Upload batch_5

batch_5/PMC2527498.json

@@ -0,0 +1,4 @@
+{
+    "id": "PMC2527498",
+    "text": "This is an academic paper. This paper has corpus identifier PMC2527498\nAUTHORS: Khalid A Al-Anazi, Abdulrahman Alshehri, Hazza A Al-Zahrani, Fahad I Al-Mohareb, Irfan Maghfoor, Dahish Ajarim\n\nABSTRACT:\nBackgroundVarious therapeutic options are available for the management of Langerhans cell histiocytosis. However, treatment administered to control this disease may be complicated by acute leukemia.Case presentationA 34 years old male was diagnosed to have Langerhans cell histiocytosis in March 1999. Unfortunately, the cytotoxic chemotherapy and radiotherapy given to control the repeated relapses and exacerbations of the primary disease predisposed him to therapy-induced myelodysplastic syndrome which transformed into acute myeloid leukemia. After achieving complete remission of his leukemia, the patient received an allogeneic hematopoietic stem cell transplant. The allograft was complicated by chronic graft versus host disease that was controlled by various immunosuppressive agents and extracorporal photophoresis.ConclusionManagement of complicated cases of histiocytosis requires various therapeutic modalities and a multidisciplinary approach. Having complications of therapy eg myelodysplasia or acute leukemia make the outcome more dismal and the management options limited to aggressive forms of treatment. High dose chemotherapy followed by an allograft may be a curative option not only for therapy-related myelodysplasia/acute leukemia, but also for frequently relapsing and poorly controlled Langerhans cell histiocytosis.\n\nBODY:\nBackgroundLangerhans cell histiocytosis (LCH), previously known as histiocytosis-X, is an uncommon reactive disorder of unknown pathogenesis, characterised by abnormal proliferation of Langerhan cells (LCs) into different body organs and tissues [1,2]. LCH has a wide range of clinical presentations from single system involvment eg skin or bone to multi-focal disease involving: liver, lungs, bone marrow and central nervous system [1]. Head and neck involvement is commonly encountered and presents a difficult management challenge [2].Current therapeutic options include: observations; aggressive local therapies eg surgical resection and radiotherapy; non-specific immunosuppression and cytotoxic chemotherapy [1,2]. However, management should be tailored according to the circumstances of individual patients and at times a multidisciplinary approach is necessary [2,3].Case presentationA 34 year old Saudi male was diagnosed to have LCH in Damascus, Syria in March 1999. He presented with 6 months history of an occipital mass causing bony destruction and bilateral cervical lymphadenopathy. After receiving 8 cycles of cylophosphomide, vinblastine and prednisone, he had regression of the occipital mass and disappearance of the cervical lymph nodes. Five months later, the patient presented to the medical oncologists at King Faisal Specialist Hospital and Research Centre (KFSH&RC) in Riyadh with progression of his disease in the form of extensive bony involvement. After receiving 2 courses of vinblastine and prednisone in addition to radiotherapy to skull, right parotid, right femur and pelvis, the disease became under control. In February 2002, the patient developed nodular lung lesions and cervical as well as inguinal lymphadenopathy. After confirming relapse of LCH, he received 8 more cycles of prednisone and etoposide, following which the second complete remission (CR) was achieved. On 29/7/2003; the patient had a localized relapse of his LCH as he presented with a new lesion behind the right ear which subsided after receiving 4 cycles of etoposide and prednisone. On 6/1/2004, this lesion increased in size so 2 more cycles of etoposide and prednisone were administered, following which the lesion disappeared. On 27/7/2004, the patient was found to have the third relapse as a new mass appeared in the right external auditory meatus that disappeared after receiving localized radiotherapy. On 6/9/2004, the patient presented with a localized relapse in the form of a tiny swelling involving the right frontal skull bone. An accidental blunt trauma caused rupture of the lesion which healed with scars. On 28/2/2005; the patient was admitted to the leukemia unit at KFSH&RC with low grade pyrexia and anemic symptoms for 2 weeks. Physical examination revealed: pallor, tiny inguinal lymphadenopathy and 2 small dark fleshy lesions, one on the forehead and one in the groin. The chest was clear and cardiovascular examination revealed no murmurs or added heart sounds. There was no abdominal tenderness or palpable organomegaly and neurological examination revealed no abnormality. Full blood count (FBC) showed: WBC: 16.3 × 109/L, Hb: 54 g/L and PLT: 40 × 109/L. Blood film revealed 21% blast cells and dysplastic changes. Bone marrow biopsy (BMB) showed a cellular marrow with 85% myeloblasts without any cytogenetic abnormablity. The renal and hepatic profiles were all within normal limits. After establishing the diagnoses of: therapy-related myelodysplastic syndrome (MDS) transforming into acute myeloid leukaemia (AML) and minimal residual LCH, the patient was commenced on an ICE induction course of chemotherapy composed of idarubicin, cytosine arabinoside and etoposide. Following this treatment, the patient achieved the first CR of his acute leukemia (AL). Meanwhile, an HLA identical sibling donor for allogeneic hematopoietic stem cell transplant (HSCT) was identified. On admission to the HSCT unit on 23/4/2005, the patient was asymptomatic and his physical examination revealed no new abnormality. Blood counts, renal and hepatic profiles were all within normal limits. A pre-allograft BMB showed no evidence of leukemia. The patient received a conditioning protocol composed of busulphan and cyclophosphamide. He was given fluconozole, acyclovir and bactrim as infection prophylaxis and methtrexate and cyclosporine as graft versus host disease (GVHD) prophylaxis. On 3/5/2005; the patient received his allograft without any complication. In the early post-HSCT period, the patient developed grade I mucositis treated with intravenous (IV) morphine infusion and one febrile neutropenic episode treated empirically with IV cefepime. No cytomegalovirus infection, acute GVHD, venoocclusive disease of the liver or hemorhagic cystitis were encountered. The patient engrafted his leucocytes on day +19 HSCT and his platelets on day +12 HSCT. After having a successful allograft, the patient was discharged on day +25 HSCT on cyclosporine, zantac and prophylactic antimicrobials. Thereafter, the patient had regular follow up at the HSCT out patient clinic. One year post-HSCT; he developed chronic GVHD of skin, nails, mouth, eyes and liver. Initially he was treated with prednisone 1 mg/kg/day but as his chronic GVHD became reactivated 5 months later, mycophenolate mofetil and extracorporal photophoresis were given. After achieving a good response to the measures taken, the immunosuppressive therapy was gradually tapered. Then the patient continued to have his regular follow up at the HSCT clinic and no new complication was encountered.DiscussionP. Langerhans first described LCs in the year 1868. LCs may be considered as distinct macrophages that have acquired antigenic differentiation in thymic and epidermal epithelia [4]. The lesions of LCH contain histiocytes similar to epidermal dendritic cells [5]. The identification of LC has been fascilitated by the discovery of specific markers: enzymes eg ATPase and endogenous peroxidase, birbek granules and immunological membrane markers shared with the macrophage-monocyte-histiocyte system [4]. The detection of clonal histiocytes in all forms of LCH indicates that the disease is probably a clonal neoplastic disorder with highly variable biological behaviour. Thus, the genetic mutations that promote clonal expansion of LCs or their precursors may now be identified [5]. Patients with extensive skin and/or multisystem involvement have high serum level of fms-like tyrosine kinase 3 ligand (FLT 3-L) and M-CSF. Therefore, early hematopoietic cytokines such as: FLT 3-L, stem cell factor and M-CSF may be relevant in the pathogensis of LCH and may be considered as novel therapeutic targets [1].LCH is a rare but rather severe neoplastic disorder characterized by focal or diffuse systemic proliferation of histiocytic cells at various degrees of differentiation in various body systems/organs eg lymph nodes, bone and bone marrow, liver, spleen and lungs [6]. Retrospective studies in patients with LCH have shown: frequent head, neck and bone involvement, with the skull being the most frequently involved location and with predilection for destructive bony lesions. Organ dysfunction was reported in 5–11%, diabetes incipidus in 10–22%, localized disease in 52.4% and multifocal disease in 47.6% of patients [3,6,7]. Theraputic options included: observation, curettage, chemotherapy eg steroids, etoposide and vinblastine, radiotherapy, HSCT and multimodality treatment [3,7]. LCH should be differentiated from: malignant lymphoma, monocytic leukemia, melanoma, lymphogranulomatosis, immunoblastic lymphosarcoma, sinus histiocytosis with massive lymphadenopathy and undifferentiated carciroma [6,7]. Good outcome and at least 3 year survival have been reported in up to 92% of patients. Single system involvement and absence of organ dysfunction are associated with favourable prognosis while multisystem involvement, organ dysfunction and young age are associated with dismal outcome [3,7,8].In the year 1985, the first HSCT was performed for a patient with malignant histiocytosis (MH) [9]. Thereafter, autologous and allogeneic HSCT have been used in the management of patients with MH and LCH [9-11]. Small numbers of patients were included in the reported studies. Standard and reduced intensity conditioning (RIC) protocols using stem cells from sibling and matched-unrelated donors were used. Total body irradiation (TBI), cytosine arabinoside, etoposide, melphalan, fludarabine, cyclophosphamide, anti-thymocyte globulin and total lymphoid irradiation were used in the conditioning regimens. TBI-based conditioning regimens proved to be effective as they resulted in sustained disease control while RIC-HSCT was shown to be a feasible option in patients with high risk LCH as treatment related mortality was considerably low [9-11].There is an association between LCH and both types of AL: AML and acute lymphoblastic leukemia (ALL). AL and LCH can either co-exist in the same patient or one of them may precede the other [12-14]. The frequency of LCH and a malignant neoplasm occurring in the same individual may be greater than previously recognized. Therapy administered to control LCH and genetic predisposition may be possible explanations [12].Therapy-related myelodysplastic syndrome and therapy-related acute leukemia (t-MDS/t-AL) are serious and rather frequent complications of immunosuppressive treatment, cytotoxic chemotherapy and radiotherapy [15]. T-MDS/t-AL were first recognised in the late 1970s and now they account for 10–20% of all cases of MDS and AL. They have been reported in patients with several malignant disorders treated with various cytotoxic chemotherapeutic agents including: etoposide, anthracyclins, alkylating agents, fludarabine and procarbazine [15]. In patients with t-MDS/t-AL, several chromosomal abnormalities have been described including: deletions and monosomies of chromosomes 5 and 7, 11q23, 21q 22, t (15,17), t (8,21) and inversion 16. Two distinct syndromes have been described: (1) t-MDS/t-AL induced by alkylating agents: characterized by an antecedent dysplasia and a long latency period of 5–7 years. (2) t-MDS/t-AL induced by anthracyclins and etoposide: characterized by absence of antecedent dysplasia, presentation with AML, short latency period of 1–3 years and specific chromosomal abnormalities eg 11q23 and 21q22 [15].There is no standard therapy for patients with t-MDS/t-AL. Treatment can be aggressive with curative intent, particularly for young and fit individuals. Aggressive chemotherapy protocols eg ICE and 3+7 (daunorubicin and cytosine arabinoside) have induced CRs in 20–100% of patients, but short-lived remissions, early relapses and resistance to chemotherapy were frequently encountered. In patients who are unable to withstand curative regimens, low dose chemotherapy is an alternative option and in elderly or infirm patients, supportive care is a legitimate choice [15]. HSCT offers the best chance of cure. Myeloablative HSCT has yielded long term survival in 30% of patients but transplant-related mortality has been reported to reach 49%, while the non-myeloablative HSCT has been associated with: frequent relapses, short survival and GVHD. Autologous HSCT has resulted in short survival and high rates of relapse [15].The patient presented suffered repeated progressions and frequent relapses of his LCH. The therapy administered to control/cure his primary disease predisposed him to t-MDS which subsequently transformed into AML. As the patient was relatively young and had no comorbidities, he was subjected to high dose chemotherapy to control his AML. Thereafter, he received an allogeneic HSCT in an attempt to cure his AML. The moderately severe chronic GVHD encountered was controlled by various immunosuppressive agents and extracorporal photophoresis. We believe that the successful allogeneic HSCT in addition to the graft versus leukemia effect of the chronic GVHD not only controlled his t-MDS/t-AL, but also had a long lasting effect on his frequently relapsing LCH.ConclusionAdult-onset LCH tends to progress and relapse. Various modalities of treatment can be used and management plan should take into account the circumstances of individual patients. However, immunosupressive and cytotoxic chemotherapy as well as radiotherapy given to control LCH may be complicated by t-MDS/t-AL. Definitive therapy for t-MDS/t-AL includes high dose cytotoxic chemotherapy followed by an allogeneic HSCT.AbbreviationsLangerhans cell histiocytosis; myelodysplastic syndrome; acute myeloid leukemia; hematopoietic stem cell transplant; graft versus host disease.Competing interestsThe authors declare that they have no competing interests.Authors' contributionsAll authors participated in the management of the patient presented. All authors read and approved the final form of the manuscript.ConsentA written informed consent was obtained from the patient for publication of this case report.\n\nREFERENCES:\nNo References"
+}

batch_5/PMC2527516.json

@@ -0,0 +1,4 @@
+{
+    "id": "PMC2527516",
+    "text": "This is an academic paper. This paper has corpus identifier PMC2527516\nAUTHORS: Jeff Clune, Dusan Misevic, Charles Ofria, Richard E. Lenski, Santiago F. Elena, Rafael Sanjuán\n\nABSTRACT:\nThe rate of mutation is central to evolution. Mutations are required for adaptation, yet most mutations with phenotypic effects are deleterious. As a consequence, the mutation rate that maximizes adaptation will be some intermediate value. Here, we used digital organisms to investigate the ability of natural selection to adjust and optimize mutation rates. We assessed the optimal mutation rate by empirically determining what mutation rate produced the highest rate of adaptation. Then, we allowed mutation rates to evolve, and we evaluated the proximity to the optimum. Although we chose conditions favorable for mutation rate optimization, the evolved rates were invariably far below the optimum across a wide range of experimental parameter settings. We hypothesized that the reason that mutation rates evolved to be suboptimal was the ruggedness of fitness landscapes. To test this hypothesis, we created a simplified landscape without any fitness valleys and found that, in such conditions, populations evolved near-optimal mutation rates. In contrast, when fitness valleys were added to this simple landscape, the ability of evolving populations to find the optimal mutation rate was lost. We conclude that rugged fitness landscapes can prevent the evolution of mutation rates that are optimal for long-term adaptation. This finding has important implications for applied evolutionary research in both biological and computational realms.\n\nBODY:\nIntroductionMutation is the ultimate source of genetic variation, and thus the rate at which spontaneous mutations appear is a fundamental evolutionary parameter. The mechanisms of DNA replication and repair are themselves genetically encoded and variable [1]–[5], making mutation rates potential targets of evolutionary optimization. Two opposing forces contribute to the evolution of mutation rates. On the one hand, most mutations with phenotypic effects are deleterious, producing a genetic load that favors organisms with low mutation rates; on the other hand, beneficial mutations are necessary for adaptation. Given this trade-off between genetic load and adaptation, there should exist an intermediate mutation rate—hereafter referred to as the ‘optimal’ rate, or Uopt—that balances these forces and maximizes adaptation over the long-term [6]–[9]. It is important, however, to note that these two forces operate at different timescales. The costs of genetic load are continuously paid in the short-term, whereas the payoffs of adaptation come in the long-term [6]–[8], [10]–[12].Experiments have shown that genotypes with increased mutation rates can be favored by selection if they face novel or changing environments [1], [13]–[21]. Similarly, recent work with RNA viruses has shown that certain high-fidelity genotypes have diminished fitness and virulence in mice [22],[23], which might reflect their restricted ability to create the genetic variability needed to escape from immune surveillance. However, another recent study with an RNA virus failed to observe a positive association between mutation rate and the rate of adaptation to a novel environment [24]. Despite their importance, these studies suffer from some unavoidable limitations. For example, it is unknown whether the observed mutation rates are the product of evolutionary optimization or, alternatively, if they are far from their optimal values. Also, it is often difficult to assess whether experimental observations reflect evolutionary equilibria or transient states.These limitations can be overcome using evolution with digital organisms owing to the speed and ease of data collection. Digital organisms are self-replicating computer programs that inhabit a virtual world where they reproduce, mutate, compete for resources, and evolve according to the same fundamental processes as biological organisms [25]. Here, we use digital organisms to study the ability of natural selection to adjust the mutation rate. We first validate the existence of an optimal mutation rate by extensively exploring a range of mutation rates and observing which rate maximizes adaptation over the long-term. Then we allow mutation rates to evolve under natural selection and assess whether the optimal rate is reached. Even in conditions highly favorable for mutation rate optimization, mutation rates systematically evolve that are far below the optimum, showing that natural selection fails to optimize mutation rates. We propose a novel hypothesis for these results based on the topology of the underlying fitness landscape, and we then proceed to experimentally test it.ResultsSelection Fails To Find the Optimal Mutation RateWe studied the evolution of mutation rates using the Avida digital evolution platform [25]–[34]. To test empirically whether there was an intermediate, optimal rate of mutation that maximized adaptation, we performed a series of evolution experiments. In each experiment, a genetically homogenous population was placed in a novel environment where it evolved for 150,000 updates (∼15,000 generations) at a constant mutation rate (see Methods). We explored 15 different mutation rates spanning six orders of magnitude (10−5 to 10 mutations per genome per generation). The final fitness values confirmed that there was an optimal mutation rate at an intermediate value, with Uopt≈4.641 (Figure 1). An analysis of the temporal dynamics of these experiments showed that this rate yielded the highest fitness from about generation 230 onward. Interestingly, for the very earliest time points (before generation 50), the lowest mutation rate (10−5) produced the highest fitness values, whereas for generations 50–230 a mutation rate of 2.2 gave the highest fitness values.10.1371/journal.pcbi.1000187.g001Figure 1Evolution of suboptimal mutation rates on a complex fitness landscape.Fitness is shown as a function of the genomic mutation rate. The solid line shows mean fitness of the final population, itself averaged over 50 runs, for 15 different static mutation rates (U = 10−5, 10−4 and from 10−3 to 10 at 1/3 log10 intervals). The shaded area represents±1 s.e.m. The optimal mutation rate—the rate that maximized final fitness—was Uopt≈4.641 (vertical dashed line). The two colored points show the mean fitness and mutation rate of the final population, averaged over 50 runs, in experiments where mutation rates freely evolved with starting values of either 10 (red) or 10−3 (blue) (error bars represent±1 s.e.m). Evolved mutation rates and fitness values were both orders of magnitude lower than those observed in the experiment with Uopt.To assess whether evolution would produce organisms with mutation rates near the long-term Uopt, we ran additional experiments in which mutation rates were allowed to change (see Methods), starting from rates either below (10−3) or above (10) the optimum. Strikingly, mutation rates evolved to levels far below the long-term Uopt, regardless of the starting value (Figure 1). In light of our observation that the optimum rate can change over time, one might hypothesize that the typical mutation rate of an evolving population had actually followed a near-optimal trajectory throughout its evolution, but that the final mutation rate is not a good indicator of the ability to optimize the mutation rate. However, this explanation can be ruled out because the final average fitness of the populations whose mutation rates could change was significantly lower than the fitness levels of the populations that evolved at a constant Uopt. The log-transformed final fitness values for treatments with changing mutation rates were 4.61±0.70 and 1.23±0.15 (mean±1 s.e.m.) for the populations starting at high and low initial rates, respectively. Both of these values are significantly lower than the 14.45±0.64 obtained for populations evolved at Uopt (Mann-Whitney tests, both P<0.001). The fitness advantage for Uopt is also clear for nearly all intermediate time points (Figure 2A). While populations starting below Uopt did experience a transient increase in their mutation rates (Figure 2B), the mutation rates still stayed more than two orders of magnitude below Uopt. For populations starting above Uopt, the results were particularly striking because selection pushed the populations through the optimal rate on their way to an evidently very suboptimal rate (Figures 1 and 2B).10.1371/journal.pcbi.1000187.g002Figure 2Evolutionary trajectories for fitness and mutation rate on a complex fitness landscape.(A) Evolution of average log-fitness±1 s.e.m. for treatments with the mutation rate fixed at Uopt = 4.641 (black) and for treatments with variable mutation rates starting at either 10 (red) or 10−3 (blue). (B) Evolution of average log genomic mutation rate±1 s.e.m. for treatments with variable mutation rates starting at either 10 (red) or 10−3 (blue). The black line indicates the mutation rate that had produced the highest average fitness for that time point.The finding that mutation rates evolved to be suboptimal was robust to diverse and substantial changes in the experimental conditions. First, we tested whether our results depended on the particular ancestral organism used. In the original experiments, the ancestor was a default, hand-coded organism. To assess whether this condition substantively influenced our results, we let a population founded by this organism adapt for 50,000 updates to an environment without any rewarded functions, using U = 4.641. The most abundant genotype at the end of this preliminary run was then used as the ancestor in repetitions of our original experiments. Second, we modified the complexity of the environment by varying the number of rewarded functions. Third, we tested the effect of environmental fluctuations by introducing periodic changes in the set of rewarded functions. In some of these experiments the non-rewarded functions were neutral, and in others performing these functions reduced fitness. The rate at which environmental fluctuations occurred was also varied. Fourth, we experimented with different implementations of how mutation rates could themselves change over time. In the original experiments, each organism's mutation rate had a constant probability Π of changing every generation, and the magnitude of any resulting change was controlled by a dispersion parameter σ, with Π = 0.5 and σ = 0.1. We conducted additional experiments in which we lowered Π, raised σ, or both by orders of magnitude. We also explored a configuration where increases in the mutation rate were more likely than decreases, as may happen in biological systems where it is more likely for mutations to harm than to improve an existing DNA repair pathway. Finally, we let the mutation rate apply reflexively to itself, such that high-fidelity genotypes rarely changed their mutation rates whereas low-fidelity genotypes did so frequently. In all of these additional experiments, mutation rates evolved to suboptimal levels (data not shown). We conclude, therefore, that selection fails to optimize mutation rates for long-term adaptation in a broad range of experimental conditions.Selection Favors Suboptimal Mutation Rates Because They Are Advantageous in the Short TermA possible explanation for why mutation rates evolved to be much lower than Uopt is that selection favored those genotypes that minimized the short-term fitness costs caused by deleterious mutations. This explanation is supported by the observations that, during the earliest generations of the evolution experiments, the lowest mutation rate yielded the highest fitness values. To test whether short-term selection would favor low mutation rates, we performed competition experiments between two kinds of organisms, designated A and B. These organisms were identical except for their mutation rate, which was set to Uopt for A and 0 for B; neither mutation rate was allowed to change during the competition. All competitions were conducted with the same environmental configurations as in the main experiments. In all of 50 runs, B drove A extinct in fewer than 40 generations. Competitions were also performed using U = 1.0 and U = 2.154 for B in order to address whether selection would also favor less extreme reductions in mutation rate. In both treatments, B drove A extinct in all 50 trials in fewer than 800 generations. These experiments confirm our hypothesis that natural selection was shortsighted and favored low mutation rates, even when such low rates precluded further adaptation.Whether an Optimal Mutation Rate Can Evolve Depends on the Ruggedness of the Fitness LandscapeWe conclude from the results presented thus far that the failure of the evolving populations to achieve or even maintain the mutation rates that maximize long-term adaptation reflect the conflict between the short-term cost of deleterious mutations and the long-term potential for adaptive evolution. We further hypothesize that the resolution of this tension may depend on the topology of the fitness landscape on which evolution occurs. In a rugged fitness landscape, where there are multiple peaks separated by maladaptive valleys [35],[36], populations at a local optimum must traverse regions of low fitness in the short-term in order to reach higher-fitness solutions in the long-term. This conflict leads us to hypothesize that the inability of natural selection to optimize mutation rates may depend on the ruggedness of the fitness landscape. The ideal test of this hypothesis requires comparing the evolution of mutation rates on fitness landscapes with and without fitness valleys. This test cannot be performed using the standard Avida setup, owing to the presence of extensive genetic interactions that make the fitness landscape complex and rugged [23]. We therefore modified Avida to allow simple, explicit, user-defined fitness functions that allowed us to manipulate the ruggedness of the fitness landscape (Methods, Figure 3). Adaptation occurs so fast when using these simple configurations that we also had to make the environment fluctuate between two ‘seasons’ in order to ensure a continual opportunity for beneficial mutations. These fluctuations mean that genotypes that are more fit in one season are less fit in the other (Figure 3).10.1371/journal.pcbi.1000187.g003Figure 3Evolution of mutation rates on simple fitness landscapes with different ruggedness.Here, fitness depended solely on the match between the environment and the number of a key instruction that organisms had in their genomes. In season A (left column) the key instruction was deleterious while it was beneficial in season B (center column). Rugged fitness landscapes with maladaptive valleys (rows 2–4) were introduced by setting the fitness of organisms with intermediate numbers of the key instruction to the minimum fitness level of one. The right-most column shows the results of evolution experiments under each of these selective regimes. Final fitness is shown as a function of genomic mutation rate for both static and dynamic mutation rates. The solid black line represents the average of the mean fitness across 10 runs for each of 100 different static mutation rates ranging from U = 0.01 to 1 in increments of 0.01. The two colored points represent the mean fitness and mutation rate, both averaged over 50 runs where the mutation rate freely evolved, with initial rates of U = 1 (red) or 10−5 (blue). Mutation rate and fitness values were time-averaged over the last 10 of 50 environmental changes. Owing to very similar final values, despite the very large initial differences, the individual colored points are indistinguishable in the first two rows, and error bars are not visible. The arrows indicate where mutation rates began and ended, on average, for the dynamic-rate experiments. Although the optimal mutation rate increases as a function of valley size (note the right-shift in the dashed line from top to bottom), the evolved mutation rates in fact decrease as a function of valley size (note the left-shift of the blue and red points from top to bottom).A quantitative investigation of mutation rates spanning orders of magnitude revealed, once again, that intermediate mutation rates were optimal over the long-term (Figure 3). We then allowed mutation rates to evolve starting at a genomic mutation rate either below (10−5) or above (1) the long-term optimum. Near-optimal values were efficiently selected in those landscapes without a fitness valley or with a narrow valley (Figure 3, rows 1 and 2). However, as the width of the valley grew, mutation rates evolved to be orders of magnitude lower than Uopt (Figure 3, rows 3 and 4). Fitness values were again used to judge the optimality of mutation rates. With no valleys or with narrow valleys, the average fitness in populations with variable mutation rates was slightly above that of populations with a constant rate of Uopt (Figure 3, rows 1 and 2, Mann-Whitney test, P<0.001 in both cases), which indicates a small benefit of adjusting mutation rates during evolution [37]. In stark contrast, for wider valleys, the average fitness in populations with variable mutation rates was far below that of populations with a constant rate of Uopt (Figure 3, , rows 3 and 4 Mann-Whitney test, P<0.001 in both cases), confirming that the evolved mutation rates were suboptimal on the these rugged landscapes.10.1371/journal.pcbi.1000187.g004Figure 4Evolutionarily stable mutation rate does not depend on the frequency with which the mutation rate changes (Π).The evolution of mutation rates in the explicit fitness landscape with a valley size of three is shown for several values of Π, as indicated by the colored key. Each curve shows the average of 20 runs; the adjacent bands represent±1 s.e.m. The value of Uopt was determined in previous experiments (see text). The rate of approach toward the evolutionarily stable mutation rate depends on Π, but the equilibrium value itself does not.These results show that there exists a conflict between short-term and long-term evolutionary strategies on rugged landscapes. In the short-term, low mutation rates are favored because they reduce the load of deleterious mutations, whereas in the long-term, high rates are favored because they increase the chance of producing beneficial mutants. Whether the short-term interests dominate, allowing genotypes with suboptimal mutation rates to spread, should be a function of the expected waiting time until the discovery of a beneficial mutant. To test this prediction, we competed genotypes with either optimal or suboptimal rates in the explicit fitness landscape with a valley size of three (Figure 3). In one set of experiments, we placed all organisms of both types on the low local fitness peak (asterisk in Figure 3) and let them compete for 300 generations (the duration of one season in the previous experiments). We then repeated the same experiments except that one of the individuals with the long-term optimal mutation rate started on the other side of the valley (triangle in Figure 3), such that the waiting time for the production of a beneficial mutant was eliminated. A comparison between these two sets of competition experiments shows that the probability that a genotype with a mutation rate that is below the long-term optimum can invade declines significantly when the waiting time to discover beneficial mutants is artificially eliminated (Table 1). This result illustrates why wider valleys, which create longer waiting times for beneficial mutants, cause the evolution of suboptimal mutation rates.10.1371/journal.pcbi.1000187.t001Table 1Outcomes of competitions between lineages with optimal (Uopt = 0.24) versus suboptimal (Usubopt) mutation rates in the explicit fitness landscape with a valley size of 3.With Waiting TimeWithout Waiting Time\nUsubopt\n\nUopt Fixed\nUsubopt FixedNeither Fixed\nUopt Fixed\nUsubopt FixedNeither Fixed\nP\n0238210249010.00820.06149247724208<0.00010.121851253229120<0.0001A total of 250 runs were performed for each treatment shown below. The two lineages started with equal numbers in all cases. The entries show the number of times that each lineage was fixed (i.e., reached 100% of the total population) or that neither lineage was fixed within 300 generations. With waiting time: all individuals started at the lower fitness peak (asterisk in Figure 3). Without waiting time: one individual belonging to the lineage with Uopt started on the other side of the fitness valley (triangle in Figure 3). Three different values of Usubopt were examined. P values are based on χ2 tests (with 2 degrees of freedom) that measured the effect of waiting time.The reader may also notice that the probability of invasion by the genotype with the suboptimal mutation rate was rather small in both sets of experiments (Table 1). This observation might seem, at first glance, to be at odds with the fact that mutation rates evolved over the long run to be extremely suboptimal (Figure 3, rows 3 and 4). This difference makes sense, however, for two interrelated reasons. First, each environmental change that follows the fixation of a mutation on one adaptive peak requires another waiting period for a beneficial mutation, which provides another opportunity for invasion by a genotype with a suboptimal mutation rate that reduces the mutational load. Second, any reductions in the mutation rate become self-reinforcing, as the lower mutation rates make it less likely to generate a beneficial mutant on a distant peak, which increases the expected waiting time for the generation of the next beneficial mutants, thereby increasing the opportunity for a genotype with an even lower mutation rate to invade.Finally, we examined whether the frequency with which the mutation rate changes (in essence, the mutation rate in the pathway that encodes the mutation rate), which we call Π, affects the evolutionarily stable mutation rate. Our intuition was that lower values of Π would make contests between lineages with different mutation rates less frequent, but that the long-term results of many such contests would remain the same. To test this prediction we again used the explicit landscape with a valley size of three. Even when Π varied over four orders of magnitude, it did not affect the final mutation rate that was reached (Figure 4). Hence, the inability of selection to optimize the mutation rate for long-term adaptation depends on the topology of the fitness landscape, but not on the frequency with which the mutation rate itself changes.DiscussionWe have shown that mutation rates evolve to near-optimal levels on extremely smooth fitness landscapes. However, if fitness landscapes are rugged, and the maladapted valleys between nearby fitness peaks are wide, then the scarcity of immediately accessible beneficial mutations tips the scale such that short-term selection favors mutation rates that are far below the optimum that would produce the fastest long-term adaptation. Moreover, this process is self-reinforcing because the lower the mutation rate, the less likely it becomes to produce a genotype on the other side of the fitness valley, thereby effectively widening the valley. The digital organisms in the standard Avida configuration used in our first set of experiments exhibit extensive and variable genetic interactions, making the fitness landscape rugged [23]. In those experiments, populations invariably evolved to have mutation rates that were far below the rate that would maximize their long-term fitness gains. We hypothesized that the ruggedness of the landscape was responsible for this inability to optimize their mutation rate for long-term adaptation. In order to test this hypothesis rigorously, we had to change the fitness landscape in Avida from one that is an emergent feature of complex interactions among many instructions to a much simpler surface that could be tuned to be either smooth or rugged. We found that evolving populations were indeed able to achieve mutation rates that maximized their rate of adaptation on smooth landscapes, whereas they became stuck at much lower mutation rates when the valleys between fitness peaks became too large, thus confirming our hypothesis. A growing body of experiments with viruses, bacteria, yeast, and higher eukaryotes shows that epistatic interactions are widespread and vary in their sign and intensity, implying that natural fitness landscapes are also often rugged [35],[36],[38]. Thus, our finding that rugged fitness landscapes can impede the optimization of mutation rates for long-term evolutionary adaptation is relevant to the natural world.Our experiments were performed under conditions that were favorable for the optimization of mutation rates. First, the organisms reproduced asexually. Both theoretical [12],[39],[40] and experimental work [15] has shown that asexuality facilitates the evolution of elevated mutation rates, because sexual recombination breaks up the linkage between mutator alleles that increase mutation rates and the beneficial mutations that are generated by the mutators. Second, to ensure that beneficial mutations were always available, our experiments used either an environment with more rewarded functions than the organisms ever evolved during a run (standard configuration) or a changing environment (explicit landscapes configuration). Third, population sizes were large and strong directional selection was imposed, so that drift was only a minor force in our experiments. Smaller populations might traverse maladaptive valleys more easily, owing to increased drift. However, small populations would be less likely to generate the multiple simultaneous mutations that would allow them to leap across these valleys in a single generation. In populations much larger than those we tested, the probability of an adaptive leap involving multiple simultaneous mutations would increase, but selection should be more powerful in preventing a multi-generation transition across a valley via drift. The effect of population size on the optimal mutation rate, and on the evolution of suboptimal mutation rates, thus remains an interesting area for future investigation. Nevertheless, while the optimal mutation rate and the precise width of the valley that is necessary to cause the evolution of a suboptimal rate may depend on population size, we would not expect that dependency to undermine the general conclusion of this paper, namely, that on sufficiently rugged fitness landscapes, mutation rates will evolve to be suboptimal for long-term adaptation.The inability of evolving populations to optimize their mutation rates for long-term adaptation, even with such favorable conditions, indicates that mutation rates will be suboptimal under a wide range of circumstances, at least when fitness landscapes are rugged and populations are far from a global fitness peak. While novel environments can promote increases in the mutation rate if many beneficial mutations become accessible [1], [13]–[21],[40], our work suggests that this rise will be temporary and, moreover, that even the elevated mutation rates may be suboptimal (Figure 2B). Also, given the difficulty of optimizing mutation rates that we have shown, it seems unlikely that stably high mutation rates, such as those for RNA viruses, are maintained primarily because of the rapid adaptive capacity they bestow, as has sometimes been argued [23],[41]. Alternative explanations are needed. For example, the evolution of mutation rates is also influenced by the costs of replication fidelity [8],[23], and recent work has suggested that this cost might explain the high mutation rates observed in RNA viruses [24],[42]. We expect that a cost of replication fidelity, all else being equal, will increase the evolved mutation rate. However, we would not expect the resulting increase to cause the optimization of mutation rates in general, although in a few fortuitous situations the cost of fidelity might increase the evolved mutation rate by just enough to push it near the optimal rate.Recent theoretical work by Gerrish et al. [43] has predicted that, contrary to our results, natural selection could favor a self-reinforcing increase in mutation rates in asexual populations. This process would continue even until a population suffered a mutational meltdown and went extinct, because a genotype with an increased mutation rate generates greater numbers of deleterious as well as beneficial mutations. Although not explicitly stated, the prediction of Gerrish et al. [43] of a run-away process toward higher mutation rates appears to assume a smooth fitness landscape. However, as we have shown here, the mutation rate typically evolves to a low value on a rugged fitness landscape, so that the runaway process explored by Gerrish et al. should not occur on such landscapes.Beyond their implications for understanding nature, our findings are also relevant for applied fields that use evolution to improve the performance of biological and computational systems, from molecular and microbial engineering to robotics and evolutionary computation [44],[45]. Researchers using evolution in computational fields have long sought to use natural selection to adjust mutation rates automatically and “on the fly”, in such a way that would sustain and even optimize long-term adaptation [46]–[48]. These efforts were successful on simple “toy” problems [46], but became frustrated when applied to more complex problems because self-adaptive mutation rates generally evolved to suboptimal levels [47],[48]. Our results suggest an explanation: the toy problems had smooth fitness landscapes, whereas the complex problems had rugged landscapes with wide valleys that favored evolutionary conservatism. Our findings also imply that high, fixed mutation rates will often outperform self-adaptive rates on more complex problems, although what the fixed rate should be will depend on the particular problem at hand.In summary, natural selection is not universally effective at optimizing mutation rates for long-term adaptation; in fact, it is very poor in this respect for populations that evolve on complex fitness landscapes. Also, our results caution against making generalizations based on analyses of simple fitness landscapes, whether one is studying natural systems or using evolution for engineering. As we have shown, the mere inclusion of fitness valleys—which are presumably common to the vast majority of fitness landscapes—can yield radically different conclusions from those based on smooth fitness landscapes.MethodsExperiment One: Standard ConfigurationA general description of the Avida software can be found elsewhere [25]. Here, each experiment started with 3,600 identical digital organisms. Genome length was held constant at 100 instructions, with 26 possible instructions per site [27]. Reproduction was asexual. To replicate, an organism first had to copy its genome line by line by repeatedly executing the copy instruction; it then had to execute a divide instruction, which took the offspring and used it to replace a random organism from the population.During replication, each genomic instruction could mutate to another with probability μ, the genomic mutation rate being U = 100×μ. All instructions were equally likely to result from any given mutation. The mutation rate was held constant in some experiments, while in others the rate could change by evolving over time. In treatments where the mutation rate could change, μ had a constant and high probability Π of changing by a small amount during any replication cycle. The magnitude of any resulting change was obtained by drawing log2(μ\noffspring/μ\nparent) values from a Gaussian distribution (0,σ\n2). For the experiments in which mutation rates were more likely to increase than to decrease, we drew log2(μ\noffspring/μ\nparent) from a Gaussian (bσ\n2,σ\n2), where b controls the upward bias, and tested values such that mutation rates were up to ∼1.6 times more likely to increase than decrease (though seemingly small, this bias has a large cumulative effect over many generations).Organisms died when another organism's offspring replaced them or when they executed 2,000 instructions without producing an offspring of their own. All experiments using the standard configuration lasted 150,000 updates. Updates are an arbitrary unit of time in Avida; they represent the time during which each organism, on average, executes 30 instructions [25]. In this configuration, an update corresponded to roughly 0.1 generations, although the precise generation time varied depending on the complexity of the evolved organisms' phenotypes.Each organism's phenotype depended on the complex rules that governed how its genomic program was executed, and its fitness depended on the interaction between the resulting phenotype and its environment [25]. More specifically, each organism had a metabolic rate that affected how fast it executed instructions, which, in turn, affected its reproduction rate. The ancestral rate doubled with every rewarded logic function that an organism performed. The ancestral organisms could self-replicate but not perform any other function. The ability to perform logic functions evolved by mutation and selection during each run. An organism's fitness, therefore, represents its expected growth rate relative to others in the population and depended on both its replication efficiency and its ability to perform computations. All fitness values are expressed relative to the ancestor. In reporting fitness data, relative fitness values were first averaged over all organisms in a population, then log10 transformed, and finally averaged over all replicate populations (independent trials) in an experimental treatment.To perform logic functions, organisms used inputs consisting of three randomly generated 32-bit strings, which they manipulated to produce an output. The manipulation of these numbers occurred as organisms moved them on and off stacks or between registers by executing instructions such as push, pop, add (combines the numbers in the two specified registers and places the result in a third), shift-r (bit shift right), and so on. A function was rewarded only if the input to output conversion conformed to one of the 77 canonical one-, two- or three-input logic operations. For example, the two-input EQU (‘equals’) function requires inputting two strings and outputting a third string that had a 1 for each of the 32 bits where both inputs had the same value and a 0 where they differed.Avida runs are inherently stochastic with respect to mutation and death. Therefore, we performed 50 replicate runs for each treatment. Those replicates had identical initial conditions except for a random number seed. That seed affects the outcome of all subsequent stochastic events.Experiment Two: Explicit Landscapes ConfigurationThe standard and the explicit Avida configurations differed in the instruction set, the fitness calculation and the mode of replication. We modified Avida to mimic a two-allele, 10-locus bit-string model used in a previous study [49]. Genome length was always 10, while each “instruction” was either A or B; the ancestral genome was entirely A. Fitness depended only on the number of A or B instructions in an organism's genome, according to the seasonal scheme shown in Figure 3. Every 300 generations the environment fluctuated between the two seasons, and the experiments ran for 15,000 generations. We found empirically that fluctuating the environment more or less frequently than every 300 generations produced smaller fitness differences between the optimal fixed mutation rate and suboptimal mutation rates (data not shown). That high mutation rates are most fit at an intermediate rate of environmental change has been previously shown [49].In the standard configuration, digital organisms had to copy their genomic instructions in order to replicate, and their fitness depended on their speed of replication as well as any rewards they obtained for performing computational functions. Under this alternative configuration, the organisms did not copy themselves, and only the number of A or B instructions mattered to their fitness. The rest of the setup, such as population size, was identical to the standard configuration.SoftwareAll experiments were performed with the Avida software, which can be downloaded for free at http://devolab.cse.msu.edu/software/avida. Default settings were used unless otherwise indicated.\n\nREFERENCES:\n1. ChangDKMetzgarDWillsCBolandCR\n2001\nMicrosatellites in the eukaryotic DNA mismatch repair genes as modulators of evolutionary mutation rate.\nGenome Res\n11\n1145\n1146\n11435395\n2. KunkelTABebenekK\n2000\nDNA replication fidelity.\nAnnu Rev Biochem\n69\n497\n529\n10966467\n3. MillerJH\n1996\nSpontaneous mutators in bacteria: insights into pathways of mutagenesis and repair.\nAnnu Rev Microbiol\n50\n625\n643\n8905093\n4. SchaaperRM\n1998\nAntimutator mutants in bacteriophage T4 and Escherichia coli.\nGenetics\n148\n1579\n1585\n9560377\n5. SchofieldMJHsiehP\n2003\nDNA mismatch repair: molecular mechanisms and biological function.\nAnnu Rev Microbiol\n57\n579\n608\n14527292\n6. FisherRA\n1930\nThe genetical theory of natural selection\nOxford, UK\nOxford University Press\n7. JohnsonTBartonNH\n2002\nThe effect of deleterious alleles on adaptation in asexual populations.\nGenetics\n162\n395\n411\n12242249\n8. OrrHA\n2000\nThe rate of adaptation in asexuals.\nGenetics\n155\n961\n968\n10835413\n9. SniegowskiPDGerrishPJJohnsonTShaverA\n2000\nThe evolution of mutation rates: separating causes from consequences.\nBioessays\n22\n1057\n1066\n11084621\n10. AndréJBGodelleB\n2006\nThe evolution of mutation rate in finite asexual populations.\nGenetics\n172\n611\n626\n16157667\n11. JohnsonT\n1999\nThe approach to mutation-selection balance in an infinite asexual population, and the evolution of mutation rates.\nProc Biol Sci\n266\n2389\n2397\n10643083\n12. KimuraM\n1967\nOn the evolutionary adjustment of spontaneous mutation rates.\nGenet Res\n9\n23\n34\n13. de VisserJA\n2002\nThe fate of microbial mutators.\nMicrobiology\n148\n1247\n52\n11988499\n14. GiraudAMaticITenaillonOClaraARadmanMFonsMTaddeiF\n2001\nCosts and benefits of high mutation rates: adaptive evolution of bacteria in the mouse gut.\nScience\n291\n2606\n2608\n11283373\n15. SniegowskiPDGerrishPJLenskiRE\n1997\nEvolution of high mutation rates in experimental populations of Escherichia coli.\nNature\n387\n703\n705\n9192894\n16. de VisserJAGMZeylCWGerrishPJBlanchardJLLenskiRE\n1999\nDiminishing returns from mutation supply rate in asexual populations.\nScience\n283\n404\n406\n9888858\n17. PalCMaciáMDOliverASchacharIBucklingA\n2007\nCoevolution with viruses drives the evolution of bacterial mutation rates.\nNature\n450\n1079\n1081\n18059461\n18. ChaoLCoxEC\n1983\nCompetition between high and low mutating strains of Escherichia coli.\nEvolution\n37\n125\n134\n19. DenamurEMaticI\n2006\nEvolution of mutation rates in bacteria.\nMol Microbiol\n60\n820\n827\n16677295\n20. KivisaarM\n2003\nStationary phase mutagenesis: mechanisms that accelerate adaptation of microbial populations under environmental stress.\nEnviron Microbiol\n5\n814\n827\n14510835\n21. KashiYKingDG\n2006\nSimple sequence repeats as advantageous mutators in evolution.\nTrends Genet\n22\n253\n259\n16567018\n22. PfeifferJKKirkegaardK\n2005\nIncreased fidelity reduces poliovirus fitness and virulence under selective pressure in mice.\nPLoS Pathog\n1\ne11\ndoi:10.1371/journal.ppat.0010011\n16220146\n23. VignuzziMStoneJKArnoldJJCameronCEAndinoR\n2006\nQuasispecies diversity determines pathogenesis through cooperative interactions in a viral population.\nNature\n439\n344\n348\n16327776\n24. FurióVMoyaASanjuánR\n2005\nThe cost of replication fidelity in an RNA virus.\nProc Natl Acad Sci U S A\n102\n10233\n10237\n16006529\n25. OfriaCWilkeCO\n2004\nAvida: a software platform for research in computational evolutionary biology.\nArtif Life\n10\n191\n229\n15107231\n26. ChowSSWilkeCOOfriaCLenskiREAdamiC\n2004\nAdaptive radiation from resource competition in digital organisms.\nScience\n305\n84\n86\n15232105\n27. LenskiREOfriaCCollierTCAdamiC\n1999\nGenome complexity, robustness and genetic interactions in digital organisms.\nNature\n400\n661\n664\n10458160\n28. LenskiREOfriaCPennockRTAdamiC\n2003\nThe evolutionary origin of complex features.\nNature\n423\n139\n144\n12736677\n29. MisevicDOfriaCLenskiRE\n2006\nSexual reproduction reshapes the genetic architecture of digital organisms.\nProc Biol Sci\n273\n457\n464\n16615213\n30. AdamiCOfriaCCollierTC\n2000\nEvolution of biological complexity.\nProc Natl Acad Sci U S A\n97\n4463\n4468\n10781045\n31. WilkeCOWangJLOfriaCLenskiREAdamiC\n2001\nEvolution of digital organisms at high mutation rates leads to survival of the flattest.\nNature\n412\n331\n333\n11460163\n32. OfriaCAdamiCCollierTC\n2003\nSelective pressures on genomes in molecular evolution.\nJ Theor Biol\n222\n447\n483\n12781743\n33. AdamiC\n2006\nDigital genetics: unravelling the genetic basis of evolution.\nNature Rev Genet\n7\n109\n118\n16418746\n34. GoingsSCluneJOfriaCPennockRT\n2004\nKin-selection: the rise and fall of kin-cheaters.\nPollackJBedauMHusbandsPIkegamiTWatsonR\nProceedings of Artificial Life Nine\nCambridge (Massachusetts)\nMIT Press\n303\n308\n35. WhitlockMCPhillipsPCMooreFBGTonsorSJ\n1995\nMultiple fitness peaks and epistasis.\nAnnu Rev Ecol Syst\n26\n601\n629\n36. WolfJBBrodieEDIIIWadeMJ\n2000\nEpistasis and the evolutionary process\nOxford, UK\nOxford University Press\n37. BäckTSchützM\n1996\nIntelligent mutation rate control in canonical genetic algorithm.\nRasZVMichalewiczM\nFoundations of Intelligent Systems, 9th International Symposium\nBerlin\nSpringer\n158\n160\n38. PoelwijkFJKivietDJWeinreichDMTansSJ\n2007\nEmpirical fitness landscapes reveal accessible evolutionary paths.\nNature\n445\n383\n386\n17251971\n39. DrakeJWCharlesworthBCharlesworthDCrowJF\n1998\nRates of spontaneous mutation.\nGenetics\n148\n1667\n1686\n9560386\n40. TaddeiFRadmanMMaynard-SmithJToupanceBGouyonPHGodelleB\n1997\nRole of mutator alleles in adaptive evolution.\nNature\n387\n700\n702\n9192893\n41. DomingoEHollandJJ\n1997\nRNA virus mutations and fitness for survival.\nAnnu Rev Microbiol\n51\n151\n178\n9343347\n42. FurióVMoyaASanjuánR\n2007\nThe cost of replication fidelity in human immunodeficiency virus type 1.\nProc Biol Sci\n274\n225\n230\n17148251\n43. GerrishPJColatoAPerelsonASSniegowskiPD\n2007\nComplete genetic linkage can subvert natural selection.\nProc Natl Acad Sci U S A\n104\n6266\n6271\n17405865\n44. ArnoldFH\n2001\nCombinatorial and computational challenges for biocatalyst design.\nNature\n409\n253\n257\n11196654\n45. LipsonHPollackJB\n2000\nAutomatic design and manufacture of artificial lifeforms.\nNature\n406\n974\n978\n10984047\n46. EibenAEHinterdingRMichalewiczZ\n1999\nParameter control in evolutionary algorithms.\nIEEE Trans Evol Comput\n3\n124\n141\n47. RandWRioloR\n2005\nThe problem with self-adaptive mutation rates in some environments.\nByerH-GO'ReilleyU-MArnoldDVBanzhafWBlumC\nProceedings of the 2005 Conference on Genetic and Evolutionary Computation\nWashington, DC\nAssociation for Computing Machinery\n1493\n1500\n48. CluneJGoingsSGoodmanEDPunchW\n2005\nInvestigations in Meta-GAs: panaceas or pipe dreams?\nByerH-GO'ReilleyU-MArnoldDVBanzhafWBlumC\nProceedings of the 2005 Conference on Genetic and Evolutionary Computation\nWashington, DC\nAssociation for Computing Machinery\n235\n241\n49. TravisJMTravisER\n2002\nMutator dynamics in fluctuating environments.\nProc Biol Sci\n269\n591\n597\n11916475"
+}

batch_5/PMC2527520.json

@@ -0,0 +1,4 @@
+{
+    "id": "PMC2527520",
+    "text": "This is an academic paper. This paper has corpus identifier PMC2527520\nAUTHORS: John Richards, Beth McNally, Xianfeng Fang, Michael A. Caligiuri, Pan Zheng, Yang Liu\n\nABSTRACT:\nBackgroundIt is well established that chronic tumor growth results in functional inactivation of T cells and NK cells. It is less clear, however, whether lymphopoeisis is affected by tumor growth.Principal FindingsIn our efforts of analyzing the impact of tumor growth on NK cell development, we observed a major reduction of NK cell numbers in mice bearing multiple lineages of tumor cells. The decrease in NK cell numbers was not due to increased apoptosis or decreased proliferation in the NK compartment. In addition, transgenic expression of IL-15 also failed to rescue the defective production of NK cells. Our systematic characterization of lymphopoeisis in tumor-bearing mice indicated that the number of the common lymphoid progenitor was significantly reduced in tumor-bearing mice.The number of B cells also decreased substantially in tumor bearing mice.Conclusions and SignificanceOur data reveal a novel mechanism for tumor evasion of host immunity and suggest a new interpretation for the altered myeloid and lymphoid ratio in tumor bearing hosts.\n\nBODY:\nIntroductionHematopoiesis is the process that generates leukocytes, erythrocytes and megakaryocytes. It has been divided into two branches[1], the lymphoid branch that generates B cells, T cells, NK cells and dendritic cells, and the myeloid branch that produces granulocytes, monocytes, dendritc cells, erythrocytes and megakaryocytes [2]. There appears to be two layers of regulation [3]. The first layer of control has been refered to as basal hematopoiesis. It is responsible for maintaining normal blood cell production and is regulated by cytokines produced within the microenvironment of the bone marrow. The second regulatory layer has been called amplified hematopoiesis. Amplified hematopoiesis is caused by physiological stress and appears to be regulated by the endocrine system. Cytokines or metabolites produced beyond the microenvironment of the bone marrow signal the marrow by a combination of diffusion and plasma transport. An example of amplified hematopoiesis is found in tumor bearing mice. Many transplantable tumors secrete different cytokines that act on bone marrow precursors among which are GM-CSF, IL-3, M-CSF, IL-6 and VEGF [4]. In response to these cytokines there is an expansion of myeloid cells that lead to immunosuppression [5].We have recently reported that natural killer (NK) cells are also affected by soluble factors that are associated with tumor growth [6]. In the aforementioned studies, NK cells were characterized as being CD11blo and had impaired function in vivo. The immature phenotype was associated with a deceased number of IL15Rα+NK1.1−CD3− cells in the bone marrow. Inhibition of NK cell development has been blocked by the intentional disruption of several genes including Flt-3L [7] and IL15 [8]. In Flt-3L and interleukin (IL)-15 null mice, the NK cell number is drastically reduced [7], [8]. Depleted numbers of NK cells in Flt3L-deficient mice is due to fewer common lymphoid progenitors [9]. In vitro culture systems have demonstrated that Flt3L enhances the expression of CD122 on human hematopoietic stem cells [10]. In a two-step culture system used to generate NK cells from murine bone marrow, Flt3L is used in the initial step to produce IL-15 responsive cells [11]. Although kit ligand (KL) is also used in these cultures and stimulates CD122 expression, mice deficient in KL have normal numbers of NK cells in the periphery [12]. Disruption of IL-15 [8] or components of the IL-15 receptor[13], [14] reduce NK cell numbers by preventing the induction of Bcl-2 [15], [16]. Thus IL-15 acts as a survival factor in NK cell development.In patients with CML, there is a progressive decrease in the number of NK cells [17]. We previously demonstrated that NK cell differentiation is inhibited in the bone marrow[6] and now hypothesize that NK cell numbers may be reduced in association with modifications of hematopoiesis. We found that as tumors progress NK cell numbers decrease. The reduced number of NK cells is not associated with an increase in apoptosis or a decrease in cellular proliferation, but a reduction in progenitor cell production. The decrease in the common lymphoid progenitor may be one early step of many which results in a dramatic decrease in lymphocyte production as there is a substantial decrease also found in B cell development.MethodsMice and TumorsC57Bl/6 and Balb/c mice were purchased from Charles River Laboratories under contract from the National Cancer Institute. B6 PLThy1<a>/cy and C57Bl/6-LySU-Pep3B were purchased from the Jackson Laboratory (Bar Harbor, ME). IL-15Tg mice have been described [18]. All mice were housed in the University Laboratory Animal Facility at The Ohio State University and University of Michigan under specific pathogen-free conditions. All experiments utilizing animals were approved by an institutional review board.Thymoma EL4, melanoma B16F1 and colon cancer cell line MC38 are syngeneic to C57Bl/6 mice. EL4 was grown in RPMI 1640 medium supplemented with 5% FBS, 100 U/ml penicillin, 100 µg/ml streptomycin and 4 mM L-glutamine. B16F1 and MC38 were all grown in DMEM medium supplemented with 5% FBS, 100 U/ml penicillin, 100 ug/ml streptomycin and 4 mM L-glutamine.AntibodiesThe following fluorochrome-conjugated antibodies were purchased from eBioscience (San Diego, CA): Fluorescein isothiocyanate (Fitc)-conjugated anti-Sca1 (clone D7); allophycocyanin (APC)-conjugated anti-CD3ε (clone 145-2C11); APC anti-CD117 (clone 2B8); phycoerythrin (PE) anti-CD127 (clone A7R34) and PE anti-GR1 (clone RB6-8C5). The following fluorochrome-conjugated antibodies were purchased from BD Pharmingen (San Diego, CA): Peridinin chlorophyll protein Cy5.5 (PercpCy5.5)-conjugated anti-NK1.1 (clone PK136); PercpCy5.5 anti-B220 (clone RA3-6B2); Fitc anti-CD11b (clone M1/70); PE-anti-CD45.1 (clone A20); PE anti-CD122 (clone TM-β1); Fitc anti-CD49b/Pan-NK cells (clone DX5), Fitc anti-BrdU (clone 3D4); Fitc IgG1 isotype control (clone MOPC-21); PE-Annexin V and streptavidin-PercpCy5.5. The Mouse Lineage Panel consisting of biotinylated CD3, B220, CD11b, GR-1 and Ter119 were purchased from BD Pharmingen.Establishment of Subcutaneous TumorSyngeneic mice were subcutaneously injected with EL4 (5×106 cells), B16F1 (1×104 cells), or MC38 (5×105 cells) viable tumor cells. Mice were monitored every 2–3 days to evaluate tumor growth and subcutaneous tumors were measured with a caliper along perpendicular axes of the tumor. Mice were sacrificed when tumors reached a size of 20 mm2.Cell Preparation and Flow CytometrySingle cell suspensions were prepared from bone marrow and spleen and were depleted of red blood cells. For flow cytometry, cells were initially incubated with 2.4G2 supernatant to block nonspecific antibody binding. Cells were then stained with four-color combinations of indicated fluorochrome-conjugated monoclonal antibodies. Stained cells were fixed and analyzed with a FACsCalibur (Becton Dickenson, San Jose, CA).ApoptosisSingle cell suspensions were surface stained as described above, in combination with fluorochrome-conjugated Annexin V and samples were analyzed immediately with a FACsCalibur.Bone Marrow TransplantC57Bl/6 mice were sublethally irradiated and injected with either EL4 tumors or PBS. 24 hours later bone marrow was obtained from 10 C57BL/6-LySU-Pep3B mice that had been treated 5-days earlier with 5FU. Approximately 1×106 bone marrow cells were intravenously injected into C57Bl/6j mice that were previously injected with either PBS or EL4. After 3 weeks, cells were isolated from the spleen and bone marrow of recipient mice and analyzed by flow cytometry.StatisticsUnless otherwise stated statistics were calculated by using an unpaired Student's T test (two tails). Significance was designated as a p-value≤0.05.ResultsTumor-associacted reduction of NK cell numberDecreased numbers of NK cells have been observed in chronic mylogenous leukemia[17]. In order to study the effects of tumor cells on NK cell numbers, a suitable model was required. To determine if the transplantable tumor cell line, EL4, would diminish NK cell numbers a time course was performed to evaluate tumor size with NK cell numbers. EL4 tumor growth was monitored every 2–3 days, and every six days a subset of EL4 bearing mice were sacrificed to evaluate NK cell numbers. The reduction of splenic NK cells is exemplified by the FACs profiles in Fig 1A that shows a 3-fold reduction in NK cell percentage. The absolute number of NK cells also decreased as the tumor increased in size (Fig 1B). In multiple experiments comparing the cellularity of splenocytes obtained from EL4 bearing mice and the control mice, the EL4-bearing mice showed no reduction in splenocyte cellularity. However, the number of NK cells in the spleen was reduced (Fig 1C). Given evidence that NK cell development occurs in the murine bone marrow [19], [20] we evaluated the number of NK cells in that organ. As in the spleen, the percentage of NK cells was reduced in the bone marrow of tumor-bearing mice Fig 1D. With the cellularity in the bone marrow being equal, the reduced percentage of NK cells coincided with a reduction in NK cell numbers (Fig 1E.) These data demonstrate that EL4 reduces NK cell numbers.10.1371/journal.pone.0003180.g001Figure 1EL4 tumor growth causes a decrease in splenic and bone marrow NK cells.Mice were injected subcutaneously with 5×106 EL4 cells and compared to mice that received PBS. The number of NK cells was calculated by multiplying the percentage of NK cells by the cellularity of the spleen or bone marrow. A. FACs profile of splenocytes stained with CD3 and NK1.1. NK cells were characterized as NK1.1+CD3−. The percentage of NK cells is given as the mean±SEM of five animals. B. Kinetics of tumor growth and reduction of NK cells. Spleens from EL4 bearing mice were obtained every six days and compared to control mice that were sacrificed on day 0. EL4 tumor growth was measured using calipers and the average diameter was obtained from the width and length of the tumor. C. Splenic cellularity and NK cell number from control and EL4-bearing mice. Each symbol represents the mean of a separate experiment. A paired T test was used to calculate the p-value. D. FACs profile of the bone marrow. NK cells are characterized as NK1.1+CD3−. The percentage of NK cells is given as the mean±SEM of five animals. E. Bone marrow cellularity and NK cell number obtained from control and EL4-bearing mice from several experiments. The number of NK cells was calculated by multiplying the percentage of NK cells by the number of bone marrow cells. Each symbol represents the mean of a separate experiment. A paired T test was used to calculate the p-value. Data in C–E were obtained when mice reached early removal criteria, i.e., the tumor reached 2 cm in diameters, usually at 4 weeks after tumor challenge.To determine if a reduction of NK cells occurs with other murine tumors, C57Bl/6 mice were challenged with MC38 colon cancer and B16F1 melanoma. Mice bearing MC38 tumors had a strong reduction of NK cells in the bone marrow with a less significant drop off of NK cells in the spleen (Table 1). Different from EL4 and MC38 the B16F1 tumor demonstrated a greater reduction in NK cell number in the spleen than in the bone marrow (Table 1). These findings suggest that multiple tumors lead to a reduction in NK cell numbers in both the bone marrow and spleen.10.1371/journal.pone.0003180.t001Table 1Multiple tumor cell lines reduce NK cell numbers.BreedNumber of MiceNumber of ExperimentsTumorSpleenp-valueBone Marrowp-valueC57Bl/6287Control1.63×106±1.9×105\n4.27×105±8.0×104\n28EL41.01×106±1.3×105\n0.005*\n2.68×105±7.1×104\n0.002*\n41Control1.42×106±2.2×105\n1.01×106±2.2×105\n4B16F10.54×106±1.6×105\n0.0100.59×106±1.2×105\n0.05392Control1.34×106±7.4×105\n2.70×105±1.8×104\n9MC380.74×106±0.2×105\n0.32*\n0.96×105±1.8×104\n0.001*\n*p-value was calculated using a paired t test.NK cell Reduction is independent of apoptosis, proliferation and IL15Patients with head and neck cancer, breast cancer and chronic myelogenous leukemia have demonstrated reduced numbers of NK cells due to an increase in NK cell apoptosis[21], [22]. To determine if apoptosis causes a reduction in NK cells numbers in EL4-bearing mice, Annexin V staining was performed on splenic NK cells. As shown in Fig. 2A, a reduced percentage of NK cells was observed in the spleen of tumor-bearing mice. The percentage of NK cells undergoing apoptosis in these mice, however, was not significantly different than that found in the control mice. Bone marrow and liver cells were also stained for Annexin V, but again, no difference was found in the percentage of NK cells undergoing apoptosis (data not shown).10.1371/journal.pone.0003180.g002Figure 2Apoptosis and NK cell proliferation do not account for NK cell depletion.Mice were injected subcutaneously with PBS or 5×106 EL4 cells. After 3 weeks, control and EL4-bearing mice were sacrificed and splenocytes were stained. A. The percentage of Annexin V+NK1.1+CD3− cells was used to determine the level of NK cell apoptosis. B. Three hours prior to euthanasia control and tumor-bearing mice were injected with BrdU. Splenic NK cells were then stained for intracellular BrdU. The splenic NK cell population was gated on CD122+CD3− cells. Data shown were means and SEM, involving 5 mice per group.An alternative explanation for the reduction in NK cell numbers seen in tumor-bearing mice is that NK cell proliferation is reduced. To examine the proliferation of NK cells we pulsed tumor-free and tumor-bearing mice with BrdU three hours prior to euthanasia. Splenocytes were stained for the incorporation of BrdU. As shown in Fig. 2B, the percentage of BrdU positive cells was actually higher among NK cells in the spleen of tumor-bearing mice. There no defect in proliferation can account for the decrease in NK cell numbers.Interleukin-15 knock out mice have reduced numbers of NK cells[8]. To determine if IL15 deficiency perpetuated the loss of NK cells seen in tumor-bearing mice, mice that over express IL15 were challenged with EL4 to determine if IL15 prevents a loss in NK cell number. PBS and EL4 were injected into IL-15Tg mice or littermate controls. As expected IL-15Tg mice displayed an increased percentage of NK cells compared to wild type control mice (Fig 3A). However, like wild type control mice, the presence of EL4 in IL-15Tg mice resulted in a reduction in the percentage of bone marrow NK cells (Fig 3A). The cellularity in the bone marrow of EL4-bearing, IL-15Tg mice was the same as IL-15Tg control mice and wild type mice (Fig 3B). Thus, IL-15Tg and wild type control mice that have EL4 tumors have decreased NK cell numbers (Fig 4B), and this decrease in NK cell numbers is independent of IL15.10.1371/journal.pone.0003180.g003Figure 3IL15 does not prevent loss of NK cell number.IL15Tg and littermate controls were injected with PBS or EL4. After 3 weeks, bone marrow was evaluated for NK cell number. A. FACs profile of NK1.1+CD3− cells obtained from WT or IL15Tg mice with or without tumor. B. Bone marrow cellularity and absolute number of NK cells. The absolute number of NK cells was calculated by multiplying the percentage of NK1.1+CD3− cells by the number of bone marrow cells. Data shown were means and SEM, involving 4 mice per group.10.1371/journal.pone.0003180.g004Figure 4NK cell progenitors are reduced in EL4 tumor-bearing mice.B6 mice were injected with PBS or EL4 and sacrificed after 21 days. NK cell progenitors were characterized as CD122+NK1.1−CD3−. A. FACs profile of NK cell progenitors. B. Absolute number of progenitors was calculated by multiplying the percentage of CD122+NK1.1−CD3− cells by the total number of NK cells. Data shown were means and SEM, involving 4 mice per group.Tumor-associated decrease in common lymphoid progenitorsIn vitro studies have demonstrated that IL15 does not induce the generation of NK cells directly from hematopoietic stem cells [23]. Instead a cocktail of cytokines that include Flt3L, KitL, IL7 and IL6 are required to induce IL-15 responsiveness, after which, the addition of IL-15 will result in the generation of NK cells [23]. The acquisition of CD122 is thought to commit cells to the NK cell lineage and IL-15 responsiveness [24]. To determine if there is a defect in NK cell precursors (NKP) we injected mice with EL4 and PBS. NK cell progenitors were initially characterized as CD122+NK1.1−CD3−. FACs analysis of the bone marrow showed a reduction in the NK1.1+CD3− population, which demonstrates a decrease in NK cells (Fig 4A). Furthermore, the CD122+NK1.1−CD3− population also decreased suggesting a decrease in NKP (Fig 4A). Upon calculating the absolute number of NKP, there was an approximate 2-fold reduction in tumor-bearing mice (Fig 4B). These data suggest a defect in the acquisition of NKP as a cause for reduced NK cells in EL4 bearing mice.Given the plausibility that NK cell progenitors are decreased in tumor bearing mice, there may be further hematopoietic defects. To evaluate this hypothesis, tumor-bearing mice were evaluated with a panel of antibodies to evaluate the number of NK cells, NK cell progenitors, common lymphoid progenitors (CLP), hematopoietic stem cells (HSCs) and B cells. HSCs belong to the Lin−CD127−cKit+Sca1+ population[25]. To evaluate the HSC pool, gated Lin−CD127− cells were evaluated for cKit+Sca1+ (Fig. 5A). The percentage of HSCs from both EL4-bearing mice and control mice were approximately equal, suggesting that the HSC pool was not affected by tumor growth. The CLP, characterized as Lin−CD127+cKitintSca1int\n[26], however, was reduced by about 50% (Fig 5A). The major point of reduction was associated with a decrease in CD127 expression as the corresponding cells that expressed intermediate levels of cKit and Sca1 were approximately equal (Fig 5A). To more stringently define NKP, bone marrow cells were stained for the NKP as marked by CD122+NK1.1−DX5−CD3−\n[24]. Bone marrow cells were gated on the CD122+CD3− cell population, which were greatly reduced in EL4-bearing animals (Data not shown). The gated population was then evaluated for NK1.1 and DX5 expression. Interestingly, there was a slight increase in the percentage of CD122+NK1.1−DX5−CD3− (Fig. 5B), although the absolute number was decreased. Since NK cells are a third subset of lymphocytes, we tested whether B cell number is also reduced, as these cells also develop in the bone marrow. Interestingly, we also observed a significant decrease in the percentage of B cells as characterized by the B cell marker B220 (Fig 5C). The absolute number of cells correlated well with the percent reduction is observed in Fig 5A–C. In several experiments HSCs were relatively equal, while there were decreases observed in the CLPs, NKPs and B cells (Fig 5D). These data suggest that lymphopoiesis is altered in EL4-bearing animals.10.1371/journal.pone.0003180.g005Figure 5Altered lymphopoiesis.B6 mice were injected with PBS or EL4 and sacrificed after 21 days. A. FACs profile of early hematopoiesis. HSCs are characterized as Lin−CD127−Sca1+cKit+ while the CLP is characterized as Lin-CD127+Sca1+cKit+. B. Characterization of NKPs as CD3−CD122+NK1.1−DX5−. C. Characterization of B cells by B220. D. Absolute numbers of HSCs, CLPs, NKPs, and B cells. Each line and symbol represents the mean of separate experiments involving 5 mice per group. A paired T test (one-tail) was used to determine whether tumor growth reduced the number of HSC, CLP, NKP and B cells. FACS data shown in A–C are representative of those obtained from 5 independent experiments, involving a total of 5 mice per group.To confirm that there is an alteration in lymphopoiesis in EL4-bearing animals a bone marrow transplant was performed. Mice were sub-lethally irradiated prior to tumor injection. One day later, 5FU treated bone marrow cells, from CD45.1 congenic mice were injected intravenously into control and tumor bearing mice and cellular differentiation was evaluated approximately three weeks later. In general, it was found that CD45.1 cells were expanded to a greater extent in tumor bearing mice (data not shown). However, the donor NK cell population was dramatically decreased (Fig 6B). Absolute numbers of donor NK cells (Fig. 6C) and B cells (Fig 6D) were also decreased.10.1371/journal.pone.0003180.g006Figure 6Decrease in CLPs, NKPS and Block in NKP progression.B6 mice were sublethally irradiated with 500 Rad prior to EL4 or PBS injection. Twenty-four hours later 5FU treated bone marrow from congenic CD45.1 mice was injected into control or EL4-bearing mice. Three weeks later bone marrow was harvested to evaluate NK cell progenitors. Donor NK cells were identified by an anti-CD45.1. A. Diagram of experiments. B. FACs profile of CD45.1+ NK1.1+ cells. Data shown are from gated CD45.1+CD3− cells. The numbers shown in the panels are means and SEM, involving a total of 5 mice per group. C. Absolute number of CD45.1+ NK cells. D. Absolute number of donor B cells. E. FACS profile of donor CLP, characterized as CD45.1+Lin−CD127+cKitint. Data are representative of 5 mice per group. F. Absolute number of CD45.1+ and CD45.1− CLP. Data shown are means and SEM, involving a total of 5 mice per group.Since NK and B cells are both derived from CLP, we tested the hypothesis that a reduction in this subset is an underlying cause for both defects. As shown in Fig. 6E and F, the donor CLP, as defined by CD45.1+Lin−CD127+cKitint markers, was reduced in both % and absolute numbers. Taken together, the congenic transfer of CD45.1+ cells into control and tumor bearing mice demonstrate that tumor growth leads to a block of differentiation into the lymphoid lineage.DiscussionIn a previous study we demonstrated that EL4 tumors secrete a soluble factor that modifies IL15Rα expression, which is associated with defective NK cell differentiation[6]. One aspect of NK cell development that was not thoroughly evaluated in that manuscript was the decrease in NK cells found in EL4-bearing mice. As we found that IL15Rα was decreased in the bone marrow, we hypothesized that modifications of hematopoietic processes may result in a decrease in NK cell number.Hematopoiesis is a tightly regulated process. Steady state hematopoiesis is regulated in the bone marrow by cytokines that act in a paracrine manner to maintain the production of all the different hematopoietic lineages[3]. It is only upon a stress in which cytokines released into circulation act in an endocrine manner to modify hematopoietic production[3]. Tumors are a prime example of a stress that results in the production of cytokines. Many transplantable tumors have been found to secrete cytokines that modify hematopoeisis[4]. Some of these cytokines include TGF-β, VEGF-A and GM-CSF amongst others[27], [28]. The ability for tumor growth depends on their ability to obtain nutrients from the circulation. To meet this demand many tumors secrete VEGF-A, which induces vasculariztion[29]. VEGF-A has been shown to act on bone marrow cells, which leads to an increase in GR-1/CD11b immature myeloid cells[5].In EL4-bearing mice, GR-1/CD11b myeloid cells were increased in both the bone marrow and spleen (Data not shown). The development of GR1/CD11b myeloid cells is associated with a large tumor burden and a state of immune suppression[30]. Immune suppression by GR1/CD11b is associated with a decrease in T cell function[30]. Interestingly, ablation of NK cells in BW-Sp3 tumor-bearing mice results in favorable growth conditions for GR-1/CD11b myeloid cells[31]. In Figure 1B we observed an inverse relationship between NK cell number and tumor growth.There have been several reports that have demonstrated a decrease in NK cell number in patients with tumors[17], [21], [22]. Most of these reports have shown that there was an increase in NK cell apoptosis. In evaluating NK cell apoptosis we observed no significant difference in cellular death. Furthermore, there was an increase in NK cell proliferation. Lymphopenic animals undergo homeostatic proliferation to restore T cell numbers and a similar phenomenon may be occurring within the NK cell compartment of tumor-bearing mice in an attempt to restore the number of NK cells [32]. With equal levels of cellular death and increased proliferation of NK cells in EL4 bearing-mice compared to control mice there should be a net increase in the NK cell number. However, we demonstrated that the pool of NK cells continually drops with tumor enlargement.Two independent studies have demonstrated that transgenic expression of Bcl2 maintains NK cell number in IL15−/− mice and that IL15 is important for NK cell survival[15], [16]. Early hematopoietic progenitor cells are not responsive to IL15 and upon upregulation of CD122 hematopoietic precursors become commited to the NK cell lineage[23], [24]. However, IL-15 transgene failed to rescue decreased levels of NK cells after tumor challenge. These data suggested that an earlier stage of NK cell development may be responsible for the decrease in NK cell numbers. The first candidate to consider was the NKP characterized as being CD122+NK1.1−DX5−CD3−\n[24]. We observed two main differences in NKPs. First, the overall number of NKPs was reduced in tumor-bearing mice compared to wildtype animals. The reduction was similar to that seen in total NK cell numbers. Therefore the decrease in NK cell number is likely due to reduction in the number of cells entering into the NK cell lineage. Second, when NKPs were gated on CD122+CD3− cells there was an approximate two-fold increase in the percentage of NK1.1−DX5− cells in tumor-bearing mice compared to control mice. This suggests that NKPs may also be blocked in their progression to become NK cells. To exemplify this point it was observed in congenic bone marrow transplants that there was a 2 to 3 fold accumulation of recipient and donor NKPs, in tumor-bearing mice compared to control mice.Given that NK cells are a third subset of lymphocytes, it is plausible that earlier progenitor cells are also impaired in tumor-bearing animals. Our data revealed that a reduction in NK cells in tumor-bearing animals also correlates with a decline in the number of CLP. As with NK cells and NKPs, the CLP is reduced by approximately 50% in tumor-bearing animals while the HSC population is constant. Furthermore, there was a substantial decrease in B cell development. Defects in B cell development have previously been reported in transgenic mice that overexpress arginase[33]. Increases in arginase concentration have been associated with the production GR-1/CD11b myeloid cells[4].The mechanism by which NK cells are affected by tumors remains elusive. We have evaluated antibodies to neutralize TGF-β in the EL4 model and blocking antibodies against VEGFR1 and VEGFR2 in the MC38 model (Data not shown). Neutralization of TGF-β, TNFα and blocking antibodies against VEGFR1 and VEGFR2 had no effect on NK cell number (data not shown). We also evaluated EL4 transcripts for GM-CSF, G-CSF, SCF and IL6 amongst others in an RNAse protection assay to determine if any cytokines that potentially inhibit NK cell development could be detected (data not shown). There was no indication of cytokines that modify hematopoietic development in EL4.Regardless of the mechanism, our study appears to be the first to show tumor growth leads to reduced lymphopoiesis, which in turn results in a decrease in NK and B cell numbers. Linking tumor growth to abnormal hematopoiesis in animal model may lead to new approaches aiming at boosting host immunity to cancer.\n\nREFERENCES:\n1. AkashiKTraverDMiyamotoTWeissmanIL\n2000\nA clonogenic common myeloid progenitor that gives rise to all myeloid lineages.\nNature\n404\n193\n197\n10724173\n2. HoangT\n2004\nThe origin of hematopoietic cell type diversity.\nOncogene\n23\n7188\n7198\n15378079\n3. ZhuJEmersonSG\n2002\nHematopoietic cytokines, transcription factors and lineage commitment.\nOncogene\n21\n3295\n3313\n12032771\n4. SerafiniPDe SantoCMarigoICingarliniSDolcettiL\n2004\nDerangement of immune responses by myeloid suppressor cells.\nCancer Immunol Immunother\n53\n64\n72\n14593498\n5. MelaniCChiodoniCForniGColomboMP\n2003\nMyeloid cell expansion elicited by the progression of spontaneous mammary carcinomas in c-erbB-2 transgenic BALB/c mice suppresses immune reactivity.\nBlood\n102\n2138\n2145\n12750171\n6. RichardsJXingCBlaserBWCaligiuriMAZhengP\n2006\nTumor growth impedes natural-killer cell maturation in the bone marrow.\nBlood\n7. McKennaHJStockingKLMillerREBraselKDe SmedtT\n2000\nMice lacking flt3 ligand have deficient hematopoiesis affecting hematopoietic progenitor cells, dendritic cells, and natural killer cells.\nBlood\n95\n3489\n3497\n10828034\n8. KennedyMKGlaccumMBrownSNButzEAVineyJL\n2000\nReversible defects in natural killer and memory CD8 T cell lineages in interleukin 15-deficient mice.\nJ Exp Med\n191\n771\n780\n10704459\n9. SitnickaEBryderDTheilgaard-MonchKBuza-VidasNAdolfssonJ\n2002\nKey role of flt3 ligand in regulation of the common lymphoid progenitor but not in maintenance of the hematopoietic stem cell pool.\nImmunity\n17\n463\n472\n12387740\n10. YuHFehnigerTAFuchshuberPThielKSVivierE\n1998\nFlt3 ligand promotes the generation of a distinct CD34(+) human natural killer cell progenitor that responds to interleukin-15.\nBlood\n92\n3647\n3657\n9808558\n11. WilliamsNSKlemJPuzanovIJSivakumarPVBennettM\n1999\nDifferentiation of NK1.1+, Ly49+ NK cells from flt3+ multipotent marrow progenitor cells.\nJ Immunol\n163\n2648\n2656\n10453005\n12. SeamanWETalalN\n1981\nNatural killing by spleen cells from W/Wv and Sl/Sld anemic mice.\nExp Hematol\n9\n691\n696\n7318972\n13. LodolceJPBooneDLChaiSSwainREDassopoulosT\n1998\nIL-15 receptor maintains lymphoid homeostasis by supporting lymphocyte homing and proliferation.\nImmunity\n9\n669\n676\n9846488\n14. SuzukiHDuncanGSTakimotoHMakTW\n1997\nAbnormal development of intestinal intraepithelial lymphocytes and peripheral natural killer cells in mice lacking the IL-2 receptor beta chain.\nJ Exp Med\n185\n499\n505\n9053450\n15. MinagawaMWatanabeHMiyajiCTomiyamaKShimuraH\n2002\nEnforced expression of Bcl-2 restores the number of NK cells, but does not rescue the impaired development of NKT cells or intraepithelial lymphocytes, in IL-2/IL-15 receptor beta-chain-deficient mice.\nJ Immunol\n169\n4153\n4160\n12370344\n16. CooperMABushJEFehnigerTAVanDeusenJBWaiteRE\n2002\nIn vivo evidence for a dependence on interleukin 15 for survival of natural killer cells.\nBlood\n100\n3633\n3638\n12393617\n17. PiersonBAMillerJS\n1996\nCD56+bright and CD56+dim natural killer cells in patients with chronic myelogenous leukemia progressively decrease in number, respond less to stimuli that recruit clonogenic natural killer cells, and exhibit decreased proliferation on a per cell basis.\nBlood\n88\n2279\n2287\n8822949\n18. FehnigerTASuzukiKPonnappanAVanDeusenJBCooperMA\n2001\nFatal leukemia in interleukin 15 transgenic mice follows early expansions in natural killer and memory phenotype CD8+ T cells.\nJ Exp Med\n193\n219\n231\n11208862\n19. HackettJJrBosmaGCBosmaMJBennettMKumarV\n1986\nTransplantable progenitors of natural killer cells are distinct from those of T and B lymphocytes.\nProc Natl Acad Sci U S A\n83\n3427\n3431\n2871552\n20. HallerOKiesslingROrnAWigzellH\n1977\nGeneration of natural killer cells: an autonomous function of the bone marrow.\nJ Exp Med\n145\n1411\n1416\n859002\n21. MellqvistUHHanssonMBruneMDahlgrenCHermodssonS\n2000\nNatural killer cell dysfunction and apoptosis induced by chronic myelogenous leukemia cells: role of reactive oxygen species and regulation by histamine.\nBlood\n96\n1961\n1968\n10961901\n22. BauernhoferTKussIHendersonBBaumASWhitesideTL\n2003\nPreferential apoptosis of CD56dim natural killer cell subset in patients with cancer.\nEur J Immunol\n33\n119\n124\n12594840\n23. WilliamsNSMooreTASchatzleJDPuzanovIJSivakumarPV\n1997\nGeneration of lytic natural killer 1.1+, Ly-49- cells from multipotential murine bone marrow progenitors in a stroma-free culture: definition of cytokine requirements and developmental intermediates.\nJ Exp Med\n186\n1609\n1614\n9348320\n24. RosmarakiEEDouagiIRothCColucciFCumanoA\n2001\nIdentification of committed NK cell progenitors in adult murine bone marrow.\nEur J Immunol\n31\n1900\n1909\n11433387\n25. SpangrudeGJHeimfeldSWeissmanIL\n1988\nPurification and characterization of mouse hematopoietic stem cells.\nScience\n241\n58\n62\n2898810\n26. KondoMWeissmanILAkashiK\n1997\nIdentification of clonogenic common lymphoid progenitors in mouse bone marrow.\nCell\n91\n661\n672\n9393859\n27. KusmartsevSGabrilovichDI\n2002\nImmature myeloid cells and cancer-associated immune suppression.\nCancer Immunol Immunother\n51\n293\n298\n12111117\n28. BronteVSerafiniPApolloniEZanovelloP\n2001\nTumor-induced immune dysfunctions caused by myeloid suppressor cells.\nJ Immunother\n24\n431\n446\n11759067\n29. ByrneAMBouchier-HayesDJHarmeyJH\n2005\nAngiogenic and cell survival functions of vascular endothelial growth factor (VEGF).\nJ Cell Mol Med\n9\n777\n794\n16364190\n30. KusmartsevSGabrilovichDI\n2006\nRole of immature myeloid cells in mechanisms of immune evasion in cancer.\nCancer Immunol Immunother\n55\n237\n245\n16047143\n31. GeldhofABvan GinderachterJALiuYNoelWde BaetselierP\n2003\nAblation of NK cell function during tumor growth favors Type 2-associated macrophages, leading to suppressed CTL generation.\nClin Dev Immunol\n10\n71\n81\n14768937\n32. PrlicMBlazarBRFarrarMAJamesonSC\n2003\nIn vivo survival and homeostatic proliferation of natural killer cells.\nJ Exp Med\n197\n967\n976\n12695488\n33. de JongeWJKwikkersKLte VeldeAAvan DeventerSJNolteMA\n2002\nArginine deficiency affects early B cell maturation and lymphoid organ development in transgenic mice.\nJ Clin Invest\n110\n1539\n1548\n12438451"
+}

batch_5/PMC2527530.json

@@ -0,0 +1,4 @@
+{
+    "id": "PMC2527530",
+    "text": "This is an academic paper. This paper has corpus identifier PMC2527530\nAUTHORS: Binh An Diep, Amy M. Palazzolo-Ballance, Pierre Tattevin, Li Basuino, Kevin R. Braughton, Adeline R. Whitney, Liang Chen, Barry N. Kreiswirth, Michael Otto, Frank R. DeLeo, Henry F. Chambers\n\nABSTRACT:\nCommunity-associated methicillin-resistant Staphylococcus aureus (CA-MRSA) strains typically carry genes encoding Panton-Valentine leukocidin (PVL). We used wild-type parental and isogenic PVL-deletion (Δpvl) strains of USA300 (LAC and SF8300) and USA400 (MW2) to test whether PVL alters global gene regulatory networks and contributes to pathogenesis of bacteremia, a hallmark feature of invasive staphylococcal disease. Microarray and proteomic analyses revealed that PVL does not alter gene or protein expression, thereby demonstrating that any contribution of PVL to CA-MRSA pathogenesis is not mediated through interference of global gene regulatory networks. Inasmuch as a direct role for PVL in CA-MRSA pathogenesis remains to be determined, we developed a rabbit bacteremia model of CA-MRSA infection to evaluate the effects of PVL. Following experimental infection of rabbits, an animal species whose granulocytes are more sensitive to the effects of PVL compared with the mouse, we found a contribution of PVL to pathogenesis over the time course of bacteremia. At 24 and 48 hours post infection, PVL appears to play a modest, but measurable role in pathogenesis during the early stages of bacteremic seeding of the kidney, the target organ from which bacteria were not cleared. However, the early survival advantage of this USA300 strain conferred by PVL was lost by 72 hours post infection. These data are consistent with the clinical presentation of rapid-onset, fulminant infection that has been associated with PVL-positive CA-MRSA strains. Taken together, our data indicate a modest and transient positive effect of PVL in the acute phase of bacteremia, thereby providing evidence that PVL contributes to CA-MRSA pathogenesis.\n\nBODY:\nIntroductionThe worldwide emergence of community-acquired methicillin resistant Staphylococcus aureus (CA-MRSA) strains has been linked to carriage of genes encoding Panton-Valentine leukocidin (PVL), a two-component leukolytic toxin [1]–[9]. The contribution of PVL to CA-MRSA pathogenesis remains controversial. No difference in virulence was detected when comparing two prevalent CA-MRSA strains, LAC (USA300 lineage) and MW2 (USA400 lineage), to their respective isogenic PVL knockout mutants in several mouse models, including subcutaneous abscess, sepsis and pneumonia models [10]–[13]. Data supporting a role for PVL in pathogenesis are derived from experiments in a mouse pneumonia model using laboratory strains of the NCTC8325 lineage lysogenized with a PVL-encoding bacteriophage [14]. Presence of PVL was associated with up-regulation of staphylococcal protein A (Spa) and other surface proteins and led the investigators to propose a model in which PVL interference with global regulatory networks culminated in overwhelming inflammation and necrosis of the murine lung [14]. Such profound effects on global gene expression raise the possibility that the experimental outcomes were due not to PVL, but were consequence of major genetic perturbations [9], perhaps due to pleiotropic mutations that occur with relatively high frequency in laboratory strains [15], [16].The conflicting data from mouse infection models and the relative insensitivity of murine polymorphonuclear leukocytes (PMNs or granulocytes) to the leukolytic effect of PVL compared with human cells prompted us to assess the role of PVL in CA-MRSA pathogenesis in a rabbit model. Importantly, the sensitivity of rabbit PMNs to the leukolytic activity of PVL mirrors that of human PMNs [17], making the rabbit an excellent model species because PMNs are a primary cellular target of PVL and the principal component of host innate immune defense. Intravenous injection of purified PVL into rabbits results in transient granulocytopenia followed by marked granulocytosis, but is not lethal [18]. Here, we tested in prevalent CA-MRSA strains whether PVL regulates global gene networks and evaluated its contribution to pathogenesis of bacteremia, a hallmark feature of invasive S. aureus disease and the most prevalent clinical syndrome of invasive CA-MRSA disease in particular [19].ResultsPVL does not impact global gene regulationIntroduction of PVL into a laboratory strain of S. aureus was reported to alter global gene regulation, resulting in increased expression of surface adhesins such as staphylococcal protein A (Spa) [14]. To assess the potential regulatory effects of PVL in clinically relevant CA-MRSA strains, we performed experiments using contemporary CA-MRSA belonging to the two prevalent lineages, USA300 and USA400 [3]–[6], [20].To examine the effect of PVL on global gene expression, we conducted transcriptional profiling of wild-type parental and isogenic Δpvl mutant USA300 (SF8300–SF8300Δpvl) and USA400 (MW2–MW2Δpvl) strain pairs. Total RNA was isolated after in vitro culture to exponential-or stationary phase of growth under conditions known to induce over-expression of PVL. Contrary to results reported by Labandeira-Rey et al. [14], only 1 of 3961 S. aureus microarray probesets met standard criteria for differentially-expressed genes (>2-fold change in transcript levels, a ratio of wild-type versus Δpvl mutant strain and P<0.05 [21], [22]) using the SF8300–SF8300Δpvl strain pair comparison. The single differentially-expressed gene, spectinomycin adenylyl-transferase (spc), was used for allelic replacement of pvl and thus present in the SF8300Δpvl mutant strain but not in the wild-type parental strain (Table 1). Only 5 of 2632 MW2 probe sets were differentially expressed in the MW2–MW2Δpvl strain pair comparison, and of these, 2 were specific for lukS-PV and lukF-PV found only in the MW2 parent strain (Table 1). Using TaqMan real-time reverse transcriptase-PCR, we confirmed that expression of PVL does not alter transcripts encoding accessory gene regulator (AgrA) and Agr-regulated virulence factors such as protein A (Spa), α-toxin (Hla), β-hemolysin (HlgABC), serine aspartate repeat protein (SdrD), serine protease (SplA), or clumping factor B (ClfB) (Figure 1A\n and Table 2).10.1371/journal.pone.0003198.g001Figure 1PVL does not alter global gene and protein expression profiles.Clinical strains of USA300 (LAC and SF8300) and USA400 (MW2) and their respective isogenic Δpvl mutant strains were cultured to mid-exponential or stationary phases of growth in TSB or CCY media. (A) TaqMan real-time RT-PCR for comparison of fold changes in transcript levels of selected Agr-regulated genes in wild type and Δpvl mutant strains. See also Table 2 for additional data derived from in vitro growth to exponential phase and stationary phase in CCY or TSB media. agrA, accessory global regulator; hla, alpha-toxin; hlgA, gamma-haemolysin component A; splA, serine protease; spa, protein A; sdrD, serine aspartate repeat protein; clfB, clumping factor B. (B) Cell extracts separated by 12% SDS-PAGE (Protean II gel, Bio-Rad) using cultures grown to stationary phase. (C) Culture supernatants, prepared from growth in TSB or CCY media, were separated by gradient 10-20% SDS-PAGE. PVL subunits were identified by automated-direct infusion tandem mass spectrometry [32]. (D, E, F) Western immunoblot analysis of supernatants and cell extracts from cultures grown to stationary or mid-exponential phase. Proteins were detected with rabbit polyclonal antibodies specific for LukF-PV, Hla (α-toxin), or Spa. The immunoblots in panel E were exposed on the same film for equal times or using a longer exposure for MW2 (inset), which produced less Spa. Protein samples for SDS-PAGE presented in panels B and C were prepared in a manner identical to those shown in panels E and F.10.1371/journal.pone.0003198.t001Table 1PVL does not alter global transcriptional profiles of USA300 and USA400.*StrainsMediaGrowth PhaseDifferentially Expressed Gene€\nNo. of GenesGene Identification¥\nSF8300 vs. SF8300Δpvl\nTSBExponential1\nspc\nSF8300 vs. SF8300Δpvl\nTSBStationary1\nspc\nSF8300 vs. SF8300Δpvl\nCCYExponential0–SF8300 vs. SF8300Δpvl\nCCYStationary1\nspc\nMW2 vs. MW2Δpvl\nTSBExponential0–MW2 vs. MW2Δpvl\nTSBStationary2\nlukF-PV and lukS-PV\nMW2 vs. MW2Δpvl\nCCYExponential0–MW2 vs. MW2Δpvl\nCCYStationary5\nlukF-PV, lukS-PV, set, lukE, plc\n€Transcriptome analyses of SF8300 vs. SF8300Δpvl mutant strains were performed using custom Affymetrix GeneChips (RMLChip1) containing 3961 probe sets from eight different S. aureus strains (COL, EMRSA16, MSSA476, RF122, TSS, 8325, Mu50, and N315); and MW2 vs. MW2Δpvl using custom Affymetrix GeneChips (RMLChip3) with 99.3% coverage of genes from MW2 (2613 probe sets of 2632 ORFs; the remaining 0.7% are represented by identical probe sets from other staphylococci). Note that RMLChip1 does not contain probesets for lukS-PV and lukF-PV, which was subsequently assayed by TaqMan real-time RT-PCR (Table 2). Table displays only probesets that met standard criteria required for differentially-expressed genes (>2-fold change in transcript levels in the wild-type vs. Δpvl strain, and P<0.05 using a unpaired Student's t test).¥\nspc, encoding spectinomycin adenylyltransferase (detected only in SF8300Δpvl by RMLChip1; probe set absent in RMLChip3); lukF-PV and lukS-PV, encoding PVL (detected only in MW2 by RMLChip3; probe sets absent from RMLChip1); set, encoding an enterotoxin homolog (MW0052), 2.4 fold-change; lukE (MW1768), encoding leukocidin E, 2.1 fold-change; and plc (MW0070), encoding 1-phosphatidyl-inositol phosphodiesterase precursor, 2.6 fold-change.10.1371/journal.pone.0003198.t002Table 2TaqMan real-time RT-PCR analysis reveals that PVL does not alter agr-regulated transcripts in USA300 and USA400 clinical strains.±\nStrainGrowth Phase\nagrA\n\nspa\n\nclfB\n\nsdrD\n\nhla\n\nhlgA\n\nsplA\n\nlukF-PV\nmean fold-change in gene transcripts for wt vs. Δpvl\nFor cells grown in TSBLAC vs. LACΔpvl\nExponential−1.25���1.391.15−1.36−1.43−1.25−1.1020769.59*SF8300 vs. SF8300Δpvl\nExponential−1.31−1.20−1.31−1.49−1.32−1.39−1.2848376.47*MW2 vs. MW2Δpvl\nExponential−1.00−1.241.40−1.18*−1.12−1.46−1.119092.34*LAC vs. LACΔpvl\nStationary1.421.361.47*1.27*1.51−1.06−1.0960925.20*SF8300 vs. SF8300Δpvl\nStationary1.161.401.35−1.041.581.221.82126494.09*MW2 vs. MW2Δpvl\nStationary1.371.521.631.391.441.951.6716747.59*For cells grown in CCYLAC vs. LACΔpvl\nExponential−1.23−1.74−1.57−1.621.13−1.64−1.432133.83*SF8300 vs. SF8300Δpvl\nExponential−1.101.27a\n−1.15−1.13−1.20−1.24*1.054008.34*MW2 vs. MW2Δpvl\nExponential−1.022.831.571.20−1.76−1.09−1.397972.35*LAC vs. LACΔpvl\nStationary1.221.131.12−1.00−1.12−1.17−1.35392409.30*SF8300 vs. SF8300Δpvl\nStationary−1.091.381.411.03−1.231.111.35781001.79*MW2 vs. MW2Δpvl\nStationary1.081.761.361.201.571.461.59146821.31*±TaqMan real-time RT-PCR was performed as described in Methods. Results are expressed as the mean fold-change of 3–5 experiments (exponential growth, Exp.) or 4–7 experiments (stationary phase of growth, Stat.) with one exceptiona (one of the TaqMan reactions failed, n = 2). The relative expression level of each transcript (dCT) was compared in parent vs. Δpvl strains using a paired Student's t-test (*\nP<0.05 versus Δpvl). Except for lukF-PV, none of the transcripts in any of the strains met standard criteria required for differentially expressed genes (>2-fold change in the wild-type vs. Δpvl mutant strain and P<0.05).\nagrA, accessory global regulator; hla, alpha-toxin; hlgA, gamma-haemolysin component A; splA, serine protease; spa, protein A; sdrD, serine aspartate repeat protein; clfB, clumping factor B; and lukF-PV, Panton-Valentine leukocidin component F.Consistent with the microarray and TaqMan data, protein profiles of cell extracts and culture supernatants were virtually identical between each of the wild-type parental and Δpvl isogenic mutant strains cultured with the same in vitro growth conditions (figure 1B and 1C\n). Although there was over-expression of PVL in supernatants of LAC, SF8300 and MW2 wild-type strains (figure 1C\n), the toxin did not modulate production of other virulence factors, including α-toxin and protein A (figure 1D–F\n). Protein A is a well-characterized proinflammatory factor mediating the development of disease in the lung [23]. However, we found that PVL does not modulate expression of protein A in CA-MRSA clinical strains, in marked contrast to what was reported by Labandeira-Rey et al. using laboratory strains of S. aureus\n[14]. Our results formally rule out the possibility that PVL contributes to CA-MRSA virulence by altering global gene and/or protein regulatory networks of S. aureus.Co-infection experiments in a rabbit model of bacteremiaAs PVL has no effect on global gene and protein expression in prevalent CA-MRSA strains, a potential direct effect of PVL to CA-MRSA pathogenesis was examined. We used a rabbit model of bacteremia to compare wild-type parental and isogenic Δpvl mutant strain pairs using a competition design. Rabbits were co-infected with a mixture of parent and Δpvl mutant strains in an approximate 1∶1 ratio. The Δpvl mutant strains contained a spectinomycin resistance cassette (spc) used to replace pvl genes, which allowed for the enumeration of the Δpvl mutant (Spc-resistant) and parent strain (Spc-sensitive). Normalized ratios of the parent to Δpvl mutant, representing competition indexes, were determined in organs harvested from rabbits that succumbed to infection or moribund rabbits with end stage bacteremia, which were euthanized between 2 and 7 days post infection (Table 3). The competition indexes in the rabbit lung, spleen, kidney and blood did not differ significantly from the null effect value of 0 for the SF8300–SF8300Δpvl (n = 17), LAC–LACΔpvl (n = 28) and MW2–MW2Δpvl (n = 25) isogenic strain pairs, indicating no contribution of PVL to bacterial colonization and persistence at the end stages of bacteremia in the competition model.10.1371/journal.pone.0003198.t003Table 3Co-infection experiments with USA300 or USA400 parental or isogenic Δpvl mutant strains assayed at end stages of bacteremia.USA300USA400SF8300 vs. SF8300Δpvl\nLAC vs. LACΔpvl\nMW2 vs. MW2Δpvl\nno. of rabbits172825mean (±sd) inoculum (log10CFU)7.42±0.497.80±0.317.45±0.25mean (±sd) inoculum wt:Δpvl ratio1\n0.81±0.051.09±0.270.92±0.05mean (±sd) survival in days3.9±1.33.9±1.23.2±1.3mean (±sd) bacterial densitylung, log10(CFU/g)2.77±1.393.31±1.463.05±1.84spleen, log10(CFU/g)2.64±1.512.97±1.352.66±1.25kidney, log10(CFU/g)3.70±1.834.95±1.694.41±1.97blood, log10(CFU/ml)1.55±1.291.14±1.151.33±1.00competition index (95% confidence interval) 2\nLung−0.35 (−1.12–0.41)0.30 (−0.14–0.74)0.07 (−0.16–0.30)Spleen−0.03 (−0.75–0.69)0.18 (−0.18–0.54)0.12 (−0.03–0.27)Kidney−0.20 (−1.10–0.70)0.66 (−0.01–1.33)0.43 (−0.24–1.09)Blood−0.35 (−1.02–0.32)−0.13 (−0.47–0.21)0.21 (−0.16–0.58)1Competition assays were used to compare three wild type-Δpvl mutant pairs: LAC-LACΔpvl (n = 28), SF8300-SF8300Δpvl (n = 17), and MW2-MW2Δpvl (n = 25), where n is the total number of animals used in each experiment. A 1∶1 mixture containing approximately 3×107 CFUs of wild type parent and 3×107 CFUs of isogenic Δpvl mutant were used to co-infect New Zealand white rabbits via the marginal ear vein. Mean bacterial densities comprising of both wild type and Δpvl mutant from vital organs and blood are shown. The competition index (CI), which is the logarithm (log10) of the output ratios of parent and isogenic mutant after correction for variations in input ratios, are shown. A positive CI value indicates enhanced tissue infectivity of the parent, whereas a negative CI value indicates enhanced tissue infectivity of the mutant; a CI = 0 is the no-effect value.2The null hypothesis (CI = 0) that there was no difference in bacterial densities between parent and isogenic Δpvl mutant in rabbit vital organs was tested using a paired Student's t test. All two-tailed P values were not statistically significant (P>0.05).Contribution of PVL to pathogenesis in a single-strain rabbit bacteremia modelAlthough PVL did not impact CA-MRSA pathogenesis in a competition bacteremia model, it is possible that secreted toxin from the parent produced a bystander effect that protected mutant cells. We assessed the potential contribution of PVL over the time course of bacteremia in the rabbit model in which either the parental strain SF8300 or the isogenic mutant SF8300Δpvl were used to inoculate individual rabbits. Bacterial densities in vital organs determined at 24, 48 and 72 hours post infection (Table 4). In this model, bacterial densities decreased over time in the lung and spleen (linear test for trend, P<0.05), but increased in the kidney (linear test for trend, P<0.05), indicating that the kidney is a target organ that supports bacterial growth. At 24 and 48 hours post infection, significantly more SF8300 than SF8300Δpvl were recovered from the kidney, but not from lung or spleen (P<0.05). In contrast, at 72 hours post infection, there was no significant difference in bacterial densities between SF8300 and SF8300Δpvl isolated from kidney (Table 4). This result is in part explained by the rapid growth of 2.5 logCFU of SF8300Δpvl in the kidney between 48 and 72 hours post infection (P<0.001), whereas this did not occur for the SF8300 parental strain (P = 0.63). The lack of difference in bacterial densities at 72 h post infection correlated with the end stages of disease in the bacteremia model, as rabbits infected with either SF8300 or SF8300Δpvl had lost >15% of the baseline weight and some also exhibited other moribund conditions (Table 4). Moreover, there were no notable differences in gross pathology of kidneys between wild-type and Δpvl mutant strains (data not shown). These results are consistent with a null effect of PVL at the end stages of bacteremia in the co-infection studies in which rabbits had a mean survival time of 3.9 days (Table 3).10.1371/journal.pone.0003198.t004Table 4Time-course single-strain infection experiments with a USA300 parental or isogenic Δpvl mutant strains.1\nSF8300wtSF8300Δpvl\n\nP valuelog10CFU/g±standard deviations24 h post infectionn = 19n = 19Lung3.41±0.573.19±0.380.179Spleen3.62±0.573.44±0.380.27Kidney4.09±1.922.63±1.76\n0.020\n48 h post infectionn = 19n = 18Lung2.88±0.562.79±0.650.65Spleen2.72±0.902.64±0.820.78Kidney4.48±1.493.20±1.91\n0.030\n72 h post infectionn = 12n = 12Lung2.70±1.402.68±1.180.96Spleen2.70±1.343.03±0.780.47Kidney4.75±1.555.50±0.790.151Blood24 h post infection1.28±0.930.96±0.130.07748 h post infection0.68±0.160.50±0.130.3772 h post infection0.71±1.261.15±1.040.361Rabbits were euthanized and log10CFU per gram of lung, spleen, and kidney were determined at 24, 48 and 72 hours post infection. It was not possible to conduct an experimental group at 96 hours post infection because rabbits loss >15% of the baseline weight by 72 hrs post infection, which is a moribund condition stipulated by UCSF animal use committee for euthanization. Two-sided P values by unpaired Student's t test are reported.DiscussionA PubMed search for articles on PVL published in 2002–2007 identified more than 300 articles, suggesting an association between PVL and CA-MRSA disease. Although compelling, epidemiological data alone are insufficient to establish whether PVL directly contributes to widespread dissemination of CA-MRSA strains or severity of infection [24], [25]. As bacteremia accounts for approximately 65% of invasive CA-MRSA disease [19], we used a rabbit model of bacteremia to study the role of PVL in CA-MRSA pathogenesis. Herein, we discovered a transient positive effect of PVL-mediated CA-MRSA pathogenesis in a rabbit bacteremia model. PVL appears to play a modest, but measurable role in pathogenesis during the early stages of bacteremic seeding of the kidney, the target organ from which bacteria were not cleared, as evidence by an increasing bacterial load over time (Table 4). During acute infection, it is possible that PVL-mediated lysis of incoming PMNs enabled better colonization and/or early survival of the parental USA300 strain. However, the early survival advantage conferred by PVL was lost by 72 hours post infection. Although it is unclear why there was no difference in bacterial densities in the kidney between parental wild-type and Δpvl mutant strains at the end stage of bacteremia, it is worth noting that PVL has been shown to prime the host innate immune system [26]–[28] and this could have resulted in enhanced clearance of parental USA300 strain. For example, sublytic concentrations of PVL, orders of magnitude lower than required for granulocyte lysis, induce release of interleukin-8, leukotriene B4, and reactive oxygen species by PMNs, which contribute to innate host defense against bacteria [26]–[28]. In contrast to the parental USA300 wild-type strain, the Δpvl isogenic mutant may not activate the immune response in this manner, allowing it to achieve significant growth in the kidney in the end stages of bacteremia and eventually compensate for any early survival advantage conferred by PVL. Alternatively, in established infection once a certain bacterial density has been achieved, PVL may play only a minimal role in maintaining infection.Our study demonstrated clearly that there is no difference in transcriptome and/or proteome profiles between prevalent CA-MRSA wild-type strains and their isogenic Δpvl mutants, indicating that a PVL effect must be direct and not mediated by interference with global regulatory networks (Figure 1 and Table 1). This stands in marked contrast to the regulatory role of PVL reported by Labandeira-Rey et al. [14] using a laboratory S. aureus strain lysogenized with a PVL-encoding phage [14]. While the reasons underlying these differences are under investigation , the absence of global gene regulatory effects of PVL in prevalent CA-MRSA strains may limit the practical applicability of the Labandeira-Rey et al. investigation [14]. Bubeck Wardenburg et al. [11], [12] found no contribution of PVL to pathogenesis in C57BL6 and BALB/c mouse pneumonia models using wild-type CA-MRSA and isogenic Δpvl mutant strains. The differences in infection outcomes between the two studies could be related to differential levels of staphylococcal protein A (Spa), which was dramatically increased in the PVL-lysogenized laboratory strains [14], but remains unaltered in PVL-harboring CA-MRSA strains (Figure 1). Protein A has well-known inflammatory effects in murine lungs and ligates tumor necrosis factor receptor-1 (TNFR1) on airway epithelial cells [23].In sum, we propose a model in which PVL exhibits a transient contribution to CA-MRSA pathogenesis in rabbit bacteremia model. Enhanced production of PVL in the early, acute phase of infection could contribute to CA-MRSA pathogenesis. This is consistent with the clinical presentation of rapid-onset, overwhelming infection that has been associated with invasive CA-MRSA disease in humans [29]. It is unclear why the effect is transient. Perhaps sublytic production of PVL in the end stages of infection could result in priming of the innate immune response that limits bacterial survival. Alternatively, once a threshold of organisms is achieved in a target organ, even if there is somewhat of a delay in getting to that threshold, factors other than PVL may be important in maintaining persistent infection.Materials and MethodsBacterial strains and cultureClinical strains LAC and SF8300 are pulsed-field gel electrophoresis (PFGE) type USA300-0114, which have been implicated in epidemiologically unassociated outbreaks in the United States [20], [30]. MW2 is PFGE type USA400, the prototype CA-MRSA strain type endemic in the U.S. Midwest [3]. The isogenic PVL knockout (Δpvl) strains, LACΔpvl and MW2Δpvl, have been described previously [10]. SF8300Δpvl was constructed as described for LACΔpvl and MW2Δpvl, in which a spectinomycin resistance cassette replaced both the lukS-PV and lukF-PV genes. Bacterial strains were cultured in tryptic soy broth containing 0.25% D-glucose (TSB, Becton, Dickenson, and Company), CCY medium (3% [wt/vol] yeast extract, 2% Bacto-Casamino acids, 2.3% sodium pyruvate, 0.63% Na2HPO4, and 0.041% KH2PO4 [pH 6.7]), or YCP medium (3% [wt/vol] yeast extract, 2% Bacto-Casamino acids, 2% sodium pyruvate, 0.25% Na2HPO4, and 0.042% KH2PO4 [pH 7.0]). Overnight cultures were diluted 1∶200 and incubated at 37°C with shaking (250 RPM). Unless specified, bacteria were cultured to mid-exponential (TSB, OD600 = 0.75; CCY and YCP, OD600 = 1.0) or stationary (TSB, CCY, and YCP, OD600 = 2.0) phases of growth.Microarray experimentsSF8300, SF8300Δpvl, MW2, and MW2Δpvl were cultured to exponential (OD600 = 1.0) and stationary (OD600 = 2.0) phases of growth in TSB or CCY medium at which time bacteria were lysed using 700 µl of RLT buffer (Qiagen, Valencia, CA) and the lysate was homogenized using an FP120 FastPrep system (Qbiogene, Carlsbad, CA). Total RNA was isolated with RNeasy kits (Qiagen). Contaminating DNA was removed using DNase (on column DNase treatment,Qiagen; off column DNase treatment, Turbo DNase). Fragmented and biotin-dUTP-labelled cDNA was generated from purified RNA as described by the Affymetrix Target Preparation protocol (www.affymetrix.com/support/downloads/ manuals/expression_s3 _manual.pdf). To synthesize cDNA, random primers at 25 ng/ml (Invitrogen), 10 mM DTT, 0.5 mM dNTPs, 0.5 U/µl SUPERase·In (Ambion) and 25 U/µl SuperScript II (Invitrogen) were added to ∼20 µg RNA in 1X first-strand reaction buffer. The remaining RNA was hydrolyzed by adding 1 N NaOH at 65°C for 30 min after which 1 N HCl was added to neutralize the reaction. cDNA purification was performed using a QIAquick PCR purification kit (Qiagen) with nuclease-free water substituted for the elution buffer. cDNA (∼5 µg/sample) was fragmented using 0.6 U of DNase I (GE Healthcare) per mg of cDNA in One-Phor-All Buffer (Amersham Biosciences). Labeling of the 3′termini of the fragmentation products was completed as described in the protocol using 7.5 mM GeneChip DNA Labeling Reagent (Affymetrix), terminal deoxynucleotidyl transferase (Promega) in 5X reaction buffer. The reaction was terminated using 0.5 M EDTA. Biotinylated S. aureus cDNA from strains SF8300 and SF8300Δpvl was hybridized to custom Affymetrix GeneChips (RMLChip1) containing 3961 probe sets from eight different S. aureus strains (COL, EMRSA16, MSSA476, RF122, TSS, 8325, Mu50, and N315). Biotinylated S. aureus cDNA from strains MW2 and MW2Δpvl was hybridized to custom Affymetrix GeneChips (RMLChip3) with 99.3% coverage of genes from MW2 (2613 probe sets of 2632 ORFs; the remaining 0.7% are represented by identical probe sets from other staphylococci). GeneChips were scanned according to standard GeneChip protocols (Affymetrix). Precise details for Affymetrix hybridization and scanning protocols can be found at the above internet address. Each experiment was replicated 3 times. Affymetrix GeneChip Operating Software (GCOS v1.4, http://www.affymetrix.com) was used to perform the preliminary analysis of custom chips at the probe-set level. All *.cel files, representing individual biological replicates, were scaled to a trimmed mean of 500 using a scale mask consisting of only the S. aureus probe-sets to produce the *.chp files. A pivot Table with all samples was created, including calls, call p-value and signal intensities for each gene. The pivot Table was imported into GeneSpring GX 7.3 (Agilent), and hierarchical clustering (condition tree) using a Pearson correlation similarity measure with average linkage was used to determine similarity of biological replicates (data not shown). The pivot Table was also imported into Partek software (Partek Inc. St. Louis, MO) to produce a Principle Component Analysis plot as a secondary check on similarity of biological replicates (data not shown). After data had passed these preliminary statistical tests, biological replicates were combined into a custom worksheet (Microsoft Excel 2003, Microsoft Corporation) used to correlate replicates of all test conditions and controls. Quality filters based upon combined calls and signal intensities (test and/or control signal intensity was required to be greater than the average background signal intensity of 27) were used in the worksheet to further evaluate individual gene comparisons. Present and Marginal calls were treated as the same. Absent calls were negatively weighted for the filters and dropped completely from further calculations. All individual genes passing the above filters and combined from all usable replicates have the ratios of test (wild-type strain)/control (mutant strain) reported with associated probability computed by a paired Student's t-test. Significance Analysis of Microarrays (SAM) was also performed using the Excel sheet with a column added containing the results. P-values obtained from ANOVA (Partek) were filtered using the False Discovery Rate (FDR). Gene lists were generated with emphasis placed upon the quality and statistical filters mentioned above. To be included in the final gene list, in addition to the above criteria, gene expression must have been changed at least 2-fold. Microarray data are posted on the Gene Expression Omnibus (GEO, www.ncbi.nlm.nih.gov/geo/, platform accession numbers GPL2129 for RMLChip1 data (SF8300 strains) and GPL4692 for RMLChip3 data (MW2 strains), series accession number GSE8677).TaqMan real-time reverse transcriptase-PCRTaqMan real-time RT-PCR analysis of 3–7 separate experiments (each assayed in triplicate) was performed with an ABI 7500 thermocycler (Applied Biosystems) using RNA samples prepared as described for the microarray experiments. Relative quantification of S. aureus genes was determined by the change in expression of target transcripts relative to the endogenous control gene, gyrB. Data were subsequently expressed as fold-change in wild-type transcript levels compared to the isogenic lukS/F-PV mutant strains (Δpvl strains set at 1.0, baseline). lukS/F-PV transcripts were undetecTable in the Δpvl strains.Analysis of S. aureus protein profiles\nS. aureus were cultured to mid-exponential (TSB, OD600 = 0.75; CCY and YCP, OD600 = 1.0) or stationary (TSB, CCY, and YCP, OD600 = 2.0) phases of growth as described above. At the desired phase of growth, 10 ml of culture was centrifuged at 2800×g for 10 min at 4°C. Supernatant (5–7 ml) was collected and concentrated to 2X using Centriplus centrifugal filters with 3,000 MW cut-off membranes (Millipore, Bedford MA). Cell pellets were washed in 1 ml of cold Dulbecco's phosphate-buffered saline (DPBS) and resuspended in 600 µL of cold DPBS. Sample was loaded into a pre-chilled FastPROTEIN BLUE tube (MP Biomedical, Solon, OH) and S. aureus were disrupted/homogenized with a FP120 FastPrep Instrument (Qbiogene) set at speed 6.0 for 20 sec. Samples were immediately returned to ice. After homogenization, tubes were centrifuged at 10,000×g for 1 min at 4°C and supernatant was transferred to a new tube. Samples were stored at −80°C until used.For SDS-PAGE, equivalent volumes of cell extract (5.6 µL of protein from exponential phase samples and 1.6 µL of protein from stationary phase samples) or culture supernatant (54 µL of 2X sample) were resolved by 12% or 10–20% SDS-PAGE (Protean II gel, Bio-Rad). Protein gels were stained with Gel-Code according to the manufacturer's instructions (Pierce). Images were adjusted for brightness and contrast in Adobe Photoshop CS (Adobe Systems Incorporated, San Jose, CA).Western blot for alpha-toxin and LukF-PV (PVL)Overnight cultures of S. aureus were diluted 1∶100 in 5 ml of YCP medium and incubated at 37°C with shaking (250 rpm). Bacteria were cultured to mid-exponential (OD600 = 1.1) or stationary (OD600 = 1.7) phases of growth and removed by centrifugation (5000×g for 10 min at 4°C). Secreted proteins were precipitated using TCA (10% w/v), washed, and solubilized in 200 µl 10 mM Tris-Cl, pH 7.6. Proteins (15 µl of each sample) were resolved with 12% SDS-PAGE, followed by transfer to nitrocellulose. Membranes were blocked overnight at 4°C with 5% non-fat milk (Bio-Rad) and then incubated with rabbit polyclonal antibodies specific for LukF-PV (1 µg/ml, a kind gift of Dr. Gerard Lina, Lyon, France) or alpha-toxin (1∶10,000 dilution, Sigma-Aldrich, St Louis, MO) for 2 h at room temperature. Immunoblots were washed and incubated with horseradish peroxidase-coupled goat anti-rabbit IgG (1∶4,000 dilution, Zymed, CA) for 1 h at room temperature. Proteins were detected with enhanced chemiluminescence (GE, Piscataway, NJ) systems and autorad film (IscBioExpress, UT). Protein A (Spa) was detected by rabbit polyclonal antibody specific for Spa.Rabbit bacteremia modelBacterial strains were grown in TSB at 37°C with shaking for 16–18 hours, harvested, and resuspended in 10% glycerol/phosphate buffered saline to approximately 5×108 colony forming units (CFUs) per ml, aliquoted into individual cryovials and immediately stored at −80°C. For competition experiments, a mixture containing approximately 3×107 CFUs of wild type parent and 3×107 CFUs of isogenic Δpvl mutant were used to co-infect New Zealand white rabbits via the marginal ear vein. A 0.5 ml volume of blood was sampled daily from the ear artery for quantitative blood culture. Rabbits were monitored twice daily to identify moribund animals (defined as those that are immobile, cannot be aroused to move from a recumbent position, and unable to access food or water) and those that have lost more than 15% of baseline weight, which are conditions stipulated by the UCSF committee on animal research for immediate euthanization. Rabbits were dead or euthanized for moribund conditions between 2 and 7 days post infection. A 0.2 to 0.3 g sample from lung, spleen and kidney was processed for quantitative bacterial culture onto blood agar plate (BAP; tryptic soy agar supplemented with 5% sheep blood; Remel, Lenexa, KS).For single-strain experiments, 8×107 CFUs of the parental strain SF8300 or the mutant strain SF8300Δpvl were injected via the marginal ear vein into New Zealand white rabbits (no difference in CFUs of SF8300 or SF8300Δpvl administered to rabbits, P>0.05 by unpaired Student's t test). Rabbits were euthanized and log10CFU per gram of lung, spleen, and kidney determined at 24, 48 and 72 hours post infection.The animal experiments reported herein were reviewed by the University of California San Francisco Institutional Animal Care and Use Committee (IACUC). Animals were housed in humane conditions in accordance with IACUC policies and procedures.StatisticsFor the rabbit co-infection model, the input ratio of the parent to mutant was determined by transferring 144 CFUs of the mixed inoculum onto TSA containing 300 µg/ml of spectinomycin. The parental strains were susceptible to spectinomycin, whereas the Δpvl mutant strains were resistant because a spectinomycin resistance cassette was used in allelic replacement of wild type pvl genes [10]. Similarly, the output ratio of the parent and mutant was determined for each rabbit tissues (lung, spleen, kidney, and blood) by transferring 144 CFUs onto TSA containing 300 µg/ml spectinomycin. For each rabbit tissue, a competition index (CI) of the two comparator strains was calculated with the following formula that corrects the output ratios for variations in the input ratios: CI = log10(output ratio/input ratio). A positive CI value indicates enhanced tissue infectivity of the parent, whereas a negative CI value indicates enhanced tissue infectivity of the mutant; CI = 0 is the no-effect value. A paired Student's t test was used to test the null hypothesis (CI = 0) that there was no difference in tissue infectivity between the parent and the Δpvl mutant. Linear trends in bacterial densities in vital organs and blood over time course of infection were explored by means of the Cuzick test (Stata 8, nptrend command) [31].\n\nREFERENCES:\n1. VandeneschFNaimiTEnrightMCLinaGNimmoGR\n2003\nCommunity-acquired methicillin-resistant Staphylococcus aureus carrying Panton-Valentine leukocidin genes: worldwide emergence.\nEmerg Infect Dis\n9\n978\n984\n12967497\n2. RobinsonDAKearnsAMHolmesAMorrisonDGrundmannH\n2005\nRe-emergence of early pandemic Staphylococcus aureus as a community-acquired meticillin-resistant clone.\nLancet\n365\n1256\n1258\n15811459\n3. NaimiTSLeDellKHComo-SabettiKBorchardtSMBoxrudDJ\n2003\nComparison of community- and health care-associated methicillin-resistant Staphylococcus aureus infection.\nJama\n290\n2976\n2984\n14665659\n4. MoranGJKrishnadasanAGorwitzRJFosheimGEMcDougalLK\n2006\nMethicillin-resistant S. aureus infections among patients in the emergency department.\nN Engl J Med\n355\n666\n674\n16914702\n5. KingMDHumphreyBJWangYFKourbatovaEVRaySM\n2006\nEmergence of community-acquired methicillin-resistant Staphylococcus aureus USA 300 clone as the predominant cause of skin and soft-tissue infections.\nAnn Intern Med\n144\n309\n317\n16520471\n6. HotaBEllenbogenCHaydenMKAroutchevaARiceTW\n2007\nCommunity-Associated Methicillin-Resistant Staphylococcus aureus Skin and Soft Tissue Infections at a Public Hospital: Do Public Housing and Incarceration Amplify Transmission?\nArch Intern Med\n167\n1026\n1033\n17533205\n7. NguyenDMMascolaLBrancoftE\n2005\nRecurring methicillin-resistant Staphylococcus aureus infections in a football team.\nEmerg Infect Dis\n11\n526\n532\n15829189\n8. DiepBACarletonHAChangRFSensabaughGFPerdreau-RemingtonF\n2006\nRoles of 34 virulence genes in the evolution of hospital- and community-associated strains of methicillin-resistant Staphylococcus aureus.\nJ Infect Dis\n193\n1495\n1503\n16652276\n9. DiepBAOttoM\n2008\nThe role of virulence determinants in community-associated MRSA pathogenesis.\nTrends Microbiol\n16\n361\n369\n18585915\n10. VoyichJMOttoMMathemaBBraughtonKRWhitneyAR\n2006\nIs Panton-Valentine Leukocidin the Major Virulence Determinant in Community-Associated Methicillin-Resistant Staphylococcus aureus Disease?\nJ Infect Dis\n194\n1761\n1770\n17109350\n11. Bubeck WardenburgJBaeTOttoMDeleoFRSchneewindO\n2007\nPoring over pores: alpha-hemolysin and Panton-Valentine leukocidin in Staphylococcus aureus pneumonia.\nNat Med\n13\n1405\n1406\n18064027\n12. Bubeck WardenburgJSchneewindO\n2008\nVaccine protection against Staphylococcus aureus pneumonia.\nJ Exp Med\n205\n287\n294\n18268041\n13. Bubeck WardenburgJPalazzolo-BallanceAMOttoMSchneewindODeLeoFR\n2008\nPanton-Valentine leukocidin is not a virulence determinant in murine models of community-associated methicillin-resistant Staphylococcus aureus disease.\nJ Infect Dis. in press\n14. Labandeira-ReyMCouzonFBoissetSBrownELBesM\n2007\nStaphylococcus aureus Panton-Valentine leukocidin causes necrotizing pneumonia.\nScience\n315\n1130\n1133\n17234914\n15. ChenJNovickRP\n2007\nsvrA, a multi-drug exporter, does not control agr.\nMicrobiology\n153\n1604\n1608\n17464075\n16. McNamaraPJIandoloJJ\n1998\nGenetic instability of the global regulator agr explains the phenotype of the xpr mutation in Staphylococcus aureus KSI9051.\nJ Bacteriol\n180\n2609\n2615\n9573143\n17. SzmigielskiSPrevostGMonteilHColinDAJeljaszewiczJ\n1999\nLeukocidal toxins of staphylococci.\nZentralbl Bakteriol\n289\n185\n201\n10360319\n18. SzmigielskiSJeljaszewiczJWilczynskiJKorbeckiM\n1966\nReaction of rabbit leucocytes to staphylococcal (Panton-Valentine) leucocidin in vivo.\nJ Pathol Bacteriol\n91\n599\n604\n4224060\n19. KlevensRMMorrisonMANadleJPetitSGershmanK\n2007\nInvasive methicillin-resistant Staphylococcus aureus infections in the United States.\nJama\n298\n1763\n1771\n17940231\n20. DiepBAGillSRChangRFPhanTHChenJH\n2006\nComplete genome sequence of USA300, an epidemic clone of community-acquired meticillin-resistant Staphylococcus aureus.\nLancet\n367\n731\n739\n16517273\n21. Palazzolo-BallanceAMReniereMLBraughtonKRSturdevantDEOttoM\n2008\nNeutrophil microbicides induce a pathogen survival response in community-associated methicillin-resistant Staphylococcus aureus.\nJ Immunol\n180\n500\n509\n18097052\n22. DiepBAStoneGGBasuinoLGraberCJMillerA\n2008\nThe arginine catabolic mobile element and staphylococcal chromosomal cassette mec linkage: convergence of virulence and resistance in the USA300 clone of methicillin-resistant Staphylococcus aureus.\nJ Infect Dis\n197\n1523\n1530\n18700257\n23. GomezMILeeAReddyBMuirASoongG\n2004\nStaphylococcus aureus protein A induces airway epithelial inflammatory responses by activating TNFR1.\nNat Med\n10\n842\n848\n15247912\n24. ChambersHF\n2005\nCommunity-associated MRSA–resistance and virulence converge.\nN Engl J Med\n352\n1485\n1487\n15814886\n25. RossneyASShoreACMorganPMFitzgibbonMMO'ConnellB\n2007\nThe emergence and importation of diverse genotypes of methicillin-resistant Staphylococcus aureus (MRSA) harboring the Panton-Valentine leukocidin gene (pvl) reveal that pvl is a poor marker for community-acquired MRSA strains in Ireland.\nJ Clin Microbiol\n45\n2554\n2563\n17581935\n26. KonigBPrevostGPiemontYKonigW\n1995\nEffects of Staphylococcus aureus leukocidins on inflammatory mediator release from human granulocytes.\nJ Infect Dis\n171\n607\n613\n7533198\n27. KonigBKollerMPrevostGPiemontYAloufJE\n1994\nActivation of human effector cells by different bacterial toxins (leukocidin, alveolysin, and erythrogenic toxin A): generation of interleukin-8.\nInfect Immun\n62\n4831\n4837\n7927762\n28. ColinDAMonteilH\n2003\nControl of the oxidative burst of human neutrophils by staphylococcal leukotoxins.\nInfect Immun\n71\n3724\n3729\n12819053\n29. GonzalezBEMartinez-AguilarGHultenKGHammermanWACoss-BuJ\n2005\nSevere Staphylococcal sepsis in adolescents in the era of community-acquired methicillin-resistant Staphylococcus aureus.\nPediatrics\n115\n642\n648\n15741366\n30. KennedyADOttoMBraughtonKRWhitneyARChenL\n2008\nEpidemic community-associated methicillin-resistant Staphylococcus aureus: recent clonal expansion and diversification.\nProc Natl Acad Sci U S A\n105\n1327\n1332\n18216255\n31. CuzickJ\n1985\nA Wilcoxon-type test for trend.\nStat Med\n4\n87\n90\n3992076\n32. BurlakCHammerCHRobinsonMAWhitneyARMcGavinMJ\n2007\nGlobal analysis of community-associated methicillin-resistant Staphylococcus aureus exoproteins reveals molecules produced in vitro and during infection.\nCell Microbiol\n9\n1172\n1190\n17217429"
+}

batch_5/PMC2527551.json

@@ -0,0 +1,4 @@
+{
+    "id": "PMC2527551",
+    "text": "This is an academic paper. This paper has corpus identifier PMC2527551\nAUTHORS: Kazuhiro Kotoh, Munechika Enjoji, Masaki Kato, Motoyuki Kohjima, Makoto Nakamuta, Ryoichi Takayanagi\n\nABSTRACT:\nBackgroundAlthough most patients with severe acute hepatitis are conservatively cured, some progress to acute liver failure (ALF) with a high rate of mortality. Based on the evidence that over-activation of macrophages, followed by disturbance of the hepatic microcirculation, plays a key role in ALF, we hypothesized that the production of serum lactate dehydrogenase (LDH) might increase in the liver under hypoxic conditions and could be an indicator to discriminate between conservative survivors and fatal patients at an early stage.ResultsTo confirm this hypothesis, we developed a new parameter with serum alanine aminotransferase (ALT) and LDH: the ALT-LDH index = serum ALT/(serum LDH - median of normal LDH range). We analyzed retrospectively 33 patients suffering acute liver injury (serum ALT more than 1000 U/L or prothrombin time expressed as international normalized ratio over 1.5 at admission) and evaluated the prognostic value of the ALT-LDH index, comparing data from the first 5 days of hospitalization with the Model for End-Stage Liver Disease (MELD) score. Patients whose symptoms had appeared more than 10 days before admission were excluded from this study. Among those included, 17 were conservative survivors, 9 underwent liver transplantation (LT) and 7 died waiting for LT. We found a rapid increase in the ALT-LDH index in conservative survivors but not in fatal patients. While the prognostic sensitivity and specificity of the ALT-LDH index was low on admission, at day 3 they were superior to the results of MELD.ConclusionALT-LDH index was useful to predict the prognosis of the patients with acute liver injury and should be helpful to begin preparation for LT soon after admission.\n\nBODY:\nBackgroundAcute liver failure (ALF) or fulminant liver failure is a disease characterized by abrupt onset and high mortality. Liver transplantation (LT) is the only effective treatment for ALF and many patients die before undergoing LT because of rapid progression of the disease [1,2]. Therefore, a prompt decision regarding LT is required following an early determination of prognosis. Among the various clinical selection criteria proposed for LT, the King's College criteria and the Model for End-Stage Liver Disease (MELD) criteria have been applied widely [3,4]. However, those criteria include some factors reflecting multiple or systemic organ failure, which means that many patients fulfilling the criteria are already too unwell for transplantation to be contemplated. The poor prognosis of ALF seems to be attributable to the definition of the disease itself. Generally, ALF is defined as an acute liver disease complicated with hepatic encephalopathy and severe coagulopathy. Considerable efforts made in the past to improve the prognosis of ALF have shown limitations. It is well known that supportive methods such as plasma exchange and hemodialysis are not necessarily efficacious once encephalopathy develops in patients suffering from severe acute hepatitis [5-8]. In order to improve the overall prognosis of ALF, it is necessary to seek ways to select patients who have the possibility of developing hepatic encephalopathy before the symptom appears, rather than struggle to cure the patients after fulfilling the ALF criteria. Of course, a new strategy is required to prevent the progression of the disease.The difficulty in predicting the prognosis of ALF is mainly attributable to incomplete elucidation of its mechanism. The most characteristic pathological finding of ALF is massive necrosis without regeneration, which implies the involvement a disturbance of the hepatic circulation in the progression of the disease. Although this idea is not novel and has not been regarded as important, we believe that it should be revisited, considering recent reports of over-activation of macrophages in the liver, which is believed to cause hepatic hypoxia as a result of disturbance of the microcirculation [9-12]. Although the importance of over-activation of hepatic macrophages in the progression of ALF may be accepted, it is difficult to demonstrate the occurrence of this phenomenon. Whilst liver biopsy is a reliable means of confirming macrophage proliferation in the liver, it carries a risk of bleeding, especially with the coagulopathy seen in ALF. Therefore, we focused on lactate dehydrogenase (LDH), which is recognized as an enzyme released in liver injury, as are aspartate aminotransferase and alanine aminotransferase (ALT). It is common to regard monitoring serum LDH as of little value because it is produced in various organs and the specificity for liver disease is low. However, in ischemic liver disease, the elevation of serum LDH is more pronounced than that of ALT [13-16]. Several pieces of evidence that the production of LDH increases in hypoxic conditions have been reported [17-19]. Another consideration regarding serum LDH in liver disease is its more rapid decline than ALT, because of its shorter half-life in serum [20]. Against the background of these findings, we hypothesized that the ALT-LDH ratio could be a marker indicating the degree of hepatic hypoxia caused by macrophage over-activation, which might be helpful to discriminate between fatal patients and conservative survivors at an early stage of ALF. In this study, we examined retrospectively the correlation between the serum ALT-LDH ratio of the patients suffering from acute liver injury and who had had the possibility of developing ALF and their outcomes, and evaluated the predictive efficacy of this new indicator compared to the MELD scoring system.ResultsComparing the backgrounds on admission between LT or death cases and the conservative survivors, the former were older and had a higher proportion of patients with ascites and hepatic encephalopathy (Table 1). The laboratory data on admission showed that the LT or death cases had significantly lower serum levels of albumin and longer PT. Concerning the enzyme activities in serum, the average values for AST and ALT were higher in conservative survivors, while LDH was lower, although the difference was not significant. The MELD score on admission was about 9 points higher in the LT or death cases; however, their average value was below 30, probably reflecting that the patients were at an early stage of their clinical courses.Table 1Characteristics of the patients.AliveLT or DeathTotalp-valueAge39.4 ± 15.348.3 ± 16.343.7 ± 16.20.0382Sex (M/F)10/78/818/150.6109Ascites (+/-)2/1510/612/210.0025Encephalopathy (+/-)2/159/711/220.0067AST (U/L)4122.4 ± 3915.13587.3 ± 4123.73862.9 ± 3963.40.5523ALT (U/L)3845.5 ± 2932.12777.8 ± 2850.63327.8 ± 2898.50.2275LDH (U/L)2668.0 ± 3431.02796.1 ± 4093.22730.1 ± 3707.30.8289ALP (U/L)543.7 ± 168.1509.4 ± 162.5527.1 ± 163.70.9139γ-GTP (U/L)293.1 ± 200.0245.8 ± 249.8270.2 ± 223.20.171Total bilirubin (mg/dL)9.1 ± 6.814.6 ± 10.011.8 ± 8.80.0689Direct bilirubin (mg/dL)6.1 ± 4.79.4 ± 6.97.6 ± 6.00.1395Albumin (g/dL)3.7 ± 0.43.3 ± 0.43.5 ± 0.40.0034PT-INR2.19 ± 1.563.38 ± 2.292.76 ± 2.000.0013Platelet (× 104/μL)14.9 ± 5.911.5 ± 5.913.3 ± 6.10.0717Creatinine (mg/dL)1.14 ± 2.031.35 ± 1.611.25 ± 1.810.5757Etiology – HAV3360.9811Etiology – HBV85130.9811Etiology – Drug2130.9811Etiology – Wilson1120.9811Etiology – Unknown3690.9811MELD score17.66 ± 9.7926.69 ± 11.8922.0 ± 11.60.0059ALT-LDH index 3.088160.8658ALT-LDH index ≥ 3.098170.8658Values were expressed by mean ± standard deviation. LT: Liver transplantation, GGT: Gamma-glutamyl transferase.The serum ALT levels on admission were over 1000 U/L in 27 patients, and decreased quickly (23 patients) or remained steady (10 patients) during the first three hospital days. This finding also indicated that the periods from the onset of the disease to admission of the patients were relatively short. There was one patient who showed re-elevation of serum ALT after the third hospital day, triggered by HBV. There was no particular tendency of ALT transition during the first five days in the conservative survivors or the LT or death cases. On the other hand, the transition of the ALT-LDH index during the same period differed between the two categories: the index increased quickly in most of the conservative survivors while it tended to remain low in the LT or death cases (Figure 1). We confirmed that there was no evidence indicating haemolysis for all enrolled patients. There were two patients with normal serum LDH but high serum ALT activity on admission, both belonged to the conservative survivors group. As shown in Figure 2, both showed rapid improvement of serum ALT and PT-INR after hospitalization, without any particular support.Figure 1Changes in the ALT-LDH index over the first 5 days after admission. In most of the conservative survivors, a rapid elevation of the index was observed. Once the serum LDH activity reached the normal range (below 229 U/L), subsequent plotting was avoided. Two patients who had serum LDH within the normal range at admission are not included in the figure. On the contrary, the index decreased or remained constant in most of the patients who died before LT or underwent LT.Figure 2Patients with normal serum LDH activity at admission. The clinical courses of the two patients who had normal serum LDH activity at admission despite high levels of serum ALT. Their liver function improved rapidly without particular intervention.The ROC curves predicting conservative survivors are illustrated with the ALT-LDH index and MELD score using data from the first and the third hospital days, respectively (Figure 3). The MELD score showed similar curves using the first and third days' data, while the ALT-LDH index for the third day showed much higher sensitivity and specificity, and was superior to the MELD score, although the curve for the first day was close to the identity line corresponding to a complete lack of discriminative power. The area under the ROC curve of the ALT-LDH index for the third day was 0.893 while that of the MELD score was 0.777 (Table 2).Table 2ROC curves with MELD score and ALT-LDH index predicting conservative survivors.Area under ROCStd. Error95% C.I.p-valueMELD (day1)0.7500.08570.582 – 0.9180.0143MELD (day3)0.7770.08410.612 – 0.9410.00779ALT-LDH (day1)0.5740.1020.373 – 0.7740.471ALT-LDH (day3)0.8930.06290.770 – 1.020.000118Figure 3ROCs using the MELD score or ALT-LDH-index. The curves from the MELD score for the first and third hospital days are similar. On the other hand, the ALT-LDH index on the third day improved in sensitivity and specificity compared with the curve for the day of admission.In past reports evaluating the predictive efficacy of the MELD score for ALF, the cut-off line was set between 30 and 35. As shown in Table 3, the MELD score from the first day data showed a high specificity of 88.24%, but a low sensitivity of 31.25%, at cut-off line 30 and this tendency did not change at cut-off line 35. On the contrary, both the sensitivity and specificity calculated by the ALT-LDH index with a cut-off of 3.0 increased from the first day to the third day: 75% and 100%, respectively.Table 3Prognostic values of MELD score and ALT-LDH index predicting conservative survivors.Sensitivity (%)Specificity (%)PPV (%)NPV (%)Efficiency (%)MELD (day1) <3031.388.271.457.760.6MELD (day1) <3518.894.17555.260.6MELD (day3) <3033.388.271.46062.5MELD (day3) <352094.17557.159.4ALT-LDH (day1) >3.05052.95052.951.5ALT-LDH (day3) >3.07510010081.087.9PPV: positive predictive value, NPV: negative predictive value.DiscussionIn this study we demonstrated the contrasting transitions of the ALT-LDH index in the early stage of acute liver injury between the conservative survivors and the patients with progressive fatal liver failure. In the former, the ALT-LDH index increased abruptly soon after the peak of serum ALT elevation, which was caused by a more rapid decrease of LDH than ALT activity. This phenomenon is convincing because the half-life of serum LDH is normally much shorter than that of serum ALT. On the other hand, in the fatal patients group, a less rapid decrease of serum LDH activity kept the ALT-LDH index low, which implied that the delayed decrease of serum LDH at an early stage of ALF may be closely related to a poor prognosis. This phenomenon might be explained by assuming hypoxic conditions in the livers of the patients with progressive ALF.Although the mechanism of ALF has not been elucidated fully, several authors recently reported that over-activation of macrophages plays a key role in the progression of ALF [9-12]. The activated and proliferating macrophages in the liver could injure endothelial cells and cause a disturbance in the hepatic microcirculation. We suppose that this may be the main process of ALF, at least for the non-acetaminophen type. Meanwhile, LDH is an essential enzyme involved in anaerobic glycolysis and is responsible for the anaerobic transformation of pyruvate to lactate. Increased expression of LDH under hypoxic conditions has been demonstrated in various cell lines [17-19]. Concerning liver diseases, it is well known that dominant elevation of serum LDH is observed in hypoxic hepatitis caused by shock or heart failure [13-16]. Although the elevation of LDH activity in acute liver injury has been simply supposed to be enzyme leakage through damaged hepatocyte membranes, as the seen with ALT, increased LDH production could also be attributable to anaerobic conditions. The hepatocytes are expected to increase the production of LDH under anaerobic conditions, until they become necrotic. From this viewpoint, the persistent low ALT-LDH index in fatal patients might be the result of increased production of LDH from residual living hepatocytes in hypoxia. Prolonged hypoxic conditions could cause massive or lobular necrosis, which coincides with the pathologic findings of ALF.When we accept the mechanism described above, acute liver injury could be supposed to consist of two different processes of cell destruction. One is direct cytotoxicity toward hepatocytes caused by various triggers. In most non-acetaminophen hepatitis, cytotoxic T cells attack hepatocytes directly. In this process, the increased release of enzymes into the serum is the result of simple leakage from injured hepatocytes, and enzyme activities decrease rapidly, according to their half-lives, as soon as the triggers are removed or inactivated. The other mechanism is hypoxic liver injury caused by disturbance of the hepatic microcirculation. The persistent low ALT-LDH index may imply the situation that the hypoxic process mainly remains after the removal of the trigger of liver injury. Most acute liver injury might be explained as a mixture of these two mechanisms, to various degrees. The patients shown in Figure 3 are supposed to be representatives of cases that almost lack a hypoxic process.In the past, many attempts have been made to predict the prognosis of ALF [21-24]. However, it is impossible to estimate the prognosis using data from a single time point at a very early stage because ALF is a disease with rapid progression and patients may present at various phases of the clinical course. The MELD score is certainly useful to predict the prognosis of patients awaiting LT. However, as shown by our results, its sensitivity remained very low over several days after admission. It is a matter of course because the MELD score was determined principally using data from patients in their end stage. On the other hand, while the sensitivity and specificity of the ALT-LDH index were rather poor on admission, both were improved dramatically beyond the MELD score at day 3. That is, the ALT-LDH index could reflect the rapid clinical change of ALF. We emphasize that the important thing is to observe the transition of clinical data, not simply a single time-point, in a disease with rapid progression, such as ALF.ConclusionIn this study, we showed the efficacy of the ALT-LDH index to predict the prognosis of patients with acute liver injury at their early stages. This index should enable us to begin preparation for LT shortly after admission. We believe that the index could be a support for other indicators, such as the MELD score. However, the number of the enrolled patients into this study was not enough. The further evaluation in larger prospective clinical studies is required.MethodsPatientsPatients with severe acute liver injury referred to our hospital for consideration for LT between April 2000 and March 2004 were analyzed retrospectively. Among them, those with serum ALT activity more than 1000 U/L or prothrombin time expressed as international normalized ratio (PT-INR) over 1.5 were enrolled into this study, amounting to 33 patients (Table 1). In order to focus on the early phase of ALF, those in whom the onset of any of clinical symptoms, such as general fatigue, appetite loss, nausea and jaundice, had begun 10 days before admission were excluded from this study. Hepatic encephalopathy grade 2 or more was seen in 11 (33%) on admission. The etiology of liver injury varied: 6 hepatitis A virus (HAV), 13 hepatitis B virus (HBV), 3 drugs other than acetaminophen, 2 Wilson's disease and 9 indeterminate. Laboratory data were checked daily in the morning. Plasma exchange was performed in the afternoon when hepatic encephalopathy was greater than grade 2 or prolonged downhill PT activity was observed. Among the enrolled patients, 17 were conservative survivors, 9 underwent LT and 7 died waiting for LT. In following analysis we considered the fatal patients and those who were transplanted as one category because the pathological examination showed that the livers of all transplant recipients were markedly atrophic and entirely necrotized, which indicated that they would have not been able to survive without LT.ALT-LDH indexThe serum ALT and LDH activities were measured using the 7500 Clinical Analyzer (Hitachi High-Technologies Corporation, Tokyo, Japan). LDH was assayed using an enzymatic rate method with lactate as the substrate (lactate-pyruvate direction). ALT assay was performed without pyridoxal phosphate supplementation. The normal ranges of ALT and LDH were 6–30 U/L and 119–229 U/L, respectively. We aimed to evaluate the increase of these enzymes above normal levels and developed a new index calculated by following formula:ALT-LDH index = serum ALT/(serum LDH - median of normal LDH range)In acute liver injury, both serum ALT and LDH commonly decrease after the peak observed in the acute phase regardless of the prognosis. However, in the patients with fatal prognosis, the decrease of serum LDH is expected to delay compared with that of serum ALT, which would be caused by microcirculation disturbance in liver. Therefore, if we use the serum LDH value as a predictive marker of ALF, it should be evaluated under connection with the serum ALT value. Although the simple ALT/LDH ratio seems to be acceptable in evaluation of the serum activity of LDH connecting to ALT, it could not reflect the degree of those enzymes' elevation from normal level when they are in relatively low levels because of the difference of their normal ranges. On the other hand, the value of ALT-LDH index distinctly increases when the serum LDH decreases close to the normal range.For the enrolled patients, this index was calculated for the first 5 days from their admission, comparing the changes in serum ALT activity during the same period. According to the normal range of our assay system, the median of the serum LDH was calculated as 174 U/L in this study.Statistical analysisDifferences in clinical backgrounds and laboratory data between conservative survivors and fatal patients, including those who underwent LT, were analyzed using the Χ2-test and Student t-test. The utilities of the ALT-LDH index and MELD score were evaluated using receiver operating characteristic (ROC) curves. The sensitivity, specificity, positive and negative predictive values (PPV and NPV), efficiency, and area under the ROC curve were calculated for each indicator.Competing interestsThe authors declare that they have no competing interests.Authors' contributionsKK conceived the design of the study and prepared the manuscript. ME and MK participated in the study design. MK analyzed clinical data. MN and RT drafted the paper. All authors read and approved the final manuscript.\n\nREFERENCES:\nNo References"
+}

batch_5/PMC2527571.json

@@ -0,0 +1,4 @@
+{
+    "id": "PMC2527571",
+    "text": "This is an academic paper. This paper has corpus identifier PMC2527571\nAUTHORS: Darby Tien-Hao Chang, Yu-Yen Ou, Hao-Geng Hung, Meng-Han Yang, Chien-Yu Chen, Yen-Jen Oyang\n\nABSTRACT:\nBackgroundThough prediction of protein secondary structures has been an active research issue in bioinformatics for quite a few years and many approaches have been proposed, a new challenge emerges as the sizes of contemporary protein structure databases continue to grow rapidly. The new challenge concerns how we can effectively exploit all the information implicitly deposited in the protein structure databases and deliver ever-improving prediction accuracy as the databases expand rapidly.FindingsThe new challenge is addressed in this article by proposing a predictor designed with a novel kernel density estimation algorithm. One main distinctive feature of the kernel density estimation based approach is that the average execution time taken by the training process is in the order of O(nlogn), where n is the number of instances in the training dataset. In the experiments reported in this article, the proposed predictor delivered an average Q3 (three-state prediction accuracy) score of 80.3% and an average SOV (segment overlap) score of 76.9% for a set of 27 benchmark protein chains extracted from the EVA server that are longer than 100 residues.ConclusionThe experimental results reported in this article reveal that we can continue to achieve higher prediction accuracy of protein secondary structures by effectively exploiting the structural information deposited in fast-growing protein structure databases. In this respect, the kernel density estimation based approach enjoys a distinctive advantage with its low time complexity for carrying out the training process.\n\nBODY:\nFindingsMotivationIn structural biology, protein secondary structures act as the building blocks for the protein tertiary structures [1,2]. Therefore, analysis of protein secondary structures is an essential intermediate step toward obtaining a comprehensive picture of the tertiary structure of a protein. In this respect, one of the main challenges is how to accurately identify the polypeptide segments that could fold to form a secondary structure. This problem is normally referred to as prediction of protein secondary structures [1,3].Though prediction of protein secondary structures has been an active issue in bioinformatics research for quite a few years and many approaches have been proposed [1,4-10], a new challenge emerges as the sizes of contemporary protein structure databases continue to grow rapidly. The new challenge, which has been addressed in several recently completed studies [9-11], is concerned with how we can effectively exploit the information implicitly deposited in the ever-growing protein structure databases and deliver ever-improving prediction accuracy. In this respect, this article proposes the Prote2S predictor designed with a novel kernel density estimation algorithm [12], which features an average time complexity of O(nlogn) for carrying out the training process, where n is the number of instances in the training dataset.Experimental resultsThis section reports the experiments conducted to investigate how Prote2S performs in comparison with the other existing predictors of protein secondary structures. The design of Prote2S is based on the relaxed variable kernel density estimator (RVKDE) that we have recently proposed [12]. In the next section, we will discuss how the RVKDE has been incorporated in the design of Prote2S and the related issues.For Prote2S, the training dataset was derived from the PDB version available at the end of May, 2007. In order to guarantee that no protein chains used to generate the training dataset is homologous to the benchmark protein chains on the EVA server [13], from which the testing dataset was extracted, BLAST [14] was invoked and the criterion of homology was set to sequence identity higher than 25%. Then, the CD-HIT clustering algorithm [15] with the similarity threshold set to 0.4 was invoked to remove redundant protein chains in the PDB. After these two processes, a total of 8006 protein chains remained. To generate the training dataset, we followed the approach employed in [6]. Accordingly, one training instance was created for each residue in the 8006 protein chains by associating the residue with the position specific scoring matrix (PSSM) computed by the PSI-BLAST software package [14] with window size set to 15. As a result, a total of 1,801,039 training instances were generated and each was labeled by DSSP [16] as one of the three types of secondary structure elements: alpha-helix, beta-strand, or coil.The testing dataset used in the following experiments was derived from the 106 benchmark protein chains released on the EVA server between September 7, 2004 and March 1, 2006. We extracted only those 89 protein chains of which the prediction results made by all the 5 predictors involved in the comparison are available on the EVA server. The testing dataset then comprises 27 long protein chains, each of which contains more than 100 residues, and 62 short protein chains.In addition to the training and testing datasets, we generated a validation dataset for tuning the parameters in Prote2S. How the validation dataset was generated and how the validation process was carried out will be elaborated in the next section.Table 1, 2, 3 show how Prote2S performs with the testing dataset in comparison with the other predictors whose results are available on the EVA server. In Tables 1 and 2, we report the accuracies deliver by alternative predictors with protein chains longer than 100 residues and with those shorter than 100 residues, respectively. One interesting observation is that most predictors delivered higher prediction accuracy with the long protein chains than with the short ones. Furthermore, Prote2S delivered the highest prediction accuracy with the long protein chains in comparison with the other predictors. If we use the rule-of-thumb proposed in [11], then the Q3 score delivered by Prote2S with long protein chains is significantly higher than those delivered by the other predictors. On the other hand, though Prote2S still leads in terms of the SOV score with long protein chains, the difference is not significant.Table 1Prediction accuracies delivered by alternative predictors with the 27 protein chains longer than 100 residues extracted from the EVA server.Q3Q3H_OQ3H_PQ3E_OQ3E_PQ3C_OQ3C_PSOVSOVHSOVESOVCProte2S80.3%76.4%78.3%60.5%75.8%84.1%76.3%76.9%77.7%64.9%75.2%Errsig2.0%3.8%3.4%9.3%7.8%2.0%2.4%2.2%3.2%9.4%2.4%PSIPRED78.2%78.0%76.4%60.6%67.3%77.0%75.3%75.0%76.2%62.7%72.0%Errsig1.2%4.1%3.8%9.0%9.4%1.8%1.9%1.4%3.7%9.0%1.8%PROFsec77.9%71.6%81.6%61.0%63.4%80.2%72.7%76.1%75.4%64.1%73.0%Errsig1.2%3.7%3.8%9.2%9.2%2.0%1.6%1.4%3.8%9.2%1.9%PHDpsi75.2%76.4%77.3%55.5%61.9%74.1%72.5%72.5%75.6%56.3%70.1%Errsig1.3%3.5%3.7%8.8%9.3%2.6%2.1%1.7%3.4%8.9%2.4%SABLE277.0%74.0%79.3%55.2%75.0%80.2%71.4%72.6%74.5%59.9%70.1%Errsig1.3%3.5%3.1%8.9%4.8%2.4%1.7%2.0%3.1%9.1%2.6%PROF_king70.7%56.6%72.7%55.8%57.8%77.6%67.1%67.5%60.9%58.6%68.2%Errsig1.5%4.6%7.8%9.1%7.2%1.8%2.1%1.6%4.6%9.1%2.2%Errsig is the significant difference margin for each score and is defined as the standard deviation over the square root of the number of proteins. Q3H/E/C and SOVH/E/C values are the specific Q3 and SOV scores of the predicted helix, strand and coil regions, respectively. Q3H_O (Q3E_O and Q3C_O, respectively) represents correctly predicted helix (strand and coil, respectively) residues (percentage of helix observed), and Q3H_P (Q 3E_P and Q3C_P, respectively) represents correctly predicted helix (strand and coil, respectively) residues (percentage of helix predicted).Table 2Prediction accuracies delivered by alternative predictors with the 62 protein chains shorter than 100 residues extracted from the EVA server.Q3Q3H_OQ3H_PQ3E_OQ3E_PQ3C_OQ3C_PSOVSOVHSOVESOVCProte2S75.1%73.1%79.4%69.7%73.7%85.3%70.6%69.4%74.7%71.8%72.4%Errsig1.5%3.5%3.6%4.4%4.7%1.6%2.2%2.5%3.5%4.3%2.1%PSIPRED77.0%78.4%80.3%69.8%76.9%77.5%77.7%73.2%75.4%72.1%72.6%Errsig1.6%3.9%3.2%4.3%3.9%1.8%2.0%2.2%3.9%4.3%2.2%PROFsec76.4%78.0%82.4%75.8%69.7%79.6%74.0%72.9%79.7%77.7%71.0%Errsig1.5%3.1%3.2%3.5%4.4%1.6%1.9%2.2%3.1%3.5%2.3%PHDpsi75.6%82.7%76.1%70.4%67.5%75.4%77.2%70.2%79.4%72.0%69.1%Errsig1.7%3.1%3.6%4.1%4.7%1.9%1.9%2.4%3.3%4.1%2.5%SABLE276.3%76.1%76.4%71.3%61.2%80.7%74.8%71.5%77.1%72.1%71.0%Errsig1.6%3.6%4.0%4.1%5.0%1.4%2.0%2.3%3.7%4.2%2.2%PROF_king72.5%67.4%83.5%72.6%66.6%79.9%70.1%65.8%67.2%72.8%68.5%Errsig1.7%4.1%3.3%4.2%4.7%1.6%2.3%2.5%4.2%4.4%2.4%Table 3Prediction accuracies delivered by alternative predictors with the 89 benchmark protein chains extracted from the EVA server.Q3Q3H_OQ3H_PQ3E_OQ3E_PQ3C_OQ3C_PSOVSOVHSOVESOVCProte2S76.7%74.1%79.1%71.4%76.6%84.9%72.3%71.7%75.6%74.2%73.3%Errsig1.3%2.7%2.7%3.2%3.5%1.3%1.7%1.9%2.6%3.2%1.6%PSIPRED77.4%78.3%79.1%71.5%78.5%77.3%77.0%73.7%75.7%73.8%72.4%Errsig1.2%3.0%2.5%3.1%2.9%1.4%1.5%1.6%2.9%3.1%1.6%PROFsec76.9%76.0%82.1%75.8%72.3%79.7%73.6%73.9%78.4%78.0%71.6%Errsig1.1%2.5%2.5%2.6%3.2%1.3%1.4%1.6%2.5%2.6%1.7%PHDpsi75.5%80.8%76.5%70.4%70.3%75.0%75.8%70.9%78.2%71.7%69.4%Errsig1.3%2.4%2.7%3.0%3.4%1.5%1.5%1.7%2.5%3.0%1.9%SABLE276.5%75.5%77.3%70.9%65.4%80.6%73.7%71.8%76.3%72.9%70.7%Errsig1.2%2.7%2.9%3.0%3.8%1.2%1.5%1.7%2.7%3.1%1.7%PROF_king72.0%64.1%82.5%72.0%66.2%79.2%69.1%66.3%65.3%73.0%68.4%Errsig1.2%3.2%2.6%3.1%3.5%1.2%1.7%1.8%3.3%3.2%1.8%Though the prediction accuracy delivered by Prote2S with long protein chains is superior, Prote2S did not perform as well with short protein chains. In fact, the prediction accuracy delivered by Prote2S with short protein chains is inferior to most predictors listed in Table 2. Accordingly, we can conclude that alternative machine learning algorithms offer different advantages and suffer some limitations. Therefore, it may be desirable to design a hybrid predictor that exploits the respective advantages of alternative predictors. For example, we may implement a hybrid predictor that invokes Prote2S when dealing with a long protein chain and invokes PSIPRED otherwise.As mentioned earlier, one of the major distinctive feature of the RVKDE-based predictor is that the average time taken to construct a predictor is in the order of O(nlogn), where n is the number of training instances. Therefore, it is conceivable that Prote2S can effectively cope with the high growth rate of the PDB and deliver ever-increasing prediction accuracy. In this respect, the experiment reported in Table 4 has been conducted to evaluate the related effects. In this experiment, we provided Prote2S and the LIBSVM package [17] with randomly generated subsets of the training dataset and testing was conducted with the 27 long protein chains in the testing dataset. The Gaussian kernel was adopted in LIBSVM and the two related parameters were set as C = 2 and γ = 0.01 based on the model selection process employed in [18]. The execution times shown in Table 4 were measured on a workstation equipped with an Intel Xeon 3.2GHz CPU and 8-GByte memory and do not include the time taken to carry out model selection or cross validation.Table 4Size of the training dataset vs. execution times taken by the Prote2S and the SVM during the training process.Prote2SSVMNumber of protein chains used to generate the training datasetTraining time (in seconds)Q3SOVTraining time (in seconds)Q3SOV5029.664.0%52.9%138.0871.3%64.3%10091.769.0%64.1%527.0274.0%68.3%250486.471.4%67.2%5105.6375.5%71.0%5001377.471.9%67.9%21040.076.8%72.3%10003887.873.9%71.1%78795.2577.4%73.3%The first observation about the experimental results presented in Table 4 is that the training time with the LIBSVM increases approximately in the order of O(n2). On the other hand, the training time with the Prote2S increases approximately in the order of O(nlogn). Accordingly, it is conceivable that simply employing the SVM might be impractical for some bioinformatics applications, in which the database involved is already large and still growing fast. Another observation with Table 4 is that LIBSVM generally delivered higher prediction accuracy than Prote2S but the difference diminishes as the size of the training dataset increases. This observation is consistent with that reported by the research team led by D.T. Jones [6]. According to their study, the SVM can deliver higher prediction accuracy than a neural network when the training dataset is small and the difference diminishes as the size of the training dataset increases.Our proposition concerning the inferior accuracies delivered by Prote2S in Table 4 is that it results from the asymptotic approach employed to establish the mathematical foundation of kernel density estimation [12,19]. Since the asymptotic approach assumes that the number of training instances approaches infinity, under circumstances in which the size of the training dataset is not sufficiently large, the mathematical model of a kernel density estimator may become inaccurate and the kernel density estimation based predictor may deliver inferior accuracy. Nevertheless, as the size of the training dataset increases, this effect should diminish.Another aspect of the execution time with a predictor is the time taken to make a prediction. In this respect, it has been shown in our recent article that the average time taken by the RVKDE-based predictor to make predictions with n' incoming objects is in the order of O(n' log n) [12]. Table 5 shows how the execution times taken by Prote2S and LIBSVM to make predictions increase with the size of the training dataset. The results show that the execution time taken by Prote2S increases slower than that taken by the SVM, which grows approximately in the same order as the size of the training dataset. In this experiment, we provided Prote2S and the LIBSVM package [17] with randomly generated subsets of the training dataset and testing was conducted with the 27 long protein chains in the testing dataset.Table 5Size of the training dataset vs. execution times taken by Prote2S and the SVM for making predictions.Prote2SSVMNumber of protein chains used to generate the training datasetTesting time (in seconds)Testing time (in seconds)5054.5146.710087.6301.0250153.3758.5500220.7990.71000333.22532.8The RVKDE based predictorAs mentioned above, the design of Prote2S is based on a novel kernel density estimation algorithm. The mathematical fundamentals of the so-called RVKDE can be found in our recent publication [12]. A kernel density estimator is in fact an approximate probability density function. Let {s1, s2..., sn} be a set of sampling instances randomly and independently taken from the distribution governed by fX in the m-dimensional vector space. Then, with the RVKDE algorithm, the value of fX at point v is estimated as follows:fˆ(v)=1|n|∑si(12π⋅σi)mexp⁡(−||v−si||22σi2), where1) σi=βR(si)π(k+1)Γ(m2+1)m;2) R(si) is the maximum distance between si and its k nearest training instances;3) Γ (·) is the Gamma function [20];4) β and k are parameters to be set either through cross validation or by the user.For prediction of protein secondary structures, one kernel density estimator is created to approximate the distribution of each class of training instances. As mentioned earlier, in our experiment, each residue is associated with a PSSM computed with the PSI-BLAST software package, and is labeled as one of the three types of secondary structure elements: alpha-helix, beta-strand, or coil, as determined by DSSP. Then, a query instance located at v is predicted to belong to the class that gives the maximum value with the likelihood function defined as follows:Lj(v)=|Sj|⋅fˆj(v)∑h|Sh|⋅fˆh(v),where |Sj| is the number of class-j training instances, and fˆj(v) is the kernel density estimator corresponding to class-j training instances. In our current implementation, in order to improve the efficiency of the predictor, we include only a limited number, denoted by k', of the nearest class-j training instances of v while computing fˆj(v).With the predictions made by the RVKDE based algorithm for the query protein chain, Prote2S carries out a smoothing process as the last step before outputting the results. The smoothing process includes two phases. In the first phase, each single-residue segment of secondary structures with its two neighboring residues belonging to the same secondary structure is examined to determine whether switching the prediction of the single-residue segment to the same secondary structure as its neighbors can form a new segment containing 4 or more residues. If yes, then the switching is carried out. Otherwise, nothing will happen. In the second phase, all the remaining single-residue segments of secondary structures except those predicted to be a coil are located and the prediction of each segment is switched to the secondary structure of its longer neighboring segment.Parameter tuningIn the experiments reported in this article, the 4 parameters in the RVKDE algorithm were set as m = 1, β = 3, k = 38, and k' = 60, through a validation process. The validation dataset was derived from the 1903 protein chains deposited into the PDB between June 1 and August 31 in 2007. In order to remove redundancy, BLAST was invoked to guarantee that the BLAST-computed e-value similarity score between any two protein chains in the validation dataset is larger than 0.1. Furthermore, we removed those protein chains that are homologous to one or more of the protein chains used to generate the training dataset with a BLAST-computed sequence identity larger than 25%. As a result, a total of 302 protein chains remained. Among these 302 protein chains, we then included those 45 chains that are longer than 100 residues to generate the validation dataset.Competing interestsThe authors declare that they have no competing interests.Authors' contributionsYJO proposed the RVKDE algorithm and conceived the study. DTHC and YYO implemented the Prote2S predictor. HGH, MHY, and CYC designed the experiments reported in this article. All authors read and approved the final manuscript.\n\nREFERENCES:\nNo References"
+}

batch_5/PMC2527575.json

@@ -0,0 +1,4 @@
+{
+    "id": "PMC2527575",
+    "text": "This is an academic paper. This paper has corpus identifier PMC2527575\nAUTHORS: Liam M McCormick, Martin Goddard, Ravi Mahadeva\n\nABSTRACT:\nIntroductionPulmonary siderosis secondary to the inhalation of iron compounds is a rare condition which, despite striking radiological and histopathological features, has not traditionally been associated with symptoms or functional impairment. Although not the first of its kind, we present an unusual case of pulmonary siderosis with symptomatic respiratory disease, most likely secondary to associated fibrosis.Case presentationA 66-year-old Caucasian man was referred to the outpatient clinic with a 2-year history of exertional breathlessness. He had worked as an engineer for 20 years where he did a significant amount of welding but always wore a face shield. Clinical, radiological and histological features were consistent with a diagnosis of pulmonary siderosis, with associated fibrosis, most likely related to his occupational welding history.ConclusionOur report illustrates that symptomatic respiratory disease due to mild peribronchiolar fibrosis can occur with pulmonary siderosis despite wearing a mask. Furthermore, it reinforces the need for all clinicians to compile a detailed occupational history in individuals presenting with breathlessness.\n\nBODY:\nIntroductionPulmonary siderosis secondary to the inhalation of iron compounds is a rare condition which was first described in 1936 [1]. Despite striking radiological and histopathological features, it has traditionally been classified as a 'benign pneumoconiosis' [2] because of the absence of associated symptoms, functional impairment or pulmonary fibrosis [3]. Uncommonly, however, symptomatic disease with interstitial fibrosis has been described in arc welders [4]. We present an unusual case of pulmonary siderosis with symptomatic respiratory disease, most likely secondary to associated fibrosis.Case presentationA 66-year-old Caucasian man was referred to the outpatient clinic with a 2-year history of exertional breathlessness. He had no other respiratory symptoms, had never smoked and was not aware of any previous asbestos exposure. He was not on any medication and had no allergies. He had worked as an engineer for 20 years where he did a significant amount of welding but always wore a face shield. A review of systems was unremarkable.On examination, he was not clubbed or cyanosed, and his chest was clear to auscultation. Pulmonary function tests showed a moderately severe obstructive defect, gas trapping and a significantly reduced gas transfer factor: forced expiratory volume in 1 second (FEV1) 1.58 (49.1%); vital capacity (VC) maximum 3.0 (75.9%); FEV1/VC maximum 52.6%; total lung capacity (TLC) 7.2 (100%); residual volume (RV) 4.3 (157%); RV/TLC ratio 140.3%; carbon monoxide transfer factor 3.57 (44.0%); carbon monoxide transfer coefficient 0.77 (66.8%). His resting oxygen saturations were 95% on room air; however, he desaturated to 89% after 4 minutes of walking, which was associated with a peak modified Borg score (perceived breathlessness score) of three, indicating moderate breathlessness. A chest radiograph showed diffuse generalised reticular nodular shadowing with a suggestion of enlarged hila (Figure 1a). Computed tomography scanning revealed multiple small nodular opacities throughout both lungs, predominantly in the mid and upper zones (Figure 1b). Transbronchial biopsies were non-diagnostic, therefore video-assisted thoracoscopic lung biopsy was performed. Microscopic examination of these specimens showed marked deposition of coarse iron granules in a centrilobular distribution, with foci of associated fibrosis (Figures 2 and 3). The appearances were consistent with pulmonary siderosis most likely related to his occupational welding history. In 3 years of follow-up his lung function and chest radiograph have not progressed.Figure 1Chest radiograph and computed tomography. (a) Chest radiograph demonstrating diffuse generalised reticular nodular shadowing. (b) Chest computed tomography scan showing bilateral tiny nodular opacities throughout both lung fields predominantly in the mid and upper zones.Figure 2Histological Analysis. (A) Low-power (×100) micrograph showing pigment accumulation in an interstitial peribronchovascular distribution. (B) Higher-power (×200) view showing pigment in the interstitium around the airway; the alveolar air spaces are empty. (C) High-power (×400) view showing the 'golden granules' of haemosiderin.Figure 3Histology – Iron and Collagen Stains. (A) Perl Prussian blue stain (×200) confirming deposition of iron. (B) High-power elastic van Gieson collagen stain (red) demonstrating significant fibrosis.DiscussionInhalation of iron compounds occurs commonly in paint factories, during welding and steelmaking, and at various stages of iron mining and iron refining. Doig and McLaughlin first described 'welders' siderosis' in 1936 when they carried out a prospective study examining the clinical and chest radiological characteristics of 16 electric arc welders [1]. All but one of these original subjects were followed for 9 years: four of these demonstrated progressive radiographic reticular changes, nine showed no radiographic changes, and in two men (both of whom had spent significantly less time welding), there was evidence of at least partial resolution of the initial radiographic opacities [3]. All subjects, however, remained in good health, leading to the conclusion that siderosis (in its pure form) was not associated with respiratory symptoms or functional impairment. This view was supported by subsequent pathological investigations of the lungs of subjects occupationally exposed to iron oxide fumes, which did not demonstrate any evidence of pulmonary fibrosis [3]. As a result, the apparently inert nature of iron compounds led to the classification of pulmonary siderosis as a 'benign pneumoconiosis' [2].Nevertheless, symptomatic disease with interstitial fibrosis has been described in arc welders [4]. Some authors have postulated that these rare cases are secondary to concomitant inhalation of silicates or asbestos that can occur in many occupations associated with exposure to iron [4]. However, Funahashi et al. challenged this view after investigating 10 symptomatic welders and performing energy dispersive X-ray analysis on lung tissue for elemental content [5]. Despite demonstrating restrictive defects in seven of their patients, and mild to moderate airway obstruction in a further two, they found no difference between the pulmonary silicon content of patients with symptomatic 'welders' pneumoconiosis' and that of age-matched control lungs [5]. Some degree of parenchymal fibrosis was present in all patients, and in 50%, this fibrosis was considered moderate to marked. As many of the iron-containing particles were seen in fibrotic alveolar septa, it was postulated that the fibrosis was a reaction to these particles rather than the co-existing silicosis [5].ConclusionOur report illustrates that symptomatic respiratory disease due to mild peribronchiolar fibrosis can occur with pulmonary siderosis despite wearing a mask. Furthermore, it reinforces the need to compile a detailed occupational history in individuals with respiratory disease. It is particularly important to obtain an accurate diagnosis of interstitial shadowing on chest radiograph as pulmonary siderosis in our patient had a relatively good prognosis compared with other interstitial and small airway disorders.AbbreviationsFEV1: forced expiratory volume in 1 second; RV: residual volume; TLC: total lung capacity; VC: vital capacity.Competing interestsThe authors declare that they have no competing interests.Authors' contributionsLM participated in the design and coordination of the case report and drafted the manuscript. MG acquired, analysed and reported on the histopathological slides. RM conceived of the case report, participated in its design and coordination, and revised it critically for important intellectual content. All authors read and approved the final manuscript.ConsentWritten informed consent was obtained from the patient for publication of this case report and accompanying images. A copy of the written consent is available for review by the Editor-in-Chief of this journal.\n\nREFERENCES:\nNo References"
+}

