pmid stringlengths 8 8 | pmcid stringlengths 8 11 ⌀ | source stringclasses 2
values | rank int64 1 9.78k | sections unknown | tokens int64 3 46.7k |
|---|---|---|---|---|---|
28929137 | PMC5600532 | pmc | 7,815 | {
"abstract": "A one-pot emulsion process produces noniridescent supraball inks made of core-shell melanin and silica nanoparticles.",
"conclusion": "CONCLUSION Inspired synergistically by nonclose packing of melanosomes in teal feathers and hollow melanosomes in turkey feathers, as well as theoretical FDTD modeling, we have designed CS-SMNPs that can self-assemble into micrometer-sized colorful supraballs through a one-pot, scalable reverse emulsion process. This control of spacing leads to supraballs with tunable colors across the entire visible spectrum. The structure of high-RI cores and low-RI shells increases reflectance to produce brighter colors. The use of melanin is critical to the success of this strategy because it provides the required RI contrast between the cores and the shells and the broad absorption that helps to enhance the color saturation by absorbing incoherent scattering. In addition to all the optical merits of using CS-SMNPs, the reverse emulsion method to fabricate supraballs is simple, fast, and easily scalable. Similar to mixing pigmentary colors, one can match a desired color by simply mixing binary CS-SMNPs. Therefore, this novel two-component strategy, melanin and silica, has the potential to revolutionize the use of structural colors in place of toxic organic- and metal-based pigments.",
"introduction": "INTRODUCTION In the colorful world in which we live, colors are significant not only for aesthetics and pleasure but also for communication, signaling, and security. Colors are produced through either absorption of light by molecules (pigmentary colors) or scattering of light by nanostructures (structural colors) ( 1 ). Structural colors are superior to pigmentary colors in many ways, because of their tunability, resistance to (photo or chemical) bleaching, and reduced dependence on toxic materials. Many recent studies have demonstrated the use of self-assembly to produce photonic crystals that generate colors across the visible spectrum ( 2 ). However, we still face significant challenges. Many traditional structural colors are iridescent and thus are not useful for wide-angle displays. Recent examples of noniridescent structural colors lack sufficient color saturation in the absence of absorbing materials (carbon black, gold nanoparticles, or black polypyrrole) to reduce incoherent scattering ( 3 – 7 ). Core-shell nanoparticles with a shell refractive index (RI) similar to water have been used to tune the spacing between cores to achieve optimal scattering for noniridescent colors, but only in solution ( 8 , 9 ). Although both bottom-up and top-down methods have been widely used ( 10 – 12 ), there is a demand for a scalable process for mass production of structural colors. Nature provides many spectacular examples of structural colors, such as green-winged teal ( Anas crecca ) wing feathers that use hexagonal nonclose-packed melanosomes ( 13 ) and wild turkey ( Meleagris gallopavo ) feathers with hollow, high-RI contrast melanosomes that brighten feather colors ( Fig. 1A ) ( 14 , 15 ). These examples inspire the design of core-shell synthetic melanin nanoparticles (CS-SMNPs) described here, for the production of bright structural colors. Further driven by the demand for scalable production of structural colors, we have developed a facile one-pot reverse emulsion process to assemble CS-SMNPs into bright and noniridescent photonic supraballs. The use of melanin as the core material can increase the brightness and saturation of supraballs because of its unique combination of high RI and broadband absorption of light. In addition, melanin is biocompatible and can dissipate almost 90% of the ultraviolet (UV) radiation into heat within a nanosecond ( 16 , 17 ), making those melanin-based supraballs suitable for cosmetics or UV-resistant inks. Fig. 1 Natural inspirations and the optical model. ( A ) Two biological examples to enhance color brightness: a green-winged teal ( Anas crecca ) wing feather and a cross-sectional transmission electron microscopy (TEM) image of a single barbule (left) and an iridescent wild turkey ( M. gallopavo ) wing feather and a cross-sectional TEM image of a single barbule (right). Scale bars, 500 nm. The photos of teal (credit to F. Pestana) and turkey (credit to T. Llovet) are from flickr.com under license numbers CCBY-SA 2.0 ( https://creativecommons.org/licenses/by-sa/2.0/ ) and CCBY 2.0 ( https://creativecommons.org/licenses/by/2.0/ ). ( B ) Normal reflectance spectra from the (111) plane of FCC lattices made of core-shell nanoparticles and homogeneous nanoparticles with similar sizes and equivalent refractive indices: high-RI core/low-RI shell nanoparticles (core: RI, 1.74; diameter, 200 nm; shell: RI, 1.45; thickness, 50 nm), equivalent homogeneous nanoparticles (RI, 1.54; diameter, 300 nm), and low-RI core/high-RI shell nanoparticles (core: RI, 1.45; diameter, 267 nm; shell: RI, 1.74; thickness, 16.5 nm). ( C ) The reflectance intensity ratio between core-shell and homogeneous structures changes as we vary the ratio of core radius to core-shell nanoparticle total radius.",
"discussion": "RESULTS AND DISCUSSION To design an optimal core-shell morphology for producing colors, we first used the FDTD (finite-difference time-domain) method to calculate the theoretical reflectance spectra (normal incidence) from the (111) plane of the most common photonic crystal with a face-centered cubic (FCC) packing composed of core-shell nanoparticles and homogeneous nanoparticles (see the Supplementary Materials for details). Relative to the lattice of homogeneous nanoparticles, the lattice of core-shell nanoparticles with high-RI cores and low-RI shells shows the maximum intensity at a similar wavelength but has higher reflectance ( Fig. 1B ). In contrast, the reverse core-shell structure consisting of low-RI cores and high-RI shells has a much lower reflectance than the lattice of homogeneous nanoparticles. These findings are consistent with previously reported photonic bandgap calculations ( 18 ). By varying the ratio of core-to-total radius, we find the highest reflectance (~130% relative to homogeneous nanoparticles) for high-RI core/low-RI shell structure when the core radius is ~60% of the radius of the whole core-shell nanoparticle. The lowest reflectance (~52% relative to homogeneous nanoparticles) for core-shell nanoparticles with low-RI cores is obtained when the core radius is ~80% of the whole core-shell nanoparticle ( Fig. 1C ). On the basis of these results, we designed nanoparticles with high-RI cores and low-RI shells to obtain higher reflectance and brighter colors. We chose synthetic melanin as the core material because it has an unusual combination of high RI (~1.74) and broadband absorption in the visible spectral region that reduces incoherent scattering and thereby enhances color purity ( 19 ). We used silica (RI, ~1.45) as the low-RI shell and used a sol-gel reaction to coat it onto synthetic melanin cores, producing CS-SMNPs ( Fig. 2A ). We used synthetic melanin cores with diameters from 120 to 160 nm and tuned the coated shell thickness from 36 to 66 nm by adjusting the reaction time and sol-gel precursor concentration (table S1). In Fig. 2B , the core diameter of 160 ± 7 nm was kept constant, and the shell thickness was changed from 0 to 66 nm (for example, core diameter/shell thickness values of 160/0, 160/36, and 160/66 nm). By serving an analogous role to keratin in teal feathers, the shell helps to control the spacing between melanin nanoparticles. Fig. 2 CS-SMNP synthesis and self-assembly. ( A ) A scheme describing the method of synthesizing silica-coated melanin nanoparticles. ( B ) TEM images of CS-SMNPs: 160/0, 160/36, and 160/66 nm, respectively. The red dashed circles represent the boundary of the core and shell. Scale bars, 100 nm. ( C ) A scheme showing the self-assembly of supraball structures via a reverse emulsion process. ( D ) An image of rainbow-like flowers, painted with supraball inks made of five different sizes of CS-SMNPs: navy blue, 123/36 nm; blue-green, 123/43 nm; olive, 160/36 nm; orange, 160/50 nm; red, 160/66 nm. We used a simple water-in-oil reverse emulsion template method to assemble CS-SMNPs into micrometer-sized supraballs (51 ± 14 μm) ( Fig. 2C ) ( 20 , 21 ). No surfactant molecules were used to stabilize the emulsion, and the transient stable emulsion droplets were formed upon shear mixing. The oil phase 1-octanol absorbed small amounts of water ( 20 ) and helped to reduce the amount of water in the aqueous phase containing CS-SMNPs. This process slowly removed the water and helped packing of CS-SMNPs into well-ordered supraballs. These supraballs produce a full spectrum of colors depending on the sizes of CS-SMNPs ( Fig. 2D ). This one-pot process is carried out at room temperature without additional posttreatment to remove water, and the supraballs can be easily separated by centrifugation. This process has a clear advantage over other emulsion-like processes used to produce colorful supraballs that require microwaves or heat to remove water ( 22 – 24 ). In contrast to microfluidic approaches, the reverse emulsion method is also easily scalable to produce larger quantities of supraball particles ( 5 , 9 , 25 ). We investigated supraballs consisting of four types of nanoparticles. Under the stereomicroscope (mostly collecting scattering light), supraballs made of CS-SMNPs (160/36 and 160/66 nm) show highly visible olive and red colors, whereas supraballs made of 160/0-nm CS-SMNPs appear almost black ( Fig. 3A ). As a control, supraballs made of pure silica nanoparticles (224 ± 16 nm) display whitish cyan color. The reflectance spectra for individual supraballs contain one dominant peak in the visible spectral range located at ~430, ~540, and ~660 nm for supraballs composed of 160/0-, 160/36-, and 160/66-nm CS-SMNPs, respectively ( Fig. 3B ). The small variation in the curves from 12 different sizes of supraballs suggests that supraball size has no obvious influence on color. The reflectance intensity of a single supraball increases with the thickness of the silica shell via reduced absorption by CS-SMNPs (fig. S1). The reflectance spectrum for silica particles has a dominant peak near 465 nm and is superimposed by a high-intensity, broad background signal that leads to a whitish color. This broad background is due to higher incoherent scattering ( 24 ). The increase in light absorption and reduction in incoherent scattering of CS-SMNPs produce more saturated colors that are visible to the naked eye. In addition, these colors are noniridescent, with clear advantages in applications such as wide-angle photonic inks ( Fig. 3C and movie S1). Fig. 3 Characterization of supraballs. ( A ) Optical images of supraballs made of four types of nanoparticles: 224-nm pure silica nanoparticles and 160/0-, 160/36-, and 160/66-nm CS-SMNPs. Scale bars, 0.5 mm. ( B ) Reflectance spectra and optical images for individual supraballs consisting of 224-nm pure silica nanoparticles (cyan curve, cyan supraball), 160/0-nm CS-SMNPs (purple curve, purple supraball), 160/36-nm CS-SMNPs (olive curve, olive supraball), and 160/66-nm CS-SMNPs (red curve, red supraball). The shaded area indicates the SD from 12 samples, plotted using pavo package in R ( 32 ). Each black box in the insets represents the size of the area probed by the optical measurements (4 × 4 μm). ( C ) Angle-resolved spectra for olive inks, as shown in Fig. 2D . The inset scheme shows the setup for angle-resolved backscattering measurements where we fixed α = 15° and varied angle θ between the source and the sample from 40° to 90°. ( D ) FDTD simulations of normal reflectance spectra from supraballs consisting of three different sizes of CS-SMNPs, where absorption of melanin was considered. We used electron microcopy to investigate the mechanistic basis of these colors. Scanning electron microscopy (SEM) results show that supraballs are spherical and composed of close-packed nanoparticles ( Fig. 4A ). High-resolution SEM images and two-dimensional fast Fourier transform power spectra reveal that the nanoparticles are quasi-ordered on the supraball outer surfaces ( Fig. 4B ). The quasi-ordered packing helps to reduce the iridescence observed in well-ordered crystalline supraballs ( 26 ). The spherical geometry of supraballs also leads to noniridescent colors ( 5 ). Cross-sectional TEM images show that supraballs are solid and filled with close-packed nanoparticles ( Fig. 4C and fig. S2). This solid morphology likely prevents the supraballs from collapsing. Fig. 4 Microstructures of supraballs. Each column represents supraballs made of different sizes of CS-SMNPs. ( A ) SEM images of whole supraball morphologies. ( B ) High-resolution SEM images of top surfaces of supraballs. ( C ) Cross-sectional TEM images of the inner structure of supraballs. Scale bars, 2 μm (A), 500 nm (B), and 500 nm (C). We compared our empirical results with theoretical predictions of colors of melanin-based supraballs using FDTD simulations. We modeled a flat FCC photonic crystal consisting of six layers and calculated the normal reflectance from the (111) crystal plane without considering the curvature of the supraball surface. To consider the absorption of melanin, we used the RI and extinction coefficient of SMNPs reported in our previous publication (fig. S3A) ( 19 ). Incorporation of absorption terms in the simulation does not shift the peak position but only reduces the reflectance intensity to a different extent, depending on the volume ratio of melanin cores (fig. S3B). The calculated spectra for the supraballs contain a maximum peak position at ~440, ~550, and ~670 nm for 160/0-, 160/36-, and 160/66-nm CS-SMNPs, respectively ( Fig. 3D ). These predictions are in close agreement with the experimental measurements shown in Fig. 3B . The simulated spectra appear narrower than the experimental measurements because we use a flat perfect photonic crystal in our calculations, which is a simplified model for the structure we have in our experiments. However, the agreement with the maximum peak position indicates that this simple model is able to capture the origin of colors of these supraball structures. Analogous to tuning pigmentary colors by mixing two types of pigments, we used the same reverse emulsion process to assemble CS-SMNPs with binary sizes (same core diameter but different shell thicknesses) into supraballs. Mixing pure SMNPs and CS-SMNPs with different shell thicknesses at a mass ratio of 1:1 resulted in a purple color similar to those produced by pure SMNP supraballs ( Fig. 5 , A and B, and fig. S4A, spectra). Both the SEM images of supraball outer surfaces and cross-sectional TEM images of supraballs demonstrate that only pure SMNPs segregate to the surfaces after mixing with CS-SMNPs ( Fig. 5 , A and B). However, mixing two sizes of CS-SMNPs with 1:1 ratio by mass results in an orange color ( Fig. 5C and fig. S4B), and nanoparticles of both sizes of CS-SMNPs are randomly mixed at the surface and in the bulk (fig. S5). The differences in blending and segregation of nanoparticles can be explained by the higher affinity of melanin than silica to the oil-water interface. Therefore, the addition of different ratios of CS-SMNPs enables us to tune colors without synthesizing new CS-SMNPs of different shell thicknesses ( Fig. 5D and fig. S6, spectra). Fig. 5 Supraballs from binary CS-SMNPs. ( A to C ) Optical images, SEM images of top surface of supraballs, and cross-sectional TEM images for supraballs consisting of 160/0- and 160/36-nm CS-SMNPs (A), 160/0- and 160/66-nm CS-SMNPs (B), and 160/36- and 160/66-nm CS-SMNPs (C). The mixing ratio was 1:1 by mass. Top: Real images of supraballs made of mixed CS-SMNPs and sketches of supraballs, illustrating the organization of CS-SMNPs. Scale bars, 500 nm. ( D ) Optical images of supraballs prepared using different mass ratios of 160/36- and 160/66-nm CS-SMNPs. Each black box represents the size of the area probed by the optical measurements (4 × 4 μm). To understand the color blending effect, we used the inverse of normalized transport mean free path A = ( k 0 l t ) −1 to calculate the scattering intensity of supraballs made of 160/36- and 160/66-nm CS-SMNPs (details are provided in the Supplementary Materials) ( 27 , 28 ). Compared with the model that assumes only independent scattering (simple summation of Mie scattering), the scattering model based on short-range order not only better captures the features of measured spectra but also predicts the color change with the variation in the mixing ratio of binary CS-SMNPs (fig. S7). Although the model considering the short-range order cannot precisely predict the reflectance peak positions, it suggests that the interference effect from the short-range order is critical for the color production in supraballs made of mixed CS-SMNPs."
} | 4,232 |
29535887 | PMC5841263 | pmc | 7,819 | {
"abstract": "Purpose The goal of this study was to develop and validate a standardized in vitro pathogenic biofilm attached onto saliva-coated surfaces. Methods Fusobacterium nucleatum (F. nucleatum) and Porphyromonas gingivalis (P. gingivalis) strains were grown under anaerobic conditions as single species and in dual-species cultures. Initially, the bacterial biomass was evaluated at 24 and 48 hours to determine the optimal timing for the adhesion phase onto saliva-coated polystyrene surfaces. Thereafter, biofilm development was assessed over time by crystal violet staining and scanning electron microscopy. Results The data showed no significant difference in the overall biomass after 48 hours for P. gingivalis in single- and dual-species conditions. After adhesion, P. gingivalis in single- and dual-species biofilms accumulated a substantially higher biomass after 7 days of incubation than after 3 days, but no significant difference was found between 5 and 7 days. Although the biomass of the F. nucleatum biofilm was higher at 3 days, no difference was found at 3, 5, or 7 days of incubation. Conclusions Polystyrene substrates from well plates work as a standard surface and provide reproducible results for in vitro biofilm models. Our biofilm model could serve as a reference point for studies investigating biofilms on different surfaces.",
"introduction": "INTRODUCTION Periodontal and peri-implant diseases are infections associated with complex biofilm structures that induce an inflammatory response, causing the destruction of connective tissue [ 1 2 ], The prevalence of periodontitis in adults is approximately 47% [ 3 ], making it the sixth most prevalent oral disease [ 4 ], while peri-implantitis was found to be present in 28% of subjects examined in a previous study [ 5 ]. Porphyromonas gingivalis is a red complex anaerobic Gram-negative bacterium, strongly associated with the advancement of both types of oral infection [ 6 7 8 ]. The mechanisms involved in bacterial colonization of natural and artificial surfaces, as well as the surrounding periodontal tissues, include direct attachment to saliva proteins and epithelial cell receptors, and/or interactions with early bacterial colonizers [ 9 10 11 12 ]. Fusobacterium nucleatum is also a Gram-negative bacterium, and is regarded as a central organism for dental biofilm maturation due to its wide ability to coaggregate with other microorganisms, such as P. gingivalis [ 13 14 15 16 ]. This pattern of coaggregation, which is known to be mutually beneficial, promotes the expression of a high number of virulence factors by both species [ 17 ]. Virulence factors may contribute to the survival, presence, and pathogenicity of these microorganisms in various oral niches [ 13 18 ]. Once bacteria are attached to a surface, the dynamic interactions between the host and the bacteria evolve into an organized and complex microbial community, protected from mechanical and chemical damage [ 19 ]. The development of promising strategies for fighting oral infections requires in vitro models of mature biofilms, which are useful for purposes such as obtaining a better understanding of the mechanism of action of certain drugs. Such models are essential for evaluating the efficiency of therapies that aim to control and prevent oral diseases caused by pathogenic biofilms. In the scientific literature, studies have reported various in vitro biofilm models used to assess the effects of specific materials, as well as to investigate the efficacy of treatments [ 20 21 22 23 ]. However, there is limited knowledge regarding the time period necessary for establishing a mature biofilm. Although oral biofilms are typically polymicrobial, mixed biofilms constructed with selected microbial species allow controlled in vitro assays, which enable a better understanding of the impact of materials and/or new treatments on pathogenic species [ 24 25 26 ]. Our purpose in this study was to develop a pathogenic dual-species biofilm model with P. gingivalis and F. nucleatum to use in further in vitro research. The contribution of each bacterium to the maturity of the biofilm was investigated through comparisons with the corresponding single-species biofilms. In this model, bacteria were grown on human saliva-coated surfaces to simulate oral conditions and to enhance bacterial attachment [ 27 28 ].",
"discussion": "DISCUSSION The success of microbiological experiments depends primarily on using the appropriate methodology to construct biofilms that respond better to in vitro investigations. Hence, the goal of this study was to present a clear step-by-step protocol for generating a robust in vitro pathogenic biofilm attached onto saliva-coated surfaces. Our data clearly documented the stages of bacterial growth in the planktonic state, and we defined the appropriate time point for the adhesion phase and subsequent steps of biofilm development. The bacteria concentration used for in vitro experiments must be standardized according to the growth curve. The bacterial growth period selected for experimental studies can obscure or interfere with the real outcomes. A growth curve includes 5 critical phases of development: the lag, exponential, stationary, death, and long-term stationary phases [ 34 35 ]. The duration of each phase is affected by various factors, primarily involving the quality of the growth culture medium, which can affect metabolic conditions. For in vitro experiments investigating bacterial susceptibility to antimicrobial agents, for example, the variability of the bacterial growth phase should be further evaluated and standardized for quantitative testing [ 36 ]. In general, the exponential phase is preferred for experimental investigations, since this period is characterized by increased metabolic activity and cell proliferation. In this study, growth was carefully monitored, using absorbance measurements of each bacterium, before designing the biofilm model. Additionally, the concentrations of CFU per milliliter at the mid-log phase were also determined. The data collected consistently showed that F. nucleatum grew earlier than P. gingivalis . After 5 hours of incubation in broth medium, F. nucleatum had already reached the exponential phase, whereas P. gingivalis required 15 hours to do so. The adhesion phases for both species of bacteria in single- and dual-species setups were then investigated by culturing the bacteria at the concentration found in the exponential phase on saliva-coated polystyrene well plates. The amount of bacterial biomass that was deposited onto the surfaces showed no difference between 24 and 48 hours of incubation for either the single-species P. gingivalis or the dual-species biofilms. However, the biomass of F. nucleatum in the single-species setups was significantly higher at the earlier time point. Thus, 48 hours of incubation led to a decreased biomass of attached F. nucleatum , indicating that there was no or slow growth and possibly cell death, as has been previously discussed [ 17 ]. This behavior can be explained by the rapid consumption of nutrients by F. nucleatum, as the peak of growth was found in the first 24 hours. In the oral cavity and in vitro models, bacterial cells irreversibly interact with natural and/or artificial substrates, or with each other, and initiate biofilm formation by extracellular polymeric matrix production. In this study, differences in biofilm development were found when P. gingivalis and F. nucleatum were grown in single-species setups. P. gingivalis exhibited slower growth and the biomass of biofilm showed only an early stage of development after 3 days of incubation. The quantitative data were also supported by microscopic images, which showed spread-out high-density areas of condensed cells that did not cover the entire surface. In contrast, F. nucleatum produced intricate networks after 3 days, which increased in size after 5 days of incubation, demonstrating a mature biofilm at this stage. Additionally, incubation for 5 or 7 days was not associated with any differences in biomass or 3-dimensional architecture. The same pattern was identified in the dual-species biofilm, indicating that F. nucleatum facilitated P. gingivalis growth based on positive interactions [ 12 15 16 ]. The methodologies used in this study successfully allowed a protocol to be developed for generating single- and dual-species pathogenic biofilms. However, since the bacteria and subsequent biofilms were grown on polystyrene surfaces, our findings might not translate into dental material substrates. We must consider that the physicochemical properties of each type of material influence the amount of bacteria that adhere to and form biofilms on it [ 21 37 38 39 40 ]; therefore, different bacterial behavior is expected on different substrates. Conversely, a growth reference for the bacterial species involved in a specific study is needed before performing a reliable in vitro experiment. Indeed, polystyrene substrates from well plates work as a standard surface and provide reproducible results for in vitro biofilm models. Thus, to obtain a better understanding of bacterial behavior and growth, our biofilm model was developed on the bottom of polystyrene plates. The in vitro biofilms described herein could serve as a reference point for studies investigating biofilms on different surfaces."
} | 2,359 |
34614246 | PMC9285811 | pmc | 7,820 | {
"abstract": "Abstract The global decline of marine foundation species (kelp forests, mangroves, salt marshes, and seagrasses) has contributed to the degradation of the coastal zone and threatens the loss of critical ecosystem services and functions. Restoration of marine foundation species has had variable success, especially for seagrasses, where a majority of restoration efforts have failed. While most seagrass restorations track structural attributes over time, rarely do restorations assess the suite of ecological functions that may be affected by restoration. Here we report on the results of two small‐scale experimental seagrass restoration efforts in a central California estuary where we transplanted 117 0.25‐m 2 plots (2,340 shoots) of the seagrass species Zostera marina . We quantified restoration success relative to persistent reference beds, and in comparison to unrestored, unvegetated areas. Within three years, our restored plots expanded ~8,500%, from a total initial area of 29 to 2,513 m 2 . The restored beds rapidly began to resemble the reference beds in (1) seagrass structural attributes (canopy height, shoot density, biomass), (2) ecological functions (macrofaunal species richness and abundance, epifaunal species richness, nursery function), and (3) biogeochemical functions (modulation of water quality). We also developed a multifunctionality index to assess cumulative functional performance, which revealed restored plots are intermediate between reference and unvegetated habitats, illustrating how rapidly multiple functions recovered over a short time period. Our comprehensive study is one of few published studies to quantify how seagrass restoration can enhance both biological and biogeochemical functions. Our study serves as a model for quantifying ecosystem services associated with the restoration of a foundation species and demonstrates the potential for rapid functional recovery that can be achieved through targeted restoration of fast‐growing foundation species under suitable conditions.",
"introduction": "Introduction Restoration of coastal foundation species has become a conservation priority because of their ecological benefits combined with their extensive global declines (Lotze et al. 2006 ). As human populations continue to grow in coastal areas, the impact of human activities on the foundation species that define coastal marine environments has intensified (Barbier et al. 2011 , Kirwan and Megonigal 2013 , Osland et al. 2019 ). Such impacts include effects of agriculture (Wasson et al. 2021 ) and urban development (Coverdale et al. 2013 ), diversion of freshwater inputs (Kennish 2002 ), and overfishing (Altieri et al. 2012 ) leading to trophic downgrading (Estes 2011 , Kéfi 2012 ). Global effects of ocean acidification on coral reefs (Bellwood et al. 2004 ), sea level rise on tidal marshes (Thorne et al. 2018 , 2018 , \n 2018 \n , \n 2018 \n ) and rising sea surface temperature effects on seagrass (Zimmerman and Hill 2015 ) and kelp forests (Muth et al. 2019 , Rogers‐Bennett and Catton 2019 ) are some prime examples of how humans are contributing to the loss of foundation species and the services they provide. To address these various multiple stressors occurring at different spatial and temporal scales, researchers and managers are applying a diversity of restoration approaches. Intervention in the form of regulatory actions that limit harvest (Hughes et al. 2009 b \n ), implement water quality standards (Kennish 2002 ) or establish perimeters of protected habitat (e.g., Marine Protected Areas) has been effective in restoring environmental conditions conducive to the recovery of foundation species (Ling et al. 2009 , Clements and Hay 2018 , Geldmann et al. 2019 ). Transplanting/seeding foundation species (marsh plants, juvenile mangroves, oysters, corals, etc.) from areas where they are thriving to areas where they are scarce is a common restoration approach (Davis and Short 1997 , Jaap 2000 , Gilman and Ellison 2007 , Pritchard et al. 2015 , Wasson et al. 2021 ) used to enhance foundation species coverage and the ecosystem services and functions they provide. Often restoration fails to bring back all the functions and services associated with foundation species, or does so slowly (Zedler and Callaway 1999 , Duarte et al. 2008 , Bayraktarov et al. 2016 ). Therefore, restoration project goals need to be developed with consideration of the life history and demography of foundation species (Montero‐Serra et al. 2018 , Yando et al. 2019 ). For example, in regions where salt marshes (cordgrass) and mangroves co‐occur, cordgrass may be the preferred foundation species to transplant due to its fast growth, expansion, and recruitment, which expedite the restoration of ecosystem functioning (Yando et al. 2019 ). Generally, succession in wetland systems (e.g., salt marsh and seagrasses) is relatively rapid, making them ideal systems for understanding recovery through restoration and the ecological responses that affect a range of functions and services. Similarly, utilizing the rapid succession of fast‐growing kelps (Dayton et al. 1992 , Tegner et al. 1997 ), active restorations in the form of artificial reefs (Reed and Schroeter 2006 ) and juvenile transplants (Carney et al. 2005 , Layton et al. 2020 ) have been key in attempts to reverse widespread deforestation. In contrast, coral reef species are typically long lived and slow growing (Young and Schopmeyer 2012 , Ladd et al. 2018 , Ladd and Burkepile 2019 ); consequently, returning reefs to predisturbance conditions can take 5+ yr (Hein et al. 2021 ), sometimes decades (Jaap 2000 , Victoria‐Salazar et al. 2017 ). Relatedly, certain foundation species, such as oysters, are often recruitment limited (Wasson 2016 ), presenting unique conservation challenges (Wasson 2020 , Ridlon 2021 ). Thus, expectations for the rate of recovery of ecosystem services should be tailored to the life history and demography of different foundation species and habitat types. Seagrasses are a group of marine foundation species that are in accelerated global decline (Orth 2006 , Waycott 2009 , Short 2011 ). As marine flowering plants, seagrasses are primarily limited by light availability, and most temperate seagrass species are restricted to the low intertidal or shallow subtidal zone (Zimmerman et al. 1997 ). Light attenuation due to poor water quality triggered by increased habitat degradation, sediment loading, eutrophication, contaminants, and other pollutants are frequently cited as the leading cause of seagrass loss (McGlathery and Sundäck 2007 , van der Heide et al. 2007 ). To combat the widespread loss of seagrass habitat, restoration efforts are on the rise (Cunha 2012 , van Katwijk 2016 ). Seagrass ecosystems are highly productive and support a suite of ecosystem functions linked to valued provisioning and regulating services (Duarte 2002 , Duffy 2006 ). For example, seagrass belowground biomass stabilizes sediment while aboveground biomass attenuates wave action (Short et al. 2011 , Ondiviela et al. 2014 ); together these two functions provide the services of mitigating erosive forces and acting as a storm buffer. Additionally, organic particulate matter that is trapped within seagrass beds is stored in its oxygen‐depleted sediments where decomposition is relatively slow, providing the service of carbon storage. Relatedly, seagrasses both transport O 2 to the rhizosphere, building a barrier against phytotoxins (Frederiksen and Glud 2006 ), and absorb large quantities of CO 2 (Orseka et al. 2020 ), the latter of which has the potential to mitigate impacts of ocean acidification (Bergstrom et al. 2019 , Ricart 2021 ). Seagrasses also provide structure that serves as nursery habitat for species of commercial importance, for example, along the U.S. West Coast, Dungeness crab ( Metacarcinus magister ) utilizes estuarine habitats, including seagrasses, during its early life stages and this particular fishery has an average annual value of over US$100 million (Hughes 2014 , Grimes et al. 2020 ). Yet it is largely unknown whether restoration of cover by these foundation species is correlated to a related restoration of associated functions and services or how long it takes to achieve such functional recovery (but see Tay Evans and Short 2005 and Orth et al. 2020 ). Given the tremendous value of seagrass ecosystems, investigations that assess both structural and functional attributes are required to evaluate whether restoration efforts can be defined as “successful.” One such study in New Hampshire’s Great Bay Estuary showed functional equivalency within 3 yr between restored and reference beds with respect to primary production, structure, and habitat use (Tay Evans and Short 2005 ). More recently Orth et al. ( 2020 ) used large‐scale seeding (70+ million seeds) to restore seagrass in Virginia’s recruitment‐limited coastal lagoons, where seagrass had been absent for over 70 yr. Over the past two decades, restored habitats in the coastal lagoons continued to expand and resulted in the recovery of multiple ecosystem services and even a related restoration of commercially harvested bay scallops (Orth et al. 2020 ). In Australia’s restored and reference seagrass ( Amphibolis antarctica ) beds, infauna abundance, belowground biomass, and sediment grain size converged within approximately 2–6 yr (Tanner et al. 2021 ) and in Korea ( Zostera marina ), stable isotope measures showed that food web structure between restored and reference beds were indistinguishable from one another 2 yr post‐transplanting (Park et al. 2021 ). Similar studies that assess both biological and biogeochemical functions are needed to gain a full understanding of ecosystem services gained from seagrass restoration. We conducted two experimental seagrass restoration efforts and monitored success in terms of multiple structural indicators and ecosystem functions and services, in a nutrient‐loaded and highly eutrophic (Wasson 2017 ) estuary on the west coast of the United States from 2015 to 2018. The goals of our study were to (1) track the temporal trajectory of restored seagrass survival, expansion, and health as a critical prerequisite to recovery of ecosystem functions and (2) quantify a suite of key ecosystem functions and determine if the restored seagrass ecosystem functionality rapidly reaches the levels of naturally existing beds. First, our investigation compared restored seagrass plots vs. naturally existing reference beds for seagrass areal expansion rates and indicators of health (productivity, canopy height, above‐ and belowground biomass, shoot densities, and epiphytic algal loads). Second, we quantified ecosystem functions including biodiversity, nursery habitat, modulation of water chemistry (pH, dissolved oxygen [DO]. and water temperature), and organic carbon stocks and compared these functions individually, and collectively using a multifunctionality index (Byrnes et al. 2014 ) across restored, reference, and adjacent unvegetated soft‐bottom habitats. Our investigation thus serves as a model for investigating the restoration trajectory of foundation species, and the ecosystem functions they support.",
"discussion": "Discussion Successful seagrass restoration Ecological restoration has been defined as “the process of assisting the recovery of an ecosystem that has been degraded, damaged, or destroyed” (SER 2004 ). A strategic approach to recovery is the ecological restoration of foundation species, which play a critical role in structuring communities and regulating key ecosystem processes (Ellison 2005 , 2019 ). As a prerequisite to restoring valued functions and services, restored foundation species must survive and grow. Seagrass restoration has proven challenging in terms of survival and growth, with a global success rate of ~37%, with success measured as an index that accounts for initial restoration survival and the longer‐term, ≥23 month, trajectory (i.e., absent, decreasing, no change, or increasing) of the restoration (van Katwijk et al. 2016 ). Our small‐scale restoration project had a success rate, calculated as the percentage of plots remaining 30–40 months post‐transplanting, of 61%, far exceeding the global average for both small (22%) and large (42%) scale seagrass restorations (van Katwijk et al. 2016 ) and we observed rapid expansion of our restored plots in total area (~8,500%). Overall transplant survivorship and spread was remarkably high in our study, indicating that our restoration efforts were successful and on par with the well‐documented outcomes of larger scale restorations (Tay Evans and Short 2005 , Leschen and Ford 2010 , McGlathery et al. 2012 ). We attribute our high success to a selection of factors linked to site and design, which compliments the findings of others that emphasize the need to address adverse conditions first or fully assess site suitability prior to transplanting or seeding to improve the likelihood of restoration success (Fonseca and Kenworthy 1982 , van Katwijk and Hermus 2000 , Bell et al. 2008 , Thom et al. 2012 , 2018 a \n , \n b \n ). Our success in Elkhorn Slough is likely due to many factors related to the site itself and our restoration design. Despite high nitrate concentrations and the proliferation of macroalgal blooms (Wasson et al. 2017 ), the natural recovery of seagrass in Elkhorn Slough over the past 30 years (Appendix S1 : Fig. S1) is evidence of an overall improvement of environmental conditions conducive to seagrass growth and expansion. Elkhorn Slough is also the only estuary along the California coast with a large (100+) population of resident sea otters, Enhydra lutris (Tinker and Hatfield 2017 ). Sea otters have been shown to have strong top‐down effects on the health of seagrass beds in Elkhorn Slough (Hughes et al. 2013 ). The return of this keystone predator to the system has been linked to natural seagrass recovery and may have contributed indirectly to our positive restoration outcomes by improving seagrass health and promoting natural expansion throughout the system. Additionally, we transplanted within a narrow depth range of 0 to −2 m MLLW. Within this tidal range, restoration plots were exposed infrequently at low tides and had sufficient light availability at high tide, nearly eliminating plot mortality linked to desiccation stress or high light attenuation. Lastly, our transplanting method of anchoring ramets (shoots with intact rhizomes) to the substrate has been shown to have higher rates of success than other transplanting methods (Park and Lee 2007 , Bell et al. 2008 , Eriander et al. 2016 , van Katwijk et al. 2016 ). All of these factors associated with site suitability and restoration design likely contributed greatly to our restoration success. By contrast, attempts to restore seagrass in other degraded systems in the region have had minimal success. For example, in Morro Bay, California. 95% of eelgrass has been lost and efforts to restore have had variable success; this may be linked to the degraded state of natural meadows (Harenčár et al. 2018 ), massive system‐wide erosion and sediment resuspension shifting tidal elevations throughout the estuary (Walter et al. 2020 ), or restoration design. Restoring multiple functions of foundation species Foundation species provide ecosystem structure (Dayton 1972 ) and structure begets ecosystem function (Dobson and Bradshaw 1997 , Bruno and Stachowicz 2003 ). For example, successful large‐scale seeding efforts in Chesapeake Bay led to the restoration of seagrass habitat and associated functions and services (i.e., decreasing turbidity levels, increasing carbon stocks, habitat provisioning; Orth et al. 2020 ). Therefore, ecological restoration of declining foundation species, such as seagrasses (Short et al. 2011 ), not only supports recovery of the vegetation itself, but potentially has the added benefit of enhancing biodiversity and ecosystem functions (Tay Evans and Short 2005 , Benayas et al. 2009 , Angelini et al. 2015 ). The primary focus of seagrass restoration monitoring has been the foundation species itself (structural attributes) and to determine how site characteristics (van Katwijk et al. 2009 , Thom et al. 2018 , 2018 , \n 2018 \n , \n 2018 \n ) and methodology (transplantation or seeding techniques; Park and Lee 2007 , Bell et al. 2008 , Eriander et al. 2016 ) affect survival. In select restoration studies, structural attributes and a few additional associated ecosystem functions (canopy friction and sediment movement [Fonseca and Fisher 1986 ], faunal communities [Fonseca et al. 1996a , 1996b , \n 1996a \n , \n 1996b \n , Leschen et al. 2010 ], carbon and nitrogen sequestration [McGlathery et al. 2012 , Greiner et al. 2013 , Reynolds et al. 2016 ]) have been assessed. Yet, few studies have simultaneously tracked multiple structural and functional attributes of biological and biogeochemical importance. Our study is the first comprehensive investigation of eelgrass restoration on the eastern Pacific Coast to track structural attributes and such a large suite of biological and biogeochemical ecosystem functions permitting a comprehensive assessment of functionality. This approach of fully characterizing the success of our ecological restoration was powerful and if sufficient funding is available for monitoring, could be broadly applied to restorations of other foundation species. To infer functional recovery in other systems, we recommend similar focal studies be conducted to validate that certain functions are enhanced with restorations elsewhere. Here, we review the highlights of the multiple functions we evaluated. Generally, biological functions recovered rapidly. This is likely due to our restoration method, which immediately added habitat structure to support species use, and the rapid growth and expansion of restored plots, which quickly resembled reference plots (i.e., shoot densities, canopy height, biomass). For example, we observed rapid colonization critical mesograzers in 2016 (Healey and Hovel 2004 , Hughes et al. 2013 , Lefcheck and Marion 2017 ) and greater species richness in restored habitats due in large part to the greater geographic area sampled in restored (strata A–E) vs. reference (strata B–C) habitats. Interestingly, by 2018, mesograzers P. resecata and P. taylori were more abundant in restored plots and P. resecata biomass was greater in restored vs. reference plots. This could be due to relatively fewer known predators (i.e., Cancridae crabs) of mesograzers found in restored plots compared to reference plots. Additionally, within 2–3 yr, restored plot nursery function, macrofaunal community composition, fish and invertebrate species density, and fish abundance, recovered to levels at or nearing those observed in reference plots. Our expectation that nursery function would be greatest for reference plots was met and this was largely driven by many more predatory crabs in reference vs. restored plots (Table 1 ). Habitat differences in nursery function likely emerged due to a preference by juveniles for structured habitat (i.e., red rock crab) and/or refugia (i.e., rockfish; Dungeness crab). Red rock crab CPUE increased with habitat structure with the highest CPUE values observed in 2015 restoration and reference plots and the lowest values in unvegetated and 2016 restoration plots: we expect CPUE in 2016 plots to increase as the plots continue to expand. Similarly, juvenile rockfish are known to prefer structured habitat (Love and Carr 1991 , Olson et al. 2019 ) and our data support these findings as rockfish were found more commonly in vegetated (restored and reference) vs. unvegetated habitat. Alternatively, Dungeness crab appear to prefer unvegetated habitat or smaller patches of seagrass (2016 restoration plots) over larger patches (2015 restoration plots) or continuous beds (reference plots). Preference for unstructured habitat by Dungeness crab is well documented (Holsman and McDonald 2006 , Holsman et al. 2010 ) and in Elkhorn Slough, sea otters, a common predator of Dungeness in the system, have depressed crab size and abundance in unvegetated habitats (Grimes et al. 2020 ). Therefore, we suspect that Dungeness crabs utilize the smaller 2016 restoration plots as temporary refuge from predators. As restored plots matured and expanded, macrofaunal diversity and community composition progressed towards reference plot levels. Restored and reference plots supported higher fish species densities than unvegetated plots, complimenting previous work (Ruesink et al. 2019 ), but fish CPUE was lower in restored plots and may be related to the size or “patchiness” of restored plots compared to the continuous reference beds. Invertebrate species richness increased with habitat structure and maturity (with unvegetated as structureless and restored as structured but less mature than reference), whereas invertebrate CPUE did not vary by habitat, suggestive that species evenness is lowest in highly structured habitats (i.e., reference). Generally, macrofaunal diversity in restored plots is moving towards levels observed in reference plots, this is further supported by the cluster diagram (Fig. 3A ), showing that similarity in macrofaunal community composition of restored plots shifted from unvegetated to reference plots over time. In summary, certain biological functions in restored plots are currently performing at or near levels observed in reference plots while others are higher than unvegetated plots and lower than reference plots. We expect such functions to edge towards reference plots as restored plots continue to expand. While biological functions were fast to emerge, biogeochemical functions either did not vary across habitats (i.e., organic carbon stocks, OC) or were more subtle and nuanced (i.e., DO, pH, water temperature). Because water chemistry is highly context dependent, an assortment of factors (incoming vs. outgoing tides, distance from mouth, amount of water in the basin, tidal height, nutrients) could mask the signal of eelgrass modulating water chemistry. Due to the wealth of data available in this study, we were able to use CDF curves to better visualize differences in pH, DO, and water temperature across restored, reference, and unvegetated habitats. By plotting data from all eight deployments, we captured a range of hydrodynamic conditions and still managed to detect differences between unvegetated and vegetated habitats. This was further supported by the K‐S test, which showed restored and reference plots were statistically more similar to one another than either were to unvegetated plots. Generally, water temperature was lower and pH and DO higher in restored and reference plots compared to unvegetated plots. Last, we expected restored and reference plots to have greater OC than unvegetated plots, and instead observed no detectable differences among habitats but a near doubling of OC with increased distance from mouth. This is likely due to a change in grain size moving upstream, from predominantly sandy to silty (Ward et al. 2021 ), and shows that in Elkhorn Slough, the ability of eelgrass to shift sediment properties is context dependent (Appendix S1 : Fig. S9). Our use of sophisticated visualizations and analyses allowed us to better evaluate habitat differences within (e.g., CDF curves, species accumulation curves) and across functions (multifunctionality index). For example, by plotting both species density and species accumulation curves, we were able to assess the expected number of species per plot type and the number of species supported per habitat, respectively. Discrepancies between species density and species accumulation curves can be due to differences in overall abundances or evenness and are informative in assessing biodiversity. As stated above, CDF curves allowed us to compare the distribution of water quality parameters between habitats in a holistic way that is biologically relevant; for example, mean pH across habitats may not be different, but if the frequency of observations for pH in vegetated habitats is higher than in unvegetated habitats, these detailed patterns may have ecological significance. Finally, the multifunctionality index was useful in visualizing our overall finding that the cumulative functional performance of restored habitats was intermediate: higher than unvegetated and lower than reference. The index also allowed us to determine, in a standardized way, the relative contributions of each function in driving differences between habitats (mainly species diversity, pH, and DO). The rapidity with which functioning was enhanced in Elkhorn Slough illustrates the potential for successful ecological restoration of a foundation species. Fast‐growing foundation species such as seagrasses are able to restore ecosystem function faster than foundation species that take years to reach maturity, or than species for which old tissue plays a large role in engineering effects (Montero‐Serra et al. 2018 ). For example, in semideciduous riparian tropical forests, restored habitats can take up to 70 yr to reach old growth forest levels of species richness (Suganuma and Durigan 2015 ). Similarly, in coral reef systems, coral transplants assist in the recovery of rugose structures and yet functional recovery is slow to follow due to slow growth of such reef‐building coral species (Ladd et al. 2019 ). In addition to contrasts among species with different life histories, there are contrasts among functions; some, such as providing structured habitat for animals, may be achieved more rapidly than others, such as carbon storage. As more restoration projects include monitoring of multiple ecosystem functions as we have done here, conservation planners can form realistic expectations of the rate of recovery of different important ecosystem services across contrasting foundation species."
