotype’: unconstrained growth . Although this leads to areas of hypoxia , cells survive in the periphery of the blood vessels and keep growing until they take over the entire domain . A plot of cell number versus time for each of these three examples are plotted in figure 5 . Unsurprisingly , the higher the vascularisation of the domain the greater the tissue size it can support . Past a certain threshold , however , the difference becomes negligible and more remarkably , the qualitative dynamics are unchanged by any change in the microenvironment . The same effect is evident in the other two parameters , the rate of symmetric vs . asymmetric division ( ) of TICs and the proliferative potential of TACs ( ) . Regardless of the vascularisation , disruption of homeostasis only occurs when the proliferative potential of TACs ( ) is below a maximum value of about 15 . For values of symmetric division ( ) above 0 . 3 , the values for in which this overgrowth occurs becomes even more restrictive with a range of approximately 10–15 . Interestingly , we observed a conserved decrease in overall tissue size for the highest value of symmetric division , , when the progenitor cells were allowed only 5 divisions ( ) . We believe this phenomenon represents a situation where the tissue is not able to grow to its potential as the stem cells themselves occupy too much space , and never allow the progenitors to contribute as much as they could to the overall population . This is a supposition however , and deserves closer study . These results are summarized in figure 3 . Of note as well: in no simulation did we observe the ‘tumour phenotype’ for a value of , suggesting something akin to a ‘phenotypic tumour suppressor’ function for this parameter . As observed biologically [10] , this rate is highly susceptible to changes in microenvironment , suggesting an extension of this minimal model to include the microenvironmental factors measured in that study . How to incorporate the changes observed in that study into a mechanistic HCA model however , is not trivial , and we reserve it for a future extension of this work . Further , our current model exists only in two dimensions . While our quantiative parameters are based on experimentally derived values , the claims we make are largely qualitative abstractions , however , we stress that the specific quantitative descriptions of cell fates are likely not yet accurate and could change if this model was in three dimensions . In this paper we have presented a simple two dimensional computational model of the HM of a TIC-driven tissue . Our results show that there are distinct regions in parameter space ( that directly correlate to the intrinsic TIC phenotype space ) that encode vastly different behaviour in the tissue ( or tumour ) arising from the TIC in question . These parameters represent different TIC phenotypes , and therefore do not represent any specific genetic mutation . In this way , we hope to generalise the intrinsic alterations which a TIC could undergo much in the same way that the hallmarks of cancer have generalised non TIC-specific alterations [21]: our end goal being the identification of treatment strategies to target these phenotypes to slow or stop the progression of TIC-driven cancers . Because of the difficulties in understanding TIC specific traits in vivo , the biological data to support these conclusions remains sparse . There have been some carefully undertaken in vitro experiments on single TICs in glioblastoma , a highly invasive and malignant brain tumour , which suggest that TIC specific division behaviour ( in our model ) is variable and changes based on environmental cues [10] . Further work has shown that the other microenvironmental cues , such as acidity [14] and hypoxia [22]–[28] can also alter the prevalence of the stem phenotype by utilising functional markers of stemness , but the mechanism for this increase is , as of yet , imperfectly understood . Our simulations do not show a significant TIC population dependence on vascular density ( ) , a surrogate for hypoxia , or a change in stem composition ( see Supplementary Table S1 ) , suggesting a flaw in the model . To rectify this , future iterations of this model should include direct feedback onto the cellular parameters from the microenvironment . We aim to parameterize this dependence by specific in vitro experiments designed to quantify this effect , rather than just elucidate its existence . Other future developments of this model should take into consideration the emerging body of work suggesting that the proportion of TICs within a tumour is directly affected by therapy and not just physiologic growth factor controls [29] . There is now evidence in several cancers to suggest that radiation increases the size of the TIC pool . Specifically , in breast cancer , it has been shown that radiation therapy induces non-stem cancer cells to de-differentiate into TICs [30] . Further , experimental studies have shown radiation increases the TIC pool in glioblastoma [31] , which has often been attributed to radiation resistance associated with differences in cell survival [16] . A new study by Gao et al . [32] , however , has shown in silico and in vitro that radiation can effect the symmetric to asymmetric division rate ( our intrinsic parameter ) , yielding further clues about the mechanism of this TIC pool expansion . Dedifferentiation due to treatment related microenvironmental factors has not yet been considered in any spatial theoretical models . Dedifferentiation due to ‘niche’ specific factors was studied by Sottoriva et al . [33] , who , using an agent based model , reported findings similar to ours: that the microenvironment made no significant change to the overall tumour growth dynamics . Beyond this single spatial study , the concept of SC dedifferentiation is gaining more and more attention in conceptual theoretical treatments [34] and has been modelled with a deterministic ordinary differential equation system for a well-mixed population of cells [35] . We , as well as others , find that the HM of tissue growth does not completely capture all the necessary dynamics that characterise cancer growth - but there is still a great deal of understanding to be gained from studying this formalism . To this end , we have performed a study of the factors related to TICs driving this dynamic and have identified several key factors which promote increased growth of the resultant tumour . Motivated by Hanahan and Weinberg [21] , who have simplified the myriad ( epi ) genetic alterations which a tumour can undergo into the hallmarks of cancer , we seek to distill the traits of TICs in a similar way . Specifically , our model suggests that the number of allowed divisions of TACs exhibits bounds outside of which tumour growth is unsustainable . This finding has been corroborated independently by recent work from Morton and colleagues [36] . Further , there is a specific balance of symmetric to asymmetric division which keeps tumours from overgrowing; almost acting as a phenotypic tumour suppressor . Indeed , changes in this rate have been recently hypothesized to underlie the increasing stem pool in glioblastoma after irradiation [32] , and could also hold a key to understanding tumour dormancy [9] . In summary , we have presented a minimal spatial Hybrid Cellular Automaton model of the HM of a TIC-driven tissue in which we have explored generalised TIC phenotypic traits and have identified several key cellular parameters which influence the overall tissue behaviour . While our model does capture a number of salient phenotypic characteristics of TICs that seem to be conserved , it fails to capture the recently observed changes in stem fraction secondary to microenvironmental perturbations . This is an indication that any computational model of a stem-hierarchical tissue , or tumour , built from this point on must not only include the physical microenvironment , but also feedback from the microenvironment onto the specific cellular parameters encoded in the HM . Therefore , this endeavour has identified the crucial point that the microenvironment must effect the behaviour of the cells within the HM , and also several conserved phenotypic hallmarks , which could be the result of any number of ( epi ) genetic alterations or microenvironmental perturbations . By focussing on mechanisms important for the HM of stem-cell driven tumour growth , we are seeking to identify common phenotypes which could be targeted in a variety of solid tumours in which TICs promote tumour maintenance - thereby reducing the number of therapeutic targets to a more tractable set . Only with this sort of distillation of the biological complexity inherent to cancer initiation ( and indeed progression ) can we hope to make progress against this disease . Our model is based on a hybrid , discrete-continuous cellular automaton model ( HCA ) of a hierarchically structured tissue . HCA models have been used to study cancer progression and evolutionary dynamics since they can integrate biological parameters and produce predictions affecting different spatial and time scales [15] , [33] , [37]–[40] . As shown in figure 6C , cells are modelled in a discrete fashion on a 500×500 2-D lattice . This comprises approximately where we assume a cell diameter of 20 micrometers [41] . The domain has periodic boundary conditions but the simulations are stopped when a cell reaches one of the boundaries . Every time step , cells are iterated in a random fashion as to avoid any bias in the way that cells are chosen . Figure 6A shows that , although all cells are assumed to have the same size and shape , they can only be one of three different phenotypes: TICs capable of infinite divisions , TACs which are capable of division