batch_5/PMC2527613.json

@@ -0,0 +1,4 @@
+{
+    "id": "PMC2527613",
+    "text": "This is an academic paper. This paper has corpus identifier PMC2527613\nAUTHORS: Digant Gupta, Carolyn A Lammersfeld, Pankaj G Vashi, Jessica King, Sadie L Dahlk, James F Grutsch, Christopher G Lis\n\nABSTRACT:\nBackgroundBioelectrical impedance analysis (BIA) is an easy-to-use, non-invasive, and reproducible technique to evaluate changes in body composition and nutritional status. Phase angle, determined by bioelectrical impedance analysis (BIA), detects changes in tissue electrical properties and has been hypothesized to be a marker of malnutrition. Since malnutrition can be found in patients with breast cancer, we investigated the prognostic role of phase angle in breast cancer.MethodsWe evaluated a case series of 259 histologically confirmed breast cancer patients treated at Cancer Treatment Centers of America. Kaplan Meier method was used to calculate survival. Cox proportional hazard models were constructed to evaluate the prognostic effect of phase angle independent of stage at diagnosis and prior treatment history. Survival was calculated as the time interval between the date of first patient visit to the hospital and the date of death from any cause or date of last contact/last known to be alive.ResultsOf 259 patients, 81 were newly diagnosed at our hospital while 178 had received prior treatment elsewhere. 56 had stage I disease at diagnosis, 110 had stage II, 46 had stage III and 34 had stage IV. The median age at diagnosis was 49 years (range 25 – 74 years). The median phase angle score was 5.6 (range = 1.5 – 8.9). Patients with phase angle <= 5.6 had a median survival of 23.1 months (95% CI: 14.2 to 31.9; n = 129), while those > 5.6 had 49.9 months (95% CI: 35.6 to 77.8; n = 130); the difference being statistically significant (p = 0.031). Multivariate Cox modeling, after adjusting for stage at diagnosis and prior treatment history found that every one unit increase in phase angle score was associated with a relative risk of 0.82 (95% CI: 0.68 to 0.99, P = 0.041). Stage at diagnosis (p = 0.006) and prior treatment history (p = 0.001) were also predictive of survival independent of each other and phase angle.ConclusionThis study demonstrates that BIA-derived phase angle is an independent prognostic indicator in patients with breast cancer. Nutritional interventions targeted at improving phase angle could potentially lead to an improved survival in patients with breast cancer.\n\nBODY:\nBackgroundIn the United States, breast cancer is the most common non-skin cancer and the second leading cause of cancer-related death in women [1].Malnutrition is a frequent manifestation in patients with advanced cancer and is a major contributor to morbidity and mortality [2]. Malnutrition is characterized by changes in cellular membrane integrity and alterations in fluid balance [3]. As a result, measurement of body composition is an important component of overall nutritional evaluation in cancer patients [4-6].Several studies have investigated the relationship between diet, physical activity, obesity and survival in breast cancer [7-11]. A prospective study was performed on 1,490 women diagnosed and treated for early-stage breast cancer between 1991 and 2000. In univariate analysis, reduced mortality was weakly associated with higher vegetable-fruit consumption, increased physical activity, and a body mass index that was neither low weight nor obese [7]. Another study investigated the influence of diet, including dietary fat (percentage energy), fiber, vegetable, and fruit intakes, and micronutrients (folate, carotenoids, and vitamin C) on overall survival in 516 postmenopausal women diagnosed with breast cancer. In the multivariate analysis, the hazard ratio of dying in the highest tertile compared to the lowest tertile of total fat, fiber, vegetable, and fruit was 3.12 (95% CI = 1.79–5.44), 0.48 (95% CI = 0.27–0.86), 0.57 (95% CI = 0.35–0.94), and 0.63 (95% CI = 0.38–1.05), respectively (P <or= 0.05 for trend, except for fruit intake) [9]. A review article summarized the evidence from clinical and epidemiologic studies that have examined the relationship between nutritional factors, survival, and recurrence after the diagnosis of breast cancer. The article reported that overweight or obesity was associated with poorer prognosis in the majority of the studies that have examined this relationship. Treatment-related weight gain also may influence disease-free survival, reduce quality of life, and increase risk for comorbid conditions [11]. Clearly, much remains to be learned about the role of nutritional factors in survival after the diagnosis of breast cancer, especially with the advent of novel techniques to assess nutritional status.Historically, nutritional status has been evaluated by various objective measures, including anthropometric (e.g. weight change, arm muscle circumference, triceps skinfold thickness) and laboratory (serum albumin, transferrin assays and nitrogen balance studies) measurements. Anthropometric methods are not ideal in a clinical setting because they are time-consuming and require well-trained staff. Some of the objective measures such as serum albumin are likely to be influenced by many non-nutritional factors [12-15]. Furthermore, some objective indicators such as serum albumin have long half-lives, thus, assessing changes in the nutritional status over a short period of time is challenging. A less common tool to assess body composition and nutritional status, called Bioelectrical Impedance Analysis (BIA), can overcome some of these challenges. BIA is an easy-to-use, non-invasive, and reproducible technique to evaluate changes in body composition.BIA has been validated for the assessment of body composition and nutritional status in a variety of patient populations including cancer [2,5,16-26]. BIA measures body component resistance (R) and reactance (Xc) by recording a voltage drop in applied current [27]. Resistance is the restriction to the flow of an electric current, primarily related to the amount of water present in the tissues. Reactance is the resistive effect produced by the tissue interfaces and cell membranes [28]. Reactance causes the current to lag behind the voltage creating a phase shift, which is quantified geometrically as the angular transformation of the ratio of reactance to resistance, or the phase angle [29].Phase angle reflects the relative contributions of fluid (resistance) and cellular membranes (reactance) of the human body. By definition, phase angle is positively associated with reactance and negatively associated with resistance [29]. Lower phase angles suggest cell death or decreased cell integrity, while higher phase angles suggest large quantities of intact cell membranes [30]. Phase angle has been found to be a prognostic marker in several clinical conditions such as human immunodeficiency virus infection, liver cirrhosis, chronic obstructive pulmonary disease, hemodialysis, sepsis, lung cancer [30-35]. Previously, we had demonstrated the prognostic role of phase angle in advanced colorectal and pancreatic cancer [36,37]. The primary objective of this study, which builds upon our prior research work in this area, was to evaluate the association of BIA-derived phase angle with survival in patients with breast cancer.MethodsA retrospective chart review was performed on a consecutive case series of 259 female breast cancer patients treated at Cancer Treatment Centers of America (CTCA)® at Midwestern Regional Medical Center (MRMC) between January 2001 and May 2006. The patients were identified from the MRMC tumor registry. Only patients with a histologically confirmed diagnosis of breast cancer were included in this study. The study was approved by the Institutional Review Board at MRMC.All patients underwent a baseline nutritional assessment, which included laboratory measurements of serum albumin, prealbumin and transferrin, subjective global assessment (SGA), and BIA. BIA was performed by a registered dietitian using a Bioelectrical Impedance Analyzer, Model BIA-101Q: RJL Systems, Clinton Township, MI, USA. BIA was conducted while patients were lying supine on a bed or exam table, with legs apart and arms not touching the torso. All evaluations were conducted on the patients' right side using the four surface standard electrode (tetra polar) technique on the hand and foot [23]. Resistance (R) and reactance (Xc) were directly measured in Ohms at 50 Khz, 800 μA using RJL BIA. One assessment of resistance (R) and reactance (Xc) was made. Phase angle was calculated using the following equation: Phase Angle = (Resistance/Reactance)*(180/π).All data were analyzed using SPSS 11.5 (SPSS Inc., Chicago, IL, USA). Patient survival was defined as the time interval between date of first patient visit to the hospital and date of death from any cause or date of last contact/last known to be alive. The Kaplan-Meier or product-limit method was used to calculate survival. The log rank test statistic was used to evaluate the equality of survival distributions across different strata. A difference was considered to be statistically significant if the p value was less than or equal to 0.05. Survival was also evaluated using univariate and multivariate Cox regression analysis. Variables evaluated included phase angle, age at diagnosis, prior treatment history and stage at diagnosis. For the purpose of univariate analysis, phase angle measurements were categorized using SPSS into 2 mutually exclusive groups with median = 5.6 as the cut-off. For the purpose of multivariate analyses, phase angle was treated as a continuous variable. Similarly, stage at diagnosis variable was treated as a dichotomous variable with 2 categories – early stage (stages I and II) and late stage (stages III and IV).ResultsAt the time of this analysis (May 07), 85 patients had expired and 174 were censored, as shown in Table 1. The cut-off date for the follow-up for all participants was May 07. The median age at diagnosis was 49 years (range 25 – 74 years). The median phase angle score was 5.6 (range = 1.5 – 8.9). Phase angle was found to be non-normally distributed. Table 2 shows the univariate survival analysis of different prognostic factors. Phase angle, tumor stage and treatment history were found to be statistically significantly associated with survival.Table 1Patient characteristicsCharacteristicCategoriesNumberPercent (%)Vital StatusExpired8532.8Censored117467.2Prior TreatmentProgressive disease17868.7HistoryNewly diagnosed8131.3Stage at DiagnosisStage I5621.6Stage II11042.5Stage III4617.8Stage IV3413.1Missing135.0Age at Diagnosis25–35261036–457227.846–558833.956–656625.666–7572.71 Patients who reached the end of their follow-up without experiencing death.N = 259Table 2Univariate Kaplan-Meier survival analysisVariableMedian survival in monthsLog-rank scoreP-valuePhase Angle • <= 5.623.1 (14.2 to 31.9)4.90.031 • >5.649.9 (35.6 to 77.8)Tumor Stage • Stage I and II54.2 (35.3 to 83.2)5.40.021 • Stage III and IV22.9 (14.0 to 31.9)Treatment History • Newly diagnosed54.6 (53.8 to 55.4)50.20.0001 • Progressive disease18.7 (14.8 to 22.5)N = 259Figure 1 shows the survival curves for the two categories of the phase angle. Patients with phase angle <= 5.6 had a median survival of 23.1 months (95% CI: 14.2 to 31.9; n = 129), while those > 5.6 had 49.9 months (95% CI: 35.6 to 77.8; n = 130); the difference being statistically significant (p = 0.031).Figure 1Survival stratified by phase angle categories with cutoff of 5.6. Each drop in a probability curve indicates one or more events in that group. Vertical lines indicate censored patients, i.e., those who reached the end of their follow-up without experiencing death.Table 3 summarizes the results of multivariate Cox regression analyses. Multivariate Cox modeling, after adjusting for stage at diagnosis and prior treatment history found that every one unit increase in phase angle score was associated with a relative risk of 0.82 (95% CI: 0.68 to 0.99, P = 0.041). Stage at diagnosis (p = 0.006) and prior treatment history (p = 0.001) were also predictive of survival independent of each other and phase angle. Phase angle was also used as a squared term in Cox regression due to its non-linear association with mortality (Table 4).Table 3Multivariate Cox proportional hazard model (phase angle as a continuous variable)Independent VariableUnit of increaseRR195% CIP-valuePhase angle1 degree0.820.68, 0.990.041Stage at DiagnosisStage I and II as referent1.91.2, 2.90.006Treatment HistoryNewly Diagnosed as referent7.94.1, 15.50.001Table 4Multivariate Cox proportional hazard model (phase angle as a squared term)Independent VariableUnit of increaseRR195% CIP-valueSquared Phase angle1 degree square0.980.96, 1.0010.06Stage at DiagnosisStage I and II as referent1.91.2, 2.90.007Treatment HistoryNewly Diagnosed as referent7.94.1, 15.20.0011 Relative risk (Cox proportional hazard)N = 259DiscussionThe current study was undertaken to investigate if BIA-derived phase angle could predict survival in breast cancer.This study demonstrated that phase angle is a strong predictor of survival in breast cancer after controlling for the effects of stage at diagnosis and prior treatment history. A similar study conducted in patients with advanced lung cancer stratified the patient cohort by the mean phase angle score of 4.5. Interestingly, patients with phase angle scores less than or equal to 4.5 had a significantly shorter survival than those with phase angle scores greater than 4.5 [38]. In our previous study in stage IV colorectal cancer patients, we found that phase angle above the median cut-off of 5.6 was associated with better survival [37]. Similarly, in stage IV pancreatic cancer, phase angle above the median cut-off of 5 was associated with improved survival [36].This study adds to the growing body of evidence regarding the clinical applications of BIA derived phase angle beyond its use in body composition equations. Although the biological meaning of phase angle is not well understood, it reflects not only body cell mass, but is also one of the best indicators of cell membrane function, related to the ratio between extracellular water and intracellular water [28]. Schwenk et al. has hypothesized that phase angle could possibly be interpreted as a global marker of malnutrition in HIV infected patients [35]. In another study conducted on HIV-infected patients, it was argued that phase angle reflects the integrity of vital cell membranes [33]. In patients with liver cirrhosis, phase angle was speculated to be a marker of clinically relevant malnutrition characterized by both increased extracellular mass and decreased body cellular mass [30]. In advanced lung cancer, phase angle was speculated to be an indicator of altered tissue electrical properties [38]. In spite of lack of standardized cut-off values, phase angle seems to play an important role as a marker of morbidity and mortality in a wide range of disease conditions, with higher phase angle reflecting a general indicator of wellness [28].Limitations of this study relate to the BIA technique and retrospective study design. This study, because of its retrospective nature, relies on data not primarily meant for research. One potential limitation of the BIA approach for estimating body composition is the reliance on regression models, derived in restricted samples of human subjects, which limits the usefulness of the derived model in other patients who differ from the original sample in which the model was developed [39,40]. However, in our study, we looked at phase angle which does not depend on regression equations to be calculated, thereby eliminating a large source of random error [3]. It has also been suggested that the variability of direct bioimpedance measures (resistance, reactance, and phase angle) depends on age, gender, and body mass characteristics of the study population which could possibly limit the extrapolation of the model [28,39,41]. A review article by Foster et al. argued that although the correlation between whole-body impedance measurements and body composition is experimentally well established, the reason for the success of the impedance technique is much less clear [42].Some other reported limitations of using BIA for assessment of body composition are hydration status and/or major disturbances of water distribution, body position during procedure, ambient air and skin temperatures, recent physical activity, conductance of the examining table, and food consumption [43]. Since the original intent of the BIA in this study was to gather estimates of body composition as part of a baseline nutritional assessment in a clinical setting, not all of these factors could realistically be controlled. Patients were free of visible edema or ascites so there was control for obvious overhydration. Body position was controlled for because all patients were in the supine position in a bed or on an exam table. Air temperature was within a controlled range in our hospital setting. Physical activity was limited in these patients due to the advanced nature of their disease. Finally, food intake was not controlled for in this clinical setting, which may have contributed to a small amount of variability.The cut-off point for phase angle in the present study was generated so as to divide the patient population into 2 equal and mutually exclusive groups. The cut-off value of phase angle in out study might differ from those in other patient populations. Although our cut-off point was in agreement with those reported by other studies [30,35,38], there is a clear need to define threshold values for phase angle as a nutritional assessment tool using Receiver Operating Characteristic analysis based on large prospective studies in advanced cancer. We also think that restricting the analysis to newly diagnosed patients (patients with no prior treatment history) would have been more accurate, since it would have allowed for evaluation of true overall survival time, i.e. time from the date of diagnosis to the date of death. However, doing so would have caused a significant reduction in the sample size. In our study, the survival time was calculated from the day of first visit at our hospital because the BIA measurements were not available at the time of diagnosis for previously treated patients. This limitation emphasizes the need for conducting prospective studies, which have nutritional information available since the date of diagnosis. No assessment of inter-rater reliability of the users of BIA was made in this study. This bias, however, was minimized by restricting the use of BIA to well-trained dietitians with an expertise in the use of this clinical technique.ConclusionIn summary, our study has demonstrated the prognostic significance of phase angle in breast cancer after controlling for the effects of stage at diagnosis and prior treatment history.Competing interestsThe authors declare that they have no competing interests.Authors' contributionsDG was the main author of the manuscript, participated in concept, design, data collection, data analysis and data interpretation. CAL, JK, and SLD participated in concept, design, data collection and writing. PGV participated in concept, design and data interpretation. JFG assisted with the statistical analysis and data interpretation. CGL participated in concept, design, writing and data interpretation. All authors read and approved the final manuscript.Pre-publication historyThe pre-publication history for this paper can be accessed here:\n\nREFERENCES:\nNo References"
+}

batch_5/PMC2527656.json

@@ -0,0 +1,4 @@
+{
+    "id": "PMC2527656",
+    "text": "This is an academic paper. This paper has corpus identifier PMC2527656\nAUTHORS: Connie J. Mulligan, Andrew Kitchen, Michael M. Miyamoto\n\nABSTRACT:\nBackgroundWe re-assess support for our three stage model for the peopling of the Americas in light of a recent report that identified nine non-Native American mitochondrial genome sequences that should not have been included in our initial analysis. Removal of these sequences results in the elimination of an early (i.e. ∼40,000 years ago) expansion signal we had proposed for the proto-Amerind population.Methodology/FindingsBayesian skyline plot analysis of a new dataset of Native American mitochondrial coding genomes confirms the absence of an early expansion signal for the proto-Amerind population and allows us to reduce the variation around our estimate of the New World founder population size. In addition, genetic variants that define New World founder haplogroups are used to estimate the amount of time required between divergence of proto-Amerinds from the Asian gene pool and expansion into the New World.Conclusions/SignificanceThe period of population isolation required for the generation of New World mitochondrial founder haplogroup-defining genetic variants makes the existence of three stages of colonization a logical conclusion. Thus, our three stage model remains an important and useful working hypothesis for researchers interested in the peopling of the Americas and the processes of colonization.\n\nBODY:\nIntroductionWe recently published a three stage model for the peopling of the Americas [1]. Specifically, we proposed that a recent, rapid expansion into the Americas was preceded by a long period of population stability in greater Beringia by the proto-Amerind population after divergence from their ancestral Asian population. We used two complementary coalescent methods, Bayesian skyline plot [2] and isolation-by-migration [3] analyses, to estimate past population growth patterns in Native American populations and to estimate a New World founder effective population size. We explicitly incorporated archaeological, geological, and paleoecological constraints into our analyses to enhance the anthropological relevance of the results and to provide a comprehensive model for the initial settlement of the Americas.Fagundes et al. [4] have published a re-analysis of the data we used in developing our three stage model for the peopling of the Americas [1]. Specifically, they identified nine mitochondrial coding region sequences that we assumed were Native American sequences, but instead are likely to derive from Asian or European individuals. Fagundes et al. are correct in this assessment, i.e. five sequences were reclassified as Asian after their publication as Native American sequences [5] and four sequences were mistakenly included in our original study. The effect of removing these sequences from the Bayesian skyline plot analysis is that the suggestion of an early expansion event in the skyline plot is no longer apparent, a finding that we have reconfirmed by re-running our original dataset without these nine genomes. It appears that the non-Native American sequences introduced additional variation that created an expansion signal that does not exist in an analysis of only Native American sequences.In light of these facts, we have now analyzed the largest dataset of Native American mitochondrial coding genomes using publicly available sequences (n = 148; [6]) in a Bayesian skyline plot analysis. We also provide an estimate for the duration of the period of population isolation required for the generation of New World founder haplogroup-defining variants. As in our previous analysis, we evaluate the significance of our results in concert with other non-genetic data.ResultsWe use a Bayesian skyline plot to visually illustrate changes in Native American female effective population size (Ne) over time. Bayesian skyline plots assume a single migration event, which makes the approach ideal for questions concerning the peopling of the Americas since it is generally agreed that there was a single migration [7]. Our new skyline plot (Fig. 1) strongly supports a large population expansion (∼1.8 orders of magnitude, or 80-fold) that occurred ∼16–12 thousand years ago (kya). This timing suggests an entry to the New World that was coincident with the retreat of the North American ice sheets, i.e. the opening of an ‘ice-free corridor’ ∼17–14 kya [8], [9]. Immediately before this expansion, there is a small drop in effective population size ∼17–16 kya (this is an insignificant change, as judged by the overlap in 95% confidence intervals at the beginning and end of the population decrease), possibly corresponding to a population bottleneck prior to entry to the Americas. Before 17 kya, the skyline plot is flat with no evidence of the early (∼40 kya) population expansion we reported previously [1]. The absence of an early expansion signal in the skyline plot may simply indicate that divergence of proto-Amerinds from the Asian gene pool was not accompanied by significant population growth. These results are highly consistent with our earlier analysis of only 20 Native American mitochondrial coding genomes [10], in support of theoretical expectations by Felsenstein [11] that increasing sample size is an inefficient way to improve the accuracy of maximum likelihood estimations from coalescent analyses of population genetic data.10.1371/journal.pone.0003199.g001Figure 1Bayesian skyline plot of 148 Native American mitochondrial coding genome sequences.The curve plots median Nef with 95% credible intervals indicated by light gray lines. The shaded gray box highlights the significant increase of Nef during the colonization of the Americas 16–12 kya. The blue box depicts the calculated time required for the generation of New World defining mitochondrial variants and its shaded region represents the variation in these estimates, i.e. 7–15 thousand years before entry to the New World (see Table 1). The green arrow identifies the date of the Yana River site of human occupation in western Beringia [21].Our new analysis (with non-Native American sequences eliminated and more Native American sequences added) shows a larger population increase (80-fold vs 16-fold) over a smaller period of time (16–12 kya vs 16–9 kya) relative to our previous analysis that inadvertently included non-Native American sequences [1]. The non-Native American sequences likely introduced additional variation that artificially increased Ne prior to the expansion. Thus, we can estimate a new Ne for the New World founding population of 1,800 (this number is multiplied by two since the skyline plot only estimates the female effective population size). This number is closer to our previous isolation-with-migration (IM)-based estimate of 1,200 [1] and thus reduces the variation around our estimate of the size of the founding population to ∼1,000–2,000 effective individuals.Prior to entry to the New World, we propose a period of isolation. A valid question remains - How long was the period of isolation? In the absence of a biphasic skyline plot, we can calculate first approximations of the time necessary to generate the defining variants for the New World mithochondrial founding haplogroups. All New World mitochondrial sequences cluster in five monophyletic clades, representing founding haplogroups that are differentiated from non-New World haplogroups by the presence of specific, defining genetic variants. The variants that occur on the branch leading to each New World founding haplogroup represent variation that evolved prior to expansion into the Americas whereas variation within each founding haplogroup, i.e. nucleotide diversity within a haplogroup, represents variation that evolved after entry to the Americas – we are interested in the variation that occurred prior to entry into the Americas. There is strong consensus on the number of New World founding haplogroup-defining variants, including both coding and non-coding hypervariable regions I and II (HVRI+II) variants [5], [12]. However, there is a wide range of substitution rates that have been estimated for both coding and non-coding variants [13]–[17]. Fagundes et al. [4], [18] tend to favor the slower substitution rates whereas we generally favor the faster substitution rates, particularly for coding variants since a faster rate (∼1.7×10−8 substitutions/site/year) has been confirmed using two independent approaches [13], [16]. However, to be complete since there is ongoing debate about the correct calculation of substitution rates most recently [19], [20], we present a series of estimates based on coding and HVRI+II variants using both fast and slow substitution rates (Table 1). As is evident from the calculations, there is a wide range of estimates for the time necessary to generate the New World defining variants, i.e. averages range from ∼6,000 to ∼25,000 years. By averaging across coding and non-coding variants and including fast and slow substitution rates, we report a range of ∼7–15 thousand years. This estimate suggests that Amerind ancestors may have experienced a period of isolation lasting at least 7–15 thousand years prior to their expansion into the Americas (see the blue box in Fig. 1).10.1371/journal.pone.0003199.t001Table 1Estimates of time necessary to generate the mitochondrial genome variants that define New World founding haplogroups.Founding haplogroups based on coding variants a\n# defining variants b\nTime necessary to generate haplogroup defining coding variants using a fast substitution rate (years) c\nTime necessary to generate haplogroup defining coding variants using a slow substitution rate (years) c\nH'grp A227,61610,276H'grp B2519,04025,690H'grp C1b13,8085,138H'grp C1c27,61610,276H'grp C1d13,8085,138H'grp D113,8085,138H'grp X2a311,42415,414Average (coding)8,16011,010Founding haplogroups based on HVRI+II variants a\nTime necessary to generate haplogroup defining HVRI+II variants using a fast substitution rate (years) c\nTime necessary to generate haplogroup defining HVRI+II variants using a slow substitution rate (years) c\nH'grp A239,06637,053H'grp D113,02212,053H'grp X2a26,04424,702Average (HVRI+II)6,04424,702Average (coding and HVRI+II)7,52515,118aThe total number of defining variants for a single founding haplogroup (H'grp) was used in each calculation. Haplogroups B2 and C1b–d do not have defining HVRI or HVRII variants and were therefore not used in the HVRI+II calculations. Averages were calculated for coding and HVRI+II variants separately as well as an average of the total number of estimates within each substitution rate.bThe number of defining variants for New World founding haplogroups was determined by Bandelt et al. [5] and Tamm et al. [12].cSubstitution rates were as follows: Coding/Fast = 1.7×10−8 substitutions/site/year→3,808 years/mutation [13], [16]; Coding/Slow = 1.26×10−8 substitutions/site/year→5,138 years/mutation [17]; HVRI+II/Fast = 4.7×10−7 substitutions/site/year→3,022 years/mutation [15]; HVRI+II/Slow = 1.15×10−7 substitutions/site/year→12,351 years/mutation [14].DiscussionOur proposal for a three stage model for the peopling of the Americas remains essentially unchanged despite the modifications to the skyline plot described above. The three stages remain; 1) divergence of Amerind ancestors from the Asian gene pool, 2) prolonged period of isolation, lasting at least 7–15 thousand years, during which time genetic variants specific to and present throughout the New World were generated, and 3) rapid expansion into the Americas ∼16 kya concomitant with a large population increase. The existence of mitochondrial New World-defining variants that are widespread throughout the Americas has been noted in numerous publications most recently [6], [12] and indicates that there must have been a period of isolation during which time these variants arose. The idea of a period of population isolation prior to expansion into the Americas was first mentioned by Bonatto and Salzano [14] and most recently supported by Tamm et al. [12]. Thus, divergence from the Asian gene pool and entry into the Americas were separated by this period of isolation, making the existence of three stages a logical conclusion.In our previous study, we suggested that the period of isolation occurred during occupation of greater Beringia [1]. The fact that Beringia is now inundated may explain why no archaeological evidence of human occupation has been found, although greater Beringia encompasses such a vast territory that more terrestrial archaeological sites may yet be discovered. The documentation of human occupation at the Yana River site ∼30 kya [21] provides independent support for the presence of humans in greater Beringia as early as 30,000 years ago [22] and strengthens our proposal of a Beringian occupation from ∼30–16 kya. Furthermore, multiple fossil sites document the presence of large mammals in Alaska and Siberia [23]–[25]. Fossil pollen and plant microfossils from eastern Beringia indicate a productive, dry grassland ecosystem [26] suggesting the entire range of Beringia was capable of supporting a large mammal fauna. Archaeological evidence and ethnographic analogy both suggest that Amerind ancestors in Beringia were skilled hunters who relied upon megafauna for sustenance and likely extended their hunting ranges in response to demographic changes in the large mammal population [27]. Thus, it is highly probable that humans inhabited the central part of greater Beringia, i.e. Beringia, for an extended period of time. In fact, the first published Bayesian skyline plot focused on the Beringian steppe bison (using 169 ancient DNA sequences and 22 modern sequences) and revealed a sharp population decline beginning ∼30 kya [2] leading us to suggest that Beringian populations of humans may have been associated with the decline in steppe bison.In conclusion, our three stage model remains an important and useful working hypothesis for researchers interested in the peopling of the Americas and the processes of colonization. We believe that divergence from the ancestral gene pool and expansion into a new territory were not simultaneous events, as is often assumed in models of population demographic history. Specifically, movement from Asia to the New World was interrupted by an extended period of population isolation and stability. Entry into the New World was mediated by a population of 1,000–2,000 effective individuals. The relevance of our model is due to its reliance on a synthetic approach that combines genetic data with multiple sources of anthropological and paleoenvironmental information. As a working hypothesis, our model is predictive. In particular, it predicts that key archaeological sites await discovery under the Bering Sea.Materials and MethodsA dataset of 148 human mitochondrial coding genomes was assembled from the publicly available sequences used by Achili et al.\n[6] and then aligned as described in Kitchen et al.\n[1]. Bayesian skyline plots [2] of the aligned coding genomes were used to estimate changes in Amerind Nef over time by providing highly parametric, piecewise estimates of Nef. In these analyses, estimates of τ (Nef×generation time) were converted to Nef by dividing by a generation time of 20 years, following convention [3]. Using a generation time of 25 years decreases Nef estimates by 20%, but does not affect the time estimates. Skyline plots were generated using the program BEAST v1.4 (http://beast.bio.ed.ac.uk). These BEAST analyses relied on the same coalescent and substitution models and run conditions as Kitchen et al.\n[10]. Markov chains were run for 100,000,000 generations and sampled every 2,500 generations with the first 10,000,000 generations discarded as burn-in.\n\nREFERENCES:\n1. KitchenAMiyamotoMMMulliganCJ\n2008\nA three-stage colonization model for the peopling of the Americas.\nPLoS ONE\n3\ne1596\n18270583\n2. DrummondAJRambautAShapiroBPybusOG\n2005\nBayesian coalescent inference of past population dynamics from molecular sequences.\nMol Biol Evol\n22\n1185\n1192\n15703244\n3. HeyJ\n2005\nOn the number of New World founders: a population genetic portrait of the peopling of the Americas.\nPLoS Biol\n3\ne193\n15898833\n4. FagundesNJRKanitzRBonattoSL\nIn Press\nA reevaluation of the Native American mtDNA genome diversity and its bearing on the models of early colonization of Beringia.\nPLoS ONE\n5. BandeltHJHerrnstadtCYaoYGKongQPKivisildT\n2003\nIdentification of Native American founder mtDNAs through the analysis of complete mtDNA sequences: some caveats.\nAnn Hum Genet\n67\n512\n524\n14641239\n6. AchilliAPeregoUABraviCMCobleMDKongQP\n2008\nThe phylogeny of the four pan-American mtDNA haplogroups: implications for evolutionary and disease studies.\nPLoS ONE\n3\ne1764\n18335039\n7. MulliganCJHunleyKColeSLongJC\n2004\nPopulation genetics, history, and health patterns in Native Americans.\nAnnu Rev Genomics Hum Genet\n5\n295\n315\n15485351\n8. HoffeckerJFPowersWRGoebelT\n1993\nThe colonization of Beringia and the peopling of the New World.\nScience\n259\n46\n53\n17757472\n9. MandrykCASJosenhansHFedjeDWMathewesRW\n2001\nLate Quaternary paleoenvironments of northwestern North America: implications for inland versus coastal migration routes.\nQuaternary Sci Rev\n20\n301\n314\n10. KitchenAMiyamotoMMMulliganCJ\n2008\nUtility of DNA viruses for studying human host history: case study of JC virus.\nMol Phylogenet Evol\n46\n673\n682\n17977749\n11. FelsensteinJ\n2006\nAccuracy of coalescent likelihood estimates: do we need more sites, more sequences, or more loci?\nMol Biol Evol\n23\n691\n700\n16364968\n12. TammEKivisildTReidlaMMetspaluMSmithDG\n2007\nBeringian standstill and spread of Native American founders.\nPLoS ONE\n2\ne829\n17786201\n13. AtkinsonQDGrayRDDrummondAJ\n2008\nmtDNA variation predicts population size in humans and reveals a major Southern Asian chapter in human prehistory.\nMol Biol Evol\n25\n468\n474\n18093996\n14. BonattoSLSalzanoFM\n1997\nA single and early migration for the peopling of the Americas supported by mitochondrial DNA sequence data.\nProc Natl Acad Sci USA\n94\n1866\n1871\n9050871\n15. HowellNSmejkalCBMackeyDAChinneryPFTurnbullDM\n2003\nThe pedigree rate of sequence divergence in the human mitochondrial genome: there is a difference between phylogenetic and pedigree rates.\nAm J Hum Genet\n72\n659\n670\n12571803\n16. IngmanMKaessmannHPääboSGyllenstenU\n2000\nMitochondrial genome variation and the origin of modern humans.\nNature\n408\n708\n713\n11130070\n17. MishmarDRuiz-PesiniEGolikPMacaulayVClarkAG\n2003\nNatural selection shaped regional mtDNA variation in humans.\nProc Natl Acad Sci U S A\n100\n171\n176\n12509511\n18. FagundesNJKanitzREckertRVallsACBogoMR\n2008\nMitochondrial population genomics supports a single pre-Clovis origin with a coastal route for the peopling of the Americas.\nAm J Hum Genet\n82\n583\n592\n18313026\n19. HoSYEndicottP\n2008\nThe crucial role of calibration in molecular date estimates for the peopling of the Americas.\nAm J Hum Genet\n83\n142\n146\n18606310\n20. FagundesNJKanitzRBonattoSL\n2008\nReply to Ho and Endicott.\nAm J Hum Genet\n83\n146\n147\n21. PitulkoVVNikolskyPAGiryaEYBasilyanAETumskoyVE\n2004\nThe Yana RHS site: humans in the Arctic before the last glacial maximum.\nScience\n303\n52\n56\n14704419\n22. GoebelT\n2007\nThe missing years for modern humans.\nScience\n315\n194\n196\n17218514\n23. EliasSShortSNelsonCBirksH\n1996\nLife and times of the Bering land bridge.\nNature\n382\n60\n63\n24. GuthrieR\n1990\nFrozen fauna of the mammoth steppe\nChicago\nUniversity of Chicago Press\n25. HopkinsD\n1982\nAspects of the paleogeography of Beringia during the Late Pleistocene.\nHopkinsDMatthewsJSchwegerCYoungS\nPaleocology of Beringia\nNew York\nAcademic Press\n3\n28\n26. ZazulaGDFroeseDGSchwegerCEMathewesRWBeaudoinAB\n2003\nPalaeobotany: Ice-age steppe vegetation in east Beringia.\nNature\n423\n603\n12789326\n27. WestFH\n1996\nBeringia and New World origins.\nWestFH\nAmerican Beginnings\nChicago\nUniversity of Chicago Press"
+}

batch_5/PMC2527799.json

@@ -0,0 +1,4 @@
+{
+    "id": "PMC2527799",
+    "text": "This is an academic paper. This paper has corpus identifier PMC2527799\nAUTHORS: M M Seenath, D Roberts, C Cawthorne, M P Saunders, G R Armstrong, S T O'Dwyer, I J Stratford, C Dive, A G Renehan\n\nABSTRACT:\nHypoxia inducible factor 1 (HIF-1) represses the transcription of pro-apoptotic bid in colorectal cancer cells in vitro. To assess the clinical relevance of this observation, HIF-1α and Bid were assessed in serial sections of 39 human colorectal adenocarcinomas by immunohistochemistry. In high HIF-1α nuclear-positive cell subpopulations, there was a significant reduction in Bid expression (ANOVA, P=0.04). Given the role of Bid in drug-induced apoptosis, these data add impetus to strategies targeting HIF-1 for therapeutic gain.\n\nBODY:\nApproximately 30–50% of patients diagnosed with colorectal cancer (CRC) (35 000 new cases per year in the United Kingdom) will either present with or develop distant metastases. For these patients, management with combination chemotherapy offers survival prolongation (Saunders and Iveson, 2006), but the disease course is frequently characterised by acquired drug resistance manifesting as tumour progression and the need to alter treatment.Tumour hypoxia and suppression of apoptosis are important drivers of drug (Makin and Dive, 2003; Shannon et al, 2003) and radiotherapy (Saunders et al, 2002) resistance in solid tumours. The transcription factor, hypoxia inducible factor-1 (HIF-1), consists of α and β subunits: the former is ubiquitinated and degraded in normoxia but stabilised in hypoxia; the latter is constitutively expressed independent of oxygen (Semenza, 2003). Under hypoxic conditions, HIF-1α is translocated to the cell nucleus where it dimerises with HIF-1β to bind hypoxia-responsive elements (HRE) in target gene promoters. Hypoxia inducible factor-1 targets encode multiple components of adaptive pathways, including increased glucose uptake, pH stabilisation (via carbonic anhydrase (CA-IX) (Wykoff et al, 2000), and regulation of apoptosis (Greijer and van der Wall, 2004).Of the Bcl-2 family of apoptosis regulatory proteins, pro-apoptotic Bid is of particular interest as it links the mitochondrial (intrinsic)- and death receptor (extrinsic)-driven apoptotic pathways (Andersen et al, 2005; Letai, 2008). In exploring links between tumour hypoxia and apoptosis, we previously demonstrated that HIF-1 repressed transcription of bid in CRC cell lines in vitro (Erler et al, 2004). Moreover, a reciprocal relationship was demonstrated between hypoxic regions and Bid expression in CRC xenografted tumours (Erler et al, 2004). This study sought to determine the clinical significance of these observations and examined the relationship between the number of cells with stabilised nuclear HIF-1α and Bid expression in ex vivo CRC tissue. To assess HIF-1 functionality, nuclear HIF-1α expression was examined in parallel with CA-IX.Materials and methodsPatientsTissues from 39 patients with colorectal adenocarcinoma were used from the joint Christie Hospital and Hope Hospital CRC tissue banks (Manchester, UK) under ethical approvals (LREC-02/051; 03/SM/449). Clinical details have been published previously (Renehan et al, 2000; Bibi et al, 2006). All samples were from luminal tumour edges and processed, formalin-fixed, and paraffin-embedded using standard operational procedures.Immunohistochemical staining and analysisAntibodies used were mouse monoclonal anti-human HIF-1α (610958, diluted 1 : 100, BD Biosciences, Oxford, UK); goat polyclonal anti-human Bid (1 : 50, Santa Cruz Biotechnology, CA, USA); and mouse anti-human CA-IX (anti-MN 75, 1 : 20, kindly provided by Bayer Diagnostics). Sections of 4  μm were stained as previously described (Wilson et al, 2000). Before HIF-1α, and Bid immunostaining, antigen retrieval was by microwaving in 10 mM citrate buffer (pH 6: 2 × 10 min), and the Tyramide Signal Amplication System (NEN Life Sciences, Boston, MA, USA) was used to optimise HIF-1α staining. All staining was completed using diaminobenzidene (EnVision HRP kit, Dako, Cambridge, USA) for 2–3 min before counterstaining with haematoxylin. A variety of approaches were used to control the experiment (see legend, Figure 1). Batch-to-batch variation was assessed using two regions with clear high or low nuclear HIF-1α expression, comparing sections from these with each batch.Hypoxia inducible factor-1α expression was heterogeneous both within and between tumours. To quantify this, a scoring system was developed with nuclear HIF-1α expression as a continuous variable, and the number of immunopositive cells expressed per area (see below). Immunopositivity was defined by strong nuclear staining; cytoplasmic staining alone was considered negative. The mean HIF-1α score was based on the average of three counts. A pilot study identified that all tumours were HIF-1α immunopositive, but that high- and low-positive subpopulations existed using a cutoff of 20 positive cells per 40 000 μm2 (200 × 200).Expression scores for Bid and CAIX were semiquantitative based on stain intensity (Rhodes et al, 2000). Bid was assigned as weak, moderate, or strong (scored as 1, 2, or 3, respectively). For CA-IX, some tumours exhibited no staining and, therefore, the scores included zero. From these scores, a consensus of three observers was taken.To quantify specifically the relationship between nuclear HIF-1α and Bid expression, a technique of paired same area serial analysis was developed. Each specimen was sectioned to produce serial slides 4 μm apart. Slides were viewed and imaged using a Nikon E600 microscope (× 100 magnification). A 200 × 200 μm2 boxed area of high HIF-1α staining was selected and an identical box placed on the corresponding area on the Bid slide. This procedure was repeated for low HIF-1α areas resulting in 78 paired analyses. A series of experiments were repeated for paired HIF-1α and CA-IX.Assessment of internal validityInterobserver variations were assessed to confirm the validity of our analytic approaches. The number of nuclear HIF-1α-immunopositive cells within each boxed area was counted independently by three observers (MMS, MPS, and CD). For Bid and CA-IX, tumour images were randomised before the same observer assessed staining within the assigned boxes.Statistical analysisDistributions of mean nuclear HIF-1α were skewed, and were therefore log-transformed. Comparisons of means used univariate and multivariate analysis of variance (ANOVA). Other immunostaining scores were treated as ordinal. Interobserver variation of continuous data (mean nuclear HIF-1α) was assessed using intraclass correlations: for ordinal data (Bid and CA-IX), κ-correlations were performed. All analyses were performed using STATA (version 9.0, College Station, TX, USA).ResultsImmunopositivity for nuclear HIF-1α was present in all adenocarcinomas but was heterogeneous within tumours. Hypoxia inducible factor-1α staining was noted adjacent to areas of necrosis but also adjacent to blood vessels (Figure 1C). Bid immunostaining was diffuse non-nuclear and positive in all sections but its intensity varied between tumours (Figure 1E).Areas of high HIF-1α immunopositivity were often associated with low Bid expression, whereas areas of low HIF-1α immunopositivity were associated with variable Bid expression. This was quantified as follows: for high-HIF-1α areas, there was a stepwise reduction in Bid expression with increasing mean HIF-1α positivity (ANOVA, P=0.04) (Figure 2A). This relationship was absent for low HIF-1α areas.Mean HIF-1α immunopositivity correlated with the total number of epithelial cells per square (r=0.86, P<0.001), that is, cell density. To assess this further, a multivariate ANOVA model was generated, and demonstrated the attenuation of the extent of inverse association between mean HIF-1α and Bid expression in the high HIF-1α subpopulations (P=0.07).In the assessment of HIF-1 functionality, there was coincident staining of mean HIF-1α immunopositivity with CA-IX immunopositivity (ANOVA, P=0.06) (Figure 2B).Internal validityThe intraclass correlation for HIF-1α immunopositivity counts (among high HIF-1α areas) was 0.77 (95 confidence interval: 0.67–0.88), with an estimated reliability of 91%. The κ-correlations for observer 1 vs 2, 1 vs 3, and 2 vs 3 were as follows: Bid: 0.72, 0.72, and 0.83; CA-IX: 0.55, 0.75, and 0.56; respectively.DiscussionWe purposefully limited our clinical study design to test a single hypothesis that HIF-1 represses bid transcription leading to reduced Bid expression in ex vivo CRC having shown that hypoxia-driven downregulation of Bid occurred via a HIF-1-dependent pathway in CRC cell lines in vitro. We had demonstrated in vitro that this mechanism was operational during hypoxic resistance to etoposide and oxaliplatin, which was overcome to a degree by forced maintained expression of Bid. When CRC lines were xenografted into CD1 nude mice, a striking reciprocal relationship was observed between the hypoxia marker pimonidazole, which formed a penumbra adjacent to necrotic regions in tumours (a location typically associated with chronic hypoxia), and Bid expression, which stained more positively away from the necrotic regions (Erler et al, 2004). However, although the tumour–xenograft model is widely used for preclinical testing of anticancer agents, it has two key limitations – first, the degree of genetic heterogeneity observed in most human tumours does not exist in xenografts; and second, the native stromal microenvironment is lacking. Thus, it was necessary to test the HIF-1α and Bid relationship in human tissues.The human data that emerged were consistent with HIF-1α-mediated repression of Bid occurring in human CRC. However, this relationship was modest, and attenuated after the adjustment for cell density. It is not surprising that HIF-1α expression varied after this adjustment, as its induction may be influenced by hypoxic and non-hypoxic mechanisms (Semenza, 2003). Hence, for example, in the human specimens, HIF-1α immunostaining was observed adjacent to oxygen-rich intratumoral blood vessels. One may expect the relationship between transcriptionally active HIF-1 and Bid to be independent of these mechanisms, but other regulators of Bid expression (e.g., p53 loss) (Sax et al, 2002) may also be operational in CRC tissue. We speculate that the absence of a relationship in low-score HIF-1α may indicate a threshold level for bid transcriptional repression to predominate.As an additional dimension of complexity, it is increasingly clear that modulation of tumour apoptosis may occur through HIF-1-independent pathways, for example, Bad activation and cell survival through an adenosine–Akt pathway in glioblastoma cells (Merighi et al, 2007).Strengths of our study were as follows: first, we used robust staining controls; second, we determined HIF-1α immunopositivity as a continuous variable allowing analyses that included adjustments for cell density; third, we developed and validated the use of the paired same area serial analysis; and fourth, we assessed HIF-1α functionality in the CRC tissues by demonstrating coincident expression with CA-IX, an established HIF-1 transcriptional target.Tumour hypoxia offers a selective approach to the treatment of cancer, as, with only few exceptions, normal tissues do not usually experience low oxygen tension. Accordingly, HIF-1 is considered an attractive anticancer drug target (Semenza, 2003; Welsh and Powis, 2003; Melillo, 2007). Hypoxia inducible factor-1 inhibition could potentially ablate several adaptive responses that allow tumour cell survival under conditions of compromised oxidative phosphorylation. Negating HIF-1 functions improves response to therapy (Hussein et al, 2006) and small-molecule HIF-1 inhibitors are currently in early clinical trial (Semenza, 2003; Melillo, 2007). The data presented here give additional impetus to targeting HIF-1 for therapeutic gain, as it is predicted to increase Bid levels and, in turn, reduce the threshold for apoptosis.\n\nREFERENCES:\n1. Andersen MH, Becker JC, Straten P (2005) Regulators of apoptosis: suitable targets for immune therapy of cancer. Nat Rev Drug Discov\n4: 399–40915864269\n2. Bibi R, Pranesh N, Saunders MP, Wilson MS, O'Dwyer ST, Stern PL, Renehan AG (2006) A specific cadherin phenotype may characterise the disseminating yet non-metastatic behaviour of pseudomyxoma peritonei. Br J Cancer\n10: 10\n3. Erler JT, Cawthorne CJ, Williams KJ, Koritzinsky M, Wouters BG, Wilson C, Miller C, Demonacos C, Stratford IJ, Dive C (2004) Hypoxia-mediated down-regulation of Bid and Bax in tumors occurs via hypoxia-inducible factor 1-dependent and -independent mechanisms and contributes to drug resistance. Mol Cell Biol\n24: 2875–288915024076\n4. Greijer AE, van der Wall E (2004) The role of hypoxia inducible factor 1 (HIF-1) in hypoxia induced apoptosis. J Clin Pathol\n57: 1009–101415452150\n5. Hussein D, Estlin EJ, Dive C, Makin GW (2006) Chronic hypoxia promotes hypoxia-inducible factor-1alpha-dependent resistance to etoposide and vincristine in neuroblastoma cells. Mol Cancer Ther\n5: 2241–225016985058\n6. Letai AG (2008) Diagnosing and exploiting cancer's addiction to blocks in apoptosis. Nat Rev Cancer\n8: 121–13218202696\n7. Makin G, Dive C (2003) Recent advances in understanding apoptosis: new therapeutic opportunities in cancer chemotherapy. Trends Mol Med\n9: 251–25512829013\n8. Melillo G (2007) Targeting hypoxia cell signaling for cancer therapy. Cancer Metastasis Rev\n26: 341–35217415529\n9. Merighi S, Benini A, Mirandola P, Gessi S, Varani K, Leung E, Maclennan S, Baraldi PG, Borea PA (2007) Hypoxia inhibits paclitaxel-induced apoptosis through adenosine-mediated phosphorylation of bad in glioblastoma cells. Mol Pharmacol\n72: 162–17217400763\n10. Renehan AG, Jones J, Potten CS, Shalet SM, O'Dwyer ST (2000) Elevated serum insulin-like growth factor (IGF)-II and IGF binding protein-2 in patients with colorectal cancer. Br J Cancer\n83: 1344–135011044360\n11. Rhodes A, Jasani B, Balaton AJ, Miller KD (2000) Immunohistochemical demonstration of oestrogen and progesterone receptors: correlation of standards achieved on in house tumours with that achieved on external quality assessment material in over 150 laboratories from 26 countries. J Clin Pathol\n53: 292–30110823126\n12. Saunders M, Iveson T (2006) Management of advanced colorectal cancer: state of the art. Br J Cancer\n95: 131–13816835584\n13. Saunders MP, Patterson AV, Stratford IJ (2002) Programming of radiotherapy and sensitization. In Oxford Textbook of Oncology, Souhami RL, Tannock I, Hohenberger P, Horiot J-C (eds) 2nd edn, Vol. 1, pp 459–474. Oxford: Oxford University Press\n14. Sax JK, Fei P, Murphy ME, Bernhard E, Korsmeyer SJ, El-Deiry WS (2002) BID regulation by p53 contributes to chemosensitivity. Nat Cell Biol\n4: 842–84912402042\n15. Semenza GL (2003) Targeting HIF-1 for cancer therapy. Nat Rev Cancer\n3: 721–73213130303\n16. Shannon AM, Bouchier-Hayes DJ, Condron CM, Toomey D (2003) Tumour hypoxia, chemotherapeutic resistance and hypoxia-related therapies. Cancer Treat Rev\n29: 297–30712927570\n17. Welsh SJ, Powis G (2003) Hypoxia inducible factor as a cancer drug target. Curr Cancer Drug Targets\n3: 391–40514683498\n18. Wilson JW, Nostro MC, Balzi M, Faraoni P, Cianchi F, Becciolini A, Potten CS (2000) Bcl-w expression in colorectal adenocarcinoma. Br J Cancer\n82: 178–18510638987\n19. Wykoff CC, Beasley NJ, Watson PH, Turner KJ, Pastorek J, Sibtain A, Wilson GD, Turley H, Talks KL, Maxwell PH, Pugh CW, Ratcliffe PJ, Harris AL (2000) Hypoxia-inducible expression of tumor-associated carbonic anhydrases. Cancer Res\n60: 7075–708311156414"
+}

batch_5/PMC2527836.json

@@ -0,0 +1,4 @@
+{
+    "id": "PMC2527836",
+    "text": "This is an academic paper. This paper has corpus identifier PMC2527836\nAUTHORS: G Lewison, S Tootell, P Roe, R Sullivan\n\nABSTRACT:\nThis study examined cancer research stories on the BBC web archive (July 1998–June 2006). There were about 260 BBC stories per year, of which about 170 were classed as relevant to reports of cancer research. The stories focused heavily on breast cancer, and over one-third of them mentioned this (compared with a cancer disease burden of 13%); the next most covered sites were lung and prostate cancers, although the former was much less mentioned than its cancer disease burden of almost 20% would have suggested. The focus of the stories was often on new or improved drugs or vaccines (20% of stories), with lifestyle choices (12%), genetic developments (9%), and food and drink (8%) also featuring fairly prominently. The BBC stories cited about 1380 research papers that could be identified as journal articles. About three-quarters of the cited papers were in the field of cancer. The papers of these authors came from over 60 countries, and 40% were from the United Kingdom and 36% from the United States. UK cancer research was heavily overcited, by about 6:1, relative to its presence in world oncology research and US research was cited about in proportion. That of most other countries, especially Japan, Germany, and Austria, was relatively undercited. These cited papers also acknowledged more funding bodies. Most of the BBC stories were put in context by external commentators, of whom the large majority was from the UK's cancer research charities.\n\nBODY:\nUK cancer research has enjoyed substantial sociopolitical focus over the last 8 years with the merger of the two largest charities and the creation of the National Cancer Research Institute. Such focus has inevitably given rise to a greater need to demonstrate the impact of national research expenditure and to influence health-care policy (Glass, 2002). Mass media are the nexus between public and policy agenda and are highly influential in shaping discourses about health and research. The way in which news media affect the public is complex and diverse. Recognised effects include informing audiences (Rees and Bath, 2000), agenda-setting, framing (Passalacqua et al, 2004) and persuading (Iyengar, 1997). There is now a substantial corpus of literature demonstrating the impact of media on shaping public opinion towards countries’ health-care systems (Benelli, 2003; Collins et al, 2006), and how newspapers targeted at particular ethnic groups can vary in their approach (Hoffman-Goetz and Friedman, 2005). In addition, there have been studies of how health-related issues have been portrayed in the media around particular diseases (Brown et al, 2001), the uptake of health care (Mintzes et al, 2003) and screening for cancer (Jones, 2004; Schroy et al, 2008), and pharmaceutical coverage (Cassels et al, 2003), particularly in certain media-conscious countries such as Australia, Canada, the United Kingdom and the United States. However, the mass media have also given misleading information about the supposedly beneficial effects of complementary and alternative therapies (Milazzo and Ernst, 2006; Weeks et al, 2007).The nature and impact of science in the media have also become a major policy concern (Conduit, 2001). Commentators on this subject have given vent to a range of complaints, such as the accuracy of media reporting (MacDonald and Hoffman-Goetz, 2002; MacKenzie et al, 2007), the pressure of commercialisation and the challenge of media hype (Ooi and Chapman, 2003; Caulfield, 2004). However, the impact of disease-specific research on the media remains a relatively underdeveloped and understudied area. This study set out to describe in a quantitative manner the impact of cancer research on a major conduit of research stories – the BBC news website (http://www.bbc.co.uk) – which is accessed by some 52% of the UK online universe, some 13.2 million people annually. This source is also used for both the national and international press, TV, and radio stories, and thus provides an ideal surrogate for the determination of overall media impact. The goals of this study were broadly to map the impact of research funding organisations and their commentators on reported cancer research stories, to determine the extent to which the media reporting of cancer research is ‘balanced’ in terms of its site-specific coverage and domains of research, and finally the degree to which media reports of cancer research by the BBC reflect the international impact and indeed whether such reporting in turn influences the citation of papers.MethodologyThe search of the BBC archive was limited to the health section and to the 8-year period from July 1998 to June 2006. The headline, date and abstract of the stories were copied from the BBC website to a spreadsheet, and they were each read through (by GL and ST). They were first coded for relevance (3 for being relevant because they cited research; 2 for being partly relevant, usually because they reported future research or some survey of patient attitudes or experiences; 1 for not relevant – often a report of an individual case). For example, stories with headlines such as ‘Vitamin D can lower cancer risk’ or ‘Virus clue to cervical cancer jab’ were coded 3, and those headed by ‘Boys less likely to eat healthily’ or ‘Ethnic minorities less breast aware’ were coded 2.Selected data (cancer site – e.g., breast or lung, scientists involved, their institution(s), the journal in which any cited paper was published, and details of any commentators) were then extracted and recorded on the spreadsheet for each story. The percentages of BBC stories focusing on different cancer sites were compared with the UK's burden of disease from these particular cancers, measured in Disability Adjusted Life Years (DALYs) (Murray and Lopez, 1996), as given by the World Health Organization for 2002, and relative to the burden from all cancer – this gives a fairer impression of the effect of different cancer types on society than the numbers of deaths. The stories were also coded for the basis of the work being reported, namely drug-related, environmental, food and drink, genetics, job, lifestyle, as 1 or 0 in each of six columns on the spreadsheet.If a story cited one (or more) research paper, then the bibliographic details of these were sought. They could usually be identified readily because the name of the journal was given, but occasionally this turned out to be given incorrectly. Some conference presentations could be identified with meeting abstracts in relevant journals, although for others there did not appear to be a corresponding publication. The addresses of all the authors were also carefully recorded, with at least three elements present: the institution name, the city name (and region/state and postcode, if present), and the country. These addresses could then be analysed by means of special macros to reveal both integer and fractional counts of countries. For example, a paper with one UK and two US addresses would be counted as unity for both countries on an integer count basis, but 0.33 and 0.67, respectively, on a fractional count basis.The research level (RL) of the cited papers was determined to see if the BBC stories primarily covered clinical work, as might have been expected, or sampled fairly the whole range of cancer research as reported in the peer-reviewed serial literature. The research level was calculated on the basis of the journal in which the papers were published as a decimal number between 1 and 4, where 1=clinical observation and 4=basic research. This was determined from the titles of the papers appearing in the given journal that had a biomedical word in their addresses (Lewison and Paraje, 2004). Over 100 ‘clinical’ title words and a similar number of ‘basic’ words were used to determine if a journal paper was clinical, basic, or both: clinical papers were given an RL of unity, basic papers an RL of 4, and ‘both’ papers an RL of 2.5. From these values, it was possible to calculate the mean RL for papers in the journal. Some examples of leading journals with their RLs are given in Table 1.The potential citation impact (PCI) of the cited papers was also based on their journal and was determined as the mean number of citations received by papers in the journal in the year of publication and four subsequent years. However, as this gives values that do not correspond well to the subjective views of both researchers and administrators (Lewison, 1995, 1998) on the relative importance of papers in different journals (which are in the range 1–5 or 6), a logarithmic function (LOG) was also calculated: 1+2log10(1+PCI), whose values range from 1 for ‘ordinary’ journals up to 5 for ‘top’ journals such as Nature. Some examples are shown in Table 1.The actual citation impact (ACI) of those cited papers published in 1998–2001 with addresses only in the United Kingdom, or in the United States (in practice, just over 300 papers) was determined by the counting of citations to them in their year of publication and 4 subsequent years. The address details of all the citing papers were also determined. This was done to gauge whether citation by the BBC influenced the papers’ impact on other scientists.The individual cited papers were also looked up to record their funding acknowledgements. Each such acknowledgement was recorded with four codes: a trigraph to identify the individual organisation (e.g., MRC=Medical Research Council), a letter to denote the type of funding (I=intramural, E=extramural, P=personal, and K=in kind), a digraph to denote the organisation's category (e.g., CH=collecting charity, FO=endowed foundation, GA=government agency, and IP=pharma industry), and another digraph to denote the country of the organisation, taken from the International Standards Organization list. A few funding bodies were European, and coded EU; and some were international and coded XN.All of these parameters – geography, RL, PCI, ACI, and funding �� were also compared with the corresponding values (both means and distributions) for world oncology research papers for relevant years so as to normalise the results and show whether the papers cited by the BBC were, or were not, unbiased samples from the larger population. The world oncology research files were derived from the Science Citation Index by means of a ‘filter’ (Cambrosio et al, 2006) that was based on specialist cancer journals and title keywords; they contained details of about 35 000 papers per year.Many of the BBC stories attempted to put the research news in context with comments from external experts. The names of these people, and their organisations – usually cancer research funders – were recorded. Some unification of the organisation names was needed, and their percentage presence among the commentators was compared with their presence among the funders of UK cancer research in 2000–2001 (Webster et al, 2003) so as to normalise the results and to see if the BBC ‘experts’ were representative of the cancer research funding community in the United Kingdom.ResultsNumbers of stories and cancer sitesFigure 1 shows the numbers of BBC cancer research stories in each of the 9 years of the study. Relevant stories categorised as ‘3’ are shown in black. The number of stories reached a peak in 2002 and has since declined.Of the relevant stories, some mentioned several cancer sites, and others did not mention any particular site. What was clear was that breast cancer dominated, with over one-third of all stories (where one or more sites were mentioned) referring to it. Lung cancer (10%) came a rather poor second, followed by prostate (8%) and skin cancer (6%). Figure 2 shows a plot of percentage of mentions of different cancer sites against percentage of total cancer DALYs for the UK for these sites. This plot makes it clear that cancers of breast, cervix, and skin are overreported in relation to the burden they cause. Of the major cancer types, the biggest ‘deficiency’ is in lung cancer, where there are far fewer stories than the burden of this disease would suggest. It causes 20% of all cancer DALYs, but is only mentioned in 10% of the BBC stories.Story featuresNew and improved drugs (and a few about vaccines) are the dominant type of story, followed by ones on lifestyle (particularly smoking and sunbathing), genetics, and food and drink (including dietary supplements such as vitamins) (Figure 3). Coverage of new drugs peaked in 2001, and then declined; the recent rise is largely due to stories about whether herceptin should be prescribed on the NHS for the early-stage breast cancer. Genetics stories rose to a peak in 2002 and have also declined somewhat. Meanwhile, stories about food and drink have steadily increased in presence; this may also reflect a generally increasing interest in food, including school dinners. But stories about occupational risks, never numerous, appear to have declined steadily; this probably reflects the continuing decline of ‘dirty’ industries in western Europe and North America and their replacement with relatively safer service jobs.Cancer research papers cited by the BBC stories: journals and geographyAbout two-thirds of the BBC stories reported cancer research advances, and some stories cited more than one research paper. In total, there were 1394 cited research papers in 253 different journals that could be identified from the information given in the story (37 papers could not be identified). Some of the leading journals are listed in Table 1, and the cancer journals are shown in bold: altogether, they account for 42 of the total, and for 443 papers, or 32%. (This is fairly typical: in most biomedical sub-fields, only about one-third of the papers are in specialist journals; Lewison, 1996.)Of the cited papers, 1036 (76% of the identified papers) fell within the oncology subfield (ONCOL), as defined by the Cancer Research UK oncology filter, described above. The cited papers included authors from 62 different countries, but the United Kingdom and the United States each had a large proportion of the total on both a fractional and an integer count bases (Figure 4).Figure 4 shows the percentage presence of 18 leading countries in the papers cited by the BBC stories plotted against their presence in oncology research. Although it looks as if the stories were dominated by research from just two countries, the US papers were cited closely in proportion to their presence in oncology research in the 5 years, 2000–2004, and it is just the UK papers that were, not altogether surprisingly, overcited. Indeed a very similar result was found for biomedical research papers cited in UK newspaper stories in 2001 (Lewison, 2002). The papers of two European countries (Denmark and Ireland) were also overcited, but not to the same extent. Research carried out in most European countries was undercited, especially Germany and Austria, as was that of Japan. If UK contributions both to papers cited by the BBC and to oncology are removed, a few more countries (Canada, Netherlands, and Switzerland) now appear to be slightly ‘above the line’.The difference between the observed presence (number of papers and fractional counts) and that expected from the oncology file can be determined as statistically significant or not. For Ireland, the difference is not significant (P∼10%), but for Denmark it is P∼0.2%, and for the larger continental countries, the (negative) difference is highly significant (P<0.01%).Cancer research papers cited by the BBC stories: RL and PCIIn this section, a comparison is made between the distribution of the RLs of the papers cited in BBC stories and those of oncology research papers from 2000–2004. Similarly, the distributions of a log function (LOG) of PCI (based on the journals in which the papers were published) have also been compared. For RL, the comparison takes the form of cumulative distribution curves (see Figure 5); for LOG, distributions are shown as a chart in Figure 6, although mean values have also been calculated and are shown in Table 2.Papers cited by the BBC stories are, on average, more clinical than the oncology papers (Figure 5) and they are also in much more highly cited journals (Figure 6). The mean values of PCI, and of its corresponding log function (LOG), have been rising slowly with time, but they are far below the mean value for the papers cited by the BBC stories. By way of comparison, a mean RL of 2.28 corresponds to Journal of Cancer Research and Clinical Oncology; a mean RL of 2.06 corresponds to Neoplasma. A mean PCI of 13.9 corresponds to British Journal of Cancer or to Journal of Mammary Gland Biology and Neoplasia, and a mean PCI of 34 corresponds to Seminars in Cancer Biology.ACI of papers cited in BBC storiesThe mass media are not only seen, heard, and read by the public, but they also attract the attention of the research community. For example, it was found that The New York Times had a marked effect on the frequency with which the research articles that it wrote about were cited from a comparison of the numbers of citations to articles in the New England Journal of Medicine published before, during, and after a 3-month strike at that newspaper (Phillips et al, 1991). We hypothesised that coverage by the BBC might have a similar effect. However, in the absence of the unique occurrence of the strike (when a paper of record was prepared but not distributed), it is more difficult to normalise and determine the counter-factual situation.We applied two tests to see whether citations by the BBC made a difference to the impact of the papers on the research community. First, we compared the mean actual citation counts with samples of oncology papers with only UK authors, or only US authors, from the year 2000 with their mean potential citation counts, that is, the expected citation scores if they were typical of papers published in the same journals. We then compared the same indicators for two sets of BBC-cited papers, again with just UK authors or just US ones. (The reason for the geographical exclusiveness is that it is known that multinational papers tend to be published in higher impact journals and to receive more citations, but the effect is somewhat dependent on the numbers of countries involved and their identity, and also on the numbers of funding sources acknowledged; Narin et al, 1991; Lewison, 2003.) The results are shown in Table 3.Both sets of oncology papers received fewer citations, on average, than might have been expected from the journals in which they were published, but both sets of papers cited by the BBC received more citations than expected. This suggests either that the BBC reporters were prescient in selecting the papers that would be highly cited or that they influenced the process – perhaps the latter explanation is more likely, as BBC coverage may well have led to stories in the newspapers as well and thus to greater publicity for the papers, and not only in the UK. They may also have been pre-selected by the journal staff, who often send out press notices to draw attention to papers thought to be of particular interest and significance.The second test was to see whether citations in BBC stories had a disproportionate influence on UK scientists, as measured by the propensity of these papers to have a higher percentage of their subsequent citations from the United Kingdom. For this purpose, we downloaded bibliographic details of all the citing papers from the above table (namely, 5466+11 007+3599+8950=29 022) and analysed them geographically so as to reveal the (fractional count) contributions of the different citing countries. The results are shown in Table 4.Data for Canadian citing authors have been provided to show that for a third country, the percentages of their citations are almost constant. The column head ‘UK, %’ shows that UK authors were more frequent citers of both groups of papers mentioned in the BBC stories, by 28.2/24.1 or by +17% for the UK papers and by 5.9/5.0 or by +18% for the US ones. The effect is not large, but the differences between the observed and expected values on the null hypothesis are statistically significant (P=0.003% for the US papers, and even less for the UK ones). The authors from US were less likely to cite the UK articles in BBC stories, but they were more likely to cite the US papers, which probably also received coverage in their own country's media.The funding of the cited research papersAltogether, funding acknowledgements were recorded for 1385 papers, of which 594 had an address in the United Kingdom and 647 had an address in the United States. Of these 1385 papers, 290 or 21% had no recorded funding, and the research was probably supported by general university funds or those of the relevant hospital or state health service based on the presence of university or hospital addresses. The others had one or more funding acknowledgements, the maximum number being 39 for two papers. Figure 7 shows the distribution of the numbers of funders, with, for comparison, the corresponding percentages found for a large structured sample of oncology papers from 2003.It is clear that the papers cited by the BBC stories have more funding acknowledgements and that, in particular, relatively few have ‘no’ funding – 21% compared with 34% for the sample of oncology papers. An analysis was then made of the sector of the funding bodies for UK and US papers. For this purpose, funders were divided into four main sectors, namely national government (including, local and regional authorities), national private non-profit, industrial, and international. There were also other funders, public and PNP sector organizations from third countries. The sectoral analysis is shown in Table 5.As would be expected from Figure 7, there are many fewer ‘unfunded’ papers than expected for both the United Kingdom and the United States. In the United Kingdom, funding is dominated by the private non-profit sector, but in the United States it is led by the government sector, with much support coming from the National Cancer Institute (NCI) and from the National Institutes of Health (NIH) more generally. Industry funds about one-sixth of the papers cited by the BBC in both countries. The international funding is primarily from the European Commission, which was acknowledged on 48 of the papers cited by the BBC, of which 35 were from the United Kingdom and 9 from the United States (all, of course, co-authored with a European Union Member State). By contrast, the World Health Organization (WHO) was acknowledged on only twelve papers cited by the BBC, of which four were from the United Kingdom and six from the United States, and two from both countries.Commentators on the research findingsA notable feature of most of the BBC stories was that external experts were asked to comment on the significance of the results, and what they might mean for the treatment of cancer patients. Altogether, there were as many as 724 different commentators over the 8-year period, and they made a total of 1842 comments on 1272 stories, with an average of 1.45 quotes on each story where a commentator was quoted. Of the 12 most frequent commentators (one of whom was RS) with 19 or more appearances, no fewer than 10 were from Cancer Research UK, and the other 2 were also from cancer research charities (Breakthrough Breast Cancer and the Prostate Cancer Charity). Public sector commentators were relatively rare. They included Julietta Patnick of the NHS Cancer Screening Programme, Mike Richards, the National Cancer Director, and also 10 Members of Parliament. The Medical Research Council (MRC) commented only on 15 stories, just over 1%. Figure 8 shows that there is no correlation at all between support for cancer research in the United Kingdom and invitations to comment, except for Cancer Research UK and the Department of Health (UK), for which the invitations are in fair proportion. Representatives of the Wellcome Trust, which, despite formally eschewing the funding of clinical cancer research in the United Kingdom (Wellcome Trust, 2008), funded about 7% of UK cancer research in 2000–2001, did not comment at all, and two cancer charities based outside London, the Yorkshire Research Campaign in Harrogate and the Association for International Cancer Research in Fife, Scotland (both funding 2.3% of UK cancer in those years) only commented once and twice, respectively.These findings suggest that the BBC (in common with other news media) see it as highly desirable to put new research findings in context. Most commentators were enthusiastic and viewed the research as very useful, but they counselled strongly against premature optimism and stressed that more research work was needed, and more time, before patients were likely to benefit. The frequency with which the BBC reporters (nearly all anonymous, unlike in the newspapers) turned to the cancer charities in London for interpretation of the research news suggests that this path has become well-worn and, in turn, that the charities have become adept at presenting themselves to the public even if only in the reflected glory of research done by others. Although the MRC and the Wellcome Trust are active in cancer research, one would not realise this from the BBC stories.DiscussionThe BBC archive, which closely approximates to the items broadcast on radio and television, is a fecund source of bibliometric data (Ghosh, 2007). It provides a new perspective on biomedical research and helps reveal its likely impact on the wider public. In this study, we have found consistently high levels of BBC cancer research stories. The impetus of charitable mergers and additional government support to cancer research appears to have been reflected in the absolute numbers of cancer research stories reaching a peak in 2002. However, the wave of activity appears to have hit a high tide mark and has been settling back down over the last 3 years to pre-2000 levels.We have found, relative to the DALY impact, a strong focus on breast cancer in these stories. Within the cancer field, breast cancer is acknowledged to have strong personal identity and worldwide advocacy stemming from activism in the late 1980s and a research track record of delivering new and effective management in the form of greater understanding of pathogenesis and risk factors (e.g., Million Women Study/Hormone Replacement Therapy) through surgical techniques (Sentinel Node Biopsy) and drugs (e.g., the wide range of endocrine therapies). This self-reinforcing cycle has filtered through to generate huge media exposure, when compared with other site-specific cancers. This has been a mixed blessing. Although this clearly provides greater breast awareness and potentially increases charitable giving to breast cancer (although there is no direct evidence that this happens), we know that women greatly overestimate their lifetime risk of breast cancer (Pohls et al, 2004). Why? The plausible hypothesis is that the greater exposure of women to breast cancer stories, particularly those associated with risk factors, leads to a resetting of the cultural milieu in which women estimate their risk.A number of site-specific cancers fare poorly in their media exposure – lung, upper GI tract, and so on. For the majority of these cancers (lymphoma being an exception to this observation), progress in controlling and curing them has been glacial. Mortality is high and survival poor. Thus, a lack of year-on-year treatment progress coupled to poor outcomes has inevitably led to poor media exposure. As art reflects society, so the media are reflecting progress – those cancers for which we are ahead of the curve (e.g., breast), those on the curve (e.g., colorectal), and those that are behind (e.g., upper GI tract and lung). Media trends in coverage of site-specific cancers over time appear from our data to be an additional tool to inform policymakers not only of active/successful research domains but also of those that have developed strong interest groups (Howlett and Ramesh, 2003).In a typical year, there will be about 240 BBC stories on cancer research, of which about 160 will cite research papers. The main focus is on new and improved drugs that could be used to treat cancer. Drugs dominate research reports and the trend is increasing. Such stories overwhelmingly emphasise breakthroughs, and, as previously reported, do a poor job in reflecting the relative contribution of the drug in question and indeed the multitude of giants on whose shoulders such new technologies have come into being (SIRC, 2001). Again the media are reflecting the dominant research paradigm in cancer – that of drug discovery and development – and providing a surrogate marker not only for the activity levels of research in this area (which are very high compared with other domains, e.g., prevention) but also the proactive sociopolitical advocacy of the organisational structures in which the majority of researchers are based, namely the pharmaceutical and biotechnology industries. The economic model of these organisations demands media communication as a tool for leveraging funding and/or influencing the stock market and shareholders. Such proactivity leads to hype and a lack of relativism in the communication by the media of cancer research stories, a constant theme in health literacy and public policy (Hayes et al, 2007). Should policymakers be concerned? In sociological terms they should be, because hype invariably leads to unmet public expectations; the knock-on effect is a backlash against such activities (or rather the funding of them) and a decline in trust (Caulfield, 2004). The latter, considering the current issues of trust around commercially driven research anyway, is particularly damaging to the public credibility of science (DeAngelis, 2000). Despite the fact that the story types probably reflect the relative research activity levels in these domains, there is a clear need to promote more balance with non-drug cancer research stories.The papers cited by the BBC, numbering about 1380, are overwhelmingly from the United Kingdom (which are overcited by the BBC by a factor of about 6) and from the United States, which together comprise about 84% of the total. Although they tend to be rather more clinical than the average for cancer research papers overall, basic research is by no means neglected. They are taken from a very wide range of journals (over 250), but three UK journals, namely The Lancet, British Journal of Cancer, and British Medical Journal, dominate. On average, the cited journals have high impact factors – the mean is more than twice that for cancer research papers overall. What these findings do suggest is that media reporting of cancer research by the BBC is, relative to global cancer research activity and outputs (publications), narrow. One argument against this might suggest that important research is (a) only published in the lingua franca of science, that is English; (b) is thus dominated by native English speaking countries, that is the United Kingdom and the United States; and (c) will inevitably be published in the higher impact journals. The problem is that a high impact journal will consequently produce research that has a high impact on the public, but the few qualitative studies in science have actually found a very poor correlation between the initial impact of a paper (as defined by the journal) and its eventual impact. The other major issue is the perceptual bias that an undue focus on a very selective group of journals and countries has on the public's view of cancer research. Such situations lend too easily to the creation of a hegemony where biases perpetuate themselves and create a public frame of reference around cancer research that has a false logic and consciousness (Sallach, 1974).Finally, we have attempted to address what impact, if any, the reporting of cancer stories by the BBC has had on the research community. Our data suggest a possible influence, but this study cannot rule out the possibility that the BBC news team consistently pick higher impact papers (although this is unlikely from other studies). How such stories impact the general public's perception of cancer research in all its permutations is a logical question that can be built upon the data presented. Similarly, the variability in which cancer research is presented through the media in different European Member States and more widely would say much about the different cultural perceptions towards this particular biomedical research domain. Cancer research does not have a well-developed ‘third culture’ (direct researcher-to-public connection), and hence is increasingly reliant on the diverse media for informing and engaging the public. If the aim is to achieve a ‘public understanding of cancer’, then we need to ask how this current interface works and whether it is achieving the aspirations of both the public and the research community.\n\nREFERENCES:\n1. Benelli E (2003) The role of the media in steering public opinion on healthcare issues. Health Policy\n63: 179–18612543530\n2. Brown P, Zavestoski SM, McCormick S, Mandelbaum J (2001) Print media coverage of environmental causation of breast cancer. Sociol Health Illn\n23(6): 747–775\n3. Cambrosio A, Keating P, Mercier S, Lewison G, Mogoutov A (2006) Mapping the emergence and development of translational cancer research. Eur J Cancer\n42: 3140–314817079135\n4. Cassels A, Hughes MA, Cole C, Mintzes B, Lexchin J, McCormack JP (2003) Drugs in the news: an analysis of Canadian newspaper coverage on new prescription drugs. Can Med Assoc J\n168(9): 1133–113712719316\n5. Caulfield T (2004) Biotechnology and the popular press: hype and the selling of science. Trends Biotech\n22(7): 337–339\n6. Collins P, Abelson J, Pyman H, Laris JN (2006) Are we expecting too much from print media? An analysis of newspaper coverage of the 2002 Canadian healthcare reform debate. Social Sci Med\n63(1): 89–102\n7. Conduit C (2001) What is ‘public opinion’ about genetics? Nat Rev Genet\n2: 811–81511584297\n8. DeAngelis C (2000) Conflict of interest and the public trust. J Am Med Assoc\n285: 2237–2238\n9. Ghosh P (2007) The BBC science correspondent, personal correspondence.\n10. Glass N (2002) UK charity to involve public in decision making for cancer research priorities. Lancet\n360(9344): 148712433528\n11. Hayes M, Ross IE, Gasher M, Gutstein D, Dunn JR, Hackett RA (2007) Telling stories: news media, health literacy and public policy in Canada. Social Sci Med\n64(9): 1842–1852\n12. Hoffman-Goetz L, Friedman DB (2005) Disparities in the coverage of cancer information in ethnic minority and mainstream mass print media. Ethn Dis\n15(2): 332–34015825981\n13. Howlett M, Ramesh M (2003) Studying Public Policy: Policy Cycles and Policy Subsystems. Oxford University Press: Don Mills, ON\n14. Iyengar S (1997) Overview. In Do the Media Govern? Politicians, Voters and Reporters in America, Iyengar S, Reeves R (eds), pp 211–216. Sage Publications. Chapter 26\n15. Jones SC (2004) Coverage of breast cancer in the Australian print media – Does advertising and editorial coverage reflect correct social marketing messages? J Health Commun\n9(4): 309–32515371084\n16. Lewison G (1995) Evaluation of national biomedical research outputs through journal-based esteem measures. Res Eval\n5: 225–235. See pp 228–229 and Table 3\n17. Lewison G (1996) The definition of biomedical research subfields with title keywords and application to the analysis of research outputs. Res Eval\n6: 25–36\n18. Lewison G (1998) New bibliometric techniques for the evaluation of medical schools. Scientometrics\n41: 5–16. pp 10–11\n19. Lewison G (2002) From biomedical research to health improvement. Scientometrics\n54: 179–192\n20. Lewison G (2003) The publication of cancer research papers in high impact journals. Aslib Proc\n55: 379–387\n21. Lewison G, Paraje G (2004) The classification of biomedical journals by research level. Scientometrics\n60: 145–157\n22. MacDonald MM, Hoffman-Goetz L (2002) A retrospective study of the accuracy of cancer information in Ontario daily newspapers. Can J Public Health\n93(2): 142–14511963520\n23. MacKenzie R, Chapman S, Holding S, McGeechan K (2007) ‘A matter of faith, not science’: analysis of media coverage of prostate cancer screening in Australian news media 2003–2006. J R Soc Med\n100(11): 513–52118048709\n24. Milazzo S, Ernst E (2006) Newspaper coverage of complementary and alternative therapies for cancer – UK 2002–2004. Support Care Cancer\n14(9): 885–88916703334\n25. Mintzes B, Baver ML, Kravitz RL, Bassett K, Lexchin J, Kazanjian A, Evans RG, Pan R, Marion SA (2003) How does direct-to-consumer advertising (DTCA) affect prescribing? A survey in primary care environments with and without legal DTCA. Can Med Assoc J\n169(5): 405–41212952801\n26. Murray CJL, Lopez AD (1996) The Global Burden of Disease. Harvard University Press: Cambridge, MA. ISBN 0-674-35448-6\n27. Narin F, Stevens K, Whitlow ES (1991) Scientific co-operation in Europe and the citation of multi-nationally authored papers. Scientometrics\n21: 313–324\n28. Ooi ES, Chapman S (2003) An analysis of newspaper reports of cancer breakthroughs: hope or hype? Med J Aus\n179: 639–643\n29. Passalacqua R, Caminiti C, Salvagni S, Barni S, Beretta GD, Carlini P, Contu A, DiCostanzo F, Toscano L, Campione F (2004) Effects of media information on cancer patients’ opinions, feelings, decision-making process and physician–patient communication. Cancer\n100(5): 1077–108414983505\n30. Phillips DP, Kanter EJ, Bednarczyk B, Tastad PL (1991) Importance of the lay press in the transmission of medical knowledge to the scientific community. N Engl J Med\n325: 1180–11831891034\n31. Pohls UG, Renner SP, Fasching PA, Lux MP, Kreis H, Ackermann S, Bender H-G, Beckmann MW (2004) Awareness of breast cancer incidence and risk factors among healthy women. Eur J Cancer Prevent\n13(4): 249–256\n32. Rees CE, Bath PA (2000) Mass media sources for breast cancer information: their advantages and disadvantages for women with the disease. J Doc\n56(3): 235–249\n33. Sallach DL (1974) Class domination and ideological hegemony. Sociol Q\n15: 38–50\n34. Schroy PC, Glick JT, Robinson PA, Lydotes MA, Evans SR, Emmons KM (2008) Has the surge in media attention increased public awareness about colorectal cancer and screening? J Community Health\n33(1): 1–918080203\n35. Social Issues Research Centre (SIRC) in partnership with the Royal Society and the Royal Institution of Great Britain (2001) Code of Practice/Guidelines on Science and Health Communications, SIRC, http://www.sirc.org/publik/cop_guidlines_1.html\n36. Webster BM, Lewison G, Rowlands I (2003) Mapping the Landscape II: Biomedical Research in the UK, 1989–2002. The City University, London, Department of Information Science. Available at http://www.ucl.ac.uk/ciber/MappingtheLandscape.php\n37. Weeks L, Verhoef M, Scott C (2007) Presenting the alternative: cancer and complementary and alternative medicine in the Canadian print media. Support Care Cancer\n15(8): 931–93817624558\n38. Wellcome Trust (2008) See http://www.wellcome.ac.uk/About-us/Policy/Policy-and-position-statements/WTD002761.htm (accessed on 14 April 2008)"
+}

batch_5/PMC2528054.json

@@ -0,0 +1,4 @@
+{
+    "id": "PMC2528054",
+    "text": "This is an academic paper. This paper has corpus identifier PMC2528054\nAUTHORS: Wei Chen, Margit Foss, Kuo-Fu Tseng, Dahong Zhang\n\nABSTRACT:\nCytokinesis is powered by the contraction of actomyosin filaments within the newly assembled contractile ring. Microtubules are a spindle component that is essential for the induction of cytokinesis. This induction could use central spindle and/or astral microtubules to stimulate cortical contraction around the spindle equator (equatorial stimulation). Alternatively, or in addition, induction could rely on astral microtubules to relax the polar cortex (polar relaxation). To investigate the relationship between microtubules, cortical stiffness, and contractile ring assembly, we used different configurations of microtubules to manipulate the distribution of actin in living silkworm spermatocytes. Mechanically repositioned, noninterdigitating microtubules can induce redistribution of actin at any region of the cortex by locally excluding cortical actin filaments. This cortical flow of actin promotes regional relaxation while increasing tension elsewhere (normally at the equatorial cortex). In contrast, repositioned interdigitating microtubule bundles use a novel mechanism to induce local stimulation of contractility anywhere within the cortex; at the antiparallel plus ends of central spindle microtubules, actin aggregates are rapidly assembled de novo and transported laterally to the equatorial cortex. Relaxation depends on microtubule dynamics but not on RhoA activity, whereas stimulation depends on RhoA activity but is largely independent of microtubule dynamics. We conclude that polar relaxation and equatorial stimulation mechanisms redundantly supply actin for contractile ring assembly, thus increasing the fidelity of cleavage.\n\nBODY:\nIntroductionThe contractile ring is a dynamic, actomyosin-based structure whose constriction generates a cleavage furrow, ultimately producing two daughter cells. To ensure accurate distribution of chromosomes to each cell, the location of the division plane and the timing of cleavage must be tightly regulated to coordinate with chromosomal segregation. In animal cells, this coordination is mediated by the spindle apparatus, which both segregates the chromosomes during anaphase and signals the cortex as to where and when the contractile ring should be assembled (reviewed in, e.g., [1–4]). In addition to its structural components, the ring contains numerous other proteins, some of which are thought to recruit or retain ring components, while others function as regulatory or signaling molecules. For example, the GTPase RhoA and its guanine nucleotide exchange factor help regulate polymerization of actin and myosin within the ring, thus promoting its constriction (reviewed in [5]). Some proteins are localized both to the central spindle and the equatorial cortex in a temporally dynamic manner. It is not known whether these proteins represent independent pools or whether the proteins diffuse or are somehow transported between locations.Spindle microtubules are known to play a critical role in specifying the cleavage plane in animal cells. Evidence from divergent organisms shows that the cleavage furrow can be initiated at the spindle equator by astral microtubules, central spindle microtubules, or both (reviewed in [1]). Furthermore, microtubules have been shown to be the only structural constituent of the spindle apparatus required for furrow induction, regardless of their original location within the spindle [6]. However, it remains unclear whether microtubules induce furrow formation by mechanisms of polar relaxation and/or equatorial stimulation of the cell cortex (reviewed in [1–4,7]).The polar relaxation model, as originally proposed by Wolpert [8], postulates that inhibitory signals are sent from astral microtubules to the polar cortex to decrease cortical tension locally. Computer simulations indicate that high microtubule density at the poles may produce the highest surface tension at the equator, where the contractile ring forms [9,10]. In sea urchin eggs, just before cytokinesis, tensile forces become elevated globally, and thus relaxation of tension at the poles may lead to furrow formation at the equator [11,12]. Furthermore, in cells with depolymerized or genetically shortened microtubules, the overall cortical contractility is higher, albeit disorganized (e.g., [13–15]). In theory, microtubules could induce polar relaxation by a number of different mechanisms. Microtubules could inhibit assembly (or promote disassembly) of polar actin filaments. Alternatively, they could inhibit contractile activity of the polar filaments. For example, microtubules might inhibit Rho-dependent actin contractility through localized sequestration and inhibition of GEF-H1, an activator of Rho [16]. Or instead, microtubules could induce polar relaxation by altering the distribution of cortical actin filaments, relocating them away from the poles in a process known as cortical actin flow. Cortical flow of actin has been observed in cultured mammalian cells [17].The equatorial stimulation model proposes that furrowing is induced by stimulatory signals sent from the central spindle midzone and/or the spindle asters to the equatorial cortex. The classic “torus experiment” by Rappaport [18] demonstrated that astral microtubules from two opposing asters, not connected by a spindle, are sufficient to induce ectopic furrow formation in echinoderm embryos. Because the cell was topologically manipulated to eliminate the poles, the implication was that polar relaxation could not be invoked as the mechanism for furrowing—although technically, a role for cortical flow has not been excluded. Similarly, astral microtubules from two neighboring spindles can define such an ectopic furrow in a fused epithelial cell [19]. For monopolar spindles, it was proposed that furrow formation was stimulated by a subset of astral microtubules stabilized by chromosomes [20], or by bundles of central spindle-like microtubules [21]. In cultured rat cells, when communication between the central spindle and equatorial cortex is disrupted by a small “perforation” (i.e., a deep indentation of the membrane, located peripherally within the equatorial plane), a furrow fails to form at the equatorial cortex [22]; the furrow initiates instead at the perforation site—as if the central spindle were now signaling to an “interior cortex” located on the spindle side of the perforation. Proteins localized to the central spindle, such as centralspindlin (consisting of a RhoGAP and a kinesin) and RhoGEF, are important for organizing midzone microtubules and signaling the equatorial cortex for furrow induction [5,23,24]. Recently, LET-99 and heterotrimeric G proteins were shown to play important roles in aster-positioned cytokinesis in Caenorhabditis elegans embryos [25]. In C. elegans zygotes, stimulatory signals from asters and the spindle midzone are both important for cytokinesis [26–28]. In Drosophila, it was very recently shown that F-actin–associated vesicles co-localize with central spindle and peripheral microtubules. Thus, it was hypothesized that the central spindle may somehow coordinate co-transportation of actin and new membrane to the ingressing cleavage furrow [29].Although it has not been directly demonstrated that polar relaxation and equatorial stimulation coexist in a single cell during cytokinesis, other groups have worked to document the relationship between astral-derived and central spindle-derived cytokinetic signals. For example, in the Drosophila asterless mutant, a suboptimal version of cytokinesis can occur despite the absence of asters from the male meiotic spindle, demonstrating that astral microtubules are dispensable for inducing cytokinesis in male meiosis [30]. Bringmann and Hyman showed that astral and central spindle signals both help position the cytokinetic furrow during the first zygotic division in C. elegans [26]; however, cortical tension was not specifically addressed. The astral signal in C. elegans was later shown to be distinct from the central spindle signal and to negatively regulate distribution of cortical myosin [28]. In contrast, myosin II in cultured mammalian cells does not undergo cortical flow, as determined by kymography, whereas actin does [17]. Furthermore, the authors postulated that cortical actin is not the sole source of equatorial actin [17]; the other source was not identified. Hu et al. recently observed HeLa cells undergoing drug-induced monopolar cytokinesis. They postulated that feedback loops (i.e., bi-directional signaling) between midzone-like spindle microtubules and various furrow components could promote polarization in these cells, and also that cortical flow could help organize microtubules [21]. Our work builds on, complements, and extends these ideas, using a novel experimental approach.If we assume that polar relaxation and equatorial stimulation can coexist in cytokinesis, it is not obvious how microtubules manage to provide opposing signals in the same cell, i.e., to relax the polar cortex yet stimulate the equatorial cortex. To address this question, and to probe the mechanistic link between cortical properties and contractile ring assembly, we used fluorescently labeled cytokinetic cells of the silkworm Bombyx mori. These cells allowed us to examine the distribution of actin filaments as driven by micromanipulated microtubules, with or without drug treatments that affect either microtubule dynamics or actin assembly. We show that loss of cortical actin may occur at any region of the cell cortex adjacent to dynamic microtubules, due to cortical flow. Stimulation, as gauged by rapid de novo assembly and delivery of actin aggregates, may also occur at any region of the cell cortex, mediated by the bundled, overlapping plus ends of central spindle microtubules. Our data show that polar relaxation and equatorial stimulation mechanisms coexist in silkworm spermatocytes and are compatible with dual roles for microtubules. Furthermore, our work provides mechanistic advances and demonstrates how both mechanisms can independently supply actin for contractile ring assembly.ResultsThe Spindle Apparatus and Cytokinesis in Spermatocytes of the Silkworm B. mori\nSpermatocytes cultured underneath a layer of halocarbon oil were relatively large (∼33 μm in diameter), optically clear, and amenable to micromanipulation. Digital-enhanced polarization microscopy on cells from late metaphase to early anaphase revealed a short but robust spindle (Figure 1A and Video S1; n = 9), whose relatively small size provided ample space for manipulation of spindle microtubules. To visualize the dynamics of microtubules and actin filaments during cytokinesis, we microinjected metaphase-stage spermatocytes with rhodamine tubulin and low-level Alexa Fluor 488 phalloidin (Figure 1B and Video S2; n = 4). The fluorescently labeled cytoskeleton was imaged using spinning disc confocal microscopy. Furrow initiation occurred during early anaphase after actin aggregates emerged at the interdigitating microtubule plus ends at the equator (Figure 1B, between 0 and 4). The aggregates gradually enlarged before coalescing with cortical actin filaments to assemble the contractile ring that constricts the cell (Figure 1B, 4–45). The time from anaphase onset to cleavage was not significantly different for the uninjected control cells (48.4 ± 2.5 min) and the injected, double-labeled cells (49.5 ± 5.0 min), suggesting that the low-level phallodin was not perturbing actin function.Figure 1Cytokinesis of Silkworm Spermatocytes(A) Polarization micrograph of a dividing spermatocyte.(B) Confocal micrographs of a spermatocyte microinjected with rhodamine tubulin (microtubules false colored green), and low-level Alexa Fluor 488 phalloidin (actin false colored red). Actin aggregates appeared at the equatorial microtubule plus ends during early anaphase (0–20), then fused into a contractile ring that bisected the cell (4–45).(C) Astral flagella, as well as naturally, asymmetrically positioned asters, were visible in dividing spermatocytes. Astral flagella were visible only in fixed, immunostained cells (a–c). Astral microtubules were prominent in fixed anaphase spermatocytes (e.g., b). One aster was loosely attached to the spindle (arrow). Occasionally, the asters (arrows) in fixed (a and c) and live (d and e) spermatocytes (arrows) were naturally positioned on the same side of the spindle. a and d, metaphase; b, c and e. anaphase. Fixed cells were double-immunolabeled for actin and tubulin; live cells were labeled as in (B).Time in min. Time 0 in (A and B), anaphase onset. Bars, 10 μm.Image stacks of fixed, immunostained cells revealed dense arrays of astral microtubules radiating from the centrosomes to the polar cortex (Figure 1C, a–c; n = 15). Notably, asters in both fixed (Figure 1C, a–c) and live (Figure 1C, d and e; n = 15; also in Figure 1B, 0) cells appeared to be attached only loosely to the spindle, which is a natural phenomenon in silkworm spermatocytes [31]. Presumably due to the centrioles' motile flagellar axonemes (Figure 1C, a–c; as seen in immunostained cells), both asters were mobile within the cell. In approximately 15% of the cells, the asters even moved to the same side or the same pole of the spindle, as documented in metaphase (Figure 1C, a and d) and anaphase (Figure 1C, c and e) stage cells. This natural phenomenon makes silkworm spermatocytes an ideal system for separating the role of asters from that of the central spindle in cytokinesis.Evidence for Polar Relaxation as Gauged by Redistribution of Cortical ActinCortical flow of actin filaments driven by spindle microtubules.In cells in which both asters (Figure 2A, arrows) happened to be located at the same pole of the spindle, cortical actin filaments appeared to flow to the opposite side of the cell during anaphase (Figure 2A, 0–6.3, and Video S3; n = 4. Note: panels were chosen to highlight cortical flow; these focal planes are not optimal for viewing actin aggregates). In such cells, the midzone of the central spindle was never aligned with the cell's equator, and these cells divided asymmetrically around the equator of the shifted spindle (Figure 2B and Video S4; n = 6). Our observation suggested that astral microtubules could relax the polar cortex by excluding cortical actin filaments from the pole. However, since the asymmetric furrow was aligned with the spindle equator, it was also possible that the spindle midzone signaled the band of cortex that encircled the spindle equator, stimulating recruitment of actin to the furrow. To discern between these alternatives, we needed to assess whether actin could be redistributed within the cortex in the absence of an intact central spindle. Thus, we used a microneedle to push the spindle poles together in anaphase, mechanically remodeling the spindle apparatus. We refer to this operation as “collapsing” the spindle. During the remodeling process, the spindle was simultaneously relocated to a region of the cell previously unexposed to any furrow cues (e.g., nascent actin aggregates). By collapsing the spindle, we accomplished three objectives. First, the act of collapsing served to reorganize the central spindle, destroying its equatorial configuration. This likely impeded its ability to stimulate the cortex in its new location, preventing the formation of a complete, circular furrow. Second, relocation of the spindle presumably prevented deposition of additional furrow cues at the original cortical site, and depending on the timing of the move, may well have prevented deposition of any cues at the original location. Third, relocation and collapsing of the spindle allowed us to induce controlled exclusion of actin filaments from a region of our choice; relocation brought the spindle close enough to the cortex to initiate cortical flow, whereas collapsing exposed or freed some microtubule plus ends that originally interdigitated with the opposite half of the spindle, potentially altering their dynamic properties (see below).Figure 2Cortical Actin Filaments Were Excluded from the Polar Cortex Bordering the Asymmetrically Distributed Asters(A) Both asters (arrows) were naturally positioned at the upper pole of the spindle. Cortical actin filaments flowed towards the lower polar cortex during early anaphase. Cells in Figure 2 were labeled as in Figure 1B. Time 0 depicts anaphase onset.(B) Cortical actin filaments, excluded by asymmetrically distributed asters (arrows), assembled a contractile ring around the equator of the naturally shifted spindle. Actin aggregates were also present by later anaphase.Time in min. Time 0, anaphase onset. Bars, 10 μm.Shortly after anaphase onset (Figure 3A, 0, and Video S5; n = 12), we pushed the spindle apparatus with a microneedle to an arbitrary region of the cell cortex while collapsing it as described (Figure 3A, 3). Shortly thereafter, cortical actin filaments began to flow away from the spindle microtubules to the opposite side of the cell (Figure 3A, 10). Because actin filaments are a major component of actomyosin contractile elements, their exclusion from the cortex near the spindle would presumably result in cortical relaxation. This microtubule-driven actin flow resulted in asymmetric distribution of cortical actin filaments (Figure 3A, 10) and initiation of contractile ring assembly (Figure 3A, 16, and Video S5) around the presumptive, newly defined zone of microtubule overlap [32]. The microtubule-driven actin flow persisted into telophase, as shown in similar experiments in which cells were manipulated while undergoing cytokinesis (Figure 3B and Video S6; n = 26). This resulted in asymmetric placement of the contractile ring (Figure 3B, 18–37). Because randomly dislocated asters (Figure 2) and arbitrarily repositioned, collapsed spindles (Figure 3A and 3B) caused cortical actin flow, we hypothesized that cortical relaxation could be induced at any region of the cell cortex by microtubules from any source. To test this idea, we mechanically remodeled cells by placing both asters in one region of the cell, separated from the asterless, collapsed spindle, as depicted in the schematic in Figure 3C (Figure 3C, 5–12, and Video S7; n = 8). As expected, the contractile ring assembled between the two noninterdigitating structures due to actin exclusion from both “poles” (Figure 3C, 5–66, and Video S7).Figure 3Cortical Flow of Actin Filaments Was Driven by Spindle Microtubules(A) During anaphase (0), when the spindle apparatus was collapsed, and pushed with a microneedle to an arbitrary region of the cell cortex (3), actin filaments flowed to the opposite side of the cell (3–25). This redistribution of actin filaments by spindle microtubules resulted in asymmetric cell division (71).(B) Repositioning and reorganization of the spindle (2) during telophase resulted in a similar scenario, preceded by regression of the original furrow.(C) An anaphase cell (with only one aster visible in this focal plane; 0) was mechanically remodeled by moving spindle microtubules to the upper portion of the cell and collapsing the spindle (shown at top of cell, 5–21). The collapsed spindle was manipulated to detach and relocate first one of its asters (dot in center of cell, 5) and then its second aster (to left of first aster; both asters marked by arrows, 12–21). Possibly due to actin exclusion (5–21) by microtubules from both structures, the contractile ring formed between the collapsed spindle and the pair of detached asters (21–66). This is shown in the schematic diagram (representing the cell in 21).(D) A schematic of the set-up, with fluorescence images of microtubule-driven actin flow blocked by a microneedle. The schematic shows the cell's outline, the location of the needle (gray), and the direction of flow of cortical actin (red), as induced by the collapsed, repositioned spindle (green, at top). The manipulation needle (arrows), which indented but did not pierce the plasma membrane, locally intercepted the actin flow; brighter fluorescence accumulated on the side of the needle that faced the repositioned spindle (1.7–4.5). The three large red objects (3.4–4.5, upper right) are cell division scars, i.e., remnants from former cytokinetic rings. These scar structures can also move around within the cortex during cortical flow. Note: despite removal of the holding needle, the spindle is not depicted as a giant aster, since the reorganization from monopolar spindle to aster takes at least several minutes. Cells in (A–D) were labeled as in Figure 1B. Time in min. Time 0 in (A), (C), and (D), mid to late anaphase; in (B), telophase. Bars, 10 μm.Microtubule-driven cortical actin flow could be intercepted by a microneedle.From the experiments shown in Figure 2 and Figure 3A–3C, it appeared that cortical actin filaments were redistributed to the equatorial cortex by astral microtubules (or by those in a collapsed spindle), rather than assembled there, de novo. To confirm this, we decided to use a microneedle to indent a region of the cortex, distant from the presumptive source of actin. We predicted the needle would intercept part of the actin flow, resulting in the accumulation of cortical actin filaments on the side of the needle-indented cortex that faced the source of actin, i.e., the side closer to the repositioned spindle (Figure 3D, diagram). We induced microtubule-driven cortical actin flow in a microinjected anaphase cell (Figure 3D, 0, and Video S8; n = 5), and then indented but did not penetrate the cell membrane with the tip of a microneedle (Figure 3D, arrows; not observable in fluorescence channels). As predicted, the needle regionally blocked actin flow, with actin fluorescence becoming increasingly brighter on the side of the needle facing the repositioned, collapsed spindle (Figure 3D, 1.7–4.5, and Video S8). As an internal control, no obstruction of actin flow was observed at the unblocked side of the cell cortex. This experiment substantiated our idea that spindle microtubules could exclude cortical actin filaments during anaphase and early telophase, thus presumably relaxing the cell cortex they contacted.Cortical stiffness correlated with local density of cortical actin.We had been using local density of cortical actin to infer cortical stiffness, and we needed to test the validity of this assumption. To assay the degree of cortical stiffness in a region of high actin density relative to one of low actin density, cortical flow of actin filaments was induced using a repositioned, collapsed spindle to generate a density differential. After cortical flow was induced, the cortex farthest from the spindle (high actin density) was probed by depression with the side of a flexible microneedle. As predicted, the needle tip was bent by the cortex (Figure 4; n = 7). When the cortex on the opposite side of the cell (low actin density) was probed using the same needle, the cortex was deformed, while the needle remained unbent (Figure 4; same seven cells). (Similar results were obtained using grasshopper spermatocytes; unpublished data (n = 6).) Thus, we concluded that regions with a high density of cortical actin are indeed stiffer (i.e., less relaxed) than regions of low density, justifying our use of the term “polar relaxation” to describe redistribution of actin via cortical flow.Figure 4Cortical Stiffness Correlates with Local Density of Cortical Actin(A) After the spindle was repositioned and collapsed to induce cortical flow, we tested the stiffness of the actin-dense region of the cortex. (a) The needle (marked with asterisks) in its initial position, tangential to the actin-rich cortex. (b) After the needle was pushed against the cortex, the stiff cortex deflected the needle.(B) After rotating the cell in (A) by 180°, the cortical stiffness of the actin-depleted region of the cortex was tested. (a) The needle (marked with asterisks) in its initial position, tangential to actin-depleted cortex. (b) After the needle was pushed against the cortex, the relaxed cortex was deformed by the needle without deflecting it. F-actin was labeled using low-level Alexa Fluor 488 phalloidin to monitor flow. Bars, 10 μm.Evidence for Equatorial Stimulation as Gauged by De Novo Assembly of ActinDe novo assembly and delivery of actin aggregates mediated by microtubule plus ends.Two sources of actin may contribute to contractile ring assembly: pre-existing actin filaments from the cell cortex and the cytoplasm, and actin aggregates that are assembled de novo at the equator. We have demonstrated that microtubule-driven cortical flow provides a mechanism by which cortical actin is excluded from the polar cortex. (Polar microtubules may also play a role in clearing cytoplasmic actin filaments from the polar region. See legend to Video S4.) We hypothesized that microtubules might stimulate de novo actin assembly at the equator, since the speckles of actin fluorescence that were detected at equatorial microtubule plus ends increased in number and size as anaphase progressed (Figure 1B). To detect de novo actin assembly, we monitored its dynamics starting at the metaphase-anaphase transition in cells with labeled microtubules and actin filaments (Figure 5A and Video S9; n = 15). Before onset of anaphase, actin fluorescence was absent from the spindle equator where the aligned metaphase chromosomes were located (Figure 5A, 0). As the cell entered anaphase (Figure 5A, 2), speckles of actin fluorescence soon emerged at the spindle midzone where microtubule plus ends overlapped (Figure 5A, 2–4). The speckles gradually grew into bigger aggregates as anaphase proceeded (Figure 5A, 4–10, and Video S9). The de novo emergence and accumulation of actin fluorescence was readily apparent when actin alone was visualized, using the rhodamine channel (Figure 5A, insets). Notably, nascent actin aggregates were assembled across the entire midzone of the central spindle, and delivered laterally to the incipient contractile ring by microtubules (Figure 5B and Video S10; n = 13). This novel means of delivery was driven by a structural reorganization of each spindle half: microtubules became more focused at their minus ends (at the spindle poles), while splaying out toward the equatorial cortex at their plus ends, thus enlarging the spindle midzone region. These splaying microtubules were organized into bundles, some of which were coated with an actin aggregate at the plus ends of the bundle, enabling its delivery to the cortex. Lateral transport appeared to begin around the time of furrow initiation, while cortical flow was already in progress. However, we could not definitively rule out an earlier starting point, since small numbers of peripheral microtubules could have escaped detection. Moreover, it was problematic to define a precise endpoint, as the trajectory of the outwardly splaying spindle was confounded by the ingressing furrow.Figure 5Interdigitating Microtubule Plus Ends Hosted De Novo Actin Assembly and Delivered Actin Aggregates to Equator(A) Emergence (0–2) and growth (2–10) of nascent actin fluorescence at the equatorial microtubule plus-end overlap region reflect de novo assembly of actin aggregates.(B) A splaying microtubule bundle (arrowheads) delivered the actin aggregate into the ingressing furrow. Cells in (A) and (B) were labeled as in Figure 1B.(C) De novo assembly of actin aggregates was not inhibited by microtubule stabilization (compare to (A) and Figure 1B). Microtubules were labeled and stabilized by Oregon green paclitaxel, and actin filaments were labeled by rhodamine phalloidin. Insets: actin channel. Time in min. Time 0 in (A and C), immediately before anaphase onset; in (B), late anaphase. Bars, 10 μm.Polar Relaxation Was Dependent on Microtubule Dynamics, whereas Equatorial Stimulation Was Largely IndependentDe novo actin assembly was independent of microtubule dynamics.Because microtubules were involved in both actin exclusion from the polar region and de novo actin assembly at the equator, we asked whether either process required dynamic microtubules. To address this question, we repeated the de novo actin assembly experiments (as in Figure 5A) in cells whose microtubules were simultaneously labeled and stabilized with Oregon green paclitaxel (Figure 5C and Video S11; n = 11). We determined whether paclitaxel-treated cells had entered anaphase by gently probing the chromosomes with a microneedle—allowing us to assess whether homologous chromosomes were physically separable or still attached to one another. Following the onset of anaphase, chromosomes in spindles that were stabilized at the metaphase-anaphase transition could be separated from their homolog, but they could not move poleward (unpublished data). However, de novo actin assembly at the midzone of microtubule plus ends was apparent (Figure 5C, 3–10; insets show the actin channel alone; and Video S11). Emergence and accumulation of actin fluorescence at the spindle midzone appeared similar in stabilized and nonstabilized cells (compare to Figure 5A), demonstrating that aggregate assembly was independent of microtubule dynamics. However, the assembled actin aggregates remained at the midzone of the spindle stabilized at the metaphase-anaphase transition. This failure of delivery suggests that at later stages, when the central spindle is remodeled to produce the polar focusing of microtubules and the splaying of bundles, dynamic microtubules may be required.Lateral transport of de novo–assembled actin aggregates by paclitaxel-stabilized microtubules.To further test how central spindle microtubules deliver de novo–assembled actin aggregates to the cell cortex, we remodeled these actin-tipped microtubules with a microneedle to expose their plus ends to the cortex (Figure 6). After anaphase onset, a central spindle (Figure 6A, a) was collapsed by pushing the spindle poles together (Figure 6A, b). Microtubules from the collapsed spindle reorganized into two lateral bundles that extended from the centrosomes outward, toward opposite sides of the cell cortex, with each bundle containing microtubules from both original half spindles (Figure 6A, c). Over time, the lateral spindle could reorganize into a monopolar spindle (Figure 6A, d) if the chromosomes and centrosomes were held together with a microneedle, or into a giant aster, if the needle was removed (Figure 6A, e). To prevent the formation of new regions of microtubule overlap [32], and to investigate dependence on dynamics, we stabilized microtubules with Oregon green paclitaxel (Figure 6B, n = 4 slowly evolving spindles; 6C, n = 12 lateral spindles; 6D, n = 6 monopolar spindles; and 6E and 6F, n = 7 giant asters).Figure 6Redistribution of Cortical Actin, but Not Equatorial Stimulation, Was Dependent on Microtubule Dynamics(A) A schematic showing remodeling of the central spindle by micromanipulation. A post-anaphase spindle (a, green) was collapsed using a micromanipulation needle (b, gray). The manipulation created two lateral microtubule bundles, with the exposed microtubule plus ends of the bundles pointing toward opposite sides of the cortex. The interior ends of the bundles flanked chromosomes (c, blue), kinetochores (c, purple) and centrosomes (c, orange). If the collapsed spindle was brought close to the cortex and held in place by a microneedle, it reorganized into a monopolar spindle (d). If the holding needle was removed, the spindle usually reorganized into a giant aster (e).(B) Actin exclusion was inhibited when the remodeled spindle was stabilized by paclitaxel. Neither the stabilized lateral spindle (8) nor the evolving monopolar spindle (12–22) induced unidirectional flow of actin filaments (compare to Figure 3). Cells in (B–F) were labeled as in Figure 5C.(C) Actin aggregates were assembled at the plus ends of stabilized, laterally oriented microtubule bundles (0–2, arrowheads), and were delivered to the non-equatorial cortex where splaying microtubule bundles were in contact with the cortex (2–12, arrows).(D) A monopolar spindle (2) gradually splayed its microtubule bundles toward the cell cortex (2–21), which delivered actin aggregates (2, arrowheads) and induced a furrow (11–21, arrows). The furrow eventually regressed (61).(E) Tracking of an actin aggregate delivered from the plus ends of one microtubule bundle to the cell cortex. The giant aster had chromosomes in its center and microtubule plus ends pointing outward, as in (A, panel e). When the aster was placed near the cell cortex using a microneedle, an actin aggregate moved away from the microtubule bundle plus ends and merged into the cortex (box, 0–8.8).(F) The region of interest in (E) is shown with additional intervening time points to highlight the increasing cortical fluorescence caused by merging of actin aggregates. Time in min. Time 0 in (B), (D), and (E), mid to late anaphase; in (C), telophase; in (F), identical to 0 in (E). Bars, 10 μm.We mechanically positioned the remodeled and stabilized spindle so that its exposed microtubule plus ends, which were coated with the de novo–assembled actin aggregates, were removed from the original equatorial cortex and thus from any cues already present there. In all spindle configurations tested (Figure 6A, b–e), actin aggregates were delivered by the bundled microtubules to the nearby cortex. Notably, microtubule splay in a lateral spindle gradually brought microtubule plus ends close to the cell cortex, and hence allowed the delivery of actin aggregates (Figure 6C and Video S13). Furrow initiation could occur in such manipulated cells, provided the actin aggregates were delivered to a cortical zone that encircled the cell (Figure 6D, arrows). Using time-lapse imaging, we followed a giant aster positioned near the cell periphery (viewed from above, as in Figure 6A, e) and documented its delivery of an actin aggregate from the plus end of a microtubule bundle to the cell cortex (Figure 6E and 6F and Videos S14 and S15). The accumulation of actin aggregates made the cell cortex significantly brighter over time (Figure 6F, 0 onward), indicating that delivery of actin aggregates can proceed even in the absence of dynamic microtubules, so long as splay of central spindle microtubules is not blocked.In the final stage of lateral transport, actin aggregates had to be released from the tips of the microtubule bundles to allow their delivery to the cortex. To examine this process in greater detail, the red and green channels were analyzed individually for signs of costaining at the tips of microtubule bundles (in transit from the central spindle to the cortex). The signal from the microtubules at the peripheral ends of the bundles was much weaker than the signal from the actin; thus, the costained regions appeared red rather than yellow in the merged images shown in Figure 6. However, close analysis revealed a correlation between the length of the costained region of the bundle and the distance from bundle tip to cortex (Table 1). This distance was actually measured from the distal end of the actin aggregate rather than from the distal end of the bundle itself. The length of the actin aggregate remained fairly constant—approximately 0.8 to 0.9 μm (Table 1). When a bundle was more than about 1.2 μm from the cortex, at least a portion of the aggregate (perhaps close to the entire aggregate) overlapped with the tip of the bundle. As the bundle, with its associated cargo, came within about 0.41–1.2 μm of the cortex, it was no longer possible for the entire length of the aggregate to fully overlap with the bundle, even taking into account the uncertainty of the measurements. In other words, the aggregate appeared to be “sliding” off the tip of the bundle, toward the cortex. Finally, as the actin-tipped bundle approached within 0.4 μm of the cortex, i.e., closer than the length of the aggregate itself, the aggregate was released completely from the bundle, and the overlap dropped to zero.Table 1Measurements of Actin Aggregate-Tipped Single BundlesCortical actin flow was dependent on microtubule dynamics.In contrast to the de novo assembly of actin aggregates, cortical flow of actin was highly dependent on dynamic microtubules. We attempted to induce cortical flow in paclitaxel-treated cells by collapsing the spindle with a microneedle while repositioning it close to the cortex (Figure 6B–6D, Video S12, and see also Figure 5C, 3–10). Cortical flow was clearly inhibited. This result supports our contention that cortical actin filaments are excluded from the polar region and redirected toward the equatorial cortex by dynamic microtubules.Equatorial Stimulation, but Not Polar Relaxation, Was Dependent on RhoA ActivityWe wanted to determine whether de novo assembly of actin aggregates at the plus ends of central spindle microtubules requires RhoA as a regulator. To do so, we microinjected late metaphase cells with C3 ribosyltransferase, a botulinum toxin that specifically inhibits Rho by ADP ribosylation [33]. We hypothesized that this inhibition would disrupt de novo actin assembly at the spindle midzone, without affecting cortical flow of actin. As expected, the C3 transferase-treated cell continued to divide but failed to accumulate actin aggregates in the central spindle (Figure 7). As seen in optical sections of control cells (n = 15), actin accumulated both at the equatorial cortex (Figure 7A, top and bottom) and at microtubule plus ends within the central spindle (Figure 7A, middle). In contrast, C3 transferase-treated cells (n = 9) accumulated actin at the equatorial cortex (Figure 7B, top and bottom), but not within the central spindle (Figure 7B, middle). This result implies that RhoA inactivation inhibits actin assembly at the plus ends of central spindle microtubules (i.e., the initial step of equatorial stimulation), but does not inhibit relocation of cortical actin from the poles to the equatorial cortex (i.e., polar relaxation).Figure 7Equatorial Stimulation, but Not Redistribution of Cortical Actin, Was Dependent on RhoA Activity(A) A cytokinetic control cell, optically sectioned lengthwise, displayed actin aggregates at the microtubule plus ends of its spindle, both at the equatorial cortex (top and bottom) and in the central spindle (middle). Cells in (A and B) were labeled as in Figure 1B.(B) In a cytokinetic cell microinjected with C3 transferase to inhibit RhoA, actin aggregates at the microtubule plus ends of the central spindle were absent.(C) Despite inhibition of RhoA, actin flow was induced (2.5–11), following manipulation of the collapsed spindle to one side of the cell (2.5). Excluded filaments assembled into a contractile ring (35). Cell was labeled as in Figure 1B, and microinjected with C3 transferase.(D) In a cell treated with paclitaxel and C3 transferase, the furrow failed to initiate due to inhibition of both pathways for redistribution of actin. Actin filaments were scattered in the cytoplasm long after the cell entered anaphase, as determined by the presence of splaying microtubules (17–73). (D) was labeled as in Figure 5C. Time in min. Time 0 in (C), late anaphase; in (D), metaphase-anaphase transition. Bars, 10 μm.Additional tests confirmed that in C3 transferase-treated cells, actin filaments were still excluded from the cortex adjacent to the dislocated spindle and incorporated into the contractile ring (Figure 7C, n = 8). However, when late metaphase cells were microinjected with C3 transferase, and subsequently treated with paclitaxel at anaphase onset, both pathways for actin redistribution were affected (Figure 7D, n = 5); specifically, delivery to the equatorial cortex from both the polar cortex and the midzone was impeded. Cells failed to divide long after entering anaphase (Figure 7D, 0–73 min), due to inhibition of both de novo assembly of nascent actin aggregates, and relocation of cortical actin filaments to the contractile ring.DiscussionPolar Relaxation and Equatorial Stimulation Both Induce Contractile Ring Assembly and Cell CleavageBy mechanically manipulating the spindles of double-labeled silkworm spermatocytes, we show how polar relaxation (as gauged by redistribution of cortical actin) and equatorial stimulation of the cortex (as gauged by de novo actin assembly and transport) both contribute to contractile ring assembly and furrow induction. We demonstrate that spindle microtubules deliver both inhibitory and stimulatory signals to the cell cortex during furrow formation, depending on their configuration and location within the cell. The dynamic astral microtubules drive actin filaments from the polar region to the equatorial cortex, which relaxes the polar cortex, while increasing cortical tension at the actin-dense equator. Meanwhile, bundles of interdigitating microtubules across the entire central spindle stimulate de novo assembly of actin aggregates at their overlapping plus ends, and deliver the aggregates to the equatorial cortex. These dual signaling mechanisms ensure that both cortical actin and central spindle-associated actin are delivered to the equatorial cortex, resulting in a robust contractile ring, and providing redundancy to safeguard an essential cellular process. Videos S4 and S9, as well as previous studies [34,35], provide hints that cytoplasmic actin filaments (perhaps vesicle bound; [29]) may constitute a third source of actin that could ultimately be delivered to the equatorial cortex by microtubules.We show that in the absence of equatorial stimulation, contractile ring assembly and furrowing can still be induced by means of actin redistribution driven by cortical flow (Figure 3A and 3B and Figure 7C). We accomplished this by displacing the entire spindle apparatus, bearing “stimulatory” furrow cues that were reorganized (via spindle collapse), or C3 transferase-suppressed, to an arbitrary region of the cortex. The existence of a cortical flow of actin filaments away from the displaced spindle microtubules was established by our ability to intercept the flow locally, using a microneedle (Figure 3D). We found that the flow of cortical actin during cytokinesis can be induced by any microtubules positioned near the cortex, provided their plus ends are dynamic and accessible (i.e., neither interdigitating nor otherwise bundled or stabilized). We hypothesize that collapsing the spindle can mechanically free a subset of plus ends; this could allow those formerly interdigitating central spindle microtubules to become more dynamic and to take on properties held by astral microtubules, such as the ability to contact the cell cortex and induce cortical flow of actin filaments. Interdigitating microtubules in an intact central spindle would be able to contact the cortex only laterally, and as incorrectly oriented bundles, which should not be conducive for promoting cortical flow. As predicted by this finding, furrow formation can occur between a collapsed asterless spindle and its two detached asters in a remodeled cell (Figure 3C). We have also shown that this microtubule-driven actin flow begins shortly after anaphase onset and persists through telophase (Figure 3A and 3B), contributing to both contractile ring assembly at the equator and furrow inhibition at the poles throughout the furrow induction and ingression stages. Cortical flow during cytokinesis has previously been inferred or observed for contractile elements including myosin II [28,36,37] and pre-existing actin filaments [17,35,38], as well as for a membrane-bound receptor-ligand complex, membrane domains, and cell surface proteins [39–41]. As the movement of certain surface proteins mirrors that of cortical actin, it has been suggested that the actin network could serve as a scaffold for these proteins, thus engineering their co-transport to the furrow region [41].How might microtubules interact with actin filaments to effect their exclusion from the poles? Since paclitaxel inhibits cortical flow (Figure 6B–6D), actin filaments are likely to be driven from the polar region towards the equatorial cortex by elongating microtubules. Following anaphase onset, the plus ends of astral microtubules can elongate toward and contact the cortex [20,42]. Such microtubules would be anchored to the spindle pole at their minus ends, and conceivably constrained at their plus ends by indirect attachment to a mobile cortical component, namely filamentous actin (analogous to cortical capture via Bim1/Kar9/Myo2 [43,44]). Given the microtubule's length, flexibility, geometrical orientation within the cell, and constraints on its direction of movement, its elongation may be sufficient to drive cortical actin toward the equator. Elongation of tethered microtubules against a fixed barrier can produce force of up to 10 pN, as gauged by microtubule buckling [45]. Presumably, this force could be harnessed to allow tip-attached cortical actin to “hitchhike” to the equatorial cortex, where it would be released, possibly by binding to myosin II [38] or anillin [46]. Alternatively, or in addition, actin filaments could be transported along the subset of microtubules that are contiguous to the cortex, and longitudinally aligned, with appropriate polarity. Motor proteins would provide the driving force for transport, with microtubule elongation serving only to extend the track toward the equator. Actin filaments could either be cargo for a plus-end–directed kinesin (attached via an adaptor protein), or else the actin could be coupled to the microtubule track via kinesin/myosinV-type hetero-motors (as modeled in [44,47]), and slide along the stationary microtubule toward the equatorial cortex, with kinesin providing directionality for the flow. The tracks could consist of the cortically apposed section of centrosome-attached astral microtubules and/or those released from the centrosomes [42,48–50] that become cortically apposed [34,47,51,52].We also show that actin aggregates can be assembled de novo within the central spindle in the absence of cortical-flow based redistribution of actin, by using paclitaxel to impede the flow. Similarly, paclitaxel-stabilized microtubules can induce a cleavage furrow in mammalian cells [53], perhaps implying mechanistic conservation. We demonstrate that cortical stimulation can be induced at any region in the cortex by microtubule plus ends of the central spindle (Figure 6). Our data reveal that de novo–assembled cytoplasmic actin aggregates can comprise a supply of actin sufficient to form the contractile ring and induce cell cleavage (Figure 6D). Newly assembled actin aggregates are precursors of the contractile ring, since they ultimately merge with the ring (Figure 5). The existence of de novo assembly of actin during contractile ring formation is consistent with evidence from diverse cell types, including grasshopper spermatocytes (unpublished data), yeast [54,55], Xenopus eggs [56], and mammalian cells [17,38]; in addition, there is precedent for the presence of actin nucleating protein formin at microtubule plus ends [57]. Assembly of actin aggregates is sensitive to the Rho GTPase inhibitor, C3 transferase (Figure 6B, middle), supporting the existence of a microtubule-dependent zone of active RhoA during cleavage plane specification [5,23,58–62]. Our findings also raise the possibility that equatorial stimulation signals could be transported along stationary central spindle microtubule tracks from the spindle pole region to the midzone (Figure 5B), before ultimately reaching the equatorial cortex, perhaps via lateral transport.Actin aggregates are assembled across the entire spindle midzone (Figure 5A and 5C); hence they must be transported laterally from the plus ends of the bundled microtubules to the equatorial cortex. We have captured this novel means of delivery using time-lapse sequences (Figures 5B, 6E, and 6F). Bundles of central spindle microtubules, some tipped with actin aggregates, splay laterally outward as the spindle changes shape, thereby conveying their cargo to the incipient furrow (Figure 5B). The actin cargo remains associated with the tips of the microtubule bundles until the bundles are within about 0.4 μm of the cortex, less than the length of the aggregate itself (Table 1). At this point, the microtubule-associated aggregates appear to be released from the microtubule bundles, in that the costaining at the microtubule tips disappears. The dissociation of the actin from the bundles may be mediated by kinesins bound to the bundle, as the aggregate appears to slide off the end of the bundle as it approaches the cortex (Table 1). This transport of actin aggregates may be largely independent of microtubule dynamics; when a post-anaphase, paclitaxel-stabilized spindle with mechanically exposed, bundled plus ends is brought close to the cortex by micromanipulation, it remains functional for delivery of actin aggregates to the cortex (Figure 6E and 6F). The induction of an ectopic furrow following cortical deposition of actin aggregates from microtubule plus ends (Figure 6D) indicates that actin-tipped microtubules may contain sufficient furrow constituents (i.e., structural constituents along with any co-transported stimulatory cues) to permit contractile ring assembly. This idea is consistent with work by Hu et al., showing that the tips of monopolar spindles contain critical stimulatory components (e.g., the Rho GEF Ect2) [21].We propose that in the bipolar spindle, other spatially dynamic midzone components may also utilize this microtubule-based lateral transport mechanism—either co-transported with actin aggregates, or else delivered via a distinct set of midzone microtubules. By ensuring timely relocation of cytokinetic components, lateral transport could play a role in processes such as midzone signaling of the equatorial cortex, assembly and maintenance of the contractile ring and abscission. For example, the midzone component centralspindlin becomes localized to the equatorial cortex during anaphase. Bundled interdigitating microtubules, tipped with centralspindlin, may become docked at the contractile ring via Anillin [46,63]. We suggest that these tipped microtubules may originate from the central spindle and deliver their cargo to the cortex via lateral transport. Similarly, lateral transport could provide a means of delivery to the cortex for a recently characterized subset of vesicles. These vesicles are associated both with actin and with the central spindle, and are thought to contribute to furrow ingression [29]. In fixed cells, the authors saw thin microtubule bundles potentially connecting the central spindle with the periphery [29], consistent with a lateral transport mechanism. As another example, by early anaphase, the chromosomal passenger complex (CPC) is at the mitotic spindle midzone, whereas later in anaphase it is also present at the equatorial cortex (reviewed in [64]), perhaps arriving there by lateral transport. Interestingly, in Dictyostelium, the CPC component INCENP has been shown to bind actin [65]. It is also worth determining whether a subset of RhoA may reach the equatorial cortex via lateral transport, given that RhoA localization may depend on microtubule organization [58].Model for Microtubule Induction of Contractile Ring AssemblyOur results explain how a single spindle component, the microtubule, can play opposing roles in two complementary pathways of contractile ring assembly. The seemingly incongruous stimulatory and inhibitory effects on the cortex are produced by different configurations of microtubules that populate different regions of the cytokinetic cell. We propose a microtubule induction model to describe how microtubules perform their dual signaling at the poles and the equator to induce contractile ring formation and cleavage furrow initiation. During early anaphase, dynamic astral microtubules at the poles inhibit ectopic furrow formation by excluding pre-existing actin filaments from the poles (Figure 8A, red dotted arrows and red lines at cortex), ultimately resulting in actin accumulation at the incipient equatorial furrow. Meanwhile, at the plus ends of the relatively stable central spindle microtubules, actin aggregates are assembled de novo, and enlarge over time (Figure 8A, red lines in midzone). As the cell progresses toward telophase, preexisting cortical actin filaments continue to flow from the polar cortex to the equator (Figure 8B, red dotted arrows). At the same time, the plus ends of bundled central spindle microtubules splay laterally to deliver the actin aggregates to the equatorial cortex (Figure 8B, red solid arrows). The newly delivered aggregates coalesce with the actin excluded from the polar cortex to assemble the contractile ring (Figure 8C).Figure 8Model for Spindle Microtubule Induction of Contractile Ring Assembly(A) During furrow induction and ingression, dynamic astral microtubules exclude pre-existing actin filaments from the spindle pole (red dotted arrows).(B) This exclusion results in mass migration of filaments to the equator. Meanwhile, overlapping spindle microtubule plus ends at the equator promote de novo assembly of actin aggregates, which are delivered (small red arrows) to the equatorial cortex by splaying bundles of central spindle microtubules.(C) Actin filaments excluded from the polar region coalesce with actin aggregates transported from the central spindle to the equatorial cortex, for assembly of the contractile ring.In summary, we show that actin is redistributed into the contractile ring via two microtubule-dependent pathways that coexist during cytokinesis of the silkworm spermatocyte: equatorial stimulation and polar relaxation. These dual signaling pathways ensure the fidelity of cytokinesis, which fails only if both mechanisms are inhibited (Figure 7D). With some modification, our model could be generalized to multiple cell types. For example, cytokinetic cells of varying sizes and spindle geometries might rely more heavily on one mechanism than the other to achieve a global balance between tension and relaxation. We have demonstrated that the entire cell cortex is potentially responsive to both stimulatory and inhibitory cues from microtubules during cytokinesis in silkworm spermatocytes. How might the cell restrict the contractile ring to a localized band of equatorial cortex? By stimulating only the equatorial cortex while concomitantly relaxing other regions of the cortex, a cell could ensure cytokinetic fidelity by building in redundancy. Polar relaxation could be thought to include a component of equatorial stimulation in that pre-existing actin filaments are transported to the equatorial cortex. Equatorial stimulation, with its de novo actin assembly, recruits cytoplasmic G-actin to the equator. Hence, the seemingly disparate models ultimately result in redistribution of all available actin resources to the equator for ring assembly.Silkworm as a Model OrganismOur work with spermatocytes of the silkworm B. mori has enabled us to document a number of novel events essential for understanding contractile ring assembly. These spermatocytes are highly amenable to mechanical manipulations and imaging techniques, making them ideal for cytokinetic studies. Equally important, the silkworm genome has been sequenced [66], paving the way for molecular and genetic studies, such as RNAi inhibition of particular genes. Thus, we believe the silkworm spermatocyte has great potential as a model system for research in cell biology, cell physiology, and developmental biology—given its innate attributes coupled with the power of molecular genetics.MethodsPrimary cell culture.The spermatocytes of silkworm Bombyx mori were isolated from the testes of a 5th instar larva and spread under inert halocarbon oil (400 oil, Halocarbon Products) to form a monolayer of cells in a glass chamber slide. Only cells in the first meiotic division were used.Microscopy.Cells were observed with an inverted Zeiss Axiovert 135 microscope modified for both digital-enhanced polarization [6] and spinning disc confocal microscopy (CARV, Carl Zeiss). The microscope is equipped with a 1.4 NA achromatic-aplanatic condenser and a 1.45 NA/100X α-Plan objective (Carl Zeiss). An EM-CCD digital camera (Hamamatsu C9100–12), Simple PCI software (C-image), and Photoshop software (Adobe Systems) were used to record and process images. All images in a given category “n” are fully representative of all unshown data in that category.Micromanipulation.Micromanipulation needles were pulled from glass tubing (outer diameter: 1.0 mm; inner diameter: 0.58 mm, World Precision Instruments) as described [67] using a microforge (Narishige, Model MF-830) and maneuvered with a Burleigh MIS-5000 series piezoelectric micromanipulator [6].Microinjection and preparation of injectants.Glass tubing (outer diameter: 1.0 mm; inner diameter: 0.75 mm) with an internal capillary (World Precision Instruments) was pulled on a Flaming/Brown P-87 micropipette puller (Sutter Instrument Company) to produce micropipettes with a tip diameter ∼ 0.1 μm. The injectant was back loaded into the micropipette using a Hamilton syringe. Microinjection was conducted at 60 psi using a custom-built pneumatic injector maneuvered with a Burleigh MIS-5000 series piezoelectric micromanipulator.\nAlexa Fluor 568 tubulin. Alexa Fluor 568 (Invitrogen) was conjugated to porcine tubulin (Cytoskeleton) following the protocol adapted from Peloquin et al. [68]. The labeled tubulin was resuspended in an injection buffer (20 mM PIPES, 0.25 mM MgSO4, 1 mM EGTA, 1 mM GTP, pH 6.8) to a final concentration of 3 mg/ml for microinjection.\nPhalloidin. Alexa Fluor 488 phalloidin or rhodamine phalloidin (Invitrogen) in methanol was concentrated using a SpeedVac concentrator (Savant) and resuspended in the injection buffer to a final concentration of ∼6.6 μM for microinjection. Although the intracellular concentration of phalloidin was low, controls were performed to ensure that the phalloidin did not adversely affect the function of the actin filaments. The timing of cell division was not slowed down in a phalloidin-treated cell relative to uninjected controls (as in Table S1).\nC3 transferase. C3 transferase (Cytoskeleton) was stored as 1 mg/ml aliquots at −70 °C in a buffer containing 500 mM imidazole (pH 7.5), 50 mM Tris HCl (pH 7.5), 1.0 mM MgCl2, 200 mM NaCl, 5% sucrose, and 1% dextran. Immediately before microinjection, the C3 transferase was mixed with Alexa Fluor 568 tubulin and Alexa Fluor 488 phalloidin to a final concentration of 0.5 mg/ml in the micropipette.Live cell labeling with the tubulin tracker and Hoechst stain.Tubulin Tracker (Invitrogen) was used as directed, with the Oregon Green 488 Taxol (paclitaxel) diluted to 50 μM in insect Ringer's solution. The diluted Tubulin Tracker was micropipetted around the target cells, resulting in a further dilution of at least 1000 fold. Hoechst 33342 (Invitrogen) was stored as 10 mg/ml stock aliquots at −20 °C and diluted in Tubulin Tracker buffer to 0.5 mg/ml before micropipetting. Immunofluorescence microscopy.Silkworm spermatocytes were fixed and stained as previously described [6]. Microtubules were stained with tubulin primary antibody (Chemicon) and Alexa Fluor 488 conjugated secondary antibody (Invitrogen); actin filaments, with rhodamine phalloidin (Invitrogen); and chromosomes, with DAPI.Labeling color scheme.In all figures, microtubules are shown in green, and actin in red (false colored, if necessary). Chromosomes, when labeled, are blue (Hoechst for live cells, DAPI for fixed).Measurement of cortical stiffness.The degree of stiffness at one region of the spermatocyte cortex relative to another region was assayed as follows. The spindle of a cell in anaphase was collapsed (by pushing the spindle poles together) and mechanically repositioned near the cortex to induce cortical flow of actin filaments. After waiting 5–10 min to allow for redistribution of actin, the first region of interest was probed with the side of a flexible microneedle, tangential to the cortex. Depending on the flexural rigidity of the needle and the local cortical stiffness, a given needle would either deform the cortex by flattening it, without deflecting the needle, or the cortex would bend the needle. The cell was repositioned (by rotating the microscope slide by 180°) to bring the second region of interest in proximity with the needle, and then reprobed.Supporting InformationTable S1Summary of Experimental Design and Results(230 KB DOC)Click here for additional data file.Video S1Cytokinesis of a Silkworm Spermatocyte as Seen by Polarization MicroscopyThe birefringence of spindle microtubules is shown with dark compensation. Video S1 corresponds to Figure 1A.(8.66 MB AVI)Click here for additional data file.Video S2Cytokinesis of a Silkworm Spermatocyte with Labeled Actin and Microtubules as Seen by Confocal MicroscopyConfocal time-lapse imaging of the cytokinesis of a silkworm spermatocyte microinjected with rhodamine tubulin (microtubules false colored green) and low-level Alexa Fluor 488 phalloidin (actin false colored red). Actin aggregates appeared at the microtubule plus ends during early anaphase, and then fused with the contractile ring that bisected the cell. Video S2 corresponds to Figure 1B.(5.75 MB AVI)Click here for additional data file.Video S3Cortical Flow of Actin Induced by Asymmetrically Distributed Asters during AnaphaseCortical actin filaments were excluded from the polar region occupied by asymmetrically distributed asters in early anaphase. Cells in Video S3 labeled as in Video S2. Video S3 corresponds to Figure 2A.(1.28 MB AVI)Click here for additional data file.Video S4Cortical Flow of Actin Induced by Asymmetrically Distributed Asters Results in Furrow Initiation during AnaphaseCortical actin filaments were excluded from the polar region occupied by asymmetrically distributed asters during anaphase. The excluded actin assembled a contractile ring around the equator of the naturally shifted spindle. The two asters were in different focal planes. In frames ∼1–13 (especially frames 6, 7, 9, and 10), actin filaments appeared to be radiating out from the microtubule aster, parallel to and possibly overlapping the astral microtubules. We hypothesize that this may be part of a mechanism by which cytoplasmic actin filaments are cleared from the polar region. These actin asters can also be seen in Video S9. Cells in Video S4 labeled as in Video S2. Video S4 corresponds to Figure 2B.(3.15 MB AVI)Click here for additional data file.Video S5Cortical Flow of Actin Driven by Microtubules of a Collapsed Spindle during AnaphaseThe redistribution of actin filaments resulted in asymmetric cell division. Cells in Video S5 labeled as in Video S2. Video S5 corresponds to Figure 3A.(1.95 MB AVI)Click here for additional data file.Video S6Cortical Flow of Actin Driven by Microtubules of a Collapsed Spindle during TelophaseThe redistribution of actin filaments resulted in asymmetric cell division. Cells in Video S6 labeled as in Video S2. Video S6 corresponds to Figure 3B.(2.99 MB AVI)Click here for additional data file.Video S7A Contractile Ring Formed between a Scrambled Spindle and Two AstersWe hypothesize that ring formation was due to actin exclusion by microtubules from both structures. Cell in Video S7 labeled as in Video S2. Video S7 corresponds to Figure 3C.(2.04 MB AVI)Click here for additional data file.Video S8Induced Cortical Flow of Actin Could Be Intercepted Using a MicroneedleThe spindle was collapsed and pushed close to the cell's upper cortex to induce actin flow. The microneedle (arrow on first frame), which indented but did not pierce the plasma membrane, locally intercepted the actin flow, with brighter fluorescence accumulating on the side of the needle facing the repositioned spindle. Cell in Video S8 labeled as in Video S2. Video S8 corresponds to Figure 3D.(2.82 MB AVI)Click here for additional data file.Video S9De Novo Assembly of Actin Aggregates at Overlapping Microtubule Plus Ends in the Central SpindleCell in Video S9 labeled as in Video S2. Video S9 corresponds to Figure 5A.(3.33 MB AVI)Click here for additional data file.Video S10Delivery of Actin Aggregates into the Ingressing Furrow by a Splaying Microtubule BundleThe splaying microtubule bundle is marked by an arrow on the first frame. Cell in Video S10 labeled as in Video S2. Video S10 corresponds to Figure 5B.(1.97 MB AVI)Click here for additional data file.Video S11De Novo Assembly of Actin Aggregates Was Not Dependent on Microtubule DynamicsCompare Video S11 to Videos S2 and S9. Microtubules were labeled and stabilized by Oregon green paclitaxel (green), and actin filaments were labeled by rhodamine phalloidin (red). Video S11 corresponds to Figure 5C.(1.86 MB AVI)Click here for additional data file.Video S12Cortical Flow of Actin Was Dependent on Microtubule DynamicsNeither the stabilized lateral spindle nor the evolving monopolar spindle induced unidirectional flow of actin filaments (compare to Videos S5–S8). Cell in Video S12 labeled as in Video S11. Video S12 corresponds to Figure 6B.(7.03 MB AVI)Click here for additional data file.Video S13Delivery of Actin Aggregates from the Tips of the Stabilized Microtubule Bundles to the CortexActin aggregates were assembled at the plus ends of stabilized microtubule bundles from a lateral spindle, and were delivered to the non-equatorial cortex where splaying microtubule bundles were in contact with the cortex (cell's lower left cortex). Cell in Video S13 labeled as in Video S11. Video S13 corresponds to Figure 6C.(5.21 MB AVI)Click here for additional data file.Video S14Delivery of an Actin Aggregate from the Tip of a Microtubule Bundle to the CortexThe microtubule bundle is boxed in the first frame of the video. The remodeled spindle had chromosomes in its center and microtubule plus ends pointing outward. Cell in Video S14 labeled as in Video S11. Video S14 corresponds to Figure 6E.(1.62 MB AVI)Click here for additional data file.Video S15A 3D View of a Giant Aster, Showing Actin-Tipped BundlesA stack of 2D images of the cell in Figure 6E was reprocessed to create a 3D image, by rotation around the y-axis. Microtubules (green) were stabilized and labeled with Oregon Green Paclitaxel, and actin (red) was labeled using low-level rhodamine phalloidin. The giant aster developed from a post anaphase spindle, which was collapsed by a microneedle. Microtubule bundles in the aster were oriented with their plus ends closest to the cortex, and many of them were tipped with actin aggregates. A subset of these actin aggregates were ultimately delivered to and merged into the cortex after the aster was maneuvered to a location near the cortex. Video S15 corresponds to Figure 6E.(924 KB AVI)Click here for additional data file.Video S16A 3D View of a Bipolar Spindle, Showing Overlapping Actin-Tipped Bundles in the Central SpindleA stack of 2D images of an anaphase cell (at a stage similar to that of the cell in Figure 1B) was reprocessed to create a 3D image, by rotation first around the y-axis, and then around the x-axis. Microtubules (green) were stabilized and labeled with Oregon Green Paclitaxel, actin (red) was labeled by low-level rhodamine phalloidin, and the nucleus (blue) was stained with Hoechst. Microtubule bundles from the two half spindles radiated outwards to the equatorial cortex, displaying tips coated with actin aggregates. Such actin coated microtubule bundles were absent on the side of the cell facing the coverslip (see the vertical rotation). Presumably because the cell was flatter on that side, the bundles had to travel a shorter distance to deliver their cargo; hence, they may have reached the cortex sooner than the bundles on the opposite side. In this cell, the chromosomes were already segregated to the poles, and some of the actin aggregates had already been delivered to the equatorial cortex by the microtubule plus ends; thus, actin aggregates could no longer be seen in the center of the spindle. Two pairs of centrosomes at the poles of the cell were clearly visible.(2.43 MB AVI)Click here for additional data file.\n\nREFERENCES:\n1. D'AvinoPPSavoianMSGloverDM\n2005\nCleavage furrow formation and ingression during animal cytokinesis: a microtubule legacy\nJ Cell Sci\n118\n1549\n1558\n15811947\n2. BurgessDRChangF\n2005\nSite selection for the cleavage furrow at cytokinesis\nTrends Cell Biol\n15\n156\n162\n15752979\n3. EggertUSMitchisonTJFieldCM\n2006\nAnimal cytokinesis: from parts list to mechanisms\nAnnu Rev Biochem\n75\n543\n566\n16756502\n4. GlotzerM\n2004\nCleavage furrow positioning\nJ Cell Biol\n164\n347\n351\n14757750\n5. PieknyAWernerMGlotzerM\n2005\nCytokinesis: welcome to the Rho zone\nTrends Cell Biol\n15\n651\n658\n16243528\n6. AlsopGBZhangD\n2003\nMicrotubules are the only structural constituent of the spindle apparatus required for induction of cell cleavage\nJ Cell Biol\n162\n383\n390\n12900392\n7. CanmanJCWellsWA\n2004\nRappaport furrows on our minds: the ASCB Cytokinesis Meeting Burlington, VT July 22–25, 2004\nJ Cell Biol\n166\n943\n948\n15490536\n8. WolpertL\n1960\nThe mechanics and mechanism of cleavage\nInt Rev Cytol\n10\n163\n216\n9. WhiteJGBorisyGG\n1983\nOn the mechanisms of cytokinesis in animal cells\nJ Theor Biol\n101\n289\n316\n6683772\n10. YoshigakiT\n1999\nSimulation of density gradients of astral microtubules at cell surface in cytokinesis of sea urchin eggs\nJ Theor Biol\n196\n211\n224\n10049616\n11. HiramotoY\n1990\nMechanical properties of the cortex before and during cleavage\nAnn N Y Acad Sci\n582\n22\n30\n2192596\n12. SchroederTE\n1981\nThe origin of cleavage forces in dividing eggs. A mechanism in two steps\nExp Cell Res\n134\n231\n240\n7195820\n13. CanmanJCHoffmanDBSalmonED\n2000\nThe role of pre- and post-anaphase microtubules in the cytokinesis phase of the cell cycle\nCurr Biol\n10\n611\n614\n10837228\n14. HirdSNWhiteJG\n1993\nCortical and cytoplasmic flow polarity in early embryonic cells of Caenorhabditis elegans\nJ Cell Biol\n121\n1343\n1355\n8509454\n15. KurzTPintardLWillisJHHamillDRGonczyP\n2002\nCytoskeletal regulation by the Nedd8 ubiquitin-like protein modification pathway\nScience\n295\n1294\n1298\n11847342\n16. KrendelMZenkeFTBokochGM\n2002\nNucleotide exchange factor GEF-H1 mediates cross-talk between microtubules and the actin cytoskeleton\nNat Cell Biol\n4\n294\n301\n11912491\n17. ZhouMWangYL\n2008\nDistinct pathways for the early recruitment of myosin II and actin to the cytokinetic furrow\nMol Biol Cell\n19\n318\n326\n17959823\n18. RappaportR\n1961\nExperiments concerning the cleavage stimulus in sand dollar eggs\nJ Exp Zool\n148\n81\n89\n14490383\n19. RiederCLKhodjakovAPaliulisLVFortierTMColeRW\n1997\nMitosis in vertebrate somatic cells with two spindles: implications for the metaphase/anaphase transition checkpoint and cleavage\nProc Natl Acad Sci U S A\n94\n5107\n5112\n9144198\n20. CanmanJCCameronLAMaddoxPSStraightATirnauerJS\n2003\nDetermining the position of the cell division plane\nNature\n424\n1074\n1078\n12904818\n21. HuCKCoughlinMFieldCMMitchisonTJ\n2008\nCell polarization during monopolar cytokinesis\nJ Cell Biol\n181\n195\n202\n18411311\n22. CaoLGWangYL\n1996\nSignals from the spindle midzone are required for the stimulation of cytokinesis in cultured epithelial cells\nMol Biol Cell\n7\n225\n232\n8688554\n23. BementWMBeninkHAvon DassowG\n2005\nA microtubule-dependent zone of active RhoA during cleavage plane specification\nJ Cell Biol\n170\n91\n101\n15998801\n24. SaintRSomersWG\n2003\nAnimal cell division: a fellowship of the double ring\nJ Cell Sci\n116\n4277\n4281\n14514883\n25. BringmannHCowanCRKongJHymanAA\n2007\nLET-99, GOA-1/GPA-16, and GPR-1/2 are required for aster-positioned cytokinesis\nCurr Biol\n17\n185\n191\n17189697\n26. BringmannHHymanAA\n2005\nA cytokinesis furrow is positioned by two consecutive signals\nNature\n436\n731\n734\n16079852\n27. MotegiFVelardeNVPianoFSugimotoA\n2006\nTwo phases of astral microtubule activity during cytokinesis in C. elegans embryos\nDev Cell\n10\n509\n520\n16580995\n28. WernerMMunroEGlotzerM\n2007\nAstral signals spatially bias cortical myosin recruitment to break symmetry and promote cytokinesis\nCurr Biol\n17\n1286\n1297\n17669650\n29. AlbertsonRCaoJHsiehTSSullivanW\n2008\nVesicles and actin are targeted to the cleavage furrow via furrow microtubules and the central spindle\nJ Cell Biol\n181\n777\n790\n18504302\n30. BonaccorsiSGiansantiMGGattiM\n1998\nSpindle self-organization and cytokinesis during male meiosis in asterless mutants of Drosophila melanogaster\nJ Cell Biol\n142\n751\n761\n9700163\n31. YamashikiNKawamuraN\n1998\nBehavior of centrioles during meiosis in the male silkworm, Bombyx mori (Lepidoptera)\nDev Growth Differ\n40\n619\n630\n9865972\n32. AlsopGBZhangD\n2004\nMicrotubules continuously dictate distribution of actin filaments and positioning of cell cleavage in grasshopper spermatocytes\nJ Cell Sci\n117\n1591\n1602\n15020685\n33. KishiKSasakiTKurodaSItohTTakaiY\n1993\nRegulation of cytoplasmic division of Xenopus embryo by rho p21 and its inhibitory GDP/GTP exchange protein (rho GDI)\nJ Cell Biol\n120\n1187\n1195\n8436590\n34. FoeVEFieldCMOdellGM\n2000\nMicrotubules and mitotic cycle phase modulate spatiotemporal distributions of F-actin and myosin II in Drosophila syncytial blastoderm embryos\nDevelopment\n127\n1767\n1787\n10751167\n35. CaoLGWangYL\n1990\nMechanism of the formation of contractile ring in dividing cultured animal cells. II. Cortical movement of microinjected actin filaments\nJ Cell Biol\n111\n1905\n1911\n2229180\n36. DeBiasioRLLaRoccaGMPostPLTaylorDL\n1996\nMyosin II transport, organization, and phosphorylation: evidence for cortical flow/solation-contraction coupling during cytokinesis and cell locomotion\nMol Biol Cell\n7\n1259\n1282\n8856669\n37. YumuraS\n2001\nMyosin II dynamics and cortical flow during contractile ring formation in Dictyostelium cells\nJ Cell Biol\n154\n137\n146\n11448996\n38. MurthyKWadsworthP\n2005\nMyosin-II-dependent localization and dynamics of F-actin during cytokinesis\nCurr Biol\n15\n724\n731\n15854904\n39. KoppelDEOliverJMBerlinRD\n1982\nSurface functions during mitosis. III. Quantitative analysis of ligand-receptor movement into the cleavage furrow: diffusion vs. flow\nJ Cell Biol\n93\n950\n960\n7119007\n40. NgMMChangFBurgessDR\n2005\nMovement of membrane domains and requirement of membrane signaling molecules for cytokinesis\nDev Cell\n9\n781\n790\n16326390\n41. BauerTMotosugiNMiuraKSabeHHiiragiT\n2008\nDynamic rearrangement of surface proteins is essential for cytokinesis\nGenesis\n46\n152\n162\n18327789\n42. RusanNMWadsworthP\n2005\nCentrosome fragments and microtubules are transported asymmetrically away from division plane in anaphase\nJ Cell Biol\n168\n21\n28\n15631988\n43. HuismanSMSegalM\n2005\nCortical capture of microtubules and spindle polarity in budding yeast - where's the catch\nJ Cell Sci\n118\n463\n471\n15673685\n44. RodriguezOCSchaeferAWMandatoCAForscherPBementWM\n2003\nConserved microtubule-actin interactions in cell movement and morphogenesis\nNat Cell Biol\n5\n599\n609\n12833063\n45. JansonMEde DoodMEDogteromM\n2003\nDynamic instability of microtubules is regulated by force\nJ Cell Biol\n161\n1029\n1034\n12821641\n46. GregorySLEbrahimiSMilvertonJJonesWMBejsovecA\n2008\nCell division requires a direct link between microtubule-bound RacGAP and Anillin in the contractile ring\nCurr Biol\n18\n25\n29\n18158242\n47. GoodeBLDrubinDGBarnesG\n2000\nFunctional cooperation between the microtubule and actin cytoskeletons\nCurr Opin Cell Biol\n12\n63\n71\n10679357\n48. KeatingTJPeloquinJGRodionovVIMomcilovicDBorisyGG\n1997\nMicrotubule release from the centrosome\nProc Natl Acad Sci U S A\n94\n5078\n5083\n9144193\n49. ZimmermanWCSillibourneJRosaJDoxseySJ\n2004\nMitosis-specific anchoring of gamma tubulin complexes by pericentrin controls spindle organization and mitotic entry\nMol Biol Cell\n15\n3642\n3657\n15146056\n50. Waterman-StorerCDueyDYWeberKLKeechJCheneyRE\n2000\nMicrotubules remodel actomyosin networks in Xenopus egg extracts via two mechanisms of F-actin transport\nJ Cell Biol\n150\n361\n376\n10908578\n51. GavinRH\n1997\nMicrotubule-microfilament synergy in the cytoskeleton\nInt Rev Cytol\n173\n207\n242\n9127954\n52. SiderJRMandatoCAWeberKLZandyAJBeachD\n1999\nDirect observation of microtubule-f-actin interaction in cell free lysates\nJ Cell Sci\n112\n Pt 12\n1947\n1956\n10341213\n53. ShannonKBCanmanJCBen MoreeCTirnauerJSSalmonED\n2005\nTaxol-stabilized microtubules can position the cytokinetic furrow in mammalian cells\nMol Biol Cell\n16\n4423\n4436\n15975912\n54. YonetaniALustigRJMoseleyJBTakedaTGoodeBL\n2008\nRegulation and targeting of the fission yeast formin cdc12p in cytokinesis\nMol Biol Cell\n19\n2208\n2219\n18305104\n55. VavylonisDWuJQHaoSO'ShaughnessyBPollardTD\n2008\nAssembly mechanism of the contractile ring for cytokinesis by fission yeast\nScience\n319\n97\n100\n18079366\n56. NoguchiTMabuchiI\n2001\nReorganization of actin cytoskeleton at the growing end of the cleavage furrow of Xenopus egg during cytokinesis\nJ Cell Sci\n114\n401\n412\n11148141\n57. FeierbachBVerdeFChangF\n2004\nRegulation of a formin complex by the microtubule plus end protein tea1p\nJ Cell Biol\n165\n697\n707\n15184402\n58. NishimuraYYonemuraS\n2006\nCentralspindlin regulates ECT2 and RhoA accumulation at the equatorial cortex during cytokinesis\nJ Cell Sci\n119\n104\n114\n16352658\n59. SomersWGSaintR\n2003\nA RhoGEF and Rho family GTPase-activating protein complex links the contractile ring to cortical microtubules at the onset of cytokinesis\nDev Cell\n4\n29\n39\n12530961\n60. YuceOPieknyAGlotzerM\n2005\nAn ECT2-centralspindlin complex regulates the localization and function of RhoA\nJ Cell Biol\n170\n571\n582\n16103226\n61. ZavortinkMContrerasNAddyTBejsovecASaintR\n2005\nTum/RacGAP50C provides a critical link between anaphase microtubules and the assembly of the contractile ring in Drosophila melanogaster\nJ Cell Sci\n118\n5381\n5392\n16280552\n62. ZhaoWMFangG\n2005\nMgcRacGAP controls the assembly of the contractile ring and the initiation of cytokinesis\nProc Natl Acad Sci U S A\n102\n13158\n13163\n16129829\n63. D'AvinoPPTakedaTCapalboLZhangWLilleyKS\n2008\nInteraction between Anillin and RacGAP50C connects the actomyosin contractile ring with spindle microtubules at the cell division site\nJ Cell Sci\n121\n1151\n1158\n18349071\n64. RuchaudSCarmenaMEarnshawWC\n2007\nChromosomal passengers: conducting cell division\nNat Rev Mol Cell Biol\n8\n798\n812\n17848966\n65. ChenQLakshmikanthGSSpudichJADe LozanneA\n2007\nThe localization of inner centromeric protein (INCENP) at the cleavage furrow is dependent on Kif12 and involves interactions of the N terminus of INCENP with the actin cytoskeleton\nMol Biol Cell\n18\n3366\n3374\n17567958\n66. XiaQZhouZLuCChengDDaiF\n2004\nA draft sequence for the genome of the domesticated silkworm (Bombyx mori)\nScience\n306\n1937\n1940\n15591204\n67. ZhangDNicklasRB\n1999\nMicromanipulation of chromosomes and spindles in insect spermatocytes\nMethods Cell Biol\n61\n209\n218\n9891316\n68. PeloquinJKomarovaYBorisyG\n2005\nConjugation of fluorophores to tubulin\nNat Methods\n2\n299\n303\n16167385"
+}

batch_5/PMC2528140.json

@@ -0,0 +1,4 @@
+{
+    "id": "PMC2528140",
+    "text": "This is an academic paper. This paper has corpus identifier PMC2528140\nAUTHORS: M-O Riener, M Bawohl, P-A Clavien, W Jochum\n\nABSTRACT:\nNo Abstract\n\nBODY:\nSir,\nWe have read with great interest the article by Kim et al (2008), in which they report the analysis of 713 cancer tissues for somatic mutations in the AKT1, AKT2 and AKT3 genes. They detected the previously reported AKT1 pleckstrin homology domain mutation (AKT1 p.E17K) in 4 out of 93 (4.3%) breast carcinomas, but not in the colorectal, lung, gastric and hepatocellular carcinomas of their series (Carpten et al, 2007). The authors concluded that the AKT1 p.E17K mutation should be further analysed in a wider range of cancers.We have studied 118 carcinomas of the biliary tract and liver (11 intrahepatic and 34 extrahepatic cholangiocarcinomas, 23 gallbladder carcinomas and 50 hepatocellular carcinomas) for AKT1 p.E17K mutations using polymerase chain reaction (PCR) and direct DNA sequencing. DNA was extracted from formalin-fixed, paraffin-embedded tumour tissues. The primers for PCR amplification and sequencing of exon 4 had the following sequence: forward 5′-CTGGCCCTAAGAAACAGCTCC-3′ and reverse 5′-CGCCACAGAGAAGTTGTTGA-3′. Reaction conditions for PCR amplification were 40 cycles of 95°C for 30 s, 60°C for 1 min and 72°C for 1 min. Polymerase chain reaction products were purified (GFX PCR DNA and Gel Band Purification Kit, Amersham Biosciences, Otelfingen, Switzerland) and analysed on an automated DNA sequencer (model 3130) using Gene Scan and SeqScape software (Applied Biosystems, Foster City, CA, USA). Nucleotide sequences were compared with the genomic sequence of AKT1 (ENSG00000142208).We did not observe AKT1 p.E17K mutations in any of the 118 carcinomas of the biliary tract and liver in our series.We have recently shown that the frequent activation of the PI3K/AKT pathway in carcinomas of the biliary tract and liver is associated with PIK3CA hot spot mutations in a small subset of these tumours (Riener et al, 2008). The low frequency of these mutations lead to the conclusion that further genetic changes have to be responsible for the activation of the PI3K/AKT pathway in these cancers. The recently identified AKT1 p.E17K mutation appeared to be a good candidate for such a mechanism. Our results in hepatocellular carcinomas confirm the findings of Kim et al in a different ethnic background and suggest that cholangiocarcinoma and gallbladder carcinoma can be added to the list of cancers that lack AKT1 p.E17K mutations.\n\nREFERENCES:\n1. Carpten JD, Faber AL, Horn C, Donoho GP, Briggs SL, Robbins CM, Hostetter G, Boguslawski S, Moses TY, Savage S, Uhlik M, Lin A, Du J, Qian YW, Zeckner DJ, Tucker-Kellogg G, Touchman J, Patel K, Mousses S, Bittner M, Schevitz R, Lai MH, Blanchard KL, Thomas JE (2007) A transforming mutation in the pleckstrin homology domain of AKT1 in cancer. Nature\n448: 439–44417611497\n2. Kim MS, Jeong EG, Yoo NJ, Lee SH (2008) Mutational analysis of oncogenic AKT E17K mutation in common solid cancers and acute leukaemias. Br J Cancer\n98: 1533–153518392055\n3. Riener MO, Bawohl M, Clavien PA, Jochum W (2008) Rare PIK3CA hotspot mutations in carcinomas of the biliary tract. Genes Chromosomes Cancer\n47: 363–36718181165"
+}