} | 6,526 |
40185870 | PMC11971309 | pmc | 7,821 | {
"abstract": "Soils harbour an incredible diversity of microorganisms that play crucial roles in ecosystem functioning. However, this biodiversity remains largely overlooked, with a poor understanding of how patterns form across landscapes. An eDNA metabarcoding approach was used to identify potential overarching patterns in fungal and bacterial communities from ultramafic ecosystems in New Caledonia, a renowned biodiversity hotspot. Our comprehensive analysis revealed several key findings, notably an important microbial diversity in the extreme environments of iron crust soils. Clear tendencies in phyla composition were also observed, with the fungal groups Ascomycota and Mucoromycota acting as potential indicators of land degradation (only in lateritic soils for Mucoromycota). For bacteria, Chloroflexi was characteristic of open vegetation, while Proteobacteria and Cyanobacteria were observed in higher relative abundances in the closed vegetation. The ectomycorrhizal fungal functional group was also found to be rich and unique, with a hypothetical endemism rate of 87%, and over-represented by the Cortinarius genus in rainforests and maquis (shrublands) dominated by ectomycorrhizal plants. Finally, each ultramafic Massif demonstrated a unique microbial community. Thus, our findings provide valuable insights into microbial ecology and emphasize the need for tailored conservation strategies for this biodiversity hotspot. Supplementary Information The online version contains supplementary material available at 10.1038/s41598-025-94915-0.",
"conclusion": "Conclusion Our eDNA metabarcoding comprehensive analysis allowed us to depict several global patterns in microbial communities from ultramafic soils in New Caledonia: \n Iron crust soils harbour a significantly higher fungal diversity than lateritic soils. The composition of microbial phyla shifts along ecological successions. Fungal Ascomycota and Mucoromycota phyla, which were prevalent in the first stages of the succession, may be markers of ultramafic land degradation. Among bacteria, the high relative abundance of Chloroflexi seems to characterize vegetation with limited plant cover, and conversely, both Proteobacteria and Cyanobacteria characterize closed maquis and rainforests. The ectomycorrhizal fungal functional group exhibits a high and unique species richness, with a hypothetical endemism rate of 87%. In addition, the Cortinarius genus was prevalent in ultramafic rainforests and maquis (shrublands) dominated by ectomycorrhizal plants. Finally, each ultramafic Massif seems to display its own soil microbial community. \n All these results highlight the relevance of further identifying the evolutionary and ecological mechanisms that have shaped the microbial diversity of ultramafic ecosystems in New Caledonia. Future investigations are also needed to better understand how the dominant groups are involved in the functioning of ultramafic systems, notably in nutrient cycling. In addition, our observations underscore the necessity of describing microbial species and assessing their threat status to establish species-conservation plans. They also support the emerging point of view of considering soil biota for identifying hotspots of diversity and defining conservation priority areas for a more holistic perception of nature conservation. This analysis represents a first step. In the coming years, expanded high-resolution investigations into soil biota, covering a wider range of vegetation and soil types (including volcano-sedimentary and calcareous soils) will provide a deeper insight into how environmental factors, including soil physicochemical parameters, climate and land use, influence soil communities. This would help to define conservation strategies and policies for this biodiversity hotspot. This will notably be achieved through the sensibilization of local populations (of all origins) to soil life and its significance, alongside the consideration of their perception of soil and nature conservation, as well as their involvement in defining priority areas.",
"introduction": "Introduction Given the high diversity of soil microorganisms 1 , 2 and their crucial roles in ecosystem functioning (e.g., nutrient cycling, soil stability and plant community dynamics) 3 , 4 , investigating soil microbial communities has become increasingly important, particularly in light of the global change. However, the belowground component of terrestrial ecosystems remains largely overlooked, especially in biodiversity hotspots. New Caledonia, a subtropical archipelago located in the southwest Pacific (Fig. 1 ), is renowned for its remarkable biological diversity 5 – 7 and recognized as one of the world’s highest priority areas for conservation 8 and restoration 9 . The long-term isolation, along with the large variety of soils that have developed over geological time, are considered as major factors that have contributed to the high richness and endemism rates encountered in this biodiversity hotspot. Soils originated from ultramafic rocks, which cover about one-third ofthe main island (Grande Terre) (ca. 5500 km²), harbour a variety of soil and vegetation types. A previous study we conducted on ectomycorrhizal fungi (symbiotic organisms associated with the roots of plants) in the South of New Caledonia, via a barcoding approach (Sanger sequencing), suggested a high diversity and endemism rate, along with the dominance of the Cortinarius genus in monodominant ultramafic rainforests (i.e., where more than 50% of the trees in the canopy layer belong to a single species) 5 . In addition to this work, our team carried out environmental DNA (eDNA) metabarcoding investigations of soil fungal and bacterial communities using a high-throughput sequencing technology (Illumina MiSeq) 4 , 10 – 12 . Combination of eDNA and next-generation sequencing is a fast and efficient approach that has revolutionised the study of biodiversity across various ecosystems 13 , particularly in soil microorganisms, which are intrinsically cryptic and largely uncultivable. Our findings revealed changes in microbial phyla and fungal functional groups’ composition along vegetation successions. Subsequently, we proposed biological indicators of succession dynamics and land degradation, the latter referring to the negative trends in land condition resulting from direct or indirect human-induced processes 14 . We also observed a structure of microbial communities according to the vegetation type 5 , 10 , 11 . Additionally, a geographical effect was strongly suspected, suggesting that each ultramafic Massif could host a unique microbial community 10 . Despite the advances made in recent years, these studies on microbial communities in ultramafic soils have been conducted on local scales, with no comparison between the predominant ultramafic soil types, particularly the two most widespread ones: iron crust and lateritic soils. Iron crust soils are composed of ironstones and gravels, and are empirically considered as harsh environments 15 . To date, no large-scale approach has been undertaken to ascertain whether local observations can be generalized, or not. This gap hinders our comprehension of the ecology of soil microorganisms in this well-renowned biodiversity hotspot. In this study, we thus conducted the first large-scale analysis of fungal and bacterial communities in ultramafic soils of New Caledonia. Both published 4 , 10 – 12 and unpublished soil eDNA metabarcoding data (generated by Illumina MiSeq sequencing) from our own work were combined here. A range of ultramafic sites from the northern to the southern regions of Grande Terre, harbouring diverse vegetation types developing on iron crust and lateritic soils, as well as a non-ultramafic site on the island of Maré (used as an outgroup) (Fig. 1 ; Table 1 , Fig. S1 ), were employed to address the following questions: (1) Is there differences in microbial diversity between iron crust and lateritic soils? (2) Are phyla and functional groups, previously identified at a local scale as markers of ecological succession and land degradation, relevant to ultramafic systems on a broader scale? (3) Is the high diversity and endemism rate in ectomycorrhizal fungi, along with the dominance of Cortinarius in monodominant rainforests previously observed, generalizable at a larger scale and on other vegetation types? And finally, (4) is there a geographical structure of soil microbial communities, with each investigated ultramafic massif exhibiting its own fungal and bacterial community?. This pioneering work provides a unique insight into the microbial dynamics of ultramafic soils and paves the way for adapted conservation strategies to this biodiversity hotspot. \n Fig. 1 ( A ) Location of New Caledonia in the southwest Pacific and ( B ) locations of the five studied ultramafic sites on the main island (Grande Terre). The non-ultramafic site (gibbsic Ferralsol) at Maré Island, used as an outgroup, is also presented. Ultramafic substrates are shown in grey. Numbers in superscript indicate the previously published research articles from which data are sourced; 1: Gourmelon et al. (2016); 2: Demenois et al. (2020); 3: Fernandez Nuñez et al. (2021); 4: Stenger et al. (2025); 5: this study). \n \n Table 1 Vegetation formations and soil types at the different study sites. Sites Vegetation formations Soil types Acronym in this work Nb. of plots and soil samples Nb. of plots retained for analyses Illumina MiSeq run ITS2 (fungi) V4 (bacteria) Goro plateau (GO) Open low maquis Iron crust soil GO.OLM 6 6 6 C Closed low maquis GO.CLM 6 6 6 C Gymnostoma deplancheanum -dominated maquis GO.TCM.Gd 6 6 6 C Preforest maquis GO.TCM.Fs 6 6 6 C Revegetated sites Iron crust and lateritic soils GO.RS 7 7 7 C Rivière Blanche (RB) Sedge-dominated maquis Lateritic soil RB.MqCy 4 0 4 A Tristaniopsis -dominated maquis* RB.MqTr 4 4 4 A Nothofagus aequilateralis -dominated rainforest RB.NaF 4 4 4 A Mixed rainforest RB.MF 4 4 4 A Bois du sud (BS) Arillastrum gummiferum -dominated rainforest Lateritic soil BS.AgF 4 4 4 B Kopéto (KO) Sedge-dominated maquis Lateritic soil KO.MqCy 4 3 4 A Tristaniopsis -dominated maquis** KO.MqTr 4 4 4 A Nothofagus aequilateralis -dominated rainforest KO.NaF 4 4 4 A Mixed rainforest KO.MF 4 2 4 A Tiébaghi (TI) Bushy maquis Iron crust soils TI.BuMq 3 3 3 B Shrubby maquis TI.ShMq 3 3 3 B Tall closed maquis TI.TCM 3 3 3 B Maré (MA) (out group) Forest Gibbsic and humic soils MA.F 5 4 4 C Short fallow MA.SF 5 5 5 C Long fallow MA.LF 5 3 4 C 91 81 89 3 Acronyms, numbers of 20 × 20 m plots retained in the analyses for the ITS2 (fungi) and V4 (bacteria) and the corresponding number of samples are presented. The Illumina MiSeq runs are detailed with, run A, samples from Rivière Blanche and Kopéto; run B, samples from Bois du Sud and Tiébaghi; and run C, samples from Goro plateau and Maré island. *: maquis dominated by Tristaniopsis glauca at Rivière Blanche site. **: maquis dominated by T. calobuxus and/or T. guillainii at Kopéto site.",
"discussion": "Discussion A higher microbial diversity on iron crust soils Surprisingly, fungal species richness was higher in iron crust soils (Goro and Tiébaghi sites). These soils are composed of ironstones and gravels and are, subsequently, empirically thought to harbour more stressful conditions than lateritic soils 15 . However, nickel content is known to gradually decrease from the bottom to the top along the weathering profiles of soils from ultramafic rocks 16 . Considering their lower nickel content and, subsequently, the potential lower nickel bioavailability, iron crust soils may thus represent less restrictive environments regarding this parameter than previously thought. One hypothesis would thus be that limitation in nickel constraints may promote fungal diversity. However, to the best of our knowledge, no studies have been performed concerning the constraints that iron crust soils have on living organisms, nor has a comparison been made with other soils originating from ultramafic rocks (e.g., in terms of nickel and the bioavailability of other heavy metals, water drainage and temperature). Another hypothesis would be an adaptation to iron crust soils, leading to fungal speciation and diversification. Further, we cannot also exclude a plant composition effect, since maquis on iron crust soils are distinct vegetation types compared to others 17 . For bacteria, greater species richness was only observed in iron crust soils at the Goro site. A previous comparison 18 of the molecular diversity of the symbiotic nitrogen-fixing bacteria genus Frankia from the root system of two Gymnostoma species (Casuarinaceae), endemic to New Caledonia, revealed that G. deplancheanum hosted a greater diversity of Frankia than G. chamaecyparis . G. deplancheanum occurs on iron crust soils, whereas G. chamaecyparis is found on hypermagnesian soils that develop at the base of ultramafic massifs. A greater selective effect of hypermagnesian soils on bacteria was given as a probable explanation 18 . This is consistent with our hypothesis of iron crust soils being less constraining in nickel content, but a vegetation composition effect must also be considered. Indeed, G. deplancheanum is a major component of the maquis in the South of the main island and could thus drive bacterial community diversity in this area. The highest values of fungal richness being observed in the G. deplancheanum -dominated vegetation suggest that the presence of this Casuarinaceae species may also locally contribute to increased soil fungal diversity, through direct (e.g., beneficial mutualistic interactions) or indirect (e.g., litter composition) effects. Soil microbial phyla composition as markers of ecological succession, land degradation and geographic origin Beyond variations in soil microbial diversity, differences in relative abundances of phyla among vegetation types were observed. Ascomycota and Basidiomycota, the two most abundant fungal phyla in the large-scale dataset, varied in their proportions, with Ascomycota being relatively well-represented in the first stages of the plant successions, especially at the Goro site (open low maquis), as also found by Fernandez Nuñez et al. 11 . At the Rivière Blanche and Kopéto sites, Ascomycota was also well present in the open vegetation types (i.e., sedge and Tristaniopsis formations); nevertheless, the pattern was not as strong as that detected in Gourmelon et al. 10 . Interestingly, we found that at these two sites, Mucoromycota were more prevalent in the first stages of the succession and decreased along the succession. Among Mucoromycota taxa, ASVs assigned to Bifiguratus adelaidae and GS23 clade were detected as the most common in terms of relative abundance and richness. B. adelaidae has been recovered mostly from soil and to a lesser extent as a plant endophyte 19 . Despite the broad geographic distribution and abundant site occurrence of this novel taxon, its functional roles are still unknown. The GS23 clade, to which most of Mucoromycota reads and ASVs were assigned, forms a new monophyletic lineage from tropical and subtropical acidic forest soils (2.5 to 4.0) 20 . Recently, fungal isolation and identification from plant roots in acidic and oligotrophic soil of northeast America led to the description of two new species, belonging to a new genus and a new family, Pygmaeomycetaceae, proposed to correspond to the Clade GS23 21 . The functions of Pygmaeomycetaceae members are not yet fully understood, however, Walsh et al. 21 suggest that these fungi may be capable of degrading diverse substrates, allowing mobilization of nitrogen and subsequently contributing to the success of their associated host plants in acidic and nutrient-poor environments. Such a hypothesis suggests that ASVs belonging to the GS23 Clade are root symbionts that strongly influence ecosystem functioning at the two distant sites that are Rivière Blanche and Kopéto, especially in vegetation with poor plant coverage and soil nutrition. This reinforces the necessity of determining to which functional groups (in terms of guilds and trophic modes) these Mucoromycota members belong to. Their unknown functionality may explain why no clear trend in fungal functional groups was detected across the successions, or conversely, along the land degradation gradients, at least at Rivière Blanche and Kopéto sites. In addition to this, akin to Ascomycota, high relative abundances of Mucoromycota may be an indicator of land degradation 10 as well as a sign of early stages of an ecological succession 11 . However, we do not know whether the prevalence of this GS23 Clade in ultramafic substrates is mainly restricted to lateritic soils in general or linked to particular vegetation types. Similarly, the relative abundances of soil bacteria varied between vegetation types. These discrepancies were overall due to the higher relative abundances of Chloroflexi in the open vegetation and Proteobacteria in the closed vegetation. Despite their low proportions, Cyanobacteria were also mostly encountered in closed maquis and rainforests. These results allow us to generalize the findings made at a local scale by Fernandez Nuñez et al. 11 , regardless of the type of ultramafic soil (i.e., iron crust soils versus lateritic soils). Indeed, the higher proportion of Chloroflexi that characterized the open low maquis at Goro 11 can be extended to the sedge-dominated maquis, two ecosystems characterized by a more or less herbaceous layer with sparse shrubs, which results in lower plant coverage compared to other vegetation types (approximately 10% for Goro’s open low maquis and 30–80% for the sedge maquis). The large relative occurrence of Proteobacteria, as well as the increased presence of Cyanobacteria, previously highlighted from Goro’s closed vegetations 11 , were also found in most of the closed maquis and rainforests. The prevalence of Chloroflexi and Proteobacteria, in three out of the four successions investigated in our analysis (i.e., at Goro, Rivière Blanche and Kopéto sites, but not at Tiébaghi), may be explained by the availability of nutrients in soil along ecological successions: Chloroflexi suspected as oligotrophs 11 , 22 , survive in nutrient-poor environments, while Proteobacteria, usually copiotrophs 23 , prefer nutrient richer substrates. Another striking finding was the high relative abundance of bacteria belonging to the Actinobacteria phyla at Tiébaghi site independent of the vegetation type. Acidothermus , Conexibacter and Mycobacterium were by far the most dominant taxa in terms of relative abundance and ASVs richness. The Acidothermus genus contains to date only one described species, A. cellulolyticus , that has been isolated and characterized from acidic hot springs 24 , and amplicons assigned to this genus have since been recovered from acidic soils from China 25 , 26 . The complete genome sequencing of A. cellulolyticus revealed the presence of several genes encoding for enzymes involved in the breakdown of plant and fungal cell wall components 27 . We, therefore, hypothesise that Acidothermus species possess the ability to use a range of carbon sources and could play a significant role in the degradation of biomass and carbon cycling. For the second most abundant actinobacterial genus, Ma et al. 28 . recently found that Conexibacter may be involved in carbon and nitrogen biogeochemical cycles in natural forests of eastern China. Both genera may therefore play a central role in nutrient cycles at the Tiébaghi site. In addition, the slow-growing and high genomic GC content observed in Conexibacter strains are suggested as potential traits that could favour the survival of this group in stressful environments 29 , 30 . It is not worthy that besides the Tiébaghi site, members of this taxonomic group were mostly recovered from Goro’s iron crust soils. The third most abundant genus was Mycobacterium which is usually positively related to the prevalence of certain iron minerals in soil 31 , 32 . Overall, the over-representation of these three genera in iron crust environments supports the interest of further characterizing and comparing environmental constraints of distinct ultramafic soils, as well as identifying the key microbial taxa involved in soil biogeochemical cycle regulation. Focus on ectomycorrhizal fungi A remarkable finding was the considerable proportion of ectomycorrhizal fungal ASVs that did not find matches at the species level to the existing databases. Out of the 317 delineated ASVs only 41 were assigned to known species (more exactly to 23 described species; several ASVs can be assigned to one species). Based on this dataset, this would indicate a hypothetical endemism rate of 87% in New Caledonia. This result, obtained from a soil eDNA and high-throughput sequencing approach, is consistent with the 95% calculation made by Carriconde et al. 5 on ectomycorrhizas and fruit bodies typed by Sanger sequencing. New Caledonia clearly hosts a high and unique ectomycorrhizal fungal diversity. This raises questions about the underlying evolutionary and ecological mechanisms that have led to this exceptional and exclusive diversity in this remote territory. The New Caledonian biodiversity is also facing various severe threats (e.g., open cast mining, pollution, frequent and extended wildfires and invasive species introduction) and fungi, as all other living forms, are prone to extinction risks. This high endemism rate points out the effort that should be made to identify fungal species under threat and establish on-the-ground species-oriented conservation plans, an aspect that has never been considered and addressed in this biodiversity hotspot. Concerning the fungal ectomycorrhizal taxonomic composition, the Cortinarius genus was over-represented in numerous ultramafic vegetation types, corroborating the observations made by Carriconde et al. 5 . on ectomycorrhizas and fruit bodies collected in N. aequilateralis , A. gummiferum and mixed rainforest stands in the south of New Caledonia. With respect to the generalization of this pattern on a larger geographical scale and to other vegetation types than rainforests, it is likely that ultramafic rainforests and shrublands dominated by ectomycorrhizal plant species are characterized by the prevalence of Cortinarius . Nevertheless, despite the limited number of samples, Cortinarius was not the main group in ectomycorrhizas collected in ultramafic substrates from Acacia spirorbis roots 33 , a shrub legume that can dominate the ecosystems where it occurs. Future in-depth examinations must be undertaken to confirm or not the pre-eminence of Cortinarius in ecosystems dominated by a single plant species in New Caledonia. Such dominance by this fungal ectomycorrhizal group has direct implications in terms of ecosystems functioning: it appears that Cortinarius has retained from their saprophyte ancestors the capacity to degrade organic matter for potentially mobilizing nitrogen 34 , 35 , a limited element in ultramafic soils 5 . Finally, it is noteworthy that, in addition to the high relative abundances of ectomycorrhizal fungi in ultramafic vegetation dominated by ectomycorrhizal trees ( N. aequilateralis or A. gummiferum ) in forests and shrubs ( Tristaniopsis spp. ) in maquis, this functional group was also prevalent in Goro’s closed vegetation, where only sparse individuals of known ectomycorrhizal plants have been recorded. Isolated ectomycorrhizal plants are unlikely to account for such high ectomycorrhizal fungal abundance and richness ( G. deplancheanum -dominated vegetation harbouring the highest ectomycorrhizal fungal richness). As stated by Fernandez Nuñez et al. 11 , these observations clearly highlight the need to determine the mycorrhizal status of the New Caledonian flora, with only 14 plant species inventoried as ectomycorrhizal to date. This will improve our understanding of the extent to which this mutualistic association influences the ecosystems functioning in this extraordinary hotspot of biodiversity. A geographical structure of soil microbial communities Finally, in our analysis, structure investigations (through PERMANOVA analyses, bipartite networks and communities’ partitioning) revealed a geographical clustering of soil microbial communities, a result in accordance with the initial work of Gourmelon et al. 10 undertaken at two distant massifs (Rivière Blanche and Kopéto). For fungi, the two studied iron crust sites, Tiébaghi and Goro, were particularly apart from the other locations and distinct from each other as well. For bacteria, only Goro appeared clearly different. Overall, such geographical structure seems to indicate that each site exhibits its own microbial community. Ultramafic outcrops are patchily distributed and represent edaphic islands. This spatial discontinuity could have led to geographical (allopatric) speciation, contributing to this species diversity 36 . Soil microorganisms, like plants 37 , may thus display microendemic patterns. In terms of soil microbial conservation, this geographical structure has a direct implication: it indicates that each ultramafic massif might be considered as a conservation unit by itself, especially for the isolated massifs such as Tiébaghi and Kopéto. For the “Massif du Grand Sud”, with a surface area of 3015 km², making it the largest continuous ultramafic outcrop in the world 38 , different hotspots of fungal and bacterial diversity likely exist and may result in many priority areas for soil conservation. Distinct and unique microbial communities may also be hosted by the other types of soils encountered in New Caledonia. Samples from Maré island collected on aluminic-rich soils (gibbsic Ferralsols) were, indeed, completely dissimilar from ultramafic soils. The interest in focusing on soil biota for establishing priority areas has been very recently spotlighted by Guerra et al. 39 at the Earth scale. However, despite the input of this work, the approach remains imprecise, with many regions being not considered, including well-known biodiversity hotspots like New Caledonia. A high-resolution investigation of microorganisms, and other soil-living organisms (e.g., protists, nematodes, earthworms, and arthropods), using high-throughput sequencing technologies, on a large array of soil types could constitute a next future research challenge. At the scale of this territory and more widely, it will undeniably help to put on the map this still largely neglected hidden biodiversity and will contribute to a more holistic perception of nature conservation."
} | 6,689 |
22823186 | null | s2 | 7,822 | {
"abstract": "Recently, the specific hybridization of DNA molecules has been used to construct self-assembled devices, such as the mechanical device to mimic cellular protein motors in nature. Here, we present a new light-powered DNA mechanical device based on the photoisomerization of azobenzene moieties and toehold-mediated strand displacement. This autonomous and controllable device is capable of moving toward either end of the track, simply by switching the wavelength of light irradiation, either UV (365 nm) or visible (>450 nm). This light-controlled strategy can easily solve one main technical challenge for stepwise walking devices: the selection of routes in multipath systems. The principle employed in this study, photoisomerization-induced toehold length switching, could be further useful in the design of other mechanical devices, with the ultimate goal of rivaling molecular motors for cargo transport and macroscopic movement."
} | 233 |
28425176 | PMC5658603 | pmc | 7,824 | {
"abstract": "Summary Polyhydroxyalkanoates ( PHA s) are natural polyesters of increasing biotechnological importance that are synthesized by many prokaryotic organisms as carbon and energy storage compounds in limiting growth conditions. PHA s accumulate intracellularly in form of inclusion bodies that are covered with a proteinaceous surface layer (granule‐associated proteins or GAP s) conforming a network‐like surface of structural, metabolic and regulatory polypeptides, and configuring the PHA granules as complex and well‐organized subcellular structures that have been designated as ‘carbonosomes’. GAP s include several enzymes related to PHA metabolism (synthases, depolymerases and hydroxylases) together with the so‐called phasins, an heterogeneous group of small‐size proteins that cover most of the PHA granule and that are devoid of catalytic functions but nevertheless play an essential role in granule structure and PHA metabolism. Structurally, phasins are amphiphilic proteins that shield the hydrophobic polymer from the cytoplasm. Here, we summarize the characteristics of the different phasins identified so far from PHA producer organisms and highlight the diverse opportunities that they offer in the Biotechnology field.",
"conclusion": "Concluding remarks The generic name of ‘phasin’ denotes a set of proteins which indeed share the ability to recognize and adsorb to PHA polyesters. They play an essential contribution in the physical stabilization of the PHA granule within the cell, ensure the correct distribution of the polyester upon cell division and assist other proteins (synthases and depolymerases) on PHA metabolism. Nevertheless, their specific role is highly dependent both on the microbial strain and on the metabolic state of the cell. Their versatility is such that they may even participate in opposite events (e.g. synthesis and degradation of the PHA polymer). Besides, their strong affinity to PHA allows their use as protein affinity tags for polymer functionalization and therefore constitutes an opportunity to develop valuable applications in biotechnology and biomedicine. Although little structural data are still available, phasins are predicted to acquire relatively simple, amphipathic, three‐dimensional structures and to bind to PHA via non‐specific hydrophobic interactions. This makes them amenable to be easily engineered to produce recombinant variants that display a modulated affinity to PHA, that may be useful both for in vivo PHA production and in vitro biotechnological and biomedical applications.",
"introduction": "Introduction Polyhydroxyalkanoates (PHAs) are natural polyesters produced and accumulated by diverse organisms from the Bacteria and Archaea kingdoms as energy and carbon storage compounds under nutrient limitation conditions (i.e. nitrogen, oxygen or phosphorus) but in the presence of an excess of carbon sources (Anderson and Dawes, 1990 ; Lee, 1996 ). These polymers have acquired notoriety in recent years because they display plastic properties similar to their oil‐derived counterparts, but show biodegradability and biocompatibility features which results in a versatile and eco‐friendly alternative (Madison and Huisman, 1999 ; Potter and Steinbuchel, 2006 ; Keshavarz and Roy, 2010 ). PHAs were first described by M. Lemoigne in France, who in the 1920s reported the presence of poly(3‐hydroxybutyrate) [P(3HB)], in the cytoplasm of Bacillus megaterium (Lemoigne, 1926 ). Since then, over 300 species, including both Gram‐positive and Gram‐negative bacteria, have been described with the metabolic ability to synthesize PHAs (Steinbuchel and Fuchtenbusch, 1998 ; Zinn et al ., 2001 ; Suriyamongkol et al ., 2007 ; Chanprateep, 2010 ; Keshavarz and Roy, 2010 ). Chemically, PHAs are polyoxoesters of R‐hydroxyalkanoic acid monomers. They are usually classified depending on the number of carbon atoms of the alkyl groups: small chain length PHAs (scl‐PHAs) contain 3–5 carbon atoms [as poly(3‐hydroxybutyrate) ‐P(3HB)‐ or poly(4‐hydroxybutyrate) ‐P(4HB)], whereas medium chain length PHAs (mcl‐PHAs) possess 6–14 carbon atoms [e.g. poly(3‐hydroxyhexanoate), ‐P(3HHx) or poly(3‐hydroxyoctanoate) – P(3HO)]. Long‐chain‐length PHAs (lcl‐PHAs) consisting of hydroxyacids with more than 14 carbon atoms are more scarcely found (Rutherford et al ., 1995 ; Singh and Mallick, 2009 ). These differences are mainly due to the substrate specificity of the PHA synthases from the particular microorganism (Park et al ., 2012 ). Moreover, the incorporation of different monomer units in the same chain gives rise to heteropolymers with new properties. The properties and functionalities of the PHAs depend on their monomer composition: whereas scl‐PHAs show thermoplastic properties similar to polypropylene, mcl‐PHAs display elastic features similar to rubber or elastomer (Keshavarz and Roy, 2010 ; Park et al ., 2012 ). Applications of PHAs in the industry are widespread, ranging from the manufacturing of packages and covers to the generation of enantiomeric pure chemicals (Philip et al ., 2007 ) or as protein immobilization supports (Draper and Rehm, 2012 ; Dinjaski and Prieto, 2015 ; Hay et al ., 2015 ). Of significant relevance is the implementation of PHAs in the biomedical discipline, especially supported by the recent FDA approval for P(4HB) to be used as suture material (Tepha Inc., MA, USA). The utility of PHAs in this field arises from their biocompatibility characteristics and has found its application in a variety of processes such as drug delivery, development of medical devices and construction of tissue engineering scaffolds (Misra et al ., 2006 ; Wu et al ., 2009 ; Wang et al ., 2010 ; Xiong et al ., 2010 ; Brigham and Sinskey, 2012 ; Martinez‐Donato et al ., 2016 ; Rubio Reyes et al ., 2016 ). The PHA polymer accumulates in the cytoplasm in the form of water‐insoluble granules (Fig. 1 ), the number per cell and size of which depend on the different species and the culture conditions (Jendrossek and Pfeiffer, 2014 ). Early studies carried out by Merrick′s group showed that these inclusions were constituted by approximately 98% (w/w) PHA, 2% granule‐associated proteins (GAPs) and 0.5% phospholipids (Griebel et al ., 1968 ). Since then, several studies have confirmed the presence of a phospholipid layer in PHA preparations (Parlane et al ., 2016 , and references therein). However, some data have put into question the actual presence of the lipid coat in vivo (Potter and Steinbuchel, 2006 ; Beeby et al ., 2012 ; Jendrossek and Pfeiffer, 2014 ), especially from electron cryotomography (Wahl et al ., 2012 ) and fluorescence microscopy (Bresan et al ., 2016 ) results, according to which the presence of the lipid layer might arise from an experimental artefact on PHA extraction and preparation. Figure 1 Scheme of the structure of PHA granules. Four different types of GAPs have been identified so far, namely PHA synthases, PHA depolymerases, phasins and other proteins (Steinbuchel et al ., 1995 ), the latter including transcriptional regulators as well as hydrolases, reductases and other enzymes involved in the synthesis of PHA monomers (Jendrossek and Pfeiffer, 2014 ; Sznajder et al ., 2015 ). Among them, phasins, which received their name in analogy to oleosins [proteins on the surface of oil globules found in oleaceous plants (Steinbuchel et al ., 1995 )], are the most abundant polypeptides in the PHA carbonosome (Mayer et al ., 1996 ). These low molecular weight proteins normally contain a hydrophobic domain, associated with the PHA, and a hydrophilic/amphiphilic domain exposed to the cytoplasm (Potter and Steinbuchel, 2005 ). On the basis of their sequence, phasins are distributed in four families according to the Pfam database ( http://pfam.xfam.org/ ), namely PF05597, PF09602, PF09650 and PF09361. A recent survey showed that a high percentage of phasins and phasin‐like proteins contains a leucine‐zipper motif in their amino acid sequences, suggesting that oligomerization is a common organization mechanism in these polypeptides (Maestro et al ., 2013 ). In the recent years, a large number of phasins have been identified, constituting a phylogenetically heterogeneous group of proteins. We will review the current knowledge on the most representative phasins participating in important biological functions (summarized in Table 1 ) such as the formation of network‐like covers on the PHA granule surface (Dennis et al ., 2003 , 2008 ; Pfeiffer and Jendrossek, 2011 ) or the regulation of the synthesis, morphology, distribution during cell division and degradation of the storage granules (Mezzina and Pettinari, 2016 ). Finally, the biotechnological potential of this group of proteins will be discussed. Table 1 List of the phasins reviewed in the text, with their most relevant characteristics Organism Phasin Molecular mass (kDa) UNIPROT accession number (localization) Most relevant characteristics and roles References \n Ralstonia eutropha \n PhaP1 Reu \n 20.0 \nAAC78327 \n(chromosome 1)\n \nHomotrimer .\nMajor phasin present in R. eutropha \n \nPlays role in the amount, size and number of granules, and in their degradation .\nBiotechnological application as immobilization tag\n (Steinbuchel et al ., 1995 ; Wieczorek et al ., 1995 ; York et al ., 2001a ; York et al ., 2001b ; Potter et al ., 2002 ; York et al ., 2002 ; Potter et al ., 2004 ; Banki et al ., 2005 ; Barnard et al ., 2005 ; Backstrom et al ., 2007 ; Kuchta et al ., 2007 ; Neumann et al ., 2008 ; Wang et al ., 2008 ; Yao et al ., 2008 ; Chen et al ., 2014 ; Sznajder et al ., 2015 ) PhaP2 Reu \n 20.2 \nAAP85954 \n(plasmid pHG1)\n Secondary participation in PHB accumulation and mobilization (Schwartz et al ., 2003 ; Potter et al ., 2004 ) PhaP3 Reu \n 19.6 \nAY489113 \n(chromosome 1)\n Secondary participation in PHB accumulation and mobilization (Potter et al ., 2004 ) PhaP4 Reu \n 20.2 \nAY489114 \n(chromosome 2)\n Secondary participation in PHB accumulation and mobilization (Potter et al ., 2004 ) PhaP5 Reu \n 15.7 \nH16_B1934 \n(chromosome 2)\n Secondary participation in PHB accumulation and mobilization (Pfeiffer and Jendrossek, 2011 ) PhaP6 Reu \n 22.7 \nH16_B1988 \n(chromosome 2)\n Secondary participation in PHB accumulation and mobilization (Pfeiffer and Jendrossek, 2012 ) PhaP7 Reu \n 16.4 \nH16_B2326 \n(chromosome 2)\n Secondary participation in PHB accumulation and mobilization (Pfeiffer and Jendrossek, 2012 ) \n Pseudomonas putida \n PhaF 26.3 Q9Z5E6 \nTetramer. \nResponsible for non‐specific binding to DNA .\nIntrinsically disordered in its majority unless bound to its ligands .\nInvolved in the PHA biosynthesis, localization of the granules in the cell and in their distribution between daughter cells during cell division .\nTranscriptional regulator\n (Prieto et al ., 1999 ; Moldes et al ., 2004 ; Ren et al ., 2010 ; Galan et al ., 2011 ; Dinjaski and Prieto, 2013 ; Maestro et al ., 2013 ) PhaI 15.4 Q9Z5E7 \nInvolved in the biosynthesis and accumulation of PHA .\nBiotechnological application as BioF affinity tag to immobilize or purify fusion proteins\n (Prieto et al ., 1999 ; Moldes et al ., 2004 ; Moldes et al ., 2006 ; Ren et al ., 2010 ; Dinjaski and Prieto, 2013 ; Maestro et al ., 2013 ) \n Pseudomonas sp. 61‐3 PhaF 25.6 Q8L3N9 Phasin bound to P(3HB‐co‐3HA) copolymers solely when granules are enriched in 3HA (C6–C12) in more than 13 mol% (Matsumoto et al ., 2002 ; Hokamura et al ., 2015 ) PhaI 15.4 Q8L3P0 Phasin bound to P(3HB‐co‐3HA) copolymers solely when granules are enriched in 3HA (C6–C12) in more than 13 mol% (Matsumoto et al ., 2002 ; Hokamura et al ., 2015 ) PhbP 20.4 A0A0K2QTP6 Phasin bound to P(3HB‐co‐3HA) copolymers solely when granules are enriched in 3HB in more than 87 mol% (Matsumoto et al ., 2002 ; Hokamura et al ., 2015 ) \n Paracoccus denitrifican s PhaP Pde \n 16.5 Q9WX81 Involved in the PHA granule formation, ensuring the correct number and size of granules by preventing coalescence and their distribution throughout the cytoplasm (Maehara et al ., 1999 ) \n Rhodococcus ruber \n GA14 14.2 \nQ53051 \n(ORF3)\n \nBinding to the PHA through two hydrophobic patches present in the C‐terminal region of the protein \nControl of the granule size\n (Pieper and Steinbuchel, 1992 ; Pieper‐Furst et al ., 1994 ; Pieper‐Furst et al ., 1995 ) \n Azotobacter sp . FA‐8 PhaP Az \n 20.4 Q8KRE9 \nTetramer. \nPHA binding by amphipathic α‐helices induces protein structuration. \nPromotes bacterial growth and PHA synthesis. \nGeneral stress‐reducting action. \nChaperone‐like mechanism\n (Pettinari et al ., 2003 ; de Almeida et al ., 2007 ; de Almeida et al ., 2011 ; Mezzina et al ., 2014 ; Mezzina et al ., 2015 ) \n Aeromonas caviae \n PhaP Ac \n 12.6 Q79EN2 Important role in biosynthesis and metabolism of PHA (Fukui et al ., 2001 ; Saika et al ., 2014 ; Ushimaru et al ., 2014 ; Kawashima et al ., 2015 ; Ushimaru et al ., 2015 ) \n Aeromonas hydrophila \n PhaP Ah \n 12.6 O32470 \nTetrameric in solution, monomeric when bound to PHA granules. \nInvolved in PHA biosynthesis. \nControls granule size and number. \nTranscription regulator of pha C gene\n (Tian et al ., 2005 ; Zhao et al ., 2016 ) \n Rhodospirillum rubrum \n ApdA 17.5 Q8GD50 55% identity with Mms16 from Magnetospirillum \n (Handrick et al ., 2004a ; Handrick et al ., 2004b ) \n Bradyrhizobium diazoefficiens \n PhaP1 Bd \n 12.3 Q89JW4 Predominantly alpha‐helical (Yoshida et al ., 2013 ; Quelas et al ., 2016 ) PhaP2 Bd \n 17.3 Q89IS9 Predominantly alpha‐helical (Yoshida et al ., 2013 ; Quelas et al ., 2016 ) PhaP3 Bd \n 12.4 Q89H66 \nPredominantly alpha‐helical. \nMinor expression\n (Yoshida et al ., 2013 ; Quelas et al ., 2016 ) PhaP4 Bd \n 15.4 Q89DP4 \nPredominantly alpha‐helical. \nC‐terminal region very rich in alanine residues. \nFavoured expressed when using yeast extract‐mannitol medium\n (Yoshida et al ., 2013 ; Quelas et al ., 2016 ) John Wiley & Sons, Ltd"
} | 3,513 |
39836325 | PMC11750901 | pmc | 7,825 | {
"abstract": "Abstract L-valine holds wide-ranging applications in medicine, food, feed, and various industrial sectors. Escherichia coli , a pivotal strain in industrial L-valine production, features a concise fermentation period and a well-defined genetic background. This study focuses on mismatch repair genes ( mutH , mutL , mutS , and recG ) and genes associated with mutagenesis ( dinB , rpoS , rpoD , and recA ), employing a high-glucose adaptive culture in conjunction with metabolic modifications to systematically screen for superior phenotypes. This approach enhances the spontaneous survival rate of stress cells and facilitates the enrichment of positive mutations. Leveraging a multi-fragment seamless recombination technique, we successfully assembled the ilvBN , ilvC , ilvE , and ilvD pathway enzyme genes, transforming E. coli from a non-producer into a proficient L-valine producer capable of generating up to 6.62 g/L. Through a synergistic application of self-evolution engineering and metabolic engineering strategies, the engineered E. coli strain exhibited significantly enhanced tolerance and demonstrated heightened accumulation of L-valine. Key points • The innovation centered on mutated genes and mismatch repair genes • By integrating modification with adaptive culture, a superior phenotype was attained • Double plasmids expressing enzymes for L-valine production in E. coli were obtained Graphical Abstract \n Supplementary Information The online version contains supplementary material available at 10.1007/s00253-024-13334-9.",
"introduction": "Introduction L-valine, also known as L-2,6-diaminocaproic acid, stands out as a superior nutritional supplement (Nielsen 2019 ; Olin-Sandoval et al. 2019 ). Its production primarily relies on microbial fermentation. Key microbial strains for L-valine production include Escherichia coli (Savrasova and Stoynova 2019 ), Corynebacterium glutamicum (Liu et al. 2021 ; Wang et al. 2024a ), and Saccharomyces cerevisiae (Ohashi et al. 2020 ; Takpho et al. 2018a , b ). However, the metabolic network of the high-yielding L-valine strain C. glutamicum is intricate (Zhang et al. 2018 ), posing significant challenges for genetic modifications due to limited genetic information. When S. cerevisiae is used as a host for metabolic engineering, it lacks a strongly regulated promoter, leading to a prolonged fermentation period, complex nutritional requirements, and unsuitability for high-density culturing. Additionally, the accumulation of nutrients and monosaccharides in the later stages of fermentation increases the cost of separation and purification (Takpho et al. 2018a , b ). On the other hand, E. coli has emerged as a preferred strain for amino acid production due to its well-defined genetics, rapid fermentation cycle, high glucose utilization, and advanced gene manipulation techniques (Hao et al. 2020 ; Zheng et al. 2018 ). Glucose serves as the primary carbohydrate source for microbial fermentation-based L-valine production (Xu et al. 2019 ). An insufficient glucose supply during fermentation can slow strain growth and reduce product yields, while excessively high glucose concentrations can increase osmotic pressure, hindering glucose and nutrient absorption, thus limiting strain growth and product synthesis. Wild-type E. coli exhibits limited glucose tolerance and metabolic synthesis capabilities. The inherent complexities of the biological system pose significant challenges in engineering novel phenotypes in E. coli (Xu et al. 2019 ). To overcome these challenges, further research is needed to enhance the glucose tolerance and metabolic efficiency of E. coli for improved L-valine production. Industrially, L-valine is primarily generated through microbial fermentation. However, achieving a phenotype capable of tolerating high glucose stress and ensuring a high yield of L-valine has emerged as a bottleneck constraining the extensive application of L-valine. Key challenges include (1) most wild-type strains lack the capacity to produce L-valine in substantial quantities, with limited yields primarily attributed to strain tolerance constraints and other factors. (2) Traditional mutation breeding serves as the primary method for developing industrial L-valine-producing strains. Nevertheless, strains obtained through these conventional breeding methods exhibit unclear and unstable genetic backgrounds and characteristics, accompanied by a low mutation breeding rate for L-valine-producing strains. (3) The intricate synthesis and metabolism networks of L-valine pose challenges, as the laborious knockout or overexpression of genes within the L-valine synthesis pathway can induce cellular damage or increase metabolic burden. Single metabolic engineering modifications result in restricted improvements in product yield. The regulation of microbial cell factory stress resistance through tolerance engineering is integral to fortifying stress defense capabilities, expediting stress responses, and enhancing damage repair mechanisms by reinforcing cellular barriers. However, owing to the intricacies of microbial system metabolism and regulatory networks, obtaining an exceptional robustness phenotype proves challenging. Consequently, tolerance evolutionary engineering has garnered increasing significance in the screening of microbial cell factories with heightened robustness (Wen et al. 2020 ; Kim et al. 2020 ). Autonomous evolutionary mutation, also known as tolerance evolutionary engineering, involves harnessing the malleability of microbial genomes to continually evolve microbial populations under specific selective pressures. Through intricate regulatory mechanisms and complex resistance strategies, microorganisms evolve to endure and adapt to diverse stress pressures, acquiring beneficial mutations (Kim et al. 2020 ). This approach can be instrumental in enhancing microbial cell growth, augmenting chemical concentration, yield, and production intensity (Choe et al. 2019 ), as well as uncovering previously unknown biological regulatory mechanisms. The genetic mutation in E. coli is governed by both fidelity systems and mutation systems. Fidelity systems, exemplified by mismatch repair (MMR) (Olivares-Hernández et al. 2022 ), possess the capability to rectify base pair mismatches, thereby ensuring high fidelity during DNA replication. On the other hand, mutation systems, such as the SOS response, can amplify the effectiveness of DNA damage repair, subsequently enhancing cell survival and mutation rates (Doong et al. 2018 ; Schroeder et al. 2018 ). During periods of rest or nutrient deprivation in E. coli , a “general stress response” is induced. Limited methods are known to elevate the spontaneous mutation rate in stressed cells by leveraging the general stress response of rpoS . These methods involve two approaches: first, the upregulation of DNA Pol IV (Moore et al. 2017 ; Zheng et al. 2018 ; Wang et al. 2019 ) and DNA Pol V, which are prone to mismatches; second, the downregulation of the MutS / MutH enzymes responsible for MMR (Hu et al. 2017 ). In the context of this study, the effective control of spontaneous mutation rates in cells enabled the screening of stable beneficial mutations conducive to the efficient biological production of organic acids and amino acids. This research transformed E. coli into a strain with elevated L-valine productivity, offering broad insights for the development of high-yield L-valine strains and the regulation of metabolic pathways.\n\nPlasmid and knockout box fragments were introduced into host E. coli by electroporation and the screen of the glucose tolerance strain The Cas9 plasmid, the mutH knockout box, and the pTargetF-sgRNA ( mutH ) plasmid were introduced into E. coli JM109 recipient cells. The knockout box fragment replaced the mutH gene in the original genome, a process well-documented in literature (Wendisch 2019 ; Tian et al. 2019 ). Single colonies were selected for PCR validation, resulting in the acquisition of a single-gene knockout strain, denoted as E. coli JM109-Δ mutH and named EC01. The construction methodology for E. coli JM109-Δ mutL , E. coli JM109-Δ mutS , and E. coli JM109-Δ recG strains followed the same procedure as described above, resulting in strains named EC02, EC03, and EC04, respectively. Primers employed in the PCR validation are detailed in Table S1 . Fig. S1 depicts the structural diagram of the pTargetF-sgRNA plasmid designed for different genes. The temporal variation in the population of different bacterial strains was monitored under varying glucose concentrations, ranging from 0 to 30–80 g/L. The mutation frequency and glucose tolerance of the strains were employed as test indices to assess the survival rates of strains exhibiting high glucose tolerance. Key genes, contributing to the highest survival rates, were identified through this evaluation. The survival rate of glucose tolerance was quantified by observing the increase in the number of glucose-tolerant mutants on the growth medium. The calculation method, as outlined in prior research (Wang et al. 2019 ), involved culturing EC01 on solid screening medium for 5 days. The survival rate of the strain, indicative of increased glucose tolerance, was expressed by the augmentation in the number of glucose-tolerant strains on the medium. Specifically, A: The total number of resistant colonies grown on the 2nd day of E. coli JM109. B: The total number of resistant colonies grown on the 5th day of E. coli JM109. C: The total number of resistant colonies grown on the 2nd day of EC01. D: The total number of resistant colonies grown on the 5th day of EC01. The increased number of E. coli JM109 colonies is calculated as (B-A), and the increased number of EC01 colonies is calculated as (D-C). The survival rate of EC01 is expressed as (D-C)/3. To quantify the improvement in the survival rate of EC01 compared to that of E. coli JM109, the formula is as follows: The survival rate of EC01 was improved by [A (D-C)—C (B-A)]/(minimum common multiple of AC). This method enables a comprehensive evaluation of the enhanced survival rate of EC01 in comparison to E. coli JM109 under high glucose conditions.",
"discussion": "Discussion Genes encoding mismatch repair (MMR) proteins, including mutH , mutL , mutS , and recG , were systematically knocked out in E. coli JM109. Additionally, emergency response (SOS)–related genes, namely, dinB , rpoS , rpoD , and recA , were individually overexpressed. High concentrations of glucose were employed as the screening conditions for strain breeding, aiming to enhance the mutation rate of E. coli and develop a strain capable of tolerating high glucose stress for L-valine production. The approach involved a combination of high glucose adaptive culture and metabolic modification to selectively screen for superior phenotypes. This led to an increase in the spontaneous high survival rate of stress-adapted cells and the enrichment of positive mutations in the E. coli strain geared towards L-valine production. This research serves as a foundational basis with universal applicability, overcoming the bottleneck that hinders large-scale L-valine production. Microbial cells, possessing inherent self-regulation mechanisms, exhibit high adaptability to environmental conditions and possess the capability for self-reproduction, leading to variations through mutation or gene recombination. The study, conducted under high glucose concentrations, resulted in enhanced tolerance of E. coli to elevated glucose levels and the acquisition of resistance to both high glucose and high osmotic pressure. Through systematic screening and genetic modifications, the L-valine synthesis pathway was engineered, yielding a strain with heightened glucose tolerance and an augmented capacity for L-valine accumulation. The traditional mutagenesis breeding strategy has led to an increase in the mutation rate of L-valine to some extent. However, this approach has limitations, including a low frequency of beneficial mutations, labor-intensive processes, and uncertainties in genotype and phenotype changes (Zhang et al. 2018 ; Wang et al. 2019 ). Strains obtained through this method often exhibit unclear genetic backgrounds, unstable genetic characteristics, and a propensity for random mutations after multiple passages (Savrasova and Stoynova 2019 ). A mutant strain, C. glutamicum VWB-1, with high L-valine yield, was achieved through multiple rounds of random mutagenesis. However, this strain demonstrated distinct characteristics in cell growth, glucose consumption, and amino acid production compared to the original strain ATCC13869. Notably, the original strain exhibited faster growth than VWB-1, indicating adverse factors for cell growth in the mutant strain VWB-1 (Zhang et al. 2018 ). Various strategies have been explored for modifying microbial cell factories to increase L-valine yield, as outlined in Table 2 . These strategies predominantly focus on rational metabolic engineering design within the metabolic synthesis network pathway of L-valine. This includes interventions at the DNA, protein, cofactor, metabolite, and cellular levels, resulting in significant improvements in L-valine production. Rational metabolic engineering interventions at the DNA level involve changes in promoter activity through site-directed mutation and alteration of promoter activity, such as the regulation of leucine-reactive protein ( lrp ) and overexpression of ilvBNC and ilvGMEDA operons (Park et al. 2011 ; Wang et al. 2022 ). Protein-level modifications include alterations to the catalytic and regulatory domains of enzymes (Zhang et al. 2018 ; Wang et al. 2024a ). Interventions at the cofactor level aim to enhance the supply of NADH and NADPH (Zhang et al. 2018 ; Sheremetieva et al. 2022 ; Wang et al. 2024a ). Metabolite-level adjustments include the removal of product feedback regulation ( ilvN knockdown), enhancement of carbon metabolic flow to L-valine (overexpression of ilvBNC and ilvGMEDA ) (Zhang et al. 2018 ; Sheremetieva et al. 2022 ; Schwentnera et al. 2018 ), and weakening of by-product synthesis pathways ( panB , ilvA , and leuA ) (Park et al. 2011 ; Wang et al. 2024b ). At the cellular level, extracellular transport of L-valine is enhanced through the deregulation of transporter genes ygaZ and the expression of inner membrane protein gene ygaH (Park et al. 2011 ). Table 2 Research status of L-valine production by different microorganisms Microbial cell factories Strategies of modification Enhanced phenotype References S . cerevisiae Constructed several variants of Ilv6 by introducing amino acid substitutions (Asn86Ala, Gly89Asp, and Asn104Ala) Valine production increased fourfold (Takpho et al. 2018a , b ) E . coli Use NADH-dependent leucine dehydrogenase (LeuDH; EC 1.4.1.9) Bcd from B. subtilis instead of the native NADPH-dependent pathway Improved the L-valine concentration 4.1 g/L from 9.1 g/L ethanol (Savrasova and Stoynova 2019 ) E . coli Gene alsS from B. subtilis was separately placed under the strong Ptrc promoter and introduced into the pseudogene ycdN locus of E. coli W3110 ΔlacI. One copy of alsS under the control of the Ptrc promoter was integrated. Overexpression of ilvED and ilvC genes. Introduction of Ptrc-controlled brnFE . Overexpression of the spoT [R290E, K292D] . Knockout ldhA , pflB , adhE , ilvC , the native copy, and the added copy were replaced with P trc -ilvC M , ilvE was replaced with bcd Improved the L-valine concentration 41.2 g/L from 84 g/L ethanol (Hao et al. 2020 ) E . coli Overexpression of PdhR and inhibition of the expression of RpoS Produced 92 g/L of L-valine (Hao et al. 2022 ) C. glutamicum The key genes involved in the biosynthesis of L-valine, ilvBN , ilvC , ilvD , and ilvE were upregulated. Downregulation of leuB and ilvA. Upregulation of the branched chain amino acid transporter genes brnFE. Upregulation of the genes involved in phosphate pentose pathway and TCA pathway and the genes coding for elongation factors and ribosomal proteins Produced 29.85 g/L of L-valine (Zhang et al. 2018 ) C. glutamicum Construct C. glutamicum ΔppcΔaceEΔalatΔpqo . An improved biosensor based on the Lrp-type transcriptional regulator and temperature sensitive. The C. glutamicum strain was mutagenized by atmospheric and room temperature plasma replication Improved the L-valine concentration 2.63 g/L from 3.2 g/L ethanol (Han et al. 2020 ) C. glutamicum Replace the NADPH-dependent transaminase BCA T with NAD-dependent leucine dehydrogenase (LeuDH) from Lysinibacillus sphaericus, ilvN gene was introduced Produced 172.2 g/L of L-valine (Sheremetieva et al. 2022 ) C. glutamicum An acetohydroxy acid synthase mutant from an industrial L-valine producer was introduced and a cofactor-balanced pathway was optimized, followed by the activation of the nonphosphoenolpyruvate-dependent carbohydrate phosphotransferase system and the introduction of an exogenous Entner–Doudoroff pathway. Anaplerotic pathways were weakened, and the tricarboxylic acid cycle via start codon substitution in icd was attenuated, encoding isocitrate dehydrogenase. An L-valine biosensor-dependent genetic circuit was established to dynamically repress citrate synthase expression The engineered strain produced 103 g/L of L-valine with a high yield of 0.35 g/g glucose (Wang et al. 2024a ) Bacillus subtilis Manipulate the native L-valine biosynthetic pathway by relieving transcriptional and allosteric regulation. Identify and eliminate factors limiting L-valine overproduction. Inactivating pdhA , encoding the E1α subunit of the pyruvate dehydrogenase complex and leuA and ilvA , respectively encoding 2-isopropylmalate synthase and L-threonine dehydratase Produced 4.61 g/L of L-valine (Westbrook et al. 2018 ) Klebsiella oxytoca Dihydroxy acid dehydratase( puDHT ) bcd was inserted at the site of the adhE , α-acetolactate decarboxylase( ilvC M ) brnFE was inserted at the site of the pflB , alsS encoding ALS from was inserted at the site of the budC , ilvC was introduced at the position of gldA , and puDHT was replaced by dhaD Improved the L-valine concentration 81.3 g/L from 122 g/L (Cao et al. 2024 ) Klebsiella pneumoniae Knockout budA , encoding an α-acetolactate decarboxylase. Knockout ldhA , encoding a lactate dehydrogenase. Knockout ipdC , encoding indole-3-pyruvate decarboxylase. Overexpression ilvE , encoding Transaminase B. Disrupting brnQ , encoding a branched-chain amino acid transporter Produced 22.4 g/L of L-valine (Wang et al. 2022 ) Cupriavidus necator H16 Citramalate synthase was introduced to simplify isoleucine synthesis pathway. Blocking poly-hydroxybutyrate biosynthesis resulted in significant accumulation of isoleucine and valine 319 mg/L valine were produced by consuming CO 2 (Wang et al. 2024b ) The existing literature on metabolic modification of E. coli for L-valine production is limited (Savrasova and Stoynova 2019 ; Zhang et al. 2018 ). Employing rational metabolic engineering, L-valine production has been enhanced based on transcriptome analysis (Cao et al. 2024 ). Park et al. conducted metabolic engineering on E. coli W, wherein the knockout of the ilvA gene directed more metabolic flux towards pyruvate, a crucial precursor of L-valine. Concurrently, the knockout of the lacI gene and the overexpression of ilvBN , lrp , and YgaZH genes, coupled with fed-batch fermentation, resulted in a noteworthy L-valine yield of 60.7 g/L after 29.5 h (Park et al. 2011 ). While metabolic modification has demonstrated its potential to increase L-valine yield, it faces metabolic bottlenecks affecting cell performance. These bottlenecks include the accumulation of toxic intermediates, cofactor imbalances, and insufficient enzyme activity. The intricate interactions within the metabolic pathways of engineered strains may contribute to these challenges (Chen et al. 2018 ). This complexity can lead to cellular damage or increased burden (Wytock and Motter 2019 ), emphasizing the need for effective screening methods to identify desired phenotypes in future studies. Consequently, the key to achieving high L-valine production lies in enhancing the positive mutation rate of E. coli . Under conditions of high glucose concentration and hyperosmolar pressure, microorganisms need to regulate their metabolism to adapt to the stressful environment. This adaptation primarily involves the production and accumulation of osmotic substances to maintain the balance of intracellular and extracellular water. The mechanism of microbial resistance to high glucose stress reveals a correlation between resistant microorganisms and extracellular compatible substances such as trehalose, glycerol, and organic acids. Additionally, in response to the need for osmotic pressure balance within and outside the cell, the cell undergoes a stress response. Oxidative stress and metabolite stress play crucial roles in microbial resistance to high glucose stress. The expression of the oxidative stress mechanism in E. coli is influenced by the activities of enzymes like intracellular superoxide dismutase (SOD) (Santomartino et al. 2019 ), intracellular catalase (Santomartino et al. 2019 ), intracellular peroxidase, cytoplasmic ATPase, and mitochondrial ATPase. Changes in the yield of the target product dynamically impact the oxidative stress response. Metabolite stress involves a non-enzymatic defense system where pressure-buffering metabolites including glutathione (Xu et al. 2017 ), thio-oxygen reducing protein (González et al. 2020 ), vitamin C (Kaźmierczak-Barańska et al. 2020 ), lysine (Nielsen 2019 ; Olin-Sandoval et al. 2019 ), and NADH/NADPH (Liu et al. 2020 ) serve as reducing agents to scavenge free radicals such as ultra-negative cations, hydroxyl radicals, organic free radicals, and organic peroxygens. Regulating the intracellular concentration of these pressure-buffering metabolites can repair damage at the metabolite level, enhancing the stress resistance of microbial cell factories. This study investigates an autonomous evolution approach, emphasizing fidelity genes such as mismatch repair protein genes mutH , mutL , mutS , and the helicase gene recG . Mutant genes include those susceptible to mutation, such as the DNA Pol IV gene dinB , stress factor-coding genes rpoS , rpoD , and the damage response protein-coding gene recA . The article also delves into the construction methodology of a high-glucose-tolerant and L-valine-producing E. coli strain. The innovation of this research lies in the focus on mutated genes and mismatch repair genes. The methodology involves utilizing high-glucose adaptive culture combined with metabolic modification to selectively screen for superior phenotypes. This approach enhances the spontaneous high survival rate of stress-adapted cells and enriches positive mutations. The breeding of E. coli strains capable of producing L-valine with heightened glucose tolerance, improved amino acid fermentation capabilities, and robust passage stability offers valuable insights for the development of E. coli strains with enhanced amino acid production and tolerance to environmental stress. In addition to glucose tolerance and growth performance, other potential factors contributing to the increased L-valine production in strains EC09 and EC10 include the following: enhanced metabolic routes with heightened enzyme activity, better intermediate accumulation, or less byproducts, bolstering L-valine synthesis. Specific gene expression patterns and altered regulatory elements in EC09 and EC10 for efficient L-valine-related gene transcription and translation. Cellular modifications in membrane, transporter, or compartmentalization influence L-valine accumulation and release. Unique genetic makeup with mutations and recombinations is related to L-valine synthesis in EC09 and EC10. To further elucidate these potential factors, the following experiments and analyses will be conducted. Sequence and compare EC09, EC10, and a reference strain’s genomes, focusing on amino acid metabolism genes. Apply RNA-seq to discern gene expression differences under varying conditions, pinpointing L-valine-related genes. Utilize metabolomics and proteomics to decipher metabolic flows and protein expressions during L-valine production, revealing key pathways and regulators. The results offer comprehensive insights for optimizing the breeding of high L-valine-producing strains and the regulation of metabolic pathways, holding paramount theoretical and practical significance. In comparison to existing technologies, the current innovation boasts noteworthy advantages. The approach facilitates targeted manipulation of mutated genes and mismatch repair genes. Utilizing a combination of high glucose adaptive culture and metabolic modification effectively screens for superior phenotypes. This strategy enhances the spontaneous high survival rate of stress-adapted cells and promotes the enrichment of positive mutations. Successful breeding of an L-valine-producing E. coli strain is achieved, showcasing favorable traits including good glucose tolerance, enhanced amino acid fermentation capabilities, and robust passage stability. The selection and modification of E. coli strains serve as a universal research foundation. This breakthrough addresses bottlenecks hindering the large-scale production of L-valine. The engineered strains exhibit elevated amino acid production levels and increased tolerance to environmental stress. Microbial cells exhibit a high degree of adaptability to the environment and possess self-regulation mechanisms, enabling them to reproduce through mutation or gene recombination. Under high concentration glucose conditions, E. coli ’s tolerance to elevated glucose concentrations is significantly enhanced, resulting in a notable improvement in E. coli ’s survival rate. In summary, this research not only contributes to the advancement of L-valine production but also establishes a versatile platform for further investigations into microbial adaptation and genetic optimization for enhanced industrial applications."