into two daughters for a certain number ( ) of generations , and TDs which cannot divide but live and consume nutrients for a specified lifetime ( ) . Modes of division for TICs include asymmetric division ( with probability ) , which is division into one TIC daughter and one TAC daughter and symmetric division , which is division into two TIC daughters ( probability ) . The continuous portion of this model is made of up the distribution and consumption of nutrients ( in this case modelled only as oxygen ) . Vessels , which are modelled as point sources and take up one lattice point ( V in Equation 1 ) , are placed randomly throughout the grid at the intiation of a given simulation , in a specified density ( ) . Each of these vessels supplies oxygen at a constant rate ( ) which then diffuses into the surrounding tissue . The diffusion speed/distance is described by Equation 1 , where is the concentration of oxygen at a given time ( ) , and place ( ) , is the diffusion coefficient of oxygen , is the rate of oxygen production from a blood vessel , , , and are the rates at which TIC , TAC and TD cells consume oxygen . The difference in time scales that govern the diffusion of nutrients and that at which cells operate is managed by updating the continuous part of the model 100 times per time step . During each update the oxygen tension in a given grid point is updated with the values of the surrounding cells using a von Neumann neighbourhood modulated by the diffusion rate ( ) . ( 1 ) Any simulation performed by this model can be characterised by the parameters found in figure 7 . The most relevant parameters for the question we are trying to address are the following: In each case , as can be seen in figure 6 , a simulation is seeded with one TIC with a given set of intrinsic parameters ( , , ) governing its and its progenys behaviour , which is placed in the centre of the computational domain . The domain is initialised with as many randomly placed oxygen source points ( vasculature ) as described by the vascular density parameter ( ) .	In this paper , we present a mathematical/computational model of a tumour growing according to the canonical cancer stem-cell hypothesis with a simplified microenvironment . We explore the parameters of this model and find good agreement between our model and other theoretical models in terms of the intrinsic cellular parameters , which are difficult to study biologically . We find , however , disagreement between novel biological data and our model in terms of the microenvironmental changes . We conclude that future theoretical models of stem-cell driven tumours must include specific feedback from the microenvironment onto the individual cellular behavior . Further , we identify several cell intrinsic parameters which govern loss of homeostasis into a state of uncontrolled growth .	[ "Abstract", "Introduction", "Results", "Discussion", "Materials", "and", "Methods" ]	[ "complex", "systems", "medicine", "oncology", "computer", "science", "computer", "modeling", "mathematics", "applied", "mathematics", "cancers", "and", "neoplasms" ]	2014	Microenvironmental Variables Must Influence Intrinsic Phenotypic Parameters of Cancer Stem Cells to Affect Tumourigenicity
Human papillomaviruses ( HPV ) activate a number of host factors to control their differentiation-dependent life cycles . The transcription factor signal transducer and activator of transcription ( STAT ) -3 is important for cell cycle progression and cell survival in response to cytokines and growth factors . STAT3 requires phosphorylation on Ser727 , in addition to phosphorylation on Tyr705 to be transcriptionally active . In this study , we show that STAT3 is essential for the HPV life cycle in undifferentiated and differentiated keratinocytes . Primary human keratinocytes containing high-risk HPV18 genomes display enhanced STAT3 phosphorylation compared to normal keratinocytes . Expression of the E6 oncoprotein is sufficient to induce the dual phosphorylation of STAT3 at Ser727 and Tyr705 by a mechanism requiring Janus kinases and members of the MAPK family . E6-mediated activation of STAT3 induces the transcription of STAT3 responsive genes including cyclin D1 and Bcl-xL . Silencing of STAT3 protein expression by siRNA or inhibition of STAT3 activation by small molecule inhibitors , or by expression of dominant negative STAT3 phosphorylation site mutants , results in blockade of cell cycle progression . Loss of active STAT3 impairs HPV gene expression and prevents episome maintenance in undifferentiated keratinocytes and upon differentiation , lack of active STAT3 abolishes virus genome amplification and late gene expression . Organotypic raft cultures of HPV18 containing keratinocytes expressing a phosphorylation site STAT3 mutant display a profound reduction in suprabasal hyperplasia , which correlates with a loss of cyclin B1 expression and increased differentiation . Finally , increased STAT3 expression and phosphorylation is observed in HPV positive cervical disease biopsies compared to control samples , highlighting a role for STAT3 activation in cervical carcinogenesis . In summary , our data provides evidence of a critical role for STAT3 in the HPV18 life cycle . Human papillomaviruses ( HPVs ) are small , non-enveloped double-stranded DNA viruses that show a tropism for squamous epithelial cells of the skin epidermis , oral and ano-genital mucosa . Infection with HPV is associated with a spectrum of clinical manifestations ranging from common warts to cancers [1] . Whilst >200 types of HPV have been identified ( https://pave . niaid . nih . gov/ ) , only a sub-set of these are classed as high-risk due to their association with malignancy . High-risk HPVs are responsible for >99% of cervical cancer cases and a growing number of oropharyngeal carcinomas [2 , 3] . In particular , high-risk HPV16 and HPV18 are detected in >70% of cervical cancer cases and over 90% of other HPV positive cancers [4] . The HPV life cycle is intrinsically linked to the terminal differentiation programme of the epithelial tissues they infect , with productive replication restricted to differentiated suprabasal cells . Following infection of mitotically active cells within the basal layer of the epithelium , HPV genomes are established as low copy ( ~100 copies ) number episomes [5] . Upon differentiation , HPV infected cells remain active in the cell cycle and re-enter S/G2 phases for virus genome amplification . In the upper layers of the epithelium , infected cells exit the cell cycle and complete differentiation , enabling transit to the late stage of infection , where the late promoter is activated to drive late gene expression prior to virion production [5] . HPV replication is dependent on host factors , which are mainly controlled by the activities of the virus encoded E5 , E6 and E7 proteins . Whilst the role of the E5 protein is less understood [6] , the E6 and E7 oncoproteins are pivotal in the productive life cycle as well as in the development of anogenital cancers [7] . E7 proteins promote S phase re-entry in the differentiated strata via an ability to bind and inactivate the pocket family proteins pRb , p107 and p130 . These interactions result in release of the transcription factor E2F , causing cell cycle progression in cells that would normally be undergoing differentiation [8 , 9] . E6 recruits the cellular ubiquitin ligase E6-associated protein ( E6AP ) into a protein complex with the tumour suppressor protein p53 , resulting in its degradation [10 , 11] . In addition , high-risk E6 proteins bind and degrade a select group of PSD95/DLG/ZO-1 ( PDZ ) domain containing proteins [12] . Disruption of either of these functions interferes with the HPV life cycle [13–16] . Despite our increased understanding of the HPV life cycle , there is still a clear need to understand how HPV and the host cell proteins it manipulates determine the outcome of HPV infection to define novel strategies to treat HPV infections . STAT3 is a member of the signal transducer and activator ( STAT ) family of transcription factors that were originally discovered through their capacity to mediate transcription in response to interferons [17] . Binding of cytokines and growth factors to cell surface receptors results in activation of STAT3 , which typically involves phosphorylation of a tyrosine ( Y705 ) residue [18] . Phosphorylation is primarily mediated by receptor-associated kinases ( e . g . Janus family kinases ( JAK ) ) and receptor tyrosine kinases ( e . g . epidermal growth factor receptor ( EGFR ) ) , but can also be accomplished by non-receptor tyrosine kinases such as Src [19 , 20] . Tyrosine phosphorylated STAT3 molecules form activated dimers and translocate to the nucleus where they initiate a programme of gene expression controlling fundamental biological processes such as proliferation , apoptosis , immune regulation and differentiation [21] . STAT3 is also subject to stimulus-dependent phosphorylation of a serine ( S727 ) residue in its transactivation domain [18] . The function of this additional phosphorylation event remains controversial , as the modification has been reported to enhance and suppress STAT3 transcriptional activity [22 , 23] . In stratified epithelia , STAT3 drives proliferation and is required to maintain cells in an undifferentiated state by actively impairing differentiation [24–26] . Genetically engineered STAT3 knockout mice show reduced epithelial proliferation and suffer a compromised wound healing response [27] . In addition , both maintenance and renewal of keratinocyte stem cells is impaired . In contrast , mice engineered to express a constitutively active STAT3 in the proliferative compartment of the epithelium exhibit a disturbed gene expression profile resulting in