batch_5/PMC2528554.json

@@ -0,0 +1,4 @@
+{
+    "id": "PMC2528554",
+    "text": "This is an academic paper. This paper has corpus identifier PMC2528554\nAUTHORS: A.M. Leis, L.C. Weeks, M.J. Verhoef\n\nABSTRACT:\nBackgroundIntegrative oncology uses both conventional and complementary medicine to meet the needs of individual patients and to focus on the whole person. The core principles of integrative oncology include individualization, holism, dynamism, synergism, and collaboration, but the nature of the evidence to guide the development of integrative oncology has been given little attention.Objectives\nTo discuss the need for evidence to support the integration of complementary therapies for integrative oncology care.To emphasize that the evidence base must be valid and respect the underlying principles of individual complementary therapies and integrative oncology practice.To suggest ways to begin developing the evidence base.Review and DiscussionAlthough the evidence for safety and efficacy seems paramount for supporting the integration of an individual complementary therapy into mainstream cancer care, the need for evidence to support the overall practice of integrative oncology has to be considered as well.We argue that developing an evidence base for integrative oncology requires a contextual and comprehensive research approach that assesses a range of outcomes over a suitable period of time that the patient and the patient’s family, in addition to the health care providers, deem important.ConclusionA whole-systems framework to the development of the evidence base for integrative oncology can guide the development of evidence that respects the complex nature of many complementary and integrative practices and their underlying principles of care delivery.\n\nBODY:\n1. INTRODUCTIONThe field of integrative oncology has emerged as a response both to cancer patients’ advocacy for holistic care and to an increasing evidence base for the safety and effectiveness of many complementary approaches, commonly called complementary and alternative medicine (cam). Cancer patients desire care that not only focuses on treating their disease, but also manages the course of their illness experience, optimizing health and enhancing well-being. Most cancer patients use complementary medicine alongside conventional medicine to meet these needs 1–4. Complementary and alternative medicine includes whole medical systems (Traditional Chinese Medicine, among others), mind–body medicine (for example, meditation), biologically-based practices (natural health products, for instance), manipulative and body-based therapies (for example, massage), and energy therapies (qi gong, among others) 5.Integrative oncology uses both conventional medicine and cam to meet the needs of individual patients and focuses on the whole person 6. At the core of integrative oncology is the need for an evidence base to support the use of conventional and complementary treatments in a collaborative and synergistic manner. The nature of the evidence to guide the development of integrative oncology has, however, been given little attention. In the present paper, we discuss the need for evidence to support the integration of complementary therapies for integrative oncology care; we emphasize that the evidence base must be valid and must respect the underlying principles of individual complementary therapies and integrative oncology practice; and we suggest ways to begin developing that evidence base.2. WHAT IS INTEGRATIVE ONCOLOGY?The goal of integrative oncology is to support cancer patients and their families throughout the cancer journey by improving quality of life, ameliorating symptoms associated with conventional cancer care, alleviating distress, and in some cases, augmenting the effectiveness of conventional treatment 7,8. Mumber defines integrative oncology as “a comprehensive, evidence-based approach to cancer care that addresses all participants at all levels of their being and experience. It represents the next step in the evolution of cancer care in that it addresses the limitations of the current system while retaining the system’s successful features” 9.Core principles of integrative oncology include individualization, holism, dynamism, synergism, and collaboration. In integrative oncology, the focus of care is on the whole person, and the aim is to promote the innate ability of each person to heal. Integrative oncology is individualized for each cancer patient over time, as each patient presents with unique symptoms and context, and as the goals of treatment change over time. Integrative care is also about compassion and caring for an individual in a holistic manner that gives voice to the patient’s values and needs. Grounded in a truly respectful partnership between patient and practitioner, a therapeutic alliance is forged that honours the patient’s informed choices. This collaborative approach to cancer care assumes that conventional and complementary practitioners—and patients—contribute their knowledge, experience, and skills to the healing encounter 8. In this context, a safe, knowledgeable, and dynamic cancer management plan is developed cooperatively, ensuring accurate monitoring and evaluation 10. Further, and in contrast to the prevailing pharmacologic model, the this cancer care approach recognizes the potential for synergy when therapies are integrated, with outcomes far exceeding the sum of the outcomes of individual therapies.Integrative oncology is usually defined as an evidence-based discipline; however, we argue that the traditional (scientific) understanding of evidence needs to be revisited and expanded.3. THE NEED FOR EVIDENCEIntegrative oncology makes a deliberate, yet fluid, distinction between complementary therapies, which are supported by evidence and used in combination with conventional cancer care, and alternative therapies, which are unproven and used as a replacement for conventional cancer care 7,11. When a strong evidence base is developed for some complementary therapies, they can potentially become part of integrative cancer care. For example, after a review of the available evidence, the Society for Integrative Oncology supports the use of acupuncture as a complementary therapy when cancer-related pain is poorly controlled 7.The practice of integrative oncology therefore depends mainly on complementary therapies meeting standards of safety and, to a lesser extent, effectiveness (because the latter may be assessed from different perspectives in real-life situations). As a result, an understanding of what constitutes appropriate evidence is crucial to the foundation and future development of the field.Despite agreement on the need for evidence to support the integration of complementary therapies into conventional cancer care 7,9,11, the discussion regarding the type of evidence required and its purpose is entirely based on evidence from specific cam treatments. We argue that this line of thinking misses an important point. To begin the discussion, a distinction must first be made between evidence for complementary therapies and evidence for integrative oncology practice. The first issue concerns evidence that supports the safety and effectiveness of individual complementary therapies, thus determining their suitability for integration into mainstream care. The second—and often forgotten—aspect concerns evidence that supports the synergistic integration of complementary and conventional practices in a collaborative and supportive manner within cancer care.3.1 Nature of the Evidence Required for Complementary TherapiesAccording to Stark, Hess, and Shaw 12, different levels of evidence are required for the safety and effectiveness of individual complementary therapies depending on the goals of treatment. These levels of evidence depend on study design and sample size, and they range from well-designed randomized controlled trials [rcts (level 1)] to preclinical in vitro and in vivo studies and traditional medicines (level 4). Level 1 evidence is required for the use of complementary therapies with anti-neoplastic goals, but lower levels of evidence, such as nonrandomized trials or observational studies, are acceptable for less-invasive procedures and preventive or supportive goals. However, this hierarchy does not capture the multidisciplinary, synergistic approach that characterizes complementary therapies and integrative oncology alike in comparison with conventional care 10. Further, while addressing the need for evidence to support the individual integration of a complementary therapy into mainstream care, the need for evidence to support the overall practice of integrative oncology is often ignored.A need arises to revisit traditional notions of evidence as they apply to complementary therapies. Traditional research methods are challenged in the attempt to evaluate complementary therapies, because these methods cannot account for the fundamental issues of individualization, synergism, and holism 8,9,13. These problems are compounded for the evaluation of integrative oncology, which involves the synergistic use of treatments from various healing paradigms and a range of physiologic, emotional, social, and spiritual outcomes.3.2 Nature of the Evidence Required for IntegrationFurther to answering whether a complementary therapy works and is safe, questions regarding the appropriateness of integration must be examined. To this end, it is critical to document the ways in which complementary therapies and conventional care are being integrated and the outcomes that are important and relevant.Integration can occur at many levels: individual, clinical, institutional, regulatory, or policy 14. Integration can also occur in many ways. For example, numerous patients are known to be integrating complementary therapies into their conventional care, but research is only starting to uncover how those patients make decisions regarding therapy selection, who is involved in the decision-making process, why the patients are integrating these therapies, and which outcomes are seen as relevant 15. Alternatively, health care providers may be the ones suggesting integration for their patients. The process of evaluation and decision-making is likely different in the two scenarios, in part because the intent may differ. For the health care professional, for example, the utmost concern is patient safety; but for patients, a decision to use cam may be driven by an attempt to minimize potential side effects and to feel empowered 16.“Integrating” must also be distinguished from “combining.” “Combining” is more akin to adding various therapies to a treatment plan without considering the overall picture. “Integrating” involves synergistically applying a range of treatments to address holistic treatment goals as they change over time and in accordance with patient needs and values. Currently, although the goals and philosophy underlying integrative oncology are well developed, practical knowledge is not available concerning the extent to which and the manner in which diverse therapies are being integrated or combined.Finally, integrative care seems to represent untapped opportunities (which so far remain understudied) for meeting the needs of cancer patients across the cancer trajectory.4. VALIDITY AS CRITERION FOR EVIDENCEAs development of an evidence base begins both for individual complementary therapies and for integrative oncology, assurance is needed that the evidence base is valid—“validity” referring to the extent that appropriate research methods were used to support a conclusion regarding the efficacy or effectiveness of a therapy.“Validity” commonly includes internal validity and external validity. Model validity is separate from both of those, but is just as important. Model validity is often overlooked because of the biomedical focus of most health care research. It refers to the extent to which the research methods used have addressed the unique theory and therapeutic context of the intervention being assessed 17.Traditional clinical research methods have been developed to assess biomedical interventions, and thus model validity can typically be assumed. For research results to be valid in the case of integrative oncology, the research methods used must address the underlying principles of integrative oncology, such as its individualized, synergistic, holistic, and collaborative nature. The same is true for research regarding individual complementary therapies that are often based on assumptions contrary to biomedicine, such as the network of channels and blood vessels connected by qi (an essential fast-flowing substance full of vigour) in an approach using Traditional Chinese Medicine.4.1 Limitations of the RCT Design for Complementary Therapies and Integrative OncologyThe double-blind rct is often upheld as the “gold standard” in clinical research, because of its strong internal validity arising from the ability to control for expected and unexpected bias, confounding factors, and error. However, it is impossible to achieve model validity while applying the blinded rct design to the practice of integrative oncology and many complementary therapies 13.Sagar 8 highlights several of the key challenges in applying the rct design to the study of select complementary therapies. Examples include difficulties in determining appropriate placebos or sham treatments, the impossibility of double-blinding when the practitioner is part of the intervention, and problems respecting the individualized approach of many complementary practices. Acupuncture provides a good example, because the choice of an appropriate sham treatment for acupuncture has been an ongoing challenge 18, not unlike that in determining an appropriate sham treatment for surgery in the realm of conventional care. Further, blinding patients and providers is difficult, because both are quite aware of whether needling has taken place, although single-blinding may be possible if simulated needling is used as a sham treatment. In practice, different needling protocols are developed for specific patients, depending on their unique symptoms and holistic context, and thus standardization of an acupuncture protocol for a rct is problematic if model validity is to be upheld.The same argument can easily be extended to integrative oncology practice, in which therapies from diverse philosophical backgrounds are combined, thus making model validity even more difficult to attain. In addition to the problems of defining a placebo, blinding, and standardization, the rct is not designed to measure the effect that each patient’s unique physical, social, and cultural context and corresponding reasons for integration may have on treatment outcomes. Further, the rct cannot assess the synergism that results from the integration of various therapies, coupled with the healing context and the clinical skills and expertise of the integrative team.4.2 Building Evidence for Integrative OncologyDeveloping an evidence base for integrative oncology requires a contextual and comprehensive research approach that assesses a range of outcomes over a suitable period of time that the patient, the patient’s family, and health care providers deem important. We next highlight some approaches that begin to meet the requirements of internal, external, and model validity. Readers should consult the papers referenced in this section for detailed descriptions of these approaches.4.2.1 Variations of the RCT DesignAlthough the rct in its classical form cannot meet the requirements for model validity when applied to complementary therapies or integrative oncology, some variations have been suggested that address many of the shortcomings. For example, pragmatic rcts do not require standardization of the intervention and thus allow for the individualized nature of the treatments to be assessed. Preference rcts account for the effect that beliefs and preferences of patients for certain treatment types will have on treatment outcomes. In a preference rct, patients with treatment preferences receive their preferred treatment; patients who do not have a preference are randomized as usual.4.2.2 The Power of Mixed MethodsAlthough quantitative research approaches have been well established in evaluating treatment interventions, the importance of qualitative research cannot be overlooked or dismissed as development begins of the evidence base for complementary therapies and integrative oncology. Qualitative research has amazing potential to explore, in depth, from a variety of perspectives (patients, oncology team, cancer care system), how integrative cancer treatment and care are experienced. Including qualitative inquiry in the evaluation of interventions should be an integral part of evidence-based medicine 19.Qualitative research aims to understand the nature of phenomena; it will prove crucial as exploration of the potential of integrative oncology for cancer care moves forward. For example, qualitative research is ideally suited to answer the question of what integrative oncology is, how it is being practiced, how it can best be practiced, and the benefits that are possible. Such exploratory work can help to elucidate the key components in integrative care and their synergistic relationships from the viewpoint of patients and of practitioners.Because qualitative research can fill an important niche in this field, it should, when possible, be nested within clinical trials or other quantitative designs 20,21 as a form of mixed-methods research. Such a design represents a further modification of the rct method that addresses the requirements of internal, external, and model validity. This approach can help to elicit whether, why, and how patients benefit from a complex intervention and can explore relevant outcomes from a variety of perspectives.4.2.3 Whole-Systems ResearchWhichever research methods are ultimately adopted to study complementary therapies and integrative oncology, investigators must systematically capture the complexity inherent in these approaches to healing. A “whole-systems” research framework is helpful to conceptualize the important issues that need attention and to design step-wise programs of research that will answer the multifaceted questions.The notion of whole-systems research has recently been given attention in the literature, both with regard to complementary therapies 13,22 and integrative oncology 23,24. In a general sense, the goal of whole-systems research “is to appropriately combine research designs and methods in a coherent research program, so that all aspects of an internally consistent approach to treatment, or a whole system, can be assessed. It acknowledges an individualized, patient centered and participatory approach to diagnosis and treatment and a process of healing that collaboratively combines patient and practitioner knowledge and skills, thus enhancing healing” 23.The relevance of a whole-systems framework to the development of the evidence base for integrative oncology should be apparent. It can guide the development of evidence that respects the complex nature of many complementary and integrative practices and their underlying principles of care delivery 25. A program of research guided by a whole-systems research framework would include rcts, variations of rcts, observational trials, and qualitative research, ideally mixing methods as appropriate, depending on the specific focus. A program of whole-systems research would thus produce research that collectively has strong validity.5. CONCLUSIONSAs cancer patients increasingly turn to cam as a way to complement their cancer care, it is crucial that health care professionals become informed about the evidence base behind this group of practices 26 and, further, that they remain critical of the need for valid research. The assessment of validity for complementary therapies and integrative oncology alike should encompass internal, external, and model validity. Because it seems unlikely that any one study design can achieve optimal levels of each type of validity, health professionals and researchers need to be open to emerging models of evidence that are not necessarily aligned to traditional ideas of evidence in biomedicine. Programs of research have to include a variety of evidence types and treat all of them as legitimate. A whole-systems research framework is helpful to guide the development of these research programs.\n\nREFERENCES:\n1. DiGianniLMGarberJEWinerEPComplementary and alternative medicine use among women with breast cancerJ Clin Oncol20022034S8S12235222\n2. MolassiotisAFernadez–OrtegaPPudDUse of complementary and alternative medicine in cancer patients: a European surveyAnn Oncol2005166556315699021\n3. SparberAWoottonJCSurveys of complementary and alternative medicine: Part ii. Use of alternative and complementary cancer therapiesJ Altern Complement Med20017281711439851\n4. LeisAFerroMComplementary and alternative medicine: responding to the needs of cancer patientsOncol Exch200655760\n5. National Institutes of Health, National Center for Complementary and Alternative Medicine (nccam)Health information > cam basics > What is cam? [Web page]Gaithersburg, MDNCCAMFebruary 2008[Available at: nccam.nih.gov/health/whatiscam; cited: March 27, 2008]\n6. BarracloughJBarracloughJIntroducing the holistic approach to cancer careEnhancing Cancer CareOxfordOxford University Press2007\n7. DengGECassilethBRCohenLIntegrative oncology practice guidelinesJ Soc Integr Oncol20075658417511932\n8. SagarSMIntegrative oncology in North AmericaJ Soc Integr Oncol20064273916737670\n9. MumberMPRakelDIntegrative oncology: an overviewIntegrative Medicine2nd edPhiladelphiaSaunders Elsevier2007\n10. BoydDBIntegrative oncology: the last ten years—a personal retrospectiveAltern Ther Health Med200713566417283742\n11. CassilethBWhy integrative oncology? Complementary therapies are increasingly becoming part of mainstream careOncology (Williston Park)200620130217024875\n12. StarkNHessSShawEMumberMPClinical research and evidenceIntegrative Oncology: Principles and PracticeLondonTaylor and Francis2005\n13. VerhoefMLewithGRitenbaughCBoonHFleishmanSLeisAComplementary and alternative medicine whole systems research: beyond identification of inadequacies of the rctComplement Ther Med2005132061216150375\n14. TatarynDVerhoefMJCombining conventional, complementary and alternative health care: a vision of integrationPerspectives on Complementary and Alternative Health CareOttawaHealth Canada2001\n15. BalneavesLTruantTKellyMVerhoefMDavisonJBridging the gap: decision-making processes of women with breast cancer using complementary and alternative medicine (cam)Support Care Cancer2007159738317609997\n16. BalneavesLGWeeksLSeelyDPatient decision-making about complementary and alternative medicine in cancer management: context and processCurr Oncol200815suppl 2S94S10018769576\n17. LewithGWalachHJonasWBLewithGWalachHJonasWBBalanced research strategies for complementary and alternative medicineClinical Research in Complementary Therapies: Principles, Problems and SolutionsEdinburghChurchill Livingstone2002\n18. WhiteAFilshieJCummingsMClinical trials of acupuncture: consensus recommendations for optimal treatment, sham controls and blindingComplement Ther Med200192374512184353\n19. StrausSRichardsonWSGlasziouPHaynesRBEvidence-based Medicine: How to Practice and Teach ebmEdinburghChurchill Livingstone2005\n20. JonasWBCrawfordCScience and spiritual healing: a critical review of spiritual healing, “energy” medicine, and intentionalityAltern Ther Health Med20039566112652884\n21. VerhoefMJCasebeerALHilsdenRJAssessing efficacy of complementary medicine: adding qualitative research methods to the “gold standardJ Altern Complement Med200282758112165185\n22. RitenbaughCVerhoefMFleishmanSBoonHLeisAWhole systems research: a discipline for studying complementary and alternative medicineAltern Ther Health Med2003932612868250\n23. VerhoefMJVanderheydenLCFonneboVA whole systems research approach to cancer care: why do we need it and how do we get started?Integr Cancer Ther200652879217101757\n24. VerhoefMWeeksLBrazierALeisAEvaluating supportive cancer care: are we missing an opportunity?Support Care Cancer200715905717609993\n25. VerhoefMLeisAFrom studying patient treatment to studying patient care: arriving at methodological crossroadsHematol Oncol Clin North Am2008226718218638695\n26. SmythJFIntegrative oncology—what’s in a name?Eur J Cancer200642572316459074"
+}

batch_5/PMC2528946.json

@@ -0,0 +1,4 @@
+{
+    "id": "PMC2528946",
+    "text": "This is an academic paper. This paper has corpus identifier PMC2528946\nAUTHORS: Kwok M. Ho, Matthew Knuiman, Judith Finn, Steven A. Webb\n\nABSTRACT:\nBackgroundLong-term survival outcome of critically ill patients is important in assessing effectiveness of new treatments and making treatment decisions. We developed a prognostic model for estimation of long-term survival of critically ill patients.Methodology and Principal FindingsThis was a retrospective linked data cohort study involving 11,930 critically ill patients who survived more than 5 days in a university teaching hospital in Western Australia. Older age, male gender, co-morbidities, severe acute illness as measured by Acute Physiology and Chronic Health Evaluation II predicted mortality, and more days of vasopressor or inotropic support, mechanical ventilation, and hemofiltration within the first 5 days of intensive care unit admission were associated with a worse long-term survival up to 15 years after the onset of critical illness. Among these seven pre-selected predictors, age (explained 50% of the variability of the model, hazard ratio [HR] between 80 and 60 years old = 1.95) and co-morbidity (explained 27% of the variability, HR between Charlson co-morbidity index 5 and 0 = 2.15) were the most important determinants. A nomogram based on the pre-selected predictors is provided to allow estimation of the median survival time and also the 1-year, 3-year, 5-year, 10-year, and 15-year survival probabilities for a patient. The discrimination (adjusted c-index = 0.757, 95% confidence interval 0.745–0.769) and calibration of this prognostic model were acceptable.SignificanceAge, gender, co-morbidities, severity of acute illness, and the intensity and duration of intensive care therapy can be used to estimate long-term survival of critically ill patients. Age and co-morbidity are the most important determinants of long-term prognosis of critically ill patients.\n\nBODY:\nIntroductionDemand for intensive care unit (ICU) services is increasing [1], and at a rate that is higher than the average for all health care services [2]. Increase in treatment and monitoring technology, patients' expectations, and ageing population all contribute to this increased demand for intensive care services [1]. Indeed, intensive care is increasingly being provided to older and sicker patients, whom in the past were not treated in intensive care [3]. Intensive care services accounted for 10% of the US$2.1 trillion total health expenditures on health care in the United States in 2006 [4] and has been estimated to cost more than £700 million in the United Kingdom in 1999 [5]. The cost of intensive care services coupled with increasing demand provides the rationale for improved modelling of outcomes of critically ill patients.Long-term survival after critical illness is increasingly being recognized as an important outcome in assessing effectiveness of new therapy [6]. In order to control for confounding and bias in assessing long-term survival of critically ill patients in a clinical trial, a risk adjustment tool that can objectively estimate long-term survival is essential. From a clinical perspective, many patients and clinicians are also interested in knowing the long-term survival outcome after critical illness, in addition to other information such as burden of treatment and quality of life after recovery, when making treatment decisions in the ICU. Although many clinicians may foretell patient hospital survival outcome more accurately than some objective prognostic models [7], treatment decisions made by clinicians do vary considerably with their practice style and work experience [8]–[10]. The strategy of continuing intensive treatment for all patients until death will reduce the need for patients and clinicians to make difficult treatment decisions and may improve the survival time of some. This strategy is, however, expensive and undesirable by imposing excessive burden of treatment on those who have a very poor prognosis [11]. For example, initiating acute renal replacement therapy in critically ill patients with less than 10% probability of 6-month survival was estimated to cost US$274,000 (£137,000) per quality-adjusted life year saved [12].The SUPPORT investigators from the United States and Wright et al. from the United Kingdom published two prognostic models that were based on age, severity of acute illness and admission diagnosis to estimate 6-month and 5-year survival of critically ill patients, respectively [13], [14]. The utility of latter model is, however, limited by its ability to classify 5-year survival probabilities only into three risk categories when the calculated risk score is either <70, 70–80, or >80 [14]. This model also did not consider the potential effect of detailed co-morbidity data on long-term survival of critically ill patients beyond the usual assessment included in the Acute Physiology and Chronic Health Evaluation (APACHE) score [14], [15]. There is currently no prognostic model that is available to estimate the survival of critically ill patients beyond 5 years after the onset of critical illness. Furthermore, the relative importance of age, co-morbidity, and severity of acute illness in determining long-term prognosis of critically ill patients also remains unknown. In this study we examined the long-term survival of 11,930 critically ill adult patients who survived at least 5 days and developed a new prognostic model (Predicted Risk, Existing Diseases, and Intensive Care Therapy: the PREDICT model) to estimate their median survival time and long-term survival probabilities.MethodsThe characteristics of the cohortThis cohort study utilized the clinical database of the ICU at Royal Perth Hospital (RPH) in Western Australia. RPH is the largest tertiary university teaching hospital in Western Australia and the 22-bed ICU admits patients of all specialties except liver transplantation and captures over 40% of all critically ill patients in Western Australia. The database analyzed in this study includes details of all ICU admissions between 1989 and 2002, including demographic factors, admission diagnosis, admission source, severity of acute illness as measured by APACHE II scores based on the worst first 24-hour ICU data [15], daily organ failure assessment and supportive therapy required [16], and ICU and hospital survival outcome.In this study the patients with a diagnosis excluded from the original APACHE II cohort (e.g. coronary artery graft surgery, burns, snake bite)[15], those who resided outside Western Australia at the time of ICU admission (who could not be followed for survival), readmissions after the first ICU readmission, patients who were younger than 16 years old, and patients who did not survive more than 5 days during their hospitalization of the index ICU admission were excluded. The data were reviewed for internal consistency annually, and there were no patients with missing hospital mortality data. Some of the other details of this cohort have been described in our previous publications [16]–[18].The ICU clinical database was linked to the Western Australian hospital morbidity and mortality databases using probabilistic matching [16], providing information on patients' co-morbidities as recorded in all private and public hospital admissions including any prior ICU admissions up to 5 years before the index ICU admission. A relatively long five-year ‘look back’ period was chosen in this study to capture all existing co-morbidities of each patient. We ascertained the presence of co-morbidities in the Charlson co-morbidity index (\nTable 1\n) using the published ICD-9-CM and ICD-10-AM coding algorithms [16], [19]. We did not assign a co-morbidity to a patient if that condition was diagnosed during the hospitalization of the index ICU admission. The proportions of invalid (false positive) and missed links (false negatives) between Western Australian hospital morbidity and mortality databases were evaluated several years ago, and both false positives and negatives were estimated to be 0.11% [20].10.1371/journal.pone.0003226.t001Table 1Charlson co-morbidity index component and its weighting.Co-morbidityWeightingMyocardial infarction1Congestive heart failure1Peripheral vascular disease1Cerebrovascular disease1Dementia1Chronic pulmonary disease1Connective tissue disease1Peptic ulcer disease1Mild liver disease1Diabetes mellitus1Hemiplegia2Moderate or severe renal disease2Diabetes with end-organ damage2Any tumour2Leukemia2Lymphoma2Moderate to severe liver disease3Metastatic solid tumour6AIDS6The survival status of the patients was assessed on 31 December 2003 and the length of follow up ranged from 1 year to 15 years with an average of 6 years. Western Australia is geographically isolated and has a very low rate of emigration (<0.03% in 2002)[16], and as such, lost to long-term survival follow-up by the Western Australian mortality database is likely to be very low. The data analyzed had the patient name and address removed and the study was approved by the RPH Ethics Committee and the Western Australian Confidentiality of Health Information Committee (CHIC).Development of the modelThe prognostic model was fitted using Cox proportional hazards regression [21], restricting predictors to factors that were likely to be important predictors of long-term survival of hospitalized patients [13], [14], [22], [23]. These pre-selected factors included age [14], [22], gender [22], APACHE II predicted mortality risk [13]–[15], Charlson co-morbidity index [19], [23], days of mechanical ventilation, acute renal replacement therapy or hemofiltration, and vasopressor or inotropic therapy during the first 5 days of the index ICU admission [13]. The APACHE II predicted mortality was chosen as a measure of severity of acute illness because it is widely used and summarizes the diagnosis, acute physiologic derangement within the first 24 hours of ICU admission, severe co-morbidities, and whether the ICU admission is after elective or emergency surgery. Our previous study also showed that the APACHE II predicted mortality has a very stable performance in this cohort over the past 10–15 years [17]. Although age and severe co-morbidities are already used to calculate the APACHE II predicted mortality [15], these two factors may still have a profound effect on long-term survival over and beyond the weightings used in the APACHE II predicted mortality [14], [22], [23]. As such, both age and Charlson co-morbidity index were used as separate predictors in additional to the APACHE II predicted mortality in this prognostic model. These seven predictors were also chosen because they are often recorded in the administrative databases of many ICUs, and as such, it is possible for other ICUs to validate this model using their data [24].The proportional hazards assumption of the continuous predictors in the Cox model was assessed and found to be acceptable (\nFigure 1a, 1b, 1c\n). During the modelling process, we avoided categorizing continuous predictors [24], [25] and allowed a non-linear relationship with hazard of death using a 6-knot restricted cubic spline function [25]. The relative contribution of each predictor was assessed using the chi-square statistic minus the degrees of freedom [25]. The discrimination performance of the model was assessed with the c-index, which is a generalization of the c-statistic or the area under the receiver-operating characteristic curve, allowing for censored data [25], [26]. In this study, a c-index between 0.70 and 0.80 was regarded as acceptable and a c-index above 0.80 was regarded as excellent [27]. Using the Design library in S-PLUS software (version 8.0, 2007. Insightful Corp., Seattle, Washington, USA), the c-index was computed and adjusted for optimism (arising from using the same data to develop the model and assess its performance) by a bootstrap technique to penalise for possible over-fitting, with 200 re-samples and at least 200 patients per risk group [25], [28]. The bootstrapping technique was used in this study instead of splitting the data into development and validation data set because this method is regarded as most data ‘efficient’ and accurate in developing a prognostic model [25]. Model calibration (similarity of predicted risks and proportions actually dying) was assessed graphically and used a bootstrap re-sampling to construct a bias-corrected calibration curve and its slope [25], [29]. Nagelkerke's R2 (a generalized measure of the percentage of the variance in survival accounted for by the model) was computed to assess the overall performance of the model [25], [30]. The performance of the model was assessed over the full 15 years of follow-up, when follow-up was restricted to a maximum of 5 years for each patient, and also when only patients admitted after 1997 were considered.10.1371/journal.pone.0003226.g001Figure 1The proportional hazards assumption of the predictors in the Cox model was assessed by plotting the logarithm of the negative logarithm of the Kaplan Meier survivor estimates and the assumption was found to be acceptable for the three pre-selected continuous predictors; APACHE II predicted mortality, Charlson co-morbidity index, and age.(a) Graph assessing the proportionality of hazards associated with severity of acute illness measured by the APACHE II predicted mortality risk categories (0–20%, 20–40%, 40–60%, 60–80%, 80–100%). (b) Graph assessing the proportionality of hazards associated with co-morbidities measured by Charlson co-morbidity index categories (0, 1, 2, 3, 4–5, >5). (c) Graph assessing the proportionality of hazards associated with age measured by age categories (16–30, 30–50, 50–60, 60–70, 70–80, >80 years old)A nomogram was developed for the model that generates the median survival time and selected annual survival probabilities by adding up the scores for each of the seven predictors [25]. The use of the nomogram and how each predictor may affect a patient's long-term prognosis is described for a selection of typical patient scenarios. In particular, these scenarios were selected to illustrate how the long-term prognosis of a patient can be different from the short-term prognosis. Nevertheless, the results of the nomogram should only be considered as an average estimate of patients with similar characteristics and not be used for individual patients.ResultsCharacteristics of the cohortThe study cohort consisted of a heterogeneous group of critically ill patients, with elective surgery including heart valve surgery, urology, gastrointestinal and spinal surgery accounting for 36.2% of all ICU admissions. The emergency admissions consisted of patients with multiple trauma (8.5%), isolated head trauma (2.5%), acute myocardial infarction, congestive heart failure, cardiac arrhythmias, or cardiogenic shock (7.4%), hypovolemic or hemorrhagic shock (0.8%), drug overdoses (7.2%), subarachnoid or intracranial hemorrhage (5.1%), sepsis (4.3%), pneumonia or aspiration (3.7%), obstructive airway diseases (2.1%), cardiorespiratory arrest (4.0%), gastrointestinal hemorrhage, perforation or obstruction (2.4%), and other medical and surgical emergencies. Details of this cohort including demographic factors, co-morbidities, severity of acute illness, and the length of ICU and hospital stay are described in \nTable 2\n.10.1371/journal.pone.0003226.t002Table 2Characteristics of the cohort (n = 11,930).VariablesMean (median, standard deviation), unless stated otherwiseAge, yrs53.8 (57.0, 19.0)Gender (male/female), no. (%)7489 (62.8)/4441 (37.2)Elective surgery admission, no. (%)4318 (36.2)APACHE II score13.7 (13.0, 6.8)APACHE II predicted mortality, %14.5 (7.0, 17.8)No. of APACHE co-morbidities0.1 (0, 0.3)(a) Cardiovascular, no. (%)592 (5.0)(b) Respiratory, no. (%)210 (1.8)(c) Renal, no. (%)109 (0.9)(d) Immunosuppressed, no. (%)197 (1.7)(e) Liver, no. (%)76 (0.6)No. of Charlson co-morbidities0.8 (0, 1.2)Charlson co-morbidity index1.0 (0, 1.7)Length of ICU stay, days5.6 (3.0, 8.3)Length of hospital stay, days20.3 (13.0, 25.9)No. of patients mechanically ventilated (%) #\n8034 (67.3)No. of patients on inotrope (%) #\n3921 (32.9)No. of patients on dialysis (%) #\n608 (5.1)No. of ICU survivor (%)*\n11557 (96.9)No. of hospital survivor (%)*\n11101 (93.1)No. of survivor/total no. of patients followed up (%)(a) at 1-year10334/11101 (93.1)(b) at 3-year8031/10019 (80.2)(c) at 5-year6109/8212 (74.4)(d) at 10-year2609/4238 (61.6)(e) at 15-year441/887 (49.7)#During the first 5 days in ICU.*Excluding patients died within 5 days of ICU admission.ICU, intensive care unit.APACHE, Acute Physiology and Chronic Health Evaluation.Effect of the Predictors on Hazard of DeathAmong all the seven pre-selected predictors in the model, age (50%), co-morbidity as measured by Charlson co-morbidity index (27%), and severity of acute illness as measured by the APACHE II predicted mortality (20%) made the strongest contributions in predicting survival time (\nFigure 2\n). After adjusting for other predictors, the log hazard of death increased linearly with age, Charlson co-morbidity index, and the number of days of vasopressor or inotropic therapy, mechanical ventilation, or hemofiltration therapy (\nFigure 3\n). The relationship between the APACHE II predicted mortality and log hazard of death was non-linear with a steep effect when the APACHE II predicted mortality was less than 10% and a smaller effect when it was more than 10%. The estimated (adjusted) hazard ratios for the seven predictors are summarized in \nFigure 4\n.10.1371/journal.pone.0003226.g002Figure 2Contribution of each predictor in predicting the survival time in the Cox proportional hazards model.10.1371/journal.pone.0003226.g003Figure 3The relationship between relative hazard and each predictor after adjusting for other predictors in the model.10.1371/journal.pone.0003226.g004Figure 4The estimated (adjusted) hazard ratios and multilevel confidence bars (0.70 as illustrated by the black bar to 0.99 as illustrated by the orange bar) for the effects of predictors in the model are summarized in the figure below.An increase of 20 years of age and an increase in Charlson co-morbidity index from 0 to 5 approximately doubled the risk of death. Doubling the APACHE II predicted mortality from 20% to 40% increased the relative risk of death by about 30 to 40%. Similarly, increased the number of days of intensive care therapy from 1 to 5 increased the relative risk of death by between 10% and 50%.Clinical Application of the Model\n\nFigure 5\n presents the model in the form of a nomogram that provides the median survival time and long-term survival probabilities corresponding to a particular total score. The total score for a patient is obtained by adding up the scores for each of the seven predictors. We use the following hypothetical but typical patients to illustrate how the nomogram is used and how the short-term prognosis of a patient can be quite different from the long-term prognosis.10.1371/journal.pone.0003226.g005Figure 5Nomogram for predicting long-term survival probabilities and median survival time.Note: gender: 2 = female, 1 = male. Predicted.mortality = APACHE II predicted mortality in %.\nPatient A:\nA 40-year old male, without pre-existing co-morbidities (ie Charlson co-morbidity index = 0), was admitted to the ICU because of septic shock with an APACHE II predicted mortality of 80%. He required vasopressor or inotropic therapy, mechanical ventilation, and hemofiltration therapy during the first 5 days in the ICU.The gender of this patient scores 5 points, age scores 28 points, Charlson co-morbidity scores zero points, the APACHE II predicted mortality or risk scores 30 points, 5 days of vasopressor or inotropic therapy scores 7 points, 5 days of mechanically ventilation scores 15 points, and 5 days of hemofiltration scores 20 points. The total score of this patient is therefore 105 which gives an estimated median survival time of about 4 years, >70% 1-year survival probability, >50% 3-year survival probability, >45% 5-year survival probability, and >20% 10-year survival probability.\nPatient B:\nA 70-year old female, with chronic obstructive airway disease and non-insulin dependent diabetes mellitus with no end-organ damage (ie Charlson co-morbidity index = 2), was admitted to the ICU because of severe community acquired pneumonia with an APACHE II predicted mortality of 30%. She required vasopressor or inotropic therapy and mechanical ventilation but not hemofiltration during the first 5 days in the ICU.The gender of this patient scores zero points, age scores 70 points, Charlson co-morbidity index scores 12 points, the APACHE II predicted mortality scores 16 points, 5 days of mechanical ventilation scores 15 points, and 5 days of vasopressor or inotropic therapy scores 7 points. The total score of this patient is therefore 120 which gives an estimated median survival time of about 2 years, 60% 1-year survival probability, 40% 3-year survival probability, 30% 5-year survival probability, and 10% 10-year survival probability.\nPatient C:\nA 80-year old male, with a history of myocardial infarction, congestive heart failure, peripheral vascular disease, cerebrovascular disease, and dementia (ie Charlson co-morbidity index = 5), was admitted to an ICU with bowel perforation and peritonitis with an APACHE II predicted mortality of 30%. He required vasopressor or inotropic therapy and mechanical ventilation but not hemofiltration during the first 5 days in the ICU.The gender of this patient scores 5 points, age scores 85 points, Charlson co-morbidity index scores 30 points, the APACHE II predicted mortality scores 16 points, 5 days of mechanical ventilation scores 15 points, and 5 days of vasopressor or inotropic therapy scores 7 points. The total score of this patient is therefore 158 which gives an estimated median survival time of <0.5 years, 25% 1-year survival probability, and 10% 3-year survival probability.Discrimination and Calibration of the Prognostic ModelThe adjusted c-index for this prognostic model was 0.757 (95% confidence interval 0.745–0.769), Nagelkerke's R2 was 0.255 and the bias-corrected calibration of the model over a 15-year period was reasonable (slope of the calibration = 0.98)(\nFigure 6\n). The Nagelkerke's R2 remained unchanged and the adjusted c-index only increased marginally when the analysis was restricted to a maximum of 5 years follow up (c-index = 0.759, slope = 0.97) or data after 1997 (c-index = 0.762, slope of the calibration = 0.97).10.1371/journal.pone.0003226.g006Figure 6Bootstrap estimate of calibration accuracy for 15-year estimates from the Cox proportional hazards model.Dots correspond to apparent predictive accuracy and x marks the bootstrap-corrected estimates.DiscussionThis study showed that age, gender, co-morbidities (Charlson co-morbidity index), severity of acute illness (the APACHE II predicted mortality), and duration of intensive care therapy or organ support within the first 5 days of ICU admission are important prognostic factors for long-term survival of critically ill patients. To the best of our knowledge, this new prognostic model (Predicted Risk, Existing Diseases, and Intensive Care Therapy: the PREDICT model) is the first preliminary prognostic model that can be used to estimate the median survival time and long-term survival probabilities of critically ill patients up to 15 years after the onset of critical illness.The current prognostic model has confirmed that age, gender, co-morbidities, severity of acute illness, and duration of intensive care therapy or organ failure are important predictors of 6 months to 5 years survival of hospitalized or critically ill patients [13], [14], [19], [22], [23]. The current model is indeed built on the results of these previous studies but further extended the significance of these risk factors in predicting survival of critically ill patients beyond 6 months to 5 years. This current model also demonstrated that most of these predictors have a relatively linear relationship to the long-term survival probability. More importantly, our results also showed that age and co-morbidities are the most important determinants of long-term prognosis of critically ill patients. This latter finding has at least two significant clinical implications. First, the factors that determine long-term survival of a critically ill patient are different from those that affect short-term prognosis. Previous evidence suggested that diagnosis and acute physiological derangement of a patient are most important in determining hospital survival [15], [31]. In our three hypothetical patients, Patient A has in fact the most severe form of acute critical illness and worst short-term prognosis. Nevertheless, because this patient is younger and has no co-morbidities, this patient has a very reasonable and better long-term prognosis than Patient B and C. If we use the prognostic model developed by Wright et al. [14] to estimate the long-term survival of our three hypothetical patients, Patient B will have the best 5-year prognosis (risk score is estimated to be 68) followed by Patient C (risk score 75) and then Patient A (risk score 87). The lack of detailed co-morbidity data and a heavy emphasis on severity of acute illness in the model developed by Wright et al. is the most likely explanation why our results are different from theirs.Many clinicians may intuitively consider the intensity of organ failure as very important in affecting a patient's prognosis [32], [33]. Our findings suggest that the effect of acute organ failure on long-term survival is not strong and mostly captured by age, co-morbidities, and the APACHE II predicted mortality on admission to ICU. Our previous studies have also showed that the intensity of organ failure alone is not as important as the APACHE II score in predicting hospital mortality [34], [35]. Therefore, our findings suggest that clinicians should be very careful not to place undue emphasis on the severity of acute illness and intensity of organ failure when making long-term prognostications of critically ill patients.Second, because the contributions by intensive care therapy are relatively small when compared to age, Charlson co-morbidity index, and the APACHE II predicted mortality, using the data after the first 24 to 48 hours of ICU stay is unlikely to underestimate the final total prediction score significantly (<20 points)(\nFigure 5\n). Therefore, early estimation of a slightly ‘optimistic’ long-term survival probability and median survival time is feasible after the first 24 to 48 hours of ICU stay; and in patients with either extremes of prognosis, this early estimation is unlikely to be significantly different from the final prediction by collecting all data after five days of intensive care therapy. Nevertheless, the current prognostic model utilizes the APACHE II predicted mortality after ICU admission as a predictor to estimate long-term survival, as such, the model cannot be used, in its current form, as a tool to triage ICU admission.This study has significant limitations. First, patients' wishes and the anticipated quality of life before and after their critical illness are important factors in making treatment decisions [36], [37]. The median survival time and long-term survival probabilities is only one of the many factors that patients and clinicians may consider in making treatment decisions. Furthermore, the c-statistics of this model is only about 0.76 and this leaves considerable uncertainty in its applicability in predicting long-term survival of individual patients. As such, the predicted survival probabilities of this prognostic model should only be considered as an average estimate of patients with similar characteristics and should not be used for individual patients. Second, evidence suggests that combining an objective prognostic model with physicians' intuition may improve the accuracy of outcome prediction [13]. Whether combining this current prognostic model with physicians' intuition will improve its predictive performance further remains uncertain, but this merits further investigation. Third, although we studied a large cohort of critically ill patients, and also the case-mix, severity of illness, and in-hospital survival of this cohort is very similar to many other ICUs in Australia [38], validation of this model by other ICUs that have access to data linkage is essential to assess its generalizability. Finally, although the APACHE II prognostic model is still widely used for risk adjustment purposes in many ICUs [39], [40], it is possible that using newer prognostic models instead of the APACHE II prognostic model may improve our current model [41]. Similarly, the performance of the current model may be improved if we consider more predictors in the model although this will also increase the complexity of the model. In this regard, we hope that the PREDICT model developed in this study will be of value to others who aim to develop a new prognostic model to enhance our understanding of long-term survival of critically ill patients.In summary, Age, gender, co-morbidities, severity of acute illness, and the intensity and duration of intensive care therapy can be used to estimate long-term survival of critically ill patients. Age and co-morbidity are the most important determinants of the long-term prognosis of critically ill patients. The current prognostic model, the PREDICT model, provides a framework for prognostications and risk adjustment when long-term survival of critically ill patients is considered.\n\nREFERENCES:\n1. Acute Health Division DoHS\n1997\nReview of intensive care in Victoria [Phase 1 report].\nMelbourne\nDepartment of Human Services\n2. HalpernNABettesLGreensteinR\n1994\nFederal and nationwide intensive care units and healthcare costs: 1986–1992.\nCrit Care Med\n22\n2001\n2007\n7988140\n3. KvåleRFlaattenH\n2002\nChanges in intensive care from 1987 to 1997 - has outcome improved? A single centre study.\nIntensive Care Med\n28\n1110\n1116\n12185433\n4. PoisalJATrufferCSmithSSiskoACowanC\n2007\nHealth spending projections through 2016: modest changes obscure part D's impact.\nHealth Aff (Millwood)\n26\nw242\nw253\n17314105\n5. The Audit Commission\n1999\nCritical to Success. The place of efficient and effective critical care services within the acute hospital.\nLondon\nAudit Commission for Local Authorities and the National Health Service in England and Wales\n6. GirardTDKressJPFuchsBDThomasonJWSchweickertWD\n2008\nEfficacy and safety of a paired sedation and ventilator weaning protocol for mechanically ventilated patients in intensive care (Awakening and Breathing Controlled trial): a randomised controlled trial.\nLancet\n371\n126\n134\n18191684\n7. SinuffTAdhikariNKCookDJSchünemannHJGriffithLE\n2006\nMortality predictions in the intensive care unit: comparing physicians with scoring systems.\nCrit Care Med\n34\n878\n885\n16505667\n8. CookDJGuyattGHJaeschkeRReeveJSpanierA\n1995\nDeterminants in Canadian health care workers of the decision to withdraw life support from the critically ill.\nJAMA\n273\n703\n708\n7853627\n9. ElsteinASChristensenCCottrellJJPolsonANgM\n1999\nEffects of prognosis, perceived benefit and decision style upon decision making in critical care.\nCrit Care Med\n27\n58\n65\n9934894\n10. GarlandAConnorsAF\n2007\nPhysicians' influence over decisions to forego life support.\nJ Palliat Med\n10\n1298\n1305\n18095808\n11. ConnorsAFJr\n1999\nThe influence of prognosis on care decisions in the critically ill.\nCrit Care Med\n27\n5\n6\n9934872\n12. HamelMBPhillipsRSDavisRBDesbiensNConnorsAFJr\n1997\nOutcomes and cost-effectiveness of initiating dialysis and continuing aggressive care in seriously ill hospitalized adults. SUPPORT Investigators. Study to Understand Prognoses and Preferences for Outcomes and Risks of Treatments.\nAnn Intern Med\n127\n195\n202\n9245224\n13. KnausWAHarrellFEJrLynnJGoldmanLPhillipsRS\n1995\nThe SUPPORT prognostic model. Objective estimates of survival for seriously ill hospitalized adults. Study to understand prognoses and preferences for outcomes and risks of treatments.\nAnn Intern Med\n122\n191\n203\n7810938\n14. WrightJCPlenderleithLRidleySA\n2003\nLong-term survival following intensive care: subgroup analysis and comparison with the general population.\nAnaesthesia\n58\n637\n642\n12790812\n15. KnausWADraperEAWagnerDPZimmermanJE\n1985\nAPACHE II: a severity of disease classification system.\nCrit Care Med\n13\n818\n829\n3928249\n16. WilliamsTADobbGJFinnJCKnuimanMLeeKY\n2006\nData linkage enables evaluation of long-term survival after intensive care.\nAnaesth Intensive Care\n34\n307\n315\n16802482\n17. HoKMDobbGJKnuimanMFinnJLeeKY\n2006\nA comparison of admission and worst 24-hour Acute Physiology and Chronic Health Evaluation II scores in predicting hospital mortality: a retrospective cohort study.\nCrit Care\n10\nR4\n16356207\n18. WilliamsTADobbGJFinnJCKnuimanMWGeelhoedE\n2008\nDeterminants of long-term survival after intensive care.\nCrit Care Med\n36\n1523\n1530\n18434893\n19. CharlsonMEPompeiPAlesKLMacKenzieCR\n1987\nA new method of classifying prognostic comorbidity in longitudinal studies: development and validation.\nJ Chronic Dis\n40\n373\n383\n3558716\n20. HolmanCDBassAJRouseILHobbsMS\n1999\nPopulation-based linkage of health records in Western Australia: development of a health services research linked database.\nANZ J Public Health\n23\n453\n459\n21. CoxDR\n1972\nRegression models and life tables (with discussion).\nJournal of the Royal Statistical Society\n34\n187\n220\n22. LeeSJLindquistKSegalMRCovinskyKE\n2006\nDevelopment and validation of a prognostic index for 4-year mortality in older adults.\nJAMA\n295\n801\n808\n16478903\n23. PompeiPCharlsonMEAlesKMacKenzieCRNortonM\n1991\nRelating patient characteristics at the time of admission to outcomes of hospitalization.\nJ Clin Epidemiol\n44\n1063\n1069\n1940999\n24. WyattJCAltmanDG\n1995\nPrognostic models: clinically useful or quickly forgotten?\nBMJ\n311\n1539\n1541\n25. HarrellFEJr\n2001\nRegression Modeling Strategies.\nNew York\nSpringer\n26. HanleyJAMcNeilBJ\n1982\nThe meaning and use of the area under a receiver operating characteristic (ROC) curve.\nRadiology\n143\n29\n36\n7063747\n27. FeringaHHBaxJJHoeksSvan WaningVHElhendyA\n2007\nA prognostic risk index for long-term mortality in patients with peripheral arterial disease.\nArch Intern Med\n167\n2482\n2489\n18071171\n28. EfronBTibshiraniR\n1993\nAn Introduction to the Bootstrap.\nNew York\nChapman & Hall\n29. HoKM\n2007\nForest and funnel plots illustrated the calibration of a prognostic model: a descriptive study.\nJ Clin Epidemiol\n60\n746\n751\n17573992\n30. NagelkerkeNJ\n1991\nA note on a general definition of the coefficient of determination.\nBiometrika\n78\n691\n692\n31. HoKMFinnJKnuimanMWebbSA\n2007\nCombining multiple comorbidities with Acute Physiology Score to predict hospital mortality of critically ill patients: a linked data cohort study.\nAnaesthesia\n62\n1095\n1100\n17924888\n32. CabréLManceboJSolsonaJFSauraPGichI\nand the Bioethics Working Group of the SEMICYUC\n2005\nMulticenter study of the multiple organ dysfunction syndrome in intensive care units: the usefulness of Sequential Organ Failure Assessment scores in decision making.\nIntensive Care Med\n31\n927\n933\n15856171\n33. NathensABRivaraFPWangJMackenzieEJJurkovichGJ\n2008\nVariation in the rates of do not resuscitate orders after major trauma and the impact of intensive care unit environment.\nJ Trauma\n64\n81\n88\n18188103\n34. HoKMLeeKYWilliamsTFinnJKnuimanM\n2007\nComparison of Acute Physiology and Chronic Health Evaluation (APACHE) II score with organ failure scores to predict hospital mortality.\nAnaesthesia\n62\n466\n473\n17448058\n35. HoKM\n2007\nCombining sequential organ failure assessment (SOFA) score with acute physiology and chronic health evaluation (APACHE) II score to predict hospital mortality of critically ill patients.\nAnaesth Intensive Care\n35\n515\n521\n18020069\n36. HoKMLiangJ\n2004\nWithholding and withdrawal of therapy in New Zealand intensive care units (ICUs): a survey of clinical directors.\nAnaesth Intensive Care\n32\n781\n786\n15648988\n37. HoKMEnglishSBellJ\n2005\nThe involvement of intensive care nurses in end-of-life decisions: a nationwide survey.\nIntensive Care Med\n31\n668\n673\n15803296\n38. FinferSBellomoRLipmanJFrenchCDobbG\n2004\nAdult-population incidence of severe sepsis in Australian and New Zealand intensive care units.\nIntensive Care Med\n30\n589\n596\n14963646\n39. HarveySHarrisonDASingerMAshcroftJJonesCM\nPAC-Man study collaboration\n2005\nAssessment of the clinical effectiveness of pulmonary artery catheters in management of patients in intensive care (PAC-Man): a randomised controlled trial.\nLancet\n366\n472\n477\n16084255\n40. FinferSBellomoRBoyceNFrenchJMyburghJ\nSAFE Study Investigators\n2004\nA comparison of albumin and saline for fluid resuscitation in the intensive care unit.\nN Engl J Med\n350\n2247\n2256\n15163774\n41. ZimmermanJEKramerAAMcNairDSMalilaFM\n2006\nAcute Physiology and Chronic Health Evaluation (APACHE) IV: hospital mortality assessment for today's critically ill patients.\nCrit Care Med\n34\n1297\n1310\n16540951"
+}

batch_5/PMC2528950.json

@@ -0,0 +1,4 @@
+{
+    "id": "PMC2528950",
+    "text": "This is an academic paper. This paper has corpus identifier PMC2528950\nAUTHORS: Vicente Sentandreu, Nuria Jiménez-Hernández, Manuela Torres-Puente, María Alma Bracho, Ana Valero, María José Gosalbes, Enrique Ortega, Andrés Moya, Fernando González-Candelas\n\nABSTRACT:\nHepatitis C virus (HCV) is a major cause of liver disease worldwide and a potential cause of substantial morbidity and mortality in the future. HCV is characterized by a high level of genetic heterogeneity. Although homologous recombination has been demonstrated in many members of the family Flaviviridae, to which HCV belongs, there are only a few studies reporting recombination on natural populations of HCV, suggesting that these events are rare in vivo. Furthermore, these few studies have focused on recombination between different HCV genotypes/subtypes but there are no reports on the extent of intra-genotype or intra-subtype recombination between viral strains infecting the same patient. Given the important implications of recombination for RNA virus evolution, our aim in this study has been to assess the existence and eventually the frequency of intragenic recombination on HCV. For this, we retrospectively have analyzed two regions of the HCV genome (NS5A and E1-E2) in samples from two different groups: (i) patients infected only with HCV (either treated with interferon plus ribavirin or treatment naïve), and (ii) HCV-HIV co-infected patients (with and without treatment against HIV). The complete data set comprised 17712 sequences from 136 serum samples derived from 111 patients. Recombination analyses were performed using 6 different methods implemented in the program RDP3. Recombination events were considered when detected by at least 3 of the 6 methods used and were identified in 10.7% of the amplified samples, distributed throughout all the groups described and the two genomic regions studied. The resulting recombination events were further verified by detailed phylogenetic analyses. The complete experimental procedure was applied to an artificial mixture of relatively closely viral populations and the ensuing analyses failed to reveal artifactual recombination. From these results we conclude that recombination should be considered as a potentially relevant mechanism generating genetic variation in HCV and with important implications for the treatment of this infection.\n\nBODY:\nIntroductionHepatitis C virus (HCV) infection affects about 170 million people worldwide, about 3% of the world's population [1], and is the major cause of liver disease and a potential cause of substantial morbidity and mortality in the future [2]. Hepatitis C virus is the only species of the genus Hepacivirus within the family Flaviviridae. It has a single stranded, positive-sense, nonsegmented RNA genome of about 9600 nucleotides (nt) with a single, long open reading frame encoding a polyprotein of about 3000 amino acids with the gene order C-E1-E2-p7-NS2-NS3-NS4A-NS4B-NS5A-NS5B. The structural proteins are C (core) and E1 and E2 (envelope glycoproteins). The function of the p7 product is presently unknown. The NS2 through NS5 regions encode the non-structural proteins [3].Six major HCV genotypes and about 50 subtypes have been described [4], [5] based on levels of sequence divergence. HCV genotypes have been shown to be distributed over distinct geographical areas and although they share most basic biological features, there seems to be some differences in their susceptibility to interferon (IFN)-based therapies [6], [7]. Genotype 1 is the predominant variant in developed countries and shows the poorest response to therapy. Patients with genotypes 2 and 3 are also common in Europe, although with a lower frequency than genotype 1, and show the best response to IFN therapy.HCV is mainly transmitted by parenteral routes and differences in their transmission rates can be an important factor to explain the differences in prevalence of a genotype/subtype in different geographic regions [8]–[10]. Needle sharing among intravenous drug users (IDUs) currently represents the most common route of acquisition of HCV in the developed world [11]. Because HCV and human immunodeficiency virus (HIV) share blood-borne transmission routes, HIV/HCV co-infection is relatively frequent, especially in regions such as Spain, where the major proportion of newly diagnosed AIDS patients belong to the IDU category (44.2%) [12].Like HIV-1, HCV is characterized by high levels of genetic heterogeneity [13], [14] which impact heavily on different aspects such as HCV persistence, susceptibility to treatment, progression of infection, among others [15]–[17]. Although it is well known that recombinant forms of HIV-1 have a relatively high prevalence all over the world [18], there has been limited evidence of HCV recombination between different genotypes/subtypes [19], suggesting that these events are rare in vivo and that the resulting recombinants are usually not viable [14], [20], [21]. In the last few years, a few natural intergenotypic recombinants of HCV have been identified (RF1_2k/1b, RF2_1a/1b and RF3_2b/1b) and the crossover points have been mapped to the NS2, NS5B and NS3 regions, respectively [22]–[25]. Very recently, a new natural intergenotypic (2/5) recombinant of HCV has been found whose crossover point is located between genes NS2 and NS3 [26]. All these reports have described HCV recombination between different genotypes/subtypes but, to date, there is only a single case of an HCV intra-subtype recombinant strain, detected by analysis of NS5A sequences from intra-patient populations belonging to six patients undergoing anti-viral therapy [27].Given the important role that recombination seems to play in the evolution of RNA viruses [28], [29], by creating genetic variation, and the important implications that the production of new pathogenic recombinant strains could have, for example, on the development of vaccines to control RNA viruses, our aim in this study has been to assess the extent and, eventually, the frequency of intragenic recombination on HCV. For this, we have retrospectively analyzed a large data set (over 17700) of HCV sequences from intra-patient viral populations. These sequences were obtained from two separate studies of our group none of which was specifically designed for this objective [10], [78], [79, and unpublished results]. One included only HCV-monoinfected patients and the other HCV/HIV coinfected patients, with a common genome region for both studies (E1-E2 region) and another, the NS5A region, analyzed only in the former study. Both studies included treatment-naïve and patients non-responding to antiviral treatment. We found evidence of intrapatient recombination in HCV sequences from over 10% of the patients thus revealing that recombination in HCV can be a much more common phenomenon than previously recognized. The possibility of this result arising from artifacts during the experimental procedure has been considered by performing an “ad hoc” experiment in which serum samples from two closely related, but clearly differentiated patients were mixed in equal proportions and the resulting mixture was subjected to the same experimental procedure used in the previous analyses. No evidence of artifactual recombination was found.ResultsSequences used in this work were derived from two previous studies on HCV genetic variation before and after antiviral treatment. One study included HCV-monoinfected patients whereas the other analyzed HCV/HIV-coinfected individuals. None of these studies was designed with the goal of detecting recombination in HCV. In both cases the E1-E2 region of the HCV genome was analyzed by sequencing viral clones and the former study also included clones from the NS5A region. We obtained sequences from the E1-E2 region from 110 patients and a total of 136 samples (Table 1), since two samples (before and after antiviral treatment) were available for 26 patients, all from the HCV-monoinfection study. For the NS5A region we obtained cloned sequences from 78 patients and a total of 98 samples, since amplification failed for 4 and 1 samples from the pre- and post-treatment groups, respectively.10.1371/journal.pone.0003239.t001Table 1Summary of patients, sequence data sets and recombination analysis results.SourcePatientsGroupAmplified samplesAverage sequence/ sampleSDTotal seqsPositive casesFreq.E1-E2 regionHCV78HCV077105.7421.43814270.091HCVT26106.155.24268040.154HCV-HIV16HCV0-01628.693.8643320.12517HCV0-T1728.887.3249130.176TOTAL11113611746160.118NS5A regionHCV78HCV07363.3318.51462370.096HCVT2553.7218.00134320.080TOTAL7898596690.091The “source” column represents the study (HCV-monoinfection, HCV/HIV-coinfection) in which samples were obtained originally. For the HCV-monoinfection study, samples were obtained from 78 patients and the numbers indicated in the “amplified samples” column yielded successful amplificates before (HCV0) and after (HCVT) antiviral treatment. For the HCV/HIV study, the two groups correspond to patients without any treatment (HCV0-0) or having been treated for HIV infection (HCV0-T). SD indicates standard deviation of the number of sequences obtained from each sample and “Freq.” denotes the frequency of samples in which at least one recombination event has been detected.The average number of clones sequenced for each region, group and patient is described in Table 1. For the E1-E2 region, 11746 cloned sequences were obtained and average number of sequenced clones per sample for the treatment-naïve mono-infected group (HCV 0) was 105.74±21.43 whereas the average for HCV treated mono-infected patients who did not respond to a combined antiviral treatment with IFN-α plus ribavirin (HCV T) was 106.15±5.24. In the case of HCV/HIV coinfected patients, an average of 28.69±3.86 and 28.88±7.32 E1-E2 viral sequences were obtained for treatment naïve (HCV 0-0) and HIV treated patients (HCV 0-T), respectively. The data set for the NS5A region comprised 5966 clonal sequences. The average number of NS5A clones sequenced per sample for the non-treated group (HCV 0) was 63.33±18.51 and 53.72±18.00 for the antiviral treated group (HCV T).Putative recombination events between the viral strains infecting the same patient were found in 20 of the 111 patients studied (18.01%). These intragenic recombination events were detected in 25 of the 234 independent samples analyzed (10.7%). The detected recombination events belonged to HCV samples from all the infection groups described and the two genomic regions analyzed (E1-E2 and NS5A), as well as to different HCV subtypes (1a, 1b and 3a). Five events were detected among the E1-E2 sequences derived from HCV/HIV coinfected patients, 11 corresponded to events in the E1-E2 region from HCV-monoinfected patients, and the remaining 9 were detected in the NS5A region. No differences between the recombination frequencies from the two genomic regions analyzed were found neither for treatment-naïve (Mann-Whitney test: z = −0.104, p-value = 0.917) nor for interferon plus ribavirin treated groups (Mann-Whitney test: z = −0.810, p-value = 0.418). Figures 1–\n\n4 show in detail the results of recombination analyses only for samples in which the presence of putatively significant recombination events was detected. Intragenic recombination analyses were performed for each locus independently using 6 different methods for the detection of recombination implemented in the program RDP 3.0. We considered as significant recombination events only those for which the corrected probability for simultaneous inference of the event was lower than 0.05 and were significantly detected by at least 3 of the 6 methods used.10.1371/journal.pone.0003239.g001Figure 1Summary of positive recombination results in the E1-E2 region (HCV-monoinfected patients).The columns represent: i) patient group before (HCV 0) or after (HCV T) antiviral treatment; ii) patient code; iii) amplified sequences for this patient and region; iv) HCV genotype; v) p-values for the different recombination detection methods implemented in RDP 3.0 (1 = RDP, 2 = Geneconv, 3 = Bootscan, 4 = Maxchi, 5 = Chimera, and 6 = Siscan) using the following color coding: non-significant p-values (white filling), p<0.05 (grey) and p<0.01 (black); vi) recombination event number; vii) Bkpt 1, location range for breakpoint 1; viii) Bkpt 2, location range for breakpoint 2; ix) parent1 sequence (% similarity to daughter sequence); x) parent2 sequences(% similarity to daughter sequence); and xi) daughter sequence(s). Only events detected as significant by at least three methods are shown.10.1371/journal.pone.0003239.g002Figure 2Summary of positive recombination results in the E1-E2 region (HCV-HIV coinfected patients).See further details in legend to Figure 1.10.1371/journal.pone.0003239.g003Figure 3Summary of positive recombination results in the NS5A region.See further details in legend to Figure 1.10.1371/journal.pone.0003239.g004Figure 4Summary of the joint analysis of two time-point samples from the same patients resulting in positive detection of recombination.See further details in legend to Figure 1.A large proportion (84%, 21/25) of the samples in which recombination was detected included more than one recombination event (Figures 1–\n\n4). Each recombination event detected in a sample was described according to the following features: i) average p-value for the event and analysis method, ii) most likely parental and daughter sequences as well as similarity among them, and iii) most likely limits for the location of breakpoint(s) in the sequenced fragment.We found 46 single recombination events (17 in the NS5A and 29 in the E1-E2 region) and 12 double recombination events (3 in the NS5A and 9 in the E1-E2 region) in the 25 alignments where recombination was detected. A single recombination event was defined by a single, significant breakpoint detected in the daughter sequence following the described methodology, while a double event was defined as two significant breakpoints identified in the same daughter sequence, delimiting the extension of the recombinant fragment. When one of the parental sequences implicated in an event could not be inferred within the sequence alignment, it was denoted as “unknown parental”.Among the methods used, those detecting a larger number of significant events were Siscan, Chimera, and Maxchi, followed by Genconv and the phylogenetic methods, Bootscan and RDP, which detected very few events.Maximum likelihood phylogenetic trees were constructed for the two regions delimited for each single breakpoint event identified. For the double events detected, three maximum likelihood phylogenetic trees were constructed; one was derived from the alignment for the region involved in the recombinant segment (delimited by the two breakpoints detected) and the others from the two resulting flanking segments. Two different tests, Shimodaira–Hasegawa (SH) and Expected Likelihood Weight (ELW), were applied to the two or three resulting topologies for each single or double event, respectively, to further verify the results obtained with RDP3. Tests for double events were made between the recombinant segment topology and each of the two topologies derived from the flanking segments independently. These tests resulted in significant differences between both topologies in all single and double event cases (detailed results of the tests are available in the Supporting Information Table S2).Recombination in the E1-E2 RegionStatistically significant recombination events were detected by at least three methods in 11.8% (16/136) of the amplified samples in the E1-E2 region. The total number of events identified in these recombinant samples was 38, corresponding to 29 single and 9 double events, belonging to 15 patients from all the studied groups. The breakpoints detected were located mainly in the segments flanking the HVR1 (Figure 5). No differences in the distribution of breakpoints were found between the different studied groups, viral subtypes or treatment response. The frequency of double recombination events detected in this region was 23.7% (9/38) and the average length of the derived recombinant fragments was 147.5 nt (ranging from 66.5 to 212 nt), representing between 27.2% and 31.2% of the total region sequenced (543 nt for HCV/HIV co-infected and 472 nt for HCV single-infected patients). In 8 of the 9 double recombination event cases the derived recombinant regions comprised a large portion of the complete HVR1, while in the remaining case the recombinant fragment involved the entire HVR3 region (Figure 5).10.1371/journal.pone.0003239.g005Figure 5Location of recombination breakpoints detected in the E1E2region.In the analysis of HCV-monoinfected samples (103 E1-E2 amplified samples from 78 patients), the frequency of significant recombination cases was 9.1% (7/77) for treatment-naïve patients (HCV 0) and 15.4% (4/26) for IFN-α plus ribavirin treated patients (HCV T, Table 1). All samples in which recombination events were detected were derived from patients infected with genotype 1b of HCV, including both responders and non-responders to antiviral treatment, with the only exception of a sample from an untreated patient infected with genotype 1a. The frequencies of samples with detected recombination events among coinfected patients were 12.5% (2/16) and 17.64% (3/17) for treatment-naïve (HCV 0-0) and HAART-treated (HCV 0-T) patients, respectively (Table 1). These five samples were obtained from individuals infected with genotype 3a (3 patients) and genotype 1b (2 patients) of HCV. No significant differences were found in the frequencies of recombination-positive samples from naïve and treated groups for HCV mono-infected and HCV/HIV co-infected patients (Mann-Whitney test for single infected patients: z = −0.894 p-value = 0.371; and z = −0.406 and p-value = 0.685, for co-infected patients), although a larger proportion of recombination events was detected among samples from treated patients both in single-infected and in co-infected individuals. Similarly, no differences were found between HCV/HIV co-infected and HCV single infected, treatment-naïve groups (z = −0.417, p-value = 0.676). Complete Mann-Whitney tests results are available in Table S3 of the Supporting Information.\nFigure 6 shows an example of a recombination event detected by phylogenetic analysis in the E1-E2 region reflected in the incongruence (reciprocal tests with SH and ELW were highly significant, see Supplementary material) between the maximum likelihood trees derived from the two regions (delimited by nucleotides 1–264 and 265–534, respectively) defined by the breakpoints assigned to the recombination event from sample EC5703. Variable positions in the alignment of the viral sequences involved in this recombination event are shown in Figure 7, marking the recombinant parental sequences involved in the detected recombination event.10.1371/journal.pone.0003239.g006Figure 6Example of recombination detection in the E1E2 region (EC5703 sample).The phylogenetic tree on the left corresponds to the analysis of the complete region whereas the other two are derived from the two regions defined by the detected recombination event.10.1371/journal.pone.0003239.g007Figure 7Example of recombination detection in the E1E2 region (EC5703 sample).Variable nucleotide positions in the daughter (EC5703-46) and two parental sequences (EC5703-34 and EC5703-36).E1-E2 Region. Analysis of Two Time-Point SamplesTwo samples were available for 24 HCV-monoinfected patients who did not respond to antiviral treatment. One sample was taken before the onset of treatment (T0 sample) and the second one was obtained when it was discontinued, 6 or 12 months later (T1 sample). For these cases an additional analysis was performed by simultaneously considering all the sequences obtained from both samples. We detected significant recombination events in 4 patients, C29, A21, G16 and G26.One recombination event was detected in the joint analysis of the two samples from patient C29 (C29T0-T1).The parental sequences from this event were derived one from the T0 sample (before treatment) and the other from the T1 sample (taken 6 months later), while the daughter sequences were detected only in the T0 sample. The same event, involving the same daughter sequences, was also detected in the analysis of sequences obtained only from the T0 sample, although in this case only one parental sequence was identified while the other was described as “unknown parental” in the analysis (Figure 1).Similarly, in the joint analysis of the two samples from patient A21, we detected one significant recombination event whose corresponding breakpoint was located between positions 210 and 214. This event was originated by two parental sequences from the T0 sample while the resulting daughter sequences from such event (A21_T1-47 and A21_T1-80) corresponded to sequences from the sample obtained after six months of unsuccessful HCV antiviral treatment (Figure 4). The same event was detected in the analysis of sequences derived only from the T1 sample but, obviously, the identified parental sequences belonged also to this sample. However, when the parental similarities for both events (the one detected only with T1 sequences and the other joining T0 and T1 sequences) were compared, the similarity between daughter and the corresponding parental sequences for both cases were lower in the T1 sample (97.4 and 99.4% for the two putative parental sequences) than in the joint T0-T1 analysis (98.4 and 100%). In consequence, T1 daughter sequences derived from this recombination event had been more likely generated by parental sequences from the T0 sample than from those in the T1 one. This would imply that the recombination event occurred at the earlier time point and the resulting daughter sequences were able to persist in the viral population six months later and under antiviral treatment. Finally, in the case of patients G16 and G26, all significant events detected in the joint analysis involved sequences from the same time-point and were also detected in the analysis of only this sample.Recombination in the NS5A RegionSignificant events of recombination were detected by at least three methods in 9.18% (9/98) of the amplified samples for the NS5A region. The total number of recombination events detected in these samples was 20, with 17 single and 3 double events, and they were identified in samples from 8 patients, all infected with HCV genotype 1b, including treatment-naïve and non-responder patients. The breakpoint(s) detected for each event were located mainly in a segment flanking the PKR-BD (protein-kinase binding domain) and within the ISDR (interferon sensitivity-determining region) regions (Figure 8). The frequency of double events detected in this region was 15% (3/20) and the average recombinant fragment length was 315 nt (ranging from 305 to 337 nt) corresponding to the 42.4% of the total sequenced region (743 nt). In 2 of the 3 double event cases, the recombinant region comprised the complete PKR-BD region, including also the ISDR, while in the other case the recombinant region involved only the non ISDR fragment of PKR-BD region (Figure 8). No differences were observed in the distribution of breakpoints between treated and non-treated samples or between treatment responses. Similar proportions of recombination events were detected in samples from treatment-naïve patients (9.6%, 7/73) than from IFN-α plus ribavirin treated patients (8.3%, 2/24) and no significant differences due to the HCV treatment were found (z = −0.236 p-value = 0.813). Figure 9 presents an example of recombination event detected in the NS5A region from the T0 sample of patient G08. In this case, the event corresponded to a single recombination breakpoint located between positions 345 and 420 of this fragment. Maximum likelihood trees obtained for each region using the GTR model of evolution showed a clear phylogenetic incongruence between the two derived region trees (Table S2 and File S1 in Supplementary material). Variable positions in the alignment of the viral sequences involved in this recombination event are shown in Figure 10.10.1371/journal.pone.0003239.g008Figure 8Location of recombination breakpoints detected in the NS5a region.10.1371/journal.pone.0003239.g009Figure 9Example of recombination detection in the NS5a region (NG08T0 sample).The phylogenetic tree on the left corresponds to the analysis of the complete region whereas the other two are derived from the two regions defined by the detected recombination event.10.1371/journal.pone.0003239.g010Figure 10Example of recombination detection in the NS5a region (NG08T0 sample).Variable nucleotide positions in the daughter (NG08-47K) and two parental sequences (NG08-33K and NG08-63K).NS5A Region: Analysis of Two Time-Points SamplesTwo samples, one taken before (T0) and the other after unsuccessful antiviral treatment (6 or 12 months later, T1 or T2), were available for 24 HCV-monoinfected patients. The joint analysis of sequences from the two samples of each patient resulted in the detection of recombination events in only two cases (8.33%), for patients C29 and G07.The joint analysis of samples from patient C29 revealed several recombination events that involved parental sequences from samples T0 and T1 and daughter sequences only from the T0 sample. All these events were also detected in the analysis of the T0 sample but, again, with presumed parental sequences from this sample time point.For patient G07, different recombination events in the joint analysis of sequences from samples T0 and T2 were found. Two of these events were particularly interesting because they involved daughter sequences from the T2 sample with inferred parental sequences from the T0 sample which was obtained one year earlier. These parental sequences (T0-66 and T0-55) were involved in other recombination events detected in the T0 sample and the T2 resulting daughters (except T2-52, only detected in the joint analysis), T2-58 and T2-76 were detected in the T2 sample too, but parental inferred sequences for these events had lower similarity or were unknown (see Figures 3 and 4). Similarly to the previously commented case of E1-E2 sequences from two time-points samples of patient A21, the T2 daughter sequences obtained from this event were more likely generated by parental sequences from the T0 sample. Again, this indicates that the recombination event most likely took place at T0 and the resulting daughter sequences were able to persist for one year under HCV treatment.Joint Analysis of Recombination in the E1-E2 and NS5A RegionsWe obtained sequences for the E1-E2 and NS5A regions from 73 patients. These were all from the HCV monoinfected, non-treated group (HCV 0). Given the observed frequency of recombination events in the two analyzed regions, we calculated the expected probability of detecting recombination events simultaneously in both genome regions for the same patient. The frequency of samples with significant recombination events detected was 9.1% (7/77) and 9.6% (7/73), for the E1-E2 and NS5A regions, respectively (Table 1). Hence, the expected frequency of obtaining a patient with recombination events detected in the two genomic regions if these events were independent was 0.87%, whereas the observed frequency was 4.1% (3/73), 4.45 times higher than the expected value. Fisher's exact probability test resulted in a significant value with p = 0.03.We have also compared the frequencies of the recombinant sequences with those of the other sequence clones. A summary of these results is shown in Table 2. While the frequencies of detected recombinant sequences varied widely, from a minimum of 0.003 to a maximum of 0.516, none of the recombinant sequences was present in more than 2 copies in the corresponding population, which corresponded to frequencies from 0.001 up to 0.074. Nevertheless, these figures must be compared with those corresponding to the frequency of the most common variant in the corresponding population, which varied between 0.030 and 0.20. In the joint analyses of two samples from the same patient in which recombinant sequences were detected, the highest frequency of a recombinant haplotype was 0.010. The most common sequence in this patient was only three times larger (0.031) but the largest number of identical sequences (111) and the highest haplotype frequency (0.547) were found in this group.10.1371/journal.pone.0003239.t002Table 2Summary of frequency analysis of recombinant sequences in the different data sets where intrapatient recombination of HCV was detected.GroupPatientTotal Seqs.Diff. HaplotypesNumb. of recombFreq. recomb.RecombinantsNon-recombinantsSeqs.Freq.Seqs.Freq.NS5a region HCV 0A28555320.03610.01820.036C29873250.05710.011160.184C35555340.07310.01820.036C36756720.02710.01330.040G05625030.04820.03230.048G07494570.14310.02020.041G08553220.03610.018170.309HCV TC05T14339100.23310.02320.047G07T24945130.26510.02020.041E1-E2 regionHCV 0A16999240.04010.01040.040A211007220.02010.010140.140A281008740.04010.01030.030C28997720.02010.01030.030C291003720.02010.010200.200C32998040.04010.01050.051G081049360.05810.01040.038HCV TA211088450.04620.01960.056G171008130.03010.01060.060G161007930.03010.01060.060G261025810.01010.010200.196HCV 0-0C062521100.40020.08030.120V0353126160.51620.06520.065HCV 0-TC23292920.06910.03410.034C30271960.22220.07430.111C57413110.02410.02480.195Two joint samplesNS5AC29T0-T11475940.02710.007340.231G07T0-T29889130.13310.01030.031E1-E2A21T0-T120915620.01020.010140.067C29T0-T12007130.01510.005200.100G16T0-T12038220.01010.0051110.547G26T0-T1-T23023110.00310.003850.281For each patient and sample(s) considered the table reports the total number of sequences analyzed, the number of different haplotypes, the total number of recombinant sequences and their frequency in the sample. Additional information is given on the absolute and relative frequencies of the most frequent haplotype among recombinant and non-recombinant sequences.Experimental Analysis of Artifactual RecombinationIn order to evaluate the level and extent of artifactual generation of recombinant sequences during the reverse transcription and PCR amplification, we performed an additional experiment using the same conditions previously described. The only difference was the starting sample used which, in this case, was a mixture of sera derived from two patients. These were chosen to maximize the possibility of detection of recombinant sequences, which requires clearly differentiated viral populations, while simultaneously maximizing the possibility of contiguous pairing enabling RNA polymerase shifting from one template to another while keeping the same position and reading-frame. This was achieved by selecting two patients HCV-1b infected from a common source who had within-patient nucleotide diversities of 0.0046 and 0.0007, respectively, with a net nucleotide differentiation of 0.0247, as estimated from a previous analysis of 10 clones from each sample (unpublished results).To ensure equimolar amounts of viral RNAs from both samples in the final mixture, we set up several mixtures with varying amounts of each sample that were subjected to the same RT and PCR-amplification procedures for the E1-E2 region described above. The resulting amplificates were directly sequenced and one mixture with equal peak heights in the automated sequencer electrophoregram in the polymorphic positions was chosen for further analysis. We cloned and sequenced PCR products from the selected mixture as described, obtaining 142 sequences. These were analyzed using the 6 methods implemented in RDP as described above and none of them detected any putative recombination event.DiscussionSeveral studies have reported recombination in different Flaviviruses [30]–[33], but until recently no evidence for recombination in natural populations of HCV had been found. Since the first identification of an intergenotypic (2k/1b) HCV recombinant in St. Petersburg [22], several intergenotypic and intragenotypic HCV recombinant strains have been identified [23]–[26], [34], [35] therefore incorporating recombination as a mechanism generating genetic variation in HCV. More importantly, the identification of these recombinant strains demonstrates that HCV is capable of successfully completing all the stages (simultaneous infection of the same cell, simultaneous replication of both viral genomes, template shift by the viral RNA polymerase while keeping the correct reading frame, encapsidation and release of the recombinant genomes) in the process. The resulting products will then be subjected to the same population processes governing the maintenance, expansion or disappearance of new variants in a heterogeneous viral population.All these reports have focused on HCV recombination between different genotypes/subtypes, but to date, there is only one single case of putative HCV intra-subtype recombinant strain, detected by the analysis of NS5A sequences from six intra-patient populations undergoing antiviral therapy [27]. In the present study we have identified a high frequency of intrasubtype recombination events (18.01% of the analyzed patients 20/111) analyzing a large data set of HCV sequences from intra-patient viral populations obtained from patients belonging to different groups: HCV-monoinfected patients, naïve and non-responding to antiviral treatment, and HCV/HIV-coinfected patients, treatment-naïve and under HAART. The relevance of recombination in HCV for its long-term evolution and its incidence in different aspects of HCV infection have not been explored yet, but these findings support a potentially significant role for recombination in the evolution of HCV by creating genetic variation through the reshuffling of independently arisen variants.Although these studies have firmly established the possibility of recombination in HCV, as in other Flaviviruses, no general mechanism has been proposed yet (but see below), and despite extensive analysis of genetic variation in HCV there has been only one report of recombination between strains from the same subtype. Hence, it seems adequate to start discussing our findings in terms of them being real or an artefact and, since we naturally accept the first option, why it has been so difficult to detect.The first question is how to discard the possibility that the detected recombination events were false positives, resulting from PCR-mediated recombination, especially since some experiments have failed to experimentally induce and detect recombination in HCV [36]. We have performed one further such experiment, also with negative results. This is no proof of absence of artifacts in our experimental results, but they clearly indicate that, if present, recombinant sequences arising from the experimental procedures cannot account for the reported results. This conclusion is based on the following arguments, derived from considering different possible estimates of the artifactual recombination rate when no such event has been observed. If no false recombination event is observed among 143 clones then the probability of any such event must be lower than 1/143 (0.0075). With this upper limit, which would be our worst case scenario, under a Poisson distribution we would have expected to observe no false recombination events in 138.5 of the 234 independent samples analyzed with an average size of 75 sequences each. The actual number of samples with no recombinant sequences detected was 210; hence the presumed rate of artifactual recombination must be lower than 0.0075. If we had based our estimate of artifactual recombination rate on the number of negative cases (210), the inferred rate using again a Poisson distribution would be 0.0014. But with this rate it is not possible to account for the observation of 21 cases with two or more recombination events (Figures 1–\n3) nor of up to 9 events when sequences from the same patient taken at two different times were combined in a single analysis (Figure 4). Finally, should we consider the 58 events observed among 17712 sequences analyzed to be artifacts then we would expect 183 and 45 samples with none and one recombination event, respectively, while we obtained 210 and 3 cases. The discrepancy for cases with 2 or more events is even larger (5, 0.5 and 0, for 2, 3 and 4 events, while 13, 4 and 4 have been observed, respectively). In summary, the observed number and distribution of recombination events cannot be explained by artifacts during the experiments. Additional arguments for this conclusion are discussed next.There are two points in our experimental procedure that have been instrumental in obtaining the reported results and building our confidence in that they are not artifacts. First, instead of analyzing the full length master sequence of the viral distribution in each sample, we concentrated our efforts in sequencing a large number of clones in each of two genome regions. These were chosen on the basis of their biological relevance and not on the location of breakpoints identified in previous reports of recombination in HCV [23]–[26], [34], [35]. Secondly, we minimized the chances of detecting false, artifactual recombination by using long extension times [37] and a proofreading DNA polymerase (Pfu) [38] in our PCRs. Additional support for the actual occurrence of recombination events within HCV infected patients is provided by the following points: i) breakpoints or recombinant regions implicated in the recombinant events are not distributed at random along the analyzed genome regions (Figures 5 and 8). On the contrary, the same breakpoints have been detected in HCV samples from different patients, indicating the presence of recombination hotspots, and entire biologically relevant regions are comprised within the recombinant fragments detected. ii) We have found evolution of the daughter sequences derived from a recombination event, with similarity to the parental sequences lower than the expected 100% if the recombination event was PCR-mediated. In addition, all recombination generated sequences preserve the polyprotein reading frame, which is not expected in PCR-mediated recombination where selection cannot purge deleterious mutants. iii) We have detected several recombination events in which the most likely parental sequences belonged to a previous time sample. Through the joint analysis of sequences from the two sample time points from a single patient, we have identified recombinant daughter sequences resulting from the cross-over between sequences sampled six or twelve months before under HCV treatment (see Figure 4, results for A21_T0-T1 and G07_T0-T2). Parental and recombinant sequences were amplified and sequenced independently from the two different samples. Hence, these events probably occurred in the period between the two samples, and therefore the recombinant sequences were able to persist in the viral population under treatment and immunological selection pressures. iv) Finally, we have found patients with evidence for recombination events in the two genome regions analyzed 4.3 times more often than expected from the assumption of complete independence between both events. In PCR-mediated recombination we would expect a similar frequency of independent recombination events in both regions, but the joint expectation would be their product. A substantially larger value such as that observed might indicate the existence of some features, probably in the viral RNA-dependent RNA-polymerase (protein NS5B in HCV) that might either facilitate or prevent recombination, thus leading to the observed discrepancy.Intergenotype recombination in HCV has been related to homologous recombination during minus-strand synthesis via template switching [23], although this proposal relies on the presence of two hairpin structures in the vicinity of the inferred breakpoint of recombination. Recombination breakpoints for the E1-E2 region were mainly located in the conserved region between hypervariable regions 1 (HVR1) and 3 (HVR3), and the recombinant fragments from the double recombinant sequences detected spanned the entire HVR1 or HVR3 regions. For the NS5A region, breakpoints were concentrated mainly at the end of the ISDR and the PKR-BD, and double recombinant fragments comprised the entire ISDR or PKR-BD regions. This distribution of breakpoints could be explained by the operation in HCV, like in other DNA or RNA viruses, of an intermolecular homologous replicative recombination system. This mechanism is associated with extensive nucleotide sequence identity between the two parental genomes around the cross-over site and copy choice or template switching during the replication process, that involves detachment from a template of the polymerase complex with a nascent product, and continuation of the copying process at the same position of another template molecule.This is the first study where recombination in HCV has been detected at the intrapatient intragenic level (the lowest possible level of diversity) by analyzing a large number of HCV sequences (17712) and samples (234) from 111 patients assigned to different clinical groups. The detection of recombinant strains in 18% of the HCV-infected patients studied implies that recombination events between the viral strains infecting the same patient may be relatively frequent, and still more if we consider that this might be an underestimate of the true frequency of HCV recombination because of the difficulty in detecting recombination events if they occur between genetically very similar variants of the same subtype or in conserved genome regions.The frequent detection of recombination events in all patient groups described makes the capability of HCV to produce recombinant forms not only relatively frequent but also effective and, depending on the recombinant strains produced, it might be selectively advantageous. However, we did not find any evidence for an increased frequency of recombinant sequences which might be explained by their presumed selective advantage. A more adequate analysis of positive selection on this same set of sequences does not show any indication of a selective advantage of these recombinant sequences (Sentandreu et al., in prep.).Given the previously reported results, a higher frequency of recombinant HCV strains than actually identified, with only 5 inter-genotypic and 1 intra-genotypic recombinants reported, might be expected. This might be explained by three different factors: Firstly, in recombination events between subtype viral strains, such as those reported here, there is a trade-off between the capability of homologous replicative recombination event to occur, which likely depend on the intrinsic recombination rate of HCV, and the intra-patient viral diversity, because homologous recombination requires a minimum length of sequence identity. Secondly, there is another trade-off between the intra-patient viral diversity and the power for the detection of recombination by the different methods used. Finally, recombination events between different genotypes/subtypes co-infecting the same patient are most probably easier to detect but, on the other hand, they are less likely to occur, given the higher differentiation between strains of different subtypes than those from the same subtype resulting in less likely template switching and, additionally, if a recombination event does happen it will likely generate recombinant sequences less viable than the parental ones. The action of some or even all these factors thus provides an explanation for the low frequency of recombinant HCV sequences reported up to date.We have detected recombination events in all the genotypes/subtypes analyzed. However, given the direct relationship between intrapatient genetic variability and our ability to detect recombination events, those viral populations with higher rates of intrapatient genetic variation are more likely to be involved in the detection of a recombination event if it ever occurs. This large genetic variation at the intrapatient level is usually associated to long persistent infections and/or to coinfection and superinfection with a strain from the same or a different subtype/genotype [39]–[41]. Furthermore, there is increasing evidence for the presence of compartmentalization of HCV populations within infected patients [42]–[47], which would further facilitate lineage divergence within a patient. Despite some failed attempts to induce superinfection in cell cultures infected with HCV [48], [49] this and previous reports of recombination in HCV clearly demonstrate that this process is not fully blocked, and more research is certainly necessary to establish under what circumstances and in which cellular types is superinfection with HCV more likely to occur.Our aim in this study has been to detect the presence of intragenic recombination and also to assess the extension and the frequency of these recombination events at the subtype level, analyzing two genome regions, E1-E2 and NS5A, from HCV-infected patients with different clinical and epidemiological backgrounds: mono-infected with HCV or coinfected with HIV, with or without antiviral treatment, and responder or non-responders to this treatment. No significant differences in the frequency of recombination events were detected between the two genomic regions studied (9.1% E1-E2 and 9.6 % NS5A, Table 1) for the treatment-naïve, HCV-monoinfected patients group (HCV 0). On the contrary, large differences were found between these two regions for the group of HCV infected patients who did not respond to interferon plus ribavirin treatment therapy (HCV T). The observed recombination frequencies, 15.4% and 8.0% for the E1-E2 and NS5A regions, respectively, were not statistically significant due to the low sample size of the HCV T group but hinted to a larger genetic variation being generated in the E1-E2 region of non-responder patients. A positive relationship between genetic variability in this region [50]–[54] and in the whole HCV genome [55] with lack of response to antiviral treatment and progression of the infection has been reported, although there are contradictory observations [17], [56]–[58]. In consequence, if this association is real, an increased rate of recombination in this region might contribute to viral resistance to treatment and, consequently, to a higher probability of detection of recombination in non-responder patients. To date, the sensitivity of recombinant forms of HCV to pegylated interferon-based therapy is still unknown, but recombinant forms for HIV do not have the same sensitivity to anti-retroviral therapy than wild type HIV-1 clade B isolates [59]. Furthermore, it has been recently suggested [60] that recombination plays an important role in the evolution of drug resistance in HIV-1 under various realistic scenarios. No significant differences were found in the frequency of recombination events in the E1-E2 region between the two treatment groups in HIV coinfected patients. Nevertheless, a higher proportion of cases of recombination were detected in the HAART treated group (12.5% for HCV 0-0 and 17.6% for HCV 0-T). This might be related to an increase in the selection pressure due to the decrease in HIV load and the restoration of the immune system in these patients.There are also reports correlating the degree of variability of the ISDR and responsiveness to interferon treatment [61]–[63], again without a complete consensus. However, the association in this case is an opposite one to that found in the E1-E2 region. Departure from a canonical sequence at ISDR has been associated to decreased response to interferon treatment, mainly in Japanese populations [64]–[66], but opposite results have been obtained for European and American ones [67]–[74]. Nevertheless, recent meta-analyses of these reports have provided further support for this relationship [75]–[77]. Hence, the reversed relationship between genetic variability (departure from the canonical sequence) at ISDR and response to interferon treatment might be counterbalanced by recombination, which would allow the maintenance of the canonical sequence at ISDR while maintaining high levels of variation at other genome locations.Given the biological relevance described about the regions involved in the recombinant fragments, and the distribution of the recombinant cross-over points, it is clearly that the reported intragenic recombinant exploratory activity producing new genomic combinations could play an important role in the HCV evolution with significant consequences for treatment efficiency and the development of vaccines.Given the obtained results with a high frequency of HCV intragenic recombinant detected strains from patients belonging to the different described groups and the biological relevance related with the regions involved in this recombinant events, we conclude that, recombination must be considered as a potentially important mechanism generating genetic variation in HCV with serious implications in the vaccine and drug treatment optimal development and the response to antiviral therapy.Materials and MethodsPatients and Samples136 serum samples from 111 HCV-infected patients were analyzed in this study. Patients belonged to two different groups: (i) infected only with HCV, either treated with IFN-α plus ribavirin (denoted HCV T) or treatment naïve (HCV 0), and (ii) HCV-HIV co-infected patients with (HCV 0-T) and without (HCV 0-0) highly active antiretroviral treatment (HAART) against HIV. Samples from the former group were included in a molecular epidemiology study of HCV in the Comunidad Valenciana and have been described in detail elsewhere [10], [78], [79]. Samples from the second group were obtained from the Hospital General de Valencia (Valencia, Spain) and informed, written consent was obtained from all the patients. Both studies were approved by the corresponding ethics committees of the institutions involved. Treatment response for all HCV treated patients is shown in Table S1.1 (Supplementary Material). For non-responder patients from the HCV T group a second serum sample taken after interruption of treatment (6 or 12 months after its start) was available and included in the study.HCV genotyping was initially performed by a commercial reverse hybridization genotyping assay (Inno-LIPA HCV II; Innogenetics) and later confirmed by nucleotide sequence comparison in the analyzed genome regions. Genotype 1b represented 61.5% of the total HCV-monoinfected patients analyzed, and genotype 1a the remainder 38.5% whereas for HCV/HIV-coinfected patients, the frequency of the different HCV genotypes were 39.4%, 30.3%, 3.0%, 18.1% and 9.1% for genotypes 1a, 1b, 2b, 3a and 4, respectively.Two HCV genome regions were considered in this study. The first one corresponds to a fragment encompassing the genes encoding envelope glycoproteins E1 and E2. This fragment spanned from positions 1322 to 1793 in the HCV reference genome sequence [GenBank accession no. AF009606, 80] for HCV mono-infected samples (472 nt) and up to position 1855 in HCV-HIV co-infected samples (534 nt). This region will be referred to as “E1–E2 region”. The second region corresponded to a 743 nt fragment from gene NS5A (positions 6742–7484 in the HCV reference genome), referred to as “NS5A region”.These two genome fragments were chosen because of the biological relevance of the regions included therein. On the one hand, three hypervariable regions are included in the E1-E2 region: HVR1, which seems to be involved in target cell recognition and virus attachment [81]; HVR2, which could be involved in cell surface receptor binding [82]; and HVR3, which could play a role in the process of binding with host cell receptors and virus entry into host cells [83]. On the other hand, two remarkable domains are included in the NS5A region: the V3 domain, seemingly involved in responsiveness to interferon [70], [84], and PKR-BD, which contains the putative interferon sensitivity determining region (ISDR) and seems to be involved in blocking the cellular antiviral response induced by interferon [61], [85]–[87].RNA Extraction, cDNA Synthesis and AmplificationVirus RNA was extracted from 200 µl serum by using a High Pure Viral RNA kit (Roche). Reverse transcriptions were performed in a 20 µl volume containing 5 µl eluted RNA, 4 µl 5× RT buffer, 0.5 mM each deoxynucleotide, 0.5 µl random hexamers, 100 U Moloney murine leukaemia virus reverse transcriptase (Promega) and 20 U RNasin ribonuclease inhibitor (Promega). The reactions were incubated at 42°C for 45 min, followed by 3 min at 95°C.A first PCR round was then carried out in a 100 µl volume containing 10 µl of the reverse transcription product, 0.2 mM each dNTP, 400 nM genomic primer, 400 nM antigenomic primer and 1.25 U Pfu DNA polymerase (Promega). For the first set of samples, i.e. those obtained from HCV-monoinfected patients, we used the primers detailed in Table 1 of [88] unless specified otherwise. These primers yielded a 472 nt fragment for the E1–E2 region, while a 543 nt fragment was obtained in this region from HCV/HIV coinfected samples using primers 1-Em1 (5′-CGCATGGCHTGGRAYATGAT), 1-Em2 (5′-GGRATATGATRAATGAAYTGGTC) and 1-Em1 (5′-GGRGTGAARCARTAYACYGG) for genotypes 1a, 1b, 2b and 4, and primers 3-Eg1 (5′-CGWATGGCTTGGGAYATGAT), 3-Eg2 (5′-GGGAYATCATGATGAAYTGGT), 3-Ea1 (5′- GGRGTRAAGCAGTABACRGG) for genotype 3. For region NS5A, subtype 1a: 1-Ng1, 2-Ng1, 1-Ng2, 2-Ng2, 1-Na and 2-Na. For the NS5A region we used primers Ng1 (59-TGGAYGGRGTRCGGYTGCACAGGTA), Ng2 (59-CAGGTACGCTCCRGYRTGCA) and Na (59-CCYTCRAGGGGGGGCAT), which yielded a 743 nt fragment. This region was analyzed only in HCV-monoinfected samples. In all cases, PCRs were performed in an Applied Biosystems 2400 thermal cycler as described [88].Cloning and Sequencing of Viral PopulationsAmplified DNA products for each region were purified and cloned directly into EcoRV-digested pBluescript II SK(+) phagemid (Stratagene). Cloned products for the E1–E2 region or NS5A region were sequenced by using vector-based primers KS and SK (Stratagene).Sequencing was carried out by using the ABI PRISM BigDye Terminator v3.0 system (Applied Biosystems) on an ABI 3700 automated sequencer. Sequences were verified and both strands were assembled using the STADEN package [89]. HCV sequences obtained in this study have been deposited in GenBank and the corresponding accession numbers are shown on Table S1 in the supplementary material along with the numbers of previously determined sequences [10], [78], [79].Sequence and Phylogenetic AnalysesSequence alignments were obtained with CLUSTALX v1.81 [90]. Optimal models of nucleotide substitution were assessed using the maximum likelihood approach implemented in Modeltest v3.7 [91]. Likelihood scores for each model were estimated in PAUP*4.0b10 [92] and the best model was determined using the Akaike Information Criterion (AIC) [93]. Maximum likelihood phylogenetic trees were obtained with PHYML 2.4.4 [94] using the previously determined models of nucleotide substitution for each genome region and sample, and support for the nodes were evaluated by bootstrapping with 1000 pseudorreplicates.Intrapatient RecombinationPutative recombination events in intrapatient sequence alignments of the two genome regions were detected using RDP 3.0b03 [95]. This program implements several methods for the identification of recombinant sequences and recombination breakpoints. We choose six of them: two phylogenetic methods, which infer recombination when different parts of the genome result in discordant topologies, RDP [96]; and Bootscanning [97]; and four nucleotide substitution methods, which examine the sequences either for a significant clustering of substitutions or for a fit to an expected statistical distribution: Maxchi [98], Chimaera [98], GeneConv [99] and Sis-scan [100].We only considered recombination events that were identified by at least three methods. Common settings for all methods were to consider sequences as linear, to require phylogenetic evidence, to polish breakpoints and to check alignment consistency. Statistical significance was set at the P<0.05 level, after considering Bonferroni correction for multiple comparisons as implemented in RDP. Consensus daughter sequences and breakpoints were determined whenever possible.In order to test the phylogenetic congruence of the two ML trees derived from each of the segments identified by the recombination breakpoints reported, we used TreePuzzle v.5.2 [101] to compare both phylogenetic trees using the SH [102] and the ELW tests [103].Supporting InformationTable S1Detailed information on patients, samples, sequences and accession numbers used in this research(0.07 MB XLS)Click here for additional data file.Table S2Summary of SH and ELW tests for alternative topologies derived from the recombination events detected(0.19 MB PDF)Click here for additional data file.Table S3Summary of Mann-Whitney tests for differences in recombination frequency between clinical groups considered in this study(0.09 MB DOC)Click here for additional data file.File S1Zip-compressed file with all the maximum likelihood trees used in the analyses(0.13 MB ZIP)Click here for additional data file.\n\nREFERENCES:\n1. ShepardCWFinelliLAlterMJ\n2005\nGlobal epidemiology of hepatitis C virus infection.\nLancet Infect Dis\n5\n558\n567\n16122679\n2. SimmondsP\n2004\nGenetic diversity and evolution of hepatitis C virus - 15 years on.\nJ Gen Virol\n85\n3173\n3188\n15483230\n3. Schulze zur WieschJSchmitzHBorowskiEBorowskiP\n2003\nThe proteins of the Hepatitis C virus: Their features and interactions with intracellular protein phosphorylation.\nArch Virol\n148\n1247\n1267\n12827459\n4. SimmondsPHolmesECChaTAChanSWMcOmishF\n1993\nClassification of hepatitis C virus into six major genotypes and a series of subtypes by phylogenetic analysis of the NS-5 region.\nJ Gen Virol\n74(Pt 11)\n2391\n2399\n8245854\n5. SimmondsPBukhJCombetCDeleageGEnomotoN\n2005\nConsensus proposals for a unified system of nomenclature of hepatitis C virus genotypes.\nHepatology\n42\n962\n973\n16149085\n6. HadziyannisSJSetteHJrMorganTRBalanVDiagoM\n2004\nPeginterferon-alpha2a and ribavirin combination therapy in chronic hepatitis C: a randomized study of treatment duration and ribavirin dose.\nAnn Intern Med\n140\n346\n355\n14996676\n7. HnatyszynHJ\n2005\nChronic hepatitis C and genotyping: the clinical significance of determining HCV genotypes.\nAntiviral Therapy\n10\n1\n11\n15751759\n8. PybusOGCochraneAHolmesECSimmondsP\n2005\nThe hepatitis C virus epidemic among injecting drug users.\nInfect Genet Evol\n5\n131\n139\n15639745\n9. ThomsonBJFinchRG\n2005\nHepatitis C virus infection.\nClin Microbiol Infect\n11\n86\n94\n15679481\n10. Jimenez-HernandezNTorres-PuenteMBrachoMAGarcia-RoblesIOrtegaE\n2007\nEpidemic dynamics of two coexisting hepatitis C virus subtypes.\nJ Gen Virol\n88\n123\n133\n17170444\n11. LauerGMWalkerBD\n2001\nHepatitis C virus infection.\nN Engl J Med\n345\n41\n52\n11439948\n12. Centro Nacional de Epidemiología\n2006\nVigilancia epidemiológica del SIDA en España.\nRegistro Nacional de casos de SIDA Informe Semestral n° 2: Accessible at http://cne.iscii.es\n13. OkamotoHKuraiKOkadaSIYamamotoKIizukaH\n1992\nFull-length sequence of hepatitis C virus genome having poor homology to reported isolates: Comparative study of four distinct genotypes.\nJ Gen Virol\n188\n331\n341\n14. SmithDBPathiranaSDavidsonFLawlorEPowerJ\n1997\nThe origin of hepatitis C virus genotypes.\nJ Gen Virol\n78\n321\n328\n9018053\n15. ZeinNN\n2000\nClinical significance of hepatitis C virus genotypes.\nClin Microbiol Rev\n13\n223\n235\n10755999\n16. FarciPShimodaACoianaADiazGPeddisG\n2000\nThe outcome of acute hepatitis C predicted by the evolution of the viral quasispecies.\nScience\n288\n339\n344\n10764648\n17. FarciPPurcellRH\n2000\nClinical significance of hepatitis C virus genotypes and quasispecies.\nSemin Liver Dis\n20\n103\n126\n10895435\n18. PeetersMCourgnaudVAbelaB\n2001\nGenetic Diversity of Lentiviruses in Non-Human Primates.\nAIDS Rev\n3\n3\n10\n19. YunZLaraCJohanssonBLorenzanadRISonnerborgA\n1996\nDiscrepancy of hepatitis C virus genotypes as determined by phylogenetic analysis of partial NS5 and core sequences.\nJ Med Virol\n49\n155\n160\n8818959\n20. SimmondsPAlbertiAAlterHJBoninoFBradleyDW\n1994\nA proposed system for the nomenclature of hepatitis C viral genotypes.\nHepatology\n19\n1321\n1324\n8175159\n21. ViazovSWidellANordenfeltE\n2000\nMixed infection with two types of hepatitis C virus is probably a rare event.\nInfection\n28\n21\n25\n10697786\n22. KalininaONorderHMukomolovSMagniusLO\n2002\nA natural intergenotypic recombinant of hepatitis C virus identified in St. Petersburg.\nJ Virol\n76\n4034\n4043\n11907242\n23. KalininaONorderHMagniusLO\n2004\nFull-length open reading frame of a recombinant hepatitis C virus strain from St Petersburg: proposed mechanism for its formation.\nJ Gen Virol\n85\n1853\n1857\n15218169\n24. ColinaRCasaneDVasquezSGarcia-AguirreLChungaA\n2004\nEvidence of intratypic recombination in natural populations of hepatitis C virus.\nJ Gen Virol\n85\n31\n37\n14718617\n25. KageyamaSAgdamagDMAlesnaETLeanoPSHerediaAM\n2006\nA natural inter-genotypic (2b/1b) recombinant of hepatitis C virus in the Philippines.\nJ Med Virol\n78\n1423\n1428\n16998890\n26. Legrand-AbravanelFClaudinonJNicotFDuboisMChapuy-RegaudS\n2007\nA new natural intergenotypic (2/5) recombinant of Hepatitis C virus.\nJ Virol\n81\n4357\n4362\n17267503\n27. MorenoMPCasaneDLopezLCristinaJ\n2006\nEvidence of recombination in quasispecies populations of an Hepatitis C Virus patient undergoing anti-viral therapy.\nVirol J\n3\n87\n17062150\n28. WorobeyMHolmesEC\n1999\nEvolutionary aspects of recombination in RNA viruses.\nJ Gen Virol\n80\n2535\n2543\n10573145\n29. MoyaAHolmesECGonzález-CandelasF\n2004\nThe population genetics and evolutionary epidemiology of RNA viruses.\nNat Rev Micro\n2\n279\n288\n30. BecherPOrlichMThielHJ\n2001\nRNA recombination between persisting pestivirus and a vaccine strain: generation of cytopathogenic virus and induction of lethal disease.\nJ Virol\n75\n6256\n6264\n11413291\n31. WorobeyMRambautAHolmesEC\n1999\nWidespread intra-serotype recombination in natural populations of dengue virus.\nProceedings of the National Academy of Sciences U S A\n96\n7352\n7357\n32. WorobeyMHolmesEC\n2001\nHomologous recombination in GB Virus C/Hepatitis G Virus.\nMol Biol Evol\n18\n254\n261\n11158384\n33. TwiddySSHolmesEC\n2003\nThe extent of homologous recombination in members of the genus Flavivirus.\nJ Gen Virol\n84\n429\n440\n12560576\n34. NoppornpanthSLienTXPoovorawanYSmitsSLOsterhausADME\n2006\nIdentification of a naturally occurring recombinant genotype 2/6 hepatitis C virus.\nJ Virol\n80\n7569\n7577\n16840336\n35. MoreauIHegartySLevisJSheehyPCrosbieO\n2006\nSerendipitous identification of natural Intergenotypic recombinants of hepatitis C in Ireland.\nVirol J\n3\n95\n17107614\n36. BernardinFHerringBPage-ShaferKKuikenCDelwartE\n2006\nAbsence of HCV viral recombination following superinfection.\nJ Viral Hep\n13\n532\n537\n37. JudoMSWedelABWilsonC\n1998\nStimulation and suppression of PCR-mediated recombination.\nNuc Acids Res\n26\n1819\n1825\n38. ShafikhaniS\n2002\nFactors affecting PCR-mediated recombination.\nEnviron Microbiol\n4\n482\n486\n12153589\n39. AccapezzatoDFravoliniFCasciaroMAParoliM\n2002\nHepatitis C flare due to superinfection by genotype 4 in an HCV genotype 1b chronic carrier.\nEur J Gastroenterol Hepatol\n14\n879\n881\n12172410\n40. BlackardJTShermanKE\n2007\nHepatitis C virus coinfection and superinfection.\nJ Infect Dis\n195\n519\n524\n17230411\n41. HerringBLPage-ShaferKToblerLHDelwartEL\n2004\nFrequent hepatitis C virus superinfection in injection drug users.\nJ Infect Dis\n190\n1396\n1403\n15378431\n42. FortonDMKarayiannisPMahmudNTaylor-RobinsonSDThomasHC\n2004\nIdentification of unique hepatitis C virus quasispecies in the central nervous system and comparative analysis of internal translational efficiency of brain, liver, and serum variants.\nJ Virol\n78\n5170\n5183\n15113899\n43. DahariHFeliuAGarcia-RetortilloMFornsXNeumannAU\n2005\nSecond hepatitis C replication compartment indicated by viral dynamics during liver transplantation.\nJ Hepatol\n42\n491\n498\n15763335\n44. Roque-AfonsoAMDucoulombierDDi LibertoGKaraRGigouM\n2005\nCompartmentalization of hepatitis C virus genotypes between plasma and peripheral blood mononuclear cells.\nJ Virol\n79\n6349\n6357\n15858018\n45. Di LibertoGRoque-AfonsoAMKaraRDucoulombierDFallotG\n2006\nClinical and therapeutic implications of hepatitis C virus compartmentalization.\nGastroenterology\n131\n76\n84\n16831592\n46. MinosseCCalcaterraSAbbateISelleriMZanirattiMS\n2006\nPossible compartmentalization of hepatitis C viral replication in the genital tract of HIV-1-coinfected women.\nJ Infect Dis\n194\n1529\n1536\n17083037\n47. BlackardJTHiasaYSmeatonLJamiesonDJRodriguezI\n2007\nCompartmentalization of hepatitis C virus (HCV) during HCV/HIV coinfection.\nJ Infect Dis\n195\n1765\n1773\n17492592\n48. SchallerTAppelNKoutsoudakisGKallisSLohmannV\n2007\nAnalysis of hepatitis C virus superinfection exclusion by using novel fluorochrome gene-tagged viral genomes.\nJ Virol\n81\n4591\n4603\n17301154\n49. TscherneDMEvansMJvon HahnTJonesCTStamatakiZ\n2007\nSuperinfection exclusion in cells infected with hepatitis C virus.\nJ Virol\n81\n3693\n3703\n17287280\n50. OkadaSAkahaneYSuzukiHOkamotoHMishiroS\n1992\nThe degree of variability in the amino terminal region of the E2/NS1 protein of hepatitis C virus correlates with responsiveness to interferon therapy in viremic patients.\nHepatology\n16\n619\n624\n1380477\n51. KoizumiKEnomotoNKurosakiMMurakamiTIzumiN\n1995\nDiversity of quasispecies in various disease stages of chronic hepatitis C virus infection and its significance in interferon treatment.\nHepatology\n22\n30\n35\n7541387\n52. Le GuenBSquadritoGNalpasBBerthelotPPolS\n1997\nHepatitis C virus genome complexity correlates with response to interferon therapy: a study in French patients with chronic hepatitis C.\nHepatology\n25\n1250\n1254\n9141446\n53. PolyakSJFaulknerGCarithersRLCoreyLGretchDR\n1997\nAssessment of hepatitis C virus quasispecies heterogeneity by gel shift analysis: correlation with response to interferon therapy.\nJ Infect Dis\n175\n1101\n1107\n9129072\n54. ChambersTJFanXDrollDAHembradorESlaterT\n2005\nQuasispecies heterogeneity within the E1/E2 region as a pretreatment variable during pegylated interferon therapy of chronic hepatitis C virus infection.\nJ Virol\n79\n3071\n3083\n15709027\n55. DonlinMJCannonNAYaoELiJWahedA\n2007\nPretreatment sequence diversity differences in the full-length Hepatitis C Virus open reading frame correlate with early response to therapy.\nJ Virol\n81\n8211\n8224\n17522222\n56. WeinerAJGeysenHMChristophersonCHallJEMasonTJ\n1992\nEvidence for immune selection of hepatitis C virus (HCV) putative envelope glycoprotein variants: potential role in chronic HCV infections.\nProc Natl Acad Sci U S A\n89\n3468\n3472\n1314389\n57. KatoNSekiyaHOotsuyamaYNakazawaTHijikataM\n1993\nHumoral immune response to hypervariable region 1 of the putative envelope glycoprotein (gp70) of hepatitis C virus.\nJ Virol\n67\n3923\n3930\n7685404\n58. TaniguchiSOkamotoHSakamotoMKojimaMTsudaF\n1993\nA structurally flexible and antigenically variable N-terminal domain of the hepatitis C virus E2/NS1 protein: implication for an escape from antibody.\nVirology\n195\n297\n301\n7686313\n59. FleuryHJToniTLanNTHungPVDeshpandeA\n2006\nSusceptibility to antiretroviral drugs of CRF01_AE, CRF02_AG, and subtype C viruses from untreated patients of Africa and Asia: comparative genotypic and phenotypic data.\nAIDS Res Hum Retroviruses\n22\n357\n366\n16623640\n60. Carvajal-RodriguezACrandallKAPosadaD\n2007\nRecombination favors the evolution of drug resistance in HIV-1 during antiretroviral therapy.\nInfect Genet Evol\n7\n476\n483\n17369105\n61. EnomotoNSakumaIAsahinaYKurosakiMMurakamiT\n1995\nComparison of full-length sequences of interferon-sensitive and resistant hepatitis C virus 1b. Sensitivity to interferon is conferred by amino acid substitutions in the NS5A region.\nJ Clin Invest\n96\n224\n230\n7542279\n62. SquadritoGLeoneFSartoriMNalpasBBerthelotP\n1997\nMutations in the nonstructural 5A region of hepatitis C virus and response of chronic hepatitis C to interferon alfa.\nGastroenterology\n113\n567\n572\n9247477\n63. SaizJCLopez-LabradorFXAmpurdanesSDopazoJFornsX\n1998\nThe prognostic relevance of the nonstructural 5A gene interferon sensitivity determining region is different in infections with genotype 1b and 3a isolates of hepatitis C virus.\nJ Infect Dis\n177\n839\n847\n9534954\n64. ChayamaKTsubotaAKobayashiMOkamotoKHashimotoM\n1997\nPretreatment virus load and multiple amino acid substitutions in the interferon sensitivity-determining region predict the outcome of interferon treatment in patients with chronic genotype 1b hepatitis C virus infection.\nHepatology\n25\n745\n749\n9049229\n65. WatanabeHEnomotoNNagayamaKIzumiNMarumoF\n2001\nNumber and position of mutations in the interferon (IFN) sensitivity-determining region of the gene for nonstructural protein 5A correlate with IFN efficacy in hepatitis C virus genotype 1b infection.\nJ Infect Dis\n183\n1195\n1203\n11262201\n66. MurayamaMKatanoYNakanoIIshigamiMHayashiK\n2007\nA mutation in the interferon sensitivity-determining region is associated with responsiveness to interferon-ribavirin combination therapy in chronic hepatitis patients infected with a Japan-specific subtype of hepatitis C virus genotype 1B.\nJ Med Virol\n79\n35\n40\n17133546\n67. HofgartnerWTPolyakSJSullivanDGCarithersRLJrGretchDR\n1997\nMutations in the NS5A gene of hepatitis C virus in North American patients infected with HCV genotype 1a or 1b.\nJ Med Virol\n53\n118\n126\n9334922\n68. KhorsiHCastelainSWyseurAIzopetJCanvaV\n1997\nMutations of hepatitis C virus 1b NS5A 2209-2248 amino acid sequence do not predict the response to recombinant interferon-alfa therapy in French patients.\nJ Hepatol\n27\n72\n77\n9252076\n69. ZeuzemSLeeJHRothWK\n1997\nMutations in the nonstructural 5A gene of European hepatitis C virus isolates and response to interferon alfa.\nHepatology\n25\n740\n744\n9049228\n70. DuverlieGKhorsiHCastelainSJaillonOIzopetJ\n1998\nSequence analysis of the NS5A protein of European hepatitis C virus 1b isolates and relation to interferon sensitivity.\nJ Gen Virol\n79\n1373\n1381\n9634077\n71. RispeterKLuMZibertAWieseMde OliveiraJM\n1998\nThe “interferon sensitivity determining region” of hepatitis C virus is a stable sequence element.\nJ Hepatol\n29\n352\n361\n9764980\n72. ChungRTMontoADienstagJLKaplanLM\n1999\nMutations in the NS5A region do not predict interferon-responsiveness in american patients infected with genotype 1b hepatitis C virus.\nJ Med Virol\n58\n353\n358\n10421401\n73. IbarrolaNMoreno-MonteagudoJASaizMGarcia-MonzonCSobrinoF\n1999\nResponse to retreatment with interferon-alpha plus ribavirin in chronic hepatitis C patients is independent of the NS5A gene nucleotide sequence.\nAm J Gastroenterol\n94\n2487\n2495\n10484013\n74. SquadritoGOrlandoMECacciolaIRumiMGArtiniM\n1999\nLong-term response to interferon alpha is unrelated to “interferon sensitivity determining region” variability in patients with chronic hepatitis C virus-1b infection.\nJ Hepatol\n30\n1023\n1027\n10406179\n75. WitherellGWBeinekeP\n2001\nStatistical analysis of combined substitutions in nonstructural 5A region of hepatitis C virus and interferon response.\nJ Med Virol\n63\n8\n16\n11130893\n76. PascuMMartusPHohneMWiedenmannBHopfU\n2004\nSustained virological response in hepatitis C virus type 1b infected patients is predicted by the number of mutations within the NS5A-ISDR: a meta-analysis focused on geographical differences.\nGut\n53\n1345\n1351\n15306598\n77. SchinkelJSpaanWJMKroesACM\n2001\nMutations in the NS5A gene and hepatitis C virus resistance to interferon therapy: an end to the controversy.\n8th International Symposium of Hepatitis C Virus and Related Viruses. P-266\n78. Torres-PuenteMCuevasJMJiménezNBrachoMAGarcía-RoblesI\n2007\nContribution of insertions and deletions to the variability of hepatitis C virus populations.\nJ Gen Virol\n88\n2198\n2203\n17622623\n79. Torres-PuenteMCuevasJMJiménezNBrachoMAGarcía-RoblesI\n2008\nGenetic variability in hepatitis C virus and its role in antiviral treatment response.\nJ Viral Hep\n15\n188\n199\n80. KuikenCCombetCBukhJShin-iTDeleageGMizokamiM\n2006\nA comprehensive system for consistent numbering of HCV sequences, proteins and epitopes.\nHepatology\n44\n1355\n1361\n17058236\n81. PeninFCombetCGermanidisGFrainaisPODeleageG\n2001\nConservation of the conformation and positive charges of hepatitis C virus E2 envelope glycoprotein hypervariable region 1 points to a role in cell attachment.\nJ Virol\n75\n5703\n5710\n11356980\n82. YagnikATLahmAMeolaARoccaseccaRMErcoleBB\n2000\nA model for the hepatitis C virus envelope glycoprotein E2.\nProteins\n40\n355\n366\n10861927\n83. TroeschMMeunierILapierrePLapointeNAlvarezF\n2006\nStudy of a novel hypervariable region in hepatitis C virus (HCV) E2 envelope glycoprotein.\nVirology\n352\n357\n367\n16781757\n84. DuranteMEFortonDMRuggieroGKarayiannisP\n2003\nHepatitis C virus E2 and NS5A region variability during sequential treatment with two interferon-alpha preparations.\nJ Med Virol\n70\n62\n73\n12629645\n85. EnomotoNSakumaIAsahinaYKurosakiMMurakamiT\n1996\nMutations in the nonstructural protein 5A gene and response to interferon in patients with chronic hepatitis C virus 1b infection.\nNew Engl J Med\n334\n77\n81\n8531962\n86. GaleMJKorthMJTangNMTanSLHopkinsDA\n1997\nEvidence that hepatitis C virus resistance to interferon is mediated through repression of the PKR protein kinase by the nonstructural 5A protein.\nVirology\n230\n217\n227\n9143277\n87. GaleMJrKatzeMG\n1998\nMolecular mechanisms of interferon resistance mediated by viral-directed inhibition of PKR, the interferon-induced protein kinase.\nPharmacol Ther\n78\n29\n46\n9593328\n88. BrachoMAGarcía-RoblesIJiménezNTorres-PuenteMMoyaA\n2004\nEffect of oligonucleotide primers in determining viral variability within hosts.\nVirol J 1\n13\n89. StadenRBealKBonfieldJ\n1999\nThe Staden package, 1998.\nMisenerSKrawetzS\nComputer Methods in Molecular Biology\nTotowa\nThe Humana Press Inc\n115\n130\n90. ThompsonJDGibsonTJPlewniakFJeanmouginFHigginsDG\n1997\nThe CLUSTAL_X windows interface: flexible strategies for multiple sequence alignment aided by quality analysis tools.\nNucl Acids Res\n25\n4876\n4882\n9396791\n91. PosadaDCrandallKA\n1998\nModeltest: testing the model of DNA substitution.\nBioinformatics\n14\n917\n918\n92. SwoffordDL\n2002\nPAUP*. Phylogenetic Analysis Using Parsimony (* and Other Methods), version 4.0beta [computer program]\nSunderland, MA\nSinauer Associates\n93. AkaikeH\n1974\nA new look at the statistical model identification.\nIEEE Trans Autom Control\n19\n716\n723\n94. GuindonSGascuelO\n2003\nA simple, fast, and accurate algorithm to estimate large phylogenies by maximum likelihood.\nSyst Biol\n52\n696\n704\n14530136\n95. MartinDPWilliamsonCPosadaD\n2005\nRDP2: recombination detection and analysis from sequence alignments.\nBioinformatics\n21\n260\n262\n15377507\n96. MartinDRybickiE\n2000\nRDP: detection of recombination amongst aligned sequences.\nBioinformatics\n16\n562\n563\n10980155\n97. SalminenMOCarrJKBurkeDSMcCutchanFE\n1995\nIdentification of breakpoints in intergenotypic recombinants of HIV type 1 by bootscanning.\nAIDS Res Hum Retroviruses\n11\n1423\n1425\n8573403\n98. SmithJM\n1992\nAnalyzing the mosaic structure of genes.\nJ Mol Evol\n34\n126\n129\n1556748\n99. PadidamMSawyerSFauquetCM\n1999\nPossible emergence of new geminiviruses by frequent recombination.\nVirology\n265\n218\n225\n10600594\n100. GibbsMJArmstrongJSGibbsAJ\n2000\nSister-scanning: a Monte Carlo procedure for assessing signals in recombinant sequences.\nBioinformatics\n16\n573\n582\n11038328\n101. SchmidtHAStrimmerKVingronMvon HaeselerA\n2002\nTREE-PUZZLE: maximum likelihood phylogenetic analysis using quartets and parallel computing.\nBioinformatics\n18\n502\n504\n11934758\n102. ShimodairaHHasegawaM\n1999\nMultiple comparisons of log-likelihoods with applications to phylogenetic inference.\nMol Biol Evol\n16\n1114\n1116\n103. StrimmerKRambautA\n2002\nInferring confidence sets of possibly misspecified gene trees.\nProc R Soc Lond Ser B\n269\n137\n142"
+}

batch_5/PMC2529286.json

@@ -0,0 +1,4 @@
+{
+    "id": "PMC2529286",
+    "text": "This is an academic paper. This paper has corpus identifier PMC2529286\nAUTHORS: Gian Beeli, Gianclaudio Casutt, Thomas Baumgartner, Lutz Jäncke\n\nABSTRACT:\nBackground\"The feeling of being there\" is one possible way to describe the phenomenon of feeling present in a virtual environment and to act as if this environment is real. One brain area, which is hypothesized to be critically involved in modulating this feeling (also called presence) is the dorso-lateral prefrontal cortex (dlPFC), an area also associated with the control of impulsive behavior.MethodsIn our experiment we applied transcranial direct current stimulation (tDCS) to the right dlPFC in order to modulate the experience of presence while watching a virtual roller coaster ride. During the ride we also registered electro-dermal activity. Subjects also performed a test measuring impulsiveness and answered a questionnaire about their presence feeling while they were exposed to the virtual roller coaster scenario.ResultsApplication of cathodal tDCS to the right dlPFC while subjects were exposed to a virtual roller coaster scenario modulates the electrodermal response to the virtual reality stimulus. In addition, measures reflecting impulsiveness were also modulated by application of cathodal tDCS to the right dlPFC.ConclusionModulating the activation with the right dlPFC results in substantial changes in responses of the vegetative nervous system and changed impulsiveness. The effects can be explained by theories discussing the top-down influence of the right dlPFC on the \"impulsive system\".\n\nBODY:\nBackgroundWhen we are watching a movie, reading a book or playing a computer game we sometimes experience these variants of virtual reality as if they were real. This subjective sensation of presence is referred as \"the feeling of being there\". From an earlier EEG (electroencephalography) study [1] we know that activations in certain brain areas (especially in the prefrontal cortex) are negatively correlated with the subjective feeling of presence in another space (spatial presence). The involvement in a virtual scene can be measured by questionnaires (e.g. MEC-SPQ [2]). Moreover, psychophysiological measures (e.g. electro-dermal activity or heart rate variations) are also used to indicate different presence states in a virtual environment (VE) of a person. Mostly, a higher involvement in the virtual reality scenario is accompanied by enhanced responses of the vegetative nervous system such as electrodermal responses and heart rate [1,3-5].In this study we will use \"transcranial direct current stimulation\" to modulate brain activation during the confrontation with a virtual reality scenario. \"Transcranial direct current stimulation\" (tDCS) non-invasively modulates the excitability of a brain region of interest by altering neuronal membrane potentials [6,7]. Anodal tDCS has been found to increase cortical excitability and the potentiation of N-methyl-D-aspartate (NMDA) receptor efficacy, while cathodal tDCS has been found to decrease cortical excitability. Several studies have shown that the effects caused by tDCS last several minutes beyond the period of tDCS application [6-9]. Until now several studies have shown that tDCS can modulate cognitive and behavioral skills associated with the targeted brain area. For example, anodal tDCS to the left prefrontal cortex was found to increase working memory performance [10] and verbal fluency [11]. Anodal tDCS to the motor cortex contralateral to stroke patients' paretic arm facilitated temporary motor recovery [12]. In addition, anodal stimulation of the left motor cortex in healthy subjects improved right-hand performance [13]. A very recent study demonstrated that anodal stimulation to the supramarginal gyrus enhanced tone memory performance in musical novices [14].In the context of these findings the question arises, whether the feeling of presence can be influenced by applying tDCS to brain areas known to be involved in the control of presence. In this study we focus on the dorso-lateral prefrontal cortex (dlPFC), which is known to be involved in controlling many higher-order behaviors. Typically it has been shown that this area is involved in selecting a possible range of responses and suppressing inappropriate ones [15]. In addition, it has been shown that this area is critically involved in the inhibition (and control) of impulsive behavior controlled by other brain-regions (e.g. the brainstem, basal ganglia; this system is sometimes called the „impulsive system\") [16].Thus, we anticipate that the dlPFC will be involved in the modulation of presence experience. If the dlPFC is activated there will be strong top-down control available inhibiting the automatically evoked presence feeling by the \"impulsive system\". When the dlPFC is deactivated the \"impulsive system\" can unfold its bottom-up activation with less top-down control of the dlPFC. If the dlPFC is indeed the critical area modulating presence feeling during the exposure of virtual environments the differential presence experience in kids, adolescents, and adults can be explained on the basis of the late maturing dlPFC [17]. The late myelination of the dlPFC can partly explain why adolescents' behavior is characterized by motivational difficulties, impulsivity and addiction (also in the context of video games and virtual scenes) [18].In our study we modulated the right dlPFC with tDCS while participants were watching a virtual roller coaster scene. In order to further evaluate the success of this modulation, we also conducted a classical Go-Nogo task. The performance in this test depends on the functioning of dlPFC [19] and indicates the degree of impulsivity. There is evidence, that the task performance in the Go-Nogo task can be influenced by tDCS application to the left dlPFC [20] and with other methods also on the right dlPFC (transcranial magnetic stimulation (TMS) [21]). In addition, it has been shown that the right dlPFC is involved in controlling risk-taking behavior [22,23] and reciprocal fairness [24].We hypothesize that the feeling of being present in the virtual environment is enhanced if the excitability (and thus the activation) of the dlPFC is decreased. In addition, lowered activation within in the dlPFC should also be accompanied by higher impulsiveness as measured with the Go-Nogo-task. On the other hand, if the excitability of the dlPFC (and thus the activation) is increased this should lead to a lowered presence experience and reduced impulsiveness.MethodsSubjectsThirty-five (17 female, 18 male) subjects participated in the experiment. Most of them being students of the University of Zurich. The mean age was 24.9 yr (standard deviation: ± 3.7 yr). All of the participants were classified as being consistent right-handed (CRH) using the Annett hand preference questionnaire [25]. No subject reported a history of neurological or psychiatric diseases and gave their informed consent for the participation in the experiment.tDCS applicationIn order to prevent an interaction between the two brain-hemispheres we decided to constrain tDCS to one hemisphere. In pilot experiments in our lab we found slightly stronger correlations between presence experience and brain activation on the right dlPFC than on the left side. Therefore, we only applied tDCS to the right dlPFC. The application side was at the FC3 electrode position of the international EEG 10–20-System. In order to constrain tDCS application to one hemisphere the reference electrode was placed on the ipsilateral mastoid. For tDCS application the \"DC stimulator\" by Eldith©  was used. The constant current was applied using two saline-soaked electrodes with a surface of 35 cm2. During the anodal tDCS mode, the anode electrode was positioned on FC3 and the cathode electrode on the ipsilateral mastoid. During the cathodal condition, the two electrodes were switched (cathode over FC3, anode over ipsilateral mastoid). tDCS application lasted 5.5 min at a constant current intensity of 1.5 mA. The system automatically turned off the stimulation when the electrical resistance was too high. For sham stimulation the stimulator was switched off.Virtual roller coasterThe subjects were sitting on a chair while watching three different rollercoaster scenarios on a 22-inch computer screen placed at a distance of 60 cm in front of them. The rollercoaster scenarios were taken from a commercially available rollercoaster simulation software . Realistic driving noises were presented on loudspeakers. Every scenario consisted of three different phases. It started with an \"ascending phase\" (30 s) followed by a \"dynamic phase\" with movements in different dimensions and very high speed (60 s) and an \"end-phase\" with low speed and without inclination (Figure 1).Figure 1Example still figures of the used rollercoaster scenario. Ascending phase (left, 30 s), dynamic phase (middle, 60 s), end phase (right, 12 s).Psychophysiological measuresDuring the roller coaster ride electro-dermal activity (EDA) and the electro-myogram (EMG) were registered. The EDA and EMG measurements were conducted using a commercially available device (PAR-PORT; Hogrefe Company, Germany). For EDA recording, electrodes were attached to the thenar and hypothenar areas on the palm of the left hand. EDA activity was quantified using two different measures. First, we measured skin conductance responses (SCR) to the roller coaster scenario. In addition, we measured skin conductance level (SCL) to measure the tonic level of skin conductance during the experimental sessions. SCL was measured as log-transformed mean EDA amplitude (log [EDAsumamp+1]). Log-transformation was used to normalize the SCL data. The EMG electrodes were attached at the left eyebrow muscle (musculus corrugator supercilii) and quantified as mean tonic EMG activity level at this site during the different experimental conditions.Go-Nogo taskThe Go-Nogo task was taken from a German standard battery used to test several executive and attentional functions (Testbatterie zur Aufmerksamkeitsprüfung, TAP, [26]). This test consisted of 5 types of stimuli including lines in different directions. The subjects were required to press a button if one of the two defined target stimuli were presented. In total 100 stimuli were presented, 40 of them were target stimuli. The number of false alarms (FA, button-press when seeing a non-target stimulus) indicates the degree of impulsivity.QuestionnairesSince presence is a subjective feeling (first person) it is also necessary to use questionnaires asking the subjects for their particular presence feeling during the different conditions. We used an adapted version of the spatial presence questionnaire MEC-SPQ [5]. The questionnaire was presented to the subjects immediately after the rollercoaster ride. Participants indicated their degree of presence on a visual analog scale. Moreover, the SAM (Self Assessment Manikin) was administered after each roller coaster ride in order to control for mood changes during the tDCS application. With the SAM experienced arousal and valence during the roller coaster presentation was measured. Although not being the main focus of this paper we used some subscales (\"thrill and adventure seeking scales\"; TAS) of the „sensation seeking questionnaire\" to control whether this trait might have an influence on presence experience [27,28].DesignWe used a repeated measurements design in which every subject was randomly assigned to the three different conditions (anodal, cathodal, sham). During each condition the subjects were exposed to the roller coaster scenario after they have received different tDCS treatments. Each condition comprised the tDCS application, followed by the Go-Nogo task, the roller coaster presentation and the final questionnaire measurement (Figure 2). Between each condition there was a break of 3.5 minutes without any task and tDCS application.Figure 2Experimental design. Sequence of the different tasks and tDCS applications. The time scale is in seconds. This sequence was repeated three times per subject for the three stimulating conditions (sham, anodal, cathodal).Statistical analysisThe number of false alarms, SCR, SCL, as well as ratings of valence, arousal, and presence were subjected to one-way repeated measurements ANOVAs with three levels (sham, anodal, and cathodal). Before ANOVA analysis the variances were evaluated for homoscedasticity and we also checked the data for normal distribution. There was no significant deviation from homoscedasticity making it unnecessary to use specific corrections (e.g., Greenhouse Geisser corrections). In addition, the data were also evaluated whether they deviate from normal distribution. Since there were no strong deviations from normal distribution we deemed the ANOVA as an appropriate method to analyze this data set. In case of significant main effects subsequent post-hoc t-tests were conducted using the Bonferroni-Holm procedure [29]. A p value < = 0.05 was used as statistical threshold.ResultsGo-Nogo taskFigure 3 shows the results of the Go-Nogo task separately for the three experimental conditions. During cathodal tDCS participants generated more often false alarms indicating a tendency for impulsive behavior. There was no change in performance during anodal stimulation. Subjecting the number of false alarms to a one-way repeated measurements ANOVA revealed a significant between-condition difference for the number of false alarms (F(2,68) = 3.653; p = 0.03). Subsequently conducted post hoc tests revealed significant differences between false alarms obtained during \"sham\" vs. \"cathodal\" (p = 0.032) and \"anodal\"- vs. \"cathodal\" (p = 0.033).Figure 3Number of false alarms (FA) in the different conditions in the Go-Nogo task. Applying cathodal tDCS to the right dlPFC led to an enhanced number of FA (p < .03) compared to sham and anodal-Stimulation. Depicted are means of FA (± SE).Psychophysiological measuresDue to artifact contamination only data of 29 participants could be used for analysis of psychophysiological measures. The EMG measure during the roller coaster ride showed no significant difference during the three tDCS conditions. For SCR a significant between-condition difference emerged. Figure 4 shows a clear SCR at the start of the virtual roller coaster ride. In the first 30 seconds of the rollercoaster ride (ascending phase), subjects showed stronger SCR during cathodal tDCS (F(2,56) = 3.237; p = 0.047; cathodal > sham: p = 0.021). The one-way ANOVA conducted for the SCL data did not reveal significant differences (F(2,56) = 3.016; p = 0.057).Figure 4Skin conductance level of the first 30 seconds of the rollercoaster ride. Cathodal tDCS application (inhibition) to the right dlPFC led to an enhanced skin conductance response (SCR).Figure 5 shows the mean peak SCL measured during the first 12 seconds of the roller coaster ride. Peak SCLs were significantly different in the three conditions (F(2,56) = 4.958 p = 0.01) with a higher peak during cathodal stimulation vs. anodal stimulation (p = 0.005) and vs. sham stimulation (p = 0.012).Figure 5Peak of skin conductance level (maximum SCL) in the first 12 seconds of the roller coaster ride. Cathodal tDCS application (inhibition of the dlPFC) leads to significantly enhanced maximum SCL (p < .01) during the ascending phase in the virtual roller coaster compared to sham and anodal stimulation. Depicted are means of the SCL (± SE).Presence- and personality questionnairesThe questionnaire data showed no significant differences between the different tDCS conditions. Interestingly, there was no significant correlation between the subscale \"thrill and adventure seeking\" (TAS) and the SCR measures (p > .33). The self-assessment-manikin (SAM) showed no differences with respect to the experienced valence of roller coaster scenarios during the different conditions (F(2,66) = 1.617; p = 0.206). However, there was a tendency for slightly increased subjective arousal levels during tDCS application compared to sham stimulation (F(2,66) = 2.532; p = 0.087).Correlation between Go-Nogo task performance and SCRThere was also a significant correlation between the number of false alarms (taken as a measure for impulsiveness) and the SCR measures (r = 0.42, p < 0.02). Thus, if participants act more impulsively in the Go-Nogo task, they also show stronger SCR measures in the ascending phase of the roller coaster ride.DiscussionOur study demonstrates that the application of cathodal tDCS to the right dlPFC modulates the degree of impulsivity (as measured with the number of false alarms in the Go-Nogo task). It follows from the current interpretation of the effect of cathodal tDCS on the neural system underlying the cathode that cathodal tDCS downregulates the dlPFC, with a resultant reduction in neural activation in this area. In line with this interpretation, we suggest that the dlPFC exerts less top-down control over the \"impulsive system\", increasing the likelihood therefore of impulsive behavior [for a summary see [30]]. Applying anodal tDCS to the dlPFC did not affect impulsiveness as indicated by Go-Nogo performance. We hypothesized at the beginning of the study that this kind of tDCS application would lead to increased neural activation of the dlPFC, this in turn would facilitate increased top-down regulation of the \"impulsive system\" in the form of reduced impulsiveness. The reason that we did not obtain this result is probably due to the fact that the task was too easy with too few false alarms, even in the sham condition. Thus, there was a kind of \"floor effect\" without any opportunity to decrease the number of false alarms. This might also explain the different findings in previous studies using a more difficult and slightly different versions of the Go-Nogo task [20,21].Besides the differential effect of tDCS on the number of false alarms, we also obtained different results for the skin conductance responses (SCR) used to indicate the reactivity of the vegetative nervous system. Application of cathodal tDCS to the dlPFC elicited increased SCRs while the subjects were exposed to the roller coaster scenario. This differential SCR was only present in the first phase of the roller coaster ride during which the virtual cab was ascending to the top of the roller coaster course (ascending phase). The strong skin conductance response during the ascent phase of the roller coaster ride might be associated with the anticipation of the following dynamic phase with its ups and downs and with the experience or expectation of bodily arousal in a real roller coaster.The correlation between the number of false alarms in the Go-Nogo task and the SCR measures indicates that impulsive behavior and autonomic responses can be influenced by tDCS application, and that both reactions might depend on the activation in the right dlPFC. However, further investigation is needed to develop a better understanding of the relationship between the inhibition of impulsive behavior and vegetative reactions.The fact that the personality trait TAS (thrill and adventure seeking) had no impact on our measures (e.g., skin conductance or number of false alarms) indicates that the application of tDCS is independent of the \"sensation seeking\" personality. Nevertheless, there might still be different effects on patients as found in patients with major depression [20].A further important result of the present study is that there are significant differences in vegetative reactions in the hypothesized direction associated with tDCS application, but that the subjective reports (measures with questionnaires) did not differ for the different conditions. This shows that subjective measures might not be reliable in the context of presence research (especially because the involvement in a VE requires low cognitive control) and that brain stimulation can lead to a change in bodily reactions without influencing subjective reports.ConclusionApplication of tDCS to the right dlPFC can influence the vegetative reactions while watching a virtual roller coaster scene as well as the number of false alarms in a standard Go-Nogo discrimination task commonly used as a behavioral measure of impulsivity. The measured vegetative effects during viewing of the virtual roller coaster ride and concomitant tDCS application had no impact on self-reported experience of presence. The cathodal (inhibiting) condition leads to enhanced impulsivity and higher skin conductance responses. There was no effect on skin conductance and impulsivity during the anodal (exciting) condition.Competing interestsThe authors declare that they have no competing interests.Authors' contributionsGB participated in the design of the study, performed the statistical analysis and drafted the manuscript. GC participated in the design, carried out the experiments and performed the statistical analysis. TB participated in the design. LJ participated in the design, the statistical analysis and drafted the manuscript. All authors read and approved the final manuscript.\n\nREFERENCES:\nNo References"
+}

batch_5/PMC2529309.json

@@ -0,0 +1,4 @@
+{
+    "id": "PMC2529309",
+    "text": "This is an academic paper. This paper has corpus identifier PMC2529309\nAUTHORS: Franck Prugnolle, Kate McGee, Jon Keebler, Philip Awadalla\n\nABSTRACT:\nBackgroundMalaria kills more people worldwide than all inherited human genetic disorders combined. To characterize how the parasites causing this disease adapt to different host environments, we compared the evolutionary genomics of two distinct groups of malaria pathogens in order to identify critical properties associated with infection of different hosts: those parasites infecting hominids (Plasmodium falciparum and P. reichenowi) versus parasites infecting rodent hosts (P. yoelii yoelii, P. berghei, and P. chabaudi). Adaptation by the parasite to its host is likely highly critical to the evolution of these species.ResultsOur comparative analysis suggests that patterns of molecular evolution in the hominid parasite lineage are generally similar to those of the rodent lineage but distinct in several aspects. The most rapidly evolving genes in both lineages are those involved in host-parasite interactions as well as those that show the lowest expression levels. However, we found that, similar to their respective mammal host lineages, parasite genomes infecting hominids are generally less constrained, evolving at faster rates, and accumulating more deleterious mutations than those infecting murids, which may reflect an historical lower effective size of the hominid lineage and relaxed host-driven selective pressures.ConclusionOur study highlights for the first time the differences in trends and rates of evolution in Plasmodium lineages infecting different hosts and emphasizes the potential importance of the variation in effective size between lineages to explain variation in selective constraints among genomes.\n\nBODY:\nBackgroundA number of useful evolutionary parameters can be estimated from between species comparisons of genome-wide divergence patterns: the magnitude of positive and negative (purifying) selection, variation in selection across different lineages, chromosomes, gene families and individual genes as well as the number of genes involved in the speciation process and adaptation to new environments.Comparative approaches have revealed that genes involved in immunity or in host defenses tend to exhibit the highest rate of evolution in the genome of different species. The arms race between hosts and parasites is generally invoked to explain this rapid evolution of genes involved in immune defense [1]. Pathogens continuously evolve to escape the defense of the host and the host, in turn, responds by modifying its defense. At the molecular level, this cycle of environmental changes means that new mutations are continuously tested and fixed by selection if adaptive, which translates into higher rates of molecular evolution in genes controlling immunity in the hosts.While accelerated evolution of genes involved in immunity is common in mammals, the relative rate of evolution of those genes may vary from one phylogenetic lineage to another. This is the case for hominids (human and chimpanzee) compared to rodents (mouse and rats) where genes involved in immune defense show an accelerated rate of evolution in murids compared to hominids, suggesting that the immune system of murids has undergone more extensive specific innovations [2].Hosts belonging to different lineages can therefore represent different environments for parasites to adapt. How do parasites preferentially infecting different host lineages respond to these different environments? Do parasites infecting different host lineages show lineage specific rates of evolution?Plasmodium is a practical case study for genome evolution of parasites specifically infecting different host lineages. The genomes of two parasite species infecting only hominids (Plasmodium falciparum [3], P. reichenowi [4]) and three species preferentially infecting rodent hosts (P. yoelii yoelii, P. chabaudi and P. berghei) [5] are now partially or completely sequenced. Although these two groups of species might be subject to similar selective pressures acting either on the genome as a whole or on genes with similar function across species, some aspects of their genomes, such as genes associated with evading host immunity, may evolve in a unique manner.In this paper we systematically analyze and compare the rate of evolution of protein-coding genes in the parasites infecting hominids (hereafter called the hominid parasite lineage) and in those infecting rodents (rodent or murid parasite lineage) (Figure 1). We explore and compare the adaptive rate of evolution of genes in both groups based on their function and timing or level of expression, factors that may explain variation in the rate of evolution among different genes between the different lineages.Figure 1Schematic representation of the phylogenetic relationship between hominid and rodent Plasmodium lineages (adapted from[27]).As for their two mammal host lineages (i.e. hominids and rodents) [2], our study reveals, in particular, that the evolution of the hominid lineage parasite genomes was less constrained than the evolution of those parasites infecting the murid lineage, which likely reflects a lower effective population size in hominid parasites (specifically P. falciparum).Results and DiscussionBy taking the complete set of coding genes of P. falciparum and aligning it to a partial genome shotgun of P. reichenowi (covering approximately 1/3 of the entire genome) available in PlasmoDB, we identified a set of 843 pairs of genes with unambiguous orthology for which it was possible to generate high quality sequence alignments covering virtually the entire coding region (see Methods and Additional file 1). The same procedure retrieved 3060 triplets of orthologues for rodent malaria parasites P. yoelii yoelii, P. berghei and P. chabaudi, distributed throughout these genomes (Additional file 2).Average rates of evolution: evolutionary constraints and selection on amino-acid sites within the hominid and murid lineagesAnalyzing estimates of ω (dN/dS) in both hominid (ωHominids) and rodent (ωrodents) parasite species independently made it possible to study how evolutionary constraints and selection vary across both clades. In the hominid lineage, the average ωHominids was estimated at 0.21 (Table 1), which is in congruence with previous estimates [4]. This excludes four genes with estimates of ω higher than 500 that had very low observed dS estimates (including these 4 genes, the average was ~4.68) and five genes with an undefined ratio (dS = 0). Among the 843 pairs of orthologues analyzed between P. falciparum and P. reichenowi, only ten pairs displayed a higher ratio than 1 but none were statistically significant (p-values less than 0.05 as determined by a likelihood ratio test). Most genes were conserved with ω significantly lower than one. A total of 719 genes, 85% of the genes analyzed, were in fact determined to be under purifying selection with ω significantly less than one.Table 1Evolutionary rates in hominid and rodent's Plasmodium lineageHominid lineageRodent (murid) lineagedN0.012 ± 0.000440.026 ± 0.00041dS0.057 ± 0.00130.20 ± 0.0027dN/dS (ω)0.21 ± 0.00680.13 ± 0.0023Maximum likelihood estimates of the rates of evolution (± standard error) in protein coding genes for hominid (P. falciparum – P. rechenowi) and rodent (P. yoelii yoelii – P. berghei – P. chabaudi) lineages. The difference between dN/dS in hominid and rodent lineages is significant. Note that the ratio of the means is not equivalent to the mean of the ratios.The average ω Murids for the 3060 genes analyzed in P. yoelii yoelii, P. chabaudi and P. berghei, was estimated as 0.13 (Table 1), significantly lower than in parasites infecting hominids (Wilcoxon test over all genes, p-value < 10-4). Only 4 genes displayed a ratio greater than 1, but none were significant. In fact, more than 98% of the genes displayed a ratio significantly lower than 1.Several observations suggest that the difference observed between the hominid and the rodent lineage is not due to the number of species aligned in each lineage nor their phylogenetic distances. First, the ω obtained between any pair of rodent parasite species, are similar, and lower than the estimate obtained for P. falciparum and P. reichenowi (data from [5]: P. yoelii yoelii and P. chabaudi: ω = 0.11, P. yoelii yoelii and P. berghei: ω = 0.16 and P. chabaudi and P. berghei: ω = 0.13). Second, the marked difference between hominid and rodent malaria parasites still held when comparisons between lineages are made in a paired way using only those genes from hominid and rodent lineages of known orthology (orthology between P. falciparum and P. yoelii yoelii was retrieved from [6]; Wilcoxon rank sum test; n (genes) = 263; p-value = 0.002). Therefore, there is an excess of about 25% of amino-acid altering substitutions, relative to synonymous substitutions, in the hominid lineage compared to the rodent lineage.Interestingly, the host lineages of these two groups of parasites show similar differences in their rate of molecular evolution. Hominids show an average ω of 0.20, over the whole genome, while murids show an average ratio close to 0.13 [2,7]. As in the case of Plasmodium parasites, the difference between both estimates is highly significant. This difference in the rate of molecular evolution among the host lineages could be the result of two different processes: 1) a relaxation in the selective constraints acting on the amino-acid sequence in the host hominid lineage, relative to the murid lineage, subsequent to a reduction in their effective population size or 2) an acceleration in the global rate of adaptive evolution in hominids. Genetic evidence favors the first hypothesis [8,2,9]. Does the same explanation hold for parasites?Comparing genetic variation at synonymous and non-synonymous sites within (polymorphism) and between species (divergence) can help distinguish between the two hypotheses above [10]. When most variation is neutral, the ratio of the number of nonsynonymous to synonymous polymorphisms observed within populations should be the same as the ratio of divergence between species [10]. Mu and collaborators [11] analyzed genome-wide polymorphism of 5 P. falciparum isolates distributed globally. Jeffares and collaborators also analyzed polymorphisms using three different African isolates [4]. In total, we obtained divergence and polymorphism data for 518 genes using Mu et al [11]'s data and 839 genes using Jeffares et al. [4]'s data. The ratio of the number of nonsynonymous to synonymous polymorphism was 2.3 for the first dataset and 2.7 for the second one. Comparatively, the ratio of divergence we observed among P. falciparum-P. reichenowi was only 1.1 and 1.3, respectively, thus indicating a 2-fold increase in the number of nonsynonymous polymorphisms. Although some amino-acid substitutions observed among species are likely adaptive, this observation supports selection being less efficient in removing segregating deleterious amino-acids in P. falciparum. A reduced effective size in hominid parasites compared to the murid lineage might be one explanation for the observed difference in ω. This scenario is supported by different studies suggesting the existence of an historical low effective population size in P. falciparum[12]. Unfortunately, no such population study exists for rodent malaria species. The evolution of both hominids and their malaria parasites appears to be less constrained than that of murids, which might reflect, for both the host and the parasite, a small historical effective size and an evolution mainly driven by slightly deleterious mutations [13].Variation in evolutionary rates across functionally different genesWe then asked whether specific groups of genes evolved differently from the rest of the genome. In particular, we searched for groups of genes that could have experienced an accelerated evolution in one lineage compared to the other, thus leading to a higher difference in their amino-acid substitution rate between lineages than expected given the difference observed across the genome.To do so, we searched for variation in ω among different functional categories of genes. No functional annotation was directly available for the rodent lineage, so we classified rodent genes using their orthology with P. falciparum (see methods). Practically, because the amino-acid divergence was, on average, higher in the hominid lineage compared to the rodent lineage (see above section: Average rates of evolution), we sought categories that showed a significantly lower or higher difference than expected on average. To do so, we used the following Linear Model: ω ~ L + Ca + L*Ca + constant, where ω corresponds to the ratio computed for each gene, L to the lineage (Hominid or Rodent) and Ca to the category to which the gene belongs (the category of interest or the rest of the genome). A difference higher than the average ω for certain categories between the hominid and the rodent lineages should translate into a significant interaction (L*Ca). We considered only those categories that contained at least 5 genes in both hominid and rodent lineages. Overall, the ω ratio was correlated among categories between hominid and rodent lineages (r2 = 0.34; p-value = 0.0019). We found no category of genes showing an accelerated evolution in one lineage compared to the other, relative to the rest of the genome (Fig. 2 and Table 2). Such a result can be interpreted in several ways. First, categories that could have been affected by such acceleration were not included in the analysis because of a lack of data. This could be the case, for instance, for categories of genes like those involved in host-parasite interactions like VAR genes which are not found in rodent parasite genomes. In our dataset, we had to exclude this category because of an insufficient number of genes belonging to it. Second, an evolutionary acceleration in the amino-acid substitution rate may not affect entire categories of genes but only some of them expressed, for instance, at only some particular stages of the parasite, thus rendering it more difficult to detect them.Figure 2Evolutionary rates (dN/dS) and Gene Ontology Processes in hominid (blue bars) and rodent (red bars) Plasmodium lineages. The numbers between parentheses are the number of genes belonging to each group. The first number corresponds to the hominid lineage; the second corresponds to the rodent one.Table 2Go categories and relative divergence rates (dN/dS) in hominid and murid lineagesGo categories within \"biological process\"dN/dS (ω) hominiddN/dS (ω) muridp-valueGO: Organelle organization and biogenesis0.0937970.1366220.34GO: Carbohydrate metabolism0.0942820.0615240.17GO: Energy pathways0.0959990.064090.25GO: Protein transport0.10006210.0569150.10GO: Cell organization and biogenesis0.1025210.111640.75GO: Cytoskeleton organization and biogenesis0.1065370.1230270.68GO: Protein biosynthesis0.108410.0764080.092GO: Transcription0.10930330.0853710.49GO: Cell growth and/or maintenance0.1125580.0781950.02GO: Cytoplasm organization and biogenesis0.1159380.1243350.80GO: Catabolism0.1176590.1077710.70GO: Transport0.1181980.0760050.059GO: Amino acid and derivative metabolism0.1184660.1053780.64GO: Protein metabolism0.1189170.090770.06GO: Biosynthesis0.1209250.090660.054GO: Physiological process0.123050.088810.0001GO: Cell proliferation0.1255430.0556010.03GO: Metabolism0.126230.08970.0001GO: Nucleobase et al*.0.1284490.079810.01GO: Cell cycle0.1371360.0551770.01GO: Response to stress0.1409270.0643590.03GO: DNA metabolism0.1544150.0629760.0028GO: Protein modification0.1703290.1166750.21GO: Lipid metabolism0.1718530.1246680.17GO: Biological_process unknown0.23530.170110.001GO: Cell communication0.24391640.1594630.17*Nucleobase, nucleoside, nucleotide and nucleic acid metabolism.Only those categories of biological processes with at least 5 genes in both lineages are listed. The p-value of the test comparing the average dN/dS (ω) ratio between hominid and murid lineages is given for each category (p-value). None of the category showed an accelerated evolution in hominid or rodent, given the average genome difference between lineages.Variation in evolutionary rates, timing and frequency of expressionWe finally addressed how variation in evolutionary rates could evolve relative to the timing and the breadth of expression of the genes. Do genes expressed at a particular stage show lineage specific evolution?As shown in Fig. 3A, hominid and rodent genes show similar patterns of evolution relative to their timing of expression. While no significant difference was observed between categories of genes for hominids (p-value = 0.3), a difference was found between categories for rodents: (p-value = 0.013). For both lineages, genes that are expressed at the merozoite stage look the most constrained while those expressed at the gametocyte stage appear the least constrained. We did not find any group of genes showing an accelerated evolution in one lineage compared to the other relative to the rest of the genome.Figure 3A, B. Evolutionary rates (dN/dS) and timing of expression. A. for all genes expressed at one stage (but that may also be expressed at another stage). B. for the genes that are only expressed at one particular stage. Blue squares: hominid lineage; Red squares: rodent lineage.The classification we used for Fig. 3A was nevertheless very broad. It included genes that were expressed at one particular stage but those same genes could also be expressed at other stages. Because this could preclude the detection of stage-specific evolution, we then re-analyzed our data keeping only the genes that were expressed at one stage. Doing so, we observed a significant difference in the rate of evolution among the different categories in the hominid lineage (p-value = 0.017) but no difference was observed in the rodent lineage (p-value = 0.59, Fig. 3B). We observed an overall difference between hominid and rodent parasites primarily due to the loci expressed at the sporozoite stage that showed an accelerated evolution in the hominid lineage compared to the rodent one (p-value = 0.018). This result suggests that genes only expressed at the sporozoite stage might be key genes in the infection process of the mammal host by malaria parasites and experienced higher adaptive evolution in the hominid lineage than in the rodent one.We then analyzed the relationship between the breadth of expression and evolutionary rates. Genes expressed at only one parasitic stage (see methods for details) were characterized as unique and those expressed at all stages were characterized as ubiquitous. For both the hominid and rodent lineages, we observed a significant relationship between the breadth of expression and the rate of non-synonymous substitutions. On average, stage specific proteins (expressed at only one stage) evolve at a higher rate relative to ubiquitous ones (expressed at all stages) (Fig. 4 and Table 3). In contrast, synonymous variation shows a very different trend regarding the breadth of expression. The rate of synonymous substitutions increases with the breadth of expression: proteins expressed in a larger number of stages evolve at higher rates at synonymous sites. The obvious corollary of these observations is a negative relationship between the breadth of expression and dN/dS as shown in Figure 4. Note that we obtained similar relationships for rodents using data on the expression of P. berghei [5] except that we did not find any significant relationship between dS and the breadth of expression (data not shown).Figure 4Substitution rates (dN/dS, dN, dS) and breadth of expression in hominid and rodent lineages.Table 3Relationship between gene expression, GC content and substitutions rates in both hominid's and rodent's Plasmodium parasitesHominid lineageRodent lineageExpression-dN/dSRho = -0.26; p = 0Rho = -0.26; p = 0Expression-dNRho = -0.13; p = 0.008Rho = -0.15; p = 0Expression-dSRho = 0.18; p = 0Rho = 0.113; p = 0.00001Expression-GC1Rho = 0.28; p = 0Rho = 0.31; p = 0Expression-GC2Rho = 0.37; p = 0Rho = 0.35; p = 0GC1-dN/dSRho = -0.37; p = 0Rho = -0.38; p = 0GC2-dN/dSRho = -0.43; p = 0Rho = -0.38; p = 0A spearman rank test was used to analyze the correlation between variables (Rho: spearman correlation coefficient; p: p-value).Our results are congruent with previous studies reporting relationships between breadth of expression and the rate of gene evolution in other organisms [14,15]. In both hominid and rodent lineages, highly expressed genes are generally more constrained than less expressed genes [16]. This observation is often attributed to the fact that proteins that are expressed in more diverse cellular environments are subjected to stronger functional constraints [14,15,17]. These results are consistent with the observation in both hominid and rodent parasites of a positive relationship between GC content at position 1 and 2 of codons (GC1-2) and the level of expression as well as a negative relationship between GC1-2 and dN/dS (Table 3). Highly and universally expressed genes are more GC-rich than lowly expressed genes which might thus reflect a codon bias, in particular for GC-rich codons. In other words, amino acids encoded by GC-rich residues are preferred and conserved in protein coding genes of the genus Plasmodium.The positive relationship observed between the rate of synonymous substitutions and expression can potentially be explained by translational selection acting on synonymous codon sites of highly expressed genes [16]. Alternatively, an increase in transcription may simply increase the level of spontaneous mutations as demonstrated in certain bacteria [18].ConclusionOur knowledge about the evolution of parasites responsible for malaria is increasing rapidly thanks to the availability of several completely sequenced genomes from species belonging to different lineages. As shown in the present study, different questions can be directly answered by comparing genomes from multiple Plasmodium species. Our comparative analysis suggests that, while there are a few aspects that are distinct among lineages, patterns of molecular evolution in the hominid parasite lineage are generally consistent with those observed in the rodent parasite lineage. In the murid lineage the most rapidly evolving genes are those involved in host-parasite interactions and those that are the least expressed. However, the evolution of the hominid lineage appears to be less constrained likely reflecting their historical lower effective size and an evolution driven by slightly deleterious mutations.While we tried to be as exhaustive as possible in our comparison of the evolution of the genome of both species, this study is still imperfect and incomplete. A definitive study will require the use of a high-quality complete sequence for P. reichenowi as well as more population data on the rodent lineages. Analyses of polymorphisms in natural populations of both hominid and rodent lineages (in the rodent lineage this information is specifically lacking) is critical to better understand the nature and intensity of selection acting on different categories of genes.Because for parasites with complex life-cycles (like Plasmodium sp.), the vector and the vertebrate host constitute very different environment, an interesting analysis would be to study the evolution of the genes exclusively expressed in the stages infecting the mammal host versus those only expressed in the stages infecting the mosquito vector. Such an analysis would however require the collection of more detailed expression profiles in both lineages.MethodsData sequences and alignmentsFor rodent malaria species, protein and nucleotide sequences for annotated genes for P. berghei, P. chabaudi and P. yoelii yoelii were obtained from The Plasmodium Genome Resource Database (Plasmodb [19]). Orthologous genes between the three species were obtained with BlastN using the criterion of best hits with scores of E < 1*10-15 and at least 70% similarity in length. Only the groups of genes for which only one gene of each species corresponded to these criteria were conserved. All groups of coding sequences were aligned using Clustal W version 1.82 [20] (default parameters) using amino acid sequences followed by back-translation into nucleotides sequences using the original sequence provided by Plasmodb.For hominid malaria species, protein and nucleotides sequences for annotated genes were obtained for P. falciparum only. For P reichenowi, Plasmodb provided only nucleotide contigs (release 09 July 2004) of a partial genome shotgun of approximately onefold coverage. The assembled contiguous sequences cover slightly less than one third of the P. reichenowi genome. Orthologous genes between the two species and their alignment were obtained following several steps. First, P. falciparum and P. reichenowi orthologues were obtained using BlastN with scores of E < 1*10-15 and at least 70% similarity in length. Only groups of genes where we obtained only one gene of each species were kept. These groups were then aligned using ClustalW (V. 1.82) using default parameters. All the alignments were then very carefully checked by eye and corrected when necessary. Introns were deleted.Synonymous and non-synonymous substitution rate analysesFor both the rodent and hominid malaria gene groups, maximum likelihood estimates of rates of non-synonymous (dN) and synonymous (dS) substitutions, averaged over all branches, were obtained using PAML version 3.14 [21]. We used a codon-based model of sequence evolution with dN and dS considered as free parameters and the average nucleotide frequencies estimated from the data at each codon position (F3 × 4 MG model). The transition/tranversion bias K was estimated for each group of orthologous sequence. Because estimates of dS > 1 are more prone to error [22], only genes with dS ≤ 1 were used for statistical calculations, yielding 843 and 3060 valid orthologues for hominid and rodents malaria groups respectively. For each gene, Likelihood Ratio Tests (LRT) were used to test whether the estimated dN/dS (ω) ratio differed significantly from 1 [23]. The tests were performed as bilateral tests of the hypothesis Ho: dN/dS =1 versus the alternative hypothesis H1: dN/dS ≠ 1 for each group of sequence. Twice the difference of the log likelihood estimated for each hypothesis was then compared to a χ2 distribution with one degree of freedom (df).Substitution rates and genomic featuresTo learn more about both synonymous and non-synonymous substitution patterns and their possible causes, we analyzed the effect of several genomic features such as the GC content, the level and timing of expression of genes and the function of proteins.The biological process of the annotated proteins of P. falciparum was determined using GO (Gene ontology) annotations. A biological process is a series of events accomplished by one or more ordered assemblies of molecular functions. Examples of broad biological process terms are cellular physiological process or signal transduction. Classification of proteins was made using the software GENERIC GENE ONTOLOGY (GO) TERM MAPPER [24]. Because no such classification was available for any of the rodent species, P. yoelii yoelii genes were classified as their orthologous P. falciparum genes defined in The TIGR Plasmodium yoelii yoelii Genome Annotation Database.For gene expression, we retrieved the mRNA abundance for genes of P. falciparum for different stages (rings, trophozoites, shizonts, merozoites, gametocytes and sporozoites) from [25]. For the rodent lineage, information on expression was available for P. berghei on a lower number of stages (asexual blood stages, gametocytes, ookinetes, oocysts and sporozoites) [5]. Such data precluded any possible rigorous comparison between the two lineages of parasites because of a lack of overlap between stages analyzed in the hominid and rodent lineages. We therefore decided to determine the timing of expression of rodent genes by using their orthology with P. falciparum, simply considering orthologous genes to be expressed at similar stages. We computed the breadth of expression for each gene in each lineage as the total number of different stages in which a gene is expressed.GC content was computed using CODONW [26]. GC content quantifies the proportion of GC inside the gene.Authors' contributionsFP: conceived of the study, designed it, performed the sequence alignment and all the statistical analyses, wrote the manuscript. KMcG and JK.: participated in the sequence alignment and the analyses. PA: conceived of the study, and participated in its design and coordination and helped to draft the manuscript. All authors read and approved the final manuscript.Supplementary MaterialAdditional file 1P. falciparum genes analyzed in the study and rates of evolution (dS, dN, dN/dS).Click here for fileAdditional file 2Orthologous genes of rodent Plasmodium species and their rates of evolution (dS, dN, dN/dS).Click here for file\n\nREFERENCES:\n1. HolmesECAdaptation and immunityPLoS Biol200422004/09/169E30710.1371/journal.pbio.002030715367941\n2. MikkelsenTSHillierLWEichlerEEZodyMCJaffeDBYangSPEnardWHellmannILindblad-TohKAltheideTKArchidiaconoNBorkPButlerJChangJLChengZChinwallaATdeJongPDelehauntyKDFronickCCFultonLLGiladYGlusmanGGnerreSGravesTAHayakawaTHaydenKEHuangXQJiHKKentWJKingMCKulbokasEJLeeMKLiuGLopez-OtinCMakovaKDManOMardisERMauceliEMinerTLNashWENelsonJOPaaboSPattersonNJPohlCSPollardKSPruferKPuenteXSReichDRocchiMRosenbloomKRuvoloMRichterDJSchaffnerSFSmitAFASmithSMSuyamaMTaylorJTorrentsDTuzunEVarkiAVelascoGVenturaMWallisJWWendlMCWilsonRKLanderESWaterstonRHConsortiumCSAInitial sequence of the chimpanzee genome and comparison with the human genomeNature20054377055698710.1038/nature0407216136131\n3. GardnerMJHallNFungEWhiteOBerrimanMHymanRWCarltonJMPainANelsonKEBowmanSPaulsenITJamesKEisenJARutherfordKSalzbergSLCraigAKyesSChanMSNeneVShallomSJSuhBPetersonJAngiuoliSPerteaMAllenJSelengutJHaftDMatherMWVaidyaABMartinDMAFairlambAHFraunholzMJRoosDSRalphSAMcFaddenGICummingsLMSubramanianGMMungallCVenterJCCarucciDJHoffmanSLNewboldCDavisRWFraserCMBarrellBGenome sequence of the human malaria parasite Plasmodium falciparumNature2002419690649851110.1038/nature0109712368864\n4. JeffaresDCPainABerryACoxAVStalkerJIngleCEThomasAQuailMASiebenthallKUhlemannACKyesSKrishnaSNewboldCDermitzakisETBerrimanMGenome variation and evolution of the malaria parasite Plasmodium falciparumNature Genetics200739112012510.1038/ng193117159978\n5. HallNKarrasMRaineJDCarltonJMKooijTWABerrimanMFlorensLJanssenCSPainAChristophidesGKJamesKRutherfordKHarrisBHarrisDChurcherCQuailMAOrmondDDoggettJTruemanHEMendozaJBidwellSLRajandreamMACarucciDJYatesJRKafatosFCJanseCJBarrellBTurnerCMRWatersAPSindenREA comprehensive survey of the Plasmodium life cycle by genomic, transcriptomic, and proteomic analysesScience20053075706828610.1126/science.110371715637271\n6. CarltonJMAngiuoliSVSuhBBKooijTWPerteaMSilvaJCErmolaevaMDAllenJESelengutJDKooHLPetersonJDPopMKosackDSShumwayMFBidwellSLShallomSJvan AkenSERiedmullerSBFeldblyumTVChoJKQuackenbushJSedegahMShoaibiACummingsLMFlorensLYatesJRRaineJDSindenREHarrisMACunninghamDAPreiserPRBergmanLWVaidyaABVan LinLHJanseCJWatersAPSmithHOWhiteORSalzbergSLVenterJCFraserCMHoffmanSLGardnerMJCarucciDJGenome sequence and comparative analysis of the model rodent malaria parasite Plasmodium yoelii yoeliiNature2002419690651251910.1038/nature0109912368865\n7. NikolaevSIMontoya-BurgosJIPopadinKParandLMarguliesEHAntonarakisSEProgramNLife-history traits drive the evolutionary rates of mammalian coding and noncoding genomic elementsProceedings of the National Academy of Sciences of the United States of America200710451204432044810.1073/pnas.070565810418077382\n8. OhtaTSynonymous and Nonsynonymous Substitutions in Mammalian Genes and the Nearly Neutral TheoryJournal of Molecular Evolution1995401566310.1007/BF001665957714912\n9. KeightleyPDLercherMJEyre-WalkerAEvidence for widespread degradation of gene control regions in hominid genomesPlos Biology20053228228810.1371/journal.pbio.0030042\n10. McdonaldJHKreitmanMAdaptive Protein Evolution at the Adh Locus in DrosophilaNature1991351632865265410.1038/351652a01904993\n11. MuJAwadallaPDuanJMcGeeKMKeeblerJSeydelKMcVeanGASuXZGenome-wide variation and identification of vaccine targets in the Plasmodium falciparum genomeNat Genet2007392006/12/13112613010.1038/ng192417159981\n12. HartlDLVolkmanSKNielsenKMBarryAEDayKPWirthDFWinzelerEAThe paradoxical population genetics of Plasmodium falciparumTrends Parasitol2002182002/05/31626627210.1016/S1471-4922(02)02268-712036741\n13. AkashiHWithin- and between-species DNA sequence variation and the 'footprint' of natural selectionGene19992381395110.1016/S0378-1119(99)00294-210570982\n14. DrummondDABloomJDAdamiCWilkeCOArnoldFHWhy highly expressed proteins evolve slowlyProceedings of the National Academy of Sciences of the United States of America200510240143381434310.1073/pnas.050407010216176987\n15. RochaEPCDanchinAAn analysis of determinants of amino acids substitution rates in bacterial proteinsMolecular Biology and Evolution200421110811610.1093/molbev/msh00414595100\n16. ChandaIPanADuttaCProteome composition in Plasmodium falciparum: Higher usage of GC-rich nonsynonymous codons in highly expressed genesJournal of Molecular Evolution200561451352310.1007/s00239-005-0023-516044241\n17. HughesALFriedmanRAmino acid sequence constraint and gene expression pattern across the life history in the malaria parasite Plasmodium falciparumMolecular and Biochemical Parasitology2005142217017610.1016/j.molbiopara.2005.02.01415978954\n18. HudsonREBergthorssonUOchmanHTranscription increases multiple spontaneous point mutations in Salmonella entericaNucleic Acids Research200331154517452210.1093/nar/gkg65112888512\n19. The Plasmodium Genome Ressource Databasehttp://www.plasmodb.org/\n20. HigginsDGSharpPMClustal - a Package for Performing Multiple Sequence Alignment on a MicrocomputerGene198873123724410.1016/0378-1119(88)90330-73243435\n21. YangZHPAML: a program package for phylogenetic analysis by maximum likelihoodComputer Applications in the Biosciences19971355555569367129\n22. Castillo-DavisCIBedfordTBCHartlDLAccelerated rates of intron gain/loss and protein evolution in duplicate genes in human and mouse malaria parasitesMolecular Biology and Evolution20042171422142710.1093/molbev/msh14315084679\n23. YangZHLikelihood ratio tests for detecting positive selection and application to primate lysozyme evolutionMolecular Biology and Evolution19981555685739580986\n24. Generic Gene Ontology (GO) Term Mapperhttp://go.princeton.edu/cgi-bin/GOTermMapper\n25. Le RochKGJohnsonJRFlorensLZhouYYSantrosyanAGraingerMYanSFWilliamsonKCHolderAACarucciDJYatesJRWinzelerEAGlobal analysis of transcript and protein levels across the Plasmodium falciparum life cycleGenome Research200414112308231810.1101/gr.252390415520293\n26. CodonWhttp://codonw.sourceforge.net/\n27. PerkinsSLSchallJJA molecular phylogeny of malarial parasites recovered from cytochrome b gene sequencesJournal of Parasitology200288597297810.1645/0022-3395(2002)088[0972:AMPOMP]2.0.CO;212435139"
+}