} | 6,608 |
30072533 | null | s2 | 7,826 | {
"abstract": "A major unresolved question in microbiome research is whether the complex taxonomic architectures observed in surveys of natural communities can be explained and predicted by fundamental, quantitative principles. Bridging theory and experiment is hampered by the multiplicity of ecological processes that simultaneously affect community assembly in natural ecosystems. We addressed this challenge by monitoring the assembly of hundreds of soil- and plant-derived microbiomes in well-controlled minimal synthetic media. Both the community-level function and the coarse-grained taxonomy of the resulting communities are highly predictable and governed by nutrient availability, despite substantial species variability. By generalizing classical ecological models to include widespread nonspecific cross-feeding, we show that these features are all emergent properties of the assembly of large microbial communities, explaining their ubiquity in natural microbiomes."
} | 240 |
26083597 | PMC4470831 | pmc | 7,827 | {
"abstract": "Single-neuron models are useful not only for studying the emergent properties of neural circuits in large-scale simulations, but also for extracting and summarizing in a principled way the information contained in electrophysiological recordings. Here we demonstrate that, using a convex optimization procedure we previously introduced, a Generalized Integrate-and-Fire model can be accurately fitted with a limited amount of data. The model is capable of predicting both the spiking activity and the subthreshold dynamics of different cell types, and can be used for online characterization of neuronal properties. A protocol is proposed that, combined with emergent technologies for automatic patch-clamp recordings, permits automated, in vitro high-throughput characterization of single neurons.",
"introduction": "Introduction \n In vitro patch-clamping is the gold standard used to investigate the intrinsic electrophysiological properties of neurons, but remains labour intensive and requires a trained experimentalist with high technical skills. In the last years, several platforms have been developed that automatize electrophysiological recordings for ion-channel screening and drug discovery [ 1 ]. Most of the existing platforms are, however, designed to record from mammalian cell lines or oocytes in which ion-channels of interest are artificially expressed [ 2 , 3 ]. In the near future, this technology is likely to be transferred to more complex setups, such as in vitro brain slices. High-throughput electrophysiology can be pushed forward with in vivo whole-cell patch-clamp recordings that are, at least partially, automatized [ 4 ]. With this technique, three to seven minutes are sufficient for a trained technician or a robot to automatically identify a cell and form a gigaohm seal of the same quality as achieved by an electrophysiologist [ 4 ]. This technological advance represents an important step towards high-throughput electrophysiology in vivo or on in vitro brain slices. To make sense of the large amount of data that automated patch-clamp can produce, adequate computational tools and experimental protocols have to be developed. Traditional protocols for single-neuron characterization rely on current-clamp injections of stimuli (e.g., square current pulses, ramps of current) that are specifically designed to extract a small number of parameters (e.g., membrane time constant, firing threshold). While this is a valid approach, the input currents adopted in these experiments are artificial and strongly differ from the signals that single neurons process in vivo . Moreover, the choice of the parameters used for single-neuron characterization is arbitrary and different parameters are generally estimated in separate sets of experiments. In this study, an alternative method is proposed in which the electrophysiological properties of neurons are characterized by means of simplified neuron models. Ideally, a single-neuron model should be sufficiently complex and flexible to capture, by a single change of parameters, the spiking activity of different neurons, but also simple-enough to allow robust parameter estimation [ 5 , 6 ]. Detailed biophysical models with stochastic ion channel dynamics can in principle account for every aspect of single-neuron activity; however, due to their complexity, they require high computational power [ 5 , 7 – 9 ]. While systematic fitting of detailed biophysical models is possible [ 10 – 15 ], most of the existing methods assume the knowledge of all the parameters that determine the dynamics of the ion channels included in the model. Overall, a reliable and efficient fitting procedure for detailed biophysical models is not known [ 6 ]. In a second class of spiking neuron models, which we call simplified threshold models, the biophysical mechanisms relevant for neural computation are not explicitly modeled, but are accounted for by phenomenological (i.e., effective) descriptions [ 16 , 17 ]. Despite their simplicity, threshold models are surprisingly good at predicting single-neuron activity [ 6 , 18 – 25 ], at least for the case of single-electrode somatic stimulation (but see [ 26 , 27 ]). Nowadays, simplified threshold models are mainly used in large-scale simulations to study the emergent properties of neural circuits [ 28 , 29 ]. By taking a different perspective, we will demonstrate that the same models can also serve an equally important purpose, namely to characterize the electrical properties of single neurons. In this view, simplified threshold models are interpreted as computational tools to automatically compress the information contained in a voltage recording into a set of unique and meaningful parameters. Summarizing the information of complex voltage recordings can in turn enable systematic comparisons, clustering and identification of cell types. Finally, in patch-clamp experiments aimed at studying detailed aspects of the neuronal dynamics, automated online identification of neurons could allow for on the fly implementation of specific stimulus sets, which are best suited for the neuron under study. After demonstrating that a limited amount of data, and little computing time, are sufficient to fit and validate our previous Generalized Integrate-and-Fire model (GIF, see [ 30 , 31 ]), we introduce an experimental protocol that, combined with automated patch-clamp technology, could make automated high-throughput single-neuron characterization possible. On the experimental side, the protocol relies on in vitro somatic injections of rapidly fluctuating currents that mimic natural inputs received in vivo at the soma of neurons. On the computational side, the protocol is based on Active Electrode Compensation [ 32 , 33 ], GIF model parameter extraction [ 30 , 31 ] and the spike-train similarity measure M d * [ 34 ]. These computational methods are combined and implemented in a Python toolbox (freely available at wiki.epfl.ch/giftoolbox). The validity of our approach is finally demonstrated with two applications: i) in silico recordings obtained by simulating the activity of a multi-compartmental conductance-based model; and ii) in vitro recordings from L5 pyramidal neurons obtained using manual patch clamping. We found that fitting and validating a GIF model takes approximatively five minutes. Considering the time required to automatically establish a patch-clamp seal, the complete characterization of a single neuron can therefore be achieved in around ten minutes. We conclude that GIF models are useful not only for network simulations, but also for rapid and systematic single-neuron characterization.",
"discussion": "Discussion The intrinsic dynamics of individual neurons strongly differ between cell types and brain areas [ 53 ]. This heterogeneity is increasingly considered as a critical feature of the brain and not as a consequence of biological imprecision [ 54 , 55 ]. Taking into account single-neuron variability may be crucial to understand how neural systems support computation. In the near future, automated electrophysiology will likely make available increasingly large amounts of data. Due to their inherent complexity (and their high dimensionality), raw data from patch-clamp recordings are difficult to interpret and cannot be directly clustered to identify electrophysiological types. GIF models are currently employed by computational neuroscientists mainly to investigate the emergent properties of neural networks. Here, we showed that these models also comprise a powerful tool to cast the information provided by voltage recordings into small sets of model parameters that can be easily interpreted and compared. More precisely, we demonstrated that the fitting procedure for GIF models we recently introduced [ 31 , 37 ] (Figs 1 and 2 ) is suitable for high-throughput analysis of intracellular patch-clamp recordings. Using an artificial dataset generated by the model itself, we first established that GIF model parameter extraction and validation can be accomplished in around five minutes given a limited amount of intracellular recordings ( Fig 3 ). Based on these results, we then designed a protocol for the characterization of the electrical activity of single neurons ( Fig 4 ). On the experimental side, the protocol relies on in vitro injections of rapidly fluctuating currents. To compensate for the artifact known to occur while delivering inputs through the recording electrode, we propose the use of Active Electrode Compensation [ 32 , 33 ] ( Fig 7 ). In AEC, estimating the electrode properties is a potentially time-consuming procedure. For this reason, in our protocol, artifacts resulting from the voltage drop across the patch-clamp electrode are removed only after the complete acquisition of the dataset used for parameter extraction ( Fig 4 ). We tested the protocol for high-throughput single-neuron characterization using both in silico data (Figs 5 and 6 ) as well as in vitro recordings obtained with conventional (i.e., manual) patch-clamping (Figs 8 and 9 ). In both cases the results confirmed our conclusion drawn from the analysis of artificial data generated by the GIF model itself (Fig 3 )X namely that a GIF model with parameters extracted from a training set with size larger than 30 seconds accurately predicts both the subthreshold and the spiking response evoked by a new input. Considering that long current-clamp recordings are generally affected by low-frequency artifacts such as drifts in resting membrane potential, access resistance and average firing rate, it seems unlikely that a training set whose size prohibits rapid characterization would improve accuracy. Intriguingly, we found that the GIF model achieves almost identical performances in predicting in silico and in vitro data (Figs 6 and 9 ), indicating that detailed biophysical models could be used in the future to guide the improvement of simplified spiking models. Analyzing the performance of the GIF model in response to dendritic inputs goes beyond the scope of this study. However, as demonstrated by a recent study [ 56 ], the mathematical framework discussed here is flexible and can in principle be extended to account for dendritic current injections. Considering the time required to automatically select a target neuron and form a gigaohm seal, our results demonstrate that, if combined with emergent technologies for automatic patch-clamping, the mathematical tools discussed in this study could be used to implement a high-throughput pipeline performing single-neuron characterization in around ten minutes. Importantly, all the computations in the protocol can be executed on the fly , while electrophysiological recordings are being performed. Consequently, the model performance in predicting the spiking activity (i.e., M d * ) and the subthreshold voltage dynamics (i.e., ϵ \n V ) could be used for online monitoring and quality control, possibly allowing for automated detection of experimental problems. Online characterization and identification of neurons may also prove useful in more detailed high-throughput characterization of neuronal cell types currently being set up in the context of several large-scale brain initiatives as this would allow for on the fly implementation of cell-specific stimulus sets. Similarly, online identification could be useful in manual patch-clamp experiments whose aim is not to perform high-throughput single-neuron characterization, but is to study other neuronal properties (e.g., the dynamics of specific ion-channels under pharmacology, the effect of neuromodulators on the response properties of neurons, connectivity, short-term and spike-timing dependent plasticity). These experiments generally start with a brief set of current injections (e.g., current-steps) aimed at identifying some basic features of the neuronal dynamics (e.g., passive membrane properties, firing patterns). Given its short duration and its limited requirements in terms of computing power, our protocol could provide an alternative in these situations. To allow for a comparison, both in silico and in vitro recordings were also fitted with a GLM [ 35 , 36 ]. Despite the fact that GLMs are more flexible than GIF models, we found that, in terms of mere spike timing prediction, the two models achieved similar performance (Figs 6 and 9 ). This result can be understood by noting that the nonparametric filter κ \n GLM ( t ) extracted with the GLM fitting procedure is well approximated by the exponential filter κ ( t ) of the GIF model. GLMs are typically considered as statistical models for spike trains and their parameters are only loosely related to biophysical cell properties. The reason for this is that GLM parameter extraction entirely relies on the likelihood maximization of the spiking data. If on one hand this fact constitutes a big advantage in case of (multi-electrode) extracellular recordings [ 22 , 36 ], the standard GLM framework is less appropriate for whole-cell current-clamp data. In contrast to GIF models, GLMs do not explicitly model the membrane potential dynamics, do not exploit all the information available in intracellular recordings and, consequently, are unable to predict the subthreshold activity of single neurons. Moreover, compared to GLMs, we found that parameter extraction for GIF models is faster. A voltage-dependent plasticity rule has recently been proposed [ 57 ] in which the subthreshold dynamics of the membrane potential plays a crucial role in explaining a large variety of experimental results obtained using different induction protocols for long-term potentiation (or depression). Among others, this finding highlights the need of spiking neuron models that accurately capture the subthreshold membrane potential dynamics. The GIF model accounts for spike-dependent adaptation using two distinct filters: a spike-triggered current η ( t ) and a spike-triggered movement of the firing threshold γ ( t ). At first glance, having two spike-dependent processes might seem redundant and unnecessary. However, while the firing threshold only affects spike probability, adaptation currents also alter the dynamics of the subthreshold membrane potential. This explains why the correct distinction between these two forms of adaptation is key to correctly predict the subthreshold response of single neurons. Supporting this claim, a reduced GIF model, in which the two processes mediating spike-frequency adaptation are combined into a single effective filter h ( t ) ( Eq 7 ), has been shown to systematically overestimate the membrane potential [ 31 ]. Since GLM parameter extraction entirely relies on spiking data (see Materials and Methods ), the linear filter κ \n GLM ( t ) also includes the effects of all biophysical processes that affect spike emission without altering the subthreshold membrane potential. In particular, the filter κ \n GLM ( t ), but not the filter κ ( t ) of the GIF model, is expected to capture a potential coupling between subthreshold voltage and firing threshold [ 58 , 59 ]. Possibly explaining the difference we found between κ \n GLM ( t ) and κ ( t ) ( Fig 8F ), both direct [ 60 ] and indirect [ 52 ] experimental evidence has been provided that such a coupling exists in cortical pyramidal neurons. Extending the GIF model to account for a coupling between membrane potential and firing threshold is beyond the scope of this study and will be presented in a separate publication. It is however worth noting that the threshold equation of the GIF model can be easily augmented as follows:\n V T ( t ) = V T * + ∑ t ^ j < t γ ( t γ ( t − t ^ j ) − t ^ j ) + ∫ t ^ last t κ θ ( s ) V ( t - s ) d s , (9) \nwith κ \n θ ( t ) being an arbitrarily shaped filter that, with a straightforward extension of the maximum likelihood method used in Step 3 (see Fig 2 , Step 3 ), could be extracted from intracellular recordings. In contrast to the GIF model, popular point-neuron models like the adaptive exponential integrate-and-fire (ADEX, [ 61 ]) or the adaptive quadratic integrate-and-fire (AQIF, [ 62 ]) feature a subthreshold adaptation current w ( t ) governed by the following differential equation\n τ w w ˙ = - w + a ( V - E L ) . (10) \n Extending the GIF model with Eq 10 would relax the assumption of having a single exponential membrane filter κ ( t ) and, depending on the parameter choice, the subthreshold dynamics of the resulting model could account for two-timescale decay or resonance [ 30 ]. In ADEX and the AQIF model, this current has been shown to play an important role in explaining the variety of firing patterns emitted by single neurons in response to a step of current [ 63 , 64 ]. In the GIF model, the lack of subthreshold adaptation is, at least partially, compensated by the fact that the spike-triggered current is not assumed to be exponential, but can have an arbitrary shape. For example, the GIF model can capture the resonate-and-fire behavior by means of a biphasic spike-triggered current. Such a current hyperpolarizes the membrane during the first milliseconds and then rapidly becomes positive, thereby favoring the emission of spikes with a particular interspike interval [ 37 ]. Our results suggest that, while increasing the complexity, extending a GIF model with a subthreshold current w ( t ) does not significantly improve the model’s performance in predicting the activity of the three main neuronal types of the mouse barrel cortex [ 30 ]. However, this might not hold true for neurons in other brain regions or in the case of more sophisticated stimulation paradigms. Performing parameter extraction with a GIF model extended with Eq 10 is possible. Once the timescale τ \n w is known, performing a least-square regression similar to Eq 17 is indeed sufficient to recover all the other parameters. Extended GIF model parameter extraction can therefore be performed by iterating on τ \n w and looking for the timescale that minimizes the sum of squared errors on the voltage derivative. Since line-search (i.e., brute-force) algorithms can be efficiently executed using parallel computing, extending a GIF model with a subthreshold adaptation current does not necessarily imply a dramatic increase of the CPU time required for parameter extraction. In the field of computational neuroscience, the last years have been characterized by the announcements of several large-scale projects aimed to build realistic models of the electrical activity of entire brains [ 8 , 9 , 65 – 67 ]. To achieve this ambitious goal, it is of crucial importance to characterize and model the diversity amongst the brain’s fundamental building blocks: the single neurons. Here, we demonstrate that, if combined with automatic patch-clamp recordings, a fitting technique for GIF models, which we recently introduced [ 31 , 37 ], can be used to build a pipeline for high-throughput single-neuron characterization and modeling."
} | 4,759 |
24122357 | null | s2 | 7,828 | {
"abstract": "We engineered a fatty acid overproducing Escherichia coli strain through overexpressing tesA (“pull”) and fadR (“push”) and knocking out fadE (“block”). This “pull-push-block” strategy yielded 0.17 g of fatty acids (C12–C18) per gram of glucose (equivalent to 48% of the maximum theoretical yield) in batch cultures during the exponential growth phase under aerobic conditions. Metabolic fluxes were determined for the engineered E. coli and its control strain using tracer ([1,2-13C]glucose) experiments and 13C-metabolic flux analysis. Cofactor (NADPH) and energy (ATP) balances were also investigated for both strains based on estimated fluxes. Compared to the control strain, fatty acid overproduction led to significant metabolic responses in the central metabolism: (1) Acetic acid secretion flux decreased 10-fold; (2) Pentose phosphate pathway and Entner–Doudoroff pathway fluxes increased 1.5- and 2.0-fold, respectively; (3) Biomass synthesis flux was reduced 1.9-fold; (4) Anaplerotic phosphoenolpyruvate carboxylation flux decreased 1.7-fold; (5) Transhydrogenation flux converting NADH to NADPH increased by 1.7-fold. Real-time quantitative RT-PCR analysis revealed the engineered strain increased the transcription levels of pntA (encoding the membrane-bound transhydrogenase) by 2.1-fold and udhA (encoding the soluble transhydrogenase) by 1.4-fold, which is in agreement with the increased transhydrogenation flux. Cofactor and energy balances analyses showed that the fatty acid overproducing E. coli consumed significantly higher cellular maintenance energy than the control strain. We discussed the strategies to future strain development and process improvements for fatty acid production in E. coli."
} | 430 |
24013352 | PMC3768014 | pmc | 7,831 | {
"abstract": "Using DNA as programmable, sequence specific ‘glues’, shape-controlled hydrogel units are self-assembled into prescribed structures. Here we report that aggregates are produced using hydrogel cubes with edge length ranging from 30 micrometers to 1 millimeter, demonstrating assembly across scales. In a simple one-pot agitation reaction, 25 dimers are constructed in parallel from 50 distinct hydrogel cube species, demonstrating highly multiplexed assembly. Using hydrogel cuboids displaying face-specific DNA glues, diverse structures are achieved in aqueous and in interfacial agitation systems. These include dimers, extended chains, and open network structures in an aqueous system; and dimers, chains of fixed length, T-junctions, and square shapes in the interfacial system, demonstrating the versatility of the assembly system.",
"introduction": "Introduction Self-assembly is the process by which small components self-organize into larger structures. Initially developed as a concept for engineering molecular complexes 1 , self-assembly has been applied to construct structures across scales using monomer units ranging from nano-scale to macro-scale dimensions 2 . Diverse techniques have been developed for mesoscale (micrometer- to millimeter-scale) self-assembly using magnetic force 3 , hydrophile-lipophile balance 4 , 5 , capillary interaction 6 and synthetic chemical binding 7 to control the assembled architecture. Increasing the complexity of mesoscale self-assembly faces a crucial challenge, namely the difficulty of engineering a large set of orthogonal specific binding interactions between the monomer units. This challenge can be potentially addressed by using DNA, biology's information carrier, as programmable “glue” to direct the assembly of mesoscale units. DNA contains four different nucleotide bases, each of which forms a base pair with another complementary base according to a set of canonical rules: adenine with thymine, and guanine with cytosine. By simply arranging the sequence of these four nucleotide bases in different DNA strands, a combinatorially large set of binding interactions can be designed as specific hybridizations between complementary DNA strands. DNA hybridization based self-assembly principles have been utilized successfully by the field of DNA nanotechnology to generate diverse complex synthetic DNA/RNA nano-structures 8 - 11 with arbitrarily prescribed geometry 12 - 29 and dynamic functions 30 - 41 . Furthermore, DNA strands can be made into hydrogels through covalent 42 - 44 or non-covalent interactions 45 . DNA has also been used as templates or glues to mediate the self-assembly of fluorophores 46 , proteins 47 , inorganic nanoparticles 48 - 51 , carbon nanotube 52 , lipid vesicles 53 , and even living cells 54 . Recently, it was reported that short single-stranded DNA probes attached to a glass surface can successfully catch 100 μm size hydrogel microspheres decorated with sequences complementary to the probes 55 . Building on these previous successes, we address the next challenge here: to fully utilize the versatile programmability of DNA to direct the self-assembly of mesoscale objects into complex higher order structures with precisely prescribed architecture and geometry. As part of our work to increase the complexity of the architectural and geometrical control of DNA direct mesoscale assembly, here, we report on combining DNA directed assembly principle with microfabrication technology to assemble mesoscale objects using shaped controlled hydrogel units. The central conceptual innovation here is the decoration of DNA glues onto the prescribed surfaces of a non-spherical hydrogel object to produce an asymmetric glue pattern. These new assembling units, by combining the molecular programmability of DNA glue and the shape controllability of microfabrication, will provide a powerful platform to achieve programmable assembly of complex mesoscale structures. To implement this strategy, a crucial technical innovation was necessary: we invent a novel strategy to use in situ rolling circle amplification to produce and attach “giant” DNA glues to the surface of hydrogel cubes. Based on this technical innovation, we demonstrate that giant DNA glue strands, but not short DNA primers, induce the assembly of hydrogel gel cubes with an edge length across scales (30 μm to 1 mm), and that they result in the self-assembly of cube dimers in a highly multiplexed fashion (25 orthogonal dimer pairs from 50 distinct cube species in one pot mixing). We then develop a method to engineer hydrogel cuboids that display giant DNA only on designated faces. Using this technology, we demonstrate the assembly, in aqueous and in interfacial systems, of diverse structures: linear chains with extended or fixed length, open networks, T-junctions, and 2×2 square structures. Thus we establish DNA directed assembly of shape-controlled mesoscale units as a promising route to produce complex structures with sophisticated geometrical and architectural control.",
"discussion": "Discussion Although short DNA primer has been reported to assemble nano-particles and microscale hydrogels, we demonstrate, for the first time, the DNA-directed self-assembly of shape-controlled hydrogel modules to build complex structures in a programmable fashion. Acting like sequence specific glue and tethered onto a microgel surface, giant single-stranded DNA exhibits a significant capability for binding objects across scales, with sizes ranging from 30 micrometers to a millimeter. Additionally, giant DNA glues offer significant diversity over current mesoscale self-assembly systems: fifty DNA sequences were designed to generate 25 orthogonal pairs of specific interactions. This is the largest of number of orthogonal binding interactions that have been used simultaneously in the same reaction system for mesoscale assembly. The designable DNA glues thus provide much richer options for programming mesoscale self-assembly. For self-assembling complex structures, the unit fabrication is crucial. As cube units uniformly carrying DNA glues assembled only into aggregates lack of architecture control, we developed a precisely controlled fabrication technique by which specific giant DNA glue is decorated on a prescribed face of a hydrogel cuboid. By changing the position of DNA glues, various structures including dimers, linear chains, and open networks were assembled. In an interfacial system, we further demonstrated that hydrogel cuboids can be fabricated with 4 different DNA glues on 4 designated faces, and by simply changing the surface DNA decoration pattern, we assembled discrete structures including dimers, T-junctions, linear chains with fixed length, and squares. We have successfully introduced programmability into self-assembling mesoscale structure. We believe that there is still room for improvement in our self-assembly system by increasing the resolution of module fabrication. As demonstrated in this study, hydrogel cuboids with a larger width ratio between the body part and the DNA pad align better with each other than cuboids with a smaller width ratio. It is also possible to improve the assembly by tuning the strength of DNA glue through controlling the DNA density and length on the gel surface. Additionally, better face-to-face alignment could be achieved by improving the gel fabrication method, for example, by minimizing the aspect ratio between the DNA carrying pads and the gel body (note that the assembly in Fig. 5 demonstrated improved face-to-face alignment over Fig. 4 ). Furthermore, advances in the mixing regimes that better regulate the hydrodynamic forces involved in the assembly process may be used to further enhance the assembly process. By coupling novel in situ DNA amplification methods and microfabrication techniques, we successfully introduced the diversity and specificity of biomolecular interaction to mesoscale assembly. DNA-directed self-assembly of shape-controlled hydrogel modules proved to be highly programmable and controllable, and will open new doors to address the challenge of building complex self-assembled 3D structures for diverse applications in materials science and especially in biomaterials. One particularly promising direction is to develop tissue engineering application, as recreating the highly defined complicate structure of tissue is a pressing challenge. For example, by encapsulating specific cells inside the hydrogel cubes, the self-assembled structures could be used to build the basic architectures of native tissues."
} | 2,142 |
23651460 | PMC3668229 | pmc | 7,832 | {
"abstract": "Background p -Hydroxycinnamic acid (pHCA) is an aromatic compound that serves as a starting material for the production of many commercially valuable chemicals, such as fragrances and pharmaceuticals, and is also used in the synthesis of thermostable polymers. However, chemical synthesis of pHCA is both costly and harmful to the environment. Although pHCA production using microbes has been widely studied, there remains a need for more cost-effective methods, such as the use of biomass as a carbon source. In this study, we produced pHCA using tyrosine ammonia lyase-expressing Streptomyces lividans . In order to improve pHCA productivity from cellulose, we constructed a tyrosine ammonia lyase- and endoglucanase (EG)-expressing S. lividans transformant and used it to produce pHCA from cellulose. Results A Streptomyces lividans transformant was constructed to express tyrosine ammonia lyase derived from Rhodobacter sphaeroides (RsTAL). The transformant produced 786 or 736 mg/L of pHCA after 7 days of cultivation in medium containing 1% glucose or cellobiose as the carbon source, respectively. To enhance pHCA production from phosphoric acid swollen cellulose (PASC), we introduced the gene encoding EG into RsTAL-expressing S. lividans . After 7 days of cultivation, this transformant produced 753, 743, or 500 mg/L of pHCA from 1% glucose, cellobiose, or PASC, respectively. Conclusions RsTAL-expressing S. lividans can produce pHCA from glucose and cellobiose. Similarly, RsTAL- and EG-expressing S. lividans can produce pHCA from glucose and cellobiose with excess EG activity remaining in the supernatant. This transformant demonstrated improved pHCA production from cellulose. Further enhancements in the cellulose degradation capability of the transformant will be necessary in order to achieve further improvements in pHCA production from cellulose.",
"conclusion": "Conclusions We demonstrated the production of pHCA from glucose and cellobiose as carbon sources using RsTAL-expressing S. lividans . The amount of pHCA produced in batch culture using our system was higher than that reported in previous studies [ 9 , 11 ]. In order to improve the production of pHCA from cellulose, we constructed a strain of S. lividans that expresses both EG and RsTAL. This transformant could secrete EG into the supernatant and produce pHCA directly from cellulose. Our results demonstrate that S. lividans can be used as a host to produce aromatic building-blocks from cellulose.",
"discussion": "Results and discussion Production of pHCA by S. lividans /pURsTAL Streptomyces lividans /pURsTAL and S. lividans /pU were cultured separately in TSB medium for 4 days, after which the culture supernatants were analyzed by HPLC. Figure 1 A shows representative chromatograms of a standard sample of pHCA solution (Lane 1), S. lividans /pURsTAL culture supernatant (Lane 2), and S. lividans /pU culture supernatant (Lane 3). In the analysis of the standard solution, the pHCA peak eluted at about 10 min (Lane 1), and a peak eluting with a similar retention time was also observed in the analysis of the S. lividans /pURsTAL culture supernatant (Lane 2). In contrast, no pHCA peak was observed on the chromatogram of the S. lividans /pU culture supernatant (Lane 3). Figure 1 Confirmation of pHCA produced by RsTAL expressing S. lividans. ( A ) HPLC analysis of pHCA. Lane 1: Standard sample of pHCA in acetonitrile:phosphate buffer (50 mM, pH 2.0) (20:80). Lane 2: S. lividans /pURsTAL culture supernatant. Lane 3: S. lividans /pU culture supernatant. ( B ) UV spectra of pHCA. Standard sample of pHCA in acetonitrile:phosphate buffer (50 mM, pH 2.0) (20:80) (solid line). The spectrum of the putative pHCA fraction isolated from the S. lividans /pURsTAL culture supernatant by HPLC is shown as a dotted line. A UV analysis of the putative pHCA HPLC fraction isolated from the culture supernatant of S. lividans /pURsTAL revealed the presence of three prominent absorption peaks in the 200–350 nm region that were consistent with the peaks produced upon analysis of standard pHCA (Figure 1 B). In addition, we carried out pHCA production using S. lividans /pURsTAL using the modified TSB medium with 1% glucose and the additional L-tyrosine, which is the precursor of pHCA. The addition of L-tyrosine into the initial culture medium increased the peak areas of pHCA (data not shown). These results confirmed that introduction of the gene encoding RsTAL into S. lividans enables the production of pHCA. In order to demonstrate effective pHCA production, S. lividans /pURsTAL was cultured in modified TSB medium with 1% glucose or cellobiose. Figure 2 A shows the time courses of pHCA production with each carbon source. The maximum concentration of pHCA produced from glucose and cellobiose was 786 and 736 mg/L, respectively. The level of pHCA produced by S. lividans was higher than that reported for batch-cultured E. coli (103 mg/L) and P. putida S12 (141 mg/L) [ 9 , 11 ]. In the present study, we carried out pHCA production using various concentrations of the additional carbon sources. As a result, 1% carbon sources are suitable for pHCA production using S. lividans transformants. In the case of using 1-5% carbon sources, the amount of produced pHCA is almost the same (data not shown). According to these results, we considered that 1% additional carbon sources are suitable for pHCA production using S. lividans transformant. In the case of without the additional carbon sources, the amount of produced pHCA was not enough, compared to using the medium with additional glucose or cellobiose. Figure 2 pHCA production and the cell growth of RsTAL expressing S. lividans . ( A ) Time-course of pHCA production in culture: S. lividans /pURsTAL cultured in modified TSB medium containing 1% glucose (closed circles) or 1% cellobiose (closed squares). ( B ) Change in the dry cell weight over time of S. lividans /pURsTAL cultured in modified TSB medium containing 1% glucose (open circles) or 1% cellobiose (open squares). Each data point shows the average of three independent experiments, and error bars represent the standard deviation. We also examined pHCA production by the control, S. lividans /pU. As expected, this strain did not produce any pHCA (data not shown). Figure 2 B shows the change in the dry cell weight over time of S. lividans /pURsTAL cultured in medium containing 1% glucose or cellobiose. The growth of S. lividans /pURsTAL in medium containing 1% cellobiose as the carbon source was almost the same as in cultures in which 1% glucose served as the carbon source, indicating that S. lividans can assimilate cellobiose and produce pHCA as well as utilize glucose, in agreement with our previous report [ 17 ]. Here, we estimated the amount of residual glucose or cellobiose using HPLC. In the case of each transformant, glucose or cellobiose added to the medium was consumed within 2 days (data not shown). These results strongly suggested that S. lividans /pURsTAL could assimilate cellobiose as well as glucose. Construction of EG- and RsTAL-expressing S. lividans Most microorganisms have difficulty degrading cellulose due to its rigid structure. In order to develop effective microbial bioprocesses utilizing cellulose as a carbon source, it is therefore necessary to introduce multiple genes encoding cellulose degradation enzymes into the genome of the organism of interest. A recent report described the introduction of four types of cellulases into Saccharomyces cerevisiae and subsequent bioethanol production from cellulose [ 22 ]. Bokinsky et al. reported the use of E. coli expressing three types of cellulases for the production of advanced biofuels from cellulose [ 23 ]. Streptomyces lividans constitutively expresses a highly active form of BGL that enables the organism to assimilate cello-oligosaccharide [ 17 ]. However, the ability of S. lividans to degrade cellulose must be improved in order to achieve effective bioconversion of cellulose to useful compounds. In this study, we demonstrated production of pHCA using cellobiose as the carbon source. To achieve effective pHCA production from cellulose, we then introduced the gene encoding EG into pHCA-producing S. lividans in order to facilitate the degradation of cellulose to cello-oligosaccharide. For this purpose we chose Tfu0901, which is a highly active EG derived from T. fusca YX, and introduced the gene for this enzyme into RsTAL-expressing S. lividans . After the Tfu0901 gene was introduced into wild-type S. lividans using the integration type vector pTYM18 [ 24 ], the resulting S. lividans mutant was able to produce pHCA after introduction of the multicopy type vector pUC702 that carries the gene encoding RsTAL. The resulting EG- and RsTAL-expressing S. lividans strain was designated S. lividans /pT09pURsTAL. Production of pHCA and expression of EG by S. lividans /pT09pURsTAL was examined in modified TSB medium containing 1% glucose. Figure 3 A shows the time courses of pHCA production by S. lividans /pT09pURsTAL in media containing various carbon sources. Approximately 753 mg/L of pHCA was produced in medium containing 1% glucose. The amount of pHCA produced by S. lividans /pT09pURsTAL was equal to that produced by S. lividans /pURsTAL, suggesting that EG expression did not affect pHCA production. Figure 3 B details the change in supernatant EG activity over time in cultures of S. lividans /pT09pURsTAL grown in the presence of various carbon sources. The maximum EG activity of S. lividans /pT09pURsTAL cultured with 1% glucose was approximately 440 U/L (Figure 3 B), indicating that S. lividans /pT09pURsTAL can produce pHCA with EG expressing. The change over time in the dry cell weight of S. lividans /pT09pURsTAL cultured with 1% glucose is shown in Figure 3 C. The growth of S. lividans /pT09pURsTAL in medium containing 1% glucose was similar to that of S. lividans /pURsTAL (Figures 2 B and 3 C), confirming that S. lividans /pT09pURsTAL expresses both EG and RsTAL. Here, as shown in Figures 2 B and 3 C, the decrease of dry cell weight of each S. lividans transformant in the stationary phase of the cell growth was confirmed. These results are corresponding to our previous report concerning low-molecular compounds production using Streptomyces [ 17 , 21 ]. Figure 3 pHCA production, EG activity and the cell growth of RsTAL and EG expressing S. lividans. ( A ) Time-course of pHCA production in culture: S. lividans /pT09pURsTAL cultured in modified TSB medium containing 1% glucose (closed circles), 1% cellobiose (closed squares), or 1% phosphoric acid swollen cellulose (PASC) (closed diamonds); S. lividans /pURsTAL cultured in modified TSB medium containing 1% PASC (closed triangles). ( B ) Change in culture supernatant endoglucanase activity over time: S. lividans /pT09pURsTAL cultured in modified TSB medium containing 1% glucose (closed circles), 1% cellobiose (closed squares), or 1% PASC (closed diamonds); S. lividans /pURsTAL cultured in modified TSB medium containing 1% PASC (closed triangles). ( C ) Change in the dry cell weight over time of S. lividans /pT09pURsTAL cultured in modified TSB medium containing 1% glucose (open circles) or 1% cellobiose (open squares). Each data point shows the average of three independent experiments, and error bars represent the standard deviation. Production of pHCA directly from cellulose by EG- and RsTAL-expressing S. lividans Using S. lividans /pT09pURsTAL, we produced pHCA from the cellulosic substrate phosphoric acid swollen cellulose (PASC). Figure 3 A shows the time course of pHCA production in modified TSB medium containing 1% PASC as the carbon source. The maximal level of pHCA production reached by S. lividans /pT09pURsTAL was 500 mg/L after 7 days of cultivation, whereas the control strain, S. lividans /pURsTAL, produced 310 mg/L after 7 days of cultivation. The change in supernatant EG activity over time with 1% PASC serving as the carbon source is shown in Figure 3 B. The maximal level of EG activity in the supernatant of S. lividans /pT09pURsTAL was approximately 210 U/L, whereas that of S. lividans /pURsTAL was only 10 U/L. Although S. lividans /pURsTAL produced some quantity of pHCA from cellulose, S. lividans is known to secrete several kinds of cellulases, including EG, and thus this result is consistent with those of our previous report [ 17 ]. In the present study, production of pHCA from cellulose was improved by using the S. lividans /pT09pURsTAL strain, which produced 500 mg/L of pHCA, a level that was 1.6-fold higher than the level of pHCA produced by S. lividans /pURsTAL. Production of pHCA was carried out using S. lividans /pT09pURsTAL cultured with 1% cellobiose as the carbon source. Figure 3 A shows the production of pHCA over time in modified TSB medium containing 1% cellobiose as the carbon source. The maximal level of pHCA produced by S. lividans /pT09pURsTAL from 1% cellobiose was 743 mg/L after 7 days of cultivation. This level was slightly higher than that produced from 1% PASC (Figure 3 A). These results indicate that S. lividans /pT09pURsTAL did not completely assimilate the 1% PASC. The amount of pHCA produced by S. lividans /pT09pURsTAL from 1% cellobiose was equal to that produced from 1% glucose, and the introduction of the gene encoding EG did not affect the ability of the organism to assimilate cellobiose. Hence, to produce pHCA from cellulose more effectively, the efficiency of the cellulose to cello-oligosaccharide degradation reaction should be improved. Due to synergism between EG and CBH, cellulose is effectively degraded to cello-oligosaccharide, which is a suitable carbon source for S. lividans [ 22 , 23 ]. To enhance pHCA productivity from cellulose, in this study we utilized an EG derived from T. fusca YX, Tfu0901, to construct EG- and RsTAL-expressing S. lividans /pT09pURsTAL, which was capable of producing pHCA from PASC (Figure 3 A) due to the increased availability of cello-oligosaccharides relative to the control strain. However, the amount of pHCA produced from 1% PASC was lower than that produced from 1% glucose or cellobiose (Figure 3 A), indicating that PASC was not completely degraded to available sugars. One promising means of achieving higher pHCA productivity from PASC would be to introduce the gene encoding CBH into S. lividans /pT09pURsTAL. Exocellobiohydrolase can act on the reducing or nonreducing ends of cellulose chains generated by EG, leading to pHCA production comparable to that obtained when either glucose or cellobiose is employed as the carbon source. Currently, our group is screening for active CBH and developing a three-gene expression system for S. lividans ."
} | 3,703 |
29064480 | PMC5739008 | pmc | 7,838 | {
"abstract": "Oxidation of methanethiol (MT) is a significant step in the sulfur cycle. MT is an intermediate of metabolism of globally significant organosulfur compounds including dimethylsulfoniopropionate (DMSP) and dimethylsulfide (DMS), which have key roles in marine carbon and sulfur cycling. In aerobic bacteria, MT is degraded by a MT oxidase (MTO). The enzymatic and genetic basis of MT oxidation have remained poorly characterized. Here, we identify for the first time the MTO enzyme and its encoding gene ( mtoX ) in the DMS-degrading bacterium Hyphomicrobium sp. VS. We show that MTO is a homotetrameric metalloenzyme that requires Cu for enzyme activity. MTO is predicted to be a soluble periplasmic enzyme and a member of a distinct clade of the Selenium-binding protein (SBP56) family for which no function has been reported. Genes orthologous to mtoX exist in many bacteria able to degrade DMS, other one-carbon compounds or DMSP, notably in the marine model organism Ruegeria pomeroyi DSS-3, a member of the Rhodobacteraceae family that is abundant in marine environments. Marker exchange mutagenesis of mtoX disrupted the ability of R. pomeroyi to metabolize MT confirming its function in this DMSP-degrading bacterium. In R. pomeroyi , transcription of mtoX was enhanced by DMSP, methylmercaptopropionate and MT. Rates of MT degradation increased after pre-incubation of the wild-type strain with MT. The detection of mtoX orthologs in diverse bacteria, environmental samples and its abundance in a range of metagenomic data sets point to this enzyme being widely distributed in the environment and having a key role in global sulfur cycling.",
"introduction": "Introduction Methanethiol (CH 3 SH; methylmercaptan, MT) is a foul-smelling gas with a low odor threshold. As a malodorous compound that can be detected by the human nose at very low concentration (odor threshold 1–2 p.p.b., ( Devos et al. , 1990 )), it has a significant role in causing off-flavors in foods and beverages and it is one of the main volatile sulfur compounds causing halitosis in humans ( Awano et al. , 2004 ; Tangerman and Winkel, 2007 ). The production and degradation of MT are major steps in the biogeochemical cycle of sulfur ( Figure 1 ). Sources of MT include the methylation of sulfide in anoxic habitats, demethiolation of sulfhydryl groups and degradation of sulfur-containing amino acids ( Lomans et al. , 2001 , 2002 ; Bentley and Chasteen, 2004 ). MT is produced in the marine environment as an intermediate of dimethylsulfoniopropionate (DMSP) degradation by the demethylation pathway. In this pathway, initial demethylation of DMSP to methylmercaptopropionic acid (MMPA) is carried out by the DMSP-dependent demethylase (DmdA) ( Howard et al. , 2006 ). Subsequent degradation of MMPA occurs via MMPA-CoA to methylthioacryloyl-CoA and then to acetaldehyde and MT by the enzymes DmdB, DmdC and DmdD, respectively ( Reisch et al. , 2011b ). MT is also produced as an intermediate of dimethylsulfide (DMS) degradation ( Lomans et al. , 1999a , 2002 ; Bentley and Chasteen, 2004 ; Schäfer et al. , 2010 ). Only few measurements of MT in the environment have been reported. Analysis of volatile sulfur compounds in freshwater ditches demonstrated that MT was the dominant volatile organic sulfur compound reaching concentrations of 3–76 n m in sediments and 1–8 n m in surface freshwater ( Lomans et al. , 1997 ). Measurements of MT concentrations in the surface ocean water are scarce. Studies reporting MT measurements in seawater suggest a typical range of ~0.02–2 n m ( Ulshöfer et al. , 1996 ; Kettle et al. , 2001 ; Xu et al. , 2001 ). Microbial uptake and degradation of MT are important sinks for MT. Despite low MT concentrations in seawater, radiotracer experiments showed that trace levels of MT (0.5 n m ) were rapidly taken up and incorporated into biomass by marine bacterioplankton ( Kiene et al. , 1999 ). Besides this assimilation, MT degradation through its utilization as a carbon and energy source in methanogenic archaea, sulfate-reducing bacteria, and aerobic bacteria ( Lomans et al. , 1999b , 2001 , 2002 ; Schäfer et al. , 2010 ) and its methylation to DMS by the recently described methyltransferase MddA (MddA: methanethiol-dependent DMS) ( Carrión et al. , 2015 ) contribute to biological MT removal. The molecular basis of MT degradation remains poorly understood. In aerobic sulfur-oxidizing and methylotrophic bacteria including strains of Thiobacillus ( Gould and Kanagawa, 1992 ; Lee et al. , 2002 ), Rhodococcus ( Kim et al. , 2000 ) and Hyphomicrobium ( Suylen et al. , 1987 ), MT is degraded by a MT oxidase (MTO) to formaldehyde, hydrogen sulfide and hydrogen peroxide; however, inconsistent data have emerged from these studies. Estimated molecular weights of MTOs characterized previously have ranged from ~29–61 kDa. The MTO from Hyphomicrobium sp. EG was reported to be a monomer of 40–50 kDa that was insensitive to metal-chelating agents ( Suylen et al. , 1987 ). In Thiobacillus thioparus ( Gould and Kanagawa, 1992 ), MTO also appeared to be a monomer with a molecular weight of ~40 kDa; however, a later study of MTO in T. thioparus reported a different molecular weight for MTO of 61 kDa ( Lee et al. , 2002 ). MTO from Rhodococcus rhodochrous was reported to have a molecular weight of 64.5 kDa ( Kim et al. , 2000 ). The genetic basis of MT degradation has not been identified, constituting a gap in fundamental knowledge of a key step in the global sulfur cycle. Here, we report new insights into the biochemistry, genetics and environmental distribution of methanethiol oxidases in bacteria. We purified and characterized MTO from Hyphomicrobium sp. VS a DMS-degrading methylotrophic bacterium that was isolated from activated sewage sludge and which has MTO activity during growth on DMS as a sole carbon and energy source ( Pol et al. , 1994 ). We identified the gene encoding MTO, mtoX , in Hyphomicrobium sp. VS and detected orthologous mtoX genes in a wide range of bacteria including methylotrophic, sulfur-oxidizing and DMSP-degrading bacteria. We then genetically analyzed its function and transcriptional regulation in a model isolate of the Rhodobacteraceae family, Ruegeria pomeroyi DSS-3, which produces MT during degradation of DMSP by the demethylation pathway ( Reisch et al. , 2011a ). The development of mtoX -specific PCR primers allowed testing environmental samples for the presence of mtoX -containing populations. This analysis suggested that the genetic potential of MT degradation is present in a wider spectrum of phylogenetic lineages than previously realized based on bacterial cultures. This was also reflected by the presence of mtoX genes from uncultivated organisms in diverse habitats based on screening of metagenomic data sets, which suggests that MTO is widely distributed in the biosphere.",
"discussion": "Discussion New insights into biochemical, genetic and environmental aspects of bacterial methanethiol oxidation presented here address a major knowledge gap in the biogeochemical sulfur cycle and the fundamental understanding of MT degradation by bacteria. Data presented here indicate that MTO is a periplasmic enzyme that is present in a wide range of bacteria, not limited to those known to produce MT as a metabolic intermediate during DMS and DMSP degradation, such as Hyphomicrobium VS, Thiobacillus sp. and R. pomeroyi DSS-3. The mtoX gene was also found in diverse cultivated bacteria that had not previously been recognized for their potential to degrade methanethiol. Homologous genes are also present in archaea and eukarya (including humans). In addition, the overall diversity of mtoX in environmental samples suggests that the potential for MT oxidation is also present in diverse uncultivated microorganisms and that MTO is a widely distributed enzyme in different terrestrial and marine environments, many of which have demonstrated potential for degradation of methylated sulfur compounds. MTO requires copper for its catalytic activity, and in R. pomeroyi , the gene encoding MTO is induced by MT. The enzyme from Hyphomicrobium sp. VS has a very high affinity for MT, with a K m (0.2–0.3 μ m ) at least 10-fold lower than those previously reported, which may explain the low MT concentrations found in the environment. Distinct molecular weights for MTOs from Hyphomicrobium , Thiobacillus and Rhodococcus strains have been reported previously. On the basis of high sequence homology of mtoX genes found in several Hyphomicrobium and Thiobacillus strains and the fact that previously purified MTOs from Hyphomicrobium sp. EG ( Suylen et al. , 1987 ) and T. thioparus ( Gould and Kanagawa, 1992 ) had similar molecular weights to the MTO of Hyphomicrobium sp. VS suggests that the previously purified MTOs are similar enzymes. Although previous studies reported MTO as a monomeric enzyme in Hyphomicrobium sp. EG and T. thioparus Tk-m ( Suylen et al. , 1987 ; Gould and Kanagawa, 1992 ), rather than a homotetramer as in this study, these differences may be due to sensitivity of the MTO’s oligomeric state to pH. At pH 8.2, we found tetrameric MTO, but when we carried out analytical gel filtration at pH 7.2, as used by Suylen et al. (1987) , MTO was detected in monomeric and tetrameric state (result not shown). Other observed differences between these MTOs may be due to different analytical approaches that were employed. For instance, a role of metals in MTO activity was previously ruled out based on chelation experiments, but these can fail to deplete the metals from the enzyme depending on variations in incubation conditions. The presence in and role of Cu for the functioning of the enzyme from Hyphomicrobium sp. VS is supported by ICP mass spectrometry analysis, changes in EPR spectra recorded with MTO in resting, reduced and oxidized state, and by chelation experiments showing a reduced activity of the enzyme. The presence of genes encoding putative Cu chaperones (SCO1/SenC) in close proximity to mtoX homologs in many bacterial genomes provides further circumstantial evidence for a role of copper in MT oxidation and provides a focus for future genetic and biochemical studies. Besides the presence of a mauG homolog, involved in maturation of a protein-derived TTQ co-factor in methylamine dehydrogenase, we found supporting evidence that the MTO also contains a TTQ co-factor. The PDB database contains the structure of the heterologously expressed SBP56 protein of Sulfolobus tokodaii (PDB entry: 2ECE). Analysis of the structure of this non-matured protein (no copper, no TTQ) made it possible to identify the putative ligands involved in copper binding (histidines) and TTQ synthesis (tryptophans) in the Sulfolobus homolog ( Supplementary Figure S11 ). Alignments of the tryptophan and histidine residues identified showed strict conservation over the three domains of life. EPR and EXAFS analyses suggest that Cu in MTO of Hyphomicrobium sp. VS is coordinated by four nitrogen atoms, which would fit with the strictly conserved histidine residues which in Hyphomicrobium sp. VS-MTO are His89, His90, His140, His412 ( Supplementary Figures S8 and S12 ). The structural information and the presence of the SCO1/senC and mauG -like genes support the presence of a TTQ co-factor and two copper atoms per monomer; further, if we assume 4 Ca and 2 Cu per monomer, the calculated mass exactly fits the Electrospray-ionisation mass spectroscopy analysis: 46 193 vs 46 186 Da. The arrangement of the genes mtoX , SCO1/senC and mauG encoding MTO, a copper chaperone, and homolog of the enzyme known to be involved in maturation of a protein-derived TTQ co-factor in methylamine dehydrogenase was highly conserved in a wide range of bacteria ( Figure 2 and Supplementary Table S7 ). The role of MTO in metabolism of MT and DMSP as well as its transcriptional regulation were demonstrated in R. pomeroyi showing that this enzyme has an important role in metabolism of DMSP. Transcriptional fusions of the IclR type regulator upstream also demonstrated that MT as well as DMSP and MMPA (which are degraded to MT) induced MTO transcription. Interestingly, despite the presence of a functional MTO, it has long been known that R. pomeroyi DSS-3 liberates MT when grown in the presence of DMSP, this being one of the products of the DMSP demethylation pathway ( Reisch et al. , 2011b ). Thus, under these circumstances, the MTO does not have sufficient activity to oxidize all the DMSP-dependent MT that is formed. However, we noted (unpublished) that the mtoX − mutant R. pomeroyi DSS-3 released more MT (~1.5-fold) when grown in the presence of DMSP than did the wild type. The identification of the gene encoding MTO in bacteria has allowed assessing the distribution of the enzyme in the environment and identified its evolutionary relationship to the selenium-binding protein family (SBP56), a protein family that has as yet an unresolved function. Metal analysis by ICP mass spectrometry did not show the presence of selenium in MTO. SBP56 is a highly conserved intracellular protein ( Bansal et al. , 1989 ). Previous reports stated that it is involved in the transport of selenium compounds, regulation of oxidation/reduction and late stages of intra-Golgi protein transport, but its exact role has remained unclear ( Jamba et al. , 1997 ; Porat et al. , 2000 ; Ishida et al. , 2002 ). Homologs of SBP56 were found in human, mouse, fish, horse, birds, abalone and plants such as Arabidopsis thaliana and maize in addition to bacteria and archaea ( Jamba et al. , 1997 ; Flemetakis et al. , 2002 ; Self et al. , 2004 ; Song et al. , 2006 ). The human SBP56 homolog has been shown to be a methanethiol oxidase (( Pol et al. ), Nat Genet , in revision). To what extent the other SBP56 have similar function to MTO needs to be addressed, but a possible relationship of SBP56 with C1 metabolism was previously pointed out based on the presence of the SBP56-encoding gene in the vicinity of genes encoding selenocysteine-containing formate dehydrogenases in the genome of Methanococcus vannielli and M. maripaludis ( Self et al. , 2004 ). Homologs of mtoX are present in a wide range of bacteria, and metagenomes from marine pelagic, coastal, hydrothermal and terrestrial environments, including DMS stable isotope probing experiments of soil and lake sediment samples. On the basis of processes that contribute to MT production in marine and terrestrial environments, a wide distribution of this enzyme is not surprising. The diversity of mtoX -containing organisms present in the environment is currently not well represented by isolated organisms, which suggests that the ability to degrade MT is more widely distributed than currently realized. This lack of environmentally relevant model bacteria limits our ability to appreciate which organisms are important as sinks for MT in different environments, how the expression of MTO in these organisms is regulated and which other degradative capabilities they may have. Using a stable isotope probing approach with 13 C 2 -DMS, we recently identified Methylophilaeceae and Thiobacillus spp. as DMS-degrading bacteria in soil and lake sediment ( Eyice et al. , 2015 ). The finding of mtoX genes in representatives of Thiobacillus and Methylophilaceae is consistent with the role that MT has as a metabolic intermediate in previously characterized DMS-degrading bacteria such as Thiobacillus spp. and adds further weight to the suggestion that certain Methylophilaceae have the metabolic potential to degrade DMS. The detection of mtoX in a saltmarsh environment is in agreement with such environments being hotspots of organic sulfur cycling ( Steudler and Peterson, 1984 ; Dacey et al. , 1987 ) based on production of DMSP and DMS by benthic microalgae, macrophytes and macroinvertebrates ( Otte et al. , 2004 ; Van Alstyne and Puglisi, 2007 ), and MT production through anaerobic processes in the sediment ( Lomans et al. , 2002 ). Overall, this study adds to our fundamental understanding of a key step in the sulfur cycle. The identification of the gene encoding this enzyme reveals its homology to a protein superfamily of which homologs are present in organisms ranging from bacteria to humans, but for which only sketchy functional information has been reported previously. The outcomes of this study will therefore facilitate future investigations of the role of MTO homologs in a wide range of organisms by providing testable hypotheses regarding its physiological relevance in these organisms. At the same time, the identification of the gene encoding MTO as well as its metal dependence will provide key foci for investigation of the diversity and distribution of MTO and potential constraints on its activity such as metal availability on MT degradation rates in the environment as well as aspects of the catalytic mechanism of MTO."