an enrichment of expression of genes associated with epithelial to mesenchymal transition [28] . Keratinocytes from these mice display an increased survival rate in response to chemical insult or exposure to UV radiation [29] . Constitutive activation of STAT3 is also a feature of many human malignancies including cervical , and head and neck cancers , and is associated with a poor prognosis in various tumours [30] . Such aberrant activation of STAT3 contributes to tumour initiation , progression , immune evasion and metastasis [21 , 31] . Despite a clear link between STAT3 activation and HPV-associated cancers , it is not known whether STAT3 contributes to the productive HPV life cycle . This is surprising given the important role of STAT3 in maintaining an undifferentiated phenotype in keratinocytes . Here , we show that STAT3 phosphorylation is increased in HPV18-positive cells , and maintained during differentiation , through the activities of JAK and MAPK family kinases . Increased STAT3 activity is associated with host gene expression changes , including over-expression of proteins required for cell cycle progression and cell survival . Importantly , blockade of STAT3 activity through small molecule inhibitors , siRNA mediated depletion or over-expression of dominant negative phosphorylation null STAT3 mutants impairs viral gene expression and viral DNA replication in undifferentiated and differentiated cells . Finally , we demonstrate that phosphorylation is essential for STAT3 function in the HPV replication cycle and is frequently increased in HPV-positive cervical disease . Thus , STAT3 is a critical host factor required during the high-risk HPV life cycle . To investigate the role of STAT3 in the HPV18 life cycle , stable cell lines harbouring HPV18 episomes were generated from two primary foreskin keratinocyte donors . To exclude donor effects , all experiments were performed using both donor lines and representative data are presented . Levels of STAT3 phosphorylation were measured in normal human keratinocytes ( NHK ) and HPV18-containing cells by western blotting . In undifferentiated cells , phosphorylation of the major tyrosine ( Y705 ) and serine ( S727 ) residues was enhanced significantly in keratinocytes harbouring HPV18 compared to NHK donor controls ( Fig 1A , compare lanes 1 and 4 , Fig 1B , showing combined data from two donors; N = 8 p<0 . 05 ) . To ascertain whether STAT3 phosphorylation was further modulated during keratinocyte differentiation , monolayer cultures of NHK and HPV18-containing cells were cultured in high calcium media for 72 hours and samples taken for western blot analysis . Keratinocyte differentiation was confirmed by increased involucrin expression ( Fig 1A ) . Importantly , whilst subject to a decline , enhanced STAT3 phosphorylation was maintained at detectable levels in the HPV18-containing cells during differentiation ( Fig 1A , compare lanes 2 and 5 , 3 and 6 ) . In contrast , HPV18 had no effect upon the total levels of STAT3 protein in the primary keratinocytes ( Fig 1A ) . Comparable STAT3 phosphorylation changes were observed in both donor samples . Next , we confirmed our findings in an additional model of keratinocyte differentiation . NHK and HPV18-containing cells were stratified in organotypic raft culture for 14 days; this method recapitulates all stages of the HPV life cycle [32] . Raft sections were stained with antibodies detecting the total and phosphorylated forms of STAT3 ( representative sections from one donor shown in Fig 1C and 1D ) . Staining of total STAT3 protein was comparable between the donor-matched NHK and HPV18-containing rafts . S727 phosphorylated STAT3 was evident in both the basal and suprabasal layers of the NHK and HPV18 containing rafts . However , the level of the S727 phosphorylated form of STAT3 was elevated in the presence of HPV18 ( Fig 1C and 1D ) . We were not able to successfully stain raft cultures with the Y705 STAT3 phosphorylation specific antibody and so could not confirm the results obtained from our calcium differentiation assays . However , our data clearly demonstrate that primary keratinocytes harbouring HPV18 genomes exhibit increased STAT3 phosphorylation . Next , we hypothesised that a virus-encoded oncoprotein ( E5 , E6 and E7 ) promotes the increased STAT3 phosphorylation observed in the HPV18 life cycle model . To determine whether one or more of the HPV18 oncoproteins was responsible for STAT3 activation , the levels of STAT3 phosphorylation were measured by western blot analysis of C33a ( an HPV-negative cervical carcinoma cell line ) cells expressing individual GFP-tagged HPV18 oncoproteins ( Fig 2A ) . Interestingly , all three oncoproteins increased Y705 STAT3 phosphorylation compared to cells expressing GFP alone ( compare lanes 1 with 2 , 3 and 4 ) ; E5 by average of 3 fold ( p<0 . 05 ) , E6 by an average of 4 . 1 fold ( p<0 . 05 ) and E7 by an average of 3 . 6 fold ( p<0 . 05 ) ( Fig 2A , lanes 2–4 ) . Overexpression of both E5 and E7 led to a small increase in S727 STAT3 phosphorylation , however , E6 expression increased S727 phosphorylation significantly by an average of 5 . 6 fold ( p<0 . 01 ) ( Fig 2A , lane 3 ) . In agreement with our observations from primary keratinocytes , the presence of the HPV oncoproteins did not alter the level of total STAT3 protein , which remained similar to the GFP control . Next , we investigated whether the viral oncoproteins are responsible for the increased STAT3 phosphorylation observed in primary keratinocytes harbouring the HPV18 genomes . For this , we silenced E6 expression in the HPV18 life cycle model using a pool of E6-specific siRNA and compared the levels of phosphorylated STAT3 to those of cells transfected with a scrambled control siRNA ( S1A Fig ) . Treatment with the E6-siRNA resulted in a loss of E6 expression coupled to an increase in levels of the E6 target p53 ( S1A Fig ) . Importantly , the loss of E6 also correlated with a significant reduction in both Y705 and S727 STAT3 phosphorylation ( p<0 . 001 ) . Our E6 depletion approach also resulted in decreased E7 expression ( average of 55% ) ( S1A Fig ) . To rule out any contributory role for this oncoprotein , we tested whether targeted siRNA-mediated depletion of E7 also affected STAT3 phosphorylation . Treatment of HPV18 genome-containing cells with E7-specific siRNA reduced E7 expression by approximately 60% without impacting on either E6 expression or STAT3 phosphorylation ( S1B Fig ) . To test the contribution of the E5 oncoprotein we utilised our recently generated primary keratinocyte lines containing an E5 knockout HPV18 genome [33] . Using this system we showed that loss of E5 did not reduce STAT3 phosphorylation in either monolayer cultures differentiated in high calcium media or organotypic raft cultures ( S1C and S1D Fig ) . In summary , these data indicate that E6 is the predominant protein responsible for increasing the STAT3 phosphorylation observed in HPV18-containing keratinocytes . To address whether increased STAT3 phosphorylation correlated with enhanced STAT3 transactivation , we measured the promoter activity of two STAT3-dependent reporter plasmids . C33a cells were co-transfected with isolated HPV oncoproteins and reporter plasmids driving firefly luciferase from the β-casein [34] and pro-opiomelanocortin ( PomC ) [35] promoters . Expression of HPV18 E5 and E7 did not significantly increase STAT3-dependent luciferase expression . Conversely , expression of HPV18 E6 led to an average 8-fold increase in β-casein promoter-driven luciferase ( p<0 . 01 ) and a 5 . 5-fold increase in PomC-driven luciferase ( p<0 . 01 ) ( Fig 2B ) . Whilst these reporter constructs have been widely used to monitor the activation of STAT3 , they can also be responsive to other members of the STAT family of transcription factors [35 , 36] . In this regard , a recent study has shown that HPV activates STAT5 , which is necessary for HPV31 genome amplification [37] . To exclude the possibility that the observed increase in luciferase expression was a result of STAT5 activation , we used a pharmacological approach to specifically block the activation of STAT3 and STAT5 ( Fig 2C ) . Transfected C33a cells were treated with the STAT5 inhibitor pimozide [37] or two chemically distinct STAT3 inhibitors , cryptotanshinone and S3I-201 . Firstly , we confirmed that cryptotanshinone ( S2A Fig ) treatment did not affect STAT5 phosphorylation in HPV18-containing keratinocytes . Importantly , cells treated with pimozide showed no reduction in luciferase expression whereas in contrast , treatment with the two STAT3 inhibitors resulted in a significant ( p<0 . 01 ) reduction in β-casein-driven luciferase expression ( Fig 2C ) . We next assessed if E6 could induce the expression of endogenous STAT3-dependent gene products . Expression of HPV18 E6 in C33a cells led to increased expression of cyclin D1 and Bcl-XL , two characterised STAT3-dependent gene products ( Fig 2D ) . To confirm that the induction of cyclin D1 and Bcl-xL was STAT3-dependent , E6 expressing cells were treated with cryptotanshinone and S3I-201 . Treatment with either inhibitor reduced STAT3 phosphorylation at both tyrosine and serine residues , but had minimal impact on levels of total STAT3 protein . In addition , inhibitor treatment reduced cyclin D1 and Bcl-xL expression to control levels ( Fig 2D ) . Quantitative reverse transcriptase-PCR ( qRT-PCR ) revealed E6-dependent increases in cyclin D1 ( ccnd1 ) and Bcl-XL ( bcl2l1 ) mRNA transcripts , and these increases were sensitive to treatment with STAT3 inhibitors ( ccnd1 plus crypto p<0 . 01 and bcl2l1 plus crypto p<0 . 05 ) ( Fig 2E and 2F ) . Additional STAT3-dependent genes , including HIF1α and Survivin ( birc5 ) , were up-regulated by E6 in a STAT3-dependent manner ( hif1α plus crypto p<0 . 