batch_5/PMC2529321.json

@@ -0,0 +1,4 @@
+{
+    "id": "PMC2529321",
+    "text": "This is an academic paper. This paper has corpus identifier PMC2529321\nAUTHORS: Olivier Bastien, Eric Maréchal\n\nABSTRACT:\nBackgroundConfidence in pairwise alignments of biological sequences, obtained by various methods such as Blast or Smith-Waterman, is critical for automatic analyses of genomic data. Two statistical models have been proposed. In the asymptotic limit of long sequences, the Karlin-Altschul model is based on the computation of a P-value, assuming that the number of high scoring matching regions above a threshold is Poisson distributed. Alternatively, the Lipman-Pearson model is based on the computation of a Z-value from a random score distribution obtained by a Monte-Carlo simulation. Z-values allow the deduction of an upper bound of the P-value (1/Z-value2) following the TULIP theorem. Simulations of Z-value distribution is known to fit with a Gumbel law. This remarkable property was not demonstrated and had no obvious biological support.ResultsWe built a model of evolution of sequences based on aging, as meant in Reliability Theory, using the fact that the amount of information shared between an initial sequence and the sequences in its lineage (i.e., mutual information in Information Theory) is a decreasing function of time. This quantity is simply measured by a sequence alignment score. In systems aging, the failure rate is related to the systems longevity. The system can be a machine with structured components, or a living entity or population. \"Reliability\" refers to the ability to operate properly according to a standard. Here, the \"reliability\" of a sequence refers to the ability to conserve a sufficient functional level at the folded and maturated protein level (positive selection pressure). Homologous sequences were considered as systems 1) having a high redundancy of information reflected by the magnitude of their alignment scores, 2) which components are the amino acids that can independently be damaged by random DNA mutations. From these assumptions, we deduced that information shared at each amino acid position evolved with a constant rate, corresponding to the information hazard rate, and that pairwise sequence alignment scores should follow a Gumbel distribution, which parameters could find some theoretical rationale. In particular, one parameter corresponds to the information hazard rate.ConclusionExtreme value distribution of alignment scores, assessed from high scoring segments pairs following the Karlin-Altschul model, can also be deduced from the Reliability Theory applied to molecular sequences. It reflects the redundancy of information between homologous sequences, under functional conservative pressure. This model also provides a link between concepts of biological sequence analysis and of systems biology.\n\nBODY:\nBackgroundAutomatic analysis of biological sequences is crucial for the treatment of massive genomic outputs. Our understanding of more than 90 % of protein sequences stored in public databases, deduced from automatic translation of gene sequences, will not result from direct experimentation, but from our ability to predict informative features using in silico workflows [1,2]. An underlying postulate is that the molecular sequences determined in biological individuals or species, which have evolved from a common ancestor sequence and are therefore homologous, have conserved enough of the original features to be similar. Popular sequence alignment methods, such as Blast [3] or Smith-Waterman [4] algorithms are used as a starting point for homology searches. All these methods computes a score s(a, b) between two sequences a and b. They use scoring matrices to maximize the summed scores of compared residues and find optimal local alignments, computed with a dynamic programming procedure [3,4]. Scoring matrices have been found to be similarity matrices as well [5]. Many similarity matrices are available [6-8] and evaluation studies led to the conclusion that all can be considered as log-odds ratio matrices, including the BLOSUM family [7] and the PAM family [6]. Log-odds ratio matrices are defined by s(i,j)=log⁡(ω(i,j)ν(i)ν(j)) where ω(i, j) is the joint probability of the amino acid pair (i, j), and ν(i) and ν(j) the probabilities of the amino acids i and j in the two aligned sequences.Because re-examination of alignments obtained after massive comparisons is not manageable, confidence in alignment score probabilities is critical for automatic sequence comparisons, clustering of orthologs and paralogs, homology-based annotations or phylogeny reconstructions based on pairwise alignments [2]. Assessing whether a computed alignment is evolutionarily relevant or whether it could have arisen simply by chance is therefore a question that has been extensively studied (for review: [9]). Two major methods have been proposed.The first and oldest method, proposed by Lipman and Pearson [10] and described extensively by Comet et al. [11] and others [12-14], uses Monte Carlo simulations to investigate the significance of a score, s calculated from the alignment of two real sequences a and b. This method consists in computing η alignments of a with sequences obtained after shuffling b [15]. The random sequence corresponding to the shuffled sequence b is termed B. The η alignments allow an estimate of an empirical mean score (μ^) and standard deviation (σ^) from the distribution of the random variable S(a, B). A Z-value is then defined as:(1)Z(a,b∗)=s−μ^σ^where * indicates the sequence that was submitted to randomization.In practice, the computation of Z(a, b*) is known to be convergent and depends on the accuracy of the estimation of μ and σ, and therefore on η, ranging usually from 100 to 1000 [11,16]. Bacro and Comet [12] showed that the asymptotic law of the Z-value (when η → ∞) was independent of the length and composition of sequences. Bastien et al. [13] further demonstrated that regardless of the distribution of the random variable S(a, B), the relation(2)P(S(a,B))≤1Z(a,b∗)2is true. This relation, known as the TULIP theorem, shows that the Z-value computed for pairwise sequence alignments 1) provides an upper bound of alignment score probability [13], 2) can be used to reconstruct molecular phylogenies [14] and 3) is an accurate clustering criterion to reduce the diversity of protein sequence databases [17]. Here we call T-value the upper bound deduced from the TULIP theorem, i.e. 1/Z(a, b*)2.Simulations of Z-value distribution [11,18] shows that it fits a Gumbel distribution, suggesting that the distribution of alignment scores might follow a Gumbel distribution as well [19].The second and most popular method proposed by Karlin and Altschul [20] is an estimate of the probability of an observed local ungapped alignment score according to an extreme value distribution (or EVD; for review: [19]), i.e. a Gumbel distribution, in the asymptotic limit of long sequences. The remarkable Karlin-Altschul formula is the consequence of interpreting the number of highest scoring matching regions above a threshold by a Poisson distribution. Briefly, considering A and B two random sequences, m and n their lengths, given the distribution of individual residues (i.e. amino acids), and given a scoring matrix, the number of distinct local alignments with score values of at least s is approximately Poisson distributed with mean(3)E(s) ≈ K.m.n.exp(-λ.s)where λ and K can be calculated from the scoring matrix and average sequence compositions based on the Poisson distribution hypothesis. E(s) is known as the E-value. As a consequence, if s is the score obtained after aligning two real sequences a and b (with m and n their respective lengths), the probability of finding an ungapped segment pair with a score lower than or equal to s, follows a Gumbel distribution:(4)P(S(A, B) ≤ s) ≈ exp(-K.m.n.exp(-λ.s))where S(A, B) is the random variable corresponding to the score of two random sequences. The P-value, defined as the probability of finding an ungapped segment pair with a score higher than s, is simply given by 1-P(S(A, B) ≤ s). Using the Taylor Expansion of equation (4), the P-value is approximated by the E-value when E(s) < 0.01. The validity of the Karlin-Altschul model depends on restrictive conditions: firstly, the residue distributions in the compared sequences should not be \"too dissimilar\" and secondly, the sequence lengths (m an n) should \"grow at roughly equal rates\" [20]. The length dependency of alignment scores has been discussed [20,21]. In particular, it has been demonstrated that the growth of the best matching score of gapped alignments was linear when gap penalties were small, becoming logarithmic when increasing sequence length and for larger gap penalties [21]. Although the Karlin-Altschul formula given by equation (4) is not valid for gapped alignments and although no asymptotic score distribution has been analytically established for local alignments allowing gaps, simulations [11,18,22,23] showed that, for both local and global alignments, the Gumbel law was well-suited to the distribution of scores after pragmatic estimation of the λ and K parameters.Noticeably, this model relies on the fact that λ is the unique positive solution to the equation ∑i,jνa(i)νb(j)exp⁡{λ.s(i,j)}=0, for the 20 × 20 combinations of i and j amino acids, with νa(i) and νb(j) the probabilities of amino acids i and j in sequences a and b respectively and s(i, j) the score in the substitution matrix. From a theoretic point of view, and regardless of the practical performance of the Karlin and Altschul [20] model, the fact that an observed distribution (the distribution of scores of real compared sequences) depends on a presupposed and pre-calculated parameter is not satisfactory. It would be more satisfactory if λ arose as a property of a biological process and/or features. We addressed therefore the question of the missing biological rationale to parameters, particularly λ and K, that proved to be valid in pragmatic terms.In this paper, we deduced biological rationale for the Gumbel-like distribution of sequence alignment scores and Z-values, based on a limited number of assumptions on sequences evolution. An ancestral sequence is the origin of a lineage of homologous sequences that are subjected to evolutionary mechanisms. We considered homologous sequences as entities sharing structural features, in particular some conserved or functionally similar amino acids detected by alignment methods. Features that are preserved in two homologous sequences are estimated by a shared amount of information (SAI). In this model, the amount of information shared between an initial sequence and the sequences in its lineage (i.e., mutual information in Information Theory) is a decreasing function of time: over time, some substitutions of amino acids by others having redundant properties (SAI at the residue level) may be permitted without functional break down, but leads to a decrease of the SAI between the sequences. Classically, molecular evolution is formalized with Markovian models for residue substitutions, allowing the backward reconstruction of sequences' evolution with the assumption that the proteins have been selected for a functional conservation. Here, proteins were considered as systems, with a high level of structural redundancy, which components may \"age\" over evolution, and \"die\" in case of loss of the initial amount of information required to operate accurately for a given biological function. Assumptions are therefore generalist regarding the process of sequence evolution, should it be strictly Markovian or not, but they give a formalism to the reliability of the sequences reflecting the functional status of the folded and maturated protein, and being a criterion on which positive selection pressure might act. We introduced therefore principles of the reliability theory of aging and longevity [24], that apply to a wide range of other systems, from artificial machines to biological population or organisms, applied here to molecular sequences. Based on the deduced model, we could provide biological basis for the Z-value Gumbel distribution, and significance for the corresponding Gumbel parameters (termed K' and λ'). Moreover, the assumption that the score between two sequences a and b should be the highest possible score between a and b is not necessary to observe an extreme values distribution for sequence alignment scores.Major points of the following demonstration are:i. The evolution of biological sequence is formalized by the evolution of the SAI between an initial sequence and sequences of its lineage. It is known that for two sequences a and b, this is measured by the mutual information I(a; b), based on Information Theory and is exactly the score s(a, b) computed with standard methods in sequences comparisons [14].ii. If a sequence evolves, the probability that it stays near its \"last\" position in the sequence space is low and the longest the sequence, the lowest this probability (consequence of the concentration in a high dimensionality space [25]). The amount of information shared between an initial sequence and the sequences in its lineage decreases with time: as a consequence, one can indifferently use I(a; b) as a measure of the divergence time.Results and discussionAssumptions for a model of sequences' evolutionA basic process in the evolution of proteins is the change of amino acids over time. In the simplest view, these changes lead to amino acid substitutions, insertions or deletions. Dayhoff et al. [6] introduced the description of this process as a continuous-time Markov chain with a matrix of transition probabilities for the substitutions of any amino acid into another through time. This model allows forward and backward expressions of sequence evolution, under time homogeneity assumption, and is therefore an important tool for phylogeny reconstructions. Given a transition matrix and an equilibrium distribution of amino acids, then a matrix of amino acid substitution scores, in the sense of sequences' comparison, can be deduced [26,27].In the generalist model described here, assumptions regarding the process of sequence evolution were not formalized, should this process be strictly Markovian or not. Given two sequences, one can, one the one hand, compute a score using dynamic algorithms [3,4] and deduce the distribution of random scores from transition matrices under the hypothesis that the two sequences have evolved according to a continuous-time Markov chains process. On the other hand, Henikoff et al. [7] demonstrated the possibility to calculate efficient log-odd matrices without the need of this assumption. Altschul [28] and Bastien et al. [14] demonstrated that log-odd matrices could be reformulated in the Information Theory framework. In particular, a score between two amino acids i and j can be interpreted as the mutual information between these two residues. At the 3D folded protein level, a molecular function emerges from the information encrypted in the amino acid sequence, and positive selection pressure acts therefore at the sequence level, maintaining a sufficient portion of the initial information, and consequently the functional status of the folded and maturated protein. We therefore focused on the evolution of the information shared between an initial sequence and the sequences of its lineage through time.Reliability theory and biological sequences evolutionThe Reliability Theory is a general theory about systems aging, in which the failure rate (the rate by which systems deteriorate) is related to the systems longevity (For review, [24]). The system can be a machine with structured components, or a living entity or population. \"Reliability\" of a system (or of one of its components) refers to its ability to operate properly according to a standard [29]. The relation between the age of a system and its failure rate shows that aging is a direct consequence of redundancies within the system. For instance, when applied to a biological system in which redundant vital structures ensure a function, damage of a component that is compensated by another redundant intact one, does not lead to a complete impairment of the system. Defects do accumulate, resulting in redundancy exhaustion and giving rise to the phenomenon of aging. As the system (or one of its components) degenerates into a system with no redundancy, new defects can eventually lead to death. Reliability of the system (or component) is described by the \"reliability function\" R(x), also named \"survival function\", which is the probability that the system (or component) will carry out its mission through time x [30], expressed as the probability that the failure time X is beyond time x:(5)R(x) = P(X > x) = 1 - P(X ≤ x) = 1 - F(x)where F(x) = P(X ≤ x) is a cumulative distribution function [24] reflecting the resistance of the system to failures (at time x, distribution of the probability that the system could have failed previously). R(x) evaluates therefore the probability that the systems becomes completely defective after a time x (x can be a direct measure of time t or an increasing function of time).The \"hazard rate\"h(x), also called \"failure rate\", is defined as the relative rate for reliability function decline:(6)h(x)=−dR(x)R(x).dx=−d(log⁡R(x))dxHazard rate is equivalent to mortality force in demography [31,32]. When h(x) is a constant h, the system does not deteriorate more often with age, and is therefore a non-aging system. In this case, a simple integration of equation (6) leads to(7)R(x) = R(0)exp(-h.x)which is the exponential distribution that characterizes non-aging systems. Interestingly, a system with redundant non-aging components can be an aging system. That is to say the hazard rate of a system of components depends can depends of time whereas the hazard rate of components do notAs discussed by Gavrilov and Gravrilova [24], the \"reliability theory\" provided explanations for some fundamental problems regarding aging, longevity, death of organisms within populations. Organisms or populations are considered as systems in which categories of components (molecules, biological processes, cells, individuals, etc.) can be highly redundant, and be key elements for the system longevity.Here, we propose to consider the particular case of protein sequences as a system, in which redundancy is ensured:i. by the number of residue positions involved in the evolution process.ii. at the residue level by the existence of functionally redundant amino acids (e.g. after a DNA damage that leads to a genetic mutation, an aspartic acid may be substituted by a functionally redundant glutamic acid), i.e. the existence of a SAI for all amino acids pairs.In this model, evolutionary time is negatively correlated to the amount of information shared between an initial sequence and sequences in its lineage (SAI decreases with time, see below).The conservation rate: a mathematical tool to study the evolution of the information shared by biological sequencesTo measure the rate of conservation of a shared structure/function relationship at time x within a system of homologous proteins (i.e. the time of observation), we considered that the decay of information shared between an original sequence and sequences of its lineage was a function of time, and therefore a mean to measure time. Evolutionary time is therefore measured here in information units. We defined an information conservation rate Ψ as follows:DefinitionGiven the cumulative distribution function F(x) = P(X ≤ x) (Probability that the system shared less than x information units with a reference), supposed continuously differentiable, the conservation rate Ψ is given by:(8)ψ(x)=lim⁡dx→0P(x−dx<X≤x/X≤x)dxThe conservation rate is simply related to the hazard function, measuring a quantity that decreases over time (shared information) instead of a quantity that increases over time (age). Given f(x) = dF(x)/dx the density function of x, this conservation rate has the following properties.(9)ψ(x)=f(x)F(x)=f(x)P(X≤x)and as corollaries:(10)ψ(x)=d(log⁡F(x))dx(11)P(X≤x)=exp⁡(−∫x+∞ψ(u)du)(12)f(x)=ψ(x)exp⁡(−∫x+∞ψ(u)du)Derivation of the distribution of sequence alignment scores based on the distribution of mutual information between amino acidsDobzhansky [33] and Wu et al. [34] established that information harbored by a protein 1) emerged from the three-dimensional self organization of its residues (i.e. the sequence of amino acids) and had to do with information harbored by amino acids, and 2) was submitted through time to evolutionary pressure (achievement of a minimal functional level fitting environmental and species survival conditions). Using previous empirical results [6,7,35], Bastien et al. [14] have shown that the alignment score of two homologous sequences a and b was proportional to the estimate of the SAI due to their common origin and parallel evolution under similar conservative pressure, i.e. the mutual information I(a; b) between the two events a and b in the sense of Hartley [36,37]:(13)s(a, b) = ξ.I (a; b)with ξ a constant defining the unity (ξ = 1, in bits) and s(a, b) the sum of the elementary scores for all aligned positions (including gap opening and gap extension penalties). Mutual information between two events a and b (differing from the mutual information defined between random variables, see [14,38]) measures the information gained by the knowledge of event a on the occurrence of event b. The mutual information being additive, I(a; b) is the sum of the mutual information of aligned residues, reflecting the magnitude of the redundancy between the sequences at the amino acid level. Mutual information between residues is therefore simply deduced from the 20 × 20 amino acid substitution matrix [6-8,35] used to compute the alignment.Inside a given sequence, mutual information was also shown to reflect the dependency of close or remote amino acids, a phenomenon known as the residue co-evolution, due to their co-contribution to the sequence function [39,40].Considering a protein as a system, which components are amino acids, we examined the mutual information between the original components and their descendants, and how amino acid mutation affected the evolution of mutual information between proteins. We simply hypothesized that an amino acid may mutate over time following random DNA mutations and look at the behavior of the entire system, namely the protein which can be measured here by the mutual information between the initial residues and the new ones, i.e. the corresponding substitution scores in a 20 × 20 substitution matrix. The substitution matrix is considered as an estimate of the mutual information between residues because it was computed from real sequences' data [6-8,35].Over time, an amino acid i is either conserved or substituted. The similarity of i in an initial sequence compared with residues at the same position in protein descendants is therefore either that of identity (the diagonal term in the scoring matrix) or a lower value(no score is higher than that of identity). In average, the magnitude of the similarity of i compared with its descendants, related to mutual information following equation (13), is therefore a decreasing function of elapsed time. On a functional point of view, the probability that i was mutated into a residue with a score Si lower than a threshold si defined to allow the component to operate like i, can be deduced from the distribution of substitution scores. For most amino acids (F, P, W, Y, V, E, G, H, I, L, K, R, N, D and C), the distribution of scores deduced from BLOSUM 62 fits an exponential distribution (see the case of valine in Figure 1A. For five amino acids (M, S, T, A and Q), the distribution of scores does not fit an exponential distribution (see the case of Threonine in Figure 1B). Taking the average situation, the distribution of scores deduced from the BLOSUM 62 matrix is exponential-like (Figure 1C) supporting a general model for amino acids mutual information distribution: The probability Pr that a residue i is mutated into a residue with mutual information below si is:Figure 1Aging properties of amino acids. Protein sequences are considered as systems, which components are amino acids. Over time, either amino acids are conserved (similarity of a residue with its descendant is that of identity, diagonal term of a substitution matrix) or modified due to random DNA mutations. Similarity decreases therefore with time, since no similarity is higher than that of identity. When the similarity falls below a threshold that is necessary for the residue to operate according to a standard (functional conservation), the component is damaged. (A) Score distribution corresponding to valine substitution. In this case, the score distribution is exponential, suggesting that valine (V) is a non-aging component. Based on BLOSUM62, residues of this type are V, F, P, W, Y, E, G, H, I, L, K, R, N, D and C (B) Score distribution corresponding to threonine substitution. The score distribution shows a peak, indicating a probable accelerated process of aging (functional damage) when the residue is substituted by random mutation in some other amino acids. Based on BLOSUM62, residues of this type are T, S, M, A and Q. (C) Score distribution in the BLOSUM62 similarity matrix. The complete distribution in the BLOSUM62 matrix is exponential (0.287.exp(-0.287.(s+4))), supporting a general model of amino acids as nonaging components. The exponential law for positive scores is characterized by the same parameter (λ' = 0.287). The original residue is termed i; its descent is termed j.(14)Pr(Si ≤ si) = 1-exp(-λi.si)where λi is the constant information hazard rate, or failure rate, for reliability function decline of the amino acid mutual information.Given a sequence a, what is the probability that any of its m residues (termed i) had previously mutated into the n residues (termed j) of a sequence b and leads to the observed mutual information between sequence a and sequence b? We can consider m ≠ n due to insertion or deletion events. If m and n are large, we can state the following asymptotic approximations: S ≈ m ⟨Si⟩, with 〈Si〉=lim⁡m→∞Sm and s ≈ m ⟨si⟩, with 〈si〉=lim⁡m→+∞sm where s (respectively S) is the score between the sequence a (respectively A) and the sequence b (respectively B) (for discussion of these approximations, see [41]). In the asymptotic limit of long sequences, we can envisage different scenarios for the evolution of a into b:In a first step (Figure 2, step 1), the probability that one residue a1 is mutated into a residue b1 with mutual information below si is given:Figure 2Computing of the probability that the amount of information shared by two sequences, S, is lower than a threshold s. Given an initial sequence a, we can envisage different scenarios for its evolution into another sequence b. In a first step (Step 1), an elementary probability is computed by taking into account the evolution of just one residue (here a1 into b1). Considering one possible evolutionary scenario (Step 2), residues are considered as independent and the probability is the product of elementary probabilities for each positions aligned in this scenario, with approximations in the asymptotic limit of long sequences. The final probability (Step 3) is then estimated by taking into account all the possible evolutionary scenarios.(15)Pr(Si ≤ si) = Pr(Si ≤ s(a1,b1)) ≈ Pr(⟨Si⟩ ≤ ⟨si⟩)Considering one possible evolutionary scenario, i.e. one alignment (Figure 2, step 2), residues are considered as independent and the probability is the product of elementary probabilities for each positions aligned in this scenario. For the alignment of the m amino acids of sequence a, we obtain the following probability:(16)Pscenario1 (S ≤ s) = (Pr(⟨Si⟩ ≤ ⟨si⟩))mAlternative scenarios are also possible (Figure 2, step 3). The final probability is therefore computed taking into account all possible evolutionary paths (all possible alignments, Figure 2, step 3) and using K'<1 a correcting factor for edge effects, deletion and insertion points:(17)P(S ≤ s) = (Pr(⟨Si⟩ ≤ ⟨si⟩))K'mnConsidering the approximation of ⟨Si⟩ and ⟨si⟩ respectively by S/m and s/m, we deduce the final formula:(18)P(S ≤ s) = (Pr(S ≤ s))K'.m.nThe density function f(s) is therefore given by:(19)f(s)=dP(S≤s)ds=K′.m.n.fr(s).(Pr(S≤s))K′.m.n−1with fr(s)=dPrds(s) the density of the probability Pr(S ≤ s) that a residue is mutated into anotherwith mutual information below s. We can then deduce the homology longevity rate Ψ, defined earlier as a function of the pairwise alignment score:(20)ψ(s)=f(s)P(S≤s)=K′.m.n.fr(s).(Pr(S≤s))K′.m.n−1(Pr(S≤s))K′.m.nUsing the expression of Pi(Si ≤ si) given by Equation (14) implies that:(21)ψ(s)=K′.m.n.λ′.exp⁡(−λ′.s).(1−exp⁡(−λ′.s))K′.m.n−1(1−exp⁡(−λ′.s))K′.m.n=K′.m.n.λ′.exp⁡(−λ′.s)1−exp⁡(−λ′.s)Asymptotically, the information conservation rate is therefore given by(22)ψ(s) = K'.m.n.λ'.exp(-λ'.s)Using equation (12), we deduce that the distribution of alignment scores should respect the general form of the Karlin-Altschul formula:(23)P(S ≤ s) ≈ exp(K'.m.n.exp(-λ'.s))Applications and ConclusionWe built a model of evolution of the information shared between an initial molecular sequence and the sequences of its lineage (i.e. homologous sequences). Sequences were considered as systems, which components are the amino acids that can independently be damaged by random DNA mutations. Residues harbor a functional redundancy reflected by the amino acid substitution scores.From these assumptions, we deduced that the pairwise sequence alignment score should follow a Gumbel distribution (equation (22)). The λ' parameter is the information hazard rate for the reliability of amino acids' mutual information: it depends 1) on the distribution of the amino acids and 2) on the distribution of amino acid similarities deduced from a substitution matrix. The K' parameter has a more complex meaning, because it depends on likelihood of an alignment of two sequences, with edge effects, gaps, length difference and repartition of the information (the local score) in the alignment. It reflects therefore internal structural constraints on the evolution of sequences.The Gumbel parameters for score alignments can be estimated by two kinds of simulations. First is by adjusting EVD to the simulated distribution of scores [19,22]. In that case, it is simpler to express the Gumbel law as(24)P(S≤s)≈exp⁡(−exp⁡(−s−θβ))with β=1λ′ and θ=1λ′log⁡(λ′.K′.m.n). The estimate of Gumbel parameters is achieved by determining β and θ, allowing an easy estimate of the λ' and K' parameters of equation (23). Second estimation of the Gumbel parameters is by computing the Z-value corresponding to the simulation of score distribution. Using the fact that for a Gumbel distribution, μ = θ + γβ and σ2=π26β2, then the Z-value allows a computation of the β and θ constants.Simulations of Z-value distribution [11,18] showed that it fitted with a Gumbel law. Based on the Gumbel distribution of scores (equations (24) and (25)) and by an appropriate change of variable with equation (1), then the distribution of Z-values should respect the following equality:(25)P(Z≤z)=exp⁡(−.exp⁡(−zπ6−γ))with γ the Euler-Mascheroni constant (γ ≈ 0.5772). Equation (25) is the precise expression of the distribution of Z-values deduced by Pearson [18] from simulations. It is important to note that this expression of the Z-value distribution is independent of sequence lengths and amino acid distributions.This consideration has practical implications, since it allows a refined estimate of the P-value based on Z-value computation, and a real gain over available methods, particularly in some documented cases where the Karlin-Altschul formula failed to assess the significance of an alignment. Table 1 shows for instance the different statistical estimates for the alignment of two homologous TFIIA gamma sequences from Plasmodium falciparum and Arabidopsis thaliana. The compositional bias in the proteome of Plasmodium falciparum, the malarial parasite, is known to limit the use of Karlin-Altschul statistics for pairwise comparisons with unbiased proteins such as those of Arabidopsis thaliana [42]. The TFIIA gamma subunit sequence of Plasmodium could not be deduced from BLASTP-based homology searches [43]. The Blastp apparent search failure was due to the overestimate of the P-value following the Karlin-Altschul formula (0.008, using unfiltered BLASTP, see Table 1). Alignment score Z-value, computed with either Blastp (P. Ortet, unpublished algorithm) or Smith-Waterman was above 10. The upper bound for the P-value based on the TULIP theorem, given by the formula T-value = 1/Z-value2 [13], was therefore below 10-2. Eventually, the P-value deduced from the Z-value Gumbel distribution was below 10-6 (see Table 1) indicating that, for both the Blastp and Smith-Waterman methods, the homology could be statistically assessed, even in the limit case of unbiased vs biased sequence comparisons. We noticed that the asymmetric DirAtPf100 matrix specified for Plasmodium vs. Arabidopsis comparisons that we developed earlier [8] allowed an additional gain in estimating this missed homology.Table 1Alignment statistics of the homologous Transcription initiation factor IIA (TFIIA) gamma chain sequences from Plasmodium falciparum and Arabidopsis thaliana.Alignment methodBlastpSmith-WatermanSubstitution matrixBLOSUM62BLOSUM62DirAtPf100StatisticsP-value (Karlin-Altschul)0.008NANAZ-value (Pearson-Lipman)101112T-value (TULIP theorem)0.018.10-37.10-3P-value (this work)1.5.10-63.7.10-71.10-7TFIIA gamma sequences from Plasmodium (UniProtKB Q8I4S7_PLAF7) and Arabidopsis (UniProtKB T2AG_ARATH) were aligned with Blastp and Smith-Waterman methods. Statistics were computed following the Karlin-Altschul model (as implemented in the Blastp algorithm) or the Lipman-Pearson Z-value model. The upper bound for the P-value based on the TULIP theorem is given following the formula: T-value = 1/Z-value2. The P-value deduced from the Z-value Gumbel distribution was computed following the model presented here. Substitution matrices were either BLOSUM62, or the asymmetric DirAtPf100 matrix specified for Plasmodium vs. Arabidopsis comparisons. NA: not applicable.Besides a theoretical support for pragmatic observations, this report shows therefore that the alignment score Gumbel distribution is a particular and general evolutionary law for molecular sequences taken as dynamical systems. This model can be parameterized using the Karlin-Altschul or the Z-Value form. If Karlin-Altschul model parameters are well-estimated (using simulations for example), both forms are equivalent in practice as reported by Hulsen et al. [44]. This model shows that an extreme value distribution of alignment scores can arise not only by considering high scoring segments pairs. Indeed, derivation of a Gumbel distribution from maximum independent random variables is a well-known technique [19] and the Karlin-Altschul theorem was first demonstrated, based on this consideration [20]. We can now state that this distribution allows a different interpretation in the light of the Reliability Theory, reflecting the redundancy of information between sequences due to both the number of residues and the shared information between these residues. The model elaboration described here additionally provides a link between concepts of biological sequence analysis and the emerging field of systems biology, with a generalization of the aging concepts to all scales of the living world.MethodsMathematical demonstrations are detailed in the Results section. Histograms and curves were built using the R package software (Statistics Department of the University of Auckland).AbbreviationsSAI: shared amount of information; TULIP: theorem of the upper limit of a score probability; EVD: extreme values distributionAuthors' contributionsOB conceived the main theoretical model, designed and developed all demonstrations and drafted the manuscript. EM participated in the design of the study and helped to draft the manuscript. All authors read and approved the final manuscript.\n\nREFERENCES:\nNo References"
+}

batch_5/PMC2530875.json

@@ -0,0 +1,4 @@
+{
+    "id": "PMC2530875",
+    "text": "This is an academic paper. This paper has corpus identifier PMC2530875\nAUTHORS: Sophie Gaudriault, Sylvie Pages, Anne Lanois, Christine Laroui, Corinne Teyssier, Estelle Jumas-Bilak, Alain Givaudan\n\nABSTRACT:\nBackgroundThe phenotypic consequences of large genomic architecture modifications within a clonal bacterial population are rarely evaluated because of the difficulties associated with using molecular approaches in a mixed population. Bacterial variants frequently arise among Photorhabdus luminescens, a nematode-symbiotic and insect-pathogenic bacterium. We therefore studied genome plasticity within Photorhabdus variants.ResultsWe used a combination of macrorestriction and DNA microarray experiments to perform a comparative genomic study of different P. luminescens TT01 variants. Prolonged culturing of TT01 strain and a genomic variant, collected from the laboratory-maintained symbiotic nematode, generated bacterial lineages composed of primary and secondary phenotypic variants and colonial variants. The primary phenotypic variants exhibit several characteristics that are absent from the secondary forms. We identify substantial plasticity of the genome architecture of some variants, mediated mainly by deletions in the 'flexible' gene pool of the TT01 reference genome and also by genomic amplification. We show that the primary or secondary phenotypic variant status is independent from global genomic architecture and that the bacterial lineages are genomic lineages. We focused on two unusual genomic changes: a deletion at a new recombination hotspot composed of long approximate repeats; and a 275 kilobase single block duplication belonging to a new class of genomic duplications.ConclusionOur findings demonstrate that major genomic variations occur in Photorhabdus clonal populations. The phenotypic consequences of these genomic changes are cryptic. This study provides insight into the field of bacterial genome architecture and further elucidates the role played by clonal genomic variation in bacterial genome evolution.\n\nBODY:\nBackgroundComparative genomics, in the study of different bacterial genera, species, and strains, leads to the definition of two DNA pools in bacterial genomes: a set of genes shared by all genomes in a taxa, namely the 'core' genome; and a set of genes containing mobile and accessory genetic elements, termed the 'flexible' gene pool. Both intergenomic and intragenomic rearrangements occur in this 'flexible' gene pool [1]. Changes in the 'flexible' gene pool are considered to be the motor of bacterial diversification and evolution [2-4].However, comparative genomic analyses of genomic variants within a clonal population are rarely undertaken because of the difficulties involved in using molecular approaches in a mixed population. Initially, researchers focused on local modifications of the DNA sequence occurring during phase variation. Phase variation is an adaptive process by which certain bacteria within a bacterial subpopulation, called phase variants, undergo frequent and reversible phenotypic changes. Phase variation is dependent on DNA sequence plasticity, generating a reversible switch between 'on' and 'off' phases of expression for one or more protein-encoding genes. Variation in the expression of certain genes in some phase variants allows the bacterial population to adapt to environmental change [5-7]. Other studies have focused on DNA sequence variations that involve large regions of the genome in a clonal population. These extensively distributed and large genomic rearrangements mostly occur through homologous recombination between repeated sequences such as rrn loci, duplicated genes, or insertion sequences, which may then lead to the inversion, amplification, or deletion of chromosomal fragments. These events can occur either under strong selective pressure - such as in vitro antibiotic selection [8], stressful high temperature [9], long-term storage [10-12], and chronic clinical carriage [13] - or without specific selective pressure [14-20].The phenotypic consequences of such large rearrangements are variable. In Streptomyces spp., genetic instability affects various phenotypical properties, including morphological differentiation, production of secondary metabolites, antibiotic resistance, secretion of extracellular enzymes, and gene expression for primary metabolism, regardless of selective pressure [20]. In other bacterial species and when stressful selective pressure is applied, large-scale genomic variation often correlates with modification of certain phenotypes: reversion from nutritional auxotrophy to prototrophy [10], variation in colony morphology [11], modification of bacterial growth features [12], and adaptation to high temperature [9]. Few data are available on phenotypic variation in the absence of strong selective pressure. A few studies suggest that large genomic architecture modifications can occur with or without slight detectable phenotypic modifications [15,16]. We studied genomic rearrangements in the entomopathogenic bacterium Photorhabdus luminescens, for which variants are frequently observed in standard growth conditions, in order to investigate further the link between genomic variation within a bacterial population and the phenotypic consequences.P. luminescens is a member of the Enterobacteriaceae; it is a symbiont of entomopathogenic nematodes and is pathogenic for a wide variety of insects [21-24]. Bacterial variants frequently arise within the Photorhabdus genus. Two types of variant exist. The phenotypic variants (PVs) are the most studied. The primary PV is characterized by the presence of numerous phenotypic traits (production of extracellular enzymes, pigments, antibiotics, crystalline inclusion bodies, and ability to generate bioluminescence) that are absent from the secondary PV. Secondary PVs are mostly obtained during prolonged in vitro culturing [25,26]. Only primary PVs support nematode growth and development both in the insect cadaver and in vitro. However, both variants are equally virulent to insect hosts [27]. This phenomenon differs from classical phase variation because it occurs at low and unpredictable frequency, it is rarely reversible, and numerous phenotypic traits are altered simultaneously [27]. Recent studies suggest that generation of PVs in P. luminescens may be controlled by several regulatory cascades, each of them involving the products of many different genes [28-31].The other common variants in Photorhabdus are colonial variants (CVs). Different colonial morphotypes can be generated from one colony subculture. This variation is unstable; indeed, each morphotype can generate all other morphotypes [32-36]. The most frequent CVs are small-colony variants (SCVs). These SCVs constitute a slow-growing bacterial subpopulation with atypical colony morphology and unusual biochemical characteristics that, in the case of clinical isolates, cause latent or recurrent infections [37]. In Photorhabdus, these SCVs can be generated from primary or secondary PV [34]. SCVs have small cells, do not produce crystalline inclusions [32-34], and have undergone changes in their proteome [33,34]. Some SCVs have modified virulence properties and do not support nematode development and reproduction [32].Previous studies, incorporating local genetic [28,38,39] or nonexhaustive genomic comparisons [33,34,40,41], have not identified genomic differences within sets of PVs or CVs. We used the recently elucidated complete nucleotide sequence of the P. luminescens subspecies laumondii strain TT01 [42] to study systematically the link between phenotypic and genomic variations in clonal Photorhabdus variants. We undertook whole-genome comparisons between the wild-type TT01 strain and six different PVs or CVs. We showed that large genomic rearrangements occurred in vivo and in vitro. We described two categories of intragenomic rearrangements: deletion events occurring in the 'flexible gene pool', and an unusual duplication of a 275-kilobase (kb) region, encompassing 4.8% of the TT01 wild-type genome. These rearrangements were not correlated with the generation of PVs, and we did not detect a functional relationship between the genes affected by rearrangements and phenotypic variation. Thus, the consequences of these genomic changes are cryptic.ResultsTT01α/I: a genomic variant isolated from the laboratory-maintained nematode Heterorhabditis bacteriophoraThe nematode Heterorhabditis bacteriophora TH01, harboring the TT01 wild-type strain, was collected in Trinidad in 1993 [43]. The nematode was maintained in the laboratory and multiplied by infestation in the Lepidopteran Galleria mellonella [44]. In 1998, a further bacterial isolate was taken from this nematode. During the course of a genetic study of the type III secretion system, we discovered that the bacterium isolated in 1998 is a genomic variant. It differs from the TT01 wild-type strain by a 250 base pair deletion at the 5' end of the gene lopT1 (Additional data file 1). This gene encodes a type III secretion system effector that appears to be involved in the depression of the insect innate immune system [45]. Both TT01 wild-type and the lopT1 genomic variant produced many of the phenotypes associated with primary PVs, including bioluminescence, lipase activity, antibiotic production, and presence of cytoplasmic crystal (Table 1). Therefore, both were primary PVs. To distinguish between them, the TT01 wild-type strain was named TT01/I and the lopT1 genomic variant, TT01α/I (Figure 1).Table 1Phenotypes of P. luminescens TT01/I, TT01α/I, and their respective variantsPhenotypeTT01/ITT01α/ITT01/IITT01α/IITT01α'/IIVARVAR*REVINTBioluminescence++-----+/wwColony morphologyConvex, mucoid,Convex, mucoid,Flat, nonmucoidFlat, nonmucoidFlat, nonmucoidFlat, nonmucoidFlat, nonmucoidConvex, mucoidSmall, convex, mucoidLipase activity on Tween 20-60+++++++++++NDLipase activity on Tween 80-85+++++/w+/w+v+v++NDPigmentation+(Orange)+(Orange)+(Yellow)++(Yellow)---+(Orange)NDAntibiotic production++-----+/wNDCrystal proteins++-----w-Coloration on TreGNO mediumGreenGreenYellowYellowYellowYellowYellowGreenGreen+, positive; -, negative; v, variable; w, weak.Figure 1Schematic representation of TT01 variants selection on TreGNO medium. TT01/I, TT01α/I, and REV colonies are green, convex, and mucoid colonies; TT01/II, TT01α/II, VAR, and VAR* colonies are yellow, flat, and nonmucoid; and the INT colonies are small, green, convex, and mucoid.Isolation and characterization of PVs and CVs from TT01/I and TT01α/IWe cultured TT01/I and TT01α/I in liquid broth and selected primary and secondary PVs on NBTA (nutrient agar supplemented with bromothymol blue and triphenyl 2,3,5 tetrazolium chloride) plates. TT01/II secondary PV was derived from TT01/I (TT01 lineage; Figure 1). TT01α/II and TT01α'/II secondary PVs were obtained from TT01α/I (TT01α lineage; Figure 1). TT01/II, TT01α/II, and TT01α'/II had classic secondary PV traits (Table 1).We developed a new agar medium, the TreGNO (nutrient agar with trehalose and and bromothymol blue) medium, for color discrimination of TT01 PVs (see Materials and methods [below] for details). PVs produce green, convex, and mucoid colonies whereas secondary PVs produce yellow, flat, and nonmucoid colonies on this medium. TT01/II and TT01α/II colonies were homogeneous and had the colonial traits of secondary PVs. However, TT01α'/II was composed of three CVs (TT01α' lineage; Figure 1). The first was a primary colonial form (green, convex, and mucoid colonies), named REV because it resembled a revertant colony, exhibiting primary PV traits (although bioluminescence, pigmentation, and crystal production were not completely restored; Table 1). The second was a secondary colonial form (yellow, flat, and nonmucoid colonies), named VAR because of its secondary PV traits (Table 1). The third form had small, green, convex, and mucoid colonies, and was named INT because of its intermediate traits or traits from both the primary and secondary PVs (Table 1). These CVs are unstable because each individual TT01α'/II colony grown in liquid broth gives rise to a mixture of the three colonial forms on TreGNO medium. We generated a stable secondary PV from the VAR colonial variant by plating a liquid subculture from an individual VAR colony on nutrient agar and picking another VAR colony for a new cycle of liquid/plate culture. We continued this enrichment process until the liquid subculture generated 95% of VAR colonies on TreGNO plates. The stable population was named VAR* (Figure 1).We PCR-amplified the lopT1 5' region from TT01/II, TT01α/II, TT01α'/II, VAR*, and REV (Additional data file 1). The lopT1 deletion was only present in the TT01α and TT01α' lineages.Virulence of TT01 variantsWe injected TT01/I, TT01/II, TT01α/I, TT01α/II, and VAR* into Spodoptera littoralis larvae to evaluate the pathogenicity of these variants in insect larvae. TT01/II, TT01α/I, and TT01α/II had the same level of pathogenicity as TT01/I; 50% mortality (LT50) was reached between 28 and 32 hours after injection for the TT01 wild-type strain and these three variants. By contrast, VAR* had a delayed LT50 of 53 hours, although 100% mortality was reached at 3 days after infection (Figure 2).Figure 2Mortality in Spodoptera littoralis. Shown is the mortality in S. littoralis infected with the TT01/I Photorhabdus luminescens wild-type strain, the genomic variant TT01α/I, the secondary variants TT01/II and TT01α/II, and the stabilized VAR* colonial variant. Bacteria obtained at the end of the exponential phase were injected into fourth-instar larvae. Mortality values are based on data obtained after injection into 20 larvae. All experiments were repeated at least twice.Extensive rearrangements in genomic architecture correlated with the variant lineagesWe examined the whole genome architecture of each variant using I-CeuI genomic macrorestriction and pulsed field gel electrophoresis (PFGE) in order to detect large rearrangement such as deletions and amplifications by recombination between rrn or deletions, amplifications, and translocations inside I-CeuI fragments. I-CeuI is an intron-encoded enzyme that specifically cleaves a 26-base-pair site in the bacterial 23S rRNA gene. The PFGE pattern obtained for the TT01/I strain matched the pattern of I-CeuI fragments predicted from the complete TT01/I genome sequence (Figure 3a, b; also see Additional data file 2 for the details of the gels). Using the TT01/I pattern used as a reference, we observed large genomic rearrangements in TT01α/I, TT01α/II, TT01α'II, VAR*, and REV. PFGE patterns revealed identical profiles for primary and secondary PVs within both TT01 and TT01α lineages (Figure 3b and Additional data file 2). Therefore, PV status (primary versus secondary) in these variant lineages is independent from global genomic architecture.Figure 3Variation in genomic architecture of the TT01 variants. (a) Schematic representation of the I-CeuI restriction map of the TT01/I Photorhabdus luminescens reference genome. (b) Schematic reconstruction of I-CeuI pulsed field gel electrophoresis (PFGE) patterns for TT01/I and the six variants representing gels presented in Additional data file 2. Fragment sizes were calculated using the TT01/I genome as a reference. Lane 1: TT01/I. Lane 2: TT01/II. Lane 3: TT01α/I. Lane 4: TT01α/II. Lane 5: TT01α'/II. Lane 6: VAR*. Lane 7: REV. (c) Clustering of the PFGE patterns. Patterns were compared using the Dice coefficient for each pair. Patterns were clustered by UPGMA.Cluster analysis of the seven observed I-CeuI patterns reveals that variant lineages are in fact genomic lineages (Figure 3c). The TT01 and TT01α lineages exhibit genomic homogeneity. The TT01α' lineage shared common genomic features with the TT01α lineage, but exhibited a more polymorphic genomic pattern than TT01 and TT01α lineages.The PFGE patterns of TT01α and the TT01α' lineages only reveal six apparent I-CeuI fragments, instead of seven fragments in the TT01/I reference chromosome; however, the intensity of the 295-kb band suggests that it may represent two different fragments. We used Southern blot analysis to confirm that the seven rrn copies are present in all the variants (Additional data file 3). Therefore, variation in I-CeuI PFGE patterns among the TT01 variants appeared to be unrelated to deletion or amplifications mediated by recombination between rrn operons.Additionally, the 465 kb faint band in the TT01α'/II pattern (white star in Additional data file 2) corresponded to a fragment in the REV pattern, suggesting the existence of a 'REV-like' chromosome subpopulation in TT01α'/II.Deletions and amplifications in the TT01α/I and VAR* variants, representative of the TT01α and TT01α' lineagesLarge genomic rearrangements were present in the TT01α and TT01α' lineages. We further evaluated the nature of these rearrangements by comparing gene content between representative variants of each lineage, TT01/I, TT01α/I and VAR*, using genomic DNA hybridization on a P. luminescens TT01/I microarray.Totals of 159 and 162 genes were absent from TT01α/I and VAR*, respectively (see Additional data file 4). We located these genes on a circular map of the TT01/I chromosome (Figure 4); they mostly clustered into eight regions absent from both the TT01α/I and VAR* genomes (regions A, C, D, E, F, G, I, and J) and one region specifically absent from the VAR* genome (region H). The deleted regions were located throughout the chromosome, with no particular symmetry around the replication origin or termination site. Several regions displayed a GC bias inversion (C, D, E, G, I, and J). Three overlapped with phagic regions (C, G, and I), suggesting that prophage excision occurred during the TT01/I to TT01α/I transition (Table 2). As well as phagic genes, the deleted regions encompass putative mobile and recombination-mediating elements such as insertion sequences and recombination hotspot (Rhs) elements (region A), and plasmid-related protein-encoding genes (region J)(Table 2). The regions C, D, E, and F potentially encode peptide synthetases involved in antimicrobial compound synthesis (Table 2). However, we did not observe any significant difference in antimicrobial activity between TT01/I and TT01α/I tested for 14 indicator strains (data not shown).A more thorough analysis of hybridization ratios revealed that 122 genes had a ratio higher than 1.4 in the VAR* genome (Additional data file 5). In contrast, comparison of the TT01/I and TT01α/I genomes revealed only four genes with a ratio higher than 1.4. These findings suggest that numerous genes are amplified in the VAR* genome. Among these potentially amplified genes, 112 are clustered in a unique and large 275-kb region, named B. This region encompasses 4.8% of the TT01/I genome (from plu0769 = mrfA to plu0980 = hpaA; Figure 4). Region B is located within the first quarter of the TT01/I chromosome and is not delimited by obvious repeat elements. According to TT01/I genome annotations, the region B may be involved in numerous and different functions (Table 2): basal cellular functions involving the DNA polymerase III ε chain (plu0943 = dnaQ), enolase (plu0913 = eno), and proteins involved in tryptophan metabolism (plu0799 = tnaA; plu0800 = mtr); and environment and/or host interactions, involving the major fimbrial biosynthesis locus (plu0769-0778 = the mrfABCDEFGHJ operon), insecticidal toxin proteins (plu0805 = tccA3; plu0806 = tccB3; plu0960 = tcc2; plu0961 = tcdB1; plu0962 = tcdA1; plu0964 = tccC5; plu0965 = tcdA4; plu0970 = tcdB2; plu0971 = tcdA2), and proteins similar to pyocins (plu0884; plu0886-0888; plu0892; plu0894).Table 2Deleted and amplified regions in the TY01α/I and VAR* genomesLocusProbable nature of eventGene regionSize (in kb)Products of interest (similarity or function)Matching GIa or EVRbADeletionplu0338-plu035518DNA cytosine, ethyl-transferase, mismatch repair endonuclease, unknown proteins, Rhs proteins, IS630 familyPart of GI plu0310-plu0373BAmplificationplu0769-plu0980275Proteins involved in basal metabolism (DNA polymerase III ε chain, enolase, tryptophan metabolism) and in interaction with environment and/or host (fimbrial biosynthesis, Tc insecticidal toxins, pyocins)Encompassed GI plu0884-plu0901, GI plu0914-plu0938, and overlapped a part of GI plu0958-plu1166CDeletionplu1086-plu112344Unknown proteins, phage regulators, peptide synthetase, transposase, bacteriophage proteinsPart of GI plu0958-plu1166DDeletionplu1861-plu187612Antibiotic biosynthesisPart of GI plu1859-plu1894EDeletionplu2191-plu220011Antibiotic synthesis and transportPart of EVR plu2179-plu2224FDeletionplu2468-plu24768unknown protein, ABC transporter, toxoflavin biosynthesis, transposaseEVR plu2468-plu2476GDeletionplu2874-plu296054Bacteriophage proteinsPart of GI plu2873-plu3038HDeletionplu3238-plu325222Unknown proteins, VgrG proteinsPart of GI plu3207-plu3275IDeletionplu3380-plu350489Bacteriophage proteinsPart of GI plu3379-plu3538JDeletionplu4324-plu432812Unknown and plasmid-related proteinsEVR plu4319-plu4332a Genomic islands described in [42]. b Enterobacterial variable regions described in [56].Figure 4Schematic representation of DNA microarray data as a circular map of the TT01/I genome. Circle 1 (from outside to inside): scale marked in megabases. Circle 2: location of transposases (red) and phage-related genes (green) location. Circles 3, 4, and 5: DNA microarray data comparing TT01/I and TT01α/I genomes (circle 3), TT01/I and VAR* genomes (circle 5), and synthesis from both experiments (circle 4). Deleted genes are represented by bars inside the circle. Amplified genes are represented by bars outside the circle. Deleted and amplified regions are circled in blue. Circle 6: GC bias (G-C/G+C). Circle 7: GC content with <32% G+C in light yellow, between 32% and 53.6% G+C in yellow, and with >53.6% G+C in dark yellow.To determine whether DNA microarray experiments explain the architectural modifications observed by macrorestriction experiments, we compared the two sets of data. The observed I-CeuI macrorestriction fragments from the TT01α lineage (36 kb, 295 kb, 295 kb, 330 kb, 465 kb, 610 kb, ~3600 kb) were similar to the theoretical I-CeuI fragments calculated after size subtraction of the eight deleted regions from the TT01/I I-CeuI fragments (36 kb, 244 kb, 266 kb, 330 kb, 462 kb, 627 kb, ~3478 kb). Therefore, large-scale deletion events appear to underlie the TT01 to TT01α lineage transition. DNA microarray experiments in the TT01α' lineage identified a 275 kb amplification of the TT01/I genome. Duplication or triplication of region B may account for the increase in genome size (~100 kb to 650 kb) observed by macrorestriction for the TT01α to TT01α' transition. Therefore, duplication appears to be mainly responsible for the TT01α to TT01α' lineage transition.Homologous recombination between long repeats led to serial deletions of the region H in the TT01α and TT01α' lineagesWe first examined the genomic deletions observed in the TT01α/I and VAR* variants. We focused on region H, which, by contrast to other deleted regions, did not exhibit typical recombination-mediating elements. Probes targeting different parts of the region H were hybridized on genomic DNA of the wild-type strain and the six variants. Hybridization patterns were identical within each variant lineage and confirmed the presence of a 25 kb deletion within the region H (from plu3237 to plu3253) for the TT01α' lineage (data not shown). Southern analysis also indicated the presence of a small deletion of about 10 kb (from plu3238 to plu3248) in the TT01α lineage. To map the deletion borders accurately, primers flanking the 25-kb deletion (R-3236 and F-3254) and the 10-kb deletion (R-3238bis and F-3249) were designed (Figure 5) and used for PCR amplification in the TT01α' and TT01α lineages. Amplified fragments of 4.8 kb and 5.2 kb were observed (data not shown). These fragments were sequenced for TT01α/I and VAR*, and the deletion was physically mapped (a genetic map of the region H is presented in Figure 5). The deletions in TT01α/I and VAR* were 12,820 bases (from coordinates 3,833,904 to 3,846,723) and 25,140 bases long (from coordinates 3,830,001 to 3,855,140), respectively.Figure 5Successive deletions between homologous repeats in the region H. Genetic map of TT01/I region H is shown (blue boxes are open reading frames [ORFs]). Location of repetition units (RPT) larger than 1 kilobase (kb) is indicated (hatched colored boxes). RPT were systematically searched on the whole TT01/I genome sequence by using Nosferatu, software that can detect approximate repeat sequences [46]. The RPTs are numbered according their position on the chromosome. DNA microarray data for the TT01α/I and VAR* genomes are indicated. '+': the gene is present. '-': the gene is absent. '?': the gene is not represented on the microarray. Schematic representation of TT01α/I and the VAR* variant deletions is shown. Deletion borders were obtained from sequencing between the R-3236 and F-3254 primers in the VAR* variant, and between the R-3238bis and F-3249 primers in the TT01α/I variant. Green and hatched gray boxes represent regions in TT01α/I and VAR* genomes variants that were found to be present or absent, respectively. The deleted regions encompassed sequence between coordinates 3.833.904 and 3,846,724 in TT01α/I genome and coordinates 3,830,001 and 3,855,141 in VAR* genome.We used Nosferatu, software that can detect approximate repeats in large DNA sequences [46]. The region H is rich in pairs of repetition units (RPT) larger than 1 kb (Figure 5). Each deletion began at the right-hand extremity of the first repetition and finished at the right-hand extremity of the corresponding second repetition (RPT179385 repetitions for the 10-kb deletion and RPT179383 repetitions for the 25-kb deletion). Therefore, successive deletions mediated by homologous recombination between RPT are likely to have occurred in the region H during the TT01/I to TT01α/I to VAR* transition, leading to genomic reduction.A single block duplication of region B is specific to the TT01α' lineageIn a second set of analyses, we focused on the gene amplification observed in region B, occurring in the TT01α/I to VAR transition. Quantitative PCR was performed for two genes in region B, mrfA (plu0769) and dnaQ (plu0943). Comparison of VAR and TT01α/I data confirmed that these two genes were duplicated in the VAR* genome (Figure 6a).In order to determine whether region B is duplicated specifically in the VAR* variant or in all variants of the TT01α' lineage, a probe covering the entire region B (the probe B) was prepared and hybridized to genomic DNA of the wild-type strain and the six variants. According to the TT01/I genome sequence, NotI hydrolysis generates 25 fragments with a unique 1,056-kb fragment containing region B. Hybridization of the probe B to NotI-hydrolyzed genomic DNA generated a unique fragment of 1,056 kb in the TT01 lineage and of 1,020 kb in the TT01α lineages (Figure 6b, c). By contrast, in the TT01α' lineage, the B probe hybridized to the 1,020-kb fragment and an additional fragment. This second fragment has a similar size in TT01α'/II and VAR* variants (610 kb) but is smaller (365 kb) in the REV variant. These findings showed that duplication of region B occurred in all TT01α' lineage variants.Figure 6Duplication of region B. (a) Quantitative PCR was carried out for mrfA (plu0769) and dnaQ (plu0943) using genomic DNA from TT01α/I and VAR* variants and specific internal primers for each gene. pilN (plu1051) and fliC (plu1954) were used for negative controls. PCR was performed in triplicate and data are presented as ratios, with gyrB as the control gene (95% confidence limits). (b, c) Pulsed field gel electrophoresis (PFGE) of NotI-hydrolyzed genomic DNA from TT01α/I and the six variants following by Southern blot and hybridization with a probe covering the region B (probe B). The PFGE conditions allow separation of NotI fragments between 50 and 400 kb (panel b) or between 350 and 1,350 kb (panel c). Gray arrows indicate fragments that hybridize with the probe B. Lane 1: TT01/I. Lane 2: TT01/II. Lane 3: TT01α/I. Lane 4: TT01α/II. Lane 5: TT01α'/II. Lane 6: VAR*. Lane 7: REV.Region B encompasses 275 kb in the TT01/I genome sequence; thus, we determined whether the resulting amplified genes were dispersed in the genome or co-localized in an unique block. The unique additional fragment detected by the probe B in the TT01α'/II and VAR* variants indicated that the product of the region B amplification is constituted either of one block or a few blocks co-localized in a genomic region whose size is smaller than 610 kb in TT01α'/II and VAR* and smaller than 365 kb in REV. The probe B was also hybridized to ApaI-hydrolyzed genomic DNA of the wild-type strain and the six variants. The seven patterns were identical and the probe B hybridized with the two main 74 and 156 kb fragments covering the major part of region B according to the TT01/I genome reference sequence (data not shown). Because the duplication did not modify the ApaI restriction pattern, we concluded that region B was amplified as a single block.DiscussionVariant lineages are genomic lineages characterized by extensive genomic rearrangementsOur study provides the first extensive investigation into genomic rearrangements in Photorhabdus variants. First, we evaluated phenotypic traits of the three variant lineages (Figure 1). The TT01 lineage is derived from the TT01/I strain, which was isolated from the H. bacteriophora TH01 nematode collected in Trinidad in 1993 [43] and whose genome is sequenced [42]. The TT01α lineage is derived from the TT01α/I genomic variant, which was collected from H. bacteriophora TH01 maintained and multiplied in the laboratory. The TT01α' lineage was derived from the TT01α/I variant after prolonged culture in synthetic medium. Each lineage is composed of PVs, whereby the primary form is characterized by the presence of typical phenotypic traits that are absent from the secondary form. The TT01α' lineage has an additional level of complexity, because the PVs exhibit features of CVs such as unstable morphotypes.We then examined the genomic architecture of each variant in macrorestriction experiments and used comparative DNA microarray hybridization experiments to analyze the genomic content of representative variants for each lineage. Our findings revealed that large genomic rearrangements characterize each variant lineage. Consequently, these findings provide insight into probable scenarios underlying each lineage transition. The whole-genome organization of the TT01 lineage is described by the TT01 reference genome [42]. Large-scale deletion events in the TT01 flexible gene pool seem to be involved in the TT01 to TT01α lineage transition. Deletion events in the TT01 flexible gene pool and a single block duplication encompassing 4.8% of the TT01 reference genome appear to underlie the TT01 to TT01α' lineage transition. The genomic clusters do not depend on the PV status (see below). Thus, each variant lineage is a genetic lineage.Deletion at new recombination hotspotTo explain the molecular mechanisms involved in the rearrangements in our variants, we investigated potential repetitive elements and recombination-mediating elements flanking the rearranged regions. Large genomic architectural changes are often driven by homologous recombination between repeated sequences. The nature of the change then depends on the relative orientation, size, and spacing of the repeated sequences [47-50]. Recombination events often occur at the rrn operon in Gram-negative bacteria, such as Salmonella, Rhizobium, Escherichia coli, and Ochrobactrum [11,13,18,51,52]. However, despite the variation detected in PFGE analysis of the rrn skeleton for the three variant lineages, we demonstrated that the rearrangements are not the result of rrn recombination.Apart from homologous recombination, rearrangements can be induced by site-specific recombination, associated with recombination-mediating elements such as mobile elements, or by illegitimate recombination, linked to shortly spaced repeats [49,50]. Most of the deleted regions in the TT01α lineage are rich in potential rearrangement-mediating elements, with both repeated sequences - including insertion sequences and Rhs elements - and mobile elements, including phagic and plasmid-related genes.Genomic annotation of the region H, which underwent successive deletions in the TT01α/I and VAR* variants, did not describe the presence of typical repetitive or recombination-mediating elements. The region H belongs to a large genomic island containing the genes vgr and hcp, initially described as genes associated with Rhs elements. Rhs elements are repeated sequences in the E. coli genome that mediate major chromosomal rearrangements [51,53,54]. Although the TT01/I genome contains Rhs-like elements [42], no Rhs element is located in the genomic island encompassing the region H. Nevertheless, we identified pairs of approximate long repeated sequences (>1 kb) in direct orientation (RPT) that corresponded to the observed deletion junction points. Therefore, successive deletions in the region H are likely to have been mediated by homologous recombination between RPT during the transition from TT01/I to TT01α/I to VAR*, leading to genomic reduction.There was a strong selective pressure during the TT01α to TT01α' lineage transition (3 months in LB broth without shaking). This environmental constraint could thus be responsible for the rearrangement leading to the region H deletion. However, the region H deletion was already initiated during the former transition (TT01/I to TT01α/I in the laboratory-maintained nematode). Therefore, the observed reduction genomic size is more likely to be the result of particular genomic features (the RPT) rather than environmental constraints.The region H is unique in the TT01/I genome. Nevertheless, some RPT elements have similarities with sequences elsewhere in the TT01/I genome, in the Photorhabdus strain W14 genome [55] or in other Enterobacteriaceae genomes such as Yersinia pseudotuberculosis IP32953 (BX936398.1), Yersinia pestis Angola (CP000901.1), Yersinia pestis Pestoides F (CP000668.1), Yersinia pestis CO92 (AL590842.1), Yersinia pestis biovar Microtus str. 91001, (AE017042.1, Yersinia pestis Antiqua (CP000308.1), Yersinia pestis Nepal 516 (CP000305.1), Yersinia pestis KIM (AE009952.1), Yersinia pseudotuberculosis IP 31758 (CP000720.1), and E. coli CFT073 (AE014075.1). Therefore, we propose that the region H represents a new type of bacterial recombination hot spot, which is vgr- and hcp-rich, but lacks Rhs elements.A new duplication classWe described a single block duplication (region B) targeting a 275-kb region of the TT01/I genome in the TT01α' lineage. This significant duplication encompasses 4.8% of the TT01/I genome. Region B is not located near the replication origin or termination and does not correspond to genomic islands or enterobacterial variable regions previously identified [42,56]. GC content or GC skew deviations are not evident.Gene amplifications can occur through three kinds of known mechanism: homologous recombination between direct repeats, illegitimate recombination, or escape replication. No repeated elements flanking region B were detected, despite the use of the Nosferatu software [46], excluding the possibility of homologous recombination underlying this duplication. Region B duplications may result from illegitimate recombination between short repeats [47,57,58]. However, amplification copy number resulting from illegitimate recombination events is often high, even for large amplicons, such as in Acinetobacter sp. ADP1 or Streptomyces kanamyceticus [59,60]. Escape replication involves amplification of large regions of the host genome (several hundred kilobases), next to phage integration sites after induction of the phage lytic cycle [61-64], or around degraded prophages without the induction of specific phage lysis [65]. Although phage remnants represent 4% of the Photorhabdus genome [42], lytic phages have not been identified in Photorhabdus strains, even after extensive investigation of lytic induction conditions [66]. We detected the presence of an 11-kb phagic segment (plu0818-plu0826) in region B, potentially representing a degraded prophage. However, whereas the copy number usually resulting from the escape replication mechanism ranges between three and ten, with its intensity decreasing symmetrically from the center, region B in the TT01α' lineage genomes represents a single block homogeneous duplication. We only identified one other previously reported example of a large duplication without repeated flanking sequences - a 250 kb duplication in Mycobacterium smegmatis mc2 155 genome [67]. Therefore, the duplication of region B is likely to belong to a new class of duplications.Observed phenotypes and global genomic architecture are not systematically correlatedLarge genomic changes such as deletions and duplications are supposed to have important fitness effects. In our study, we firstly demonstrated that the PV status (primary or secondary) is independent from global genomic architecture. This was consistent with previous studies analyzing specific genetic regions [28,38,39] and with partial genome studies [33,40,41], but this is the first time it has been demonstrated using a whole-genome approach.We showed that the overall genomic pattern corresponds to the variant lineage. Both the phenotype and pathogenic traits of the primary PV (or the secondary PVs) are indistinguishable between the TT01 and TT01α lineages. Therefore, changes in the genomic architecture of these strains did not lead to observable changes in the phenotype. Furthermore, certain regions that were deleted in the TT01α lineage potentially encode biosynthesis pathways for antimicrobial compounds. However, we did not observe any difference in antimicrobial activity between TT01/I and TT01α/I. This finding suggests that some TT01/I genes are redundant. Indeed, genes encoding proteins potentially involved in the biosynthesis of antimicrobial compounds are over-represented in TT01/I genome [42]. Moreover, the encoded proteins in the deleted regions may be adaptive factors required for specific conditions that are not encountered in the laboratory or in our antibiosis assays.The TT01α' lineage differs from the two other lineages due to its polymorphic genomic pattern. Furthermore, this lineage is composed of three unstable CVs and the virulence of the stabilized VAR* variant is attenuated in insects. This is consistent with previously reports of CVs isolated from the Photorhabdus genus [33]. Therefore, changes in genomic architecture might be correlated to phenotypic changes in variants of this lineage. The main rearrangement observed in the TT01α' lineage is the region B duplication. According to TT01/I genome annotation, region B may be involved in both basal cellular functions and environment and/or host interactions. Gene duplication events can underlie modification of phenotypes [58]. However, we did not detect any modification of gene transcription in region B using transcriptomic microarray comparison between the VAR* and TT01α/I variants (Gaudriault S, unpublished data). Thus, this duplication does not appear to modify gene expression in the VAR* variant. Therefore, the attenuation of virulence of the VAR* variant is not likely to be due to amplified expression in region B. Rather, it is more likely that the 'cost' to the bacteria of the increased genome size is decreased virulence in insects.We conclude that the observed phenotypes and overall genomic architecture are not systematically correlated in TT01, TT01α, and TT01α' lineages. It is likely that this result is general in the field of bacterial genomic architecture. Similar observations were previously made between strains of the Pseudomonas aeruginosa species [68], but also inside a clonal bacterial population of a wide range of bacterial groups such as Yersinia pestis [19], Pseudomonas aeruginosa [17], and Sinorhizobium meliloti [16].Stability and plasticity of bacterial genome architectureDo large genomic rearrangements occur randomly or are they shaped by drastic selective evolutionary forces? Several years of comparative genomics between whole bacterial genomes showed that the prokaryotic genome is a heterogeneous entity, with regions of stability and flexibility [4,49,50]. Genomic stability is subject to selective pressures such as functional replication [69], gene essentiality [70], or translation [71]. The three main routes of evolution of genome repertoire are lateral gene transfer, when several bacterial communities share a same ecological niche, deletions, and duplications [4,49,50]. The dynamism of genome repertoire inside a clonal population only arises by the last two phenomena, as illustrated by our study on Photorhabdus variants.In E. coli, the chromosome is organized in structured macrodomains, limiting genome plasticity. Whereas some genomic rearrangements between these macrodomains have only moderate effects on cell physiology, others have detrimental effects [72]. The rearrangements that we observed in our variants may have been selected to preserve chromosomal configurations that are not detrimental for bacterial fitness in the laboratory or in the nematode. We believe that structured macrodomains that restrict chromosome plasticity are likely to exist in other bacterial genus. Identification of structured macrodomains in P. luminescens genome would provide better knowledge on evolutionary forces modeling bacterial genome.Clonal variation, environmental adaptation, and bacterial evolutionThe major genomic variations described in TT01 variants have cryptic consequences in our laboratory conditions. The absence of associated phenotypes makes them difficult to identify, explaining why such genomic variations are rarely observed. However, further studies of such genomic variations may be crucial for a better understanding of bacterial adaptation and evolution.Indeed, we observed that the extensive genomic rearrangements in Photorhabdus variants were often associated with several genomic subpopulations in the same culture. Similar observations were previously made for a P. luminescens TT01/I locus encoding a phage tail-like structure [73] and the mrf locus of the P. temperata strain K122 [74]. In Sinorhizobium meliloti, Yersinia pestis, and Pseudomonas aeruginosa, extensive variations of genome architecture, without obvious changes in phenotype, were also observed during bacterial growth in broth medium [16,17,19]. Different pre-existing chromosomal forms in a clonal bacterial population are likely to give this population an adaptive capacity. It is therefore possible that bacterial populations maintain various subpopulations with different genomic structures as a way to cope with different environments during its life cycle.Additionally, deletion events in TT01α and TT01α' lineages are located within the TT01/I 'flexible' gene pool. Whereas intragenomic recombination in the 'flexible' gene pool have been widely studied using comparative genomics for different bacterial genera, species, and strains [1-4], similar reports for clonal variants are rare. Gene repertoires of the 'flexible' gene pool may evolve through variations in bacterial subpopulations and then become fixed after bacterial speciation. Such pre-existing or currently existing genomic variations have an important role in evolutionary patterns of natural eukaryotic populations [75]. They may also have a determinant role in bacterial evolution.ConclusionThe study of molecular mechanisms underlying genomic plasticity in clonal populations is challenging because classical molecular tools only detect the major genomic state of the population. Such studies are easier in bacterial species with a high rate of bacterial variants. With our model, P. luminescens, we identified two new genomic rearrangements, allowing a new research axis for gaining a comprehensive knowledge of bacterial chromosome plasticity. The cryptic consequences of large genomic rearrangements in our model also allow prospective comprehensive analysis of bacterial genome evolution. Therefore, we propose that the P. luminescens TT01 strain represents a new bacterial model for study of genomic plasticity.Materials and methodsStrains, plasmids, primers, and culture mediaAll bacterial strains and plasmids used in this study are listed in Additional data file 6. Primers are listed in Additional data file 7. P. luminescens was grown at 28°C in LB broth or on nutrient agar 1.5% (BD Difco™, Franklin Lakes, New Jersey, USA) for 48 hours. Escherichia coli was grown at 37°C in LB broth or on LB supplemented with 1.5% agar (BD Difco™, Franklin Lakes, New Jersey, USA). Strains were stored at -80°C in LB broth containing 16% glycerol (vol/vol). Secondary variants were obtained by prolonged culture of primary variants at 28°C for 10 days in Schneider's insect medium (Cambrex Bio Science, Walkersville, Maryland, USA) with shaking (TT01/II [30]), for 10 days in LB broth with shaking (TT01α/II), or for 3 months in LB broth without shaking (TT01α'/II). Secondary variant phenotypes were evaluated from culture on NBTA (nutrient agar 1,5%, 25 mg/l bromothymol blue and 40 mg/l triphenyl-2,3,5-tetrazolium chloride) plates and on TreGNO plates (see below) at 28°C. Secondary variants were identified by performing phenotypic tests as previously described [76] and controlled by PCR-restriction fragment length polymorphism of the 16S rRNA gene [77].Analysis of phenotypic variants on a new selective medium: TreGNOXenorhabdus and Photorhabdus secondary variants are typically selected on NBTA plates to distinguish red secondary variants colonies from blue primary colonies [76]. Because of the high level of pigmentation of Photorhabdus colonies, the use of color assays does not allow clear distinction between primary and secondary variants for Photorhabdus genus. We found that TT01 secondary variants were able to undergo trehalose fermentation, whereas primary variants can not. On nutrient agar plates supplemented with trehalose (10 g/l) and bromothymol blue (25 mg/l), secondary colonies acidified the bromothymol blue and became yellow at 28°C after 48 hours. Primary colonies remained green. Furthermore, secondary colonies were flat and large with irregular borders. This new medium was named TreGNO medium and was routinely used for the discrimination of Photorhabdus luminescens strain TT01 phenotypic variants.PFGE and DNA electrophoresisIntact genomic DNA was extracted in agarose plugs as follows. Bacterial cells grown on nutrient agar plates were suspended in phosphate-buffered saline (GIBCO® Invitrogen, Carlsbad, California, USA) to a turbidity of 1.25 at 650 nm, included in 1% (vol/vol) low melting agarose (SeaPlaque® GTG, FMC BioProducts, Rockland, Massachusetts, USA) solution and then subjected to lysis as described previously [78].NotI and ApaI hydrolysis were performed by incubation of the agarose plugs overnight with 40 units of the endonuclease in buffer recommended by the supplier (New England Biolabs, Hertfordshire, UK), at 37°C for NotI and 25°C for ApaI. PFGE was carried out in a contour-clamped homogeneous field electrophoresis apparatus CHEF-DRII (Bio-Rad, Hercule, California, USA) in a 0.8% agarose gel in 0.5× Tris-borate-EDTA (TBE) at 10°C. PFGE conditions were as follows: for NotI fragments, a 35 to 5 second pulse ramp for 47 hours followed by a constant pulse time of 50 seconds for 6 hours at 4.5 V/cm; and for ApaI fragments, 35 to 5 seconds for 35 hours, followed by 5 seconds to 1 second for 4 hours at 4.5 V/cm.I-CeuI hydrolysis was performed as described previously [79]. For the separation of I-CeuI fragments, different electrophoresis conditions were selected according to fragment size: a pulse ramp from 5 to 50 seconds for 24 hours at 6 V/cm for fragments with size below 700 kb; and a pulse ramp from 150 to 400 seconds for 45 hours at 4.5 V/cm for I-CeuI fragments for fragments between 700 kb and 1 megabase. For I-CeuI fragments larger than 1 megabase, PFGE was performed on Rotaphor apparatus (Biometra, Goettingen, Germany) using 0.7% agarose gels in 0.5× TBE buffer. The electrophoresis conditions used were as follows: 50 to 47 V (linear ramp), 6,000 to 1,000 seconds decreasing pulses (logarithmic ramp), with a increasing angle from 96 to 105°, buffer temperature 11°C, for 240 hours. I-CeuI PFGE patterns were compared by calculating the Dice coefficient for each pair [80]. Patterns were clustered by UPGMA using the Phylip program package [81].HindIII-hydrolyzed DNA was subjected to electrophoresis for 3 hours at 2.6 V/cm in a 0.8% agarose gel in 0.5× TBE using SubCell apparatus (Bio-Rad) [13].Southern blotting, probes, and hybridization experimentsElectrophoresis gels were transferred onto a Nytran N SuperCharge nylon membrane (Schleicher and Schuell, Dassel, Germany) by vacuum blotting in 20 × SSC (Euromedex, Souffelweyersheim, France).A digoxigenin-labeled probe targeting 16S rRNA gene was obtained by PCR from genomic DNA of P. luminescens strain TT01/I, using primers 27f and 1492r with a dNTP mixture containing 0.1 mmol/l digoxigenin-dUTP [13].Probes B and H were obtained using respectively small fragment insert from plg2711 and large fragment inserts from plbac4g08, plbac6h12, plbac3a10, plbac3c04, and plbac2f12. Fragment inserts were purified, sonicated into fragments of between 1 and 10 kb if insert size was higher than 10 kb, and labeled with digoxygenin by random priming (Dig DNA labeling Kit; Roche, Meylan, France). Hybridization of the probes was detected using a CSPD chemiluminescent system (Roche).Standard DNA manipulationsGenomic DNA was extracted as previously described [56] and stored at 4°C. We PCR-amplified the lopT1 deletion region with Taq polymerase (Invitrogen, Carlsbad, California, USA), in accordance with the manufacturer's recommendations, using the PlopT1.fw an PlopT1.rev primers. The region H was amplified by PCR with the Herculase Enhanced DNA polymerase (Stratagene, Amsterdam Zuidoost, Pays Bas), in accordance with the manufacturer's recommendations, using the R-3236, F-3249, R-3238bis, and F-3254 primers. For sequencing region H deletions, we purified the 4.8 kb and 5.2 kb fragments using the Montage PCR kit (Millipore, Guyancourt, France) and sequenced using PCR primers and chromosome walking (Millegen, Toulouse, France). Sequencing of the 5.2 kb fragment central region of the fragment failed probably because of the presence of repetitions. A 3.2 kb central region was therefore amplified by PCR with PstIdMutF and XbaIdMutR primers. The amplicon was hydrolyzed by PstI and XbaI, ligated into PstI- and XbaI-hydrolyzed pUC19, and inserted into E. coli XL1blue by transformation. The resulting plasmid was purified by Nucleobond AX-100 kit (Macherey-Nagel, Hoerd, France), and the insert was sequenced with PstIdMutF and XbaIdMutR primers and then by chromosome walking.DNA microarray hybridization and analysisDNA microarray hybridization and analysis were performed as previously described [56].Quantitative PCR analysisQuantitative PCR was performed in triplicate using the LightCycler FastStart DNA MasterPLUS SYBR Green I kit from Roche Diagnostics with 1 ng genomic DNA and 1 μmol/l specific primers targeting fliC (L-1954 and R-1954), mrfJ (L-0778 and R-0778), dnaQ (L-0943 and R-0943), and pilN (L-1051 and R-1051). The enzyme was activated for 10 minutes at 95°C. Reactions were performed in triplicate at 95°C for 5 seconds, 60°C for 5 seconds and 72°C for 10 seconds (45 cycles), and monitored in the Light Cycler (Roche). Melting curves were analyzed for each reaction; all reactions exhibited a single peak. The amount of PCR product was calculated with standard curves obtained from PCR with serially diluted TT01/I genomic DNA. All data are presented as ratios, with gyrB (primers L-0004 and R-0004) as a control (95% confidence limits).Sequence analysisSequence annotation of the TT01/I genome was obtained from the MaGe database [82]. We evaluated amino-acid and nucleotide similarity using BLASTP and BLASTN software [83]. We used Repseek software, previously Nosferatu [46], to detect approximate repeats in large DNA sequences.Pathogenicity assaysIn vivo infection assays were performed as previously described [45]. We performed three independent experiments for each variant. Statistical analysis were performed as previously described [84].Antibiosis plate assaysAntibiosis assays were performed as previously described [76] with the following bacterial species: Micrococcus luteus, Staphylococcus epidermidis CIP 6821, Staphylococcus aureus CIP 7625, Escherichia coli CIP 7624, Proteus vulgaris CIP 5860, Pseudomonas aeruginosa CIP 76.110, Corynebacterium xerosis, Ochrobactrum intermedium LMG 3301T, Ochrobactrum anthropi ATCC 49188T, Ochrobactrum sp. FR49, Erwinia amylovora CFBP1430, Pseudomonas sp. BW11M, Salmonella enterica 14028s, and Yersinia enterocolitica serotype 08.AbbreviationsCV, colonial variant; kb, kilobase; NBTA, nutrient agar supplemented with bromothymol blue and triphenyl-2,3,5-tetrazolium chloride; PCR, polymerase chain reaction; PFGE, pulsed field gel electrophoresis; PV, phenotypic variant; Rhs, recombination hotspot; RPT, repetition units; SCV, small-colony variant; TBE, Tris-borate-EDTA; TreGNO, nutrient agar with trehalose and bromothymol blue.Authors' contributionsSG, SP, and AG characterized bacterial variants. SG, AL, and CL provided molecular materials. SG and AL performed microarray analysis. SG, CT, and EJ-B provided PFGE analysis. SG analyzed sequence data. SG wrote the paper with contributions from AG and EJ-B.Additional data filesThe following additional data files are available with this paper. Additional data file 1 is a figure showing the deletion in the lopT1 gene in TT01/I strain and the six variants. Additional data file 2 is a figure showing PFGE of I-CeuI-hydrolyzed genomic DNA of TT01/I strain and the six variants. Additional data file 3 is a figure showing the copy number of 16S rDNA in TT01/I and the six variants. Additional data file 4 is a table listing the TT01/I missing genes in TT01α/I and VAR* variants according whole-genome comparison using DNA microarray. Additional data file 5 is a table listing the TT01/I amplified genes in TT01α/I and VAR* variants, according to whole-genome comparison using DNA microarray. Additional data file 6 is a table listing strains and plasmids used in this study. Additional data file 7 is a table listing primers used in this study.Supplementary MaterialAdditional data file 1Presented is a figure showing the deletion in the lopT1 gene in TT01/I strain and the six variants.Click here for fileAdditional data file 2Presented is a figure showing PFGE of I-CeuI-hydrolyzed genomic DNA of TT01/I strain and the six variants.Click here for fileAdditional data file 3Presented is a figure showing the copy number of 16S rDNA in TT01/I and the six variants.Click here for fileAdditional data file 4Presented is a table listing the TT01/I missing genes in TT01α/I and VAR* variants according whole-genome comparison using DNA microarray.Click here for fileAdditional data file 5Presented is a table listing the TT01/I amplified genes in TT01α/I and VAR* variants, according to whole-genome comparison using DNA microarray.Click here for fileAdditional data file 6Presented is a table listing strains and plasmids used in this study.Click here for fileAdditional data file 7Presented is a table listing primers used in this study.Click here for file\n\nREFERENCES:\nNo References"
+}

batch_5/PMC2531080.json

@@ -0,0 +1,4 @@
+{
+    "id": "PMC2531080",
+    "text": "This is an academic paper. This paper has corpus identifier PMC2531080\nAUTHORS: Marcelo Moreno, Janete Eunice Wiltgen, Benito Bodanese, Ricardo Ludwig Schmitt, Bianca Gutfilen, Lea Mirian Barbosa da Fonseca\n\nABSTRACT:\nBackgroundThe detection of sub-clinical breast lesions has increased with screening mammography. Biopsy techniques can offer precision and agility in its execution, as well as patient comfort. This trial compares radioguided occult lesion localization (ROLL) and wire-guided localization (WL) of breast lesions. We investigate if a procedure at the ambulatorial level (ROLL) could lead to a better aesthetic result and less postoperative pain. In addition, we intend to demonstrate the efficacy of radioguided localization and removal of occult breast lesions using radiopharmaceuticals injected directly into the lesions and correlate radiological and histopathological findings.MethodsOne hundred and twenty patients were randomized into two groups (59 WL and 61 ROLL). The patients were requested to score the cosmetic appearance of their breast after surgery, and a numerical rating scale was used to measure pain on the first postoperative day. Clearance margins were considered at ≥ 10 mm for invasive cancer, ≥ 5 mm for ductal carcinoma in situ, and ≥ 1 mm for benign disease. Patients were subsequently treated according to the definitive histological result. When appropriate, different statistical tests were used in order to test the significance between the two groups, considering a P value < 0.05 as statistically significant.ResultsWL and ROLL located all the occult breast lesions successfully. In the ROLL group, the specimen volume was smaller and there were more cases with clear margins (P < 0.05). There were significant differences in mean time of hospital stay between WL and ROLL (21.42 vs. 2.56 hours), but not in operative time (39.4 vs. 29.9 minutes). There were significant differences in the subjective ease of the procedures as rated by the patients (cosmetic outcomes and postoperative pain).ConclusionROLL is an effective method for the excision of non-palpable breast lesions. It enables more careful planning of the cutaneous incision, leading to better aesthetic results, less postoperative symptoms, and smaller volumes of excised tissue.\n\nBODY:\nBackgroundThe diagnosis of breast cancer has changed over the last years. Previously, about 50 to 70% of breast cancers were diagnosed through physical examination [1]. The detection of subclinical lesions has increased with screening mammography [2]. Thus, the need arose to develop minimally invasive techniques of locating and histological confirming small alterations [3,4]. Wire localization (WL) is a well-known technique in breast surgery where a malleable needle with a spear at its distal extremity is used to locate a lesion. Under mammography or ultrasound visualization, a needle is placed directly into an area suspicious as per the nature of the lesion [3,4]. There is a risk of needle displacement during the period between its positioning and retreat, mainly in breasts with a predominant fatty component [2,4]. This can represent an important complication in patients with mammary prosthesis, for example. In dense breasts, difficulty in positioning the needle localization device can occur. Cases of transected needles, pneumothorax, and other accidents have been described [2,5]. Needle localization of occult lesions is usually done under general anesthesia due to patient discomfort when the needle localization device is manipulated [5-7].Radioguided occult lesion localization (ROLL) is a method that has been used since 1996 [8]. It was developed at the European Institute of Oncology in Milan, and is currently the standard of care in many breast surgery services. In this procedure, a radioactive labeling substance is used at the suspect site (under ultrasound or mammography guidance). The gamma-detecting probe guides the localization of a suspicious opacity or microcalcification cluster during the surgical procedure [6,9]. The cutaneous incision can be planned with better aesthetic results. In this method, a spear is not used; instead, a small portion of liquid makes the process less traumatic for patients. Local anesthesias for ROLL and patients' opinions as to the pain and postoperative aesthetic results have not been previously studied for effectiveness and patient acceptability [7,9].The goal of this paper is to show the feasibility of performing the ROLL technique in an ambulatory setting, with shorter operative time and less patient morbidity, through careful surgical planning and the extraction of a smaller mammary sample. Therefore, these advantages make it the preferred method for occult breast lesion localization with diagnostic intention.MethodsOne hundred and twenty patients with suspicious breast opacity or microcalcification cluster requiring diagnostic excision were randomized and submitted to guided surgical biopsy. WL was performed in 59 patients (49.2%) with standard techniques [2]. For ROLL (61 patients), 0.15 mCi (5.55 MBq) of 99mTc-labeled macro albumin aggregate in 0.2 mL of saline was used. On the day of surgery, this solution was injected into the non-palpable lesion under mammography or ultrasound guidance. In addition, 0.1 mL of water-soluble non-ionic iodinated contrast medium was administered to check the exact position of the radiotracer at the time of injection. One hour later, the patient was submitted to front and lateral view planar scintigraphic images using a 99mTcO4 flood to check the radiographic correlation (Figure 1). The patient was taken to the operating room for excision of the lesion. Localization of the area of highest radioactivity was performed with a hand-held gamma probe (Navigator GPS™ – United States Surgical/Tyco Healthcare) to choose the most cosmetically acceptable site to incise. The specimen was excised after locating the highest radioactivity point and this hot spot was removed. It was located in the center of the specimen with a resection margin, with no excessive removal of normal breast parenchyma. The parenchyma bed was verified with the probe to rule out residual areas of high radioactivity. During surgery, a radiological study was performed to confirm total resection of cases previously demarcated by mammography. The use of local anesthesia for the ROLL procedure was proposed, considering that, contrary to the WL method, the wire is not maintained in the breast during the procedure. In all ROLL cases, local anesthesia (mean of 16 mL/patient of lidocaine with epinephrine-1:200000) was used in the skin and breast parenchyma close to the lesion. The hospital stay considered the period (in hours) between the beginning of the surgery and the discharge from the hospital. The procedure time was considered as the mean time of surgery in minutes. The patients were requested to score the cosmetic appearance of their breast as excellent, good, or poor in the first month after surgery. In addition, a numerical rating scale was used to measure pain on the firs postoperative day, considering a variation between 0 (no pain) and 10 (worst pain) [10,11]. Different statistical tests were used, when appropriate, in order to test the significance between the two groups, considering a P value < 0.05 as statistically significant. Clearance margin was considered as ≥ 10 mm for invasive cancer, ≥ 5 mm for ductal carcinoma in situ, and ≥ 1 mm for benign disease. All specimens were included with transversal serial cuts, with the margin size defined as the distance between the lesion and closest margin. Patients were subsequently treated according to the definitive histology result. This study was performed as per a protocol approved by the Ethical Review Board of Chapeco University.Figure 1Lateral (A) and front (B) view planar scintigraphic images to check the radiographic correlation of a breast lesion (arrow).ResultsFifty-nine patients were randomized to WL and 61 to ROLL. The mean age of the two groups was 51.3 and 49.6 years, respectively. All procedures (in both groups) were done with diagnostic intention. The clinical and radiological characteristics of the two groups are summarized in Tables 1 and 2. There were no significant differences in the lesion sites and in the auxiliary localization technique in both groups. The ROLL technique did not increase the number of aesthetic incisions. The hospital stay was significantly longer in the WL group due to the use of general anesthesia (mean of 19.82 h vs. 2.04 h). This difference was statistically significant, but the result presents the bias that all the patients submitted to the WL technique needed a period of recovery from the anesthetic, i.e., a longer hospital stay. Procedure time was significantly shorter in the ROLL group (37.2 min vs. 26.06 min, respectively). The mean volume of the excised specimen was significantly smaller in the ROLL group than in WL group (8.70 cm3 vs. 23.15 cm3, respectively). There was a difference in the margin status of the surgical specimens (P < 0.05), but these findings were not significant when considering only malignant lesions (Table 2). Postoperative wound infections between the groups were not significant. The mean of the numerical rating pain scale was different in both groups (2.20 for WL vs 1.62 for ROLL), and according to patients' opinions, there were better cosmetic outcomes with the ROLL technique (Table 3). In two cases of WL, there was wire dislodgement and the procedure needed repositioning.Table 1Clinical and radiological characteristics of WL and ROLL groups.WLROLLPNumber of patients5961Mean of age (yr)49.9 (35–77)50.7 (32–76)£ 0.540Micro-calcifications4049Micro-Calcifications + Stromal deformity23Stromal deformity30Nodule149£ 0.080Right Breast1720Left Breast4241£0.351SL Quadrant2227SM Quadrant2119IM Quadrant64IL Quadrant810Central21£0.948Estereotaxy Localization3127US Localization2824£0.130£ Chi-Square Test with Yates' correctionTable 2Comparison of surgical and pathological features of WL and ROLL groups.WLROLLPLocal of cutaneous incision Peri-areolar2131 Others3830§ 0.022Histopathological Diagnosis Fibrocystic changes2534 Fibroadenoma1214 Invasive carcinoma62 Carcinoma in situ108 Others63£0.047Size of specimen (mean in cm3) All lesions23.159.70¥0.001 Benign lesion14.808.70¥0.002Invasive and not invasive carcinomas20.309.45¥0.170Period of hospital stay (mean in hours)18.73.06¥<0.001AnesthesiaGeneralLocalTime of procedure (mean – min)37.226.06¥0.719Margins of all lesions Clear5157 Involved84£0.010Carcinoma Margins Clear149 Involved21#0.670 Postoperative wound infection12£0.270§ Chi Square Test. £ Chi-Square Test with Yates' correction. ¥ T-test. # Fisher exactTable 3Dependent variables on the patient opinion.TechniquesWLROLLPCosmetic outcome Excellent4957 Good1004 Poor--§ <0.001Pain (mean of numerical scale)2.201.62¥ 0.021§ Chi Square Test. ¥ T – testDiscussionNowadays, the diagnosis of subclinical breast lesions is very common due to easy access to standard mammography in most places. Many techniques, such as core biopsy, fine needle aspiration, and mammotomy are used for the histological study of clinically occult breast lesions. Sometimes it is necessary to excise all occult lesions in order to choose the adequate treatment. WL is a method used in many places as standard preoperative localization of non-palpable lesions. However, the problems reported with this technique are well known: wire transection, difficulties in wire repositioning in dense or fatty breasts, dislodgement, interference with the surgical approach, and patient discomfort during wire positioning and during patient transportation from the radiological center to the operating room [2,9,12].Since 1996, when the first paper presented the advantages of ROLL, other authors have reported the same findings and have documented some characteristics of this technique: it is a radiologically and surgically easier procedure to perform, and the lesion can be identified in three dimensions affording greater flexibility in making a cosmetic incision [13]. ROLL is also appropriate for combination with sentinel lymph node mapping in which the occult breast cancer and sentinel lymph node can be excised in the same procedure [9,14,15]. Until now, the methodology for evaluation of postoperative pain has not mentioned the ROLL procedure, despite some works reporting postoperative pain when the WL is carried out [6,8,14,16]. The evaluation of pain on the first postoperative day and of the cosmetic outcomes was used as a parameter for comparing patients' opinions about both procedures, and there was a difference between the two groups. This is due to a better choice of an incision site (radioguided) and the fact that the size of the ROLL specimen is smaller. Hospital stay was shorter in the ROLL group due to the ambulatory characteristics of this procedure. Moreover, with the WL procedure, the patient needed a time to recover from general anesthesia. The failure rate of the wire guided technique (i.e. incomplete cancer resection) has been reported in the range of 40–50% [17]. The duration of the procedure was shorter in the ROLL procedure, which could be explained by better radioguided planning of the method; however, there were no significant findings (p > 0.05). The specimen size (mean in cm3) was smaller in patients submitted to ROLL and there were more cases with compromised margins with the WL procedure (p < 0.05), which again reflects better planning to include all lesions at the same time, and the specimen excised is the smallest possible. This difference, however, could affect the results for rates of cases with involved margins due to the different criteria of histological categories. To date, there have been no descriptions of a comparison between the use of local anesthesia in the ROLL procedure and use of general anesthesia in WL, especially comparing them as to aesthetic results and pain measurement.ConclusionROLL can provide diagnosis or treatment of the breast lesion with a shorter hospital stay, shorter operative period, less breast tissue excised, and consequently, better aesthetic outcomes and fewer procedure-related symptoms. It can result in lower costs and a better acceptance on the part of patients.List of AbbreviationsROLL: Radioguided Occult Lesion Localization; WL: Wire-guided Localization; US: Ultrassonography.Conflict of interestsThe authors declare that they have no competing interests.Authors' contributionsMM carried out all surgeries, participated in the interpretation of the study and drafted the manuscript. JEW participated in the execution of all ROLLs procedures by US or estereotaxy. BB participated in the execution and in the interpretation of the study. RLS participated in the design of study and performed the statistical analysis. BG aided to write the manuscript. LMBF conceived of the study, and participated in its design and coordination. All authors read and approved the final manuscript.\n\nREFERENCES:\nNo References"
+}

batch_5/PMC2531104.json

@@ -0,0 +1,4 @@
+{
+    "id": "PMC2531104",
+    "text": "This is an academic paper. This paper has corpus identifier PMC2531104\nAUTHORS: Stephanie A Tuck, David Ramos-Barbón, Holly Campbell, Toby McGovern, Harry Karmouty-Quintana, James G Martin\n\nABSTRACT:\nAccidental chlorine (Cl2) gas inhalation is a common cause of acute airway injury. However, little is known about the kinetics of airway injury and repair after Cl2 exposure. We investigated the time course of airway epithelial damage and repair in mice after a single exposure to a high concentration of Cl2 gas. Mice were exposed to 800 ppm Cl2 gas for 5 minutes and studied from 12 hrs to 10 days post-exposure. The acute injury phase after Cl2 exposure (≤ 24 hrs post-exposure) was characterized by airway epithelial cell apoptosis (increased TUNEL staining) and sloughing, elevated protein in bronchoalveolar lavage fluid, and a modest increase in airway responses to methacholine. The repair phase after Cl2 exposure was characterized by increased airway epithelial cell proliferation, measured by immunoreactive proliferating cell nuclear antigen (PCNA), with maximal proliferation occurring 5 days after Cl2 exposure. At 10 days after Cl2 exposure the airway smooth muscle mass was increased relative to controls, suggestive of airway smooth muscle hyperplasia and there was evidence of airway fibrosis. No increase in goblet cells occurred at any time point. We conclude that a single exposure of mice to Cl2 gas causes acute changes in lung function, including pulmonary responsiveness to methacholine challenge, associated with airway damage, followed by subsequent repair and airway remodelling.\n\nBODY:\nIntroductionChlorine (Cl2) gas is a common inhalational irritant, encountered both occupationally and environmentally[1,2]. The acute effects of Cl2 gas inhalation can range from mild respiratory mucus membrane irritation to marked denudation of the mucosa, pulmonary oedema, and even death. Recovery from Cl2-induced lung injury requires repair and/or regeneration of the epithelial layer. The repair process after Cl2 exposure may not restore normal structure and function as cases of subepithelial fibrosis, mucous hyperplasia, and non-specific airway hyperresponsiveness have been reported in persons after recovery from Cl2 injury[3,4]. Repeated exposure to chlorine through swimming appears to be a significant risk factor for airway disease manifesting as asthma[5].The airway epithelium is the first target of inhaled Cl2 gas. Although the exact mechanism of epithelial damage is unknown, oxidative injury is likely involved as Cl2 gas can combine with reactive oxygen species to form a variety of highly reactive oxidants [6]. Direct oxidative injury to the epithelium may occur immediately with exposure to Cl2, but further damage to the epithelium may occur with migration of inflammatory cells such as neutrophils into the airway epithelium and the subsequent release of oxidants and proteolytic enzymes.Limited information is available regarding the time course of injury and repair of the epithelium after acute Cl2 gas exposure. Bronchial biopsies from humans have shown epithelial desquamation from 3 to 15 days after accidental Cl2 exposure followed by epithelial regeneration, characterized by proliferation of basal cells at two months post-exposure[7]. Animal studies of Cl2 exposure have furthered our understanding of the time course of injury and repair. However, these studies have been primarily descriptive in nature. Rats acutely exposed to high concentrations of Cl2 gas demonstrated bronchial epithelial sloughing 1 hour after exposure with epithelial regeneration occurring by 72 hrs after exposure[8]. Recently, we have described the response of A/J mice to a single exposure to varying concentrations of Cl2 exposure[9]. Exposure to the highest concentration of Cl2 gas (800 ppm for 5 minutes) resulted in marked epithelial loss and airway hyperresponsiveness to methacholine 24 hrs after exposure.Airway remodelling is a feature of asthma that has the potential to explain the induction and chronicity of the disease. Generally animal models have focussed on allergen-driven changes in airway structure which are of uncertain relevance to irritant-induced asthma. For this reason we wished to explore the injury and repair processes involved in irritant-induced asthma. To do this we characterized the time course of airway injury and repair after a single exposure to Cl2 gas in mice using quantitative measures of epithelial damage and repair. Markers of epithelial damage were apoptosis, assessed by terminal dUTP nick end labelling (TUNEL) staining, and the presence of protein and epithelial cells in the bronchoalveolar lavage fluid. Epithelial repair was assessed by quantifying cell proliferation using the proliferation marker proliferating cell nuclear antigen (PCNA). PCNA is a DNA polymerase-δ cofactor located in the nuclear compartment of proliferating cells [10,11]. Airway remodelling was assessed by quantification of airway smooth muscle mass using standard morphometric techniques on smooth muscle specific α-actin immunostained tissue sections and by scoring of airway fibrosis on Picrosirius red stained tissue sections. Goblet cell numbers were assessed by light microscopy and standard morphometric techniques. Airway histology was also used to qualitatively assess the time course of damage and repair to the airways. We wished to relate these markers of damage and repair to functional consequences of Cl2-induced injury in terms of airway mechanics and airway responsiveness to methacholine.MethodsAnimals and chlorine exposureMale A/J mice (23–27 g) were purchased from Harlan (Indianapolis, Indiana) and housed in a conventional animal facility at McGill University. Animals were treated according to guidelines of the Canadian Council for Animal Care and protocols were approved by the Animal Care Committee of McGill University.Forty-eight mice were exposed to either room air (control) or 800 ppm Cl2 gas diluted in room air for 5 minutes using a nose-only exposure chamber. This concentration of Cl2 gas was chosen as it was previously shown to result in severe airway damage but with minimal animal mortality[9]. Mice exposed to Cl2 were studied at 12 hrs, 24 hrs, 48 hrs, 5 days (d), or 10 d after Cl2 exposure (n = 8 at each time point). The control mice were studied 24 hrs after exposure to room air (n = 8).Bronchoalveolar lavage, lung histology and morphometryThe chest was opened, the left main bronchus clamped, and 0.3 ml of sterile saline followed by four separate 0.5 ml instillations were washed into the right lung. Fluid recovered from the first wash was centrifuged at 1500 rpm for 5 minutes at 4°C and the supernatant used for protein quantification. The cell pellet was pooled with the remaining lavage samples and total live and dead cells were counted using trypan blue exclusion. Cytospin slides were prepared using a cytocentrifuge (Shandon, Pittsburgh, PA) and stained with Dip Quick (Jorgensen Labs Inc., Loveland, CO). Differential cell counts, including epithelial cells, were determined on 300 cells/slide. Total protein in the BAL supernatant was quantified using a dye-binding colorimetric assay (Bio-Rad, Hercules, CA), and determined by spectrophotometry at 620 nm and quantified using a bovine serum albumin standard curve.Tissue preparationFollowing BAL, the lungs were removed and the left lung was fixed with an intratracheal perfusion of 10% buffered formalin at a constant pressure of 25 cmH2O for a period of 24 hrs. Histology and immunohistochemistry were performed on 5 μm thick paraffin-embedded sections taken from the parahilar region. Adjacent sections were either stained with hematoxylin-eosin (H&E), periodic acid Schiff (PAS), or processed for immunohistochemistry.ImmunohistochemistryCells undergoing proliferation were detected in tissue sections by immunostaining for proliferating cell associated nuclear antigen (PCNA. Following deparaffination in xylene and rehydration through graded ethanol solutions, the tissue sections underwent a high temperature epitope unmasking treatment by a modified version of the microwave boiling method. An acidic antigen retrieval buffer (Vector Laboratories, Burlingame, CA) was microwave pre-heated to 95°C, and the slides were incubated in it for 30 minutes using a pre-warmed coplin jar protected with styrofoam. After cooling for 20 minutes, a membrane permeabilization treatment was applied by immersing the slides for 20 minutes in a 0.2% dilution of Triton X-100 (Sigma Chemical Co., St. Louis, MO) in pH 7.6 Trizma base (Sigma) buffered saline. The tissues were then blocked for 1 hour using a blocking reagent designed for immunohistochemistry using mouse primary antibodies on mouse tissues (Vector Laboratories). Primary murine anti-PCNA antibody was applied at a concentration of 2.5 μg/ml and the sections were incubated for 30 min. at room temperature. A biotinylated anti-mouse antibody (1:250 dilution; Vector Laboratories) was applied for 10 min. followed by a 45-min. incubation with an avidin-biotin complex-alkaline phosphatase reagent (ABC-AP). Rat intestine was used as a positive control and mouse lung sections incubated with isotype control mouse IgG were used as a negative control. PCNA-positive cells were visualized with Vector Red chromogen (Vector Laboratories) and the tissue was counterstained using methyl green (Sigma). Finally, the sections were dehydrated and mounted under glass coverslips with VectaMount (Vector Laboratories).To determine the amount of airway smooth muscle by morphometry, airway smooth muscle was detected by immunostaining for smooth muscle α-actin. The lung sections were prepared as described above with the exception of high temperature antigen unmasking, and incubated with monoclonal antibody to smooth muscle α-actin (1A4, 1:1000 dilution; Sigma) for 30 minutes followed by biotinylated anti-mouse IgG antibody and ABC-AP steps as above.PCNA was colocalized with smooth muscle α-actin in order to detect cell proliferation in the airway smooth muscle. Immunohistochemistry for PCNA was done first as described above, and the signal developed with BCIP/NTB chromogen (Vector Laboratories) instead. The sections were then incubated with anti-smooth muscle α-actin antibody (1A4, 1:1000 dilution, Sigma) for 30 min. at 37°C, followed by the biotinylated anti-mouse antibody and ABC-AP steps as above. The smooth muscle α-actin signal was developed with Vector Red, and the tissues counterstained with methyl green.Detection of apoptotic cells in situTo detect apoptotic cells in lung tissue sections we used a TUNEL technique (ApopTag peroxidase detection kit; Intergen, Purchase, NY). The sections were deparaffinized, pretreated with 20 μg/ml proteinase K (Intergen) for 15 min at 37°C, and endogenous peroxidase activity was quenched with 3% hydrogen peroxide for 5 min This was followed by polymerization of digoxigenin-labeled UTP on nicked DNA ends and application of anti-digoxigenin peroxidase conjugate, using ApopTag kit components as per manufacturer's instructions. The signal was developed with DAB chromogen, and the tissues counterstained with methyl green.Quantitative morphology on airway sectionsQuantification of PCNA-positive cells was performed on parahilar lung sections. Cross-sectioned airways, with a major/minor diameter ratio < 2.5, were selected for analysis. The number of PCNA+ cells in the epithelium and sub-epithelial layers were quantified under a light microscope using a 40× objective. The airway basement membrane length was measured by superimposing the image of the airway onto a calibrated digitizing tablet (Jandel Scientific, Chicago, IL), with a microscope equipped with a camera lucida projection system (Leica Microsystems, Richmond Hill, ON, Canada). The numbers of proliferating cells corrected for airway size were expressed as PCNA+ cells/mm of basement membrane perimeter (PBM).Quantification of ASM mass and proliferationASM mass was measured on control, 5 d, and 10 d post-exposure groups by tracing the ASM bundles, as defined by positive staining for smooth muscle α-actin, using a camera lucida and digitizing system. The sum of the ASM bundle areas was calculated for each airway and referenced to PBM2 for airway size correction. To determine if airway smooth muscle cells expressed PCNA, co-localization of PCNA with smooth muscle α-actin was done in a subset of animals. The number of PCNA+ cells in the epithelial and sub-epithelial layers of each airway with a major/minor diameter ratio < 2.5 was quantified and expressed per mm of PBM for epithelium or PBM2 for subepithelial cells.Goblet cell quantificationThe number of goblet cells was assessed on PAS stained tissue sections. A total of 118 airways from 28 animals representing animals from the different exposure times was analyzed and cells were expressed as cell numbers per mm of PBM.Semiquantitative assessment of collagen depositionTo address whether chlorine exposure could affect the development of subepithelial fibrosis, lung sections were stained with Picrosirius red and collagen deposition scored in airways. Scoring by two blinded observers of collagen deposition in airways was performed independently using a scale from 1 to 3. The cumulative score for each mouse was averaged according to treatment group.The quantity of airway smooth muscle (ASM) was quantified by the camera lucida technique. Images of the airways were traced using a microscope side arm attachment and areas of the α-actin positive smooth muscle bundles were digitized using commercial software. The area of ASM was standardized for airway size using the PBM, with the quantity of ASM expressed as ASM/PBM2 (mm2). Morphometric assessments were made on all airways in the tissue section that met the above criterion for its aspect ratio.Methacholine responsivenessIn a separate group of sixty mice, airway responsiveness to methacholine was measured at similar time points after room air or Cl2 exposure (n = 10 at each time point). Animals were sedated with xylazine hydrochloride (10 mg/kg i.p.) and anaesthetized with sodium pentobarbital (40 mg/kg i.p). A flexible, saline-filled cannula (PE-10 tubing) was inserted into the jugular vein for administration of drugs and the trachea was cannulated with a snug-fitting metal cannula. Animals were connected to a computer-controlled small animal ventilator (flexiVent, Scireq, Montreal, PQ, Canada) and paralysed using pancuronium chloride (0.8 mg/kg i.v.). Mice were ventilated in a quasi-sinusoidal fashion with a tidal volume of 0.18 ml at a rate of 150 breaths/min. A positive end-expiratory pressure (PEEP) of 1.5 cmH2O was used. Measurements of pulmonary mechanics were made using a 2.5 Hz sinusoidal forcing function with an amplitude of 0.18 ml. The perturbation was applied after cessation of regular ventilation and expiration by the animal to functional residual capacity. Respiratory system resistance (Rrs) and dynamic elastance (Ers) was derived from the relationship between airway opening pressure, tidal flow and volume After initial baseline measurements of Rrs and Ers, doubling doses of methacholine chloride (Sigma;10 μg/kg to 320 μg/kg i.v.) were administered. Rrs and Ers were measured every 15 seconds after methacholine infusion until peak Rrs was reached. Thirty seconds after peak Rrs was reached, the next highest dose of methacholine was administered. The peak Rrs and Ers at each methacholine dose were used to construct a dose-response curve. After completion of all methacholine doses, animals were euthanized by i.v. pentobarbital overdose. Airway responses were evaluated as the difference between the peak in Ers after 160 μg/kg methacholine and baseline Ers (ΔErs). Changes in Ers rather than Rrs were chosen to represent airway responsiveness because methacholine-induced changes in elastance are affected to a greater degree in mice after Cl2 exposure[9].Statistical analysisOne-way analysis of variance was used to determine the effect of time on the dependent variables except ASM/mm2. The significance of the post-hoc comparisons was determined using Dunnett's test versus control at the p < 0.05 level. The effect of Cl2 on ASM/PBM2 (in mm2) at different times after exposure was tested using the Kolmogorov-Smirnoff test.ResultsHistological and immunohistochemical evaluation of airwaysNormal airway structure and basal levels of proliferation and apoptosis in airway epithelium are shown in Figures 1A, 2A, 3A. Histological examination from samples obtained 12 hrs after exposure showed severe injury to the bronchial epithelium with extensive detachment of the epithelium from the basement membrane and complete denudation of the epithelium in some airways (Figure 1B). Cell cycle was inhibited at this time point after chlorine exposure, as indicated by the virtual absence of positive staining for PCNA (Figure 2B). The TUNEL technique produced cytoplasmic staining of the injured epithelium, but not a signal conforming to usual histopathological criteria for the identification of apoptosis, suggesting that a mechanism other than apoptosis accounts for the rapid and massive epithelial disaggregation following Cl2 gas exposure (Figure 3B). At 24 hrs after Cl2 exposure, most of the detached airway epithelial cells were cleared and airway epithelial cell proliferation was re-established (Figure 3C). In this phase, some clusters of basal cells undergoing apoptosis alternated with proliferating cells, overlying a preserved basement membrane (Figure 3D). Epithelial regeneration was evident at 48 hrs with flattened cells with elongated nuclei lining the basement membrane and an increased frequency of PCNA positive cells. Co-localisation of PCNA and smooth muscle α-actin provided evidence of airway smooth muscle proliferation (Figure 2F). Five days following chlorine exposure, the airway epithelium was evenly re-populated with cells showing an intense proliferative activity, and the frequency of apoptotic cells was similar to baseline levels. Ten days after chlorine exposure, the epithelium was reconstituted and the airway wall was thickened (1 D). Cl2 exposure did not induce goblet cell metaplasia as determined by PAS staining at any time point (data not shown). Only 4 of 118 airways analyzed from 28 mice, sampled at all time points showed any PAS positive cells and these were very infrequent.Figure 1Effects of Cl2 exposure on lung histology. A: Normal mouse lung showing a large airway in cross section, an accompanying artery and two terminal bronchioles (Tb) that open into their respective alveolar ducts. B: Lung histology 12 h after a single 800 ppm Cl2 exposure. Partial or complete detachment of airway epithelium, as seen in this example, occurred in all airways. C: 10 d post-exposure, the epithelium is reconstituted and the airway wall is thickened. D: 10 d post-exposure, high magnification detail showing fully reconstituted airway epithelium. Stain: H&E. Scale bars: 100 μm in A-C; 25 μm in D.Figure 2Effect of Cl2 exposure on cell proliferation as detected by PCNA immunostaining. A: Control mouse airway, showing baseline airway epithelial cell proliferation. PCNA positive cells are indicated by open arrowheads. B: 12 h post-exposure. There is an absence of PCNA positive events, suggesting inhibition of cell cycle. C and D: 24 h post-exposure. Proliferation of airway epithelial cells (C) is re-established. Endothelial cell proliferation (En) is also observed at this time point (D). E: 48 h post-exposure. An increase in PCNA positive epithelial cells is observed. F: Co-localisation of smooth muscle α-actin (red cytoplasmic signal) and PCNA (dark-violet nuclear signal), 48 h post-exposure. PCNA positive cells can be seen in the airway epithelium, smooth muscle layer, and adventitia. The inset shows an example of a PCNA positive airway myocyte at high magnification. G: 5 d post-exposure. The airway epithelium is evenly re-populated with cells undergoing intense proliferative activity. Scale bars: 50 μm (25 μ in F inset). Pn: Pneumocytes; SM: Smooth muscle.Figure 3Effect of Cl2 exposure on airway cell apoptosis; TUNEL technique. A: Control mouse airway, showing baseline airway epithelial cell apoptosis (arrowheads). B: 12 h post-exposure. Cytoplasmic TUNEL signal in damaged epithelium. The high magnification inset details the cytoplasmic localisation of the TUNEL stain on cells with methyl green counterstained nuclei. These cells lack a TUNEL signal attributable to apoptosis-related DNA fragmentation. The arrowheads indicate examples of cells that appear truly apoptotic. C: 24 h post-exposure. Some clusters of basal cells undergoing apoptosis are visible. Inset shows high magnification detail. D: 5 d post-exposure. The frequency of TUNEL positive cells at 5 d is back to baseline level. Scale bars: 100 μm in I; 50 μm in A, B, C inset and D.Cl2 exposure did affect the quantity of ASM as determined by morphometry (Figure 4). 10 days after Cl2 exposure, a shift was observed in the distribution of airways with small amounts of ASM. For example, the proportion of airways with values of ASM area > 0.0015 (ASM/mm2 of BM) was approximately 50% for control animals, but < 10% for the 10 day post-exposure group.Figure 4Cumulative distribution of airway smooth muscle mass per mm2 of basement membrane (ASM/mm2 of PBM). The values plotted are individual airway measurements. 2–8 airways were quantified per animal. The distribution of the 10 day group was significantly different from both the control and 5 day groups (p < 0.05). n = 38, 40, and 31 for control, 5 days, and 10 days.Quantification of PCNAThe number of PCNA+ cells in the airway epithelium and sub-epithelium is shown in Figure 5. A baseline frequency of epithelial and sub-epithelial proliferation was detectable in control animals. Twelve hours after Cl2 exposure, epithelial PCNA expression tended to be lower than control values although the difference did not reach statistical significance. Epithelial PCNA expression was significantly elevated by 48 hrs after chlorine exposure, increasing approximately 14-fold from control levels (p < 0.05) and over 30-fold by 5 d post-exposure (p < 0.05). Although the majority of the PCNA+ cells in the airways were epithelial cells, a significant amount of sub-epithelial PCNA expression was also observed after Cl2 exposure. Subepithelial PCNA expression was significantly elevated at 5 d post-exposure. By 10 d post-exposure, both epithelial and subepithelial PCNA immunoreactivity had returned to control levels. No significant correlation was found between airway size (as determined by basement membrane length) and PCNA index at any of the time points.Figure 5Time course of PCNA expression in the epithelium (A) and subepithelium (B) of airways in mice exposed to air (control) or Cl2 gas. Data is expressed as PCNA-positive cells/mm basement membrane. The number of airways evaluated at each time point ranged from 25 to 57. Values are means ± S.E. *significantly different from control (p < 0.05).Determination of airway fibrosisAssessment of collagen deposition using Picrosirius red staining demonstrated a significant increase in collagen in the airways 10 days following chlorine exposure (Figure 6). There was no significant difference in the amount of collagen at 24 hours or 5 days. Twenty nine animals were analyzed and assessed by two observers independently.Figure 6Illustrative photomicrograph showing collagen in the airway walls by Picrosirius red staining (two left panels). Quantitative analysis of degree of staining by semi-quantitative scoring at different time points after Cl2 gas exposure.Bronchoalveolar lavageThe recovery of BALF averaged 90% and did not differ significantly among groups. Total cell counts were significantly elevated at 5 d and remained elevated at 10 d post-exposure relative to controls (Table 1). Differential cell counts showed no significant change in eosinophils or lymphocytes after Cl2 exposure (Figure 7), but neutrophils were significantly elevated relative to controls at 5 d post-exposure (0.02 ± 0.01 (SE) × 104 cells in controls, 4.76 ± 1.94 at 5 d post-exposure; p < 0.05) and macrophages were significantly elevated at both 5 d and 10 d post-exposure (12.0 ± 1.9 × 104 in controls, 32.2 ± 7.7 at 5 d, 33.7 ± 3.3 at 10 d, p < 0.05 versus controls). Dead cells in the BALF, identified by trypan blue, were markedly elevated from 12 hrs to 48 hrs post-exposure (Table 1); these cells were almost exclusively comprised of epithelial cells, identified by their cuboidal shape and cilia. Similarly, the number of epithelial cells counted during differential cell counting of cytospin slides was markedly elevated at 12 and 24 hr (p < 0.05) but had returned to control levels by 48 hr (Figure 7). The amount of total protein in BALF supernatant, a marker of airway microvascular permeability and epithelial damage, was significantly elevated 12 hrs after chlorine exposure, and remained elevated up to 5 d post-exposure (Table 1).Table 1Time course of protein, live and dead cell counts in BALF after Cl2 exposure.Control12 hr24 hr48 hr5 d10 dLive cells (×104/ml BALF)12.3 ± 1.99.4 ± 1.914.0 ± 1.133.1 ± 6.338.0 ± 9.8*36.9 ± 3.4*Dead cells (×104/ml BALF)1.3 ± 0.592.6 ± 11.2*106.2 ± 9.7*54.1 ± 17.0*5.8 ± 0.73.5 ± 0.6Protein (g/ml)69.4 ± 8.1612.8 ± 178.6*391.7 ± 102.2*251.5 ± 29.5*221.0 ± 42.7*116.7 ± 6.3Values are means ± S.E. * p < 0.05 versus controlFigure 7Time course of BALF differential cell counts after a single Cl2 gas exposure. At each time point, n = 8. Values are means ± S.E. * significantly different from control (p < 0.05).Airway mechanics and responsiveness to methacholineCl2 exposure altered respiratory mechanics as reflected by changes in baseline Ers and Rrs. The initial response to Cl2 exposure was an elevation of Ers and Rrs, which persisted up to 48 hrs post-exposure (Ers = 51.1 ± 3.09 cmH2O/ml in control mice vs. 70.9 ± 3.23, 67.5 ± 2.16, and 61.5 ± 1.67 cmH2O/ml at 12, 24, and 48 hrs post-exposure respectively, p < 0.05; Rrs = 0.98 ± 0.05 cmH2O/ml/sec in control mice vs. 1.32 ± 0.06 and 1.23 ± 0.05 cmH2O/ml/sec at 12 and 24 hrs post-exposure respectively, p < 0.05) (Figure 8). Airway mechanics returned to baseline levels by 5 d, but at 10 d post-exposure, Ers levels fell significantly below control levels (Ers = 51.1 ± 3.09 cmH2O/ml in control mice vs. 40.7 ± 0.97 cmH2O/ml at 10 d post-exposure, p < 0.05). Airway responsiveness to methacholine, as determined by ΔErs, increased after Cl2 exposure compared to control, and was significantly higher at 12 hrs and 5 d post exposure (ΔErs = 100 ± 19.7 in control mice vs. 257 ± 45.3 and 269 ± 34.0 at 12 hrs and 5 d post-exposure respectively, p < 0.05) (Figure 9). ΔRrs was not significantly altered at any time point after Cl2 exposure, although a trend for ΔRrs to be lower 24 hrs after Cl2 exposure was observed (p = 0.055).Figure 8Time course of baseline respiratory elastance (Ers) and resistance (Rrs) in mice exposed to Cl2 gas. Ers and Rrs were measured using a 2.5 Hz sine-wave perturbation with an amplitude of 0.18 ml. At each time point, n = 10. Values are means ± S.E. * significantly different from control (p < 0.05).Figure 9Time course of airway responsiveness of elastance (Ers) and resistance (Rrs) to methacholine in mice exposed to Cl2 gas. Responsiveness is expressed as the peak Ers or Rrs after administration of 160 μg/kg methacholine minus baseline Ers or Rrs. Values are means ± S.E. * significantly different from control (p < 0.05).DiscussionThis study describes the time course of airway epithelial damage and repair in A/J mice following a single exposure to a high concentration of Cl2 gas. Cl2 exposure resulted in marked damage to the airways, as indicated by epithelial cell sloughing, increased protein in BALF, an inflammatory response with neutrophil and macrophage recruitment into the airways, and altered lung mechanics. Subsequent airway repair was characterized by increased epithelial and subepithelial cell proliferation, complete restoration of the epithelial layer, increases in the quantity of ASM and modest airway hyperresponsiveness. There was also evidence of airway fibrosis at 10 days after the Cl2 exposure.A pronounced feature of the acute injury phase after Cl2 exposure was extensive and synchronous loss of airway epithelial cells. Programmed cell death was not likely the mechanism of the generalized loss of epithelial cells, since the TUNEL technique did not produce a nuclear signal consistent with apoptosis. The explanation for the diffuse cytoplasmic staining observed in the detached epithelium is not clear but may have been caused by the highly reactive chlorine molecules. As opposed to apoptosis, disruption of the intercellular junctions and the attachments of the epithelial cells to the basement membrane by the Cl2 gas may have been the mechanism responsible for detachment of the epithelium. Other oxidants such as hypochlorous acid (HOCl) and ozone can disrupt cell adhesion via damage to extracellular matrix proteins and β-1 integrins[12,13], thus Cl2 gas may act via similar mechanisms.Acute loss of epithelial barrier function resulted from the extensive sloughing of the airway epithelium, as reflected by the increased protein concentration in BALF. Changes in baseline respiratory mechanics (resistance and dynamic elastance) paralleled the time course of BALF protein concentration with the most pronounced alterations occurring 12 hrs after exposure followed by resolution of these changes over the 10 d study period. Pulmonary edema and alveolar flooding may have contributed to the acute decreases in lung elastance in this model, as has been demonstrated in other species after Cl2 gas exposure[14,15]. However, heterogeneous airway narrowing may have also contributed.Exposure to chlorine gas exposure had a direct toxic effect on airway epithelium as severe airway damage was observed at early time points in the absence of an inflammatory response. When inhaled, chlorine gas combines with water to form hydrochloric and hypochlorous acids (Cl2 + H2O → HCl + HOCl). HOCl is unstable and breaks down into HCl and free oxygen. Oxidant injury due to this nascent oxygen is thought to be the primary mechanism of cytotoxicity, with the acid production being secondary. In a similar study from our laboratory, positive staining for 3-nitrotyrosine residues, a marker of oxidative stress, was observed in mouse airways 24 hrs after exposure to 800 ppm Cl2 gas, supporting oxidative injury as a mechanism in this model[9].A modest neutrophil and macrophage inflammation did subsequently develop after Cl2 exposure and the inflammatory cells themselves could also have contributed to airway damage. Activated neutrophils can produce reactive oxygen species and myeloperoxidase, a neutrophil-specific enzyme that catalyses the formation of hypochlorous acid/hypochloride (HOCl/OCl-) from hydrogen peroxide. Neutrophils can also release proteolytic enzymes such as collagenase and elastase which could also contribute to the airway damage.Following the acute airway injury induced by Cl2 gas exposure, tissue repair and restoration of the barrier function of the epithelium occurred. One mechanism by which an epithelial layer can be repaired is by migration of healthy epithelial cells from an area adjacent to the damaged epithelium. Studies of mechanical de-epithelialisation in vivo demonstrate that this is a quickly occurring process, with initial migration of adjacent epithelium to the wound site occurring within 8–15 hrs[16]. The relevance of migration as a mechanism, however, is questionable in cases of near to complete denudation of the epithelium, as was observed in many airways in this study. In this instance, growth and differentiation of local progenitor cells is another mechanism by which the epithelial layer can be repopulated. In the trachea and bronchi, basal cells constitute a separate layer of cells attached to the airway basement membrane. In response to epithelial injury, these cells can turn into a highly proliferative cell phenotype and can become flattened and cover the basement membrane[17]. In smaller bronchioles, Clara cells likely play the role of progenitor cell after injury[18] Intriguing new evidence suggests a possible role for circulating bone marrow stem cells in bronchiolar repopulation after injury[19]. Ortiz et al. [19] have demonstrated that murine mesenchymal stem cells are able to home to the lung after injury and adopt an epithelium-like phenotype. It is uncertain at this time as to which specific cell population may have acted as progenitor cells for the airway epithelium in this study.The time course of epithelial repair after Cl2 gas exposure was assessed by quantifying the amount of cellular proliferation occurring in the airway. Increased levels of PCNA immunoreactivity were detectable by 48 hrs and maximal proliferative activity in the airways occurred 5 d post-exposure. Compared to other studies reporting dynamics of epithelial repair after acute airway injury, the recovery of murine airways from Cl2 damage was relatively prolonged. In rats, peak cell proliferation occurred 26 to 36 hrs after mechanical injury of tracheal epithelium[20,21] and at 24 to 48 hrs after acute ozone exposure[22]. In mice, epithelial cell proliferation after desquamation of airway epithelium by naphthalene treatment was maximal 2 to 7 days post-treatment depending on mouse strain[23]. The time course of epithelial repair after damage is likely related to the severity of injury, and therefore is difficult to compare among these different models.Increased cellular proliferation after Cl2 exposure was not limited to the airway epithelium as significant PCNA immunoreactivity was also observed in the sub-epithelial layer of airways. Using immunohistochemical co-localization, we provide evidence of airway smooth muscle cell proliferation. This finding, together with the quantification of ASM mass, suggests that chlorine exposure in this model results in ASM hyperplasia. This is in agreement with the study of Demnati et al [8] who reported an increase, albeit transient, in ASM quantity in rats after acute exposure to Cl2 gas.The signals involved in repair and in the repopulation of the epithelium after Cl2-induced injury are unclear. Epidermal growth factor (EGF)-dependent mechanisms may be important as mediators such as epidermal growth factor (EGF) and TGF-α can bind to EGF receptors located on both basal cells and epithelial cells and stimulate cell migration, proliferation and differentiation[24]. The absence of goblet cells is however somewhat surprising if indeed EGF receptor ligands are important in repair as stimulation of the EGF receptor has been repeatedly demonstrated to cause goblet cell differentiation in the airways[25]. EGF-independent factors may also be important. Neutrophils, for example, may contribute to signalling of repair processes as neutrophil defensins, antimicrobial peptides present in the neutrophil, may also stimulate proliferation[26]. Interestingly, the maximal proliferative activity of the airway epithelium at 5 d corresponded to the time of maximal neutrophil influx in the BALF.Restoration of the airway epithelial layer, as assessed histologically, was complete by 10 days after Cl2 exposure. However, not all variables had returned to control levels after 10 days; inflammatory cells in the BALF were still elevated and baseline elastance was lower than control levels. Therefore complete resolution of the Cl2-induced damage may not have occurred in the timeframe of this study. Also the timeframe of this study may not have been long enough to fully evaluate remodelling processes. As we only detected changes in ASM quantity at our latest time point, 10 days after exposure, the possibility remains that further remodelling may take place at even later time points. There was also an increase in collagen deposition in the airway wall at this same time point. The epithelium is a source of fibrogenic cytokines[27] and it is potentially the cause of the collagen deposition. Although the changes were not significant there appeared to be a trend for a reduction in airway smooth muscle mass at 5 days after Cl2 exposure, suggesting that damage may have penetrated beyond the epithelium to the ASM layer.Persistent airway hyperresponsiveness occurs in a small percentage of people after acute Cl2 gas exposure[28]. In this study, mice receiving a single exposure to a high concentration of Cl2 gas did display modest increase in dynamic elastance in response to methacholine but it was transient in nature. That responsiveness of pulmonary dynamic elastance to methacholine was affected to a greater degree by Cl2 gas exposure than was responsiveness of pulmonary resistance is consistent with results from a previous study[9]. This suggests that changes in responsiveness to methacholine after Cl2 gas exposure in mice may be dominated by abnormalities in the peripheral lung, as opposed to central airways. Perhaps also the trend for a reduction in responsiveness to methacholine may reflect injury to the airway smooth muscle from the high levels of Cl2 used for exposure.In conclusion, this study describes the time-course of injury and repair after an acute exposure of mice to a high concentration of Cl2 gas. Severe epithelial injury was induced quickly after exposure with loss of the epithelial barrier function and acute alterations in respiratory mechanics. Epithelial repair processes were apparent by 24 hrs and restoration of the epithelium was complete by 10 d. Recovery from the Cl2-induced damage was associated with modest airway hyperresponsiveness and alterations in airway smooth muscle mass. Whether comparable airway remodelling is associated with lesser degrees of repeated exposures remains to be explored.Competing interestsThe authors declare that they have no competing interests.Authors' contributionsST was involved in the design and performance of the experiments and wrote the manuscript. DRB was responsible for the planning and oversight of all immunohistochemistry and contributed to the manuscript. HC assisted in the performance of measurements of airway responsiveness and tissue harvesting. TM performed histochemical staining for goblet cells and collagen and performed quantification of same. HKQ assisted in the analysis of histochemical images for goblet cells and collagen and assisted in editing the manuscript. JGM was responsible for the questions being tested and for the design of the experiments. He reviewed all phases of analysis and finalized the writing of the manuscript. All of the authors have read and approved the manuscript.\n\nREFERENCES:\nNo References"
+}

batch_5/PMC2531108.json

@@ -0,0 +1,4 @@
+{
+    "id": "PMC2531108",
+    "text": "This is an academic paper. This paper has corpus identifier PMC2531108\nAUTHORS: Sean A Martin, Matthew T Haren, Anne W Taylor, Sue M Middleton, Gary A Wittert\n\nABSTRACT:\nBackgroundAn increasing proportion of Australia's chronic disease burden is carried by the ageing male. The aim of this study was to determine the prevalence of asthma, cancer, diabetes, angina and musculoskeletal conditions and their relationship to behavioural and socio-demographic factors in a cohort of Australian men.MethodsSelf-reports of disease status were obtained from baseline clinic visits (August 2002 – July 2003 & July 2004 – May 2005) from 1195 randomly selected men, aged 35–80 years and living in the north-west regions of Adelaide. Initially, relative risks were assessed by regression against selected variables for each outcome. Where age-independent associations were observed with the relevant chronic disease, independent variables were fitted to customized multiadjusted models.ResultsThe prevalence of all conditions was moderately higher in comparison to national data for age-matched men. In particular, there was an unusually high rate of men with cancer. Multiadjusted analyses revealed age as a predictor of chronic conditions (type 2 diabetes mellitus, angina, cancer & osteoarthritis). A number of socio-demographic factors, independent of age, were associated with chronic disease, including: low income status (diabetes), separation/divorce (asthma), unemployment (cancer), high waist circumference (diabetes), elevated cholesterol (angina) and a family history of obesity (angina).ConclusionSocio-demographic factors interact to determine disease status in this broadly representative group of Australian men. In addition to obesity and a positive personal and family history of disease, men who are socially disadvantaged (low income, unemployed, separated) should be specifically targeted by public health initiatives.\n\nBODY:\nBackgroundAlmost four out of five Australians have at least one long-term health condition [1]. National expenditure on chronic disease and associated care now accounts for over two thirds of the entire health care budget [2]. Many of these diseases are preventable through the modification of the risk factors that contribute to their development [3]. In Australia, the National Chronic Disease Strategy (NCDS) has driven a renewed focus on the determinants and settings that promote the development of chronic conditions, particularly those that relate to the designated National Priority Areas of asthma, cancer, cardiovascular disease, diabetes and musculoskeletal conditions [4]. Much of this focus has centred on eliminating many of the health inequalities that persist in our population. To date however, the disproportionate disease burden carried by men, particularly the ageing male, has received little attention.In Australia, as elsewhere, men display poorer health outcomes when compared with women [5,6]. The problems of male health in Australia reflect those increasingly noted in other developed nations. Men in Australia have a lower life expectancy than women (76.6 years compared with 82.0 years) with higher rates of mortality at all ages, a discrepancy that begins from birth [7]. There is a disproportionate level of chronic physical and psychological disease in Australian men [7] and higher rates of illness-related disability [1]. In addition, men display a higher prevalence of the major risk factors – smoking, lack of physical activity, poor nutrition and alcohol abuse – linked to the development of most major chronic diseases [3,8]. When combined with a reduced likelihood of adopting a healthy lifestyle [9] and a resistance to public health messages [10], the health and behaviour of ageing males in Australia should be an urgent concern.Of the disease groups targeted by the NCDS, only musculoskeletal conditions have a lower prevalence in men at all ages as compared to Australian women [1] (with current indications that the incidence of rheumatoid- and osteo-arthritis in men aged over 60 is increasing faster than that of age-matched females [11]). Cardiovascular disease (CVD), the leading cause of death and disability in Australia, strikes more men than women across the entire age spectrum, with death rates among males aged 25–74 years two to three times that of females [12]. The rates of asthma in Australia are amongst the highest in the world and whilst the prevalence is greater in males than females during childhood, this reverses through the adult years. Men, however, aged 65 years and over in Australia are more likely to suffer disability and death as a result of their asthma [13]. Three times as many men as women over the age of 65 report some type of cancer [3], including a higher prevalence of those sex-independent cancers nominated as national priorities (colorectal cancer, lung cancer, melanoma, non-melanoma skin cancers and non-Hodgkin's lymphoma [1,14]). Men also present a significant challenge in the global diabetes epidemic, showing a higher prevalence of diabetes overall and an increased rate of mortality and complications arising from the condition [3,15].Despite all these disparities, the health of men and the related changes in biological, psychological and social settings through ageing remains one of the most understudied areas of health research in Australia. Recognition of this is feeding a groundswell of support for men's health issues and policy initiatives in Australia, including a call by a number of peak and government bodies for a comprehensive men's health longitudinal study [16-18]. The Florey Adelaide Male Ageing Study (FAMAS) is a multi-disciplinary population cohort study of 1195 men, aged 35–80 years at recruitment and living in the north-west regions of Adelaide, Australia. We report here, in an analysis of the baseline cross-sectional data of the men in the cohort, the relationships between biological, social and demographic factors and the presence a number of chronic conditions considered to be of national priority.MethodsThe Florey Adelaide Male Ageing Study (FAMAS)Details of the FAMAS design, procedures and participants have been published elsewhere [19]. Briefly, subjects were recruited at random from the Electronic White Pages (EWP) using six digits of the standard eight digit telephone number in addition to prefixes and exchanges for the North Western Suburbs of Adelaide. Randomly selected households were sent an introductory letter and brochure. Approximately two weeks later a call was made to the household and the male person aged between 35 and 80 years to last have his birthday was invited for a Computer-Assisted Telephone Interview (CATI) lasting no more than 15 minutes, together with an invitation to participate in the study. A series of questions relating to age, other socio-demographic variables, history of disease and presence of risk factors enabled comparison of responders and non-responders. Of those eligible to participate, 70.7% agreed to be interviewed (Participation Rate) and 45.1% ultimately attended a clinic (Final Response Rate). Non-responders were more likely to live alone, be current smokers, and had a higher prevalence of self-reported diabetes and stroke and lower prevalence of hypercholesterolemia. A comparison with 2001 Census data showed that participants matched the local and national population for most key demographics, but younger age groups and never married men under-represented and married men and elderly participants were over-represented [19].Recruitment of participants occurred during two phases-from August 2002 to July 2003 and from June 2004 to May 2005 and complete clinical follow-up is scheduled to reoccur every five years from baseline throughout the life of the cohort. Follow-up questionnaires are designed by a multi-disciplinary investigative team using standard measures where available and are completed annually.All protocols and procedures were approved by the Royal Adelaide Hospital Research Ethics committee and, where appropriate, the Aboriginal Health Research Ethics Committee of South Australia.Demographic and lifestyle measuresDemographic and lifestyle information were obtained by self-report questionnaires [19]. Socio-economic status was assessed using the relative advantage/disadvantage index from the Socio-Economic Indexes for Areas (SEIFA) by the Australian Bureau of Statistics. Higher scores indicate community-dwelling areas with a relatively high proportion of people with high incomes or a skilled workforce, as well as a relatively low proportion of people with low incomes and relatively few unskilled people in the workforce. Medical conditions were assessed by the following question item, 'Have you ever been told by a doctor that you have any of the following conditions?' Smoking status was determined using question items from recent Australian National Health Surveys [20]. Leisure-time physical activity was determined using items from the National Physical Activity Survey 1999 [21]. The semi-quantitative food frequency questionnaire (FFQ) developed by the Cancer Council of Victoria was used to estimate usual energy-providing macronutrient intakes [22].Anthropometric measuresAnthropometry was performed using standard protocols [23] in the morning, prior to breaking an overnight fast with participants barefoot and in light clothing. Height was measured using a wall-mounted stadiometer (Seca Model No. 220, Humberg, Germany). Body weight was obtained using digital platform scales (Wedderburn UW OFWB, Taiwan, CN) and maintained by on-site engineering services. Waist circumference was assessed using a fiberglass tape measure (Gulik II, Country Technology, Wisconsin, USA). Waist circumference was measured in triplicate, taken at the level of the narrowest point (or midway) between the lower costal border and the top of the iliac crest and read in the mid-axillary line, and the mean of the three measurements was used in analyses. Coefficients of variation for triplicate waist circumference measurements were less than 2.2% for 99.7% of the sample. Standard BMI cut-offs were used to classify participants in normal/underweight (<25 kg/m2), overweight (25–29.9 kg/m2), and obesity (≥ 30 kg/m2) categories.Hormone measuresBlood samples were drawn between 8 and 11 am after a 12-hour overnight fast. Fasting plasma glucose and lipids (triglyceride, total cholesterol, HDL, LDL) were measured on auto-analysers in the Diagnostic Services laboratory (IMVS) on a 24-hour basis. Determination of serum lipids was done enzymatically using a Hitachi 911 (Boehringer, Germany). The inter-assay CV's for the measurement of serum lipids are as follows; triglyceride 3%, total cholesterol 2.3%, HDL 6.7% and LDL 3.7%. Glucose was determined using an automated chemistry analyser system (Olympus AU5400, Olympus Optical Co Ltd. Japan). The inter-assay CV's for this assay are 2.5% at 3.5 mmol/L and 3.0% at 19.6 mmol/L. Glycated haemoglobin (HbA1c) was measured by high-pressure liquid chromatography (HPLC) using a spherical cation exchange gel (CV 2% at 6% of total haemoglobin).Chronic disease modellingThe conditions examined were based on those National Health Priority Areas covered in the recent National Chronic Disease Strategy [4], namely: cardiovascular disease, asthma, cancer, T2DM and musculoskeletal conditions. In the case of cardiovascular disease, angina was specifically selected for this model (as per [1]), given the significant burden of this condition in ageing men. The use of the term 'angina' refers to the presence (current or past) of any chronic stable angina or unstable angina as diagnosed by physician. Osteoarthritis and rheumatoid arthritis were chosen as representative of musculoskeletal conditions in this cohort, given the relatively low proportion of men with osteoporosis. Exposure variables were selected from established or suspected associations with the relevant outcome. Descriptive and tabular analysis was initially conducted to examine data distributions and observe patterns in exposure-outcome relationships.Relative risks and 95% confidence intervals were estimated by binomial regression with the log link function using the GENMOD procedure in SAS (SAS Institute, Cary, NC). Firstly, the relative risk of each potential confounder on the outcome was examined to determine individual main effects. Next, bivariate regression models were fitted with the exposure variable and age to get age-adjusted relative risks. Age was then checked as a modifier for each of the exposure variables by fitting exposure, age and the age-exposure interaction. Age was considered a modifier for an exposure variable if the interaction had a p-value < 0.005 (to account for the multiple testing).Those variables whose age adjusted relative risks had a p value < 0.25 were included in the final multivariate model [25]. This initial higher p value was selected to account for the confounding effect of age in the model (i.e. inclusion of age in any model of this type would decrease the overall probability of other variables reaching significance).For all conditions, the socio-demographic variables investigated were: age, income, employment status, pension status, current smoker/ever smoked, physical activity, SEIFA (Index of Disadvantage), body mass index (BMI), and waist circumference. The choice of family history and co-morbidity variables varied according to outcome.ResultsAnginaOf the men examined, 6.5% (n = 78) reported to having been diagnosed with angina by a physician. The majority of cases were detected in men over 55 years (n = 60). Age was found to strongly associate with the condition, peaking in 55–64 year old men (Additional file 1). There was a reduced risk observed for men with higher gross household incomes (most notably in the $80, 000+ category), although this relationship was largely lost when controlled for age (Additional file 1). Widowed men were found to have an increased risk of angina when compared with men currently married or living with a partner, again this effect was not seen in the age-standardized model (Additional file 1). Not being in the workforce strongly increased the risk of developing angina in both models and those currently on a pension displayed an increased risk of angina (Additional file 1). When factored for the age of participants, there was a notable increase in the risk of angina for those with a family history of obesity and hypertension (although the latter was fractionally outside of the nominal significance range). There was a strong interaction observed between the presence of angina and other conditions. Men that had either diabetes, hypercholesterolaemia or hypertension all displayed an increased risk of also developing angina. With the exception of elevated cholesterol, this observation held when responses were age-adjusted (Additional file 1).When all qualifying variables were included in the final model, age proved an extremely robust determinant of angina. When compared with men from the youngest bracket, the risk of developing angina near doubled in successive age groups (55–64 years: RR = 9.75 (3.35, 28.37); 65–80 years: RR = 18.79 (6.44, 54.79)). Having a family history of obesity also increased the likelihood of developing angina (RR = 1.84 (1.12, 3.01)). Again, a strong risk was observed if men were also hypercholesterolemic (RR = 2.97 (1.44, 6.10)) (Additional file 3).AsthmaThe prevalence of men diagnosed with asthma in the cohort was 9.5% (n = 113). The majority of respondents (51.4%) with asthma were in the youngest age group (35–44 years).A binomial regression model was applied to all selected exposure variables (Additional file. 1). When controlled for age, only waist circumference was found to associate with asthma. Being born in regions other than Australia or New Zealand, and being separated/divorced, were the only other factors nearing significance. All age-adjusted exposure variables with a p-value lower than 0.25, were included in the final model, in which the prevalence of asthma was associated with larger waist circumference (RR = 1.01 (1.00, 1.03)) and a history of separation/divorce (RR = 1.75 (1.18, 2.60)) (Fig. 3).CancerWithin the cohort, 10.3% of men (n = 123) reported having some form of cancer. A variety of factors was shown to associate with cancer status following the application of regression models (Additional file 1). In the unadjusted binomial model, the strongest association was observed for age, with older age groups demonstrating increased cancer prevalence. Also, men not in the work force, currently receiving some form of pension and with hypertension were all found to have increased risk of all-types of cancer. When controlled for age, none of these associations held significance, with only men born outside of Australia/New Zealand showing a trend towards a lowered cancer risk (p = .03). Men from the highest SEIFA quartile were at increased risk of being diagnosed with some type of cancer, an observation that held after adjustment for age. In the converged model (Additional file 3), an increasing age (55–64 years: RR = 2.35 (1.24, 4.44); 65–80 years: RR = 3.88 (1.93, 7.79), and being unemployed (RR = 3.47 (1.29, 9.36), p = .014) were associated with cancer risk.Type 2 Diabetes Mellitus (T2DM)Within the cohort, 15.6% of men (n = 187) either self-reported being diagnosed with T2DM or were diabetic by established clinical indicators (see Methods). There were numerous associations with diabetic status. Age showed a clear positive relationship with T2DM, with higher age groups showing elevated risk of disease. Income levels showed a particularly strong linear relationship with risk of T2DM, with lower risks observed for higher incomes in both the full and age-adjusted models. Being widowed was also shown to have a strong relationship with having T2DM in both models. Men who were either not in the work force or not on the pension displayed an increased risk of diabetes (although this was not significant for pensioners when adjusted for age). Current smokers tended to be less likely to be diabetic (although this was not observed in the age-adjusted model), whereas past smokers had an increased likelihood of diabetes. Obese participants showed an increased probability of T2DM, when compared to participants in normal weight ranges. In addition, participants with T2DM were also found to have larger waist circumferences. Men in the cohort who reported a family history of diabetes, showed a robust association with current T2DM status. As well, when controlled for age, those with a family history of obesity were moderately more likely to have T2DM. Finally, those diagnosed as hypertensive, were also at increased risk of T2DM (Additional file 2).When all of the eligible effectors were entered in to the final model, there were numerous determinants of T2DM detected. There was a slight influence of age on diabetes observed in the cohort, with men aged above 55 tending towards an increased risk for the condition when compared with their younger counterparts. The overall trend of men with higher incomes showing a reduced possibility of T2DM neared significance. Widowed men were at greater risk of developing diabetes when compared with married/partnered men (RR = 1.91 (1.18, 3.08)). The waist circumference relationship observed initially was also conserved in the final model, with men with larger waists again showing an elevated risk of diabetes (RR = 1.02 (1.01, 1.04)). A family history of diabetes was again shown to strongly increase risk of having the condition (RR = 1.64 (1.21, 2.24)).Osteoarthritis & Rheumatoid ArthritisThe prevalence of the musculoskeletal conditions under consideration were 9.7% for osteoarthritis (n = 118) and 5.0% for rheumatoid arthritis (n = 60). In both cases, almost four out of five cases occurred in men over 55 years (77.6% for osteoarthritis and 76.7% for rheumatoid arthritis). There is a noted impact of age on the association with the selected exposure variables and these musculoskeletal conditions. When age was accounted for in the binomial models, there was a reduced risk for osteoarthritis in men born overseas and those with lower waist circumferences (Additional file 2).The predictive model of osteoarthritis (Additional file 3) confirms age as a determinant of this condition, with the eldest age group showing a significantly increased risk (RR = 2.98 (1.49, 5.93)) (and a strong trend in the 55–64 age group). Of note, being obese (whilst showing a weak association in the binomial models) had a strong association with osteoarthritis in the final model (RR = 3.78 (1.35, 10.58)) (Additional file 3).In the case of rheumatoid arthritis, none of the selected factors proved to be significant, although there was a slight tendency for such men to be unemployed (p = .028).DiscussionThe prevalence of the chronic diseases examined in this study was generally similar to available estimates from national and local data sources. The higher cancer prevalence in comparison to age-matched men nationally was the noticeable exception, and this has previously been demonstrated in two other studies of the sampling area [24,25].There was a clear effect of age on the conditions examined (with the possible exception of asthma) as has been shown previously in both men and women [26-28], however data from this study suggest that a number of social and demographic factors also contribute significantly to disease risk for each of the conditions studied, over and above age.AnginaThe higher rate of mortality from cardiovascular disease is one of the major drivers of the life-expectancy gap between men and women. In this cohort, angina was reported by 6.5% of all participants, a similar prevalence to a previous study undertaken in the same region at approximately the same time (5.7% of participants aged 35+) [30], but higher than that reported in the 2004–5 NHS (4.2% of males 35–80) [1]. The high prevalence of obesity, T2DM and other risk factors for cardiovascular disease in this cohort may account for the discrepancy.The effect of age was most noticeable in participants with angina, with men aged 65+ showing a markedly elevated risk of episodes of angina relative to younger men in accordance with observations of other population based studies [31,32].Both hypertensive and diabetic men in the cohort showed an increased age-adjusted risk for angina. Likewise, elevated cholesterol was identified as a predictor of ever having angina in this multiadjusted model. All are well established risk factors for angina in men [1,3]. We also confirmed the independent relationship between obesity and angina. Zdrackovic (2007) in a twin registry study has recently pointed to a stronger genetic influence on the relationship between obesity and angina than previously reported [33]. Accordingly, a combination of family history of not only angina, but also obesity should prompt the assessment of and aggressive management of risk factors.AsthmaThe prevalence of men with asthma in Australia (9.0% of the total population) is amongst the highest in the developed world. In this cohort, 9.4% of men reported having being ever diagnosed with asthma, with over half in the youngest age group. Whilst this prevalence is slightly higher than that observed in comparable studies of the region [29,34] and population surveillance data [1], it does provide limited support to recent observations of rising levels of asthma in ageing men in the face of overall decreases in other sub-groups [35]. It is suggested that this is, at least in part, related to the increase in obesity that has been observed in such men [3,36]. Asthmatic men in our study had higher waist circumferences in all level of analyses, independent of BMI. Such a finding is consistent with other epidemiological studies involving asthmatic men (see [37] for summary).Whilst there have been numerous studies demonstrating a protective effect of marriage on developing asthma, to our knowledge this has been the first cohort study that has specifically shown an age-independent increase in asthma susceptibility for divorced or separated men. In an analysis of a group of middle aged British men, Ebrahim (1995) [38] speculated that the observed increase in asthmatic symptoms amongst recently separated men was most likely a combination of an increase in detrimental behaviour (alcohol consumption, smoking and physical inactivity) and the removal of spousal support. This suggestion is supported by recent data indicating that separated men aged 55+ are least likely to adhere to a written asthma plan [39].The absence of any significant effect of smoking behaviour on asthma prevalence in this study is not in accordance with the common view of smoking being linked to adult-onset asthma. It is possible that smokers with airway disease had been classified by their treating practitioners as having chronic obstructive airways disease, and that those with asthma had been successfully convinced not to smoke. Also, there are a relatively high proportion of current smokers in the cohort who could be termed heavy consumers (i.e. greater than 25/day [40]). It has been suggested that such smokers, through a complex interaction of inherited and environmental factors may not display asthmatic-type symptoms [41].CancerThe rising proportion of elderly men with cancer is leading to an increased investigation into the conditions which promote its development. Numerous demographic and behavioural factors have been found to associate with various types of cancer. A high proportion of men in this cohort had been diagnosed with some form of cancer when compared with age-matched men from local (4.1% of men aged over 35 years) and national surveillance data (3.3% of Australian men over 35 years). Whilst interpretation of this disparity is limited by the comparatively low number of participants in the study, a specialized health atlas of the region has identified that four out of the six highest Standardised Incidence Ratios for cancer are found in this study's catchment area [42]. This is currently the focus of intense investigation by local health authorities [43]. Interestingly we observed that age-matched men from the highest SEIFA quintile were at an increased risk of ever having cancer. In surveillance studies, lower SEIFA quintiles have generally been shown to associate with increased morbidity and mortality [44]. It has been argued that improvements in recent years in cancer screening and awareness may have created an artificially high impression of cancer prevalence amongst affluent American men [45]. Accordingly those men in areas of high social advantage within the study area (and better access to health services) may have an increased rate of cancer detection by physician. In contrast, men not currently employed, who have previously been shown to have a low rate of health care utilization [46,47], had one of the lowest rates of cancer in this study. It is tempting to speculate that these men may have cancers diagnosed late and therefore be prone to an increased cancer-related mortality.Type II diabetes mellitus (T2DM)T2DM is a common condition of the older male. In this cohort, 15.6% of the men examined at clinic, had T2DM by either self-report or fasting glucose or HbA1c, a figure slightly higher than the prevalence reported by the AusDiab study (12.3% of men aged over 35 years [48]). Our data was collected 3 years later than AusDiab and the higher prevalence in our study most likely relates to the very high prevalence of obesity in the cohort [19].In all levels of the analyses there were strong associations with the obesity (BMI, waist circumference, family history) in an order consistent with many other studies (see [49-51] for summary). One of the strongest associations observed with diabetes in this cohort, was the increased risk in widowed men. Following the loss of spousal support, widowed men (both middle aged and elderly) have been shown to exhibit poor self care, for example a reduction in physical activity and sub-optimal macronutrient intake [52]. Currently, there are some limited multi-disciplinary programmes for widowed men in Australia (support groups, 'tool-shed' social meetings etc.), which in part seek to demonstrate appropriate management of diabetes and appear to be having a positive effect on presenting complications [53].There is also considerable evidence that supports the inverse association between income and T2DM, particularly in elderly men [54,55], as has been reported by others in both men and women in the same geographic region [56]. Whilst multifactorial in nature, a large portion of this effect appears to reside in the (perceived and real) expense of maintaining a well-balanced diet, and an over consumption of high fat/sugar foods at lower cost [57]. The other factors found to associate with diabetes in this study (namely, smoking and hypertension) were broadly consistent with other studies (see [58,59] for summary).Osteoarthritis & Rheumatoid ArthritisThe prevalence of musculo-skeletal conditions (particularly osteoarthritis, rheumatoid arthritis and osteoporosis) has been increasing in both in Australia [3] and abroad [60,61]. The proportion of elderly men with such conditions is also increasing, and to a greater extent than in ageing females [62]. In this study, the prevalence rates of ever-diagnosed osteoarthritis and rheumatoid arthritis for men in the sampling region was the same as that reported from national sources [3].As expected, the strongest determinant of both osteoarthritis and rheumatoid arthritis was an increasing age [3]. Given the projected increased prevalence of these conditions in an ageing population and the associated impact on disability and public health costs, there is an urgent need for research identifying earlier markers and preventative treatments for these conditions. There is also a large body of evidence that identifies obesity as a risk factor for developing osteoarthritis [63-65]. Moreover once osteoarthritis is established, its progression and consequent limitations are more pronounced in obese men [66]. Whilst it was generally assumed that this effect is mediated through a decrease in physical activity [67], our findings are consistent with recent studies showing a minimal or non-existent relationship between osteoarthritis and physical activity patterns [68]. The mechanism by which obesity increases the risk of osteoarthritis is likely more than a simple mechanical effect [69]. The combined problem of an ageing and increasingly obese population and the morbidity and costs associated with osteoarthritis emphasise the importance of further study and primary prevention.In a study of this type there are several limitations of the design that may limit findings. Firstly, the use of self-report data (whilst common to most health surveys and epidemiological studies) likely underestimates the true prevalence of any conditions, an effect repeatedly observed in male respondents [3]. Obviously, the type of self-report data used (current vs. 'ever-diagnosed by doctor' conditions) necessitates caution when comparing rates of disease amongst different sources. The prevalence of any condition can often differ substantially by region (even between local and statistical divisions). Indeed data from the latest National Health Survey advises standard errors of up to 25% for some conditions (e.g. angina, cancer) [1]. Second, the use of point-prevalence data in trying to examine associations with such dynamic diseases is not ideal, although this is common in most epidemiological studies of this type. Third, whilst the study included a wider range of variables than most, the complex nature of the conditions examined meant that much of the variability remains unexplained. Whilst including too many predictors can dilute an analysis of this type, inclusion of additional data might have further clarified factors associated with disease risk (e.g. alcohol consumption, nutrient, and energy intake). Finally, the directionality of many of the observed relationships cannot be determined in cross-sectional studies of this sort.Conclusions & implicationsThis analysis was designed to give an indication of the chronic disease burden in one of the few cohort studies specific to the ageing male worldwide, and the biological, behavioural and environmental settings that associate with these conditions. Whilst many proposals and strategies have been developed to address the problem of chronic disease and ageing populations, the unique challenges posed by elderly men still receives disproportionate attention.This study demonstrated that a high proportion of men are currently suffering from chronic disease, including most of the conditions recently identified as National Health Priorities in Australia. It is clear from this study and others like it, that these conditions are further exacerbated with age. Furthermore, the prevalence of conditions examined in this study were largely equivalent to available data for age-matched men from the region, providing further qualified support to the representativeness of the cohort to the local population.Obesity was associated with most of the diseases studied, and for some, having one or more obese parents added to the risk (angina, diabetes), even if those affected were not themselves obese, (at least at the time point examined). Many men had multiple, often related, conditions (angina, cancer, diabetes) and this multi-disease state is particularly prominent in elderly men [70], underscoring the importance of primary prevention. Apart from age and obesity a number of social and demographic factors were associated with chronic disease prevalence. Spousal support is demonstrably important; separated and widowed men have an increased prevalence of asthma and diabetes, respectively. Men from low-income households were at greater risk of diabetes. Participants not in the work force (most likely due to their disease) were shown to be more likely to have angina or cancer.The ongoing collection of data from this cohort, with a high retention rate to date [19], will provide clearer information about cause and effect relationships. Since DNA has also been collected and stored, rapid advances in technology will permit an assessment of gene-environment interactions in the future.The challenges that the ageing male presents in the current global fight against chronic disease have important policy and public health implications. Men still occupy the majority of the workforce, and in the background of an ageing population, understanding the many facets of the development of disease and disability are vital for any functional and productive society.Competing interestsThe authors declare that they have no competing interests.Authors' contributionsGAW & MTH conceived of the study. SAM participated in the design, management and coordination of the study and drafted the manuscript. AWT provided crucial initial infrastructural support and contributes to the management of the study. GAW acts as Chief Investigator and continues to manage the study. MTH participated in the study design and coordination. SMM performed statistical analyses. All of the authors read and approved the final manuscript.Pre-publication historyThe pre-publication history for this paper can be accessed here:Supplementary MaterialAdditional file 1Table 1. Risk of angina, asthma and cancer by personal, behavioural and socioeconomic factors (attached).Click here for fileAdditional file 2Table 2. Risk of type 2 diabetes, osteoarthritis and rheumatoid arthritis by personal, behavioural and socioeconomic factors (attached).Click here for fileAdditional file 3Table 3. Multivariate model of personal, behavioural and socioeconomic predictors of selected chronic diseases (attached).Click here for file\n\nREFERENCES:\nNo References"
+}

batch_5/PMC2531109.json

@@ -0,0 +1,4 @@
+{
+    "id": "PMC2531109",
+    "text": "This is an academic paper. This paper has corpus identifier PMC2531109\nAUTHORS: Elisabetta Pandolfi, Maria C Graziani, Roberto Ieraci, Giovanni Cavagni, Alberto E Tozzi\n\nABSTRACT:\nBackgroundImproving immunisation rates in risk groups is one of the main objectives in vaccination strategies. However, achieving high vaccination rates in children with chronic conditions is difficult. Different types of vaccine providers may differently attract high risk children.AimTo describe the characteristics of two populations of children who attended a private and a public immunisation provider in the same area. Secondarily, to determine if prevalence of patients with underlying diseases by type of provider differs and to study if the choice of different providers influences timeliness in immunisation.MethodsWe performed a cross-sectional study on parents of children 2 – 36 months of age who attended a private hospital immunisation service or a public immunisation office serving the same metropolitan area of Rome, Italy. Data on personal characteristics and immunisation history were collected through a face to face interview with parents of vaccinees, and compared by type of provider. Prevalence of underlying conditions was compared in the two populations. Timeliness in immunisation and its determinants were analysed through a logistic regression model.ResultsA total of 202 parents of children 2–36 months of age were interviewed; 104 were in the public office, and 98 in the hospital practice. Children immunised in the hospital were more frequently firstborn female children, breast fed for a longer period, with a lower birthweight, and more frequently with a previous hospitalisation. The prevalence of high risk children immunised in the hospital was 9.2 vs 0% in the public service (P = 0.001). Immunisation delay for due vaccines was higher in the hospital practice than in the public service (DTP, polio, HBV, and Hib: 39.8% vs 22.1%; P = 0.005). Anyway multivariate analyses did not reveal differences in timeliness between the public and private hospital settings.ConclusionChildren with underlying diseases or a low birthweight were more frequently immunised in the hospital. This finding suggests that offering immunisations in a hospital setting may facilitate vaccination uptake in high risk groups. An integration between public and hospital practices and an effort to improve communication on vaccines to parents, may significantly increase immunisation rates in high risk groups and in the general population, and prevent immunisation delays.\n\nBODY:\nBackgroundThe control of vaccine preventable diseases is one of the major advances of public health [1]. Although immunisation uptake is high for most routine immunisations in western countries, yet high risk groups, including children with underlying diseases, have often low immunisation coverage [1]. Chronic diseases such as neurological and cardiovascular disorders are associated with high hospitalisation rates [2,3], and some immunisations including influenza and conjugate pneumococcal vaccines may prevent admission into hospital, medical visits, and other negative effects in these patients [4,5]. Despite mathematical models suggest that focusing immunisations on high risk groups may be suitable [6,7], parents of children at risk may underestimate incidence and severity of vaccine preventable diseases and may not be appropriately informed about safety and efficacy of available vaccines [7-10]. Moreover immunisation delays often occur because of false contraindications linked to an underlying condition that, on the contrary, may represent an indication to immunisation [10-12]. Immunisation rates are usually monitored by immunisation registries [13]; in the U.S. vaccine delivery is shifting from the public sector to the private sector, with an emphasis on vaccination in the context of primary care and the medical home [14]. Patients with chronic diseases may refer to private providers, that often do not submit data to a national or regional registry [13,15]. The reasons leading parents to choose a private practice for immunisation have not been well studied yet. This study aims, primarily, to describe the characteristics of two populations attending a private or a public immunisation service in the same metropolitan area. Secondarily this study has the objective to determine if the prevalence of patients with chronic diseases by type of service differs, and to study if the choice of different providers is associated with timeliness in immunisation.MethodsBackground on local immunization policiesIn Italy immunisations against diphtheria, tetanus, pertussis, hepatitis B, poliomyelitis, Haemophilus influenzae type b, mumps-measles-rubella are universally offered to all infants [16]. Children with underlying diseases are recommended to receive influenza, pneumococcal, meningococcal and varicella vaccines [17,18]. A national survey performed in 2003 showed that immunisation coverage within 24 months of age for influenza, pneumococcal, and varicella vaccines was less than 3% in the general population, and less than 10% in high risk groups [19,20]. Most immunisations are delivered in Italy by the public health service while private immunisation practices administer nearly 3% of all vaccines with wide differences among Regions [20].The national immunisation schedule includes: three doses of DTaP, Polio (IPV), HepB, Hib in the first year of life; a first dose of MMR between the 12th and the 14th month; a second dose of MMR at 5 years of age. Conjugate pneumococcal, conjugate meningococcal C, and varicella vaccines are recommended at national level for selected risk groups only.SettingWe conducted the present study in a private paediatric research hospital with nearly 400 beds, and in a public immunisation service in the same area. The two facilities offer the same immunisations with the same schedule but with different charges for families (Table 1).Table 1Comparison of the immunisation services included in the study.Public immunisation serviceHospital providerApproximate average number of immunisations per year50.0003.000DTaP, Polio, Hib, HepBFreeFully chargedPneumococcal conjugate vaccineCopaymentFully chargedMeningococcal C conjugate vaccineCopaymentFully chargedInfluenza vaccineCopaymentFully chargedVaricella vaccineCopaymentFully chargedStudy designWe performed a cross-sectional study on parents of children 2 – 36 months of age in a private hospital immunisation service attended by outpatients or to a public immunisation office serving the same metropolitan area of Rome, Italy, with approximately 53.000 inhabitants.One interviewer (EP) visited the two practices two days a week during office hours, and systematically performed a face to face interview to parents of vaccinees from January to July 2006 until reaching the desired sample size (Figure 1; Figure 2).Figure 1Co-operation rate of parents to the interview in the public immunization service.Figure 2Co-operation rate of parents to the interview in the Pediatric hospital.A questionnaire including information on children immunisation history, the reasons for delay or incomplete immunisation, and the reasons for choosing a public or a private hospital service, was administered to parents before the child was immunised after obtaining informed consent.DefinitionsFor influenza vaccine, coverage was calculated as the proportion of children older than 6 months who had received at least one dose in the past season. For conjugate pneumococcal and meningococcal C vaccines, coverage was the proportion of children older than three months who received at least one dose of vaccine. For varicella vaccine, coverage was the proportion of children older than 12 months who ever received at least one dose of vaccine.Delay in immunisations was defined for DTP, Polio, HBV and Hib as: a) first dose received later than 105 days of age; OR b) second dose received later than 70 days after the first dose; OR c) third dose received after 380 days of age. For MMR and varicella delay was defined as an immunisation after the 15th month of age. These time intervals were calculated according to national recommendations for vaccine administration [20].Sample sizeThe sample size was calculated to detect a difference in the prevalence of high risk children in the two settings. Assuming a prevalence rate of children with underlying diseases of 3% in the general population [19] and a risk ratio of 5, a total population of 200 children (100 per group) was considered sufficient to detect a difference in prevalence between those observed in the public and in the hospital practice with a power of 80% and a level of significance of 5%.AnalysisChi square or Fisher exact test for categorical variables and the Student's t test for continuous variables were used for assessing statistical significance at the univariate level. A logistic regression model was used to asses the role of different variables as determinants of timeliness in immunisation. Odds ratios and their 95% confidence intervals were used as measures of effect in the logistic regression model.ResultsA total of 202 parents of children 2–36 months of age were interviewed, 104 (51.5%) in the public office, and 98 (48.5%) in the hospital practice. (Figure 1a,1b).The general features of parents interviewed and their children in the two groups are shown in Table 2.Table 2Characteristics of vaccinees and their families recorded by type of practicePublic immunisation service (N = 104)Hospital provider (N = 98)Total (N = 202)P-valueMother's mean age, years (range)34.7 (19 – 47)34.7 (24 – 45)34.7 (19–47)0.96Mother from foreign country, n (%)17 (16.3%)7 (7.2%)24 (11.9%)0.04Graduated mother, n (%)59 (56.7%)40 (41.2%)99 (49.2%)0.03Working mother, n (%)75 (72.1%)67 (69.0%)142 (70.3%)0.04Father's mean age, years (range)37.3 (26 – 52)37.8 (23 – 63)37.6 (23–63)0.48Father from foreign country, n (%)13 (12.5%)5 (5.1%)18 (8.9%)0.06Graduated father, n (%)47 (45.2%)37 (37.7%)84 (41.6%)0.30Working father, n (%)103 (99%)96 (97.9%)199 (98.5%)0.50Children age, months, mean (range)9.8 (2–22)11.8 (2–35)10.8 (2–35)0.02Male children, n (%)62 (59,6%)40 (40.8%)102 (50.5%)0.008Firstborn child, n (%)56 (53.8%)69 (70.4%)125 (61.9%)0.01No. of households, mean (range)3.5 (2 – 5)3.4 (2–6)3.5 (2–6)0.50Children with Previous hospitalisations (%)15 (14.4%)23 (23.7%)38 (18.8%)0.09Birthweight, g, mean (range)3270.0(1500–4400)3065.0(1180 – 4500)3170 (1180–4500)0.01Breast feeding duration, months, mean (range)4.21 (0 – 15)5.3 (0 – 14)4.7 (0–15)0.04Although several characteristics of parents and children seen in the two practices were similar, foreign parents and graduated mothers were more represented in families who requested immunisation in the public service, whereas families interviewed in the hospital practice had more frequently firstborn female children to be vaccinated, who were breast fed for a longer period, who had a lower birth weight, and who were more frequently hospitalised in the previous period.The overall immunisation coverage for influenza vaccine was 1.9% (2 patients) in children in the public office and 1.0% (1 patient) in those immunised in the hospital practice (P = 0.52). Immunisation with conjugate pneumococcal vaccine was administered to 70 (67.3%) children in the public practice and 63 (64.0%) in the hospital office (P = 0.65), whereas 59 (56.7%) of them in public service and 49 (50.0%) in hospital office were immunised with conjugate meningococcal C vaccine (P = 0.33). Finally, none of the children in the public practice and 5 (5.10%) in the hospital office were immunised against varicella vaccine (P = 0.02)Nine (9.2%) out of the 98 patients immunised in the hospital had an underlying disease which was an indication for influenza, pneumococcal, and meningococcal C immunisation. Out of them only 1 child was immunised against influenza; 6 against pneumococcus, and 4 against meningococcus C at the time of the interview. Underlying diseases included: three cases of congenital heart disease, two cases of a neurological disease, and one case of a genetic syndrome. None of the children immunised in the public service had an underlying disease (P = 0.001).The Italian Ministry of Health also recommends influenza immunisation of households of patients belonging to risk groups [20]. Among the parents of these children only 2/9 mothers and 4/9 fathers received influenza immunisation.Although the number of children with an immunisation delay was substantial in both practices, they were nearly twice in the hospital practice than in the public service (DTP, polio, HBV, and Hib: 39.8% vs 22.1%; P = 0.005; MMR: 20.0% vs 12.5%; P = 0.15). Causes of delayed or missing immunisation reported from parents were mostly related to misinformation (58%), and child illness (15.6%).Determinants of delayed immunisation was studied through a logistic regression model (Table 3). The univariate analysis showed that a lower education of parents, a low birth weight, and immunisation in the hospital predicted an immunisation delay, but none of these variables were significant in the multivariate model. None of the variables included in the analysis were associated as well with delay of MMR immunisation, while children with a father from a foreign country were more likely to receive pneumococcal immunisation (OR: 3.18; 95% CI: 1.08 – 9.42).Table 3Likelihood of on-time vaccination (multivariate analysis results are reported only for those variables who resulted significant at the univariate analysis)DTP, HBV, POLIO, HibUnivariate analysis, OR (95% CI)Multivariate analysis, OR (95% CI)Mother's age < 30 yrs0.52 (0.22–1.24)Mother from foreign country1.17 (0.43–3.13)Mother degree0.46 (0.24–0.90)1.387 (0.662;2.903)Working mother0.88 (0.44–1.79)Father's age0.29 (0.08–1.01)Father from foreign country0.86 (0.25–2.75)Father degree0.41 (0.20–0.82)2.102 (0.963;4.589)Working father0.22 (0.01–3.11)Child age < 12 months0.74 (0.40–1.34)Vaccinee's male gender0.97 (0.51–1.85)Child with previous hospitalization2.19 (1.00–4.81)1.757 (0.823;3.750)Child birthweight < 2500 gr0.41 (0.14–1.13)1.000 (1.000;1.001)Child breastfeeding duration < 6 months0.79 (0.38–1.65)Firstborn child1.44 (0.73–2.85)Private immunisation office2.32 (1.22–4.54)The most frequent reason for choosing the hospital service was that parents felt safe in a hospital environment (39%). On the other hand parents whose children were immunised in the public service felt that the office was easy to reach (20.2%) or were advised by the family paediatrician (19.2%).Even if a higher proportion of children with chronic conditions requiring immunisation in the hospital was expected, it should be noted that 38% of families interviewed in the hospital service had previously visited a public office for immunisation and 38% of their children were not immunised due to a false contraindication, that most of all was represented by the child's underlying disease. Another reason for attending the hospital practice was unavailability of some vaccines due to a temporary shortage of Meningococcal, Pneumococcal or Varicella vaccines, because of their different policies compared with universally recommended vaccines. Finally 8% of children immunised in the hospital had previous contacts with the same hospital.Discussion and conclusionThis study shows that children with underlying diseases, which often represent an indication to certain immunisations, were more likely to attend the hospital immunisation provider rather than the public office of their own district.Families of children with low birth weight preferred more often immunisation in the hospital and were more likely to be immunised against influenza. Parents who chose to immunise their children in the hospital have the perception of hospital as a safer environment; this finding underlines how lack of information or misinformation and the perception of an underlying disease as a contraindication, play a substantial role in the choice of health care providers and inappropriate consultation.Despite recommendations in place, coverage for immunisations indicated in high risk groups is low in our country [18,19]. Other studies performed in hospital settings and on high risk group children showed that immunisation coverage among these groups remains low and immunisations are often delayed. [21-28].We found that families of children with immunisation delay attending the hospital practice had often previously attempted to vaccinate their children. This observation underlines the potential to improve immunisation timeliness through simple information and educational activities.Different approaches have been proposed for overcoming immunisation barriers in high risk groups. There is evidence that enhanced information focusing on vaccines benefits and how to manage their potential adverse effects, increased availability of vaccine offices, use of missed opportunities, and gratuity of immunisation are efficacious in increasing immunisation uptake [11-13].We also observed that being immunised in a hospital setting was a predictor of delayed immunisation for routine vaccines. We speculate that this finding may be associated with misinformation and false contraindications, and it is in line with the frequent parents' perception of hospital as a safe environment.Regarding the determinants of appropriateness of immunisation we did not find significant associations with any of the variables included in the analysis. However the study may have not been adequately powered to address this issue. Nevertheless, we found a signal toward an association between an higher education of parents and timeliness of immunisation.Although the number of vaccinees may be small in a private hospital practice, this finding underlines how offering immunisations indicated in high risk groups in a hospital setting may result in a significant benefit. Hospitals are likely to attract high risk groups which often are immunised late.One advantage of this study was the inclusion of different immunisation settings serving the same population although these results are not necessarily generalizable to other settings. Our findings show that an integration between public and hospital settings, and an effort to improve communication on vaccines to parents, may significantly increase immunisation rates in high risk groups and in the general population and prevent the delays in immunisation.Competing interestsThe authors declare that they have no competing interests.Authors' contributionsAll Authors participated to the design and coordination of the study and helped to draft the manuscript. All authors read and approved the final version of the manuscript.Pre-publication historyThe pre-publication history for this paper can be accessed here:\n\nREFERENCES:\nNo References"
+}