} | 4,269 |
34969851 | PMC8740587 | pmc | 7,839 | {
"abstract": "Significance Persistently diverse microbial communities are one of biology’s great puzzles. Using a modeling framework that accommodates high mutation rates and a continuum of species traits, we studied microbial communities in which antagonistic interactions occur via the production of, inhibition of, and vulnerability to toxins (e.g., antibiotics). Mutation size and mobility enhanced microbial diversity and temporal persistence to extraordinarily high levels. These findings—including the discovery that the duration of the transient phase in community assembly provides a guide to equilibrial diversity—highlight the potentially critical role that antagonistic interactions play in promoting the diversity of bacterial systems. Such interactions, together with resource-driven interactions and spatial structure, may drive the enigmatic levels of biodiversity seen in microbial systems.",
"discussion": "Discussion A continuous species model with simple rules for interactions based on antibiotic production, inhibition, and vulnerability can produce a wide array of complex dynamics and generate hyper-diverse, persistent communities. Despite enormous heterogeneity in behavior, certain general patterns in the assembly and stability of these hypothetical microbial communities stand out. The diversity patterns seen across time were expressed in the form of mean SD and its associated variance. The phase space of these two metrics shows a nonlinear relationship between the metrics. RF regressions revealed that the mutation size and the growth radius were the best predictors of mean SD diversity and its variance ( Fig. 3 B and C ). In other words, the mutation size and mobility of species jointly determine how diversity dynamics play out in these theoretical communities. Mobility controls diversity dynamics in numerous previous studies ( 6 , 20 , 27 , 28 ). In particular, small amounts of mobility enhance diversity, whereas large amounts of mobility jeopardize it ( 6 , 28 ). Reduced mobility results in spatially structured populations in which coexistence is easier to maintain as seen in experimental works on competing Escherichia coli strains ( 26 , 54 , 55 ). Recent work has shown that even in cases of high mobility (i.e., well-mixed communities), one can observe coexistence if higher-order interactions such as antibiotic production and degradation are considered ( 20 ). This is the case for in vivo experiments with bacterial colonies in the intestines of cocaged mice; these systems can be considered locally well mixed and have high levels of coexistence ( 30 ). High mutation rates, and processes such as horizontal gene transfer (HGT), have been long known to affect diversity in experimental microbial populations, especially during the initial phase of community assembly ( 56 , 57 ). However, cyclic dominance models have tended to overlook the role of mutations and HGT in coexistence. Such studies have focused instead on the detailed understanding of a small number of discrete species under a low mutation regime ( 47 ). Recently, studies have focused on understanding the effects of high mutational regimes in community assembly, which better represent microbial systems ( 38 ) and can generate frequent noncyclic interactions ( 40 ). Our efforts have considered an array of mutational regimes and characterized the importance of noncyclic patterns of interactions in the assembly and maintenance of microbial communities. Mutation size and the growth radius (a proxy for mobility) were good predictors of SD mean and SD variance but yielded no discernable patterns in the phase space ( Fig. 3 D and E ). In contrast, CFT partitioned the space cleanly into four regions with distinct properties ( Figs. 3 F and 4 A ). Communities that have low diversity (SD mean) and small fluctuations in diversity (i.e., low SD variance) (group 1) assemble quickly, and mobility enhances diversity and coexistence in this scenario ( Fig. 4 A and B ). In contrast, mobility negatively affects diversity and coexistence for communities with high diversity (SD mean) and larger fluctuations (high SD variance) (group 2), and these communities assemble slower than group 1. For communities that take a long time to stabilize (groups 3 and 4), diversity is generally high (SD mean), but it fluctuates less, and mobility has a negligible effect. Mutation sizes were weakly correlated with CFT, as observed previously ( 38 ), but exhibited substantial variation in trends due to interactions with other model parameters. For low SD mean and variance (group 1) and high SD mean and variance (group 2), mutation size positively affects CFT ( Fig. 4 B ). This implies that communities with larger mutations take longer to stabilize but only when the diversity outcome corresponds to these two groups ( Fig. 4 B ). If the SD mean is high but the SD variance is low, mutation affects CFT negatively (i.e., higher mutation helps communities converge faster) ( Fig. 4 B ). Community spatial structure is also important for coexistence ( 28 ). However, rather than focusing on the final spatial structure, we examined the degree of spatial disturbance that occurred throughout the assembly process, which provides insight into the extent of mixing/migration that might occur in these communities. The categorization of spatial disturbance regimes was best predicted by species-level parameters (the kill margin and the inhibit margin) rather than by other spatial parameters in our model. This is somewhat surprising, but higher-order interactions are known to influence spatial patterns of coexistence ( 20 ). Importantly, we found a strong correspondence between the regimes of spatial disturbance and the groups produced by CFTs. The communities that formed quickly (group 1) tended to have low spatial disturbance ( Fig. 4 C ). Communities that took more time to stabilize and ended up with high diversity through large overall diversity fluctuations (high SD mean and SD variance, group 2) have primarily a medium level of spatial disturbance. The communities that took longest to stabilize were characterized by high diversity (SD mean) and lower average fluctuations in diversity (SD variance) (groups 3 and 4) but have larger spatial disturbances. All these observations point toward a complex interplay of mutation and mobility, which affects the assembly time of a community (CFT), and, in turn, controls the diversity of the assembled community. Although mobility (or, more broadly, dispersal) is known to enhance diversity in ecological communities in both theoretical ( 58 , 59 ) and experimental ( 60 , 61 ) settings, especially at local scales ( 62 ), we found that mobility is linked to diversity only in communities with short and intermediate CFT (i.e., those that equilibrated relatively quickly). Mobility enhances diversity and coexistence in group 1 in which a subset of dominant (high relative abundance) species appear to set community dynamics. In this capacity, dispersal would appear to act in the classic disruptive fashion, permitting coexistence where it would otherwise not occur ( 63 , 64 ). In contrast, in group 2 in which communities are characterized by both high diversity and high fluctuations, mobility has a negative effect on diversity and coexistence in keeping with the capacity for dispersal to homogenize otherwise diverse systems ( 65 , 66 ). This dichotomy, together with the absence of an important role for mobility in group 4 in which communities are characterized by long-term nonequilibrial dynamics, offers an intriguing target for future integrative research. The mechanistic light shed upon these patterns by studying the Shannon equitability index (which measures the distribution of relative abundance of species in a community) and Simpson’s index (which measures the distribution of the “dominance” of species in a community) is also worth mentioning. The more equitable a community (higher Shannon equitability), the lower the chance of a particular species dominating the interactions (lower value of Simpson’s index) and, therefore, the longer it takes for the community dynamics to stabilize (longer CFTs) as seen in the behavior of the four groups ( Fig. 4 D – G ). The importance of species dominance to the maintenance of diversity in this model, which is structured via nonresource-based antagonistic interactions, is intriguingly similar to the critical role that species dominance plays in both real and theoretical communities structured via resource-based competition ( 67 – 70 ). Collectively, these results demonstrate the rich spatiotemporal dynamics that are possible when large numbers of microbial species with limited but heterogeneous rules for aggressive, inhibitory, and vulnerable interactions live in a common space. Although these findings are focused primarily on nonresource-based antagonistic interactions between microbes, such dynamics may also be relevant for coral communities and other spatially structured systems featuring diverse types of interspecific interactions ( 71 – 73 ). Our results also point toward the important yet often-overlooked role that transient dynamics play in the behavior and structure of ecological systems ( 53 ). Usually, models of ecological dynamics use asymptotically stable behavior or values at stability to explore patterns in the system ( 53 ). Admittedly, we have followed this approach in our exploration of how different parameters affect microbial communities’ long-run behavior (i.e., after CFT is reached). In addition, however, we also focused on categorizing spatiotemporal disturbance regimes and connecting these transient phenomena to system diversity. Great opportunities exist for future work investigating temporal diversity dynamics and how they relate to spatial heterogeneity and system processes. Such investigations can shed light on how transient microbial dynamics and assembly history affects the species diversity seen in microbial communities ( 74 ). Even though our model introduced the use of continuous axes for species traits and interactions, we note that the post hoc binning procedure creates a bridge back to the matrix/network framework that has characterized community ecology studies for decades ( 2 , 4 , 13 , 14 , 21 , 23 ). In this case, however, a multilayered network perspective would be necessary to accommodate the different kinds of antibiotic-mediated interactions (production, inhibition, and vulnerability). Future work could also incorporate other important forms of microbial interactions (e.g., mutualism, cross-feeding, resource competition) into the continuous species framework we have developed. Such investigations would allow exploration of how the interplay between resource-based and antagonistic interactions jointly shape diversity dynamics. However, to do this, we need to understand the interrelationships between resource-based and antagonistic interactions across species ( 11 , 75 , 76 ). Such investigations could be highly beneficial by providing a more complete view of how higher-order, intransitive interactions shape community assembly, stability, and diversity in natural systems."
} | 2,801 |
40307287 | PMC12043821 | pmc | 7,840 | {
"abstract": "The rhizosphere microbiota recruited by plants contributes significantly to maintaining host productivity and resisting stress. However, the genetic mechanisms by which plants regulate this recruitment process remain largely unclear. Here, we generated a comprehensive dataset, including 27 root transcriptomes, 27 root metabolomes, and 54 bulk or rhizosphere soil 16S rRNA amplicons across nine poplar species from four sections grown in nutrient-poor natural soil, along with eleven growth phenotype data. We provided a thorough description of this dataset, followed by a comprehensive co-expression network analysis example that broke down the wall of the four-way relationship between plant gene-metabolite-microbe-phenotype, thus identifying the links between plant gene expression, metabolite accumulation, growth behavior, and rhizosphere microbiome variation under nutrient-poor conditions. Overall, this dataset enhances our understanding of plant and microbe interactions, offering valuable strategies and novel insights for resolving how plants regulate rhizosphere microbial compositions and functions, thereby improving host fitness, which will benefit future research."
} | 295 |
38172438 | PMC7615830 | pmc | 7,841 | {
"abstract": "For both humans and machines, the essence of learning is to pinpoint which components in its information processing pipeline are responsible for an error in its output, a challenge that is known as ‘credit assignment’. It has long been assumed that credit assignment is best solved by backpropagation, which is also the foundation of modern machine learning. Here, we set out a fundamentally different principle on credit assignment called ‘prospective configuration’. In prospective configuration, the network first infers the pattern of neural activity that should result from learning, and then the synaptic weights are modified to consolidate the change in neural activity. We demonstrate that this distinct mechanism, in contrast to backpropagation, (1) underlies learning in a well-established family of models of cortical circuits, (2) enables learning that is more efficient and effective in many contexts faced by biological organisms and (3) reproduces surprising patterns of neural activity and behavior observed in diverse human and rat learning experiments.",
"discussion": "Discussion Our paper identifies the principle of prospective configuration, according to which learning relies on neurons first optimizing their pattern of activity to match the correct output and then reinforcing these prospective activities through synaptic plasticity. Although it was known that in energy-based networks the activity of neurons shifts before weight update, it has been previously thought that this shift is a necessary cost of error propagation in biological networks, and several methods have been proposed to suppress it 11 , 12 , 14 , 20 , 21 to approximate backpropagation more closely. By contrast, we demonstrate that this reconfiguration of neural activity is the key to achieving learning performance superior to that of backpropagation and to explaining experimental data from diverse learning tasks. Prospective configuration further offers a range of experimental predictions distinct from those of backpropagation (Supplementary Figs. 11 and 12 ). Together, we have demonstrated that prospective configuration enables more efficient learning than backpropagation by reducing interference, demonstrates superior performance in situations faced by biological organisms, requires only local computation and plasticity and matches experimental data across a wide range of tasks. Our theory addresses a long-standing question of how the brain solves the plasticity-stability dilemma, for example, how it is possible that, despite adjustment of representation in the primary visual cortex during learning 43 , we can still understand the meaning of visual stimuli we learned over our lifetime. According to prospective configuration, when some weights are modified, compensatory changes are made to other weights to ensure the stability of correctly predicted outputs. Thus, prospective configuration reduces interference between different weight modifications while learning a single association. Previous computational models have proposed mechanisms that reduce interference between new and previously acquired information while learning multiple associations 34 , 44 . It is highly likely that such mechanisms and prospective configuration operate in the brain in parallel to minimize both types of interference. Prospective configuration is related to inference and learning procedures in statistical modeling. If the ‘energy’ in energy-based schemes is variational free energy, prospective configuration can be seen as an implementation of variational Bayes that subsumes inference and learning 45 . For example, dynamic expectation maximization 46 , 47 can be regarded as a generalization of predictive coding networks in which the D-step optimizes representations of latent states (analogously to relaxation until convergence during inference) while the E-step optimizes model parameters (analogously to weight modification during learning). Other recent work 48 , 49 also noticed that the natural form of energy-based networks (‘strong control’ in their words) performs different learning than backpropagation. Their analysis concentrates on an architecture of deep feedback control, and they demonstrated that a particular form of their model is equivalent to predictive coding networks 49 . The unique contribution of our paper is to show the benefits of such strong control and explain why they arise. The principle of prospective configuration is also present in other recent models. For example, Gilra and Gerstner 50 developed a spiking model in which feedback about the error on the output directly affects the activity of hidden neurons before plasticity takes place. Haider et al. 51 developed a faster inference algorithm for energy-based models that computes a value to which the activity is likely to converge, termed latent equilibrium 51 . Iteratively setting each neuron’s output based on its latent equilibrium leads to much faster inference 51 and enables efficient computation of the prospective configuration. Predictive coding networks require symmetric forward and backward weights between layers of neurons, so a question arises concerning how such symmetry may develop in the brain. If predictive coding networks are initialized with symmetric weights (as in our simulations), the symmetry will persist because the changes in weight between neurons A and B are the same as those for feedback weight (between neurons B and A). Even if the weights are not initialized symmetrically, the symmetry may develop if synaptic decay is included in the model 52 because then the initial asymmetric values decay away, and weight values become more influenced by recent changes that are symmetric. Nevertheless, weight symmetry is not generally required for effective credit assignment 53 , 54 . Here, we assumed for simplicity that the convergence of neural activity to an equilibrium happens rapidly after the stimuli are provided so that the synaptic weight modification after convergence may take place while the stimuli are still present. Nevertheless, predictive coding networks can still work even if weight modification takes place while the neural activity is converging. Specifically, Song et al. demonstrated that if neural activities are only updated for the first few steps, the update of the weights is equivalent to that in backpropagation 14 . As a reminder, we demonstrate here that if the neural activities are updated to equilibrium, the update of the weights follows the principle of prospective configuration and possesses the desirable demonstrated properties. Thus, a learning rule where neural activities and weights are updated in parallel will experience a weight update that is equivalent to backpropagation at the start and then move to prospective configuration as the system converges to equilibrium 55 . Furthermore, predictive coding networks have been extended to describe recurrent structures 56 – 58 , and it has been shown that such networks can learn to predict dynamically changing stimuli even if weights are modified before the activity converged for a given ‘frame’ of the stimulus 57 . The advantages of prospective configuration suggest that it may be profitably applied in machine learning to improve the efficiency and performance of deep neural networks. An obstacle for this is that the relaxation phase is computationally expensive. However, recent work demonstrated that by modifying weights after each step of relaxation, the model becomes comparably fast to backpropagation and easier for parallelization 55 . Most intriguingly, it has been demonstrated that the speed of energy-based networks can be greatly increased by implementing the relaxation on analog hardware 59 , potentially resulting in energy-based networks being faster than backpropagation. Therefore, we anticipate that our discoveries may change the blueprint of next-generation machine learning hardware, switching from the current digital tensor base to analog hardware and being closer to the brain and potentially far more efficient."
} | 2,019 |
24371363 | null | s2 | 7,844 | {
"abstract": "The breakdown of organic nitrogen in soil is a potential rate-limiting step in nitrogen cycling. Arbuscular mycorrhizal (AM) fungi are root symbionts that might improve the ability of plants to compete for organic nitrogen products against other decomposer microbes. However, AM uptake of organic nitrogen, especially in natural systems, has traditionally been difficult to test. We developed a novel quantitative nanotechnological technique to determine "
} | 113 |
37223735 | PMC10124867 | pmc | 7,845 | {
"abstract": "Abstract Nucleotide-derived signalling molecules control a wide range of cellular processes in all organisms. The bacteria-specific cyclic dinucleotide c-di-GMP plays a crucial role in regulating motility-to-sessility transitions, cell cycle progression, and virulence. Cyanobacteria are phototrophic prokaryotes that perform oxygenic photosynthesis and are widespread microorganisms that colonize almost all habitats on Earth. In contrast to photosynthetic processes that are well understood, the behavioural responses of cyanobacteria have rarely been studied in detail. Analyses of cyanobacterial genomes have revealed that they encode a large number of proteins that are potentially involved in the synthesis and degradation of c-di-GMP. Recent studies have demonstrated that c-di-GMP coordinates many different aspects of the cyanobacterial lifestyle, mostly in a light-dependent manner. In this review, we focus on the current knowledge of light-regulated c-di-GMP signalling systems in cyanobacteria. Specifically, we highlight the progress made in understanding the most prominent behavioural responses of the model cyanobacterial strains Thermosynechococcus vulcanus and Synechocystis sp. PCC 6803. We discuss why and how cyanobacteria extract crucial information from their light environment to regulate ecophysiologically important cellular responses. Finally, we emphasize the questions that remain to be addressed.",
"conclusion": "Concluding remarks In recent years, several studies have uncovered that c-di-GMP is the master molecule governing the survival strategy of cyanobacteria through the control of many aspects of cellular physiology in an ecologically relevant context. Specifically, c-di-GMP is emerging as a crucial regulator of cyanobacterial collective behaviours, such as phototaxis and cell aggregation, processes that usually involve groups of cells. Since cyanobacteria are one of the founders of multicellularity (Hammerschmidt et al. 2020 ), c-di-GMP signalling systems could have been fundamental for the development of multicellular bacterial consortia. Future studies should provide a more comprehensive understanding of the c-di-GMP signalling network in cyanobacteria and will uncover the respective effectors and their downstream targets. These developments will be crucial for comprehending the tremendous ecological and evolutionary success of cyanobacteria and their interaction with other phototrophic and non-phototrophic microorganisms in many different ecosystems.",
"introduction": "Introduction Cyanobacteria are obligate phototrophs and ancestors of the chloroplasts of algae and plants. It is not surprising that cyanobacteria are considered an excellent model for studying oxygenic photosynthesis and related cellular functions, such as light harvesting, energy and carbon metabolism, and its regulation. However, the ecology and behavioural responses of cyanobacteria have not been studied at a comparable molecular level. Nucleotide-based signalling molecules are known to control such responses in many bacteria at the transcriptional, translational, and post-translational levels. In cyanobacteria, studies on the role of prototypical second messengers cAMP, cGMP, and (p) ppGpp, as well as the later discovered cyclic dinucleotides c-di-GMP and c-di-AMP, have lagged behind. However, in the last 10 years, based on the availability of more sequencing data from cyanobacteria, it has become clear that these phototrophs encode a large variety of proteins that, based on their domains, are potentially involved in second messenger signalling. Notably, cyanobacteria are particularly rich in photoreceptors that contain output domains similar to those involved in c-di-GMP synthesis and degradation (Agostoni et al. 2013 ). For other bacteria, it is well established that intracellular c-di-GMP levels coordinate specific aspects of bacterial lifestyle, such as biofilm formation, aggregation, virulence, and motility (Jenal et al. 2017 ). High intracellular concentrations of c-di-GMP are usually associated with inhibition of motility and induction of biofilm formation. These behavioural processes have rarely been studied in cyanobacteria at the molecular level, for various reasons. On the one hand, most marine cyanobacteria that were intensively studied in their ecological context (e.g. marine Prochlorococcus and Synechococcus species) do not encode enzymes related to c-di-GMP metabolism. On the other hand, research on well-established and genetically tractable model cyanobacteria, which contain a large variety of such proteins, has focused on other general or unique characteristics of these strains, such as nitrogen fixation, photosynthesis, the circadian clock, and toxin production. Moreover, the most widespread cyanobacterial laboratory strains in research [e.g. Synechococcus elongatus PCC 7942 and Synechocystis sp. PCC 6803 (henceforth Synechocystis )] have lost many natural behavioural responses that are targeted by c-di-GMP due to the accumulation of mutations in key genes encoding components of cellular appendages such as type IV pili. In this review, we highlight the major advances in our understanding of the role of c-di-GMP in cyanobacterial behaviour, focusing on the two model cyanobacterial strains Thermosynechococcus vulcanus ( T. vulcanus , recently also termed Thermostichus vulcanus ) and Synechocystis (Fig. 1 ). It is noteworthy that most of the analysed behavioural responses require the participation of multiple cells, which Menon et al. ( 2021 ) refer to as ‘collective behaviour’. Figure 1. Regulation of cellular behaviour by c-di-GMP in cyanobacteria. Light, especially blue light, controls many c-di-GMP-dependent lifestyle decisions in cyanobacteria. However, other external factors might contribute to c-di-GMP-dependent regulation. In Synechocystis (coccoid bacteria), blue light leads to an overall increase in the cellular c-di-GMP concentration, which induces flocculation and biofilm formation and inhibits motility. A high c-di-GMP concentration leads to cellulose-dependent aggregation of T. vulcanus (a rod-shaped bacterium). C-di-GMP influences phototaxis reversals of T. vulcanus and heterocyst development in the filamentous cyanobacterium Anabaena sp. PCC 7120."
} | 1,567 |
28701966 | PMC5488766 | pmc | 7,847 | {
"abstract": "Coherent collective behavior emerges from local interactions between individuals that generate group dynamics. An outstanding question is how to quantify group coordination of non-rhythmic behavior, in order to understand the nature of these dynamics at both a local and global level. We investigate this problem in the context of a small group of four pedestrians walking to a goal, treating their speed, and heading as behavioral variables. To measure the local coordination between pairs of pedestrians, we employ cross-correlation to estimate coupling strength and cross-recurrence quantification (CRQ) analysis to estimate dynamic stability. When compared to reshuffled virtual control groups, the results indicate lower-dimensional behavior and a stronger, more stable coupling of walking speed in real groups. There were no differences in heading alignment observed between the real and virtual groups, due to the common goal. By modeling the local speed coupling, we can simulate coordination at the dyad and group levels. The findings demonstrate spontaneous coordination in pedestrian groups that gives rise to coherent global behavior. They also offer a methodological approach for investigating group dynamics in more complex settings.",
"introduction": "Introduction Collective behavior in humans and other animals is thought to arise from local interactions between individuals that are coupled by sensory information. This coupling may be modulated by factors such as environmental context (e.g., presence of predators, food sources), motivation (e.g., metabolic state, goals), and cognitive or social constraints (e.g., strategies, group membership, dominance relations). To understand the emergence of collective behavior, researchers must characterize both the local coupling between individuals and the global patterns of coordination. Such an approach calls for a set of analytic tools that can quantify the degree and stability of spatio-temporal coordination at both the individual and collective levels. The purpose of this paper is to investigate coordination in human collective behavior, beginning with the analysis of local and global coordination in small pedestrian groups. By way of introduction, consider the flocking behavior of a murmuration of starlings. Each bird is visually coupled to nearby neighbors, and this local coupling influences an individual's behavior in accordance with a particular set of “rules;” we call them control laws to emphasize their continuous dynamical as opposed to logical form. These local interactions give rise to coordinated behavior between neighbors, which in turn feeds back to involve more individuals, so the coordination pattern propagates through the flock. The end result is a self-organized pattern of global motion that emerges from local interactions. The exact nature of the control laws that govern these local interactions and how they generate coherent flocking behavior is an active area of research (Ballerini et al., 2008 ; Cavagna et al., 2010 ; Hildenbrandt et al., 2010 ; Lukeman et al., 2010 ). It is difficult to infer the local control laws based solely on the observed global behavior, however. An important theoretical result is that different sets of interaction rules can generate the same pattern of coherent flocking (Vicsek and Zafeiris, 2012 ); thus, the local control laws are underdetermined by analysis of the global behavior. This finding implies that direct experimental study of interactions between individuals is required to model the control laws, which can then be used to simulate coordination patterns. Therefore, a complete account of collective behavior demands an approach that combines a local-to-global (bottom-up) perspective, in which empirically-grounded control laws are used to predict global behavior, and a global-to-local (top-down) perspective, in which measurements on global behavior are analyzed and compared with the predictions (Sumpter et al., 2012 ). We are pursuing this dual approach to understand the collective behavior of human crowds. The program of research includes characterizing the control laws by which visual information guides locomotion, a pedestrian model that generates locomotor trajectories, and multi-agent simulations of the emergent crowd dynamics. Warren ( 2006 ) proposed a behavioral dynamics framework that aims to characterize how stable low-dimensional behavior emerges on-line from the interactions between an agent and its environment. Goal-directed behavior such as locomotion is regulated by perceptual information in accordance with task-specific control laws (Gibson, 1979 ; Warren et al., 2001 ; Warren and Fajen, 2004 ). Within this framework, Fajen and Warren ( 2003 , 2007 ) and Warren and Fajen ( 2008 ) developed a pedestrian model that successfully characterizes locomotor behavior such as steering to stationary and moving goals, and avoiding stationary and moving obstacles. This model has recently been extended from agent-environment interactions to interactions between pairs of pedestrians (dyads), including pursuit and evasion, following, and walking side-by-side (Cohen et al., 2010 ; Bonneaud and Warren, 2012 ; Page and Warren, 2013 ; Rio et al., 2014 ). In certain contexts, two pedestrians may have the goal of walking together, in which case they visually coordinate their velocity, i.e., walking speed and direction of travel (heading). During pedestrian following, Rio et al. ( 2014 ) found that the follower matches the leader's speed, independent of their interpersonal distance (1–3 m); this is accomplished by nulling the optical expansion of the leader (see also Lemercier et al., 2012 ; Bruneau et al., 2014 ). A similar speed-matching strategy was observed in side-by-side walking, with a similar coupling strength (Page and Warren, 2013 ). In addition, Dachner and Warren ( 2014 ) found that pedestrians match the walking direction of a neighbor, independent of interpersonal distance (1, 2, 4 m), with a comparable coupling strength in following and side-by-side walking. They recently proposed that speed and heading are jointly controlled by nulling both the optical expansion and the change in bearing direction of the leader (Dachner and Warren, 2017 ). These results indicate that pedestrian dyads utilize visual information to adopt a common speed and direction over a range of distances and positions. This research has established a preliminary set of control laws that govern pedestrian interactions. An outstanding question is whether they scale from dyads to groups, and ultimately, can account for the self-organization of collective crowd behavior. Answering this question requires methods for quantifying the emergent patterns of coordination at both the local and global scales. This is a particularly difficult problem given that pedestrian locomotor trajectories are a continuously evolving, aperiodic behavior. Accordingly, it requires analysis tools that can identify the temporal pattern of non-rhythmic coordination between dyads at a local level, as well as group coherence at a global level. As a first step, the system must be operationalized. In previous work, two behavioral variables have been used to describe a locomotor trajectory: (1) the agent's direction of heading (Φ), and (2) the agent's speed ( s ), which together define the agent's velocity in an allocentric coordinate frame. This operationalizes a pedestrian as having two degrees of freedom (DoF), which may be coupled between neighbors. Similarly, Riley et al. ( 2011 ) proposed that behavioral coordination between two agents arises from the coupling of their DoF. It is believed that agents couple the DoF of a system via shared information variables, so that the DoF directly regulate one another. Hence, the control of behavior at the level of the group emerges via functional, information-based linkages between the behavioral variables of individual agents. When framed in terms of behavioral dynamics, collective behavior can be considered a problem of informationally coupling the appropriate behavioral variables to yield a stable solution of the global behavioral dynamics. For the task of locomotion, each pedestrian is operationalized as a two DoF system with the state variables Φ and s . Each additional individual in a group of N pedestrians would add two more state variables to the collective system, so the total DoF = 2N. Thus, the state space of the system has 2N dimensions. Once the behavioral variables are identified, the next step is to quantify the degree of coordination at the collective level. From a global perspective, the degree of coordination among a set of pedestrians would be reflected in a reduction of the effective DoF of the system to a value between 2N, such that all individuals move independently, and 2, such that all individuals move with the identical speed and direction. One way to measure the reduction in a system's DoF is to quantify the dimensional compression of the observed behavior. Principle Components Analysis (PCA) is a valuable tool in this regard (Riley et al., 2011 ). PCA can be used to identify collective variables, or principle components, based on the relations among observations in a high-dimensional state space (cf. Haken and Wunderlin, 1990 ). It also indexes the load magnitude of each state variable on the identified principle components, which can help uncover the coupling between behavioral variables. The strength of PCA is its ability to include many variables of a complex system in a single analysis and to provide an output that quantifies the degree of relation, or even coordination, between the component variables. Its limitation is that PCA is a linear analysis, and therefore assumes linear relations among the system's variables. PCA provides the first part of the analysis by quantifying group coherence at the global level. At the local level, the next step is to quantify the degree of coordination between pairs of individuals in a group, to reveal the coupling strength as a function of variables such as neighbor distance and position. One approach is to compute the linear cross-correlation between the time series of speed (or heading) for two pedestrians. The limitation of this analysis is that it assumes that individuals are coupled at a single time-scale and that behavior is stationary (i.e., a constant delay). It therefore has limited utility in analyzing more complex systems, such as bidirectional coupling at multiple time-scales and non-stationary behavior that evolves over time. Cross-recurrence quantification (CRQ), is well-suited to the latter type of data and has proven useful in analyzing interpersonal coordination (cf., Shockley et al., 2003 ; Richardson, D. C. et al., 2007 ; Ramenzoni et al., 2012 ). CRQ is a non-linear analysis that indexes repeating patterns in a pair of time series at multiple temporal scales (Webber and Zbilut, 1994 ; Shockley et al., 2002 ). In particular, the output measure “cross-maxline” (CML) has proven to be a reliable estimate of the temporal stability of coordination, associated with coupling strength, between two movements (Richardson, M. J. et al., 2007 ; Page and Warren, 2013 ). However, these local analyses are limited to a pairwise comparison of dyads in a group. Finally, to determine whether a model of the local coupling can account for the observed patterns of coordination, agent-based simulation methods can be used to try and reproduce the data. In particular, we investigate the mechanism of coordination by testing whether our model of the local “rule” for speed matching, derived from data on pairs of pedestrians, generalizes to coordination in a group, and can explain the adoption of a common collective speed and heading. Our goal in the present paper is to measure the degree of coordination in pedestrian groups at the global and local levels, and to model the local coupling that generates such coordination. Establishing the emergence of coordinated behavior is prerequisite to modeling the informational control laws, characterizing the conditions for the emergence of such behavior, and eventually investigating the roles of other cognitive and social variables. In the present experiment, groups of four pedestrians walked toward one of three goals, while the group's initial density (interpersonal distance) was varied on each trial (see Figure 1 ). The role of density is important due to its potential contribution to self-organization: if coupling strength is distance-dependent, higher densities would create stronger local interactions and promote coherent crowd formation. Previous results have shown that, for an individual pedestrian, the coupling to obstacles decays exponentially with distance, asymptoting at 3–4 m (Fajen and Warren, 2003 ), but on the other hand, the coupling between pairs of pedestrians appears to be independent of distance, at least up to 3–4 m (Dachner and Warren, 2014 ; Rio et al., 2014 ). In the present experiment, we explored interpersonal distances of 0.5–2.5 m within groups of four people. Figure 1 The four possible starting positions for each of the four possible starting densities (left). Note the dotted “trigger” line 1 m from the midpoint between the front two participants that represents when the experimenter “goal” command was given. The visual couplings of the six possible dyads (center) with double arrows indicating bi-directional vs. unidirectional (single arrow) coupling. The six dyads are highlighted in the right pane. As described above, we analyzed two behavioral variables: the walking speed s and walking direction Φ for each agent. This resulted in a total of eight state variables, or DoF, for the four-agent system. To determine whether the observed coordination is a consequence of the informational coupling between individuals and is not due to other task constraints, we compared the real groups with virtual groups that were constructed by randomly sampling the same four pedestrians from four different trials. At the global level, we hypothesized that the real groups would exhibit dimensional compression in all conditions, compared to the virtual groups. We also investigated whether dimensionality would be reduced more in the higher density conditions. At the local level, we hypothesized that the coupling strength would be greater between real dyads than virtual dyads, and we asked whether it would increase as a function of group density. Finally, we tested whether Rio et al.'s ( 2014 ) speed-matching model generalizes to the observed speed coordination between individuals in a group and can explain the emergence of a common speed.",
"discussion": "Discussion The present experiment investigated the degree of coordination in pedestrian groups during goal-directed walking, with the aim of analyzing the effects of a visual coupling, group density, and neighbor position on collective behavior. We analyzed the behavioral variables heading Φ and speed s in a four-pedestrian group, yielding an eight DoF system. We then submitted the behavioral variables to a global (collective) analysis: (1) PCA to index the dimensional compression of group behavior; and to local (pairwise) analyses: (2) linear cross-correlation to estimate the coupling strength between dyads in a group, and (3) non-linear CRQ to measure the dynamic stability of the local coupling. Our main finding is that most analyses yielded evidence of spontaneous coordination in walking speed due to the visual coupling in real groups, compared to reshuffled virtual groups. It is important to point out that the external task constraints in this experiment (common goal, simultaneous go signal, simultaneous goal command, similar preferred walking speeds) by themselves induced similar behavior across individuals, which we estimated using the shuffled virtual groups. We expect that emergent heading and speed coordination would be observed in less restricted contexts, and research is under way to study spontaneous coordination in both heading and speed. At the global level of analysis, the PCA indicated that visually coupled pedestrian groups exhibited significant dimensional compression across all experimental conditions. Note that the external task constraints accounted for a reduction of ~2.2 DoF (from 8 to 6.2) in the virtual groups, a 23% reduction in DoF. Yet the visual coupling produced a further reduction of ~2.6 DoF (from 6.2 to 3.6) in the real groups, or an additional 33% reduction in DoF. This is indicative of a functional reorganization of DoF via the informational coupling of behavioral variables, consistent with the emergence of collective coordination. These results are similar to those of Ramenzoni et al. ( 2012 ), who demonstrated dimensional compression in an interpersonal supra-postural task, and support the reduction of DoF in interpersonal coordination proposed by Riley et al. ( 2011 ). The analysis of the composition of PC1 offers preliminary evidence of a new collective variable underlying the emergence of group coordination in the context of the current task. The loading of behavioral variables on PC1 suggests that speed coordination is a primary contributor to the collective behavior, whereas heading coordination was no greater in the real than the virtual group. Further, the analysis of the composition of PC2 demonstrated that the heading and speed loading is not simply dichotomous, as evidenced by the bimodal distribution for the heading coefficients in the real group (Figure 5B ) and the negatively skewed unimodal distribution of the speed coefficients (Figures 6B ). This indicates that the heading behavioral variable was a relatively weak contributor to the first two PCs overall. Thus, the remaining discussion focuses on the analysis and modeling of speed coordination. At the local level of analysis, the cross-correlations for speed indicated a high visual coupling strength within the groups. Specifically, a significantly higher mean correlation was found for the real group ( r = 0.84) compared to the virtual group ( r = 0.36), independent of dyad. This can be explained similarly to the PCA results, in that the visual coupling increased the speed correlation for all dyads. It appears that local coupling strengths can be reliably estimated by pairwise linear correlations. However, the pairwise cross-correlations did not reveal a significant difference between types of dyads. This could be due, in part, to the possibility that back participants were influenced by more than one neighbor at a time. We are currently developing a neighborhood model that allows us to estimate the combined influence of multiple neighbors. The non-linear CRQ analysis provided further evidence regarding the strength and stability of the local coupling. Speed coordination exhibited a longer CML in real groups than in virtual groups, indicating that the visual coupling was dynamically stable. Specifically, real dyads were stably coupled for almost two full seconds (i.e., 111.13 samples at 60 Hz), at some point in each 6–8 s trial. Taken together, the PCA, cross-correlation, and CRQ results indicate that the global coordination in the present task is due in large part to the local coordination of speed, which in turn emerges from the visual coupling between individual pedestrians. Finally, we tested whether an empirical model of the local speed coupling could reproduce the observed coordination patterns. The simulation results supported this interpretation, for the coordination of dyads in a group is reproduced by the speed-matching model. The simulation results show that the speed-matching model generalizes from pairs of pedestrians to pedestrian groups, and imply that the local coupling is sufficient to explain the adoption of a common speed. We conclude that the local visual coupling can account for the pattern of global coordination. Somewhat to our surprise, we did not observe a consistent effect of density on the degree of coordination. In fact, no measures yielded significant density effects, consistent with our previous finding that speed coordination in following is independent of interpersonal distance over 1–3 m (Rio et al., 2014 ). It is possible that the range of densities tested (0.5–2.5 m spacing) was insufficient to reveal an effect, or that the external task constraints, combined with a short walking distance, limited the degree of variation in the data. Research is in progress to test a wider range of densities (up to 4 m spacing) over longer walking distances, without a common goal or timing signals. Finally, we would like to mention that we also performed an uncontrolled manifold (UCM) analysis on the eight-dimensional Φ and s data (Scholz and Schöner, 1999 ), as another way to estimate the reduction in effective DoF. This approach was unsuccessful, and it is instructive to consider why that was the case. A UCM analysis depends on the existence of reciprocal compensation between two or more behavioral variables in the system, which is considered a signature of motor synergies. But in retrospect, there is no reason to expect reciprocal compensation in collective group behavior: the acceleration of one agent would not be expected to produce a compensatory deceleration by a coupled agent to maintain the mean speed, but rather a coordinated acceleration; similarly, a change in heading direction by a subset of agents would not be expected to yield compensatory heading changes in the other direction, but a coordinated turn by the group. This observation suggests that reciprocal compensation may not be a general characteristic of all forms of interpersonal coordination in human groups (cf. Riley et al., 2011 ). The present work is a starting point for understanding collective behavior in pedestrian groups. We began by analyzing the local coupling in dyads, on the hypothesis that this generic coordination mechanism would scale up to small groups, large crowds, and even flocks or schools in other species. Expanding the methodological framework of interpersonal coordination (Riley et al., 2011 ; Ramenzoni et al., 2012 ) to the behavior of small groups, we obtained evidence of dimensional compression and speed coupling. The present framework provides a foundation for the analysis and modeling of local and global coordination in future research. It is likely that other factors may also constrain group coordination. For example, cognitive processes such as decision-making and motivation, and social factors such as group membership, dominance relations, and social communication, may influence the selection of goals, neighbors, walking speeds, and control laws and shape the emergent crowd dynamics. The present experiment evaluates ways of quantifying local and global coordination in many of these contexts, and offers an approach to characterizing emergent collective behavior."