05 and birc5 plus crypto p<0 . 001 ) ( Fig 2G and 2H ) . Based on these results , we conclude that the E6 oncoprotein increases the transcription of STAT3-dependent genes . Whilst E6 proteins lack intrinsic enzymatic activities , those E6 proteins encoded by high-risk HPVs are able to interact with key cellular partners including E6AP , p53 and a number of PDZ-domain containing proteins to modulate cellular functions [38–40] . Therefore , we tested whether these interactions were required for the E6-mediated increase in STAT3 phosphorylation observed in HPV-positive cells . To this end , STAT3 phosphorylation was measured in cells expressing wild type and mutant HPV18 E6 proteins deficient in their ability to bind p53 , E6AP or PDZ domains . Valine substitution of phenylalanine at amino acid position four ( F4V ) generates an E6 protein incapable of destabilizing p53 and in a HPV18 life cycle model is deficient in supporting viral DNA amplification [41] . Whilst deficient for inhibiting p53 destabilization , the E6 F4V mutant enhanced STAT3 phosphorylation to levels comparable with wild type E6 ( Fig 3A; compare lanes 2 and 4 ) . E6 expression leads to the degradation of p53 by virtue of its interaction with the E6AP E3 ubiquitin ligase . Expression of an E6 mutant unable to interact with E6AP ( L52A ) [42] impaired p53 degradation , but retained the ability to enhance STAT3 phosphorylation ( Fig 3A; compare lanes 2 and 5 ) . To test whether E6 interaction with PDZ proteins was required for increased STAT3 phosphorylation , we engineered HPV18 E6 ΔPDZ , which lacks the C-terminal four amino acid PDZ-binding motif and cannot bind to PDZ domains [13] . When expressed in C33a cells , this mutant protein induced STAT3 phosphorylation to wild type E6 levels ( Fig 3A; compare lanes 2 and 3 ) . Finally , we confirmed that the PDZ-binding domain of E6 was not required for STAT3 phosphorylation in the context of the entire HPV genome in a differentiating epithelium . Organotypic raft cultures were generated from NHK harbouring wild type and ΔPDZ HPV18 genomes [13] . Rafts were stained with an antibody that detects the pS727 form of STAT3 . A similar pattern of STAT3 S727 phosphorylation was observed throughout the basal and suprabasal layers of the HPV18 wild type and ΔPDZ containing rafts ( Fig 3B ) . Together these data demonstrate that STAT3 phosphorylation is increased in cells expressing HPV18 by a mechanism independent of the p53 destabilizing and PDZ binding functions of E6 . Janus family receptor-associated tyrosine kinases are the most common kinases responsible for mediating STAT3 Y705 phosphorylation , which is deemed necessary for STAT3 activation [30] . Previous studies have shown that treatment of SiHa HPV16-positive cervical cancer cells with AG490 , a non-specific JAK2 inhibitor , prevented STAT3 Y705 phosphorylation [43] , however , such studies have not been performed in primary keratinocytes . To address this , we first confirmed that JAK were active in primary keratinocytes harbouring HPV18 by western blot analysis for tyrosine phosphorylated JAK2 . Levels of phosphorylated JAK2 were higher in HPV18-containing keratinocytes compared to NHK controls , and were retained during calcium-mediated differentiation ( Fig 4A ) . Next , we showed that increased JAK phosphorylation was E6-dependent by overexpressing a GFP HPV18 E6 fusion protein in C33a cells . Western blot analysis showed an increase in JAK phosphorylation in C33a cells expressing GFP-18E6 compared to GFP control ( Fig 4B; compare lanes 1 and 2 ) . Lastly , we incubated HPV18-containing keratinocytes with the highly specific clinically available JAK1/2 inhibitor Ruxolitinib [44] , or the JAK2 inhibitor Fedratinib [45] . Treatment with either inhibitor led to a marked reduction in STAT3 Y705 phosphorylation without affecting S727 phosphorylation ( Fig 4C; compare lane 1 to 2 and 3 ) . These results indicate that JAK2 mediates the STAT3 Y705 phosphorylation in HPV18-containing primary keratinocytes . A number of candidate STAT3 S727 kinases have emerged including; CDK5 [46] , mTOR [47] , NEMO-like kinase ( NLK ) [48] and several PKC isoforms [49 , 50] . However , the strongest evidence indicates that MAPK members mediate the serine phosphorylation of STAT3 [18] . The plethora of kinases capable of phosphorylating STAT3 implies that under physiological conditions this event might be restricted by cell type or the nature of the stimulus . We therefore set out to determine which kinases were responsible for STAT3 S727 phosphorylation in keratinocytes harbouring HPV18 genomes . We focused on the canonical MAPK members since S727 is embedded within a strong MAPK consensus sequence ( 725PMSP728 ) . To determine whether MAPK members could directly phosphorylate STAT3 , purified recombinant JNK1 , ERK2 and p38α were subject to an in vitro kinase assay by using bacterially expressed and purified STAT3 as the substrate . All three kinases successfully phosphorylated STAT3 ( S3 Fig ) . Whilst data generated from in vitro kinase assays demonstrates the ability of a kinase to directly phosphorylate a substrate , it is essential to confirm that the kinase in question is active in cell culture . To this end , we tested the activation status of each of the canonical MAPK members in differentiating keratinocytes . In these studies , levels of total ERK1/2 were increased in HPV18-containing keratinocytes compared to NHK controls ( Fig 5A; compare lanes 1–3 and 4–6 ) . In accordance with a key role in keratinocyte proliferation , keratinocytes displayed detectable ERK1/2 phosphorylation [51] ( Fig 5A ) . As expected , both basal and differentiation-induced ERK1/2 phosphorylation was greater in HPV positive cells ( compare lanes 1 and 4 , and lanes 3 and 6 ) . Whereas total p38 protein expression did not alter in the presence of HPV18 , the levels of p38 phosphorylation were greater in both undifferentiated and differentiated HPV18-containing keratinocytes ( Fig 5A; compare lanes 1 and 4 , 3 and 6 ) . In contrast with ERK1/2 and p38 , less is known of JNK1/2 regulation by HPV18; however , studies demonstrate that JNK1/2 signalling can prevent keratinocyte differentiation [52] . In keeping with this role , levels of JNK1/2 phosphorylation were highest in undifferentiated NHKs , and decreased rapidly upon differentiation in high calcium media ( Fig 5A; compare lanes 1 and 3 ) . JNK1/2 phosphorylation also declined in differentiating HPV18 positive keratinocytes but overall levels were noticeably higher than in NHK ( compare lanes 1 and 4 , lanes 3 and 6 ) . Together , these data demonstrate the presence of active MAPK members in HPV18 genome-containing cells . Moreover , in transient transfection experiments the phosphorylation of all three MAPK members was increased in C33a cells expressing GFP-18E6 compared to a GFP control ( Fig 5B ) . To test the roles of endogenous MAPK in the regulation of STAT3 , we compared the levels of STAT3 S727 phosphorylation in undifferentiated HPV18-containing cells treated with the p38 inhibitor VX-702 [53] , the MKK1/2 inhibitor UO126 ( prevents ERK1/2 phosphorylation and activation ) [54] and the JNK1/2 inhibitor JNK-IN-8 [55] , alone or in combinations ( Fig 5C and 5D ) . The STAT3 inhibitor cryptotanshinone served as a positive control in these experiments and the Mitogen and Stress activated protein Kinase ( MSK ) inhibitor SB-747651 acted as a negative control , given that in our tests MSK did not phosphorylate STAT3 in vitro ( S3 Fig ) . As shown in Fig 5C , STAT3 S727 and Y705 phosphorylation were significantly ( p<0 . 001 ) impaired by cryptotanshinone ( lane 2 ) but unaffected by the MSK inhibitor SB-747651 ( lane 3 ) . STAT3 Y705 phosphorylation remained intact in cells treated with the MAPK inhibitors , indicating that tyrosine phosphorylation of STAT3 is not dependent on active MAPK in keratinocytes . STAT3 S727 phosphorylation was largely unaffected in cells treated with VX-702 ( p38 inhibitor ) and JNK-IN-8 , and was only partially suppressed by the MKK1/2 inhibitor UO126 ( lanes 5 ( p<0 . 05 ) and 7 ( p<0 . 01 ) ) . Importantly , the combination of UO126 , VX-702 and JNK-IN-8 , blocking all three MAPK members , was required to significantly ( p<0 . 001 ) reduce STAT3 S727 phosphorylation ( lane 8 ) ( Fig 5D ) . Western blot analysis of the phosphorylated forms of the MAPK substrates MAPKAP-K2 ( p38 ) , MSK ( ERK1/2 ) and c-Jun ( JNK1/2 ) were used to demonstrate the efficacy of inhibitor treatment . These data indicate that several MAPK family members can phosphorylate STAT3 S727 in HPV18-positive keratinocytes . Given that STAT3 activation and the expression of STAT3-dependent genes were increased in HPV18-positive cells , we assessed if STAT3 was necessary for the virus life cycle . For this , STAT3 was depleted from primary keratinocytes harbouring HPV18 genomes using a panel of four commercially validated siRNAs . Each siRNA produced a reproducible depletion of STAT3 by an average of 38% compared to a scrambled control , and did not reduce levels of the STAT5 protein ( Fig 6A and S2B Fig ) . Despite the modest depletion of STAT3 , expression of the HPV18 E6 and E7 proteins was greatly reduced compared to the scrambled control ( Fig 6A ) . To assess if the inhibition of HPV18 oncoprotein expression occurred at the level of transcription , HPV18-positive keratinocytes were transfected with a pool of the STAT3 siRNA , and total RNA isolated and assayed by qRT-PCR for the levels of STAT3 , E6 and E7 transcripts . STAT3 depletion ( p<0 . 05 ) caused a significant decrease in early transcript expression ( E6 p<0 . 05 , E7 p<0 . 