batch_5/PMC2531123.json

@@ -0,0 +1,4 @@
+{
+    "id": "PMC2531123",
+    "text": "This is an academic paper. This paper has corpus identifier PMC2531123\nAUTHORS: Javaid Iqbal, Iftikhar Ahmed, Wazir Baig\n\nABSTRACT:\nIntroductionMyocardial abscess is a rare and potentially fatal condition. Metastatic myocardial abscess in the setting of infective endocarditis has been infrequently reported in the medical literature. To the best of the authors' knowledge no case of myocardial abscess affecting the free wall of the left ventricle secondary to infective endocarditis of a right-sided heart valve has been reported previously.Case presentationWe report a case of tricuspid valve endocarditis caused by Staphylococcus aureus and resulting in a myocardial abscess on the posterior wall of the left ventricle, far from the active valvular infection. We also briefly discuss the role of different investigation modalities including cardiac magnetic resonance imaging in diagnosing myocardial abscess.ConclusionMyocardial abscess is a life-threatening illness. A high index of clinical suspicion is required to make a prompt diagnosis. Final diagnosis may need multi-modality imaging. An early diagnosis, aggressive medical therapy, multidisciplinary care and timely surgical intervention may save life in this otherwise fatal condition.\n\nBODY:\nIntroductionMyocardial abscess (MA) is a suppurative infection of the myocardium, endocardium, native or prosthetic valves, perivalvular structures or the cardiac conduction system. It is a potentially life-threatening disease, where early recognition and institution of appropriate medical and surgical therapy is vital for patient survival. The overall mortality rate associated with Staphylococcus aureus endocarditis is 42%. If treated with appropriate antibiotics and surgery, the mortality rate falls to 25%. The presence of an intracardiac abscess results in a 13.7-fold increase in mortality. In the past, most cases of MA were found during autopsy; however, detection of MA can now be achieved antemortem, using noninvasive diagnostic modalities including transthoracic echocardiography (TTE), transoesophageal echocardiography (TOE), radionuclide scintigraphy, computed tomography (CT) scan and cardiac magnetic resonance imaging (CMRI).Case presentationA 28-year-old intravenous drug user was admitted in a district general hospital with a 2-week history of fever, malaise and myalgia. He had no past medical history of note. On examination he was pyrexial but haemodynamically stable. His cardiovascular examination revealed signs of tricuspid regurgitation. His respiratory, abdominal and neurological examination was normal. Clinically, the diagnosis of infective endocarditis (IE) was suspected. Three sets of blood cultures were drawn and empirical intravenous antibiotic treatment commenced.His blood tests showed leukocytosis with predominant neutrophilia and mild normochormic, normocytic anaemia. His electrocardiogram revealed non-specific ST-changes but no conduction abnormality. His chest X-ray was unremarkable. TTE confirmed vegetations on the tricuspid valve with severe regurgitation. All other valves were normal. Blood cultures grew S. aureus and vigorous antibiotic treatment was continued appropriately. However, the patient's condition continued to deteriorate with spiking fever and raised inflammatory markers. He was referred to the regional cardiothoracic centre for evaluation of valve surgery in view of uncontrolled infection.On arrival at the cardiothoracic centre, the patient was acutely unwell with a temperature of 38.5°C, pulse of 120 beats per minute, blood pressure of 100/70 and respiratory rate of 26 breaths per minute. He had signs of severe tricuspid regurgitation and right heart failure. His repeat chest X-ray showed multiple cavitating lesions depicting metastatic pulmonary abscesses. There was also evidence of splenic abscesses on his abdominal ultrasound scan. Repeat TTE confirmed vegetations on the tricuspid valve with severe regurgitation but additionally it showed a small echo-free space in the wall of the left ventricle, raising suspicion of an MA (Figure 1). TOE was planned to evaluate this further but the patient was unable to tolerate it. Urgent CMRI was obtained, which revealed a 4.5 cm diameter left ventricular posterior wall abscess contained by only a 2 mm thin layer of myocardium (Figure 2 and Additional file 1). Urgent surgical intervention was planned but, unfortunately, the patient had a cardiac arrest prior to surgery and could not be resuscitated.Figure 1Transthoracic echo – Short axis view showing abscess cavity.Figure 2Cardiac magnetic resonance imaging. A 4.5 cm diameter left ventricular posterior wall abscess contained by only a 2 mm thin layer of myocardium.DiscussionMA has been reported in about 20% of patients with IE [1]. They are usually adjacent to the area of valve infection and represent a direct extension of the infection [2]. Rarely, embolization of septic material results in a metastatic myocardial abscess remote from the main focus [3-6], as was the case here. The normal appearance of left-sided valves and concurrence of abscesses in extracardiac organs lead us to the conclusion that the left-sided MA had occurred as the result of embolization from the right-sided endocarditis.Antemortem diagnosis of MA remains a challenge and a high index of clinical suspicion is required. TTE has a sensitivity of 23% and specificity of 98.6% in diagnosing MA [7]. At present, TOE is considered the investigation of choice but a recent prospective study of 115 patients revealed that it has only 48% sensitivity in diagnosing MA [8]. CMRI is a noninvasive imaging modality with high temporal and spatial resolution. To the best of our knowledge, no studies have compared the diagnostic value of TOE and CMRI in such cases. However, there are a few case reports and studies which suggest good diagnostic yield with CMRI in diagnosing annular abscess [9], sub-valvular abscess [10] and pseudo-aneurysm [11] in the setting of complicated IE. There may also be a complementary role for radionuclide imaging in diagnosing MA, where it can reveal a focal area of increased uptake in the myocardium suggesting the location of an abscess. It has low sensitivity but can be helpful in cases of prosthetic valve endocarditis where echocardiography may show too much scatter. Different radioisotopes including gallium-67, technetium-99 and indium-111 have been used in clinical practice with variable success [12,13].Patients with this lethal disease can be saved by aggressive antibiotic treatment and prompt surgical intervention [5]. This can be best achieved by multidisciplinary care involving cardiologists, microbiologists, cardiac radiologists and cardiothoracic surgeons. Urgent surgery is recommended in most cases of MA since the perioperative risk and chances of rupture increase with the delay to surgery. However, the decision to perform emergency (same day) or urgent (1–2 days) surgery has to be made in individual cases depending on the clinical status of the patient, size of the abscess and thickness of the abscess wall. CMRI can provide useful morphological evaluation to help make this decision [10].ConclusionIn conclusion, MA is a life-threatening illness. A high index of clinical suspicion is required to make a prompt diagnosis. Final diagnosis may need multimodality imaging. Many of these patients may present to district hospitals where appropriate imaging and surgical facilities may not be available, and an urgent transfer to a specialist cardiothoracic centre is imperative. An early diagnosis, aggressive medical therapy, multidisciplinary care and timely surgical intervention may save the patient's life in this otherwise fatal condition.AbbreviationsCMRI: cardiac magnetic resonance imaging; IE: infective endocarditis; MA: myocardial abscess; TOE: transoesophageal echocardiography; TTE: transthoracic echocardiography.Competing interestsThe authors declare that they have no competing interests.Authors' contributionsJI collected data, performed the literature search and drafted the manuscript. IA was involved in the literature search and manuscript review. WB supervised this patient's care and contributed to the preparation of the manuscript. All authors read and approved the final manuscript.ConsentWritten informed consent was obtained from the patient's next of kin for publication of this case report and the accompanying images. A copy of the written consent is available for review by the Editor-in-Chief of this journal.Supplementary MaterialAdditional file 1Cardiac magnetic resonance imaging of our patient showing morphological features of myocardial abscess.Click here for file\n\nREFERENCES:\nNo References"
+}

batch_5/PMC2531183.json

@@ -0,0 +1,4 @@
+{
+    "id": "PMC2531183",
+    "text": "This is an academic paper. This paper has corpus identifier PMC2531183\nAUTHORS: Cécile Esnault, Stéphane Priet, David Ribet, Odile Heidmann, Thierry Heidmann\n\nABSTRACT:\nBackgroundAPOBEC3 cytosine deaminases have been demonstrated to restrict infectivity of a series of retroviruses, with different efficiencies depending on the retrovirus. In addition, APOBEC3 proteins can severely restrict the intracellular transposition of a series of retroelements with a strictly intracellular life cycle, including the murine IAP and MusD LTR-retrotransposons.ResultsHere we show that the IAPE element, which is the infectious progenitor of the strictly intracellular IAP elements, and the infectious human endogenous retrovirus HERV-K are restricted by both murine and human APOBEC3 proteins in an ex vivo assay for infectivity, with evidence in most cases of strand-specific G-to-A editing of the proviruses, with the expected signatures. In silico analysis of the naturally occurring genomic copies of the corresponding endogenous elements performed on the mouse and human genomes discloses \"traces\" of APOBEC3-editing, with the specific signature of the murine APOBEC3 and human APOBEC3G enzymes, respectively, and to a variable extent depending on the family member.ConclusionThese results indicate that the IAPE and HERV-K elements, which can only replicate via an extracellular infection cycle, have been restricted at the time of their entry, amplification and integration into their target host genomes by definite APOBEC3 proteins, most probably acting in evolution to limit the mutagenic effect of these endogenized extracellular parasites.\n\nBODY:\nBackgroundThe APOBEC family of cytosine deaminases includes numerous members that can deaminate cytosine to uracil within DNA and/or RNA molecules. Among these enzymes, the APOBEC3 sub-family has been discovered when human APOBEC3G (hA3G) was reported to restrict HIV replication ([1]; reviewed in [2]). Human hA3G has been shown to trigger extensive deamination of cytosine in the negative viral DNA strand during reverse transcription and to lead to deleterious G-to-A mutations considered as the hallmark of APOBEC3-editing activity. Subsequently, several other human APOBEC3 proteins – including APOBEC3A (hA3A) [3], APOBEC3B (hA3B) [4,5], APOBEC3C (hA3C) [5], APOBEC3DE (hA3DE) [6], APOBEC3F (hA3F) [7-9] and APOBEC3H (hA3H) [10] – have been shown to exhibit antiviral effects against a variety of viruses, including numerous retroviruses – i.e. HIV, SIV, MLV, HTLV and foamy viruses –, hepatitis B virus and adeno-associated virus (AAV) (for review [11]). In contrast to humans, the mouse genome encodes only one APOBEC3 (mA3) protein, which, like human APOBEC3 proteins, displays antiviral effects [12]. Aside from the antiviral function of APOBEC3 proteins against exogenous viruses, some inhibitory effects have been reported on intracellular targets (for review [2]) and several studies support the notion that the primary function of APOBEC3 proteins could be to prevent the propagation of mobile elements. Indeed, mammalian genomes have accumulated numerous transposable elements which account for > 45% of the genomic DNA [13,14]. These elements can be grouped into two main classes: the strictly intracellular non-LTR (Long Terminal Repeat) retrotransposons, namely long interspersed nuclear elements (LINEs) and short interspersed nuclear elements (SINEs), which account for ~30% of each mammalian genome, and the LTR-containing retroelements (including the endogenous retroviruses, ERVs), accounting for ~10% of the genomes and closely related to retroviruses. The life cycle of ERVs includes the formation of virus-like particles (VLPs) that, in several instances – but not systematically – can remain strictly intracellular as observed for the well-characterized murine intracisternal A-particle (IAP) and MusD elements (the so-called \"intracellularized\" ERVs, [15-18]), or that can bud at the cell membrane for an extracellular cycle as observed for the recently identified murine intracisternal A-particle-related envelope-encoding (IAPE; [18]) and the human endogenous retrovirus HERV-K(HML2) elements [19,20]. Although most of these elements are no longer active due to the accumulation of inactivating mutations, some of them are still functional and have been cloned, thus allowing direct ex vivo assay of the effect of APOBEC proteins on their mobility. Accordingly, several APOBEC3 proteins, including hA3A, hA3B, hA3C and hA3F have been demonstrated to restrict the retrotransposition of the human LINE-1 (L1) elements [3,21,22], as well as the L1-dependent transposition [23] of the human Alu SINE elements [24]. Moreover, although no effect on the retrotransposition of L1 elements was observed in the presence of hA3G [21,25-27], reports have shown that hA3G can prevent the retrotransposition of Alu elements [27,28] by sequestering Alu RNAs in cytoplasmic high-molecular-mass (HMM) ribonucleoprotein complexes [28]. Similarly, the cloning of active copies for the intracellular murine IAP and MusD elements [15,17] made possible to demonstrate susceptibility of these retroelements to murine APOBEC3 and to most of the human APOBEC3 proteins [24,26,29]. In addition, in silico analyses of the naturally present genomic copies of these elements in the murine genome have revealed \"traces\" of APOBEC3 editing on these elements ([26]; see also [30]), thus supporting the physiological relevance of the observed ex vivo assays, and the genomic impact of APOBEC3 protein activity.Here we take advantage of the recent identification of the infectious progenitor of the intracellularized IAP retrotransposon, namely IAPE, to analyze the possible restriction of a bona fide murine ERV, in a state close to that at the time of its initial endogenization step when the element still behaved as an infectious retrovirus, having not yet reached its highly adapted \"intracellularized\" state [18]. In parallel, we performed a similar analysis on the human progenitor of the HERV-K(HML2) family members that we had \"reconstituted\", resulting in the Phoenix element which proved to be a bona fide endogenous retrovirus, the element being able to enter cells by infection and integrate with all the characteristic features of the genomic copies presently found in the human genome [19]. These two functional human and murine \"extracellular\" ERVs were used to assess the effects of APOBEC3 proteins on mammalian endogenous retroviruses in appropriate ex vivo assays, and refined in silico analyses of the naturally present copies of these elements in their target host genomes finally unambiguously demonstrated \"traces\" of APOBEC3 editing, with identifiable signatures. Altogether, the data show that APOBEC3 proteins play a role not only on the intracellular retrotransposons found in humans and mice, but also on their retroviral \"progenitors\" endowed with an extracellular life style, thus de facto filling the gap between the described effects of APOBEC3 proteins on bona fide exogenous retroviruses on the one hand and intracellular retroelements on the other.Results and discussionRestriction of murine and human infectious ERVs by APOBEC3 proteinsTo assay whether the mouse IAPE element is restricted by APOBEC3 proteins, we used the previously described functional copy of IAPE-D (; mm9 July 2007 Assembly: chr12: 24,282,555–24,290,874) [18] that was cloned under the control of the CMV promoter, and in which a neo resistance gene was inserted in reverse orientation into the env gene (Figure 1A). The effect of APOBEC3 proteins on HERV-K was analyzed by using the \"reconstituted\" Phoenix element cloned under the control of the CMV promoter, in which the env gene is stopped and an anti-sense-oriented neo resistance gene is inserted into its 3'-LTR (Figure 1). Proviral clones of IAPE-D or HERV-K (4.5 μg), complemented with an expression vector for a functional IAPE or VSV-G Env (0.5 μg) respectively, and the murine (mA3) or human (hA3A-G) APOBEC3 proteins or a control plasmid (5 μg), were transfected in 293T cells. Supernatants were harvested 48 h post-transfection, filtered through 0.45-μm pore-size PVDF membranes, supplemented with Polybrene (4 μg/ml), and transferred onto HeLa target cells. To increase sensitivity, target cells were subjected to spinoculation at 1.200 × g for 2.5 h at 25°C. Infection events were detected after G418 selection of target cells and viral titers expressed as the number of G418R clones per mL of supernatant. As illustrated in Figure 1, mA3 and hA3G protein expression leads to a dramatic decrease in both the IAPE-D and HERV-K viral titers (Figure 1B). In the case of the murine IAPE-D element, only a limited effect – if any – was observed with the human APOBEC3 proteins other than hA3G, with for instance no effect of hA3A which otherwise has a strong effect on the rate of retrotransposition of its intracellular counterpart, i.e. the IAP element (Figure 1). In the case of the human HERV-K, at variance with what is observed for the murine IAPE-D element, almost all the APOBEC3 proteins (with the exception of hA3C) have an effect, the highest activity being observed with hA3B and hA3F.Figure 1Murine and human APOBEC3 proteins inhibit endogenous retroviruses. (A) Rationale of the assay for detection of infection events by endogenous retroviruses in the presence of APOBEC3 proteins. The IAPE-D and HERV-K elements used in the assay are marked with the neo reporter gene – inserted in reverse orientation – and carry their own functional genes, except for the env gene which is supplied in trans, thus allowing only for single rounds of infection. Human 293T cells are co-transfected with the indicated expression vectors for APOBEC3 family members, the supernatants collected 2-days post-transfection to infect HeLa target cells, and infection events detected upon G418 selection. (B) Analysis of the activity of murine and human APOBEC3 proteins on the indicated endogenous retroviruses. Viral titers are given as percentages relative to a control (no apobec: expression vector with a nonfunctional hA3G; 622 and 549 G418R clones/ml for IAPE-D and HERV-K, respectively). Data are the means ± standard deviations (s.d.) for at least three independent experiments. Bottom: retrotransposition frequency of an active autonomous IAP element marked with a neo indicator gene for retrotransposition [17] in the presence of the corresponding APOBEC3 proteins; the assay was performed by cotransfection of HeLa cells with the marked IAP and APOBEC expression vector as previously described [26]; values are the means ± standard deviations (s.d.) for at least three independent experiments and are given as percentages relative to the control (no apobec; 1.3 × 10-3 G418R clones/cell).We further assessed whether the observed decrease in viral titers was associated with editing of the viral DNA by sequencing a 800 or 1600 bp fragment of the de novo integrated IAPE-D or HERV-K proviral DNA copies, respectively, in 20–25 individual G418R clones. As illustrated in Figure 2 numerous G-to-A transitions were observed in the presence of mA3 or hA3G in both ERVs, as expected for an APOBEC3-mediated editing. For HERV-K, G-to-A editing was also observed with hA3B, hA3DE and hA3F, but not with hA3A, as expected from previous characterization of this enzyme ([3,21,24,29]; reviewed in [11]). Furthermore, mA3 and hA3G editing leads to G-to-A mutations in a GXA or GG context, respectively, which are the hallmarks previously described for each enzyme [26,31,32]. For hA3B and hA3F, G-to-A editing was observed in the GA context [2,33]. In addition, in spite of a low number of G-to-A mutations, hA3DE editing seems to preferentially take place in the GA/T context as expected [6]. It has to be stressed that the editing rate is probably underestimated because too heavily mutated neo genes present in these ERV DNAs can no longer confer G418 resistance after integration.Figure 2APOBEC3 proteins induce specific G-to-A hypermutations. Two-entry tables showing nucleotide substitution preferences in the presence of the indicated APOBEC3 proteins for the IAPE-D and HERV-K integrated proviruses. n, total number of bases sequenced. The adjacent graphs represent the relative frequencies of observed G-to-A mutations as a function of the G neighboring nucleotides (+2 position for the expected mA3 footprint, +1 position for the other APOBEC3s); for the two-entry tables, p-values calculated by a Poisson regression in a log-linear model for the occurrence of the G-to-A versus C-to-T mutations yielded p < 0.03 in all cases (except for hA3DE (p = 0.18) due to the low number of mutations); for the adjacent graphs, p-values calculated by a chi square test were p < 0.01 in all cases (except again for hA3DE, p = 0.7); similar levels of significance (or even higher) were obtained using the Kruskal Wallis test.Traces of APOBEC3 past activity on resident IAPE and HERV-K elements in the murine and human genomeSince the murine IAPE-D and the human HERV-K elements are found to be restricted by APOBEC3 proteins in the ex vivo assay above, we asked whether APOBEC3 proteins might have actually impaired the in vivo amplification of these elements in the past, by searching for evidence of APOBEC3-editing on the endogenous copies residing in the murine and human genome, respectively. Accordingly, an in silico analysis was performed to assess the levels of G-to-A mutations in two sets of full-length genomic IAPE elements, originating from two different subfamilies, namely IAPE-A and IAPE-D, and on full-length HERV-K elements. Both the murine IAPE-D subfamily and the human HERV-K elements have most probably been amplified by reinfection of the germline and therefore could have been subjected to APOBEC3 editing. Conversely, the IAPE-A subfamily has most probably been amplified via gene duplication, with several elements – essentially on the Y chromosome – disclosing identical flanking sequences [34,35], and therefore should not have undergone APOBEC3 editing: this family of elements – closely related to IAPE-D – can therefore be used as an internal control for the in silico genomic analyses. For all three families of elements, we selected by BLAST analysis a set of twenty copies displaying the closest sequence similarity to their cognate \"master\" copy: to the functional \"Phoenix\" element for HERV-K, to the functional copy used in the ex vivo assay for IAPE-D, and to the unique full-length copy with preserved open reading frames for IAPE-A. A consensus sequence was then derived for each family of elements, and each family member was analyzed for mutations to the consensus. As illustrated in Figure 3A, numerous mutations can be found for the three families of elements, consistent with the million years of genome evolution that have elapsed since the initial infection and/or amplification events. However, a specific increase in G-base mutations can be observed for both the IAPE-D and the HERV-K copies, not observed for the IAPE-A copies. These mutations are essentially G-to-A substitutions, with the effect being most probably \"strand-specific\", since the number of such mutations is almost twice that of the C-to-T substitutions. In addition, this bias is not observed for the IAPE-A elements, as expected for a duplicated element which has amplified by chromosomal DNA duplication, without a reverse transcription step prone to APOBEC3 mutagenesis. Interestingly, as illustrated in Figure 3B, the observed G-to-A changes are not randomly distributed but seem to be influenced by the neighbouring nucleotides: the GXA triplet is the most frequent \"target\" for the G-to-A substitutions in the IAPE-D elements (see arrow in Figure 3B), in agreement with previous reports – and data in Figure 2 – indicating that mA3 preferentially targets GXA trinucleotide motifs [26,31,33]. On the other hand, the G-to-A substitutions in the HERV-K copies are most frequently observed in the GG context (see arrow in Figure 3B), which corresponds to the footprint of hA3G editing [31] and data in Figure 2. There is no clear-cut evidence for G-to-A substitutions in the GA and GT context, excluding any significant contribution of hA3B, hA3DE or hA3F. Noteworthily, a \"non-specific\" bias can be observed for the endogenous IAPE-A, -D and HERV-K elements, which favors G-to-A mutations in CG dinucleotides (Figure 3B), most probably reflecting an APOBEC3-independent (since it is also observed for the duplicated IAPE-A elements) deamination of methylated-CpG islands. Finally, examples of sub-genomic regions of IAPE-D and HERV-K elements enriched in G-to-A substitutions, are shown in Figure 3C, where the di- or tri-nucleotide sequences specific for the hA3G and mA3 APOBEC proteins, respectively, are underlined. Altogether, these in silico data strongly suggest that the IAPE and HERV-K elements have been subjected to editing by specific APOBEC3 proteins during their retroviral cycle of amplification and insertion into their target host genome.Figure 3Distribution of the nucleotide substitutions in the IAPE-D and HERV-K genomic copies residing in the mouse and human genomes. Endogenous sequences were extracted from the mouse and the human genome databases, aligned and compared to the derived consensus. (A) Upper panels: percentage of substitutions for each nucleotide, for the endogenous IAPE-D and HERV-K elements (with the IAPE-A elements used as a control). Lower panels: two-entry tables showing nucleotide substitutions preferences, with the G-to-A values in bold (higher that the \"non-specific\" C-to-T value for IAPE-D and HERV-K, and identical in the case of the IAPE-A control). n, total number of nucleotides analyzed. (B) Influence of nucleotides at position -2, -1, +1 and +2 on G-to-A mutations (the mutated G is at position 0). Data represent the percentage of indicated target di- or trinucleotide sequences bearing G-to-A mutations. X represents any nucleotide. P-values calculated for the two-entry tables (in A) by a Poisson regression in a log-linear model for the occurrence of the G-to-A versus C-to-T mutations yielded p < 0.003 for IAPE-D and HERV-K; in B, p-values calculated by a chi square test were p < 0.001; similar levels of significance were obtained using the Kruskal Wallis test. (C) Example of G-to-A mutations present in twenty IAPE-D (upper panel) and HERV-K (lower panel) sequences. GXA trinucleotides and GG dinucleotides are underlined in the consensus sequence of IAPE-D and HERV-K, respectively.We further explored APOBEC3 editing by analyzing more specifically the G-to-A substitutions at the mA3 and hA3G target sites for each of the twenty IAPE-D and HERV-K proviruses, respectively. As shown in Figure 4A–B, for each proviral element, both the total number of G-to-A mutations (grey plus hatched grey) and the number of G-to-A mutations at the mA3- and hA3G-specific sites (hatched grey) were measured, together with the number of C-to-T \"non-strand-specific\" mutations as an internal control (dark bars; also used to order the copies in the Figure). Figure 4A–B then clearly shows that i) the total number of G-to-A mutations is for most proviruses higher than that of the \"control\" C-to-T mutations, ii) this increase is essentially due to \"specific\" mutations at the respective APOBEC3 sites, and iii) the extent of the observed mutations is highly variable depending on the proviral copy. Actually, for both the IAPE-D and HERV-K proviruses, more G-to-A mutations than C-to-T mutations can be observed, consistent with a strand specificity that can only have occurred prior to integration; in addition, this excess of G-to-A mutations is in general observed at GXA triplet positions for the murine, mA3-sensitive IAPE-D (> 40% of the G-to-A mutations for the majority of the proviruses, namely thirteen out of twenty), and at GG doublet positions for the human, hA3G-sensitive HERV-K elements. For the latter, it should be noted that the extent of specific G-to-A mutations is rather limited (seventeen out of the twenty HERV-K proviruses display < 30% of their G-to-A mutations at GG positions), except for two proviruses (ch3-1271 and ch21-0189) which are specifically hypermutated (Figure 4B–C), with > 70% of their G-to-A mutations in the GG context, without any evidence for a clear-cut gradient along the proviral sequence (Figure 4C). These results indicate that HERV-K can indeed be severely edited by hA3G, and that APOBEC3G protein expression at different times of HERV-K amplification in the human genome must have been quite variable.Figure 4Variability of the number of G-to-A mutations within the endogenous IAPE-D and HERV-K proviruses depending on the element. The total numbers of G-to-A mutations (plain + hatched grey bars) for each IAPE-D (A) and HERV-K (B) proviruses are represented, together with that of the C-to-T (black bars) \"none-strand-specific\" mutations, given as an internal control (also used for ordering the elements) indicative of the genetic drift-associated age-dependent amount of mutations for each copy (same rank order as the sum of all non-G-to-A base substitutions). The number of G-to-A mutations specifically associated with the mA3 or hA3G APOBEC footprints is indicated with hatched grey. (C) Positioned G-to-A mutations (red bars) specifically associated with the hA3G APOBEC footprint (\"GG\") for the individual HERV-K elements in (B); yellow bars correspond to deletions in the proviruses (relative to the Phoenix consensus sequence). The data and the image shown in figure 4C were generated using the Hypermut 2.0 software available at the  website.ConclusionThe restriction effects of APOBEC proteins on endogenous retroelements have essentially concerned retrotransposons with a strictly intracellular life cycle, namely the LINE/SINE non-LTR retrotransposons, and LTR-retrotransposons including the yeast Ty1 element [36,37], and the IAP and MusD murine elements [3,24,26,33]. In these cases severe restriction has been observed, both in ex vivo assays and by in silico analysis of the traces that APOBEC proteins have left through DNA edition in the course of reverse transcription of the retroelements [26]. Here we show that similar effects take place at the level of endogenous retroviruses with an extracellular life cycle, with an unambiguous restriction of the murine IAPE by a murine APOBEC3 protein, and of the human HERV-K element by a human APOBEC3 protein. Taking into account that an infectious IAPE retrovirus with an extracellular life cycle has been the progenitor of the IAP element, the restriction observed for IAP by mA3 appears simply to be the consequence of the restriction that initially controlled the progenitor infectious IAPE invading the rodent ancestor, with the effect being maintained in the evolution of the endogenized IAP retroelements. Although it concerns a heterologous – and therefore not necessarily very relevant-situation, it is noteworthy that the human hA3A protein can control the murine intracellular IAP retroelement, a property not observed for the IAPE infectious progenitor. This is most probably relevant to the localization of the hA3A protein – in the nucleus – and to its rather atypical mode of action – not involving editing of the reverse transcribed DNA – which identifies this restriction factor as more specifically devoted to intracellular retrotransposons, consistent with the absence of reported effects of this factor on – most – infectious retroviruses (reviewed in [2]). Finally, in silico analysis of the genomic copies of the elements demonstrates that APOBEC3 editing has taken place in evolution for these amplified elements, with clear-cut evidence for a severe heterogeneity in the extent of the editing process.MethodsPlasmidsThe human (HERV-K) and murine (IAPE-D) neo-marked ERV copies (pBS CMV-Kcons Stop Env neoAS and pCMV RU5 IAPE neoAS, respectively), the VSV-G and IAPE-D env expression vectors, and the neo-marked autonomous murine IAP retrotransposon (pGL3-IAP92L23 neoTNF) have been previously described [17-19]. The APOBEC3 expression plasmids were obtained from M. Malim (hA3A), the NIH AIDS Research and Reference Reagent program (hA3B, hA3C and hA3F), Open Biosystems (hA3DE), A. Hance (hA3G), and N. Landau (mA3). A plasmid expressing a defective hA3G gene (with a premature stop codon) was used as a negative control. All the APOBEC3 ORF-containing fragments were re-cloned into the pcDNA6 expression plasmid (Invitrogen).Retrotransposition and infection assaysRetrotransposition assays with the neo-marked IAP were as described previously [38]. For the infection assays, 293T cells seeded in 60-mm-diameter plates were transfected using the Lipofectamine Plus kit (Invitrogen) with 4.5 μg of the neo-marked env-defective murine or human ERV, 0.5 μg of the IAPE or VSV-G env expression vector, and 5 μg of the APOBEC3 expression vector to be tested. Supernatants were harvested 48 h post-transfection, filtered through 0.45-μm pore-size PVDF membranes, supplemented with Polybrene (4 μg/ml), and used to infect HeLa target cells by spinoculation (1.200 g for 2.5 h at 25°C). Infection events were detected upon G418 selection of target cells and viral titers quantified as the number of G418R clones per mL of supernatant [18,19].Analysis of integrated proviral DNAsCellular DNA from 20–25 individual G418R clones was used to PCR-amplify a 996 bp fragment encompassing the env to neo gene region (nt 6783–7779) of the IAPE-D element and a 2049 bp fragment spanning the neo to gag gene region (nt 1093–3142) of the HERV-K element (initial 3 min denaturation step at 94°C; 40 cycles: 94°C, 50 sec; 60°C, 50 sec; 68°C, 150 sec). PCR reactions were performed with sets of appropriate primers in 50 μl containing 0.5 μg of cellular DNA, 1× Buffer II and 1.5 U AccuPrime Taq DNA polymerase (Invitrogen). The PCR products were electrophoresed on agarose gels, purified with the Nucleospin Extract II kit (Macherey-Nagel) and a ~800 bp or a ~1600 bp fragment was sequenced (Applied Biosystem sequencing kit) for IAPE-D and HERV-K, respectively.Human and Mouse genome analysesIAPE-D, IAPE-A and HERV-K endogenous retroviruses were extracted from the mouse and human genome sequence databases (Mouse GoldenPath mm8, February 2006 assembly and Human GoldenPath hg18, March 2006 assembly; ) by using as a querying probe the sequence of the previously described functional IAPE-D1 copy [18], the sequence of the IAPE-A copy with intact gag-pol open reading frames (chr14-0436, [18]), and the sequence of the HERV-K element (Phoenix-derived; [19]) used in the cell-based infection assay. Twenty sequences displaying the highest homology to their cognate probe were selected for the IAPE-D and HERV-K elements. Twenty sequences with the highest homology to the IAPE-A sequence and localized on the Y chromosome were also selected to be used as a control (see Results). Alignments were performed using the ClustalW and Editsequence softwares and consensus sequences generated. Quantitative analysis of the nucleotide substitutions within the IAPE-A, IAPE-D and HERV-K elements was performed using Excel and Hypermut 2.0 (available at the  website) softwares, on the full-length retroviruses. The localization of the analyzed sequences within the mouse and human genomes are given in additional file 1.Statistical analysesSignificance levels for the data in Figures 2 and 3 were calculated using the Kruskal Wallis test (GrapPrism software package). More refined analyses for the occurrence of the G-to-A versus C-to-T mutations were performed using a Poisson regression in a log-linear model. The genmod procedure of the SAS software was used (version 9.1, SAS Institute Inc, Cary, NC). The observed distributions of the G-to-A mutations among the GA, GC, GG and GT contexts for HERV-K or the GXA, GXC, GXG and GXT contexts for IAPE were compared to the distribution of these di- or trinucleotides by the chi square test.Competing interestsThe authors declare that they have no competing interests.Authors' contributionsCE, SP and DR carried out the experimental work and drafted the manuscript. OH performed the in silico analyses and drafted the manuscript. TH conceived the study and drafted the manuscript. All authors read and approved the final manuscript.Supplementary MaterialAdditional file 1table 1. localization of the analyzed sequences within the mouse and human genomes.Click here for file\n\nREFERENCES:\nNo References"
+}

batch_5/PMC2532678.json

@@ -0,0 +1,4 @@
+{
+    "id": "PMC2532678",
+    "text": "This is an academic paper. This paper has corpus identifier PMC2532678\nAUTHORS: Rebecca L Tooher, Philippa F Middleton, Caroline A Crowther\n\nABSTRACT:\nBackgroundRecruitment of eligible participants remains one of the biggest challenges to successful completion of randomised controlled trials (RCTs). Only one third of trials recruit on time, often requiring a lengthy extension to the recruitment period. We identified factors influencing recruitment success and potentially effective recruitment strategies.MethodsWe searched MEDLINE and EMBASE from 1966 to December Week 2, 2006, the Cochrane Library Methodology Register in December 2006, and hand searched reference lists for studies of any design which focused on recruitment to maternal/perinatal trials, or if no studies of maternal or perinatal research could be identified, other areas of healthcare. Studies of nurses' and midwives' attitudes to research were included as none specifically about trials were located. We synthesised the data narratively, using a basic thematic analysis, with themes derived from the literature and after discussion between the authors.ResultsAround half of the included papers (29/53) were specific to maternal and perinatal healthcare. Only one study was identified which focused on factors for maternal and perinatal clinicians and only seven studies considered recruitment strategies specific to perinatal research. Themes included: participant assessment of risk; recruitment process; participant understanding of research; patient characteristics; clinician attitudes to research and trials; protocol issues; and institutional or organisational issues. While no reliable evidence base for strategies to enhance recruitment was identified in any of the review studies, four maternal/perinatal primary studies suggest that specialised recruitment staff, mass mailings, physician referrals and strategies targeting minority women may increase recruitment. However these findings may only be applicable to the particular trials and settings studied.ConclusionAlthough factors reported by both participants and clinicians which influence recruitment were quite consistent across the included studies, studies comparing different recruitment strategies were largely missing. Trials of different recruitment strategies could be embedded in large multicentre RCTs, with strategies tailored to the factors specific to the trial and institution.\n\nBODY:\nBackgroundDifficulty with recruitment to randomised controlled trials is a significant obstacle to their successful completion. Trials frequently fail to recruit the number of participants required or require extensions of the recruitment period. A recent study suggests as few as one third of UK trials recruited the required sample size in the planned period for recruitment and another third needed to extend the recruitment period [1]. Such trials may then be underpowered to detect clinically meaningful differences in important outcomes [2], substantially reducing trial precision [3]. If the recruitment period is extended in order to reach the target it is possible that clinical practice may change before the results of the trial become available [2,4]. Problems with recruitment can also lead to selective enrolment, reducing the generalisability of trial results [3].Randomised trials in perinatal medicine face some additional hurdles to successful recruitment. Clinical outcomes of importance may be rare, therefore very large sample sizes are required to detect significant differences in health outcomes for the mother or baby [5]. Consequently, many maternal and perinatal trials need to be multicentre, adding additional complexity to the recruitment task. The need for large sample sizes may also result in situations where the same women and their babies are asked to participate in more than one trial. However, consent for maternal and perinatal trials is often challenging as women and parents are very vulnerable at the time consent is required and may have difficulty in making fully informed decisions [5,6].We reviewed the literature regarding recruitment to maternal and perinatal trials in order to identify barriers and enablers to successful recruitment and strategies which may be effective in enhancing the recruitment effort. This literature review was used to provide an evidence resource for two workshops based on recruitment convened by the WOMBAT (Women and Babies Health and Wellbeing: Action Through Trials) Collaboration in November 2006 and March 2007.MethodsLiterature reviewWe searched MEDLINE and EMBASE from 1966 to December Week 2 2006 and hand searched reference lists of relevant articles and conference proceedings for studies of any design, including qualitative research, which focused on recruitment to perinatal trials. We also searched the Cochrane Library Methodology Register in December 2006. Studies were included in the review if they obtained data from either participants (women and/or parents), clinicians, or others involved in the recruitment of participants for perinatal trials. Studies which focused on the consent process were considered for inclusion, as recruitment and consent in maternal and perinatal research may be closely linked. If no studies of maternal or perinatal research could be identified, studies which focused on recruitment to trials in other areas of healthcare were also included, if it was felt the information would be relevant to maternal and perinatal trialists. Studies of nurses' and midwives' attitudes to research in general were included as no studies specifically about trials were located. Papers which reported primarily anecdotal evidence, opinion or commentary were excluded, as were papers which did not directly add to the body of evidence collated. Such exclusions were determined by consensus after discussion between the authors. Only papers in English were included. One author reviewed all titles and abstracts and identified potentially eligible papers for inclusion in the analysis. These were then checked by a second author for relevance and any disagreements were resolved by discussion. Search terms included subject* or patient* recruit*, enrol*, participat*, enlist*, trial*, study, research, pregnancy, childbirth, neonat*, obstetric*.Demographic details and results for each study were extracted by one author, and checked by a second. As there were no data suitable for statistical pooling, we synthesised the data in narrative form, using a thematic analysis [7], with themes derived from the literature and after discussion between the authors. A thematic analysis \"...involves the identification of prominent or recurrent themes in the literature, and summarising the findings of the different studies under thematic headings\" [7] (p.47).WOMBAT Workshops on recruitment to maternal and perinatal trialsHalf-day workshops were held in November 2006 in Sydney and March 2007 in Melbourne. Participants were trialists with a range of experience from novice to expert. This literature review was used as reference material by presenters who focused on Hot Topic areas in recruitment to perinatal trials. These presentations provided a foundation for the subsequent small group discussion on factors that influence successful recruitment and strategies to improve recruitment. Outcomes of the small group discussion were reported back to all the workshop participants and collected by the authors as a resource for future trialists. These results are reported below.ResultsResults of the literature searchThe studies included in the review have been summarised in Table 1. We included 53 studies, of which 22 used a questionnaire design, 11 used a qualitative design, 11 were systematic or other reviews and 9 reported recruitment data. There were 21 studies which focused on participant (women/families or babies) factors (see Table 2) and 24 studies which focused on clinician factors (see Table 3). We identified eleven studies which considered strategies to improve recruitment, including four systematic reviews. Although more than half of the included studies (29/53) were specific to the maternal and perinatal healthcare context, we identified only one study which focused on barriers and enablers for clinicians working in maternal and perinatal medicine [8] and only seven studies considered recruitment strategies specific to maternal and perinatal research: two reviews [9,10]; one cluster RCT [11]; three observational studies [12-14]; and one before and after design.[15]Table 1Summary of papers included in the literature review*Study topicSystematic and other reviewsQuestionnaires/surveysQualitative studiesRecruitment data recordedParticipants factors (Women)East 2006 [30] n = 600 womenBaker 2005 [22] n = 17 womenJefferies 2006 [31] n = 48 womenRodger 2003a [21] n = 50 womenDaniels 2006 [29] n = 262 womenVnuk 2000 [32] n = 106 womenKenyon 2006 [16] n = 22Mohanna 1999 [24] n = 18 womenRodger 2003a [21]Participants factors (Babies)Burgess 2003 [17] n = 72 parentsHoehn 2005 [34] n = 34 parentsMorley 2005 [35] n = 50 mothers and 48 fathersMason 2000 [18] n = parents of 200 infantsSinghal 2002 [23] n = 52 parents of babies in NICU and 106 parents of babies in normal nurserySnowdon 1999 [33] n = 44 parentsZupancic 1997 [20] n = 140 parentsSnowdon 2006 [19] n = 78 parentsClinician factors (Doctors)Rendell 2006 [48] Cochrane review of 11 observational studiesSinghal 2004b [8] n = 64 doctorsCaldwell 2005 [26] n = 250 paediatricians and 250 physiciansFallowfield 1997 [25] n = 357 oncologists and surgeonsSomkin 2005 [27] n = 199 oncologistsClinician factors (Nurses/midwives)Singhal 2004b [8] n = 50 nursesMeah 1996 [44] n = 32 midwivesHicks 1995 [41] Hicks 1995 [42] n = 395 midwivesBrown 2002 [28] n = 27 women recruitersMcSherry 1997 [45] n = 297 nurses and midwivesRoxburgh 2006 [46] n = 7 nursesAdamsen 2003 [40] n = 79 nursesKuuppelomaki 2003 [43] n = 400 community nursesWatson 2005 [47] n = 485 staffStrategies to improve recruitmentGates 2004 [9] n = 0 studies locatedDuggal-Beri 2006 [11] n = 27 nurses and 22 medical staffHomer 2000 [12] n = 1089 womenHundley 2004 [10] n = 11 studiesKenyon 2000 [15] n = 64 UK maternity hospitals (total births 230000–240000Bryant 2005 [57] n = 3 cross-sectional surveysMoore 1997 [13] n = 3159 women eligible for studyMapstone 2002 [58] n = 15 RCTs or quasi-RCTsPeindl 2003 [14] n = 283 and 306 women screenedWatson 2006 [59] n = 14 studiesTrials in other areas of healthcareAbraham 2006 [39] n = 94 studies of recruitment to surgical trialsAbraham 2006 [49] n = 18 surgeons and 113 patientsGillan 2000 [4] n = 2 trialsFayter 2006 [52] n = 56 studies of cancer trialsLing 2000 [50] n = 1206 patients suitable for palliative care trialsMcDonald 2006 [56] n = 114 trialsMcDaid 2006 [55] n = 8 studies of cancer trialsSullivan 2004 [51] n = 16 health professionalsMinority populationsBartlett 2005 [36] n = 52 trials and 134598 patients (record linkage)Hussain-Gambles 2004 [53] n = 20 health professionals and 75 patientsRuggiero 2003 [37] n = 958 eligible womenWendler 2006 [54] n = 20 studies of consent decisionsRochon 2004 [38] n = 280 RCTs*Perinatal specific papers are shown in bold type. a – Rodger 2003 reports both questionnaire and interview data; b – Singhal 2004 reports data for doctors and nurses.Table 2Participant factors which influence recruitmentThemes and specific findingsUnderstanding of risk▪ Women and parents may underestimate the risks involved in trial participation and overestimate the benefits [16-19,23]▪ Typically risks to the baby dominate over risks to the mother in decisions to participate in trials [19,21-24]Recruitment process and procedures▪ The process of recruiting women and babies into trials has an impact on the decision to participate [17,18,22,23]▪ Communication skills of recruiters are important both to those recruiting and to potential research participants [16,22,24-29]▪ There is conflicting evidence about the provision of written information such as patient information sheets [16-18,22,24]▪ The timing and method of approach may impact on women's and parent's decision to participate [17,19,22,30]▪ Practical issues faced by potential participants include: - work and childcare commitments[28,36] - transport issues [28,36] - privacy and confidentiality concerns [28] - practical concerns about the trial medications/treatments [22,28,31]Participants understanding of the research process and methodological issues▪ Women and parents may not understand specific elements of trial design such as randomisation, blinding and the use of placebos [8,16,17,21-23,32,33]Patient characteristics▪ Characteristics of patients may be related to women's and parent's decisions to participate in research: - altruism [20,22,23,28,34,35] - attitudes and beliefs about research [28,36] - cultural background [12,37,38] - language barriers [12,37,38]Table 3Clinicians factors which influence recruitmentThemes and specific findingsClinicians' attitudes to research and trials▪ Clinicians who are more research oriented are more likely to be involved in trials and to recruit participants [8,26,39]▪ Although nurses and midwives typically report moderate to strong research orientation this does not generally translate into research activity or involvement mainly due to lack of sufficient research training and insufficient time for research [41-47]▪ Doctors who believe trial participation affects patient-doctor relationship are less likely to recruit patients into trials [48]▪ Doctors choosing not to participate in trials may believe that trial participation restricts ability to individualise care [26]▪ Doctors with a strong preference for one or other therapy are less likely to enrol patients in trials [39,49]▪ Evidence unclear whether discussing uncertainty is a barrier to recruitment for clinicians [9,26,39,48]Issues related to the trial protocol and methodology▪ Aspects of trial protocol can affect clinicians' participation in trials and recruitment activity [26,27,39,50,51] - treatments more aggressive than standard - use of placebo - complex protocols - strict eligibility criteria▪ Relevance of trial, especially local relevance [27,39,52]Clinician beliefs about potential participants▪ Younger patients with better prognosis are more likely to be asked to enrol [39,48]▪ Women and babies from minority groups are less likely to be recruited [53,54]▪ Women and babies thought to be unable to participate or to consent are less likely to be recruited [10,39,50,52]Organisational/institutional issues▪ Lack of time is a significant barrier to trial involvement for doctors and nurses [9,25-27,39,41-43,45,46,52,54]▪ A positive organisational culture and material support for trial activity increases clinician participation [9,26,27,39,41-47]▪ A range of practical barriers to setting up and running a trial have been identified [9,10,14,39,50,52,55,56] - identifying and contacting eligible patients - ethics approvals - setting up trial treatment proceduresQuality of included studiesAs the majority of papers included in this literature review were descriptive studies of factors which influence recruitment we have not formally assessed their quality, as there is no generally accepted or validated method of doing so for these study designs. Eleven studies focused on strategies to improve recruitment and hence tested an intervention of some type. The four systematic reviews were of good quality but were limited in their conclusions by the poor quality of the studies included in them and the poorly populated evidence base (so that many of strategies considered were only tested in one or two RCTs). The seven studies which focused on recruitment strategies for maternal and perinatal trials were limited by the designs of the studies used. Although Duggal-Beri [11] was a cluster RCT, the results reported were in abstract form only and did not provide data regarding changes in recruitment rates. One of the two reviews [9], was a commentary and therefore did not provide details of search strategies or inclusion criteria; and the other [10] provided insufficient methodological detail to confidently rule out bias. The other four studies [12-15] did not use a control group for comparison and therefore it is difficult to determine the relative effectiveness of the strategies employed.Participant factors which influence recruitmentPerception of riskParticipant assessment of risk is very important in terms of the decision to consent or refuse participation in RCTs; both for women in deciding about their own participation, and for parents in deciding about the participation of their baby. Participants (particularly parents) frequently underestimate the risks their baby might face by participating in a perinatal randomised controlled trial. A number of studies found that parents thought there would be minimal or no risk for their baby in trial involvement [16-19]. Giving consent and the speed of decision making have also been found to be related to perceptions of low or no risk for the baby [19,20]. In making the decision to participate women and parents weigh up the risks of participation against the possible benefits. Typically, the mother's/parent's duty to their (sometimes unborn) child will be given a higher priority than either consideration of the mother's own health [21] or any altruistic motives that the woman or family may have about research participation [22,23]. When risk is perceived to be too high consent will typically be refused [21,24] possibly with a resultant loss of trust in the doctors providing care [19].Recruitment process and proceduresThe processes by which recruitment is achieved can have an important bearing on women's decisions to participate in maternal and perinatal trials. Recruiters, in particular those also providing women's clinical care, should be aware that women and parents may feel vulnerable and coerced by the recruitment process [17,22,23]. Parents may feel pressured to make a decision quickly, and fear the baby will receive less than optimal care if consent is refused [18,22]. The communication skills of recruiters have been found to impact on the success of recruitment in terms of providing information about a trial to potential participants, obtaining informed consent and discussing uncertainty [23-27]. Developing a personal relationship with the study participant and individualising the recruitment approach for each woman or family may facilitate recruitment and ongoing involvement in research [19,22,28,29].The timing and method of approach also has important implications for the success of recruitment. Consideration of women's situations before presenting research information should ensure that requests for trial participation are not made when a woman is in a particularly vulnerable position [22]. Women and parents may prefer to have information earlier, or have more time to consider their decision to participate in the trial [17,30]. The provision of written information such as Patient Information leaflets may assist in recruiting participants [18,22,24], ideally used together with verbal information to support the relationship between the recruiter and participant [17,22,24]. Careful consideration given to the content of information sheets, with excessive jargon avoided where possible, may also enhance recruitment [17].A range of practical issues which impact on participation were also identified. These included work and childcare commitments, holiday plans, and transportation issues [22,28,31]. Privacy and confidentiality concerns in small communities may also be a barrier to involvement in research [28]. Other practical barriers relate to trial treatment schedules and medications [21,32].Participants' understanding of the research process and methodological issuesWhile potential participants may understand the purpose of the research and the procedures involved, many do not appear to understand why a randomised design has been used or what the implications of this may be [8,17,22,23,33]. Specific elements such as randomisation, blinding, and the use of placebos may be poorly understood, with participants demonstrating a preference for designs which are unblinded [16], do not include a placebo [21], or do not involve random allocation to treatment [32].Individual beliefs and attributes related to decision to participateA range of personal beliefs and attributes may be related to the decision of women and parents to participate in randomised controlled maternal and perinatal trials. Altruism is commonly reported as a reason for research participation [20,22,23,28,34,35]. Positive or negative beliefs about research including the level of trust held in research and clinical governance also influence trial participation [28,36]. Cultural background and language barriers have also been found to influence the participation of women from minority groups [12,37,38].Clinician factors which influence recruitmentClinicians' attitudes to research and trialsThere is a relationship between a clinician's research orientation and their research involvement including recruitment to clinical trials [25,26]. Higher research orientation is generally correlated with research experience, research involvement and recruitment to trials [8,26,39]. Although nurses and midwives typically report a moderate to strong research orientation this does not always translate into research activity or involvement mainly due to lack of sufficient research training and insufficient time for research activities [40-47].Doctors' attitudes and beliefs about trials may affect trial participation and recruitment. Doctors who believe trial participation affects the patient-doctor relationship are less likely to recruit [48]. Doctors who believe trial participation restricts their ability to individualise patient care are less likely to participate in trials [26], as are doctors with a strong preference for one of the treatment arms [39,49]. Clinicians who are motivated to participate in a clinical trial by a personal relationship with the investigator(s) are less likely to recruit than those motivated by other factors [48], suggesting some clinicians may feel pressured to agree to trial participation by their personal acquaintance with researchers, without a having a strong commitment to the trial question or processes involved. Doctors' handling of uncertainty may affect trial participation and recruitment, however, the evidence is conflicting [9,26,39,48].Issues related to the trial protocolAspects of the trial protocol can affect clinicians' participation and recruitment activity. Trials involving treatment more aggressive than the standard treatment, trials involving a placebo, complex trial protocols (that require extra time to learn about eligibility and treatment), and strict eligibility criteria have all been cited as barriers to trial participation by clinicians [26,27,39,50,51]. Relevance of the trial, especially local relevance, was also identified as barrier to participation by clinicians [27,39,52]. Trials with more pragmatic designs in line with standard practice and that were easier to explain to patients and logical extensions of previous trials may increase clinicians' involvement [52].Clinician beliefs about potential trial participantsPatient characteristics may affect clinicians' decisions to offer trial participation or recruit eligible patients. Younger patients and those with better prognosis are more likely to be invited to participate [39,48]. Patients who clinicians believe have higher intelligence are also thought to be easier to communicate with about trials [25]. Clinician gate-keeping of patients judged to be unable to participate in a trial, or provide informed consent at the time of recruitment, has also been identified as a significant barrier to recruitment [10,39,50,52]. Assumptions about the willingness of women from minority backgrounds to participate in trials has also led to an under-representation of these women in research generally and in clinical trials [53,54].Institutional/organisational issuesLack of time is a significant barrier to trial involvement for doctors and nurses/midwives. Specifically clinicians report a lack of time available for recruitment, for data management, to learn about protocol eligibility and treatment requirements, and to obtain informed consent [9,25-27,39-43,45,47,52,55]. Organisational culture and support for trials may also influence participation of clinicians and to an extent participation of women and babies in randomised trials. Lack of expert support staff to handle recruitment and data management has been cited in several studies of barriers to involvement of clinicians in randomised trials [9,27,39]. Furthermore, in one study of recruitment to 114 UK trials [56], better recruitment was significantly associated with the presence of a dedicated clinical trial manager (OR3.8, 95%CI:0.79 to 36.14, p = 0.087). Lack of financial reward, either for individuals or departments involved in trials, together with the expense and financial implications involved in trial participation has also been identified as a barrier to trial involvement [4,9,26,27]. The presence of a culture which encourages and supports randomised trial activity also impacts positively on clinicians' participation in trials. Nurses and midwives commonly report lack of support for research activities from management as a significant barrier to research involvement [41-47].Practical barriers in setting up and running a trial can also impact negatively on recruitment. Such barriers include: identifying eligible patients; trials competing for the same patients; the need to engage and maintain the interest of the whole clinical team in the trial; and a lack of awareness of ongoing trials and eligibility criteria [52,55]. Accurate estimates of the number of patients eligible for participation in trials may also be a barrier to recruitment [56]. In trials which target or include women from minority backgrounds, investigators should keep in mind that it may be difficult to contact the women of interest [13]. Administrative barriers such as problems with staff, the ethics approval process, and implementation of study treatment procedures have also been identified as barriers to clinician involvement in trials and have been found to impact negatively on recruitment [9,56].Strategies to enhance recruitment to randomised trialsEvidence from systematic reviewsThe evidence-base for strategies about improving recruitment to trials is not well populated. Although four systematic reviews were identified [55,57-59], which included 33 unique studies, few strong conclusions were able to be made. None of the reviews could identify any strategies that were clearly found to increase recruitment to trials, but conversely none of the strategies identified could be unequivocally said to be ineffective at increasing recruitment due to small numbers of studies testing each strategy and methodological weaknesses. Only one of the studies included in the four reviews addressed a maternal or perinatal topic.A number of strategies were identified as being possibly effective in the four reviews (Table 4). However, the evidence for these strategies was either weak, or conflicting. Furthermore, conclusions about the strategies sometimes differed between the reviews due to differences in their inclusion criteria. Two of the reviews [55,57] did not identify any effective strategies. One review [59] found that recruitment was increased in trials without a placebo arm or which were not blinded. However, the other review [58] found no difference in recruitment in two studies which used no placebo or a partial randomisation design. Sending a questionnaire related to the study with the request to participate increased recruitment to a home safety trial [58,59]. A telephone reminder to non-respondents increased recruitment in one study [59]. Financial incentives were found to increase recruitment of teenage girls to a quit smoking intervention and helped to retain them in the study [58,59]. However, financial incentives to general practitioners did not increase recruitment to two primary care trials and when surveyed, financial incentives were considered of minor importance by recruiters [57]. Socioculturally specific interventions such as training lay recruiters to recruit women from particular ethnic minorities and multifaceted interventions increased recruitment in two trials, however, the increase in recruitment was small compared to the effort required [59].Table 4Summary of systematic review findings about recruitment strategiesStrategies that might improve recruitment (but evidence weak or conflicting)▪ Using a trial design without blinding or a placebo group▪ Sending a questionnaire related to the trial with the request to participate▪ Telephone reminder to non respondents▪ Financial incentives for participants▪ Interventions tailored to meet the needs of particular minority groupsStrategies not shown to significantly improve recruitment▪ Warning potential participants about an impending request for participation▪ Using a personalised letter together with a flyer▪ Changing information available to potential recruits▪ The professional background of the recruiter (doctor vs nurse)▪ Visiting trial sites to encourage recruitment▪ Changes to consent process▪ Collecting patient trial data by internet vs paper methodsThis table summarises results from four systematic reviews which altogether considered 33 unique studies comparing different recruitment strategies. [54,57-59] The reviews all noted that there is insufficient evidence to make strong conclusions about the relative effectiveness of different strategies.All four systematic reviews also identified a number of strategies which did not significantly alter recruitment rates. All authors highlighted the need to reserve judgement about the effectiveness of these strategies due to methodological weaknesses in the studies and the small size of the evidence base for each (often only one study).These strategies included:• warning potential participants about an impending request for participation• using a personalised letter together with a flyer• changing information available to potential recruits• the professional background of the recruiter (doctor vs nurse)• visiting trial sites to encourage recruitment• changes to consent process• collecting patient trial data by internet vs paper methodsEvidence about strategies to improve recruitment to perinatal trialsA cluster randomised trial using DVD training about consent for a trial of intravenous immunoglobulin for neonatal sepsis resulted in high levels of confidence and knowledge about the trial [11]. The study was reported only as an abstract and did not report effects on recruitment or results of the control groups. A training intervention for midwives recruiting to a trial about antibiotics for premature rupture of the membranes (in which local midwives were employed for 3 hours per week to provide training and motivation to local staff about the trial) resulted in a significant increase in the number of women recruited out of all births (from 0.31% prior to the intervention to 0.68% after, p < 0.0001) [15]. A study of recruitment to two postpartum mental health trials (a treatment and a prevention trial) found that referral from a health professional accounted for almost half of participants for both prevention and treatment trials, and in the treatment trial 32% of participants came from mass mailing. However, media appearances and advertisements on local TV and radio resulted in many women being screened but did not translate into a large number of extra participants [14].One non-systematic review of factors affecting recruitment to multicentre trials in maternal and perinatal health found no high quality evidence on which to base recommendations about strategies to improve recruitment [9]. It was suggested that recruitment would be improved if clinicians were trained to regard recruitment to trials as part of their normal clinical duties, and noted that the strategies used need to be targeted to the particular barriers associated with each trial or trial site. A review of recruitment to intrapartum studies found problems with all three of the most common strategies used for recruitment [10]. Antenatal recruitment involved a significant delay between enrolment and the intervention and resulted in many women being consented who would subsequently be found ineligible for trial participation. Recruitment and randomisation during labour was subject to a significant degree of clinician gate-keeping so that many potentially eligible participants were not approached for trial participation. Staged randomisation in which consent was gained antenatally but randomisation was done after labour commenced or at the time of the intervention resulted in substantial pre-randomisation losses. It was not possible to determine whether this was a result of participants changing their mind or clinician gate-keeping [10].Two observational studies focused on strategies to increase recruitment of minority women (either from non-English speaking or low-income backgrounds) to maternal and perinatal trials [12,13]. Both studies found that a range of specific strategies to increase recruitment of these women resulted in high levels of participation and a sample representative of the population from which the women were drawn. Strategies included: engagement of the community; use of interpreters and translation of written materials; financial incentives; and multiple approaches and pre-warning of impending requests for participation [12,13].Strategies identified at WOMBAT workshops on trial recruitmentThe factors which influence recruitment and strategies for improvement that were identified in the literature are broadly similar to those collected by WOMBAT during the two workshops. Table 5 summarises strategies discussed to improve recruitment at these workshops.Table 5Summary of strategies to improve recruitment discussed in WOMBAT Collaboration workshopsFor participants: • provide information that is important to participants (not researchers) • use a personalised approach in terms of method and timing of approach and request for consent to participate • make it easy to participate by making trial protocol not too onerous • use different methods for reaching participants including pamphlets, telephone and mass media • increase transparency of information provided about treatment and research to help potential participants understand need for trial • assure potential participants there will be no compromise in care if they choose not to participateFor clinicians and participating clinical units/centres: • education for staff including communication skills training for potential recruiters • provide feedback and information to all involved • recognition of contribution through acknowledgement in any publications or where appropriate joint authorship • incentives (usually tangible) for reaching recruitment targets • use of mass media • communication and support • make recruitment easy for recruiters • develop an important and clinically relevant question • have multidisciplinary input (especially unit directors) into trial design to help ensure 'buy-in' from all relevant stakeholders • reduce impact of participation on units by having a dedicated research team and a centrally funded researcher who specific role is trial recruitment • build relationships within participating centres by identifying and nurturing local contacts and local facilitators and also people likely to influence othersOrganisational culture • research should be seen as 'standard care' therefore recruitment to clinical trials seen as normal part of clinical practice • institution is committed to high quality researchThis table summarises results from two workshops held by the WOMBAT Collaboration which focused on recruitment in November 2006 and March 2007.DiscussionThere is considerable consistency in the types of factors reported by women and families which influence their participation in maternal and perinatal trials. A key theme identified was participant estimation of risk, which was found to be frequently underestimated leading them to believe that there would be little or no risk involved in trial participation. Women and parents were also found to have difficulty in understanding some of the methodological aspects of trial design, in particular the use of randomisation and placebos. It appears that parents may not clearly differentiate research and treatment [19]. This lack of understanding points to the need for better communication about trial aims and design. However, it is not possible to say whether improvements in communication of information about trials will lead to increased recruitment. Indeed, there was a suggestion from the evidence that increasing the information available to potential participants could result in fewer agreeing to participate in perinatal trials. However, better informed participants may be more likely to remain in the trial and adhere to trial treatment schedules, although this is also currently unknown.Practical barriers to trial involvement such as difficulties with childcare and transportation appear to be relatively easily ameliorated. However, whether simple financial incentives would be sufficient to compensate women and families for trial participation remains unclear. The beliefs of minority women and families about participation in maternal and perinatal trials also require further study. Women and babies from ethnic minorities are likely to be under-represented in perinatal research including trials, despite the fact that these populations often bear a higher burden of disease for many conditions than the general population that is usually recruited for clinical trials.The studies reviewed also suggested considerable consistency in the factors reported by clinicians which influence their recruitment efforts. Unfortunately a lack of research in this area limits our ability to determine whether there are specific issues for health professionals caring for women and their babies during and after pregnancy. It does seem that more generic issues relating to the design of the trial and some of the organisational issues identified are likely to be shared by clinicians in every area of healthcare. Some of these issues, in particular, those related to the design of the trial protocol could be addressed by investigators with relatively little additional effort, and probably no real increase in costs. In designing trial protocols, investigators could aim to use simpler rather than more complex designs and ideally develop research questions which are relevant in the localities in which the trial is to be conducted [27,39,52]. When possible, the trial design could use standard care as the basic treatment model, and limit the amount of clinical practice behaviour change which the trial would require [27,39]. Taking into consideration these types of issues at the design phase of the study may also have the added benefit of making the trial easier to manage with less risk of protocol violations.A recruitment plan could be developed which takes into consideration the types of information to be conveyed to potential participants and the timing of requests for participation and for obtaining consent, as this has been shown to impact on women's and families' decisions to participate [17,19,22,30]. Although, we currently have no direct evidence about the impact of such a recruitment plan on the participation of clinicians in trials, it is likely that a recruitment process which fits into the standard clinical practices of the units responsible for doing the recruiting will result in more requests for participation. Investigators may need to spend some time exploring with recruiters exactly what the recruitment plan should be. A more carefully developed and streamlined recruitment process may also reduce the time demands associated with trial participation and therefore increase the willingness of clinicians to participate in trials.Support from a clinical trial coordinator or research nurse with responsibility for trial recruitment was found to be positively linked to recruitment in two studies (one in a perinatal context) [15,56]. However, this strategy has not been tested in a randomised trial design. A supportive organisational and professional culture for research and trial activity, was suggested by many authors as a key to improving recruitment. What constitutes a positive research culture and how such a culture can be encouraged remains undefined. Furthermore, changing organisational and professional culture is likely to be beyond the influence of most individual investigators. Here networks of researchers and clinicians are needed, with support from organisational management. Time for research is a barrier which can really only seriously be tackled by changes in the way healthcare is organised and delivered.The extent to which maternal and perinatal trials are reliant on funding from public rather than industry or commercial sources, is likely to limit the amount of money which can be directed towards improving the recruitment effort. Therefore, it is important that research is undertaken to determine the best strategies or mix of strategies to use to improve recruitment. Trials or studies of different recruitment strategies could be nested within larger clinical intervention randomised trials. Strategies which could be tested include:- timing of requests for consent (timing would depend on the nature of the trial);- use of financial or other compensatory incentives for either participants and/or clinicians or hospital units;- provision of information for participants especially about risk;- increasing support available for trial recruitment through the provision of a recruitment officer, or protected time for recruitment.The strategies to be tested should ideally be tailored to the factors which influence recruitment identified for a specific trial and location. A tool which assists investigators to identify these factors for both for participants and clinicians or units, would assist in the selection of appropriate strategies to include in the trial recruitment plan. A similar tool has been developed by the National Institute of Clinical Studies (NICS) in Australia for diagnostic assessment of barriers to the implementation of best available evidence into clinical practice [60].The NICS Barrier tool enables users to work systematically through the process of identifying which people or groups of people are responsible for a particular practice and then determining what the barriers to change for each group may be. Users may work individually, or preferably in a small group, to brainstorm these issues and then identify strategies which would address the barriers identified. The Barrier Tool is accompanied by a number of information sheets and a booklet describing methods of obtaining information about barriers, including survey, consensus processes and interview techniques. The WOMBAT Collaboration is currently working on modifying the NICS Barrier Tool to create a trial recruitment tool.ConclusionThe factors we identified which influence recruitment for both participants and clinicians were quite consistent across the included studies. However, studies which compared different strategies were largely missing from the literature. Trials of different recruitment strategies could be embedded in large multicentre RCTs, to enable assessment of this area with minimal additional effort. Ideally the strategies used should be tailored to the factors specific to the trial and institution. Such methodological research is urgently needed to provide the evidence base for effective strategies to optimise recruitment into randomised trials.Competing interestsThe authors declare that they have no competing interests.Authors' contributionsRT participated in the design of the study, carried out data collection and data extraction, performed the data synthesis, and drafted the manuscript. PM carried out data extraction and both PM and CC participated in the design of the study, the interpretation of data and critically revised the manuscript for important intellectual content. All authors have given final approval of the version to be published.Pre-publication historyThe pre-publication history for this paper can be accessed here:\n\nREFERENCES:\nNo References"
+}

batch_5/PMC2532684.json

@@ -0,0 +1,4 @@
+{
+    "id": "PMC2532684",
+    "text": "This is an academic paper. This paper has corpus identifier PMC2532684\nAUTHORS: Susan Bassham, Cristian Cañestro, John H Postlethwait\n\nABSTRACT:\nBackgroundGene duplication provides opportunities for lineage diversification and evolution of developmental novelties. Duplicated genes generally either disappear by accumulation of mutations (nonfunctionalization), or are preserved either by the origin of positively selected functions in one or both duplicates (neofunctionalization), or by the partitioning of original gene subfunctions between the duplicates (subfunctionalization). The Pax2/5/8 family of important developmental regulators has undergone parallel expansion among chordate groups. After the divergence of urochordate and vertebrate lineages, two rounds of independent gene duplications resulted in the Pax2, Pax5, and Pax8 genes of most vertebrates (the sister group of the urochordates), and an additional duplication provided the pax2a and pax2b duplicates in teleost fish. Separate from the vertebrate genome expansions, a duplication also created two Pax2/5/8 genes in the common ancestor of ascidian and larvacean urochordates.ResultsTo better understand mechanisms underlying the evolution of duplicated genes, we investigated, in the larvacean urochordate Oikopleura dioica, the embryonic gene expression patterns of Pax2/5/8 paralogs. We compared the larvacean and ascidian expression patterns to infer modular subfunctions present in the single pre-duplication Pax2/5/8 gene of stem urochordates, and we compared vertebrate and urochordate expression to infer the suite of Pax2/5/8 gene subfunctions in the common ancestor of olfactores (vertebrates + urochordates). Expression pattern differences of larvacean and ascidian Pax2/5/8 orthologs in the endostyle, pharynx and hindgut suggest that some ancestral gene functions have been partitioned differently to the duplicates in the two urochordate lineages. Novel expression in the larvacean heart may have resulted from the neofunctionalization of a Pax2/5/8 gene in the urochordates. Expression of larvacean Pax2/5/8 in the endostyle, in sites of epithelial remodeling, and in sensory tissues evokes like functions of Pax2, Pax5 and Pax8 in vertebrate embryos, and may indicate ancient origins for these functions in the chordate common ancestor.ConclusionComparative analysis of expression patterns of chordate Pax2/5/8 duplicates, rooted on the single-copy Pax2/5/8 gene of amphioxus, whose lineage diverged basally among chordates, provides new insights into the evolution and development of the heart, thyroid, pharynx, stomodeum and placodes in chordates; supports the controversial conclusion that the atrial siphon of ascidians and the otic placode in vertebrates are homologous; and backs the notion that Pax2/5/8 functioned in ancestral chordates to engineer epithelial fusions and perforations, including gill slit openings.\n\nBODY:\nBackgroundOhno's classical model to explain the fate of genes after gene duplication [1] predicts that one gene duplicate preserves the original gene function while its paralog either disappears by accumulation of detrimental mutations (called nonfunctionalization [2]) or occasionally acquires beneficial mutations that confer novel, positively selected functions (called neofunctionalization [2]). The duplication-degeneration-complementation (DDC) model predicts a third alternative, in which the two duplicate genes become permanently preserved as a consequence of complementary, degenerative mutations that result in partitioning of ancestral subfunctions, so that the sum of the functions of two paralogs equals the functions of the original gene prior to the duplication [2,3]. The DDC model also predicts that after the initial preservation of the two duplicates, whether by subfunctionalization or from neofunctionalization, further partitioning of redundant subfunctions may occur. Studies show that functional constraints on genes duplicated in whole-genome duplications are relaxed, compared with singletons, for tens of millions of years [4]. In addition, novel functions may originate over time, their evolution facilitated by the relaxation of pleiotropy occasioned by fewer tasks in each descendent gene duplicate compared with its single copy gene ancestor [2,5-9].Understanding the evolution of duplicated genes is important because of the hypothesis that gene duplicates provide opportunities for the evolution of reproductive barriers that lead to lineage divergence [10], and for the origin of evolutionary novelties [1,11,12]. In vertebrates, many sets of paralogous genes arose in two rounds of genome duplication (R1 and R2) that took place in early vertebrate evolution [12-18]. An additional round of genome duplication (R3) occurred in the teleost lineage after ray-fin fish diverged from lobe-fin fish, and provided additional gene family members observed in many fish models [19-23]. It has been suggested that the R1 and R2 genome amplifications facilitated the origin of vertebrate innovations [24], and the R3 event may have facilitated the teleost species radiation [9,25].Non-vertebrate chordates often have single copies of vertebrate gene families because their lineages diverged from the vertebrate lineage before the R1 and R2 genome duplication events. Recent phylogenomic analyses converge on the conclusion that the chordate subphylum Urochordata, which includes the classes Larvacea and Ascidiacea, are the closest living relatives of the vertebrates, constituting the group Olfactores (vertebrates + urochordates), while the subphylum Cephalochordata, including the amphioxus, diverged basally among chordates ([7,26-30], reviewed in [9]). As gene duplication is pervasive [31,32], non-vertebrate chordates sometimes have genes that duplicated independently in the cephalochordate [33-35] or urochordate lineages [35-41].We propose that the comparative analysis of gene expression patterns in a gene family that experienced lineage-specific, independent duplication events, interpreted in a phylogenetic context and with respect to subfunction partitioning, can help in the inference of ancient gene functions and in the identification of gene functions that arose by neofunctionalization, and thus may be important for lineage divergence and the origin of developmental novelties. To test this proposition, we examined the Pax2/5/8 gene family. Pax2/5/8 genes encode transcription factors with conserved motifs, including a paired domain, homeodomain, and octapeptide, and are associated with mechanosensory development in mammals [42], frogs [43], fish [44,45], flies [46,47], ascidians [48] and in mollusks [49], while in a cnidarians, the apparent Pax2/5/8 homolog is associated with nerve and sensory cell differentiation [50,51].Pax2/5/8 genes duplicated independently in different chordate lineages. Within the vertebrates, tetrapods have three members of the Pax2/5/8 gene family (Pax2, Pax5 and Pax8), and teleosts additionally have duplicate pax2 genes ([45] and references therein). In contrast to vertebrates, the basally diverging cephalochordate amphioxus possesses a single Pax2/5/8 gene that is equally related to Pax2, Pax5 and Pax8 [52], and urochordates have two Pax2/5/8 genes (Pax2/5/8a and Pax2/5/8b) [38,53,54] that originated in a duplication event prior to the divergence of larvacean and ascidian lineages [38]. The chordate Pax2/5/8 family embodies a full spectrum of gene evolutionary events: non-duplication (in amphioxus); independent duplications (within urochordates and vertebrates); gene loss (for example, tetrapods have just three of the four paralogs expected from two rounds of genome duplication); neofunctionalization (for example, vertebrate Pax5 in lymphocyte development [55]); and ancestral subfunctions appear to have been partitioned between paralogs within a lineage and, further, differently partitioned between lineages (for example, Pax2 and Pax8 in fish and mammal thyroids [56]).The independent duplication of the stem olfactores' Pax2/5/8 gene in vertebrate and urochordate lineages provides replicate evolutionary experiments to explore the principles of subfunction partitioning and the origin of novel functions. To exploit this opportunity, we provide here a detailed description of expression patterns for Pax2/5/8 paralogs during development of the larvacean urochordate Oikopleura dioica. We then compare our results with expression patterns of Pax2/5/8 orthologs in ascidians, with independently duplicated Pax2/5/8 paralogs in vertebrates, and with the non-duplicated Pax2/5/8 gene in amphioxus as an outgroup. We discuss the expression of the Pax2/5/8 gene family during development of the heart, endostyle, pharynx, and sensory organs, and provide new insights that reconcile previous conflicting hypotheses about the homology of the ascidian atrial primordia and the vertebrate otic placode. This work thus illustrates the power of comparative analyses of independently duplicated genes to infer ancestral gene subfunctions, modules that can segregate independently from each other to evolving gene duplicates.ResultsFunctional motif variation and gene structure of chordate Pax2/5/8 paralogsPax2/5/8 proteins are transcription factors, whose sequence is poorly conserved across Chordata outside of two functional motifs: the paired domain, which interacts with the DNA of target genes (dark blue in Figure 1), and an octapeptide motif (dark green in Figure 1), which is conserved with other Pax proteins [57], functions in repression of Pax transactivation [58], and interacts with other transcriptional cofactors [59]. Sequence alignment of various chordate Pax2/5/8 proteins (Figure 1) reveals that while the sequence of the DNA-interacting paired domain is highly conserved across all chordate Pax2/5/8 proteins analyzed, the octapeptide motif is less conserved. In urochordates, for instance, the octapeptide is present in the ascidian Halocynthia roretzi [48], and, contrary to a previous report [60], our alignment reveals it is also present in Ciona intestinalis Pax2/5/8a (Figure 1A). The octapeptide sequence of the Oikopleura Pax2/5/8a gene is poorly conserved, and the octapeptide motif is absent from the expected position in all urochordate Pax2/5/8b proteins (Figure 1A). Interestingly, we have identified a new motif (light green in Figure 1B) that is conserved among all ascidian Pax2/5/8b paralogs, located further toward the carboxyl end. As this newly recognized sequence shows some similarities with the octapeptide motif, we have called it the 'octapeptide-like' motif (bottom Figure 1A). We could not identify an octapeptide-like motif in the Oikopleura Pax2/5/8b. Our alignment also reveals a lysine-arginine-rich domain (in red in Figure 1) that is conserved in vertebrate Pax2 and amphioxus Pax2/5/8, is variable among vertebrate Pax5 and Pax8 paralogs, and is present in urochordate Pax2/5/8b but not Pax2/5/8a. This protein motif marks the beginning of an amino acid range identified as important for greatly increasing vertebrate Pax2 protein's transactivation activity [58].Figure 1Comparison of chordate Pax2/5/8 proteins and gene structures. (A) Alignment of the chordate Pax2/5/8 proteins showing the conserved DNA-binding paired-domain (dark blue), the octapeptide motif (dark green), the octapeptide-like motif (light green), and the lysine-arginine (KR) rich region (red). Arrowheads indicate the positions of introns. (B) Exon-intron organization deduced from the comparison of ESTs and genomic regions available in public databases (Ghost: , JGI: , and NCBI: ). Numbers indicate the length of the exons (boxes) in base pairs. The position of the conserved domains shown in (A) is indicated with the same code of colors. Exons containing the putative beginning of the coding sequence are labeled in orange. Exons with low degree of sequence conservation and which are hardly alignable among different organisms are labeled in light blue. Exon regions containing 5' and 3'UTR are labeled in grey. Analysis of EST sequences suggested the presence of multiple splice variants, revealing exons (in pink) that were absent from other EST sequences for the same gene; these alternates include the exon harboring the poorly conserved octapeptide motif from Ciona intestinalis. The arrow indicates the totally conserved position of the intron within the paired domain. For a phylogenetic analysis of the chordate Pax2/5/8 proteins, see Additional file 1. Bfl: Branchiostoma floridae; Cin: Ciona intestinalis; Csa: Ciona savignyi; Dre: Danio rerio; Hro: Halocynthia roretzi; Hsa: Homo sapiens; Odi: Oikopleura dioica; Pma: Phallusia mammillata.Comparison of gene structures revealed that, while most chordate Pax2/5/8 genes have 8 to 11 exons, ascidian Pax2/5/8b paralogs have 20 exons, which code for a protein of about 1300 amino acids, three times longer than the approximately 400 amino acid-long Pax2/5/8 proteins from other chordates. This difference in size is due mainly to an exon more than 1200 nucleotides long located upstream of the paired domain. Our analysis of publicly available EST sequences confirmed the presence of this large exon in gene models predicted for Pax2/5/8b in Ciona intestinalis (EMSBL ID ENSCINT00000012344) and in Ciona savignyi (ENSCSAVG00000001640) (data not shown). This large exon is absent from O. dioica Pax2/5/8 paralogs and may have evolved in the ascidian lineage after the divergence of urochordates.In conclusion, various Pax2/5/8 paralogs appear to have lost ancestral features and evolved new structural motifs or new exons, as would be predicted by the DDC model applied to Pax2/5/8 paralogs evolving after independent gene duplication events. The identification of these structural features focuses attention on regions to test for function to learn the roles each motif may have played in the origin of lineage-specific morphologies.Developmental expression of the Oikopleura paralogs Pax2/5/8a and Pax2/5/8bTailbud stagesAt the end of gastrulation, Pax2/5/8a and Pax2/5/8b were both broadly expressed in the Oikopleura embryo. Pax2/5/8a was expressed mainly in the ectoderm of the trunk (Figure 2A), while Pax2/5/8b was expressed primarily in the interior of the trunk (Figure 2B). In early tailbud stages, Pax2/5/8b expression continued in its broad internal expression domain (Figure 2D), but Pax2/5/8a expression became restricted to two domains: a few medial cells at the boundary of the anterior brain and the pharynx (Figure 2C,C' yellow arrowheads), and a bilateral pair of ectodermal cells in the posterior trunk (Figure 2C,C' red arrowheads). By mid- and late-tailbud stage, the signal of the Pax2/5/8a expression domains strengthened (Figure 2E,E' yellow and red arrowheads), and a new domain appeared in the rostral-most ectoderm of the trunk (Figure 2E, E\" black arrowhead), probably labeling the first two cells fated to invaginate into the stomodeum and form the oral epithelium (see below; Figure 3A).Figure 2Expression of Pax2/5/8 paralogs at tailbud stages in Oikopleura dioica. Whole mount in situ hybridization of Pax2/5/8a (A, C, E) and Pax2/5/8b (B, D, F) at incipient-tailbud stage (A, B), early-tailbud stage (C, D), and mid-tailbud stage (E, F) in lateral views (anterior to the left and dorsal to the top). Specific aspects of the expression domains are shown in ventral (A', C' and E') and frontal views (E\") in optical sections in the plane of the dashed white lines. Colored arrowheads label expression domains (black, stomodeum, st; yellow, anterior pharynx, ph; red, placodal precursor of the Langerhans receptors, LR). Scale bar = 10 μm.Figure 3Expression of Pax2/5/8 paralogs in Oikopleura dioica hatchlings. Whole mount in situ hybridization of Pax2/5/8a (A, C) and Pax2/5/8b (B, D) in early-hatchling stage (A, B) and mid-hatchling stage (C, D). Lateral (A, B, C, D) and ventral (A', B', C' and D') views for each stage. Anterior is to the left and dorsal to the top. Insets (A\", B\", and C\") show details of the regions on the surface of the embryo where the Langerhans receptors eventually form. Colored arrowheads label expression domains (black, stomodeum, st; yellow, anterior pharynx, ph; blue, posterior pharynx, ph; white, endostyle, en; purple, presumptive cell precursors of the gills, g; red, placodal precursor of the Langerhans receptors). The position of the anterior tip of notochord (n) is demarcated (white line). Scale bar = 10 μm.Early hatchling and mid-hatchling stagesAt the early hatchling stage, Pax2/5/8a expression maintained its late-tailbud stage pattern. The rostral-most expression domain in the stomodeal region expanded internally from the surface to form a group of contiguous internal cells (Figure 3A, black arrowheads), which were separated from the internal expression domain in presumptive pharynx/endostyle precursor cells (Figure 3A, yellow arrowhead). The bilateral Pax2/5/8a expression domain in the posterior part of the trunk included at least four ectodermal cells in the area where the Langerhans mechanoreceptors eventually develop (Figure 3A, red arrowheads), at a level slightly anterior to the tip of the notochord and adjacent to the homolog of the vertebrate hindbrain [38]. During early hatchling stages, the internal broad expression domain of Pax2/5/8b that was present at tailbud stages began to fade at the same time that two separate expression domains became distinct, one in the ectoderm at about the same bilateral position as the Pax2/5/8a ectodermal domain (Figure 3A', B', red arrowheads) and the other within the pharynx (Figure 3B, yellow arrowhead). Although Pax2/5/8a and Pax2/5/8b were both expressed in the ectoderm at the eventual position of the Langerhans receptors, close inspection revealed differences in the number and morphology of the Pax2/5/8a and Pax2/5/8b positive cells (Figure 3A\", B\"). By mid-hatchling stages, the bilateral Pax2/5/8a expression domain expanded to at least six ectodermal cells (Figure 3C\"), while the brief flash of Pax2/5/8b expression in the ectoderm became undetectable (Figure 3C', D', purple arrowheads). Simultaneously, the distal portion of each gill pouch began to express Pax2/5/8b. It is not known if ectodermal cells contribute to the formation of the gill pouches at these early stages when the ectoderm is still forming its epithelial character.Expression of the two Oikopleura Pax2/5/8 paralogs during mid-hatchling stages appeared to be complementary in the pharynx in time and space: while Pax2/5/8a expression was diminishing in the stomodeum and the rostral pharynx (Figure 3C, C'), new Pax2/5/8b expression domains appeared (Figure 3D, D', black and yellow arrowheads). The proximal endoderm of the gill pouches began to express Pax2/5/8a while the distal portion of the gill pouches expressed Pax2/5/8b (Figure 3C, C', D', blue arrowheads). Presumptive dorsal precursor cells of the endostyle first began to express Pax2/5/8a, while presumptive ventral endostyle cells expressed Pax2/5/8b (Figure 4C, C', D, white arrowheads).Figure 4Expression of Oikopleura dioica Pax2/5/8 paralogs during organogenesis. Whole mount in situ hybridization of Pax2/5/8a (A, C) and Pax2/5/8b (B, D) in late hatchling stages during the initial (A, B) and advanced state (C, D) of the expansion of internal organ cavities during organogenesis. Lateral aspect (A, B, C and D) and integrated ventral view (A', B', C' and D') for each stage are shown. Anterior is to the left and dorsal to the top. Details of the Pax2/5/8 expression domains in the gills (dashed squares and lines in C and D) are shown in lateral (C\" and D\") and frontal (C\"', D\"') views. The position of the ciliary rings (cr) is labeled with an arrow. Colored arrowheads label some of the domains (black, stomodeum, st; yellow, anterior pharynx, ph; blue, posterior pharynx; white, endostyle, en; purple, gills, g; orange, medial wall of the heart, h; green, anus and rectum, r; red, base of the Langerhans receptors, LR, and posterior part of the trunk). Scale bar = 10 μm.Late hatchling stagesBy the late hatchling stage, organs are distinct and internal cavities have started to expand. At this stage, the gills and mouth have opened to the outside, and the heart and the ciliary rings of the gills have both begun to beat. The direct development of larvacean urochordates permits the tracing of gene expression domains in organ rudiments from hatchlings until fully mature adults. In contrast, in ascidian urochordates, many organs begin to develop only at metamorphosis, just as many chordate-specific characters disappear.Pax2/5/8a expression in late hatchling stages overlaps Pax2/5/8b in some domains: in the distal part of the gills from the ciliary rings (cr) to the external gill opening (Figure 4, purple arrowheads), and, transiently, in the posterior trunk epidermis (Figure 4A, B, C, red arrowheads). Other domains are unique to each gene, and in several regions expression of the paralogs appears complementary. For example, while the lips of the mouth express Pax2/5/8a (Figure 4A, C, black arrowheads), the pharynx just internal to the lips strongly expresses Pax2/5/8b (Figure 4B, D, black arrowheads); while dorsal cells of the endostyle express Pax2/5/8a (Figure 4A, C, white arrowheads), ventral cells of the endostyle express Pax2/5/8b (Figure 4B, D, white arrowheads); and while the most posterior section of the rectum expresses Pax2/5/8a (Figure 4A, C, green arrowheads), the anus expresses Pax2/5/8b (Figure 4D, green arrowhead). Except for the distal part of the gills, the posterior pharynx exclusively expresses Pax2/5/8b, including the gill endoderm where the two gill pouches meet medially (Figure 4B, D, blue arrowheads). And, strikingly, only one layer of the two-ply heart, the pericardium, expresses Pax2/5/8b (Figure 4B', D', orange arrowheads); the muscular myocardium does not.A subtle metamorphic event called 'tailshift', the rapid reorientation of the tail from a posterior to a ventral position with respect to the trunk, signals the beginning of the juvenile stage in Oikopleura. Just prior to tailshift, the two Pax2/5/8 paralogs continue to show mostly non-overlapping expression patterns, now in clearly differentiated organs. Pax2/5/8a (Figure 5A) is still expressed around the mouth (including ciliated sense organs in the upper lip), in the dorsal, iodine-binding corridor cells of the endostyle, in the gill endoderm, in the rectum, and in a row of cells at the posterior margin of the oikoplastic epithelium (a specialized secretory tissue that covers much of the trunk). Expression of Pax2/5/8b (Figure 5B) is also a continuation of domains identified earlier, including the pharynx and gills, the ventral endostyle, the anus, and the pericardium. New domains, however, now appear in dorsal, ventral and lateral fields of the oikoplastic epithelium, the tissue that will soon secrete the first filter-feeding 'house' the animal inhabits [61]. Figure 5C is a schematic diagram summarizing the expression domains of both Oikopleura Pax2/5/8 paralogs in late hatchlings. Table 1 summarizes by tissue and developmental stage all Pax2/5/8 expression domains.Figure 5Expression of larvacean Pax2/5/8 paralogs at pre-tailshift stage, when organogenesis nears completion. (A) Pax2/5/8a. (B) Pax2/5/8b. (C) Schematic representation summarizing the non-overlapping Pax2/5/8a (red) and Pax2/5/8b (blue) expression domains. Arrows indicate perforations where epidermal fusions occur and Pax2/5/8 paralogs are expressed. Abbreviations: a, anus; ab, anterior brain; cr, ciliary ring; en, endostyle; es, esophagus; h, heart; lr, Langerhans receptor; m, mouth; ph, pharynx; r, rectum; s, stomach. Scale bar = 10 μm.Table 1Summary of Pax2/5/8 expression domains during Oikopleura developmentTailbudEarly hatchlingMid-hatchlingLate hatchlingStomodeumaaa/ba/bPharynxaa/ba/bbLangerhans receptorsaa+baaEndostylea/ba/bGillsba/bAnusa/bPosterior trunk epidermisa+bHeartba, Pax2/5/8a; b, Pax2/5/8b; a/b, expression of paralogs does not spatially overlap or only partly overlaps; a+b, both paralogs spatially overlap.DiscussionThese investigations of the structure and expression of Pax2/5/8 gene duplicates in Oikopleura, when analyzed in a comparative context with shared and independent Pax2/5/8 gene duplications in other chordate lineages, illuminate a variety of problems in the evolution of chordate developmental mechanisms and the principles that govern change in duplicated gene function over time. The work resolves alternative hypotheses for the homologies of the ascidian atrial siphon, illuminates evolution of the thyroid, identifies candidates for ancestral gene subfunctions, and defines the origin of functional innovations in this gene family.Pax2/5/8 and the evolution and development of the chordate heartThe shared expression of cardiac developmental genes (for example, Nkx-2.5, dHand, and Mesp1) supports the homology of bilaterian hearts from chordates to flies [62-69]. The evolutionary relationship of the single-chambered heart of non-vertebrate chordates to the individual chambers of the innovative multi-chamber vertebrate heart, however, is still unclear (reviewed in [70]). Our work reveals unprecedented Pax2/5/8 expression in the developing heart of a chordate, and may provide a late ontogenetic marker for assessing homology of tissue layers among simple, urochordate hearts.In contrast to the single-layered heart of cephalochordates, urochordates have a two-layered heart. The simple heart of Oikopleura dioica consists of two epithelial layers lying just medial to the left stomach lobe. In Oikopleura species, the lateral wall of the heart contains the muscle fibers while the medial wall is a thin pericardial membrane (called 'procardium' in [71]). Peristaltic heart contractions, which periodically reverse, cause the hemolymph to course between the heart and the stomach wall [71,72]. In comparison, the ascidian heart rolls up to become tubular rather than planar, but also has a contractile layer and a non-muscular, pericardial layer (reviewed in [70]). The larvacean pericardial membrane, therefore, is the likely homolog of the ascidian pericardium.In ascidians, heart development can be broadly divided in two phases (reviewed in [70]). In the first phase, heart cell precursors become fated during early cleavage and embryonic stages, and after a complex process of muscle cell migration from the tail to the trunk, heart development temporarily arrests. In the second phase, after metamorphosis, heart development re-initiates and after final differentiation, the heart starts to beat and becomes functional. In larvaceans, the transparency and absence of a drastic metamorphosis provides a complementary model system for the study of heart development. In O. dioica, the heart starts beating less than 24 hours post fertilization, before tailshift. In late larvacean hatchlings, when heart differentiation is at or nearing completion, the pericardium expresses Pax2/5/8b. In ascidians, however, neither Pax2/5/8 paralog has been shown to be expressed in the heart, perhaps because most published analyses of Pax2/5/8 expression in Ciona, Halocynthia and Herdmania [48,53,60,73,74] have not included late developmental stages after metamorphosis (3 to 4 days after fertilization). Therefore, whether Pax2/5/8 expression in the heart – and more specifically in the pericardium – is a shared feature among urochordates remains to be investigated.Cephalochordates have a single-layered, muscularized blood vessel ventral to the gut that propels the hemolymph by peristalsis [75,76]. During amphioxus development, the primordium of this muscularized vessel expresses a homolog of vertebrate NK2 class genes – a group that includes several genes necessary for vertebrate cardiac development – suggesting that the amphioxus vessel is homologous to the vertebrate heart [69]. Pax2/5/8 is not reported to be expressed in the circulatory vessels of amphioxus [52], and likewise, amphioxus has no structure that appears to be morphologically comparable to the Pax2/5/8-expressing larvacean pericardium or to the pericardia of ascidians or vertebrates.In vertebrates, mouse Pax2 and Pax5 are not expressed in the heart [77,78], and although Pax8 expression has been detected in the heart of newborn mice, its level was equal to or less than all other tissues tested except kidney [79], suggesting that it may represent background levels. No role for Pax2, Pax5, or Pax8 has been demonstrated in the heart in knockout mutations in mouse [80-82]. Consistent with these results, our survey of EST databases in NCBI  reveals the absence of Pax2/5/8 genes in the heart transcriptomes of human, mouse, frogs and zebrafish. Thus, we can infer that Pax2/5/8 expression in the heart was either already present in the last common ancestor of olfactores (that is, urochordates + vertebrates) and lost in the vertebrate lineage, or was a neofunctionalization event within the urochordates, and perhaps within larvaceans.Evidence of Pax2/5/8 subfunction partitioning in thyroid homologsThe thyroid is an endocrine gland located in the neck ventral to the pharynx in humans and other vertebrates, and it regulates energy production and growth by synthesizing tyrosine-based, iodine-containing hormones (T3 and T4). The presumed homolog of the thyroid in filter-feeding, non-vertebrate chordates is the endostyle, an organ located ventral to the pharyngeal floor that functions in iodine binding and secretion of food-trapping mucus [83]. The homology between the thyroid and the endostyle is further supported by the expression of similar molecular markers, including members of the Pax2/5/8 family ([84] and references therein). Our results show that the larvacean endostyle expresses Pax2/5/8 genes in a pattern similar to that described for amphioxus [52], ascidians [84], and vertebrates [56,85]. The last common ancestor of all three chordate subphyla, therefore, probably employed a Pax2/5/8 gene in the development of an endostyle-like organ that evolved into the endostyle of cephalochordates and urochordates and the thyroid of vertebrates (Figure 6).Figure 6Hypothesis for the evolution of chordate Pax2/5/8 subfunctions. In stem chordates, Pax2/5/8 already played a role during the development of the endostyle, the homolog of the vertebrate thyroid, as well as in controlling genes for making epithelial fusions and perforations (for example, gill openings). The Pax2/5/8 subfunction related to the development of neurogenic placodes (for example, otic placode) appears to be restricted to the olfactores; like vertebrate orthologs, urochordate Pax2/5/8a genes are expressed from just after neurula stage in paired thickenings overlapping the early expression of other placode-marking genes [39]. Inferred origins for some of the functions (circles in stems) performed by the pleiotropic Pax2/5/8 gene family and inferred subfunction partitioning are schematized in the context of the chordate phylogeny. Gene duplication events (stars) in different chordate lineages permitted independent partitioning of endostyle/thyroid (red circles) and otic placode (blue circles) subfunctions among gene paralogs. Half semicircles denote dorsal and ventral expression domains. Pale blue circles denote paralogs whose expression is inferred to have become delayed, transient or spatially narrowed in development of the otic system. White circles indicate inferred subfunction losses. [43,45,48,52,56,73,84,85,118,119].Pax2/5/8 subfunctions related to vertebrate thyroid development have likely suffered independent processes of subfunction partitioning among paralogs (Figure 6). For instance, in mouse, Pax8 is expressed during thyroid development but Pax2 is not [56,85]; in frogs, however, subfunctions of these genes are reversed, as Pax2 but not Pax8 is involved in the development of the thyroid [43]. These results suggest that ancestral vertebrate thyroid regulatory subfunctions of Pax2/5/8 were preserved by both Pax2 and Pax8 in stem amniotes, and resolved independently in the amphibian and mammalian lineages. In zebrafish, both pax8 and pax2a (previously called pax2.1) retain thyroid expression, although they are activated at different times in development [56].Although the duplication of Pax2/5/8 paralogs in urochordates occurred before the divergence of the larvacean and ascidian lineages [38] (Additional File 1), the endostyle of Oikopleura expresses both Pax2/5/8 paralogs (Pax2/5/8a dorsally, and Pax2/5/8b ventrally), while the endostyle of ascidians does not express Pax2/5/8b, but expresses Pax2/5/8a in both the dorsal and ventral domains, the sum of the pattern for the two Oikopleura duplicates [84]. These lineage-specific differences in expression patterns of orthologous Pax2/5/8 duplicates suggest that in the last common ancestor of larvaceans and ascidians, both Pax2/5/8a and Pax2/5/8b were expressed both dorsally and ventrally, and that after lineages diverged, in larvaceans the dorsal and ventral endostyle subfunctions partitioned to different Pax2/5/8 genes, but in ascidians both subfunctions remained associated with the Pax2/5/8a paralog and both were lost by the Pax2/5/8b paralog. As vertebrate Pax8 can directly bind the promoters of thyroid follicle cell-specific genes such as thyroperoxidase [86], we raise the hypothesis that urochordate Pax2/5/8a genes, which are expressed in peroxidase-producing, dorsal cells of the endostyle [87,88], play a homologous role while urochordate Pax2/5/8b genes have lost this function.Evidence of Pax2/5/8 subfunction partitioning in the anterior pharynx and stomodeumComparison between ascidians, larvaceans and cephalochordates suggests that stem chordates expressed Pax2/5/8 in the stomodeum and anterior pharynx, but that this expression domain was lost in the vertebrate lineage. Larvaceans and ascidians both have ciliated mechanosensory organs that ring the mouth and are involved in a filter-feeding, particle-rejection response [39,89,90]. Although the oral sensory cell types themselves are morphologically different in the two urochordate classes, these organs may share a common origin in a stomodeal sensory placode that has been lost in the vertebrate lineage [39,73,90-94]. Although both larvacean Pax2/5/8 paralogs are expressed in this stomodeal domain, their patterns are distinct: Pax2/5/8a expression in the rostral pharynx of early hatchlings is replaced with Pax2/5/8b by mid-hatchling stages. In mid-hatchlings, Pax2/5/8a expression is in the external ectoderm surrounding the mouth opening, including the two bristle-bearing cells of the upper lip, while Pax2/5/8b is expressed just inside the mouth, including the ciliated sensory cells of the circumoral organ. Although the two mechanosensory cell types fall into expression domains of different Pax2/5/8 paralogs, they are innervated by the same branched sensory axons emanating from a pair of rostral brain cells [39,89].The single amphioxus Pax2/5/8 gene is expressed both externally around the developing mouth and internally in the pharynx [52]; this expression domain corresponds to the sum of expression domains occupied by different larvacean paralogs, as expected by subfunction partitioning [2] if the last stem chordate had separate regulatory elements governing the contiguous external and internal pharyngeal domains that partitioned to different larvacean paralogs after the Pax2/5/8 duplication event. Ascidian Pax2/5/8a is expressed in the invaginating stomodeum but Pax2/5/8b is expressed in the buccal cavity, a pattern comparable with the late hatchling expression domains of the Oikopleura orthologs. At least some subfunctions, therefore, appear to have partitioned between Pax2/5/8 paralogs before the larvacean and ascidian lineages diverged. The ascidian Pax2/5/8a gene, however, appears to lack expression comparable with the early, internal expression of Oikopleura Pax2/5/8a in the rostral pharynx, a difference that may derive from the developmental delay and incomplete development of endodermal organs experienced by ascidian larvae compared with the uninterrupted and complete endodermal ontogeny in larvaceans.Pax2/5/8 expression and the Langerhans receptor, ascidian atrial siphon, and vertebrate otic placodeThe expression of Pax2/5/8 genes in the Langerhans receptors of Oikopleura helps us understand apparently conflicting interpretations of the ancestral role of Pax2/5/8 in the origin of the vertebrate otic placode. Previous evidence supports the notion that larvacean Langerhans receptors are homologous to hair cell-like sensory organs in the ascidian atrium, the chamber surrounding the branchial basket [39,73,95]. Expression of Pax2/5/8a in the ascidian atrium and expression of Pax2 and Pax8 in the vertebrate hair cell-producing otic placode suggested the hypothesis that the atrium of ascidians is homologous to the otic vesicles of vertebrates [48].An alternative to the 'placode hypothesis', however, arose from the finding that in amphioxus, the developing gill slits and mouth express Pax2/5/8. This alternative suggests that Pax2/5/8 expression in the atrial primordium of ascidians reflects an ancient gene function in the perforation, adhesion and fusion of epithelial layers, such as in the epithelia of gill openings, rather than the homology of atrial primordia and vertebrate placodes [52]. The 'epithelial fusion hypothesis' is consistent with the finding that among vertebrates, Pax2 apparently plays a role in gill slit perforation in Xenopus [96] and Pax8 is necessary for vaginal opening in mouse [97]. Furthermore, Pax2 and Pax8 regulate genes that control the composition of the extracellular matrix in epithelial fusions [42,98].Evidence to resolve this dilemma comes from examination of gene expression and organ structure in a phylogenetic context. Our analysis of larvacean Pax2/5/8 expression teases apart epithelial fusion from placode formation and suggests that Pax2/5/8 likely functioned in both processes in ancient chordates. In Oikopleura, several Pax2/5/8 expression domains can be grouped into two overlapping categories: sites of epithelial fusion and sites of sensory cell development. At the sites of epithelial fusion, larvacean Pax2/5/8 paralogs are expressed at the junction of the gill pouch with the epidermis, the joining of the rectum to the epidermis at the anus, and the fusion of the stomodeum and pharynx at the mouth. At sites of sensory cell development, Oikopleura's Pax2/5/8 paralogs are expressed at the stomodeal placode and the putative acousticolateralis placode homolog (the Langerhans organ primordia). Therefore, in contrast to ascidians, in which hair cell-like organs and sites of perforation are conflated in the atrium [95], Oikopleura's paired mechanosensory organs are topographically separate from the gill openings.In agreement with the perforation argument [52,99], expression of Oikopleura Pax2/5/8 as in amphioxus, in the gill openings and anus, which lack sensory cell types, supports the hypothesis of an ancient function for Pax2/5/8 in the adhesion and fusion of epithelial layers and in the promotion of epithelial perforations. On the other hand, expression of Oikopleura Pax2/5/8a at early tailbud stage in the primordia of the Langerhans organs occurs long before the fusion of the gill endoderm with the ectoderm, which happens at the late hatchling stage. This timing gap argues against the interpretation that early expression of Pax2/5/8 in the Langerhans domain is associated only with remodeling the extracellular matrix for gill perforation. Instead, early Pax2/5/8 expression defines the location of paired, thickened sensory organ primordia, in agreement with the hypothesis of Wada et al. [48] that Pax2/5/8 marks a urochordate placode. Larvacean Pax2/5/8 expression adds to a growing body of morphological and gene expression data that supports the origin of cranial placodes in early chordates, rather than in vertebrates, as had been long assumed [39,48,73,93-95], [100-105], and strengthens the case for an early origin specifically of the otic or acousticolateralis placode.Therefore, analysis of Pax2/5/8 paralogs in Oikopleura has allowed us to reconcile the 'placode hypothesis' and 'epithelial fusion hypothesis', supporting an evolutionary scenario in which the role of Pax2/5/8 in epithelial fusions and perforations was already present in stem chordates, and the role of Pax2/5/8 related to placode development is likely also ancient, though perhaps restricted to Olfactores (Figure 6).Evolution of Pax2/5/8 genetic pathwaysPax, Eya, Six and Dach genes form a genetic network in several biological processes, including in sensory placode development (reviewed in [81,106]). Pax2, Pax5, Pax8, Eya1, Six1, and Dach1 are co-expressed during fish otic vesicle development (see for example [107,108]), where they interact to specify otic tissue and maintain its continued ontogenesis. Larvacean Pax2/5/8 genes are expressed in presumptive sensory tissues previously reported also to express Eya and Six orthologs [39], namely around the mouth and in the mechanosensory Langerhans receptors. The two Pax2/5/8 paralogs may interact with different subsets of Eya-Six-Dach genes in different tissues. For instance, in developing sensory tissues, Pax2/5/8a expression in tailbud and early hatchling stages is similar to that of Eya and Six3/6a in the developing Langerhans receptor primordia, and to Six1/2 and Six3/6a expression around the mouth, including sensory cilia-bearing cells of the upper lip. Endodermal Pax2/5/8b expression, on the other hand, most strongly overlaps that of Six3/6a in the rostral pharynx and Eya in the gill endoderm in mid-hatchling stages. Similarly, ascidian Pax2/5/8a expression overlaps Eya and Six1/2 in the atrial primordia and Eya, Six1/2, and Six3/6 in the stomodeal domain, but Pax2/5/8b exhibits broad expression in the ectoderm and may also overlap Eya, Six1/2, and Six3/6 in a way that larvacean Pax2/5/8b does not [73]. Such differences between presumed Pax2/5/8 paralogs and orthologs is further evidence that, though the Pax-Six-Eya-Dach 'module' is conserved at the level of gene families, paralogs within each module may differ both within a developmental program and between lineages [39,99]. Not surprisingly, then, sequence analysis shows that several known protein interaction domains differ between the larvacean Pax2/5/8 paralogs and between the larvacean and ascidian orthologs (Figure 1), consistent with divergence of the molecular pathways in which these Pax proteins participate.Pax2/5/8 gene duplications and the evolution of gene functionsA requirement for partitioning of ancestral gene functions is that the ancestral gene must have multiple independently mutable functions either in protein-coding domains or in regulatory elements that drive restricted expression: in other words, units that are by definition 'subfunctions'. Direct evidence for independently mutable Pax2/5/8 subfunctions in an extant gene comes from the Drosophila ortholog, called D-Pax2, in which mutations in separate regulatory elements affect development of either ommatidia or sensory bristles [109]. The known functions of vertebrate Pax2/5/8 genes are diverse and include the establishment of the midbrain-hindbrain border [110-113], specification and mature function of thyroid follicular cells [85], specification and morphogenesis in the pronephros [80], differentiation of interneuron subtypes in the central nervous system (see for example [114]), promotion of correct axon guidance in the optic nerves [115], and morphogenesis and sensory cell specification in the epibranchial, otic and lateral line sensory placodes (see for example [96]). It is not known, however, how many of these functions, segregated among extant genes, resulted from independently mutable ancestral subfunctions.From the partially overlapping expression patterns of vertebrate Pax2/5/8 genes, we can infer that subfunction partitioning has occurred in this gene family. Next, comparative analysis of expression domains in the chordate phylogenetic context helps us infer when subfunctions arose. Parallel, independent partitioning of Pax2/5/8 endostyle/thyroid functions in both vertebrate and urochordate lineages suggests that the single Pax2/5/8 gene of stem olfactores had already acquired independently mutable regulation for endostyle/thyroid expression. Likewise, paralleling segregation of mammalian otic placode subfunctions to Pax2 and Pax8 [42,45], larvacean otic placode-like tissues express only one Pax2/5/8 paralog early and in a pattern overlapping orthologs of other vertebrate placode markers of the Six and Eya gene families [39] (Figure 6). It remains possible, though, that despite the similarity between vertebrate and urochordate Pax2/5/8 gene expression patterns, each lineage separately evolved independently mutable functions in the endostyle/thyroid and otic placode development and that the ancestral gene did not already bear these subfunctions.Within the urochordate lineage, further parsing of gene functions may represent subfunctions that were present in the ancestral urochordate Pax2/5/8 gene. For example, at some sites of epithelial fusion, larvacean Pax2/5/8 paralogs exhibit complementary expression, for example just outside (Pax2/5/8a) or just inside (Pax2/5/8b) the mouth and just outside (Pax2/5/8b) or just inside (Pax2/5/8a) the anus. Therefore, though a role in epithelial fusions is probably ancient in Chordata, urochordate paralogs may have taken on separate refinements of the same function.In addition to the partitioning of ancestral functions, we might also expect to detect apparent neofunctionalizations of gene duplicates, particularly when we compare deeply diverging lineages such as vertebrates and urochordates. The role of vertebrate Pax5 in the B-lymphoid lineage of the immune system might be one example of a vertebrate neofunctionalization [55]. Similarly, because no comparable expression can be found outside of Urochordata, a role for Pax2/5/8b in the pericardium of the heart could be a novel deployment of Pax2/5/8 exclusive to the urochordates or even to the larvacean lineage.Though observed Pax2/5/8 expression patterns are consistent with a hypothesis of ancestral subfunction partitioning, this analysis detects only transcriptional regulatory differences and might underestimate the extent of actual subfunction partitioning. For example, mRNA expression of the paralogs may overlap in a given tissue, but the structurally different proteins the genes encode may not have redundant functions in those tissues. In addition, post-transcriptional regulation would also be missed in our analysis. Partitioned genes likely continue to evolve after their initial partitioning, and such divergence could complicate distinguishing subfunctionalization from neofunctionalization when comparing gene functions in deeply diverging lineages. Nonetheless, functional analysis of urochordate Pax2/5/8 genes could help distinguish between ancestral chordate subfunction partitioning and convergent patterns of neofunctionalization. For example, if Pax2/5/8 proteins carry out a similar function in the vertebrate thyroid as in the urochordate endostyle, this would be more concrete evidence that an ancient function was partitioned to paralogs in both chordate subphyla.ConclusionThe present work shows how analyzing the evolution of gene families that have experienced multiple independent gene duplication events in related lineages can improve understanding of the evolution of the genetic mechanisms underlying the development of homologous structures of anatomically divergent organisms (for example the endostyle of non-vertebrate chordates and the thyroid of vertebrates), and can help to identify gene subfunctions that otherwise may be difficult to recognize because of extensive pleiotropy (for example, Pax2/5/8 subfunctions in epithelial fusions around perforations and development of placode derivatives). Analysis of the evolution of subfunctions in a phylogenetic context identifies lineage specific innovations (for example, placode derivative homologs in olfactores). The modular fashion in which gene subfunctions partition after independent gene duplication events in various lineages implies the existence of independent regulatory elements that control temporal and spatial expression for each subfunction. For instance, comparison of Pax2/5/8 expression patterns in the endostyles of amphioxus, Oikopleura, and ascidians predicts the presence of separable regulation for expression in the iodine-binding dorsal component and the supporting ventral component of the endostyle/thyroid, a hypothesis that remains to be tested. The identification of the complete set of gene orthologs and paralogs in gene families in an increasing number of completely sequenced genomes will likely reveal additional cases of independent gene duplications and subfunctionalization events, the analysis of which will improve our understanding of the evolution of gene functions and the implications in the evolution and diversity of living forms.MethodsBiological materialsOikopleura dioica animals were collected in the Pacific Ocean near the Oregon Institute of Marine Biology (Charleston, OR), and were cultured in the laboratory at the University of Oregon (Eugene, OR, USA) at 13°C in 10 μm-filtered seawater for several generations. The transparency of Oikopleura embryos and adults allows non-invasive study of internal anatomy at the level of individual cells and the tracing of organs from embryo to adult. For some images, we merged DIC optical sections using Adobe Photoshop software to integrate images of structures that spanned multiple focal planes.Whole-mount in situ hybridizationWhole-mount in situ hybridization was performed as described [116,117] with minor modifications. Fixed, dehydrated embryos were dechorionated manually with glass needles before re-hydration; to reduce background, Tween-20 concentration was increased from 0.1% to 0.15% in the hybridization buffer, PBT solution and post-hybridization washing buffers. Embryos were mounted in 80% glycerol for microscopy. Riboprobes for detecting the expression of Pax2/5/8a and Pax2/5/8b (Genbank accession numbers, respectively: AY870648, AY870649) genes are described in [38].List of abbreviationsDDC: duplication-degeneration-complementation.Authors' contributionsSB identified, cloned and characterized the expression pattern of Oikopleura Pax2/5/8b, performed part of the comparative analysis and wrote part of the manuscript. CC identified, cloned and characterized the expression pattern of Oikopleura Pax2/5/8a, performed part of the comparative analysis and wrote part of the manuscript. JHP supervised the project, performed part of the comparative analysis and wrote part of the manuscript. All authors read and approved the final manuscript.Supplementary MaterialAdditional file 1Phylogenetic relationships among chordate Pax2/5/8 proteins. The data provided illustrates the independent gene family expansions that permitted parallel histories of subfunction partitioning among vertebrate paralogs (Pax2, Pax5 and Pax8) and among urochordate paralogs (Pax2/5/8a and Pax2/5/8b).Click here for file\n\nREFERENCES:\nNo References"
+}

batch_5/PMC2532685.json

@@ -0,0 +1,4 @@
+{
+    "id": "PMC2532685",
+    "text": "This is an academic paper. This paper has corpus identifier PMC2532685\nAUTHORS: Natalia I Minaeva, Evgeny R Gak, Danila V Zimenkov, Aleksandra Yu Skorokhodova, Irina V Biryukova, Sergey V Mashko\n\nABSTRACT:\nBackgroundThe development of modern producer strains with metabolically engineered pathways poses special problems that often require manipulating many genes and expressing them individually at different levels or under separate regulatory controls. The construction of plasmid-less marker-less strains has many advantages for the further practical exploitation of these bacteria in industry. Such producer strains are usually constructed by sequential chromosome modifications including deletions and integration of genetic material. For these purposes complex methods based on in vitro and in vivo recombination processes have been developed.ResultsHere, we describe the new scheme of insertion of the foreign DNA for step-by-step construction of plasmid-less marker-less recombinant E. coli strains with chromosome structure designed in advance. This strategy, entitled as Dual-In/Out, based on the initial Red-driven insertion of artificial φ80-attB sites into desired points of the chromosome followed by two site-specific recombination processes: first, the φ80 system is used for integration of the recombinant DNA based on selective marker-carrier conditionally-replicated plasmid with φ80-attP-site, and second, the λ system is used for excision of inserted vector part, including the plasmid ori-replication and the marker, flanked by λ-attL/R-sites.ConclusionThe developed Dual-In/Out strategy is a rather straightforward, but convenient combination of previously developed recombination methods: phages site-specific and general Red/ET-mediated. This new approach allows us to detail the design of future recombinant marker-less strains, carrying, in particular, rather large artificial insertions that could be difficult to introduce by usually used PCR-based Recombineering procedure. The developed strategy is simple and could be particularly useful for construction of strains for the biotechnological industry.\n\nBODY:\nBackgroundEscherichia coli is widely used in fundamental investigations and in modern biotechnology for production of biologically active compounds such as recombinant proteins, amino acids, vitamins etc. Construction of plasmid-less marker-less strains has advantages for extending the practical exploitation of these bacteria in industry [1-5]. Such producer strains are usually constructed by sequential chromosome modifications, mainly including deletions and integration of genetic material. For gene deletions, the Red/RecET recombination method developed by several groups [6-11] is considered as the most useful now. This method has been named Recombinogenic Engineering or Recombineering and reviewed in several papers [12-15]. It usually (but not always) based on λRed- or RecET-mediated recombination between bacterial chromosome and amplified DNA fragment carrying the removable selective marker, in which PCR primers provide the rather short homology to the targeted sequence. The integrated marker could be excised out of the chromosome by site-specific recombination. The Recombineering approach based on generated PCR products might be used not only for target genes disruption, but for integration of relatively short DNA fragments, for example, for substitution of recombinant regulatory regions (i.e. promoter, RBS) of a particular gene for its native regulatory region [16-19]. Although the special modifications of Recombineering procedure have been already developed for integration of large DNA fragments carrying several genes/operons (see, [20], for instance), more often special tools based on modified transposons [1,21-23], non-replicative [3] or conditionally-replicative plasmids [4,24-28] are using for the same purposes. Different transposons and phage Mu systems are exploited for introduction of the DNA cassettes into random points of the bacterial chromosome [1,21,29]. Integration into the native coliphages attB sites [30] or into artificially inserted recombinogenic sequences [3] is based on exploitation of corresponding site-specific recombination systems. By using cloned fragments of chromosomes as so-called \"guides\" [4] it is possible to integrate the cassette by general homologous recombination. In addition, combinations of different systems in one integration strategy are also used. For example, site-specific insertion of cassettes can be carried out in preliminary randomly integrated recombinogenic sites [3], or the \"marked\" cassettes can be randomly integrated, followed by excision of the marker by a site-specific recombination system [1].For expansion of this group of methods, we propose a new strategy of DNA fragments integration in the process of plasmid-less marker-less recombinant E. coli strain construction. It initiates from Red-driven insertion of the antibiotic resistance marker flanked by φ80-attL/R sites, into the desired point of bacterial chromosome (this part of the total procedure is named as – the first \"In\") followed by φ80-Int/Xis-mediated excision of the marker (the first \"Out) with retaining of φ80-attB-site. After that two sequential site-specific recombination processes are provided: first, the φ80-Int is used for integration (the second – \"In\") of the recombinant DNA constructed on the basis of conditionally-replicated plasmid with φ80-attP-site, and second, the λ-Int/Xis system [1,10,19] is used for excision of inserted vector part, including the plasmid ori-replication and the selective marker, that flanked by λ-attL/R-sites (the second – \"Out\"). So, the new strategy schematically presented in Fig. 1, could be named Dual-In/Out.Figure 1Dual In/Out method for plasmid-less marker-less strain construction. A: deletion of the native φ80-attB by Red recombination. B: Red integration of the φ80-removable marker to the desired loci of MG-Δ(φ80-attB) chromosome. C: curing of the marker by φ80-Int/Xis-system. D: φ80-driven integration of the CRIM plasmid with target cassette into the different of φ80-attB sites. E: construction of \"marker-less\" cassette-carrier-strains by λ-Int/Xis excision of the vector part of integrative plasmids.ResultsConstruction of strains with different locations of the φ80-attB siteThe native φ80-attB-site is located at ~28 min of E. coli MG1655 chromosome ( and ). Insertion of the artificial φ80-attB into the chromosome of strain with deleted φ80-attB native site will allow us to introduce the DNA cassette into the new loci by φ80-Int-dependent system. If cassettes, integrated in the set of strains in different points, possess excisable selective markers, it will be possible to bring them in one strain by P1vir-mediated generalized transduction (P1-duction) due to removing the marker from the recipient genome before the next step of transduction.The desired locations of φ80-attB sites can be spread in non-essential parts of bacterial genome at different distances from oriC. In this case, the level of expression of the same integrated cassette will depend on its position due to the θ-like structure of replicating chromosome and gene-dosage effect: the cassettes closer to oriC will be expressed at higher level than those located near the terminus of the chromosome replication [31].We consider as good candidates for the further integration points the positions of the \"native\" IS elements (in our case IS5.7–IS5.11) since insertions into the genes already disrupted by IS elements will not be detrimental to cell viability and IS elements are almost randomly distributed along the chromosome of MG1655 [32]. Accordingly, the set of genes disrupted by IS5-elements insertion was chosen as the future locations of the artificial φ80-attB sites. These points are rather far from each other, so the future integrated cassettes can be combined in one strain by independent acts of P1-duction.Realization of this strategy includes several steps. At first, the deletion of the native φ80-attB was carried out by Recombineering between E. coli MG1655 chromosome and the constructed \"λ-excisable\" CmR-marker amplified by PCR from pMW118-(λattL- CmR-λattR) [19]. After Red recombination, the DNA locus modification was verified by PCR, and followed by λ-Int/Xis-mediated excision of the marker from the chromosome. The marker-less strain, MG-Δ(φ80-attB), was used as the recipient for the insertion of the artificial φ80-attB. The latter includes:1) construction and cloning of the cassette [(φ80-attL) - KmR - (φ80-attR)] in the plasmid;2) using the obtained plasmid, pMWattphi, as the template for PCR amplification of φ80-removable KmR-marker flanked by the 36 bp arms homologous to the desired loci on the MG1655 chromosome;3) Red integration of the markers into the chromosome of MG-Δ(φ80-attB) for construction of the desired library of \"marked\" strains;4) each obtained strain was cured from the marker by φ80-Int/Xis-system [30]. In this way the \"unmarked\" part of the library was constructed;5) new strains and MG1655 (as a control) were tested for φ80-driven integration/excision of the \"conditionally-replicated integrative and modular (CRIM) plasmid\" pAH162 carrying φ80-attP site [30], using φ80-Int- or φ80-Int/Xis-helper plasmids, respectively. Here, the efficiencies of integration/excision In or Out of the artificial points were practically the same as for In/Out of the native site.Although the efficiency of CRIM plasmids excision from all tested sites in our experiments was 15–30% (not 100%, as reported in [30]), nevertheless, selection the marker-less clones was not the problem. We noticed, as well that, in our hands, the efficiency of λ-Int/Xis-driven excision under the same conditions was significantly higher and exceeded 80%.Construction of φ80-integrative CRIM plasmid with λ-removable \"vector part\"The φ80-cognate CRIM plasmids with different selective markers can be used for integration of cassettes in the obtained strains that differ in location of φ80-attB. A recombinant strain that contains multiple insertions can be constructed by P1 transduction. However the presence of plasmids' markers in the bacterial chromosome cannot satisfy \"marker-less\" criteria for the practical application of the engineered strain. CRIM plasmids constructed by Haldimann and Wanner [30], only allow the site-specific excision of the entire recombinant structure initially inserted in the chromosome. It is useful to modify CRIM plasmids to allow excising of the vector part after site-specific integration of recombinant DNA. In this way, the new φ80-cognate CRIM plasmid (Fig. 2) with a λ-removable vector part was obtained (for details see Methods).Figure 2New φ80-cognate CRIM plasmids with λ-removable part. A. Map of pAH162-λattL-TcR-λattR. This plasmid could be used as a vector for molecular cloning of the genes of interest followed by φ80-Int-dependent integration of the recombinant plasmid in bacterial chromosome and λ-Int/Xis-dependent excision of the selective marker-carrier vector part. B. Map of pAH162-λattL-TcR-λattR-CmR. The recombinant plasmid constructed on the basis of new CRIM-vector, that contains cat-gene under the transcriptional control of phage T7 A2-promoter, as a model gene for integration in the chromosome according to Dual-In/Out strategy.The new plasmid pAH162-λattL-TcR-λattR (Fig. 2) retains from its progenitor pAH162 the non-modified fragment carrying φ80-attP and the multi-cloning site flanked by bacterial (rgnB) and phage λ (tL3) transcription terminators. This fragment is bracketed by λ-attL/R sites which allow λ-Int/Xis excision of the vector part, including conditionally-replicative origin of R6K (oriRγ) and the selective marker, TcR (from Tn10). We have confirmed the expected properties of pAH162-λattL-TcR-λattR by its integration into the native φ80-attB of MG1655 chromosome, followed by λ-Int/Xis excision of the marked vector part flanked by λ-attL/attR. Of 80 tested clones cures from λ-Int/Xis-helper plasmid, more than 80% have lost the TcR marker of integrated CRIM plasmid, as well.Exploiting the new system for multiple integration of cat-gene into E. coli chromosomeTo test the Dual-In/Out strategy of marker-less strain construction, we used the efficiently expressed variant of cat-gene as the model cassette. This experiment includes several steps:1) cloning of cat-gene into pAH162-λattL-TcR-λattR, using E. coli CC118 (λpir+) strain as the recipient;2) φ80-driven integration of the recombinant CRIM plasmid into the strains that differ in location of φ80-attB;3) construction of \"marker-less\" strains with single cat-cassette by λ-Int/Xis excision of TcR-containing DNA fragment flanked by λ-attL/R-sites, followed by determination of cat expression levels in these strains;4) combining the set of cat-cassettes in one \"marker-less\" strain by sequential P1-duction of the \"marked\" cassettes followed by curing the TcR-marker from the recipient strain before the next stage of transduction.The previously constructed [19] pMW118-(λattL-CmR-λattR) plasmid was chosen as a template for PCR amplification of cat-gene. In this plasmid the structural part of cat-gene with its native RBS from E. coli Tn9 is under the transcriptional control of rather strong phage T7 A2 promoter [33] that is recognized by E. coli RNA polymerase with σ70 in a constitutive manner. This cat-gene was amplified by PCR with the primers that carry restriction sites for cloning in pAH162-λattL-TcR-λattR. The transformants of E. coli CC118 (λpir+) carrying the recombinant plasmid of interest, were selected on the medium supplemented with Cm. The expected structure of the plasmid was verified by restriction analysis and PCR.At the next stage this plasmid was integrated by φ80-Int system into MG1655-derived strains with a different location of φ80-attB, using TcR-marker for selection (Fig. 3). The correct integration was confirmed by PCR, and the corresponding strains were entitled MG-TcR-cat-(i), where (i) varied from 1 to 6 depending on the number of φ80-attB sites. Subsequently, the vector part of the integrated plasmids in all MG-TcR-cat-(i) strains was excised by λ-Int/Xis site-specific recombination. After checking by PCR, the corresponding recombinant strains retaining the cat-gene in their chromosomes, was entitled MG-cat-(i).Figure 3Integration of plasmid pAH162-λattL-TcR-λattR-CmR into MG1655-derived strains with different location of φ80-attB.The efficiency of cat-gene expression in MG-cat-(i)-strains was evaluated by determination of Cat enzymatic activity (Fig. 4, Table 1). As expected (see explanation above and reference [31]), the level of cat-gene expression correlates with the distance between the integration point and oriC: the strains with the cat-gene position closer to oriC have higher Cat activity.Table 1Cat-activity of tested MG-cat-(i) and MG-cat-(i+j) strains. The results were averaged over three independently grown cultures for each clone; the scatter was 5–15%.strain MG-cat-(i)IS elementActivity, nmol/minxmgMG-cat-(1)IS 5.7180 ± 15MG-cat-(2)IS 5.8180 ± 15MG-cat-(3)IS 5.9240 ± 30MG-cat-(4)IS 5.10280 ± 15MG-cat-(5)IS 5.11270 ± 15MG-cat-(6)native φ80-attB200 ± 30MG-cat-(i+j)MG-cat-(1+2)IS 5.7 IS 5.8330 ± 15MG-cat-(1+3)IS 5.7 IS 5.9420 ± 30MG-cat-(2+3)IS 5.8 IS 5.9420 ± 30MG-cat-(1+2+3)IS 5.7 IS 5.8 IS 5.9520 ± 15MG-cat-(4+5)IS 5.10 IS 5.11500 ± 30MG-cat-(4+6)IS 5.10 native φ80-attB370 ± 15MG-cat-(5+6)IS 5.11 native φ80-attB400 ± 15MG-cat-(4+5+6)IS 5.10 IS 5.11 native φ80-attB540 ± 30Figure 4The cat-gene expression in MG-cat-(i)-strains. The dependence of the expression level of the cat-carrier cassetes (T7A2-cat) on their positions in the chromosome is indicated by arrows. Cat-activity of the MG-cat-(6) strain in which Cassette (T7A2-cat) is located in native φ80-attB was taken as 100%.Several cat-cassettes were combined in one strain by sequential P1-duction of Tc-marked fragments from MG-TcR-cat-(i) strains into MG-cat-(j) (where i ≠ j), followed by λ-Int/Xis-driven curing of TcR-marker that gives an MG-cat-(i + j) \"marker-less\" strain carrying two cat-cassettes in (i) and (j) positions; a third cassette was added by an analogous procedure. In addition, all steps used in increasing the number of cat-cassettes in the chromosome of recombinant \"marker-less\" strain were controlled by determination of Cat activity. The data presented in Table 1 show that the total Cat activity of multi-integrants correlates with the expected sum of the Cat activities of the corresponding single-integrants.DiscussionIn this paper we have described the development of a new strategy for the integration of genes/operons of interest during plasmid-less marker-less recombinant strain construction, which we have named Dual-In/Out. It combines Red-mediated insertion of the artificial φ80-attB site into desired point of bacterial genome, φ80-Int-dependent site-specific integration of recombinant DNA of interest constructed on the basis of specially constructed CRIM plasmid followed by λ-Int/Xis-mediated excision of the plasmid's vector part flanked by λattL/R, out of the chromosome.In this study a library composed of five strains that differ in the position of inserted φ80-removable KmR-marker and five isogenic marker-less strains for the possible φ80-dependent integration of a cassette has been obtained. (In fact, the sixth strain from this library is the MG1655 wild type with the native φ80-attB site). The \"marked\" part of this library can be used to select, in advance, the desirable points for the cassette integration that can be useful for construction (modification) of the producer strain. This is achieved by checking that transduction of the marker to a particular location does not affect the producer features (growth rate, desirable product yield, etc.). After preliminary selection of the members of the library based on testing \"marked\" strains, the corresponding \"unmarked\" variants can be directly used for the φ80-driven integration of cassette(s).This library can be extended by the Red-mediated insertion of φ80-attB sites into new points on MG1655 chromosome with deleted native φ80-attB. The plasmid pMWattphi can be used as the template for PCR with the primers for this purpose. The designing of the primers can be based on the known \"native\" insertions in the E. coli genome, as described in this paper. On the other hand, φ80-attB integration can be used for simultaneous inactivation of an \"undesirable\" gene whose expression decreases the performance of a producer strain. It is important to keep in mind that the integrated cassette \"ter rgnB – (genes of interest) – tL3\" could be downstream from a native transcription unit in the chromosome. In this case, terminator tL3 prevents transcription of chromosomal genes originating in the cassette, and ter rgnB protects the cassette from the bacterial transcription read through.In earlier techniques, the library of E. coli strains with different positions for site-specific (Flp-dependent) integration was constructed by random insertions of corresponding recombinogenic (FRT) sites by the Tn5-system [3]. At first glance, this library may have the same applications as the library described in this paper. Moreover, its improvement requires only determination of the FRT integration points for already obtained several hundreds of independent Tn5-driven integrants [3]. On the other hand, the expansion of our library depends on the separate Red-driven integrations, each leading to construction of only one new member. However, Red-driven integration is now used as a routine procedure and, according to our experience, can be provided in a quantity of several tens per week. At the same time our library has two advantages: (i) due to the initially designed points of the φ80-attB insertion we can exclude the interference between sequentially introduced cassettes, while random insertions more often localize near oriC due to a gene dosage effect produced by replication of the bacterial chromosome [29]; (ii) it is possible to combine distribution of site-specific integration points simultaneously with deletion of some \"undesirable\" genes.As already mentioned, the Dual-In/Out strategy allows construction of marker-less strain carrying different cassettes and/or several copies of the same cassette. The presence of several copies might be necessary to increase the level of the cassette expression. However, transduction of additional copies into the same strain may lead to possible chromosome rearrangements due to general recombination events between repeated sequences.These questions of general recombination can be divided in two groups. The first one is the question of the additional insertion of the new \"marked\" cassette during the transduction process, not into the point corresponding to its location in the donor genome, but into the \"unmarked\" cassette in the recipient chromosome. This will lead to substitution of the new cassette for the previous one. Theoretically, this process can not be excluded, but its probability is extremely low due to the small distance between φ80-attP and λ-attR in the plasmid pAH162-λattL-TcR-λattR which is used as the vector for the cassettes cloning (Fig. 2), and so, the size of the possible \"arm\" for envisaged general recombination would be too small. At least, we have never detected substitutions instead of expected amplification in tested clones. In any case, verifying the amplification of the integrated cassettes by PCR is desirable.Concerning the stability of maintaining identical cassettes in the genome, direct or inverted repeats should be considered. In case of directly repeated cassettes the general recombination between them would lead to the deletion of part of the bacterial genome. According to the proposed design of the integration points, the distance between the cassettes will exceed the size of the transducing phage genome (about 100 kb in case of P1 phage). Only a few parts of the E. coli MG1655 chromosome more than 100 kb in length, do not contain any of the ca. 230 absolutely essential genes. If the designed points of integration are outside these few parts, the strains with the deleted regions between directly repeated Cassettes will not survive. General recombination between long inverted repeats leads to the chromosomal inversions that, in turn, can change the properties of the strain [34-36]. Although these events are rather rare, it could be recommended to avoid construction of the strains with inversely repeated copies of the cassette. This is not a difficult task, because the developed Dual-In/Out strategy of strain construction allows the determination, in advance, not only of location, but also of the orientation of the desired insertions.ConclusionSumming up, the developed Dual-In/Out strategy is rather straightforward, but convenient combination of previously developed methods. Previous approaches of integration of rather large DNA fragments, usually, use only one high-performance site-specific recombination system. When the site-specific recombination is used for insertion of the fragment, the selective marker remains in the chromosome [30]. When it is used for excision of the selective marker, the initial integration of the cassette is carried out by general recombination [3]. The combination of Red/ET-driven and two site-specific recombination systems in one strategy for integration cassettes carrying several genes/operons, during construction of marker-less strains with desired structure is rather obvious, and, probably, it will be useful for fundamental and applied fields of microbiology and biotechnology.MethodsStrains and plasmidsEscherichia coli K-12 MG1655, a wild-type strain with sequenced genome [32], was used as the recipient for the insertion of the artificial φ80-attB. CRIM plasmids were propagated in CC118 λpir+ strain [37]. E. coli W3350(80) lysogenic for phage φ80 was obtained from GosNIIGenetika collection and used as a template for PCR-driven amplification of φ80-attL/R sites.pKD46 was used as a donor of λRed-genes for providing Red-dependent recombination according to the described procedure [8].pAH123 and pAH129-helper plasmids [30], GenBank accession numbers AY048726 and AY048727 respectively. These helper plasmids were used for φ80-dependent integration/excision procedures.pMWts-λInt/Xis-helper plasmid is similar to the plasmid pMP955A described in [1]. It has low-copy-number thermo-sensitive replicon pSC101, genes xis and int of phage λ under the control of λPR, thermo-sensitive repressor cIts857. It is used for λ-Int/Xis-mediated excision of the DNA fragments flanked by λattL/R.pMWattphi – this recombinant plasmid was constructed on the base of pMW118 (GenBank accession number AB005475). This plasmid is used as the template for PCR amplification of fragment (φ80-attL) - KmR - (φ80-attR) flanked with 36 bp arms homologous to targeted site in MG1655 DNA. Hybrid φ80-attL and φ80-attR sites were obtained by PCR amplification from purified chromosome of E. coli W3350(80) using primers: P1 – P2 for attL and P3 – P4 for attR. Amplified fragments were restricted with EcoRI-BamHI and XbaI-PstI restrictases and cloned into corresponding sites flanking kan on plasmid to give pMWattphi (Fig. 3).P1 5'-atagaattcgaaaggtcatttttcctgaatatgc-3'P2 5'-ataggatccatcattgaatgggtacacatttttg-3'P3 5'-atattctagagatttgaatagcgagcgtaccttag-3'P4 5'-atactgcagtcgtttgttgacagctggtccaatg-3'pAH162-λattL-TcR-λattR – integrative plasmid. Construction of this plasmid included several steps with isolation and analysis of recombinant DNA intermediates. The structure of this plasmid is shown in Fig. 2. Sequence landmarks: 1) from 6 to 1031 – fragment from pAH162 (GenBank accession number AY048738) which contains conditional-replication origin oriRγ; 2) from 1038 to1145 – fragment contains attL of phage λ from plasmid pMW118-(λattL-tetA-tetR-λattR) which structural similar to the plasmid pMW118-(λattL-CmR-λattR) [19]; 3) from 1153 to 2274 – fragment from pAH162 which contains bacterial terminator rgnB, multiple cloning sites MCS, phage λ terminator tL3, phage attachment attP phi80; 4) from 2281 to 2462 – fragment contains attR of phage λ from plasmid pMW118-(λattL-tetA-tetR-λattR); 5) from 2456 to 4463 – fragment from plasmid pMW118-(λattL-tetA-tetR-λattR) which contains the Tn10-encoded tetracycline resistance gene tetA and the repressor gene tetR.pAH162-λattL-TcR-λattR-CmR was constructed by cloning of the SphI-SacI cat-carrier DNA fragment amplified from pMW118-(λattL-CmR-λattR) [19] with primers P5 – P6.P5 5'-cagtaagcatgcgcggccgcccggataagtagacagcctgataag-3'P6 5'-cagtaagagctcgcggccgcttacgccccgccctgccactc-3'Molecular biology methodsRestriction analysis of the recombinant plasmids and Ca2+-dependent transformation of E. coli cells were performed in accordance with the routine experimental protocols [38]. Commercially available preparations of restrictases, T4 DNA ligase and the Klenow fragment of E. coli DNA polymerase I (Fermentas, Lithuania) were used. PCR fragment for cloning were generated by using AccuTaq DNA polymerase (Sigma, USA). Sigma (USA) products were used for the isolation of plasmid DNA, extraction of DNA fragments from agarose gels.Construction of strains with different locations of the φ80-attB siteThe deletion of the native φ80-attB was carried out by Red-dependent integration of \"λ-excisable\" CmR-marker which has been amplified by PCR from pMW118-(λattL-CmR-λattR) [19] with primers P7 – P8. DNA locus modification was verified by PCR with primers P9 – P10.P7 5'-gtaatcaaaggatttgagcgagcaactgtacctcagcgctcaagttagtataaaaaagctgaac-3'P8 5'-acatttagcacgtttacagttactgcatgatgaaggtgaagcctgcttttttatactaagttgg-3'P9 5'-tgcagcgcgtgaatgtgtta-3'P10 5'-ctcaagacaaagctgatagcc-3'The obtained marker-less strain, MG-Δ(φ80-attB), was used as the recipient for the insertion of the artificial φ80-attB. pMWattphi was used as the template for PCR amplification of fragment (φ80-attL) - KmR - (φ80-attR) which integrated into the chromosome of MG-Δ(φ80-attB) to the desired loci – the set of genes disrupted by IS5-elements insertion. For these purposes the following primers were used:IS5.7 P11 5'-tcctaaagaaagtatctattctgatacggttgttgagaaaggtcatttttcctgaatatg-3'P12 5'-aagccatttacacgcacaaaatctgaaaaacgtacctcgtttgttgacagctggtccaatg-3'P13 5'-gtcttctcacgggaacggtt-3'IS5.8 P14 5'-gagggtatcagtacattgaaatgaatggcgccgcaggaaaggtcatttttcctgaatatg-3'P15 5'-tctggtttgccgcgccacccatttgaacaatttgattcgtttgttgacagctggtccaatg-3'P16 5'-cctcccttttcgatagcgacaa-3'IS5.9 P17 5'-gggcgtattaccgcgcaaatagataccttgcaccgcgaaaggtcatttttcctgaatatg-3'P18 5'-ctgcggatcatcaatggcgtcaatcatgccgaaatg-tcgtttgttgacagctggtccaatg-3'P19 5'-gttcaatatgcgcggcatacca-3'IS5.10 P20 5'-tatcaattgacgttaaggtgactctggaagctgcaggaaaggtcatttttcctgaatatg-3'P21 5'-tattgactgaatgactaccgaagttaacaactccgctcgtttgttgacagctggtccaatg-3'P22 5'-ttccggtggtcatactatccattc-3'IS5.11 P23 5'-attattaaccattaatgacaaccttttacgagcaaagaaaggtcatttttcctgaatatg-3'P24 5'-tatgaaagattggttatcctggcctctaaaaatttatcgtttgttgacagctggtccaatg-3'P25 5'-ctttttcattaggcagtggcctc-3'The integration of fragment was verified by PCR with primers P13, 16 – P26 and P19, 22, 25 – P27.P26 5'-tgtttcgggcggaccaaatgata-3'P27 5'-gccatggcagaatctgctccatgcggg-3'Plasmid integration and excision of the vector part of integrated plasmidFor testing the new MG1655 strains with artificial φ80-attB sites and CRIM plasmid integration/excision, procedures were driven by standard protocols using helper plasmids pAH123 (φ80-Int) and pAH129 (φ80-Int/Xis) respectively [30]. The vector part of CRIM plasmids was excised by λInt/Xis – site-specific recombination using helper plasmids pMWts-λInt/Xis by standard protocols [1]. The integration of plasmid was verified by PCR with primers P13, 16, 19, 22, 25, 28 – P30 and P 26, 27, 29 – P31. P28 and P29 are primers for native φ80-attB [19]; P30 is a primer annealing on vector part of integrated plasmid [19] and P31 is annealing on tetR gene.P28 5'-taaggcaagacgatcagg-3'P29 5'-ctgcttgtggtggtgaat-3'P30 5'-acgagtatcgagatggca-3'P31 5'-gtaaactcgcccagaagctagg-3'Cat activity assaysThe Cat activity was assayed using a spectrophotometric method (UVmini 1240; Shumadzu, Japan). Log-phase cells harvested at OD595 = 0.8 were resuspended in potassium phosphate buffer (50 mM; pH 7.5). Cell lysates were prepared by sonication. The quantity of protein was determined by the Bradford method [39]. Assays were performed in 1 ml (1 cm light path) cuvettes at room temperature. The reaction mixture in each cuvette contained 100 μl of 1 M Tris-hydrochloride, pH 7.5, 100 μl of 1 mM acetyl CoA (Sigma, USA), 100 μl of 10 mM 5, 5'-dithio-bis-2-nitrobenzoic acid (DTNB; Sigma, USA), 0.05 mg of protein, H2O for a total volume of 0.99 ml. 10 μl of 10 mM Cm (Sigma, USA) was added to start the reaction, and thionitrobenzoic acid (TNB) production was followed at 412 nm. Enzyme activity was calculated in terms of nmol of thionitrobenzoic acid produced per min per mg of protein. The results were averaged over three independently grown cultures for each clone; the scatter was no more than 15%.AbbreviationsPCR: polymerase chain reaction; Cm: chloramphenicol; Km: kanamycin; Tc: tetracyclin.Authors' contributionsNIM obtained the library of strains with different locations of the φ80-attB site, performed the integration of plasmid pAH162-λattL-TcR-λattR-CmR to these strains, carried out Cat-assay experiments and edited the manuscript. ERG and DVZ designed the primers and the construction scheme and constructed pMWattphi and pAH162-λattL-TcR-λattR plasmids. AYS and IVB participated in the design of the study and helped draft the manuscript. SVM supervised and coordinated the work and edited the manuscript. All authors read and approved the final manuscript.\n\nREFERENCES:\nNo References"
+}

batch_5/PMC2532955.json

@@ -0,0 +1,4 @@
+{
+    "id": "PMC2532955",
+    "text": "This is an academic paper. This paper has corpus identifier PMC2532955\nAUTHORS: S G Haworth\n\nABSTRACT:\nPulmonary hypertension is relatively common in children and has many causes. The management of the condition has changed dramatically in the past 5 years with the introduction of new medicines. However, diagnosis, investigation and choice of therapy remain a challenge. In 2002 the United Kingdom Pulmonary Hypertension Service for Children was established and this has become the mainstay of management in this country. This service, based at Great Ormond Street Hospital for Children, provides advice, expertise and infrastructure support for the most severely affected patients, particularly those with idiopathic pulmonary arterial hypertension for whom chronic intravenous prostacyclin remains the most effective medication. New medicines are being developed which, rather than focussing on dilating a diseased pulmonary vascular bed, aim to structurally remodel the pulmonary vasculature towards normal.\n\nBODY:\nPulmonary hypertension is more common in children than in adults. A modest elevation in pulmonary arterial pressure (PAP) is tolerable, but a high pressure is associated with obstructive pulmonary vascular disease leading to right heart failure and death. Our perspective on this condition has changed dramatically since I wrote an editorial for this journal in 1998.1 Improved diagnostic techniques facilitate assessment, drug discoveries have revolutionised management and the establishment of a UK Service for the Care of Children with Pulmonary Hypertension has facilitated diagnosis and management and provided infrastructure support to improve the quality of life of the most severely affected children. We now have a clinically useful classification of pulmonary hypertension which includes all the types of pulmonary hypertension encountered in childhood (box 1).2 We have also established internationally recognised diagnostic and management guidelines.3 4 This review is based on the experience gained in establishing and running the UK Service for Children, a service commissioned by the Specialist Commissioning Group. The service is a structured clinical network designed to treat children with pulmonary hypertension rapidly and to provide immediate access to new medicines.Box 1 Classification of pulmonary hypertension1. Pulmonary arterial hypertension (PAH)1.1 Idiopathic (IPAH)1.2 Familial (FIPAH)1.3 Associated with (APAH)1.3.1 Connective tissue disease1.3.2 Congenital systemic to pulmonary shunts1.3.3 Portal hypertension1.3.4 HIV1.3.5 Drugs and toxins1.3.6 Other1.4 Associated with significant venous or capillary involvement1.4.1 Pulmonary veno-occlusive disease1.4.2 Pulmonary capillary haemangiomatosis1.5 Persistent pulmonary hypertension of the newborn2. Pulmonary hypertension associated with left heart diseases2.1 Left sided atrial or ventricular disease2.2 Left sided valvular heart disease3. Pulmonary hypertension associated with respiratory disease and/or hypoxia3.1 Chronic obstructive pulmonary disease3.2 Interstitial lung disease3.3 Sleep disordered breathing3.4 Alveolar hypoventilation disorders3.5 High altitude3.6 Developmental abnormalities4. Pulmonary hypertension due to chronic thrombotic/embolic disease5. MiscellaneousTumour, and othersIndicated in italics are the only conditions NOT to have been encountered in the UK Pulmonary Hypertension Service for Children.Pulmonary hypertension is defined as a mean PAP equal to or greater than 25 mm Hg at rest or 30 mm Hg on exercise, a definition which applies to all but the youngest infants. The aetiology is more diverse in paediatric than adult patients (box 1). Pulmonary arterial hypertension is much more common than venous hypertension, particularly in childhood, and the new therapies target arterial rather than venous disease. Pulmonary arterial hypertension is subdivided into idiopathic pulmonary arterial hypertension (IPAH), formerly known as primary pulmonary hypertension, and associated pulmonary arterial hypertension (APAH) when associated with other disorders such as congenital heart disease, connective tissue or lung disease.2 There is as yet no evidence of ethnic variability in the different types of pulmonary hypertension seen in British children.In 1998 the only drugs available to treat pulmonary hypertension were calcium channel antagonists and intravenous epoprostenol, although epoprostenol was only shown to be efficacious in children with IPAH in 1999.5 Now there are effective endothelial receptor antagonists, phosphodiesterase inhibitors and prostacyclin analogues, the first two groups of drugs being effective orally.DIAGNOSTIC STRATEGYClinical suspicion of pulmonary hypertension (fig 1)Pulmonary hypertension should be suspected in any child who is unduly short of breath, tires easily or is syncopal when there is no evidence of heart or lung disease. Children occasionally complain of chest pain. The disorder should also be suspected in those known to suffer from these diseases when increasing shortness of breath cannot be explained by the underlying disease process itself. The physical signs of pulmonary hypertension include a left parasternal, right ventricular lift, an accentuated pulmonary component of the second heart sound and sometimes cool extremities. A diastolic murmur of pulmonary regurgitation may be present, but signs of overt right heart failure are a late event in young children.Confirming the clinical suspicion of pulmonary hypertensionAn accurate and complete diagnosis is essential, remembering that dual pathology is not uncommon in children with pulmonary hypertension. A chest x ray, ECG and a transthoracic Doppler echocardiogram are mandatory (fig 1). The classical findings include a chest x ray showing enlarged central pulmonary arteries and diminished peripheral pulmonary vascular markings. ECG findings include evidence of right ventricular dilatation and hypertrophy which can be confirmed by transthoracic echocardiography. Doppler interrogation can estimate pulmonary arterial systolic and diastolic pressures. The right ventricular systolic pressure (RVSP) is estimated from the systolic regurgitant tricuspid flow velocity v and the estimated right atrial pressure (RAP) in the Bernouille equation: RVSP = 4v2+RAP. A tricuspid regurgitant jet is present in the majority of pulmonary hypertensive patients. Echocardiographic measures of right ventricular function include the Tei index which is the sum of the isovolumetric contraction time and isovolumetric relaxation time divided by the ejection time. It assesses both systolic and diastolic function. The tricuspid annular plane systolic excursion (TAPSE) correlates with the right ventricular ejection fraction.Figure 1Investigating pulmonary hypertension: diagnostic strategy. CXR, chest x ray; PH, pulmonary hypertension; V/Q, ventilation/perfusion.Blood tests may show hyperuricaemia, particularly in Eisenmenger syndrome and elevation of brain natriuretic peptide, although less reliably than in adults.6Functional capacity is graded according to the NYHA/WHO classification, from the asymptomatic patient in class I to the severely disabled in class IV. Shortness of breath is a common complaint and an objective assessment of exercise capacity is helpful. In a co-operative child aged 6 years or more the results of a 6 min walk test can be compared with those in normal children of the same age and sex.7 Cardiopulmonary exercise testing to determine maximum VO2, work rate and anaerobic threshold can also be helpful, but this test should be carried out after the severity of the pulmonary hypertension has been ascertained at cardiac catheterisation, and stressing the child in this way is thought to be safe.Classification of pulmonary hypertension: determining causationIf the chest x ray excludes parenchymal lung disease, and echocardiography excludes an intracardiac anomaly, then the child probably has IPAH (fig 1). Rarely, pulmonary veno-occlusive disease (PVOD) can masquerade as IPAH. Evaluation must exclude a potentially remedial cause of pulmonary hypertension. The age of the child will obviously determine the ease and success with which certain investigations, such as lung function tests, can be carried out.Any child suspected of having IPAH or PVOD/pulmonary capillary haemangiomatosis (PCH) should be referred immediately to the UK Pulmonary Hypertension Service for Children.IPAHPresentationPatients can present throughout childhood, even in infancy. Symptoms vary, are age related and often non-specific. Syncope is relatively common. The commonest misdiagnosis is asthma. Parents report shortness of breath without wheeze, unresponsive to bronchodilator therapy. The familial form of IPAH, which accounts for 6% of all cases, shows genetic anticipation, presentation occurring at a younger age in successive generations.8 9 Mutations in the bone morphogenetic receptor-II gene account for the majority of cases of familial IPAH and 26% of sporadic cases.10 Pulmonary hypertension also occurs in families with hereditary haemorrhagic telangectasia (HHT) caused by mutations in the activin-like kinase 1 (ALK-1) and endoglin genes.11 Taking a careful family history and examining specifically for evidence of HHT is important.InvestigationThe systemic arterial oxygen saturation is normal in the absence of an atrial septal defect. Desaturation can occur in the presence of an atrial communication when the PAP, right ventricular and atrial pressures increase and right to left shunting occurs. The ECG findings are characteristic, and include right axis deviation, evidence of right ventricular hypertrophy with or without a strain pattern and right atrial enlargement. In addition to an abnormal chest x ray, contrast enhanced spiral CT of the lungs is indicated in older patients to exclude thrombus in the central pulmonary arteries and chronic thrombo-embolic disease. It is also helpful in distinguishing IPAH from PVOD. Ventilation/perfusion scans can be normal in IPAH but may show small peripheral non-segmental perfusion defects. Magnetic resonance imaging can help assess right ventricular function but is not yet part of the routine investigation in children with IPAH. The echocardiogram generally reveals dilated right heart chambers, right ventricular hypertrophy with posterior bowing of the ventricular and, in the absence of an atrial communication, the atrial septum.12 The left ventricle may be severely compressed. Right atrial size and left ventricular eccentricity index are predictive of outcome in adults and are routinely assessed in children. Lung function testing can sometimes demonstrate small airways obstruction.In addition to routine haematology and biochemistry, thyroid function tests, a thrombophilia screen including antiphospholipid antibodies (lupus anticoagulant, anticardiolpin antibodies) and an auto-immune screen are carried out. Children may seroconvert when they are older. Some children have an immunoglobulin deficiency. They may have a low level of antithrombin III, protein S or protein C, which may be genetic in origin or result from a consumption coagulopathy.Screening of other siblings for the familial form of the disease is necessary.APAHSuccessful treatment of the pulmonary hypertension depends on the rapidity with which the physician treating the underlying disease recognises that pulmonary hypertension is a major complicating factor in the clinical picture.Presentation and the investigation of children with APAH are determined by the underlying disease process. The principle aetiologies are as follows.Chronic hypoxiaChronic hypoxia leads to pulmonary hypertension. Chronic lung disease of prematurity is a relatively common cause of moderate pulmonary hypertension. Interstitial lung disease, as strictly defined, is uncommon in children, as are the pulmonary complications of connective tissue disease. When pulmonary hypertension does occur in association with connective tissue disease in children, it is usually severe. Of the developmental abnormalities, the commonest association is congenital diaphragmatic hernia. In these patients pulmonary hypertension is relatively common in the neonatal period. It usually abates but can persist indefinitely, be severe and require long term treatment. The extent to which lung disease is causing or contributing to an elevated PAP is revealed by pulmonary function tests, the blood gases, ventilation/perfusion and high resolution CT scans.Persistent pulmonary hypertension of the newborn (PPHN)This condition is usually self-limiting unless there is a significant irreversible underlying problem such as chronic lung disease of prematurity. Alveolar hypoplasia, usually accompanied by a degree of dysplasia, is signalled by persistent failure to wean off the ventilator and an open lung biopsy may be needed to guide management. PPHN may also be the first indication of IPAH.Congenital heart diseaseSustained pulmonary hypertension is sometimes recognised following a technically successful intra-cardiac repair and can be severe. Typically, the child improves clinically after the repair but then becomes short of breath and tires easily. The more severely affected children are investigated and treated as though they had IPAH. Classical Eisenmenger syndrome in the patient who initially had a large left to right shunt and then slowly developed pulmonary vascular disease with shunt reversal, central cyanosis and clubbing is not usually a problem in childhood. Most patients begin to deteriorate in their teenage years. Any child who appears to have Eisenmenger syndrome should be evaluated carefully to ensure that the cardiac diagnosis is correct and complete and that remedial surgery is not feasible. Sleep studies may be indicated in those with trisomy 21. Paediatric cardiology units are familiar with the small group of children who had a high pulmonary vascular resistance when first seen and were therefore deemed inoperable. In such children the resistance probably remained high from birth. They frequently become symptomatic in childhood and may benefit from medical therapy. Complex surgically naive and post-operative cases for whom curative or palliative surgery is not feasible frequently need reappraisal including cardiac catheterisation with a view to palliation by medical therapy.Liver diseasePulmonary hypertension is a recognised complication of chronic liver disease. Portal hypertension rather than hepatocellular disease appears to be the main determinant. It is uncommon in childhood but when present can influence the timing of liver transplantation. In adults liver transplantation is contra-indicated if the mean PAP exceeds 35 mm Hg and the pulmonary vascular resistance is greater than 250 dynes/s/cm.13PVOD and PCHThese conditions are uncommon. They can present like IPAH, but recognition is important because their management differs from that of IPAH. PVOD usually presents with shortness of breath, sometimes accompanied by small, frequent haemoptysis. The patient may be desaturated, clubbed and have minimal rales on auscultation. The ECG and echocardiogram can be indistinguishable from IPAH but the lung diffusing capacity is lower and high resolution CT scans can be diagnostic to the experienced radiologist. In addition to the expected features of pulmonary hypertension, contrast CT with vascular imaging shows a patchy centrilobular pattern of ground glass opacities, the lobules having a “pavement appearance” because their margins are demarcated by thickened septal lines caused by oedema. Typically, mediastinal lymphadenopathy is prominent. The contrast scan also confirms the echocardiographic findings of unobstructed venous return in the large pulmonary veins. These children are usually very ill indeed and should be listed for transplantation without delay. Cardiac catheterisation may be indicated to confirm the diagnosis. If the child is relatively well and stable, it can be helpful to do an open lung biopsy in order to distinguish those with PCH who may be amenable to treatment for a short period of time, although such treatment with anti-cancer drugs must be regarded as experimental. Medical treatment of PVOD with pulmonary hypertension specific therapies can be hazardous and is contraindicated.Cardiac catheterisation to confirm the diagnosis, assess disease severity and guide therapeutic management and the role of lung biopsyThe purpose of cardiac catheterisation in children with pulmonary hypertension is to confirm the diagnosis and to ensure that the conclusions drawn from the non-invasive tests were complete and accurate. The catheter also determines disease severity by determining the PAP accurately, is the only reliable means to date of determining pulmonary vascular resistance and tests the vasoreactivity of the pulmonary vasculature. The main determinant of treatment is the response to vasodilator testing with nitric oxide at cardiac catheterisation. Pulmonary vascular resistance can only be determined using the Fick principle if the pulmonary blood flow can be determined accurately by measuring, rather than assuming, the oxygen consumption.The risk of both cardiac catheterisation and general anaesthesia are increased in the presence of pulmonary hypertension and therefore the procedure is carefully planned after discussion with the child’s parents. Adequate sedation, optimal ventilation and meticulous attention to acid base status and blood loss is mandatory. The clinical history and the haemodynamic findings may indicate the need for further interventions such as the creation of an atrial septostomy and/or insertion of a Hickman line for continuous infusion of epoprostenol. These procedures are best done under the one anaesthetic, but the family need to be fully informed of their likelihood and significance and consent to these procedures, or not, before the catheterisation study can take place.Lung biopsy is rarely justified in pulmonary hypertension. The exceptions are suspicion of PVOD or PCH, of alveolar hypoplasia/dysplasia in PPHN, and very rarely in children with complex congenital heart disease in whom it might still be possible to operate.Treatment of pulmonary arterial hypertensionImmunisation schedules should be maintained, and young children need respiratory syncytial virus prophylaxis with palivizumab. Anaesthesia for any general surgical or dental procedure requires particular care.The aim of medical treatment is to dilate the pulmonary vasculature and reverse the abnormal remodelling characteristic of pulmonary vascular disease. The practical difficulties encountered in treating children influence management and include their age, level of understanding, size and in some, the presence of other anomalies. Three signalling pathways are targeted: the prostacyclin, endothelin and nitric oxide pathways.Prostacyclin and its analoguesThe most effective therapy is a continuous intravenous infusion of epoprostenol, the sodium salt of prostacyclin. It has a short half-life of 3–5 min, is unstable and the infusion has to be prepared every 24 h. The principle side effects are jaw pain and diarrhoea. The child is dosed according to response. Children need much higher doses than adults. A more stable analogue of prostacyclin, treprostinil, can also be given intravenously but is associated with more prominent side effects, headaches and leg pain. Meticulous care of the Hickman line is essential to prevent local and systemic infections, the latter being extremely uncommon. Treprostinil can also be given by continuous subcutaneous infusion but is painful, the drug stimulating nerve endings and causing induration and sometimes ulceration of the skin.14 It is not used in children. Iloprost is a prostacyclin analogue which can be given by inhalation but small, tired children find it difficult to inhale an effective dose every 2 h. The drug is effective for less than 2 h.Endothelin receptor antagonistsThe dual endothelin (ET) receptor antagonist bosentan (Tracleer) was the first oral drug shown to be efficacious in IPAH and has been used extensively in this and other types of pulmonary hypertension since its introduction in 2002.15 It is efficacious in children.16 Its principle side effect is elevation of liver enzymes which necessitates a monthly blood test. Drug interactions can occur. Bosentan decreases effective exposure to warfarin because of induction of CYP3A4 and/or CYP2C9. The newer selective ET-A receptor antagonists, sitaxentan and ambrisentan have not yet been studied in children. Both drugs affect the liver less than bosentan and drug interaction is probably less likely with ambrisentan.Phosphodiesterase inhibitorsSildenafil was the first drug of this class and is still the most commonly used, particularly in young children with APAH. The principle side effects include erections and systemic hypotension when high doses are used. The dose is 0.5–1 mg/kg/dose, given three to four times a day, rarely more.AnticoagulationPatients with pulmonary vascular disease are prone to develop thrombosis in situ. Older children are given warfarin and younger ones usually receive aspirin. The INR must be monitored particularly closely in those on endothelin receptor antagonists.OxygenOxygen is a potent pulmonary vasodilator. Nocturnal supplemental oxygen is indicated if there is nocturnal systemic arterial oxygen desaturation and can benefit those with a high PAP.Atrial septostomyThis procedure is indicated in children with IPAH and post-operative pulmonary hypertension suffering from syncope and/or severe right heart failure.17 It reduces the effect of a sudden increase in pulmonary arterial and right heart pressures and maintains a left ventricular output.Lung and heart–lung transplantationThe policy of the UK Pulmonary Hypertension Service is to refer children on intravenous epoprostenol for assessment by the transplantation service when they are established on treatment and are still well. They are then reviewed by the transplantation service as indicated.WHICH CHILDREN WILL BENEFIT FROM WHICH DRUG?It is impossible to generalise about who should and should not be treated. All children with IPAH need urgent treatment, but in those with APAH it is the severity of the pulmonary hypertension and the extent to which even a modest increase in pressure influences their well-being and prognosis which determines the need to treat.Treating IPAHWithout treatment the expected survival is less than 1 year.5 The main determinant of treatment is the response to vasodilator testing with nitric oxide at cardiac catheterisation. In those with a positive response the PAP and PVR (pulmonary vascular resistance) must fall to a near normal level with no fall in cardiac output. Only patients with a positive vasodilator response can be treated with a calcium channel antagonist. This applies to less than 10% of those with IPAH. Children who improve and are stable on a calcium channel antagonist need repeat cardiac catheterisation after 1 year or less, because they can become resistant to the drug at any time and need escalation of therapy before they deteriorate. The majority of negative responders present in NYHA/WHO class III–IV and frequently need to be started on intravenous epoprostenol therapy immediately. In a minority it is feasible to initiate therapy with an endothelial receptor antagonist. Most children however, require dual therapy. Sildenafil has not been proven to be efficacious in children. Syncopal children need an urgent atrial septostomy. After discharge from hospital close monitoring, principally by echocardiography, is mandatory since urgent intensification of therapy is frequently necessary.Treating APAHCongenital heart diseaseChildren with severe post-operative pulmonary hypertension are treated like those with IPAH. Symptomatic children with classical Eisenmenger syndrome are treated either with an endothelin receptor antagonist or sildenafil, according to age, sex and maturation. Children with a high PVR who have never been operable are usually more symptomatic at an early age and need similar treatment.Chronic lung and connective tissue diseaseThese patients are treated with bosentan which is thought to have an anti-fibrotic effect in addition to treating pulmonary vascular disease. Young and less severely affected children receive sildenafil. Those with connective tissue disease may require intravenous epoprostenol.PPHN persisting beyond the neonatal periodChildren with a modest increase in PAP are treated with sildenafil unless it is apparent that resolution is not occurring when therapy is intensified.In both chronic lung disease and PPHN the aim is to encourage normalisation of the pulmonary vasculature and so be able to stop treatment with relatively new, powerful drugs whose long term effects on children are unknown.Bone marrow transplantationPulmonary hypertension can occur as a result of the condition indicating bone marrow transplantation, be a complication of transplantation or be an incidental finding. In any event, treatment with bosentan is contraindicated because it interacts with cyclosporine and tacrolimus. Sildenafil can be given to the less severely affected, but intravenous epoprostenol may be indicated.Quality of lifeAll children are encouraged to return to school quickly. For those on intravenous therapy, dedicated school carers are given appropriate training by the clinical nurse specialists of the UK Pulmonary Hypertension Service who visit the school regularly. All drugs and equipment for intravenous, inhalational and oral therapy are delivered directly to the child’s home. Psychological support may be necessary, particularly with respect to managing relationships in a family with a chronically sick child. The patients’ organisation, The Pulmonary Hypertension Association UK, is a great source of comfort and support. Hospice care also helps many families. It is the responsibility of the UK Pulmonary Hypertension Service for Children to establish support networks in the community for these families.Improving quality of life must be the first priority when treating an incurable disease. In the Quality of Life questionnaires completed by the parents and older children cared for by the UK Pulmonary Hypertension Service the scores for physical performance were low, as expected.18 The psychosocial scores were significantly higher, almost normal, a result which probably indicates the extent to which expectations change in chronic disease.Lung transplantationPatients who fail to respond to medical therapy are offered the possibility of lung transplantation. The predicted survival for lung transplantation in children is 4.3 years with a 75% survival at 1 year.19Results of treatmentIn IPAH, the UK Pulmonary Hypertension Service showed survival figures of 84% at 1 year and 76% at 3 years.18 In APAH the survival rates were 89% at 1 year and 79% at 3 years. These figures compare favourably with those in international adult and paediatric studies.New and emerging therapiesThe therapeutic goal is the remodelling of the pulmonary vasculature back to normal, the restoration of endothelial function and the growth of new peripheral pulmonary arteries. New modes of administration, new formulations and new analogues of existing drugs, new endothelin receptor antagonists and new PDE 1/5/6 plus PDE-3 and PDE-1 inhibitors are in development now. Statins, RhoA inhibitors, anti-growth factor drugs, metalloproteinase inhibitors, K channel openers, stem cell and gene therapy and vasoactive intestinal peptide are all under investigation. At present we can stabilise patients for many years, but new, radical medicines used in combination offer the hope of cure and greater promise of tailoring the treatment to the individual child.\n\nREFERENCES:\n1. HaworthSG Primary pulmonary hypertension in childhood. Arch Dis Child 1998;79:452–510193263\n2. SimonneauGGalieNRubinLJ  Clinical classification of pulmonary hypertension. J Am Coll Cardiol 2004;4312 Suppl S:5S–12S15194173\n3. GalieNTorbickiABarstR  Guidelines on diagnosis and treatment of pulmonary arterial hypertension. The Task Force on Diagnosis and Treatment of Pulmonary Arterial Hypertension of the European Society of Cardiology. Eur Heart J 2004;25(24):2243–7815589643\n4. British Cardiac Society Guidelines and Medical Practice Committee Recommendations on the management of pulmonary hypertension in clinical practice. Heart 2001;86(Suppl 1):I1–1311473937\n5. BarstRJMaislinGFishmanAP Vasodilator therapy for primary pulmonary hypertension in children. Circulation 1999;99(9):1197–120810069788\n6. NagayaNNishikimiTUematsuM  Plasma brain natriuretic peptide as a prognostic indicator in patients with primary pulmonary hypertension. Circulation 2000;102(8):865–7010952954\n7. LammersAEHislopAAFlynnY  The six-minute walk test: normal values for children of 4–11 years of age. Arch Dis Child 2007 Aug 3 [Epub ahead of print].\n8. LaneKBMachadoRDPauciuloMW  Heterozygous germline mutations in BMPR2, encoding a TGF-beta receptor, cause familial primary pulmonary hypertension. The International PPH Consortium. Nat Genet 2000;26(1):81–410973254\n9. LoydJEButlerMGForoudTM  Genetic anticipation and abnormal gender ratio at birth in familial primary pulmonary hypertension. Am J Respir Crit Care Med 1995;152(1):93–77599869\n10. ThomsonJRMachadoRDPauciuloMW  Sporadic primary pulmonary hypertension is associated with germline mutations of the gene encoding BMPR-II, a receptor member of the TGF-beta family. J Med Genet 2000;37(10):741–511015450\n11. TrembathRCThomsonJRMachadoRD  Clinical and molecular genetic features of pulmonary hypertension in patients with hereditary hemorrhagic telangiectasia. N Engl J Med 2001;345(5):325–3411484689\n12. GalieNHinderliterALTorbickiA  Effects of the oral endothelin-receptor antagonist bosentan on echocardiographic and Doppler measures in patients with pulmonary arterial hypertension. J Am Coll Cardiol 2003;41(8):1380–612706935\n13. KrowkaMJPlevakDJFindlayJY  Pulmonary hemodynamics and perioperative cardiopulmonary-related mortality in patients with portopulmonary hypertension undergoing liver transplantation. Liver Transpl 2000;6(4):443–5010915166\n14. SimonneauGBarstRJGalieN  Continuous subcutaneous infusion of treprostinil, a prostacyclin analogue, in patients with pulmonary arterial hypertension: a double-blind, randomized, placebo-controlled trial. Am J Respir Crit Care Med 2002;165(6):800–411897647\n15. RubinLJBadeschDBBarstRJ  Bosentan therapy for pulmonary arterial hypertension. N Engl J Med 2002;346(12):896–90311907289\n16. MaiyaSHislopAAFlynnY  Response to bosentan in children with pulmonary hypertension. Heart 2006;92(5):664–7016216850\n17. MichelettiAHislopAALammersA  Role of atrial septostomy in the treatment of children with pulmonary arterial hypertension. Heart 2006;92(7):969–7216278272\n18. HaworthSGFlynnYHislopAA Survival and quality of life in children with severe pulmonary hypertension. Heart 2006;92Supple II:A14\n19. BoucekMMAuroraPEdwardsLB  Registry of the International Society for Heart and Lung Transplantation: tenth official pediatric heart transplantation report--2007. J Heart Lung Transplant 2007;26(8):796–80717692783"
+}