} | 5,739 |
28861471 | PMC5566163 | pmc | 7,848 | {
"abstract": "Mimosa Origami: Large-scale dynamic self-assembly of soft materials powered by\ncapillary-driven propagation of a pinpoint stimulus.",
"conclusion": "CONCLUSIONS In summary, we have demonstrated a new self-organization mechanism that, over time,\nenables the directional large-scale reconfiguration of soft materials. The observed\nself-assembly dynamics occur through a cascade of thermodynamic states that are\nindividually accessible by dosing the water volume supplied to the Janus bilayer. As a\nresult, this Mimosa Origami regime can overcome some of the limitations of purely\nelastocapillary systems and can theoretically self-assemble over unlimited lengths.\nExemplification of this concept in microfluidics demonstrates record-high response time,\nas compared to conventional microfluidics ( 27 ), with near-ideal capillary velocities. Moreover, the\nself-assembly is reversible, being capable of unfolding and recovering the initial\nsurface properties. This orthogonal propagation of stimulus and response demonstrated by\nthe Janus bilayers is a powerful mechanism that can be exploited in numerous research\nareas and commercial applications, including stimuli-responsive materials ( 10 , 32 ), fog harvesting ( 33 ), artificial muscles ( 9 , 25 ), sensors ( 34 ), switches ( 32 ), and power-independent devices ( 18 ).",
"introduction": "INTRODUCTION Various biological systems in nature orchestrate a high level of adaptability to their\nenvironments through the use of smart material interfaces. These can be distinguished\nunder two overarching categories, namely, static and dynamic self-assembly ( 1 , 2 ). Static self-assembly is constrained by equilibrium\nthermodynamics ( 3 ). This is\nillustrated by the elegant self-cleaning of the lotus leaves ( 4 ) and the crystallization-driven ( 5 ) construction of intricate shells\nby marine invertebrates ( 6 ). More\nexciting is the dynamically responsive nature of living organisms that often manifests\nin spontaneous motion ( 7 ). For\nexample, the Mimosa pudica , a thigmonastic plant, can react to the\nslightest contact pressure with a very rapid protective folding of its leaflets. This\ncentimeter-long, negative tropism is transduced by a cascade of electrical potentials\nand osmotic pressure waves ( 8 ).\nAlthough the specific mechanisms vary largely, the structural and functional properties\nin nature exhibiting such large-scale reconfigurations provide important insights for\nthe rational design and creation of new classes of self-organizing materials for\npotential applications in biotechnology ( 7 ), micromechanics ( 9 ), microelectronics ( 10 ), photonics ( 11 ), and fluidics ( 1 ). To date, the engineering of inorganic systems capable of spontaneous motion relies\nlargely on static self-organization mechanisms ( 12 , 13 ). In these systems, the material self-organization is\nlocalized around/in the proximity of the initial stimulus droplet, limiting the\nself-assembly scale. For example, in classical elastocapillarity, where a thin polymer\nsheet folds around a water droplet, the water droplet’s surface provides both the\nenergy for the initial folding and the propagation of the folding stimulus to the\nresidual polymer sheet. As a result, the scale of the self-assembled structure is\ncomparable to the droplet size and limited to ca. 10 mm, a very small fraction of that\nobserved in natural systems ( 12 ). Here, we report the directional dynamic self-organization of soft materials into\nlarge-scale geometries by a rapid cascade folding mechanism that is reminiscent of the\n M. pudica ’s leaflet folding. We engineer a hybrid Janus\nbilayer structure with enhanced and precisely controlled surface chemistry, morphology,\nand mechanical properties. These soft materials are capable of imparting directional\nspontaneous motion in response to a pinpoint stimulus. This self-organization mechanism\nrelies on the rapid propagation of a pinpoint stimulus and an orthogonal local material\nresponse. The longitudinal reconfiguration (stimulus propagation) rate (maximum of 7.8\ncm/s) is driven by capillary/Laplace pressure ( 14 ). The elastocapillary-driven orthogonal material\nresponse, observed here, has much faster kinetics (folding at ca. 23.8 cm/s) and is in\nline with previous studies ( 15 – 17 ). We use this system to induce the reversible\nself-assembly of three-dimensional (3D) microfluidic channels and spontaneous liquid\nself-propulsion, with velocities approaching pneumatically actuated systems. To the best\nof our knowledge, this Mimosa Origami regime represents the first large-scale\nself-assembly of a material powered by capillary-driven propagation of a pinpoint\nstimulus across a predetermined path.",
"discussion": "RESULTS AND DISCUSSION The material layout involves a stack of multifunctional layers (fig. S1, A to C)\ndesigned to impart efficient transformation of surface energy into directional kinetic\nand elastic energy. This is enabled through a stimulus-responsive Janus interface. The\nuse of Janus materials has been well documented for nanoparticles, where two distinct\nand sometimes opposite properties, such as hydrophilic-hydrophobic, are synergistically\nexploited ( 18 ). Here, a cohesive\nJanus bilayer is obtained by an interconnected network of highly wettable\npolycaprolactone (PCL) nanofibers adhering to the bottom layer of polyvinyl chloride\n(PVC) microfibers ( Fig. 1A ). The adhesion of the\nPVC and PCL layers is attributed mainly to van der Waals interaction. Sequential\ndeposition of PCL and PVP led to very weak bonding and layers that were easily peeled\noff, suggesting that mechanical interlocking is not the main adhesion mechanism. The PVC\nis designed to be highly superhydrophobic and flexible, serving as a water impenetrable\nbackbone to the PCL layer. Moreover, to attain sufficient mobility for vertical\nself-organization while suppressing in-plane wrinkling, the Janus bilayer is hosted on a\nsuperhydrophobic substrate (fig. S2A) with low affinity to PVC ( Fig. 1B ). This substrate is composed of polystyrene (PS) nanofibers\non a dense polydimethylsiloxane (PDMS) film (fig. S3A). Fig. 1 Preparation and characterization of the superhydrophilic-superhydrophobic\nJanus bilayer. ( A ) Schematic illustration of the Janus bilayer assembly: a\nmultifunctional stack is fabricated by sequential electrospinning of a protective\nPVP, a superhydrophilic PCL, and a superhydrophobic PVC nanofiber layers on paper.\nThis stack is shaped in a functional geometry and completed by adhering a PS\nnanofiber layer to a flexible PDMS substrate on the PVC surface by van der Waals\n(VDW) interaction. The protective PVP layer and paper are easily peeled off by\nhand. ( B ) Optical photographs show the isolated Janus bilayer and its\ncohesive and stretching properties. ( C and D ) SEM\nanalysis at low-magnification (8.8k) and high-magnification (70k) images (insets,\nbottom right) of the Janus bilayer PVC and PCL surfaces and their contrasting\nwetting (insets, upper right). ( E ) FTIR spectroscopic analysis of the\nmultilayer stack and isolated Janus bilayer confirming its PCL (orange line) and\nPVC (green line) composition. a.u., arbitrary units. ( F ) Dynamic\nmechanical stress-strain analysis (tension mode) of the Janus bilayer showing a\nsoft rubbery nature, with a Young’s modulus ( E ) of 4.85\nMPa. This multilayer stack is easily assembled on paper using a sacrificial polyvinyl\npyrrolidone (PVP) layer as a protective film for the in situ deposition of the top (PCL)\nsurface of the Janus bilayer ( Fig. 1A and fig. S1).\nIn terms of wettability, the PCL layer has a Wenzel hemiwicking (fig. S4) character,\nwith the water contact angle approaching 0° ( Fig.\n1D , inset). This is achieved by the careful engineering of a network of\ninterwoven PCL nanofibers with an average diameter of 192 ± 49 nm ( Fig. 1D ). Similarly, the PVC backbone of the Janus\nbilayer is fabricated in situ by deposition of submicrofibers with an average diameter\nof 671 ± 305 nm ( Fig. 1D ) on the PCL layer.\nThis porous PVC structure is superhydrophobic, with a water contact angle of 155°\n± 7° and a contact angle hysteresis of 30° ± 10°\n( Fig. 1C , inset). The functional stack is\ncompleted by van der Waals stacking of the PS-PDMS substrate on the PVC layer. The Janus\nbilayer can be easily isolated from the protective PVP film (fig. S5) and the PS-PDMS\nsubstrate ( Fig. 1B ) by sequential peel-off. The\nstructural integrity and composition of the isolated bilayer are confirmed by its\nchemical signature ( Fig. 1E ). The Fourier transform\ninfrared (FTIR) spectroscopic spectra of the multilayer stack is characterized by five\nsharp peaks located at 1656, 1726, 612, 701, and 789 cm −1 that are\nattributed to the C=O ring of PVP, carbonyl C=O stretch of PCL, C–Cl gauche of\nPVC, C–H aromatic ring of PS, and Si–C with CH 3 rocking\nvibrations of PDMS, respectively ( 19 ). The dominant presence of PCL and the lack of PVP in the\nfinal Mimosa Origami structure (PCL-PVC-PS-PDMS) confirm successful removal of the\nsacrificial layers ( Fig. 1E ). Similarly, chemical\nsignatures of freestanding Janus bilayers (PCL side) confirm the clean separation of\nJanus bilayers from the PS-PDMS substrate. The key structural and chemical properties of the Janus bilayer, such as its\nelastocapillary length, surface roughness ( r ), and energy\n( E S ) can be tuned far beyond that of conventional dense\npolymers ( 20 ). Optimization of\nthe PCL and PVC layer thickness leads to self-supported, flexible, and highly cohesive\nfilms (fig. S1B). Scanning electron microscopy (SEM) and gravimetric analysis reveal\nthat the PCL has a surface roughness of 68 (Supplementary Materials). This is\nsignificantly higher than that ( r = 2 to 6) achieved by microtexturing\nof dense films ( 21 ) and can be\nfurther enhanced by increasing the PCL layer thickness and decreasing the nanofiber\ndiameter. Dynamic mechanical analysis of the optimal Janus bilayer reveals a unique\nrubbery stress-strain nature ( Fig. 1F ) with a\nYoung’s modulus of 4.85 MPa. This is two to three orders of magnitude lower than\nthat of bulk PVC (2700 to 3000 MPa) ( 22 ) and PCL (252 to 430 MPa) ( 23 ). Considering the total PVC and PCL\nlayers’ thickness of 50 μm, this results in a very low bending rigidity\n( K b ) of 68 nNm and an elastocapillary length\n( L EC ) of only 1 mm, where L EC = K b γ LV (1) and γ LV is the surface\nenergy density of water (0.072 Nm −1 ). Figure S6 illustrates the transient elastocapillary response of the Janus bilayer to\nwater. When a water droplet is gently placed on the superhydrophilic side of the\ncircular-shaped bilayer, the latter partially detaches from the PS-PDMS substrate and\nencapsulates it by folding symmetrically (movie S1). For a circular surface of 79\nmm 2 , this process takes less than 33 ms, ultimately resulting in the\nformation of a bulb containing the initial water volume. Note that the presence of the\nPS-PDMS substrate and nonwetting superhydrophobic (PVC) backbone of the Janus bilayer\nare also essential for the successful folding and subsequent leak-proof water\nencapsulation. Without the PVC layer, the non-Janus superhydrophilic PCL layer is\nsusceptible to unwanted effects, such as uncontrolled in-plane wrinkling and eventual\nwater leakage (figs. S7 and S8). Without the PS-PDMS substrate, the self-assembly is\nadversely affected by pinning to the hosting surface (figs. S7 and S8). The rapid folding response of the Janus bilayer is attributed to its unique\nelastochemical properties. Notably, whereas the folding of thin dense films around a\nwater droplet has been previously showcased as an exemplary application of\nelastocapillarity, here we show that utilization of highly porous layers is challenging\nbecause water leaks rapidly (fig. S7) through the hydrophilic porous structure. The\nsuperhydrophilic-hydrophobic Janus layout significantly improves the material response,\navoiding wrinkling and containing the water droplet within its volume. Our rough\nnanostructured morphology enables significantly higher surface energy density than that\nof 2D textured dense films. The Janus bilayer’s surface energy density was\nestimated at 185 J kg −1 (Supplementary Materials). This is comparable\nto that of artificial muscles ( 9 ,\n 24 , 25 ) and large enough to easily overcome the\ncounteracting bending rigidity (68 nNm) of the Janus bilayer. Together, this unique\nJanus bilayer architecture extends the working regime of classical capillary origami and\nrenders the folding of films with more than 10 times larger thickness ( 12 ) while preserving a very small\nelastocapillary length through exceptionally high surface roughness. The Janus bilayer’s properties can be exploited to induce an unprecedented\ndirectional self-organization of soft materials into functional 3D structures. Figure 2A shows the spontaneous construction of a\nstraight microchannel with a length of 6.5 cm. This is achieved by placing a water\ndroplet with a diameter of 0.42 cm on the circular terminal of a rectangular strip of\nthe Janus bilayer (figs. S5A and S9 and movie S2). This directional folding response is\nreminiscent of the mimosa’s tropism in nature ( Fig.\n2B ), though the stimulus propagation mechanism of the Janus bilayer is\ndifferent. The reversibility of this self-organization state is achieved by\nreinstatement of the initial surface energy equilibrium. Figure 2C illustrates selected snapshots of the spontaneous unfolding\nprocess. Here, we used low–surface tension ethanol liquid to wet both the\nsuperhydrophobic and superhydrophilic sides of the Janus bilayer. Spectroscopy maps the\nsurface composition of the Janus bilayers during the folding-unfolding cycles and\nsuggests clean desorption of both water and ethanol from the material during cyclic use,\nwith preservation of the initial chemical compositions ( Fig. 2D ). Subsequent desorption of the water on the PCL side restores the\nsymmetry of the Janus bilayer surface energy ( Fig.\n2E ) and unfolds the microchannel back into its original flat shape. The\nunfolded Janus bilayer is easily reactivated (Materials and Methods) and capable of\nmulticycle self-assembly ( Fig. 2 , C and E). Fig. 2 Demonstration of directional self-organization via Mimosa Origami\nself-assembly. ( A ) Optical photographs of the spontaneous directional\nself-organization response of a rectangular-shaped Janus bilayer. A pinpoint water\ndroplet stimulus results in the immediate self-assembly of a centimeter-long\nmicrochannel. ( B ) This rapid motion is reminiscent of the\nstimulus-response propagation during the negative tropism of the M.\npudica ’s leaflets. ( C ) The folded Janus bilayers\nare spontaneously unfolded by immersion in an ethanol bath. Restoration of the\ninitial surface properties allows a novel folding cycle, demonstrating the full\nreversibility of this self-organization state. ( D ) FTIR spectroscopic\nanalysis showing the variation in the surface composition of the Janus bilayer\nduring the folding and unfolding cycle. ( E ) Schematic illustrations\nof capillary-induced unfolding of the self-assembled microchannel. Figure 3 (A and B) explains the mechanism of the\nMimosa Origami self-assembly. A water-filled bulb initially forms (<33 ms) in\nresponse to the wetting of the Janus bilayer’s circular end, and then the liquid\nfront advances into the rectangular strip in a relatively slow manner due to the PCL\nlayer hemiwicking character. When a critical amount of water has accumulated at the\nbulb-strip junction (<110 ms), the wetted strip folds into a quasi-cylindrical\nmicrochannel. The formation of this 3D hollow architecture gives rise to strong\ncapillary force that propels water into the adjacent dry section in a rapid manner (fig.\nS9). Most notably, the folding signal is transported at an average rate of 400 ms\ncm −1 or an average velocity of 2.5 cm s −1 over a\nstrip length of 6.5 cm. For a droplet of 40 μl and a strip width of 2 mm, the\ninstantaneous stimulus propagation rate decreases linearly from initially 7.8\ncm −1 to standstill over the length of 6.5 cm ( Fig. 4C ). Distinct from static self-organization, this axial\npropagation is orthogonal to the local elastocapillary potential that drives the folding\nof the strip. This rapid propagation of the pinpoint water stimulus and the orthogonal\nfolding response ( Fig. 3B ) results in a cascade of\ncross-sectional folding and directional mass transport. The effective capillary pressure\ndecreases during self-assembly (fig. S4C). In addition, the stimulus propagation is also\ncountered by elastic folding and viscous capillary forces. The dropping capillary\npressure and increasing elastic and viscous forces decrease the stimulus propagation\nrate ( Fig. 4C and fig. S13), ultimately halting the\nself-assembly, although some water is still available in the bulb. As a result, the\ninitial scale of the self-assembly is determined by the initial droplet volume, and the\nself-assembly can be restarted by the supply of additional liquid to the water bulb\n(movies S3 and S4). Fig. 3 Mimosa Origami self-assembly mechanism and theoretical analysis. ( A ) Optical photographs of the directional self-assembly of the Janus\nbilayers into a closed microchannel. ( B ) Schematic description of the\nself-assembly process: initially, a water-tight bulb is formed by the rapid\nfolding (33 ms) of the Janus bilayer terminal around a water droplet. Thereafter,\nthe waterfront slowly advances from the bulb to the dry PCL surface. Once\nsufficient water has collected, the wet Janus bilayer strip folds rapidly, forming\na hollow 3D cross section. This leads to the Mimosa Origami propagation (400 ms\ncm −1 ) of the folding stimulus by longitudinal propulsion of\nthe waterfront and orthogonal folding of the Janus bilayer strip. ( C )\nTheoretical model of the minimal strip width required for the spontaneous Mimosa\nOrigami self-assembly regime as a function of the surface roughness and\ncharacteristic contact angle (θ e ). Fig. 4 Application of the Mimosa Origami directional self-organization to\nmicrofluidics. ( A ) Waterfront displacement from the bulb during Mimosa Origami\nself-assembly as a function of the strip width and time. ( B ) Maximal\ndisplacement and velocity as a function of strip width and 1/width fit.\n( C ) Water instantaneous velocity as a function of the time since\nwater droplet release on the Janus bilayer terminal surface and comparison against\nthe LWR equation for an ideal circular capillary. ( D to\n G ) Exemplary modular microfluidic designs obtained by the\nself-assembly of functionally shaped Janus bilayer strips, including (D) mixing\nbulb channel, (E) curved tapering channel, (F) T junctions, and (G) U turns. We derived a mathematical model to determine the range of material and geometrical\nproperties for the spontaneous Mimosa Origami regime ( Fig.\n3C ). This is based on the extension of the equations of McHale et\nal . ( 15 ) to an\ninfinitesimally small length of the rectangular strip of the Janus bilayer, assuming\nthat the top and bottom surfaces of the Janus bilayer stay in the Wenzel and\nCassie-Baxter states, respectively ( 15 , 16 ). Material properties ( Fig.\n1F ) and equations are described in the Supplementary Materials. We found that\nthe spontaneous formation of a 3D hollow cross section necessitates a minimal critical\nwidth ( w c ) of the Janus bilayer strip. This critical width\nis a function of the elastocapillary length ( L EC ), the\ncharacteristic contact angle (θ e ), and roughness factor\n( r ) of the Janus bilayer top surface ( 26 ). It can be estimated as w c = L E C 2 1 + r cos ( θ e ) 2 Π (2) The roughness ( r ) of the nanofibrous PCL layer was computed from the\nratio of its total surface area to its geometric surface area, resulting in a surface\nroughness of 68 r = 4 m ∅ ︀ π ρ D 2 (3) where m is the mass\n(3.74 × 10 −3 kg m −2 ) of the monolayer PCL per\nm 2 , ⌀ is the average circumference of a nanofiber (601 ×\n10 −9 m), ρ is the density of PCL (1145 kg\nm −3 ), and D is the average diameter of a nanofiber\n(192 × 10 −9 m). On the basis of these calculations, the PCL\nlayer has a surface roughness of 68. Figure 3C shows contour plots of the minimal strip\nwidth for spontaneous folding as a function of the contact angle and roughness factor\nfor hydrophilic films (θ e < 90°) and a constant\nelastocapillary length (1 mm). On the basis of this theoretical model, the minimal width\nfor Mimosa Origami decreases significantly with increasing surface roughness ( Fig. 3C ). For dense flat films ( r =\n1), it is impossible to fully fold strips less than 4 mm in width. In stark contrast,\nfor a film having comparable roughness ( r = 68) to the top Janus\nbilayer surface, spontaneous complete folding is expected down to a strip width of 1.3\nmm. This is extremely close to the elastocapillary length of 1 mm. As a result, for\nthese nanorough Janus bilayers, the small amount of liquid transferred from the bulb to\nthe dry interface by hemiwicking is sufficient to trigger the self-assembly and initiate\nthe folding stimulus. Furthermore, it should be noted that there exists an upper limit\nfor the strip width beyond which the self-assembled hollow cross section would partially\ncollapse under the self-generated capillary tension. A prompt and distal based motion that mimics the M. pudica ’s\nmechanical response represents an essential improvement over state-of-the-art\nself-organization of soft materials ( 12 ). Here, we have further optimized the self-assembly\nkinetics by the Janus bilayer’s geometrical design. For a constant water droplet\nvolume, the maximal self-assembly length is inversely proportional to the width of the\nstrips ( Fig. 4A ). This is in line with the\ntheoretical and dynamic analysis of the self-organization process ( Figs. 3 and 4B ) and confirms\nthat during Mimosa Origami, the flow is driven by the Laplace pressure of the\nself-assembled hollow cross section. For a rectangular strip with a width of 2 mm, the\nfolding stimulus propagated through the complete strip length (6.5 cm) with an average\nflow velocity of 2.5 cm s −1 ( Fig.\n4 , A and B). Notably, for this optimal geometry, the self-assembly length is\nonly limited by the initial size of the strip. Significantly longer lengths (ca. 200%)\nwere easily achieved by increasing the path length (movie S4). Increasing the strip\nwidth to more than 3 mm partially disrupts the shape of the hollow cross section and\ndecreases the maximal length of the self-assembled microchannels ( Fig. 4A and movie S5). This is attributed to the partial\nself-collapse of the Mimosa Origami effect for strip size significantly above the\nelastocapillary length. The average folding-stimulus propagation velocity measured for a\n2-mm-wide and 6.5-cm-long strip is 2.5 cm s −1 , which is comparable to\nthe travel speeds (2 to 3 cm s −1 ) of electrical signals in the\n M. pudica ( 8 ). Remarkably, in an exemplification of bio-inspired microfluidics, the optimized Janus\nbilayers conveyed fluids at an estimated initial volumetric flow rate of 14.7 μl\ns −1 . This is up to 10 times faster than state-of-the-art\nmicrofluidic propulsion systems based on wicking, evaporation, and degassing ( 27 ). Notably, the optimal\nself-assembling Janus bilayer has an initial flow velocity up to 81% of that of an ideal\nLucas-Washburn-Rideal (LWR) capillary due to the small delay in the time required for\nthe self-assembly of the capillary structure. The subsequent decrease in instantaneous\nvelocity (stimulus propagation rate) scales as the ideal LWR capillary ( Fig. 4C ) but eventually ceases because of the\ndecreasing effective capillary pressure and counteracting elastic folding and viscous\nforces. These speeds also rival some of the fastest pumpless microfluidic devices based\non etched superhydrophilic V-shaped grooves ( 28 ). The self-organization potential of this multilayer\nstructure extends beyond previous studies on the utilization of water surface tension to\nconstruct complex but static 3D structures based on polymers ( 12 , 29 ), silicon ( 30 ), and other materials. This is exemplified by\ncontrolling the directionality and geometry of Janus bilayer self-assembly into several\nfunctional shapes. Various key microfluidic modules with increasing degree of difficulty\nare easily obtained. This includes bulb mixing, tapered curves, and single and double\nright corners with a demonstrated self-assembly length of 10 cm ( Fig. 4 , D to G) that can be used for fabricating flexible modular\nmicroflow devices ( Fig. 4 , D to G, and movies S6 to\nS8). From a fundamental perspective, these structures are more than an order of\nmagnitude larger than that previously achieved by static elastocapillary self-assembly\n( 12 , 31 )."
} | 6,165 |
35409097 | PMC8998989 | pmc | 7,849 | {
"abstract": "While chemical fertilisers and pesticides indeed enhance agricultural productivity, their excessive usage has been detrimental to environmental health. In addressing this matter, the use of environmental microbiomes has been greatly favoured as a ‘greener’ alternative to these inorganic chemicals’ application. Challenged by a significant proportion of unidentified microbiomes with unknown ecological functions, advanced high throughput metatranscriptomics is prudent to overcome the technological limitations in unfolding the previously undiscovered functional profiles of the beneficial microbiomes. Under this context, this review begins by summarising (1) the evolution of next-generation sequencing and metatranscriptomics in leveraging the microbiome transcriptome profiles through whole gene expression profiling. Next, the current environmental metatranscriptomics studies are reviewed, with the discussion centred on (2) the emerging application of the beneficial microbiomes in developing fertile soils and (3) the development of disease-suppressive soils as greener alternatives against biotic stress. As sustainable agriculture focuses not only on crop productivity but also long-term environmental sustainability, the second half of the review highlights the metatranscriptomics’ contribution in (4) revolutionising the pollution monitoring systems via specific bioindicators. Overall, growing knowledge on the complex microbiome functional profiles is imperative to unlock the unlimited potential of agricultural microbiome-based practices, which we believe hold the key to productive agriculture and sustainable environment.",
"introduction": "1. Introduction With the global population projected to reach 9 billion by 2050, farm productivity is also required to be increased up to 70–100% ideally to meet food, fuel, and fibre demands [ 1 ]. For that, excessive usage of chemical fertilisers and pesticides has been unavoidable to improve agricultural productivity [ 2 ], which unfortunately has caused environmental pollution, soil infertility, and the loss of biodiversity, affecting the overall ecosystem sustainability and potentially being hazardous to human health [ 3 ]. Therefore, as the current agricultural practices have become increasingly unsustainable, there is an essential need to rectify this issue. Globally, the Sustainable Development Goals (SDGs) formulated by the United Nations in 2015 have been widely promoted to ensure development sustainability, with the 17 SDGs forming the backbone of this plan. The SDGs are adopted by factoring in the relationship of the key ecological processes and relevant human behavioural activities, which are highly interdependent [ 4 ]. It is well established that ecological processes are primarily mediated and regulated by microbiomes, the predominant form of life on the planet. Overtaking the advantage of the existing synergistic microorganisms in regulating the biogeochemical cycle in the environment [ 5 ], a subtle way of exploiting the beneficial microbes and the carried essential genes involved in certain processes could be embarked upon in order to resolve the negative impacts of aggressive agricultural activity and potentially advance the related SDGs such as SDG 2 (Zero Hunger), SDG 6 (Clean Water and Sanitation), SDG 7 (Affordable and Clean Energy), SDG 9 (Industry, Innovation, and Infrastructure), SDG 11 (Sustainable Cities and Communities) and SDG 12 (Responsible Consumption and Production) through microbiome-based innovations ( Figure 1 ). With this motion, researchers are steering the application of next-generation sequencing (NGS) towards achieving a comprehensive understanding of the role and potential of beneficial microbes in sustainable agricultural practices where productivity and profit are maximised, environmental damage is minimised, and natural resources are preserved. NGS started about two decades ago with a microbial taxonomic diversity analysis using amplicon sequencing (16S/18S rRNA sequences) [ 6 , 7 ]. Amplicon sequencing is mainly used to characterise the diversity and structural compositions of microbial communities, and many studies have successfully employed this approach to explore the microbial taxonomies and phylogeny in different environments, such as surface water [ 8 ], agriculture wastewater treatment ponds [ 9 ], and activated sludge reactors [ 10 ]. Amplicon-based gene sequencing gained immense popularity due to its low cost, simple sample preparation protocols, and wide accessibility of bioinformatics tools [ 11 ]. While gene amplicon sequencing provides unparalleled insight into the microbiome’s nature, this technique can only provide profiling of community structures without having an accurate overview of their functionality [ 12 ]. As technology advances, researchers have started to work with whole-genome shotgun (WGS) metagenomics, contributing to the discovery of both the taxonomic and functional diversity of a particular community [ 13 , 14 , 15 ]. Shotgun metagenomic sequencing is a powerful technique to infer the taxonomical structure by sequencing all genomic DNA fragments in a community while avoiding the biases observed in amplicon sequencing due to the non-requirement of amplification before sequencing [ 11 ]. However, as WGS is a type of descriptive-based analysis, it can only provide a predictive functional profile [ 16 ]. Aside from that, shotgun metagenomics is also unable to differentiate the active from the inactive members of the microbiome [ 17 ]. Consequently, it cannot address a part of the research question relating to which community contributes to the observed ecosystem activity and which are merely present in a dormant state [ 6 ]. Therefore, the demand for a more comprehensive approach to infer the microbiome’s functional profiles brought forth the emergence of metatranscriptomics. Metatranscriptomics is defined as the assessment of a gene’s expression in a population or a whole community [ 16 ]. Metatranscriptomics is used to expand our knowledge of the microbial community’s functions in terms of gene expression, regulations, and pathways, which significantly correspond to environmental changes. In the last decade, various studies have employed large-scale metatranscriptomics analyses for complex environmental samples such as the study of soil fertility [ 18 ], biofertilisers [ 19 ], crop disease [ 20 ], and second-generation biofuel [ 21 ]. As metatranscriptomics applications are gaining more attention, it is important to understand how we can utilise the knowledge on microbial genes and functionality to achieve a sustainable environment ( Figure 2 ). Thus, the current applications of environmental metatranscriptomics in microbiome-based innovations and its possible contribution towards realising environmentally sustainable agriculture by harnessing the microbiome’s potential in developing healthy soil and pollution bioindicators will be further discussed in this review."
} | 1,744 |
29030778 | PMC5895687 | pmc | 7,850 | {
"abstract": "Femtosecond transient absorption was used to study excitation decay in monomeric and trimeric cyanobacterial Photosystem I (PSI) being prepared in three states: (1) in aqueous solution, (2) deposited and dried on glass surface (either conducting or non-conducting), and (3) deposited on glass (conducting) surface but being in contact with aqueous solvent. The main goal of this contribution was to determine the reason of the acceleration of the excitation decay in dried PSI deposited on the conducting surface relative to PSI in solution observed previously using time-resolved fluorescence (Szewczyk et al., Photysnth Res 132(2):111–126, 2017). We formulated two alternative working hypotheses: (1) the acceleration results from electron injection from PSI to the conducting surface; (2) the acceleration is caused by dehydration and/or crowding of PSI proteins deposited on the glass substrate. Excitation dynamics of PSI in all three types of samples can be described by three main components of subpicosecond, 3–5, and 20–26 ps lifetimes of different relative contributions in solution than in PSI-substrate systems. The presence of similar kinetic components for all the samples indicates intactness of PSI proteins after their deposition onto the substrates. The kinetic traces for all systems with PSI deposited on substrates are almost identical and they decay significantly faster than the kinetic traces of PSI in solution. We conclude that the accelerated excitation decay in PSI-substrate systems is caused mostly by dense packing of proteins. Electronic supplementary material The online version of this article (doi:10.1007/s11120-017-0454-z) contains supplementary material, which is available to authorized users.",
"introduction": "Introduction In recent years, photosynthetic pigment–protein complexes gained great attention in fabrication of bio-inorganic devices. Therefore, deep insight into the nature and behavior of these complexes in different environments (e.g., immobilized onto various substrates) is crucial for their utilization. Among all photosynthetic particles, Photosystem I (PSI) is one of the most commonly used photosensitive materials in miscellaneous prototype bio-devices, due to its overall high stability and high photon-to-electron conversion quantum yield near unity (Gobets and van Grondelle 2001 ). In nature, it catalyzes the trans-membrane electron transfer upon irradiation, and thus it can be considered as a natural optoelectronic device in nanoscale. The structure and properties of this complex are well documented in the literature (Fromme and Grotjohan 2008 ; Caffarri et al. 2014 ; Nelson and Junge 2015 ). Among 12 protein subunits and 127 cofactors per cyanobacterial PSI monomer, which include 96 chlorophylls a (Chls a ), 22 carotenoids (Cars), two phylloquinones, three iron–sulfur clusters, four lipids, one Ca 2+ metal ion, and 201 water molecules, few of them compose the reaction center (RC) where charge separation process occurs (Jordan et al. 2001 ). The RC cofactors are the following: 6 Chls (two of them forming P700—primary donor, two accessory Chls labeled A, and two primary electron acceptors labeled A 0 ), two phylloquinones, and three terminal iron–sulfur cofactors (Fx, Fa, Fb). The role of the remaining pigments is to form an antenna system which collects the light energy. Although primary donor absorption maximum is at 700 nm, a great majority of Chls absorb at around 680 nm (“bulk” Chls). In various PSI complexes, a small pool of pigments that absorb at wavelengths longer than 700 nm is present (“red” Chls). Their nature, origin, and presumptive location in Synechocystis were extensively studied in the literature and briefly discussed previously (Szewczyk et al. 2017 ). A recent, detailed review on red Chls in different cyanobacteria can be found in Karapetyan et al. ( 2014 ). The usefulness of PSI complex in various bio-hybrid devices has been previously demonstrated in many studies. These include different substrates (for working electrodes) such as gold, TiO 2 , ZnO, graphene, and conductive polymer systems (for instance, Ciesielski et al. 2010 ; Mershin et al. 2012 ; Feifel et al. 2015 ; Carter et al. 2016 ; Robinson et al. 2017 ). In these contributions, it was proposed that electron flow direction in devices based on high-carrier concentration electrodes (gold, modified graphene, FTO—fluorine-doped tin oxide, or ITO—indium-doped tin oxide) is determined mostly by orientation of the PSI near the electrode, due to the unidirectional electron transfer in PSI—from P700 to Fb. Theoretically, in the case of disordered assembly of the PSI monolayer onto surface, the observed net photocurrent density can be decomposed into two major, opposing contributions, which correspond to cathodic (electron injected from the working electrode substrate to P700 + ) or anodic (electron injected from Fb − to the substrate) photocurrent. In the aforementioned studies, different net photocurrent, under various experimental conditions, was observed, which may be related to the degree of asymmetry of cathodic and anodic photocurrents. However, there may be other factors limiting the overall photocurrent related to electrical contact between proteins and substrate, charge recombination reactions, and, last but not least, very basic energy and electron transfer properties of PSI immobilized on the substrate. In our previous paper, we reported that immobilization of PSI on inorganic substrate, FTO conductive glass (a plate of glass covered with a thin, optically transparent, conductive layer of FTO), modifies the spectral characteristics of some chlorophyll pools and accelerates the decay of PSI antenna excitation: from mean lifetime of t av = 16 ps for PSI in solution to 11 ps for PSI deposited on FTO glass (Szewczyk et al. 2017 ). A possible reason was proposed to be an additional quenching process resulting from electron injection from PSI to the conductive FTO substrate. It is well established that excited states of dyes being in contact with semiconducting surfaces are quenched very efficiently due to electron transfer between the dye molecule and the substrate (for instance, Tachibana et al. 1996 ; Grätzel et al. 2005 ; Sobuś et al. 2015 ; Ashford et al. 2015 ). The electron injection rates for dyes immobilized on semiconducting surfaces like TiO 2 , ZnO, ZrO 2 , or Al 2 O 3 are often extracted from measurements of dye-excited states’ lifetimes (Wenger et al. 2005 ; Koops et al. 2009 , Sobuś et al. 2014 , 2015 ; Idigoras et al. 2015 ). Still, however, the nature and timescale of electron injection from dyes most commonly used in dye-sensitized solar cells photovoltaics, ruthenium or indoleum, are a matter of debate. Some contributions considered biphasic injection kinetics—the faster component with lifetime < 100 fs, and the slower one occurring on picosecond to nanosecond timescale (Koops et al. 2009 ; Idigoras et al. 2015 ; Sobuś et al. 2015 ), related with electron transfer from the singlet and triplet state of the dye, respectively. Other studies proposed that slower component is a result of a transfer from loosely attached, more distant, or aggregated molecules (Wenger et al. 2005 ). The exact values of electron injection rates depend on the surface type and experimental conditions. However, acceleration of the dye-excited state quenching was generally correlated with more efficient electron injection into a particular substrate (Koops et al. 2009 ; Sobuś et al. 2014 ). Another considered cause of accelerated excitation dynamics in immobilized PSI (Szewczyk et al. 2017 ) was radical disparity in environmental conditions (PSI in water + detergent solution vs. PSI dialyzed in order to get rid of detergent and then deposited onto semiconductor and dried). Due to drying and formation of a densely packed multilayer PSI film during the deposition process, new interactions between complexes may arise which change their spectroscopic (and energetic) properties. Red chlorophylls are examples of natural states that appear as a result of strong interaction between pigments, leading to a mixing of excitonic and charge transfer states (Romero et al. 2008 ; Novoderezhkin et al. 2016 ). An artificial formation of three extra red states in densely packed PSI monomers that are deposited on FTO glass and dried was demonstrated (Szewczyk et al. 2017 ). To our knowledge, there is no literature describing the influence of dehydration on the first steps of energy transfer in PSI; however, it was shown that water density near electron transfer cofactors may affect the protein flexibility, charge screening, and P700 + reduction rate (Dashdorj et al. 2005 ). A possibility of altering hydrogen bonds and, therefore, changes in structure which may lead to aggregation was discussed in the aforementioned contribution, but interestingly the FTIR analysis of severely dehydrated sample excluded changes in secondary structure of PSI (Sacksteder et al. 2005 ). In another contribution (Malferrari et al. 2016 ), it was demonstrated that the relative humidity indeed modulates electron transfer to iron–sulfur clusters. Despite this, the dried PSI complexes remained stable over a long period. In the case of purple bacteria, it was shown that the decay rate of the excited bacteriochlorophyll dimer (P*) was slightly increased due to drastic drying of the purple bacterial RC film (Yakovlev et al. 2012 ). In order to reveal the factors responsible for the observed accelerated de-excitation in immobilized PSI and taking into account the above considerations, we decided to investigate excitation dynamics in PSI, using femtosecond transient absorption in 3-ns time window, under four different environmental conditions: (1) PSI suspended in aqueous solution; (2) PSI immobilized and dried on FTO conductive glass (Fig. 1 a); (3) PSI immobilized and dried on non-conducting surface of silanized FTO glass (Fig. 1 b); and (4) PSI immobilized on FTO conductive glass and being in contact with aqueous solution (photovoltaic cell-like system with a thin layer of aqueous solution between PSI-covered FTO glass and a coverglass; Fig. 1 d). The experiments were performed for both trimeric and monomeric forms of PSI from Synechocystis sp. PCC 6803. \n Fig. 1 Illustration of the PSI-immobilized systems used in the study. a PSI deposited on the conductive surface of the FTO glass. b PSI deposited on the silane-covered conductive surface of the FTO glass. c PSI deposited on the silane-covered non-conductive surface of the FTO glass. d PSI deposited on the conductive surface of the FTO glass in the cell-like system with aqueous electrolyte \n Femtosecond transient absorption measurements for wild-type PSI from Synechocystis in aqueous solution have been performed previously (Hastings et al. 1994 ; Melkozernov et al. 2000 ; Savikhin et al. 2000 ; Shelaev et al. 2010 ). In general, the excitation dynamics in PSI core complex can be described using three major components: (1) subpicosecond, which describes excitation equilibration, (2) 2–6 ps related to both photochemical quenching of bulk chlorophylls and equilibration between bulk and red chlorophylls, and (3) 21–26 ps ascribed to excitation trapping in reaction center (photochemical quenching of excitation equilibrated over bulk and red chlorophylls). Experiments described in the aforementioned studies were performed under excitation in the red region (predominantly 660 nm), for trimeric PSI form, in a range of excitation pulse energies.",
"discussion": "Results and discussion Transient absorption measurements—kinetics at 690 nm Figure 3 a presents transient absorption kinetics at 690 nm of monomeric and trimeric PSI in solution and immobilized onto the conductive layer of FTO glass (Fig. 1 a). Acceleration of the overall excitation decay in the immobilized PSI versus PSI in solution is well noticeable and consistent with previous, time-resolved fluorescence results (Szewczyk et al. 2017 ). On the other hand, there are no major differences in kinetic traces between monomeric and trimeric forms, in respective systems. In the following, we focus on trimeric PSI. Similar results for monomeric PSI obtained under identical experimental conditions are shown in supplementary information. Figure 3 b presents kinetic traces of the trimeric PSI immobilized onto different substrates/in different systems: PSI deposited and dried on bare FTO (conditions with possible electron injection from PSI to FTO), PSI deposited and dried on silanized FTO (conditions with blocked electron injection by insulating layer of silane), and PSI deposited on bare FTO and being in contact with aqueous buffer. One can see that the traces are almost identical, except for the longest component, the lifetime and amplitude of which slightly vary for different samples (see below). A common feature of these three systems is the formation of a film of crowded PSI complexes. On the other hand, they differ either in electrical contact between PSI and FTO or in degree of hydration. Thus, we hypothesize that the major factor responsible for the acceleration of excitation decay in immobilized PSI complexes is their crowding on the substrate, and not drying or electron injection into the substrate. As shown in the inset of Fig. 3 b, the experiments with PSI on silanized FTO and on silanized glass brought also almost identical results. This observation is interpreted in terms of the lack of electron transfer from PSI to the FTO conductive layer through the insulating silane layer by hypothetical tunneling effects or by the silane layer discontinuities. \n Fig. 3 Transient absorption kinetic traces at 690 nm (wavelength of maximal absorbance changes) recorded at an excitation wavelength of 400 nm. The original traces were multiplied by factor (− 1) and normalized to unity. a Comparison of raw signals obtained for monomeric and trimeric PSI in solution and immobilized directly onto the conductive surface of FTO glass. Inset: fits to the corresponding data. b Data obtained for trimeric PSI immobilized onto substrates in three different systems. Inset: comparison of the decay kinetic for trimeric PSI attached to two different silanized surfaces \n Transient absorption measurements—PSI in solution—global analysis Figure 4 c shows the results of the global analysis performed for trimeric PSI in solution. The dynamics of the PSI complex can be described by three major decay-associated spectral (DAS) components of 0.3, 3.3, and 26 ps. The non-conservative shape of the fastest component with a smaller negative band at about 670 nm and a larger positive band at 690 nm indicates two different processes occurring on the same time scale: (1) relaxation from the Soret to Q y band of Chls (causing the appearance of stimulated emission in the Q y region—manifested as the positive band) and (2) excitation equilibration within the bulk antenna (energy transfer from “blue bulk Q y ” to “red bulk Q y ” Chl states, seen as the negative band at ~ 670 nm and a small contribution to the positive band at ~ 690 nm) (Fig. 5 ). The second component (3.3 ps) was also characterized with two bands, the negative one at about 685 nm and the positive one at about 710 nm. The existence of both positive and negative bands, with similar amplitudes (and integrated areas), indicates energy transfer, which was assigned to equilibration of bulk red Chls. The third component (26 ps) with one negative band at about 690 nm and a bump at about 705 nm was ascribed to photochemical quenching in the RC of the excitation equilibrated over bulk (690 nm) and red (705 nm) Chls. The additional, forth component with a small amplitude was assigned to uncoupled or loosely attached chlorophylls (Uchls), because of their slow excitation decay and blue-shifted spectrum. Similar DAS shapes and lifetimes for the closed RC were reported previously (Savikhin et al. 2000 ). The slight differences in the spectra reported in our study and those by Savikhin et al. may come from the model used (four- vs. five-component fit), different excitation wavelengths, and different ways of controlling the RC state (chemical vs. strong illumination). \n Fig. 4 Time-resolved absorption results for the trimeric PSI in solution and immobilized in different systems. The first column ( a, d, g, j ) presents the model underlying target analysis, estimated molecular lifetimes, and initial distribution of the excitation between well-coupled and uncoupled Chls; the middle column ( b, e, h, k ) presents species-associated spectra (SAS) resulting from the target analysis; the third column ( c, f, i, l ) presents the results of global analysis (decay-associated spectra, DAS) \n \n Fig. 5 Graphical explanation of the mixed character of the fastest, subpicosecond kinetic DAS and SAS components shown in Fig. 4 as “Soret → Bulk” transition. Within the specified subpicosecond lifetimes in Fig. 4 , both relaxation from Soret to Q y states of bulk Chls and excitation transfer between different subpopulations of bulk Chls being in Q y states occur \n Transient absorption measurements—PSI in solution—target analysis In order to obtain more physical description of the excitation energy dynamics in PSI in solution and next in immobilized PSI, the target analysis was performed. Within the given signal-to-noise ratio, the four-compartment model was sufficient to characterize basic transitions for all the systems under study. This model contained the following compartments: “Soret”, “Bulk”, “Red”, and uncoupled (“Uchls”; Fig. 4 a). It is essentially the same model as was used previously for analysis of time-resolved fluorescence data (Szewczyk et al. 2017 ) but with an extra “Soret” compartment which was possible to identify due to better temporal resolution of the transient absorption experiment. As a result, molecular lifetimes of each transition (reciprocals of molecular rate constants; Fig. 4 a; Table 1 ) and spectral distributions of the states within each of the compartments (species-associated spectra (SAS)—Fig. 4 b) were obtained. Excitation quenching by closed RC occurs with lifetime t \n 1 = 18.3 ps (Fig. 4 a). Excitation energy transfer from bulk to red Chls with lifetime t \n 2 = 13.6 ps is coupled with backward transfer which is nearly three times faster than forward reaction: t \n 3 = 4.7 ps. The SAS band minima for bulk and red Chls are at 685.5 and 706 nm, respectively (Fig. 4 b). The negative band of the “Red” SAS is of bigger amplitude than that of the “Bulk” SAS. Similar effect was observed also for monomers in solution (see Fig. S1 in Supplementary information). The blue-shifted spectrum with a maximum at about 680 nm and lifetime t \n 4 = 5 ns was assigned to uncoupled Chls. The last molecular lifetime, t \n 5 = 0.3 ps, was ascribed to “Soret” → “Bulk” transition as discussed above (Fig. 5 ). \n Table 1 Parameters estimated from transient absorption measurements of trimeric PSI complexes Sample Average lifetime at 690 nm t \n av (ps) SAS band minimum wavelength (nm) \n Δλ (nm) \n δ = ± 0.5 nm \n ΔH \n 0 (meV) \n δ = ± 2.6 meV \n t \n 1 (ps) \n t \n 2 (ps) \n t \n 3 (ps) \n t \n 4 (ps) \n t \n 5 (ps) \n ΔG \n 0 (meV) \n δ = ± 1 meV \n N \n r \n eff \n Bulk λ \n b \n Red λ \n r \n Trimers solution 20 685.5 706 20.5 53 18.3 13.6 4.7 5000 0.3 − 27 4.3 ± 0.7 (4.2)* (3.9) # \n Trimers bare FTO 14 689 710 21 53 14.3 22 7.9 454 0.7 − 26 4.3 ± 0.7 (3.9) Trimers silanized FTO 13 688.5 709 20.5 52 13.2 17.9 6.4 285 0.5 − 26 4.5 ± 0.8 Trimers cell system 13 686.5 706 19.5 50 14 15.5 5.2 390 0.4 − 28 4.6 ± 0.7 Average transient absorption decay lifetime, t \n av , was calculated from the equation: t \n av = ( t \n 2 \n A \n 2 + t \n 3 \n A \n 3 \n ) / (A \n 2 + A \n 3 \n ) , where t \n i are the lifetimes and A \n i are the amplitudes (at 690 nm) of the two DAS components (Fig. 4 ). Bands’ minima were read out from the respective SAS (Fig. 4 ) and molecular lifetimes, t \n i , presented in Fig. 4 a were rewritten from Fig. 4 . Δ λ is the difference between wavelengths of the minima of red and bulk Chls’ SAS. Enthalpy difference (Δ H \n 0 ), free energy difference (Δ G \n 0 ), and effective number of red chlorophylls ( N \n r \n eff ) were calculated according to Eqs. S1–S4 (or Fig. S3). In the last column, values in the brackets are the numbers of red chlorophylls reported previously on the basis of time-resolved fluorescence (Szewczyk et al. 2017 ; indexes “*” and “#” stand for trimeric PSI with open and closed RCs, respectively). The uncertainty of molecular lifetimes necessary to estimate δ Δ G \n 0 was taken as ± 0.5 ps \n Transient absorption measurements—PSI immobilized on substrates—global analysis Figure 4 f, i, l shows the results of the global analysis obtained for trimeric PSI immobilized in three different systems—onto conductive surface of FTO glass, onto conductive surface of FTO glass covered with insulating silane layer, and onto conductive surface of FTO glass in the cell system, respectively (compare to Fig. 1 a, b, d). After PSI immobilization, the three-exponential character of the excitation dynamics is well preserved in all cases. The overall dynamics in all these systems is similar, in line with the results shown in Fig. 3 b, and can be described by three major DAS components of 0.4–0.7, 3.6–5.1, and 20–23.7 ps. The most noticeable effects of immobilization are changes in relative amplitudes between the second and the third component (~ 3–5 and ~ 20–24 ps, respectively; compare Fig. 2 c with Fig. 2 f, i, l). In general, greater contribution of the faster phase over the slower one together with acceleration of the third component after immobilization implies the acceleration of the overall excitation energy decay. Although this effect is partly compensated by increasing lifetime of the second component after immobilization (from 3.3 ps in solution to up to 5.1 ps on bare FTO), direct comparison of the kinetics (Fig. 3 ) and average lifetimes (Table 1 , t \n av ) demonstrates the overall acceleration effect. The same effect was observed for PSI monomers (Fig. S1 and Table S1 in supplementary information). The non-conservative shape of the second DAS component (3.6–5.1 ps) in the case of immobilized PSI (Fig. 4 f, i, l) suggests mixing of two processes: (1) energy transfer assigned to equilibration of bulk red Chls, the same as that for PSI in solution (Fig. 4 c) and (2) photochemical quenching of the excitation in the closed RC, absent in solution on this time scale. The fourth, slowest component is for each PSI-substrate sample characterized by much shorter lifetime (~ 300–500 ps) than in solution (5 ns). These results are in line with those obtained previously with time-resolved fluorescence (Szewczyk et al. 2017 ) and are discussed below. Transient absorption measurements—PSI immobilized on substrates—target analysis After immobilization of trimeric PSI onto substrates, few main changes relative to PSI in solution can be observed in target analysis results (Fig. 4 ). The first difference is shortening of the lifetime t \n 1 —from 18.3 ps in solution to 13–14 ps in immobilized PSI. The second one is the weaker coupling between bulk and red Chls reflected by increased values of t \n 2 and t \n 3 lifetimes: from 13.6/4.7 ps for PSI in solution to 22/7.9 ps, 17.9 /6.4 ps, and 15.5/5.2 ps for PSI on bare FTO, silanized FTO, and in the cell system, respectively. This effect is quite large for PSI-bare FTO sample, intermediate for PSI-silanized FTO, and weak for PSI in the cell system. The third effect is related to the positions of “Bulk” and “Red” SAS band maxima after PSI immobilization. In solution, the respective bands are at 685.5 and 706 nm. They are red-shifted to 689 and 710 nm for bare FTO and similarly to 688.5 and 709 nm for silanized FTO, respectively. Oppositely, the positions of SAS band maxima for the PSI in the cell system almost did not change in comparison to solution—they are at 686.5 and 706 nm. Also, the differences in amplitudes of SAS are firmly visible. Under “dry” conditions (bare and silanized FTO), the “Red” SAS shows strongly reduced amplitude (compare Fig. 4 e, h–b). In immobilized PSI complexes in the cell system (“wet” conditions), the “Red” SAS is also of reduced amplitude albeit to a lower extent (compare Fig. 4 k–b). This effect may indicate that the oscillator strength of the red Chls decreases if PSI is immobilized and densely packed. Similar effect was observed in the case of monomeric PSI (Fig. S1) and previously in aggregated (densely packed) LHCII particles in solution (Gruszecki et al. 2006 ). Immobilization of PSI in all systems causes similar changes in the lifetime and shape of the uncoupled Chls’ SAS (Fig. 4 e, h, k and Fig. S2C). The lifetime is shortened by one order of magnitude as noticed above (global analysis). Apart from main blue-shifted band with maximum at about 680 nm (present also in PSI in solution; Fig. 4 b) being a fingerprint of unconnected Chls (Melkozernov et al. 2000 ), an additional small band at about 705–710 nm can be distinguished. This observation suggests that after immobilization some of the uncoupled Chls undergo transition from “blue” to “red” form. All the described modifications of excitation dynamics induced by immobilization of PSI on the solid substrate are very much consistent with previously published data obtained using time-resolved fluorescence method (Szewczyk et al. 2017 ). To sum up, immobilization of PSI causes similar effects in all the systems under study although in the case of the cell system some of the features of SAS (Fig. 4 and Fig. S1) and also DAS (Fig. 4 ) are intermediate between those for PSI in solution and under the “dry” conditions (bare and silanized FTO). Estimation of the effective numbers of red Chls Trimeric PSI As shown previously (Szewczyk et al. 2017 ) and in the supplementary information, results of target analysis (molecular lifetimes and spectral positions of SAS; Fig. S3) may be used to estimate energetic parameters of bulk and red Chls (standard enthalpy difference, Δ H \n 0 , and standard free energy difference, Δ G \n 0 , between bulk and red Chls), and from those, effective number of bulk and red Chl states may be extracted. The results of these calculations as well as input data taken from the target analysis are shown in Table 1 for PSI trimers and in Table S1 for PSI monomers. The estimated effective number of red Chls per monomer in PSI trimers (4.3–4.6, Table 1 ) is independent of the system (PSI in solution and different systems with immobilized PSI), which is the same as that reported in the previous fluorescence studies (Szewczyk et al. 2017 ), from which also a similar number of 3.9–4.2 red Chls per monomer in trimeric PSI was estimated. The results shown in Table 1 consistently demonstrate that despite little spectral red-shift of red (and also bulk) Chls (see also Fig. S1) and the reduction of oscillator strength of red Chls (see above and Fig. 4 ), immobilization of PSI trimers influences neither the energetic parameters nor the numbers of red Chls. Monomeric PSI In the case of the monomeric PSI, the results of global and target analyses are generally similar to those for trimers (Fig. S1), except for some differences regarding the red Chls. The estimated effective number of red Chls was increased from 3.1 per monomer in solution to 4.3–5.3 for systems with immobilized PSI (Table S1). Similar tendency was observed in fluorescence studies where immobilization caused an increase in the effective number of red Chls from 3–3.4 in solution to 6.3. In both absorption and fluorescence studies, these changes are related to modifications of energetic parameters, in particular standard free energy difference between bulk and red Chls, Δ G \n 0 , whose absolute value decreases after immobilization. This observation confirms the previous hypothesis, that Chl–Chl interactions being most likely the origin of the extra red Chl states are more susceptible to modifications for PSI monomers than for trimers after immobilization/dense packing on the substrate (Szewczyk et al. 2017 ). Appearance of additional low-energy Chls, as a result of aggregation, was also observed previously in LHCII particles (Vasil’ev et al. 1997 ; Gruszecki et al. 2006 ; Andreeva et al. 2009 ; see also below). Origin of the acceleration of the excitation energy decay in immobilized PSI We have performed careful comparative analysis of PSI trimers and monomers immobilized onto different surfaces, conductive and non-conductive, dried, or being in contact with aqueous solution. In all cases, immobilization of PSI complexes caused similar acceleration of excitation decay within the protein. This result suggests that the acceleration of the PSI excitation decay is caused neither by electron injection into the substrate nor by drastically changed hydration state of the proteins, but most likely it is due to dense packing of PSI on the substrate. This conclusion is supported by earlier studies on a different pigment–protein photosynthetic complex, LHCII. For that system, accelerated excitation decay as well as the formation of a few red-shifted chlorophyll species as a result of aggregation, caused by low detergent concentration, was reported (Vasil’ev et al. 1997 ; Gruszecki et al. 2006 ; Andreeva et al. 2009 ). Furthermore, it was proposed that the origin of the new electronic low-energy levels is related to exciton coupling of protein-bound photosynthetic pigments (Gruszecki et al. 2006 ). Similar effects were observed by us: acceleration of the overall decay—for monomeric and trimeric PSI, and formation of additional red Chls—for monomeric PSI. Moreover, apart from the acceleration of overall excitation decay of Chls well coupled to RC occurring on 10–20-ps time scale (see t \n av values in Table 1 and S1), we also observed ~ 10-fold acceleration of excitation decay within a very minor pool of Chls uncoupled to RC: from ~ 5 ns in solution to ~ 500 ps after immobilization on the substrate. For these reasons, we propose that dense packing of PSI on the substrate resembles the “dense packing” of LHCII complexes within the aggregates, although the exact mechanism of the observed acceleration remains to be discovered. The above consideration was performed for hypothetical homogenous packing, although scenario in which different “clusters” of packed proteins are formed cannot be excluded. The parameters retrieved from the analysis would then represent average/mean values. Finally, we conclude that despite the described spectral and dynamic modifications of PSI complexes immobilized in different systems, these proteins remain fully functional in terms of excitation energy transfer."