05 ) , indicating that STAT3 has a role in HPV gene expression in undifferentiated cells ( Fig 6B ) . To rule out the possibility that siRNA studies perturbed a crucial STAT3 protein scaffolding function [48] , we used small molecule inhibitors to specifically block STAT3 phosphorylation and transactivation whilst retaining levels of total STAT3 protein . For this , we monitored the level of HPV protein expression in undifferentiated HPV containing keratinocytes treated with increasing concentrations of the STAT3 inhibitor cryptotanshinone . As shown in Fig 6C , inhibitor treatment reduced STAT3 phosphorylation in a dose dependent manner without affecting total STAT3 levels . Importantly , cryptotanshinone treatment had no adverse effect upon keratinocyte viability during the term of the experiment ( S4 Fig ) . Concentrations of cryptotanshinone that blocked STAT3 phosphorylation also decreased E6 and E7 protein ( Fig 6C ) and mRNA transcript levels ( p<0 . 05 ) ( Fig 6D ) . Next , we wished to demonstrate the contribution of the individual Y705 and S727 phosphorylation events to STAT3 regulation of HPV oncogene expression . This was important given that an emphasis has been placed on Y705 phosphorylation as a critical step in STAT3 activation; however , under certain circumstances S727 phosphorylation is necessary for full STAT3 transactivation [18] . Dominant negative forms of STAT3 were used in which Y705 was replaced with a phenylalanine or S727 substituted with an alanine , generating mutants with a single unphosphorylatable site [56 , 57] . The STAT3 mutants were expressed in HPV18-containing keratinocytes , and over-expression of the Y705F mutant confirmed by western blot with an anti-FLAG antibody ( the pFugW-Y705F plasmid encodes a FLAG-epitope tagged STAT3 ) ( Fig 6E , lane 2 ) . Expression of the Y705F mutant corresponded with a loss of Y705 phosphorylation but had no overall impact on S727 phosphorylation ( Fig 6E , compare lanes 1 and 2 ) . Likewise , expression of S727A ablated the observed S727 phosphorylation but did not reduce Y705 phosphorylation ( Fig 6E , compare lanes 3 and 4 ) . Over-expression of the phosphorylation site mutants had negligible impact on STAT5 phosphorylation ( S2C Fig ) . Notably , expression of either of the phosphorylation site mutants led to a substantial reduction in HPV18 E6 and E7 protein expression ( Fig 6E ) . Taken together , these data demonstrate that STAT3 phosphorylation at both Y705 and S727 is essential for HPV oncogene expression in undifferentiated keratinocytes . We hypothesized that STAT3 might control HPV early gene expression by binding directly to the HPV genome . Indeed , recent chromatin immunoprecipitation sequencing ( ChIP-Seq ) data observed STAT3 binding to the long control region ( LCR ) of integrated virus genomes in the HPV18-positive HeLa cervical cancer cell line [58] . The LCR is bound by a number of host transcription factors and is essential for driving early viral transcription [59] . We used ChIP followed by qPCR to investigate whether STAT3 bound to the LCR of HPV18 episomes in primary keratinocytes [60] . In contrast to observations in the cancer cell line , we failed to detect significant STAT3 enrichment at the LCR of HPV18 in undifferentiated cells ( Fig 6F ) . Our combined data demonstrates a requirement for STAT3 phosphorylation and activation in HPV oncogene expression; likely independent of a direct interaction between STAT3 and the viral LCR . Considering that STAT3 is well known for its proliferative effects , we asked whether STAT3 contributes towards E6 and E7 gene expression by facilitating cell proliferation . We investigated the effects of cryptotanshinone on the STAT3-dependent gene product and key cell cycle regulator , cyclin D1 . Western blot analyses were performed to determine the expression of cyclin D1 in undifferentiated HPV-positive keratinocytes after 24 hours of cryptotanshinone treatment . Fig 7A showed that cryptotanshinone decreased the expression of cyclin D1 in a dose-dependent manner . Lower levels of cyclin D1 protein expression correlated with significantly ( p<0 . 01 ) lower levels of cyclin D1 mRNA in these cells as confirmed by qRT-PCR ( Fig 7B ) . In keeping with inhibitor studies , depletion of total STAT3 protein by siRNA ( Fig 7C ) or expression of phosphorylation site dominant negative forms of STAT3 ( Fig 7D ) also resulted in reduced cyclin D1 expression in HPV positive keratinocytes . STAT3 also inhibits cell cycle arrest and senescence through inhibition of the cyclin dependent kinase ( CDK ) inhibitor p21WAF1/CIP1 . Suppression of p21WAF1/CIP1 expression is facilitated indirectly by suppression of the p53 tumour suppressor protein [61] , or thought to occur directly by formation of a STAT3-cyclin D1 transcriptional repressor complex bound to the p21WAF1/CIP1 promoter [62] . Loss of STAT3 expression and activity led to an up-regulation of p21WAF1/CIP1 expression at both the protein and transcript level ( Fig 7A–7D ) . Given that cell cycle progression depends on the sequential expression of stage-specific cyclins and the activity of their corresponding CDKs we examined the cell cycle profiles of HPV positive keratinocytes in which STAT3 activity was impaired . In the presence of cryptotanshinone , more keratinocytes were observed in S phase ( p<0 . 05 ) with a concomitant decrease ( p<0 . 001 ) in the fraction of cells in G2/M ( Fig 7E ) . Of note , cryptotanshinone treatment had no impact on the cell cycle profile of normal human keratinocytes ( Fig 7F ) . To rule out any non-specific effects of inhibitor treatment we confirmed a similar cell cycle block in HPV18-containing keratinocytes when STAT3 was depleted by siRNA ( p<0 . 001 ) ( Fig 7G ) . These observations are consistent with a delay in transit of cells from S to G2 phase upon inhibition of STAT3 in HPV18-containing cells , but not in normal human keratinocytes . The importance of STAT3 for maintaining efficient cell cycle progression prompted us to assess whether STAT3 was necessary for stable HPV genome maintenance in undifferentiated cells . HPV containing keratinocytes were grown in undifferentiated monolayer culture and incubated with cryptotanshinone for 48 hours to impair STAT3 activity or transfected with a pool of STAT3-specific siRNA to deplete total protein . Total DNA was isolated from these cells and analysed by Southern blotting for levels of virus genomic DNA . There was a reduction in HPV episome levels when STAT3 activity was ablated by small molecule inhibitor ( 62% ) or total STAT3 protein removed by siRNA ( 30% ) ( Fig 7H ) . Together , these data demonstrate a role for STAT3 in maintaining cell cycle progression and stable genome maintenance in undifferentiated keratinocytes . Given that levels of STAT3 phosphorylation were maintained upon keratinocyte differentiation , we wished to assess if STAT3 was also needed for the differentiation-dependent stages of the HPV18 life cycle . HPV18-containing keratinocytes were co-cultured for 48 hours in media containing high calcium to induce differentiation and cryptotanshinone to block STAT3 activation . Whilst levels of phosphorylated STAT3 and total E6 and E7 early proteins were maintained upon differentiation in control cells , they were suppressed by cryptotanshinone ( Fig 8A; compare lanes 2 and 3 ) . By maintaining high levels of cyclins , HPV-containing cells remain active in the cell cycle upon differentiation to allow for virus genome amplification . In accordance with this , upon differentiation high levels of cyclin D1 were maintained , and expression of the CDK inhibitor p21WAF1/CIP1 was not elevated ( Fig 8A; compare lanes 1 and 2 ) . Blockade of STAT3 activation reduced cyclin D1 levels in differentiated cells ( Fig 8A; compare lanes 2 and 3 ) , and consistent with the observed down-regulation of cell cycle pathways , cryptotanshinone increased the expression of p21WAF1/CIP1 ( Fig 8A; compare lanes 2 and 3 ) . The switch between keratinocyte proliferation and differentiation is tightly controlled and STAT3 has been identified as a regulator of this process [25] . Therefore , we asked whether STAT3 contributed to the suppression of keratinocyte differentiation observed during the HPV life cycle . In untreated keratinocytes harbouring HPV18 genomes , levels of the differentiation-specific proteins involucrin and filaggrin did not increase significantly upon incubation in high calcium media ( Fig 8A; compare lanes 1 and 2 ) . In contrast , cells treated with cryptotanshinone displayed enhanced expression of these differentiation-dependent proteins ( Fig 8A; compare lanes 2 and 3 ) . These experiments indicate that STAT3 regulates the cell cycle and differentiation marker expression in differentiating HPV-positive keratinocytes . To investigate if STAT3 was also important for differentiation-dependent HPV18 genome amplification , HPV18 containing cells were treated with STAT3 inhibitors and differentiated by suspension in methylcellulose prior to harvesting the DNA for Southern blotting . Western blot analysis confirmed similar effects of STAT3 inhibition on HPV18 gene expression and keratinocyte proliferation and differentiation marker expression in the methylcellulose differentiated keratinocytes ( Fig 8B ) . Moreover , treatment with cryptotanshinone significantly reduced amplification of HPV18 genomes upon keratinocyte differentiation ( Fig 8C and 8D ) . These data indicate that STAT3 is required for differentiation-dependent stages of the HPV life cycle . Our monolayer differentiation assays suggested that the ability of HPV to promote cell cycle re-entry and delay differentiation was dependent on active STAT3 . To address this in the context of a fully stratified epithelium , NHK and HPV18-containing keratinocytes were transduced with a lentivirus expressing the dominant negative phosphorylation site Y705F STAT3 mutant or an empty vector control ( pFugW ) . The Y705F mutant was utilized as it appeared marginally more effective at impairing HPV gene expression compared to the S727A mutant ( Fig 6D and 6E ) . Forty-eight hours after transduction cells were seeded onto collagen plugs and grown as organotypic raft cultures , and sections stained with haematoxylin and eosin ( Fig 9A ) . In comparison to NHK epithelium , the presence of HPV18 genomes was associated with a thickening of the parabasal and spinous cell layers as previously shown [13] . Whilst the over-expression of a dominant negative STAT3 protein had no impact on the morphology of NHK rafts , the overall thickness of rafts produced from HPV18-containing cells expressing STAT3-Y705F was consistently reduced , and their appearance more closely resembled the morphology of the structures produced from the normal donor cells ( higher magnification image in S5A Fig ) . Amplification of the viral genome occurs in suprabasal cells that have entered a protracted G2 phase of the cell cycle , and these cells can be identified by accumulation of cytoplasmic cyclin B1 [63] . Thus , to assess cell cycle status of suprabasal cells , rafts were stained for cyclin B1 . In NHK , cyclin B1 was only observed in the basal layers of the raft and this was unaffected by transduction with the Y705F STAT3 mutant ( S5B Fig ) . In contrast , we observed a significant ( p<0 . 