batch_5/PMC2533104.json

@@ -0,0 +1,4 @@
+{
+    "id": "PMC2533104",
+    "text": "This is an academic paper. This paper has corpus identifier PMC2533104\nAUTHORS: Konstantinos D. Pantazis, Ioannis S. Elefsiniotis, Dimitrios Papaioannou, Hero Brokalaki, Gerasimos Bonatsos, Christos Mavrogiannis\n\nABSTRACT:\nNo Abstract\n\nBODY:\nGastrointestinal disorders (especially diarrhea) are observed in a proportion of patients treated with the\ncurrently approved combination treatment for chronic hepatitis C (CHC), pegylated-interferon alpha (PEG-IFNa) plus ribavirin (RIB) [1]. Several reports\nsuggest that the presence of inflammatory bowel disease (IBD) is not a\ncontraindication for interferon-alpha- (IFNa-) based treatments [2, 3].\nFurthermore, a randomised placebo-controlled trial of PEG-IFNa in patients with active ulcerative colitis\nconcluded that PEG-IFNa is safe but not effective treatment for these patients\n[4]. On the contrary, several reports \n[5, 6, 7, 8] revealed that\ntreatment of chronic viral hepatitis with IFNa or PEG-IFNa with or without RIB\nwas related with the onset of clinically and histologically confirmed acute\ncolitis of the IBD type. To our\nknowledge the effect of chronic hepatitis C virus (HCV) infection or PEG-IFNa\nplus RIB combination treatment on large intestine histopathology has not been\ninvestigated, within a clinical trial. The principal aim of our study was to investigate\nthe effect of PEG-IFNa plus RIB treatment on the large intestine histology of\ntreated CHC patients.Twenty-four treatment-naïve CHC patients with\nserologically (antiHCV-positive, Abbott Laboratories, Abbott Park, Ill, USA), virologically (serum HCV-RNA detection, Cobas Amplicor HCV test, version 2, Roche Diagnostics, Branchburg, NJ,\nUSA), and histologically (liver biopsy-Ishak scoring system) confirmed CHC and no contraindication of receiving\ncombination treatment with PEG-IFNα2b and RIB were enrolled in this pilot study. All patients were treated with weight-based dosing of pegylated\ninterferon-a2b (Peg-Intron, 1.5μg/kg/week) and genotype-related ribavirin dose\n(Rebetol, 800 mg/day for genotype 2/3 and 1000–1200 mg/day for genotype\n1/4-infected patients, depending on baseline body weight < or ≥ 85 kg,\nresp.) for 24 weeks. Patients were evaluated for the presence of gastrointestinal symptoms,\nby receiving a detailed history, before the beginning of treatment, during the\ntreatment period and at week 24 of treatment. The study population underwent\nrecto-sigmoidoscopic examination prior to the beginning of the treatment\nschedule. Three to five biopsy specimens were taken from the rectosigmoid area.Histological findings characteristic of inflammation as well as the presence of architectural disorders and the quantity of\nmucus production, the presence of ulcerations and Paneth cells as well as the\nnumbers of lymphoid follicles in every biopsy specimen were evaluated. Fifteen\npatients underwent the same procedure after the completion of 24-week treatment\ncourse and three of them were also evaluated during treatment because of\ndiarrhea. Ten age-, sex-, and BMI-matched healthy subjects (control group)\nunderwent recto-sigmoidoscopic examination and biopsy. All controls had no\nhistory of inflammatory bowel disease or other gastrointestinal diseases, did\nnot receive any medication and did not report symptoms from the\ngastrointestinal tract. Written informed consent was obtained from each\nparticipant for his or her participation in the study. The study conformed to\nthe ethical guidelines of the 1975 Declaration of Helsinki.No pathological macroscopic (endoscopic)\nfindings were identified at baseline as well as in endoscopic re-evaluation in\neither patients or controls. CHC patients had no statistically significant\ndifference in the presence of bowel inflammatory infiltration compared to\ncontrols as shown in Table 1. It is important to note that \na respectable proportion of CHC patients\nexhibited architectural disorder and decreased mucus production (25%) as well as presence\nof Paneth cells (16.7%), compared to controls, but this does not reach\nstatistical significance possibly due to the small sample size of the study\npopulation. This finding needs further investigation because of the documental\nlymphotropism of HCV and the well-known HCV-related autoimmune manifestations.No patient reported diarrhea before the initiation of treatment whereas 3 patients\nreported diarrhea during the treatment course. In particular, the first patient\nreported diarrhea at week 6 of treatment, the second one at week 12, and the\nthird one at week 14. Baseline histology was normal in all three patients.\nHistopathological evaluation of the large intestine at the onset of symptoms\nrevealed mild chronic nonspecific colitis in the first patient (Figure 1) and\nnormal large intestine histology in the other two patients, respectively. The\nsame findings were repeated in these patients in the histological re-evaluation\nat week 24 of treatment. Diarrhea spontaneously resolved in all of them within\n5–7 days and none of them followed treatment discontinuation.There was no statistically\nsignificant difference between the percentage of patients with large intestine\ninflammation before the initiation and at week 24 of treatment (66.7% versus\n53.3% resp., P = .71). Particularly, 6/15 patients exhibited mild\ninflammatory infiltration before the initiation as well as at week 24 and 3/15\nexhibited absence of inflammation both before and at week 24 of treatment.\nMoreover, 4/15 positive patients for the presence of inflammation before\ntreatment had normal large intestine histology at week 24 and finally, 2/15\npatients had no findings of inflammation before treatment but exhibited mild\nnonspecific inflammatory lesions at week 24. No statistically significant\ndifferences were observed before the beginning and at week 24 of combination\ntreatment regarding architectural disorder and decrease of mucus production (P = .70),\npresence of Paneth cells (P = .10), and the number of lymphoid follicles\nin every biopsy specimen (P = .97) as shown in Table 2. Interestingly none\nof 4 patients with detectable Paneth cell at baseline evaluation exhibits\npresence of them at week 24. This finding needs further investigation in\nlarge-scale studies because of the proposed importance of these multifaceted\ncells in the pathophysiology of IBD [9].In conclusion, according to the preliminary results of our\npilot study, it seems that the immunomodulatory-antiviral treatment of CHC with\nPEG-IFNa2b plus ribavirin for 24 weeks possibly does not significantly affect\nthe large intestine histology of treated patients, despite the appearance of\nsymptoms from the gastrointestinal tract in a subgroup of them. The effect of\nchronic HCV infection in the intestine histopathology needs further\ninvestigation.\n\nREFERENCES:\n1. FriedMWShiffmanMLReddyKRPeginterferon α-2a plus ribavirin for chronic hepatitis C virus infectionNew England Journal of Medicine20023471397598212324553\n2. CottoneMMaglioccoATralloriGClinical course of inflammatory bowel disease during treatment with interferon for associated chronic active hepatitisItalian Journal of Gastroenterology1995271347795286\n3. BargiggiaSThorburnDAnderloniAIs interferon-α therapy safe and effective for patients with chronic hepatitis C and inflammatory bowel disease? A case-control studyAlimentary Pharmacology & Therapeutics200522320921516091058\n4. TilgHVogelsangHLudwiczekOA randomised placebo controlled trial of pegylated interferon α in active ulcerative colitisGut200352121728173314633951\n5. YamamotoYSakataniNYanoSInterferon induced IBD-like acute colitis—two cases of patients with chronic active hepatitisNippon Shokakibyo Gakkai Zasshi1995929129312967474487\n6. MavrogiannisCPapanikolaouISElefsiniotisISPsilopoulosDIKaramerisAKarvountzisGUlcerative colitis associated with interferon treatment for chronic hepatitis CJournal of Hepatology200134696496511451187\n7. VillaFRumiMGSignorelliCOnset of inflammatory bowel diseases during combined α-interferon and ribavirin therapy for chronic hepatitis C: report of two casesEuropean Journal of Gastroenterology & Hepatology200517111243124516215439\n8. SprengerRSagmeisterMOffnerFAcute ulcerative colitis during successful interferon/ribavirin treatment for chronic hepatitisGut2005543438439\n9. PorterEMBevinsCLGhoshDGanzTThe multifaceted Paneth cellCellular and Molecular Life Sciences200259115617011846026"
+}

batch_5/PMC2533298.json

@@ -0,0 +1,4 @@
+{
+    "id": "PMC2533298",
+    "text": "This is an academic paper. This paper has corpus identifier PMC2533298\nAUTHORS: Ashiq Masood, Kanan H Hudhud, AZ Hegazi, Gaffar Syed\n\nABSTRACT:\nPlasmacytoma are extramedullary accumulations of plasma cells. Most extramedullary Plasmacytomas are associated with the upper respiratory tract. The mediastinum is rarely involved. We report a rare case of mediastinal plasmacytoma with multiple myeloma. The patient is 66 year old woman presented with bone pains and mediastinal mass on CT scan and MRI. The preliminary diagnosis of occult lung cancer with mediastinal involvement, and widespread skeletal metastasis was made, although lymphoproliferative disorder along with germ cell tumor was also kept in differentials.The diagnosis of mediastinal plasmacytoma with multiple myeloma was made after extensive investigations\n\nBODY:\nIntroductionPlasmacytoma, a neoplastic proliferation of plasma cells, is a form of plasma cell dyscrasia that may manifest as multiple myeloma, primary amyloidosis, or monoclonal gammopathy of unknown significance. Plasmacytoma may be primary or secondary to disseminated multiple myeloma and may arise from osseous (medullary) or nonosseous (extramedullary) sites. Primary extramedullary plasmacytoma can be solitary or multiple [1]. The mediastinum is rarely involved by extramedullary plasmacytoma. We report a case of mediastinal plasmacytoma with multiple myeloma which is extremely rare in clinical practice. Our case highlights mediastinal plasmacytoma as differential diagnosis for mediastinal masses and aggressive search for multiple myeloma.Case presentationA 66 year old lady referred to our oncology clinic for the management of mediastinal mass. Two weeks prior to her visit, she started with severe low back pain radiating to lower extremities, nausea, retching, and mild dyspnea. She was admitted in the hospital, thorough investigation revealed mediastinal mass on CT scan (Figure 1). MRI of spine showed L4–L5 disc herniation along with significant spinal canal stenosis. She received treatment and was discharged in stable condition after 5 days of admission.Figure 1CT scan of chest showing mediastinal mass and pathological fracture of the right 8th rib.The patient's history was significant for GERD, Osteoporosis, and Hypercholestemia. She has been smoking less than a pack a day for 40 years and quit a month ago. Family history was remarkable for lung cancer (In her father, who died at the age 76) and sarcoidosis (in her only daughter). Current medications include rosuvastatin, omeprazole, Ibandronate, and ibuprofen.On her first visit to our clinic, the patient was little uncomfortable due to bone pains, her blood pressure was 104/60 mmHg, heart rate of 84 beats per minute, respiratory rate 12 breathes per minute, and body temperature 99.6 F.The cardiac and lung examination showed no murmurs, gallops, wheeze or rhonci. The rest of the examination which included HEENT, neck, abdomen, lymph nodes, and musculoskeletal was unremarkable.The MRI showed 3.9 × 4.2 cms mass in azygoesophageal recess, the pathologic fracture of 8th right rib, abnormal bone marrow signal in multiple areas of right sacroiliac joint, left superior acetabulum, and right greater trochanter. The PET scan confirmed these findings with uptake in all of these regions (Figure 2).Figure 2PET scan showing hypermetabolic activity in azygoesophageal recess and right 8th rib.The provisional diagnosis of occult lung cancer with mediastinal involvement and wide spread skeletal metastasis was made, although the diagnosis of lymphoproliferative disorder and germ cell tumor was kept in differentials.The patient underwent biopsy of mediastinal mass, which was found to be consistent with plasmacytoma (lambda light chain restricted). A bone marrow examination showed marrow involvement by plasma cell neoplasm (Overall 10% of the total cellularity). Figure 3, 4, and 5Figure 3H&E stain illustrating trilineage hematopoiesis.Figure 4CD138 stain demonstrating increased plasma cells.Figure 5Plasma cells show excess lambda chain expression.Laboratory studies showed normochromic anemia (Hemoglobin level 107 g/l, MCV 86 fl), elevated ESR (61 mm/hr) and normal white cell and blood count. Beta microglobulin was elevated (5.2 mg/L) and BUN/Creatinine (21/1.3) was normal. Serum protein electrophoresis and immunofixation electrophoresis were negative but 24 hours urine for protein electrophoresis and immunofixation electrophoresis were consistent with free lambda chain measuring 85 mg/dl.Further workup included a cytogenetic/FISH analysis (showed abnormal result for chromosome 14, consistent with the presence of clonal lymphoid hematologic malignancy) Figure 6. Flow cytometry detected CD 56 + monoclonal plasma cells (1% of nucleated cells) also consistent with plasma cell disorder. The diagnosis of indolent multiple myeloma was made. The patient was put on chemotherapy with bortezomib and dexamethasone and is planned for autologous stem cell transplant (ASCT).Figure 6FISH showing abnormal result for chromosome 14 consistent with multiple myeloma.DiscussionExtramedullary plasmacytoma (EMP) is a rare plasma cell neoplasm of the soft tissue and constitutes about 4% of all plasma cell tumors [2]. The most common site for extramedullary involvement is the upper aerodigestive tract [3]. The mediastinum is rarely involved by extramedullary plasmacytoma [3]. Only 5% of patients with EMP's have coexistent multiple myeloma [4]. However, in our case, a diagnosis of multiple myeloma was established within a month after the diagnosis of the mediastinal mass.The sequences of proceedings suggest that the mediastinal plasmacytoma provided an early hint to the diagnosis of multiple myeloma, and therefore we conclude that the multiple myeloma was coexisting with the mediastinal lesion [5]. Our case is unique in the sense mediastinal plasmacytoma with multiple myeloma presenting simultaneously is extremely rare. The plasmacytoma in our case is aggressive revealed by the enhancer uptake on PET scan in distinction to the lack of increased uptake in indolent plasmacytoma [6,7]. Another aspect of our case is that it presented like a diagnostic dilemma; we initially thought that the diagnosis of occult lung cancer with mediastinal involvement with widespread skeletal metastasis. However, diagnosis of plasmacytoma with multiple myeloma was reached after extensive investigations.After diagnosis of plasmacytoma aggressive search for multiple myeloma is vital as the management is entirely different for both types of Plasma Cell DyscrasiasPlasmacytomas are treated with radiotherapy, surgery or both [8]. Chemotherapy may be considered for patients with refractory or relapsed disease [8] whereas multiple myeloma is mostly treated with chemotherapy [9]. Extranodal plasmacytoma in a patient with multiple myeloma carries a poor prognosis and treatment, which includes chemotherapy or autologous hematopoietic cell transplantation which is directed towards the underlying disease [10]. Our case also demonstrates the clinical usefulness of PET/CT scan in imaging plasmacytoma [6].AbbreviationsFISH: Fluorescent in situ hybridization; ASCT: Autologous stem cell transplantation.Patient consentWritten informed consent was obtained from the patient for publication of this case report and accompanying images. A copy of the written consent is available for review by the Editor-in-Chief of this journal.Competing interestsThe authors declare that they have no competing interests.Authors' contributionsKHH conceived the study and provided substantial contributions to the analysis and interpretation of data. AM, the lead author involved in carrying out the literature search, study design and writing of case report. AZH and GS assisted with writing the paper. AZH also provided valuable insights in study design. KHH was involved in the diagnosis and management of case. All authors have read and approved the final manuscript.\n\nREFERENCES:\nNo References"
+}

batch_5/PMC2533302.json

@@ -0,0 +1,4 @@
+{
+    "id": "PMC2533302",
+    "text": "This is an academic paper. This paper has corpus identifier PMC2533302\nAUTHORS: Carmen Alaez, Ling Lin, Hilario Flores-A, Miriam Vazquez, Andrea Munguia, Emmanuel Mignot, Reyes Haro, Harry Baker, Clara Gorodezky\n\nABSTRACT:\nBackgroundNarcolepsy-cataplexy is characterized by excessive daytime sleepiness with recurrent episodes of irresistible sleep, cataplexy, hallucinations and sleep paralysis. Its aetiology is unknown, but it is positively associated with the human leukocyte antigens (HLA) in all studied populations. The purpose of the present study was to investigate the association of HLA class II DRB1/DQB1 alleles with narcolepsy-cataplexy in Mexican Mestizo patients.MethodsThis is a case-control study of consecutive patients and ethnically matched controls. We included 32 patients diagnosed with typical narcolepsy-cataplexy, of the National Institute of Neurology, of the Institute of Psychiatry and at the Center of Narcolepsy at Stanford University. As healthy controls, 203 Mexican Mestizos were included. DRB1 alleles were identified using sequence based typing. A PCR-SSOP reverse dot blot was used for DQB1 typing. Allele frequency was calculated by direct counting and the significance of the differences was assessed using the Yates Chi square. Odds ratio and confidence intervals were evaluated.ResultsHLA-DRB1*1501 (OR = 8.2; pc < 0.0001) and DQB1*0602 (OR = 8.4; pc < 0.0001) were found positively associated with narcolepsy. When deleting DQB1*0602+ patients from the analysis, DQB1*0301 was also found increased (OR = 2.7; p = 0.035; pc = NS). DQB1*0602/DQB1*0301 genotype was present in 15.6% of the cases (OR = 11.5; p = 0.00035), conferring a high risk. DRB1*0407 (OR = 0.2; p = 0.016 pc = NS) and DQB1*0302(OR = 0.4; p = 0.017, pc = NS) were found decreased in the patients. The gender stratification analysis showed a higher risk in females carrying DRB1*1501 (OR = 15.8, pc < 0.0001) and DQB1*0602 (OR = 19.8, pc < 0.0001) than in males (OR = 5.0 for both alleles; p = 0.012, pc = NS for DRB1 & p = 0.0012, pc = 0.017 for DQB1). The susceptibility alleles found in Mexicans with narcolepsy are also present in Japanese and Caucasians; DRB1*04 linked protection has also been shown in Koreans. A stronger HLA association is suggested in females, in accordance with the sexual dimorphism claimed previously.ConclusionThis knowledge may contribute to a better understanding of the disease pathogenesis in different populations. The evaluation of the risk to develop narcolepsy-cataplexy in carriers of the described alleles/genotypes may also be possible. A larger sample should be analysed in Mexican and in other Hispanic patients to confirm these results.\n\nBODY:\nBackgroundNarcolepsy is a serious chronic neurological sleep disorder affecting between 0.02–0.05% of the general Caucasian population [1]. It is characterized by excessive daytime sleepiness, cataplexy, hypnagogic hallucination, sleep paralysis, and nocturnal fragmented/disorganized sleep [2,3]. Narcolepsy causes cognitive dysfunction, low academic performance and interpersonal problems [4]. Cataplexy is a pathognomonic clinical symptom required for the diagnosis of narcolepsy, according to the International Classification of Sleep Disorders [5]. It is a symptom that occurs when the muscle tension, in various areas of the body, is suddenly decreased involuntarily and lasts from few seconds to several minutes. Cataplexy is generally induced by laughter, excitement anger, and other emotional changes [5]. Mutations in genes of the hypocretin (orexin) neurotransmitter system cause narcoleptic symptoms in animal models [6]. Although most patients with narcolepsy-cataplexy have a reduction of hypocretin concentration in the cerebrospinal fluid [7], mutations or polymorphisms in hypocretin-related genes are extremely rare [8]. A potential autoimmune mechanism has been suggested, supported by the finding that in most narcolepsy patients, 80–90% of hypocretin cells in the hypothalamus were destroyed [9]. The lost of hypocretin neurons [10] has also been shown in a few post-mortem cases. In addition, the transcription of preprohypocretin mRNA was significantly decreased in the brain of these patients [11]. This wake-promoting neuropeptide is involved in sleep regulation, energy homeostasis, reward-seeking, learning, and memory; and it is also involved in the body temperature regulation, endocrine function, and the cardiovascular system, among other systems [12]. Recent studies also indicate that hypocretin/orexin neurons can alter their intrinsic electrical activity according to fluctuations in the levels of nutrients and appetite-regulating hormones [13].Narcolepsy transmission is polygenic, environmentally influenced and genetic factors play an important role in its expression. Family studies indicate the presence of a 20–40 times increased risk of disease expression in first-degree relatives, and monozygotic twin studies showed that concordance is partial (25–31%) [1]. Early onset French and Chinese patients suffering from narcolepsy have shown a positive family history as compared with late-onset patients [14,15], suggesting that the disease may be more likely due to genetic factors in this subgroup of patients.Based on the strong HLA association, family study segregation [16,17] and the finding of reduction of hypocretin-1 levels in the cerebrospinal fluid of DQB1*0602 positive patients [18,19], an autoimmune mediated destruction of hypothalamic neurons secreting hypocretin has been suggested as the cause of the disease. However, serum autoantibody markers have not been detected and no immunological abnormalities have been found in patients with narcolepsy [20].The disease is strongly associated in Caucasians and Japanese with the DRB1*1501-DQA1*0102-DQB1*0602 haplotype. In African Americans, it is associated with DQB1*0602 haplotypes bearing different DRB1 alleles (*1101 and *1503) suggesting that DQA1 and DQB1 play a primary role in susceptibility [17]. DQB1*0602 rather than DRB1*1501 has been found increased in the patients, indicating that the disease susceptibility allele or locus is within, or, in the vicinity of the DQ region [21]. Haplotype analysis and contiguous genomic sequencing across the region have identified no other candidate gene [22]. Two-five fold increased risk in DQB1*0602 homozygous vs. heterozygote patients has been demonstrated in different ethnic groups [23]. DQB1*0301/*0602 carriers are also at an increased risk, whereas DQB1*0602/*0601 and DQB1*0602/*0501 heterozygotes have a lower disease risk [23-25].Several authors reported that 85 to 95% of patients with narcolepsy carry DQB1*0602 when cataplexy is clinically typical or severe. However, only 40–60% of the patients are DQB1*0602 when mild, atypical or no cataplexy exists [25,26]. TNF alpha region has also been claimed to be involved in susceptibility independently of class II loci [27].Genetic factors in other chromosomes have also been implicated [28]. A gender dimorphism and a strong effect of the catechol-O-methyltransferase (COMT) genes seem to influence symptoms. COMT genotype distribution between male and female patients was associated with the response to modafinil in Caucasians, since the optimal dose of modafinil was approximately 100 mg lower in females with narcolepsy, suggesting that females are better responders to the drug [29].The HLA genetic profile of Mexican and other Central American patients has not been published; therefore the aim of this study was to investigate the class II-DRB1/DQB1 allele distribution in a group of sporadic Mexican Mestizo patients with narcolepsy and to explore if the HLA association is gender related.MethodsSubjectsA total of 32 patients (14 males (45.17%) and 17 females (54.83%) were included. (The gender of one patient was unknown). All the cases were chosen and diagnosed based on the International Classification of Sleep Disorders (ICSD) [5] using clinical histories, nocturnal polysomnography, and Multiple Sleep Latency Tests. It is important to emphasize that the inclusion and exclusion criteria were given to every participant in the Narcolepsy Component of the 13th International Histocompatibility Workshop, and every centre must have used the same questionnaire and the same classification given by the experts [17]. The study was approved by the Ethics and Research Committees of each hospital. An informed consent was signed by every patient included in the study. Nine patients were selected from the Sleep Clinic at The Instituto Nacional de Neurologia in Mexico City, 12 were diagnosed at Stanford University Sleep Centre, and 11 were diagnosed at the Hospital Medica Sur in Mexico City. All patients were selected under exactly the same criteria in terms of ethnicity and diagnosis. Two hundred and three healthy controls belonging to the same population were included for comparison. These controls were all selected at the Department of Immunology and Immunogenetics in Mexico City and had no personal or family antecedents of narcolepsy. The patients and controls were all Mexican Mestizos defined according to the criteria of the Instituto de Investigaciones Antropologicas of the Universidad Nacional Autonoma de Mexico, UNAM [30]. All of them were Mexicans with at least parents, grandparents and great-grandparents born in Mexico; having Hispanic last names. Any individual with a non Hispanic background was excluded from the study. This ethnic group is the result of the admixture of Mediterranean, Black and Native genes [30,31]. The patients diagnosed in the USA followed exactly the same criteria, and most of them were in fact, patients from Mexico City from a previous collaborative done by Mignot E and Baker H. (Personal Communication). The controls were healthy subjects with no history of HLA confirmed associated diseases. Subjects with excessive daytime sleepiness, other sleep disorders, circadian or mental disorders, medication or substance abuse were carefully excluded. Each subject underwent a detailed clinical interview and completed a set of questionnaires according to The International Classification of Sleep Disorders [5].HLA-DRB1 &DQB1 typingAll patients and controls were typed at the Immunogenetics Department in Mexico City with the same technology and under the same technical conditions for each technique used. Ten mL peripheral EDTA blood were drawn from each patient coming from the Mexican Institutions. DNA was isolated using proteinase K digestion; purification was achieved with phenol chlorophorm and isopropanol was used for precipitation. DNA of patients diagnosed at Stanford University, were sent to our laboratory in Mexico City for HLA typing. DQB1 was typed on DNA samples from 32 patients, using a commercial kit based on Polymerase Chain Reaction and hybridization with membrane immobilized Sequence Specific Oligonucleotide Probes (PCR-SSOP reverse dot blot). DRB1 Sequence Based Typing (SBT) was performed in all patients using AlleleSEQR DRB1 kits, kindly donated by Atria Genetics. DRB1 and DQB1 alleles were typed in the control samples by PCR-SSOP using a chemiluminescent detection method designed for the 13th IHW (International Histocompatibility Workshop) [32].DRB1 typing was performed in 27/32 samples because the amount of DNA was not enough. In one patient gender was unknown; therefore gender stratification analysis included only 31 patients.Haplotype assignment was done on the basis of the very well known DRB1-DQB1 linkage disequilibrium data published by us in Mexican Mestizos [31]. It was assumed that no blanks were present; when a single HLA allele was found; in this case, the subject was considered homozygous. Rare haplotype associations were taken into account, only when the complementary haplotype was perfectly defined.Statistical AnalysisAllele frequency was calculated by direct counting. The Chi square test with Yates correction was used to assess the statistical difference of the HLA allele distribution between patients and controls. Bonferroni correction of the p value (pc) was done multiplying the p value by the number of comparisons made (equal to the number of tested alleles). This correction made more stringent the statistical significance of the results. Odds ratio (OR), was calculated when a significant association between the particular allele and the disease was found. Confidence interval (CI) (95%) was defined for every statistical deviation and is shown in each table. HLA related gender stratification was done using the SPSS 11 software.Analysis of DQB1*0602 negative patientsTo assess whether HLA alleles, other than DQB1*0602 were associated with narcolepsy, the allelic frequency between patients and controls was compared, after taking out from the analysis the DQB1*0602 positive patients.ResultsHLA AssociationOnly two DRB1 alleles were found significantly deviated in the patients (Table 1). DRB1*1501 was significantly increased (OR = 8.2; CI = 3.9–17.6; p < 0.0001; pc < 0.0001). The frequency of DRB1*0407 was significantly decreased in them (OR = 0.2; CI = 0.02–1.2; p = 0.016), although the significance was lost after Bonferroni correction. DQB1*0602 which is in linkage disequilibrium with DRB1*1501 was associated with susceptibility (OR = 8.4; CI = 4.3–16.6; p < 0.001; pc < 0.001) and the frequency of DQB1*0302 was decreased in the patients (Table 2).Table 1HLA-DRB1 distribution in Mexican Mestizo patients with narcolepsy and in healthy controls.LocusTotal PatientsTotal ControlsDRB1*N = 27†AF%N = 203AF%X2YOR(CI)ppc010100112.70.6010200174.21.301030020.50.3100111.830.70110135.61230.411020041011030010.21.4110423.792.2012010041013010092.20.3130211.8102.50130323.7411130411.810.20.313050010.21.414010051.20140223.7143.40.1140600245.92.315011527.8184.435.68.2(3.9–17.6)<0.0001<0.0001150223.751.20.6150323.720.52.6150411.8001.4160100410160211.8215.20.5030135.6184.4003020030.70.104010010.21.404020071.70.1040335.671.71.7040435.6174.200405001230.7040711.86014.85.80.2(0.02–1.2)0.016NS04100020.50.3041123.71230070147.4276.7008010061.50.1080235.65112.61.6080411.8001.4080611.8001.409010020.50.3AF: allele frequency; X2Y: Chi2 value with Yates correction; OR: Odds Ratio; CI: Confidence Interval; pc: Bonferroni correction (the p value was multiplied by the total number of alleles tested); NS: Not significant; †5/32 patients were not typed for DRB1 locus.Table 2HLA-DQB1 distribution in Mexican Mestizo patients with narcolepsy and in healthy controls.LocusTotal PatientsTotal ControlsDQB1*N = 32AF%N = 203AF%X2YOR(CI)ppc0201812.54310.60.103011625.010024.60.00302812.511127.35.70.4(0.2–0.9)0.017NS030323.151.20.4040257.86014.81.7050123.1358.61.605020051.20.105030051.20.1060100410.006022132.8225.446.78.4(4.3–16.6)<0.0001<0.000106030071.70.306040071.70.3060511.610.20.2060911.610.20.2AF: allele frequency; X2Y: Chi2 value with Yates correction; OR: Odds Ratio; CI: Confidence Interval; pc: Bonferroni correction (the p value was multiplied by the total number of alleles tested); NS: Not significant.Upon withdrawal from the analysis of DQB1*0602 positive patients and controls, DQB1*0301 was found associated with the disease (Table 3) since 9/11 remaining patients were carriers of this allele (OR = 2.7; CI = 1.1–6.5; p = 0.0035; pc = Non significant (NS). We calculated the DQB1 genotype distribution in order to know the risk conferred by different combinations (Table 4). Only two genotypes were associated with narcolepsy: the highest risk was conferred by the combination DQB1*0602/DQB1*0301 (OR = 11.5; CI = 2.6–50.7; p = 0.00035), present in 15.6% (5/32) of the patients. DQB1*0602/X (excluding *0301) was also found significantly increased with a lower risk (OR = 9.5; CI = 4.1–21.9; p < 0.0001). This combination was positive in 50% of patients (16/32). No DQB1*0602 homozygotes were found.Table 3HLA-DQB1 distribution in Mexican Mestizo DQB1*0602 negative patients and controls.LocusPatientsControlsDQB1*ΛN = 11AF%ςN = 181AF%X2YOR(CI)ppc0201418.23810.50.6030111**509726.84.42.7(1.1–6.5)0.035NS0302313.610428.71.703030051.40.2040229.15515.20.2050114.5339.10.105020051.40.205030051.40.206010041.10.306030071.9006040071.90060514.510.31.406090010.31.4AF: allele frequency; X2Y: Chi2 value with Yates correction; OR: Odds Ratio; CI: Confidence Interval; pc: Bonferroni correction (the p value was multiplied by the total number of alleles tested); NS: Not significant; Λ: 11/32 total patients were DQB1*0602 negative; ς: 181/203 total controls were DQB1*0602 negative. ** this number includes two DQB1*0301 homozygotes.Table 4HLA-DQB1 genotypes distribution in Mexican Mestizo patients with narcolepsy and controls.DQB1* GenotypesN = 32GF%N = 203GF%X2YOR(CI)p0602/X(not *0301)1650.0199.432.99.5(4.1–21.9)<0.00010602/0301515.631.512.811.5(2.6–50.7)0.000350301/030126.3157.40.010301/X(no 0602 or 0301)928.18240.41.3X/X26.39948.818.70.1(0.0–0.4)<0.0001GF: genotype frequency; X2Y: Chi2 value with Yates correction; OR: Odds Ratio; CI: Confidence Interval.Gender StratificationAlthough DRB1*1501 and DQB1*0602 were increased in both, females and males, the risk was higher in females: DRB1*1501 (OR = 15.8 CI = 4.5–55.7, p < 0.0001, pc < 0.0001 in females) vs. (OR = 5 CI = 1.6–15.4; p = 0.012, pc = NS in males) and DQB1*0602, (OR = 19.8, CI = 5.9–66.9, p < 0.0001, pc < 0.0001 in females) vs. (OR = 5, CI = 1.9–13, p = 0.0012, pc = 0.017 in males). DQB1*0302 was found significantly decreased only in males (OR = 0.3, CI = 0.1–1, p = 0.043, pc = NS).DiscussionThe results of this study are considered preliminary, due to the small sample size. However, it is important to mention that we are not \"Hypothesis generating\" but \"hypothesis testing\". As demonstrated by many statistician experts in the HLA field: \"If an association is detected in the first case, it can be tested and confirmed in the latter without having to correct in multiple comparisons\" [33]. Nevertheless, future studies in Hispanic admixed populations are needed to confirm the presence of other HLA allele associations. It is worth to mention, that DRB1/DQB1 association has been tested and repeatedly demonstrated by many authors in different ethnic groups [23-25,34]; therefore, the associations shown here, are real. In this study, we report on the genetic profile in a sample of narcoleptic Mestizo patients. DRB1*1501 (OR = 8.2; pc < 0.0001) and DQB1*0602 (OR = 8.4; pc < 0.0001) were the strongest associated alleles found in narcoleptic Mexicans, similarly to African, White Americans and different Oriental groups [23-25,34-37]. Fifteen of the 27 DRB1 typed patients, were positive for DRB1*1501 (allele frequency = 27.8 vs. 4.4% in the controls) and 21 of the 32 DQB1 typed patients were positive for DQB1*0602 (allele frequency = 32.8% vs. 5.4% in the controls). In five DQB1 typed patients, DRB1 was not typed because of insufficient DNA, thus we were not able to assemble the DRB1-DQB1 combinations in them. Interestingly, the DRB1*1501-DQB1*0602 haplotype was present in most of the DQB1*0602 positive patients but other DR2 allele combinations were also found. Indeed, two narcoleptic patients had DRB1*1502-DQB1*0602 and two had DRB1*1503-DQB1*0602. A DRB1*1503 but not DRB1*1502 association with DQB1*0602 has been reported in patients from Martinique [38] and in African Americans [23]. These results show, beyond doubt, that DQB1*0602, rather than DRB1*1501 is the major narcolepsy susceptibility allele in Mestizos. Interestingly, none of the DR52 associated-DQB1*0602 haplotypes were found in Mexican patients, but their frequency is low in the general healthy population. As an example, in 160 Mexican Mestizo healthy individuals typed in our laboratory, the haplotype frequency was 0.31% for each of the following combinations: DRB1*1101-DQB1*0602, DRB1*1201-DQB1*0602 and DRB1*1301-DQB1*0602 (unpublished data). Some of these haplotypes have been reported in narcoleptic patients in other populations, most notably in African Americans [23].No increase in DQB1*0602 homozygosity was found, as previously reported in White Americans and in African American patients, in whom a two to four fold higher risk has been described, compared to heterozygotes [23]. The same has been shown in Japanese patients [24]. This fact cannot be explained only based on the different allele frequencies of DQB1*0602 across ethnic groups, since the frequency shown in Japanese (AF = 6.4%) [24] was similar to the one found in the control group of the present study (AF = 5.5%). The lack of homozygote patients may be due to the reduced number of cases. Thus, again, a larger number is needed to confirm these results.As in other populations, other DQB1 alleles, beside DQB1*0602, influence narcolepsy susceptibility [24,39]. The analysis of DQB1 distribution in DQB1*0602 negative patients showed that DQB1*0301 (OR = 2.7, p = 0.03) was significantly increased in this subgroup of patients. DQB1*0301 had also the second strongest susceptibility effect, after DQB1*0602 in Africans, Japanese and White Americans [24]. In Mexican patients, DQB1*0301 occurred in the context of several HLA haplotypes that included DRB1*1101, *1303, *1304, *0806 and *1602; however the number of patients in this group was insufficient to perform additional comparisons and to explain the possible independent contribution of the mentioned DRB1 alleles to susceptibility. None of these patients had the DRB1*04-DQB1*0301, found associated in White Americans [24]. Genotype distribution in patients and controls showed that DQB1*0602/DQB1*0301 conferred the highest risk for susceptibility (OR = 11.4) compared to DQB1*0602/X (non *0301) (OR = 9.4). The former combination was also described as the one with the highest risk for the development of narcolepsy, across three different ethnic groups [24].DRB1*0407, which is the most frequent allele in Mexican population, seemed to be linked to protection in the present study. This allele is in strong linkage disequilibrium with DQB1*0302 in Mexicans and the haplotype frequency of DRB1*0407-DQB1*0302 is 13.1% [31]. In Koreans DRB1*0406-DQB1*0302 was found protective in patients with narcolepsy [39]. None of these studies confirmed a possible effect of DRB1*04 in susceptibility as previously shown in Whites and Japanese [24]. It may be claimed that the DQB1 locus may also be involved in protection, since in Korean as well as in Mexican patients, DQB1*0302 was decreased, although combined with *0406 in Koreans and with *0407 in Mexicans. DQB1*0601 has also been associated with protection in Koreans [25,39] and Japanese [40], but the latter was not related with protection in Mexicans, perhaps due to its low frequency (AF = 0.5%) [31]. Similar findings have been published recently in Koreans, where again, DRB1*0406 was found negatively associated and DQB1*0301 was described as a susceptibility allele. The authors claim based on their own work, and previous work, that a remarkable consistency of the HLA association pattern across multiple ethnic groups and cultures exists [41]. Thus, even if our sample size is small and the results of protection and secondary association may be regarded as preliminary, our data are consistent with those published [21,23-26,39,41].Interestingly, the overall positive rate for DQB1*0602 in Mexican patients was 65.6%, while in Japanese, White Americans and African Americans; the rate is between 75–80% [24]. To analyse if this difference was significant or not, we performed a statistical comparison between patients and controls from the present study and those from the Mignot et al. study [24]. No significant difference, regarding DQB1*0602 distribution, was found between Japanese and Mexican controls but we did find a significant deviation when comparing Mexican controls with White or African Americans healthy people (p = 0.001 and p = 0.00000001, respectively). These differences are due, undoubtedly, to the lower frequency of DQB1*0602 existing in the general Mexican population [42], which is similar to the one found in Orientals, but it is lower than the frequency in Caucasian and African Americans [43]. The lower frequency of the DQB1*0602 allele in Mexican patients compared with African (p = 0.0007), Caucasian (p = 0.003) and Japanese (p = 0.0009) patients, is due to our small number of cases.Gender stratification showed a differential distribution of DRB1 and DQB1 alleles. The patients selection was unbiased since no significant difference was found when distribution of males Vs. females was compared (Z = 0.561, p = 0.29). To demonstrate if the gender selection among the controls was biased, we performed several analyses comparing the mean allele frequency in the overall control group with the frequency for HLA DQB1*0602 among male and female patients. However, the distribution of DQB1*0602 alleles among female and male controls was found in the limit of the significance (X2Y = 3.841, p = 0.05; data not shown). Therefore, it must be mentioned that even if the p value was p = 0.05, the control selection may have been biased. When HLA distribution was compared matching for gender, DRB1*1501 (OR = 15.8; pc < 0.0001) and DQB1*0602 (OR = 19.8; p < 0.0001) showed a higher risk in female patients, although the same alleles were significantly increased in males, but with less intensity. This stronger female HLA association was described for the first time. However, it is important to confirm these data with a larger sample size and in different ethnic groups. Thus, the gender association reported here should be regarded as preliminary. Even if we have a stronger HLA DRB1/DQB1 association in females, this does not imply, by any means, that females are more or less affected than males; this only means that there is a stronger HLA genetic predisposition to narcolepsy in females than in males. A better response in affected women to certain drugs [29] is not contradictory at all, with the HLA/female association.The explanation for this gender specific association is not clear, but a sexual dimorphism and a strong effect of the COMT genotype on disease severity and response to modafinil have been shown [44]. Females with narcolepsy with high COMT activity fell asleep twice as fast as those with low COMT activity during the multiple sleep latency test, while the opposite was true for men [45]. A gender difference in body weight gain and leptin signaling in hypocretin/orexin deficient mouse models has been also claimed [46]. Obesity was more prominent in females in both preprohypocretin knockout mice and orexin/ataxin-3 transgenic narcoleptic mice and was associated with higher serum leptin levels, suggesting a partial leptin resistance [46].ConclusionIn conclusion, this is the first study reported in Mexican patients with narcolepsy-cataplexy; the profile of HLA class II allele associations was found to be complex. As in other diseases such as type I diabetes, multiple sclerosis and others, studied by our group in Mexican Mestizos, the HLA pattern in these diseases was somewhat distinct from the association typically found in Caucasians or Blacks and Orientals [47,48], illustrating the importance of analysing MHC associations in populations with different ethnic backgrounds. We also demonstrated an HLA associated sexual dimorphism in this population and a protective allele effect which was also shown in some Oriental populations. Insights on the different HLA associations in different ethnic groups may prove to be an important asset in the investigation of genetic factors and the molecular mechanisms of disease expression. This knowledge may be important for the design of predictive, therapeutic and perhaps preventive approaches.Competing interestsThe authors declare that they have no competing interests.Authors' contributionsCA Drafted the manuscript and supervised the laboratory work. LL Collaborated in drafting the manuscript and extracted the DNA at Stanford University. HF-A Carry out the statistical analysis and revised the drafted manuscript. MV Collaborated with the technical HLA DNA typing and revised the drafted manuscript. AM Collaborated with the technical HLA DNA typing and revised the drafted manuscript. EM Selected the clinical cases and wrote the criteria for all Centres and for the 13th International Histocompatibility Workshop and revised the drafted manuscript. RH Diagnosed and selected the cases from The Instituto Nacional de Neurologia and revised the drafted manuscript. HB Diagnosed and selected the cases from and revised the drafted manuscript. CG Coordinated all the work, revised the results and statistical analysis, revised the drafted manuscript thoroughly; revised and corrected the final version. All authors read and approved the final manuscript.Pre-publication historyThe pre-publication history for this paper can be accessed here:\n\nREFERENCES:\nNo References"
+}

batch_5/PMC2533306.json

@@ -0,0 +1,4 @@
+{
+    "id": "PMC2533306",
+    "text": "This is an academic paper. This paper has corpus identifier PMC2533306\nAUTHORS: Michiel M Winter, Flip JP Bernink, Maarten Groenink, Berto J Bouma, Arie PJ van Dijk, Willem A Helbing, Jan GP Tijssen, Barbara JM Mulder\n\nABSTRACT:\nBackgroundThe method used to delineate the boundary of the right ventricle (RV), relative to the trabeculations and papillary muscles in cardiovascular magnetic resonance (CMR) ventricular volume analysis, may matter more when these structures are hypertrophied than in individuals with normal cardiovascular anatomy. This study aimed to compare two methods of cavity delineation in patients with systemic RV.MethodsTwenty-nine patients (mean age 34.7 ± 12.4 years) with a systemic RV (12 with congenitally corrected transposition of the great arteries (ccTGA) and 17 with atrially switched (TGA) underwent CMR. We compared measurements of systemic RV volumes and function using two analysis protocols. The RV trabeculations and papillary muscles were either included in the calculated blood volume, the boundary drawn immediately within the apparently compacted myocardial layer, or they were manually outlined and excluded. RV stroke volume (SV) calculated using each method was compared with corresponding left ventricular (LV) SV. Additionally, we compared the differences in analysis time, and in intra- and inter-observer variability between the two methods. Paired samples t-test was used to test for differences in volumes, function and analysis time between the two methods. Differences in intra- and inter-observer reproducibility were tested using an extension of the Bland-Altman method.ResultsThe inclusion of trabeculations and papillary muscles in the ventricular volume resulted in higher values for systemic RV end diastolic volume (mean difference 28.7 ± 10.6 ml, p < 0.001) and for end systolic volume (mean difference 31.0 ± 11.5 ml, p < 0.001). Values for ejection fraction were significantly lower (mean difference -7.4 ± 3.9%, p < 0.001) if structures were included. LV SV did not differ significantly from RV SV for both analysis methods (p = NS). Including structures resulted in shorter analysis time (p < 0.001), and showed better inter-observer reproducibility for ejection fraction (p < 0.01).ConclusionThe choice of method for systemic RV cavity delineation significantly affected volume measurements, given the CMR acquisition and analysis systems used. We recommend delineation outside the trabeculations for routine clinical measurements of systemic RV volumes as this approach took less time and gave more reproducible measurements.\n\nBODY:\nBackgroundThe number of adult patients with a congenital heart defect is steadily increasing with the constant improvement of cardiac surgery. A substantial portion of these patients has a morphologic right ventricle (RV) supporting the systemic circulation (e.g. patients with a congenitally corrected transposition of the great arteries (ccTGA) or a complete transposition of the great arteries (TGA) after an atrial switch operation). Medium term survival in patients with a systemic RV is relatively good. However, long-term outcome is unknown, and morbidity is worrisome, with RV dysfunction, tricuspid valve regurgitation, and arrhythmias being the main constituents. [1-4]Assessment of ventricular function is of particular importance, as European guidelines consider it an important part of preoperative assessment, (peri-operative) management and follow-up of patients with any congenital heart defect. [5] Moreover, decisions on the timing of surgical intervention in patients with a systemic RV are frequently based on RV function. [6,7] Therefore, the importance of having an accurate and reproducible diagnostic tool for the evaluation and follow-up of systemic RV volumes and function is evident.For the assessment of subpulmonary and systemic RV volumes and function Cardiovascular Magnetic Resonance (CMR) is considered the gold standard. [8-15] In individuals with normal cardiac anatomy the influence of trabeculations and papillary muscles on measured ventricular volumes seems of marginal importance. The differences in measured ventricular volumes and function when these structures are included in the ventricular cavity, compared to when structures are excluded, are small and unlikely to influence clinical decision making.[16,17] However, patients with a systemic RV pose a challenge, as the method of delineating the cavity relative to the hypertrophied trabeculations and papillary muscles could affect RV volume and function measurements. [18]To our knowledge, no study has ever addressed the issue of CMR analysis methods in this patient group. Aim of the present study was to evaluate the impact of trabeculations and papillary muscles on systemic RV measurements, by comparing a CMR analysis method in which trabeculations and papillary muscles were included in the RV volume to an analysis method in which these structures were excluded from RV volume (figure 1). Additionally, differences in analysis time and intra- and inter-observer reproducibility between analysis methods were evaluated.Figure 1Four chamber image from a multi-phase steady-state free precision sequence of the highly trabeculized systemic RV in a patient with an atrially switched TGA (a) and in a patient with a congenitally corrected TGA (b).MethodsStudy populationA total of 29 adult patients (69% male, mean age 34.7 ± 12.4 years) with a systemic RV underwent CMR for the evaluation of RV volumes and function. Twelve patients had a ccTGA, 17 patients an atrially switched TGA. The Institutional Review Boards of all three participating tertiary referral centers approved the study protocol. Written informed consent was obtained from all patients prior to participation in the study.Image acquisitionImage acquisition was performed by CMR, using a 1,5 Tesla scanner (Siemens Avanto, Erlangen, Germany), using standardly available sequences to assess ventricular volumes. After visualizing the long and short axes of the heart, a multi-phase steady-state free precession sequence (SSFP) with retrospective electrocardiographic triggering was applied to visualize two-chamber, three-chamber and four-chamber views. Guided by these views, a multislice and multiphase SSFP sequence was applied perpendicular to the ventricular septum, encompassing the total heart. These sequences were individually adjusted to acquire short axis slices with optimal spatial and temporal resolution. Typical parameters were: flip angle: 50–70 degrees; repetition time: 3–4 msec; echo time: 1–2 msec; temporal resolution: 40 msec, 1–2 × 1–2 mm/pixel in-plane spatial resolution, 8 mm slice thickness, and 1 mm interslice gap. This resulted in 9 to 15 slices to cover the whole heart. CMR images were acquired during repeated end-expiratory breath holds.Image analysesFor CMR image analysis two independent observers (MW, FB) used MASS Analytical Software System (Medis, Leiden, the Netherlands). Cine loops were used to choose end diastole (ED) and end systole (ES). ED was defined as the phase with the largest RV (and left ventricular (LV)) area and ES as the phase with the smallest RV (and LV) area. The slices at the base of the heart were considered to be in the ventricle if the blood was at least half surrounded by ventricular myocardium. Cine loop movies in phase and slice were used in case the distinction between the ventricles, atria and great vessels was unclear. Moreover, four-chamber views in phase with the short axis views were available. Tracing was performed manually on each ED and ES short-axis view.The sums of the traced contours in ED en ES were used to calculate ED volume (EDV) and ES volume (ESV) using a disc summation technique. EDV and ESV were used to calculate Stroke Volume (SV) and Ejection Fraction (EF). SV was defined as EDV - ESV, and EF as [(EDV - ESV)/EDV] × 100%.CMR analysis methodsAll contours were traced twice, using two different tracing methods. Both the systemic RV, and the subpulmonary LV were subjected to both analysis methods. Method A: Contour tracing was first performed including the papillary muscles and trabeculations in the ventricular cavity, by tracing immediately within the apparently compact layer of the myocardium. A continuous movie display of the slice being evaluated was used to enhance differentiation between trabeculations, papillary muscles and the ventricular free wall. Although the exclusion of complex RV trabeculations and papillary muscles is relatively easy in ED, optimal differentiation is especially important in the ES phases as trabeculations and papillary muscles compress and fold during systole making the end-systolc border of trabeculations and blood volume less distinct. This method resulted in smooth contours (figure 2, method A). Method B: Contour tracing was then performed excluding the papillary muscles and trabeculations from the ventricular volume. This was done by tracing around these structures if attached to the ventricular wall and by tracing them separately if not attached to the ventricular wall. This resulted in irregular endocardial contours (figure 2, method B).Figure 2Assessment of systemic right ventricular volumes using two different analysis protocols. Short axis view from a multi-phase steady-state free precision sequence in end systole (left) and end diastole (right) obtained in a patient with an atrially switched TGA demonstrating the two analysis protocols. Method A depicts the inclusion of trabeculations and papillary muscles in the ventricular cavity. Method B depicts the exclusion of trabeculations and papillary muscles from the ventricular cavity. LV = left ventricle.To ensure that found differences would only be due to the impact of papillary muscles and trabeculations both analysis methods were performed on the same ED and ES phases, and the same slices for each patient. Duration of RV volume and function analysis was recorded for both analysis methods.Statistical methodsFor statistical analyses SPSS 12.0.1 (SPSS Inc., Chicago, Illinois) for Windows was used. P values < 0.05 were considered statistically significant. The shift between the two tracing methods was compared with the two-tailed paired t test, calculating mean, standard deviation and statistical significance of the differences. The agreement between the two tracing methods was assessed and visualized with the method and plots as described by Bland and Altman. [19] Agreement between RV SV and LV SV for both analysis methods was assessed using a two-tailed paired t test.Intra- and inter-observer reproducibility of the two analysis methods was determined from the mean value and the differences between the two measurements. The coefficient of variability was calculated as the standard deviation of the difference of the paired measurements divided by the mean of the average of the paired measurements, and expressed as a percentage. An extension of the Bland-Altman method was used to assess the statistical significance of differences in intra- and interobserver reproducibility between the two analysis methods. A log transformation of the squared differences between the two measurements was performed. If the squared difference was zero, we replaced the value by the next smallest value multiplied by 0.5. A two-tailed paired t test of the logged squared differences was performed thereafter.[16]ResultsDifferences in measured volumes and function (Additional file 1)Including trabeculations and papillary muscles in the systemic RV volume (Method A) resulted in significantly higher outcome measures for EDV with a mean difference of 28.7 ml (95% CI 24.7 – 32.7, p < 0.001), and for ESV with a mean difference of 31.0 ml (95% CI 26.7 – 35.4, p < 0.001), compared to when structures were excluded from the ventricular volume (Method B). This resulted in a significantly lower calculated systemic RV EF with a mean difference of 7.4% (95% CI -8.9 – -5.9, p < 0.001). No significant changes were found for SV and CO. Bland-Altman plots were used to visualize the systematic differences between Method A and B (figure 3).Figure 3Bland-Altman plots demonstrating the systematic differences in measured systemic right ventricular volumes and function between Method A and Method B. Bland-Altman plots demonstrate on the X-axis the mean value of Method A and Method B for each parameter (a. mean end diastolic volume; b. mean end systolic volume; c. mean stroke volume; d. mean ejection fraction), and on the Y-axis the difference between the two analysis methods for the same parameter. The solid line represents the mean value of the difference for each method, the dotted lines represent ± 2 SD. RVEDV = right ventricular end diastolic volume; RVEF = right ventricular ejection fraction; RVESV = right ventricular end systolic volume; ml = millilitre; SD = standard deviation; RVSV = right ventricular stroke volume. The 'A' or 'B' behind parameters indicate that values were obtained using Method A or Method B respectively.We found no statistically significant differences between RV SV and LV SV, when trabeculations and papillary muscles were included in the RV and LV volumes (75.5 ± 18.5 ml vs. 71.1 ± 23.9 ml; p = NS), neither when structures were excluded from the RV and LV volumes (77.8 ± 18.4 ml vs. 69.3 ± 22.3 ml, p = NS).Analysis time was significantly shorter when using Method A (20 ± 3 min) compared to Method B (26 ± 4 min); p < 0.001.Observer reproducibilityAnalysis Method A lead to a lower coefficient of variability for all outcome measures compared to Method B. This indicates superior intra- and inter-observer reproducibility when including trabeculations and papillary muscles in the ventricular volume compared to excluding these structures. Although these differences were not found to be statistically significant for the intra-observer measurements, we found statistically significant difference in the inter-observer reproducibility of systemic RV SV (p < 0.05) and EF (P < 0.01), favoring the inclusion of structures in the ventricular volume (Additional file 2).DiscussionIncluding trabeculations and papillary muscles in the systemic RV cavity lead to a substantially higher measured EDV and ESV and a substantially lower calculated EF, compared to excluding these structures from the volume of the cavity. Although the influence of these structures on measured ventricular volumes in individuals with normal cardiac anatomy seems of marginal clinical importance,[16,17] their influence on systemic RV volumes was found to be striking. Moreover, there were statistically significant differences in analysis time and in reproducibility between CMR analysis methods.Although CMR is considered the most accurate diagnostic tool for the assessment of ventricular volumes and function, [11-13,20,21], there is no consensus in the literature on the role of trabeculations and papillary muscles. In anatomically normal hearts, Lorenz et al. excluded trabeculations and papillary muscles from the LV and the RV cavity,[12] whereas Rominger et al. chose to include these structures.[13] Most authors, however, refrained from specifying the role of trabeculations and/or papillary muscles in their analysis method. [22-25] Similar incompleteness in methodology of CMR analysis is seen in literature regarding patients with congenital heart defects, and with systemic RVs in specific. Although Helbing et al. excluded papillary muscles and the moderator band from the RV cavity of patients with congenital heart defects,[10] and Lidegram et al. chose to include both structures in the cavity of the systemic RV,[26] most authors are less specific on the role of these structures in their analysis method.[11,21,27]The influence of trabeculations and papillary muscles on LV and RV measurements in healthy subjects and patients with known cardiac disease (patients with congenital heart defects were excluded) has been studied previously, using similar analysis protocols as were used in our study.[16,17] In anatomically normal hearts, both Sievers et al. and Papavassiliu et al. found significant differences in measured left and right ventricular volumes and function when comparing both CMR analysis protocols. However, both authors concluded that the observed differences in ventricular volumes and function were too small to influence clinical decision making, and advised the inclusion of all structures in the ventricular cavity. Including structures not only shortened analysis time, Papavassiliu et al. also demonstrated superior reproducibility for several outcome measures when using this analysis protocol.[16,17] Our observations in systemic RVs differ from those of Sievers and Papavassiliu, as we found systematic and large differences in measured systemic RV volumes and function between the two analysis methods. This study indicates the importance of a consistent approach to cavity delineation relative to the trabeculations and papillary muscles, to avoid misinterpretation of measurements and erroneous clinical decision making.[5,6]Although the true values of systemic RV volumes and function remain unknown, and in spite of tricuspid valve regurgitation in some patients, we found no significant differences between the measured RV and LV SV by either method of analysis. However, delineation of the RV cavity boundary outside the trabeculations and papillary muscles had the advantages of shorter analysis time and better inter-observer reproducibility. We therefore recommend the use of this approach in routine CMR measurements of systemic RV volumes, at least when comparable systems for CMR acquisition and volume analysis are being used.Study limitationsThe sample size of 29 patients in two distinct clinical categories is relatively small. The methods used were not suitable for determining which of the analysis approaches measured systemic RV volumes more accurately. Measurements of systemic RV mass were not attempted, and remain challenging given the relative amount of trabeculated RV myocardium. The slice thickness of 8 mm may not have been optimal for clear delineation of the trabeculations, making the definition of the boundaries between trabeculations and blood, and between trabeculations and apparently compact myocardium hard to define in some cases. Moreover, as myocardial boundaries were first defined at end diastole, detecting the corresponding boundary at end systole could be difficult, due to elimination of blood from the inter-trabecular spaces at end systole. The important issue of inter-study reproducibility was not addressed by this study. Between studies, volume measurements might be affected by variation in the relative positioning of the basal short axis slice, and by variables such as the shimming of the magnet and the reliability of ECG triggering.ConclusionWe found the method of systemic RV cavity delineation to affect the measurements of cavity volume, given the CMR acquisition and analysis systems used. We recommend cavity delineation inside the apparently compact myocardium of the RV but outside the trabeculations and papillary muscles for routine clinical measurements of systemic RV volumes as this approach took less time and gave more reproducible values.AbbreviationsccTGA: congenitally corrected transposition of the great arteries; CHD: congenital heart defect; CMR: cardiovascular magnetic resonance; CO: cardiac output; EDV: end diastolic volume; EF: ejection fraction; ESV: end systolic volume; LV: left ventricle; RV: right ventricle; SV: stroke volume; TGA: transposition of the great arteriesCompeting interestsThe authors declare that they have no competing interests.Authors' contributionsMW designed the study format, carried out the literature search, the data collection, and analysis, the statistical analysis and the interpretation of data, carried out the manuscript writing, and gave final approval. FB designed the study format, carried out the literature search, the data collection, and analysis, the statistical analysis and the interpretation of data, carried out the manuscript writing, and gave final approval. MG substantially contributed to the conception and design of the study, performed data acquisition, critically revised the article, and gave final approval. BB substantially contributed to the conception and design of the study, the analysis of data, critically revised the article, and gave final approval. AD substantially contributed to the conception and design of the study, the analysis of data, critically revised the article, and gave final approval. WH substantially contributed to the conception and design of the study, the analysis of data, critically revised the article, and gave final approval. JT substantially contributed to the conception and design of the study, the statistical analysis of data, critically revised the article, and gave final approval. BM substantially contributed to the conception and design of the study, the analysis of data, the interpretation of data, critically revised the article, and gave final approval.Supplementary MaterialAdditional file 1Measurements of systemic right ventricular volumes and function by mean of CMR.Click here for fileAdditional file 2Intra- and inter-observer reproducibility of measurements.Click here for file\n\nREFERENCES:\nNo References"
+}

batch_5/PMC2533337.json

@@ -0,0 +1,4 @@
+{
+    "id": "PMC2533337",
+    "text": "This is an academic paper. This paper has corpus identifier PMC2533337\nAUTHORS: James Demetrious, Gregory J Demetrious\n\nABSTRACT:\nBackgroundThe objective of this case report is to describe the clinical presentation of a patient who complained of shoulder pain and was diagnosed with carcinoma of the scapula and spine that metastasized from the lung.Case presentationA 76-year-old man without a history of cancer sought chiropractic care for right shoulder pain. Careful evaluation, radiographs, and subsequent imaging revealed primary and metastatic lung cancer. The patient was referred to his primary care physician for immediate medical care. Diagnostic images are included in this case to provide a comprehensive depiction of the scope of the patient's disease.ConclusionMusculoskeletal symptoms are commonly encountered in chiropractic practice. It is important to recognize that primary lung cancer may be unidentified, and musculoskeletal symptoms may reflect the first sign of primary or metastatic pulmonary disease. Thoughtful evaluative procedure and clinical decision making, combined with the use of appropriate diagnostic tests may allow timely identification of primary or metastatic disease.\n\nBODY:\nBackgroundIn the USA, more people die from lung cancer than any other type of cancer [1]. This is true for both men and women. In 2004, lung cancer accounted for more deaths than breast cancer, prostate cancer, and colon cancer combined [2].Lung cancer can metastasize to virtually any bone, although the axial skeleton and proximal long bones are most commonly involved [3]. The primary symptom resulting from bone involvement is pain, which may have a pleuritic component when the ribs are involved. Bone pain is present in up to 25% of all patients at presentation [3].Patients commonly seek chiropractic care with musculoskeletal complaints [4,5]. Through history and examination, chiropractic physicians have an opportunity to assess patients and determine whether serious conditions are present that may necessitate medical referrals.Patients with previously identified or yet to be identified cancer may seek care with chiropractic physicians. This case report demonstrates previously undiagnosed lung cancer with widespread metastatic foci.Case presentationCase reportA 76-year-old male sought chiropractic care for complaints of right shoulder pain and mild right arm weakness. The onset of pain was insidious and of one week's duration. Pain was rated 8/10 on a visual analogue scale (0 = no pain, 10 = the worst pain of one's life). The pain was described as severe and worsened with movement. Additional symptoms included mild shortness of breath and posterior thoracic pain on respiration.The patient's past medical history included headache, degenerative joint disease affecting the cervical spine, and a benign thyroid nodule. The patient reportedly smoked tobacco products for 50 years. He was a retired electrician.The patient was afebrile. Vital signs were normal. Respirations were 18 cycles per minute. The lungs were clear to auscultation. The patient reported upper thoracic pain on inspiration.A non-tender, mild decrease in active range of motion of the cervical spine was noted in all planes. No tenderness was elicited on palpation of the cervical spine. Cervical compression and Soto-Hall tests were negative. Valsalva maneuver was negative. Neurologic examination revealed no focal deficits.Examination of the right shoulder revealed exquisite tenderness on palpation of the lateral border of the scapula with muscle spasm affecting the ipsilateral infraspinatus, teres major, and teres minor muscles. Active ranges of shoulder motion were restricted and painful in abduction, internal, and external rotation.Plain film radiographs of the right shoulder (AP with internal and external rotation views) and thoracic spine (AP and lateral views) were performed. Disruption of the cortical margin of the lateral border of the right scapula was noted as evidenced by an indistinct lucency (see Figure 1). In addition, a suspicious mass was noted in the hilar region of the right lung. Complete loss of the right hilar vascular detail secondary to the tumor mass effect were noted with visualized subsegmental infiltrate densities. No evidence of pleural effusion was noted.Figure 1AP radiograph of the right scapula reveals a focal indistinct lucency and lytic destruction of the lateral scapular cortical margin.The initial diagnostic impression included: suspicious right lung pathology and apparent lytic process affecting the scapula of an unknown origin. The patient was referred for imaging evaluations that included chest x-ray (CXR) and computed tomographic (CT) evaluation of the chest. He was referred to his primary care medical physician.The CXR and CT examination of the chest, abdomen and pelvis revealed:1. A large mass in the right upper lobe of the lung with associated mediastinal and hilar adenopathy (see Figures 2 and 3).Figure 2PA chest radiograph reveals a right hilar mass.Figure 3CT of the chest reveals a large mass in the right upper lobe of the lung with associated mediastinal and hilar adenopathy.2. Metastatic disease of the scapula (see Figure 4).Figure 4CT of the chest reveals cortical lucency, expansile destruction, and medullary invasion due to metastatic lung carcinoma affecting the right scapula.3. Metastatic liver disease.Subsequent bone scintigraphy revealed abnormal increased accumulation of radiopharmaceutical along the lateral aspect of the right scapula (see Figure 5). MRI evaluation revealed additional metastatic foci including the cervical, thoracic and lumbar spinal regions as evidenced by multiple regions of decreased signal intensity are visualized on T1 weighted images (see Figures 6 and 7). Biopsy confirmed a primary lung carcinoma origin. Unfortunately, the patient succumbed to the disease within 3 months of its diagnosis.Figure 5Bone scintigraphy of the right scapula reveals increased uptake where metastatic lung carcinoma is present.Figure 6MRI sagittal T1WI reveals scattered foci of decreased signal intensity reflective of metastatic disease affecting the cervical and thoracic spine regions.Figure 7MRI sagittal T1WI reveals scattered foci of decreased signal intensity reflective of metastatic disease affecting the thoraco-lumbar spine.DiscussionChiropractic considerationsThe identification of primary or secondary metastatic cancer requires careful consideration with regard to history and physical examination. A key objective for the chiropractic physician is to identify \"red flags\" as quickly as possible. This is especially true for any disease process that may weaken bone.The application of directed force into spinal or osseous structures inherent to the chiropractic adjustment mandate careful evaluative procedure. Janse defined the adjustment as a specific form of articular manipulation using long or short lever techniques with specific contacts and is characterized by a dynamic thrust of controlled velocity, amplitude and direction [6].While chiropractic physicians are challenged with the responsibility of attempting to identify relative and absolute contraindications to spinal adjustments, sometimes early onset, insidious and seemingly innocuous symptoms may delay early identification [7,8].Clinical considerationsWhen primary cancer is not yet identified, metastatic extension to skeletal structures can at times be difficult to detect [7,8]. As was illustrated in this case, clinical considerations that may assist or delay the identification of metastatic bone disease include:1. Early in the course of the disease progression, important red flag identifiers may not initially be present and can delay early identification.2. Initial pain presentations may be suggestive of common clinical conditions that are less aggressive.3. Patients may or not be aware of, or report, the existence of a primary cancer.4. Pain can be initially mild to severe and is often progressive in nature and unremitting despite therapeutic interventions.5. It is sometimes extremely difficult to positively identify metastatic disease due to complex clinical factors [7,8].Red flag indicators for metastatic bone disease include: age over 50 or under 20 years, a history of cancer, constitutional symptoms including unexplained weight loss, pain worse at night or in atypical areas, no significant improvement after > 1 month of conservative (non-invasive) care, pain that has no mechanical exacerbating or remitting factors, and severe disabling pain affecting a child or adolescent [9].Diagnostic imaging considerationsHumphrey reported that about 25% of people with lung cancer do not have symptoms from advanced cancer when their lung cancer is found [10]. Maghfoor reported that 7–10% of patients with lung cancer are asymptomatic and their cancers are diagnosed incidentally after a CXR was performed for other reasons [11]. Numerous studies have shown that the chest radiograph lacks sensitivity in detecting mediastinal lymph node metastases and in detecting chest wall and mediastinal invasion [12].CT has become the major imaging modality of choice in the evaluation of patients with bronchogenic carcinoma [13]. Traditionally, chest CT for staging of lung cancer is extended into the abdomen to include the adrenal glands. Whether this requires intravenous contrast material is debatable [13]. Patz et al. [14] concluded that contrast-enhanced CT extended to include the liver rarely adds to routine nonenhanced CT through the adrenal glands and does not influence management decisions.The evaluation of the mediastinum with magnetic resonance imaging (MRI) is approximately equal to that of CT with regard to the staging of bronchogenic carcinoma and MRI is significantly more accurate for detecting direct mediastinal invasion [15]. Other studies have confirmed the usefulness of MRI, particularly in the evaluation of chest wall invasion and the local staging of superior sulcus tumors [16,17]. The general conclusion of these studies is that MRI has advantages in the assessment of both chest wall and mediastinal invasion [13].Indications for the use of whole body positron emission tomography imaging in lung cancer using 18-fluorodeoxyglucose (FDG-PET) in patients with non-small cell lung cancer include high clinical index of suspicion of high grade malignancy and radiographic evidence of nodal enlargement [13]. In addition, PET scans may be helpful in centers where mediastinoscopy is not readily available and in patients with significant comorbid conditions who are borderline candidates for surgery, with locally advanced disease, solitary brain metastasis, and cases of local recurrence that might qualify for reoperation [18,19].Bone scintigraphy in the detection of metastatic disease has significant limitations. Although it has high sensitivity, it is noted for having very low specificity that ranges from 50%–60% [13]. Bone scintigraphy should probably be limited to cases in which patients have specified clinical indicators of bone metastasis [20].When evaluating suspected pulmonary metastasis, CXR and CT of the chest are rated by the American College of Radiology (ACR) scale as: \"9 – most appropriate\" (Rating Scale: 1-Least appropriate, 9-Most appropriate) [21]. It is generally accepted that chest radiography, with posteroanterior (PA) and lateral views, should be the initial imaging test in patients without known or suspected thoracic metastatic disease [22-24]. Compared with chest radiography, CT is much more sensitive for detecting pulmonary nodules, because of its lack of superimposition and its high contrast resolution [22-24].ConclusionLung cancer is a significant and aggressive primary cancer with a predilection for skeletal metastasis. When primary lung cancer is not previously identified, metastatic disease to skeletal structures may initially manifest as musculoskeletal complaints. Careful diagnostic evaluation and decision making may allow for earlier diagnosis.Competing interestsThe authors declare that they have no competing interests.Authors' contributionsJD conceived the study and drafted the manuscript. GJD participated in the care of the patient and provided data related to the case. Both authors read and approved the final manuscript.\n\nREFERENCES:\nNo References"
+}

batch_5/PMC2533348.json

@@ -0,0 +1,4 @@
+{
+    "id": "PMC2533348",
+    "text": "This is an academic paper. This paper has corpus identifier PMC2533348\nAUTHORS: Vijay Ganji, Jennifer Kuo\n\nABSTRACT:\nBackgroundCardiovascular disease is the leading cause of death in women and men. Psyllium, a soluble fiber has been known to reduce serum lipids. In this pilot study, we evaluated whether menopausal status would affect the serum lipid responses to psyllium fiber in women.MethodsEleven post-menopausal and eight pre-menopausal women with serum total cholesterol >200 mg/dL were included in the study. Subjects consumed their habitual diet and 15 g psyllium/d for 6 weeks. Psyllium was incorporated into cookies. Each cookie contained ≈5 g of psyllium fiber. Subjects ate one cookie in each meal.ResultsWith psyllium fiber, total cholesterol concentration was significantly lower (≈5.2%, P < 0.05) in post-menopausal women but not in pre-menopausal women (≈1.3%). Also, there was a significant decrease in HDL-cholesterol in post-menopausal women (≈10.2%, P < 0.05). There were no significant changes observed in concentrations of LDL-cholesterol, triglycerides, apolipoprotein A1, and apolipoprotein B in both pre- and post-menopausal women with psyllium.ConclusionIn this pilot study, post- and pre-menopausal, hypercholesterolemic women responded differently to psyllium fiber supplementation. Post-menopausal women would benefit from addition of psyllium to their diets in reducing the risk for heart diseases. The results of this study should be used with caution because the study was based on a small sample size.\n\nBODY:\nIntroductionCardiovascular disease (CVD) is the leading cause of death among women and men. About 13 million Americans have CVD and approximately 450,000 people die each year in the US from CVD [1]. Risk factors for CVD in women include post-menopausal status, age, hypertension, smoking, and diabetes [2]. The onset of menopause coincides with elevated serum lipids such as total cholesterol, LDL-cholesterol, and triglycerides and decreased HDL-cholesterol [3]. Elevated serum lipids in post-menopausal women are partly due to the loss of estrogen [4]. Hormone therapy showed a reduction in LDL-cholesterol and an increase in HDL-cholesterol [5]. It also frequently increased triglycerides by increasing the production of triglyceride-rich VLDLs [6]. Additionally, hormone therapy increased the risk for stroke, breast cancer, and cholecystitis [7]. Hormone replacement therapy is not for all post-menopausal women [8].Generally, relative to drug therapy, dietary intervention is less expensive and cost effective for primary intervention of heart diseases [9]. Consumption of foods containing dietary fiber, may improve the long-term maintenance of low atherogenic LDL-cholesterol [10]. Psyllium (Plantago ovata) seed husk fiber, a widely used soluble fiber, has been known to reduce serum total cholesterol and LDL-cholesterol [11-13]. Additionally, psyllium fiber supplementation with 10 mg of simvastatin (hypocholesterolemic drug) was as effective as 20 mg of simvastatin alone [14]. This suggests that psyllium fiber is an effective adjuvant hypocholesterolemic agent.There is a paucity of data on how fiber therapy modulates the risk for CVD in women. Previous studies conducted in women indicated that dietary therapy for lowering serum lipids differs according to menopausal status [15]. In men compared to post-menopausal women, a greater decrease was observed in total cholesterol after 4 months of high fiber diet [16] suggesting that gender differences exist in hypocholesterolemic effect of dietary fiber. Limited studies were conducted comparing lipid responses to psyllium fiber intervention in pre- and post-menopausal women. Therefore, we tested the hypothesis of whether serum lipid responses to psyllium fiber intake in post-menopausal, hypercholesterolemic women differ from pre-menopausal, hypercholesterolemic women.MethodsSubjectsA total of 25 hypercholesterolemic women (13 pre-menopausal and 12 post-menopausal) were recruited for the study. Hypercholesterolemia was defined as having serum cholesterol >200 mg/dL. Subjects were recruited from San Francisco and surrounding communities through posting of flyers. Participants lived in their homes during the study. The exclusion criteria were smoking, diabetes, pregnancy, breast feeding, allergic to psyllium fiber, and blood cholesterol ≤200 mg/dL. Individuals with thyroid and gastrointestinal diseases, CVD, chronic alcohol and drug abuse, individuals who were taking lipid-lowering and/or anti-hypertensive medications, and individuals who were on estrogen replacement therapy were excluded from the study. Also, strict vegetarians (vegans) were excluded from the study because their habitual fiber intake is generally higher than non-vegetarians. Post-menopausal status was defined as women who were into menopause at least one year from their last menstruation at the beginning of the study. Subjects were asked to continue their habitual diet and to maintain their pre-study physical activities and body weights. Study protocol and informed consent were approved by the Human Subjects Protection Committee at San Francisco State University.Administration of psyllium fiberSubjects were asked to consume psyllium fiber enriched cookies for 6 weeks. We incorporated 15 g of psyllium fiber into cookies (≈5 g/cookie). Subjects consumed 3 cookies/d (one in each meal). Psyllium fiber enriched cookies were made in the departmental food laboratory, according to the formulation given in Table 1. Previous studies documented that 4–6 week feeding period was adequate to observe changes in serum lipids with psyllium fiber intake [17,18]. The reason that psyllium fiber was incorporated into cookies rather than administering as supplements because fiber has shown to produce maximum cholesterol-lowering effect when it is taken mixed with foods [19]. Numerous products (muffins and powdered drink mixes) incorporating psyllium were tested for palatability and cookies yielded the most appetizing product. Cookies were individually supplied to subject's home in a fresh or frozen form on a weekly basis. The rationale for the amount of fiber used (15 g/d) was based on the recommendation of 25–35 g/d of fiber intake for adults [20]. Since Americans consume about 15 g/d of fiber [11], by adding 15 g/d subjects would meet the recommendations for fiber. During the study, subjects were asked to refrain from consuming fiber supplements. If a subject was a social drinker that subject was asked to maintain her usual alcohol intake level.Table 1Composition of psyllium fiber-enriched cookies*IngredientsAmount#Psyllium fiber, g250All purpose flour, g252Sugar, g294Brown sugar, g316.5Shortening, g484Eggs, g200Baking soda, g4.68Vanilla extract, mg3.5* = Procedures: Mixed dry and wet ingredients well. Weighed the dough and divided by 50. Baked at 330°F for 11–14 minutes until cookies became light golden brown. # = Made 50 cookies. Each cookie contained approximately 5 g of psyllium fiber.Serum lipid analysisTwelve-hour overnight fasting blood samples were obtained from venipuncture in the arm at the beginning and at the end of the 6-week period by a certified phlebotomist. Blood samples were centrifuged and serum was separated before lipid analysis. Serum samples were analyzed for total cholesterol, LDL-cholesterol, HDL-cholesterol, triglycerides, apolipoprotein A-1, and apolipoprotein B using an automatic analyzer (Quest Diagnostic Laboratories, Dublin, CA).Data analysisData were presented as mean ± standard deviation. Paired t-test was used to determine the significant difference between the baseline and the post-fiber treatment values in pre- and post-menopausal women (Microsoft Excel for Windows, 1999). Statistical significance was set at P < 0.05.ResultsOf 25 subjects, 19 (8 pre- and 11 post-menopausal) women completed the study. Reasons for discontinuation of participation included gastrointestinal discomfort, increased frequency of bowel movements, softer stools, and personal travel and time conflicts. Subjects' characteristics were summarized in Table 2. As expected, the post-menopausal women were significantly older compared to pre-menopausal women (p < 0.05). Body mass index (BMI) at baseline between pre- and post-menopausal women was not significantly different. BMI for all subjects were within the desirable low health risk range. There were no significant differences in body weight between baseline to post-study measurements in both pre- and post-menopausal women (data are not shown).Table 2Characteristics of subjects*VariablePre-menopausal womenPost-menopausal womenRace-ethnicity White/Hispanic/Asian, n1/1/60/0/11Age, y34.6 ± 11.552.9 ± 2.8#Weight, kg63.2 ± 9.357.6 ± 5.6@Height, m1.6 ± 0.11.6 ± 0.04@BMI, kg/m224.2 ± 4.222.6 ± 2.2@* = Mean ± standard deviation. # = Significant difference from pre-menopausal women (p < 0.05).@ = No significant difference from pre-menopausal women.The baseline and post-study serum lipid concentrations for pre- and post-menopausal women are presented in Table 3. Mean serum total cholesterol concentration with psyllium fiber intake was significantly lower compared to baseline values in post-menopausal women (218.7 mg/dL vs. 230.8 mg/dL; p < 0.05) but not in pre-menopausal women (240.5 mg/dL vs. 243.6 mg/dL). The decrease in post-menopausal women with respect to total serum cholesterol was ≈5.2%.Table 3Serum lipid responses to psyllium fiber intake in pre- and post-menopausal women*Serum lipidPre-menopausal women (n = 8)Post-menopausal women (n = 11)Total cholesterol Baseline, mg/dL243.6 ± 40.3230.8 ± 24.4 Post-fiber, mg/dL240.5 ± 41.2218.7 ± 17.0#LDL-cholesterol Baseline, mg/dL156.0 ± 46.0133.4 ± 32.3 Post-fiber, mg/dL151.0 ± 42.0127.3 ± 14.3HDL-cholesterol Baseline, mg/dL61.8 ± 12.265.0 ± 17.0 Post-fiber, mg/dL61.4 ± 24.158.4 ± 14.8#Triglycerides Baseline, mg/dL129.5 ± 52.1139.5 ± 72.1 Post-fiber, mg/dL139.4 ± 77.4165.4 ± 85.2Apolipoprotein A-1 Baseline, mg/dL152.5 ± 30.3150.6 ± 21.8 Post-fiber, mg/dL154.4 ± 39.2152.3 ± 23.8Apolipoprotein B Baseline, mg/dL122.3 ± 34.7110.6 ± 19.7 Post-fiber, mg/dL117.1 ± 31.5106.3 ± 16.5* = Mean ± standard deviation. # = Significantly different from pre-menopausal women, paired t-test (p < 0.05).Mean HDL cholesterol was significantly lower in post-menopausal women with fiber intake (65 mg/dL to 58.4 mg/dL) (p < 0.05). In contrast, no change was observed in HDL-cholesterol concentrations in pre-menopausal women with psyllium. Changes in LDL-cholesterol in pre- and post-menopausal women from baseline to post-study were not significant although there was a slight decline in values with psyllium fiber intake. The decline was ≈4.6% in post-menopausal women and was ≈3.2% in pre-menopausal women.There was a ≈7.6% and ≈18.6% increase in serum triglycerides in pre- and post-menopausal women, respectively, with psyllium fiber supplementation. However, these changes were not statistically significant. Similarly, there were no significant changes in response to psyllium fiber intake in serum apolipoprotein A-1 and apolipoprotein B concentrations in both pre- and post-menopausal women.DiscussionWe investigated the effect of psyllium fiber intake at 15 g/d dosage level for 6 weeks on serum lipids in free-living, pre- and post-menopausal, hypercholesterolemic women. Psyllium fiber significantly lowered serum total cholesterol in post-menopausal women. Decreased total serum cholesterol was primarily due to changes in the HDL-cholesterol. Although LDL-cholesterol was lower (≈4.6%) with fiber intake, the decrease was not statistically significant when compared to baseline values in post-menopausal women. Concentrations of serum triglycerides, apolipoprotein A-1, and apolipoprotein B were not affected by psyllium fiber intake in post-menopausal women. Also, psyllium fiber intake had no effect on serum total cholesterol, LDL-cholesterol, HDL-cholesterol, triglycerides, apolipoprotein A-1, and apolipoprotein B in pre-menopausal, hypercholesterolemic women.Previously, we documented cholesterol-lowering effect of psyllium in normocholesterolemic humans [18]. Other studies have shown that consumption of psyllium fiber reduced serum total cholesterol and LDL-cholesterol in hypercholesterolemic subjects [12,13]. Several other investigators reported a 3–17% reduction of serum total cholesterol and a 4–20% reduction of LDL-cholesterol with psyllium fiber [12,19,21,22]. In a meta-analysis, Anderson et al. (2000b) reported that consumption of 10.2 g psyllium/d lowered total cholesterol by 4%. Whereas, in our study, we achieved a ≈5.2% decrease in total serum cholesterol concentrations at a dosage level of 15 g psyllium/d. Most of the psyllium fiber used in previous studies was either in the form of ready-to-eat cereal or as fiber supplement. In our study, we administered psyllium fiber in cookies. Thus, the method of administration of psyllium might account for some differences in cholesterol-lowering property of psyllium fiber in various studies.Davidson et al. [10] reported that consumption of psyllium fiber lowered total cholesterol without modifying HDL-cholesterol concentrations. In contrast, our study did not reveal that psyllium fiber reduced serum total cholesterol without lowering HDL-cholesterol. The discrepancy in results between this study and other studies may likely be due to differences in characteristics of subjects. Also, our subjects consumed self-selected diets during the study. Thus, the diet intake might have some influence on the study outcomes.In this study, we found the hypocholesterolemic effect of psyllium fiber in post-menopausal women but not in pre-menopausal women indicating that pre- and post-menopausal women responded differently to psyllium fiber. Lack of significant hypocholesterolemic effect of psyllium fiber in pre-menopausal women suggests that the elevated cholesterol concentrations in those subjects might be due to genetic causes and thus are unresponsive to dietary modification. On the other hand, elevated cholesterol in post-menopausal women is often due to deranged lipid metabolism (non-genetic causes) arising from loss of estrogen function and thus is responsive to psyllium fiber intervention. In post-menopausal women with established CVD, higher intake of fiber has been associated with less progression of coronary atherosclerosis [23]. Also, psyllium fiber significantly reduced post-prandial concentrations of serum triglycerides, glucose, and insulin in persons with diabetes [24] suggesting a role for psyllium fiber in the management of diabetes.Previous studies on the effect of psyllium fiber on triglycerides yielded equivocal results [22,25,26]. Psyllium fiber intake significantly reduced plasma triglycerides in men, but in post-menopausal women, it resulted in a significant increase with no change in pre-menopausal women [26]. In our study, we found no significant effect of psyllium fiber intake on serum triglycerides in both pre- and post-menopausal women, although there was a trend towards increased serum triglycerides.The lack of effect of psyllium fiber intake on serum lipids in pre-menopausal women might be due insufficient number of subjects recruited to observe statistically significant differences. Large sample size might reveal significant total cholesterol and/or LDL-cholesterol reduction. Therefore, the results of this study must be used with caution. Although, we have instructed the subjects not to alter their dietary habits during the study, it is possible that our results may have confounded by the fiber from unintended food sources. Because psyllium fiber was incorporated into cookies, consumption of psyllium enriched cookies lead to an additional intake of ≈30 g/d fat. This may have blunted the hypocholesterolemic effect of psyllium in both pre- and post-menopausal women. Additionally, all subjects in the post-menopausal group were Asians. It is not known how this has affected the cholesterol-lowering effect of psyllium.In conclusion, post- and pre-menopausal women differed with regard to lipid responses to psyllium fiber. Serum total cholesterol was responsive to psyllium fiber in post-menopausal women. Post-menopausal women would likely benefit from addition of psyllium fiber to their diets in reducing the risk of heart diseases because every 1% reduction in serum total cholesterol concentration results in 2% reduction in risk for heart diseases [27]. Differences in cholesterol-lowering response to psyllium intake may be related to the differences in hormonal status between pre- and post-menopausal women [15,26]. Additionally, consumption of psyllium fiber provides a low-cost adjunct to the National Cholesterol Education Program diets for hypercholesterolemia.Competing interestsThe authors declare that they have no competing interests.Authors' contributionsVG and JK directed the research. VG was responsible for getting the support for the project. Both authors are responsible for collecting the data, analysis of data, and drafting the manuscript. All authors read and approved the manuscript before submission.\n\nREFERENCES:\nNo References"
+}

batch_5/PMC2533435.json

@@ -0,0 +1,4 @@
+{
+    "id": "PMC2533435",
+    "text": "This is an academic paper. This paper has corpus identifier PMC2533435\nAUTHORS: Nikolas Pietrzik, Daniel Schmollinger, Thomas Ziegler\n\nABSTRACT:\nCopper-catalyzed, thermal or microwave promoted 1,3-dipolar cycloaddition (Click Reaction) of 2-propynyl and 3-butynyl 2,3,4-tri-O-acetyl-6-azido-6-deoxy-glycopyranosides in the D-gluco, D-galacto and D-manno series afford the corresponding dimeric cycloaddition products.\n\nBODY:\nIntroductionOur ongoing interest in constructing combinatorial libraries of highly glycosylated beta-peptides that can mimic specific oligosaccharide-protein interactions prompted us to further search for efficient routes toward glycosylated amino acid building blocks derived from asparaginic acid in which the glycon is bound to C-1 of the asparaginic acid through variable spacers (Figure 1). Previously, we have prepared a series of glycosylated asparaginic acid building blocks containing as spacers either simple alkyl chains [1], or amino alcohols [2–3]. Such building blocks have been shown to be well suited for combinatorial solid phase or spot synthesis of libraries of highly glycosylated peptides, some members of which were indeed shown to behave like oligosaccharide mimics capable to specifically bind lectins [1,4].Figure 1Schematic representation of glycosylated building blocks for the combinatorial synthesis of glycopeptides.In order to increase the structural diversity of the aforementioned building blocks, we contemplated using as the spacer entity 1,2,3-triazoles which are known to be easily generated through a copper-catalyzed 1,3-dipolar cycloaddition of an organic azide and an alkynyl derivative (Click Reaction) [5–7]. For review articles on copper-catalyzed Click Reactions see references [8–11]. Recently, we applied this approach to a series of 1,2,3-triazole containing per-O-acetyl-glycosides which were prepared by copper-catalyzed 1,3-dipolar cycloaddition either between fully acetylated propargyl 1-thio-glycosides and t-butyl (S)-4-azido-3-fluorenylmethyloxycarbamido-butyrate or between Fmoc-L-Asp(OtBu)-propargyl amide and 2,3,4,6-tetra-O-acetyl-glycosyl azides and ethyl 2,3,4-tri-O-acetyl-6-azido-6-deoxy-1-thio-glycosides, respectively [12]. In order to increase the structural diversity of glycosyl amino acid building blocks containing 1,2,3-triazole spacers even more, we next looked at the possibility to use glycosides bearing both, azido and alkynyl groups in copper-catalyzed 1,3-cycloadditions. The results are presented here.Results and DiscussionFirst, 2-propynyl 6-azido-6-deoxy-2,3,4-tri-O-acetyl-β-D-glucopyranoside (4a) was prepared by the following sequence. 2-Propynyl 2,3,4,6-tetra-O-acetyl-β-D-glucopyranoside (1a) [13] was Zemplén-deacetylated with a catalytic amount of sodium methanolate in methanol. Next, thus obtained crude 2-propynyl β-D-glucopyranoside was regioselectively tosylated at position 6 [14] followed by chromatographic purification to afford 6-O-p-tolylsulfonyl-glucoside 2a in 66% yield. Acetylation of the latter with acetic anhydride in pyridine gave crude tri-O-acetyl-6-O-p-tolylsulfonyl-glucoside 3a which was sufficiently pure for the next step. Treatment of 3a with NaN3 in DMF finally afforded 6-azido-6-deoxy-glucoside 4a in 38% yield. When glucoside 4a was reacted with asparaginic propargyl amide derivative 5 [12] in the presence of (EtO)3PCuI as catalyst and with or without microwave irradiation [15], the induced 1,3-dipolar cycloaddition between the alkynyl and azide moieties (Click Reaction) afforded compound 6 in variable medium yields of approximately 60%. The yield depended on the reaction conditions under which the cycloaddition was carried out. Several byproducts were formed during this cycloaddition reaction which, however, could not be separated and characterized. The amount of these byproducts increased at higher reaction temperatures or upon irradiation with microwave. It was anticipated that the byproducts which lowered the yield of compound 6 might be decomposition products of the starting material 4a. Therefore, the more stable benzoylated glucoside 3a' was prepared from 2a, and converted into the azide 4a' in 89% and 84% yield, respectively. Treatment of 4a' with 5 under Cu(I)-catalysis, however, only resulted in a complex mixture of reaction products from which no uniform product could be isolated. Therefore, it was concluded that 4a and 4a' may have reacted with themselves resulting in products of oligomerization. Indeed, when 4a was treated with a catalytic amount of (EtO)3PCuI , TLC (ethyl acetate/n-hexane 1:1) revealed the formation of one faster moving product along with a complex mixture of slower moving products with mobility similar to those previously observed. Careful inspection of the products revealed that dimerisation of 4a occurred, affording the dimeric glycoside 7a beside products of oligomerization (Scheme 1). The reaction proceeded significantly slower than the coupling of 4a and 5. A faster reaction occurred upon irradiation with microwave, which also gave a higher yield (54%) of 7a. Benzoylated glycoside 4a' did not give any product of dimerization though. Only oligomers 8 were observed in this case (for details see Supporting Information File 1).Scheme 1Synthesis and reaction of compounds 4a and 4a'.At first, it was unclear whether 7a was formed by an intramolecular cyclization or a dimerization of 4a since its concentration-dependent ESI-MS and MALDI-TOF-MS spectra both showed peaks corresponding to the molecular mass of 4a and 7a, respectively. However, the dimeric structure of compound 7a was finally unambiguously assigned by NMR spectroscopy and field desorption (FD) mass spectrometry. The NMR spectra of 7a showed no conformative anomalies of the pyranose ring what would have been expected if 7a would have been the product of intramolecular 1,3-dipolar cycloaddition of the azido group and the 2-propynyl aglycon in the starting material 4a.The oligomerization of glycosides containing both, an azido and an alkynyl group upon copper-catalyzed Click-Reaction had been observed previously in two instances. Gin and coworker recently found that 2,3,6-tri-O-benzyl-4-O-(2-propynyl)-α-D-mannopyranosyl azide affords a cyclic trimer upon 1,3-dipolar cycloaddition of its azido moiety to its propynyl moiety while the corresponding α-1,4-linked manno-disaccharide afforded a cyclic dimer similar to compound 7a [16]. Jarosz et al. also recently reported about the copper catalyzed reaction of 6-azido-1',2,3,3',4,4'-hexa-O-benzyl-6-deoxy-6'-propargyl-sucrose to afford either a product of intramolecular cyclization or a dimeric product, depending on the reaction conditions [17]. Likewise, Vasella reported the thermal intramolecular 1,3-dipolar cycloaddition of protected 2-azidoethyl 45-O-(2-propynyl)-malto-hexaoside, giving the corresponding isomeric macrocyclic derivatives [18]. In the light of Gin's and Jarosz's results and our own unexpected finding that 4a can form cyclic dimers upon copper-catalyzed Click-Reaction, we investigated several other 2-propynyl and 3-butynyl 6-azido-6-deoxy-glycosides 4 in order to probe their ability to form similar cyclic dimers 7.First, an alternative route to 2-propynyl 6-azido-6-deoxy-glucoside 4a was attempted (Scheme 2). Compound 1a was deacetylated and treated with N-bromosuccinimide and triphenylphosphine in DMF according to Hanessian's procedure [19] followed by reacetylation of the OH-groups with acetic anhydride in pyridine to afford 2-propynyl 6-bromo-6-deoxy-2,3,4-tri-O-acetyl-β-D-glucopyranoside (3a'') in 60% yield. Next, the latter was stirred with NaN3 in DMF (48 h, 65 °C) to afford 4a in 44% yield. The preparation of compound 4a\nvia the corresponding tosylate 3a was somewhat more convenient than the synthesis via the 6-bromo-6-deoxy counterpart 3a'' and resulted in a similar overall yield. Therefore, all other 6-azido-6-deoxy-glycosides 4 were prepared via the corresponding tosylates 3 as described above. Scheme 2 summarizes the yields for the preparation of the tosylates 3 and 6-azido-6-deoxy-glycosides 4. Starting materials 1 were prepared following known procedures for 1a [13,20], 1b [21], 1d [13], 1f [20,22] and 1g [21]. 2-Propynyl 2,3,4,6-tetra-O-acetyl-α-D-glycopyranosides 1c and 1e have not been described previously. They were prepared from D-glucose and D-galactose in 20% and 22% yield, respectively via classical Fischer-Glycosylation in 2-propynol as the solvent under acidic conditions followed by acetylation of the intermediate glycosides and chromatographic separation of the anomeric acetates.Scheme 2Preparation of compounds 4a–g.Next, glycosides 4a–g were submitted to dimerization by 1,3-dipolar cycloaddition reaction. As the catalyst, 10 mol% (EtO)3PCuI was applied and used along with three equivalents diisopropyl ethylamine in toluene [15]. Microwave irradiation [23] reduced the reaction time significantly but also resulted in decomposition of the starting material in some cases. Table 1 summarizes the results for the dimerization of 4a–g to 7a–g.Table 1Dimerization of Glycosides 4a–g under Cu-Catalysis.EntryGlycoside 4Product 7ConditionsYield14a\n7a12 h rt1 h 80 °C, 20 W MW54%20%24b\n7b12 h rt1 h 80 °C, 20 W MW-32%34c\n7c12 h rt1 h 80 °C, 20 W MW14%-44d\n7d12 h rt1 h 80 °C, 20 W MW28%-54e\n7e12 h rt1 h 80 °C, 20 W MW--64f\n7f12 h rt1 h 80 °C, 20 W MW-30%74g\n7g12 h rt1 h 80 °C, 20 W MW-53%Yields of the copper-catalyzed dimerizations were low to medium (14–54%) and depended on the sugar moiety, the anomeric configuration and the ring size which was formed during the Click-Reaction. In general, no other cyclization product could be isolated from the reaction mixtures although significant amounts of byproducts were formed. These byproducts were slower moving compounds on TLC (ethyl acetate/n-hexane 1:1) and appeared to be products of oligomerization of the starting material 4. In the case of benzoylated glycoside 4a' where no cyclic dimer could be isolated from the complex reaction mixture, FAB MS of the purified mixture indeed revealed the presence of linear dimers, trimers and tetramers.α-Galactoside 4e did not give any isolable dimer 7e at all (cf.\nTable 1, entry 5). Similarly, α-glucoside 4c resulted in a lower yield of the corresponding dimer compared to β-glucoside 4a (cf.\nTable 1, entries 1 and 3). This may be attributed to a significant ring-strain in the α-linked dimers. For example, the 1H NMR of compound 7c showed an unusually small coupling constant between H-1 and H-2 (<1.0 Hz) and H-2 and H-3 (3.1 Hz) which is indicative that the sugar moieties in 7c are no longer in a chair conformation (see Table 1 in the Supporting Information File 1). No such effects were observed in the manno series though (cf.\nTable 1, entries 6 and 7). Here, the corresponding dimers 7f and 7g showed regular coupling constants in their NMR spectra.The effect of microwave irradiation on the outcome of the dimerization is somewhat confusing. In general, microwave irradiation resulted in a faster reaction, i.e. faster disappearance of the starting material (cf.\nTable 1, entry 1). Similar accelerations of Click-Reactions upon microwave irradiation had been observed previously as well [15]. However, the higher temperature associated with the microwave irradiation also resulted in a more pronounced decomposition of the starting material, and thus resulted in a lower yield of the dimers (cf.\nTable 1, entries 1, 3 and 4) while heating of the reaction mixture alone resulted in complex product mixtures from which no dimerization products could be isolated. In the case of compounds 4b and 4e–g, no reaction occurred at room temperature (cf.\nTable 1, entries 2 and 5–7).ConclusionWe describe for the first time the copper-catalyzed dimerization of simple acetylated 2-propynyl and 3-butynyl 6-azido-6-deoxy-glycosides in the gluco, galacto and manno series leading to macrocyclic rings containing two sugar moieties and two 1,2,3-triazole moieties. For instance, such compounds may function as novel ligands for the preparation of metal complexes [24]. Further examples for cyclizations of other azido-alkynyl-glycosides are under investigation.Supporting InformationFile 1Experimental Data\n\nREFERENCES:\n1. ZieglerTRöselingDSubramanianL RTetrahedron: Asymmetry20021391191410.1016/S0957-4166(02)00212-4\n2. SchipsCZieglerTJ Carbohydr Chem20052477378810.1080/07328300500326859\n3. ZieglerTSchipsCUmsetzung von Aminoalkoholen mit sauren, organischen Substraten nach Art einer Mitsunobu-ReaktionGerman Patent102004046010B32005128Chem. Abstr.2005, 144, 1282957.\n4. ZieglerTSchipsCNat Protoc200611987199410.1038/nprot.2006.30717487187\n5. HuisgenRKnorrRMöbiusLSzeimiesGChem Ber1965984014402110.1002/cber.19650981228\n6. RostovtsevV VGreenL GFokinV VSharplessK BAngew Chem, Int Ed2002412596259910.1002/1521-3773(20020715)41:14<2596::AID-ANIE2596>3.0.CO;2-4\n7. TornøeC WChristensenCMeldalMJ Org Chem2002673057306410.1021/jo011148j11975567\n8. BockV DHiemstraHvan MaarseveenJ HEur J Org Chem2006516810.1002/ejoc.200500483\n9. BinderW HKlugerCCurr Org Chem2006101791181510.2174/138527206778249838\n10. DedolaSNepogodievS AFieldR AOrg Biomol Chem200751006101710.1039/b618048p17377651\n11. AngellY LBurgessKChem Soc Rev2007361674168910.1039/b701444a17721589\n12. PietrzikNSchipsCZieglerTSynthesis200851952610.1055/s-2008-1032150\n13. MereyalaH BGurralaS RCarbohydr Res199830735135410.1016/S0008-6215(97)10104-5\n14. CramerFOtterbachHSpringmannHChem Ber19599238439110.1002/cber.19590920221\n15. Pérez-BalderasFOrtega-MuñozMMorales-SanfrutosJHernández-MateoFCalvo-FloresF GCalvo-AsínJ AIsac-GarcíaJSantoyo-GonzálezFOrg Lett200351951195410.1021/ol034534r12762694\n16. BodineK DGinD YGinM SOrg Lett200574479448210.1021/ol051818y16178563\n17. JaroszSLewandowskiBListkowskiASynthesis200891391610.1055/s-2008-1032198\n18. HoffmannBBernetBVasellaAHelv Chim Acta20028526528710.1002/1522-2675(200201)85:1<265::AID-HLCA265>3.0.CO;2-1\n19. HanessianSPonpipomM MLavalleePCarbohydr Res197224455610.1016/S0008-6215(00)82258-2\n20. KaufmanR JSidhuR SJ Org Chem1982474941494710.1021/jo00146a023\n21. TietzeL FBotheUChem–Eur J199841179118310.1002/(SICI)1521-3765(19980710)4:7<1179::AID-CHEM1179>3.0.CO;2-F\n22. Fernandez-MegiaECorreaJRodríguez-MeizosoIRigueraRMacromolecules2006392113212010.1021/ma052448w\n23. SavinK ARobertsonMGernertDGreenSHembreE JBishopJMol Diversity2003717117410.1023/B:MODI.0000006801.27748.3b\n24. ZieglerTHermannCTetrahedron Lett2008492166216910.1016/j.tetlet.2008.01.081"
+}

batch_5/PMC2533524.json

@@ -0,0 +1,4 @@
+{
+    "id": "PMC2533524",
+    "text": "This is an academic paper. This paper has corpus identifier PMC2533524\nAUTHORS: Francesco Sofi, Francesca Cesari, Rosanna Abbate, Gian Franco Gensini, Alessandro Casini\n\nABSTRACT:\nObjective To systematically review all the prospective cohort studies that have analysed the relation between adherence to a Mediterranean diet, mortality, and incidence of chronic diseases in a primary prevention setting. Design Meta-analysis of prospective cohort studies.Data sources English and non-English publications in PubMed, Embase, Web of Science, and the Cochrane Central Register of Controlled Trials from 1966 to 30 June 2008. Studies reviewed Studies that analysed prospectively the association between adherence to a Mediterranean diet, mortality, and incidence of diseases; 12 studies, with a total of 1 574 299 subjects followed for a time ranging from three to 18 years were included.Results The cumulative analysis among eight cohorts (514 816 subjects and 33 576 deaths) evaluating overall mortality in relation to adherence to a Mediterranean diet showed that a two point increase in the adherence score was significantly associated with a reduced risk of mortality (pooled relative risk 0.91, 95% confidence interval 0.89 to 0.94). Likewise, the analyses showed a beneficial role for greater adherence to a Mediterranean diet on cardiovascular mortality (pooled relative risk 0.91, 0.87 to 0.95), incidence of or mortality from cancer (0.94, 0.92 to 0.96), and incidence of Parkinson’s disease and Alzheimer’s disease (0.87, 0.80 to 0.96).Conclusions Greater adherence to a Mediterranean diet is associated with a significant improvement in health status, as seen by a significant reduction in overall mortality (9%), mortality from cardiovascular diseases (9%), incidence of or mortality from cancer (6%), and incidence of Parkinson’s disease and Alzheimer’s disease (13%). These results seem to be clinically relevant for public health, in particular for encouraging a Mediterranean-like dietary pattern for primary prevention of major chronic diseases.\n\nBODY:\nIntroductionThe Mediterranean diet, representing the dietary pattern usually consumed among the populations bordering the Mediterranean sea, has been widely reported to be a model of healthy eating for its contribution to a favourable health status and a better quality of life.1\n2 Since the first data from the seven countries study,3 several studies in different populations have established a beneficial role for the main components of the Mediterranean diet on the occurrence of cardiovascular diseases and chronic degenerative diseases.2\n4 However, research interest in this field over the past years has been focused on estimating adherence to the whole Mediterranean diet rather than analysing the individual components of the dietary pattern in relation to the health status of the population.5 This because the analyses of single nutrients ignore important interactions between components of a diet and, more importantly, because people do not eat isolated nutrients. Hence, dietary scores estimating adherence to a Mediterranean diet, devised a priori on the basis of the characteristic components of the traditional diet of the Mediterranean area, have been found to be associated with a reduction of overall mortality and mortality from cardiovascular diseases and cancer.6 The aim of this study was to do a systematic review with meta-analysis of all the available prospective cohort studies that have assessed the association between adherence to a Mediterranean diet and adverse outcomes, in order to establish the role of adherence to a Mediterranean diet in primary prevention.MethodsData sourcesWe focused on prospective studies investigating the association between adherence to a Mediterranean diet and health outcomes. We searched PubMed, Embase, Web of Science, and the Cochrane Central Register of Controlled Trials databases up to 30 June 2008, using a search strategy that included both truncated free text and exploded MeSH terms. MeSH headings included “Mediterranean”, “diet”, “dietary pattern”, “disease”, “health”, “cardiovascular disease”, “cerebrovascular disease”, “coronary heart disease”, “degenerative diseases”, “cancer”, “neoplasm”, “prospective”, “follow-up”, or “cohort”, and their variants. The search strategy had no language restrictions. We also consulted references from the extracted articles and reviews to complete the data bank. When multiple articles for a single study were present, we used the latest publication and supplemented it, if necessary, with data from the most complete or updated publication. We assessed the relevance of studies by using a hierarchical approach based on title, abstract, and the full manuscript.Study selectionWe identified studies that prospectively evaluated the association of an a priori score used for assessing adherence to a Mediterranean diet and adverse clinical outcomes. We excluded the studies if they had a cross sectional or case-control design, if they analysed adherence to a non-specific dietary pattern or to a recommended dietary guideline and not to a Mediterranean diet, if they evaluated a cohort of patients with a previous clinical event (that is, secondary prevention), if they did not adjust for potential confounders, and if they did not report an adequate statistical analysis.Figure 1 shows the process of study selection. Our initial search yielded 62 reports, of which we excluded 20 on the basis of the title or abstract. Of the remaining 42 articles, we excluded 26 for the following reasons: a non-specific dietary pattern, instead of a Mediterranean diet, was evaluated (n=3); cross sectional or case-control design was used (n=18); and the study population was in secondary prevention (n=5). We excluded four additional articles because they represented duplicate studies, so we included only the latest or the more complete paper in the final analysis. Finally, 12 articles fulfilled our inclusion criteria.w1-w12Fig 1 Process of study selectionData extractionWe extracted the following baseline characteristics from the original reports by using a standardised data extraction form and included them in the meta-analysis: lead author, year of publication, cohort name, country of origin of the cohort, sample size of the cohort and number of outcomes, duration of follow-up, age at entry, sex, outcome, components of the score for adherence to a Mediterranean diet, and variables that entered into the multivariable model as potential confounders (table 1). Two investigators (FS and FC) collected the data, and disagreements were solved by consensus and by the opinion of a third author (AC), if necessary. Outcomes of interest were overall mortality, mortality from cardiovascular diseases, incidence of or mortality from cancer, as well as occurrence of Parkinson’s disease and Alzheimer’s disease.Table 1 Study characteristicsAuthor, yearCountryNo of outcomes/No in cohortOutcomeFollow-up (years)Age at entry (years)SexComponents of scoreAdjustmentTrichopoulou et al, 1995w1Greece53/182Overall mortality4-5>70 (mean 75.4)M/F1. High legumes; 2. High cereals; 3. High fruits; 4. High vegetables; 5. High MUFA:SFA ratio; 6. Moderate alcohol; 7. Low meat and meat products; 8. Low milk and dairy productsAge, sex, smoking habit, total diet scoreKouris-Blazos et al, 1999w2Australia36/330Overall mortality4-6≥70M/F1. High legumes; 2. High cereals3. High fruits; 4. High vegetables; 5. High MUFA:SFA ratio; 6. Moderate alcohol; 7. Low meat and meat products; 8. Low milk and dairy productsAge, sex, smoking habit, ethnic originLasheras et al, 2000w3Spain96/161Overall mortality9.565-80M/F1. High legumes; 2. High cereals; 3. High fruits; 4. High vegetables; 5. High MUFA:SFA ratio; 6. Moderate alcohol; 7. Low meat and meat products; 8. Low milk and dairy productsAge, sex, total diet score, albumin, dieting in response to chronic conditions, BMI, self assessment of health, physical activityTrichopoulou et al, 2003w4 (EPIC)Greece275/22 043; 54/22 04Overall mortality; CHD mortality3.720-86M/F1. High legumes; 2. High cereals; 3. High fruits/nuts; 4. High vegetables; 5. High fish; 6. High MUFA:SFA ratio; 7. Moderate alcohol; 8. Low meat and poultry; 9. Low dairy productsAge, sex, smoking habit, years of education, BMI, waist to hip ratio, energy expenditure score, energy intake, consumption of potatoes and eggsKnoops et al, 2004w5 (HALE project: SENECA and FINE)Belgium, Denmark, Finland, France, Greece, Hungary, Italy, Netherlands, Portugal, Spain, Switzerland935/2339; 122/2152; 371/2152; 233/2152Overall mortality; CHD mortality; CVD mortality; cancer mortality1070-90M/F1. High legumes/nuts/seeds; 2. High cereals; 3. High fruits; 4. High vegetables/potatoes; 5. High fish; 6. High MUFA:SFA ratio; 7. Low meat and meat products; 8. Low dairy productsAge, sex, smoking habit, physical activity, BMI, dietary habits, alcohol, years of education, study populationTrichopoulou et al, 2005w6 (EPIC-elderly)Denmark, France, Germany, Greece, Italy, Netherlands, Norway, Spain, Sweden, UK3810/67 228 (after exclusion of Greek patients)Overall mortality7.4>60M/F1. High legumes; 2. High cereals; 3. High fruits/nuts; 4. High vegetables; 5. High fish; 6. High MUFA+PUFA:SFA ratio; 7. Moderate alcohol; 8. Low meat and poultry; 9. Low dairy productsAge; sex; smoking habit; diabetes; education; BMI; waist to hip ratio; physical activity; energy; consumption of potatoes, eggs, sugar, and confectioneryLagiou et al, 2006w7Sweden572/42 237; 280/42 237Overall mortality; cancer mortality1230-49F1. High legumes; 2. High cereals; 3. High fruits/nuts; 4. High vegetables; 5. High fish; 6. High MUFA:SFA ratio; 7. Moderate alcohol; 8. Low meat and meat products; 9. Low dairy productsAge; height; BMI; smoking habit; physical activity; education; energy; consumption of potatoes, eggs, PUFA, sweets, and non-alcoholic beveragesFung et al, 2006w8 (nurses’ health study)USA3580/71 058Breast cancer1830-55F1. High legumes; 2. High wholegrain products; 3. High fruits; 4. High nuts; 5. High vegetables; 6. High fish; 7. High MUFA:SFA ratio; 8. Moderate alcohol; 9. Low red and processed meatsAge, smoking habit, BMI, multivitamins, energy, physical activity, family history of breast cancer and of benign breast disease, menopause, HRT, weight change since age 18 yearsScarmeas et al, 2006w9USA85/2258Alzheimer’s disease4Mean 77.2M/F1. High legumes; 2. High cereals; 3. High fruits; 4. High vegetables; 5. High fish; 6. High MUFA:SFA ratio; 7. Moderate alcohol; 8. Low meat; 9. Low dairy productsAge, sex, cohort, ethnicity, smoking habit, BMI, education, energy, ApoE genotype, comorbidity indexGao et al, 2007w10 (health professionals and nurses’ health studies)—menUSA508/49 692Parkinson’s disease1640-75M1. High legumes; 2. High wholegrain products; 3. High fruits; 4. High nuts; 5. High vegetables; 6. High fish; 7. High MUFA:SFA ratio; 8. Moderate alcohol; 9. Low red and processed meatAge, smoking habit, BMI, use of NSAIDs, energy, caffeine intakeGao et al, 2007w10 (health professionals and nurses’ health studies)—womenUSA190/81 676Parkinson’s disease1640-75F1. High legumes; 2. High wholegrain products; 3. High fruits; 4. High nuts; 5. High vegetables; 6. High fish; 7. High MUFA:SFA ratio; 8. Moderate alcohol; 9. Low red and processed meatAge, smoking habit, BMI, use of NSAIDs, energy, caffeine intakeMitrou et al, 2007w11 (National Institutes of Health diet and health study)—menUSA18 126/214 284; 2425/214 284; 3717/214 284Overall mortality; CVD deaths; cancer deaths1050-71M1. High legumes; 2. High wholegrain products; 3. High fruits; 4. High nuts; 5. High vegetables; 6. High fish; 7. High MUFA:SFA ratio; 8. Moderate alcohol; 9. Low red and processed meatAge, race, smoking habit, energy, BMI, education, physical activityMitrou et al, 2007w11 (National Institutes of Health diet and health study)—womenUSA9673/166 012; 1026/166 012; 2268/166 012Overall mortality; CVD deaths; cancer deaths1050-71F1. High legumes; 2. High wholegrain products; 3. High fruits; 4. High nuts; 5. High vegetables; 6. High fish; 7. High MUFA:SFA ratio; 8. Moderate alcohol; 9. Low red and processed meatAge, race, smoking habit, energy, BMI, education, physical activity, HRTBenetou et al, 2008w12 (EPIC)Greece851/25 623Incident cancers (excluding non-melanoma skin cancers)7.920-86M/F1. High legumes; 2. High cereals; 3. High fruits/nuts; 4. High vegetables; 5. High fish; 6. High MUFA:SFA ratio; 7. Moderate alcohol; 8. Low meat and poultry; 9. Low dairy productsAge; sex; smoking habit; years of education; BMI; height; physical activity; total energy; consumption of potatoes, eggs, confectionery, and non-alcoholic beveragesApoE=apolipoprotein E; BMI=body mass index; CHD=coronary heart disease; CVD=cardiovascular disease; HRT=hormone replacement therapy; MUFA=mono-unsaturated fatty acids; NSAIDs=non-steroidal anti-inflammatory drugs; PUFA=polyunsaturated fatty acids; SFA=saturated fatty acids.We assessed the quality of the studies according to the number of participants, the duration of follow-up, and adjustment for potential confounders. We considered studies with a high number of participants; long duration of follow-up; and adjustment for confounders including demographic, anthropometric, and traditional risk factors to be of high quality.Definition of adherence to Mediterranean dietAdherence to a Mediterranean diet was defined through scores that estimated the conformity of the dietary pattern of the studied population with the traditional Mediterranean dietary pattern. Values of zero or one were assigned to each dietary component by using as cut offs the overall sex specific medians among the study participants. Specifically, people whose consumption of components considered to be part of a Mediterranean diet (vegetables, fruits, legumes, cereals, fish, and a moderate intake of red wine during meals) was above the median consumption of the population were assigned a value of one, whereas a value of zero was given to those with consumptions below the median. By contrast, people whose consumption of components presumed not to form part of a Mediterranean diet (red and processed meats, dairy products) was above the median consumption of the population had a value of zero assigned, and the others had a value of one. However, some differences among the studies existed, especially in relation to the food category of vegetables (grouped with potatoes in one studyw5), meat and meat products (grouped with poultry in some studiesw4 w6), and nuts and seeds (grouped with fruits in some studies,w4 w6 w7 w12 with legumes in one study,w5 and considered a group by themselves in some othersw8 w10 w11), as well as milk and dairy products (not present in some studiesw8 w10 w11) and fish (present only in more recent studiesw4-w12). Thus, the total adherence scores (estimated as the sum of the above indicated scores of zero and one) varied from a minimum of 0 points indicating low adherence to a maximum of 7-9 points reflecting high adherence to a Mediterranean diet.Statistical analysisWe used RevMan, version 4.2 for Windows by the Cochrane Collaboration to analyse data. We used the results of the original studies from multivariable models with the most complete adjustment for potential confounders; table 1 shows the confounding variables included in this analysis. We used a random effects model that accounts for interstudy variation and provides a more conservative effect than a fixed model. We calculated random summary relative risks with 95% confidence intervals by using an inverse variance method. We grouped the studies according to the different clinical outcomes (mortality from all causes, mortality from cardiovascular diseases, incidence of or mortality from cancer, and incidence of Parkinson’s disease and Alzheimer’s disease). We assessed the potential sources of heterogeneity by using the standard χ2 test. In addition, we used the I2 statistic to investigate heterogeneity by examining the extent of inconsistency across the study results. To examine the potential source of heterogeneity across studies evaluating overall mortality, we did sensitivity analyses according to some characteristics of the studies—sex (male, female), country of origin (European countries, United States, other countries), follow-up time (below or above the median follow-up time of the studies: 8 years), and the quality of the studies (low, high). To assess the presence of publication bias, we computed the “failsafe N” for each of the main outcomes; this value is an estimate of the number of studies with null results that would need to be added to the meta-analysis to reduce the overall observed significant result to non-significance.ResultsCharacteristics of study cohortsSample sizes varied between 161 and 214 284, with a follow-up time ranging from 3.7 to 18 years. Outcomes of interest were overall mortality, cardiovascular mortality, incidence of or mortality from neoplastic disease, and incidence of Parkinson’s disease and Alzheimer’s disease. Only six out of 12 studies were done in Mediterranean populations.w1 w3-w6 w12 The remaining cohorts comprised US populations,w8-w11 northern Europeans,w5-w7 and a cohort of Europeans living in Australia.w3 The total number of subjects in the included studies was 1 574 299.Main outcomesAccording to the different clinical outcomes, overall mortality was evaluated in eight cohorts (nine studies) for a total of 514 816 subjects and 33 576 deaths, cardiovascular mortality in three cohorts (four studies) including a total of 404 491 subjects and 3876 fatal events, cancer incidence/mortality in five cohorts (six studies) comprising 521 366 subjects and 10 929 events, and incidence of Parkinson’s disease and Alzheimer’s disease in two cohorts (three studies) for a total of 133 626 subjects and 783 cases.Figure 2 shows the cumulative analysis for studies that analysed overall mortality as the primary clinical outcome. Using a random effects model, we found that a two point increase in score for adherence to a Mediterranean diet was significantly associated with a reduced risk of mortality from any cause (relative risk 0.91, 95% confidence interval 0.89 to 0.94; P<0.0001). Significant heterogeneity was present among the studies (I2=48.8%; P=0.05). However, after exclusion of the paper by Trichopoulou et al 2003 that analysed the same cohort as Trichopoulou et al 2005,w4 w6 the significant association with overall mortality remained (relative risk 0.92, 0.91 to 0.94; P<0.0001), showing no significant heterogeneity (I2=18.3%; P=0.3).Fig 2 Risk of all cause mortality associated with two point increase in adherence score for Mediterranean diet. Squares represent effect size; extended lines show 95% confidence intervals; diamond represents total effect sizeSimilarly figure 3 shows that a greater adherence to a Mediterranean diet significantly reduced the risk of mortality from cardiovascular diseases (relative risk 0.91, 0.87 to 0.95; P<0.0001) with non-significant heterogeneity (I2=32.6%; P=0.2). Furthermore, greater adherence to a Mediterranean diet significantly reduced the occurrence of and mortality from neoplasm (relative risk 0.94, 0.92 to 0.96; P<0.0001) (I2=0%; P=0.5) (fig 4). Finally, the overall analysis showed a significant reduction in incidence of Parkinson’s disease and Alzheimer’s disease associated with a higher score of adherence to a Mediterranean diet (relative risk 0.87, 0.80 to 0.96; P=0.004), with no heterogeneity among the studies (I2=0%; P=0.5) (fig 5).Fig 3 Risk of mortality from cardiovascular diseases associated with two point increase in adherence score for Mediterranean diet. Squares represent effect size; extended lines show 95% confidence intervals; diamond represents total effect sizeFig 4 Risk of occurrence of or mortality from cancer associated with two point increase in adherence score for Mediterranean diet. Squares represent effect size; extended lines show 95% confidence intervals; diamond represents total effect sizeFig 5 Risk of Parkinson’s disease and Alzheimer’s disease associated with two point increase in adherence score for Mediterranean diet. Squares represent effect size; extended lines show 95% confidence intervals; diamond represents total effect sizeSensitivity analysesStudies included in this meta-analysis varied in some characteristics. Because such heterogeneity of studies is likely to produce heterogeneity of effect sizes across studies, we did some sensitivity analyses. Table 2 shows the different subgroup analyses on studies evaluating overall mortality as clinical outcome. These analyses showed no significant influence of any variable (country of origin of the study, sex, follow-up time, quality of the studies) on the overall results of the meta-analysis.Table 2 Sensitivity analysisVariable (No of studies)Relative risk (95% CI)Sex: Male (7)0.89 (0.85 to 0.94) Female (8)0.90 (0.86 to 0.94)Country of origin: Europe (6)0.87 (0.81 to 0.94) United States/Australia (3)0.93 (0.91 to 0.94)Follow-up time: <8 years (4)0.82 (0.69 to 0.97) >8 years (5)0.92 (0.91 to 0.94)Study quality: Low (3)0.89 (0.83 to 0.99) High (6)0.92 (0.90 to 0.94)Publication biasTo assess the presence of publication bias, we computed the failsafe N for each of the main outcomes. Each failsafe N (580 for studies evaluating overall mortality as main outcome, 68 for studies with cardiovascular mortality as main outcome, 72 for incidence of or mortality from cancer, and 43 for incidence of degenerative diseases) far exceeded Rosenthal’s recommendation (failsafe N>5k+10, where k is the number of studies included in the analysis) for a robust effect of the overall analysis.DiscussionThis meta-analysis shows, in an overall analysis comprising more than 1.5 million healthy subjects and 40 000 fatal and non-fatal events, that greater adherence to a Mediterranean diet is significantly associated with a reduced risk of overall mortality, cardiovascular mortality, cancer incidence and mortality, and incidence of Parkinson’s disease and Alzheimer’s disease. The cumulative analysis of 12 cohort studies shows that a two point increase in the score for adherence to a Mediterranean diet determines a 9% reduction in overall mortality, a 9% reduction in mortality from cardiovascular diseases, a 6% reduction in incidence of or mortality from neoplasm, and a 13% reduction in incidence of Parkinson’s disease and Alzheimer’s disease. To the best of our knowledge, this is the first report that has systematically assessed, through meta-analysis, the possible association between adherence to a Mediterranean diet, mortality, and the occurrence of chronic diseases in the general population.Diet and diseaseThe effect of diet on human health has been amply reported in many epidemiological, population based, and randomised clinical trials, providing evidence that a dietary pattern rich in some beneficial food groups such as fruit, vegetables, whole grains, and fish can reduce the incidence of cardiovascular and neoplastic diseases.7 However, until now, the vast majority of studies followed the approach of assessing single nutrients or food groups in relation to the occurrence of disease.4\n8\n9 This approach seems to have several conceptual and methodological limitations, because food components of diet present synergistic and antagonist interactions and because people eat a complex of nutrients.5 Therefore, over the past few years, researchers have shifted their attention from the evaluation of single nutrients to the analysis of dietary pattern as a whole.6\nw1-w10 As a result, an increasing number of studies have been done by summing foods considered to be important for health to provide an overall measure of dietary quality—that is, a quality diet score.6In this context, a prominent position has been occupied by studies evaluating adherence to a Mediterranean diet, because of its well known and evidence based beneficial effects on human health. Indeed, since the early 1970s many investigators have reported the beneficial role of the Mediterranean diet, as originally reported by Keys in the pioneering seven countries study.3 A diet rich in fruits, vegetables, legumes, and cereals, with olive oil as the only source of fat, moderate consumption of red wine especially during meals, and low consumption of red meat has been shown to be beneficial for all cause and cardiovascular mortality, lipid metabolism, blood pressure, and several different disease states such as endothelial dysfunction and overweight.7Practical implicationsIn this study we aimed to systematically analyse all the prospective cohort studies that evaluated the effect of a computational score estimating adherence to a Mediterranean diet on health status. From the overall analysis of 11 cohort studies, of which eight assessed the risk of overall mortality, four assessed cardiovascular mortality, six assessed incidence of or mortality from neoplasm, and three assessed incidence of Parkinson’s disease and Alzheimer’s disease, we report a significant reduction in risk of all the main clinical outcomes with an increasing score for adherence to a Mediterranean diet. This observation seems to show that a score based on a theoretically defined Mediterranean diet is an effective preventive tool for measuring the risk of mortality and morbidity in the general population.A Mediterranean diet has been shown to have a beneficial effect on the occurrence of diseases in industrialised and non-industrialised countries. All the major scientific associations, in fact, strongly encourage people to consume a Mediterranean-like dietary pattern to reduce their risk of disease.10\n11\n12 Unfortunately, despite this worldwide promotion of the Mediterranean diet, a progressive shift to a non-Mediterranean dietary pattern, even in countries bordering the Mediterranean sea, has progressively developed.13 It thus seems urgent to identify an effective preventive strategy to decrease the risk burden related to dietary habits in the general population; the use of such a tool could be important in increasing the implementation of dietary guidelines.LimitationsSome limitations of this study can be identified. The Mediterranean diet is not a homogeneous pattern of eating, and heterogeneity on the score items exists. How to group some food categories such as legumes, nuts, and milk and dairy products; the real importance of different types of meat; and the establishment of the moderate amount of alcohol intake are still matters of dispute among researchers and can differ among the selected studies. None the less, the key characteristics of a Mediterranean diet were present in all the studies, and the overall analysis seemed not to be significantly influenced by these differences. In addition, the use of a score for estimating a dietary pattern is limited by subjectivity, conditioned by the available data and the main objectives of the study, and so possibly determining a great variability in the interpretation of the results.Finally, a further limitation exists in the different adjustment for potential confounders seen among the included studies. This difference could have determined a residual confounding within the studies, especially for the non-Mediterranean cohorts. However, the sensitivity analysis according to the quality of the studies, which also included the presence or not of adjustment factors, showed no significant influence of residual confounding on the overall findings of our meta-analysis.ConclusionsThis meta-analysis shows that adherence to a Mediterranean diet can significantly decrease the risk of overall mortality, mortality from cardiovascular diseases, incidence of or mortality from cancer, and incidence of Parkinson’s disease and Alzheimer’s disease. These results seem to be clinically relevant in terms of public health, particularly for reducing the risk of premature death in the general population, and are strictly concordant with current guidelines and recommendations from all the major scientific associations that strongly encourage a Mediterranean-like dietary pattern for primary and secondary prevention of major chronic diseases.What is already known on this topicThe Mediterranean diet is a well known model of diet for primary and secondary prevention of major chronic diseasesAn adherence score can be used to assess the adherence of a specific population to the rules of a traditional Mediterranean dietWhat this study addsGreater adherence to a Mediterranean diet confers a significant protection for overall mortality, as well as cardiovascular disease mortality and incidence of cancer and degenerative diseasesThe adherence score based on a theoretically defined Mediterranean diet could be an effective preventive tool for reducing the risk of mortality and morbidity in the general population\n\nREFERENCES:\n1. Willett WC, Sacks F, Trichopoulou A, Drescher G, Ferro-Luzzi A, Helsing E, et al. Mediterranean diet pyramid: a cultural model for healthy eating. Am J Clin Nutr1995;61:1402-6.\n2. Serra-Majem L, Roman B, Estruch R. Scientific evidence of interventions using the Mediterranean diet: a systematic review. Nutr Rev2006;64:S27-47.16532897\n3. Keys A. Seven countries: a multivariate analysis of death and coronary heart disease. Cambridge, MA: Harvard University Press, 1980.\n4. De Lorgeril M, Salen P, Martin J, Monjaud I, Delaye J, Mamelle N. Mediterranean diet, traditional risk factors and the rate of cardiovascular complications after myocardial infarction: final report of the Lyon diet heart study. Circulation1999;99:799-85.\n5. Hu FB. Dietary pattern analysis: a new direction in nutritional epidemiology. Curr Opin Lipidol2002;13:3-9.11790957\n6. Bach A, Serra-Majem L, Carrasco JL, Roman B, Ngo J, Bertomeu I, et al. The use of indexes evaluating the adherence to the Mediterranean diet in epidemiological studies: a review. Public Health Nutr2006;9:132-46.16512961\n7. Willett WC. The Mediterranean diet: science and practice. Public Health Nutr2006;9:105-10.16512956\n8. Kromhout D, Bosschieter EB, de Lezenne Coulander C. The inverse relation between fish consumption and 20 year mortality from coronary heart disease. N Engl J Med 1985;312:1205-9.3990713\n9. Genkinger JM, Koushik A. Meat consumption and cancer risk. PLoS Med 2007;4:e345.18076281\n10. WHO Study Group. Diet, nutrition, and the prevention of chronic diseases. Geneva: WHO, 2003:916.\n11. American Heart Association Nutrition Committee, Lichtenstein AH, Appel LJ, Brands M, Carnethon M, Daniels S, et al. Diet and lifestyle recommendations revision 2006: a scientific statement from the American Heart Association Nutrition Committee. Circulation2006;114:82-96.16785338\n12. US Department of Health and Human Services, US Department of Agriculture. Dietary guidelines for Americans 2005. 6th ed. Washington, DC: US Government Printing Office, 2005 (available at www.healthierus.gov/dietaryguidelines).\n13. Sofi F, Vecchio S, Giuliani G, Martinelli F, Marcucci R, Gori AM, et al. Dietary habits, lifestyle and cardiovascular risk factors in a clinically healthy Italian population: the “Florence” diet is not Mediterranean. Eur J Clin Nutr2005;59:584-91.15741987"
+}

batch_5/PMC2533645.json

@@ -0,0 +1,4 @@
+{
+    "id": "PMC2533645",
+    "text": "This is an academic paper. This paper has corpus identifier PMC2533645\nAUTHORS: Hong Ju, Alice R Rumbold, Kristyn J Willson, Caroline A Crowther\n\nABSTRACT:\nBackgroundThe impact of borderline gestational diabetes mellitus (BGDM), defined as a positive oral glucose challenge test (OGCT) and normal oral glucose tolerance test (OGTT), on maternal and infant health is unclear. We assessed maternal and infant health outcomes in women with BGDM and compared these to women who had a normal OGCT screen for gestational diabetes.MethodsWe compared demographic, obstetric and neonatal outcomes between women participating in the Australian Collaborative Trial of Supplements with antioxidants Vitamin C and Vitamin E to pregnant women for the prevention of pre-eclampsia (ACTS) who had BGDM and who screened negative on OGCT.ResultsWomen who had BGDM were older (mean difference 1.3 years, [95% confidence interval (CI) 0.3, 2.2], p = 0.01) and more likely to be obese (27.1% vs 14.1%, relative risk (RR) 1.92, [95% CI 1.41, 2.62], p < 0.0001) than women who screened negative on OGCT. The risk of adverse maternal outcome overall was higher (12.9% vs 8.1%, RR 1.59, [95% CI 1.00, 2.52], p = 0.05) in women with BGDM compared with women with a normal OGCT. Women with BGDM were more likely to develop pregnancy induced hypertension (17.9% vs 11.8%, RR 1.51, [95% CI 1.03, 2.20], p = 0.03), have a caesarean for fetal distress (17.1% vs 10.5%, RR 1.63, [95% CI 1.10, 2.41], p = 0.01), and require a longer postnatal hospital stay (mean difference 0.4 day, [95% CI 0.1, 0.7], p = 0.01) than those with a normal glucose tolerance.Infants born to BGDM mothers were more likely to be born preterm (10.7% vs 6.4%, RR 1.68, [95% CI 1.00, 2.80], p = 0.05), have macrosomia (birthweight ≥4.5 kg) (4.3% vs 1.7%, RR 2.53, [95% CI 1.06, 6.03], p = 0.04), be admitted to the neonatal intensive care unit (NICU) (6.5% vs 3.0%, RR 2.18, [95% CI 1.09, 4.36], p = 0.03) or the neonatal nursery (40.3% vs 28.4%, RR 1.42, [95% CI 1.14, 1.76], p = 0.002), and have a longer hospital stay (p = 0.001). More infants in the BGDM group had Sarnat stage 2 or 3 neonatal encephalopathy (12.9% vs 7.8%, RR 1.65, [95% CI 1.04, 2.63], p = 0.03).ConclusionWomen with BGDM and their infants had an increased risk of adverse health outcomes compared with women with a negative OGCT. Intervention strategies to reduce the risks for these women and their infants need evaluation.Trial registrationCurrent Controlled Trials ISRCTN00416244\n\nBODY:\nBackgroundThe prevalence of gestational diabetes mellitus (GDM) is increasing all over the world [1,2]. In Australia the recent prevalence estimates for GDM ranged from 5.2% to 8.8% [3]. The risks for both mothers with GDM and their infants are well-documented. For the infants, these include an increased risk of macrosomia, birth injuries such as shoulder dystocia, bone fracture and nerve palsies, hypoglycaemia, and hyperbilirubinaemia [4-7]. Women with GDM are at increased risk of developing pre-eclampsia and have an increased chance of need for induction of labour and caesarean section. Gestational diabetes is also a strong risk factor for later development of type 2 diabetes [8].Although the risks associated with GDM are well recognised, the impact on maternal and infant health outcomes is less clear for borderline gestational diabetes mellitus (BGDM), which is characterised by values of glucose tolerance intermediate between normal and gestational diabetes. A recent 10 year audit examining the influence of different levels of glucose tolerance on pregnancy complications, [9] revealed a significantly increased risk of pre-eclampsia, caesarean section, neonatal hypoglycaemia and hyperbilirubinaemia for women with BGDM compared with women with normal glucose tolerance. The results are consistent with other literature reports, which identified an increasing risk of adverse maternal and infant outcomes with increasing plasma glucose values [10-12].It is estimated that 6.6% of pregnant women or approximately 16,500 women have BGDM each year in Australia [9]. Given the uncertainty surrounding BGDM, we assessed data from participants in the Australian Collaborative Trial of Supplements with antioxidants Vitamin C and Vitamin E to pregnant women for the prevention of pre-eclampsia (ACTS) [13] to compare the maternal demographic, pregnancy and infant health outcomes of women who had BGDM (screened positive for GDM on oral glucose challenge test (OGCT) but their subsequent oral glucose tolerance test (OGTT) was normal) with women who screened negative on OGCT for GDM.MethodsThe study population included women participating in the ACTS trial [13], a multi-centre randomised placebo controlled trial of antioxidant (vitamins C and E) supplements for the prevention of perinatal complications, who had an OGCT as screening for gestational diabetes. The methods and results of this trial have been reported previously [13]. Briefly, eligible women were: nulliparous, with a singleton pregnancy between 14 and 22 weeks of gestation with a normal blood pressure at the time of recruitment and who gave informed consent. Women with any of the following were ineligible: known multiple pregnancy, known lethal fetal anomaly, known thrombophilia, chronic renal failure, antihypertensive therapy or contraindication to vitamin C or E therapy including haemochromatosis or anticoagulant therapy. Randomisation was performed through a central telephone randomisation service. Women assigned to the vitamin group were provided a daily dose of 1000 mg vitamin C and 400 IU vitamin E until birth, and women in the control group were provided a matching placebo. An OGTT was offered between 24–30 weeks gestation, for those women who screened positive on OGCT test. The study protocol was approved by the research and ethics committees at the nine collaboration hospitals around Australia.We compared demographic, obstetric and neonatal outcomes between women with BGDM and those who screened normal on OGCT. As the ACTS found no significant differences between the antioxidant and placebo groups for the risk of pre-eclampsia, intrauterine growth restriction or other serious outcomes for the infant, the analyses include the combined populations of women who received either antioxidant or placebo supplements.Data collectionPregnancy outcome data including OGCT and OGTT results were collected prospectively from women's medical records. Sociodemographic variables were collected either from women's medical records or self-completed questionnaires at trial entry and included: maternal age, ethnicity, body mass index (BMI), social-economic status as measured by socio-economic index for area (SEIFA) score [14], maternal education, smoking status, blood pressure at trial entry, and family history of pre-eclampsia. Complete outcome data were available for all 1877 women randomised.Outcome variablesBGDM was defined as a positive OGCT (blood glucose ≥7.8 mmol/L 1 hour after a 50 g glucose load) and normal 75 g OGTT (fasting blood glucose <5.5 mmol/L and 2 hour blood glucose <7.8 mmol/L). Pregnancy outcomes assessed included: maternal adverse outcomes (a composite outcome defined as any of the following until six weeks postpartum: death, pulmonary oedema, eclampsia, stroke, thrombocytopenia, renal insufficiency, respiratory arrest, placental abruption, abnormal liver function, preterm prelabour rupture of membranes, major postpartum haemorrhage, postpartum pyrexia, pneumonia, deep-vein thrombosis, or pulmonary embolus requiring anticoagulant therapy) [13]; pregnancy induced hypertension (PIH); pre-eclampsia (defined as systolic blood pressure ≥140 mmHg or diastolic blood pressure [Korokoff V] ≥90 mmHg on at least two occasions four or more hours apart, or both arising after 20 weeks' gestation and one or more of the following: proteinuria, renal insufficiency, liver disease, neurological problems, haematologic disturbances, or fetal growth restriction) [15]; antenatal hospitalisation; preterm prelabour rupture of the membranes; induction of labour; mode of birth; postnatal complications such as postpartum haemorrhage and infection; and length of hospital stay.Neonatal outcomes included a composite outcome of death or infant adverse outcome defined as: stillbirth or death of a liveborn infant before hospital discharge, birthweight <3rd centile for gestational age, severe respiratory distress syndrome, chronic lung disease, intraventricular haemorrhage grade 3 or 4, cystic periventricular leukomalacia, retinopathy of prematurity grade 3 or 4, necrotizing enterocolitis, 5 minute Apgar score <4, seizures before 24 hours of age or requiring 2 or more drugs to control, hypotonia for ≥2 hours, stupor, decreased response to pain or coma, tube feeding for ≥4 days, care in the neonatal intensive care unit (NICU) >4 days, or use of ventilation for ≥24 hours [13]; gestational age at birth; preterm birth (<37 weeks); 5 minute Apgar score <7, infant body size at birth (weight, length and head circumference), small and large-for-gestational age (defined as a birth weight below the 10th percentile or above 90th percentile for gestation according to fetal sex on standardized birth-weight charts, respectively), macrosomia (defined as birthweight ≥4.5 kg), admission to NICU or neonatal nursery, respiratory distress syndrome, mechanical ventilation, antibiotics use after birth, encephalopathy (Sarnat 2 or 3 score) and length of hospital stay.Statistical analysisStatistical analysis was carried out using SAS software, version 9.1. Dichotomous variables were analysed using log-binomial regression and presented as relative risks, with 95% confidence intervals; and continuous variables, if normally distributed, were analysed using Student's t-test and presented as mean differences, with 95% confidence intervals; non-parametric tests were used for skewed data. Analyses were then adjusted for maternal age and BMI given the strong association of these factors with GDM. A p value of 0.05 or less was considered to indicate statistical significance.ResultsOf the 1877 women enrolled in the ACTS trial, 1804 (96%) did not have a fetal loss and underwent screening using a 50 g oral glucose challenge test for gestational diabetes. Of the women screened 1596 (88%) had a normal OGCT screening result, 68 (4%) had an abnormal OGTT and 140 (8%) had BGDM (screened positive on OGCT, normal OGTT).Overall, women with BGDM and women with a normal OGCT had similar characteristics at entry to the study including ethnicity, socio-economic status and educational attainment (Table 1). Compared with women with a normal OGCT, women with BGDM were older (mean difference 1.3 years, [95%CI 0.3, 2.2], p = 0.01), less likely to have a normal BMI (RR 0.82, [95%CI 0.67, 0.99], p = 0.04) and almost twice as likely to be obese (RR 1.92, [95%CI 1.41, 2.62], p < 0.0001) (Table 1). There was no statistically significant difference found between these groups in the number of women who smoked or in their mean systolic or diastolic blood pressure at study entry.Table 1Demographics of women with borderline GDM compared with women with a normal OGCTCharacteristicsBorderline GDM n = 140 (%)Normal OGCT n = 1596 (%)Relative risk [95% CI]p valueAgea (years)27.5 ± 5.426.3 ± 5.81.3 [0.3, 2.2]0.01Race Caucasian129 (92.1)1517 (95.1)0.97 [0.92, 1.02]0.22 Asian5 (3.6)47 (2.9)1.21 [0.49, 3.00]0.68 Other6 (4.3)32 (2.0)2.14 [0.91, 5.02]0.08BMI Underweight (<18.5)3 (2.3)59 (4.0)0.59 [0.19, 1.80]0.34 Normal (18.5 – <25)59 (45.7)818 (55.9)0.82 [0.67, 0.99]0.04 Overweight (25 – <30)32 (24.8)380 (26.0)0.96 [0.70, 1.31]0.78 Obese (≥30)35 (27.1)207 (14.1)1.92 [1.41, 2.62]<0.0001SEIFAb Low40 (28.6)429 (26.9)1.06 [0.81, 1.40]0.66 Low-Mid25 (17.9)288 (17.9)1.00 [0.69, 1.44]0.99 Mid-High37 (26.4)391 (24.5)1.08 [0.81, 1.44]0.61 High38 (27.1)490 (30.7)0.88 [0.67, 1.17]0.39Education Secondary or lower54 (39.7)704 (45.0)0.88 [0.71, 1.09]0.25 TAFE or equivalent38 (27.9)361 (23.1)1.21 [0.91, 1.61]0.19 University44 (32.4)498 (31.9)1.02 [0.79, 1.31]0.91Smoking37 (26.4)340 (21.3)1.24 [0.93, 1.66]0.15BP at trial entrya (mmHg) Systolic BP110.7 ± 11.2110.1 ± 10.50.6 [-1.3, 2.4]0.55 Diastolic BP66.4 ± 9.065.3 ± 8.01.1 [-0.3, 2.5]0.12a Values are mean ± standard deviation, and the comparisons are mean difference (95% CI)b Lower scores indicate lower socioeconomic statusBMI, Body mass index; BP, Blood pressure; SEIFA, Socio-economic index for areaIn unadjusted analyses women with BGDM were more likely to experience a maternal adverse outcome (RR 1.59, [95%CI 1.00, 2.52], p = 0.05) and to develop pregnancy induced hypertension (RR 1.51, [95%CI 1.03, 2.20], p = 0.03) compared with women with a normal OGCT. These differences were not seen when adjustment was made for maternal age and BMI. There was no significant difference in the rate of pre-eclampsia between the two comparison groups (Table 2).Table 2Clinical outcomes among women with borderline GDM compared with women with a normal OGCTOutcomeBorderline GDM n = 140 (%)Normal OGCT n = 1596 (%)Unadjusted relative risk [95% CI]p valueAdjusted relative risk [95% CI]p valueMaternal adverse outcome18 (12.9)129 (8.1)1.59 [1.00, 2.52]0.051.47 [0.92, 2.34]0.11Pregnancy induced hypertension25 (17.9)189 (11.8)1.51 [1.03, 2.20]0.031.31 [0.90, 1.90]0.16Pre-eclampsia9 (6.4)86 (5.4)1.19 [0.61, 2.32]0.601.08 [0.56. 2.10]0.82Antenatal hospitalisation29 (20.7)287 (18.0)1.15 [0.82, 1.62]0.421.17 [0.83, 1.65]0.36PPROM6 (4.3)41 (2.6)1.67 [0.72, 3.86]0.231.54 [0.66, 3.57]0.32Induction of labour49 (35.0)498 (31.2)1.12 [0.88, 1.42]0.341.06 [0.84. 1.34]0.62Vaginal birth94 (67.1)1184 (74.2)0.90 [0.80, 1.02]0.100.96 [0.86, 1.08]0.48 Normal vaginal birth70 (50.0)865 (54.2)0.92 [0.78, 1.10]0.361.00 [0.85. 1.17]1.00 Instrumental vaginal birth24 (17.1)319 (20.0)0.86 [0.59, 1.25]0.420.87 [0.60, 1.27]0.48Caesarean section46 (32.9)412 (25.8)1.27 [0.99, 1.64]0.061.13 [0.89. 1.43]0.33 Elective10 (7.1)101 (6.3)1.13 [0.60, 2.11]0.701.01 [0.54, 1.88]0.98 Emergency36 (25.7)311 (19.5)1.32 [0.98, 1.78]0.071.17 [0.87, 1.56]0.30Caesarean section for fetal distress24 (17.1)168 (10.5)1.63 [1.10, 2.41]0.011.43 [0.97, 2.11]0.07Major postpartum haemorrhage4 (2.9)42 (2.6)1.09 [0.40, 2.98]0.870.96 [0.35, 2.66]0.94Postpartum pyrexia3 (2.1)13 (0.8)2.63 [0.76, 9.12]0.132.33 [0.66, 8.17]0.19Maternal length of staya (days)3.5 ± 2.03.1 ± 1.70.4 [0.1, 0.7]0.010.3 [-0.0. 0.6]0.06a Value is mean ± standard deviation, and the comparison is mean difference (95% CI).The rate of induction of labour was similar and the overall caesarean section rate did not differ between groups. In the unadjusted analyses significantly more women with BGDM gave birth by caesarean section for fetal distress (RR 1.63, [95%CI 1.10, 2.41], p = 0.01) compared with women with a normal OGCT although this was not significant in the adjusted analyses. The length of postnatal hospital stay was significantly longer (mean difference 0.4 days, [95%CI 0.1, 0.7], p = 0.01) for women with BGDM compared to women with normal OGCT in the unadjusted analyses but not when adjusted for maternal age and BMI (Table 2).Overall there was no difference in the risk of death or infant adverse outcome between the two groups (Table 3). In unadjusted analyses infants born to women with BGDM were at increased risk of being born preterm (RR 1.68, [95%CI 1.00, 2.80], p = 0.05) and were also significantly more likely to be macrosomic (birthweight ≥4.5 kg) (RR 2.53, [95%CI 1.06, 6.03], p = 0.04) compared with infants born to women with a normal OGCT. When adjusted for maternal age and BMI the association with an earlier gestational age at birth and the risk of being macrosomic remained for infants born to women with BGDM compared with infants born to women with a normal OGCT (Table 3).Table 3Clinical outcomes among babies born to women with borderline GDM compared with women with a normal OGCTOutcomeBorderline GDMNormal OGCTUnadjusted relative risk [95% CI]p valueAdjusted relative risk [95% CI]p valueBirthsn = 140 (%)n = 1596 (%)Infant death or adverse outcome18 (12.9)162 (10.2)1.27 [0.80, 2.00]0.311.25 [0.79. 1.98]0.34Stillbirth1 (0.7)13 (0.8)0.88 [0.12, 6.65]0.900.79 [0.10, 6.08]0.82Neonatal death05 (0.3)--------Perinatal death1 (0.7)18 (1.1)0.63 [0.09, 4.71]0.660.56 [0.07, 4.21]0.57GA at birtha (weeks)39.7 (38.5–40.9)40.1 (39.0–41.0)--0.004--0.003Preterm birth (GA <37 weeks)15 (10.7)102 (6.4)1.68 [1.00, 2.80]0.051.64 [0.97, 2.75]0.06Very preterm birth (GA <34 weeks)4 (2.9)33 (2.1)1.38 [0.50, 3.84]0.541.40 [0.50, 3.91]0.53Extremely preterm birth (GA <28 weeks)012 (0.8)--------Apgar 5 minute <73 (2.1)33 (2.1)1.04 [0.32, 3.33]0.951.05 [0.32, 3.40]0.94Birthweightb (g)3375 ± 626.33388 ± 593.4-13.0 [-116, 89.5]0.80-28.8 [-132, 73.9]0.58Birth lengthb (cm)50.2 ± 2.550.3 ± 3.3-0.06 [-0.6, 0.5]0.82-0.09 [-0.7, 0.5]0.75Birth head circumferenceb (cm)34.3 ± 1.834.4 ± 1.9-0.10 [-0.4, 0.2]0.55-0.17 [-0.5, 0.2]0.34Livebornsn = 139 (%)n = 1583 (%)SFGA (Birthweight <10th percentile)10 (7.2)153 (9.7)0.74 [0.40, 1.38]0.350.76 [0.41. 1.42]0.39LFGA (Birthweight ≥90th percentile)19 (13.7)153 (9.7)1.41 [0.91, 2.20]0.131.29 [0.83, 2.00]0.27Macrosomia (Birthweight ≥4.5 kg)6 (4.3)27 (1.7)2.53 [1.06, 6.03]0.042.27 [0.97, 5.34]0.06Length of staya (days)3 (3–5)3 (2–4)--0.001--0.01Admission to nursery56 (40.3)450 (28.4)1.42 [1.14, 1.76]0.0021.35 [1.09, 1.68]0.01Admission to NICU9 (6.5)47 (3.0)2.18 [1.09, 4.36]0.032.05 [1.02, 4.13]0.04RDS014 (0.9)--------Mechanical ventilation2 (1.4)33 (2.1)0.69 [0.17, 2.85]0.610.65 [0.16, 2.71]0.56Antibiotics <48 hours14 (10.1)78 (4.9)2.04 [1.19, 3.51]0.012.14 [1.24, 3.68]0.01Sarnat stage 2 or 3 encephalopathy18 (12.9)124 (7.8)1.65 [1.04, 2.63]0.031.69 [1.06, 2.69]0.03a Values are median (IR range). b Value is mean ± standard deviation, and the comparison is mean difference (95% CI). GA, gestational age; SFGA, small for gestational age; LFGA, large for gestational age; NICU, neonatal intensive care unit; RDS, respiratory distress syndromeThe hospital stay was significantly longer (unadjusted p = 0.001, adjusted p = 0.01) for infants born to BGDM mothers compared with infants born to mothers with a normal OGCT (Table 3). Infants born to BGDM mothers were more than twice as likely to be admitted to NICU (unadjusted p = 0.03, adjusted p = 0.04) and more likely to be admitted to the neonatal nursery (unadjusted p = 0.002, adjusted p = 0.01). Antibiotic use less than 48 hours after birth was significantly greater among infants born to BGDM mothers (unadjusted and adjusted p = 0.01) and more infants born to the BGDM women had Sarnat stage 2 or 3 encephalopathy (unadjusted and adjusted p = 0.03) compared with infants born to women with a normal OGCT (Table 3).DiscussionIn this cohort of primiparous women in Australia, 8% were found to have BGDM. In this study, associations with BGDM were identified for maternal obesity and increasing maternal age, similar to those identified for gestational diabetes in other literature [16-18].In our study, women with BGDM had a higher risk of adverse health outcomes overall, and were more likely to develop pregnancy induced hypertension, require a caesarean section for fetal distress and have a longer postnatal hospital stay. However, we did not detect a statistically significant increase in the risk of pre-eclampsia or caesarean section overall among women with BGDM, which has been reported by previous studies [9-11]. Increasing maternal age and BMI are strongly associated with adverse maternal health outcomes. When these factors were adjusted for, no differences were seen for health outcomes between women with BGDM and normal women.We identified an increased risk of preterm birth amongst BGDM mothers. The reason for this is not readily apparent, given that there is no difference in the rate of induction of labour between the two groups. Infants of BGDM mothers were more likely to require a NICU and/or nursery admission and longer hospital stays. This may be explained by the higher rate of pregnancy induced hypertension, caesarean section for fetal distress, preterm birth and encephalopathy (Sarnat stage 2 or 3) in this group. Infants born to BGDM mothers in both unadjusted analyses and when adjusted for maternal age and maternal BMI were also at higher risk of macrosomia, which is consistent with previous studies [10,11].Our study has identified increased risks of maternal adverse health outcomes overall and a range of infant adverse health outcomes associated with BGDM. In Australia, there are over 250,000 births annually [19]. Our data suggest that a substantial number of Australian pregnant women, over 20,000 each year, will have BGDM and therefore maternal and infant adverse health outcomes that are directly or indirectly attributable to BGDM. Evidence from the Australian Carbohydrate Study in Pregnant Women (ACHOIS) trial [20] confirmed that untreated mild GDM is associated with relatively rare but nonetheless significant adverse perinatal outcomes. The trial demonstrated that the risk of these outcomes can be reduced with standard treatment consisting of individual dietary and lifestyle advice during pregnancy. There is, however, insufficient evidence regarding the benefits and harms of similar intervention for women with BGDM, with only one small clinical trial identifying a significantly reduced risk of large-for-gestational age infants with dietary advice and regular blood glucose monitoring for women with borderline glucose intolerance [21]. Data from our analysis highlight the need for well-designed large randomised clinical trials to investigate the benefits and harms of such treatment for women with BGDM.Competing interestsThe authors declare that they have no competing interests.Authors' contributionsAll authors contributed to the study design, interpretation of the data and preparation of the drafts of the manuscript. In addition CAC and ARR coordinated the study and the collection of data. KJW performed the data analyses. All authors read and approved the final manuscript.Pre-publication historyThe pre-publication history for this paper can be accessed here:\n\nREFERENCES:\nNo References"
+}

batch_5/PMC2533655.json

@@ -0,0 +1,4 @@
+{
+    "id": "PMC2533655",
+    "text": "This is an academic paper. This paper has corpus identifier PMC2533655\nAUTHORS: Mylène Riva, Philippe Apparicio, Lise Gauvin, Jean-Marc Brodeur\n\nABSTRACT:\nBackgroundIn health and place research, definitions of areas, area characteristics, and health outcomes should ideally be coherent with one another. Yet current approaches for delimiting areas mostly rely on spatial units \"of convenience\" such as census tracts. These areas may be homogeneous along socioeconomic conditions but heterogeneous along other environmental characteristics. This heterogeneity can lead to biased measurement of environment characteristics and misestimation of area effects on health. The objective of this study was to assess the soundness of census tracts as units of analysis for measuring the active living potential of environments, hypothesised to be associated with walking.ResultsStarting with data at the smallest census area level available, zones homogeneous along three indicators of active living potential, i.e. population density, land use mix, and accessibility to services were designed. Delimitation of zones ensued from statistical clustering of the smallest areas into seven clusters or \"types of environment\". Mapping of clusters into a GIS led to the delineation of 898 zones characterised by one of seven types of environment, corresponding to different levels of active living potential. Homogeneity of census tracts along indicators of active living potential varied. A greater proportion (83%) of variation in accessibility to services was attributable to differences between census tracts suggesting within-tract homogeneity along this variable. However, census tracts were heterogeneous with respect to population density and land use mix where a greater proportion of the variation was attributable to within-tract differences. About 55% of tracts were characterised by a combination of three or more \"types of environment\" suggesting substantial within-tract heterogeneity in the active living potential of environments.ConclusionSoundness of census tracts for measuring active living potential may be limited. Measuring active living potential with error may lead to misestimation of associations with walking, therefore limiting the correctness of inference about area effects on walking. Future studies should aim to determine homogeneity of spatial units \"of convenience\" along environment characteristics of interest prior to examining their association with health. Further evidence is needed to assess the extent of this methodological issue with other indicators of environment context relevant to other health indicators.\n\nBODY:\nBackgroundResidential areas are proximal to everyday life and are therefore likely to influence health of local populations through the possibility they provide for leading healthy lives [1,2]. An accumulating body of research shows evidence for variation in health across residential areas and the significance of area context for explaining this variation, independently of the characteristics of individuals [3-5].Different scales, or spatial units, may be relevant to specific contextual conditions and to specific heath outcomes [6,7], as illustrated by studies reporting varying strength and magnitude of area effects on health according to the operational definition of areas [8-15] or to contextual conditions [16-19]. Nonetheless current approaches for delimiting areas mostly rely on spatial units \"of convenience\" such as census tracts, boroughs, or wards [3,5]. These spatial units are certainly useful because they can easily be linked to data from censuses and other surveys that can be used for measuring contextual conditions. Also, they are often designed to be homogeneous along socioeconomic conditions of populations, thus being appropriate spatial units to operationalise the socioeconomic context of areas [20] (this may not hold for other administrative units, e.g. postal code areas which are design for postal delivery purposes and may be very heterogeneous in terms of population composition). However it is to be considered that through time, the composition of the units may change leading to modification of the socioeconomic conditions which may become more heterogeneous.Yet, other contextual dimensions relevant for health may not be optimally defined within administrative spatial units. For example, conduciveness of areas to physical activity or geographic accessibility to health services may operate on different scales than socioeconomic factors. Operationalising relevant spatial units for studying area effects on health remains a conceptual and methodological challenge [4,5,7,21-26] giving rise to issues of validity and soundness of areal units as units of analysis [27].Operationalising small areas: issues of validity and soundness units of analysisConstruct validity refers to whether or not the measurement instrument operationalises the concept of interest. In area effects on health research, construct validity is a matter of establishing 1) the soundness of units of analysis, i.e., whether or not area boundaries are aetiologically meaningful for studying the association between area characteristics and a given health indicator, and 2) whether or not data constitute appropriate operationalisations of exposure variables, i.e. the characteristics of areas [27]. Ideally, definitions of areas, the characteristics of these areas, and the health outcome(s) being studied should be coherent with one another [7].Measures of area characteristics derived from population censuses and other surveys, e.g. socioeconomic position, although easily accessible, provide only partial information on the context of areas and may in fact be endogenous to the composition of the areas as they are determined by individual characteristics of residents [28]. Collecting and measuring \"true\" or \"integral\" area data, i.e. data only measurable at the area level through procedures such as ecometrics and spatial analyses has been underscored as critical for measuring unbiased area-level variables [2,7,28,29]. Likewise, defining aetiologically meaningful areas in coherence with the specific purposes of the study, either in terms of health outcomes, characteristics of environment, or associations between the two [23,28,30,31] is important for understanding the significance of residential areas for health. Measurement errors can result if the spatial patterning of environmental characteristics does not correspond to the spatial units chosen for operationalising areas and their context [27].Defining relevant geographic areas becomes salient in light of the modifiable areal unit problem, i.e. the fact that analytical results are sensitive to the definition of spatial units at which data are aggregated [32,33]. In other words, area effects may be observed only at certain scales, i.e. scales at which data are collected and aggregated and may vary or be absent when observed at other scales. Imposing arbitrary spatial units on a continuous spatial process, e.g. characteristics of environments, may lead to the delineation of artificial spatial patterns. In such cases, environment characteristics may be measured with error. As a result the internal validity of the study, i.e. whether or not observed associations are unbiased, may be threatened.In addition, as per spatial autocorrelation, areas will share similar contextual conditions as a function of their proximity in space [34]. By using spatial units of convenience, it is assumed that contextual conditions within one area are different and influence health independently of conditions in neighbouring areas [4,5,21,22,24-26,35], when in fact these conditions are clustered in space. Furthermore, for any area effects to be detected there must be variation in the exposure being studied [36]. Yet the variation of environment characteristics may be smoothed out by the definition of area units used to measure them. For example, if spatial units encompass environments that are both conducive to walking and others that are less so, averaging values of conduciveness over census tracts could potentially lead to mismeasurment of exposures. Within area homogeneity along the contextual conditions under examination is thus required for minimising measurement error. Correspondingly, for inferring about area effects on health, between area differences must be maximised: if data are collected in contiguous and heterogeneous areas, variations in both characteristics of environments and health outcomes, and their association, may be misestimated. As area effects on health have been observed to be stronger in more homogeneous areas [37,38], homogeneity of areas may thus influence the estimation of area effects and therefore the validity of conclusion.In Figure 1, we propose a template that could be useful for establishing the soundness of spatial units \"of convenience\" to operationally define areas for specific research questions. For example, the template could be used to guide the decision as to whether or not census tracts are the most appropriate spatial units of analysis for measuring associations between area-level socioeconomic position (SEP) and obesity. That is, if they allow for measuring indicators of SEP without bias (and ultimately for estimating non-biased association with health outcomes) by showing homogeneity in the distribution of indicators and optimising their spatial patterning. In the methods section, we propose an approach for achieving this end. Intuitively, it can be expected that census tracts are appropriate units for undertaking such a study as they are, as mentioned above, initially designed to be homogeneous along socioeconomic conditions. But across time, the socioeconomic composition of census tracts may change as people migrate in and out of areas, potentially introducing heterogeneity in the socioeconomic make-up of the area. This could result in a \"dilution\" of the true level of deprivation. Averaging indicators of SEP over census tracts thus may mask \"pockets\" of poverty. The exercise of establishing the soundness of census tracts as units of analysis would be important here, as it would allow to measure with less error indicators of SEP and their association with health outcomes. In multilevel studies, mismeasurement of environment characteristics may influence the strength of the observed association between environment characteristics and health indicators [39]. As such, associations may not be detected or may be spurious, therefore limiting the precision of research findings for informing public health and public policy actions to tackle social and geographical inequalities in health.Figure 1Template for deciding upon the soundness of \"spatial units of convenience\" to operationally define small area units of analysis.Establishing the soundness of spatial units of analysis chosen for operationalising area boundaries and measuring area context is an important methodological consideration, but it is often overlooked. Alternatively, designing spatial units of analysis maximising homogeneity of selected environment characteristics may prove to be a viable strategy for advancing the understanding of processes linking place to health [7].ObjectivesThe aim of this investigation is to assess the soundness of census tracts as units of analysis for studying associations between a specific exposure and a specific health outcome, namely the active living potential of residential environments and walking behaviours. Active living potential refers to the conditions of areas that encourage the likelihood of integrating physical activity into daily routines [29]. Census tracts were selected as spatial units \"of convenience\" because of extensive use of this spatial unit of analysis in current research on health and place [4,5]. In Canada, census tracts are small and relatively stable geographic areas with populations ranging in size between 2500 and 8000 inhabitants; at the time of their creation, census tracts were homogeneous in terms of socioeconomic characteristics, e.g. economic status and social living conditions [40].To establish the soundness of census tracts as a unit of analysis, we developed and tested a comprehensible method for designing optimal and homogeneous spatial units espousing the spatial distribution of selected environment characteristics linked to the concept of density of destinations that is the physical and social characteristics of residential areas related to land use pattern [29]. Three indicators were used to operationalise the construct of active living potential: population density, land use mix, and geographic accessibility to proximity services. The specific objectives of the study are to examine whether or not: 1) census tracts are homogeneous units of analysis along indicators of active living potential; 2) active living potential and socioeconomic indicators follow a similar spatial distribution; and 3) census tracts encompass smaller areas with different (or similar) levels of active living potential.Active living potential was chosen because of increasing research reporting associations between this environmental construct and walking [19,41-53], an important public health indicator [54-56]. This choice was also motivated by availability of spatial datasets allowing for the operationalisation of integral measures of land use mix and geographic accessibility to services in geographical information systems, and by the availability of individual-level data on walking behaviours (to be examined in future analyses).MethodsThe methodology section includes two parts. First, we present criteria and methods for designing homogeneous areas (henceforth designated as \"zones\"). Second, we present analyses undertaken to assess the soundness of census tracts as units of analysis for measuring the active living potential of residential areas.Designing optimal, homogeneous zonesZone design refers to the placement of areal unit boundaries [9]. It can be achieved discursively (manually) by grouping basic spatial units into larger ones [57-59], by combining social, statistical, and spatial analysis methods [60,61], and automatically through computationally intensive automated zoning software [9,15,37,62-65].Three criteria guided the choice of the method for zone design. First, we wanted to design zones based on the spatial distribution of environmental characteristics related to active living potential, namely population density, land use mix, and geographic accessibility to selected proximity services. We had no requirement regarding population and area sizes as zones were defined on the basis of the spatial distribution of these characteristics. Second, the method for zone design had to be optimal, i.e. to maximize variation between zones and to minimize variation within zones in the selected characteristics. In other words, the aim was to design zones that were internally homogeneous on the three indicators of active living potential, but different (heterogeneous) amongst themselves. Finally, we wanted a method that was rigorous but comprehensible and easy to implement. We opted for an approach that combined a statistical classification method, K-means clustering, to mapping applications in geographic information system. This three-step approach is described in greater details in the following sections.Step 1: Measuring environment characteristics at the smallest area levelThe study area is the Island of Montreal, Canada, an urban centre with 1 812 723 residents. As of January 2006, on the Island of Montreal, there are 15 municipalities, in addition to the municipality of Montreal which includes 19 boroughs [66]. The Island of Montreal is further divided into 521 census tracts and 3222 dissemination areas. Dissemination areas (DAs) were used as basic spatial units for designing zones because they are the smallest standard geographic areas for which Canadian census data are available (population size between 400 and 700 residents) [40]. On the Island of Montreal, their average size is 0.15 km2 (ranging between 364 m2 and 18 km2) with an average population of 562 individuals (ranging between 44 and 2138 residents). DA values for population density, land use mix, and accessibility to services were computed in a geographical information system (ArcGIS 9.2) [67].Population density refers to number of individuals per unit area. It was computed by dividing the total number of residents of a DA by its area size (km2) [68].Land use mix relates to the diversity or variety of land uses within an area. It was computed using an entropy index [47,69,70] which measures the homogeneity or diversity of land uses within a spatial unit. The index is defined as follow:(1)Ej=−∑i=1n[(Aij/Dj)ln⁡(Aij/Dj)]/ln⁡nWhere Aij is the surface area of land use i in dissemination area j, Dj is the surface area of dissemination area j, and n is the total number of possible land uses which in the current case corresponds to 16, the number of different land uses characterising the Island of Montreal [71]. The index values range between 0 and 1, where 1 corresponds to a highly mixed area, and 0 to a homogeneous area, that is an area characterised by only one type of land use (e.g. low density housing). This index has been used in many studies to measure land use mix [47,72].Geographic accessibility to proximity services refers to geographic distance to or from destinations, here to supermarkets, pharmacies, banks, and libraries. These services were selected because they are most likely to be used on a regular basis, conveying the idea of proximity services potentially accessible through walking. There are many measures of geographic accessibility [73,74]. In this study, geographic accessibility was defined in terms of the number of the selected services within an area, conferring the notion of the offer of services provided by the immediate surroundings. Supermarkets, pharmacies, banks, and libraries were geocoded at the parcel level [75]. In order to minimise aggregation errors [73,76], accessibility was measured by computing distances of services located within a one kilometre (network distance) radius [77] from the centroid of census blocks (n = 14 527) comprised within any one DA; the distances were than averaged and weighted by the total population of each census blocks.Characterisation of DAs along the three indicators resulted in a sample of 3206 DAs. Measures of land use mix and accessibility to services were normally distributed; population density was normalized using a LOG10 transformation [78]. Population density was significantly and positively correlated to accessibility to services (r = 0.45, p < 0.001), and negatively to land use mix (r = -0.32, p < 0.001). Land use mix and accessibility to services were not significantly correlated (r = 0.03, p > 0.500). Prior to cluster analyses, these variables were standardized to a mean of 0 and a standard deviation of 1, higher values representing greater levels of population density, land use mix, and accessibility to services.Step 2: Classifying smallest areas into clusters, e.g. \"types of environments\", using K-means clusteringK-means statistical clustering techniques using SAS (version 9.1) for Windows [79] was applied to classify DAs into k number of optimal clusters homogeneous in terms of active living potential. In social sciences, notably in geography, K-means is largely employed to classify areas (e.g. geodemographics [80]). The method uses an allocation/re-allocation algorithm to optimally reassign objects, here DAs, to the nearest cluster centroid [81-83]. The goal is to maximize between cluster variations and to minimize within cluster variations. The aim of this second step was to group DAs with similar values of population density, land use mix, and accessibility to services into k types of environments that are internally homogeneous but different among them. These types of environments correspond to different levels of active living potential. For K-means clustering, the number of clusters (k) must be determined at the onset of analyses; as we had no a priori for such number, we conducted analyses for k = 4 to k = 20.Step 3: Mapping the clusters to create optimal and homogeneous zonesIn a final step, the k types of environments were imported into ArcGIS 9.2 and mapped out. This lead to the delineation of n homogeneous zones i.e., units of analysis, characterised by one of k active living potential.Statistical analyses: Assessing the soundness of census tracts as units of analysis for operationalising active living potentialThe soundness of census tracts for operationalising indicators of active living potential was assessed through three series of analyses.First, to assess the homogeneity of census tracts, variation in indicators of active living potential was estimated and decomposed between and within areas. Population density, land use mix, and accessibility of services were measured continuously at the DAs level (level 1: n = 3206). In separate two-level multilevel models, DAs were nested into zones (n = 898) and into census tracts (n = 506 with valid population and socioeconomic data). Between-area variation in indicators of active living potential was estimated using the intraclass correlation coefficient (ICC) from unconditional (null) multilevel models using HLM software Version 6.04 [84]. The ICC indicates the proportion of variation in a dependent variable that is attributable to differences between area units. Greater ICC values indicate that variation of a variable is greater between units than within, i.e. units are different among them but internally homogeneous. Using the same analytical approach, homogeneity of zones and census tracts along indicators of socioeconomic position was assessed and compared. DA-level data on the proportion of low-income households, of people with less than high school education, and of people with a university degree were obtained from the 2001 Canadian census.Second, analysis of variance was performed to examine the proportion of variation across zones in socioeconomic variables explained by the k types of environment. Indicators of SEP at the DA-level were aggregated (weighted by population) at the zone-level. These analyses were performed to examine whether or not socioeconomic and active living indicators follow a similar spatial distribution as is implicitly assumed when measured within the same area unit of analysis.Finally, descriptive statistics were employed to assess the extent to which the spatial distribution of the different types of environment coincides with the boundaries of census tracts. These analyses were conducted to examine if census tracts encompassed environments with differing levels of active living potential. The numbers of zones straddling over one or more census tracts, and the number of types of environment encompassed within census tracts were computed. To examine whether or not the spatial distribution of more mixed or more homogeneous census tracts (i.e. the number of types of environments encompassed within census tracts) was structured in space, global values of spatial autocorrelation were computed using Moran I with a first-order contiguity matrix [85,86]. Values for Moran I vary between -1 and 1, where negative values indicate negative spatial autocorrelation, i.e. neighbouring spatial units have different values, and positive values indicate positive spatial autocorrelation, i.e. neighbouring units have similar values. The covariance in Moran I is the covariance over space for neighbouring spatial units, and will not be computed unless two units are contiguous (first order); also, only one variable is considered [85], here the number of types of environments included in census tracts.ResultsDescription of types of environment and zonesFigure 2 illustrates results of the K-means clustering, which show that the 3206 DAs were optimally classified into 7 clusters or \"types of environments\" as indicated by peaks [87] in both the Pseudo-F statistic [88] and the Cubic clustering criterion [89]. These clusters explain 72.8% of the total variation in the three indicators of active living potential. Thus, differences among the seven clusters and similarity of DAs comprised within the same cluster, i.e. within-cluster homogeneity, were both maximized. The seven types of environments correspond to seven different levels of active living potential. They encompassed more suburban to more central urban types of environments defined by different values of population density, land use mix, and accessibility to services. The types of environments are described in Figure 2 and Figure 3.Figure 2Statistical proximity of the seven types of environment (clusters).Figure 3Description of types of environment (clusters) and zones.Low-density and mid-density suburban areas are characterised by lower values of population density and accessibility to services. Diverse central urban areas and central urban areas with high accessibility are more densely populated and have greater access to services than any other types of environment. Although population density and accessibility to services follow to some extent an increasing gradient from more suburban to more urban areas, the pattern of land use mix is more complex: there are low values in urban areas and high values in suburban areas. Dissemination areas are designed to be similar in population size (among other characteristics); thus the area size required to reach the set population threshold (i.e. between 400 to 700 residents [40]) will be larger in less densely populated areas and smaller in more urban areas. As a consequence, larger dissemination areas are more likely to encompass different land use than are smaller dissemination areas located in urban areas.Figure 2 also presents the statistical proximity (Euclidian distance) of the centroids of clusters (cluster mean values), i.e. types of environment, in a three dimensional graph where the axes correspond to the three indicators of active living potential. With respect to their spatial distribution, the types of environment are positively correlated in space indicating that contiguous zones were characterised by similar types of environment.Mapping of the clusters into the GIS led to the delineation of 898 zones or units of analysis characterised by one of the seven types of environments, i.e. active living potential, as illustrated in Figure 3. Zones are significantly smaller than census tracts, an average of 0.54 km2 (SD = 3.50) compared to 0.96 km2 (SD = 1.98) (t = -2.46; p < 0.05), but the variation of their area size is not statistically different (F = 0.68; p = 0.409). Zones are significantly smaller than census tracts in population size, an average of 1960 (SD = 3867) residents compared to 3554 (SD = 1647) (t = -10.73; p < 0.001), and there is significantly greater variability in population size across zones than across census tracts (F = 11.40; p < 0.01). Zones characterised by more suburban contexts are on average larger and have relatively smaller population counts than urban zones.Soundness of census tracts as units of analysis for measuring active living potentialHomogeneity of census tracts along active living potential indicatorsResults of homogeneity of zones and census tracts along active living indicators appear in Figure 4. The variation in indicators is not uniform across census tracts. A greater proportion (83%) of variation in accessibility to services is attributable to differences between census tracts, as indicated by a higher ICC value, suggesting within census tract homogeneity along this indicator. Yet about half of the variation in population density is between census tracts (52%), whereas there is greater variation in land use mix within census tracts (85%), indicating greater heterogeneity of tracts along these indicators. The degree of homogeneity of tracts therefore varies according to the indicator examined. For population density and land-use mix, but not for accessibility of services, variation between zones is greater than variation between census tracts. This shows that the method was successful in designing areas or units of analysis that were more homogeneous than census tracts along dimensions of active living potential.Figure 4Decomposition of variation in indicators of active living potential and socioeconomic position across zones and across census tracts.The degree of homogeneity of census tracts and zones along socioeconomic indicators shows that for the selected variables, variability is larger between census tracts than between zones (Figure 4). Census tracts are relatively homogeneous areas in terms of the socioeconomic environment, especially for proportion of population with a university education.Spatial distribution of active living potential and socioeconomic indicatorsExamining variation in socioeconomic indicators across zones shows that they follow a different spatial distribution than that of active living potential indicators. Results of analyses of variance (results not shown) revealed that 15.2% of the variation in the proportion of low-income households was explained by the seven types of environment whereas these proportions were 5.2% for the proportion of people with less than high school and 3.8% for the proportion of people with a university education.Types of environments encompassed within census tracts boundariesOverall, zones are not well contained within census tracts. As shown in Figure 5, only 30.5% of zones are completely located within the boundaries of one census tract. Forty-eight percent of zones straddle two or three census tracts whereas, 21.5% spread over more than four tracts. Correspondingly, there is considerable variability in types of environment within census tracts.Figure 5Proportion of zones straddling different numbers of census tracts across the Island of Montreal.As illustrated in Figure 6, 11.2% of census tracts encompass only one type of environment and 34.3% encompass two types. About 28% of census tracts are characterised by three different types of environment, whereas 26.3% comprise 4 or more different types. Among census tracts encompassing two types of environment (n = 175), about two-thirds (66.3%) comprise types that are statistically similar as indicated by distances between their centroids (two or less distance lag as indicated in the distance matrix in Figure 2; results not shown). For example, census tracts often comprise a combination of low-density suburban and suburban/urban axial zones (26.3%), or a grouping of diverse and high accessibility central urban areas (35.4%). Globally, the number of types of environment encompass within census tracts is positively correlated in space (Moran I = 0.26; p < 0.001), suggesting that more homogeneous or more mixed census tracts are often contiguous in space (Figure 6). More heterogeneous census tracts are located mainly on the periphery of central urban areas and in the eastern part of the Island of Montreal, and to a lesser extend in the west-end suburbs.Figure 6Number of types of environment encompassed within census tract boundaries.DiscussionThe objective of this study was to assess the soundness of census tracts as units of analysis, i.e. their degree of homogeneity in terms of the active living potential of residential environments associated with walking. In order to do so, homogeneous zones that optimised the spatial patterning of active living potential indicators hypothesised to be associated with greater involvement in walking, namely population density, land-mix use, and accessibility to services, were successfully designed. This was done through the application of an easy-to-use method combining a classification method called K-means clustering with basic mapping applications of geographical information systems. The degree of soundness of census tracts as units of analysis was established through a series of analyses comparing them to the newly-designed zones.First the distribution of the three active living indicators between and within census tracts was assessed. Although census tracts were homogeneous in terms of accessibility to services, they were less homogenous in population density; for this indicator within and between census tracts variations were about equal. Census tracts were clearly not homogeneous in terms of land use mix as the variability within tracts largely exceeded the variability between tracts. In contrast, census tracts were homogeneous along socioeconomic variables. These results suggest that the spatial patterning of the active living potential of environments do not neatly follow in the delineation of census tracts, which may be more suitable as units of analysis for operationalising socioeconomic contexts.Then, findings revealed that the spatial distribution of active living and socioeconomic indicators followed different spatial distribution. At the zone-level, types of environment explained a small proportion of variation of socioeconomic variables. This indicates that processes underlying the distribution of active living and SEP indicators, although potentially linked [2,6], operate at different scales and thus require different units of analysis.In the final set of analyses, within tract variability in terms of what we labelled \"types of environment\" was examined. This allowed for the assessment of whether or not census tracts encompassed environments that were substantively different among them in terms of their active living potential. Census tracts comprising two different types of environments (34.3%) were not considered necessarily as problematic, given that some types of environment were more similar than others and were often contiguous in space. For example, diverse and high accessibility central urban zones were often contiguous in space and were statistically most similar (as indicated by statistical distances between clusters; Figure 2). However, census tracts comprising three or more types of environment raised concerns; such a situation was observed in more than half of census tracts. These tracts encompass environments that are simultaneously most conducive to walking and others that are least so. Averaging values of conduciveness to walking could potentially lead to significant errors when measuring active living potential at the census tract-level.The approach for defining areas or units of analysis differs from those involving the definition of strictly \"ecologically meaningful\" or \"natural\" neighbourhoods, i.e. neighbourhoods imbued with meaning for residents [21] or as consisting of a group of homes sharing a commonly defined residential area often having name [20]. Defining such units of analysis is important when the notion of commonly shared territory is related to the contextual condition of interest, for example social capital or collective efficacy [7,27]; this notion is not conjured up by active living potential. Designing zones based on the spatial distribution of active living indicators empirically linked to greater involvement in walking leads to the definition of areas that are more appropriate units of analysis and increases the internal validity of study design examining the environmental determinants of walking.Future studies are needed to assess the impact of the choice of other environmental characteristics for designing zones relevant to other health indicators, and to other geographical areas. For example, areas relevant for studying the social and environmental determinants of overweight and obesity may be delimited according to the distribution of active living variables and food provision (accessibility of both healthy and non-healthy food). For studying mental health outcomes, social dimensions of area context such as social support and opportunities for social participation, may be more relevant. It is to be expected that designing zones using other indicators of contextual conditions associated with other health outcomes will lead to different spatial configuration of area units of analysis.Homogeneous zones are designed with the aim of optimising the study of a phenomenon or for the purpose of uncovering the aetiology underlying associations between area context and health. As such, the configuration of zones should not be viewed as other \"spaces\" of actions for public health and policy interventions. Rather, they may be useful for informing on viable interventions and policy strategies that may be health promoting.LimitationsResults of this study should be considered in light of some limitations. First, there is a seven year time lag (2000 to 2006) between the dates of creation of the different datasets used to characterise dissemination areas in terms of their active living potential and socioeconomic position. Although changes in the built environment may have taken place during this period, the speed at which changes occur is not well documented; however over a seven-year period, changes in the built environment can be expected to be modest.Other indicators of active living potential could be examined in designing homogeneous areas, such as street connectivity, safety, and accessibility to other services or resources such as parks. In this study, the measurement of land use mix was dependent on the size of disseminations areas which are defined in part by a population size threshold: because of lower population density in suburban areas, DAs are likely to span a greater territory and therefore encompass more types of land use. Other scales for measuring land use mix could be considered [47].ConclusionFor studies concerned with the social and environmental determinants of health and more specifically of physical activity, results of this study have several implications. Delimiting areas is a key conceptual and methodological challenge in research on health and place. In this paper, we developed an easy-to-use method for establishing homogeneous units of analysis in terms of specific environmental characteristics hypothesised to be linked to a specific health indicator. The focus was on active living potential of areas and walking behaviours. Using these homogeneous zones as comparison, the objective was to assess the soundness of spatial units \"of convenience\", i.e. census tracts, to operationalise contexts for which they were not purposely developed. The methods developed in this study add to the growing literature on alternative ways to conceptualise and define the boundaries of area units for studying the determinants of health.Findings showed that although census tracts may be homogeneous along independent indicators of active living potential, they were most often characterised by a combination of types of environment that were substantively different in terms of their active living potential. For this reason, census tracts should be used with caution as units of analysis when operationalising active living potential for studying determinants of walking. But census tracts or other administratively defined areas may be appropriate area units, i.e. may be homogeneous enough, when processes hypothesised to be operating on health are linked to the socioeconomic context of an area, for example affluence or poverty.In this study, zones were delimited for methodological and aetiological purposes with the aim of minimising measurement errors of environmental characteristics and increasing internal validity of study design for measuring area effects on health. As can be expected, the zones are context-specific and cannot be exported to other geographic areas. Rather they are representations of the local realities of processes relating environmental characteristics to health. As suggested by others, the geographical aspects of the study design should be considered prior to conducting analyses [15]. Establishing the soundness of spatial units \"of convenience\" for representing the environmental and spatial processes under investigation should be part of the empirical approach for conceptualising, operationalising, and measuring area effects on health.Competing interestsThe authors declare that they have no competing interests.Authors' contributionsMR conceptualised the study. She carried out spatial and statistical analyses, mapping of results, and drafted the manuscript. PA, LG and JMB participated in the conceptualisation the study, and in data analyses. All authors critically revised the paper, and read and approved the final manuscript.\n\nREFERENCES:\nNo References"
+}