} | 7,725 |
30687292 | PMC6333693 | pmc | 7,851 | {
"abstract": "Long-term continuous soybean cropping can lead to the aggravation of soil fungal disease. However, the manner in which the fungal community and functional groups of fungi are affected by continuous soybean cropping remains unclear. We investigated the fungal abundance, composition and diversity during soybean rotation (RS), 2-year (SS) and long-term (CS) continuous soybean cropping systems using quantitative real-time PCR and high-throughput sequencing. The results showed that the fungal abundance was significantly higher in CS than in SS and RS. CS altered the fungal composition. Compared with RS, SS had an increase of 29 and a decrease of 12 genera in fungal relative abundance, and CS increased 38 and decreased 17 genera. The Shannon index was significantly higher in CS and SS than in RS. The result of principal coordinate analysis (PCoA) showed that CS and SS grouped together and were clearly separated from RS on the PCoA1. A total of 32 features accounted for the differences in fungal composition across RS, SS, and CS. The relative abundance of 10 potentially pathogenic and 10 potentially beneficial fungi changed, and most of their relative abundances dramatically increased in SS and CS compared with RS. Our study indicated that CS results in selective stress on pathogenic and beneficial fungi and causes the development of the fungal community structure that is antagonistic to plant health.",
"conclusion": "Conclusion In this study, CS increased soil fungal abundance compared with the RS system and SS. The Shannon index indicated that both long- and short-term continuous soybean cropping could increase soil fungal diversity. Continuous soybean cropping influenced the fungal community structure in species composition and the abundance of fungi with a clear separation between continuous soybean cropping and the RS system on the PCoA1 axis. The separation between 2-year and long-term continuous soybean cropping on the PCoA2 axis indicated that the continuous cropping duration also affected the soil fungal community. The change in the fungal community structure was primarily driven by 14 genera, and Guehomyces, Alternaria , and Metacordyceps contributed more to the variation in the RS and 2-year and long-term continuous soybean cropping, respectively. The relative abundance of most of the potentially pathogenic and beneficial fungi, including potential fungi antagonistic to plant pathogens and the parasitic fungi of insect and plant parasitic nematodes, increased during 2-year and long-term continuous soybean cropping. Overall, this study suggested that CS increased the fungal abundance and diversity and changed the fungal community composition compared to the RS. In addition, CS can result in the development of a fungal community structure that is antagonistic to plant health.",
"introduction": "Introduction Soil microbes play a vital role in the agroecosystem, because they are involved in nutrient cycling, organic matter decomposition and soil-borne disease development ( Heijden et al., 2008 ). Continuous or monoculture cropping, a practice of repeated cultivation of the same crop for multiple years, is a common agricultural practice over the world. During long-term continuous soybean ( Glycine max Merr.) cropping, the crop roots secrete the same exudates for a long time, and some of the crop root secretions such as phenolic acids can alter the microbial community structure and result in an increase in the abundance of pathogenic fungi ( Guo et al., 2011 ; Wang et al., 2013 ). Due to the industrial demand and the cultivation habits of farmers, CS has increased in recent years which often causes significant decline in soybean yields and quality as a consequence of severe occurrences of soil-borne diseases ( Liu and Herbert, 2002 ). Therefore, understanding how CS affects the soil fungal community may help in developing practices to relieve soil-borne diseases and obstacles associated with soybean cultivation. Soil fungi can be designated as harmful or beneficial groups based on their functions in agroecosystem. Some fungi can infect crops and cause plant diseases, while others can suppress pathogenic fungi to the plant, benefit plant growth or decompose plant residues ( Setälä and McLean, 2004 ). Continuous cropping can change the fungal composition; for instance, continuous soybean cropping increased the population of Fusarium and tended to increase the susceptibility to root rot ( Jie et al., 2015 ). Compared with soybean and maize ( Zea mays L.) rotations, CS decreased the populations of Trichoderma and Gliocladium that may play roles in the biocontrol of soil-borne pathogens ( Meriles et al., 2009 ; Pérez-Brandán et al., 2014 ). Previous studies in the same fields found that the populations of Fusarium spp. and Heterodera glycines during 20-year continuous soybean cropping were significantly lower than during 3-year continuous soybean cropping, while the populations of Pochonia chlamydosporia, Paecilomyces lilacinus , and Pseudomonas fluorescens were higher than those in the 3-year continuous soybean cropping and the rotation of soybean/maize/wheat ( Triticum aestivum L.) ( Wei et al., 2015 ). The populations of Fusarium equiseti, Pochonia chlamydosporia and Purpureocillium lilacinum that were isolated from parasitized cysts during 21-year continuous soybean cropping were distinctly larger than those during 3-year continuous soybean cropping ( Song et al., 2017 ). The percentage of Pochonia chlamydosporia and Purpureocillium lilacinum that were isolated from the cysts of the soybean cyst nematodes during the 23-year continuous soybean cropping were notably higher than that during the SS ( Song et al., 2016 ). However, the general changes of the fungal composition and potentially pathogenic and beneficial fungi have yet to be addressed during CS. Recently, a growing number of studies have focused on the effect of long-term continuous cropping on the soil microbial community structure using different techniques ( Vargas et al., 2011 ; Li et al., 2014b ; Wu et al., 2015 ). A denaturing gradient gel electrophoresis (DGGE) analysis showed an obvious difference in the DGGE bands between the continuous and the rotation soybean cropping systems ( Li et al., 2010 ). The phospholipids fatty acid (PLFA) signatures indicated that the microbial community structure of the rhizosphere changed within the span of continuous soybean cropping, and the daidzein and genistein released by the soybeans had a significant influence on the fungal community ( Guo et al., 2011 ). The terminal restriction fragment length polymorphism (T-RFLP) profiles demonstrated clear differences in the relative abundance, logarithmic transformation and Bray-Curtis dissimilarity matrix of the fungal community during a three-year gradient continuous Rehmannia glutinosa cropping ( Wu et al., 2015 ). Using 454 pyrosequencing, Li et al. (2014b) found that continuous peanut ( Arachis hypogaea Linn) cropping caused a significant accumulation of the pathogenic fungi Fusarium oxysporum, Phoma and Bionectria ochroleuca , while the relative abundance of the beneficial fungi Trichoderma and Mortierella elongate was significantly decreased. Because of the limitation of techniques in previous studies, only certain dominant microbial groups were detected in continuous soybean cropping. It is necessary to conduct systematic and comprehensive research on the change of the soil fungal community structure in CS. With the development of high throughput sequencing technology, abundant information of soil microbial structure, diversity and function can be obtained, which can be used to thoroughly investigate changes in soil fungal community. Thus, in this study, the main hypotheses were that the CS altered the fungal community structure and functions. We used high-throughput sequencing methods to examine the fungal composition and functions during RS, 2-year short-term and 27-year long-term continuous soybean cropping systems in northeast China. The objectives of this study were (1) to reveal the fungal abundance and community composition in the three different soybean cropping systems; (2) to compare the changes of fungal diversity among the three different soybean cropping systems, and (3) to investigate the changes of potentially pathogenic and beneficial fungi in the three different soybean cropping systems. Understanding the effects of continuous soybean cropping on fungal community will provide theoretical guidance for development of improved agricultural management strategies.",
"discussion": "Discussion Shifts in Fungal Abundance and Community Composition in Response to Long-Term Continuous Soybean Cropping In this study, CS led to an increase in soil fungal abundance compared with the RS system and SS (Figure 1 ). This is consistent with another study which showed that the cultivable fungal population in the continuous soybean cropping was higher than that in the RS and that the imbalance of soil microbial flora occurred in the continuous cropping soil ( Meriles et al., 2009 ). Zhou and Wu (2012) used q-PCR analysis and found that continuous cucumber cropping increased the fungal abundance in the rhizosphere. The increased fungal abundance may be related to the root exudates and different response of the microbes to root exudates or rhizodepositions. In this study, soybean roots exuded a number of flavonoids and phenolic acids ( Colpas et al., 2003 ; Guo et al., 2011 ), which were lower in the wheat and maize root exudates of the RS or SS. It was found that flavonoids and phenolic acids had a significant effect on the fungal community, especially fungal pathogens ( Qu and Wang, 2008 ; Wang et al., 2013 ). Overall, our study suggested that CS promotes soil fungal growth. The effects of CS on the soil fungal community composition were distinctly observed at the phylum and genus level. At the phylum level, Ascomycota was the most abundant, and its relative abundance increased in SS and CS compared to RS (Figure 2A ), suggesting the ubiquity and important role of Ascomycota in the soybean ecosystems. This is consistent with a previous study that found that Ascomycota was the most important phylum in continuous soybean cropping soils ( Li et al., 2010 ). Both SS and CS enhanced the relative abundance of Zygomycota (Figure 2A ), which is a saprophytic phylum and plays an important role in decomposing plant debris ( Richardson, 2009 ). This agrees with the findings of ( Li Z. et al., 2016 ), who found that the relative abundance of Ascomycota, Zygomycota, and Glomeromycota increased with continuous Piper nigrum L. (black pepper) cropping duration. At the genus level, our results demonstrated that CS and SS enhanced the relative abundance of some genera, such as Fusarium, Humicola, Alternaria, Clonostachys , and Mortierell a (Figure 2B ). A previous study also reported that short-term continuous soybean cropping caused an increase in the relative abundance of Fusarium, Humicola , and Alternaria ( Bai et al., 2015 ). Continuous cropping of other plants, such as Pseudostellaria heterophylla , also increased the relative abundance of Fusarium, Clonostachys and Mortierella ( Wu et al., 2016 ). Our results indicated that CS affected the fungal composition and altered the dominant genera. The causes of variation in the microbial community composition are multifaceted. For example, they can include soil nutrient imbalances, soil physical and chemical property deterioration and allelopathic autotoxicity accumulation ( Kong et al., 2008 ; Li et al., 2010 ). However, many studies indicated that the changes of the microbial community composition are caused by the indirect ecological effect of root exudates rather than direct allelopathic autotoxicity ( Guo et al., 2011 ; Wu et al., 2015 ). Owing to the different components of root exudates, some of them may promote the reproduction of certain microorganisms, such as soil-borne pathogens, while others may benefit the reproduction of beneficial microbes, such as antagonistic microorganisms in this manner, resulting in an increase of corresponding chemotaxis microbes ( Li et al., 2014a ). Fungal Diversity in Response to Long-Term Continuous Soybean Cropping An increase in the Shannon index was observed in SS and CS, suggesting that continuous soybean cropping increased the soil fungal diversity (Figure 3A ). Bai et al. (2015) and Xiong et al. (2015) also reported that the continuous cropping of soybean and vanilla ( Vanilla planifolia ) dramatically enhanced the fungal diversity. However, there are also contrasting findings that the fungal community diversity was not different between the continuous cropping of soybean/potato ( Solanum tuberosum L.) and their corresponding rotation systems ( Li et al., 2010 ; Liu et al., 2014 ). This might due to inconsistent technology or crop and soil environments, since different crops have diverse root exudates that affected the microbial community ( Rengel, 1997 ). In this study, Bray_Curtis and Binary_Euclidean were used to evaluate the beta-diversity of the fungal community. The Bray_Curtis considered both species composition and the fungal abundance, while the Binary_Euclidean only considered fungal species composition. The Bray and Binary principal coordinate analyses indicated that continuous soybean cropping resulted in the greatest impact on the variation of the fungal community structure. Along the PCoA1, both Bray_Curtis and Binary_Euclidean showed a clear separation between the continuous soybean cropping and the RS system, indicating that continuous soybean cropping changed the fungal community structure both in species composition and the abundance of fungi (Figures 4A,B ). However, 2-year and long-term continuous soybean cropping were separated on the PCoA2 axis, suggesting that the soil fungal communities are influenced by the duration of continuous cropping. This finding is generally consistent with the recent investigation in continuous peanut cropping systems, which demonstrated that the fungal community composition and structure of 5-year and 10-year continuous peanut were similar compared to 1 year of continuous peanut cropping, but 5-year and 10-year continuous peanut cropping were also separated on the PC2 axis of the principal component analysis ( Li et al., 2014b ). This is likely due to the differential accumulation in the amount of crop root exudates and residues in the rhizosphere microenvironment with continuous cropping duration, since crop root exudates and residues can influence the microbial community structure by serving as substances for microbes. Subsequent LEfSe analysis showed that the change in the fungal community structure was primarily driven by 14 genera, among which Guehomyces, Alternaria , and Metacordyceps played an important role in RS, SS, and CS, respectively (Supplementary Figure S1 ). The Pearson correlation analysis between the soybean yield and fungal community diversity showed a negative relationship between the soybean yield and the Shannon index and the PCoA1 of the Bray and Binary analyses (Table 1 ). The Bray_Curtis considered both species composition and the fungal abundance. Thus, this result suggested that the soybean productivity is negatively correlated with the fungal species composition and abundance. Changes in Potentially Pathogenic and Beneficial Fungi in Response to Long-Term Continuous Soybean Cropping This study found that the relative abundance of potentially pathogenic fungi significantly increased in SS and CS, indicating continuous soybean cropping benefits the proliferation of specific pathogenic fungi. Our finding demonstrated that SS and CS significantly increased the relative abundance of Fusarium and Alternaria (Figure 6 ). Fusarium is well known to be an pathogenic fungus that leads to soybean Fusarium root rot ( Zhang et al., 2010 ; Chang et al., 2018 ), and Alternaria can infect various crops and cause corresponding diseases, such as soybean Alternaria leaf spot, tomato and carrot black rots, citrus fruit gray rot and cereal black point ( Logrieco et al., 2009 ; Li and Yang, 2009 ). In our study, Alternaria also showed a positive correlation with the soybean root disease incidence. This finding suggested that continuous soybean cropping could exacerbate crop disease. This is consistent with a previous study that the dominant pathogenic fungi were Fusarium and Alternaria in soybean fields, and their abundance was higher in 1-year and 2-year continuous soybean cropping than in RS soil ( Bai et al., 2015 ). Fusarium showed a significantly positive correlation with the ECs but non-significant correlation with soybean root disease incidence (Table 2 ). This may be due to the complexity of soybean root rot, caused by a variety of pathogens; Fusarium is a pathogen but can also infect the cysts or eggs of soybean cyst nematodes ( Song et al., 2017 ). In this study, the relative abundance of Volutella, Cylindrocarpon , and Ganoderma was more enriched in SS and CS and much less abundant in the RS (Figure 6 ), especially Ganoderma was significantly positively correlated with the soybean root disease incidence (Table 2 ). These three fungi were designated to potentially pathogenic fungi based on FUNGuild and relative literature 6 ( Gilbertson and Ryvarden, 1987 ; Tedersoo et al., 2014 ). Some fungi of Volutella , such as V. colletotrichoides , were described in diseased alfalfa ( Medicago sativa L.) and other forage legumes in Iowa ( Cannon et al., 2012 ). Species of the fungus Cylindrocarpon , such as C . destructans var. destructans , can cause black root rot of Sanqi ( Panax notoginseng ) and grapevine ( Vitis sp.) ( Alaniz et al., 2007 ; Mao et al., 2014 ). Ganoderma , such as G . boninense , is a pathogen of oil palm ( Elaeis guineensis ) ( Najihah et al., 2015 ). Both Boeremia and Lectera were not detected in RS, while their relative abundances were dramatically increased in SS and CS (Figure 6 ), and they were positively correlated with the soybean root disease incidence (Table 2 ). Some studies have reported that Boeremia caused stem rot of Origanum dubium in Oregano ( Origanum dubium Boiss) and black rot of artichoke ( Cynara scolymus ) in California ( Koike et al., 2016 ; Samouel et al., 2016 ). Lectera , a new genus of Plectosphaerellaceae , is a legume pathogen ( Cannon et al., 2012 ). The relative abundance of most of the potentially pathogenic fungi increased in SS and CS compared to RS. However, there were relative abundances of some potentially pathogenic fungi, such as Ustilago, Bipolaris and Sarocladium , which decreased in SS and CS (Figure 6 ). This is likely due to the tendency of Ustilago, Bipolaris and Sarocladium to infect graminaceous crops. Therefore, their relative abundances decreased during continuous soybean cropping but increased in a RS system of wheat/maize/ soybean. Many studies have reported that these three fungi cause diseases of graminaceous crops, such as rice sheath rot caused by Sarocladium oryzae ( Saravanakumar et al., 2009 ), maize smut disease caused by Ustilago maydis and maize leaf spot caused by Bipolaris ( Kämper et al., 2006 ; Li G.F. et al., 2016 ). The above findings can also explain the negative correlation between Ustilago and the soybean root disease incidence (Table 2 ). Our finding also suggested that continuous soybean cropping duration results in selective pressure on some pathogenic fungi of crop. Biocontrol microbes have been proposed to be a sustainable alternative to chemical control ( Lecomte et al., 2016 ). In this study, the relative abundance of the most potentially beneficial fungi exhibited an increased trend in SS and CS (Figure 7 ). These beneficial fungi included fungi antagonistic to plant pathogen and parasitic fungi of insect and plant parasitic nematodes. Some species of Mortierella and Clonostachys are fungal antagonists to plant pathogens, and they can protect banana ( Musa sp.) and soybean from Fusarium wilt disease and Fusarium root rot ( Pan et al., 2013 ; Shen et al., 2018 ). Hamid et al. (2017) also found that Clonostachys was abundant during continuous soybean cropping and gradually increased with the duration of continuous cropping. Some species of Metacordyceps, Metarhizium , and Beauveria genera are parasitic fungi of insects, and they also have potential value as pharmaceuticals ( Roberts and St. Leger, 2004 ; Kepler et al., 2012 ). A recent study demonstrated that Beauveria bassiana had negative effects on cotton ( Gossypium hirsutum ) aphid reproduction ( Lopez et al., 2014 ). The fungi Hirsutella, Purpureocillium and Pochonia are parasitic fungi of plant parasitic nematodes. A previous study reported that Hirsutella combined with chitosan suppressed the infestation of soybean cyst nematodes in soybean roots ( Mwaheb et al., 2017 ). Purpureocillium is not only a well-known beneficial agent against various plant pathogenic in agriculture but also a commercialized agent to control plant parasitic nematodes ( Wang et al., 2016 ). Hamid et al. (2017) reported that the relative abundance of Purpureocillium gradually increased with the duration of soybean monoculture, and they also detected Hirsutella in CS soils. In this study, the relative abundance of Pochonia decreased in SS and CS compared with the RS (Figure 7 ). Song et al. (2016) found that the percentage of Pochonia chlamydosporia was significantly higher in CS system than in a RS system. These inconsistent results might be due to different experimental methods and the isolated positions of the fungi. Song et al. (2016) used the traditional method to isolate fungi from the cysts of the soybean cyst nematodes, while we used the high-throughput sequencing technique to analyze the relative abundance and composition of the fungi from the soybean field soils. Current results revealed that the relative abundance of Penicillium gradually decreased in RS, SS, and CS (Figure 7 ). This is consistent with the previous study of Wu et al. (2016) who found that continuous Pseudostellaria heterophylla cropping markedly reduced the relative abundance of Penicillium . Although some Penicillium species can cause crop diseases, many species of Penicillium have been found to be beneficial fungi, which can inhibit the growth of pathogenic organisms to control plant diseases ( Van Wees et al., 2008 ). In this study, the relative abundance of Clonostachys, Metarhizium and Hirsutella significantly increased in CS compared to SS (Figure 7 ), which indicated that these potentially beneficial fungi were sensitive to continuous soybean cropping duration. The Pearson correlation analysis demonstrated that Purpureocillium had positive correlation with the soybean root disease severity and the FC (Table 2 ), which suggested that the potentially beneficial fungi will increase accompanying the increase of pathogenic agents. The Mortierella, Metacordyceps, Clonostachys, Metarhizium , and Hirsutella were positively correlated with the ECs (Table 2 ). This is likely due to the parasitism of these potential beneficial fungi on the cysts or eggs of the soybean cyst nematodes ( Song et al., 2017 ). This result is in agreement with previous study that Hirsutella can suppress soybean cyst nematodes ( Mwaheb et al., 2017 ). Overall, these findings suggested that CS can increase the both pathogenic and beneficial fungi of plant and lead to an antagonistic development of the fungal community structure on plant health."
} | 5,931 |
38315845 | PMC10873592 | pmc | 7,852 | {
"abstract": "Significance Interspecies interactions play a crucial role in ecological community assembly, but we still lack a quantitative perspective on how they influence the fitness of each species in complex communities. This knowledge gap arises from our limited understanding of how interspecies interactions present in diverse communities differ from those studied in a pairwise manner. To address this, we quantified 1,344 interspecies interactions among seven bacterial strains and found that the changes of pairwise interactions follow consistent patterns. These patterns could be learned from trio combinations, highlighting the importance of observing simple beyond-pairwise combinations to capture actual interspecies relationships. This work illustrates how we can deal with the enormous complexity of interspecies interactions toward improved understanding and management of multispecies ecosystems.",
"discussion": "Discussion Incorporating three-member combinations enabled a reasonable prediction of interspecies interactions that are actually present in the diverse microbial communities. This is probably because trio combinations can capture higher-order effects that cannot be informed by pairwise coculturing results. In addition, including combinations of four, five, and six members only marginally improved prediction accuracy compared to the trio-based predictions, probably because higher-order effects on each pairwise interaction follow consistent patterns that can largely be captured by trio combinations, the smallest unit of higher-order effects. Indeed, our trio-based prediction of the six- or seven-member SBC structure yielded similar accuracy (Bray–Curtis dissimilarity of ca. 0.1 to 0.25) to the non-bottom–up deep-learning model trained on diverse community structure data (ca. 0.02 to 0.28 for 5-member SBC and ca. 0.06 to 0.67 for real human gut microbiota) ( 40 ). Predicting microbial community poses a significant challenge that has been pursued based on pairwise interactions ( 9 , 41 ) and leave-one-out communities ( 14 ). In this study, we demonstrate that learning higher-order effects from simple beyond-pairwise combinations, such as trios, is a viable approach that quantitatively links the community structure and interspecies interactions. While observing all beyond-pairwise combinations within diverse communities is unfeasible, we suggested that implementing the proposed bottom–up community prediction is achievable even with limited data input. Thus, by leveraging high-throughput culture techniques and evaluating hundreds or thousands of combinations, it could become possible to gain improved insight into the assembly of real microbial communities in diverse environments, including gut, soil, and marine ecosystems. While existing analytical methods for studying microbial communities focus either on individuals or the community as a whole (e.g., single-cell and omics analyses), adopting beyond-pairwise combinations as the unit of observation would provide a fresh perspective to the field. To unlock the full possibility of this approach, the development of data collection techniques for massive microbial combinations and the advancements in data analytical methodologies are imperative. Recent theoretical studies have increasingly focused on higher-order interactions and their critical roles in species coexistence and community stability ( 31 – 34 ). Although empirical studies examining these aspects have been scarce (but see refs. 17 , 22 , and 32 ), our dataset of 1,344 interaction coefficients could provide unique insights. In particular, we found that the majority of the remarkable changes in the pairwise relationships were not solely induced by specific third-party species, but were rather triggered by multiple species across different taxa. Consequently, each pairwise interaction exhibited a tendency to change in the same direction, with limited regard for the community composition and diversity. This finding provides a simple intuition that the occurrence of higher-order effects is mostly explained as the property of pairwise interactions, such as being robust or being easily weakened, rather than as the sporadic phenomena involving a specific set of ≥3 members. In addition, we have demonstrated a clear trend where higher-order interactions tend to counterbalance pairwise interactions, preventing species overabundance or extinction. Although most theoretical models treat higher-order interactions as independent variables from pairwise interactions and other factors (e.g., background abundance of affected species) ( 32 , 38 , 39 ), aligning these empirical knowledges with theoretical perspectives could catalyze an improved understanding of the ecological community assembly. Finally, we outline potential factors to consider when extending our results and the proposed approach to different ecosystems. First, since this study focused on SBCs dwelling on the surface of duckweed, the observed microbial interactions possibly include indirect interactions via the effects on the host physiology, similar to the related studies on gut microbiota ( 17 , 22 , 41 ). Although our model does not consider this fact, it cannot be ruled out that some host-specific factors may have influenced the properties of observed microbial interactions. Second, the budding-based growth of duckweed continuously provides new space and substrates for the microbes, which results in a deterministic SBC structure that is largely independent of the inoculation order and initial abundance of the SBC members ( 26 , 27 ). This suggests that some key factors of the community assembly, especially regarding temporal aspects (e.g., priority effect) and stochasticity ( 42 , 43 ), were not considered in our analyses. Last, the seven bacterial strains used in this study belong to relatively distant taxonomic groups (different families) that constantly coexist in natural duckweed microbiomes ( 44 – 46 ). As a result, the investigated interspecies interactions may be biased toward the moderate ones, rather than exclusive competition among species vying for the same niche. These facts emphasize the importance of conducting relevant investigations in several other ecosystems and evaluating the viability of the bottom–up approach under various temporal, spatial, and taxonomic contexts."
} | 1,588 |
37820728 | null | s2 | 7,854 | {
"abstract": "Transferable plasmids play a critical role in shaping the functions of microbial communities. Previous studies suggested multiple mechanisms underlying plasmid persistence and abundance. Here, we focus on the interplay between heterogeneous community partitioning and plasmid fates. Natural microbiomes often experience partitioning that creates heterogeneous local communities with reduced population sizes and biodiversity. Little is known about how population partitioning affects the plasmid fate through the modulation of community structure. By modeling and experiments, we show that heterogeneous community partitioning can paradoxically promote the persistence of a plasmid that would otherwise not persist in a global community. Among the local communities created by partitioning, a minority will primarily consist of members able to transfer the plasmid fast enough to support its maintenance by serving as a local plasmid haven. Our results provide insights into plasmid maintenance and suggest a generalizable approach to modulate plasmid persistence for engineering and medical applications."
} | 276 |
16553966 | PMC1435936 | pmc | 7,855 | {
"abstract": "Background Successful realization of a \"systems biology\" approach to analyzing cells is a grand challenge for our understanding of life. However, current modeling approaches to cell simulation are labor-intensive, manual affairs, and therefore constitute a major bottleneck in the evolution of computational cell biology. Results We developed the Genome-based Modeling (GEM) System for the purpose of automatically prototyping simulation models of cell-wide metabolic pathways from genome sequences and other public biological information. Models generated by the GEM System include an entire Escherichia coli metabolism model comprising 968 reactions of 1195 metabolites, achieving 100% coverage when compared with the KEGG database, 92.38% with the EcoCyc database, and 95.06% with iJR904 genome-scale model. Conclusion The GEM System prototypes qualitative models to reduce the labor-intensive tasks required for systems biology research. Models of over 90 bacterial genomes are available at our web site.",
"conclusion": "Conclusion The GEM System facilitates systems biology research by prototyping a metabolic pathway simulation model from a genome. Given a complete genome, all modeling procedures are automated with configurable options, generating stoichiometric models in SBML format that are readily usable by cell simulators. In comparison with the KEGG organism-specific databases, the qualitative modeling step has high accuracy, with few false positives and negatives. More than 90 models generated from complete bacterial genomes are available for download online, with visualized pathway maps and gene lists.",
"discussion": "Discussion We have developed the GEM System, automated software for the rapid construction of draft simulation models of cell-wide metabolic pathways from genome sequence information by integration of public biological databases. Automatic generation of the models is currently limited to metabolism in bacteria, and depends on the availability of EC numbers in public databases, but we have shown that qualitative models of the metabolic pathways of bacteria can be generated with low false positives and negatives, as validated by the comparison with KEGG, EcoCyc, and Reed et al .'s model. Although the generated models are draft models and thus still require expert curation to ensure the accuracy of simulations, manual involvement is minimized. There are, however, several limitations of this approach. Firstly, although EC numbers are generally effective for enzyme data representation for well known pathways, certain number of reactions have no EC number assigned, and therefore majority of the transporters are identified as genes but not included as reactions in GEM System, making a large fraction of the model different from iJR904. Secondly, some EC numbers are incomplete and therefore ambiguous, and some become quickly obsolete, being assigned to new EC numbers. This resulted in more than 40 inconsistent enzyme assignments in GEM System. Thirdly, since the GEM System identifies enzymes and the corresponding reactions based on the genome information, it cannot identify reactions that are experimentally observed but with no corresponding gene found. To overcome these problems, more general nomenclature for enzymes should be used in addition to the EC numbers and integrate necessary information that have no link to the gene sequences. The system generates a stoichiometric simulation model in SBML format, which is readily applicable to flux-based analyses on a number of simulation platforms. The stoichiometric models can be used for metabolic flux analyses by supplying experimental data for exchange fluxes as reported elsewhere [ 33 , 34 ]. One potential application of GEM System using this stoichiometric matrix is for dynamic large-scale simulation of metabolic pathways with hybrid dynamic/static simulation method [ 35 ]. Using this method, quasi-dynamic simulation is achieved by subdividing the model into multiple \"static modules\" connected by \"dynamic modules\", and by calculating the flux distribution of static modules using the stoichiometry and boundary flux of the dynamic module that is modeled with traditional enzyme kinetics methods. In this way, necessary kinetic equations and parameters are significantly reduced while maintaining simulation accuracy. Most reactions with high elasticities can be included in the static module, for which the stoichiometric matrix generated by the GEM System is directly applicable. The GEM System can generate models automatically from public databases, but can also utilize private databases if such experimental data becomes available. Mining of high-throughput data by bioinformatics may facilitate the quantitative modeling step; for example, it should be possible to take advantage of recent progress in \"metabolomics\". Once genome-wide metabolome data becomes available via high-throughput techniques such as the capillary electrophoresis – electrospray ionization – mass spectrometry (CE-ESI-MS) method, metabolome data can be used to add unknown pathways, to supply the initial values of the metabolites, and to optimize kinetic parameters. Parameter fitting of time-series metabolite concentration data to general dynamic equations such as Generalized Mass Action is a possible substitution for accurate kinetic modeling, at least in the given time frame of the data set used for parameter optimization. Our next step is to model the gene expression layer, including transcription, translation, and degradation processes. The GEM System is a powerful platform for this purpose, in no small part because the genome-based approach enables a link to databases of different fields based on the nucleotide sequences already described. Because the GEM System has been based on a generic bioinformatics workbench, that is, the G-language Genome Analysis Environment [ 36 ], the system can directly access genome sequences and perform computational genome data-mining. Required parameters or information such as the structure of a promoter can be directly obtained from the genome sequence as the simulation takes place. In this respect, GEM System can be extended to be applicable for the modeling of eukaryotes, by identifying protein subcellular localizations from database reference and with predictable methods [ 37 , 38 ]. Although the parameters in the functional annotation process should be revised to cope with the information availability and the existence of a multitude of duplicate gene paralogs, by selecting tissue specific gene expression pattern with expressed sequence tags (EST) or microarray data, the general approach of the GEM System should also be applicable for tissue specific cellular models of higher eukaryotes. In sum, the rapid accumulation of biological information now allows the realization of integrative systems biology, but at the same time makes manual modeling unrealistic; therefore, a genome-based automatic modeling procedure is a crucial step forward for the grand challenge to construct life in a computer."
} | 1,760 |
38645395 | PMC11026606 | pmc | 7,857 | {
"abstract": "Understanding belowground plant-microbial interactions is important for biodiversity maintenance, community assembly and ecosystem functioning of forest ecosystems. Consequently, a large number of studies were conducted on root and microbial interactions, especially in the context of precipitation and temperature gradients under global climate change scenarios. Forests ecosystems have high biodiversity of plants and associated microbes, and contribute to major primary productivity of terrestrial ecosystems. However, the impact of root metabolites/exudates and root traits on soil microbial functional groups along these climate gradients is poorly described in these forest ecosystems. The plant root system exhibits differentiated exudation profiles and considerable trait plasticity in terms of root morphological/phenotypic traits, which can cause shifts in microbial abundance and diversity. The root metabolites composed of primary and secondary metabolites and volatile organic compounds that have diverse roles in appealing to and preventing distinct microbial strains, thus benefit plant fitness and growth, and tolerance to abiotic stresses such as drought. Climatic factors significantly alter the quantity and quality of metabolites that forest trees secrete into the soil. Thus, the heterogeneities in the rhizosphere due to different climate drivers generate ecological niches for various microbial assemblages to foster beneficial rhizospheric interactions in the forest ecosystems. However, the root exudations and microbial diversity in forest trees vary across different soil layers due to alterations in root system architecture, soil moisture, temperature, and nutrient stoichiometry. Changes in root system architecture or traits, e.g. root tissue density (RTD), specific root length (SRL), and specific root area (SRA), impact the root exudation profile and amount released into the soil and thus influence the abundance and diversity of different functional guilds of microbes. Here, we review the current knowledge about root morphological and functional (root exudation) trait changes that affect microbial interactions along drought and temperature gradients. This review aims to clarify how forest trees adapt to challenging environments by leveraging their root traits to interact beneficially with microbes. Understanding these strategies is vital for comprehending plant adaptation under global climate change, with significant implications for future research in plant biodiversity conservation, particularly within forest ecosystems.",
"conclusion": "7 Conclusions In conclusion, this review summarizes the general changes in the relative abundance and activity of soil microbial communities, alongside alterations in root traits and metabolites across soil horizons, and their associations with climatic drivers of change. We have concluded that different soil horizons, root characteristics, and environmental conditions collectively influence microbial communities in forest ecosystems. It explains that various soil layers, each with distinct physical and chemical properties, harbor specific microbial populations. The diversity in microbial life is largely dependent on the soil’s organic matter content and nutrient availability, with particular emphasis on the distinct distributions of fungi such as ectomycorrhizal and saprotrophic species across different soil horizons. Root exudates, significantly influenced by root morphology including diameter, tissue density, and specific root length, are crucial in shaping the composition and activity of these soil microbial communities. Furthermore, the review underscores the role of environmental factors like drought, temperature, and nutrient availability in altering root traits and exudation patterns. These environmental changes, in turn, affect the structure and function of microbial communities. Stated simply, linking rhizospheric microbial interactions to root morphology and metabolomics may help us to reserve below ground resources for maintenance and enhancement of natural ecosystem services in the context of climate change. Furthermore, this review explains the adaptive strategies of plants by harnessing root trait plasticity and microbial interactions under global climate change scenarios. However, there remain some open questions regarding the environmental effects on plant traits and their implications for the rhizosphere microbiome in forest ecosystems. These questions are particularly pertinent to plant functional traits concerning the functions of root metabolites, changes in root morphology, and their relationships to microbial functional groups.",
"introduction": "1 Introduction Forest ecosystems, which cover one-third of the world’s land area and encompass more than 24.6 million square kilometers, are crucial for global ecosystem services and carbon cycling, particularly under the influence of climate change ( Viet et al., 2022 ). These ecosystems, predominantly temperate and boreal forests, consist of diverse deciduous and coniferous trees that are well-adapted to a broad range of environmental conditions, especially in terms of temperature and precipitation ( Tasenkevich et al., 2023 ). The soils of these forests support a rich array of prokaryotic and eukaryotic microbial communities, including bacteria and fungi. These organisms are essential for maintaining ecosystem health and are key players in ecosystem carbon and nutrient cycling ( Bardgett et al., 2014a ; Baldrian, 2017 ). Some microbes are particularly adept at decomposing refractory soil organic matter (SOM) and cellulose, a capability linked to their filamentous growth and ability to excrete extracellular enzymes ( Bodeker et al., 2014 ; Gómez et al., 2020 ). Furthermore, these forest soils are home to diverse communities of saprotrophic and mycorrhizal fungi, integral to the ecosystem’s functioning and heavily reliant on the interactions within the belowground rhizosphere ( Voříšková et al., 2014 ). Understanding these complex microbial interactions and the dynamics of below-ground carbon allocation via root exudation is vital for predicting carbon balance in terrestrial ecosystems, especially in the face of ongoing global changes ( DeAngelis et al., 2015 ). Fine roots of forest trees, highly active metabolic zones within the soil, are key in the interactions with subterranean microorganisms ( \n Figure 1 \n ) ( King et al., 2023 ). The communication between plants and soil microbes is predominantly mediated through chemical signals, as indicated by Pascale et al. (2020) and Williams and de Vries (2020) . Root systems of forest trees excrete various chemical compounds, collectively known as metabolites, which are essential for diverse biological processes ( \n Figure 1 \n ). These metabolites encompass primary metabolites such as amino acids, organic acids, and sugars, along with secondary metabolites like phenolic chemicals, and volatile compounds including terpenoids and sulfides. Such root exudations of forest trees have a significant impact on numerous ecosystem functions. For example, they affect soil microbial dynamics ( Jing et al., 2023 ), enhance plant resilience to non-living stress factors ( Williams and de Vries, 2020 ; Jiang et al., 2023 ), and contribute to nutrient cycling and the stability of soil structure ( Mommer et al., 2016 ; Wang et al., 2021a ; Wang et al., 2021b ). Additionally, roots release various other metabolites, accounting for 10–50% of the carbon fixed by plants ( Nguyen, 2003 ; Massalha et al., 2017 ; Prescott and Grayston, 2023 ), thereby playing an integral role in the global carbon cycle. Figure 1 General overview of the metabolic niche of the rhizosphere microbial functional group in forest ecosystems. Different soil horizons differ in the activity of ectomycorrhizal fungi (ECM), arbuscular mycorrhizal fungi (AM), saprotrophic fungi, and bacterial communities as well as exudation of root metabolites and nutrient availability, demonstrated by currently available research. Microbial diversity and abundance in the rhizosphere are affected by plant developmental stage, genotypes, and soil parameters ( Ofek-Lalzar et al., 2014 ; Schlaeppi et al., 2014 ; Bulgarelli et al., 2015 ; Paredes and Lebeis, 2016 ). For instance, the main soil parameters, e.g. changing soil organic matter distribution with soil depth, influence root exudation as well as microbial diversity and functioning in temperate forest ecosystems ( Tückmantel et al., 2017 ; Prada-Salcedo et al., 2022 ) ( \n Figure 1 \n ). However, this assemblage of microbes in different soil horizons may change due to the changes in root morphology and exudation of forest trees along a wide range of environmental gradients, for example drought or temperature ( Meier et al., 2020 ; Veach et al., 2020 ). It has even been hypothesized that plants dynamically recruit soil microbes by secreting metabolites in the rhizosphere that ideally stimulate rhizosphere microorganisms or endophytic fungi and bacteria that are advantageous to plant growth by helping them cope with abiotic stresses ( Sasse et al., 2018 ; Hildebrand et al., 2023 ). Knowledge about the responses of root metabolites and microbial assemblages in the roots and rhizosphere in different horizons of forest soil along environmental gradients of drought and temperature will increase our aptitude to forecast the global change effects on soil organic matter decomposition and C cycling ( Steidinger et al., 2019 ). Therefore, a better understanding of rhizosphere microbial interactions and below ground C allocation via root exudation is important for predicting C balance in the terrestrial ecosystem under global change scenarios. In this review, we address the existing knowledge about the metabolic niche of rhizosphere microbes and questions which were previously unanswered. How microbial activity differ in different soil horizon of forest ecosystems? How do root traits and exudation change within soil profiles? What are the impacts of different metabolites on microbial functional groups? How is metabolite change linked to root traits to influence microbial functional groups in forest soil? And finally, how do climate factors influence root traits, root exudation and microbiome dynamics in forest ecosystem? The goal of this review is to answer these questions for forest ecosystems to summarize the current information about the interactions among root traits, rhizosphere metabolites and microbial communities, and about their response to climate change, which have particular implications for future research investigations into the microbial ecology of forest soils."
} | 2,668 |
34429430 | PMC8384870 | pmc | 7,859 | {
"abstract": "Development of a versatile, sustainable and efficient photosynthesis system that integrates intricate catalytic networks and energy modules at the same location is of considerable future value to energy transformation. In the present study, we develop a coenzyme-mediated supramolecular host-guest semibiological system that combines artificial and enzymatic catalysis for photocatalytic hydrogen evolution from alcohol dehydrogenation. This approach involves modification of the microenvironment of a dithiolene-embedded metal-organic cage to trap an organic dye and NADH molecule simultaneously, serving as a hydrogenase analogue to induce effective proton reduction inside the artificial host. This abiotic photocatalytic system is further embedded into the pocket of the alcohol dehydrogenase to couple enzymatic alcohol dehydrogenation. This host-guest approach allows in situ regeneration of NAD + /NADH couple to transfer protons and electrons between the two catalytic cycles, thereby paving a unique avenue for a synergic combination of abiotic and biotic synthetic sequences for photocatalytic fuel and chemical transformation.",
"introduction": "Introduction Abiotic–biotic hybrid systems that combine light-driven artificial catalysis with biosynthetic enzymes at the same location have emerged as attractive and versatile avenues for light-trap fuel and chemical transformation with high efficacy and selectivity 1 – 5 . Recent advances have demonstrated that coupling solar fuel synthesis with value-added dehydrogenation may enhance economic and environmental benefits sans the expense of sacrificial reagents, while avoiding the harsh conditions required for the reforming processes 6 , 7 . Ethanol is a promising hydrogen storage chemical that can be effectively dehydrogenated by alcohol dehydrogenase (ADH) with the assist of coenzymes 8 , 9 . However, due to the inherent two-electron reduction characteristic 10 , 11 , the use of NADH (reduced nicotinamide adenine dinucleotide) to mediate artificial photoinduced proton reduction with enzymatic conversions remains a steep challenge in homogeneous system due to issues related to kinetic synergy and catalytic compatibility 12 , 13 . Precise matching of the kinetics of multiple electron transfer steps between abiotic and biotic components is a prerequisite to restrain the competitive reaction of NADH radical aggregation with photosensitizer radicals or itself 14 , 15 . Consequently, the integration of photosensitizer, coenzyme, and catalyst into one working module via the host–guest approach is promising to co-localize the essential components within the catalytic pocket of ADH and manipulate biomimetic catalysis at the molecular level 16 – 18 . Of the reported artificial supramolecular catalysts, metal-organic cages are superficially reminiscent of enzymes by modulating the microenvironment to accommodate and interact with substrates 19 – 23 . Dye-containing metal-organic cages exhibit profound effects in regulating and promoting the light-driven hydrogen evolution and related hydrogenation 24 , 25 . Incorporation of dye-containing metal-organic cages into the ADH catalytic pocket was expected to eliminate inherent communication barriers as well as mutual inactivation between the abiotic and biotic systems and promote the delivery of matters and energy, thereby facilitating the combination of NAD + -mediated dehydrogenation and NADH-modified hydrogen evolution 12 , 13 , 26 , 27 . Successful realization of paradigmatic structural fitness and kinetic compatibility requires careful orchestration of organic dye encapsulated by a potential-matching redox-active metal-organic cage for inclusion into the ADH catalytic pocket in a matryoshka fashion 26 , 27 . Here, we report a cobalt dithiolene-embedded pillared cage capable of trapping the shape and size matching photosensitizer and the coenzyme NADH as the middle layer of matryoshka, thereby combining abiotic photocatalytic hydrogen production with biotic dehydrogenation of alcohol within the ADH catalytic pocket (Fig. 1 ). The coexistence of the planar dye, 2-phenyl-4-(1-naphthyl)-quinolinium ( PNQ ) 28 , and the coenzyme NADH, within one redox-active microenvironment can enforce close proximity between these components to enhance the efficacy of photoinduced electron transfer inside metal-organic cage 29 , 30 , while simultaneously allowing efficient photocatalytic hydrogen production to directly produce NAD + in analog to the natural hydrogenase 31 . While situated inside the ADH catalytic pocket, the coenzyme is in direct contact with two catalytic cycles in situ, which enables it to maintain a closed loop of electrons and protons, thereby allowing the formation of a versatile redox-neutral photosynthesis system to actuate a non-photoactive natural enzyme for solar chemical conversion. Fig. 1 Schematic of the combination of artificial and natural enzymatic system. Construction of the molecular triangular prism Co 3 TPS 2 , the dye-containing cage, the NADH-dye-cage ternary supramolecular system, and the host–guest semibiological system comprising metal-organic cage Co 3 TPS 2 and natural enzyme ADH via non-covalent interactions, representing the assumed major binding conformation of the cage in the ADH enzymatic pocket from the docking study and the potential communication between artificial proton reduction and enzymatic alcohol dehydrogenation via the NAD + /NADH couple.",
"discussion": "Discussion In summary, a redox-active metal-organic cage Co 3 TPS 2 as a hydrogenase analog was embedded into the catalytic pocket of natural enzyme ADH through supramolecular interactions for solar alcohol splitting. The abundant non-covalent interaction sites in the artificial host allowed it to form an integrated host–guest species with a photosensitizer and an electron donor, constraining the photocatalytic hydrogen production from alcohol dehydrogenation inside the supramolecular host. The direct proton and electron delivery at close range between the two redox catalytic cycles provided positive feedback to the alcohol dehydrogenation processes. The attempt to associate artificial enzyme with non-photoactive natural enzyme in a host–guest approach achieved the optimized allocation of matter and energy by forming regional cooperation and division and insured electron transfer more efficient and controllable, illuminating the superiority of this supramolecular host–guest approach for redox-neutral artificial photosynthesis and providing a potential way to reduce carbon dioxide and even nitrogen."
} | 1,638 |
36875220 | PMC9975684 | pmc | 7,860 | {
"abstract": "Collective motion behaviour such as the movement of swarming bees, flocking birds or schooling fish has inspired computer-based swarming systems. They are widely used in agent formation control, including aerial and ground vehicles, teams of rescue robots, and exploration of dangerous environments with groups of robots. Collective motion behaviour is easy to describe, but highly subjective to detect. Humans can easily recognise these behaviours; however, it is hard for a computer system to recognise them. Since humans can easily recognise these behaviours, ground truth data from human perception is one way to enable machine learning methods to mimic this human perception. Hence ground truth data has been collected from human perception of collective motion behaviour recognition by running an online survey. In this survey, participants provide their opinion about the behaviour of ‘boid’ point masses. Each question of the survey contains a short video (around 10 seconds), captured from simulated boid movements. Participants were asked to drag a slider to label each video as either ‘flocking’ or ‘not flocking’; ‘aligned’ or ‘not aligned’ or ‘grouped’ or ‘not grouped’. By averaging these responses, three binary labels were created for each video. This data has been analysed to confirm that it is possible for a machine to learn binary classification labels from the human perception of collective behaviour dataset with high accuracy."
} | 362 |
37520767 | null | s2 | 7,862 | {
"abstract": "Capacitated Vehicle routing problem is NP-hard scheduling problem in which the main concern is to find the best routes with minimum cost for a number of vehicles serving a number of scattered customers under some vehicle capacity constraint. Due to the complex nature of the capacitated vehicle routing problem, metaheuristic optimization algorithms are widely used for tackling this type of challenge. Coronavirus Herd Immunity Optimizer (CHIO) is a recent metaheuristic population-based algorithm that mimics the COVID-19 herd immunity treatment strategy. In this paper, CHIO is modified for capacitated vehicle routing problem. The modifications for CHIO are accomplished by modifying its operators to preserve the solution feasibility for this type of vehicle routing problems. To evaluate the modified CHIO, two sets of data sets are used: the first data set has ten Synthetic CVRP models while the second is an ABEFMP data set which has 27 instances with different models. Moreover, the results achieved by modified CHIO are compared against the results of other 13 well-regarded algorithms. For the first data set, the modified CHIO is able to gain the same results as the other comparative methods in two out of ten instances and acceptable results in the rest. For the second and the more complicated data sets, the modified CHIO is able to achieve very competitive results and ranked the first for 8 instances out of 27. In a nutshell, the modified CHIO is able to efficiently solve the capacitated vehicle routing problem and can be utilized for other routing problems in the future such as multiple travelling salesman problem."
} | 409 |
31360812 | null | s2 | 7,863 | {
"abstract": "Can we create robots with the behavioral flexibility and robustness of animals? Engineers often use bio-inspiration to mimic animals. Recent advances in tissue engineering now allow the use of components from animals. By integrating organic and synthetic components, researchers are moving towards the development of engineered organisms whose structural framework, actuation, sensing, and control are partially or completely organic. This review discusses recent exciting work demonstrating how organic components can be used for all facets of robot development. Based on this analysis, we propose a Robotic Taxonomic Key to guide the field towards a unified lexicon for device description."