001 ) decrease in the population of cyclin B1 positive suprabasal cells following staining of the Y705F transduced HPV18 rafts , relative to empty vector transduced HPV18 rafts ( Fig 9B and 9E ) . These data illustrate the failure of cells lacking active STAT3 to maintain cell cycle activity in the upper stratified layers . Immunofluorescent staining of the differentiation marker involucrin was used to assess the differentiation state of the rafts . As an intermediate stage differentiation marker , involucrin is found in the suprabasal compartment of a normal epithelium ( S5C Fig ) . Empty vector transduced HPV18 rafts showed a delayed staining pattern , with expression restricted to the upper suprabasal layers of the epithelium ( Fig 9C ) . In contrast , abundant involucrin staining was noted throughout the spinuous layers of the epithelium in rafts transduced with the dominant negative Y705F phosphorylation mutant . Sections were also stained for the viral protein E1^E4 , a marker of productive infection [32 , 64] . E1^E4 staining was observed in the mid and upper suprabasal layers of HPV18 control rafts , but was significantly ( p<0 . 001 ) reduced in rafts expressing STAT3 Y705F ( Fig 9D and 9E; a wide field image available in S5D Fig ) . Taken together , these data indicate that active STAT3 is dispensable for the formation of a stratified epithelium in NHK , but is necessary for increased suprabasal cellular proliferation and cell cycle progression in cells harbouring HPV18 genomes . STAT3 function also contributes to the delay in differentiation and expression virus proteins observed in HPV-positive cells . Although STAT3 phosphorylation was increased in primary keratinocytes containing HPV18 , we queried whether aberrant STAT3 phosphorylation correlated with cervical cancer initiation and progression . Initially , we utilized the W12 in vitro model system [65] . The W12 system was developed from a polyclonal culture of cervical squamous cells naturally infected with HPV16 , derived by explant from a low-grade squamous intraepithelial lesion ( LSIL ) . At early passages these cells recapitulate an LSIL in organotypic raft cultures . However , following long-term culture these cells mirror the events associated with cervical cancer progression , with phenotypic progression to high-grade intraepithelial lesions ( HSIL ) and squamous cell carcinoma . Organotypic raft cultures were generated from NHK cells and a W12 clone representing a HSIL phenotype and stained for STAT3 S727 phosphorylation . In contrast to the NHK control , high levels of S727 phosphorylation was observed in the nuclei of cells throughout the basal and suprabasal layers of the HSIL raft ( Fig 10A ) . These data confirm that STAT3 S727 phosphorylation is greater in a high-grade lesion compared to NHK control cells . They also confirm that increased STAT3 S727 phosphorylation is observed in the context of a high-risk HPV16 infection . Next , cervical liquid based cytology samples from a cohort of HPV16 positive patients representing the progression of disease development ( CIN1-CIN3 ) and HPV negative normal cervical tissue controls were collected from the Scottish HPV archive and examined for the levels of STAT3 protein and phosphorylated STAT3 . Western blot analysis of the samples showed that in normal cervical tissue STAT3 expression and phosphorylation was low ( Fig 10B ) . However , STAT3 expression and its phosphorylation increased in the disease samples and this significantly correlated with the degree of disease severity ( Fig 10B–10E ) . Finally , we used immunofluorescence to examine phosphorylated STAT3 S727 levels in a series of human cervical sections classified as LSIL ( CIN1 ) , LSIL with foci of HSIL ( CIN1/CIN2 ) and CIN3 . An analysis of the staining pattern showed a clear difference between the three groups ( Fig 10F ) , with greater STAT3 S727 phosphorylation in the high-grade neoplasia compared with lower grade lesions . Together , these findings demonstrate that STAT3 phosphorylation is increased in HPV positive keratinocytes including those of a natural HPV infection and correlate with cervical disease progression . This study identifies the STAT3 transcription factor as a critical regulator of the HPV life cycle . Using calcium and methylcellulose-mediated differentiation in monolayer , and organotypic raft culture models of primary human keratinocytes harbouring HPV18 episomes , we discovered that STAT3 was necessary for HPV18 gene expression and viral DNA replication in undifferentiated cells and in stratified epithelium . In undifferentiated cells , a loss of STAT3 transcription factor activity—brought about through the use of small molecule inhibitors or expression of dominant negative STAT3 phosphorylation mutant proteins—led to a reduction in HPV oncogene expression . This required the transcription factor function of STAT3 , but it did not entail direct binding of STAT3 to the viral LCR . Our findings are in contrast to ChIP-Seq studies gathered from the HPV18-positive HeLa cervical cancer cells . This may in part be due to differences in STAT3 expression levels between primary keratinocytes and cancer cell lines . It is also possible that additional changes to the host cell signalling machinery or the integration of the viral genome , allow for STAT3 binding . The mechanism by which STAT3 activates HPV gene expression is therefore likely to be indirect . One possible mechanism would be to generate a cellular milieu favourable for virus transcription . In this regard , STAT3 is a driver of cell cycle progression and keratinocyte proliferation . Our results show that undifferentiated HPV-containing keratinocytes express cyclin D1 and this is largely STAT3-dependent as STAT3 inhibitor treatment , or depletion by siRNA resulted in a loss of proliferative marker expression and an increase in expression of the cell cycle arrest protein p21WAF1/CIP1 . Furthermore , these changes correlated with induction of S phase arrest . This likely results from the reduced expression of cyclin D1 , which in actively cycling cells must be increased in order for cells to transition into G2 . A similar S-phase arrest is observed in Epstein Barr virus infected lymphoblastoid cell lines ( LCLs ) , where STAT3 is required to relax the intra-S phase checkpoint [66] . Intriguingly , STAT3 inhibition had no discernable impact on the cell cycle profile of normal human keratinocytes , indicating that STAT3 is non-essential in HPV negative cells . STAT3 regulates the stable maintenance of HPV episomes in undifferentiated cells and inhibition of STAT3 transcription factor activity reduces stable episome copy numbers . This can also be explained by the change in cellular proteins brought about by blockade of STAT3 transcriptional activity , which might deprive the virus of host factors essential for genome maintenance . The loss of HPV early protein expression associated with inhibition of STAT3 activity is also likely to contribute to the deficit in virus transcription and genome maintenance . Both E6 and E7 have been shown to be critical for cell cycle progression in keratinocytes and loss of either protein impairs genome maintenance [15] . In particular , the ability of E6 to stimulate the degradation of the p53 tumour suppressor protein is important for episomal maintenance in undifferentiated keratinocytes [13 , 67] . We observed an increase in p53 expression in cells lacking E6 expression and also noted a significant up-regulation of the p53 target gene product p21WAF1/CIP1 . Whilst p21WAF1/CIP1 expression is negatively regulated by STAT3 , it is plausible that p53 contributes to the increases observed in cells lacking active STAT3 . Finally , there may be additional yet to be defined , viral or cellular proteins that are regulated by STAT3 , which are important for episomal maintenance . Since STAT3 is important for regulating stem cell like proliferative abilities along with controlling differentiation [24 , 68] , we investigated whether STAT3 played a role in these properties during the HPV life cycle . As expected , our results show that HPV18-containing keratinocytes are differentiation resistant and retain their proliferative potential . This enhanced capacity is largely dependent on STAT3 as inhibitors of transcription factor activity or dominant negative STAT3 phosphorylation mutants reduced cyclin D1 expression in