batch_5/PMC2533678.json

@@ -0,0 +1,4 @@
+{
+    "id": "PMC2533678",
+    "text": "This is an academic paper. This paper has corpus identifier PMC2533678\nAUTHORS: Haobing Li, René Vaillancourt, Neville Mendham, Meixue Zhou\n\nABSTRACT:\nBackgroundResistance to soil waterlogging stress is an important plant breeding objective in high rainfall or poorly drained areas across many countries in the world. The present study was conducted to identify quantitative trait loci (QTLs) associated with waterlogging tolerance (e.g. leaf chlorosis, plant survival and biomass reduction) in barley and compare the QTLs identified across two seasons and in two different populations using a composite map constructed with SSRs, RFLP and Diversity Array Technology (DArT) markers.ResultsTwenty QTLs for waterlogging tolerance related traits were found in the two barley double haploid (DH) populations. Several of these QTLs were validated through replication of experiments across seasons or by co-location across populations. Some of these QTLs affected multiple waterlogging tolerance related traits, for example, QTL Qwt4-1 contributed not only to reducing barley leaf chlorosis, but also increasing plant biomass under waterlogging stress, whereas other QTLs controlled both leaf chlorosis and plant survival.ConclusionImproving waterlogging tolerance in barley is still at an early stage compared with other traits. QTLs identified in this study have made it possible to use marker assisted selection (MAS) in combination with traditional field selection to significantly enhance barley breeding for waterlogging tolerance. There may be some degree of homoeologous relationship between QTLs controlling barley waterlogging tolerance and that in other crops as discussed in this study.\n\nBODY:\nBackgroundWaterlogging is one of the major restrictions for barley production in high rainfall areas. It causes chlorophyll, protein and RNA degradation and also decreases the concentration of nutrients such as nitrogen, phosphorus, metal ions and minerals in barley shoots. These can occur rapidly after the onset of flooding, precede leaf chlorosis [1-3], and consequently reduce shoot and root growth, dry matter accumulation and final yield [4-8]. The average yield loss due to waterlogging is estimated to be 20–25% and can exceed 50% depending on the stage of plant development affected [9].Barley cultivars differ in their tolerance to waterlogging. The barley collections from China, Japan and Korea contained many tolerant cultivars while those from North Africa, Ethiopia and southwest Asia showed few tolerant cultivars [10]. Fufa and Assefa [11] suggested that locally adapted landraces could be major sources of tolerance. Our previous studies showed some Chinese cultivars showed significantly better tolerance than Australian cultivars [12-14]. Thus it is possible to breed for tolerance. However, waterlogging tolerance is likely to be a complex trait affected by several mechanisms and complicated by confounding factors such as temperature, plant development stage, nutrient availability, soil type and sub-topography. Direct selection on grain yield has low effectiveness since the heritability of yield after waterlogging has been reported to be very low [15]. Different traits have been used as indirect selection indices for waterlogging tolerance. Among them, leaf chlorosis after waterlogging is one of the major indices used by researchers in different crops such as wheat (Triticum spp.) [16-19], soybean (Glycine max) [20]and barley [21]. Waterlogging tolerance has been found to be controlled by one dominant gene in common wheat [18], Makha wheat (Triticum macha) [22]and maize (Zea mays ssp. mays) [23]. In barley, based on leaf chlorosis, waterlogging tolerance was found to be a quantitative trait and mainly controlled by additive genetic variation [12,24]. Even though the heritability was relatively high for leaf chlorosis [12] and early generation selection could be efficient, well-controlled waterlogging conditions are still crucial for the precise evaluation of this trait. In practice, it is very difficult for breeders to control the multiple confounding environmental factors in a field experiment over thousands of barley genotypes. Development of molecular markers associated with barley waterlogging tolerance and marker assisted selection (MAS) could effectively avoid environmental effects. QTL analysis has proven to be very useful in identifying the genetic components of the variation for important economic traits [25]. A molecular marker closely linked to the target gene or QTL can act as a \"tag\" which can be used for indirect selection of the gene(s) in a breeding programme [26]. Great progress in molecular mapping of economically important traits in barley has been made [27]. Little progress, however, has been made in mapping QTLs controlling waterlogging tolerance in barley because it is affected by many factors in the natural environment [28]. With recent research showing that leaf chlorosis and some other physiological traits may be practical to use in the evaluation of waterlogging tolerance in barley [13,14], QTL identification has become possible. In this paper, we report on the identification of QTLs for waterlogging tolerance in two barley double haploid (DH) populations based on leaf chlorosis, plant survival and biomass reduction after waterlogging and comparisons were made between different populations and under different growing seasons.MethodsPopulations used for QTL analysisThe first population consisted of 92 doubled haploid (DH) lines from a cross between TX9425 and Franklin. TX9425 is a feed barley with waterlogging tolerance and originates from China, while Franklin is an Australian malting barley and is susceptible to waterlogging. The two parents also differ in malting quality, resistance to some diseases and several agronomic traits[29]. The second population consisted of 177 doubled haploid lines from the barley cross between Yerong and Franklin. Yerong is an Australian six-rowed variety with good tolerance to waterlogging stress.Map constructionDArT protocolGenomic representations and preparation of the \"discovery arrays\" and \"polymorphism-enriched arrays\" were the same as explained by Wenzl et al. [30]. A quality parameter Q, which is the variance of the hybridization intensity between allelic states as a percentage of the total variance, was calculated for each marker. Only markers with a Q and call rate both greater than 80% were selected for linkage analysis.SSR analysis142 SSR primers were screened for polymorphism between the four parents of the two populations and 104 primers showed polymorphisms. Twenty-eight polymorphic primers were selected for genotyping the DH populations using four well-separated primers for each of the seven chromosomes.AFLP analysisAFLP markers were assayed only in the Franklin/TX9425 population. AFLP methodology was performed following Vos et al [31] with minor modification: Genomic DNA (250 ng) from the two parents and the DH lines was restricted with 2.5 u each of EcoRI and MseI in a 20 μL reaction mixture for 2 hours at 37°C. Ligation mixtures of 20 μL containing the EcoRI and MseI adaptors, 1 U T4 DNA ligase, 0.4 mM ATP in 10 mM Tris-HCl (pH 7.5), 10 mM magnesium acetate, and 50 mM potassium acetate were added. Ligation mixtures were incubated at 16°C overnight. The reagents and thermo-cycling conditions for pre-selective and selective amplification followed Vos et al [31]. Pre-selective primers (EcoRI +A, MseI +C) and selective amplification primers (EcoRI +3, MseI +3) were described by Freeman et al [32]. The selective EcoRI (+3) primers were fluorescently labelled with TET for detection by a Gel Scan 2000. AFLP samples from the selective amplification were combined with two volumes of formamide B-blue loading buffer (98% v/v formamide, 10 mM EDTA, 0.25% w/v bromophenol blue, 0.25% w/v xylene cyanol) and denatured at 90°C for 3 min. Two μL of each sample was loaded onto 18 cm 6% w/v denaturing polyacrylamide gel with 7.0 M urea and electrophoresed in a 1% v/v TBE buffer at 1400 V for 1.5 h. Gene Profiler 4.03{3} software was used to extract data and score the traces. AFLP fragments were given a three-point confidence rating denoting their quality and ease of scoring. All AFLP markers were named using a code for each primer combination, followed by sequential numbers for scored bands e.g. p3b1.Linkage analysisThe segregation signatures of each of the two individual datasets were imported into JoinMap 3.0 to distribute loci into linkage groups. LOD thresholds (from LOD 3 to LOD 10) were tested to group the markers, until a LOD threshold was obtained for each population that resulted in the optimum number of markers in linkage groups in which linkage order and distances were maintained. Marker order analyses were conducted with a JMMAP LOD threshold of 0.1 and a REC threshold value of 0.45. In order to obtain a rigorous marker order, framework maps were constructed using only non-distorted markers. Some distorted markers were then added into the data set gradually and integrated into the map frameworks. In most cases, the introduction of distorted markers did not affect the statistical confidence of marker order, or just changed the order of markers within small regions with high marker density. The genetic linkage map from the population of TX9425/Franklin comprised 412 DArT, 80 AFLP and 28 microsatellite markers and the map from the population of Yerong/Franklin comprised 496 DArT and 28 microsatellite markers.Evaluation of waterlogging tolerance of the DH linesFour replicates of ten seeds for each DH and parental line were sown in soil in 3.5 L pots (one pot of each line per replicate) filled with soil from a frequently waterlogged site (Cressy Research Station) in Tasmania. Several measures were adopted to reduce the effects of variation in the degree of soil compaction across pots and also other sources of variation on the waterlogging conditions. First, the same type of pots was used through all the experiments. Second, we measured the same amount of soil for each pot and made sure the soil was packed to the same level in each pot. Third, the bottom of the water tanks or pools were checked to ensure they were flat and level. Finally, seeds were sown at the same depth in each pot.After germination, five plants were kept in each pot and grown in a glasshouse under natural daylight but temperature controlled to less than 24°C. Waterlogging treatments were conducted in children's paddling pools. Each replicate was placed into a different pool and the two populations were placed in pools of different size. A randomised design was used for each pool. Three replicates were subjected to waterlogging and one replicate was not waterlogged as a control for the experiments. Waterlogging was achieved by filling the pool with water to just cover the soil surface in the pots. Waterlogging was started at the 3-leaf stage, and lasted three to eight weeks depending on the trait measured. This experiment was carried out in 2004 and repeated in 2005 using fresh soil and seeds.The first trait measured was the proportion of each leaf that had lost its green colour (was yellow), this trait was called leaf chlorosis. Leaf chlorosis was chosen as the main indicator for waterlogging tolerance because other studies have found it to be correlated with yield reduction resulting from waterlogging stress [33]. This trait was measured three times for each population across the two experimental years (Table 1). Leaf chlorosis was measured as follows: the proportion of yellowing or chlorosis on each leaf was visually scored, then the length of each leaf was measured to weight the overall average proportion of chlorosis in each plant. Then an average was calculated for all the plants in each pot. The control plants of both populations in both years had no leaf chlorosis.Table 1Traits measured in the two barley mapping populations.Traits measured in each populationYear of measurementDuration of waterlogging stress20042005Franklin/TX9425 Leaf chlorosis 1.1×two weeks Leaf chlorosis 1.2×four weeks Leaf chlorosis 2.1×two weeks Plant survival×eight weeks Plant biomass reduction×three weeksFranklin/Yerong Leaf chlorosis 1.1×two weeks Leaf chlorosis 2.1×two weeks Leaf chlorosis 2.2×four weeks Plant survival×eight weeks Plant biomass reduction×three weeksThe second trait measured was plant biomass reduction. This trait was measured in 2004 for the Franklin/Yerong population and in 2005 for the Franklin/TX9425 population (Table 1). After three weeks of waterlogging treatments, barley plants were cut at ground level and dried at 60°C for four days in an electric oven. The average plant dry weight was measured for each replicate in both the control and in waterlogging treatments. Plant biomass reduction was calculated by subtracting the average dry weight of plants grown in waterlogging conditions from that in the control, then dividing by the average dry weight in the control. The third measured trait was plant survival. After eight weeks of waterlogging, dead plants in each pot were counted after the water was drained. Measurements were done in 2004 for Franklin/TX9425 and in 2005 for Franklin/Yerong (Table 1). Plant survival was calculated as the numbers of surviving plants divided by the initial number of plants in each pot.Statistical analysisStatistical analysis was undertaken to detect significance of genetic effects for each trait in each population and also to calculate broad-sense heritability. For each experiment, the following mixed-effects model was used: Yij = μ + ri + gj + wjj. Where: Yij = observation on the jth genotype planted in the ith replication; μ = general mean; ri = effect due to ith replication; gj = effect due to the jth genotype; wij = error or genotype by replication interaction, where genotype was random and replicate treated as a fixed effect in analysis conducted using PROC MIXED of SAS. As part of the model checking procedure, SAS PROC UNIVARIATE was used to verify that the residuals were normally distributed. Broad-sense heritabilities were calculated for each trait as the ratio of the genetic variation (genotype) divided by phenotypic variation (due to genotype and residual). In order to calculate least square means for each genotype by trait by population by experiment combinations, PROC GLM was used with the same model as above, except that genotype was treated as a fixed effect. The normality of each trait distribution was checked using SAS PROC UNIVARIATE for skewness and kurtosis.Using the software package MapQTL5.0 [34], QTLs were first analysed by interval mapping (IM), followed by composite interval mapping (CIM). The closest marker at each putative QTL identified using interval mapping was selected as a cofactor and the selected markers were used as genetic background controls in the approximate multiple QTL model (MQM) of MapQTL5.0. Logarithm of the odds (LOD) threshold values applied to declare the presence of a QTL were estimated by performing the genome wide permutation tests implemented in MapQTL version 5.0 using at least 1000 permutations of the original data set for each trait, resulting in a 95% LOD threshold between 2.7 and 3.0. One or two LOD support intervals around each QTL were established, by taking the two positions, left and right of the peak, that had LOD values of one and two less than the maximum [34], after performing restricted MQM mapping which does not use markers close to the QTL. The percentage of variance explained by each QTL (R2) was obtained using restricted MQM mapping implemented with MapQTL5.0.ResultsPhenotypic and genetic variation among the DH lines of the two populationsLeaf chlorosis, plant survival and plant biomass reduction following waterlogging stress showed normal distributions for both populations with no significant skewness and kurtosis. Summary statistics for each trait are presented in Table 2 for both populations. Transgression beyond the parental values was observed for all traits including those for which parental values hardly differed. There was significant variation between DH lines (genetic variation) in each population for all the measured traits (Table 2). The effect of replication was not significant for traits measured early in the experiments, but was significant for most traits measured later. The broad sense heritabilities of the various traits ranged from 0.71 to 0.11 in the Franklin/TX9425 population and from 0.57 to 0.20 in the Franklin/Yerong population (Table 2). Biomass reduction was the ratio of the biomass of waterlogged plants divided by their control. Since the control consisted of only one replicate, due to limited glasshouse space, the results for biomass reduction need to be treated with caution.Table 2Descriptive statistics of the investigated waterlogging traits in the Franklin/TX9425 and Franklin/Yerong DH populations, with means for each parent, minimum/maximum/mean values of DH lines, standard deviation (SD) and probability (Prob Z) of significant variation among DH lines, and estimated broad-sense heritability (H2).Mean for parentsDH linesTraitsFranklinOther parentMin.MaxMeanSDProb ZH2Franklin/TX9425 Leaf clorosis 1.10.100.340.040.400.190.08< 0.00010.56 Leaf chlorosis 1.20.210.340.100.540.300.09< 0.00030.11 Plant survival0.930.740.001.000.550.28< 0.00050.31 Leaf chlorosis 2.10.050.340.020.350.160.09< 0.00010.71 Plant biomass reduction0.370.510.180.710.430.110.00750.30Franklin/Yerong Leaf chlorosis 1.10.130.190.040.270.140.05< 0.00010.34 Plant biomass reduction0.280.44-0.051.050.390.19< 0.00010.22 Leaf chlorosis 2.10.050.240.000.270.090.06< 0.00010.20 Leaf chlorosis 2.20.280.380.150.650.340.08< 0.00010.57 Plant survival0.220.200.001.000.300.230.0030.25Identification of QTLs associated with waterlogging tolerance in Franklin/TX9425Three QTLs (tfy1.1-1, tfy1.1-2 and tfy1.1-3) controlling leaf chlorosis after two-weeks of waterlogging stress (2004) were identified (Table 3, Figure 1). For all these QTLs, the Franklin alleles increased leaf chlorosis while the TX9425 alleles decreased it. One QTL (tfy1.2-1) was identified for leaf chlorosis after four-weeks waterlogging (2004) treatment. This is likely to be the same QTL as tfy1.1-2 as it was mapped to the same position and the Franklin allele also increased leaf chlorosis. Two QTLs (tfy2.1-1 and tfy2.1-2) were found for leaf chlorosis in the experiment carried out in 2005. QTL tfy2.1-1 is likely to be the same as tfy1.1-2 and tfy1.2-1 as it is in the same position and again the Franklin alleles increased leaf chlorosis.Figure 1The Franklin/TX9425 chromosomes showing the locations of QTLs for the traits analyzed. Each linkage group consists of a vertical bar on which the map positions and names of loci are indicated. QTL positions are shown through their support interval on the right of each chromosome. One LOD support intervals are the inner intervals, while the outer intervals represent the two LOD support intervals. Prefix \"bPb\" and \"p\" signify a DArT marker and a AFLP marker, respectively. The other markers on the map are microsatellites.Table 3Characteristics of the detected QTLs explaining waterlogging related traits in the Franklin/TX9425 population.TraitQTLChr.One LOD support interval (cM)LOD scoreR2 (%)Leaf chlorosis 1.1tfy1.1-12H31–359.2123.3(two weeks stress, 2004)tfy1.1-23H68–717.5933.4tfy1.1-31H61–672.757.1Leaf chlorosis 1.2tfy1.2-13H67–717.3136(four weeks stress, 2004)Leaf chlorosis 2.1tfy2.1-13H68–739.2834.1(two weeks stress, 2005)tfy2.1-27H72–983.6216Plant biomass reductiontfmas4H47–782.7516.3Plant survivaltfsur-12H49–653.2919tfsur-22H15–182.7513.2Although the difference in the reduction of plant biomass due to waterlogging stress between TX9425 and Franklin was small (Table 2), one QTL (tfmas) was identified for plant dry weight reduction after three-weeks of waterlogging stress (Table 3). This QTL was mapped to chromosome 4H. Compared to the TX9425 allele, the Franklin allele led to a greater reduction of plant biomass following waterlogging.Two QTLs (tfsur-1 and tfsur-2) were found for plant survival rate after eight weeks continuous waterlogging stress (Table 3). Both of these were located on chromosome 2H. These QTLs were located onto different regions of chromosome 2H compared with the QTLs for leaf chlorosis. This confirms the statistical analysis results showing no significant correlation between these two traits (results not shown). For the detected QTLs, the Franklin allele increased the survival rate of the plant at tfsur-1 locus, whereas TX9425 allele increased plant survival at the locus of tfsur-2. This may explain the strong transgressive segregation found for this trait.Identification of QTLs associated with waterlogging tolerance in Franklin/YerongTwo QTLs (yfy1.1-1 and yfy1.1-2) controlling leaf chlorosis after two-weeks of waterlogging stress (2004) were found on chromosome 2H and 5H. The Franklin alleles increased leaf chlorosis at the yfy1.1-1 locus, whereas at the yfy1.1-2 locus the Yerong allele increased leaf chlorosis (Table 4, Figure 2). Three QTLs (yfy2.1-1, yfy2.1-2 and yfy2.1-3) were found for leaf chlorosis after two weeks of waterlogging in the experiment carried out in 2005, these QTLs were located on chromosome 7H, 3H and 4H. The Franklin alleles increased leaf chlorosis in all three cases. Three QTLs (yfy2.2-1, yfy2.2-2 and yfy2.2-3) were found for leaf chlorosis after four weeks of waterlogging stress in the 2005 experiment, these QTLs were located on chromosome 3H, 1H and 4H. The Franklin allele increased leaf chlorosis at yf2.2-1 and yf2.2-3 loci, whereas the Yerong allele did so at the yf2.2-2 locus. QTL yfy2.2-1 is likely to be the same as yfy2.1-2 as it is in an identical position on chromosome 3H. The same applies to QTL yfy2.1-1 and yfy2.2-3 on chromosome 4H.Figure 2The Franklin/Yerong chromosomes showing the locations of QTLs for the traits analyzed. Each linkage group consists of a vertical bar on which the map positions and names of loci are indicated. QTL positions are shown through their support interval on the right of each chromosome. One LOD support intervals are the inner intervals, while the outer intervals represent the two LOD support intervals. Prefix \"bPb\" and \"p\" signify a DArT marker and a AFLP marker, respectively. The other markers on the map are microsatellites.Table 4Characteristics of the detected QTLs explaining waterlogging related traits in Franklin/Yerong population.TraitQTLLinkage groupsOne LOD support interval (cM)LOD scoreR2 (%)Leaf yellowingyfy1.1-12H46–552.905.8proportion 1.1 (twoyfy1.1-25H38–533.947.6weeks stress, 2004)Leaf yellowingyfy2.1-17H64–733.726.7proportion 2.1 (twoyfy2.1-23H42–526.4111.9weeks stress, 2005)yfy2.1-34H104–1129.2518.5Leaf yellowingyfy2.2-13H43–524.509.5proportion 2.2 (fouryfy2.2-21H53–682.775weeks stress, 2005)yfy2.2-34H104–11410.3722.4Reduction of plant biomassyfmas4H91–1203.038.2Plant survivalyfsur-12H34–613.157.1yfsur-25H42–585.0513.1One QTL (yfmas) was identified for the reduction of plant biomass following waterlogging in this population (Table 4). This QTL mapped on chromosome 4H to almost the same position as QTL yfy2.2-3 and yfy2.1-3 and is probably due to pleiotropy. This was supported by the significant correlation between leaf chlorosis and plant biomass reduction in this population (results not shown).Two QTLs (yfsur-1 and yfsur-2) were identified on chromosome 2H and 5H for plant survival rate after 8 weeks of continuous waterlogging stress. The Yerong allele increased plant survival rate at the yfsur-1 locus while the Franklin allele increased plant survival rate at the yfsur-2 locus. Yfsur-1 was located near yfy1.1-1 while yfsur-2 was located near yfy1.1-2 and again this may be because of pleiotropy.Comparison of waterlogging tolerance QTLs between populationsIn order to compare the QTLs identified in different populations, the markers flanking the one LOD support intervals for each QTL were relocated on the consensus map [35] including the two populations used in this study. Comparison of the identified QTLs between the two populations (Table 5; Figure 3) showed that many of the QTLs identified in Franklin/TX9425 mapped to similar chromosomal regions compared to those identified in Franklin/Yerong (such as QTLs identified on chromosome 3H and 7H), or mapped very close to one another with almost touching or overlapping two LOD support intervals (such as QTLs identified on chromosome 1 H, 2H and 4H) (Figure 3).Figure 3Comparison of quantitative trait loci (QTLs) identified for waterlogging tolerance in two different barley doubled haploid populations: tf = Franklin/TX9425; yf = Franklin/Yerong. Markers flanking the one LOD support interval of each QTL identified in the individual population were re-located on a barley composite map [35] so that their relative position could be compared. Centromeres are indicated as in [35]. A general name (such as Qwt1-1) was given to each chromosome region associated with waterlogging tolerance, the first number was the chromosome number and the second number was the serial number of regions identified on that chromosome.Table 5Comparison of QTLs identified in the two populations after the flanking markers for each QTL were placed on the barley consensus map.ChromosomeFranklin/YerongFranklin/TX9425QTLsChromosome interval (cM)Effect (%)QTLsChromosome interval (cM)Effect (%)1Hyfy2.2-249–665tfy1.1-368–737.12Hyfy1.1-147–565.8tfy1.1-173–8223.3yfsur-145–497.1tfsur-182–11519.1tfsur-226–3313.23Hyfy2.1-259–9011.9tfy1.1-263–8533.4yfy2.2-159–909.5tfy1.2-178–9736tfy2.1-163–6834.14Hyfy2.1-3114–13618.6tfmas80–11316.3yfy2.2-394–12322.4yfmas114–1368.25Hyfy1.1-295–987.6yfsur-286–9813.27Hyfy2.1-174–796.7tfy2.1-250–8316Discussion and conclusionLeaf chlorosis in green plants is a complex and highly regulated process that occurs as part of plant development or that can be prematurely induced by stress. Recent analysis of the signalling pathways involved with different stress responses has indicated that these have considerable cross-talk with senescence related gene expression [3]. In wheat, many of the studies on waterlogging tolerance have been based on leaf chlorosis or leaf/plant death [18,36,37]. Leaf chlorosis has been found to be highly negatively correlated with grain yield which was regarded as the final criterion for waterlogging tolerance in wheat [33]. In barley, Hamachi et al [21] found that screening for waterlogging tolerance by the amount of dead leaf was a useful criterion and that the tolerance was under polygenic control, while Setter et al [9] concluded that severity of leaf chlorosis was not a good criterion. However, our preliminary yield trials using the same genetic material as used in our crosses (unpublished data) showed that under waterlogging conditions, the yield reductions of Franklin (which also has high leaf chlorosis under waterlogging) and TX9425 (low leaf chlorosis under waterlogging) were 86% and 28% in a pot experiment and 61% and 39% in a controlled field experiment (data not shown). Since leaf chlorosis after waterlogging showed high heritability [12], this trait was used as the major criterion to test for waterlogging tolerance along with plant survival, and plant biomass reduction in the current study.The QTL analysis of two doubled haploid populations (Figure 3) found at least seven distinct QTLs for waterlogging tolerance. It was also demonstrated that some QTLs controlling leaf chlorosis were very stable and were validated under different stress duration, between different experiments and different populations (for example QTLs on chromosomes 1H, 3H and 7H). Some QTLs affected multiple waterlogging tolerance related traits, for example, the allele on chromosome 4H from the tolerant parent contributed not only to reducing barley leaf chlorosis, but also to increasing plant biomass under waterlogging stress, whereas other allelles such as those on chromosomes 2H and 5H controlled both leaf chlorosis and plant survival. This result suggested that leaf chlorosis is an important stable selection criterion for barley waterlogging tolerance, which can be used practically in breeding programs.Waterlogging tolerance is a complex trait affected by several mechanisms and complicated by confounding factors such as temperature, plant development stage, nutrient, soil type and sub-topography. The current experiment was conducted under well controlled environmental conditions. The soil, obtained from a waterlogged site in Tasmania, was well mixed before being evenly packed into pots. Waterlogging treatments were conducted in the early vegetative growth stage to avoid the effect of variation in development rate on waterlogging tolerance. As indicated in the Material and Methods, the parents of both populations differ in many developmental traits including ear emergence in both populations and plant height in the Franklin/TX9425 population. One major QTL located on chromosome 2H was found for plant height and ear emergence in the Franklin/TX9425 population and two major QTLs located on chromosomes 2H and 7H were found for ear emergence in the Franklin/Yerong population (data not shown). The locus controlling row type in the Franklin/Yerong population was located on chromosome 2H, which is in a similar position to that reported in other studies [38]. None of these loci were within the confidence intervals of the QTLs controlling waterlogging tolerance detected in the current study.Accuracy of QTL mapping is important in implementing marker-assisted selection (MAS) for polygenic traits, but small confidence intervals for QTL positions are not easily obtained [39,40], although typical approximate 95% confidence intervals for QTL positions are of the order of 20 cM [41,42]. Van Ooijen [40] recommended using a two LOD support interval as an approximation of the 95% confidence intervals. Using only the one LOD support interval in this study, we observed significant overlap in QTL positions across populations. The results of this study showed that one LOD support intervals around QTLs identified in the Franklin/Yerong population were smaller than those in the Franklin/TX9425 population, this is because the Franklin/Yerong population was larger and further reduction in size of confidence intervals will require the use of larger populations [43].There is only one published report of QTLs for waterlogging tolerance in barley. Qian et al [44] found one SSR marker (WMC1E8) correlated with waterlogging tolerance based on chlorophyll content of the second top leaf in an F2 population by constructing two DNA (tolerant and susceptible) bulks. The identified QTL explained 29.9% of the total variation [44], and the authors deduced that this QTL was located on chromosome 1H based on the published barley linkage maps [45]. In our study we identified QTLs controlling leaf chlorosis in both populations on chromosome 1H. However, the position of the QTLs found in our study were different from that of WMC1E8 reported by Qian et al [44] according to the consensus map [35].Different segregating populations of rice, maize, wheat, and barnyard grass have been studied for diverse waterlogging related characteristics or criteria, such as plant survival, leaf senescence, the extent of stimulation of shoot elongation caused by stress [46], waterlogged shoot growth and waterlogged root growth [47], adventitious root formation and leaf injury [48,49]. QTLs controlling many of these traits have been identified. Comparison of genetic mechanisms of waterlogging or flooding tolerance among different crops remains difficult because different waterlogging related traits were used for QTL analysis in these studies. Another difficulty for comparing QTLs identified for waterlogging tolerance in different species is the lack of common markers among different genetic linkage maps, sometimes even among different populations within the same species. Different marker nomenclature among researchers also contributes to the difficulties with comparative mapping.Despite these difficulties, comparative mapping across cereals can provide interesting information. For example, a major QTL controlling waterlogging tolerance based on dry matter production in maize was located on chromosome 1 [50]. In our experiment, a QTL controlling plant biomass under waterlogging stress was identified on chromosome 4H, which comparative mapping has shown to be highly homoeologous to chromosome 1 in maize [51,52]. QTLs controlling percent plant survival in rice under submergence stress were mapped to chromosome 7, 9 and 10, and the QTL located on chromosome 9 was the most significant one [46]. According to comparative mapping in the grass family, rice chromosome 9 had a homoeologous relationship with wheat chromosome 5L and maize chromosome 2 [51]. Maize chromosome 2 is in part homoeologous to wheat chromosome 2 [51], so it can be deduced that rice chromosome 9 is homoeologous with barley chromosome 2H and 5H [52]. In barley, the QTLs contributing to plant survival were located on chromosomes 2H and 5H. These QTLs identified for controlling plant survival could be the same as the QTL identified on chromosome 7 and 9 in rice.Improving waterlogging tolerance in barley is at an early stage compared with other traits. The future use of marker assisted selection (MAS) in combination with traditional field selection could significantly enhance barley breeding for waterlogging tolerance. As demonstrated in this study, and in other previously published studies [53], diversity array technology (DArT) is very efficient for whole-genome profiling [30]. Although this technique is still limited to only a few laboratories at this stage, barley consensus maps [35] have been constructed to link DArT markers with many SSR and RFLP markers which have been previously developed and applied widely in barley mapping studies and to provide plant breeders with practically useful molecular markers for improving barley waterlogging tolerance. DArT markers can easily be sequenced and to obtain stronger support for the microsynteny of the QTLs (or genes) for waterlogging tolerance among grass species, further research should involve direct comparison of DNA sequence of markers (those linked to QTLs) to that of the genome sequence of rice [54,55] and other species.Authors' contributionsHBL selected barley genotypes and made crosses between them for DH population construction, performed SSR and AFLP assays, prepared DNA samples for DArT assays, built the component maps and the composite map, screened waterlogging tolerance on the DH lines, conducted statistical analysis and QTL analysis, drafted the manuscript, figures and tables. REV supervised this project and provided technical guidance. NM supervised this project. MXZ constructed DH populations for Franklin/TX9425 and Franklin/Yerong crosses and supervised this project.\n\nREFERENCES:\nNo References"
+}

batch_5/PMC2535567.json

@@ -0,0 +1,4 @@
+{
+    "id": "PMC2535567",
+    "text": "This is an academic paper. This paper has corpus identifier PMC2535567\nAUTHORS: Lidia Ruiz, Magali Traskine, Irene Ferrer, Estrella Castro, Juan F. M. Leal, Marcelline Kaufman, Amancio Carnero\n\nABSTRACT:\nBackgroundP53 activation can trigger various outcomes, among them reversible growth arrest or cellular senescence. It is a live debate whether these outcomes are influenced by quantitative or qualitative mechanisms. Furthermore, the relative contribution of p53 to Ras-induced senescence is also matter of controversy.Methodology/Principal FindingsThis study compared situations in which different signals drove senescence with increasing levels of p53 activation. The study revealed that the levels of p53 activation do not determine the outcome of the response. This is further confirmed by the clustering of transcriptional patterns into two broad groups: p53-activated or p53-inactivated, i.e., growth and cellular arrest/senescence. Furthermore, while p53-dependent transcription decreases after 24 hrs in the presence of active p53, senescence continues. Maintaining cells in the arrested state for long periods does not switch reversible arrest to cellular senescence. Together, these data suggest that a Ras-dependent, p53-independent, second signal is necessary to induce senescence. This study tested whether PPP1CA (the catalytic subunit of PP1α), recently identified as contributing to Ras-induced senescence, might be this second signal. PPP1CA is induced by Ras; its inactivation inhibits Ras-induced senescence, presumably by inhibiting pRb dephosphorylation. Finally, PPP1CA seems to strongly co-localize with pRb only during senescence.ConclusionsThe levels of p53 activation do not determine the outcome of the response. Rather, p53 activity seems to act as a necessary but not sufficient condition for senescence to arise. Maintaining cells in the arrested state for long periods does not switch reversible arrest to cellular senescence. PPP1CA is induced by Ras; its inactivation inhibits Ras-induced senescence, presumably by inhibiting pRb dephosphorylation. Finally, PPP1CA seems to strongly co-localize with pRb only during senescence, suggesting that PP1α activation during senescence may be the second signal contributing to the irreversibility of the senescent phenotype.\n\nBODY:\nIntroductionAmong the different methods cells have to monitor external or internal stresses, the surveillance mechanism associated with the p53 gene is central. Numerous molecular studies over the years have presented p53 as an essential controller of cellular and genome integrity [1]. p53 is a master transcription factor, functionally inactive under normal conditions due to its rapid degradation by the ubiquitin ligase MDM2. A chain of events triggered in response to cellular stress upsets this precise balance, leading to the uncoupling of MDM2-driven degradation and to the ultimate accumulation and activation of p53 [2]. p53 works mostly as a transcriptional activator, with few molecules in each cell [3]. However, p53 might also act as a repressor in some instances [4]. The p53 transcriptional program includes the activation of a number of cell cycle inhibitors and proapoptotic proteins, which results in apoptosis, reversible proliferative arrest or cellular senescence [5], [6], [7].In principle, the various outcomes of p53 activation might be influenced by quantitative or qualitative mechanisms [8]. Some studies suggest that the level of p53 output determines whether cells will enter cell cycle arrest or apoptosis. Consistent with this view, only a subset of the genes induced by high p53 levels are induced by lower p53 levels [9]. Introduction of high p53 levels into tumor cell lines induces apoptosis, while the introduction of lower levels induces only cell cycle arrest [10]. However, other studies suggest that the outcome of p53 activation is determined by factors controlled by the tissue type or by the cell genotype.Oncogenic Ras can activate p53 to promote cellular senescence, limiting the transforming potential of excessive signalling [11]–[16]. This study and others have demonstrated that conditional activation of p53 in mouse embryonic fibroblast cells (MEFs) produces reversible cell cycle arrest, whereas activation of p53 in the presence of oncogenic Ras leads to a permanent cell cycle arrest with features of cellular senescence [17], [18]. Although oncogenic Ras may increase p53 levels, it is not clear whether this increase is sufficient to explain the induction of senescence.Two different, though not mutually exclusive, models have been proposed to explain the different biological outcomes associated with p53 activation. The quantitative model implies that p53 levels are sufficient to determine the outcome. Thus, low p53 levels induce a reversible cell cycle arrest while higher p53 levels induce senescence or apoptosis. This model is supported by studies in which p53 levels may be artificially controlled with the appropriate expression systems [9], [10]. One potential mechanism that could explain such an effect is based on differential p53 affinity for p53 response elements, such that genes required for a reversible cell cycle arrest have protein products with greater affinities than those required for senescence or apoptosis.A qualitative model of p53 action implies that non-quantitative factors controlled by a stimulus, either the tissue origin or the cell genotype influence the outcome of p53 activation. Again, two non-mutually exclusive mechanisms might support the published data. First, certain collateral signals might directly modulate p53 activity by changing the conformation of p53 or its association with various coactivators, perhaps leading to the expression of different subsets of p53 target genes. Consistent with this possibility, ionizing radiation and UV light have been shown to induce expression of different subsets of p53-dependent target genes in the same cell type [9]. Interestingly, these two stimuli induce different p53 modifications [19]–[21], raising the possibility that the activating signal may modulate p53 activity in a qualitative manner by directing p53 to different promoters [22]. Similarly, the ability of oncogenes to promote either apoptosis or senescence is correlated with different p53 modifications.Furthermore, oncogenic Ras induces p53 phosphorylation on serine 15 and induces senescence, whereas the E1A oncoprotein does not induce serine 15 phosphorylation and promotes apoptosis. The E1A effect is dominant, since cells coexpressing E1A and Ras do not contain p53 that has been phosphorylated on serine 15, and these cells are prone to apoptosis [23], [24]. Whether this effect leads to the expression of different p53 target genes has yet to be determined. Second, it is possible that the signal produced by p53 activation is the same in different contexts and that the outcome of p53 activation is determined by how this signal is interpreted by the cell. One may envision several mechanisms by which this might occur, but an obvious possibility involves the combined action of p53 and other transcription factors such that the action of p53 on outcome-specific targets is influenced by the presence or absence of these other factors. These other factors, in turn, would be the targets for the hypothetical collateral signal. One precedent for this involves the integration of p53 and interferon signaling on the p21 promoter, which contains both p53 and IRF-1 response elements that act to synergistically induce p21 expression during a DNA damage response [25]. How different signal transduction pathways integrate to produce new biological outcomes is an important biological problem that may also have an impact on the understanding of p53.How does oncogenic Ras convert p53 to a senescence inducer? Although it seems likely that a component of this response results from the ability of oncogenic Ras to produce quantitative increases in p53 activity via ARF-mediated inhibition of MDM2, there is compelling evidence for collateral signals that modify the outcome of p53 activation leading to senescence [17], [26]. Following the discussion above, it is formally possible that oncogenic Ras directly modulates p53 activity or, instead, produces cellular changes that reinterpret the p53 signal.One potential mechanism may involve the ability of Ras to induce PPP1CA (the catalytic subunit of PP1α) expression, regulating senescence in a pRb-dependent manner [18]. pRb is involved in the SAHF, maintaining long-term inhibition of E2F-dependent transcription through changes in the packaging status of chromatin [27].To characterize the p53 response during growth arrest and senescence, this series of experiments compares p53-dependent transcription in different situations involving proliferation, reversible arrest, replicative senescence or Ras-induced senescence.Resultsp53 levels and phenotypeTo assess if differences in p53-dependent transcription play a role in reversible arrest or senescence, this study took advantage of the mouse embryo fibroblast (MEF) cell system that allows easy manipulation of cellular stresses in otherwise homogeneous conditions. For instance, conditional activation of the p53 pathway in MEFs is known to trigger reversible cell cycle arrest, whereas activation of p53 in the presence of oncogenic Ras leads to permanent cell cycle arrest with features of replicative senescence (Figure 1A) [16].10.1371/journal.pone.0003230.g001Figure 1Experimental system.A) Molecular markers used to identify senescence. B) Scheme of the procedure (see M & M) and 2 representative images of the dot blot obtained after hybridization. C) Comparison of several conditions of well-known activated targets of p53. Western blot showed no variation among p53 levels under comparable conditions.To induce replicative senescence, wild-type and p53-null (p53 −/−) embryos were generated from crosses between heterozygous p53 knock-out mice. From wild-type embryos, MEFs were generated and grown until replicative senescence was reached (approximately at passage 5, corresponding to 10 population doublings). We extracted mRNA under these conditions, i.e., terminally arrested with senescence features (P5), and also from exponentially growing MEFs (early passage, P3). Other stress conditions leading to senescence were produced as follows. Wild-type MEFs growing at early passage were infected with retroviruses carrying oncogenic Ras (Val12-Hras). Cells were selected for retrovirus insertion and once they reached senescence (corresponding to approximately passage 3), mRNA was extracted (P3+ras). P53-null MEFs were infected with viruses carrying the 135V thermosensitive mutant of p53 that induces cellular arrest at permissive temperature (32°) [28]. These cells (p53ts), while maintained at restrictive temperature (39°C), were infected with viruses carrying oncogenic Ras (p53ts-ras), which induces senescence when shifted to permissive temperature [17]. For a summary of conditions and the resulting phenotypes see Table 1.10.1371/journal.pone.0003230.t001Table 1Summary of cell lines and conditions used in this study.Cell lineGenotypephenotype\np53−/−\np53−/− MEFsGrowth at 32° and 39°\np53−/−;ts (p53ts)p53−/− MEFs with p53val135Growth at 39° reversible arrest at 32°\np53−/−;ts-ras (p53ts-ras)p53−/− MEFs with p53val135Growth at 39°; senescence at 32°\np3\nnaïve MEFs, 6 population doublingsGrowth at 37°\np5\nnaïve MEFs, 10 population doublingssenescence at 37°\nP3+Ras (Ras)naïve MEFs+Ras-val12, 6 population doublingssenescence at 37°The abundance of p53 did not change among different passages reaching replicative senescence, or between the restrictive or permissive status in the case of the overexpression of the thermosensitive mutant of p53 (Figure 1C). Therefore, this study first measured broad p53-dependent transcription (Figure 1B). We measured the expression of 122 p53 target genes using Dot Blot arrays in the different proliferating and arrested cellular scenarios discussed above (See Figure S1 for a list of the 122 genes analyzed). The increased transcription rates of important p53 target genes such as Bax, GADD45, p21 and PIG8 confirm the activation of p53 in both senescence systems (Figure 1C).We observed that the arrest of MEFs at senescence (P5) and after Ras-induced senescence (P3+ras) correlated with a net increase in p53-dependent transcription (Figure 2A). Similarly, cells arrested after p53 activation (ts and ts-Ras at 32°C) also showed, as expected, a significant increase in p53-dependent transcription (Figure 2A). Therefore, we had a genetically homogeneous system with different levels of p53 activity measured with respect to 122 p53 transcriptional targets. It was possible to ascribe a phenotype to each level of p53 activity (Table 1). There were three conditions of proliferating cells: (1) P3, (2) cells with mutant p53 at restrictive temperature (null p53), and (3) cells showing basal levels of p53 activity. There was also one condition of replicative senescence with a moderate increase of p53 activity (P5). Oncogenic Ras activation seems to induce higher levels of p53 activation with similar senescent phenotypes (P3+ras). However, elevated levels of p53 do not always induce senescence as p53ts cells at permissive temperature are reversibly arrested, but p53 activity is higher than in the two previous conditions displaying senescence. As before, oncogenic Ras expression switches the cell from arrest to senescence, also increasing the relative p53-dependent transcription (Table 1 and Figure 2A).10.1371/journal.pone.0003230.g002Figure 2Analysis of the p53-dependent transcriptional signature.A) Comparison of the levels of 122 transcriptional targets of p53 at different conditions of p53 activity. See text. Data normalized against β-actin were compared to the proliferative condition P3 to evaluate its statistical significance. Statistical analysis was performed by paired t-test. (*) = p<0.05; (**) = p<0.005; (***) = p<0.001. B) and C) Analysis of the expression values of 122 transcriptional targets of p53 under different cellular conditions, which led either to proliferation or to growth arrest. Clustering analysis. Hierarchical clustering was performed using the function hcluster (package amap) of the free statistical software R. See M & M. The expression level of each gene, relative to its median expression level across all conditions, was represented by a color, with red representing expression greater than the median, green representing expression less than the median, and the color intensity representing the magnitude of the deviation from the median. D) Feature selection. Since the number of genes is much greater than the number of conditions, we used penalized regression methods. See text and M & M for more details. E) Validation of the feature selection by quantitative PCR. See text and M&M. We determined Lats2, DKK1, pRb and PTEN mRNA levels after 24 hts treatment of HCT116 cells with the indicated treatment, by quantitative PCR. Cyclophilin (ref. 4326316E), an endogenous control, was used to normalize variations in cDNA quantities from different samples. Each reaction was performed in triplicate with cDNA from normal and tumor tissue from each patient studied. C shows untreated samples. E: Etoposide, D: Doxorubicin, T:Taxol, U: UCN-01, F:Flavopiridol. Data shows average of three determinations.Therefore, arrest vs. senescence is not determined by the relative levels of p53 activity alone.Specificity of the senescence responseTo study the expression pattern of p53-responsive genes during arrest or senescence in order to compare both processes and to ascertain what gene or genes may play a crucial role in the proliferating or arrested cell phenotypes, we performed a hierarchical clustering of the different cellular conditions on the basis of pattern similarity (see Materials and Methods). In Figures 2B and 2C we observed that the conditions are separated into two groups corresponding to the arrested (right side) and proliferating (left side) phenotypes. The cell lines are grouped together on the cluster dendogram by the activation or inactivation of p53 and not by the presence or absence of the Ras oncogene. This is clear in wild-type MEFs growing at passage 3 (P3), which have low levels of p53 activation compared to arrested wild-type MEFs in passage 5 (P5), which have p53 highly activated. However, it is interesting that the most extreme condition, p53 activation in the presence of oncogenic Ras, triggers an enhanced transcriptional response (Figures 2A and 2C, lane p53ts-Ras [32°]). See below.Although all the physiological conditions that lead to growth-arrest onset are clustered together and all the transcripts considered are p53-dependent, it is clear that there are some genes whose enhanced activation (relative to their median expression level over all cellular conditions) is specific to each particular condition (Figure 2C and Table 2). These genes might serve as specific marker genes. However, no concurrent senescence signature could be observed, indicating that the senescence program is not determined by the specificity of the p53 response.10.1371/journal.pone.0003230.t002Table 2Condition-specific genes.P5 ReplicativeRas Oncogenic Stressp53ts(32°) Growth arrestIGF-Rp63GMLMAP4CycB1BakZAP70Krt2-8BaxWig1Krt1-15PTGFPIG8Pmaip1Igfbp6IL6DKK1Bcl6P73PUMAPPM1DLats2MgmtTyrBaxPold1MDR1JunLats2Thbs1LRDD/PIDDKai1PthlhBtg2Waf1MAP4Igfbp3TstMST1RB1RB1IGFRHic1p14-ARFThis table represents the genes that are the most representative (relative to their median level–based on all conditions) in each particular arrested condition. A threshold equal to 2.30 was chosen for the ratio. Genes are arranged from the lowest to highest ratio.Next, applying a penalized least-squares regression technique with an L1-type penalty to the expression data (see Materials and Methods) it was possible to identify four p53 target genes among the 122 genes studied as the most relevant markers for predicting the proliferating or arrested phenotype of each cellular condition. These four relevant genes are: MAP4, PTEN, Lats2 and Rb1 (Figure 2D). Furthermore, combining L1- and L2-norm penalties allowed small subgroups of additional genes that are highly correlated with the main predictors to be extracted. This study identified five more genes closely related to MAP4 behavior: p63, caspase1, DKK1, Bcl2 and Gtse1; as well as LRDD, related to Rb1. This robust set of p53 target genes molecularly defines a minimal footprint to identify a p53-dependent arrest.In order to confirm the p53-dependent arrest footprint defined by these markers, we measured the p53-dependent transactivation of 4 among the selected genes by qRT-PCR in HCT 116 p53+/+ cells treated with different DNA-damaging agents. p53 protein is present at low levels in resting cells but after exposure to those agents as well as to other stressing stimuli, it is stabilized and activated by a series of post-translational modifications. These modifications leave p53 free from mdm2, an E3 ubiquitin ligase that ubiquitinates it and facilitates its degradation by the proteasome [5]. p53 stabilization and activation is followed by cell-cycle arrest. To ascertain whether the transcription of this set of genes also depends on other chemotherapeutic drugs that act through p53-independent mechanisms, we also treated the cells with compounds that do not directly cause DNA breaks. Only the treatment with the topoisomerase inhibitors Etoposide and Doxorubicin induced an activation of the transcription of PTEN, Lats2, Rb1 and MAP4 (Figure 2E). However, we did not detect increase of these genes by Taxol, flavopiridol or UCN-01.Downregulation of p53 response without senescenceP53 transcription seems to define only arrest, and not senescence, suggesting the existence of a p53-independent signal necessary to convert the reversible arrest into senescence. To explore this, we analyzed whether sustained p53 activation might induce senescence without a second signal. In the same p53-induced transcriptional setting, we analyzed the activation of p53 during long periods and its correlation with the appearance of senescence. After an initial activation, general p53-induced transcription seems to decay at 24 hrs; this downregulation is maintained for long periods even in the presence of Ras activation (Figure 3A). However, senescence features are only maintained in the p53ts-Ras cells incubated at 32°C (Figure 3B). We found that in cells carrying activated p53 only, senescence is not induced despite a long period of growth arrest (up to 6 days). These data support the finding of Ferbeyre et al. [17], that growth arrest and senescence are two independent phenotypes; the permanence of growth arrest does not induce senescence unless another signal is involved.10.1371/journal.pone.0003230.g003Figure 3P53 activity is downregulated maintaining senescence.Cells were plated in 10-cm-diameter plates. Cells were grown at 39°C (i.e., never incubated at 32°C) or arrested for the indicated times at 32°C. Cells were harvested and RNA collected for A, or stained for SA β-GAL for B. A) Comparison of the levels of 122 transcriptional targets of p53 at different times after p53 activation. Data normalized against β-actin was compared to the proliferative conditions at 39°C to evaluate statistical significance. Statistical analysis was performed by paired t-test. (*) = p<0.05; (**) = p<0.005; (***) = p<0.001. B) More than 400 cells were visually analyzed for SA β-GAL staining as described in Figure 1A. Data represent the percentage of cells showing SA β-GAL staining.Finally, to confirm that the cells in long-term arrest have not suffered molecular changes that might indicate a switch to senescence, we analyzed 53BP1 and γH2AX phosphorylation at the senescence-associated DNA foci. As before, p53ts and p53ts-Ras cells were cultured at 39°C, then were moved to 32°C and maintained for up to 6 days at restrictive temperature. Cells were taken at different time points and analyzed for the presence of DNA-damage foci labeled by 53BP1 and γH2AX phosphorylation as markers for cellular senescence (Figure 4). One or two 53BP1 and γH2AX foci appear with cell proliferation, and the same number of foci was maintained in p53ts arrested cells even after 6 days of growth arrest. However, p53ts-Ras cells showed a strong increase in the number of foci per nuclei after 48 hrs of arrest (Figure 4); this was maintained despite p53-transcriptional downregulation.10.1371/journal.pone.0003230.g004Figure 4Enforced growth arrest does not induce senescence.P53ts or p53ts-Ras cells were grown at 39°C or incubated at 32°C for different times as indicated. Cells were fixed and stained with DAPI to identify the nuclei, or with antibodies against 53BP1 or phosphorylated gH2AH. A) Representative picture. B) Foci of >60 nuclei of each condition were counted and data represented as the average of the number of foci per nuclei. Bars = StDev.These data, which are consistent with previous observations [17], [26], indicate that initial p53 activation is required to induce growth arrest. However, a second Ras-dependent signal seems to be required to stabilize the arrest as irreversible senescence.PPP1CA contributes to growth arrest stabilization in senescenceThe application of a retroviral-based genetic screen yielded an antisense RNA fragment against PPP1CA, the catalytic subunit of PP1α. Loss of PPP1CA function bypasses Ras/p53-induced growth arrest and senescence [18]. It was found that oncogenic Ras promotes an increase in the intracellular level of ceramides, which may increase PPP1CA activity, contributing to senescence. PP1α has been identified as the protein phosphatase responsible for the dephosphorylation of pRb [29]; this has been related to the growth arrest response [30]–[32]. When cells are actively growing, the hyperphosphorylated form of the Rb protein (ppRb) predominates. On the contrary, when cells are delayed in their growth, the hypophosphorylated form of the Rb protein (pRb) is the most abundant. Thus, enforced pRb dephosphorylation might contribute to the arrest to senescence transition [27], [33].PPP1CA protein levels increase upon Ras activation (Figure 5A) [34], [18], but not mRNA (Figure 5B). PP1 phosphatase activity also increases upon oncogenic ras expression (Figure 5C), paralleling protein levels of PPP1CA. Expression of a specific shRNA against PPP1CA impairs pRb dephosphorylation, thus bypassing p53-induced arrest. When p53ts-Ras cells were shifted at 32°C, pRb became hypophosphorylated, in accordance with the growth-arrest induced by thermo-sensitive p53 at this permissive temperature. In contrast, p53ts-Ras cells stably transduced with shRNA against PPP1CA showed an increase in the hyperphosphorylated form of Rb protein when kept for 24 h at 32°C (Figure 5D). These data show that downregulation of PPP1CA maintains pRb in the hyperphosphorylated state, even in the presence of active p53, therefore allowing cell growth (Figures 5D and 5E). While p53ts-Ras cells at 32°C show mostly the senescent phenotype, only 22% of cells carrying the PPP1CA shRNA showed senescence features, confirming the relevance of PP1α activity to the senescence phenotype (Figure 5F). This was further confirmed by immunofluorescence studies (Figure 6). In proliferating cells, PPP1CA and pRb levels are low, increasing slightly upon growth arrest. However, these proteins showed diffuse distribution (Figure 6). Under conditions inducing senescence (p53ts-Ras at 32°C), cells increase pRb and PPP1CA levels, which showed nuclear co-localization, strengthening evidence for their functional relationship to senescence.10.1371/journal.pone.0003230.g005Figure 5A)Oncogenic Ras increased PPP1CA protein levels. P53ts (control) or p53ts-Ras (+ras) cells were grown at 39°C or incubated at 32°C for 24 hrs. Then PPP1CA protein levels were analyzed by western Blot. α-tubulin was used as a loading control. The data are representative of three independent experiments. B) PPP1CA mRNA levels were not dependent on the expression of oncogenic ras. mRNA levels were analyzed by Northern blot. A labeled probe able to specifically recognize PPP1CA isoform was used as described in M&M. C) Oncogenic ras increased PP1 activity. Exponentially growing cells were keep growing or switch for 24 hrs at 32°C as indicated. Then were starved and PP1 phosphatase activity was measured as described in M&M. C shows the remaining activity after 100 nM okadaic acid treatment to inhibit PP1 and PP2A activity. D, E, and F) P53ts-Ras cells carrying the shRNA against PPP1CA (shRNA) or vector alone (control) were grown at restricted temperature (39°C), or permissive temperature (32°C) as indicated, for 24 hrs. Cells were harvested for protein extraction (for D), fixed and stained with crystal violet (for E) or for SA β-GAL (for F). D) Downregulation of PPP1CA inhibits p53-induced pRb hypophosphorylation. Cells were processed for western blot, showing hyperphosphorylated (ppRb) and hypophosphorylated (pRb) forms of the protein. α-tubulin was used as a loading control. The data are representative of three independent experiments. Bottom panel shows quantification of pRb bands. E) and F) Downregulation of PPP1CA bypasses p53/ras-induced senescence. Cells (104) were seeded and grown at 39° or 32°C for 1 week, then fixed and stained for colony formation with crystal violet (E) or SA β-GAL (F). In F, numbers show the percentage of cells with SA β-GAL staining.10.1371/journal.pone.0003230.g006Figure 6PPP1CA and pRB co-localize during p53/ras-induced senescence.P53ts or p53ts-Ras were grown at 39°C or incubated at 32°C for 24 hrs. Cells were fixed and labeled with DAPI to identify the nuclei, as well as antibodies against PPP1CA (red) or pRb (green).Oncogenic stress enhances p53-dependent transcriptionWe also observed that oncogenic Ras enhances p53-dependent transcription (Figures 2A and 2C). To study this effect in detail, we selected three different p53-responsive promoters, Bax, p21waf1 and the synthetic p53 response element (x13). We engineered a construct fusing the different promoters 5′ of the luciferase reporter gene and compared the effect of p53 alone to the effect of the combination of p53 and Ras-val12 (Figure 7A). Oncogenic Ras enhances p53-dependent transcription in all cases, but does not alter transcription when transfected alone (Figure 7A). These effects are dependent on p53 and Ras doses (Figure S2).10.1371/journal.pone.0003230.g007Figure 7Oncogenic Ras enhances p53-transcriptional activation. A, B and C)p53(−/−), p53(−/−); MDM2(−/−) or p19(−/−) MEFs were transfected as indicated with plasmids carrying luciferase and the indicated genes. Luciferase activity was measured as indicated in the M & M. A) Assays were performed in p53(−/−) MEFs using different promoters responding to p53 (p21waf1, Bax and synthetic p53x13 carrying only the p53 binding site repeated 13 times). B) Only the Bax promoter was used in p53(−/−) MEFs. C) Only the Bax promoter was used in p53(−/−), double p53(−/−); MDM2(−/−) or p19(−/−) MEFs. Data show the averages of three independent experiments.To further study this effect, we selected the Bax promoter and investigated dependence of the phenomenon on Ras. To that end, we tested the N17 mutant of Hras-val12. This mutant lacks the ability to bind to Ras effectors and therefore acts as a dominant negative mutant. The N17 mutant does not alter the p53 response (Figure 7A), indicating that the Ras effect is dependent upon activation of Ras effectors. To directly discriminate between the two main effector pathways involved in this effect, the same experiment was performed with active PI3K or Raf pathway mutants. We co-transfected p53 and an active mutant of AKT (AKT-CA) (PI3K pathway), or an active mutant of Raf (BXB-Raf-CAAX). We were able to reproduce the ras-enhancing effect (Figure 7B), indicating that a strong activation of either pathway may provoke the enhancement of p53 transcription.Ras, acting through the Raf pathway, may activate p53 through p19ARF, either dependent upon or independently of MDM2, while PI3K may inhibit p53 through MDM2 phosphorylation [7], [35], [36]. To determine whether MDM2 or p19 was involved in the effect, the experiment was performed in p19-null or MDM2-null cells (Figure 7C). We observed that the p53-enhancing effect observed in the Ras oncogenic signal was dependent upon p19ARF but not on MDM2. A similar observation was made with the activated Raf oncogene. However, activated AKT showed p53-enhancing effects independent of p19ARF and MDM2 (Figure 7C). These data match those previously reported [11], [37]–[39]; Ras and Raf oncogenes require the p19ARF protein to activate p53.DiscussionThe data presented in this study elucidate the regulation of p53-responsive genes during proliferation and senescence. We have clearly demonstrated that Ras effects on p53-dependent transcriptional activation result in quantitative rather than qualitative changes. Therefore, the senescence response depends on factors other than p53 activation. p53 activation seems to be necessary but not sufficient to induce senescence, as other signals may be needed for the full onset of senescence. We have shown that Ras-induced activation of PPP1CA, the catalytic subunit of PP1α, is necessary to induce Ras-dependent senescence [18]. It is therefore possible to split the senescence response into two physiological processes. The first of these involves induction of growth arrest and is dependent on p53 activation or other physiological signals activating a proliferative brake similar to that of p53, such as p73 or p63. The second process occurs later, acting on pRb to stabilize its active unphosphorylated form, independent of p53. Unphosphorylated pRb will bind and inactivate E2F factors blocking cell cycle progression and altering local chromatin [27]. PPP1CA activation will take part in this second process, contributing to irreversible proliferative arrest by enforcing pRb dephosphorylation.Since senescence is a safeguard mechanism that may prevent pretumoral cells from further expansion, many studies have recently emphasized the relevance of this possible new therapeutic tool against cancer (reviewed in [41]–[43]). Our work has identified a set of p53 target genes that affect growth arrest in response to p53 activation. Although our work only identify these 4 genes as the minimal footprint to differentiate growing from p53-arrested cells, these 4 genes have been broadly studied and its relevance in growth arrest and senescence has been established.Tumor suppressor Lats2 has been shown to be necessary for culture-induced replicative senescence in MEFs, since Lats2−/− MEFs bypass this process [48]. Furthermore, cells lacking Lats2 showed increased prevalence of micronuclei, chromosomal defects and aneuploidy [48], [49]. Lats2 and p53 establish a positive feedback loop that prevents tetraploidization of cells treated with the microtubule poison nocodazole [49]. Most important, miRNA-372 and miRNA373 microRNAs directly target Lats2 expression and have been shown to cooperate with oncogenic Val12-Ras in a way that resembles p53 inactivation, acting as oncogenes in testicular germ cell tumors [50]. Finally, Lats2 has been shown down-regulated through promoter hypermethylation [51], [52], in association with poor prognosis human breast cancers and acute lymphoblastic leukemia. Lats2 might have a role against cancer development, probably through the induction of senescence, and this could explain the link between its down-regulation and tumoral progression. The tumor-suppressor gene RB1 can suppress S phase entry and cause a transient G1 arrest following DNA damage [53]–[55] and the mutations in Rb1 pathway-related genes are associated with poor prognosis in many tumor types. The PTEN/PI3K pathway is also regarded as an effector of cellular senescence [56] through p27kip1 cell cycle inhibitor activation.The key findings obtained in this study may contribute to the current understanding of the molecular basis of senescence and should be of great interest in future senescence studies.Materials and MethodsCell CulturePrimary MEFs from p53−/− mice were derived from day 13.5 embryos. Cells expressing murine p53val135 were generated by retrovirus-mediated gene transfer of p53val135 into p53−/− MEFs (p53−/− ts). Cells expressing Val12-Ras were generated by retrovirus-mediated gene transfer of pWZLBlast hVal12-Ras into wild-type MEFs at passage 3 (P3-Ras) and p53−/− ts (p53−/− ts Ras cells). Cells were cultured in Dulbecco's Modified Eagle's medium (GIBCO) supplemented with 10% fetal bovine serum (Sigma), 1% penicillin G- streptomycin sulfate (GIBCO), 0.5% fungizone-amphotericin B (GIBCO) and 5 µg/ml plasmocin (InvivoGen).P53−/− MEFs, p53−/− MDM2−/− MEFs and p19−/− MEFs were cultured in Dulbeccós Modified Eaglés medium (GIBCO). All media was supplemented with 10% fetal bovine serum (Sigma), 1% penicillin G-streptomycin sulphate (GIBCO), and 0.5% fungizone-amphotericin B (GIBCO) in a humidified CO2 incubator at 37°C.Retroviral Vectors and Gene TransferThe following retroviral vectors were used: p53val135 mutant cDNA in pWZLHygro and pWZLBlast hVal12-Ras. Retrovirus-mediated gene transfer was performed as previously described [44]. Briefly, 5×106 LinXE retrovirus producer cells were plated in a 10 cm dish, incubated for 24 h and then transfected by calcium-phosphate precipitation with 20 µg of retroviral plasmid (16 h at 37°C). The medium was changed and the plates were maintained at 32°C for 48 h to increase viral stability. Virus-containing supernatant was filtered through a 0.45 µm filter and supplemented with 8 µg/ml polybrene (Sigma) and an equal volume of fresh medium. Prior to infection, 8×105 target fibroblasts were plated per 10 cm dish and incubated for 24 h. For infections, the culture medium was replaced by the viral supernatant, and then the culture plates were centrifuged (1 h at 1,500 rpm) and incubated at 37°C for 16 h. The medium was changed and cells were split 24 h later. Infected cell populations were selected in hygromycin (20 µg/ml) for pWZLHygro-based vectors and in blasticidin (2 µg/ml) for pWZLBlast-based vectors.Northern assaysTotal RNA was isolated from cells using the TRI-REAGENT method (Molecular Research Center, Cincinnati, OH) according to the manufactureŕs instructions. A reverse transcription was done for each sample (20 µg of total RNA) using MMLV reverse transcriptase (Promega), oligo dT primer and dCTP32-labeling nucleotide.The cDNA 32P-labeled probes were hybridized to the p53 target gene array membrane (TranSignal, Panomics, CA, USA) at 42°C overnight. After removing excess substrate by gently washing twice with 2×SSC+0.5% SDS and 0.1×SSC+0.5% SDS at 62°C, the membranes were exposed to BioMax Films (Eastman Kodak Company, NY, USA). The assay normalization was done selecting β-actin as the control housekeeping gene. Analysis was done using the GS-800 Calibrated Densitometer® and the Quantity One® program from Bio-Rad.Each experiment for each condition was performed independently at least twice, the data quantified and normalized for the value of β-actin (a gene with transcription that is independent of p53).Raw data for all conditions were normalized against an internal control, β-actin, and then compared to normal proliferating MEFs.PPP1CA Northern BlotTotal RNA was extracted using RNAzolB. 10 µg of total RNA were run in formaldehyde-agarose gels and transferred to a Hybond membranes. The membrane was pre-hybridized during 4 hours at 65°C. The probe was labeled by PCR with 50 µC of redivue dCTP32 (Amersham), using specific primers for mouse PPP1CA. The purified probe was denatured and added to the hybridization solution. The hybridization was performed overnight at 65°C. After extensive whashing, the membrane was exposed to a Biomax MS film (Kodak).Data AnalysisThe data consisted of the expression values of 122 transcriptional targets of p53 in different cellular conditions, which led either to proliferation or to growth arrest.Clustering analysisHierarchical clustering was performed using the function hcluster (package amap) of the free statistical software R (Ihaka and Gentleman, 1996). Before statistical analysis, gene expression levels were standardized gene by gene across all conditions using the median and interquartile range (IQR). The cellular conditions were clustered using Ward linkage and uncentered Pearson metric tests. The results were visualized and analyzed with TreeView (M. Eisen; http://www.microarrays.org/software). The expression level of each gene, relative to its median expression level across all conditions, was represented by a color, with red representing expression greater than the median, green representing expression less than the median, and color intensity representing the magnitude of the deviation from the median.Feature selectionThe problem of extracting a robust set of predictors for the proliferating status of the different cellular conditions has been formulated as a least squares regression problem. Since the number of genes is much larger than the number of conditions, we used penalized regression methods. The standard penalty used in so-called ridge regression is given by the L2-norm of the vector containing the regression coefficients. Such penalty allows stabilizing the ordinary least squares estimate, but typically will retain all regression coefficients so that no selection of the relevant variables (genes) may be done. To perform the selection task, we used an L1-norm penalty, as is done in lasso regression. This type of penalty is known indeed to promote sparsity, i.e., to force many regression coefficients to be zero; this obviates the need for pre-selection of the data. However, a known drawback of the L1 penalty for variable selection is that in a group of highly correlated genes, it may pick up only one representative. We therefore also used combined L1- and L2-norm penalties to select sparse groups of highly correlated genes; this is done in the so-called “elastic net” proposed in Zou and Hastie, 2005, [46]. To compute the corresponding penalized least-squares solutions, we applied the iterative thresholding algorithm developed in Daubechies et al, 2004. [47], which is simple to implement, robust to measurement errors and works well for high-dimensional data. Despite the small number of conditions, some standard validation tests (such as leave-one-out, label and gene permutation, bootstrap sampling) were performed.Transcriptional AssaysFor transient transfection of cells, we seeded 2–4×105 H1299 cells per well in six-well plates. After 24 h, transfections were performed by the calcium chloride method and JetPEI reagent (Polytransfection, Illkirch, France) according to the manufacturer's recommendations. For both transfection methods, we used 1.5–2 µg each of the reporter plasmids pGL3-13X, pGL3-Bax and pGL3-p21 in the presence or absence of pBABE puro p53 wt (0.6–0.75 µg) and pLSXN Ras val 12, or active mutants of the PI3K or Raf pathway (0.6–0.75 µg).Renilla luciferase plasmid was used as an internal control for transfection efficiency. The total amount of DNA within the experiments was kept constant by adding empty vector plasmid DNA to the transfection mixtures.Reporter gene assays were performed with the Dual-Luciferase® Reporter Assay System (Promega, USA) 48 h after transfection and the results were measured with a Victor2V luminometer. The activity of the reporter luciferase was expressed relative to the activity in renilla vector-transfected cells. Similar results were obtained in at least three different experiments. All results were compared to the control and are shown in the figures as the mean±S.D. of independent triplicate cultures.Western BlotCells were prepared in lysis-buffer and proteins were separated on SDS-PAGE gels, transferred onto PVDF membranes (Immobilon-P, Millipore) and immunostained. The following primary antibody was used: anti-p53FL393 (Santa Cruz 6243, diluted 1∶1000), anti-PP1α (from Calbiochem) anti-Rb: G3-245 from BD Pharmingen; and horseradish peroxidase-labeled rabbit anti-mouse (Promega diluted 1∶5000) and goat anti-rabbit (Calbiochem 401315, diluted 1∶4000) secondary antibodies. Proteins were visualized using the ECL detection system (Amersham Biosciences, Buckinghamshire, UK).ImmunofluorescenceImmunostaining and confocal analysis for 53BP1 and γH2AX fociCells were seeded onto glass cover slips and cultured for 8 h at 39°C. Then we placed the cells at 39°C and 32°C. After 24 h (cells at 39°C and 32°C) and 48 h, 96 h and 144 h (cells at 32°C), cover slips were fixed in 4% paraformaldehyde for 5 min at room temperature, washed twice with PBS, permeabilized in Triton X-100 0.5% in PBS for 5 min and washed twice more with PBS. Samples were incubated in blocking solution (PBS containing 3% bovine serum albumin) at 37°C for 15 min, followed by incubation for 30 min at 37°C with anti-phospho-Histone H2A.X (Ser139) antibody (Millipore 05-636) or anti-53BP1 antibody (Novus Biologicals NB100-304) diluted 1∶100. After washing with PBS, cells were incubated with species-specific Alexa 488-conjugated secondary antibody diluted 1∶100 in blocking buffer for 30 min at 37°C in the dark. The nuclei were stained with Hoechst 33258 diluted 1∶1000 for 3 min at room temperature in the dark prior to mounting with mowiol (Calbiochem). Images were collected by confocal laser microscopy (model TCS-SP2-AOBS, Leica, Germany).Immunostaining and confocal analysis for PPP1CA and pRb co-localizationCells were seeded onto glass cover slips and cultured for 8 h at 39°C. Then we placed the cells at 39°C and 32°C. After 24 h (cells at 39°C and 32°C) and 48 h (cells at 32°C), cover slips were fixed in 4% paraformaldehyde for 5 min at room temperature, washed 2 times with PBS, permeabilized in Triton X-100 0.5% in PBS for 5 min and washed again 2 times with PBS. Samples were incubated in blocking solution (PBS containing 3% bovine serum albumin) at 37°C for 15 min, followed by incubation for 30 min at 37°C with anti-human Retinoblastoma Protein (RB) monoclonal antibody (BD Pharmingen 554136) diluted 1∶100. After washing with PBS, cells were incubated with species-specific Alexa 488-conjugated secondary antibody diluted 1∶100 in blocking buffer for 30 min at 37°C in the dark. Then, cells were incubated with anti-Protein Phosphatase 1α, C-terminal antibody (Calbiochem 539517) diluted 1∶100. After washing with PBS, cells were incubated with species-specific Alexa 633-conjugated secondary antibody diluted 1∶100 in blocking buffer for 30 min at 37°C in the dark. The nuclei were stained with Hoechst 33258 diluted 1∶1000 for 3 min at room temperature in the dark prior to mounting with mowiol (Calbiochem). Images were collected by confocal laser microscopy (model TCS-SP2-AOBS, Leica, Germany).SA ß-Gal activitySenescence-associated (SA) ß-galactosidase (ß-Gal) activity was measured as previously described [45], except that cells were incubated in 5-bromo-4-chloro-3-indolyl-ß-D-galactopyranoside (XGal) at pH 5.5 to increase the sensitivity of the assay in MEFs. The percentage of cells expressing SA ß-Gal was quantified by inspecting >400 cells per 10-cm-diameter plate three times.Protein phosphatase assaysPP1 activity was determined according to standard procedures as previously described [57]. PP activity was assayed using 32P-labeled phosphorylase a as a substrate which detects both PP1 and PP2A activities. To selectively measure PP1 activity we used 2 nM okadaic acid to selectively inhibit PP2A. The cell pellet was homogenized in the extraction buffer containing 20 mM Tris-HCI, pH 7.5, 5 mM EDTA, 10 mM EGTA, 15 mM -mercaptoethanol, 0.25 M sucrose, 0.3% Triton X-100, 5 µg/ml leupeptin, and 5 µg/ml aprotinin and centrifuged to give a soluble supernatant. The PP activity in the clear supernatant was determined by measuring the trichloroacetic acid-soluble counts released after incubation of the 32P-labeled phosphorylase a in the cell extract. The PP activity was linear up to assay times of 10 min and 5 µg protein of the cell extract. Routinely, incubation for PP activity was carried out for 10 min with an extract containing 5 µg of protein as determined by the Bio-Rad assay (Bio-Rad, Hercules, CA). Negative controls were obtained incubating with 100 nM Okadaic acid to inhibit PP1 and PP2A activity. One unit (U) of activity is defined as the amount that catalyzes the release of 1 nmol Pi from phosphorylase a per min at 30°C.Real time PCR (qRT-PCR) experimentsTotal RNA were isolated form HCT 116 p53 +/+ cells (a generous gift from B. Vogelstein) treated with 400nM and 1 microM Etoposide, 0.6 µg/ml Doxorubicin, 10nM Paclitaxel (Taxol), 100nM UCN-01, 15 µM PD98059 for 8 hours. After DNAse treatment, reverse transcription was performed with 20 µg of mRNA using MMLV reverse transcriptase (Promega) and oligo dT primer according to the manufacturer's recommendationsQRT- PCR experiments were carried out using SYBR® Green PCR Master Mix (Applied Biosystems, USA). Reaction mixtures contained: 5 µl cDNA sample (1/10 dilution RT product), 1.5 µl primer mix (sense and antisense, 0.6 µM final concentration), and 12.5 µl SYBR® Green PCR Master Mix. The final volume should be 25 µl. The following primers were used to amplify regions: LATS2 forward 5′- AACAGCCTCAACGTGGACCTGTATGAA-3′ and reverse 5′-CAGGGCATGCTCCTCCTTGGCGTCGAA- 3′; PTEN forward 5′-CAGAAAGACTTGAAGGCGTAT-3′ and reverse 5′- GTAACGGCTGAGGGAACT C-3′; RB1 forward 5′-TCTGCATTGGTGCTAAAAGTTTCTTGGA-3′ and reverse 5′-CCTGTTCTGACCTCGCCTGGGTGTTCGA- 3′; MAP4 forward 5′-TGATCCCTTTAAGATGTACCATGATGAT-3′and reverse 5′-AATGCTTGTGCTGGTGGCCTCTCTTCTG-3′ and β-actine forward 5 -AGGCCAACCGCGAGAAGATGAC-3 and reverse 5 -GAAGTCCAGGGCGACGTAGCA-3′. The samples were amplified according to the following protocol: 10 min 95°C, 50 cycles: 15 sec 95°C, 30 sec 56°C–62°C (depending on the primer), 1 min 72°C. Then in order to get the dissociation curve, a stage was added: 15 sec 95°C, 15 sec 60°C and 15 sec 95°C.The normalized values were analyzed using SDS2.2.2 program (Applied Biosystems, USA). All samples were measured in duplicates and the right formation of the products was verified by 1% agarose gel electrophoresis (data not shown).Supporting InformationFigure S1List of 122 p53 target genes used int his study(0.11 MB EMF)Click here for additional data file.Figure S2Oncogenic ras increases p53-induced transcription in a dose-dependent maner(0.07 MB TIF)Click here for additional data file.\n\nREFERENCES:\n1. EfeyanASerranoM\n2007\np53: guardian of the genome and policeman of the oncogenes.\nCell Cycle\n6\n1006\n1010\n17457049\n2. LavinMFGuevenN\n2006\nThe complexity of p53 stabilization and activation.\nCell Death Differ\n13\n941\n950\n16601750\n3. NakanoHYonekawaHShinoharaK\n2007\nThreshold level of p53 required for the induction of apoptosis in X-irradiated MOLT-4 cells.\nInt J Radiat Oncol Biol Phys\n68\n883\n891\n17544001\n4. MaedaYHwang-VersluesWWWeiGFukazawaTDurbinML\n2006\nTumour suppressor p53 down-regulates the expression of the human hepatocyte nuclear factor 4alpha (HNF4alpha) gene.\nBiochem J\n400\n303\n313\n16895524\n5. LevineAJ\n1997\np53, the cellular gatekeeper for growth and division.\nCell\n88\n323\n331\n9039259\n6. LohrumMAVousdenKH\n2000\nRegulation and function of the p53-related proteins: same family, different rules.\nTrends Cell Biol\n10\n197\n202\n10754563\n7. OrenM\n2003\nDecision making by p53: life, death and cancer.\nCell Death Differ\n10\n431\n442\n12719720\n8. SionovRVHauptY\n1999\nThe cellular response to p53: the decision between life and death.\nOncogene\n18\n6145\n6157\n10557106\n9. ZhaoRGishKMurphyMYinYNottermanD\n2000\nAnalysis of p53-regulated gene expression patterns using oligonucleotide arrays.\nGenes Dev\n14\n981\n993\n10783169\n10. ChenXKoLJJayaramanLPrivesC\n1996\np53 levels, functional domains, and DNA damage determine the extent of the apoptotic response of tumor cells.\nGenes Dev\n10\n2438\n2451\n8843196\n11. LinAWBarradasMStoneJCvan AelstLSerranoM\n1998\nPremature senescence involving p53 and p16 is activated in response to constitutive MEK/MAPK mitogenic signaling.\nGenes Dev\n12\n3008\n3019\n9765203\n12. LinAWLoweSW\n2001\nOncogenic ras activates the ARF-p53 pathway to suppress epithelial cell transformation.\nProc Natl Acad Sci U S A\n98\n5025\n5030\n11309506\n13. CarboneRPearsonMMinucciSPelicciPG\n2002\nPML NBs associate with the hMre11 complex and p53 at sites of irradiation induced DNA damage.\nOncogene\n21\n1633\n1640\n11896594\n14. InsingaAMonestiroliSRonzoniSCarboneRPearsonM\n2004\nImpairment of p53 acetylation, stability and function by an oncogenic transcription factor.\nEmbo J\n23\n1144\n1154\n14976551\n15. PearsonMCarboneRSebastianiCCioceMFagioliM\n2000\nPML regulates p53 acetylation and premature senescence induced by oncogenic Ras.\nNature\n406\n207\n210\n10910364\n16. SerranoMLinAWMcCurrachMEBeachDLoweSW\n1997\nOncogenic ras provokes premature cell senescence associated with accumulation of p53 and p16INK4a.\nCell\n88\n593\n602\n9054499\n17. FerbeyreGde StanchinaELinAWQueridoEMcCurrachME\n2002\nOncogenic ras and p53 cooperate to induce cellular senescence.\nMol Cell Biol\n22\n3497\n3508\n11971980\n18. CastroMEFerrerICasconAGuijarroMVLleonartM\n2008\nPPP1CA contributes to the senescence program induced by oncogenic Ras.\nCarcinogenesis\n29\n491\n499\n18204081\n19. KapoorMLozanoG\n1998\nFunctional activation of p53 via phosphorylation following DNA damage by UV but not gamma radiation.\nProc Natl Acad Sci U S A\n95\n2834\n2837\n9501176\n20. LuHTayaYIkedaMLevineAJ\n1998\nUltraviolet radiation, but not gamma radiation or etoposide-induced DNA damage, results in the phosphorylation of the murine p53 protein at serine-389.\nProc Natl Acad Sci U S A\n95\n6399\n6402\n9600977\n21. WebleyKBondJAJonesCJBlaydesJPCraigA\n2000\nPosttranslational modifications of p53 in replicative senescence overlapping but distinct from those induced by DNA damage.\nMol Cell Biol\n20\n2803\n2808\n10733583\n22. OdaKArakawaHTanakaTMatsudaKTanikawaC\n2000\np53AIP1, a potential mediator of p53-dependent apoptosis, and its regulation by Ser-46-phosphorylated p53.\nCell\n102\n849\n862\n11030628\n23. FerbeyreGde StanchinaEQueridoEBaptisteNPrivesC\n2000\nPML is induced by oncogenic ras and promotes premature senescence.\nGenes Dev\n14\n2015\n2027\n10950866\n24. LoweSWRuleyHEJacksTHousmanDE\n1993\np53-dependent apoptosis modulates the cytotoxicity of anticancer agents.\nCell\n74\n957\n967\n8402885\n25. TanakaNIshiharaMLamphierMSNozawaHMatsuyamaT\n1996\nCooperation of the tumour suppressors IRF-1 and p53 in response to DNA damage.\nNature\n382\n816\n818\n8752276\n26. CastroMEdel Valle GuijarroMMoneoVCarneroA\n2004\nCellular senescence induced by p53-ras cooperation is independent of p21waf1 in murine embryo fibroblasts.\nJ Cell Biochem\n92\n514\n524\n15156563\n27. NaritaMNunezSHeardELinAWHearnSA\n2003\nRb-mediated heterochromatin formation and silencing of E2F target genes during cellular senescence.\nCell\n113\n703\n716\n12809602\n28. MichalovitzDHalevyOOrenM\n1990\nConditional inhibition of transformation and of cell proliferation by a temperature-sensitive mutant of p53.\nCell\n62\n671\n680\n2143698\n29. NelsonDAKrucherNALudlowJW\n1997\nHigh molecular weight protein phosphatase type 1 dephosphorylates the retinoblastoma protein.\nJ Biol Chem\n272\n4528\n4535\n9020179\n30. AlbertsASThorburnAMShenolikarSMumbyMCFeramiscoJR\n1993\nRegulation of cell cycle progression and nuclear affinity of the retinoblastoma protein by protein phosphatases.\nProc Natl Acad Sci U S A\n90\n388\n392\n8380637\n31. BerndtNDohadwalaMLiuCW\n1997\nConstitutively active protein phosphatase 1alpha causes Rb-dependent G1 arrest in human cancer cells.\nCurr Biol\n7\n375\n386\n9197238\n32. RubinETamrakarSLudlowJW\n1998\nProtein phosphatase type 1, the product of the retinoblastoma susceptibility gene, and cell cycle control.\nFront Biosci\n3\nD1209\n1219\n9835651\n33. MooiWJPeeperDS\n2006\nOncogene-induced cell senescence–halting on the road to cancer.\nN Engl J Med\n355\n1037\n1046\n16957149\n34. AyllonVMartinezACGarciaACaylaXRebolloA\n2000\nProtein phosphatase 1alpha is a Ras-activated Bad phosphatase that regulates interleukin-2 deprivation-induced apoptosis.\nEmbo J\n19\n2237\n2246\n10811615\n35. SherrCJMcCormickF\n2002\nThe RB and p53 pathways in cancer.\nCancer Cell\n2\n103\n112\n12204530\n36. WeberJDJeffersJRRehgJERandleDHLozanoG\n2000\np53-independent functions of the p19(ARF) tumor suppressor.\nGenes Dev\n14\n2358\n2365\n10995391\n37. PalmeroIPantojaCSerranoM\n1998\np19ARF links the tumour suppressor p53 to Ras.\nNature\n395\n125\n126\n9744268\n38. EfeyanAGarcia-CaoIHerranzDVelasco-MiguelSSerranoM\n2006\nTumour biology: Policing of oncogene activity by p53.\nNature\n443\n159\n16971940\n39. MartinsCPBrown-SwigartLEvanGI\n2006\nModeling the therapeutic efficacy of p53 restoration in tumors.\nCell\n127\n1323\n1334\n17182091\n40. ChristophorouMAMartin-ZancaDSoucekLLawlorERBrown-SwigartL\n2005\nTemporal dissection of p53 function in vitro and in vivo.\nNat Genet\n37\n718\n726\n15924142\n41. CampisiJ\n2005\nSuppressing cancer: the importance of being senescent.\nScience\n309\n886\n887\n16081723\n42. ColladoMSerranoM\n2005\nThe senescent side of tumor suppression.\nCell Cycle\n4\n1722\n1724\n16294043\n43. EvanGIChristophorouMLawlorEARingshausenIPrescottJ\n2005\nOncogene-dependent tumor suppression: using the dark side of the force for cancer therapy.\nCold Spring Harb Symp Quant Biol\n70\n263\n273\n16869762\n44. CarneroAHudsonJDHannonGJBeachDH\n2000\nLoss-of-function genetics in mammalian cells: the p53 tumor suppressor model.\nNucleic Acids Res\n28\n2234\n2241\n10871344\n45. DimriGPLeeXBasileGAcostaMScottG\n1995\nA biomarker that identifies senescent human cells in culture and in aging skin in vivo.\nProc Natl Acad Sci U S A\n92\n9363\n9367\n7568133\n46. ZouHHastieT\n2005\nRegularization and variable selection via the elastic net.\nJ ROY STAT SOC SER B-STAT MET\n67\n301\n320\n47. DaubechiesIDefriseMDe MolC\n2004\nAn iterative thresholding algorithm for linear inverse problems with a sparsity constraint.\nCommunications on Pure and Applied Mathematics\n57\n1413\n1457\n48. McPhersonJPTamblynLEliaAMigonEShehabeldinA\n2004\nLats2/Kpm is required for embryonic development, proliferation control and genomic integrity.EMBO J\n23\n3677\n88\n49. AylonYMichaelDShmueliAYabutaNNojimaHOrenM\n2006\nA positive feedback loop between the p53 and Lats2 tumor suppressors prevents tetraploidization.\nGenes Dev\n20\n2687\n700\n17015431\n50. VoorhoevePMle SageCSchrierMGillisAJStoopH\n2006\nA genetic screen implicates miRNA-372 and miRNA-373 as oncogenes in testicular germ cell tumors.\nCell\n124\n1169\n81\n16564011\n51. TakahashiYMiyoshiYTakahataCIraharaNTaguchiT\n2005\nDown-regulation of LATS1 and LATS2 mRNA expression by promoter hypermethylation and its association with biologically aggressive phenotype in human breast cancers.\nClin Cancer Res\n11\n1380\n5\n15746036\n52. Jiménez-VelascoARomán-GómezJAgirreXBarriosMNavarroG\n2005\nDownregulation of the large tumor suppressor 2 (LATS2/KPM) gene is associated with poor prognosis in acute lymphoblastic leukemia.\nLeukemia\n19\n2347\n50\n16208412\n53. SunABagellaLTuttonSRomanoGGiordanoA\n2007\nFrom G0 to S phase: a view of the roles played by the retinoblastoma (Rb) family members in the Rb-E2F pathway.\nJ Cell Biochem\n102\n1400\n4\n17979151\n54. VidalACarneiroCZalvideJB\n2007\nOf mice without pockets: mouse models to study the function of Rb family proteins.\nFront Biosci\n12\n4483\n96\n17485390\n55. GoodrichDW\n2006\nThe retinoblastoma tumor-suppressor gene, the exception that proves the rule.\nOncogene\n25\n5233\n43\n16936742\n56. BringoldFSerranoM\n2000\nTumor suppressors and oncogenes in cellular senescence.\nExp Gerontol\n35\n317\n29\n10832053\n57. RajeshDSchellKVermaAK\n1999\nRas mutation, irrespective of cell type and p53 status, determines a cell's destiny to undergo apoptosis by okadaic acid, an inhibitor of protein phosphatase 1 and 2A.\nMol Pharmacol\n56\n515\n25\n10462539"
+}

batch_5/PMC2535588.json

@@ -0,0 +1,4 @@
+{
+    "id": "PMC2535588",
+    "text": "This is an academic paper. This paper has corpus identifier PMC2535588\nAUTHORS: Stefan Malmqvist, Charlotte Leboeuf-Yde\n\nABSTRACT:\nBackgroundThe Finnish chiropractic profession is young and not fully accepted by Finnish healthcare authorities. The demographic profile and style of practice has not been described to date. However, as the profession seems to be under rapid development, it would be of interest to stakeholders, both chiropractic and political, to obtain a baseline description of this profession with a view to the development of future goals and strategies for the profession. The purpose of this study was to describe the chiropractic profession in Finland in relation to its demographic background, the demographics of their clinics, practice patterns, interactions with other health care practitioners and some of the professions' plans for the future.MethodsA structured questionnaire survey was conducted in 2005, in which all 50 members of the Finnish Chiropractic Union were invited to participate.ResultsIn all, 44 questionnaires were returned (response rate 88%). Eighty percent of the respondents were men, and 77% were aged 30 to 44 years old, most of whom graduated after 1990 with either a university-based bachelors' or masters' degree in chiropractic. Solo practice was their main practice pattern. The vast majority described their scope of practice to be based on a musculoskeletal approach, using the Diversified Technique, performing Soft Tissue Therapy and about two-thirds also used an Activator Instrument (mechanical adjusting instrument). The mean number of patient visits reported to have been seen weekly was 59 of which nine were new patients. Most practitioners found this number of patients satisfactory. At the initial consultation, 80% of respondents spent 30–45 minutes with their patients, 75% spent 20–30 minutes with \"new old\" patients and on subsequent visits 80% of respondents spent 15–30 minutes. Interactions with other health care professions were reasonably good and most of chiropractors intended to remain within the profession.ConclusionThe Finnish chiropractic profession is relatively young. Consequently, many of the practitioners have a university-degree, which reflects recent developments in undergraduate chiropractic education. Their practice profile and the manner in which they practice appear to be fairly traditional.\n\nBODY:\nBackgroundFinland is a country situated in the north of Europe, consisting of approximately 5 million inhabitants. The climate is characterized by cold winters and relatively warm summers. Most inhabitants, 95%, speak Finnish, a language that is difficult to learn for foreigners since its vocabulary lacks common roots with most other European languages, and has a structure that differs significantly from the classical languages. English was not taught as a foreign language to all primary school pupils until the early 1970s. As a result the Finnish people was not strongly influenced by the Anglo-American culture.The previous main sources of income were forestry, agriculture, fishery and heavy industry. In the period between its liberation from Russia in 1917 until World War II, the economy was weak. Finland was deeply scarred by its participation in World War II, and the economy improved only slowly thereafter, partly due to the large post-war \"fine\" that had to be paid to the Soviet Union. Over the past decades, an advanced electronic industry has developed and much of the rural population has become urbanized. The standard of living is now high, as is the educational standard, and much emphasis is put on various public health measures [1-3].Traditionally, folk medicine has played a large role, and still does, particularly among people in the rural areas. It is therefore not surprising that Finland was late to \"discover\" chiropractic. Not until in the early 1950s did the first North American-educated chiropractor set up practice, followed by a second chiropractor 20 years later. Subsequently, the profession grew very slowly with competition from folk healers and manual therapists, who were typically trained through weekend courses.Attempts have been made to ensure a world-wide minimum standard for all chiropractic institutions, through the use of an eductional control organisation, the Council on Chiropractic Eduction (CCE). There are branches in different parts of the world, and the European branch (ECCE) inspects and certifies all European chiropractic institutions, whether private or state funded. Chiropractic educational institutions that have achieved full certifiction by the ECCE are the Anglo-European College of Chiropractic, UK, the University of Glamorgan, UK, l'Institut Franco Européen de Chiropratique, France, and the University of Southern Denmark, Denmark. There is no ECCE-approved chiropractic education in Finland, so a chiropractic degree must be obtained in foreign countries.During the late 1980s and early 1990s legislation for chiropractors was introduced in the other Nordic countries. In Finland, a new law introduced in 1994 licenced chiropractors with an academic degree from foreign chiropracic educational institutions, and grants were made available for students to study chiropractic abroad. The legal situation for the chiropractic profession has improved, but the working conditions are still unsatisfactory. Although the professional title recently became protected by law, chiropractors are unable to refer patients to other health care providers, cannot perform their own radiological examination, do not have direct access to imaging services, and may not prescribe sick leave. Unlike in the other Nordic countries, chiropractic patients in Finland are not entitled to government-subsidized reimbursement.Despite this situation, the size of the Finnish Chiropractic Union membership has, according to personal communication with its secretariat, increased five-fold during the last 15 years. To date there are, according to communication with the Finnish National Authority for Medicolegal Affairs, 70 chiropractors in Finland with either a DC degree or an academic chiropractic degree, of which 52 (Jan 2008) are members of the Finnish Chiropractic Union.It was the purpose of this study to describe the chiropractic profession in Finland, in terms of demographic and educational background. The demographics of their clinics included location, practice pattern, interactions with other health care practitioners, and some of the chiropractors' plans for the future.MethodsA survey was conducted using a structured questionnaire [Additional file 1: Demographic questionnaire for chiropractors in FCU]. All chiropractors who at the time of the study were members of the Finnish Chiropractic Union (N = 50) were invited to participate. The selection of participants was limited to members of the Finnish Chiropractic Union to ensure participation of graduates from CCE/ECCE accredited educational institutions.Appropriately qualified graduates who were non-members of the Finnish professional association were not approached because they, from experience, are unwilling to participate in any communal activities. The questionnaire was first tested in a pilot study on 10% of the members for face validity, and the main survey was then administered in June 2005.Approval was sought from the Helsinki University Ethics Committee, but because the survey was considered a quality assurance project approval was not needed. However, all questionnaires were coded to avoid recognition of respondents and the code key was destroyed when data collection was completed. Return of questionnaire implied consent from the participant. In order to respect the anonymity of the participants, in such a small group of practitioners, no comparison was made between responders and non-responders.Data were entered into the SPSS 11.0 spreadsheet by a person experienced in data entry. Eleven questionnaires were randomly selected and each item was manually checked versus the entered data. No errors in the data entry were identified, which was considered satisfactory.Therefore it was not considered necessary to undertake a double entry.Analysis was done using SPSS 11.0 and Minitab. Some of the variables were grouped into fewer categories, based on the frequency of responses. The results were reported as descriptive data in tables and summarized in the text.ResultsDescription of study sampleForty-four of the 50 distributed questionnaires were returned, a response rate of 88%. Eighty percent of the respondents were men and 77% were aged 30 to 44 years [Additional file 2]. Eighty percent practised in a city suburb or city center. Forty-eight percent had one practice only, followed by 40% with two practices, and 12% with more than two practices. However, about one third expected to enter a partnership with a colleague within the next two years. Forty-five percent worked together (in the same clinic) with another health care provider and another 25% expected to do so within two years. Two of the respondents expected not to be working as a chiropractor at that time.Fourteen percent employed a (non-chiropractic) assistant and a further 23% expected to do so within the next two years. About half had access to a receptionist. The majority (77%) graduated between 1990 and 2004 and 53% reported to have practised actively for a maximum of 10 years [Additional file 3]. Only 32% had a diploma of Doctor of Chiropractic, the remaining had either a university-based bachelor or master degree. Nine percent would consider undertaking an additional university degree. In addition, almost half of the members subscribe to a professional journal, usually the Journal of Manipulative and Physiological Therapeutics. In relation to interactions with other health care professions, these seem to be reasonably good [Table 1]. For example, the mean number of conversations/phone calls with other health care personnel in the past week was 3.3.Table 1A description of professional interactions between 44 Finnish chiropractors and other health care practitioners.VariablesSubgroupsFrequencyPercentageReceived at least one referral last week from...Medical practitioner2864Physiotherapist1227Masseur2352Sent at least one report in relation to referral last weekYes1841No2659Had at least one conversation/phone call with other health care personnel last weekYes2352No2148Quality of co-operation with other health care providersMainly good2148Both good and bad1227Mainly lack of co-operation1125Scope of practiceThe survey instrument also included questions on scope of practice, type of technique used, and adjunctive therapies used [Additional file 4]. The vast majority described their scope of practice to be based on a musculoskeletal approach. Almost all used the Diversified Technique, the vast majority performed Soft Tissue Therapy, and about two-third also made use of an Activator Instrument.Various adjunctive therapies were used but none of these was used by all or even by the majority. Ice was most commonly reported, by 46%. Seventy-seven percent had a viewing box for radiology readings, 40% had the possibility to, indirectly, refer patients for X-ray examination via medical practitioners, of which 9% could refer for MRI or CT scans, whereas the use of ultrasound was very rare.PatientsThe mean and median patient numbers, respectively, during the third week of 2005 was reported to be 59 and 47. However, there was a wide range, from 5 to 228. The mean and median number of new patients in that week was 9 and 2, respectively, indicating that the spread of data were skewed. The number of new patients in the past week preceding the study was also stated to be 9.Eighty percent of the participants spent 30–45 minutes with their patients at the first visit, 75% spent 20–30 minutes on \"new old\" patients, whereas in subsequent visits 80% of respondents spent 15–30 minutes. At one extreme, one respondent reported spending one minute only on subsequent visits. The number of patient seen wasconsidered to be \"about right\" by 55%, and 9% reported they were seeing more patients than they would like to. However, about one third (36%) would have been happy to see some more patients.Eighty percent had the (mandatory) malpractice insurance, 73% had a private pension scheme and almost as many (68%) had a private health care insurance.DiscussionThe Finnish Chiropractic profession is relatively young and small, compared to other national chiropractic associations in Europe, and the demographics and practice procedures of the profession have never previously been documented [4]. Perhaps for this reason, Finnish chiropractors were eager responders to this survey with 88% returning the questionnaire.According to the present survey, the members of the Finnish Chiropractic Union consisted of mainly young men (80%), who, in the majority of cases, graduated from university based or university affiliated chiropractic institutions. Information acquired from the administrative offices of corresponding associations in Sweden, Norway and Denmark reveals a different gender distribution, with 70%, 71% and 51%, respectively of men in the three countries. The proportion of male practitioners was lower (63%) also in a recent study of German chiropractors (response rate 72%) [5].Despite the young age of the Finnish chiropractors, their current practice pattern was similar to that of the early years of chiropractic in Finland. Typically a Finnish chiropractor is working in solo practice, sharing his time between one or two practices. Sixty percent reported working in a solo practice, whereas, according to a previous study (response rate 70%), the estimated proportion was 41% among European chiropractors in general [6]. In the more recent German study, 45% of the respondents worked in a solo practice setting [5].The Finnish Chiropractic Union subscribe to the Chiropractic Report for its members, a cost that is included in the membership fee. Additionally does almost 50% of the members subscribe to one more professional scientic journal. The Finnish chiropactor thus seem to be academically updated. However, regarding future development is only a small number interested in further education at a university level. This is understandable at the present time, considering the isolated position of the chiropractic profession, and the absence of chiropractic academic institutions in Finland. The time, money and effort spent on further education, would lead to no additional career possibilities.Respondents were satisfied with the number of patients and they seemed to enjoy reasonable contacts with other health care practitioners. This may indicate that their professional activities are felt to be fulfilling. Nevertheless, two of the respondents were planning to leave the profession, although they were not close to retirement age.Most reported to have a musculoskeletal approach, using mainly Diversified Manipulation Technique, Soft Tissue Techniques and Activator Instrument. These are methods previously reported frequently to be used in Europe [5-7].The use of adjunctive therapies showed a less distinct pattern, perhaps because chiropractors determined that different patients require different approaches. It was also interesting that about one-third of the respondents had some sort of rehabilition equipment in their clinic, indicating that they also have the facility to assist patients with general or specific training following the acute treatment stage.Regarding professional activities, only some of our data are comparable with information from previous European surveys, such as time spent with patients. The time spent on the first visit appeared to be similar in Finland and the Netherlands (36 and 41 minutes, respectively) [8]. Subsequent visits took 22 minutes in Finland and 15 minutes in the Netherlands.Most of the Finnish chiropractors had made sure that they were covered with insurances both for pension scheme and private healthcare but, a small number appeared not to have the obligatory malpractice insurance.The limitations of this study are that, despite the high response rate, not all chiropractors with a CCE/ECCE-approved education are members of the Finnish Chiropractors' Union and that not all members of the professional association participated in the survey. It is possible that non-participants in the study have a profile that differs from that of the responders. Other limitations are, of course, that the questionnaire was not exhaustive. For example, the description of practice procedures might have been designed differently by other groups of researchers, and the participants were not encouraged to extend their answers beyond the the questions stated in the questionnaire. Therefore, it is possible that some nuances of clinical practice failed to be recorded. However, the results of the pilot testing of the survey instrument did not indicate that the questionnaire failed to provide meaningful answering options.ConclusionThe educational background of the chiropractic participants in this study reflects the recent development in chiropractic education, with university affiliations and masters degrees. Although the Finnish chiropractic profession is relatively young, these chiropractors appeared to have a traditional practice profile: solo practice, a musculoskeletal approach, allowing good time for examination and treatment.Competing interestsThe authors declare that they have no competing interests.Authors' contributionsSM was responsible for planning and executing the demographic survey, participated in the data collection and drafted the manuscript. CL–Y supervised the process. Both SM and CL–Y participated in the design of the study and performed the analysis. Both authors read, finalized and approved the final manuscript.Supplementary MaterialAdditional file 1Demographic questionnaire for chiropractors in FCU. A translation of the original Finnish questionnaire.Click here for fileAdditional file 2Table 2. Description of 44 Finnish chiropractors and their practice patterns, I.Click here for fileAdditional file 3Table 3. Description of 44 Finnish chiropractors and their practice patterns, II.Click here for fileAdditional file 4Table 4. A description of scope of practice, techniques and adjunctive therapies used according to a survey of 44 Finnish chiropractors.Click here for file\n\nREFERENCES:\nNo References"
+}

batch_5/PMC2535590.json

@@ -0,0 +1,4 @@
+{
+    "id": "PMC2535590",
+    "text": "This is an academic paper. This paper has corpus identifier PMC2535590\nAUTHORS: Louise C Bruce, Robert Jellison, Jörg Imberger, John M Melack\n\nABSTRACT:\nThe significance of the transport of nutrient-rich hypolimnetic water via the benthic boundary layer (BBL) to the productivity of Mono Lake was studied using a coupled hydrodynamic and ecological model validated against field data. The coupled model enabled us to differentiate between the role of biotic components and hydrodynamic forcing on the internal recycling of nutrients necessary to sustain primary productivity. A 4-year period (1991–1994) was simulated in which recycled nutrients from zooplankton excretion and bacterially-mediated mineralization exceeded sediment fluxes as the dominant source for primary productivity. Model outputs indicated that BBL transport was responsible for a 53% increase in the flux of hypolimnetic ammonium to the photic zone during stratification with an increase in primary production of 6% and secondary production of 5%. Although the estimated impact of BBL transport on the productivity of Mono Lake was not large, significant nutrient fluxes were simulated during periods when BBL transport was most active.\n\nBODY:\nBackgroundThe transport of nutrient-rich water from benthic to pelagic regions has been linked to increased levels of primary productivity in stratified lakes [1-3]. Ostrovsky et al.. (1996) suggest that seiche activity in the boundary layer of Lake Kinneret sustained a vertical flux between the hypolimnetic and epilimnetic waters enhancing biological productivity in the lake. MacIntyre et al. (1999) calculated the upward fluxes of ammonium across the nutricline in Mono Lake and suggested nearshore boundary fluxes could be the dominant pathway supplying ammonium to the deep chlorophyll maxima. Eckert et al. (2002) used microstructure measurements of temperature, oxygen and hydrogen sulphide in Lake Kinneret to conclude that following the onset of stratification, the flux of benthic nutrients to the water column controls primary productivity. In this study we have defined BBL transport as that which occurs in the layer bordering the sediments of a lake [4,5] alternatively referred to as the bottom boundary layer [6].The development of basin-scale internal waves arising from wind-induced energy are responsible for large scale water motions and most of the turbulence caused by these large-scale motions occurs in the BBL [7,8]. In order to differentiate between boundary and internal modes of vertical transport Yeates and Imberger (2004) parameterized the split between mixing in the internal and benthic boundary layer (BBL) using values of Lake number, LN [9] and Burger number, BN [10]. The LN is a measure of the amplitude of basin-scale internal waves in response to surface wind forcing, and BN describes waves that evolve from simple seiches [4]. Simulations performed on a number of monomictic lakes indicated that fluxes through the BBL were dominant during strong wind events occurring during period of stratification [4].A number of studies, aimed at identifying sources and sinks of nutrients in the photic zone have focused on bacterial mineralization [11], regeneration through planktonic organisms [11-13], nitrogen fixation [14], hypolimnetic flux and inflows and outflows [15]. Although the occurrence of BBL transport and its potential impact on primary productivity has been examined, the upward mixing of nutrient-rich hypolimnetic waters via the BBL and the consequent effect on lake-wide ecological processes deserves further analysis.Mono Lake is a nitrogen-limited saline lake with a relatively simple food web [16] and is subjected to wind-driven boundary-layer mixing events [1]. Yeates and Imberger (2004) simulated a BBL thickness in Mono Lake of 10–15 m during a sequence of strong wind event suggesting an active role in the development of the thermal structure of the lake. These features make it well-suited for examining the role of BBL-supplied nutrients and the influence of these nutrients on the seasonal plankton dynamics and overall productivity of the lake.The objective of the present study is to investigate the role of BBL transport in the supply of nutrients to the photic zone and its consequent impact on the lake's ecology. A coupled hydrodynamic and ecological model was used to quantify nitrogen biogeochemistry during a 4-yr period from 1991–1994 when the lake mixed to the bottom during the winter. Initially, we calibrated the model parameters and processes to ensure an acceptable representation of the field data. The simulated output was then used to calculate the sources and sinks of nitrogen to the photic zone. A comparison could then be made between the roles of recycled and external sources on the primary and secondary productivity in the lake. To enable quantification of the significance of BBL transport for ecological processes, a series of simulations were run in which this mechanism was switched off allowing a comparison between lake behavior with and without BBL transport.Study SiteMono Lake (38°N: 119°W) is a large saline lake with a salinity of 85–92 g kg-1, a maximum depth 45 m, mean depth 17 m and surface area approximately 160 km2 (Fig. 1). The lake was monomictic during the period studied (1991–1994), and vertically mixed in winter (December to February) with thermal stratification beginning in early spring and persisting through autumn [17]. At other times following large runoff years, the lake experienced multi-year periods of chemical stratification (i.e., meromixis; 1982–1988, Jellison and Melack 1993b; 1995–2003, Jellison unpublished data). The present study examines four monomictic years (1991–94) to assess the effects of BBL on nutrient cycling and productivity during stratified and holomictic periods.Figure 1Mono Lake. Bathymetric map of Mono Lake showing sampling stations. Depth contours are in meters.Figure 2Model schematic. Schematic representation of (A) the model layer structure (B) internal and boundary layer mixing in the physical model DYRESM. (BBL: benthic boundary layer; Internal: internal cells; BC: benthic boundary layer cells) and (C) the carbon and nitrogen fluxes represented in the ecological model, CAEDYM. Dotted lines indicate that these variables are not included in model.The planktonic community of Mono Lake has few species as is typical of hypersaline waters. The phytoplankton is dominated by a newly described picoplanktonic (2–3 μm) green alga, Picocystis salinarum Lewin (Lewin et al., 2000), and several bacillarophytes, mainly Nitzschia spp. (20–30 μm) (Lovejoy & Dana, 1977; Mason, 1967). A brine shrimp, Artemia monica Verill, is the only macrozooplankter (Lenz, 1980; Lenz, 1984). While pelagic ciliates and rotifers may also be present at times (Mason, 1967; Jellison et al. 2001), they contribute a negligible amount to the total zooplankton biomass.There is a strong seasonal pattern in the nutrient and plankton dynamics of Mono Lake [18]. The seasonal patterns are driven by biotic and abiotic forces affecting productivity via bottom-up and top-down controls. Water temperatures of the surface mixed-layer ranged from 2–5°C in winter to 12–22°C in summer. Seasonal stratification and high productivity result in anoxic conditions in the hypolimnion where ammonium accumulates. The flux of this ammonium to the photic zone is limited until winter overturn mixes the whole lake providing nutrients for a pronounced spring algal bloom. Daily primary productivity rates are relatively high (Jellison and Melack 1993a).The lake's only macrozooplankter, A. monica, produces over-wintering cysts that lie dormant on the bottom during the winter and hatch during early spring (February-April) [19]. A. monica biomass usually peaks in the late spring, remains high during the summer and gradually declines during the autumn as food is scarce and temperatures decline. The spring growth of A. monica biomass is associated with a simultaneous decline in phytoplankton biomass due to grazing and rise in surface concentrations of ammonium from zooplankton excretion. Phytoplankton biomass remains low during the summer and only increases toward the end of the year when grazing pressure is reduced [20].As phosphorus concentrations are always high (>400 μM; Jellison et al. 1993), nitrogen limits primary production in the photic zone (Jellison & Melack 1993a, 2001). Nitrogen inputs from inflowing streams and planktonic nitrogen fixation are very low relative to internal fluxes where the main sources are from sediment release in the hypolimnion, phytoplankton and zooplankton excretion, and bacterial mineralization of particulate detrital organic nitrogen. Peak concentrations in the photic zone are observed at the breakdown of stratification as nutrient-rich hypolimnetic waters become entrained into the epilimnion. Towards the end of the mixed period and onset of stratification ammonium levels are generally low. When the zooplankton become abundant in late spring, grazing reduces phytoplankton biomass and internal phytoplankton nitrogen is converted to ammonium via the zooplankton grazing and excretion. Zooplankton excretion and reduced ammonium uptake due to low phytoplankton biomass results in an increase in epilimnetic ammonium concentrations.ResultsNutrient concentrations, phytoplankton and zooplankton biomassThe seasonal ammonium pattern of low winter concentrations and high summer values is reproduced by the model (Fig. 3). Similarly, peak concentrations of ammonium apparent in the observed data coinciding with the arrival of A. monica in the spring are matched in magnitude and timing by the model results. At the breakdown of stratification, the model simulated reduced ammonium concentrations corresponding to increased phytoplankton biomass (Fig. 3). However, the isolated high spikes in ammonium concentration observed in the field data during full circulation were generally not captured by the model (Fig. 3).Figure 3Model simulations. Comparison of model simulation results (lines) and field data (crosses) for Mono Lake from 1991 to 1994 for 9-m depth integrated averages of ammonium (NH4), total phytoplankton carbon (phytoplankton), total organic nitrogen (TPON), total organic carbon (TPOC) and dissolved oxygen (DO), vertical net tows of Artemia monica (Artemia).The simulated values of phytoplankton biomass follow the low summer concentrations, timing and slope of the autumn recovery and spring decline observed in the field data (Fig. 3). However, the elevated values of phytoplankton biomass observed in the field at the end of the mixing periods (early 1992, 1993 and 1994) are not captured by the model (Fig. 3).The particulate organic nitrogen (PON) and particulate organic carbon (POC) data observed in the field closely followed that of the phytoplankton, with elevated values during the winter in the absence of grazing and low values during the summer months. These patterns were captured by the model, although elevated levels of PON were underestimated by the model during periods of full circulation (Fig. 3). However, elevated levels of POC were generally captured by the model which suggests that the model overestimated the detrital component of the particulate carbon pool (Fig. 3).The timing and slope of the early spring peak in A. monica biomass observed in the field was matched in the simulated results across the four year simulation period (Fig. 3).Simulated concentrations of dissolved oxygen are similar to those measured in the field during stratified periods (Fig. 3). The model, however, under predicted concentrations in late spring for both 1991 and 1992.Productivity and nitrogen fluxesPrimary productivity in Mono Lake has been estimated using a numerical interpolative model incorporating photosynthetic uptake rates and measured vertical attenuation of PAR [21]. During the non-meromictic conditions of 1989 and 1990, Jellison and Melack (1993a) estimated an average daily productivity of 1.6 g C m-2 d-1. This matches the value simulated by DYRESM-CAEDYM for the 1991–1994 monomictic period. During periods of stratification an average daily productivity of 1.7 g C m-2 d-1 was simulated and 1.3 g C m-2 d-1 during periods of full circulation.Average rates of lake-wide nitrogen deposition measured in 1986 and 1987 ranged from approximately 5.9 Mg N d-1 (ca. 2.5 mmol m-2 d-1) during the summer to 2.7 Mg N d-1 (ca. 1.2 mmol m-2 d-1) during the winter (Jellison et al. 1993). These rates are similar to those simulated by the model, i.e., 3.7 Mg N d-1 and 2.1 Mg N d-1 averaged during periods of stratification and full circulation, respectively. Areal average lake-wide nitrogen fluxes from the sediments were calculated by the model as 12.5 Mg N m-2 d-1 and 6.2 Mg N m-2 d-1 averaged during period of stratification and full circulation, respectively. Jellison et al. (1993) estimated the rate of ammonia release from the sediments based on sediment cores collected in 1988 as 58–162 Mg N m-2 d-1 (ca. 3.6–10.1 mmol m-2 d-1). Although greater than those predicted by the model these estimates were derived under anoxic conditions so should be taken as an upper estimate.Measures of model performanceThe calculated values of normalized mean absolute error, correlation coefficient and slope are presented in Table 3 for each of the main state variables over the full simulation period from 1991 to 1994 and compared to the calibration period from 1991 to 1992. Calculations of correlation coefficients are all equal to or greater than 0.8 with the exception of ammonium and dissolved oxygen.Sensitivity analysisThe five parameters that displayed the greatest sensitivity to annual estimates of lake-wide nitrogen fluxes were: (1) release rate of NH4 from sediments (SdNH4); (2) the fraction of zooplankton grazing excreted (fex); (3) background attenuation coefficient (Kd); (4) internal nitrogen to carbon ratio of the phytoplankton (INcon) and (5) the fraction of zooplankton grazing egested (feg). The optimal parameter value (determined by the model best fit), and the upper and lower bounds used to determine alternative parameter sets are listed in Table 4.Nitrogen budgetFive major nitrogen fluxes were extracted from the model to compare the various component of the nitrogen budget (Fig. 4). These fluxes were: (1) phytoplankton uptake, (2) sediment to water exchange, (3) bacterially mediated mineralization, (4) phytoplankton excretion, and (5) zooplankton excretion. The model results are expressed as mass flux per day with respect to the whole lake, with phytoplankton uptake as a negative flux (sink) and the other four terms as positive fluxes (source) (Fig. 4). The results indicate that mineralization of particulate nitrogen made the greatest contribution to phytoplankton uptake in the winter and zooplankton excretion during the summer (Fig. 4). Sediment-released nitrogen fluxes are comparatively low, although significant in making up the difference between phytoplankton uptake and excretion (Fig. 4).Figure 4Nitrogen fluxes. Nitrogen fluxes (Mg N day-1) for total phytoplankton uptake (PhyUp) against sediment flux (SedFlux), mineralization of PON (Mineral), phytoplankton excretion (PhyEx), and zooplankton excretion (ZoopEx). Corresponding periods of stratification and full circulation are demarked by dashed lines.Boundary layer mixingIn the absence of BBL transport a greater buildup of ammonium in the hypolimnion was simulated, the difference being greatest in the early part of the stratified period (Fig. 5). However, the difference in the epilimnion is not so pronounced. Similarly, the simulated results of the 9 m depth averaged concentrations of ammonium, PON, POC, dissolved oxygen and phytoplankton and A. monica biomass indicated little difference between the alternative scenarios of BBL mixing (Fig. 6).Figure 5BBL ammonium transport. Comparison of NH4 (g m-3) depth profiles for the scenarios of BBL transport activated (solid line) and absent (dotted line) and field data (solid dots) for selected dates from 1991.Figure 6Effect of BBL transport. Comparison of model simulation results with BBL transport activated (solid lines) and absent (dotted lines) from 1991 to 1992 for 9-m depth integrated averages of ammonium (NH4), total phytoplankton carbon (phytoplankton), organic nitrogen (PON), organic carbon (POC) and dissolved oxygen (DO) and vertical net tows of Artemia monica (Artemia).Calculations based on model output indicate that for 1991 to 1994 BBL transport was responsible for a 53% increase in upwards flux of ammonium across the thermocline during periods of stratification. For the corresponding periods, the simulated increase in primary production was calculated as 6% and secondary production as 5%. The model results, averaged over periods of autumn and winter mixing for the 4 years of simulation, indicated a reduction in upward ammonium flux of 28% when the BBL transport was active. This corresponded with a simulated decrease in primary production of 7% and negligible increase in secondary production of 1% for the same periods. The estimated net increase for 1991–1992 in ammonium flux across the thermocline due to BBL transport was 9%, primary productivity was 2% and secondary productivity was 3%.To place the differences in upward ammonium flux due to BBL transport in the context of the nitrogen cycle, the five major nitrogen fluxes were compared for both scenarios (Fig. 7). Almost no differences were found in the rates of regenerated nutrients, sediment flux and settling when BBL transport is inactive. Model results indicate that when the BBL transport was active ammonium flux across the thermocline accounts for 11% of the nitrogen sources to the photic zone during stratified periods. This compares to 5% when BBL transport is inactive.Figure 7Lakewide nitrogen fluxes. Comparison of lake-wide nitrogen fluxes for phytoplankton uptake (Phy Uptake), total regenerated sources (Tot Regen), sediment flux (Sed Flux), flux across the thermocline (Hyp Flux) and settling of particulate nitrogen (Settling) averaged annually, during the stratified periods and during the mixed periods from 1991 to 1994 from the boundary mixing on (black bars) and off (white bars) scenarios. Error bars indicate one standard deviation from mean.DiscussionSeveral aspects of the modeling require further examination. The step temperature function used to represent the process responsible for the hatching and initial growth of over-wintering A. monica cysts simulated well the timing and slope of the early spring peak in A. monica biomass. However, experiments have demonstrated that increases in salinity can influence the hatching process [22]. It is anticipated, therefore, that an additional salinity factor would be required before the model could be used to predict A. monica dynamics under alternative salinities. Although simulated mid-summer concentrations of A. monica compare favorably with those observed in the field, the autumn decline was difficult to simulate well (Fig. 3). The model included three processes responsible for decreases in biomass during this period; limited grazing at low temperatures, end of life mortality and grebe predation. Improved understanding of the combination of triggers responsible for the autumn decline in A. monica will aid in the model representation of these processes. Alternatively a cohort model such as that proposed by [23] may be required to accurately represent the autumn decline.Differences between measures of fit comparing the calibration and validation periods are small. Although the ecological dynamics of the model during the validation period are similar to that of the calibration period, this result is an indication of model stability. However, this stability only relates to the representation of the interactions between the main processes responsible for determining the ecological patterns observed in the lake over the four years studied. Comparison to measures of fit for other lake ecosystem models is difficult as quantitative measures are rarely given. However, our overall measurement of NMAE compare favorably to Ross et al. (1994) (0.65) and Bruce et al. (2006) (0.52).Since we defined sensitivity in relation to estimates of lake-wide nitrogen fluxes, it follows that the parameters showing the most sensitivity are related to the nitrogen cycle. Since the inflow of nitrogen into the lake is negligible, it follows that for Mono Lake, sediment release is a critical source of nitrogen to the water column. Similarly both the fractions of zooplankton grazing that goes into either egestion (the bulk of which is deposited into sediments and thus lost from the photic zone) or excretion (providing nitrogen in a form for primary production) have a direct effect on the proportion of phytoplankton nitrogen that is recycled. The ratio of phytoplankton internal nitrogen to carbon controls both the uptake of inorganic nitrogen by phytoplankton and the flux of nitrogen recycled via the zooplankton grazing and excretion pathway. Background attenuation influences nitrogen fluxes indirectly by controlling the amount of light available for primary productivity.Model results indicated that during the summer stratified periods the N demand by phytoplankton in the surface to 9-m of Mono Lake is predominantly met by zooplankton excretion, phytoplankton leakage of dissolved organics, and bacterially mediated mineralization. Of these, the model predicted that the dominant source was zooplankton excretion. Since zooplankton biomass was well represented by the model including timing and magnitude of the initial peak, it follows that during these peaks the model has the closest fit to the ammonium data. Midwinter spikes in ammonium during period of reduced phytoplankton biomass were not reproduced in the model output. The model simulated almost constant phytoplankton biomass during the winter months that is inconsistent with the field data. From this we would conclude that the processes of phytoplankton ammonium uptake and release are not well represented by the model during winter conditions of high algal biomass and light-limitation. One of the limitations of this study was the assumption (for simplicity) of a constant internal phytoplankton C:N ratio. It is anticipated that modeling the internal nitrogen as a dynamic variable would improve the simulation of the phytoplankton-ammonium interactions particularly during periods of full circulation.This study employed optimization techniques to determine a series of parameter sets to best represent field data as described by the processes included in the current model formulation. Some field data were better represented than others and misrepresentation of field data by the simulation will serve to direct improvements in future model generations. Although the modeled fluxes sometimes over or underestimated the measured concentrations, the general seasonal patterns were captured by the simulations and thus used to provide insight into the processes that determine the ecosystem dynamics of Mono Lake.Bruce et al. (2006) in their study of the role of zooplankton in the nutrient cycles of Lake Kinneret, Israel, found zooplankton excretion to be the dominant source of dissolved nitrogen during winter overturn and sediment release the dominant source during summer stratification. For Mono Lake we also found that zooplankton excretion was most influential in the summer stratified period. Although the simulated rate of ammonium flux from the sediments was higher in Lake Kinneret [24], the main reason for finding sediment-released nutrients relatively less important in Mono Lake is due to two-fold higher rates of primary productivity and greater recycling due to zooplankton excretion in Mono Lake.MacIntyre and Jellison (2001) suggested that transport of nutrient-rich hypolimnetic water via the BBL layer is responsible for increased ammonium flux across the thermocline and consequential increase in productivity. By comparing the simulation results from the two scenarios we found that, although the increase in upward ammonium flux across the thermocline during the stratified periods of 1991–1994 due to BBL transport was 53% (± 4%), primary productivity for the same period increased only 6% (± 4%). Since the model suggested that 87% of the N demand by phytoplankton is met by regenerated sources, it is not unexpected that an increase in external supply has a limited impact. MacIntyre et al. (1999) highlighted the importance of the flux of BBL transported ammonium across the thermocline in sustaining primary productivity to the deep chlorophyll maxima. As a percentage of phytoplankton demand during the stratified periods the upward flux of ammonium across the thermocline was calculated as 12% with BBL on and 5% with BBL off. MacIntyre et al. (1999) reached a similar conclusion and, assuming that 5–10% of primary productivity occurs in the deep chlorophyll maximum during the summer, suggested that BBL may be the dominant mechanism supplying ammonium to the deep chlorophyll maximum.As anticipated, during stratified periods simulation results indicate that BBL transport leads to an increase in ammonium transport across the thermocline and concomitant increase in primary productivity. However, this pattern was reversed during periods of mixing. A greater build up of ammonium in the hypolimnion occurred during stratification in the case where BBL transport is absent (Fig. 6). Although in the absence of BBL transport, less flux was available in the photic zone during stratification, at overturn a greater mass of ammonium led to greater upwards flux of ammonium and concomitant increase in primary productivity. As a result, on an annual average, primary productivity was similar under both scenarios.Our model results have illustrated the importance of timing of BBL transport and its subsequent effect on primary and secondary productivity. The model used in this study did not include algorithms to represent differences in generations using a stage-structured zooplankton model. Inclusion of a stage structured model might enable us to determine whether the timing of BBL transport events and concomitant increases in primary productivity effect the timing and magnitude of successive generations of A. monica in Mono Lake.It is apparent that one of the reasons the transport of ammonium via the BBL does not have a significant impact on the productivity of Mono Lake is that sediment released nutrients are not a major component of the nutrient cycle. Model results have confirmed previous studies indicating that productivity is predominantly sustained by recycled nutrients (Jellison et al.. 1993). Furthermore, simulated estimates of BBL volume from 1991–1994 revealed that, on average, the benthic boundary layer comprised only 1% by volume and stored only 1% of the lake-wide nitrogen mass. To investigate the potential importance of BBL transport for shallower lakes where the volume of BBL may be larger in proportion to the lake volume we ran three additional simulations. The same Mono Lake input files for 1991–1994 were used, lowering the surface level of the lake to simulate initial depths of 35 m, 30 m and 22 m. Combining the results of these simulations we plotted the flux of ammonium transported via the BBL as a fraction of N demand by phytoplankton against daily average values of Lake Number (LN) and primary productivity (Fig. 8). Simulated output indicated that the fraction of N demand met by hypolimnetic nutrients transported upwards in the BBL rarely exceeds 50% and only when primary productivity is minimal or for LN close to 1. The LN is inversely proportional to the thermocline height and is both a measure of the energy available at the thermocline from wind induced surface stress and the volumetric importance of the hypolimnion [9].Figure 8Ammonium flux versus lake number. Flux of ammonium transported via the BBL as a fraction of lake-wide vertical fluxes (closed circles) versus daily average values of Lake Number (LN) and Burger Number (BN).For Lake Kinneret, estimates of monthly primary productivity fall between 0.5 and 1.7 g C m-2 d-1 (Bruce et al. 2006). Mean daily values of LN estimated for Lake Kinneret range from 10-2 to 100 with a period of low LN associated with strong wind events (Yeates and Imberger 2004). Given these ranges, it is predicted that the importance of BBL transport in Lake Kinneret may be greater than the 6% predicted for Mono Lake. In Lake Geneva, a study investigating the effect of internal waves on basin exchange indicated up to 40% of the hypolimnetic volume was exchanged following episodes of strong winds [25]. Primary productivity in Lake Geneva is relatively high [26]. For Lake Constance, estimates of LN during stratification are relatively high (Yeates and Imberger 2004) and productivity is less than 1 g C m-2 d-1 [27] suggesting that for Lake Constance the transport of nutrients through the BBL may be less important to overall lake productivity but potentially significant during episodic events associated with low LN.The results of this study have indicated that the relative importance of BBL transport as a source of nutrients sustaining productivity in the photic zone is determined by productivity and morphology. Future studies will be focused on comparing the effect of BBL transport on the ecology of other lakes. By differentiating between physical and ecological process we will be able to determine what limnological features alter the importance of BBL transport.MethodsModel descriptionThe model used in this study is a modified version of the Computational Aquatic Ecosystem Dynamics Model (CAEDYM) [28,29] coupled to the Dynamic Reservoir Model (DYRESM) [4]. In DYRESM the lake is represented as a series of homogeneous horizontal layers of variable thickness [4]; as inflows and outflows enter or leave the lake, the affected layers expand or contract, respectively, and those above move up or down to accommodate the volume change. Mass, including that of the ecological state variables, is adjusted conservatively each time layers expand, contract, merge or are affected by inflows and outflows. The main processes modelled in DYRESM are surface heat, mass and momentum transfers, mixed layer dynamics, hypolimnetic mixing, benthic boundary layer mixing, inflows and outflows.Local meteorological data are used to determine heating due to short-wave radiation and surface heat fluxes due to evaporation, sensible heat, long-wave radiation and wind stress. The surface wind field introduces both momentum and turbulent kinetic energy to the surface layer contributing to vertical mixing. In addition to surface layer mixing, DYRESM includes algorithms that account for internal mixing (encompassing the effects of internal wave energized shear mixing) and benthic boundary layer (BBL) mixing (determined by the turbulent kinetic energy budget and parameterized by Lake number and the Burger number). The total volume of water (FiT) exchanged by deep water mixing and transport processes for layer i is determined by the following equation:(1)FiT=200Ni2AiKM∆tLNNmax2(δi+δi+12)where N2 is the buoyancy frequency, A is the layer area (m2), KM is the molecular diffusion coefficient for heat, Δt is the time step (seconds), LN is the Lake number and δi the layer thickness of layer i (m) [4]. In this way mass transfer is enabled from hypolimnetic layers to the thermocline region internally and via the BBL. A recent modification of the DYRESM code is the separation of these mass transfers described in detail by Yeates and Imberger (2004). The Lagrangian layers have been separated into internal and BBL cells so volume exchange occurring beneath the surface mixing layer can be separated into that associated with internal mixing (between internal cells) and that associated with benthic boundary layer mixing (between BBL cells; Fig. 2A&B). The volume exchange is partitioned into BBL (FiB) and internal (FiI) using the following equation:(2)FiI={FiTtanh(BN)(LN−1)LNLN>10otherwiseandFiB=FiT−FiIwhere BN is the Burger number [4].The ecological model CAEDYM was set up in the form of an 'N-P-Z' (nutrients-phytoplankton-zooplankton) model (Fig. 2C) with resolution to the level of individual species or groups of species [24]. In the present study it is used to simulate phosphorus and nitrogen in both particulate and dissolved inorganic forms (POP and PO4, PON, NO3, NH4), dissolved oxygen (DO), particulate organic carbon (POC), dissolved organic carbon (DOC), one phytoplankton group representing Picocystis sp. and one zooplankton group representing A. monica. A series of ordinary differential equations is used in CAEDYM to describe changes in concentrations of nutrients, detritus, dissolved oxygen, phytoplankton and zooplankton as a function of environmental forcing and ecological interactions for each cell represented by DYRESM (Table 1). The variables of irradiance, temperature, salinity and density are also passed to CAEDYM at each 1-hr time step and used in equations to determine rates of change of biomass and chemical constituents for each of the ecological state variables. The two models CAEDYM and DYRESM share the same layer structure including the division of hypolimnetic layers into two cells, BBL and internal. The BBL cells are considered adjacent to sediment cells so that sediment exchange of nutrients and dissolved oxygen occurs to and from these cells. The physical transfer of ecological variables between adjacent cells due to various mixing processes is accounted for in DYRESM. Further details of the structure of CAEDYM are given in Robson and Hamilton (2004) and Romero et al. (2004).Table 1Model process equations. Equations used to describe the processes included in the ecological model CAEDYM∂Zi/∂t = [GiAif(Z)if1(T)(1-fex-feg) - (Ri+Mi)f2(T) - Predi]Zi = (assimilation - excretion - egestion) - (respiration + mortality) - predation∂P/∂t = [Pmax,jf1(T)min(f(I),f(P),f(N)) - (Rj)f2(T) - Predj]Pj ± Sj = photosynthetic uptake - (respiration + excretion + mortality) - predation ± settling∂POC/∂t = Σ[Gif(Z)if1(T)i((1-Ai) + Aifeg) + Mif2(T)i]Zi + Σ[Rj(1-fres)(1-fDOM)f2(T)]Pj - PredPOCPOC - RPOCf(DO)f1(T)POC ± SPOM = (unassimilated zooplankton food + zooplankton egestion + zooplankton mortality) + phytoplankton mortality - zooplankton predation - POC decomposition ± settling∂DOC/∂t = Σ[Rj(1-fres)fDOMf2(T)]Pj + RPOCfPOC(POC)f (DO)f2(T)POC - RDOCf (DO)f2(T)DOC = phytoplankton excretion + POC decomposition - DOC mineralisation∂POP/∂t = Σ[Gif(Z)if1(T)i((1-Ai) + Aifeg) + Mif2(T)i]IPZiZi + Σ[Rj(1-fres)(1-fDOM)f2(T)]IPj - PredPOCPOP - RPOPf(DO)f1(T)POP ± SPOM = (unassimilated zooplankton food + zooplankton egestion + zooplankton mortality) + phytoplankton mortality - zooplankton predation - POP decomposition ± settling∂DOP/∂t = Σ[Rj(1-fres)fDOMf2(T)]IPj + Σ[AifexGif(Z)if1(T)i]IPZiZi + RPOPf (DO)f1(T)POP - RDOPf (DO)f1(T)DOP = phytoplankton release + zooplankton excretion + POP decomposition - DOP mineralisation∂PO4/∂t = RDOPf (DO)f2(T)DOP - Σ[UNmax,jf1(T)jf(IP)jf(P)j]Pj + SdPO4f(DO)f2(T)LA/LV = DOP mineralisation - phytoplankton uptake + PO4 sediment flux∂PON/∂t = Σ[Gif(Z)if1(T)i((1-Ai) + Aifeg) + Mif2(T)i]INZiZi + Σ[Rj(1-fres)(1-fDOM)f2(T)]INj - PredPONPON - RPONf(DO)f1(T)PON ± SPOM = (unassimilated zooplankton food + zooplankton egestion + zooplankton mortality) + phytoplankton mortality - zooplankton predation - PON decomposition ± settling∂DON/∂t = Σ[Rj(1-fres)fDOMf2(T)]INj + Σ[AifexGif(Z)if1(T)i]INZiZi + RPONf(DO)f2(T)PON - RDONf(DO)f2(T)DON = phytoplankton release + zooplankton excretion + PON decomposition - DON mineralisation∂NH4/∂t = RDONf(DO)f1(T)DON - Σ[UNmax,jPNf1(T)jf(IN)jf(N)j]Pj - RNOf(DO)f2(T)NH4 + SdNH4f(DO)f2(T)LA/LV = PON mineralisation - phytoplankton uptake - nitrification + NH4 sediment flux∂NO3/∂t = RNOf(DO)f2(T)NH4 - RN2f(DO)f2(T)NO3 - Σ[UNmax,j(1-PN)f1(T)jf(IN)jf(N)j]Pj = nitrification - denitrification - phytoplankton uptake∂DO/∂t = kO2(DO_atm - DO) + Σ[Pmax,jf1(T)jmin(f(I),f(P),f(N)) - Rjf2(T)j]PjYO2:C Σ[Rif2(T)i]ZiYO2:C - RDOCf (DO)f1(T)DOCYO2:C - RNOf (DO)f2(T)NH4 - SdO2f(DO)f2(T)LA/LV = atmospheric flux + (phytoplankton oxygen production - phytoplankton respiratory consumption) - zooplankton respiratory consumption - utilisation of oxygen in mineralisation of DOM - utilisation of oxygen in nitrification - sediment oxygen demand.Temperature functionsf1(T) = θT-20 - θk(T-a) + bwhere k, a and b are constants solved numerically to satisfy the following conditions:f1(T) = 1; at T = Tsta∂f1(T)/∂T = 0; at T = Toptf1(T) = 0; at T = Tmaxf2(T) = θT-20Limitation equationsf(Z)i=(ΣPj+ΣZk+POC)/(Ki+ΣPj+ΣZk+POC)f(I)j = I/Is exp(1-I/Is)f(IP)j = [IPmax/(IPmax-IPmin)] [1-IPmin/IP]f(IN)j = [INmax/(INmax-INmin)] [1-INmin/IN]f(DO) = DO/(KDO+DO)f(P) = PO4/(KPO4+PO4)f(N) = (NH4+NO3)/(KN2+NH4+NO3)PN = (NH4 NO3)/[(NH4+KN)(NO3+KN)] + (NH4 KN)/[(NH4+KN)(NO3+KN)]SettlingSj = (ws/Δz)PjSPOM = (g(ρPOM - ρw)(DPOM)2/18μ)/Δz)POMPredationPredi = Σ(Gkf(Z)kf1(T)kZkPzZOOk,i)Predj = Σ(Gif(Z)if1(T)iZiPzPHYi,j)Abbreviations: Z, zooplankton; P, phytoplankton; POC, particulate organic carbon; DOC, dissolved organic carbon; POP, particulate organic phosphorus; PO4, phosphate; PON, particulate organic nitrogen; NH4, ammonium; NO3, nitrate; POM, particulate organic matter (C, N or P); IPzi, zooplankton internal phosphorus; INzi, zooplankton internal nitrogen; IPj, phytoplankton internal phosphorus; INj, phytoplankton internal nitrogen; DO, dissolved oxygen; DOatm, concentration of oxygen in the atmosphere; LA, layer area; LV, layer volume; Δz, layer thickness; ρw, density of water; μ, viscosity of water; kO2, oxygen transfer coefficient. Subscripts: i, zooplankton group; j, phytoplankton group; k, zooplankton predator group.The major nutrient fluxes represented in CAEDYM are uptake of dissolved inorganic nutrients by phytoplankton, release of dissolved nutrients from phytoplankton excretion, grazing, egestion and excretion of nutrients by zooplankton, nitrification and denitrification of inorganic nitrogen, sedimentation of nutrients in particulate form, mineralization of organic nutrients and release of dissolved nutrients from sediments (Table 1).Net change in carbon concentration of the phytoplankton at each model time step is calculated as the difference between the increment due to gross primary production and losses due to sedimentation, grazing by zooplankton, respiration, excretion and mortality. These terms are calculated using equations parameterized to represent the physiology of the main phytoplankton species. Losses due to grazing by zooplankton are calculated by multiplying the food assimilation rate for the zooplankton by a preference factor for phytoplankton over detrital POC.Net zooplankton growth is calculated as a balance between food assimilation and losses from respiration, excretion, egestion, predation and mortality. Food assimilation is calculated as the product of the maximum potential rate of grazing, assimilation efficiency, and temperature and food concentration functions. A constant internal nutrient ratio is assumed and excretion of nutrients calculated to maintain this ratio at each time step. Advective movement of zooplankton is carried out in DYRESM.Bacteria have not been directly simulated as they were not measured during the study period. However the nutrient pathways mediated by bacteria were included as mineralization of the particulate organic pools (POC, POP and PON). The POC, POP and PON pools available for zooplankton grazing include bacteria. Predation of zooplankton by grebes was included by an additional predation term for the months of August to November estimated from predation studies (Cooper et al. 1984).The advantage of using a depth resolved model DYRESM linked to the ecological model CAEDYM is that we could explore the effect of transport and mixing between the epilimnion, metalimnion and hypolimnion on the ecological processes in the lake. Of most relevance to this study is the exchange of nutrient-rich hypolimnetic waters to the photic zone via the BBL and its consequential effect on the primary productivity. The separation of internal and BBL cells in the layered structure of the current DYRESM allowed us to differentiate between the transport of nutrients in the internal and BBL and to determine the relative importance of each process on the mixed layer ecological dynamics. The ecological model also returns the attenuation coefficient (as a function of the concentration of both phytoplankton and particulate organic matter) to the hydrodynamic model at each one-hour time step. This variable is used to determine the extent of light and heat penetration that in turn governs the deepening of the surface mixed layer and the timing of winter turnover. In this way the feedback on a sub-daily time scale between the ecological and physical models is instrumental in the application of the model to aid in understanding of the interaction of various lake processes.Field sampling and analytical analysisSeasonal and year-to-year variations in the physical, chemical, and biotic environments were monitored fortnightly from March through October and monthly during November through January. Water temperature and conductivity were measured at nine buoyed, pelagic stations (2, 3, 4, 5, 6, 7, 8, 10 and 12) (Fig. 1). Profiles were taken with a high-precision, conductivity-temperature-depth profiler (CTD) (Seabird Electronics model Seacat 19) equipped with a submersible photosynthetically available radiation (PAR) (LiCor 191S), fluorescence (695 nm) (WETLabs WETStar miniature fluorometer), and transmissivity (660 nm) (WETlabs C-Star Transmissometer). Specific conductivity, salinity, and density were all calculated based on equations derived from measurements on Mono Lake brine [30]. Dissolved oxygen was measured at one centrally located station (Station 6) with a Yellow Springs Instruments temperature-oxygen meter (YSI, model 58) and probe (YSI, model 5739). The oxygen electrode was calibrated at least once each year against Miller titrations of Mono Lake water (Walker et al. 1970).Ammonium and chlorophyll profiles were determined by sampling 7–10 discrete depths at two pelagic stations (2 and 7; Fig. 1), while A. monica abundance was determined via vertical net tows collected at 10 (1991–1992) or 20 (1993–1994) pelagic stations (Fig. 1). Chlorophyll a was also determined in the upper water column from samples collected with a 9-m integrating tube sampler at 5 pelagic stations (2, 6, 7, 10, 11; Fig. 1).Nutrient and phytoplankton samples were immediately passed through a 120-μm net to remove all stages of A. monica and a sub-sample filtered through Gelman A/E glass fiber filters for analysis of nutrients (Jellison and Melack 1993a). Ammonium concentrations were measured with the indophenol blue method as described by Jellison et al. (1993). Nitrate and nitrite concentrations were measured but were always low (< 1 μM) [21,31] and thus not considered in this study. Phosphate concentrations are orders of magnitude greater than the half saturations constants for phytoplankton so were not considered in this study [21]. Phytoplankton chlorophyll a was determined by spectrophotometric analysis as described by Jellison and Melack (1993a). Conversion of chlorophyll to carbon units were made by assuming a C:Chl a ration of 50 (see Jellison & Melack 2001). Subsamples were filtered onto precombusted Gelman A/E filters for the determination of particulate organic carbon (POC) and nitrogen (PON). Duplicate carbon and nitrogen filters were acid fumed for 12 hours over concentrated HCl, and then dried at 40–50°C before determination by combustion in a Perkin-Elmer 240B elemental analyzer standardized with acetanilide. A. monica were collected using vertical net tows (120-μm mesh) to within 1-m of the bottom or well below the oxycline depending on stratification. A. monica biomass (dry weight) was estimated from stage-specific abundance, adult female length data, and weight-length relationship determined in the laboratory simulating in situ conditions of food and temperature [32]. Conversion from dry weight to carbon assumed 0.4 g C/g dry weight [33,34].Model inputsModel input files included data for initialization, meteorology, inflows and outflows. The initialization file was prepared from field data collected on 13 January 1991. On this day the temperature of the lake ranged from 2.5 to 3°C and the salinity from 88 to 88.5 g kg-1. Inflow data included the daily volume, temperature and salinity for two inflows, one representing total surface inflows (streams and direct runoff) and the other, hydrothermal springs. The volume of the hydrothermal springs was set at 3888 m3 day-1, based on a 3He mass balance of Mono Lake [35]. The surface inflows were calculated based on a water mass balance using measured values of water depth and evaporation calculated by DYRESM. As ammonium, phytoplankton and zooplankton concentrations are negligible in the inflows, they were set to zero (Jellison and Melack 2001). Meteorological input data included hourly short- and long-wave radiation, air temperature, vapor pressure, wind speed and precipitation [36]. Air temperature, vapor pressure (converted from relative humidity), wind speed and precipitation were collected at a meteorological station located on Paoha, a central island (Fig. 1). Radiation data were collected from a meteorological station located approximately 7 km from the southwest shore of the lake (Fig. 1).The physical parameters used to simulate the hydrodynamics of Mono Lake were either physical constants or ones fixed according to the dimensions of the lake [4].The formulation of CAEDYM used here to describe the ecological variables and processes required 57 parameters determined by several methods (Table 2). Most phytoplankton parameters were derived from experimental analysis on the predominant phytoplankton species of Mono Lake [21,37]. The zooplankton parameters were determined where available from experiments conducted on A. monica or alternative Artemia species (Table 2). If parameters were not available, a series of model runs were performed to calibrate the simulation results against field data, maintaining parameter values within the bounds of literature values measured in other lakes.Table 2Model parameters. Parameters used in CAEDYM to simulate ecological variables in Mono Lake.GeneralParameterDescriptionUnitsAssigned valueValues from field/litKdBackground extinction coefficientm-10.350.29–0.34aSourcea Calculated from unpub data on in-situ light measurementsPhytoplanktonParameterDescriptionUnitsAssigned values:Values from field/literaturePmaxMaximum potential growth rated-15.967.2aIKParameter for initial slope of PI curveμEm-2s-12525bKepSpecific attenuation coefficientm2 g C-10.0080.008cKPHalf saturation constant for phosphorus uptakemg L-10.001Low value as not P limitedKNHalf saturation constant for nitrogen uptakemg L-10.0573CalibratedINconConstant internal N ratiomg N (mg C)-10.09260.17dIPconConstant internal P ratiomg P (mg C)-10.0260.048dθjTemperature multiplier for growth1.061.07eTstaStandard temperature°C19ToptOptimum temperature°C22TmaxMaximum temperature°C39.5RjMetabolic loss rate coefficientd-10.302CalibratedθRTemperature multiplier for metabolic loss1.05CalibratedfresFraction of respiration relative to total metabolic loss0.693CalibratedfDOMFraction of metabolic loss rate that goes to DOM0.291CalibratedwsSettling velocitym d-10.0080.04–0.013fSourcesaJellison and Melack 1993a, based on maximum value of carbon uptake measured from lake samples 1983–1990 assuming 50 g C g Chl a-1b Jellison and Melack 1993a, based on minimum value of IK measured from lake samples 1983–1990.cJellison and Melack 1993a.dJellison and Melack 2001, estimated from seston ratios during the summer period from monomictic years 1991–1995 1984eJellison and Melack 1993a, based on Q10 of 1.95.fJellison et al.. 1993.ZooplanktonParameterDescriptionUnitsAssigned values:Values from field/literatureGiGrazing rateg C m-3 (g C m-3)-1 d-11.121.26aAziGrazing efficiency-1.0Close to 1 as filter feedersRiRespiration rate coefficientd-10.1130.035–0.1bMiMortality rate coefficientd-10.01070.0033c 0.0262dfegFecal pellet fraction of grazingd-10.096Kfz+kez = 0.36–0.68efexExcretion fraction of grazingd-10.49DOmzMinimum DO tolerancemg L-10.00–1.2fθiTemperature multiplier for growth1.0551.22gTminMinimum temperatureDeg C66.8–9.0hθRiRespiration temperature dependence1.10KiHalf saturation constant for grazingg C m-31.122.96iINziInternal ratio of nitrogen to carbon.g N g C-10.2080.197/0.218jIPziInternal ratio of phosphorus to carbong P g C-10.020.0135kPzPHYPreference of zooplankton for phytoplankton0.8PzPOCPreference of zooplankton for POC0.2Sourcesa [33,44] (Artemia fransiscana optimal food 11 days old)b[33,44] (Artemia fransiscana range of food 11 days old)cJellison et al.. 1993 (based on survival rate over 30 days)dDana and Lenz 1986 (based on survival rate over 26 days)eEvjemo et al.. 2000, Artemia fransiscana.fDO concentration at depth of deep Chl a maxima (unpub data).gJellison et al.. 1993 (best fit to temperature function used in model)h Jellison unpub data 1991–1994. (based on temperature at which total biomass < 0.01 before Spring growth)i Evjemo and Olsen 1999 (Artemia fransiscana 11 days old, 26–28°C, Holling Type II)j Jellison unpub data (Females/Males)k [45]Dissolved Oxygen and NutrientsParameterDescriptionUnitsAssigned valuesValues from field/literatureSdDODO sediment exchange rateg m-2d-10.053KDO_sedHalf saturation constant for DO sediment fluxmg O L-10.537KDO_POMHalf saturation constant for dependence of POM/DOM decomposition on DOmg O L-11.46fanBAerobic/anaerobic factor-0.357θPOMTemperature multiplier-1.031.02–1.14aRPOCMineralisation rate for POC to DOCd-10.12RPOPMineralisation rate for POP to DOPd-10.10.01–0.1aRPONMineralisation rate for PON to DONd-10.40.01–0.03aDPOMDiameter of POM particlesm0.000009ρPOMDensity of POM particleskg m-31109KePOCSpecific light attenuation coefficient for POCm2 g-10.00943RDOCMineralisation rate for DOCd-11Set to 1 to eliminate DOP pool for simplicityRDOPMineralisation rate for DOP to PO4d-11Set to 1 to eliminate DOP pool for simplicityRDOPMineralisation rate for DOP to PO4d-11Set to 1 to eliminate DOP pool for simplicityRDONMineralisation rate for DON to NH4d-11onset to 1 to eliminate DON pool for simplicity.KeDOCSpecific light attenuation coefficient of DOCm2 g-10.001RN2Denitrification rate coefficientd-10.0008640.1aθN2Temperature multiplier for denitrification-1.081.045aKN2Half saturation constant for denitrification dependence on oxygenmg N L-11.75RNONitrification rate coefficientd-10.005530.1–0.2aθNOTemperature multiplier for nitrification-1.081.08aKNOHalf saturation constant for nitrification dependence on oxygenmg O L-10.5θsedTemperature multiplier for sediment nutrient fluxes-1.05SdNH4Release rate of NH4 from sedimentsg m-2 d-10.07120.054–0.18bKDO_SdNH4Controls sediment release of NH4 via oxygen – Half saturation constant for sediment NH4 release dependence on DOg m-30.565SourcesaJorgensen and Bendoricchio 2001b Jellison et al.. 1993Table 3Normalised mean absolute error. VariableNMAESD/Meanr2SlopeNH40.58 (0.56)0.88 (0.72)0.37 (0.59)0.29 (0.61)Phytoplankton0.45 (0.44)1.15 (1.04)0.79 (0.85)0.61 (0.80)Artemia monica0.30 (0.30)0.85 (0.95)0.79 (0.83)0.80 (0.87)TPON0.35 (0.34)0.83 (0.82)0.90 (0.94)0.52 (0.57)TPOC0.43 (0.42)0.91 (0.88)0.86 (0.92)1.02 (1.15)Dissolved oxygen0.31 (0.31)0.48 (0.48)0.64 (0.64)0.32 (0.32)Average0.40 (0.40)0.85 (0.82)0.72 (0.79)0.59 (0.72)Results of normalised mean absolute error (NMAE) calculations applied to compare simulated to field data for simulated years 1991–1994. The values in brackets represent the same calculations made over the 1991–1992 calibration period.Table 4Sensitivity analysis. ParameterOptimalLower boundUpper boundNMAE (lower bound)NMAE (upper bound)SdNH40.060.010.100.490.42fex0.500.050.700.490.44Kd0.300.350.250.410.43INcon0.090.070.220.470.49feg0.160.050.200.470.42The minimum and maximum values of the five most sensitive parameters and the corresponding results of normalised mean absolute error (NMAE) calculations applied to compare simulated to field data for simulated years 1991–1994. The values in brackets represent the same calculations made over the 1991–1992 calibration period.A variety of quantifiable measures of model fit are described in Alewell and Manderscheid (1998). We choose the average absolute error normalized to the mean (NMAE):(3)NMAE=∑t=1n(|st−ot|)no¯where st is the simulated value at time t, ot is the observed value at time t, ō is the mean of the observed values over the simulation period and n is the number of observed values. NMAE is a measure of the absolute deviation of simulated values from observations, normalized to the mean; a value of zero indicates perfect agreement and greater than zero an average fraction of the discrepancy normalized to the mean. To compare the extent of variability within the observed data, we also calculated for each state variable, the standard deviation of observed data normalized to the mean over the simulation period (Table 3). In addition, the correlation coefficients and associated slope for the direct comparison of observed against simulated values for each state variable were calculated.The period from 1991–1992 was used for initial parameter calibration. Comparisons of field and model data were made for six major model state variables: (1) ammonium (NH4); (2) particulate organic nitrogen (PON); (3) particulate organic carbon (POC); (4) dissolved oxygen (DO); (5) phytoplankton carbon and (6) zooplankton carbon. PON and POC refer to the sum of phytoplankton and detrital particulate nitrogen and carbon, respectively. Lake-wide averages of the surface to 9-m integrated concentrations of NH4, PON, POC, DO and phytoplankton carbon were compared. For zooplankton, lake-wide averaged biomass (g C m-2) as determined by vertical net tows were compared to vertically-integrated model output.A manual calibration procedure was initially applied whereby individual parameters were adjusted and the model response observed. The particular features of the observed data that were used to adjust individual parameters were dependent on the parameter adjusted. For example, the minimum temperature for A. monica growth was adjusted to gain best representation of the timing of the spring zooplankton peak and the grazing rate adjusted to gain best representation of the magnitude of this peak. Individual parameters were adjusted in this way until an overall model average NMAE (calculated using the field data from the five variables listed above) of less than 0.5 was achieved.Once a reasonable fit was achieved through trial-and-error, the local parameter space optima was determined by applying a Levenberg-Marquant (L-M) method of optimization [38] using a predefined Matlab® function (The MathWorks Inc., Natick, MA). In this function, parameters were adjusted to optimize the sum of the NMAE values for the same five variables listed above. Additional bounds were placed on simulated values of primary productivity and nitrogen sedimentation to fall within the ranges of those estimated by Jellison et al. (1993). The years 1993–1994 were used for model validation.Uncertainly in model predictions arises from different sources including those associated with process representation, parameter estimation, uncertainty in inputs and observed data [39-42]. While a full analysis of model uncertainty is beyond the scope of this paper, we made an estimate of the uncertainty associated with parameter estimation by comparing output from simulations using ten different parameter sets. Initially, the five most sensitive parameters to model output were established by the sensitivity analysis described below. We then determined alternative parameter sets by fixing the lower and upper bound of each parameter and then optimizing the remaining parameters via the L-M method described above until appropriate calibration was achieved. A benchmark NMAE value of 0.5 was selected so that calibration was deemed successful if the NMAE was less than 0.5 (Table 4). The lower and upper bounds for each sensitive parameter were determined by experimental or literature ranges. Model output from this suite of parameter sets was then used as an estimate of the relative uncertainty in model output. As the upper and lower bound of the five most sensitive parameters were used, this should provide a conservative estimate of the model uncertainty associated with parameter estimation.To determine the five parameters most sensitive to model output, a sensitivity analysis was performed on each of the CAEDYM parameters listed in Table 2. Sensitivity coefficients (sij) to assess the relative sensitivity of variable i to parameter j were calculated according to:sij=Δcic¯iΔβjβ¯jwhere Δcj is the change in output variable i from the reference value ci and Δβj is the change in parameter j from the reference value βj [43]. Because this study is concerned with the role of physical transport mechanisms on lake-wide nitrogen fluxes, we focused on the response of the five major nitrogen fluxes (phytoplankton uptake, sediment flux, zooplankton regeneration, settling and upward flux into surface mixed layer) to parameter manipulation. Each parameter was adjusted by ± 10% or by ± 0.01 in the case of the temperature multipliers. A time variant array of sensitivity parameters was calculated for each flux and then the average taken and used to rank the parameters according to sensitivity.To enable us to quantify the significance of BBL transport on the ecological processes of the lake, a series of simulations were run in which this mechanism was switched off allowing a comparison between lake behavior with and without BBL transport. In DYRESM the lake-wide vertical fluxes are partitioned into internal and BBL contributions (see eq.36 in Yeates and Imberger (2004)). The BBL contribution was set to zero with all other parameters (physical and ecological) remaining the same. The simulated output of nitrogen fluxes and primary and secondary production were then analyzed and compared against base line output.List of abbreviations usedBBL: Benthic boundary layer; CAEDYM: Computational aquatic ecosystem dynamic model; DYRESM: Dynamic reservoir simulation model.Declaration of competing interestsThe authors declare that they have no competing interests.Authors' contributionsAll authors participated in the conception and design of the study. RJ collected data used in model validation while LCB conducted model simulations and drafted the initial manuscript. All authors contributed to revising and editing the manuscript, and have read and approved the final version.\n\nREFERENCES:\n1. MacIntyreSFlynnKMJellisonRRomeroJRBoundary mixing and nutrient fluxes in Mono Lake, CaliforniaLimnology and Oceaonography1999443USA , American Society of Limnology and Oceanography512529\n2. OstrovskyIYakobiYZWallinePKalikhmanISeich-induced mixing: Its impact on lake productivityLimnology and Oceanography1996412323332\n3. EckertWImbergerJSaggioABiochemical response to physical forcing in teh water column of a warm monomictic lakeBiogeochemistry20026129130710.1023/A:1020206511720\n4. YeatesPSImbergerJPseudo two-dimensional simulations of internal and boundary fluxes in stratified lakes and reservoirs.International Journal of River Basin Research200414123\n5. GoudsmitGHPeetersFGloorMWuestABoundary versus diapycnal mixing in stratified natural watersJournal of Geophysical Research199710227903-27914\n6. LorkeAPeetersFWuestAShear-induced convective mixing in bottom boundary layers on slopesLimnology and Oceanography200550516121619\n7. WuestALorkeASmall-scale hydrodynamics in lakesAnnual review of Fluid Mechanics20033537341210.1146/annurev.fluid.35.101101.161220\n8. ImbergerJImberger JFlux paths in a stratified lake: A review.Physical Processes in Lakes and Reserviors1998 American Geophysical Union117\n9. ImbergerJPattersonJCPhysical limnologyAdvanced Applied Mechanics199027303475\n10. Pedlosky2nd Geophysical fluid dynamics1987 Springer\n11. HudsonJJTaylorWDMeasuring regeneration of dissolved phosphorus in planktonic communities.Limnology and Oceanography199641715601565\n12. LehmanJTRelease and cycling of nutrients between planktonic algae and herbivores.Limnology and Oceanography1980254620632\n13. WenYHPetersRHEmpirical models of phosphorus and nitrogen excretion rates by zooplankton.Limnology and Oceanography199439716691679\n14. TonnoIOttKNogesTNitrogen dynamics in the steeply stratified, temperate Lake Verevi, EstoniaHydrobiologia2005547637110.1007/s10750-005-4145-y\n15. CaracoNFColeJJLikensGENew and recycled primary production in an oligotrophic lake: Insights for summer phosphorus dynamics.Limnology and Oceanography1992373590602\n16. JellisonRMillerLGMelackJMDanaGLMeromixis in hypersaline Mono Lake, California. 2. Nitrogen fluxesLimnology and Oceaonography19933810201039\n17. RomeroJRMelackJMSensitivity of vertical mixing in a large saline lake to variations in runoff.Limnology and Oceanography1996415955965\n18. JellisonRMelackJMNitrogen limitiation and particulate elemental ratios of seston in hypersaline Mono Lake, California, U.S.A.Hydrobiologia200146611210.1023/A:1014525805934\n19. DanaGLJellisonRMelackJMEffects of different natural regimes of temperature and food on survival, growth and development of Artemia monica VerrillJournal of Plankton Research199517112117213010.1093/plankt/17.11.2117\n20. JellisonRMelackJMPhotosynthetic activity of phytoplankton and its relation to environmental factors in hypersaline Mon oLake, CaliforniaLimnology and Oceanography19881586988\n21. JellisonRMelackJMAlgal photosynthetic activity and its response to meromixs in hypersaline Mono lake, CaliforniaLimnology and Oceaonography1993384USA , American Society of Limnology and Oceanography, Inc.818837\n22. DanaGLLenzPHEffects of increasing salinity on Artemia population from Mono Lake, California.Oecologia19866842843610.1007/BF0103675128311791\n23. JellisonRDanaGLMelackJMZooplankton cohort analysis using systsm identification techniques.Journal of Plankton Research199517112093211510.1093/plankt/17.11.2093\n24. BruceLCHamiltonDPImbergerJGalGGophenMZoharyTHambrightKDA numerical simulation of the role of zooplankton in C, N and P cycling in Lake Kinneret, Israel.Ecological Modelling2005In press\n25. UmlaufLLemminUInterbasin exchange and mixing in the hypolimnion of a large lake: The role of long internal wavesLimnology and Oceanography200550516011611\n26. PelletierJPPhytoplankton as the basis of the trophic web in Lake GenevaArchives des Sciences19985113744\n27. TilzeMMGaedkeUSchweizerABeeseBWieserTInterannual variability of phytoplankton productivity and related parameters in Lake Constnace - No response to decreased phosphorus loadingJournal of Plankton Research199113475577710.1093/plankt/13.4.755\n28. RomeroJRAntenucciJPImbergerJOne- and three-dimensional biogeochemical simulations of two differing reservoirs.Ecological Modelling200417414316010.1016/j.ecolmodel.2004.01.005\n29. RobsonBJHamiltonDPThree-dimensional modelling of a Microcystis bloom event in the Swan River estuary, Western Australia.Ecological Modelling200417420322210.1016/j.ecolmodel.2004.01.006\n30. JellisonRMacIntyreSMilleroFThe density characteristics of Mono Lake brineInt J Salt Lake Res19998415310.1007/BF02442136\n31. JellisonRMelackJMMeromixis in hypersaline Mono Lake, California. 1. Stratification and vertical mixing during the onset, persistence, and breakdown of meromixisLimnology and Oceanography19933810081019\n32. DanaGLJellisonRMelackJMRelationships between Artemia monica life history characteristics and salinityHydrobiologia199326312914310.1007/BF00006264\n33. EvjemoJOOlsenYEffect of food concentration on the growth and production rate of Artemia franciscana feeding on algae (T-iso)Journal Of Experimental Marine Biology And Ecology1999242227329610.1016/S0022-0981(99)00104-5\n34. OlsenAIAttramadalYJensenAOlsenYInfluence of size and nutritional value of Artemia franciscana on growth and quality of halibut larvae (Hippoglossus hippoglossus) during the live feed periodAquaculture19991791-447548710.1016/S0044-8486(99)00181-7\n35. ClarkJFHudsonGBQuantifying the flux of hydrothermal fluids into Mono Lake by use of helium isotopes.Limnology and Oceanography2001461189196\n36. RomeroJRMelackJMSensitivity of vertical mixing in a large saline lake to variations in runoff. . 41: 955–965. Limnol Oceanogr199641955965\n37. RoeslerCSCulbertsonCWEtherideSMGoerickeRKieneRPMillerLMOremlandRSDistribution , production, and ecophysiology of Picocystis strain ML in Mono Lake, California.Limnology and Oceanography200247440452\n38. MarquardtDWAn algorithim for least-square approximation of nonlinear parametersJournal for the Society for Industrial and Applied Mathematics196311243144110.1137/0111030\n39. OmlinMReichertPForsterRBiogeochemical model of Lake Zürich: model equations and results.Ecological Modelling20011417710310.1016/S0304-3800(01)00256-3\n40. SolidoroCCriseACrispiGPastresRAn a priori approach to assimilation of ecological data in marine ecosystem modelsJournal of Marine Systems200340-41799710.1016/S0924-7963(03)00014-9\n41. AlewellCManderscheidBUse of objective criteria for the assessment of biogeochemical ecosystem models.Ecological Modelling199810721322410.1016/S0304-3800(97)00218-4\n42. PowerMThe predictive validatiaon of ecological and environmental modelsEcological Modelling199368335010.1016/0304-3800(93)90106-3\n43. ChenCJiRSchwabDJBeletskyDFahnenstielGLJiangMJohengenTHVanderploegHEadieBWells BuddJBundyMHGardnerWCotnerJLavrentyevPJA model study of the coupled biological and physical dynamics of Lake MichiganEcological Modelling200215214516810.1016/S0304-3800(02)00026-1\n44. EvjemoJOVadsteinOOlsenYFeeding and assimilation kinetics of Artemia franciscana fed Isochrysis galbana (clone T. Iso)Marine Biology200013661099110910.1007/s002270000306\n45. MartinJHPhytoplankton-Zooplankton Relationships In Narragansett Bay .3. Seasonal Changes In Zooplankton Excretion Rates In Relation To Phytoplankton AbundanceLimnology Oceanography19681316371"
+}