} | 172 |
38146695 | PMC10866086 | pmc | 7,864 | {
"abstract": "Abstract Polyhydroxyalkanoates (PHAs) are biodegradable polyesters produced by a wide range of microorganisms, including extremophiles. These unique microorganisms have gained interest in PHA production due to their ability to utilise low‐cost carbon sources under extreme conditions. In this study, Halomonas alkaliantarctica was examined with regards to its potential to produce PHAs using crude glycerol from biodiesel industry as the only carbon source. We found that cell dry mass concentration was not dependent on the applying substrate concentration. Furthermore, our data confirmed that the analysed halophile was capable of metabolising crude glycerol into poly(3‐hydroxybutyrate‐ co ‐3‐hydroxyvalerate) copolymer within 24 h of the cultivation without addition of any precursors. Moreover, crude glycerol concentration affects the repeat units content in the purified PHAs copolymers and their thermal properties. Nevertheless, a differential scanning calorimetric and thermogravimetric analysis showed that the analysed biopolyesters have properties suitable for various applications. Overall, this study described a promising approach for the valorisation of crude glycerol as a future strategy of industrial waste management to produce high value microbial biopolymers.",
"introduction": "INTRODUCTION Industrial biotechnology continues to drive innovation, improve bioprocess efficiency and contribute to a more sustainable and circular bioeconomy. In the last few decades, microbial processes are attracting interests of the scientific community as the sources of value‐added bio‐based products. Among them, polyhydroxyalkanoates (PHAs) are especially attractive as ecological alternative to conventional polymers derived from fossil fuels. They are a group of biopolymers that are produced by certain microorganisms in the form of intracellular granules. For many years, PHAs were thought to be a way to store carbon and energy (Policastro et al., 2021 ). However, recent studies demonstrated that the biological role of PHAs is more complex. The data suggested that the accumulation of PHAs provide stress robustness of bacteria, helping them maintain their structural integrity and function in challenging conditions (Obruca et al., 2020 ). PHAs possess unique properties such as biodegradability, non‐toxicity, biocompatibility; therefore, they are considered as innovative biopolymers that can be used in many novel medical or agricultural applications (Ashori et al., 2019 ). The specific details of PHA production can vary depending on the chosen microorganism, carbon source and desired PHA properties. The industrial production of PHA is still scarce, due to the high production costs (Koller & Mukherjee, 2022 ). Therefore, researchers continue to screen microorganisms that efficiently synthesise PHAs and explore ways to reduce their production costs. Extremophiles are of great interest to scientists because of their numerous applications in the fields of biotechnology (Kaur et al., 2019 ). The study of extremophiles for PHA production is still relatively in its early stages, but it holds promise for sustainable biopolymers production. Saline environments are widely distributed around the world. They are a source of halophilic bacteria that have a great potential to produce biomolecules such as PHAs. Moreover, in order to develop sustainable and economically viable microbial processes of PHA production, the requirement of the usage of cheap and renewable carbon source is essential. One of industrial feedstocks that can be applied as a substrate for microorganisms, is crude glycerol generated during biodiesel production. Due to the impurities, this residue has limited direct applications and the purification process is too expensive. For this reason, it has become very important to find an alternative method of its utilisation. Moreover, crude glycerol is one of the few waste sources that can be used directly as the carbon source for PHA production (Ray et al., 2016 ). Therefore, it is considered a promising alternative substrate in reducing production cost (Vicente et al., 2023 ). Moreover, based on the concept of next‐generation industrial biotechnology, extremophilic production strains have potential to transform the current PHA production into more competitive bioprocess (Yu et al., 2019 ). In this context, halophiles are currently considered as candidates for PHA production due to several advantages, including the utilisation of low‐cost carbon sources, reduced risk of contamination and potential for producing tailor‐made PHA compositions with unique properties (Yin et al., 2015 ). To the best of our knowledge, only in two publications the PHA biosynthesis process using Halomonas spp. grown on biodiesel byproducts were investigated. Shrivastav et al. ( 2010 ) analysed Halomonas hydrothermalis towards poly(3‐hydroxybutyrate) [P(3HB)] production using Jatropha biodiesel byproduct. Also, algal biodiesel waste residue was investigated for P(3HB) synthesis by Halomonas daqingensis (Dubey & Mishra, 2021 ). The above‐mentioned reports revealed that Halomonas spp. are capable of producing P(3HB) homopolymer. However, none of the study evaluated in details properties of the extracted biopolyesters. Moreover, there is still a lack of studies that showed the potential of Halomonas spp. to synthesise PHA copolymers from waste feedstock. In general, P(3HB) homopolymer is stiff and brittle due to its high crystallinity that limits its commercialization. Whereas, poly(3‐hydroxybutyrate‐ co ‐3‐hydrovalerate) copolymer [P(3HB‐ co ‐3HV)] was reported to be more desirable than P(3HB) because their melting point is much lower, and they are less crystalline (Możejko‐Ciesielska & Kiewisz, 2016 ). Therefore, the aim of the present study was to evaluate the capability of Halomonas alkaliantarctica to produce P(3HB‐ co ‐3HV) copolymer using biodiesel‐derived glycerol without additional precursors. Furthermore, we investigated the effect of the waste substrate concentration on biomass rate and PHA productivity. In addition, the extracted PHA copolymers were comprehensively studied based on their physico‐thermal properties.",
"discussion": "RESULTS AND DISCUSSION \nEffect of crude glycerol concentration on growth and PHA production\n Extremophiles have been recently discovered to be capable of utilising renewable feedstocks into high value products (Joulak et al., 2022 ). The study of halophiles has attracted scientific interest due to their ability to survive in extreme environments and their potential applications in the production of biomolecules. There are no reports that evaluated the potential of H. alkaliantarctica to grow and produce PHAs during fermentation process with biodiesel‐derived glycerol. Moreover, any Halomonas spp. has not been investigated for production of PHA copolymers using crude glycerol derived from vegetable oil so far. Therefore, the aim of this study was to evaluate whether H. alkaliantarctica is capable of converting crude glycerol into PHA copolymers and to characterise them taking into consideration their physico‐chemical properties. Our results confirmed that the crude glycerol did not hamper the growth of the analysed bacterial species (Figure 1 ). We found that CDM values were not dependent on the applying substrate concentration. Shake flasks yielded comparable biomass value at all applied crude glycerol concentrations. Nevertheless, the CDM values were higher at 48 and 72 h in comparison to 24 h of all experimental variants. The biomass concentration reached the maximum value in the cultivation with 80 g/L of crude glycerol in 72 h (1.7 g/L). Lower CDM rates (below 0.5 g/L) were determined by Dubey and Mishra ( 2021 ) who cultured H. daqingensis in the fermentor supplemented with algal biodiesel waste residue. Lower bacterial cell density was also reported by Shrivastav et al. ( 2010 ) in the cultivation of H. hydrothermalis with Jatropha biodiesel byproduct (about 0.4 g/L). FIGURE 1 Growth of Halomonas alkaliantartica in the medium supplemented with crude glycerol. Mean values are calculated from triplicate measurements. As shown on Figure 2A,B , H. alkaliantarctica was able to produce PHAs in all conducted experiments within 24 h of the cultivation. PHA content was the highest in bacterial cells cultivated in BMB medium supplemented with 50 g/L of biodiesel‐derived glycerol (about 17% of CDM). At higher level of this feedstock, the PHA content in CDM decreased (Figure 2A ). However, in all experimental variants the PHA content was higher compared to the values reported in shake‐flasks culture of H. halophila , H. salina or H. meridiana grown on waste frying oil which synthesised only 0.38%, 0.66% and 2.96% of CDM, respectively (Pernicova et al., 2019 ). Moreover, our results showed that the highest PHA concentration (0.27 g/L) and PHA productivity (5.63 mg/(L·h)) was reached in the cultivation with 50 g/L of crude glycerol in 48 h. We also observed that the further increase of the waste substrate concentration (above 50 g/L) resulted in a decrease of PHAs productivity at all measured time‐points (Figure 2B ). However, Liu et al. ( 2022 ) suggested that NaCl concentration can play a role in PHA production. The authors showed that pathways involved in salt tolerance can be blocked or weakened in H. cupida J9 cells when using glycerol as a feedstock. In a consequence, these bacterial cells can respond to a high salt concentration by increased production of PHA as an effective protectant against salt stress. Also, Kucera et al. ( 2018 ) reported that the concentration of NaCl used during cultivation of H. halophila influenced PHA productivity. The highest P(3HB) yield was observed in the cultivation supplemented with 60 g/L of NaCl, whereas at lower and higher salt concentration (20, 40, 80 and 100 g/L), the homopolymer productivity decreased. The same observations were made by Rodríguez‐Contreras et al. ( 2016 ) in the culture of halophilic bacterium Bacillus megaterium uyuni S29. The authors proved that NaCl concentration is an essential factor influencing PHA productivity when employing halophilic bacteria. FIGURE 2 Polyhydroxyalkanoate (PHA) production by Halomonas alkaliantartica in the medium supplemented with crude glycerol. (A) PHA content in cell dry mass (CDM), (B) PHA productivity in 24, 48 and 72 h of the cultivations. Mean values are calculated from triplicate measurements. \nEffect of crude glycerol concentration on PHA composition\n The results from gas chromatography coupled with mass spectrometry analysis showed that H. alkaliantarctica is capable of producing copolymers contained higher content of 3HB monomer and lower content of 3HV fraction (Table 1 ). We observed that the repeat units content in the purified PHA copolymers depended on the concentration of biodiesel‐derived glycerol. Moreover, the content of the monomers varied in different cultivation time. The concentration of 3HV fraction was higher in 48 h than in 24 h in all experimental variants. Halomonas spp. are known to be able to produce P(3HB) (Dubey & Mishra, 2022 ; Kawata & Aiba, 2010 ). Nevertheless, PHA copolymers are of high interest due to their favourable thermomechanical properties compared to PHA homopolymers. Incorporation of 3HV monomer to the polymer chain lowers melting temperature, reduces crystallinity, improves flexibility that are important parameters for their future application (Hammami et al., 2022 ). Most bacterial species required precursors to produce P(3HB‐ co ‐3HV) such as Cupriavidus necator (Grousseau et al., 2014 ), Bacillus aryabhattai (Balakrishna Pillai et al., 2020 ) or Halomonas sp. YLGW01 (Kim et al., 2023 ). To our knowledge, the analysed H. alkaliantarctica is the only halophilic bacterial strain reported to be able to synthesise P(3HB‐ co ‐3HV) copolymers using biodiesel‐derived glycerol without any co‐substrates. TABLE 1 Monomeric composition of extracted PHAs. Crude glycerol (g/L) 24 h 48 h 72 h 3HB (Mol%) 3HV (Mol%) 3HB (Mol%) 3HV (Mol%) 3HB (Mol%) 3HV (Mol%) 10 98.95 1.05 97.23 2.77 97.17 2.83 30 99.03 0.97 97.80 2.20 98.03 1.97 50 98.78 1.22 98.18 1.82 97.77 2.23 70 99.01 0.99 97.93 2.07 98.25 1.75 80 99.00 1.00 98.35 1.65 98.17 1.83 Abbreviations: 3HB, 3‐hydroxybutyrate; 3HV, 3‐hydroxyvalerate; PHA, polyhydroxyalkanoate. The chemical structure of the extracted biopolymers was also confirmed by Fourier transform infrared spectroscopic spectra (FTIR spectra) (Figure 3 ). The observed bands were characteristic for P(3HB‐ co ‐3HV) copolymer. Similar signals were reported in earlier studies suggesting the chemical structure of the above‐mentioned PHA copolymer (Volova et al., 2013 ). The band in the range of 3100–2700 cm −1 was associated with the presence of ‐CH 3 and ‐CH 2 groups in macromolecules. A single peak band with a maximum of ~1722 cm −1 was derived from the C=O carbonyl groups found in macromolecules. Also, Abd El‐malek et al. ( 2020 ) showed the strongest band at 1720 cm −1 for PHA extracted from Halomonas pacifica ASL10 and Halomonas salifodiane ASL11. Additionally, our results confirmed that in the spectrum range of 1500–800 cm −1 , which is the fingerprint region characteristic for PHAs, several bands with numerous peaks are characteristic for the type of the produced copolymers. Peaks with maxima of ~1460, ~1379, ~980, ~900 and ~825 cm −1 were associated with the presence of ‐CH 3 and ‐CH 2 groups in macromolecules, as well as C‐C bonds. It has been previously proved that a strong vibration at 1379 cm −1 is characteristic for PHA (Sathiyanarayanan et al., 2017 ). Moreover, in the obtained FTIR spectra a series of peaks appeared at ~1278, ~1228, ~1180, ~1132, ~1101 and ~1053 cm −1 were associated with bonds in the C‐O‐C region. FIGURE 3 FTIR spectra of selected copolymers extracted from Halomonas alkaliantartica cells: (A) P(97.23% HB‐co‐2.77% 3HV) from the cultivation with 10 g/L of crude glycerol; (B) P(98.18% HB‐co‐1.82% 3HV) from the cultivation with 50 g/L of crude glycerol; (C) P(98.35% HB‐co‐1.65% 3HV) from the cultivation with 80 g/L of crude glycerol. \nEffect of crude glycerol concentration on the properties of extracted PHAs \n Besides chemical structure, the potential applicability of PHAs also depends on its thermal characteristic. Therefore, we investigated the effect of crude glycerol concentration on thermal properties of the extracted co‐polymers (Table 2 ). Properties studies on the biopolyesters extracted from Halomonas spp. cells cultured with biodiesel‐derived glycerol as a sole carbon source have not been reported so far. To characterise comprehensively the PHAs biofilms, the copolymers extracted in 48 h of the cultivations were chosen for further analyses. TABLE 2 Thermal properties of P(3HB‐ co ‐3HV) copolymers extracted from the cultivation of Halomonas alkaliantarctica grown on crude glycerol in 48 h. Crude glycerol (g/L) PHA composition Differential scanning calorimetry Thermogravimetric analysis \n T \n g [°C] \n T \n cc [°C] Δ H \n cc [J/g] \n T \n m [°C] Δ H \n m [J/g] \n X \n c [%] \n T \n d [°C] \n T \n max [°C] 10 P(97.23% HB‐ co ‐2.77% HV) −1.5 45.0 49.2 163.6 50.6 1.3 193.9 271.0 30 P(97.80% HB‐ co ‐2.20% HV) −1.0 48.8 56.4 164.8 62.1 5.2 188.4 285.0 50 P(98.18% HB‐ co ‐1.82% HV) −0.2 50.3 74.0 165.7 81.7 7.1 212.1 286.1 70 P(97.93% HB‐ co ‐2.07% HV) 0.0 42.7 41.9 165.5 53.7 10.9 213.5 285.9 80 P(98.35% HB‐ co ‐1.65% HV) 0.7 41.3 41.3 165.0 56.1 13.6 213.1 288.3 Abbreviations: P(3HB‐ co ‐3HV), poly(3‐hydroxybutyrate‐ co ‐valerate); T \n cc , cold crystallisation temperature; T \n d , decomposition temperature; T \n g , glass transition temperature; T \n m , melting temperature; T \n max , maximum decomposition temperature; X \n c , degree of crystallinity; Δ H \n cc , change in enthalpy of the cold crystallisation process; Δ H \n m , change in enthalpy melting process. For all tested materials, three phase transitions are observed on the DSC curves: glass transition, cold crystallisation and melting (Figure 4 ). As can be seen in Table 1 , the content of 3HV units in individual polymers is very similar. The changes in thermal properties must therefore result from another parameter or material characteristic. It was found that the recorded thermal parameters were strongly influenced by the crude glycerol concentration. We observed that the higher crude glycerol concentration the higher T \n g value. The glass transition temperature ( T \n g ) of the tested materials ranged from −1.5 to 0.7°C. The increase in T \n g was probably due to the change in the degree of crystallinity of the copolymers which also increased with the increase in the concentration of crude glycerol. At a concentration of 10 g/L, the calculated degree of crystallinity was 1.3%, while at a concentration of 80 g/L the degree of crystallinity increased to 13.6%. The melting temperature ( T \n m ) of the analysed biopolymers was also affected by degree of crystallinity and reached about 165°C. This point value is similar to the P(3HB‐ co ‐3HV) isolated from Halomonas campisalis cells in the cultivation with bagasse as sole substrate (Kulkarni et al., 2015 ). Higher T \n m value (178°C) was reported for P(3HB) produced by Halomonas sp. SF2003 cultured on agro‐industrial effluents (Lemechko et al., 2019 ). It could be suggested that the concentration of crude glycerol influences the structure of polymer chains in a way that favours the crystallisation process. The increase in the degree of crystallinity influences the shift in the glass transition and melting temperatures. FIGURE 4 DSC spectra of selected copolymers extracted from Halomonas alkaliantartica cells: (A) P(97.23% HB‐co‐2.77% 3HV) from the cultivation with 10 g/L of crude glycerol; (B) P(98.18% HB‐co‐1.82% 3HV) from the cultivation with 50 g/L of crude glycerol; (C) P(98.35% HB‐co‐1.65% 3HV) from the cultivation with 80 g/L of crude glycerol. Furthermore, we observed that the degree of crystallinity increasing with the concentration of crude glycerol influenced on the thermal resistance of the extracted biopolymers (Figure 5 ). P(3HB‐ co ‐3HV) isolated from the H. alkaliantarctica cells grown on higher concentration of crude glycerol were characterised by higher decomposition temperatures ( T \n d ), that is, the parameter adopted as the degree of thermal resistance and representing the temperature of loss of 5% of the sample weight. The highest thermal resistance was detected for the PHA copolymer produced at the highest carbon source concentration ( T \n d = 213.1°C). Similar observation was made for maximum degradation temperature ( T \n max ), that is, the parameter determining the temperature at which the thermal degradation process proceeds in the most intensive range. Kavitha et al. ( 2016 ) showed that P(3HB) homopolymer produced by Botryococcus braunii degraded at 240°C. Our results indicated that the carbon source concentration influenced also on the rate of the degradation process of the extracted copolymers. As the concentration of crude glycerol increased, the T \n max value of analysed PHA copolymers also increased. The T \n max value for PHA biofilm extracted from the cultivation supplemented with 80 g/L reached 288.3°C. Stanley et al. ( 2020 ) reported that the maximum degradation for P(3HB) synthesised by H. venusta occurred at 323.3°C. FIGURE 5 TG and DTG curves of selected copolymers extracted from Halomonas alkaliantartica cells: (A) P(97.23% HB‐co‐2.77% 3HV) from the cultivation with 10 g/L of crude glycerol; (B) P(98.18% HB‐co‐1.82% 3HV) from the cultivation with 50 g/L of crude glycerol; (C) P(98.35% HB‐co‐1.65% 3HV) from the cultivation with 80 g/L of crude glycerol. The exploration of halophilic bacteria for PHA production has many benefits such as short fermentation time, lack of pathogenicity and ability to produce this biomolecule using renewable resources (Mitra et al., 2016 ). Our results proved that H. alkaliantarctica is able to utilise crude glycerol for the growth and P(3HB‐ co ‐3HV) production. Also, we found that PHA productivity, monomers content and thermal properties were dependent on the biodiesel‐derived glycerol concentration. In addition, this bacterium produced the scl‐copolymer without any 3HV precursors and make this biotechnological process a valuable strategy and industrially very important. Furthermore, the obtained data suggest that P(3HB‐ co ‐3HV) copolymers have properties that are important for industrial applications. Further work should include metabolic engineering along with genome editing to improve the ability of H. alkaliantarctica to metabolise broader range of waste feedstocks. A pressing need also arises to optimise the culture parameters in bioreactor for PHA production at industrial scale."
} | 5,230 |
35718767 | PMC9208136 | pmc | 7,868 | {
"abstract": "Background Lycopene is increasing in demand due to its widespread use in the pharmaceutical and food industries. Metabolic engineering and synthetic biology technologies have been widely used to overexpress the heterologous mevalonate pathway and lycopene pathway in Escherichia coli to produce lycopene. However, due to the tedious metabolic pathways and complicated metabolic background, optimizing the lycopene synthetic pathway using reasonable design approaches becomes difficult. Results In this study, the heterologous lycopene metabolic pathway was introduced into E. coli and divided into three modules, with mevalonate and DMAPP serving as connecting nodes. The module containing the genes ( MVK, PMK, MVD, IDI ) of downstream MVA pathway was adjusted by altering the expression strength of the four genes using the ribosome binding sites (RBSs) library with specified strength to improve the inter-module balance. Three RBS libraries containing variably regulated MVK, PMK, MVD, and IDI were constructed based on different plasmid backbones with the variable promoter and replication origin. The RBS library was then transformed into engineered E. coli BL21(DE3) containing pCLES and pTrc-lyc to obtain a lycopene producer library and employed high-throughput screening based on lycopene color to obtain the required metabolic pathway. The shake flask culture of the selected high-yield strain resulted in a lycopene yield of 219.7 mg/g DCW, which was 4.6 times that of the reference strain. Conclusion A strain capable of producing 219.7 mg/g DCW with high lycopene metabolic flux was obtained by fine-tuning the expression of the four MVA pathway enzymes and visual selection. These results show that the strategy of optimizing the downstream MVA pathway through RBS library design can be effective, which can improve the metabolic flux and provide a reference for the synthesis of other terpenoids. Supplementary Information The online version contains supplementary material available at 10.1186/s12934-022-01843-z.",
"conclusion": "Conclusion The goal of this study was to use synthetic biology concepts to increase the efficiency of lycopene production. The one-step OLEM method, a semirational approach that can streamline the library creation procedure, was employed to construct three libraries containing RBS with varying intensities for each of four genes of the lycopene pathway based on various plasmid backbone. The visual selection of lycopene-accumulating colonies based on its vivid red hue simplified the subsequent screening steps even further in a glucose-rich medium plate. Following that, the association between RBS intensity and lycopene accumulation was evaluated in recombinant strains harboring various pathways. Although there was no clear proportion relationship between these four genes and lycopene metabolic flux, there was a difference in lycopene output between the ten best-ranked samples from each of the three libraries for shaker flask fermentation, ranging from 20.3 to 219.7 mg/g DCW. This finding suggests that it is feasible to balance the metabolic flux of the entire MVA pathway by leveraging a combination optimization strategy to optimize expression level of the downstream MVA pathway.",
"introduction": "Introduction Lycopene, a carotenoid with a red color and significant antioxidant effects due to its conjugated polyene structure [ 1 ], plays a vital role in human health. As an antioxidant, it exerts protection on the cardiovascular system by reducing the risk of myocardial infarction, lowering blood pressure, and preventing the oxidation of low-density lipoprotein cholesterol [ 2 – 4 ]. Additionally, research has shown that lycopene may be beneficial in preventing some types of malignant tumors, including prostate, lung, uterus and breast cancer [ 5 , 6 ]. Furthermore, in synthetic biology, the lycopene enables high-throughput screening based on color variation and serves as an excellent model system for researching isoprenoids biosynthesis pathways [ 7 ]. Lycopene is currently directly extracted from natural resources [ 8 ]. Although the products extracted by this method are of high quality and have biological activity, the extraction method is relatively expensive due to the low content of lycopene in tomatoes and the complex multi-step process [ 9 , 10 ]. Chemical synthesis has the advantage of low raw material cost, but it will bring health and food safety issues. Therefore, the microbial production of lycopene provides an attractive alternative because it can completely synthesize lycopene from cheap carbon sources, which has the potential to increase yield and sustainability and reduce production costs. Furthermore, as micro metabolic engineering and synthetic biology have progressed, it has become easier to modify, design, and optimize host microorganisms into advanced microbial cell factories [ 11 – 13 ]. Escherichia coli , as an excellent resident microorganism, is widely used for this purpose [ 14 – 17 ]. The mevalonate (MVA) pathway, which is native to eukaryotes and archaea, and the 2-C-methyl-D-erythritol 4-phosphate (MEP) pathway, which is native to most of the bacteria and plant plastids [ 18 , 19 ], were employed to synthesize the precursors of lycopene isopentenyl diphosphate (IPP) and dimethylallyl diphosphate (DMAPP) during the biosynthesis process. In the MVA pathway, two molecules of acetyl-CoA are condensed to form 3-hydroxy-3-methylglutaryl-CoA (HMG-CoA), followed by a reduction reaction that generates mevalonate. After the production of mevalonate, two phosphates are added sequentially to the molecule by mevalonate kinase (MVK) and phosphomevalonate kinase (PMK). Finally, mevalonate-5-diphosphate is decarboxylated by mevalonate diphosphate decarboxylase (MVD) to form IPP. Once produced, IPP is converted reversibly to DMAPP by isopentenyl diphosphate isomerase (IDI). Subsequently, IPP and DMAPP were condensed to generate farnesyl diphosphate (FPP), which then was catalyzed by GGPP synthase, lycopene synthase, and lycopene desaturase, encoded by Erg20 , CrtE , CrtB, and CrtI, respectively, to form geranylgeranyl diphosphate (GGPP, C20), colorless phytoene (C40), and red-colored lycopene. With the advancement of synthetic biology technologies in recent years, the capacity to synthesis lycopene has been enhanced by introducing and optimizing heterologous metabolic pathways in E. coli . Due to the shortage of precursors in E. coli with native MEP pathway, the supply of IPP and DMAPP is usually enhanced by improving the MEP pathway or introducing a heterologous MVA pathway. Many studies have shown that introducing the exogenous MVA pathway into E. coli to improves IPP and DMAPP precursor supply. When the entire MVA pathway from Streptomyces sp. CL190 was introduced into E. coli with only a native MEP pathway, lycopene production increased by twofold [ 20 ]. Zhu et al. adopted a new target engineering strategy to optimize the MVA pathway in E. coli for improving the supply of precursor IPP and DMAPP, and the lycopene titer eventually reached 1.23 g/L in fed-batch fermentation [ 21 ]. In addition, Zhang et al. increased lycopene production 36-fold by altering the order of key enzymes crtE, crtB and crtI in the lycopene synthesis pathway [ 22 ], suggesting that a wrong gene arrangement can lead to a severe imbalance of enzymes in this pathway. Furthermore, ribosome-binding site libraries may be employed as a modifying approach to boost lycopene output, with the best strain producing 3.52 g/L in fed-batch fermentation [ 23 ]. In a previous study [ 24 ], we discovered that optimizing the ratios of the MVA pathway's five enzymes ( MVK\\PMK\\MVD\\IDI\\Isps4 ) in vitro significantly improved the conversion efficiency of mevalonate to isoprene. However, due to the apparent intricacy of the cellular metabolic microenvironment, the results of in vitro are not appropriate to intracellular. Therefore, the development of synthetic biology methods can enable us to optimize the expression of the four genes of MVA in vivo to increase the metabolic flux of the MVA pathway. Manipulation of promoter, replication origin of plasmid, and RBS was thought to be the most straightforward approach for fine-tuning expression at transcriptional and translational levels [ 25 ]. Among them, RBS is essential for the translational control of enzyme activity [ 26 ]. It is preferable to fine-tune genes inside MPMI modules via RBS library engineering. RBS calculators have been designed to produce RBS sequences with precise intensities, which may then be used to build multi-enzyme pathways with great precision [ 27 , 28 ]. As a result, further boosting lycopene synthesis necessitates fine-tuning of the complex metabolic networks that surround inter-and intra-modules and metabolic optimization at the transcriptional and translational levels. In this work, the capacity of microbial biosynthesis to be optimized by combining the coordinated expression of multiple genes in the module and the combinatorial approach between modules. To begin, the lycopene synthesis pathway was divided into three modules: module ESE and module MPMI and module EBI (Fig. 1 A). Subsequently, the RBS library with defined strength for adjusting the expression levels of four genes in module MPMI was constructed based on different plasmid backbones with the variable promoter and replication origin using an oligo-linker mediated assembly (OLEM) method [ 26 ]. Finally, the RBS library was then transformed into engineered E. coli BL21 (DE3) /pCLES&pTrc-lyc to obtain a lycopene producer library and employed high-throughput screening based on lycopene color to obtain the required metabolic pathway. Finally, the MVA pathway obtained through screening increased lycopene yield, thus achieving the goal of \"what you see is what you get.\" Fig. 1 Library construction and visual selection based on-plate and 96 deep-well plate selections. A Lycopene accumulation pathway in E. coli by heterogenous MVA pathway. The whole pathway was divided to three modules: module ESE, the upstream MVA pathway; module MPMI, the downstream MVA pathway; module EBI, the lycopene synthesis pathway containing Erg20 , CrtE , CrtB, and CrtI , In the module2, an RBS library and combinatorial constructs based on three plasmid backbones (with the pT7 promoter and the p15A origin, with pTrc promoter and p15A origin, or with pTrc promoter and pSC101) were employed to screen the optimal expression levels for the four genes MVK, PMK, MVD, and IDI to achieve the purpose of optimizing the lycopene metabolic flux. B On-plate and 96-deep-well plate visual selection of pHM-library. Each colony on the plate or 96-deep-well plate showed different colors indicating different lycopene accumulation",
"discussion": "Results and discussion Mevalonate pathway optimization by RBS library construction In the multienzyme pathway, the simple overexpression of one or more enzymes risks causing an imbalance between enzymes and upstream and downstream pathway modules, resulting in growth inhibition or poor yield owing to the cytotoxicity of accumulated intermediates, as well as the formation of new bottleneck nodes in the metabolic pathway [ 29 , 30 ]. Liu et al. found that the upstream and downstream modules of the MVA pathway have severe metabolic imbalances, resulting in the output of mevalonate in the upstream module reaching about 84.0 g/L, while the production of isoprene, the product catalyzed by the downstream module, is only about 8.0 g/L [ 31 ]. The downstream module's four enzymes (MVK, PMK, MVD, IDI) and isoprene synthase were then expressed and purified in vitro using mevalonate as a substrate to catalyze isoprene synthesis. Adjusting the proportion of the five enzymes in vitro significantly improved the conversion efficiency of mevalonate to isoprene conversion [ 24 ]. The findings provided a novel ideal for optimizing the MVA pathway in E. coli : balancing the upstream and downstream modules by fine-tuning the expression levels of the four enzymes of the downstream module to enhance the metabolic flux. Lycopene is suitable for validating the high-throughput MVA metabolic pathway due to its vivid red hue. The lycopene metabolic pathway was divided into three modules using mevalonate and DMAPP as connecting nodes. The module MPMI containing the genes ( MVK, PMK, MVD, IDI ) of downstream MVA pathway was adjusted by altering the expression strength of the four genes using the ribosome binding sites (RBSs) library with specified translation initiation rate (TIR), which was predicted by RBS calculator [ 27 ]. The semi-rational OLMA strategy was used to build an RBS library with defined strengths for each of the four genes MVK, PMK, MVD, and IDI [ 26 ]. This strategy can construct a library involving different variables such as promoters, RBSs, CDSs, and terminators in one step. The library is constructed based on three plasmid backbones: one medium copy (ori p15A) with strong promoter T7 (library HM, copy number 10–12, promoter T7), one medium copy (ori p15A) with medium strength promoter Trc (library MM, copy number 10–12, promoter Trc) and a low copy number plasmid (ori pSC101) with medium strength promoter Trc (library LM, copy number 1–2, promoter Trc). Each correctly assembled plasmid in the library contains four RBS randomly selected from four groups of RBS libraries. The library has a capacity of 10,000 combinations because each group has 10 different RBSs. Three libraries of 10,000 combinations (HM, MM, and LM) were constructed using OLMA method (Fig. 1 A). Then, the obtained library was co-expressed in E. coli with the upstream MVA pathway and the lycopene synthesis pathway and spread on the LB agar plate with inducers. After culturing for 72 h, red colonies with various intensities appeared on the plates (Fig. 1 B), indicating their varying capacity for lycopene production. PCR analysis was performed on ten randomly selected red colonies of various intensities, and the gel electrophoresis revealed a 100% positive rate (Additional file 1 : Fig S1). The results demonstrate that by assembling an RBS library and performing a simple color-based pre-screening, lycopene metabolic pathways with a diverse spectrum of metabolic fluxes can be obtained with minimal effort. High-throughput screening of recombinants with lycopene pathway Colonies were chosen based on the depth of the colony color, which is a result of the color rendering of lycopene. In other words, the colonies grown on the LB-Glu solid medium exhibited a bright red hue compared to the colonies that accumulated little or no lycopene. 500 colonies from the pHM-library, 500 colonies from the pMM-library, and 500 colonies from the pLM-library, all of which were classified according to varying degrees of redness, were chosen using the visual selection method. All the selected colonies and three reference strains HM-CT, MM-CT, and LM-CT (positive control, which was a recombinant containing ESE, EBI, and MPMI modules with native RBSs under the three plasmid backbones) were grown in M9 medium on 96 deep-well plates. The culture medium showed various intensities of red after 96 h of cultivation (Fig. 1 B), indicating the strains with different lycopene accumulation performance in the liquid medium. Lycopene was extracted from the samples with acetone and measured at 472 nm using absorbance and corrected with OD 600 . As shown in Fig. 2 A–C, the relative lycopene production of the selected strains in 96-deep-well showed a 14 ~ 40-fold range (pHM-library 30-fold, pMM-library 14-fold, and pLM-library 40-fold), which indicated the combinatorial RBS library yield a large number of functional constructs with different lycopene production level. Fig. 2 The results of 96-deep-well plate culture of 500 randomly picked strains in pHM-library, pMM-Library, and pLM-Library, respectively. The Y axis represents the absorbance of lycopene at 472 nm corrected using OD 600 . The orange dot represents reference strain. A The results of 96-deep-well plate culture of 500 strains in pHM-library. B The results of 96-deep-well plate culture of 500 strains in pMM-library. C The results of 96-deep-well plate culture of 500 strains in pLM-library. D Boxplots of the strains based on 500 randomly selected colonies per library Furthermore, when the module ESE and module EBI were kept constant, the lycopene production was significantly increased by using the RBS library to fine-tune the four genes expression of the module MPMI (Fig. 3 D). More than a quarter of the strains screened from the three libraries with various plasmid backbones were capable of producing more lycopene than the control strain. In the pHM-library, the best sample with ABS 472 /OD 600 of 0.58 showed 3 times stronger than HM-CT and 30 times stronger than the lowest sample with ABS 472 /OD 600 of 0.02. The highest one showed 2.5 times higher than MM-CT in the pMM-library and 14 times stronger than the lowest sample with ABS 472 /OD 600 of `0.05. In the LM library, 50% of samples showed higher than LM-CT, the highest sample with AbS 472 /OD 600 of 0.9, was higher 1.2-fold and 30 times than the LM-CT and the lowest sample, respectively. The LM library, which was under the plasmid backbone with the medium promoter and low copy, as shown in Fig. 3 D, enables the screening of transformants with significant lycopene accumulation (ABS 472 /OD 600 = 0.9). Low copy number and weaker promoters of the MVA downstream pathway operon were more favorable to lycopene accumulation, according to the findings (Fig. 2 D). The explanation for this may be that decreasing the expression level of the midstream module can better match the upstream and the downstream module of lycopene synthesis, reducing the accumulation of intermediate and therefore increasing lycopene metabolic flux. Furthermore, this result revealed that altering enzyme activity generated by the strength of RBS has a substantial impact on the activity of the lycopene pathway. It was able to comprehensively investigate the activity of the MVA pathway since the RBS library constructed in this research caused a continuous shift in lycopene accumulation from zero to a high level (Fig. 2 D). More significantly, this method can be used to synthesize additional terpenoids. Fig. 3 Characterization of the downstream MVA pathway (module2) activity and correlation between lycopene accumulation and RBS strengths. Lycopene accumulation was showed by orange columns. The reference strain with the native RBS was showed by red columns. RBS strengths were indicated by shades of color. A module MPMI activities and corresponding RBS strengths in pHM-library. B module MPMI activities and corresponding RBS strengths in pMM-library. C module MPMI activities and corresponding RBS strengths in pLM-library. D RBS strengths of RBS mvk (green), RBS pmk (red), RBS mvd (blue), and RBS idi (purple) indicated by shade of color. The results represent the means ± S.D. of three independent experiments Characterization of MVA pathway activities To investigate the association between the RBS strengths and lycopene accumulation, 50 samples from each library were chosen and confirmed the RBS strengths. RBS PMK and RBS MVD have no apparent relationship with pathway activity, whereas RBS MVK has a modest positive correlation with pathway activity, especially in the pHM-library and pMM-library. RBS IDI had a significant positive correlation with pathway activity in the three libraries. As shown in Fig. 3 A, B, the ten best-ranked strains of lycopene accumulation, eight strains in pHM-library and nine strains in pMM-library showed stronger TIR of RBS MVK , while all the strains showed stronger TIR of RBS IDI . Similarly, among the ten strains with the lowest lycopene production, seven strains in pHM-library and pMM-library showed weaker TIR of RBS MVK , eight strains showed weaker TIR of RBS IDI . IDI activity appeared to be critical for a high lycopene pathway. The findings were consistent with previous reports in which the MVK and IDI were identified as the crucial enzymes in the mevalonate pathway [ 32 , 33 ]. Unfortunately, there was no clear proportion relationship involving these four genes, indicating that optimizing the MVA pathway by entirely rational design is challenging due to the sophisticated interactions among heterogeneous pathways and complex metabolic context. Given that there were 10,000 possible combinations, but only 500 colonies were selected for analysis, the small sample size resulted in the absence of a clear association between the RBS strengths and lycopene accumulation. Furthermore, the library developed in this study could be used to synthesize other products that could be selected via high-throughput visualization (colored products), such as carotene, astaxanthin, etc. Shake-flask fermentation of recombinants with selected pathways For further validation in shake flask fermentation, the three reference strains and ten top-ranked samples from each of the three libraries were chosen. We begin by comparing the lycopene accumulate and cell density in the three reference strains grown in shake-flask. As shown in Fig. 4 , the lycopene yield of the MM-CT strain, in which the operon of midstream module was initiated by medium strength promoter (Trc), was 60.4 mg/g DCW, which was barely 20% greater than that of HM-CT (48.3 mg/g DCW) with the stronger promoter (T7). However, as compared to MM-CT strain with the medium-copy plasmid (10 ~ 12), the lycopene production of the LM-CT strain harboring low-copy plasmid (1 ~ 2) increased the lycopene production by 100% to 90.2 mg/L DCW. The biomass of LM-CT is significantly higher than that of HM-CT and MM-CT. This result showed that at constant module ES and module EBI expression, lycopene production increased in lockstep with the decreasing strength of module MPMI. The trends of lycopene accumulation of reference strains in shake flasks were consistent with the results in 96-deep well. Fig. 4 The lycopene production and biomass analysis. Ten best ranked samples selected from per library and their respective reference strains were further cultivated in shake flasks. Lycopene accumulation was showed by orange columns. The biomass was showed light yellow. A The lycopene accumulation and biomass of strains in pHM-library and of corresponding reference strain. B The lycopene accumulation and biomass of strains in pMM-library and of corresponding reference strain. C The lycopene accumulation and biomass of strains in pLM-library and of corresponding reference strain. The results represent the means ± S.D. of three independent experiments Figure 4 shows that replacing RBSs with native ones increased lycopene yield by over 120 percent in all three libraries without changing any other sequence in the operon. Figure 4 A shows that the sample pHM25 with the highest yield of 120 mg/g DCW in pHM-library was 250% stronger than the control strain HM-CT (48.3 mg/g DCW). In Fig. 4 B, C, we found that the lycopene yield of sample MM35 (143.6 mg/g DCW) in MM-library and LM22 (219.7 mg/g DCW) in LM-Library increased by 140% and 120% compared to the reference strain MM-CT and LM-CT, respectively. The biomass of the LM-Library was significantly higher than that of the pHM-Library and pMM-library, which was consistent with the reference strain trend. While 70%-80% of the selected samples accumulated more lycopene than the control strain containing natural RBS in shake flask fermentation, 20% ~ 30% samples in each library perform worse than the control strains, which could be the inconsistency between high-throughput screening and shake-flask fermentation conditions. Over the course of many projects, strain productivity and growth rate may or may not improve when transferred from deep-well plates to shake flasks due to stochastic effects. With constant upstream module and downstream module, as the promoter of the midstream module changed from pT7 to pTrc with decreasing strength, lycopene accumulation increased correspondingly. Furthermore, as the replication origin of the midstream module changed from p15A to pSC101 with reducing the copy number, lycopene production further improved likewise. The results demonstrate that the downregulation of the midstream module enhances the flux of the entire lycopene pathway, hence boosting product biosynthesis, which is consistent with previous research [ 34 ]. Excessive midstream module flux harmed lycopene accumulation and cell growth, which could be due to the downstream module's limited capacity or the potential cytotoxicity of accumulated intermediate metabolites like GPP and FPP [ 29 , 35 ]. The four genes of the midstream module were fine-tuned based on these foundations by the construction of the RBS library, which further optimized the lycopene metabolic pathway and improved lycopene accumulation by 120% ~ 250% over the reference strains. In the case of limited samples (500 samples was selected in each library), the high-yielding lycopene (219.7 mg/g DCW) strain in shake flask fermentation in LM-library was screened. These results indicate that coordinating the expression of all the genes of the midstream module was an effective strategy for optimizing the lycopene synthesis pathway. Although the strain obtained through screening has a lower yield than the previously reported 448 mg/g DCW [ 36 ], a strain with increased lycopene accumulation can be created by increasing the screening quantity of the library. Supposing the screening of library samples is broadened, for example, by using the automated screening tool QPix 400 microbial colony pickers. In this case, it may be feasible to determine the relationship between the coordinate expression level of the four genes of the midstream module and the lycopene pathway flux. Simultaneously, we can screen for strains that produce color-producing terpenoid pigments like carotene, astaxanthin, and others by optimizing the RBS of the entire MVA pathway."
} | 6,530 |
33796861 | PMC7968483 | pmc | 7,870 | {
"abstract": "Since the first invention of triboelectric nanogenerators (TENGs) in 2012, many mechanical systems have been applied to operate TENGs, but mechanical contact losses such as friction and noise are still big obstacles for improving their output performance and sustainability. Here, we report on a magnet-assembled cam-based TENG (MC-TENG), which has enhanced output power and sustainability by utilizing the non-contact repulsive force between the magnets. We investigate the theoretical and experimental dynamic behaviors of MC-TENGs according to the effects of the contact modes, contact and separation times, and contact forces (i.e., pushing and repulsive forces). We suggest an optimized arrangement of magnets for the highest output performance, in which the charging time of the capacitor was 2.59 times faster than in a mechanical cam-based TENG (C-TENG). Finally, we design and demonstrate a MC-TENG-based windmill system to effectively harvest low-speed wind energy, ~4 m/s, which produces very low torque. Thus, it is expected that our frictionless MC-TENG system will provide a sustainable solution for effectively harvesting a broadband of wasted mechanical energies.",
"introduction": "1. Introduction A “hyperconnected society” refers to the connection of all things, such as people, machines (devices), and space, wherein information is created, shared, and utilized between the connected objects [ 1 – 4 ]. The number of connected devices exceeded the world population in 2008, and it is expected to increase nearly four-fold by 2023 [ 5 ]. Connected devices such as PCs, TVs, smartphones, and wearable devices are already familiar, and they are connected via the Internet of Things (IoT) to provide services such as digital payments, online healthcare, and smart homes [ 6 ]. The IoT, first introduced as the US military's sensor network technology, has evolved over the past 40 years and is now commonly found in buildings, traffic systems, and smart cities [ 7 – 9 ]. In the IoT network, connected devices must be able to generate and transmit information anytime, anywhere. Therefore, they must be wireless devices to have lower cost, more mobility, and more scalability than wired devices. With the development of wireless communication technology, there are various solutions for data transmission without wires [ 10 ]. To bypass another wire, the power line, most connected devices currently use batteries. However, batteries have critical limitations such as limited lifetime, the need for frequent replacement or recharging, and attendant high cost and high risk for maintenance. Therefore, a connected device needs an independent, self-powered system for improved operation time and little or no maintenance. Triboelectric nanogenerators (TENGs) are emerging energy-harvesting devices that generate useful electric power by the movement of tribo-surfaces [ 11 – 18 ]. Since their debut in 2012, many novel designs and applications for harvesting energy from various mechanical sources have been introduced [ 19 – 26 ]. The power density has reached 1,200 Wm −2 , the volume density has reached 490 kWm −3 , and an energy conversion efficiency of 50%-85% has been demonstrated [ 27 ]. Hybrid nanogenerators have been also introduced to broadening the working frequency ranges by utilizing more than two energy harvesting mechanisms in one structure [ 28 , 29 ]. The triboelectrification between the tribo-surfaces in a TENG is caused by their contact and separation. The movements of the tribo-surfaces can be vertical motions repeated perpendicular to the tribo-surface, or sliding motion repeated horizontally to the tribo-surface. In the latter case, a tribo-surface can be a disc or a cylindrical structure, so repetitive contact and separation are possible even with the unidirectional motion of the energy source. Furthermore, adjustment of the triboelectrification frequency can be achieved by resizing the repeated tribo-surface, such as a grating structure, resulting in higher power output. However, since a sliding contact has to be made, the selection of materials is limited due to mechanical wear, and the design and manufacture of precise motion guides are required to maintain sliding contact between the tribo-surfaces during operation. In the case of vertical motion, the structure is simple and there is no sliding, so damage between tribo-surfaces is extremely small, leading to long unit life (i.e., high sustainability). However, unlike in sliding motion, there is a disadvantage in that an energy source that repeats the movement up and down is required, and the corresponding operating frequency is relatively low because the linear contact-separation motion is normally slow. In our previous study, a mechanical cam-based TENG (C-TENG) was implemented as a contact and separation mode by using high-speed rotational energy [ 30 ]. By using a rotating cam, repetitive triboelectrification was provided for the unidirectional motion of the energy source. The advantages of long unit life and high power output were simultaneously realized by changing the number of cams, which changed the frequency of the triboelectrification. In practical and industrial applications, the sustainability of a TENG is critical, and this is meaningful only if it has a longer lifespan than battery replacement or recharge cycles. Our C-TENG has less wear between the tribo-surfaces than a sliding-mode TENG, but it still has mechanical parts with limited life—the rotating cam and spring. Especially, the cam contact causes mechanical energy losses such as noise, impact, and further continuous wear that can lead to the destruction of the TENG (Figure S1(a-b) ). More importantly, such a mechanical cam has a limitation in its pushing force: a higher pushing force produces higher friction, which can cause the cam to stick (Figure S1(c) ). Herein, we propose a magnet-assembled, cam-based TENG (MC-TENG) which can provide superior sustainability and boosted output power. Instead of mechanical cam contacts, magnets provide non-contact pushing and repulsive forces to overcome the limitations of energy loss, noise, friction, and damage between the cam and the top plate. Furthermore, we can expect a significant enhancement of power output compared to a C-TENG because of the reduced contact-separation time (Δ t ) in magnetic contact mode, and an augmented pushing force without the stuck cam problem. We investigate these behaviors with a high-speed camera and magnified the time span of the output power. Finally, all the mechanical pushing and repulsive parts of the C-TENG were replaced by permanent magnets, thus resulting in high sustainability and boosted power. We also design and demonstrate an effective MC-TENG-based windmill system to harvest energy from low-speed wind producing very low torque. Thus, our MC-TENG could be considered a cornerstone technology to improve sustainability and boost power for effectively utilizing TENGs in practical environmental conditions.",
"discussion": "3. Discussion In this study, we explored the reduction of energy loss, noise, friction, shock, and damage using permanent magnets in a cam-based TENG. Even though a C-TENG is an effective design to harvest high-speed rotational energies, its poor structural sustainability might be a significant handicap in real environment applications. Therefore, we studied the performance of a magnetic force-driven, cam-based TENG (MC-TENG) considering various types of cam contact and types of repulsion springs, the contact-separation time, and the contact force. We developed a wear-free, sustainable MC-TENG by changing the mechanical interaction parts of a conventional mechanical C-TENG to magnetic interaction parts utilizing permanent magnets. We experimentally verified the relation between the contact force and the power output from the MC-TENG and found that a larger magnetic pushing force generated more electric power. The larger repulsive force of the magnetic spring produced less electric power, indicating that the lower repulsive force of the magnetic spring was better to boost the output power. The maximized electric power generation can be achieved through a suitable combination of permanent magnets that produce the maximum contact force which is the difference between the magnetic pushing force by the cam and the repulsive force by the magnetic spring. Unlike in a C-TENG, the electric power generation from an MC-TENG can be easily adjusted by replacing the permanent magnets of different sizes. It makes the design of TENG more flexible since any rearrangement or reassembly process is not required. We also demonstrated capacitor charging and LED bulb lighting performance to show the possibility of using an MC-TENG as the power source for a self-powered system. Finally, we applied an MC-TENG in a windmill system design to verify its ability to generate electric power. The benefits of its non-contact magnets and its ability to generate power at low energy input (low wind speed) proved its practical applicability. The effects of the cam diameter and the number of magnets on the power output were also demonstrated. Based on our results, we expect that our frictionless MC-TENG offers a more sustainable solution and ample output performance for effective applications in our real environment."