monolayer culture . Consistent with this notion , loss of STAT3 activity imparted a reduction in the proliferative capacity of a HPV-containing stratified epithelium resulting in increased expression of keratinocyte differentiation markers . Rafts expressing the dominant negative STAT3 protein retained the ability to stratify but were significantly altered in the expression of cell cycle regulators and differentiation markers . In particular , suprabasal cells were no longer active in the cell cycle as judged by the reduction in cyclin B1 staining . As HPV genome amplification is reliant on a prolonged G2 phase of the cell cycle , the loss of suprabasal cyclin B1 protein expression might account for the reduction in HPV genome amplification upon differentiation in cells lacking active STAT3 [63 , 69] . Alternatively , whilst poorly characterised , both E6 and STAT3 expression have been linked to the activation of the ATM-dependent DNA damage response , a key player in HPV DNA replication [70 , 71] . It would be interesting to determine whether E6-mediated activation of STAT3 might contribute to this key process . Despite the loss of hyperplasia in the HPV18-containing rafts , the overall morphogenesis of the epithelium appeared intact and reminiscent of a normal epithelium . This suggests that whilst STAT3 phosphorylation was essential for increased proliferation , it was dispensable for normal skin development . The minimal impact on raft morphology observed in normal keratinocytes expressing the dominant negative STAT3 protein further illustrates that STAT3 appears not to be required for normal epithelial stratification . These results are in accordance with studies using gene-targeted mice containing a keratinocyte specific STAT3 knockout . These mice display no overt gross morphological differences in epithelial development but are profoundly impaired in their capacity to undergo effective wound healing in response to injury [27] . In contrast , mice engineered to express a constitutively active STAT3 in the proliferating epithelial compartment exhibit hyperplasia and a perturbed differentiation programme [28] . They are also predisposed to develop skin malignancies . Our findings clearly suggest that active STAT3 is necessary for suprabasal proliferation . Our studies show that HPV infection leads to enhanced STAT3 Y705 and S727 phosphorylation that are maintained at high levels during keratinocyte differentiation without an alteration to the total levels of STAT3 . Interestingly , we demonstrated that expression of all three HPV oncoproteins could induce STAT3 phosphorylation to a certain degree , but only E6 was sufficient to induce the dual phosphorylation and activation of STAT3 . This is consistent with previous observations showing increased STAT3 Y705 phosphorylation in HPV-positive cancer cell lines , and leading to increased levels of STAT3-dependent gene products [43 , 72] . A number of cellular pathways converge to phosphorylate and activate STAT3 [18 , 20] . Conventional dogma for STAT3 activation implicates Y705 phosphorylation as necessary and sufficient to activate STAT3 in response to a number of stimuli [18] . In addition , phosphorylation of S727 within the transactivation domain of STAT3 has also been documented to activate STAT3 signalling under some conditions . Using phosphorylation null dominant negative STAT3 proteins our results clearly demonstrate that mono-phosphorylation of either site is insufficient to activate the STAT3 dependent genes analysed in this study within keratinocytes harbouring HPV and that dual phosphorylation is essential for the virus life cycle . Moreover , our work does not support previous findings that S727 phosphorylation is detrimental to Y705 phosphorylation [73] . Rather , we suggest a model of interdependent phosphorylation between Y705 and S727 sites , in which mutation of one site does not impede the phosphorylation of the other . Our work also sheds light on the identity of the host kinases that phosphorylate these sites in HPV containing primary keratinocytes . Using validated pharmacological inhibitors we demonstrate that Y705 phosphorylation is critically dependent on JAK2 activity . Whilst JAK kinases are the best characterised mediators of Y705 phosphorylation in other cellular systems , to the best of our knowledge no study has shown that E6 up-regulates their activity . Since they function downstream of growth factor and gp130 cytokine receptor pathways , it is possible that E6 up-regulates an upstream component which subsequently activates JAK2 to phosphorylate STAT3 Y705 . In this regard , E6 activates the EGF receptor and increases IL-6 and oncostatin-M expression , which mediate their effects via gp130 receptors [74–76] . On-going experiments are testing whether these pathways contribute to STAT3 phosphorylation in primary keratinocytes . The mechanisms that mediate S727 phosphorylation are less clear due to a wide field of potential candidate kinases . As a first step we focused on MAPK proteins , given the presence of a strong consensus motif adjacent to S727 . We found that similar to what has been reported for p38 MAPK , the phosphorylation of all three canonical MAPK proteins ( ERK1/2 , p38 and JNK1/2 ) is dysregulated in HPV containing cells [77] . Our studies suggest that inhibition of all three MAPK members is required to fully impair STAT3 S727 phosphorylation . Given that E6 can activate all three kinases , our results suggest that functional redundancy exists between the different MAPK members , providing an opportunity for S727 phosphorylation throughout differentiation . Whilst E6AP-mediated p53 degradation and PDZ domain binding are well-characterised attributes of high-risk E6 proteins , neither of these functions was required for the increase in STAT3 phosphorylation . Studies are now uncovering a wealth of additional E6 binding partners and the functional consequences of these interactions [78 , 79] . Use of further E6 mutants defective for binding to additional cellular targets will allow further investigation of the molecular basis for activation of these pathways . Although the crucial role of STAT3 in both tumour cells and the tumour microenvironment is evident , gaps remain in our understanding of the regulation of STAT3 signalling in cancer . In particular , how S727 contributes to cancer development has not been fully elucidated . Given the crucial role of phosphorylation of this site for HPV containing keratinocyte proliferation , we further investigated STAT3 protein expression and phosphorylation levels in HPV associated cancers . A clear trend was evident showing an abundance of STAT3 protein and increased levels of STAT3 phosphorylation in high-grade HPV lesions compared to low-grade lesions , or control cervical samples . Moreover , staining showed obvious STAT3 S727 nuclear localization in high-grade lesions , suggesting active STAT3 . Whilst STAT3 inhibitors are not yet clinically available , it might be possible to treat HPV-associated malignancies either by targeting STAT3 directly or an upstream target such as JAK2 . Overall , these studies demonstrate that in addition to being activated in HPV-associated cancers , STAT3 is critically important for the productive HPV life cycle and a possible target for therapeutic intervention . Neonate foreskin tissues were obtained from local General Practice surgeries and foreskin keratinocytes isolated in S . Roberts’ laboratory under ethical approval no . 06/Q1702/45 . The STAT3 inhibitor S3I-201 was purchased from AdooQ BioSciences and used at a final concentration of 10 μM . This cell permeable compound binds to the STAT3 SH2 domain to prevent phosphorylation and activation . Cryptotanshinone was purchased from LKT Laboratories and used at a final concentration of 10 μM to inhibit STAT3 dimerisation and activation . The STAT5 inhibitor Pimozide was purchased from Calbiochem and used at a final concentration of 10 μM , as previously described [37] . UO126 is a selective MKK1/2 inhibitor , and is used to inhibit activation of ERK1/2 . It was added to cells at a final concentration of 20 μM and purchased from Calbiochem [80] . VX-702 was purchased from Tocris and used to inhibit p38 kinase activity . It is highly specific and used at a final concentration in cells of 10 μM [53] . The JNK1/2 inhibitor JNK-IN-8 was purchased from Cambridge BioSciences and used at a final concentration of 3 μM in cells [55] . SB-747651 was used to inhibit MSK1/2 and used at a final concentration of 5 μM in cells [81] . The JAK1/2 inhibitor Ruxolitinib , and JAK2 inhibitor Fedratinib were kindly provided by Dr Edwin Chen , University of Leeds and used at a final concentration of 10 μM [44] . All compounds were used at concentrations required to minimise potential off-target activity . Plasmids expressing HPV16 oncoproteins fused to an amino-terminal GFP protein were previously described [82 , 83] . GFP18 E5 was previously described [83] . E6 and E7 sequences were amplified from the HPV18 genome and cloned into peGFP-C1 with SalI and XmaI and SalI and BamHI restriction enzymes . Mutations were engineered into HPV18 E6 to disable interactions with E6AP , p53 and PDZ domain containing proteins . A single point mutation was introduced to create HPV18 E6 F4V and a C-terminal truncation lacking the final four amino acids created HPV18 E6 ΔPDZ . These mutants were generated using Q5 DNA polymerase and cloned into peGFP using SalI and XmaI . The L52A mutagenesis was performed by Genewiz ( New Jersey , USA ) and cloned into peGFP using EcoRI and BamHI restriction sites . The plasmids driving Firefly luciferase from the β-casein promoter and a constitutive Renilla luciferase reporter ( pRLTK ) were previously described [34] . pRc/CMV-STAT3 S727A ( 8708 ) [56] , pcFugW ( 14883 ) , pcFugW-EF . STAT3DN . Ubc . GFP ( 24984 ) [57] and pGL3-PomC ( 17553 ) [35] were purchased from Addgene ( Cambridge , MA , USA ) and we thank the principle investigators Jim Darnell , David Baltimore , Linzhao Cheng and Domencino Accili for depositing them . The transfection of primary human foreskin keratinocytes ( NHK ) isolated from neonate foreskin tissues ( ethical approval no . 