batch_5/PMC2535594.json

@@ -0,0 +1,4 @@
+{
+    "id": "PMC2535594",
+    "text": "This is an academic paper. This paper has corpus identifier PMC2535594\nAUTHORS: Liu Yang, Janet C Effler, Brett L Kutscher, Sarah E Sullivan, Douglas N Robinson, Pablo A Iglesias\n\nABSTRACT:\nBackgroundMany cellular processes involve substantial shape changes. Traditional simulations of these cell shape changes require that grids and boundaries be moved as the cell's shape evolves. Here we demonstrate that accurate cell shape changes can be recreated using level set methods (LSM), in which the cellular shape is defined implicitly, thereby eschewing the need for updating boundaries.ResultsWe obtain a viscoelastic model of Dictyostelium cells using micropipette aspiration and show how this viscoelastic model can be incorporated into LSM simulations to recreate the observed protrusion of cells into the micropipette faithfully. We also demonstrate the use of our techniques by simulating the cell shape changes elicited by the chemotactic response to an external chemoattractant gradient.ConclusionOur results provide a simple but effective means of incorporating cellular deformations into mathematical simulations of cell signaling. Such methods will be useful for simulating important cellular events such as chemotaxis and cytokinesis.\n\nBODY:\nBackgroundMany cellular processes are characterized by substantial shape changes. For example, chemotaxing cells become polarized, assuming a highly elongated form, and crawl across solid substrates in the direction of increasing concentrations of chemoattractant [1]. During cytokinesis, a single cell undergoes significant cytoskeletal deformation, reforming into two daughter cells [2]. These cellular processes are fundamentally mechanical, utilizing force generation at the molecular scale to generate shape changes. Properly simulating cellular shape change requires that we have a description of the underlying mechanical properties of the cell.To understand fully the mechanisms that regulate these cell shape changes requires knowledge of the signaling pathways as well as their effect on the mechanical properties of cells. For example, a complete model of chemotaxis would require a description of the gradient sensing capability of cells together with a physical model for the cellular migration [3]. Few such models exist, even though it is now appreciated that the response of cell-signaling pathways can be regulated in response to alterations in cell size and shape [4]. The traditional method of simulating cellular deformations is by specifying the boundary of the cell explicitly through a finite-element model (FEM) [5-7]. One problem is that simulation of biological shape deformations – which invariably involves solving partial differential equations on moving boundaries – can be computationally expensive particularly when the cellular deformations are not small. During many processes including cytokinesis and chemotaxis, cellular shape deformations tend to be large and occur rapidly. Here, we demonstrate how the Level Set Method (LSM) can be used to couple mechanical models of the cell with biochemical models of signaling pathways to simulate large cellular deformations.We briefly contrast the LSM approach to other methods that have been used to account for cellular deformations.The immersed boundary method (IBM), introduced by Peskin [8] was developed to simulate the interaction of flexible tissues with the surrounding incompressible fluid. It has been used to simulate cell shape changes during motility [9]. In the IBM, the Navier-Stokes equation describing the fluid flow can be solved on a fixed grid, simplifying this computationally expensive step. The membrane and cytoskeleton is discretized by assigning a series of nodes that are connected by viscoelastic elements. As the cell deforms, nodes and their corresponding links have to be inserted or deleted. This book-keeping comes at a considerable computational complexity. For this reason, the IBM may best be used in situations where the cell shape does not change considerable [10].More recently, the cellular Potts model (CPM) has become a popular vehicle to simulate cell shape changes [11]. In the CPM, a cell is described by a connected domain of pixels on a regular grid. The shape of the cell is evolved by updating each pixel based on a set of probabilistic rules. This method does not use an explicit viscoelastic description of the cell. Instead, cell shape is constrained by minimizing an energy function that penalizes size-deformations as well as membrane bending. Cellular Potts models have been used to simulate two-dimensional (2-D) models of cell motility in fish keratocytes [12] and amoebae [13]. Unlike FEM or IBM, modeling large changes in the shape of the cell is no more computationally expensive than small changes. One drawback, however, is that the mechanical description of cells in the CPM framework is not as tightly integrated with experimentally-based measurements as the method presented here.Models of cellular shape changes have all been derived based on explicit descriptions of the cell morphology that are updated based on the simulated behavior of the underlying cytoskeleton. For example, Rubinstein et al. provide a detailed 2-D computational model of the lamellipodium keratocyte motility [14]. In this model, the cellular domain is updated at each time step based on the protrusive and retractive forces (actin polymerization and acto-myosin contraction) and re-gridded. This avoids the necessity for nodes and keeping track of the mechanical state of the system. However, the model relies on an elastic (rather than viscoelastic) network which may be appropriate for the thin keratocyte, but is not likely to be applicable to thicker cells.The rest of the paper is organized as follows. We first provide some necessary background. We then develop a mechanical model within the LSM that accounts for the viscoelastic nature of the cell. We fit this model to experimental data obtained through micropipette aspiration experiments on Dictyostelium cells. We incorporate this viscoelastic model into a level set framework and illustrate how large-scale shape deformations can be accounted by the model. This is done through simulations by showing that the model accurately captures the behavior of the aspirated cell. Finally, using a simple gradient sensing model to generate internal force profiles, we simulate the changing morphology of a cell chemotaxing in response to an externally applied chemoattractant gradient. Using the framework developed here, we obtain the force profiles needed to achieve stable migrating cell morphologies observed for several strains. The methods developed here allow us to link forces acting on the cell and mechanical properties of the cytoskeleton to cell shape deformation explicitly, and will prove useful in studying cellular processes undergoing large-scale shape changes.Biological backgroundCells derive their mechanical properties from actin, actin-associated proteins, and motor proteins such as myosin-II [2], which are components of the cytoskeleton. Though distributed throughout the cell, the actin cytoskeleton is concentrated along the periphery of the cell underneath the membrane, particularly in Dictyostelium, and is the molecular machinery that generates cellular shape changes during cell division and chemotaxis.Cytoskeletal networks exhibit viscoelastic behavior, having both viscous and elastic properties [15-17]. Actin filaments alone do not create significant mechanical resistance; instead, cross-linking of actin filaments by various actin binding proteins imparts mechanical rigidity to the cell. Under applied load, cross-linked actin networks behave similarly to an elastic solid and can be described using Hooke's law. However, because cross-linking proteins bind to and dissociate from actin filaments, actin-based networks may also exhibit viscous flow. Myosin-II, in filament form, also binds to actin filaments and provides mechanical resistance of the cell, as well as influencing the binding kinetics of various actin crosslinkers [2,18]. The interior of the cell also contains cytoskeletal polymers, as well as organelles, a nucleus, and cytoplasmic fluid. Thus, owing to their viscoelastic nature, cells exhibit a time-dependent deformation in response to mechanical force.Introduction to level set methodsCell motion has been traditionally simulated by: discretizing the cell boundary, computing the displacement of each of the points according to the local velocity, and forming a new boundary with the displaced points (Fig. 1A). This method may run into difficulties when the spatial or temporal resolution of the simulations is not sufficiently fine, or when changes in topology occur (Fig. 1B). The Level Set Method (LSM) can be used to overcome these difficulties [19]. LSM is a numerical technique for tracking interfaces and shapes which has been widely used in various fields including computer graphics [20], image processing [21], computational fluid dynamics [22] and material science [23].Figure 1Introduction to level set methods. A. The traditional method of tracking moving boundaries involves discretization of the boundary (dotted red) into a set of points, moving each point x = (x, y) according to the local velocity (v(x, t)), leading to a new boundary at the new locations (solid red). B. Difficulties can arise, however, when the geometry of the boundary becomes irregular. In this case, the point tracking method often fails to preserve the boundary topology. Special attention is required to resolve these errors, increasing computational costs. C. In the Level Set Method (LSM), the boundary Γ(t) is embedded into a higher dimensional potential function (ϕ(x, t)) as the zero-contour. Γ(t)moves as ϕ(x, t)evolves in time. D. Because the boundary is defined implicitly, the LSM framework overcomes some of the difficulties of boundary point tracking. E. This example illustrates how an arbitrary cell shape (black contour) can be embedded into a signed distance function to form the level set potential function ϕ(x, t). In this case, the potential function is given by the Euclidean distance to the cell boundary, with positive (resp. negative) sign when outside (resp. inside) the cell.Suppose that the cell boundary at time t is described by the closed-contour Γ(t). The LSM requires a potential function (Fig. 1C.), denoted as ϕ(x, t), that is related to Γ(t) according to:Γ(t) = {x|ϕ(x, t) = 0}.Thus, Γ(t) is the zero-level set of ϕ(x, t). It follows that, in the LSM, the cell membrane is represented implicitly through the potential function which is defined on a fixed Cartesian grid, thus eliminating the need to parameterize the boundary. This allows the LSM to handle complex boundary geometries efficiently (Fig. 1D).One candidate for the potential function is the signed distance function [24], defined by:(1)signd(x,Γ)={−d(x,Γ),d(x,Γ),0,if x∈S,if x∉S,if x∈Γ,where S identifies the area occupied by the cell and d(x, Γ) is the distance of position x to the curve Γ; see Fig. 1E for an example of a cell shape embedded in a potential function derived from the signed distance function.We now manipulate Γ(t) implicitly through the function ϕ(x, t) according to the equation:(2)∂ϕ(x,t)∂t+v(x,t)⋅∇ϕ(x,t)=0.The vector v(x, t) is the velocity of the level set moving in the outward normal direction. In our case, v(x, t) intrinsically describes the cell's membrane protrusion and retraction velocities. These velocities can be driven by externally applied forces on the cell membrane (e.g. from a micropipette aspirator), or internally generated mechanical forces (e.g. actin polymerization or myosin-II retraction), or both. To determine how these forces translate to membrane velocity, however, first requires a mechanical model of the cell.As the potential function corresponding to the cell shape is evolved, it can become quite steep or flat. To reduce the numerical errors caused by these effects, we re-initialize the potential function periodically [24]. This can be done using the re-initialization equation [25]:∂ϕ(x,t)∂t=S(ϕ(x,0))(|∇ϕ(x,t)|−1),where S(ϕ(x, 0)) is taken as +1 inside the cell, -1 outside the cell and zero on the cell membrane.Results and DiscussionViscoelastic model of cell deformationThe LSM relies on a continuum description of the material properties of the cell [2,26]. We use mechanical models to describe the viscoelastic behavior of the cell [27]. Our mechanical model is based on a representation of cells that assumes a viscoelastic cortex surrounding a viscous core. For cells where intracellular components, such as the nucleus, take a considerable fraction of the cellular volume and play an active role in determining cell shape, the method described here will not be applicable without explicitly modeling these internal structures.We model the cortex connecting the cell membrane and the cytoplasm with a Voigt model, which consists of the parallel connection of an elastic element kc (nN/μm3) and a viscous element τc (nNs/μm3). The latter describes the association and dissociation dynamics of the cross-linkers. We model the cytoplasm by a purely viscous element, τa (nNs/μm3), which is placed in series with the Voigt model (Fig. 2A). The element τa includes contributions from the interior of the cell as well as adhesion, friction and cytoskeletal reorganization. Strains of the cortex and cytoplasm are described by the variables xm and xc, respectively. Note that, in our simulations, we use pressure rather than force to induce the cellular deformations; this accounts for the extra μm2 found in the denominators of the parameters in our model.Figure 2Viscoelastic model of cell. A. Representation of the viscoelastic model of the cell. xc and xm denote the location of the cell cytoplasm and membrane, respectively; τc and kc define the mechanical model of the cell cortex; τa includes the viscous deformation of the cytoplasm as well as other components including adhesion. B. To validate our model and determine model parameters, we utilized micropipette aspiration technique. Relevant parameters include the radius of the micropipette (Rp), the radius of the cell (Rc), and the length of protrusion into the micropipette (Lp). C. Example of a Dictyostelium cell being aspirated into the micropipette at 0.65 nN/μm2. Time stamps are in seconds, scale bar shows 10 μm. D. Protrusion into the pipette was measured and was accounted for by the model. Different colored circles represent data from 22 individual experiments. Solid line represents the deformation defined by Eqn. 6 with parameters kc = 0.098 nN/μm3, τc = 0.064 nNs/μm3 and τa = 6.09 nNs/μm3.As shown below, this combined Voigt-dashpot viscoelastic model reasonably approximates the mechanical properties of Dictyostelium where cross-linking proteins are predominantly enriched in the cortex. Extending our framework to other cell types may require different viscoelastic models to describe the cell of interest. For example, aspiration of chondrocytes suggests that these cells obey a Kelvin model (similar to the Voigt element, but includes an elastic component in series with the viscous element) [28]. Once the appropriate viscoelastic model is developed, the implementation in the LSM framework introduced here is straightforward.Experimental determination of model parametersTo determine appropriate parameters for the viscoelastic model, we used micropipette aspiration to apply step pressures (rapid increase of pressure from 0 to 0.65 nN/μm2) to individual cells [29,30]. In this technique, a small negative hydrostatic pressure is created at the tip of a micropipette. By bringing the micropipette into close proximity of the cellular surface, the cell is aspirated into the micropipette.We applied step pressures to wild type interphase cells and measured cellular deformation as a function of time (Fig. 2B). Deformation is quantified by the length of cellular protrusion into the pipette tip, denoted Lp (Fig. 2C). We aspirated 22 cells with a radius of 4.3–6.1 μm, a pipette radius of 3.1 μm and a pressure of 0.65 nN/μm2. The cells deformed in two distinct phases (Fig. 2D). Within the first several seconds after application of the aspirator, the cellular deformation increased sharply, with the length of the aspirated cortex increasing to an average value of 4 μm. The deformation during this phase can be interpreted as being dominated by the elastic characteristics of the cytoskeletal network. Thereafter, the trajectory was dominated by slow cellular flow into the micropipette, increasing, on average, about 2 μm over the next 25 s.The pressure applied by the micropipette aspirator is not the only pressure experienced by the cell. At rest, the cell is also under pressure from cortical tension (γten), which maintains the spherical shape of the cell. Under the cortical shell-liquid drop model [31], we assume that the cortical tension arises as surface tension (ignoring tangential stress). Following the Young-Laplace equation for liquid interfaces, the equilibrium pressure experienced by a spherical cell of radius Rc is(3)Peq = 2γten/Rc.The cell's protrusion into the micropipette is driven by the aspiration pressure. As the cell is aspirated, the portion of the cell inside the micropipette will be a spherical cap of radius Rcap <Rc. Given a measured length of protrusion Lp, the radius of the spherical cap Rcap can be obtained (see Additional file 1). The cap's smaller radius gives rise to higher local curvature, creating a rounding pressure:Pround(LP) = 2γten/Rcap - Peq,to oppose the aspirationAt the critical aspiration pressure, Pcrit, the cell extends a perfect hemispherical projection with radius Rp into the micropipette and does not protrude any further under this constant pressure. Thus, the critical pressure is:(4)Pcrit = Pround(Rp) = 2γten(1/Rp - 1/Rc).The cortical tension has been measured to be 1–1.5 nN/μm in passive, wild type Dictyostelium cells [18,31,32]. Here, assuming cortical tension of γten = 1 nN/μm, pipette radius of Rp = 3.1 μm and cell radius of Rc = 5.1 μm, we can compute Pcrit to be approximately 0.25 nN/μm2. Because the applied pressure was greater than the critical pressure, the cell was continuously aspirated into the pipette. Cells were only tracked for 30 s, as longer timescales are dominated by cortical remodeling and turnover [33,34].Pascal's law dictates that the hydrostatic pressure, Pext, in the micropipette is normal to the cell membrane inside the micropipette and has the same magnitude in all directions. Similarly, the cell's equilibrium pressure is normal to the cell membrane everywhere with the same magnitude. We used the total pressure, Ptotal = Pext - Pround, as the input to the cell's mechanical model. This pressure is applied to the cell membrane region around xm and is transferred directly to the cell's cortex, formed of cytoskeleton and its cross-linkers, just beneath the cell membrane. The corresponding mathematical model is:Ptotal=τc(x˙m−x˙c)+kc(xm−xc−w0)Ptotal=τax˙c,where w0 represents the initial position of the cell cortex when no force is applied to the system. We define ℓ such that:xc = xm - ℓ - w0.With this variable change, the transformed system can be written as:(5)[x˙mℓ˙]=[0−kc/τc0−kc/τc][xmℓ]+[1/τa+1/τc1/τc]Ptotal.Using the Voigt-dashpot model of Eqn. 5 to account for the viscoelastic response of the cell to an applied step pressure, the aspirated cellular length into the pipette, Lp, is given by:(6)Lp(t)=xm(t)=Ptotal(1kc(1−e−kct/τc)+tτa).Data from all 22 cells were combined. The following parameters in the viscoelastic model were obtained using a least squares fit (using Matlab's curve fitting toolbox): kc = 0.098 ± 0.007 nN/μm3, τc = 0.064 ± 0.018 nNs/μm3, and τa = 6.09 ± 1.44 nNs/μm3 (the ± value refer to a 95% confidence interval). With these parameter values, the 1-D model was able to capture the deformation trends observed in the experimental data (Fig. 2D). Note that the elastic constant obtained, when applying the methods of Theret et al. [35] and Hochmuth [30], is equivalent to an elastic Young's modulus of 70 pN/μm2, which is similar to the value of 95 pN/μm2 measured for Dictyostelium using different techniques [18].Implementation of micropipette aspiration simulationDuring micropipette aspiration, the cell's velocity is generated by externally applied pressure, as well as internally generated cellular pressures such as cortical tension. We now outline how the contribution of each pressure is computed and applied to the cell's potential function ϕ(x, t).We choose to do the simulations in two dimensions. The level set method is directly applicable to three dimensions (3-D), and all of the level set equations either carry over without change into 3-D, or have natural extensions. In practice, however, the computational burden of 3-D simulations is significant and hence we restrict ourselves to two dimensions. To differentiate the forces (and hence pressures) which are 2-D, from the scalar pressures used above, we use bold characters.Evolving the cell membraneThe simulation accounts for the effects of three pressures: those generated externally by the micropipette; those generated internally to maintain constant volume; and rounding pressure corresponding to the cell's cortical tension. Together, these pressures generate a velocity field that evolves the cell's membrane.Externally applied pressureTo account for the force generated by the hydrostatic pressure in the micropipette, the pressure Pext, uniform in magnitude and normal to the cell membrane, is used in the LSM simulation. This force exists only inside the inner boundaries of the pipette.Pressure due to volume conservationWe assume that, during aspiration, the cellular volume (V) remains constant. To enforce this constant volume condition numerically, we implement a pressure, acting normal to the surface:(7)Pvol = Kvol(Vresting - Vactual)nwhere n is the outward normal. The cell's volume is evaluated by assuming the cell is radially symmetric:(8)Vactual = ∫cell lengthπr(x)2dx.To ensure that the cell's volume does not change during the course of the aspiration requires that Kvol be large. In our simulations, we set Kvol = 0.1 nN/μm5, which was sufficiently high to ensure that volume changes were minimal (Fig. 3G) while maintaining stability of the simulations.Figure 3Simulation of micropipette aspiration. A. To account for the solid surface of the micropipette, we introduce a mask potential function (Ψ) defined by the micropipette walls (black line). B. A cross section illustrates how the masking potential function is used to clip the evolving level set potential function. Based on the driving equations, the potential function evolves from ϕ(t)(solid blue) to ϕ(t + Δt)(dashed blue). However, this new position makes the cell cross into the pipette (defined by the mask function Ψ – green line). The level set function is then clipped to ϕ(t+Δt)¯ to account for the solid surface. C. Parallel spring-dashpot units are used to represent the viscoelastic state of the cell as the boundary evolves from Γ(t) to Γ(t + Δt). Each component consists of a viscoelastic model as defined in Fig. 2A. D. Simulation of micropipette aspiration implementing the viscoelastic model of the cell in the LSM (using the adjusted parameters; see main text). Shown is an overlay of simulation frames at t = 0 s (spherical cell, light grey), 0.5 s, 1 s, 5 s and 20 s (farthest protrusion, black). E. Measurements from aspiration simulations. Assuming an aspiration pressure of 0.65 nN/μm2, the protrusion into the cell from the simulation (black line) can account for the experimental data (grey dots; mean square error, MSE, is 0.74 μm; coefficient of determination, R2, of 0.78) nearly as well as the data fit (dotted line) from Fig. 2D. (MSE: 0.73, R2: 0.79). With slightly different parameters (see main text) the simulation (red line) overlaps the fitted data better (MSE: 0.73, R2: 0.79). Aspiration forces near the critical pressure (0.35 nN/μm2) can deform the cell initially, but do not draw it in further (green line). F. After 20 s of micropipette aspiration, the pressure in the aspirator is dropped, leading to a relaxation in the protrusion distance; a typical example is shown here. Time stamp indicate seconds after release of aspiration pressure, scale bar corresponds to10 μm. Note that the cell does not fully retract the aspirated portion. G. In a LSM simulation, the cell's retraction can be shown as the decrease of length of protrusion. Simulation of cell relaxation accurately demonstrates the lack of complete retraction observed (red). Also shown is the cell's volume (blue) during the simulation demonstrating that any volume changes are minimal.Rounding pressure due to cortical tensionResting cells experience cortical tension [31] which generates pressure, Peq, as shown in Eqn. 3. When a spherical cell is aspirated, the cell's cortex resists deformation.The pressure generated depends on the local surface curvatureκ(x)=∇⋅(∇ϕ|∇ϕ|),and a material property of the cortex referred to as the cortical tension (γten) according to:(9)Pten(x) = 2γtenκ(x)n.Therefore the rounding pressure produce by the cell is Pround = Pten - Peq. This acts inward normal to the membrane.We have chosen to define Pround as the difference between the tension and an equilibrium pressure. This is accordance to experiments on neutrophils that found that cortical tension depends on surface area [36]. However, the latter term can also be incorporated into the volume conservation term. In particular, combining Pvol and Pround leads to:Pvol−Pround=Kvol(Vresting−Vactual)n−(2γtenκ(x)n−2γten/Rc)n=Kvol(Vresting−V^actual)n−(2γtenκ(x))n,where V^actual = Vactual + 2γten/(RcKvol).The coefficient 2 in the pressure equation is introduced to account for the fact that our curvature calculation is based on the 2-D representation of cell shape, as the curvature of a sphere of radius r is 2/r, but the curvature of a 2-D circle is only 1/r. In the computation of curvature, spline-based interpolation was used to smooth out discretization noise.Total pressure and cell evolutionIn the above formulations, the total pressure outward normal to the cell membrane is:(10)Ptotal = Pext + Pvol - Pround.The formulation of x˙m in Eqn. 5 provides us with the pressure-velocity relationship:(11)v=x˙m=−(kc/τc)l+(1/τa+1/τc)Ptotal.The velocity vector, v, is defined for points on the cell membrane. This needs to be extrapolated to a velocity field to evolve the potential function ϕ. It is only the velocity variations tangential to a given interface that dictate the interface motion [37]. A velocity field that minimizes the normal component of the field variation is achieved by extrapolating the membrane velocity with the nearest neighbor method. In other words, the velocity v(x) at a point x can be set equal to the membrane velocity v(x¯) at the membrane location x¯ closest to the point x. It has been shown that a signed distance function tends to stay a signed distance function when the closest neighbor extrapolation method is used [38]. We can now use this velocity field to evolve the cell membrane according to Eqn. 2.Eqn. 11 points to a difference between the LSM model of cellular deformation and the one-dimensional (1-D), scalar model used to obtain the viscoelastic parameters (Eqn. 6). In the latter, the pressure is co-aligned with the direction of the viscoelastic components, implying that the direction of motion is also always inline with the direction of the applied pressure. In the LSM simulation, the pressure is applied normal to the cell membrane, but the viscoelastic component, l, does not have to have the same directionality, and the resultant velocity vector is not always normal to the cell membrane. While providing us with good starting point for the parameter estimation, the 1-D formulation therefore can not be expected to explain the 2-D simulation completely.Restricting cell shape inside micropipetteAs the cell's level set potential function moves into the micropipette, its shape is restricted to remain inside the micropipette. This is achieved by first defining a mask potential function [39], Ψ, for the micropipette (Fig. 3A). The effect of the mask is to correct for the cell's potential function by clipping it (Fig. 3B) according to:ϕ(t+Δt)¯=min⁡(ϕ(t+Δt),ψ).This restriction guarantees that the cell boundary never moves outside of the inner walls of the virtual micropipette. After this restriction, the net change in ϕ is: ϕ(t+Δt)¯ - ϕ(t), which translates (see Additional file 1) to an effective velocity that is normal to the cell membrane:(12)v¯=−ϕ(t+Δt)¯−ϕ(t)Δt∇ϕ|∇ϕ|2.Thus, wherever clipping by the micropipette mask occurs, we must use this effective velocity to evolve the potential functions in simulation.Evolution of the viscoelastic state of the cellIn our simulations, the cell can be represented by a series of parallel viscoelastic systems with the same parameters (Fig. 3C). These sub-systems are not interconnected, and the applied pressure on each system, Ptotal as defined in Eqn. 10, is normal to the cell membrane. We argue that applying total pressure to the parallel unconnected spring damper systems used in this model closely approximates cellular behavior when the following conditions are met:1. Membrane pressure profile is piecewise smooth. This is a reasonable assumption as, in practice, pressure profiles are piecewise smooth. Even when a point force is applied to a particular location of the cell membrane, membrane elasticity will diffuse this force and make the pressure smooth locally.2. Simulation grid density is dense enough for simulation stability, but not much denser than the discretization of the membrane pressure profile. With this assumption, the interpolation nature of level set method acts like a low pass filter, where effects of artificial abrupt jumps in the pressure profile are smoothed.Let l(x, t), x ∈ Γ(t) be the viscoelastic state of the cell at time t and at a position x on the membrane. That is, |l| represents the length of the numerous parallel unconnected spring-damper systems. At a given position, x, on the membrane, there is a vector with length given by |l(x)| = |ℓ| in Eqn. 5, representing the state of a single spring-damper system. Thendldt=∂l∂x∂x∂t+∂l∂y∂y∂t+∂l∂t=[Dl]v+∂l∂t,where D is the Jacobian operator, [Dl]v represents the displacement of the whole cell membrane, and ∂l∂t=ℓ˙n as defined in Eqn. 5. The equation describing the evolution of l is:(13)−kcτcl+1τcPtotal=[Dl]v+∂l∂t.Testing of model: Micropipette aspiration simulationTo summarize, the flow chart of the simulation steps is shown in Fig. 4. The implementation is derived from the Level Set Toolbox [39] and is coded in Matlab (Mathworks, Natick, MA). The simulations were implemented on a fixed grid of 10 μm in size, with density of 20 points/μm and 4 ms time steps. Simulating 15 seconds of aspiration takes approximately 8 h on a desk-top computer.Figure 4Algorithm for LSM simulation of micropipette aspiration.We simulated the micropipette aspiration experiment under several different aspiration pressures. Using an aspiration pressure of 0.65 nN/μm2 (the pressure used to obtain our viscoelastic model parameters), our simulation reproduced the trend observed in real cells (black line in Fig. 3E). The result of this simulation did not completely overlap the least-squares fitted data, though the fit to the experimental data is nearly as good. The fitted data has a mean square error (MSE) of 0.73 μm and a coefficient of determination (R2) of 0.79; the simulation has 0.74 μm and 0.78 respectively. Using different parameter values: kc = 0.1 nN/μm3, τc = 0.08 nNs/μm3, and τa = 4.6 nNs/μm3, we were able to reproduce the fitted data slightly more accurately (Fig. 3D and red line in Fig. 3E; MSE of 0.73 μm and an R2 value of 0.79).Using 0.35 nN/μm2 of pressure, the cell was rapidly and partially aspirated into the pipette. Thereafter, it remained nearly immobile. This simulation recreates the observed behavior of Dictyostelium cells at aspiration pressures near the critical pressure.To test our model further, we simulated the relaxation of an aspirated cell and compared this to experimental results in which a cell is aspirated into the micropipette for approximately 20 s at which point the applied pressure is released. The cell responds by rapidly retracting the aspirated portion (Fig. 3F). The retraction gradually slows to a near halt, with a significant portion of the cell remaining inside the micropipette. This behavior was reproduced in our simulations. The simulated cell retraction from the micropipette is measured in the reduction of length of protrusion (Fig. 3G), matching the retraction behavior seen in live cells. As shown in Fig. 3G, the variation in cell volume during these simulations was less than 1%.Simulating Dictyostelium cell shape changes using a simplified chemotaxis modelHaving established that we can recreate the cellular shape during micropipette aspiration, in which externally applied pressures are driving cell shape changes, we consider a situation in which the pressures arise as a response to external stimuli. To this end we simulated the cell shape behavior of chemotactic Dictyostelium cells.Dictyostelium cells have the ability to detect spatial differences in the concentration of the extracellular chemoattractant cAMP. They interpret these spatial differences and respond by localizing signaling molecules. These signaling molecules in turn bias the locations of actin polymerization driven protrusions and myosin-II motor mediated retractions, generating internal mechanical forces to deform the cell as well as propel the cell towards the chemoattractant [1,40].Our goal in these simulations is not to propose new chemotaxis signaling mechanisms, or even to analyze the large number of proposed mechanisms (reviewed in [3]). Rather, it is to illustrate how cellular signaling can be coupled to the LSM framework to drive cellular deformations. Thus, we purposely implement a simple model connecting chemoattractant gradients with intracellular markers.Implementation and testingWe base our model for pressure generation on a previously published signaling model that accounts for receptor mediated localization of phosphatidylinositol (3,4,5)-trisphosphate (PI(3,4,5)P3) [41]. Though recent experimental data suggests that cells employ multiple parallel pathways to regulate chemotaxis [42,43], localization of this membrane lipid has been correlated with the appearance of pseudopods [40]. Moreover, elevated levels of PI(3,4,5)P3 correlate temporally with increased levels of actin polymerization [44].Rather than implementing the complete reaction-diffusion equations describing the PI(3,4,5)P3 model, we simplify it by using a steady-state distribution of PI(3,4,5)P3 along the cellular membrane. It was shown that the membrane concentration of PI(3,4,5)P3 is an amplified response of the relative cAMP concentration observed on the membrane [41,45]:(14)PI(3,4,5)P3 ∝ [cAMP/mean(cAMP)]3.Next, we compute the pressure components contributing to cell motion, which include protrusion, retraction, volume conservation, and cortical tension pressures. To compute protrusion pressure, we first assume that actin polymerization creates a pressure wherever the PI(3,4,5)P3 concentration is above its mean level:(15)Pprot=Kprotmax⁡(0,PI(3,4,5)P3−mean(PI(3,4,5)P3)max⁡(PI(3,4,5)P3)−mean(PI(3,4,5)P3))n.Similarly, we assume myosin-II retraction occurs wherever PI(3,4,5)P3 concentration is below its mean level:(16)Pretr=−Kretrmax⁡(0,mean(PI(3,4,5)P3)−PI(3,4,5)P3mean(PI(3,4,5)P3)−max⁡(PI(3,4,5)P3))n.Both of these act normal to the cell membrane. We let the proportionality constant in Eqn. 14 be absorbed into constants Kprot and Kretr. Eukaryotic cells can generate actin mediated protrusion pressures of 0.5–5 nN/μm2 [46]. We chose Kprot = 0.5 nN/μm2 and Kretr = 1 nN/μm2.When computing the conservation of volume pressure, we assume that the cell is flat with uniform thickness. Thus, volume conservation is equivalent to conserving the 2-D area of the cell:Parea = Karea(A0 - A)n.The flat cell assumption also implies that the pressure generated by cortical tension depends only on the 2-D local surface curvature and the 2-D equilibrium pressure. The rounding pressure due to cortical tension is therefore given by:(17)Pten = Kten(κ(x) - 1/Rc)n.Values of Karea = 0.2 nN/μm4 and Kten = 1 nN/μm were used in these simulations.Summing all these components yields the total force normal to the cell membrane:Ptotal = Pprot + Pretr + Parea - Pten.Finally, the membrane velocity is computed using Eqn. 11, with the same viscoelastic parameters τa, kc and τc. The simulation algorithm is similar to the micropipette aspiration case, and is summarized in Fig. 5.Figure 5Algorithm for LSM simulation of cell shape changes in response to external chemotaxis gradients. This algorithm includes only general steps required to generate the pressure profile. To simulate chemotaxis also requires that the chemoattractant gradient be generated and that the protrusive (Pprot) and retractive (Pretr) pressures be computed. These would be determined by specific models of chemotactic response. In our simulations, these were generated by Eqn. 15 and Eqn. 16, respectively.This simulation successfully generated chemotaxis behavior (Fig. 6). In response to a chemoattractant gradient, the cell, whose shape was initialized as a circle, changed shape and migrated in the direction of the chemoattractant gradient (Fig. 6A). The pressure profile (Fig. 6B) and displacement (Fig. 6C) are shown as functions of local cAMP concentration and time, respectively. The cell achieved a velocity of 11.7 μm/min, which is similar to published velocities of Dictyostelium cells (e.g. 11.8 μm/min[47]). During the simulation, the cellular area (and hence volume) remained nearly constant (Fig. 6C).Figure 6LSM simulations of cell shape changes during chemotaxis. A. We simulated the change in cellular morphology of a Dictyostelium cell exposed to a point source of chemoattractant (1 μM of cAMP). Shown is the resultant chemoattractant field (computed by solving the diffusion equation) as well as the location of the cell at times 0, 1.5, 10, 40, 60, 80 and 100 s. Initially, the cell is assumed to be round (red circle). B. In this model, pressure was determined by the concentration of PI(3,4,5)P3 on the membrane as described in the text. The maximum and minimum refer to the concentrations experienced by the cell around the membrane. C. The position of the cell (blue) was plotted as a function of time, showing fairly constant velocity (11.7 μm/min). Also shown is the cell's area (red) during the simulation demonstrating that changes are also minimal.Membrane pressure profile and cell shapeWhile our simulations of Dictyostelium recreate the motion of the cell in response to the chemoattractant gradient, the resultant cell shape change is small and the steady-state morphology does not resemble that observed experimentally in chemotaxing. Wild type chemotaxing Dictyostelium cells become elongated (Fig. 7A). Other strains, including the amiB- mutants [48] can move stably in fan-like shapes that are reminiscent of keratocytes (Fig. 7D). Without determining the underlining molecular methods, we hypothesized that the difference in cell shape can be accounted for by the way that the force generation is distributed along the cell membrane. Our LSM simulation framework allows us to determine how these forces are distributed along the cell to generate the resulting cell shapes, both for wild type and mutants. To this end, we set out to replace our initial model, described by Eqn. 15 and 16, by one based on the observed morphologies.Figure 7Pressure profile drives cell shape. A. During chemotaxis, wild type Dictyostelium cells acquire a polarized, elongated morphology. B. Eqn. 19 was used to compute the pressure profile (red dots) necessary to maintain the elongated cell shape (inset) along the cell membrane, and this is plotted as a function of the local chemoattractant (cAMP) concentration. The maximum and minimum refer to the concentrations experienced by the cell around the membrane. Pressure profile used in the simulations (blue line) was obtained by fitting the computed pressure profile, details are given in the Additional file 1. C. Chemotaxing cell using the pressure profile of panel B. The shapes of the cell are shown at times 0, 1.5, 10, 20, 40, 60, 80 and 100 s. Other details in the simulation are as in Fig. 6. D-F. Simulations of chemotaxis in Dictyostelium amiB- cells. These mutant cells acquire a fan-like morphology (panel D) and move along their broad axis. This form of movement was recreated using the pressure profile of panel E (colors as in panel B). F. Chemotaxing cell using the force profile of panel E. Times of the shapes are as in panel C.Given a stable cell shape Γ0 traveling at velocity u, we let Γu be the displaced cell at time Δt, and ϕ0 and ϕu be the potential functions representing Γ0 and Γu respectively. The effective velocity field necessary for this displacement is:(18)u¯=−(ϕu−ϕ0)/Δt|∇ϕ|n.If the cell shape is at steady state, we can assume that the internal viscoelastic network is also in steady state, that is, dldt=0. Therefore, from Eqn. 5, we compute the viscoelastic steady state ℓ = Ptotal/kc.Moreover, the membrane speed at steady state is expressed as x˙m = Ptotal/τa. Combined with Eqn. 18, we find Ptotal:Ptotal=τa(ϕu−ϕ0)/Δt|∇ϕ|nTaking into account the effect of cortex tension, and assuming that there is no cell volume deviations, we can compute:(19)Pprot+Pretr=τa(ϕu−ϕ0)/Δt|∇ϕ|n+Pten,where Pten is the cortical tension-driven rounding pressure defined in Eqn. 17. Using this formula, and a cell velocity of 10 μm/min, we calculated the pressure profiles required to generate cell shapes seen in wild type cells as well as in amiB- cells.Obtaining these pressure profiles is straight-forward computationally, taking less than one minute of CPU time on a desk-top computer. It does require, however, a smooth shape. Thus, a certain amount of image processing is needed when using segmented images from experiments. Moreover, the formula in Eqn. 18 is based on a steady-state shape. Handling transient cell shape changes, such as protrusions or retractions, needs a local description of the velocity v(x).Our results indicate that to generate polarized cell morphologies observed in wild type Dictyostelium cells, the protrusive forces must be primarily concentrated along the anterior ≈ 25% portion of the cell; see Fig. 7B. This is reminiscent of the PI(3,4,5)P3 threshold observed in cells [45,49]. At the sides, a smaller and less localized retractive force gives the cell its elongated shape. When this pressure profile was used to simulate a chemotaxing cell (Fig. 7C), the resulting virtual cell achieved an elongated shape and chemotaxed successfully to the source of chemoattractant achieving a stable velocity of 11.1 μm/min.Clearly, a different pressure profile is needed to generate a fan like shape as observed in amiB- cells (Fig. 7D). Here, the maximum protrusive force is spread out considerably more at the front, while large amount of retraction force is still needed to pull the tail region along. Using this pressure profile in the chemotaxis simulation led to a migrating cell with stable shape similar to that seen experimentally (Fig. 7F). The resultant fan-shaped cell achieved the stable velocity of 9.7 μm/min.ConclusionWe have shown that the simulation framework we have developed can be used to model cell shape deformations as well as cell motility. The simulations can produce deformations seen during micropipette aspiration experiments. This requires parameter values for the viscoelastic model which can be obtained experimentally. It should be noted, however, that 2-D simulations using parameters based on a 1-D model may not reproduce the 1-D model simulation precisely.In the simulations of cell shape changes during chemotaxis, we saw that our simple model for generating the cell's protrusive and retractive forces in response to a chemoattractant gradient does not produce experimentally observed cell shapes. However, our techniques allow us to work backwards from shape to obtain the required forces. We determined that generating the elongated cell shape requires a large protrusive force at the front (the pressure profile there is positive). At the sides, there is a large retractive force (the pressure profile there is negative). While measuring this pressure profile directly would be difficult, it is possible to image fluorescently-tagged myosin-II to infer a measure of the forces acting on the cell. Under the assumption that the retractive force is being generated by myosin-II, we expect that myosin-II would be greatly enriched at the sides. Quantitatively, the spatial distribution of myosin could be used to estimate how much force is being generated along the membrane (as has been done during cytokinesis [50]).Authors' contributionsLY implemented the LSM simulations and drafted the manuscript. JCE performed experiments to measure the viscoelastic properties of cells, under the guidance of DNR. BLK and SES participated in the implementation of the LSM algorithm. PAI conceived of the study, and participated in its design and coordination. LY, JCE, DNR and PAI wrote the manuscript which was read and approved by all the authors.Competing interestsThe authors declare that they have no competing interests.Supplementary MaterialAdditional file 1This document presents detailed derivation of several of the formulae in the text.Click here for file\n\nREFERENCES:\nNo References"
+}

batch_5/PMC2535632.json

@@ -0,0 +1,4 @@
+{
+    "id": "PMC2535632",
+    "text": "This is an academic paper. This paper has corpus identifier PMC2535632\nAUTHORS: Marianne R. Hopkins, Adrienne S. Ettinger, Mauricio Hernández-Avila, Joel Schwartz, Martha María Téllez-Rojo, Héctor Lamadrid-Figueroa, David Bellinger, Howard Hu, Robert O. Wright\n\nABSTRACT:\nBackgroundGiven the association between iron deficiency and lead absorption, we hypothesized that variants in iron metabolism genes would predict higher blood lead levels in young children.ObjectiveWe examined the association between common missense variants in the hemochromatosis (HFE) and transferrin (TF) genes and blood lead levels in 422 Mexican children.MethodsArchived umbilical cord blood samples were genotyped for HFE (H63D and C282Y) and TF (P570S) variants. Blood lead was measured at 24, 30, 36, 42, and 48 months of age. A total of 341 subjects had at least one follow-up blood lead level available and data available on covariates of interest for inclusion in the longitudinal analyses. We used random-effects models to examine the associations between genotype (HFE, TF, and combined HFE + TF) and repeated measures of blood lead, adjusting for maternal blood lead at delivery and child’s concurrent anemia status.ResultsOf 422 children genotyped, 17.7, 3.3, and 18.9% carried the HFE H63D, HFE C282Y, and TF P570S variants, respectively. One percent of children carried both the HFE C282Y and TF P570S variants, and 3% of children carried both the HFE H63D and TF P570S variants. On average, carriers of either the HFE (β = 0.11, p = 0.04) or TF (β = 0.10, p = 0.08) variant had blood lead levels that were 11% and 10% higher, respectively, than wild-type subjects. In models examining the dose effect, subjects carrying both variants (β = 0.41, p = 0.006) had blood lead 50% higher than wild-type subjects and a significantly higher odds of having a blood lead level > 10 μg/dL (odds ratio = 18.3; 95% confidence interval, 1.9–177.1).ConclusionsIron metabolism gene variants modify lead metabolism such that HFE variants are associated with increased blood lead levels in young children. The joint presence of variant alleles in the HFE and TF genes showed the greatest effect, suggesting a gene-by-gene-by-environment interaction.\n\nBODY:\nDespite efforts to reduce lead in the environment by removing lead in gasoline and banning lead-based paint, an estimated 310,000 U.S. children 1–5 years of age have elevated blood lead levels [Centers for Disease Control and Prevention (CDC) 2005]. In developing countries, lead exposure is a concern because of continued use of lead-containing products and lack of regulations or enforcements policies (Meyer et al. 2003). Furthermore, research suggests that lead exerts its neurotoxic effects in children at blood levels lower than the current CDC action level of 10 μg/dL (Canfield et al. 2003; Jusko et al. 2008; Lanphear et al. 2005). There is growing interest in identifying host factors that increase the risk of elevated blood lead levels in children. Gene variants within metabolic pathways that influence lead absorption may be such a susceptibility factor and could place children at increased risk of lead poisoning even at low environmental levels (Onalaja and Claudio 2000).Because of the well-described inverse relationship between iron stores and lead absorption (Kwong et al. 2004), iron metabolism genes are potential candidates to modify lead absorption and stores. Bradman et al. (2001) and Hammad et al. (1996) showed an inverse relationship between dietary iron and blood lead levels, and Wright et al. (1999, 2003) found an association between biomarkers of iron deficiency and elevated blood lead levels both cross-sectionally and longitudinally.Iron absorption is regulated by iron stores and erythropoiesis (Bothwell et al. 1979) and is influenced by dietary iron intake and hypoxia (Morgan and Oates 2002). Hereditary hemochromatosis (HH) is an autosomal recessive disorder leading to excessive iron stores secondary to increased intestinal iron absorption (McLaren et al. 1991). Two predominant variants of the hemochromatosis (HFE) gene account for most cases: the C282Y and H63D variants, which are common in the U.S. population, with a prevalence of 7–17% and 10–32%, respectively (Bradley et al. 1998).Transferrin forms a stable complex with the HFE protein to facilitate iron transfer (Feder et al. 1998; Lee et al. 1999). It has been suggested that the effects of HFE gene on iron absorption depend on the complex relationship between HFE and the transferrin receptor (TfR) (Morgan and Oates 2002). Feder et al. (1998) demonstrated that the HFE H63D variant altered transferrin binding, leading to a loss of HFE-repressor function for transferrin uptake, and thereby increasing iron transport into cells. A common missense variant of the transferrin gene (TF) is P570S, with a prevalence rate of about 15% in the general population (Lee et al. 1999).Previous reports suggest that subjects with clinical hemochromatosis have higher (Barton et al. 1994) or equivalent (Åkesson et al. 2000) blood lead levels compared with normal controls. These studies were not population based, but used case–control designs comparing adult subjects with clinical disease with those without. In a population-based cohort, our group reported lower bone lead levels among elderly men carrying at least one copy of H63D or C282Y (Wright et al. 2004). The differences in the design, sex, and age of the different cohorts studied might account for the seemingly disparate findings, with older, male subjects with HFE variants being most likely to have high body iron stores and down-regulated lead absorption. In our 2004 report (Wright et al. 2004), we hypothesized that variants in iron metabolism genes might predict higher blood lead levels in young children over time because of their greater dietary iron needs and lower body iron stores, which would up-regulate iron and lead absorption differentially among HFE variant carriers and children with wild-type genotypes. In this study, we sought to determine whether genetic variants of iron metabolism (HFE C282Y, HFE H63D, TF P570S) were associated with blood lead levels in children.Materials and MethodsStudy populationStudy participants were identified from among pregnant women receiving antenatal care between 1994 and 1995 at three hospitals in Mexico City that serve low- to middle-income populations. The women were approached before giving birth to participate in a randomized trial of calcium supplementation to lower blood lead levels over the course of lactation (Hernández-Avila et al. 2003). Infant development was also measured as part of this study, which was conducted as part of an established interinstitutional collaborative effort between researchers in the United States (Harvard University) and Mexico [National Institute of Public Health and American British Cowdray (ABC) Hospital]. The research protocol was approved by the Human Subjects Committees of the Harvard School of Public Health and the National Institute of Public Health and the participating hospitals in Mexico and has complied with all federal guidelines governing the use of human participants.Data collection methods and exclusion criteria have been described in detail elsewhere (González-Cossío et al. 1997). Interviewers explained the study to and obtained written informed consent from eligible pregnant women who were willing to participate. Anthropometric data from the mother and newborn as well as umbilical cord and maternal venous blood samples were gathered within 12 hr of delivery. Information on estimated gestational age and characteristics of the birth and newborn period was extracted from the medical records. Baseline maternal information on reproductive and health status, social and demographic characteristics, risk factors for environmental lead exposure, and dietary assessment of nutrient intake was collected from all eligible participating mothers.Infants of participating mothers identified before delivery had umbilical cord blood samples collected at birth (n = 520). Children were subsequently assessed for neurocognitive development, and blood lead levels were obtained at 24, 30, 36, 42, and 48 months of age. Data on the neurocognitive test performance are presented elsewhere (Gomaa et al. 2002). The present analysis is limited to data from 422 infants who had archived umbilical cord blood available and were successfully genotyped. After genotyping, the full data set was anonymized to protect the confidentiality of study subjects and to conform with current institutional review board policies. A total of 341 subjects had at least one follow-up blood lead level available and data available on covariates of interest for inclusion in the longitudinal analyses.Blood measurementsBlood lead measurements were performed using graphite furnace atomic absorption spectrophotometry (model 3000; Perkin-Elmer, Wellesley, MA, USA) at the ABC Hospital Trace Metal Laboratory in Mexico City according to a technique described by Miller et al. (1987). The laboratory participates in the CDC blood lead proficiency testing program administered by the Wisconsin State Laboratory of Hygiene (Madison, WI, USA), which provided external quality control specimens varying from 2 to 88 μg/dL. Our laboratory maintained acceptable precision and accuracy over the study period [correlation = 0.98; mean difference = 0.71 μg/dL; SD = 0.68].Complete blood count (red blood cells, white blood cells, hematocrit, hemoglobin, mean corpuscular volume) in manually diluted samples of whole blood (Beckman/Coulter CBC-5 Hematology Analyzer; Block Scientific Inc, Holbrook, NY, USA) and serum ferritin (Ferritin RIA Kit, Kodak Clinical Diagnostics, Ltd. Amersham, Bucks, UK) were measured using standard clinical methods at the ABC Hospital.HFE and TF genotypingWe extracted high-molecular-weight DNA from white blood cells of archived umbilical cord blood with commercially available PureGene Kits (Gentra Systems, Minneapolis, MN, USA). After DNA quantification, samples were adjusted to TE buffer (containing Tris, a common pH buffer, and EDTA, a chelating agent), partitioned into aliquots, and stored at −80°C. Multiplex polymerase chain reaction (PCR) assays were designed using Sequenom SpectroDESIGNER software (Sequenom, Inc., San Diego, CA) by inputting sequence containing the single nucleotide polymorphism (SNP) site and 100 base pairs of flanking sequence on either side of the SNP. The extension product was then spotted onto a 384-well spectroCHIP (Sequenom, Inc.) before being flown in the MALDI-TOF (matrix-assisted laser desorption ionization–time of flight) mass spectrometer (Sequenom, Inc.).For this study, we included three SNPs: hemochromatosis (HFE) C282Y (rs1800562) and H63D (rs1799945) [Reference Sequence NM_139011, GenBank (http://www.ncbi.nlm.nih.gov/entrez/viewer.fcgi?db=nuccore&id=91718887)], and transferrin (TF) P570S (rs1049296) [Reference Sequence NM_001063, GenBank (http://www.ncbi.nlm.nih.gov/entrez/viewer.fcgi?db=nuccore&id=21536430)]. Specifically, the following primers were used in the multiplex assay:For HFE H63D (rs1799945): forward PCR primer, 5′-ACGTTGGATGTCTACTG-GAAACCCATGGAG-3′; reverse PCR primer, 5′-ACGTTGGATGTTGAAGC-TTTGGGCTACGTG-3′; extension primer 5′-GCTGTTCGTGTTCTATGAT-3′For HFE C282Y (rs1800562): forward PCR primer, 5′-ACGTTGGATGTACCCCA-GATCACAATGAGG-3′; reverse PCR primer, 5′-ACGTTGGATGTGGATAAC-CTTGGCTGTACC-3′; extension primer 5′-GAAGAGCAGAGATATACGT-3′For TF P570S (rs1049296): forward PCR primer, 5′-ACGTTGGATGTGAGTTG-CTGTGCCTTGATG-3′; reverse PCR primer, 5′-ACGTTGGATGATCTTTC-CGTGTGACCACAG-3′; extension primer, 5′-CGCATACTCCTCCACAG-3′.Statistical analysisWe examined distribution of HFE and TF alleles and genotypes and tested frequencies using a chi-square statistic to compare observed and expected counts according to principles of Hardy–Weinberg equilibrium. A priori the two HFE variants (H63D and C282Y) were combined into a single indicator term (i.e., presence of one or two copies of either gene’s variant allele), and subsequent analyses compared carriers of HFE or TF variants with wild-type subjects, thus assuming dominant effects. We calculated summary statistics for child characteristics, stratified by genotype. Bivariate associations between children’s blood lead, ferritin, and hemoglobin levels by genotype (wild type vs. carrier) at each time point were compared using Student’s t-test.Blood lead levels followed a log-normal distribution and were log transformed for the analyses; thus, beta coefficients from regression models represent percent change in blood lead. We used multivariate linear regression to model the effect of genotype on blood lead in cross-sectional analyses at each time point, adjusting for maternal blood lead level at delivery (as a measure of prenatal lead exposure) and child’s concurrent anemia status (as a marker of dietary iron intake). Anemia was defined as a hemoglobin concentration < 12.1 g/dL, based on recommendations for children 2 to < 5 years of age living at an altitude of 7,000–7,999 feet above sea level (CDC 1998). We used random-effects models with unstructured covariance to model the association between HFE and TF genotype and repeated measures of log-transformed blood lead in separate models. These models are flexible with respect to imbalance in the data and, in addition, take into account the correlation between repeated measures on subjects. To explore a possible gene–gene interaction between HFE and TF genotypes in predicting blood lead, we modeled combined HFE plus TF joint genotype as a dichotomous variable (any variant present vs. both wild type) and then, to assess allele “dose” effects, as an ordinal variable (both variants present, TF variant/HFE wild type, TF wild type/HFE variant, both wild type) with wild type for both variants as the reference group. Separate models were repeated controlling for concurrent hemoglobin and concurrent ferritin to account for potential differences in dietary iron intake. We also constructed two-way interaction terms between HFE and TF genotype, and between each gene and lead, hemoglobin, and ferritin levels at each time point to explore potential interactions. Finally, we used logistic regression (PROC NLMIXED) to examine the odds of having a blood lead level ≥ 10 μg/dL associated with presence of gene variants. All statistical analyses were performed using SAS software version 9.1.3 (SAS Institute Inc., Cary, NC, USA).ResultsFour hundred twenty-two children were genotyped, and 17.7%, 3.3%, and 18.9% carried the HFE H63D, HFE C282Y, and TF P570S variants, respectively (Table 1). In addition, 1% of children carried both the HFE C282Y and the TF P570S variants, and 3% of children carried both the HFE H63D and the TF P570S variants (Table 2). There were no statistically significant differences in mean gestational age, sex, umbilical cord blood lead levels, or anthropometric measures at birth between carriers and wild-type subjects (Table 3).Unadjusted mean blood lead levels were consistently higher for carriers of either the HFE or TF variant compared with wild-type subjects at each age (24, 30, 36, 42, and 48 months), but these differences were non-significant (Table 4). There were no statistically significant differences in hemoglobin or ferritin levels at any age between carriers and wild-type subjects for either gene variant (results not shown). Unadjusted mean blood lead levels of subjects carrying any variant or both variants (combined HFE + TF genotype) were consistently higher (at any age) than subjects who were wild type, although these differences were also not statistically significant (Table 4).In cross-sectional analyses, adjusting for maternal blood lead level at delivery and child’s concurrent anemia status, the relationship between genotype (either HFE or TF) and blood lead level (at any age) was non-significant (data not shown). However, in the longitudinal analysis (using repeated measures of blood lead at 24, 30, 36, 42, and 48 months of age) adjusting for covariates (maternal blood lead level at delivery and child’s concurrent anemia status), the relationship between HFE genotype and blood lead was statistically significant (β = 0.11, p = 0.04), and the relationship between TF genotype and blood lead was borderline significant (β = 0.10, p = 0.08) (Table 5). Thus, on average, carriers of either the HFE or TF variant had blood lead levels that were 11% and 10% higher, respectively, than wild-type subjects.There was a statistically significant unadjusted gene-by-gene interaction (βunadj = 0.37, p = 0.02) between the HFE and TF genotypes and, when adjusted for maternal blood lead level at delivery and child’s concurrent anemia status, this interaction term was marginally significant (βadj = 0.31, p = 0.06) (data not shown). There were no significant interactions between either gene with concurrent hemoglobin or with concurrent ferritin concentration (data not shown).We then examined the data to explore whether there was an interactive effect of the presence of both HFE and TF variants on blood lead levels. We first examined the cross-sectional association between having any variant present (those with HFE and/or TF variant) versus both wild type. The dichotomous combined HFE + TF genotype was not significantly related with blood lead (at any age) in cross-sectional models (data not shown). Next, we defined a combined joint genotype by grouping subjects into four categories (both variants, either HFE and/or TF variant, both wild type) to examine the dose effect for presence of gene variant(s). This specification did not reveal any statistically significant associations with blood lead (at any age) in cross-sectional models (data not shown).In longitudinal models, the result of having any variant present (β = 0.08, p = 0.07) was similar to having either HFE (β = 0.11, p = 0.04) or TF (β = 0.10, p = 0.08) variant compared with subjects who were wild type for both variants (Table 5). When examining the variant dose effect, compared with subjects who were wild type for both variants, subjects who carried both HFE + TF variants had higher blood lead levels (β = 0.41, p = 0.006) than subjects who carried either TF variant/HFE wild type (β = 0.04, p = 0.5) or HFE variant/TF wild type (β = 0.06, p = 0.3) (Table 5). Therefore, subjects carrying both variants had blood lead 50% higher than subjects who were wild type for both HFE and TF. These results were unchanged when adding concurrent ferritin or concurrent hemoglobin (in place of anemia status) as covariates in our models (data not shown).Figure 1 shows the risk of elevated blood lead levels (≥ 10 μg/dL) by presence of variant allele(s). Those subjects with either the HFE or TF (any) variant present had significantly higher odds of having a blood lead greater than or equal to 10 μg/dL [odds ratio (OR) = 2.3; 95% confidence interval (CI), 1.0–5.5] compared with those who were wild type. Those subjects with both variants present had significantly higher odds of having a blood lead ≥ 10 μg/dL (OR = 18.3; 95% CI, 1.9–177.1) compared with those wild type for both HFE and TF, though the wide CIs are attributed to the fact that < 5% of our study population had both variants present.DiscussionIn our current study of Mexican children, carriers of the HFE variant genotype had blood lead levels 11% higher than wild-type subjects. Furthermore, carriers of both HFE and TF variant alleles had 50% higher blood lead levels compared with wild-type subjects in models comparing the joint effect of combined HFE + TF genotype on blood lead levels. Those subjects with both gene variants present also had significantly higher odds of having a blood lead level ≥ 10 μg/dL. Our results suggest that genes affecting iron metabolism also affect lead metabolism, and this impact may be magnified by the presence of multiple genotypic variant alleles.This is the first study to examine the association between iron metabolism genes and lead exposure in children. There are at least three previous reports in adults or studies of mixed ages. Barton et al. (1994) found higher blood lead levels in subjects (children to young/middle-aged adults) with HH compared with normal controls, although they used family history and clinical data to determine case status rather than genotype and therefore some subjects in the control group were likely HFE variant carriers. In another study, Åkesson et al. (2000) found no difference in blood lead levels between adult subjects (average age, 55.5 years) with hemochromatosis and controls, but found increased blood lead levels in HH patients who had increasing number of years of treatment by phlebotomy, suggesting an up-regulation of absorption (due to lowered iron stores) with a subsequent increase in lead absorption.In a previous population-based cohort study (Wright et al. 2004), we found lower blood/bone lead levels in hemochromatosis variant (H63D or C282Y) carriers in a group of elderly men. These results contrast particularly with those of Barton et al. (1994). We hypothesized that the results of these three studies may have differed partly because of variations in body iron stores, which correlate with age and sex. Because of menstrual losses, women tend to have lower body iron stores than men, which would up-regulate iron/lead absorption (Hallberg and Rossander-Hulten 1991). In addition, because no mechanisms for iron excretion exist, in the absence of chronic bleeding disorders, iron stores tend to increase with age. We postulated that the HFE gene effect on blood lead level may vary by age because of these differences in iron stores/needs that correlate with age/sex, and that in a younger population with higher body iron needs, the effect of HFE variants may be to increase lead absorption among variant carriers. Conversely, in an older population of elderly males, the HFE variant effect may be attributable to down-regulation of iron and lead absorption, because iron stores in men are higher than in women. Our current study supports this hypothesis; in a population of young children with high body-iron needs, we found higher blood lead levels among HFE variant carriers. This study is among the first to present evidence that gene environment interactions may vary by life stage.In addition, our current study explored the interactive effect of HFE + TF combined genotype on blood lead levels in children. Our results suggested that iron metabolism and body lead burden are affected more profoundly by the joint presence of genotypic variant alleles in both HFE and TF. Given the relatively small sample size to detect gene–gene or gene–environment interactions, we used longitudinal random–effects models, because these models are flexible with respect to imbalance in the data. We were able to include subjects with incomplete data to increase the power of the study to detect these effects.There are several limitations to our study. Our sample was restricted to a homogeneous sample of Mexican children with a lower prevalence of the C282Y variant (heterozygotes, 3.3%) than people from Europe (9.2%) and the Americas (9.0%), but a higher prevalence than people from Africa/Middle East (0.2%), the Indian subcontinent (0.5%), Asia (0%), and Australia (0%) (Hanson et al. 2001). Therefore, because of the wide variability in prevalence rates of the C282Y variant (heterozygotes), it will be important to replicate this study in different populations. The potential for misclassification should also be considered, because there may be other polymorphisms in the HFE or TF genes or in a proximal gene that is in tight linkage disequilibrium with these genes that could account for our findings. Residual confounding is always a concern in observational studies. We chose covariates for this analysis based on biological plausibility as confounders (i.e., factors likely to be associated with both HFE genotype and blood lead levels). Among common predictors of blood lead, few are likely associated with HFE or TF genotype. For example, HFE and TF are not X-linked; therefore, sex should not be a confounder because it is not associated with HFE or TF genotype, and any sex-related differences due to menstruation are not yet an issue in a pre-pubescent population. Genotype should also be independent of environmental lead levels. Iron status is plausibly related to blood lead and HFE genotype (Wright et al. 2004), but should theoretically be a modifier of the association of HFE with blood lead rather than a confounder. Our results did not demonstrate evidence of effect modification by serum ferritin or hemoglobin. Race and socioeconomic status were accounted for in our study design, because we enrolled Mexican children from hospitals that serve a relatively homogeneous group of low-to-moderate income families. Therefore, to account for potential confounding, we controlled for maternal blood lead level at delivery (to account for differences in prenatal lead exposure) as well as child’s concurrent anemia status (to account for differences in dietary iron intake).With respect to the gene-by-gene interaction findings, a limitation of this study is that carriers of both variants represent < 5% of our study population, and even though the result is statistically significant, this could be a chance finding. However, there is biological plausibility to the relationship between HFE and TF, and ours is not the first study to find synergy between these two variant alleles. The HFE gene product regulates the binding of transferrin to transferrin receptors, which in turn regulates the transfer of iron (and potentially other transferrin-bound metals) across cell membranes (Roy et al. 1999; Salter-Cid et al. 1999). Furthermore, lead has been shown to bind transferrin, down-regulating transferrin gene expression (Adrian et al. 1993). Further evidence of this epistatic relationship is seen in clinical studies. Robson et al. (2004) found that carriers of the C2 variant of the TF gene and the C282Y allele of the HFE gene were at five times greater risk for Alzheimer disease compared with all others, whereas neither allele alone had any effect on risk for Alzheimer disease. Although outside the scope of this study, Beckman et al. (1999) examined the association between genetic variants of iron metabolism (HFE, TF, and the two combined) and risk for multiple myeloma, breast cancer, and colorectal cancer. In their study, the HFE and TF genotypes tested separately were not associated with any of these neoplastic disorders, but there was a significant difference between patients and controls with respect to the two genotypes combined.HFE genotype is associated with higher blood lead levels in Mexican children over time. Our results also suggest an interaction between HFE and TF genotype in predicting higher blood lead in young children. These results differ from reports in elderly adults in which HFE variants predicted lower blood lead levels, demonstrating that genetic effects differ by life stage.In conclusion, iron deficiency has been associated with increases in absorption and deposition of lead (Barton et al. 1978); however, the relationship between iron and lead is complex and not completely understood. There is no evidence that iron supplements themselves change lead levels after lead exposure. In a randomized control trial controlling for initial blood lead level, Mexican children who were administered iron (or iron plus zinc) did not have lower blood lead concentrations than the placebo group. Iron supplementation of these lead-exposed children significantly improved iron status but did not reduce blood lead levels (Rosado et al. 2006). Based on previous research, it has been recommended that iron supplementation should be prescribed only to iron-deficient children, regardless of blood lead levels, and not as a treatment for lead poisoning in children (Wright 1999). Thus, children with genetic variants affecting iron metabolism may present a unique challenge: They may be more susceptible to low environmental levels of lead exposure because of increased lead absorption and may also be resistant to interventions such as iron supplementation. Another study evaluating the association between blood lead concentration and a vitamin D receptor (VDR) gene polymorphism found the VDR-Fok1 variant to be an effect modifier of the relationship of floor dust lead exposure and blood lead concentration (Haynes et al. 2003). In combination, these results highlight that some children may be more susceptible to lead exposure or absorption, which emphasizes the importance of further reducing lead exposure to protect the most at-risk children around the world. In addition, these children might represent a vulnerable population that may benefit from promising new interventions such as environmental enrichment, as defined by a combination of “complex inanimate objects and social stimulation” (Guilarte et al. 2003; van Praag et al. 2000). Guilarte and colleagues exposed rats pre- and postnatally to lead, and then randomized them to an enriched environment (cage with a greater space allowance per rat, more varied toys and fixed objects, and increased human handling) versus a standard laboratory cage with minimal handling. The rats reared in the enriched environment showed improvement in spatial learning performance, compared with rats reared in the standard laboratory cage, suggesting the enriched environment mitigated some of the neurotoxic effects of lead exposure (Guilarte et al. 2003). These results are particularly applicable to children, as risk factors for increased lead exposure in children include low socioeconomic conditions such as poor housing and decreased maternal education. Therefore, programs aimed at defining and providing environmental enrichment for children at increased risk of lead exposure might not only reduce risk but also mitigate the neurotoxic effects of exposure.\n\nREFERENCES:\n1. AdrianGSRiveraEVAdrianEKBuchananJHerbertDCWeakerFJ1993Lead suppresses chimeric human transferrin gene expression in transgenic mouse liverNeurotoxicology142–32732828247401\n2. ÅkessonAStålPVahterM2000Phlebotomy increases cadmium uptake in hemochromatosisEnviron Health Perspect10828929110753085\n3. BartonJCConradMENubySHarrisonL1978Effects of iron on the absorption and retention of leadJ Lab Clin Med924536547712193\n4. BartonJCPattonMAEdwardsCQGriffenLMKushnerJPMeeksRG1994Blood lead concentrations in hereditary hemochromatosisJ Lab Clin Med12421931988051482\n5. BeckmanLEVan LandeghemGFSikströmCWahlinAMarkevärnBHallmansG1999Interaction between haemochromatosis and transferrin receptor genes in different neoplastic disordersCarcinogenesis2071231123310383894\n6. BothwellTHCharltonRWCookJDFinchCA1979Iron Metabolism in ManOxford, UKBlackwell Scientific Publications256283\n7. BradleyLAJohnsonDDPalomakiGEHaddowJERobertsonNHFerrieRM1998Hereditary haemochromatosis mutation frequencies in the general populationJ Med Screen5134369575458\n8. BradmanAEskenaziBSuttonPAthanasoulisMGoldmanLR2001Iron deficiency associated with higher blood lead in children living in contaminated environmentsEnviron Health Perspect1091079108411675273\n9. CanfieldRLHendersonCRJrCory-SlechtaDACoxCJuskoTALanphearBP2003Intellectual impairment in children with blood lead concentrations below 10 microg per deciliterN Engl J Med348161517152612700371\n10. CDC (Centers for Disease Control and Prevention)2005Blood lead levels—United States, 1999–2002MMWR Morb Mortal Wkly Rep542051351615917736\n11. CDC (Centers for Disease Control and Prevention)1998Recommendations to prevent and control iron deficiency in the United States. Centers for Disease Control and PreventionMMWR Recomm Rep47RR-3129\n12. FederJNPennyDMIrrinkiALeeVKLebrónJAWatsonN1998The hemochromatosis gene product complexes with the transferrin receptor and lowers its affinity for lig-and bindingProc Natl Acad Sci USA954147214779465039\n13. GomaaAHuHBellingerDSchwartzJTsaihSWGonzález-CossíoT2002Maternal bone lead as an independent risk factor for fetal neurotoxicity: a prospective studyPediatrics1101 Pt 111011812093955\n14. González-CossíoTPetersonKESaninLHFishbeinEPalazuelosEAroA1997Decrease in birth weight in relation to maternal bone-lead burdenPediatrics10058568629346987\n15. GuilarteTRToscanoCDMcGlothanJLWeaverSA2003Environmental enrichment reverses cognitive and molecular deficits induced by developmental lead exposureAnn Neurol531505612509847\n16. HallbergLRossander-HultenL1991Iron requirements in menstruating womenAm J Clin Nutr54104710581957820\n17. HammadTASextonMLangenbergP1996Relationship between blood lead and dietary iron intake in preschool children. A cross-sectional studyAnn Epidemiol6130338680621\n18. HansonEHImperatoreGBurkeW2001HFE gene and hereditary hemochromatosis: a HuGE review. Human Genome EpidemiologyAm J Epidemiol154319320611479183\n19. HaynesENKalkwarfHJHornungRWenstrupRDietrichKLanphearBP2003Vitamin D receptor Fok1 polymorphism and blood lead concentration in childrenEnviron Health Perspect1111665166914527848\n20. Hernández-AvilaMGonzález-CossíoTHernández-AvilaJERomieuIPetersonKEAroA2003Dietary calcium supplements to lower blood lead levels in lactating women: a randomized placebo-controlled trialEpidemiology14220621212606887\n21. JuskoTAHendersonCRJrLanphearBPCory-SlechtaDAParsonsPJCanfieldRL2008Blood lead concentrations < 10 μg/dL and child intelligence at 6 years of ageEnviron Health Perspect11624324818288325\n22. KwongWTFrielloPSembaRD2004Interactions between iron deficiency and lead poisoning: epidemiology and pathogenesisSci Total Environ3301–3213715325155\n23. LanphearBPHornungRKhouryJYoltonKBaghurstPBellingerDC2005Low-level environmental lead exposure and children’s intellectual function: an international pooled analysisEnviron Health Perspect11389489916002379\n24. LeePLHoNJOlsonRBeutlerE1999The effect of transferrin polymorphisms on iron metabolismBlood Cells Mol Dis255–637437910660486\n25. McLarenGDNathansonMHJacobsATrevettDThomsonW1991Regulation of intestinal iron absorption and mucosal iron kinetics in hereditary hemochromatosisJ Lab Clin Med11753904012019794\n26. MeyerPAMcGeehinMAFalkH2003A global approach to childhood lead poisoning preventionInt J Hyg Environ Health2064–536336912971691\n27. MillerDTPaschalDCGunterEWStroudPED’AngeloJ1987Determination of lead in blood using electrothermal atomisation atomic absorption spectrometry with a L’vov platform and matrix modifierAnalyst11212170117043445938\n28. MorganEHOatesPS2002Mechanisms and regulation of intestinal iron absorptionBlood Cells Mol Dis29338439912547229\n29. OnalajaAOClaudioL2000Genetic susceptibility to lead poisoningEnviron Health Perspect108suppl 1232810698721\n30. RobsonKJLehmannDJWimhurstVLLiveseyKJCombrinckMMerryweather-ClarkeAT2004Synergy between the C2 allele of transferrin and the C282Y allele of the haemochromatosis gene (HFE) as risk factors for developing Alzheimer’s diseaseJ Med Genet41426126515060098\n31. RosadoJLLópezPKordasKGarcía-VargasGRonquilloDAlatorreJ2006Iron and/or zinc supplementation did not reduce blood lead concentrations in children in a randomized, placebo-controlled trialJ Nutr13692378238316920858\n32. RoyCNPennyDMFederJNEnnsCA1999The hereditary hemochromatosis protein, HFE, specifically regulates transferrin-mediated iron uptake in HeLa cellsJ Biol Chem274139022902810085150\n33. Salter-CidLBrunmarkALiYLeturcqDPetersonPAJacksonMR1999Transferrin receptor is negatively modulated by the hemochromatosis protein HFE: implications for cellular iron homeostasisProc Natl Acad Sci USA96105434543910318901\n34. van PraagHKempermannGGageFH2000Neural consequences of environmental enrichmentNat Rev Neurosci1319119811257907\n35. WrightRO1999The role of iron therapy in childhood plumbismCurr Opin Pediatr11325525810349106\n36. WrightROShannonMWWrightRJHuH1999Association between iron deficiency and low-level lead poisoning in an urban primary care clinicAm J Public Health8971049105310394314\n37. WrightROTsaihSWSchwartzJWrightRJHuH2003Association between iron deficiency and blood lead level in a longitudinal analysis of children followed in an urban primary care clinicJ Pediatr142191412520247\n38. WrightROSilvermanEKSchwartzJTsaihSWSenterJSparrowD2004Association between hemochromatosis genotype and lead exposure among elderly men: the normative aging studyEnviron Health Perspect11274675015121519"
+}

batch_5/PMC2535641.json

@@ -0,0 +1,4 @@
+{
+    "id": "PMC2535641",
+    "text": "This is an academic paper. This paper has corpus identifier PMC2535641\nAUTHORS: Edith Chen, Michael Brauer\n\nABSTRACT:\nNo Abstract\n\nBODY:\nWe thank Clougherty and Kubzansky for their thoughtful review of our article (Chen et al. 2008). We view our article, as well as their article on exposure to violence, air pollution, and asthma etiology (Clougherty et al. 2007), as suggestive regarding how the social and physical environments operate in asthma. Although the nature of the interaction effects were different in these two studies, the broader point—that there are interactive effects between the social and physical environments in asthma —is consistent and is the key message that we wish to emphasize.We would like to address their specific comments. First, regarding temporal issues, Clougherty and Kubzansky raise the possibility that stress increases susceptibility to subsequent pollution. We agree that this is possible; we also recognize the possibility that chronic pollution exposure could heighten responses to subsequent stressors. As we stated in our “Discussion” (Chen et al. 2008), the time frame of assessments that were available to us for these analyses was not ideal, and future studies should more specifically coordinate the timing of exposures to both stress and air pollution.Second, we agree it is possible that saturation effects may occur at high levels of pollution exposure. However, because pollution levels in Vancouver (British Columbia, Canada) are not extreme (the range in our sample was 10–30 ppb nitrogen dioxide), we think this is an unlikely explanation.Third, regarding spatial covariance, in our study (Chen et al. 2008), family stress was measured at the individual level; thus, we do not have neighborhood-level stress maps or information on spatial patterns in stress. Although spatial covariance between socioeconomic status and air pollution has the potential to lead to confounding, the availability of individual measures of stress and air pollution exposure estimates at the resolution of individual addresses allowed us to evaluate interactions. Our longitudinal findings also diminish the likelihood of confounding. Further, previously published pollution maps (Henderson et al. 2007) have shown that, in our study area, air pollution levels are not spatially correlated with neighborhood socioeconomic status [e.g., see UBC (University of British Columbia) Centre for Health and Environment Research 2008].Fourth, we presented information about disease characteristics in Table 1 (Chen et al. 2008). We also controlled for asthma severity and medication use in all analyses, as described in our article under “Potential confounders.”Finally, we agree that it would be interesting to know whether stress by air pollution effects vary by age. However, given the limited sample size in our study, we were unable to test this possibility.\n\nREFERENCES:\n1. ChenESchreierHMCStrunkRBrauerM2008Chronic traffic-related air pollution and stress interact to predict biologic and clinical outcomes in asthmaEnviron Health Perspect11697097518629323\n2. CloughertyJELevyJIKubzanskyLDRyanPBSugliaSFCannerMJ2007Synergistic effects of traffic-related air pollution and exposure to violence on urban asthma etiologyEnviron Health Perspect1151140114617687439\n3. HendersonSBBeckermanBJerrettMBrauerM2007Application of land use regression to estimate long-term concentrations of traffic-related nitrogen oxides and fine particulate matterEnviron Sci Technol412422242817438795\n4. UBC (University of British Columbia) Centre for Health and Environment Research2008Annual Average Air Pollution Levels, Asthma Rates and Neighbourhood Income Levels in VancouverAvailable: http://www.cher.ubc.ca/UBCBAQS/images/Asthma_Vancouver.gif[accessed 21 July 2008]"
+}

batch_5/PMC2535655.json

@@ -0,0 +1,4 @@
+{
+    "id": "PMC2535655",
+    "text": "This is an academic paper. This paper has corpus identifier PMC2535655\nAUTHORS: No authors listed\n\nABSTRACT:\nNo Abstract\n\nBODY:\nA Question of Balance: Weighing the Options on Global Warming PoliciesWilliam D. NordhausNew Haven, CT:Yale University Press, 2008. 256 pp. ISBN: 978-0-300-13748-4, $28Arsenic Pollution: A Global SynthesisPeter Ravenscroft, Hugh Brammer, Keith RichardsHoboken, NJ:John Wiley & Sons, Inc., 2008. 640 pp. ISBN: 978-1-405-18601-8, $49.95Changes in the Human-Monsoon System of East Asia in the Context of Global ChangeCongbin Fu, J.R. Freney, J.W.B. Stewart, eds.Hackensack, NJ:World Scientific Publishing Co., 2008. 384 pp. ISBN: 978-981-283-241-2, $88Climate Solutions: A Citizen’s GuidePeter BarnesWhite River Junction, VT:Chelsea Green Publishing, 2008. 120 pp. ISBN: 978-1-6035-8005-2, $9.95Dire Predictions: Understanding Global WarmingMichael E. Mann, Lee R. KumpNew York:DK Publishing, 2008. 208 pp. ISBN: 978-0-7566-3995-2, $25Emissions Trading: Institutional Design, Decision Making and Corporate StrategiesRalf Antes, Bernd Hansjürgens, Peter Letmathe, eds.New York:Springer, 2008. 274 pp. ISBN: 978-0-387-73652-5, $119Energy and Climate Change: Creating a Sustainable FutureDavid ColeyHoboken, NJ:John Wiley & Sons, Inc., 2008. 672 pp. ISBN: 978-0-470-85313-9, $50Environmental Law, Policy, and Economics: Reclaiming the Environmental AgendaNicholas A. Ashford, Charles C. CaldartCambridge, MA:MIT Press, 2008. 1,088 pp. ISBN: 978-0-262-01238-6, $90Global Warming, Natural Hazards, and Emergency ManagementGeorge Haddow, Jane A. Bullock, Kim HaddowBoca Raton, FL:CRC Press, 2008. 280 pp. ISBN: 978-1-420-08182-4, $59.95International Documents on Environmental LiabilityHannes Descamps, Robin Slabbinck, Hubert BockenNew York:Springer, 2008. 372 pp. ISBN: 978-1-4020-8366-2, $189Malformed Frogs: The Collapse of Aquatic EcosystemsMichael LannooBerkeley:University of California Press, 2008. 288 pp. ISBN: 978-0-520-25588-3, $65Natural Climate Variability and Global WarmingRick Battarbee, Heather Binney, eds.Hoboken, NJ:John Wiley & Sons, Inc., 2008. 288 pp. ISBN: 978-1-4051-5905-0, $95Not One Drop: Betrayal and Courage in the Wake of the Exxon Valdez Oil SpillRiki OttWhite River Junction, VT:Chelsea Green Publishing, 2008. 256 pp. ISBN: 978-1-933-39258-5, $21.95Potential Impacts of Climate Change on U.S. Transportation: Special Report 290Committee on Climate Change and U.S. Transportation, National Research CouncilWashington, DC:National Academies Press, 2008. 296 pp. ISBN: 978-0-309-11306-9, $37Protecting Health in Europe from Climate ChangeB. Menne, F. Apfel, S. Kovats, F. RacioppiGeneva:WHO Press, 2008. 51 pp. ISBN: 978-928-907187-1, $15Regulating Water and Sanitation for the Poor: Economic Regulation for Public and Private PartnershipsRichard Franceys, Esther GerlachLondon:Earthscan, 2008. 320 pp. ISBN: 978-1-8440-7617-8, $97.50Retaking Rationality: How Cost Benefit Analysis Can Better Protect the Environment and Our HealthRichard Revesz, Michael LivermoreNew York:Oxford University Press, 2008. 262 pp. ISBN: 978-0-19-536857-4, $34.95\n\nREFERENCES:\nNo References"
+}

batch_5/PMC2536674.json

@@ -0,0 +1,4 @@
+{
+    "id": "PMC2536674",
+    "text": "This is an academic paper. This paper has corpus identifier PMC2536674\nAUTHORS: Peter Humburg, David Bulger, Glenn Stone\n\nABSTRACT:\nBackgroundTiling arrays are an important tool for the study of transcriptional activity, protein-DNA interactions and chromatin structure on a genome-wide scale at high resolution. Although hidden Markov models have been used successfully to analyse tiling array data, parameter estimation for these models is typically ad hoc. Especially in the context of ChIP-chip experiments, no standard procedures exist to obtain parameter estimates from the data. Common methods for the calculation of maximum likelihood estimates such as the Baum-Welch algorithm or Viterbi training are rarely applied in the context of tiling array analysis.ResultsHere we develop a hidden Markov model for the analysis of chromatin structure ChIP-chip tiling array data, using t emission distributions to increase robustness towards outliers. Maximum likelihood estimates are used for all model parameters. Two different approaches to parameter estimation are investigated and combined into an efficient procedure.ConclusionWe illustrate an efficient parameter estimation procedure that can be used for HMM based methods in general and leads to a clear increase in performance when compared to the use of ad hoc estimates. The resulting hidden Markov model outperforms established methods like TileMap in the context of histone modification studies.\n\nBODY:\n1 BackgroundHigh density oligonucleotide tiling arrays allow the investigation of transcriptional activity, protein-DNA interactions and chromatin structure across a whole genome. Tiling arrays have been used in a wide range of studies, including investigation of transcription factor activity [1] and of histone modifications in animals [2] and plants [3], as well as DNA methylation [4]. Analyses of these data are usually based either on a sliding window [1,5], or on hidden Markov models (HMMs) [6-8]. Other approaches have been suggested, e.g., by Huber et al. [9] and Reiss et al. [10], but are less common.Parameter estimates for sliding window approaches as well as hidden Markov models are typically ad hoc. Although there are some notable exceptions in gene expression studies [8,11], no established procedures exist to obtain good parameter estimates from tiling array data, especially in the context of chromatin immunoprecipitation (ChIP-chip) experiments. Attempts have been made to obtain parameter estimates by integrating genome annotations into the analysis [12]. While this may provide good results when investigating transcriptional activity in well studied organisms, it is limited by the quality of available annotations. For ChIP-chip studies the required annotation data is unavailable. A method for the localisation of transcription factors from ChIP-chip experiments by Keleş [13] does obtain the required parameter estimates from the data and allows for variations in length of enriched regions.Methods designed for the analysis of ChIP-chip data focus almost exclusively on the study of transcription factors [[6,7,10], and [13]]. While this is an important class of experiments, ChIP-chip studies are not limited to transcription factors, and the analysis of other ChIP-chip experiments may require new methods. One other area of active research that utilises ChIP-chip experiments is the study of histone modifications and chromatin structure [3]. Although both types of experiment employ the same technology, there are several important differences between them. Most importantly, the 147 bp of DNA bound by a histone complex are considerably longer than the typical transcription factor binding site, and the histone modifications of interest are expected to affect several neighbouring histones. Consequently the ChIP fragments derived from a transcription factor binding site all originate from a small region containing the given binding site while regions affected by histone modifications can be much longer than the ChIP fragments used. As a result of this, the data from histone modification experiments usually contain long regions of interest encompassing several non-overlapping ChIP fragments, rather than the short and relatively isolated peaks produced by transcription factor studies.Here we consider the analysis of data from a histone modification study in Arabidopsis [3]. These data consist of four ChIP samples for histone H3 with lysine 27 trimethylation (H3K27me3) and four histone H3 ChIP samples that act as a control. The aim of this analysis is to identify and characterise regions throughout the genome that exhibit enrichment for H3K27me3. It is desirable to use a method which is specifically designed for the analysis of histone modifications or flexible enough to accomodate the varying length of enriched regions. Furthermore, the method should obtain all parameter estimates from the data without the use of genome annotations and be robust towards outliers. Amongst the methods discussed above TileMap [7] comes closest to these requirements. Although it was developed with transcription factor analysis in mind it is general enough that it should provide useful results for other ChIP-chip experiments. This is emphasised by its application to histone modification [3] and DNA methylation [4] data as well as transtription factor analysis [14,15]. TileMap obtains some, but not all, of the required parameter estimates from the data. To provide a method which meets the requirements oulined above we develop a two state HMM with t emission distributions. All parameter estimates for the model are obtained by maximum likelihood estimation using the Baum-Welch algorithm [16] and Viterbi training [17]. These methods have the advantage that no prior knowledge about parameter values is required and one need not rely on frequently unavailable genome annotations. To assess the performance of our model, we apply it to simulated and real data. Results are compared to those produced by TileMap. The remainder of this article is structured as follows. In Section 2 the hidden Markov model is developed and MLEs for all parameters are derived in Section 4. The performance of the resulting model is assessed in terms of sensitivity and specificity on simulated data in Sections 2.3.3–2.3.6. In Section 2.3.7 the model is used to analyse a public ChIP-chip data set [3] and results are compared to the original analysis of these data.2 Results and discussionTiling array data consists of a series of measurements taken along the genome. Typically, microarray probes are designed to interrogate the genome at regular intervals. Design constraints such as probe affinity and uniqueness cause differences in probe density along the genome and can lead to large gaps between probes. Here we assume that the probe density is homogeneous except for a number of large gaps where the distance between adjacent probes is larger than max_gap. In the following analyses we use max_gap = 200 bp. This is identical to the value used by Zhang et al. [3], allowing for a direct comparison of results. Consider a ChIP-chip tiling array experiment with two conditions, a ChIP sample X1 targeting the protein of interest and a control sample X2. Each sample Xl has ml replicates (l = 1, 2) providing measurements for K genomic locations. The measurements for each probe are summarised by the \"shrinkage t\" statistic [18]:(1)yk=x¯1k−x¯2kv1k∗m1+v2k∗m2,where vlk∗ is a James-Stein shrinkage estimate of the probe variance obtained by calculating(2)vlk∗=λˆl∗slmedian2+(1−λˆl∗)slk2,and slk2 are the usual unbiased empirical variances and λˆl∗ is the estimated optimal pooling parameter(3)λˆl*=min(1,∑k=1KVar︿(slk2)∑k=1K(slk2−slmedian2)2).Other moderated t statistics have been suggested and could be used instead, most notably the empirical Bayes t statistic used by Ji and Wong [7] and the moderated t of Smyth [19]. All of these approaches are designed to increase performance compared to the ordinary t statistic by incorporating information from all probes on the microarray into individual probe statistics. Here we choose the \"shrinkage t\" because it does not require any knowledge about the underlying distribution of probe values while providing similar performance compared to more complex models [18].2.1 Hidden Markov ModelTo detect enriched regions we use a two state discrete time hidden Markov model with continuous emission distributions and homogeneous transition probabilities (Figure 1), i.e., the transition probabilities depend only on the current state of the model. The use of homogeneous transition probabilities assumes equally-spaced probes within each observation sequence as well as a geometric distribution of the length of enriched regions. As discussed above there will be some variation in probe distances. Using a relatively small value for max_gap ensures that the assumption of homogeneity holds at least approximately. The two states of the model correspond to enrichment or no enrichment in the ChIP sample. The model is characterised by the set of states S = {S1, S2}, the initial state distribution p, the matrix of transition probabilities A and the state specific emission density functions fi, i = 1, 2. The emission distribution of state Si is modelled as a t distribution with location parameter μi, scale parameter σi, and νi degrees of freedom.Figure 1Hidden Markov model for the analysis of ChIP-chip tiling array data.The use of t distributions has the advantage that their sensitivity to outliers can be adjusted via the degrees of freedom parameter, making them more robust and versatile than normal distributions. This is particularly useful when ν is estimated from the data [20]. It should be noted that the yk modelled here are from a t-like distribution (Equation (1)). While this in itself might suggest the use of t distributions for the fis, they are primarily chosen for their robustness. In the following we will refer to this model by its parameter vector θ = (θ1, θ2), where θ1 is the ordered pair (p, A) and θ2 the ordered triple (μ, σ, ν).Given a hidden Markov model θ and an observation sequence Y, it is possible to compute the sequence of states Q = q1q2...qK that is most likely to produce Y. There are several approaches to obtaining Q [21]. Usually Q is computed either by maximising the posterior probabilities P(qk = Si|Y; θ), k = 1, ..., K or by calculating the sequence that maximises P(Q|Y; θ). The latter provides the single most likely sequence of states and can be computed efficiently by the Viterbi algorithm [22]. For the particular model used here both approaches are equivalent.2.2 Parameter EstimationIn this section we will discuss two different approaches to estimate θ for the model described in Section 2.1. The methods under consideration are the EM algorithm, which is usually known as the Baum-Welch algorithm in the context of HMMs, and Viterbi training. While the Baum-Welch algorithm is guaranteed to converge to a local maximum of the likelihood function, it is computationally intensive. Viterbi training provides a faster alternative but may not converge to a local maximum.2.2.1 Initial EstimatesBoth optimisation algorithms discussed here require initial parameter estimates. These are obtained from the data by first partitioning the vector of observations Y into two clusters using k-means clustering [23]. From these clusters the location and scale parameters of the corresponding states are obtained as the mean and variance of the observations in the cluster. In the following, ν1 = ν2 = 6 is used as initial estimate for the degrees of freedom parameters.2.2.2 Baum-Welch AlgorithmThe Baum-Welch algorithm [16] is a well established iterative method for estimating parameters of HMMs. It represents the EM algorithm [24] for the specific case of HMMs. This algorithm can be used to optimise the transition parameters θ1 as well as the emission parameters θ2. Each iteration of the algorithm consists of two phases. During the first phase, the current parameter estimates are used to determine for each probe statistic in the observation sequence how likely it is to be produced by the different states of the model. In the second phase, parameters for the emission distributions of each state are estimated using contributions from all observations, according to the probability that they were produced by the respective state of the model. The state transition parameters are updated in a similar fashion, accounting for the probability of transitions between states based on the observation sequence and the current model. After each iteration this procedure results in a model which explains the observed data better than the previous one, approaching a locally optimal solution. Using this method parameter estimates are updated until convergence is achieved. The details of the resulting algorithm are outlined in Section 4.1.This method of parameter estimation is computationally expensive and time-consuming for a typical tiling array data set. The computing time can be reduced by fixing the degrees of freedom for the emission distributions in advance, thus avoiding the root-finding required for the estimation of these parameters. While this does not provide the same flexibility as estimating the required degree of robustness from the data it reduces the complexity of the optimisation problem. It is noted by Liu and Rubin [25] that attempts to estimate the degrees of freedom are more likely to produce results which are of little practical interest. The impact on classification performance of this choice is investigated in Section 2.3.The formulation of the Baum-Welch algorithm used in this article is based on the description given by Rabiner [21] and on the EM algorithm derived by Peel and McLachlan [26] for fitting mixtures of t distributions.2.2.3 Viterbi TrainingWhile the Baum-Welch algorithm described in Section 2.2.2 is expected to provide good parameter estimates, it is computationally expensive. A faster model-fitting procedure can be devised by replacing the first phase of the Baum-Welch algorithm with a maximisation step. This method was introduced in [17] as segmental k-means and is now commonly referred to as Viterbi training. Unlike the Baum-Welch algorithm which allows each probe statistic to contribute to the parameter estimates for all states, Viterbi training assigns each observation to the state that is most likely to produce the given probe statistic. Thus each observation contributes to exactly one state of the model. While each iteration of this method is faster than one iteration of the Baum-Welch algorithm some iterations may decrease the likelihood of the model, thus failing to advance it towards a useful solution. See Section 4.2 for further details on the implementation of Viterbi training used here.2.3 Testing2.3.1 Simulated DataTo assess the ability to distinguish between enriched and non-enriched probes of the models obtained by the different parameter estimation methods discussed in Section 2.2, we simulate data with known enriched regions. To ensure that the simulation study is providing meaningful results, it is based as closely on real data as possible. To this end, two independent analyses of the H3K27me3 data published by Zhang et al. [3] are carried out, one using TileMap [7], the other based on our model. The result of each analysis is used to generate a new dataset with known enriched regions. See Section 4 for further details. In the following these data are referred to as datasets I and II respectively. Since the simulation procedure is likely to bias results towards the model that was used in the process, we concentrate on the analysis of dataset I, with some results for dataset II presented for comparison. The use of data based on both models allows us to consider their performance under advantageous and disadvantageous conditions.2.3.2 Performance MeasureThe performance of different models on these data is determined in terms of false positive and false negative rates at probe level. While the relative importance of false positives and false negatives depends on the experiment under consideration, they are often equally problematic in the context of ChIP-chip experiments, especially when considering experiments which investigate differences between different cell lines or developmental stages, where all incorrect classifications are of equal concern. In this context, we define false positives as probes that are classified as non-enriched by the analysis of the real data but are called enriched in the subsequent analysis of simulated data, and vice versa for false negatives.The output of each model is the estimated posterior probability of enrichment for each probe. In practice, probe calls (\"enriched\" or \"non-enriched\") are generated from this posterior probability based on a 0.5 cut-off. For any given model, classification performance will change with the chosen threshold. Thus we assess model performance across a range of cut-offs, reporting the relative number of false positives and false negatives as well as the error rate. The latter is also used to determine the cut-off that minimises incorrect classification results, and model performance is judged on the numbers of incorrect classifications at this optimal cut-off and at the usual 0.5 cut-off, and on the distance between the optimal cut-off and 0.5. The trade-off between sensitivity and specificity provided by the different models is characterised with ROC curves and the associated AUC values.Another measure of interest is the ability to characterise the length distribution of enriched regions correctly. When studying chromatin structure the extent of structural changes is of interest; this is the case for the data studied in Section 2.3.7. This property of the different models is investigated in Section 2.3.6.2.3.3 Estimating Degrees of FreedomWe now consider the performance of both the Baum-Welch procedure and Viterbi training when all model parameters, including the degrees of freedom ν, are estimated from the data. Both parameter estimation methods are used to fit an HMM to datasets I and II, and the performance of resulting models is assessed in terms of the achieved error rate (Figure 2), ROC curves (Figure 3) and their associated AUC (Table 1) for both datasets. To assess how well these methods perform in comparison to an established algorithm, we also fit a TileMap model to the two simulated datasets. The three models are compared to each other, as well as an ad hoc model which simply uses, without optimisation, the initial parameter estimates used by the two parameter optimisation methods. When comparing the performance of these models on both simulated datasets, it is important to consider that the simulation procedure introduces a bias towards the underlying model.Figure 2Error rate for different models on datasets I and II. Error rate resulting from the different models on dataset I (left) and II (right). When the total number of incorrect probe calls is considered, both parameter estimation procedures outperform TileMap on dataset I for cut-offs larger than 0.2. Both Baum-Welch and Viterbi training provide models with an optimal cut-off close to 0.5, while TileMap significantly underestimates the posterior probability resulting in an optimal cut-off of 0.19. The models with optimised parameters show similar performance on both datasets. On dataset II TileMap's performance is reduced in comparison to the results on dataset I. The main differences between the models considered here occur at error rates of 0–0.08. The relevant area of the figures in the top row is magnified in the plots below.Figure 3ROC curves for different models on datasets I and II. TileMap and the models with Baum-Welch and Viterbi training parameter estimates show similar performance on dataset I (left) with a small advantage for the models with optimised parameters. Comparison with a model using ad hoc parameter estimates highlights the performance increase achieved by optimising model parameters. On dataset II (right) TileMap performs similarly to the model with ad hoc parameter estimates. Figures on the bottom provide a close-up view of the plots above.Table 1AUC for different models on both simulated datasets.TileMapBaum-WelchViterbi-TrainingViterbi-EMad hocdataset I0.99860.99980.99970.99980.9869dataset II0.97490.99950.99940.99950.9728All models with optimised parameters outperform TileMap on both simulated datasets. While TileMap performs well on dataset I it is only slightly better than the model with ad hoc parameter estimates.Estimating all parameters from the data with either the Baum-Welch algorithm or Viterbi training leads to models with high sensitivity, producing fewer false negatives than TileMap for any given cut-off [see Additional file 1]. At the same time they lead to an increased number of false positives [see Additional file 2] compared to TileMap, indicating a slight reduction of specificity. When considering the error rate it becomes apparent that both Baum-Welch and Viterbi training provide a favourable trade-off between sensitivity and specificity. These models reduce the number of incorrect classifications compared to TileMap both at the usual 0.5 cut-off and at the optimal cut-off. Moreover, while the Baum-Welch algorithm and Viterbi training both lead to models with an optimal cut-off close to 0.5 (0.51 and 0.42 respectively), TileMap provides an optimal cut-off of 0.19, indicating that it underestimates the posterior probability of enrichment. This becomes even more apparent when considering the result for dataset II where the optimal cut-off for TileMap is at 0.002 compared to 0.5 for Baum-Welch and 0.41 for Viterbi training. This result suggests that TileMap is more tuned towards avoiding false positives than false negatives. From the above results we estimate that the weight given to false positives by TileMap is approximately 3.2 and 26 times larger than the weight for false negatives on datasets I and II respectively. The ROC curves (Figure 3) provide further evidence that the models with MLEs outperform TileMap. Although all three models perform well on dataset I, both parameter optimisation methods lead to better results than TileMap. The benefits of optimising parameter estimates are further highlighted by the performance of the model with ad hoc estimates that is used as starting point for the optimisation procedures. On both datasets, optimised parameters provide a notable increase in performance, with TileMap performing only slightly better than the ad hoc model on dataset II.2.3.4 Fixed Degrees of FreedomEstimating ν, the degrees of freedom, for t distributions from the data is time-consuming and may not be very accurate, especially for relatively large values of ν. In this section we investigate the effect of fixing ν a priori for both states of the model. Only the case ν1 = ν2 is considered here. The remaining parameters are estimated from the training data using the Baum-Welch algorithm and Viterbi training with ν = 3, 4, ..., 50. For each value of ν, we report the error rate (Figure 4) as well as the AUC (Figure 5) on the simulated data.Figure 4Model performance for different choices of ν. The Baum-Welch model (red) performs better for relatively small values of ν while Viterbi training (blue) favours larger ν. For the optimal choice of ν the Baum-Welch parameter estimates lead to an optimal cut-off close to 0.5.Figure 5AUC for different choices of ν and increasing numberof iterations. Change in AUC for different choices of ν (left). The Baum-Welch model performs better for relatively small values of ν while Viterbi training favours larger ν. Improvements in AUC with increasing number of iterations (right). The performance of the Viterbi trained model improves substantially during the first five iterations. Further iterations only produce small changes in the AUC. The Baum-Welch method requires more iterations to obtain the same AUC as as the Viterbi model. After 20 iterations the Baum-Welch model starts to outperform the Viterbi model.For the best combination of ν and cut-off, both parameter estimation methods result in models with a classification performance comparable to the case of variable degrees of freedom (Figure 2). While the Baum-Welch algorithm tends to produce models with an optimal cut-off close to 0.5, Viterbi training only achieves this for large values of ν. Notably, the best classification performance of the Viterbi trained model is achieved with 14 degrees of freedom and a 0.37 cut-off compared to 7 degrees of freedom and a 0.49 cut-off from Baum-Welch. This results in a decreased performance of the Viterbi model relative to the Baum-Welch model at the 0.5 cut-off.2.3.5 ConvergenceTo reduce the time required for parameter estimation it is useful to limit the number of iterations. While each iteration of the Baum-Welch algorithm is guaranteed to improve the likelihood of the model, small changes to the parameter values do not necessarily lead to significant changes in the classification result. Furthermore, Viterbi training is not guaranteed to converge to a local maximum of the likelihood function and a likelihood based convergence criterion may not be appropriate for this method. Here we investigate the convergence of both algorithms based on the error rate and AUC to gauge the number of iterations required to achieve good classification results. Parameter estimation is performed with 60 iterations for both algorithms. Current estimates are used to classify the test data at every 5th iteration and AUC (Figure 5) and error rate (Figure 6) are determined.Figure 6Error rate at optimal and 0.5 cutoff for increasing number of iterations. Parameter estimates obtained by the Baum-Welch algorithm (filled symbols) and Viterbi training (open symbols) improve model performance with increasing nuber of iterations. Viterbi training quickly approaches its optimal solution and initially outperforms Baum-Welch. The final model produced by the Baum-Welch algorithm provides a lower error rate than Viterbi training.The most striking difference in the convergence behaviour of the two methods is that Viterbi training appears to obtain good parameter estimates within a small number of iterations. Further iterations of the algorithm do not improve results substantially, whereas the Baum-Welch procedure provides parameter estimates that are better than the ones obtained by Viterbi training, both in terms of likelihood and classification performance, but takes substantially longer to obtain these estimates. The Baum-Welch algorithm not only requires more iterations than Viterbi training, but the time required for each iteration is also longer.2.3.6 Length Distribution of Enriched RegionsWhen studying histone modifications one possible characteristic of interest is the length of enriched regions. To assess how accurately the different methods reflect the length distribution of enriched regions, we compare the length of regions predicted by TileMap and by the model (using Baum-Welch parameter estimates) to the length distribution of enriched regions in the simulated data (the \"true length distribution\"). Note that this length distribution may vary from the one found in real data. Nevertheless this comparison highlights some of the differences between the two models. Quantile-quantile plots of the respective length distributions show that TileMap systematically underestimates the length of enriched regions (Figure 7 (bottom left) and Figure 8 (bottom left)). While this effect is relatively small on dataset I there is some indication that it increases with region length and long regions may not be characterised appropriately by TileMap (Figure 7 (top left)). This observation is further supported by the length distribution of enriched regions produced by TileMap on dataset II (Figure 8 (left)). Enriched regions in dataset II are generally longer than regions in dataset I. This difference is not captured by TileMap. Both TileMap and the Baum-Welch trained model produce several regions that are shorter than the shortest enriched region in the simulated data (Figure 7 (bottom)). There are two possible explanations for these short regions. They may be caused by underestimating the length of enriched regions, possibly splitting one enriched region into several predicted regions, or they may represent spurious enriched results produced by the model. In each case there is the possibility that the occurrence of extremely short regions is caused either by an intrinsic shortcoming of the model or by artifacts introduced during the simulation process. Since the simulation relies on TileMap to identify enriched and non-enriched probes it is inevitable that some probes will be misclassified. Subsequently these probes may be included in the simulated data, causing short disruptions of enriched and non-enriched regions. A sufficiently sensitive model could detect these unintended changes between enriched and non-enriched states.Figure 7Length distribution of enriched regions from dataset I. Quantile-quantile plots comparing length distributions of enriched regions found with TileMap (left) and with the model based on maximum likelihood estimates (right) to the true length distribution of enriched regions in dataset I. Figures on the bottom provide a close-up view of the plots above. Each dot represents a percentile of the length distributions.Figure 8Length distribution of enriched regions from dataset II. Quantile-quantile plots comparing length distributions of enriched regions found with TileMap (left) and with the model based on maximum likelihood estimates (right) to the true length distribution of enriched regions in dataset II. Figures on the bottom provide a close-up view of the plots above. Each dot represents a percentile of the length distributions.To investigate further which of these is the case, we first examine the number of enriched probes contained in the short regions found by the Baum-Welch model and by TileMap respectively. The model with Baum-Welch parameter estimates found 126 regions with less than 10 probes. These regions contain a total of 866 probes of which 717 are in enriched regions. While this indicates that the majority of short regions is due to underestimating the length of enriched regions, several spurious probe calls remain. TileMap produced 249 regions with less than 10 probes, containing a total of 1781 probes, of which 1753 are in enriched regions. This is strong evidence that almost all of these short regions are caused by underestimating the length of enriched regions, and is consistent with the above observation that TileMap systematically underestimates the length of enriched regions.To investigate whether the spurious short regions produced by the Baum-Welch model are due to an intrinsic shortcoming of the model or are artifacts introduced by the simulation procedure, we turn to real data. Here we focus on enriched regions containing only a single probe, which are most likely to be false positives. On dataset I the Baum-Welch model produced six of these extremely short regions. One of these probes is a true positive from an enriched region containing ten probes, i.e., the length of this region is underestimated by the Baum-Welch model. Of the remaining five probes three are identical, leaving three unique probes to be investigated further. For each of these three probes, we determine its position in the real data and its distance from enriched regions identified by TileMap and by our model (Section 2.3.7). Two of the probes are found to be located close to enriched regions identified by TileMap (142 and 391 bp) and all three probes are contained within enriched regions identified by our model [see Additional file 3]. This suggests that these probes may have been misclassified by TileMap during the original analysis, leading to an overestimation of the number of false positives produced by the Baum-Welch model on dataset I.2.3.7 Application to ChIP-Chip DataTo investigate the performance of our model further, we apply it to the data of [3] and compare the result to the original analysis. Based on the results of the simulation study (Sections 2.3.3–2.3.6) we use the following procedure:1. Quantile normalise and log transform data;2. Calculate probe statistics (Equation (2));3. Obtain initial estimates (Section 2.2.1);4. Use 5 iterations of Viterbi training to improve initial estimates;5. Use 15 iterations of Baum-Welch algorithm to obtain maximum likelihood estimates;6. Apply resulting model to data to identify enriched regions.This results in the detection of 5285 H3K27me3 regions covering 12.9 Mb of genomic sequence. Of these enriched regions, 3962 (~75%) are overlapping at least one annotated transcript. A total of 4982 or about 18.9% of all annotated genes are found to be enriched for H3K27me3. While most of the enriched regions cover a single gene, some regions are found to contain up to seven genes (Figure 9(b)). Enriched regions are predominantly longer than 1 kb with some extending over more than 20 kb (Figure 9(c)).Figure 9Analysis of ChIP-chip data. (a) Gene density in areas surrounding genes that contain H3K27me3 enriched regions and genes that do not contain enriched regions. (b) Number of genes found in H3K27me3 regions. While most enriched regions cover a single gene, there is a substantial number of H3K27me3 regions that cover several genes and enriched regions are found to contain up to seven genes. (c) Length distribution of H3K27me3 regions.To assess whether there is a difference between regions of the genome that show H3K27me3 enrichment and the rest of the genome, we investigate the density of genes in the neighbourhood of genes that appear to be regulated by H3K27me3, and compare this to the gene density in other regions of the genome. For this purpose we obtain the gene density for the 50 kb upstream and downstream of each gene as (bp annotated as genes)/100 kb. The resulting gene densities for genes with and without enriched regions are summarised in Figure 9(a). There are visible differences between the two distributions which we test for significance with a two sided Kolmogorov-Smirnov test; this results in an approximate p-value of 2 × 10-15. The significance of this result is further confirmed by a resampling experiment: the smallest p-value obtained from a series of 10000 resampled datasets is 1 × 10-6.3 ConclusionWith the use of MLEs for all model parameters, our model clearly improves classification performance on simulated data compared to ad hoc estimates, and outperforms TileMap. While our model produced some short regions that appear to be false positives, they are readily explained as a result of the simulation process. Comparison of results on simulated and real data suggests that TileMap produced a large number of false negatives in the original analysis used as the basis for the simulation. Inevitably, these false negatives were selected as part of non-enriched regions during the simulation process. The fact that the model with Baum-Welch parameter estimates was able to identify these isolated enriched probes despite the non-enriched contexts where they appeared emphasises the high sensitivity of the model.TileMap's apparent tendency to penalise false positives more than false negatives clearly contributes to its relatively low performance in our comparisons which are based on the assumption that both types of error are equally problematic. While this is the case for the application considered here, one may argue that false positives are indeed of greater concern in some cases. When this is the case, TileMap's trade-off between sensitivity and specificity may lead to better results. However, it should be noted that the relative weights given to false positives and false negatives by TileMap can vary substantially between datasets. The parameter estimation procedure used for our model on the other hand provides consistent performance at the chosen cut-off.The model-fitting procedure derived from the results of the simulation study (Sections 2.3.3–2.3.6) provides a fast and reliable approach to parameter estimation. This method retains all the favourable properties of the Baum-Welch algorithm while utilising the reduced computing time provided by Viterbi training. The use of MLEs ensures that model parameters are appropriate for the data. Results from the simulation study show that estimating model parameters from the data improves the model's ability to recognise enriched regions of varying length and generally improves classification performance.3.1 Future WorkThe analysis of the H3K27me3 data (Section 2.3.7) largely confirms the analysis of [3] although there are some notable differences. Most importantly, the H3K27me3 regions detected by our analysis are longer than the ones determined by TileMap (Figure 10). While Zhang et al. [3] found few regions longer than 1 kb, our analysis indicates that over 70% of enriched regions have a length of at least 1 kb, with the longest region spanning over 20 kb. Accordingly we find more regions that extend over several genes (Figure 9(b)). This may have implications for conclusions about the spreading of H3K27me3 regions in Arabidopsis.At this stage, the biological significance of the observed difference in gene density in the neighbourhood of enriched and non-enriched genes is unclear. However, it indicates that the two groups of genes differ in a significant way. This suggests that the partition into enriched and non-enriched genes produced by our analysis is indeed meaningful.Figure 10Length distribution of enriched regions from real data. Length distribution of enriched regions as determined by TileMap (blue) and Baum-Welch (red). Region length is determined in terms of probes per region. Both distributions were truncated at 10 for the simulation, ensuring that all regions in the simulated data contain at least ten probes.The hidden Markov model presented in this article uses homogeneous transition probabilities, assuming that all probes are spaced out equally along the genome. To satisfy this assumption at least approximately, we use a fixed cut-off of 200 bp to partition the sequence of probe statistics such that there are no large gaps between probes. This arbitrary cut-off could be avoided by using a continuous time hidden Markov model.4 Methods4.1 Baum-Welch AlgorithmThe Baum-Welch algorithm [16] used to estimate parameters for our model is outlined in Section 2.2.2; further details are given below. Computing the likelihood of the long observation sequences produced by tiling arrays involves products of many small contributions. This typically results in likelihoods below machine precision. To avoid this effect computations are carried out in log-space, using the identity(4)ln(x + y) = ln(x) + ln (1 + eln(y)-ln(x)).In the following we use ln∑ to denote summations which should be computed via Equation (4). The sequence of probe statistics Y is split into D observation sequences Y (d) such that the distance between probes within each observation sequence is at most max_gap and the distance between the end points of different observation sequences is greater than max_gap.The emission distribution of state Si is given as(5)fi(yk;μi,σi,νi)=Γ(νi+12)Γ(νi2)−1σiπνi(1+(yk−μi)2σiνi)νi+1.For a given parameter set θ we can obtain new parameter estimates for transition probabilities by calculating(6)ξkijd=ln[P(qkd=Si,qk+1d=Sj|Y(d);θ)](7)=αkid+ln[aij]+ln[fi(yk+1d;θ2)]++β(k+1)jd−ln[P(Y(d);θ)].Here αk and βk are known as forward and backward variables. For observation sequence d, d = 1, ..., D, they are defined as(8)α1id=ln⁡[pi]+ln⁡[fi(y1d;θ2)],(9)α(k+1)jd=(ln∑i=1N(αkid+ln[aij]))+ln[fj(yk+1d;θ2)],where 1 ≤ i ≤ N, 1 ≤ j ≤ N, 1 ≤ k <Kd and(10)βKdid=0,(11)βkid=ln∑j=1N(ln[aij]+ln[fj(yk+1d;θ2)]+β(k+1)jd),where 1 ≤ i ≤ N, 1 ≤ i ≤ N, k = Kd - 1, ..., 1. Note that ln [P(Y (d); θ)] is given by ln⁡∑i=1NαKdd We then calculate(12)γkid=ln⁡[P(qkd=Si|Y(d);θ)](13)=αkid+βkid−ln⁡[P(Y(d);θ)](14)=ln⁡∑j=1Nξkijd.Combining the estimates from all observation sequences we obtain new parameter estimates for the transition probabilities:(15)ln⁡[aˆij]=ln⁡∑d=1Dln⁡∑k=1Kd−1ξkijd−ln⁡∑d=1Dln⁡∑k=1Kd−1γkid,(16)ln⁡[pˆi]=ln⁡∑d=1Dγ1id−ln⁡[D].Calculations for the re-estimation of θ2 may involve negative values and cannot be carried out in log-space.To obtain the required parameter estimates we first define ln⁡[τkid]=γkid and then compute(17)ukid=νi+1νi+(ykd−μi)2,(18)μˆi=∑d=1D∑k=1Kdτkidukidykd∑d=1D∑k=1Kdτkidukid,(19)σˆi=∑d=1D∑k=1Kdτkidukid(ykd−μˆi)2∑d=1D∑k=1Kdτkid.There is no closed form estimate for νi. To obtain νˆi one has to find a solution to the equation(20)[−ψ(νi2)+ln⁡(νi2)+1++1∑k=1Kdτkid∑k=1Kdτkid(ln⁡(ukid)−ukid)++ψ(νi+12)−ln⁡(νi+12)]=0where ψ is the digamma function. Standard root-finding techniques are employed to find a solution to (20).4.2 Viterbi TrainingViterbi training provides a faster alternative to the Baum-Welch algorithm. See Section 2.2.3 for a high level description of the algorithm. Details of the parameter estimation procedure are given below. Instead of calculating the conditional expectation of the complete data log likelihood, this algorithm first computes the most likely state sequence Q given the observation sequence Y and the current model θ. The sequence Y is partitioned according to Q, assigning each observation to the state that it most likely originated from. New estimates for θ1 are then obtained by calculating(21)pˆi=1D|{d=1,...,D:q1d=Si}|,(22)aˆij=|{d=1,...,D:qkd=Si and qk+1d=Sj}|∑d=1D(Kd−1).Updates for μ and σ are obtained as in Section 2.2.1. The degrees of freedom ν can be either fixed in advance or estimated from the data using Equation (20) by setting τkid=1 if (qkid,qk+1d)=(Si,Sj) and τkid=0 otherwise.4.3 Simulated DataIn a first step following the original analysis by [3], TileMap [7] is used with the HMM option to define enriched and non-enriched probes. Note that, although this classification of probes is not perfect, it can be assumed that most probes are assigned to the correct group. The length distribution of enriched and non-enriched regions detected by TileMap is used to determine the length distributions for the simulated data after removing all regions that contain less than 10 probes (Figure 10). Data are generated by first determining the length of enriched and non-enriched regions from the empirical length distributions and then sampling data points from the respective TileMap generated clusters. Following this procedure, 600 sequences with one to ten enriched regions in each sequence are generated. A second dataset is generated by applying the model described in Section 2. Note that, although this procedure relies on the classifications produced by the respective models, the resampling procedure will place individual probe values in a new context of surrounding probes, which may lead to different probe calls in the analysis of the simulated data. Prior to analysis all data are quantile normalised.5 AvailabilityThe parameter estimation methods used in this article are available as part of the R package tileHMM from the authors' webpage  and from CRAN. The simulated data used in this study is available from the authors' web page.6 Authors' contributionsPH conducted the research and wrote the manuscript. DB critically revised the manuscript. GS conceived the project. DB and GS provided supervision to PH. All authors have read and approved the final manuscript.Supplementary MaterialAdditional file 1False negative probe calls resulting from different models. For any given cut-off TileMap produces more false negatives than the Baum-Welch and Viterbi trained models.Click here for fileAdditional file 2False positive probe calls resulting from different models. For any given cut-off TileMap produces fewer false positives than the Baum-Welch and Viterbi trained models.Click here for fileAdditional file 3Origin of isolated enriched probes in dataset I. The isolated enriched probes identified in dataset I by the Baum-Welch model originate from enriched regions identified by the Baum-Welch model in the real data. Two out of three probes are located close to enriched regions identified by TileMap.Click here for file\n\nREFERENCES:\nNo References"
+}

batch_5/PMC2536676.json

@@ -0,0 +1,4 @@
+{
+    "id": "PMC2536676",
+    "text": "This is an academic paper. This paper has corpus identifier PMC2536676\nAUTHORS: Sanja Rogic, Ben Montpetit, Holger H Hoos, Alan K Mackworth, BF Francis Ouellette, Philip Hieter\n\nABSTRACT:\nBackgroundSecondary structure interactions within introns have been shown to be essential for efficient splicing of several yeast genes. The nature of these base-pairing interactions and their effect on splicing efficiency were most extensively studied in ribosomal protein gene RPS17B (previously known as RP51B). It was determined that complementary pairing between two sequence segments located downstream of the 5' splice site and upstream of the branchpoint sequence promotes efficient splicing of the RPS17B pre-mRNA, presumably by shortening the branchpoint distance. However, no attempts were made to compute a shortened, 'structural' branchpoint distance and thus the functional relationship between this distance and the splicing efficiency remains unknown.ResultsIn this paper we use computational RNA secondary structure prediction to analyze the secondary structure of the RPS17B intron. We show that it is necessary to consider suboptimal structure predictions and to compute the structural branchpoint distances in order to explain previously published splicing efficiency results. Our study reveals that there is a tight correlation between this distance and splicing efficiency levels of intron mutants described in the literature. We experimentally test this correlation on additional RPS17B mutants and intron mutants within two other yeast genes.ConclusionThe proposed model of secondary structure requirements for efficient splicing is the first attempt to specify the functional relationship between pre-mRNA secondary structure and splicing. Our findings provide further insights into the role of pre-mRNA secondary structure in gene splicing in yeast and also offer basis for improvement of computational methods for splice site identification and gene-finding.\n\nBODY:\nBackgroundSplicing of precursor mRNA is one of the essential cellular processes in eukaryotic organisms. Although this process has been extensively studied since the discovery of splicing three decades ago [1,2], resulting in a thorough understanding of the splicing pathway and identification of the numerous components of the splicing machinery, there are still many unanswered questions. For example, while the ability of pre-mRNA to form intramolecular interactions between short complementary segments in long yeast introns was initially suggested 20 years ago [3], the role of pre-mRNA secondary structure in splicing is not well understood.Introns in S. cerevisiae are known to have bimodal length distribution [4] and can be classified into short and long introns based on their length. The distance between the 5' splice site and the branchpoint sequence, also known as the 'lariat length' or 'branchpoint distance' (we also refer to it as linear branchpoint distance), is tightly correlated with intron length (with a Pearson correlation coefficient of r = 0.99 [5]) and can also be used to classify introns into long (5'L) and short (5'S) [3]. It was hypothesized that 5'L introns, for which the branchpoint distance is greater than 200 nt, can fold into secondary structures to optimize the positioning of the 5' splice site and branchpoint sequence to one that is optimal for spliceosome assembly [3]. This hypothesis was confirmed for a limited number of yeast introns by comprehensive biological experiments that demonstrated that the existence of such secondary structure elements is essential for splicing efficiency [6-11]. Structural elements that exhibit a similar effect on splicing efficiency were also found in introns of Drosophila melanogaster and related species [12]. Furthermore, in mammalian cells, folding of long intron sequences is facilitated by protein binding and interactions, which presumably shortens the long distance between essential splicing sequences [13].The nature of the base-pairing interactions within introns and their effect on splicing efficiency were most extensively studied in S. cerevisiae's ribosomal protein gene RPS17B, previously known as RP51B (YDR447C). It was shown that secondary structure interaction between two sequence segments located downstream of the 5' splice site and upstream of the branchpoint sequence promotes efficient splicing of the RPS17B pre-mRNA [7]. This interaction was further tested by comprehensive mutational and structure-probing analysis to determine the structure of the stem formed in the wildtype intron and the sensitivity of splicing efficiency to the alterations in this stem [8,9]. These studies demonstrated that complementary pairing between two ends of the RPS17B intron, but not necessarily the formation of the described stem, is essential for its efficient splicing in vitro and in vivo.While the authors of the previous studies speculated that the function of the complementary pairing is to shorten the branchpoint distance, they did not attempt to determine the secondary structure of the intron and the resulting 'structural' branchpoint distance. Thus a functional relationship between this distance and the splicing efficiency remains unknown.In this paper we use computational RNA secondary structure prediction to investigate the secondary structures of wildtype and mutant intron sequences within the S. cerevisiae RPS17B pre-mRNA. We present a unique algorithm for measuring 'structural' distance between two bases in an RNA secondary structure and use it to compute the distance between the 5' splice site and the branchpoint sequence based on the predicted secondary structure. Our analysis show that there is a tight correlation between structural branchpoint distances and splicing efficiency levels for all mutants examined.ResultsSecondary structure of RPS17B intron and the efficiency of splicingThe first goal of our study was to determine if the splicing efficiency results previously reported for RPS17B intron [8] can be correlated with the computationally predicted secondary structures of wildtype and mutant intron sequences.In this study the sensitivity of splicing to alterations in the stem formed in the RPS17B intron was tested by introducing mutations in the interacting regions designated UB1 (upstream box 1) and DB1 (downstream box 1). The assumption behind the mutant design was that any mutation within the stem would disrupt it and change the intron secondary structure in such a way that the resulting structural branchpoint distance (ds) would be greater than for the wildtype intron. The authors created 9 mutant introns within the RPS17B gene: 3mUB1 (3 nt mutation), 4mUB1 (4 nt), 5mUB1 (5 nt), 6mUB1 (6 nt) and 8mUB1 (8 nt), where mutations fall in the UB1 region; 3mDB1 (3 nt) and 5mDB1 (5 nt), where mutations fall in the DB1 region and are designed to restore the base-pairing disrupted by the mutations in the 3mUB1 and 5mUB1, respectively; and 3mUB1_3mDB1 and 5mUB1_5mDB1, which are double mutants. All of the single mutants are expected to disrupt the secondary structure, while the double mutants are predicted to restore it. The RPS17B intron was inserted into the coding region of the copper resistance gene (CUP1), which served as a reporter gene. Thus, yeast cells grown on copper containing medium will be viable only if the intron-containing Cup1 mRNA is spliced. The results of this assay suggested that for all single mutants except 8mUB1, splicing was reduced. Surprisingly, 8mUB1 had a similar growth rate on copper media as the wildtype intron suggesting that splicing was as efficient. Out of two double mutants, 5mUB1_5mDB1 was able to partially rescue copper resistance, while 3mUB1_3mDB1 did not. The authors hypothesized that these unexpected results were the result of some secondary structure rearrangements; however, the secondary structure of the mutants 8mUB1 and 3mUB1_3mDB1 was not explored.In order to investigate if the differences in the splicing efficiency levels are due to the differences in secondary structures, we computed the minimum free energy (MFE) structures of the introns using mfold [14,15], one of the most frequently used RNA secondary structure prediction tools. The comparative RNA secondary structure prediction, which is considered more reliable, requires a certain number of orthologous sequences which were available only for the wildtype RPS17B intron and not for the mutants created in [8].According to the mfold MFE predictions, the introduced mutations have the desired effect of disrupting the stem in all single mutants, but the compensatory mutations fail to restore it in two double mutants. Focusing on the positioning between the donor site and the branchpoint sequence, we compared the part of the structure that contains these two sites across all the mutants. The specified structural domain was almost identical for the 3mUB1, 5mUB1, 8mUB1, 3mDB1, 3mUB1_3mDB1 and 5mUB1_5mDB1 mutants, some of which have very different splicing efficiency levels (see Additional file 1). Moreover, the full secondary structures of the 3mUB1 and 3mUB1_3mDB1 mutants were almost identical with only three base-pairs difference, while the copper resistance experiment suggested significant differences in splicing efficiency. Therefore, it appears that differences in the splicing efficiency of Libri et al.'s [8] mutants cannot be attributed to differences in the computed MFE secondary structures of introns.However, considering only a single, minimum free energy secondary structure prediction of an intron might not be the appropriate approach. While functional, non-coding RNAs, such as tRNAs and rRNAs, have a strong evolutionary pressure to maintain their unique, functional structure, it is believed that mRNAs, whose primary role is to carry the protein coding information to the translation apparatus, do not have functional constraints on their global structure. Thus, instead of always folding into unique MFE structure, it is likely that mRNAs exist in a population of structures [16-18]. Another reason for considering suboptimal structures, especially when using computational prediction methods, is that RNA secondary structure prediction algorithms have limited accuracy and sometimes the correct structure is buried among the suboptimal predictions with free energies very close to the MFE [15,19,20].Structural branchpoint distances of suboptimal secondary structures and the efficiency of splicingBased on these considerations, we modified our approach to include not only the optimal, i.e., MFE structure, but also near-optimal predictions whose free energies are within 5% of the optimum. There is an exponential relationship between the free energy of a structure and its probability in the ensemble of all possible structures for a given sequence. The probability of a structure Si in the Boltzmann ensemble of all possible structures (S1, S2,...) for a given RNA sequence is given by:(1)P(Si)=e−ΔG(Si)/RTQwhere ΔG(Si) is the free energy of structure Si, Q = ΣS e-ΔG(S)/RT the partition function for all possible secondary structures for the given sequence, R is the physical gas constant, and T is the temperature. The probability of a secondary structure is also called the Boltzmann weight of that structure.From the equation we can see that the lower the free energy of a structure the higher its probability, thus, the predictions within 5% from the MFE also represent the most probable structures for a given sequence, with the MFE prediction being the one with the highest probability.We used RNAsubopt algorithm [20] to sample 1000 suboptimal structures within 5% of the MFE for each considered intron. RNAsubopt first calculates all suboptimal structures within a user defined energy range and then produces a random sample of structures, drawn with probabilities equal to their Boltzmann weights. Therefore, RNAsubopt computes a representative sample of the secondary structure space within 5% of the MFE.Since the pair-wise structure comparison and distance estimation approach that we used for MFE structure predictions were not applicable to large number of structures we had to devise a new way to quantify the structural distance between the donor site and the branchpoint sequence. We designed an algorithm that converts an RNA secondary structure into a graph and then applies a shortest-path algorithm from graph theory to compute the shortest distance between two bases in the secondary structure. To the best of our knowledge this is the first algorithm for structural distance computation. More details are given in Materials and Methods.For each secondary structure prediction, we computed the exact distance between the donor site and the branchpoint sequences (ds) using the shortest-path algorithm. The average structural branchpoint distances are given in Table 1. We assigned descriptive splicing efficiency labels based on the gel images in Figure 2A in [8]. The distributions of computed structural branchpoint distances for each of the RPS17B mutants are given in Figure 1.Table 1Average structural branchpoint distances for the wildtype (wt) RPS17B intron and Libri et al.'s [8] intron mutants.mutantaverage dssplicing efficiencywt26.67efficient (1)3mUB127.67slightly reduced (2)5mUB128.42slightly reduced (2)8mUB127.94efficient (1)3mDB137.55inhibited (4)5mDB139.19inhibited(4)3mUB1_3mDB137.44inhibited (4)5mUB1_5mDB133.81slightly reduced (2)6mUB146.31inhibited (4)4mUB132.08reduced (3)Levels of splicing efficiency were approximated from the gel images in Figure 2A in [8]. The numbers within parentheses correspond to numerical values assigned to descriptive splicing efficiency labels.Figure 1Distribution histograms of structural branchpoint distances for (a) wt, (b) 3mUB1, (c) 5mUB1, (d) 8mUB1, (e) 3mDB1, (f) 5mDB1, (g) 3mUB1_3mDB1, (h) 5mUB1_5mDB1, (i) 6mUB1, and (j) 4mUB1 introns.Figure 2Cumulative distributions of structural branchpoint distances for all Libri et al.'s [8] intron mutants.These results suggest an interesting correlation between the average structural branchpoint distance and the splicing efficiency levels: sequences that are more efficiently spliced (wildtype, 3mUB1, 5mUB1, 8mUB1, 5mUB1_5mDB1, and 4mUB1) have lower values for the average distance than those that are poorly spliced. After assigning numerical values to the descriptive splicing efficiency labels (efficient = 1, slightly reduced = 2, reduced = 3 and inhibited = 4) we obtain a Pearson correlation coefficient of 0.87.The histograms in Figure 1 offer further insights into the relationship between structural branchpoint distances of introns and their efficiency of splicing; introns that are spliced efficiently or with slightly reduced efficiency have large frequency of suboptimal structures with ds < 10. Mutant 5mUB1_5mDB1, which does not have this prominent peak in its distribution histogram and mutant 4mUB1, which has reduced splicing efficiency, but not completely inhibited, still have higher frequency of structures with ds < 20 than the remaining, poorly spliced mutants. The correlation coefficient between splicing efficiency level and the proportion of structures with ds < 20 is 0.85.Finally, the cumulative distribution plot of structural branchpoint distances for all mutants, where lines are labeled according to the splicing efficiency levels (efficient – blue, slightly reduced – green, reduced – black and inhibited – red) shows a clear separation of spliced and unspliced mutants (Figure 2).Upon closer inspection we noticed that most of the structures with ds < 10 have ds = 4. Analysis of the secondary structures of these sequences reveals that this distance corresponds to a structural conformation where the donor and branchpoint sequences have two base-pairing interactions between them (see Section 2.1.3). The observed base-pairing interactions are not necessarily inconsistent with established models of the splicing process, according to which spliceosomal snRNAs interact with the donor site and the branchpoint sequence, since the base-pairing can be easily disrupted after the splicing factors have been aligned properly.Structural branchpoint distances and the efficiency of splicing for other published RPS17B mutantsIn order to test the generality of the observed correlation between splicing efficiency levels and structural branchpoint distances we also analyzed the RPS17B intron mutants described in [9]. These are mut-UB1i, which has an inverted UB1 sequence; mut-DB1i, which has an inverted DB1 sequence; mut-UB1iDB1i, which has both UB1 and DB1 sequences inverted to make them complementary to each other; mut-5, which reduces the consecutive pairing region to 5 base-pairs, mut-12; which improves pairing to 12 consecutive base-pairs (eliminating one one-nucleotide bulge); and mut-18, which extends pairing to 18 consecutive base-pairs (eliminating all three bulges in the pairing region). The authors compared splicing efficiency of the wildtype and mutant introns by analyzing the formation of spliceosomal complexes. Based on their gel images, we assigned descriptive and numerical splicing efficiency labels to the tested sequences (see Table 2). The average structural branchpoint distances of 1000 suboptimal structures sampled from within 5% of the MFE for each mutant are given in Table 2.Table 2Average structural branchpoint distances for the wildtype (wt) RPS17B intron and Charpentier and Rosbash's [9] intron mutants.mutantaverage dssplicing efficiencywt26.67normal (2)mut-UB1i42.51reduced (3)mut-DB1i35.95reduced (3)mut-UB1iDB1i26.39improved (1)mut-532.14reduced (3)mut-1224.82improved (1)mut-1825.30improved (1)We inferred levels of splicing efficiency based on Figures 2 and 3 and Table 1 in [9]. The numbers within parentheses correspond to numerical values assigned to descriptive splicing efficiency labels.The branchpoint distance results for these mutants are similar to those of Libri et al.'s [8] mutants; the average structural branchpoint distances are lower for the sequences that are efficiently spliced (wildtype, mut-UB1iDB1i, mut-12, and mut-18). After assigning numerical values to the descriptive splicing efficiency labels (improved = 1, normal = 2 and reduced = 3), we obtain the correlation coefficient as 0.85. This, again, corresponds to the ability of these sequences to fold in such a way as to bring the donor site and the branchpoint sequences close to each other; each of the efficiently spliced sequences has a large fraction of predicted secondary structures for which ds < 10 (Figure 3 and Additional file 2). The mutants that show reduced splicing have very few of these structures (0.02% for mut-UB1i and 0.33% for mut-5), except for mut-DB1i, which has 11.3% of structures with ds < 10. However, this is still significantly lower than for the efficiently spliced mutants. Again, the cumulative distribution plot clearly separates mutants based on their splicing efficiency (Figure 3).Figure 3Cumulative distributions of structural branchpoint distances for all Charpentier and Rosbash's [9] intron mutants.Base-pairing probabilities of the RPS17B intron and the efficiency of splicingThe branchpoint distance analysis of S. cerevisiae's RPS17B intron suggests that the ability to form highly probable secondary structures (within 5% of the MFE) with short distance between the donor site and the branchpoint sequence seems to be required for efficient splicing of the intron. The short structural branchpoint distance for the RPS17B intron results from two base-pair interactions: between the first intron base (G) and the third base of the branchpoint sequence (C); and between the second base in the intron (U) and the second base of the branchpoint sequence (A) (see Figure 4). It is possible to compute the probability of these base-pairing interactions directly using a dynamic programming algorithm that computes the partition function [21]. The base-pair probability reflects a sum of all probability-weighted structures in which the chosen base-pair occurs. Thus, these base-pairing probabilities also take into account the structures that were not within 5% from the MFE, eliminating the necessity to chose an arbitrary percent suboptimality value. The base-pair probabilities can be computed using RNAfold [22], another frequently used program for RNA secondary structure prediction.Figure 4A part of the wildtype RPS17B intron secondary structure that shows base-pairing between the donor site and the branchpoint sequence. The highlighted stem is the same as the one identified in [9] using experimental structure probing.The base-pair probability values for the wildtype RPS17B intron and all of Libri et al.'s [8] mutants are given in Table 3. The probability values for the two base-pairs (G-C and U-A) are identical up to second decimal place for each intron sequence and that is why only one number is shown in the table. It can be observed that all of the efficiently spliced sequences have higher base-pair probabilities than the poorly spliced sequences (r = -0.92). The correlation is not strictly linear since, for example, the mutant sequence 8mUB1 has almost the same base-pair probability value as 3mUB1 and 5mUB1, although it is more efficiently spliced than these two. Similarly, the double mutant 5mUB1_5mDB1 is more efficiently spliced than 4mUB1, but this is not reflected in the base-pair probability values.Table 3Base-pairing probabilities of contact conformation (Figure 4) for the wildtype (wt) RPS17B intron and Libri et al.'s [8] intron mutants.mutantbase-pairing probabilitysplicing efficiencywt0.40efficient3mUB10.33slightly reduced5mUB10.31slightly reduced8mUB10.34efficient3mDB10.01inhibited5mDB1< 0.01inhibited3mUB1_3mDB10.01inhibited5mUB1_5mDB10.11slightly reduced6mUB10.05inhibited4mUB10.18reducedFor Charpentier and Rosbash's mutants, the base-pair probabilities are also higher for the sequences that are more efficiently spliced (Table 4): all of the sequences that are efficiently spliced (wildtype, mut-UB1iDB1i, mut-12, and mut-18) have base-pair probabilities of 0.40, while the other sequences have lower values (r = -0.85).Table 4Base-pairing probabilities of contact conformation for the wildtype (wt) RPS17B intron and Charpentier and Rosbash's [9] intron mutants.mutantbase-pairing probabilitysplicing efficiencywt0.40normalmut-UB1i0.04reducedmut-DB1i0.25reducedmut-UB1iDB1i0.40improvedmut-50.04reducedmut-120.40improvedmut-180.40improvedOverall, based on the results for Libri et al.'s [8] and Charpentier and Rosbash's [9] mutants it seems that, at least for RPS17B intron, base-pair probabilities for the two base-pairs formed between the first two bases of the intron and the second and third base of the branchpoint sequence are good indicators of splicing efficiency. We will see in the following sections that this is not a general requirement for all genes. Taken together with the observed correlation between the splicing efficiency levels and structural branchpoint distances the results are consistent with the following hypothesis: the existence of highly probable secondary structures that have short branchpoint distance is required for efficient splicing of yeast introns.Experimental testing of the hypothesisIn order to test the validity of the proposed hypothesis, we designed and functionally tested in vivo a series of RPS17B intron mutants. To assay the effect of these mutations on splicing we opted to introduce the mutated intron sequences at their endogenous locus, instead of within the CUP1 gene as was previously done [8,9]. This allows us to analyze the splicing of this intron within its normal context of flanking DNA sequences. We estimated the splicing efficiency directly from protein expression levels, which were quantified using a fluorescence imaging system.Using protein expression as a measurement of splicing efficiency requires that: 1) the level of protein abundance is proportional to the mRNA abundance (for a given gene) in the cell and, 2) the abundance of mRNA in the cell reflects any change in splicing efficiency. To demonstrate that RPS17B follows these general rules, we analyzed a number of Libri et al.'s [8] mutants that have previously documented changes in mRNA levels for their protein expression levels. The sequences tested were the wildtype RPS17B intron, and the 5mUB1, 3mUB1, 8mUB1, 5mDB1, and 3mDB1 mutated introns. The levels of protein expression, as shown in Figure 5, are proportional to the levels of copper-resistance in the copper growth assay in [8]. Moreover, our approach is able to provide a quantifiable measure for mutants such as 3mDB1 and 5mDB1, which did not support any growth in the copper growth assay. Thus, using changes in protein expression levels in the context of different intron sequences to assay the effects of mutations on splicing efficiency is a valid approach.Figure 5Protein expression levels for the RPS17B gene containing some of Libri et al.'s [8] mutant introns. Expression levels are normalized with respect to the internal loading control and plotted as a fraction of the wildtype expression level. Shaded boxes represent the mean value for several different samples and error bars represent +/- 1 standard deviation for these samples. The error bar for the wildtype intron comes from the comparison of two different wildtype samples.New RPS17B intron mutantsWe designed 8 new RPS17B intron mutants for the purpose of testing our current model of correlation between intronic pre-mRNA secondary structure and splicing efficiency. The most important structural characteristic used for mutant design was structural branchpoint distance (ds) of its MFE and suboptimal structures. Four mutants that are predicted to splice efficiently were designed to have multiple suboptimal structures with contact conformation (Figure 4) and short average structural branchpoint distance (these mutants are labeled with letter 'S', which stands for short ds). The only exception is mutant rps17b-S2, which does not have any suboptimal structures with contact conformation, but still exhibits a short structural branchpoint distance (most of the suboptimal predictions have ds = 10). This mutant was designed to test whether contact conformation, rather than the resulting short structural branchpoint distance, is important for splicing. Four mutants that are predicted to have reduced splicing were designed not to have any structures with contact conformation or otherwise short structural branchpoint distances (these mutants are labeled with letter 'L', which stands for long ds).The mutant design was based on mfold predictions, while RNAsubopt predictions where used post-experimentally to analyze the results. Mfold also samples the suboptimal space of secondary structures, however it does not compute all possible structures and the sample is much smaller. Although the distribution of ds computed based on structure predictions by mfold is similar to the one based on RNAsubopt predictions, the average distances for RNAsubopt predictions are not as distinct between 'S' and 'L' mutants as ones based on mfold predictions.Table 5 shows average ds for newly designed mutants based on RNAsubopt predictions and base-pair probabilities computed by RNAfold. The analogous table based on mfold suboptimal predictions, which was used in the design process is given in Additional file 3.Table 5Characteristics of newly designed RPS17B mutants.mutantavg(ds)bp probwt26.670.40rps17b-L143.630.0rps17b-L241.110.0rps17b-L334.050.21rps17b-L432.980.04rps17b-S124.550.40rps17b-S229.620.03rps17b-S312.650.80rps17b-S49.270.70avg(ds) – average structural branchpoint distances of 1000 suboptimal structures predicted by RNAsubopt; bp prob – base-pairing probability of interaction between the donor site and the branchpoint sequence based on the partition function.As seen in Figure 6, mutants rps17b-L1, rps17b-L2 and rps17b-L4 have reduced protein expression levels when compared to the wildtype as expected. Mutant rps17b-L3 has reduced splicing efficiency but not as much as the other three mutants with long structural branchpoint distances. As previously explained, this mutant was designed to have reduced splicing based on suboptimal predictions by mfold, which failed to predict any structures with ds < 10. However, RNAsubopt, which does a more rigorous sampling of the suboptimal space, detected a small fraction of suboptimal structures that have ds < 10 (see Additional file 4). This is in agreement with the relatively high probability of base-pairing interaction between the donor site and the branchpoint sequence (0.21).Figure 6Protein expression results for the RPS17B gene containing the newly designed mutant introns.Mutants rps17b-S1, rps17b-S2, and rps17b-S3 are all spliced efficiently, as predicted. The efficient splicing of mutant rps17b-S2, which has short structural branchpoint distance (ds = 10) without contact conformation in many of the predicted structures, suggests that a specific structural arrangement between the donor site and the branchpoint sequence is not required for efficient splicing. Mutant rps17b-S4 shows reduced levels of protein abundance, which is in disagreement with our prediction. The mutated sequence for this mutant has the same location as the mutated sequence for the mutant rps17b-S3, which is efficiently spliced, thus we can exclude the possibility that the discrepancy in splicing is sequence-based. A possible explanation for this phenomenon may be the existence of a very thermodynamically stable stem (with free energy ΔG = -36.6 kcal/mol) that holds the 5' splice site and the branchpoint together (analogous stems in wildtype introns have much higher free energy, see Section 2.3). This stem may be too stable to be disrupted, which might prevent the spliceosome to bind to the splice signals [8]. Overall, the results on the new RPS17B intron mutants are consistent with the proposed model of the role of intronic secondary structure in gene splicing in yeast.Selecting additional genes for experimental validationTo further validate our hypothesis regarding the role of intron secondary structure in splicing, we selected additional yeast, intron-containing genes to test our model. The selection criteria were: the linear distance (number of nucleotides) between the donor site and the branchpoint sequence is greater than 200 nt (5'L introns); the intron does not contain an snRNA gene; the gene is not essential (i.e., cells are viable if the gene is mutated or deleted); and the protein product has relatively high abundance in the cell, is amenable to c-terminal tagging, and has molecular weight between 20–120 kDa (to facilitate manipulation).From our initial dataset of 98 yeast genes that contain 5'L introns (see Materials and Methods), 18 genes matched the selection criteria (17 of these were ribosomal protein genes). We selected two of these for the experiments: the ribosomal protein gene RPS6B (YBR181C) and the amino-peptidase gene APE2 (YKL157W).The RPS6B gene contains one intron of length 352 nt, with a linear branchpoint distance (the distance between the 5' splice site and the branchpoint sequence) of d = 329 nt. The computed structural branchpoint distance (ds) is 18 for the MFE and all the suboptimal computationally predicted secondary structures within 5% of the MFE. Thus for this intron, unlike for the RPS17B intron, the donor and branchpoint sequences are not base-paired.The APE2 gene contains one intron of length 383 nt, with a linear branchpoint distance of d = 327 nt. One of the suboptimal structures within 5% of the MFE has a structural branchpoint distance of 6 and the others have greater distances. In the suboptimal prediction that has ds = 6 there is no base-pairing interactions between the donor and branchpoint sequences.RPS6B intron mutantsWe designed intron mutants for the RPS6B gene in a similar manner as for the RPS17B gene: the mutants that are supposed to have efficient splicing were designed to have similar structural branchpoint distances as the wildtype intron, and the mutants that are supposed to have reduced splicing were designed to have longer distances (see Additional file 5). Table 6 shows average structural branchpoint distances for a sample of 1000 suboptimal predictions within 5% of the MFE and the probability of short branchpoint distance derived form the base-pairing probabilities. The reported probability is the highest base-pair probability between the first donor nucleotide and any nucleotide within 20 bases away from the branchpoint adenosine. This guarantees that the branchpoint distance in a secondary structure that contains that base-pair will be no longer than 20.Table 6Characteristics of newly designed RPS6B mutants.mutantavg(ds)bp probwt18.060.84rps6b-L136.740rps6b-S119.080.65rps6b-S218.040.84rps6b-S318.040.83rps6b-S418.090.84rps6b-S522.000avg(ds) – average structural branchpoint distance; bp prob – probability of ds < 20.From Figure 7 we can see that all of the 'S' mutants, which have structural branchpoint distances similar to the wildtype intron, are expressed at levels similar to the wildtype. Mutant rps6b-L1, which has avg(ds) = 40 shows a reduction in splicing efficiency. The probability of ds < 20 also correlates well with the protein expression data except for mutant rps6b-S5 for which ds > 20 for all suboptimal predictions. Thus, for the RPS6B gene, structural branchpoint distances slightly longer than 20 seem to be still optimal for splicing. To summarize, the protein expression data for the RPS6B gene containing designed intron mutants are compatible with our proposed model of splicing efficiency dependence on the structural branchpoint distance.Figure 7Protein expression results for the RPS6B gene containing the newly designed mutant introns.APE2 intron mutantsUsing the same selection criteria as before, we designed six APE2 intron mutants. The values for average ds and the probabilities of structural branchpoint distance shorter than 20 are given in Table 7, and the histograms of structural branchpoint distance distributions are given in Additional file 6.Table 7Characteristics of newly designed APE2 mutants.mutantavg(ds)bp probwt27.900.37ape2-L175.730ape2-L269.680ape2-S18.930.82ape2-S223.330.50ape2-S324.600.45ape2-S425.140.42ape2-S54.100.99avg(ds) – average structural branchpoint distance; bp prob – probability of ds < 20.The experimental results are consistent with our prediction for five out of seven mutants: mutants ape2-S1, ape2-S2, ape2-S3 and ape2-S5 all have a level of protein abundance similar to the wildtype (Figure 8) and mutant ape2-L1 shows significantly reduced expression as expected. Mutant ape2-L2, which was expected to have reduced protein abundance as a consequence of reduced splicing efficiency, is expressed at the same level as the wildtype. Also, mutant ape2-S4 has reduced splicing despite the fact that it has a similar distribution of structural branchpoint distances as the wildtype intron. Since this mutant has the mutation at the same location as ape2-L1 (see Materials and Methods), it is possible that the intron segment that we mutated was important for splicing (e.g., contained a splicing enhancer). Overall, the results for APE2 mutants support our hypothesis of the role of structural branchpoint distance in gene splicing.Figure 8Protein expression results for the APE2 gene containing the newly designed mutant introns. Protein expression level is normalized with respect to wildtype expression level. Shaded boxes represent the mean value for several different samples and error bars represent +/- 1 standard deviation for these samples.Shortening of branchpoint distances by zipper stemsThe splicing efficiency study of RPS17B, RPS6B and APE2 genes containing wildtype and mutant introns supports our hypothesis that short structural branchpoint distances are required for efficient splicing. Although these distances are computed in the context of the secondary structure of the entire intron, our hypothesis is still consistent with the original hypothesis [3] that attributes the shortening of a long branchpoint distance to a single stem. Such stems, which we will refer to as 'zipper' stems, since they 'zip' the intron, are probably essential for achieving a short structural branchpoint distance. If we analyze the computed secondary structures of the RPS17B, RPS6B and APE2 wildtype introns we can easily identify stable stems whose 3' and 5' constituents are close to the donor site and the branchpoint sequence (Figure 9). The zipper stem labeled in the RPS17B intron is the same as the one identified in [9] using experimental structure probing.Figure 9Portions of the RPS17B, RPS6B and APE2 introns containing computationally identified zipper stems. The free energy values (ΔG) for the shaded zipper stem are given in parentheses. Stems conserved between Saccharomyces sensu stricto group are also labeled.To further test the functional importance of the identified zipper stems we performed comparative structure analysis using several closely related yeast species (S. paradoxus, S. mikatae, and S. bayanus, as well as S. cerevisiae, all belonging to the Saccharomyces sensu stricto group). We used multiple sequence alignments to extract the orthologous intron sequences for our three genes [23,24]. Both RPS17B and RPS6B intron alignments contain three sensu stricto sequences. The multiple sequence alignment for APE2 contains all four sequences; however, these are not intronic sequences but sequences from the exon 2 of the APE2 gene. This error is due to the old S. cerevisiae annotation which mapped two genes to the location of the current APE2 gene [25].We computed the consensus structure of RPS17B and RPS6B introns using Alifold [26]. The previously indicated zipper stems were found in the consensus structures for both genes (Figure 9), thus suggesting evolutionary conservation of these structural elements.DiscussionThe hypothesis that secondary structure interactions within yeast introns are needed for efficient splicing was proposed two decades ago [3]. Since then, experimental evidence in support of this hypothesis was found for several of S. cerevisiae's introns [6-11]. These studies identified complementary segments located downstream of the donor site and upstream of the branchpoint sequence whose base-pairing interactions are essential for splicing. It is conjectured that the function of the formed stem is to bring the donor site and the branchpoint sequence closer together so that they are in optimal alignment for spliceosome assembly.In this paper we use computational RNA secondary structure prediction to study structural requirements for efficient splicing in yeast. Our approach considers a representative sample of suboptimal structures with free energies close to the MFE and it also considers the entire secondary structure of an intron, rather than a single stem, both of which are more consistent with the nature of RNA molecules. Furthermore, the approach includes a calculation of the structural branchpoint distance, which is used to quantify the effect of the secondary structure on the distance between the donor site and the branchpoint sequence and can easily be correlated with splicing efficiency measurements. Using this method we were able to identify structural characteristics of the RPS17B intron and its mutants that seem to be responsible for their splicing differences. Notably, mutants that are likely to have a short structural branchpoint distance are spliced more efficiently.Based on our model of structural requirements for efficient splicing we computationally designed intron mutants for three S. cerevisiae genes, RPS17B, RPS6B and APE2, and experimentally tested their splicing efficiency. The results were mostly consistent with our model, with a few exceptions (rps17b-L3, rps17b-S4, ape2-L1 and ape2-S4) which may be due to some structural characteristics of mutants that are not considered by the current model or some inherent approximations in the model that are discussed below. Some of the intron mutants that were designed to have different structural characteristics and splicing efficiencies have mutations at the same locations (e.g., rps17b-L3 and 8mUB1; rps17b-S3 and 3mDB1; rps6b-L1 and rps6b-S3). The experimental results that confirm differences in splicing between these pairs of mutants indicate that the secondary structure of a pre-mRNA, rather than the underlying primary sequence, is responsible for differences in splicing.We also tested our model on the YRA1 gene intron, whose splicing efficiency had previously been studied by Preker and Guthrie [27]. The published experimental results were in agreement with our model; the efficiently spliced mutants (ΔL10 and ΔTCC/GGA) had higher base-pair probabilities than the poorly spliced sequences (wildtype intron and mutants ΔR/L10, TCC ΔL10, GGA ΔL10 and TCC+GGA ΔL10) (data not shown).Our current model is simplified in the sense that the secondary structure of an intron is computed disregarding its flanking sequences, and the three dimensional branchpoint distance is estimated from secondary structure interactions. However, we believe that folding intronic sequences in isolation is appropriate, partly because of the existence of co-transcriptional splicing, where splicing occurs before the entire pre-mRNA has been synthesized [28-30]. Therefore, the precise part of the pre-mRNA that serves as the splicing substrate is not known. The region upstream of the transcribed intron, which consists of the 5' UTR and the first exon, is also not precisely defined due to the fact that the transcription start sites have not been unambiguously mapped [31]. In addition, 5'UTRs are known to associate with a number of protein factors [32,33] which are likely to have an effect on the structure formation, but these interactions are not currently modelled by computational RNA secondary structure approaches. A preliminary investigation, in which we considered some of the upstream region yielded inconclusive results (data not shown). Thus, we believe that folding only intronic sequences gives us a reasonable approximation of the secondary structure of an intron at the time of the splicing reaction.The approximation of the three dimensional branchpoint distance using pre-mRNA secondary structure is necessary since there are no reasonably reliable algorithms for predicting RNA tertiary structure. However, it is believed that RNA secondary structure plays a crucial role in tertiary structure formation, since most tertiary interactions are thought to arise after the formation of a stable secondary structure, when the molecule is able to bend around the flexible, single-stranded regions [34,35]. Moreover, the tertiary structure interactions that arise in the later stages of folding are usually too weak to disrupt secondary structure that has already formed. Therefore, we believe that the structural branchpoint distance based on the secondary structure interactions provides a reasonable approximation of the true spatial distance.ConclusionOur computational study offers further insights into the role of pre-mRNA secondary structure in gene splicing in yeast. We show that it is necessary to consider near-optimal structure predictions to be able to detect structural differences between intron mutants that have different splicing efficiencies. We also propose a novel method for quantifying a distance between two bases in an RNA secondary structure and apply this to compute structural branchpoint distances in the studied intron mutants. Positive experimental results on three different yeast genes suggest that our model of structural requirements for efficient splicing can be applied universally to all 5'L yeast introns. Additional laboratory experiments are needed to refine the current model by determining the upper bound of the structural branchpoint distance needed for efficient splicing and acceptable thermodynamic stability of the stems adjacent to splicing signals. Considering that several biological studies indicate that shortening of the branchpoint distance, either by formation of secondary structure or by protein interactions, is important for efficient splicing in Drosophila melanogaster and some mammalian species [12,13], it might be possible to extend our model to define structural requirements for efficient splicing in other eukaryotes. Another possible application of our findings is in gene-finding, where structural characteristics of identified long introns can be used to distinguish between real and false positive predictions.MethodsComputational RNA secondary structure predictionIn this work we used four different RNA secondary prediction tools: mfold [14,15], RNAsubopt [20], RNAfold [22] and Alifold [26].Mfold was used for predicting MFE secondary structures for Libri et al.'s [8] mutants and for predicting suboptimal structures within 5% of the MFE during the mutant design process. Mfold uses dynamic programming to identify the MFE secondary structure and a set of suboptimal structures within a user defined percentage from the MFE for a given RNA sequence. We used both, the web (3.2) and command line (3.0) versions of mfold with default parameters.RNAsubopt was used to compute a sample of 1000 suboptimal structures within the 5% from the MFE. Unlike mfold, it computes all suboptimal secondary structures within a user defined energy range or percentage from the MFE for a given RNA sequence. It can also draw a random sample of the computed suboptimal structures using their Boltzmann weights. We used the command line version of RNAsubopt with options \"-ep 5 -p 1000 -noLP\", which specify the percentage from the MFE (5%), random sample size (1000) and disable prediction of helices of length 1.RNAfold was used to compute partition function and base-pair probabilities. RNAfold uses dynamic programming to compute the MFE secondary structure of a given RNA sequence, but when run with option '-p' it also computes base-pair probabilities.Alifold was used to compute consensus secondary structure for RPS17B and RPS6B introns based on the alignment of introns in Sensu stricto species. It uses modified dynamic programming algorithms that add a covariance term to the standard energy model to compute a consensus secondary structure for a set of aligned RNAs.RNAsubopt, RNAfold and Alifold are part of the Vienna RNA secondary structure package [22] (we used version 1.7). All four algorithms use free energy calculation based on Turner's nearest neighbour energy model [15,36-38].Distance calculation in an RNA secondary structureCalculating the spatial distance between two nucleotides in a folded RNA molecule requires knowledge of the tertiary structure of the molecule. Since currently there are no reasonably reliable algorithms for predicting RNA tertiary structure, our distance calculation is based solely on RNA secondary structure. Considering that secondary structure is generally believed to play a crucial role in tertiary structure formation [34,35], this approach should give us a good approximation of the true spatial distance.To calculate the structural branchpoint distance ds, we consider a predicted secondary structure of the intronic pre-mRNA as an undirected graph whose vertices are nucleotide bases and whose edges correspond to the bonds between the nucleotides. These bonds can be either sugar-phosphate bonds between the nucleotides in the RNA chain or the hydrogen bonds between paired bases in a given RNA secondary structure. Figure 10 shows the conversion from an RNA secondary structure to the secondary structure graph representing it. To compute the distance between two vertices in the graph, we employed Dijkstra's shortest-path algorithm [39]. Since Dijkstra's algorithm requires a directed graph, we represent each non-directed edge (u, v) as two directed edges, (u, v) and (v, u). All edges in the RNA secondary structure graph have uniform weight w(u, v) = 1.Figure 10Conversion from the RNA secondary structure to the graph representing it. (a) Graphical representation of the secondary structure of an intron produced by mfold (filled-in circles represent base-pairing interactions, i.e., hydrogen bonds). (b) Graph representing the RNA structure in (a). The bolded path between the source and target vertices is the one found by the algorithm to be the shortest (ds = 11).In our implementation of the algorithm, the inputs to the program are a pseudoknot-free RNA secondary structure in dot-bracket notation (Vienna format) and the locations of two bases for which the distance needs to be calculated. These bases are the first nucleotide of the intron and the bulging A in the branchpoint sequence (UACUAAC). The output of the program is the shortest distance between these two bases, which we consider as the structural branchpoint distance (ds) for the given intron secondary structure. The program is available at .Mutant sequencesWe used two basic strategies for designing intron mutants with desired structural characteristics. To obtain mutants with long structural branchpoint distances we aimed to disrupt a zipper stem that was bringing the donor site and the branchpoint sequence close together in the wildtype intron. Conversely, for the mutants designed to have efficient splicing we aimed to stabilize the zipper stem found in the wildtype intron. With these strategies in mind, we used a combination of a trial-and-error approach and secondary structure designs computed by RNA Designer [40] to obtain mutant sequence with desired structural characteristics.Most of the intron mutants that we designed have segment substitutions around 20–30 nt long. Sequence segments of this size allowed us to rearrange the secondary structure of a mutant in a desired way. The exception is mutant rps6b-S5 which has three short insertions (8 nt in total) in the polypyrimidine tract of RPS6B intron. Mutant rps17b-L3 is a result of two 3-nucleotide-segment substitutions in Libri et al.'s [8] mutant 8mUB1 (the middle sequence of lower case letters represents the original 8mUB1 mutation). Similarly, mutant rps17b-S3 is a result of a 4-nucleotide-segment substitution in the 3mDB1 mutant (the first segment of lower case letters represents the original 3mDB1 mutation). Table 8 gives the location and sequence of mutant substitutions and Figure 11 depicts mutant locations with respect to the secondary structure of the introns we studied.Table 8Specifications for the new intron mutants used in our study.mutantsegment locationoriginal sequencesubstitution/insertionrps17b-L1258–268 (11 nt)UGAAGAGAGGUaugagacaacurps17b-L2138–157 (20 nt)GAUUAGAAAACUCCAUUACUcuuaaguuaguaaauaccucrps17b-L322–47 (26 nt)UGAAGCCGGAUAUGAUGGACUGGGCuuaAGCCGcuacuacuUGGACUGucgrps17b-L4167–189 (23 nt)AGAAGAGCGCUCAAUGAAGUAGUuggcuuggguuaguaggugccucrps17b-S1217–231 (15 nt)AAUUGCUUUCGAAUGuuucauguguucagcrps17b-S2280–286 (7 nt)UAAGUUGuacguacrps17b-S3246–253 (8 nt)AUCCAAUGuagCggcurps17b-S4244–253 (10 nt)UUAUCCAAUGcuucaucaacrps6b-L121–54 (34 nt)CCUUAGAAUUCUAAUGAAUCAGCACGCGCUAACCguauuuugggugugucccuguuauaaauaauaccrps6b-S119–29 (11 nt)AUCCUUAGAAUuuuguuaguaarps6b-S287–113 (27 nt)CACAAAUUAGUGCACUAUAAUAAAAAUuuauaaauagugauaccauuugguaaarps6b-S321–57 (37 nt)CCUUAGAAUUCUAAUGAAUCAGCACGCGCUAACCGGCaaauuccaacguuucccugcaacaugccuuucuuccgrps6b-S438–55 (18 nt)AUCAGCACGCGCUAACCGauucccaacagacuguccrps6b-S5337–345GUAUUAUUUGgUguucAUUAUUacaUape2-L1159–175 (17 nt)UGUUACCCUCAUAUUCUggguacaauuaauagagape2-L2237–252 (16 nt)GCAAUAGCUUAGGUAAccuucguacuuuugggape2-S123–37 (15 nt)CAAAGAAACAAGGAAagggcagaaauagaaape2-S243–57 (15 nt)AUACAUAAUAUAAAUaacugguagguacguape2-S3237–252 (16 nt)GCAAUAGCUUAGGUAAcaaugaaugagaacucape2-S4159–175 (17 nt)UGUUACCCUCAUAUUCUaaauauuaccuaagcuaape2-S5300–322 (23 nt)CUCGUUACCGACCUUUGAGUUCUuuaagcuuuuguguuugagaacaThe upper case letters represent the original sequences and the lower case letters represent substitution or insertion sequences. The first base of an intron is numbered 1.Figure 11Location of mutations with respect to the secondary structure for (a) RPS17B, (b) RPS6B, and (c) APE2 introns. The two lines for each mutant indicate the beginning and end of the sequence segment that was modified.Generation and assaying of intron mutantsUsing the TRP1 gene as a selectable marker, RPS17B, RPS6B and APE2 were tagged at their genomic locus with a -13MYC fragment to generate C-terminal protein fusions in yeast strains derived from a s288c background [41]. Western blotting with a MYC antibody (Covance Research Products) confirmed expression of the correct size protein product in each strain. The intron of the selected gene plus 5' and 3' flanking sequences were deleted through homologous recombination with the URA3 selectable marker in each of these tagged strains. Intron DNA containing sequences homologous to regions 5' and 3' of the URA3 insertion plus the selected intron mutations were created by PCR. Transformation of these fragments into the appropriate intron deletion strain results in recombination, removal of the URA3 gene, and insertion of the mutant intron sequence. The URA3 gene product leads to cell death when placed on 5-fluoroorotic acid (5-FOA) due to the conversion of 5-FOA to a toxic by product [42]. After transformation, cells can be selected on 5-FOA for those strains that have lost URA3 via insertion of the mutant intron, and thus can grow in the presence of 5-FOA. PCR was used to confirm that 5-FOA resistant strains were the result of insertion of the mutated intron in place of URA3. Each intron mutation was subsequently confirmed by sequencing. Strains containing the correct intron mutations were mated with a strain carrying a 13 MYC epitope tagged protein of a different molecular weight as an internal control and assayed for protein expression levels by western blotting. Western blotting was performed using 20–200 ng of whole cell lysate with a MYC antibody (Covance Research Products) and was quantitated after being developed with ECL Plus Western Blotting Detection reagent (Amersham Bioscience) using a Storm Imaging system (Amersham Bioscience). For each mutant assayed, the internal control was used to normalize protein loading, and the experiments were performed a minimum of 2 times on two independently derived mutant isolates.Yeast intron datasetIn order to obtain a high quality yeast intron dataset we consulted three databases: the Ares lab Yeast Intron Database [43], the Yeast Intron DataBase [44], and the Comprehensive Yeast Genome Database [45]. For additional information, we used the Saccharomyces Genome Database (SGD) [46]. We constructed our dataset by including introns that have consistent annotations between at least two of the three databases. We considered only introns from single-intron genes (which represent the majority of intron-containing genes in S. Cerevisiae) that interrupt the gene's coding region (this excluded introns found in the 5' UTR region). The number of introns found to have a consistent annotation between at least two databases was 214 (there are ~240 introns in the yeast genome). Eleven of these were excluded because they were not supported by the latest comparative genomic study [23], which labeled them as possible misannotations. The final dataset contains 203 yeast introns, 155 of which are experimentally verified and 48 are putative introns. There are 98 long (5'L) and 105 short (5'S) introns. We call this dataset the STRuctural INtron (STRIN) dataset. The STRIN dataset is available at .Authors' contributionsSR conceived the study, performed computational experiments and drafted the manuscript. HHH and AKM participated in the design and coordination of the study and together with BFO supervised the research project. BM designed and performed laboratory experiments and helped draft the manuscript. PH helped design laboratory experiments. All authors analyzed the results and reviewed drafts of the manuscript. All authors read and approved the final manuscript.Supplementary MaterialAdditional file 1Minimum free energy structures for Libri et al.'s [8] mutants predicted by mfold.Click here for fileAdditional file 2Distribution histograms of structural branchpoint distances for (a) wt, (b) UB1i, (c) DB1i, (d) UB1iDB1i, (e) mut-5, (f) mut-12, and (g) mut-18 introns.Click here for fileAdditional file 3Structural characteristics of newly designed RPS17B mutants based on mfold predictions: ds – structural branchpoint distances for MFE and all suboptimal predictions within 5% from the MFE; avg – average ds; bp prob – base-pairing probability of interaction between the donor site and the branchpoint sequence based on the partition function.Click here for fileAdditional file 4Distribution histograms of structural branchpoint distances for (a) rps17b-L1, (b) rps17b-L2, (c) rps17b-L3, (d) rps17b-L4, (e) rps17b-S1, (f) rps17b-S2, (g) rps17b-S3, and (h) rps17b-S4 mutants.Click here for fileAdditional file 5Distribution histograms of structural branchpoint distances for (a) RPS6B wildtype intron, (b) rps6b-L1, (c) rps6b-S1, (d) rps6b-S2, (e) rps6b-S3, (f) rps6b-S4, and (g) rps6b-S5 mutants.Click here for fileAdditional file 6Distribution histograms of structural branchpoint distances for (a) APE2 wildtype intron, (b) ape2-L1, (c) ape2-L2, (d) ape2-S1, (e) ape2-S2, (f) ape2-S3, (g) ape2-S4, and (h) ape2-S5 mutants.Click here for file\n\nREFERENCES:\nNo References"
+}