06/Q1702/45 ) was performed in S . Roberts’ laboratory as described previously [13] . Briefly , plasmids containing the HPV18 genome were digested with EcoRI to release the genome , which was then re-circularised with T4 DNA ligase . The genomes were co-transfected with a plasmid encoding resistance to neomycin into low passage NHK in serum free medium . One day later , the cells were harvested and seeded onto a layer of γ-irradiated J2-3T3 fibroblasts and selected with G418 in complete E media containing foetal calf serum ( FCS , Lonza ) and epidermal growth factor ( EGF , BD BioSciences ) for 8 days . Cell colonies were pooled and expanded on γ-irradiated J2-3T3 fibroblasts . To account for donor-specific effects , NHK from two donors were used . Keratinocytes containing wild type HPV18 genomes or the mutant E6ΔPDZ genome were grown in organotypic raft cultures by seeding the keratinocytes onto collagen beds containing J2-3T3 fibroblasts [13] . Once confluent the collagen beds were transferred onto metal grids and fed from below with FCS-containing E media without EGF . The cells were allowed to stratify for 14 days before fixing with 4% formaldehyde . The rafts were paraffin-embedded and 4 μm tissue sections prepared ( Propath UK , Ltd . , Hereford , UK ) . For analysis of Phospho-STAT3 ( S727 ) ( ab32143 , abcam ) , Total STAT3 ( C-20: sc-482 , Santa Cruz Biotechnology and 9132 , CST ) , involucrin ( SY5 , Santa Cruz Biotechnology ) and HPV18 E1^E4 ( mouse monoclonal antibody 1D11 [84] ) expression , the formaldehyde-fixed raft sections were treated with the sodium citrate method of antigen retrieval . Briefly , sections were boiled in 10 mM sodium citrate with 0 . 05% Tween-20 for 10 minutes . Sections were incubated with appropriate antibodies and immune complexes visualized by using Alexa 488 and 594 secondary antibodies ( Invitrogen ) . The nuclei were counterstained with the DNA stain 4’ , 6-diamidino-2-phenylindole ( DAPI ) and mounted in Prolong Gold ( Invitrogen ) . Total genomic DNA was extracted from cell culture by phenol chloroform extraction and analysed on a 0 . 8% agarose gel and DNA transferred to GeneScreen nylon membrane . Complete HPV18 genome was released from the pGEMII backbone vector by EcoRI digestion , purified and labelled with [α-32P]-CTP . The membrane was incubated with this radiolabelled linear probe at 42°C overnight . Following washing the membrane was exposed to auto-radiograph film [13] . Untransfected NHK and HPV18 containing keratinocytes were grown in complete E media until 90% confluent . Media were changed to serum free keratinocyte media without supplements ( SFM medium , Invitrogen ) containing 1 . 8 mM calcium chloride . Cells were maintained in this media for between 48–72 hours before lysis and analysis . Alternatively , keratinocytes were resuspended in E-Media containing 1 . 5% methylcellulose and cultured for 120 hrs prior to harvesting . Transient transfections were performed with a DNA to Lipofectamine 2000 ( ThermoFischer ) ratio of 1:4 . 48 h post transfection , cells were lysed in Leeds lysis buffer for western blot [85] . Human recombinant STAT3 ( Sigma SRP2062 ) was incubated with 10 mU active JNK1 , ERK2 or p38α in 50 mM Tris-HCl pH 7 . 5 , 0 . 1 mM EGTA , 10 mM magnesium acetate , 0 . 1 mM Na3VO4 , 0 . 1% ( v/v ) 2-mercaptoethanol and 0 . 1 mM [γ32P] ATP for 30 minutes at 30°C . The reaction was terminated by the addition of 5x SDS sample buffer ( 0 . 24 M Tris-HCl pH6 . 8 , 8% ( w/v ) SDS , 40% ( v/v ) glycerol , 5% ( v/v ) 2-mercaptoethanol ) . Incorporation of phosphate was determined following SDS-PAGE and autoradiography or immunoblotting with pS727 STAT3 antibody . For autoradiography , the gel was stained with Instant Blue ( Expedeon ) for 1h at room temperature , destained in water and exposed to film at -80°C . Protein bands were excised from the gel and γ32P measured by Cerenkov counting in a liquid scintillation counter . The pSTAT3 western blot and the Coomassie stained gel were imaged using a LiCor Odyssey and quantified using Image Studio software . Total protein was resolved by SDS-PAGE ( 10–15% Tris-Glycine ) , transferred onto Hybond nitrocellulose membrane ( Amersham biosciences ) and probed with antibodies specific for Phospho-STAT3 ( S727 ) ( ab32143 , abcam ) , Phospho-STAT3 ( Y705 ) ( 9131 , Cell Signalling Technology ( CST ) ) , Total STAT3 ( C-20: sc-482 , Santa Cruz Biotechnology ) , Phospho-STAT5 ( Y694 ) ( 9314 , CST ) , Phospho-JAK2 ( Y1007/1008 ) ( 3776 , CST ) , Total JAK2 ( 3230 , CST ) , involucrin ( SY5 , Santa Cruz Biotechnology ) , HPV18 E6 ( G-7 , Santa Cruz Biotechnology ) , HPV18 E7 ( 8E2 , Abcam ( ab100953 ) , HPV 16/18 E6 ( C1P5 , Santa Cruz Biotechnology ) , HPV 16 E7 ( ED17 , Santa Cruz Biotechnology ) , Phospho-ERK1/2 ( Thr202/Tyr204 ) ( 43705 , CST ) , Phospho-JNK ( Thr183/Tyr185 ) 4668 , CST ) , Phospho-p38 ( Thr180/Tyr182 ) ( 9211 , CST ) , Bcl xL ( H-62 , Santa Cruz Biotechnology ) , Cyclin D1 ( A-12 , Santa Cruz Biotechnology ) p53 ( FL-393 , Santa Cruz Biotechnology ) , p21 ( 2947 , CST ) , FLAG ( F3165 , Sigma ) , GFP ( B-2: sc-9996 , Santa Cruz Biotechnology ) and GAPDH ( G-9 , Santa Cruz Biotechnology ) . Western blots were visualized with species-specific HRP conjugated secondary antibodies ( Sigma ) and ECL ( Thermo/Pierce ) . Keratinocytes were incubated for 48 hours with STAT3 inhibitors or transfected with STAT3 siRNA , harvested and fixed in 70% ethanol overnight . The ethanol was removed and cells washed with PBS containing 0 . 5% BSA . Cells were stained with PBS containing 0 . 5% BSA , 50 μg/ml propidium iodide ( Sigma ) and 5 μg/ml RNase ( Sigma ) and incubated in this solution for 30 minutes at room temperature . Samples were processed on an LSRFortessa cell analyzer ( BD ) and the PI histograms analyzed on modifit software . Lentivirus plasmids were transfected into HEK 293TT cells with envelope and GAG/polymerase plasmids ( kindly provided by Professor Greg Towers , University College London ) using PEI transfection reagent . After 48 hours the medium was removed from the HEK 293TT cells and added to keratinocytes for 3 hours . After this time , the complete E medium was replaced and the cells incubated for 48 hours . Total RNA was extracted from NHK using the E . Z . N . A . Total RNA Kit I ( Omega Bio- Tek ) according to the manufacture’s protocol . One μg of total RNA was DNase treated following the RQ1 RNase-Free DNase protocol ( Promega ) and then reverse transcribed with a mixture of random primers and oligo ( dT ) primers using the qScript cDNA SuperMix ( Quanta Biosciences ) according to instructions . qRT- PCR was performed using the QuantiFast SYBR Green PCR kit ( Qiagen ) . The PCR reaction was conducted on a Corbett Rotor-Gene 6000 ( Qiagen ) as follows: initial activation step for 10 min at 95°C and a three-step cycle of denaturation ( 10 sec at 95°C ) , annealing ( 15 sec at 60°C ) and extension ( 20 sec at 72°C ) which was repeated 40 times and concluded by melting curve analysis . The data obtained was analysed according to the ΔΔCt method using the Rotor-Gene 6000 software [86] . Specific primers were used for each gene analysed and are shown in S1 Table . U6 served as normaliser gene . ChIP experiments were carried out in primary foreskin keratinocytes containing episomal HPV18 genomes as previously described [60] . ChIP efficiency throughout the HPV18 genome was assessed by quantitative PCR ( qPCR ) using SensiMix SYBR master mix ( Bioline ) . Primer sequences are available on request . STAT3 association with the cyclin D1 promoter was assessed in each individual ChIP experiment using previously described primer sequences [87] . Where indicated , data was analyzed using a two-tailed , unpaired Student’s t-test .	Human papillomaviruses ( HPV ) are the leading cause of viral induced cancers worldwide . HPV are the causative agents of cervical cancers and an increasing number of head and neck cancers . HPV infections are dependant on the manipulation of the host cell for their replication and this may result in diseases such as cancer . The STAT3 transcription factor , a known driver of cancer progression , is often over active in HPV-associated cancers; however , its role in the life cycle of HPV has not been studied . Using primary cell culture models we provide the first evidence demonstrating that HPV increases both the phosphorylation and activity of STAT3 and that this is required for viral gene expression and replication . Importantly , inhibition of STAT3 by small molecule inhibitors or expression of STAT3 mutants that cannot be phosphorylated impairs the HPV life cycle . Finally , we demonstrate that STAT3 phosphorylation is increased during cervical disease progression , highlighting the potential of STAT3 as a novel therapeutic target in HPV-associated cancers .	[ "Abstract", "Introduction", "Results", "Discussion", "Methods", "and", "materials" ]	[ "phosphorylation", "urology", "keratinocytes", "medicine", "and", "health", "sciences", "gene", "regulation", "cell", "cycle", "and", "cell", "division", "cell", "processes", "cell", "differentiation", "epithelial", "cells", "developmental", "biology", "sexually", "tran...	2018	STAT3 activation by E6 is essential for the differentiation-dependent HPV18 life cycle

Subsets and Splits

No community queries yet

The top public SQL queries from the community will appear here once available.