} | 2,329 |
24798206 | PMC4010524 | pmc | 7,871 | {
"abstract": "Ammonia-oxidizing archaea (AOA) are ubiquitous and abundant and contribute significantly to the carbon and nitrogen cycles in the ocean. In this study, we assembled AOA draft genomes from two deep marine sediments from Donghae, South Korea, and Svalbard, Arctic region, by sequencing the enriched metagenomes. Three major microorganism clusters belonging to Thaumarchaeota , Epsilonproteobacteria , and Gammaproteobacteria were deduced from their 16S rRNA genes, GC contents, and oligonucleotide frequencies. Three archaeal genomes were identified, two of which were distinct and were designated Ca . “Nitrosopumilus koreensis” AR1 and “Nitrosopumilus sediminis” AR2. AR1 and AR2 exhibited average nucleotide identities of 85.2% and 79.5% to N . maritimus , respectively. The AR1 and AR2 genomes contained genes pertaining to energy metabolism and carbon fixation as conserved in other AOA, but, conversely, had fewer heme-containing proteins and more copper-containing proteins than other AOA. Most of the distinctive AR1 and AR2 genes were located in genomic islands (GIs) that were not present in other AOA genomes or in a reference water-column metagenome from the Sargasso Sea. A putative gene cluster involved in urea utilization was found in the AR2 genome, but not the AR1 genome, suggesting niche specialization in marine AOA. Co-cultured bacterial genome analysis suggested that bacterial sulfur and nitrogen metabolism could be involved in interactions with AOA. Our results provide fundamental information concerning the metabolic potential of deep marine sedimentary AOA.",
"conclusion": "Conclusions Metagenomic analyses enabled the assembly of two distinct deep marine sediment-derived AOA genomes, AR1 and AR2, and the determination of genetic similarities and differences between these organisms and previously sequenced AOA. Many key genomic features were conserved between AR1 and AR2 and other AOA, including genes pertaining to energy metabolism and carbon fixation. Nevertheless, genomic variations were also apparent, including: 1) Large GIs comprising ∼15% of the total genomes were found in AR1 and AR2; 2) Approximately 24% of CDS in AR1 and AR2 were unique; and 3) High-affinity phosphate uptake genes were absent in AR1 and AR2. In addition, a urease operon was found in the AR2 genome, but not the AR1 genome, suggesting potentially distinctive strategies for resource utilization between the two deep marine sedimentary AOA strains. The availability of the genome sequences of deep marine sedimentary AOA will provide a foundation for evolutionary, biochemical, and ecophysiological studies that will contribute to the understanding of niche adaptations in marine AOA.",
"introduction": "Introduction Aerobic nitrification is a key process in the nitrogen cycle that converts ammonia to nitrate via nitrite and is catalyzed by aerobic autotrophic ammonia-oxidizing and nitrite-oxidizing microorganisms. The first step in autotrophic nitrification, the oxidation of ammonia, was long thought to be exclusive to Proteobacteria in the domain Bacteria \n [1] ; however, more recently, metagenomic analyses of terrestrial [2] and marine environments [3] revealed that ammonia oxidation is also associated with Archaea . Moreover, critical evidence for the existence of autotrophic ammonia-oxidizing archaea (AOA) was obtained through characterization of the first ammonia-oxidizing archaeon, Nitrosopumilus maritimus SCM1, which was isolated from a marine aquarium [4] . This discovery was followed by the successful cultivation of diverse AOA of Thaumarchaeota \n [5] , [6] from marine (group I.1a) [4] , [7] , [8] and soil (group I.1a and I.1b) [9] – [11] environments. Furthermore, molecular ecological studies indicate that AOA often predominate over ammonia-oxidizing bacteria in marine environments such as the North Sea and coastal sediments [8] , [12] . The seafloor comprises approximately two-thirds of the Earth’s surface and is therefore one of the most extensive of all microbial habitats. Quantitative assessments of subsurface microbial populations indicate that prokaryotes constitute a large portion of the Earth’s overall biomass, and that marine sediment processes may therefore substantially contribute to the global nitrogen budget. Research into nitrification, a key step in the nitrogen cycle, has focused on water-column, and studies regarding marine sediment nitrification are minimal. Investigations into the metabolic properties and nitrification potential of sedimentary AOA are therefore necessary to understand the nitrogen cycle in marine environments. Fundamental information about microorganisms and their metabolic features can be revealed via metagenomic and genomic techniques. Analysis of the genome sequence of an amoA -encoding archaeon Ca . “Cenarchaum symbiosum” from a marine sponge [13] , [14] and a marine ammonia-oxidizing archaeon N . maritimus \n [15] provided valuable insights into the evolution of nitrogen and carbon metabolism in marine AOA of the Nitrosopumilus lineage (also called group I.1a). Comparative analyses of group I.1a AOA genome sequences from low-salinity aquifers and terrestrial environments have revealed several genetic traits likely to be adaptations to such habitats, such as motility and protection from osmotic stress [16] , [17] . AOA metagenomic information obtained from the water column of the Gulf of Maine has shed light on the metabolic potential of planktonic AOA [18] . Although the genomes of two AOA enriched from low-salinity sediments have been sequenced [19] , [20] , genomic data from deep marine sedimentary AOA are not yet available. AOA are widespread and dominant ammonia-oxidizers in marine sediment [12] . One of the main difficulties in obtaining axenic AOA cultures is their dependence on co-cultured bacteria, as described in AOA characterization reports [10] , [11] , [21] , [22] . Sedimentary AOA were, however, successfully enriched when co-cultured with sulfur-oxidizing bacteria (SOB) in a technique that facilitated characterization of the AOA [7] . Here, we analyzed metagenomes from enrichment cultures and were able to assemble the genomes of two deep marine sedimentary AOA. The aims of this study were to investigate the genomic features of deep marine sedimentary AOA through comparisons with the genomes of other AOA and to assess possible microbial interactions between deep marine sedimentary AOA and co-cultured bacteria.",
"discussion": "Results and Discussion Metagenome analysis, assembly, and binning We obtained 536.8 Mb and 308.2 Mb of metagenomic sequences from two independently enriched ammonia-oxidizing cultures containing thaumarchaeotal group I.1a archaeal strains, named AR (from Svalbard, Arctic region) and SJ (from Donghae, South Korea), respectively. General features of the metagenome datasets are as indicated in Table S1 . The GC% profiles of the raw reads from the two enrichment metagenomes were very similar to one another ( Figure S1 ). Single reads of 16S rRNA genes recovered from the metagenome dataset (n = 1,100 in AR and n = 908 in SJ cultures) were used to analyze the compositions of the microbial communities that were enriched in the two cultures ( Figure S2 ). The most frequently recovered 16S rRNA gene sequences were affiliated to Epsilonproteobacteria (60–62%), Thaumarchaeota (13–17%), and Gammaproteobacteria (10–18%), with the proportions of these three taxa being similar in the two cultures ( Figure S2 ). Most of the 16S rRNA gene sequences of Epsilonproteobacteria were affiliated with the sulfur-oxidizing genus Sulfurovum . More than 10% of the 16S rRNA gene reads from each metagenome were affiliated with Thaumarchaeota , and, specifically, the genus Nitrosopumilus . Gammaproteobacteria sequences were related to those of diverse Gammaproteobacteria (e.g., Marinobacter , Marinobacterium , and Neptuniibacter ). Overall, this analysis suggested that the proportion of 16S rRNA genes from archaea was approximately 20%, which was lower than the proportion of archaea observed by previous fluorescence in situ hybridization analysis of the SJ and AR cultures [7] . This discrepancy could have arisen due to the presence of multiple rRNA operons in bacterial genomes [23] by contrast with the single rRNA operon in the genome of N . maritimus ( Thaumarchaeota ) [15] . Indeed, Nakagawa et al. [24] reported that the genome of Sulfurovum sp. NBC37-1 ( Epsilonproteobacteria ), a close relative of the dominant bacterium in the SJ and AR cultures, has three copies of the rRNA operon. Data obtained from 16S rRNA gene reads were complemented by comparing the entire metagenome dataset of functional genes to homologous genes of known microbial genomes using the MG-RAST server ( Figure S3 ). Assembly of the metagenomic data produced 15,155 and 2,595 contigs from the AR and SJ metagenomic sequences, respectively ( Table S1 ). We filtered the contigs, selecting only those that were ≥ 5 Kb in length (n = 118 for AR and n = 91 for SJ) and which yielded consistent hits to a single high-level taxon (e.g., Thaumarchaeota , Epsilonproteobacteria and Gammaproteobacteria ). An examination of GC% versus length in the selected contigs indicated they comprised three clusters ( Figure S4 ). Moreover, principal component analysis of the oligonucleotide frequencies also revealed three distinct clusters in each enriched sample ( Figure 1 ). Based on BLAST analysis of the genes, we assigned clusters 1, 2, and 3 to Thaumarchaeota , Epsilonproteobacteria , and Gammaproteobacteria , respectively, which was consistent with results obtained from the 16S rRNA analysis ( Figure S2 ). The GC% range in cluster 1 ( Thaumarchaeota ) ( Figure 1 ) was similar in both the AR and SJ assemblies (27–37% in AR and 32–35% in SJ). With the exception of Ca . “C. symbiosum” (57%) [13] and Ca . “Nitrososphaera gargensis” (48%) [25] , all other previously analyzed AOA, including N . maritimus , had GC contents of 32–34% [15] – [17] . The amounts of sequence obtained for cluster 1 differed between the two clusters: 3.44 Mb in AR and 1.65 Mb in SJ. Considering the size of the N . maritimus genome (1.64 Mb), the 1.65 Mb size of the archaeal cluster from the SJ metagenome assembly potentially represented a draft genome of a single AOA. However, the 3.44 Mb of contigs in cluster 1 of the AR metagenome suggested that two putative archaeal draft genomes had been assembled. 10.1371/journal.pone.0096449.g001 Figure 1 Principal component analysis of oligonucleotide frequencies in assembled contigs from two archaeal enrichment cultures (A) AR culture, and (B) SJ culture. Reference genomes are shown as larger circles. The total number of contigs for each group ( Gammaproteobacteria , Epsilonproteobacteria, and Thaumarchaeota ), total length, mean length, and GC content range are also indicated. The contig types and published genomes are as follows: orange, Gammaproteobacteria ; yellow, Thaumarchaeota ; green, Epsilonproteobacteria ; light green, assembled contigs including viral coding sequences; gray, not identified; red, Ca . “Cenarchaum symbiosum” A (CsymA); fuchsia, Ca . “C. symbiosum” B (CsymB); lime, Nitrosopumilus maritimus SCM1 (Nmar); blue, Ca . “Nitrosoarchaeum koreensis” MY1 (MY1); cyan, Ca . “Nitrosoarchaeum limnia” (Nlim); violet, Ca . “Nitrososphaera gargensis” (Ngar); teal, Sulfurovum sp. NBC37-1 (Sul); and purple, Thiomicrospira crunogena XCL-2 (Tcr). The GC content of cluster 2 was approximately 43%, which corresponded to that of Sulfurovum sp. NBC37-1 (43.8%) [24] . The expected genome size of cluster 2 (2.12 Mb) was slightly smaller than that of Sulfurovum sp. NBC37-1 (2.56 Mb). We were unable to detect the 16S rRNA gene within cluster 3, which contained the gammaproteobacterial contigs, and so were unable to definitively determine phylogenetic position. BLAST analysis indicated that cluster 3 contig genes were most similar to genes in Gammaproteobacteria genomes such as Oceanospirillum . The Average Nucleotide Identity (ANI) [26] of the gamma- and epsilonproteobacterial clusters in the two metagenome sets indicated that they were nearly identical (∼99%). Some features of the binned contigs from both metagenomic datasets are summarized in Table 1 . 10.1371/journal.pone.0096449.t001 Table 1 Features of binned contigs for genomes of thaumarchaeota, epsilon- and gammaproteobacteria (≥ 5 Kb contigs). \n Thaumarchaeota \n \n Epsilonproteobacteria \n \n Gammaproteobacteria \n AR SJ AR SJ AR SJ Size (Mbp) 3.44 1.65 2.12 2.12 0.47 3.00 No. of predicted genes 4,148 1,934 2,136 2,138 512 2,907 No. of contigs 58 15 11 13 49 63 Average contig size (Kb) 59 110 193 163 9 47 Average GC content (%) 33.83 34.31 39.37 39.39 52.37 53.42 Average gene length (bp) 737 760 903 903 818 924 Coding percentage (%) 88.9 89.1 91.0 91.0 89.4 89.7 Genome coverage (X) 34 42 71 67 7 12 RNA genes 23S 2 1 ND ND ND ND 16S 2 1 ND ND ND ND 5S 2 1 ND ND ND ND ND, not detected. Establishing draft genome assemblies for three deep marine sedimentary archaea and defining their unique characteristics The binning and assembly procedures described above were used to define three AOA draft genomes. We hypothesized that the cluster 1 (thaumarchaeotal) sequences from culture AR (3.44 Mb) represented two genomes, henceforth termed AR1 and AR2. Cluster 1 sequences from culture SJ (1.65 Mb) appeared to represent a single genome. Genomic diversity in a microbial population can be determined by analyzing sequence variations in metagenome reads. We used the Strainer program ( http://www.bioinformatics.org/strainer/wiki/ ) to assess variation in the archaeal populations of the metagenome datasets. Archaeal diversity in the AR and SJ cultures was assessed by analyzing the ammonia monooxygenase gene (ammonia monooxygenase alpha subunit, amoA ), which is involved in ammonia oxidation, and the 16S-23S rRNA intergenic spacer (ITS) region. The amoA and ITS sequences were examined in raw reads (data not shown), and the results fully supported the above hypothesis that the metagenomic data captured a single draft archaeal genome in the SJ culture and two draft archaeal genomes in the AR culture. Archaeal contigs in the AR culture clearly separated into two distinct groups based on contig alignment with N . maritimus using Mauve [27] and ANI analysis with N . maritimus . We propose that our assembled genomes warrant draft genome status for the following reasons: (i) Each draft genome features 97–98% of the archaeal genes used by the NIH Human Microbiome Project as criteria for complete draft genomes ( http://hmpdacc.org/tools_protocols/tools_protocols.php ) [28] . These archaeal genes are known to be highly conserved between the genomes of free-living Archaea and comprise 104 core gene groups. Additionally, the majority of the core archaeal genes are found in the complete or nearly complete genomes of several published AOA ( Ca . “C. symbiosum”, 92%; Ca . “Na. koreensis”, 98%; N . maritimus , 100%; and one exception, Ca . “N. gargensis”,74%); (ii) The two draft genomes of SJ and AR1 were independently sequenced and assembled but were nearly identical to one other, as recognized by gene content and synteny comparisons; (iii) A high degree of genomic similarity was observed between the three draft archaeal genomes and the completed N . maritimus genome. Furthermore, the number of tRNAs (n = 44) was identical in the draft genomes of SJ, AR1, and AR2, and the complete genome of N . maritimus . The two AR1 and AR2 archaeal genomes exhibited approximately 80% ANI with each other and ANIs of 85.2% and 79.5% with N . maritimus , respectively. The ANI of the AR1 archaeal bins with those of the SJ culture was ∼99%; no significant differences were observed between the SJ and AR1 archaeal contigs with respect to gene content or local synteny. On the basis of these results, we concluded that the SJ and AR1 assembled archaeal genomes were indistinguishable and might have originated from very closely related microorganisms. Therefore, our further analyses focused on two of the three archaeal genomes: AR1 (synonymous with the archaeon from culture SJ) and AR2. Despite the strong similarities (>99.5%) between the 16S rRNA gene sequences in N . maritimus and in the AOA obtained from our enrichments ( Table S2 and Figure S5 ), the low ANI (<85%) indicates high genomic variation within this cluster of marine AOA. The proposed cutoff for defining separate species is 94% ANI between two genome sequences [26] . This criterion suggests that each archaeal strain (AR1 and AR2) can be considered a separate species distinct from N . maritimus . We propose that these genomes represent two new marine AOA within the genus Nitrosopumilus, named Ca . “Nitrosopumilus koreensis” (AR1 and SJ) [29] and “Nitrosopumilus sediminis” (AR2) [30] . Genetic differences between AOA genomes and their adaptive implications Most of the putative coding sequences (CDS) in the AR1 and AR2 genomes (71.9% and 65.1%, respectively) had homology to N . maritimus genes, and most of the genes were syntenic with those in the N . maritimus genome ( Figure S6 ). However, 20.5% and 24.4% of the putative CDS of the AR1 and AR2 genomes, respectively, had no similarity to genes in other known organisms. We hypothesized that the adaptive traits of deep sedimentary AOA in our enrichment cultures might contrast with those of water-column AOA. To address this, a recruitment analysis was performed in which nucleotide-sequence fragments from the planktonic Sargasso Sea metagenome dataset of the global ocean sampling (GOS) database [3] were mapped onto the AR genomes ( Figure S7 ). Many of the genes that were present in the AR genomes but absent in the Sargasso Sea metagenome dataset were clustered in genomic islands (GIs) of >15 Kb ( Figure S7 , and Tables S3 and S5 ). GIs were a major feature of the AR1 and AR2 genomes ( Tables S3 and S5 ) and comprised approximately 15% of the total AR1 (six GIs) and AR2 genomes (12 GIs). Most of the GIs in the AR1 and AR2 genomes were different from one another and were absent from the N . maritimus genome, and gene functions can be putatively inferred for approximately half of the genes in the GIs. Most GI genes in both the AR1 and AR2 genomes were related to cell-wall biosynthesis, osmotic stress tolerance, antibiotic resistance, sensory signal transduction, and phage proteins. In addition, the GIs of both genomes comprised genes with high anomalies in codon usage, indicating that they might have been obtained via horizontal transfer events, as suggested by Rusch et al. [31] . The Clusters of Orthologous Genes (COG) classification of the GI genes from the two genomes indicated that genes belonging to COG class M (cell wall/membrane/envelope biogenesis), K (transcription), and T (signal transduction mechanisms) were abundant ( Figure S8 ). This is in partial contrast to the COG classes found in the GIs of other archaeal genomes, which are predominantly M or Q (secondary metabolite biosynthesis, transport, and catabolism) [32] . The proteinaceous surface layers of AOA have an abundance of reactive surface sites that are conceivably related to their oligotrophic adaptations [33] . The frequent observation of COG class M genes in the GIs of the AR1 and AR2 genomes could contribute to variations in cell surface structure, which might be important factors for niche specialization in AOA ecotypes. Overall, the identified GIs might constitute strain-specific (hyper)variable regions or sedimentary AOA-specific regions. Ammonia oxidation, electron transfer, and carbon fixation for the deep marine sedimentary AOA Pathways for ammonia oxidation, electron transport, and carbon fixation were assembled from the AR1 and AR2 archaeal genomes and compared with other reference AOA genomes. The AR1 and AR2 archaeal strains held key metabolic traits in common with other AOA, including N . maritimus ( Table S4 ). Ammonia oxidation and electron transport chain All of the putative ammonia monooxygenase genes ( amo ; amoA , amoB , and amoC ) were found in the AR1 and AR2 genomes. The gene arrangement [ amoA -hypothetical gene (named amoX )- amoC - amoB ] was similar to that in other AOA of the Nitrosopumilus cluster (e.g., N . maritimus ) as well as into Ca . “N. devanaterra” [34] , but differs from the gene arrangements in group I.1b AOA [9] , [25] . For example, the amo genes in some group I.1a marine lineages and in most of the soil lineages (group I.1b) were not consecutive, but were interrupted by other genes. In most AOA, another small protein encoding a transmembrane protein and referred to as amoX was linked to the amoA gene [35] . Although AOA produce nitrite as the final product of ammonia oxidation, homologs of the heme-containing hydroxylamine oxidoreductase ( hao ) gene of ammonia-oxidizing bacteria (AOB) were absent from the AR1 and AR2 genomes, as in other AOA genomes [14] , [15] , [17] , [25] . However, Vajrala et al. [36] observed hydroxylamine-induced oxygen consumption and ATP production in the marine ammonia-oxidizing archaeon N . maritimus . The number and sequences of six putative genes encoding copper-containing oxidases, which were suggested to function as possible hydroxylamine oxidoreductases (HAOs) [15] , were conserved between N . maritimus and strains AR1 and AR2, encoding proteins with 88% amino acid identity on average. The number of putative genes encoding copper-containing oxidases found in the AOA genomes was six for Ca . “N. gargensis” and 3–4 for Ca . “Na. koreensis”, Ca . “Na. limnia”, and Ca . “C. symbiosum”. A putative gene for copper-containing oxidase was highly conserved (average 83% amino acid identity) between soil strain Ca . “Na. koreensis” (MY1_0289) and the marine AOA genomes (Nmar_1131, AR1_298, and AR2_318), and warrants further investigation as a possible HAO candidate. The other putative copper-containing oxidase gene, nirK , was highly conserved in all AOA, which might be involved in nitrifier denitrification [37] . A TATA box and parts of a BR element (transcription factor B recognition element), 23 nt or 25 nt upstream of the nirK gene ( Figure S9 ), were observed as in the archaeal amo gene [35] , suggesting that the nirK gene could be expressed independently under the control of its own promoter. As in other AOA genomes, strains AR1 and AR2 appear to encode a complete respiratory chain with complexes I–V, which are used for energy generation and reverse electron transport. The components have ∼93% amino acid identity to those of N . maritimus . Complex V is an archaeal type ATPase that is known to use both Na + and proton gradients to generate ATP [38] . Na + is frequently used instead of H + in gradient formation during electron transport in oligotrophic or energy-stressed environments, since Na + is usually less permeable to the cellular membrane. Like other AOA genomes, the genomes of AR1 and AR2 lack homologs of cytochrome c proteins [15] – [17] , [25] , and therefore blue copper-containing proteins ( Table S6 ) might be involved in the transfer of electrons from complex III. Known homologs encoding essential genes for heme biosynthesis ( ahb-nirJ1 and ahb-nirJ2 ) were missing [39] and putative genes for heme-containing proteins were rare in the AOA genomes. The only heme-containing gene detected in the AOA genomes (including AR1 and AR2) was that encoding the cytochrome b/b6 family protein of respiratory complex III. Since heme uptake by prokaryotes from the environment is not plausible [40] , AOA genomes require further screening and analysis to characterize gene sets for heme biosynthesis. The variability in iron availability in marine and terrestrial environments suggests that the abundance of copper-containing oxidases for redox reactions in both soil (e.g., Ca . “Na. koreensis”) and marine AOA might be an evolutionary trait of Thaumarchaeota rather than a functional or environmental adaptation of the AOA. The high abundance of multicopper-containing proteins and blue copper-containing proteins in AOA, rather than heme-containing proteins, implies that ammonia oxidation pathways and respiratory chains in AOA groups I.1a and I.1b may be novel and conserved. Carbon fixation Most AOA characterized to date are able to grow chemolithotrophically using inorganic carbon (carbon dioxide and/or bicarbonate) as their sole carbon source [4] , [7] , [9] – [11] , [22] . By contrast with their bacterial counterparts, AOA genomes do not contain key genes for the Calvin-Bassham-Benson cycle [41] , [42] , but might instead utilize the 3-hydroxtpropionate/4-hydroxybutyrate pathway. The genes encoding the three main proteins for this pathway, 4-hydroxybutyrate-CoA dehydratase, acetyl-CoA carboxylase, and methylmalonyl-CoA epimerase, were present in the AR1 and AR2 genomes and the putative proteins had 80–98% amino acid identity to the N . maritimus homologs. Stable-isotopic and molecular studies raised questions regarding the mixotrophic nature of the marine lineage of group I.1a [43] , [44] . Ammonia oxidation and growth of N . viennensis (a soil lineage of group I.1b) was supported by pyruvate and some pyruvate carbons were incorporated into archaeal cells [9] . Genes encoding proteins that are possibly involved in the transport of organic compounds, such as carbohydrates, amino acids, oligo/dipeptides, and nucleosides, were evident in the AR1 and AR2 genomes and in other AOA genomes. However, there has been no direct biochemical and physiological evidence from cultivated AOA to support the hypothesis that the marine lineage of group I.1a is mixotrophic. The Ca . “N. gargensis” genome encodes alanine dehydrogenase and an array of pyruvate transformation genes [25] , suggesting that Ca . “N. gargensis” might utilize pyruvate or alanine as an alternative carbon source, by contrast with other AOA. Pyruvate phosphate dikinase, which is involved in the transformation of pyruvate to phosphoenolpyruvate for gluconeogenesis, was encoded in the genomes of marine AOA, including the AR1 and AR2 strains. Genomic traits of the deep marine sedimentary AOA Urea utilization A complete set of genes involved in urea utilization was identified in the AR2 genome ( Figure 2 ). This was absent from other marine (AR1 and N . maritimus ) and soil/low-salinity AOA ( Ca . “Na. koreensis” and Ca . “Na. limnia”) genomes. Urease operons were identified in the genomes of Ca . “C. symbiosum” [14] , N . viennensis \n [45] , Ca . “N. salaria” [19] and Ca . “N. gargensis” [25] , and in a scaffold from a recent ocean metagenomic study [18] , with 46–86% amino acid identities to the AR2 operon, respectively. Moreover, two copies of a urea transporter gene were identified in the AR2 genome that were 50–76% identical to the dur3 gene from Ca . “C. symbiosum”, Ca . “N. gargensis”, and to the dur3 gene from the Pacific Ocean metagenome recovered from a 4,000 m depth at station ALOHA [46] . A recruitment analysis comparing the AR2 genome to a Sargasso Sea metagenome showed that the archaeal urease utilization trait was widespread in water-column archaea. Since urea comprises a significant proportion of the dissolved nitrogen compounds in the surface layer of marine sediment [47] , the capacity for urea utilization within sedimentary AOA may confer a selective advantage within that niche. Moreover, Alonso-Sáez et al. [48] suggested that deep water Thaumarchaeota in the Arctic and Antarctic oceans use urea as an energy source in nitrification. 10.1371/journal.pone.0096449.g002 Figure 2 Comparison of the Ca . “Nitrosopumilus sediminis” AR2 genomic region containing genes for urea utilization with those of Ca . “Cenarchaeum symbiosum” and environmental metagenomes. \n Ca . “N. sediminis” AR2 genome is central, with the Ca . “C. symbiosum”, Ca . “Nitrososphaera gargensis”, and environmental metagenomic regions above and below, respectively. Homologous genes are connected with shaded regions, and the shaded color indicates the percent identity as determined by TBLASTX. Ectoine synthesis Ectoine is a compatible solute that is found in a wide range of bacteria. The AR1 and AR2 genomes (as well as that of N . maritimus \n [49] ) contained all four genes in the archaeal ectoine biosynthesis cluster ( ectA , ectB , ectC , and ectD ). In AR1 and AR2, the ectoine gene clusters were located in the centers of GI 6 and GI 3, respectively and the codon usage in these islands deviated markedly from the conserved core genes in the AR genome ( Table S3 ). Recruitment analysis did not find ectoine biosynthesis genes in the Sargasso Sea metagenome or the Ca . “Na. limnia”, Ca . “Na. koreensis”, Ca . “N. gargensis”, or Ca . “C. symbiosum” genomes [13] , [16] , [17] , [25] . Instead, Ca . “Na. limnia”, Ca . “Na. koreensis”, and Ca . “N. gargensis” employ mechanosensitive ion channels (MS channels; mscS and mscL genes) for regulating osmotic pressure. The AR1, AR2, and N . maritimus genomes also harbored genes for a small-conductance MS channel ( mscS ), but no large-conductance MS channel gene ( mscL ) was apparent; thus the ability to synthesize ectoine might be an important osmotic adaptation in members of the genus Nitrosopumilus . Clustered regularly interspaced short palindromic repeats (CRISPRs)/Cas system The CRISPR/Cas system mediates resistance against phages, and is found in the majority of investigated Archaea genomes [50] . Possible spacer-repeat arrays were identified in the AR1 (n = 3) and AR2 (n = 1) genomes, but only a single CDS exhibited similarity to a gene encoding a Cas protein (CAS1-like) (see GI 4 and 6, respectively, in Table S3 ). It is unclear whether the putative CRISPR spacers observed in AR1 and AR2 are artifacts or instead represent remnants of previous CRIPSR-loci. By contrast with the wide distribution of CRISPR in archaea, only one thaumarchaeon ( Ca . “N. gargensis”) has so far been found to contain a CRISPR-locus and associated CAS-genes [25] . Phosphate assimilation High-affinity phosphate uptake genes are often found in AOA, including the recently published Ca . “N. gargensis” genome [25] , but we were unable to identify a high-affinity, high-activity phosphate uptake operon ( pstSCAB ) in either of the AR1 or AR2 genomes. The absence of these genes in the deep marine sedimentary AOA metagenome datasets may reflect habitat-specific circumstances. It is likely that sufficient phosphate is available in marine sediment as phosphate levels up to 100 µM were previously noted [51] ; this is 50-fold higher than phosphate concentrations in the marine water column (∼2.0 µM) [52] . Chlorite degradation Perchlorate (ClO 4 \n − ), chlorate (ClO 3 \n − ) and chlorite (ClO 2 \n − ) are important pollutants in groundwater, surface waters, and soils [53] . Several (per)chlorate-reducing bacteria, including Dechloromonas aromatic, Idenella dechloratnas, and nitrite-oxidizing bacteria [54] , contain a cld gene, which encodes enzymes that degrade chlorite (ClO 2 \n − ) to chloride (Cl − ) and oxygen (O 2 ). Although cld genes are not present in AOB genomes, they are contained in all AOA genomes examined to date, including the AR1, AR2, N . maritimus , Ca . “Na. koreensis”, Ca . “Na. limnia”, Ca . “N. gargensis”, and Ca . “C. symbiosum” genomes. Cld proteins in AR1 and AR2 exhibited 35–68% and 50–87% identity, respectively, with those of other AOA. Cld in AOA may be necessary for chlorite detoxification, since chlorite is a selective inhibitor of ammonia oxidation [55] . This concurs with our previous results [7] , [10] showing that group I.1a AOA tolerated higher concentrations of chlorite than Nitrosomonas europaea \n [7] , [10] , [55] . Genomic features of co-cultured SOB Successful cultivation of sedimentary AOA reportedly depends upon co-cultivation with SOBs [7] . Epsilonproteobacterial and gammaproteobacterial genomes were major constituents of the AR and SJ culture sequences, as detailed herein. Because the metagenomic features of the Epsilonproteobacteria (cluster 2) and Gammaproteobacteria (cluster 3) from the AR and SJ cultures were nearly identical (reciprocal ANI 99%), we selected epsilonproteobacterial (cluster 2) and gammaproteobacterial (cluster 3) bins from the AR and SJ cultures, respectively, for further analysis. These are designated “EP_AR” and “GM_SJ”, and their metabolic capabilities as determined by genomic analysis are discussed below and summarized in Table S4 . Strain EP_AR was affiliated with chemolithoautotrophic SOB. Several key enzymes involved in sulfur oxidation (e.g., sulfur-compounds oxidation system, SOX) were encoded within the EP_AR genome [56] ( Table S4 ). The putative SOX proteins had 55–92% amino acid identity to those of the close relatives Sulfurovum sp. NBC37-1 [24] and Sulfurimonas denitrificans DSM 1251 [57] . Strain GM_SJ resembled a typical marine heterotroph since no genes related to sulfur oxidation or carbon fixation were observed in the genome ( Table S4 ). Microbial interactions play a critical role in shaping niches for microorganisms in natural environments. Sedimentary AOA and SOB occupy similar niches in sediment redox gradients [58] , since AOA and SOB at oxic-anoxic interfaces consume ammonia and sulfide, respectively, diffused from the anoxic layers of marine sediment. Joye and Hollibaugh [59] reported that sulfide (<100 µM) inhibits nitrification in marine sediments. The prevalence of AOA may therefore be assisted by SOB detoxification of sulfides. The unusually tight associations between AOA and SOB were described in a terrestrial cold sulfidic spring [60] , and thaumarchaeotal strains were physically associated with SOB in sulfide-rich mangrove swamps [61] . Sulfide-quinone reductase ( sqr ), sulfite:cytochrome c oxidoreductase ( dsrAB ), and the SOX system genes ( soxYZABCFHL ) in the EP_AR genome could mediate sulfide oxidation reactions [62] . This suggests that strain EP_AR might be a natural co-habitant of sedimentary AOA, and, although we used thiosulfate instead of sulfide for enrichment in this study [7] , interactions between SOB and AOA might be exploited for the successful enrichment of SJ and AR in the laboratory. AOB have a low efficiency for N 2 O production during nitrifier denitrification and most NO is emitted to an extracellular environment [63] , [64] . Excess NO is therefore potentially toxic to the nitrifier itself and to other bacteria. Nitric oxide is suggested as an intermediate during bacterial [65] , [66] and archaeal nitrification. Archaeal NO production was suggested by genomic analysis [67] in this study and by Walker et al. [15] and is supported by the inhibition of AOA by NO scavengers [68] . N 2 O emissions during archaeal ammonia oxidation [69] , [70] provide indirect evidence of the involvement of NO in archaeal nitrifier denitrification [10] , [11] . A putative gene encoding toxic NO-detoxifying flavohemoglobin [NO dioxygenase, NOD, 51.4% amino acid identity with that in Aquifex aeolicus VF5 [71] ] was observed in strain EP_AR ( Figure S10 ), while no homolog was found in the genome of the closest relative, Sulfurovum sp. NBC37-1 ( Table S4 ). A gene-encoding phage integrase [48% amino acid identity with that in Sulfurimonas denitrificans \n [57] ] located upstream of the NOD gene suggests that the NOD gene may have been acquired through horizontal gene transfer. Catalytic NO dioxygenation occurs most effectively via NOD under aerobic conditions [72] , while nitric oxide reductase would be active under anoxic conditions [73] . The NOD in co-cultured SOB might therefore play a role in stimulating AOA growth. Genomic analysis of co-cultured SOB suggested that sulfur and nitrogen metabolism might be involved in the interactions between sedimentary AOA and co-cultured bacteria. Further systematic investigations are warranted to determine the response of sedimentary AOA to nitric oxide scavengers and generators."
} | 9,068 |
25258165 | null | s2 | 7,872 | {
"abstract": "The use of lignocellulosic biomass as a feedstock for microbial fermentation processes presents an opportunity for increasing the yield of bioproducts derived directly from glucose. Lignocellulosic biomass consists of several fermentable sugars, including glucose, xylose, and arabinose. In this study, we investigate the ability of an E. coli Δpgi Δzwf mutant to consume alternative carbon sources (xylose, arabinose, and glycerol) for growth while reserving glucose for product formation. Deletion of pgi and zwf was found to eliminate catabolite repression as well as the ability of E. coli to consume glucose for biomass formation. In addition, the yield from glucose of the bioproduct D-glucaric acid was significantly increased in a Δpgi Δzwf strain."
} | 189 |
31822352 | null | s2 | 7,875 | {
"abstract": "Bacterial natural products (NPs) and their analogs constitute more than half of the new small molecule drugs developed over the past few decades. Despite this success, interest in natural products from major pharmaceutical companies has decreased even as genomics has uncovered the large number of biosynthetic gene clusters (BGCs) that encode for novel natural products. To date, there is still a lack of universal strategies and enabling technologies to discover natural products at scale and speed. This review highlights several of the opportunities provided by genome sequencing and bioinformatics, challenges associated with translating genomes into natural products, and examples of successful strain prioritization and BGC activation strategies that have been used in the genomic era for natural product discovery from cultivatable bacteria."
} | 212 |
21656812 | null | s2 | 7,876 | {
"abstract": "Persistence and survival under various environmental stresses has been attributed to the capacity of most bacteria to form biofilms. In aquatic environments, the symbiotic bacterium Vibrio fischeri survives variable abiotic conditions during its free-living stage that dictates its ability to colonize the squid host. In the present study, the influence of different abiotic factors such as salt concentration, temperature, static/dynamic conditions, and carbon source availability were tested to determine whether biofilm formation occurred in 26 symbiotic and free-living V. fischeri strains. Statistical analysis indicate that most strains examined were strong biofilm producers under salinity concentrations that ranged between 1-5%, mesophilic temperatures (25-30 °C) and static conditions. Moreover, free-living strains are generally better biofilm formers than the symbiotically competent ones. Geographical location (strain origin) also correlated with biofilm formation. These findings provide evidence that abiotic growth conditions are important for determining whether mutualistic V. fischeri have the capacity to produce complex biofilms, allowing for increased competency and specificity during symbiosis."
} | 304 |
40229808 | PMC11998424 | pmc | 7,878 | {
"abstract": "Rhodococcus jostii RHA1 is an oleaginous bacterium that has attracted considerable attention due to its capacity to use different carbon sources to accumulate significant levels of triacylglycerols that might be converted into biofuels. However, this strain cannot transform xylose into lipids reducing its potential when growing on saccharified lignocellulosic biomass. In this work, we demonstrate that wild type R. jostii RHA1 can be evolved by adaptive laboratory evolution (ALE) to metabolize xylose without engineering heterologous metabolic pathways in the host. We have generated a phenotypically adapted ALE-xyl strain able to use xylose as the sole carbon and energy source more efficiently that an engineered recombinant strain harbouring heterologous xylA and xylB genes encoding a xylose isomerase metabolic pathway. The R. jostii RHA1 ALE-xyl strain accumulates lipids very efficiently using xylose as substrate, but even more importantly it can consume glucose and xylose at the same time. Transcriptomic analyses of ALE-xyl strain growing with glucose or xylose revealed the existence of a silent pentose metabolizing operon that is overexpressed in the presence of xylose. The detection of a xylose reductase activity together with the presence of xylitol in the cytoplasm of ALE-xyl strain suggests that xylose is consumed by a reductase pathway. This study demonstrates that, in cases where a clear phenotypic selection method is available, ALE can be used to improve very efficiently industrial microbial strains without using genetic engineering tools. Strategies focused to exploit the silent phenotypic flexibility of microorganisms to metabolize different carbon sources are powerful tools for the production of microbial value-added products using saccharified lignocellulosic wastes. Supplementary Information The online version contains supplementary material available at 10.1186/s13036-025-00503-1.",
"conclusion": "Conclusions This work shows that the oleaginous strain R. jostii RHA1 can acquire by ALE the ability to metabolize xylose without adding heterologous genes. Our results demonstrate that RHA1 contains silent capabilities to metabolize pentoses that can be unsilenced by adaptation to a xylose medium. The evolved ALE-xyl strain is no only able to efficiently use xylose as carbon source to grow and produce lipids, but also capable of using glucose and xylose at the same time to accumulate lipids with a high efficiency. Therefore, this evolved strain emerges as a promising microbial factory for industrial bio-oils production from saccharified lignocellulosic materials.",
"introduction": "Introduction Hemicellulose represents a group of complex polysaccharides that is the second largest category of plant cell wall polysaccharides in nature after cellulose. It has a heterogeneous chemical structure, composed of glucose, xylose, and other sugars in less proportion such as L-arabinose. Xylose derived from hemicellulose can account for 18–30% of the monosaccharides present in lignocellulose hydrolysates, and therefore, it is considered a relevant renewable resource [ 1 ]. However, only a limited number of bacteria are able to metabolize xylose in order to produce different bio-based products as biofuels and chemicals [ 2 ]. Microorganisms have evolved different metabolic pathways to assimilate xylose [ 1 ]. The isomerase pathway is the most commonly present in bacteria (e.g. Escherichia coli , Bacillus subtilis ), where xylose is initially converted to xylulose by xylose isomerase (XylA), then xylulose is phosphorylated by xylulokinase (XylB) to xylulose-5-phosphate which is metabolized by the pentose phosphate pathway or cleaved into acetyl phosphate and glyceraldehyde-3-phosphate by a phosphoketolase [ 1 ] (Fig. 1 ). In contrast, most yeasts and mycelial fungi (e.g., Saccharomyces cerevisiae , Candida spp. ) metabolize xylose through the reductase pathway, where xylose is converted into xylitol and subsequently into xylulose which is further phosphorylated by a xylulose kinase. Xylose can also be metabolized by a non-phosphorylative metabolic pathway, called the Weimberg pathway, discovered a few years ago in Caulobacter crescentus , which uses α-ketoglutarate as a key intermediate [ 3 ] (Fig. 1 ). Moreover, xylose can be converted into pyruvate and glycolaldehyde, using 2-keto-3-deoxy-xylonate as intermediary, through the Dahms pathway [ 4 , 5 ] (Fig. 1 ). Finally [ 6 ], described a novel non-phosphorylative pathway of xylose metabolism in Herbaspirillum huttiense that renders pyruvate and glycolate (Fig. 1 ). \n Fig. 1 Natural xylose metabolic pathway described in different microorganisms \n Xylose metabolism also requires xylose uptake and a regulatory systems [ 1 ]. Xylose transporters have been identified in different microorganisms within the ATP-binding cassette (ABC) family (e.g., XylFGH), or within the major facilitator superfamily (MFS) (e.g., XylE, AraE, or XylT) [ 1 ]. In many microorganisms, xylose metabolism is severely repressed by glucose through the carbon catabolite repression complex CRP-cAMP or other specific and pleiotropic regulators (e.g., XylR, CcpA) [ 1 , 7 ]. The isomerase metabolic pathway to assimilate xylose has been reconstructed in bacteria with industrial interest such as Corynebacterium glutamicum , Rhodococcus opacus , Pseudomonas putid a or Zymomonas mobilis . For this purpose, xylA gene encoding a xylose isomerase has been used alone or in combination with xylB gene encoding a xylulokinase [ 8 – 18 ]. In addition, the Weimberg pathway for xylose assimilation has been also engineered in C. glutamicum [ 19 ]. Among bacteria with industrial interest are the oleaginous species, Rhodococcus jostii RHA1 and R. opacus PD630 which have attracted considerable attention due to its capability to utilize different compounds as carbon sources, and to accumulate significant levels of triacylglycerols as intracellularly microbial oils that can be utilized for biofuels [ 20 ]. Since these strains cannot use xylose or arabinose as carbon sources they have been engineered to use these sugars to produce lipid from saccharified lignocellulosic materials [ 10 , 14 , 21 ]. Interestingly, in R. opacus and R. jostii it has been demonstrated that only the expression of a heterologous xylA gene is necessary to facilitate their growth on xylose, since these strains contain two xylB genes encoding enzymes with xylulokinase activity [ 10 , 14 ]. Given that R. jostii RHA1 has a xylulokinase activity, we hypothesize that perhaps this strain might have a silenced capacity to metabolize xylose that could be revealed by performing an adaptive laboratory evolution (ALE). This approach uses Darwinian rules for artificial selection of desired traits in microorganisms [ 22 ]. Although ALE has been successfully used to improve the production of lipids in Rhodococcus [ 23 ], it has not been used so far to evolve the capacity to metabolize xylose without the assistance of genetic engineering techniques. The aim of this work was to use ALE to confer the wild type R. jostii RHA1 strain the capacity to use simultaneously glucose and xylose as carbon sources to produce lipids. In addition, we have compared its performance with an engineered version of R. jostii RHA1 carrying heterologous xylA and xylB genes without further improvements. Moreover, transcriptomic analyses revealed that RHA1 contains a silent operon most probably involved in the metabolism of pentose sugars that can be unsilenced by ALE to allow the growth of this strain and the production of lipids on xylose.",
"discussion": "Discussion Although recombinant organisms can be designed à la carte , they do not always meet al.l the requirements for their industrial use. In addition, the use of GMOs (Genetically Modified Organisms) requires the utilization of containment procedures during the whole process even if they are classified as BSL-1. Moreover, when using plasmids for such metabolic engineering designs, antibiotics are usually required to maintain the plasmid stability. Consequently, in many cases the industry tends to use improved microorganisms that do not involve GMOs. One of the non-recombinant strategies that can be used to improve the efficiency of industrial microorganisms is ALE [ 22 ]. This adaptation occurs because of the redundancy of these microorganisms in genes/proteins/enzymes functions. Furthermore, genes mutate through natural mechanisms, changing their original functions or regulations, or phenotypic changes occur that only remain as far as the pressure persists [ 39 , 40 ]. In this work we show that it is possible to evolve by ALE a relevant oleaginous microorganism, such as R. jostii RHA1, to confer it the ability of metabolizing xylose. The selection of this evolved strain of R. jostii RHA1 was relatively easy, probably due to the fact that this strain has a large genome (10 Mb) that enable the presence of many flexible phenotypes. Actually, after few weeks of culture in a minimal medium containing xylose, a significant amount of biomass was observed facilitating the isolation of the evolved ALE-xyl strain. However, considering that the capacity to metabolize xylose was due to a phenotypical adaptation, ALE-xyl cells must be always cultured in minimal medium containing xylose as sole carbon and energy source to keep the xyl phenotype. Although the heterologous xylA and xylB genes are sufficient to confer the ability to metabolize xylose, the RHA1 recombinant strain should have other metabolic bottlenecks for an efficient consumption of xylose. Most likely these metabolic burdens are related to pentose transport, but we cannot rule out that other fine metabolic setting related to energy and redox balances are required to optimize xylose metabolism. Apparently, these fine-tuned changes can be more easily achieved by ALE since these settings usually depend of the acquisition of different mutations that can be accumulated and selected at the same time during the ALE process. Therefore, we assume that a recombinant strain like R. jostii RHA1 (pNVSxylABatf1) must be evolved by ALE to increase its xylose metabolic efficiency. In fact, the utilization of xylose has been improved by ALE in other engineered microorganisms [ 41 , 42 ], such as, Pseudomonas putida S12 [ 11 ] or yeast strains [ 43 , 44 ]. Probably the most interesting result of this work, is the fact that the ALE-xyl strain was able to metabolize at the same time glucose and xylose with no catabolic repression by glucose, and therefore, we have not observed a biphasic growth curve that might hinder the lipid production fermentation strategy when using saccharified lignocellulosic media. In this sense, Millán et al. (2020) used ALE to improve the assimilation of xylose and glucose in Azotobacter vinelandii , however, in this case, the co-utilization of both appears to be fundamental since xylose is poorly used in the absence of glucose [ 45 ]. Moreover, Jin et al. (2019) accelerated by ALE the simultaneous conversion rate of glucose and non-glucose sugars derived from lignocellulose biomass in Gluconobacter oxydans . Interestingly, this strain is naturally able to metabolize xylose and arabinose, but with low efficiency, demonstrating that even in a natural consumer a fine tuning of metabolism provides a further advantage [ 46 ]. In addition, the co-consumption of glucose and xylose has been also improved by ALE in yeast strains [ 47 ]. Remarkably, we have determined that the production of lipids in the ALE-xyl strain using xylose as carbon source is at least comparable to, or even superior than the production of lipids with glucose, suggesting that xylose metabolism couples perfectly to lipid biosynthesis in this strain. Trying to understand how ALE-xyl strain has evolved to consume xylose we performed an in-silico analysis of R. jostii RHA1 genome, but it did not reveal the presence of a putative complete xylose metabolic pathway in this strain. In this sense [ 14 ], have demonstrated that RHA1_ro02812 ( xylB1 ) and RHA1_ro02901 ( xylB2 ) genes encode functional xylulokinases enzymes (XylB), but to metabolize xylose R. jostii RHA1 requires the cloning and expression of an heterologous xylose isomerase (XylA). Interestingly, Xiong et al. (2012) showed that the R. jostii RHA1 recombinant strain harbouring the heterologous XylA gene grows poorly and it must be evolved by ALE to improve its xylose consumption, suggesting that other genes are required to consume xylose efficiently [ 14 ]. The transcriptomic analysis revealed that the pen operon RHA1_ro02898-RHA1_ro02909 , containing xylB2 ( RHA1_ro02901 ), was overexpressed in xylose. The pen operon also encodes an ABC pentose transporter that according to our hypothesis can facilitate the uptake of xylose, but it does not encode a putative xylose isomerase (XylA) that could allow us to propose the existence of a silent xylose isomerase pathway. Therefore, according to the gene expression levels in ALE-xyl (Fig. 6 ), we propose that xylose can be transformed into xylulose by a reductase pathway using a xylose reductase. In agreement with this hypothesis, we have detected a low but significant xylose reductase activity in the crude extract of ALE-xyl. Moreover, we have detected the presence of small amounts of xylitol in the crude extract indicating that xylose is transformed into xylitol in this evolved strain. Interestingly, a similar finding that reinforces our hypothesis was observed by Sekar et al. (2016) when they evolved S. oneidensis to consume xylose [ 35 ]. They were able to activate a silent reductase xylose pathway of S. oneidensis after 12 months of ALE evolution demonstrating that a mutation in an MFS transporter that might facilitate the xylose uptake was crucial to metabolize xylose. Moreover, they also demonstrated that an aldolase/keto reductase has a xylose reductase activity and transform xylose into xylitol. Although, they could not detect a specific xylitol/xylulose dehydrogenase activity, they propose that there are many alcohol dehydrogenases that can fulfil this role [ 35 ]. Therefore, the key factor to un-silent the xylose metabolic pathway in S. oneidensis appears to be an efficiently xylose uptake as appears to be the case in ALE-xyl strain. This is, in the ALE-xyl strain, we have observed that the RHA1_ro04589 gene that encodes a putative xylose reductase, and the RHA1_ro02809 gene that encodes a putative xylitol dehydrogenase are highly expressed in RHA1 both in glucose and xylose and thus, the critical factor that can render ALE-xyl strain the capacity to consume xylose is the overexpression of the penABC genes that encode a pentose transporter. In this sense, although [ 35 ] were able to activate by ALE a silent xylose metabolic pathway of S. oneidensis , in contrast with ALE-xyl strain which grows very efficiently in xylose, the S. oneidensis strain had to be engineered to improve its growth on xylose [ 18 ]. Therefore, we speculate that the high efficiency to metabolize xylose in the ALE-xyl strain can be probably due to the high efficiency of the ABC transport system induced in this strain when compared to the mutated MSF transporter in S. oneidensis . Other studies have highlighted the crucial role of xylose transporters in preventing glucose catabolic repression, as demonstrated in Ogataea polymorpha , where enhanced xylose assimilation was achieved by introducing a hexose transporter (HXT1*) with higher affinity for xylose, allowing for efficient co-utilization of glucose and xylose in mixed-sugar conditions [ 48 , 49 ]. Although the analysis of the genes encoded by the pen operon RHA1_ro02898-RHA1_ro02909 does not allow to propose a clear function for this operon, the presence of putative pentose ABC transporter ( penABC ) and the penHIJ genes encoding the putative xylulokinase, phosphoglycolate phosphatase and ribose-5-phosphate isomerase (RipA) enzymes suggests its implication in a phosphorylating pentose pathway. A similar pentose metabolizing function of this operon in RHA1 has been proposed by Iino et al. (2012) [ 50 ]. In spite of the fact that we have detected the presence of xylitol in the ALE-xyl strain suggesting that xylose is consumed by a reductase pathway we cannot completely discard that PenJ (RipA), i.e., the second copy of a ribose 5-phosphate isomerase in RHA1, might transform xylose into xylulose, assuming a different role for this enzyme. In this sense, it has been demonstrated that ribose-5-phosphate isomerase (RipA) from Ochrobactrum can isomerize several non-phosphorylated pentoses [ 51 ]. Therefore, the second copies of RipA (PenJ), XylB (PenH) found in the pen operon of RHA1 together with the PenABC pentose transporter might be part of a new pentose metabolizing pathway. In this sense, it is worth to mention that pen operon contains other 5 genes, penDEFGK , that can be partially correlated by BLAST analyses with the genes found in non-phosphorylating pentose metabolizing pathways reinforcing the assumption that this operon is related to pentose metabolism [ 6 , 52 ]."
} | 4,301 |
39036058 | PMC11260027 | pmc | 7,879 | {
"abstract": "Graphical abstract",
"conclusion": "6 Conclusion This paper describes step-by-step instructions on how to build an electromechanical tensile test equipment (DIY EMTT). We successfully reduced the equipment cost from several thousand dollars to several hundred dollars by using the DIY methods. In principle, the DIY EMTT integrates displacement, resistance tester, load cell, and linear stage modules to simultaneously measure the electromechanical properties of stretchable conductive material. The DIY EMTT is currently designed for soft conductive material with a maximum tensile load of 10 kg. The limit of this equipment lies in the loadcell module limit and the motor stepper module. We can advance the DIY EMTT to measure higher stress values by changing the motor stepper and the loadcell module. There are also shortcomings in this equipment, such as the data rate, which is stable at 7 data/sec. We further debug the system to maximize its data rate. Furthermore, there is also potential vibration from the equipment that can induce unwanted noise in the output data. To overcome this potential problem, we can add an anti-vibration mat under the equipment to reduce the vibration. This anti-vibration mat cost around $83 (including the shipping price) or 47 % of current equipment total price. We expect our DIY equipment to contribute to the development of stretchable sensors and electronics. The low-cost system and easy GUI interface should enable any newcomer researchers in soft robotics to advance the system based on their requirements."
} | 383 |
36618653 | PMC9816334 | pmc | 7,880 | {
"abstract": "In this work we compiled information on current and emerging microbial-based fertilization practices, especially the use of cell-free microbial culture filtrates (CFs), to promote plant growth, yield and stress tolerance, and their effects on plant-associated beneficial microbiota. In addition, we identified limitations to bring microbial CFs to the market as biostimulants. In nature, plants act as metaorganisms, hosting microorganisms that communicate with the plants by exchanging semiochemicals through the phytosphere. Such symbiotic interactions are of high importance not only for plant yield and quality, but also for functioning of the soil microbiota. One environmentally sustainable practice to increasing crop productivity and/or protecting plants from (a)biotic stresses while reducing the excessive and inappropriate application of agrochemicals is based on the use of inoculants of beneficial microorganisms. However, this technology has a number of limitations, including inconsistencies in the field, specific growth requirements and host compatibility. Beneficial microorganisms release diffusible substances that promote plant growth and enhance yield and stress tolerance. Recently, evidence has been provided that this capacity also extends to phytopathogens. Consistently, soil application of microbial cell-free culture filtrates (CFs) has been found to promote growth and enhance the yield of horticultural crops. Recent studies have shown that the response of plants to soil application of microbial CFs is associated with strong proliferation of the resident beneficial soil microbiota. Therefore, the use of microbial CFs to enhance both crop yield and stress tolerance, and to activate beneficial soil microbiota could be a safe, efficient and environmentally friendly approach to minimize shortfalls related to the technology of microbial inoculation. In this review, we compile information on microbial CFs and the main constituents (especially volatile compounds) that promote plant growth, yield and stress tolerance, and their effects on plant-associated beneficial microbiota. In addition, we identify challenges and limitations for their use as biostimulants to bring them to the market and we propose remedial actions and give suggestions for future work.",
"introduction": "Introduction Plant’ growth and development are influenced by microorganisms occurring in the phytosphere that communicate with plants by exchanging chemical signals ( Hartmann et al., 2014 ). Some of these microorganisms can benefit host plants in a variety of ways, a scenario of utmost interest when searching for new and efficient agricultural approaches based on manipulation of plant-associated microbiota. Beneficial microorganisms can directly promote plant growth through mechanisms involving production of bioactive compounds (e.g. phytohormones, volatile compounds, peptides, etc.), dinitrogen fixation, solubilization of minerals and organic material and enhancement of water and nutrient uptake and use ( Tsavkelova et al., 2006a ; Rodríguez et al., 2007 ; Francis et al., 2010 ). These microorganisms can also indirectly promote plant growth by antagonism/antibiosis against pathogens, alleviation of stress caused by environmental pollutants or other stressful abiotic conditions (e.g. drought and salinity), or by triggering in the host plant enhanced defense capacities against pathogen attack. A decline in natural resources and the environmental damage caused by practices relying on the excessive and inappropriate application of fertilizers and depletion of soil and water resources have become major limitations in conventional agriculture. A more sustainable and eco-friendly agriculture requires increases in product yield quality, while reducing the negative environmental impact of agrochemicals on soil fertility and biodiversity; potential solutions may be fostered by microbial-based approaches ( Calvo et al., 2014 ). The aim of this review was to compile information on current and emerging microbial-based fertilization practices, particularly the use of microbial inoculants, microbial-derived compounds and microbial culture filtrates (CFs), to promote plant growth, yield and stress tolerance, and their effects on plant-associated beneficial microbiota. In addition, we identify challenges and limitations to bring microbial CFs to the market as biostimulants compliant with scientific requirements of the official regulations for fertilizer products."
} | 1,120 |
30194350 | PMC6128925 | pmc | 7,883 | {
"abstract": "Estimates suggest that at least half of all extant insect genera harbor obligate bacterial mutualists. Whereas an endosymbiotic relationship imparts many benefits upon host and symbiont alike, the intracellular lifestyle has profound effects on the bacterial genome. The obligate endosymbiont genome is a product of opposing forces: genes important to host survival are maintained through physiological constraint, contrasted by the fixation of deleterious mutations and genome erosion through random genetic drift. The obligate cockroach endosymbiont, Blattabacterium – providing nutritional augmentation to its host in the form of amino acid synthesis – displays radical genome alterations when compared to its most recent free-living relative Flavobacterium . To date, eight Blattabacterium genomes have been published, affording an unparalleled opportunity to examine the direction and magnitude of selective forces acting upon this group of symbionts. Here, we find that the Blattabacterium genome is experiencing a 10-fold increase in selection rate compared to Flavobacteria . Additionally, the proportion of selection events is largely negative in direction, with only a handful of loci exhibiting signatures of positive selection. These findings suggest that the Blattabacterium genome will continue to erode, potentially resulting in an endosymbiont with an even further reduced genome, as seen in other insect groups such as Hemiptera.",
"conclusion": "Conclusions Our findings indicate that the Blattabacterium genome is experiencing elevated rates of both positive and negative selection when compared to its free-living relative Flavobacterium , approaching a 10-fold increase in selection rate at the significance level p ≤ 0.05 across 304 individual genes. In combination with previous studies elucidating the evolutionary patterns in other insect endosymbionts, we conclude that the Blattabacterium genome is shaped by similar evolutionary mechanisms. Previous studies have outlined the current state of the Blattabacterium genome, which is drastically reduced from its ancestral state and possesses a very strong bias towards A + T nucleotide base pairs. Analysis of these trends indicate that Blattabacterium are experiencing an accumulation of slightly deleterious mutations through the continued effects of random genetic drift resulting from consecutive population bottlenecks throughout Blattabacterium’s evolutionary history, with physiological constraint acting to maintain genes important to bacterial survival and host fecundity. Additionally, Blattabacterium has lost many of the genes involved in DNA repair, likely through similar mechanisms discussed here, thus exacerbating this evolutionary bias towards slightly deleterious mutations. That these mutations cannot be repaired increases functional protein evolution rates within this endosymbiont. The patterns discussed here are highly similar to those evolutionary and genomic trends observed in other intracellular insect endosymbionts 34 , 45 , 61 , 90 . Additionally, our analyses also provide insight into the direction of selection of loci within the genome. A vast majority of loci in all Blattabacterium genomes analyzed here show signs of negative selection. Only a small fraction of loci ( miaB, Holliday Junction, rplY, atpG ) show signs of positive selection. These observations are in accordance with our previous understanding of the evolutionary history of Blattabacterium , as well as its function within its cockroach host as a nutritional endosymbiont aiding in the recycling of nitrogenous waste and the production of both essential and nonessential amino acids. The analysis presented here could be augmented through a robust analysis of genome reduction within Blattabacterium . Using a parsimony approach, the ancestral genome of another primary insect endosymbiont, Buchnera-Ap , was reconstructed by Moran and Mira 32 . The results of Moran and Mira’s analysis indicated that much of the ancestral Bucnhera genome was lost during a relatively small number of large deletion events shortly after this bacteria’s transition to an intracellular lifestyle. While it is likely that that the Blattabacterium genome was reduced through similar mechanisms, a similar reconstruction within this group would offer us a more complete picture of the evolutionary origins of this unique cockroach endosymbiont.",
"introduction": "Introduction Comprised of over one million species, Class Insecta is the most speciose group among animals; at least half of extant genera are estimated to harbor obligate bacterial mutualists 1 – 3 . While some intracellular bacteria can be harmful or even lethal to their insect host, many others play an important role in host survival and fecundity 3 – 8 . These primary bacterial symbionts exist obligately within the cells of the insect, and are often required for the survival and reproduction of their host organism 1 , 7 – 9 . An intercellular lifestyle affords endosymbiotic bacteria relative safety from competition and exploitation, in exchange for increased ecological flexibility imparted onto the host species. In many cases, these obligate bacterial mutualists function in the provisioning, recycling, or degradation of essential nutrients, and are vital to those insect species that subsist on nutritionally narrow diets, such as those composed primarily of woody material, plant sap, mammalian blood, or decaying organic material 8 , 10 , 11 . However, within some insect species primary bacterial endosymbionts also function in non-nutritional roles such as parasitoid defense 12 . With the exception of a single cave-dwelling genus, Noticola (Blattodea, Nocticolidae), all cockroach species contain endosymbiotic bacteria within their fat bodies 1 , 5 , 13 , 14 . These obligate endosymbionts belong to the genus Blattabacterium (Class Flavobacteria, Phylum Bacteriodetes) 15 , 16 . Phylogenetic reconstruction suggests that cockroaches acquired these endosymbionts in a single infection event, dating between 300 million years ago - the approximate age of the first fossil roaches from the Carboniferous - and 140 million years ago, when currently extant families last shared a common ancestor 17 , 18 . Initially, the function of these endosymbionts was subject to speculation, owing to their recalcitrance to culture outside their host. However, modern DNA-sequencing techniques have allowed for the study of a number of Blattabacterium genomes. From these genomes, it was discovered that the function of Blattabacterium is primarily the synthesis of amino acids and vitamins from the nitrogenous waste products of the cockroach host 16 , 19 . Cockroaches store excess nitrogen as uric acid within their fat body cells 20 . The decaying organic matter on which cockroaches typically feed is poor in nitrogen content. Thus, a mechanism for recycling nitrogenous waste would be beneficial to any organism whose diet is nitrogen-deficient. Unlike most insects, which excrete waste nitrogen as uric acid, cockroaches excrete ammonia 21 . Blattabacterium are capable of utilizing both urea and ammonia because they contain an active urease as well as a functioning urea cycle that converts host urea to ammonia 22 – 24 . In addition, increases in dietary nitrogen intake by host cockroaches correlates with increases in uric acid buildup within that host’s fat bodies 20 , 21 , 23 . Cockroaches represent an evolutionary lineage consisting of diverse and ancient taxa that have adapted to many habitats and exhibit broad nutritional ecology; their endosymbionts, therefore, represent an excellent system in which to assess relationships between these traits. To date, eight Blattabacterium genomes have been sequenced from the following cockroach host species: Periplaneta americana 19 , Blatta germanica 25 , Cryptocercus punctulatus 26 , Blaberus giganteus 27 , Blatta orientalis 28 , Panesthia angustipennis 29 , Nauphoeta cinerea 30 and the termite, Mastotermes darwiniensis 31 . While these genomes share similar gene composition and genome architecture, each also displays unique capacities for metabolic and physiological function. Thus, while the results of phylogenetic analysis support the hypothesis of co-cladogenesis between the endosymbionts and hosts 17 , 18 , gene composition of Blattabacterium is not directly congruent with host phylogeny; rather it varies likely as a function of host nutrition, its relative importance in the mutualism, and the interaction between phenotypic constraint, environmental natural selection, and genetic drift. An intracellular lifestyle strongly influences the selective pressures and evolutionary trajectories of bacterial endosymbionts 32 . Evolution of the bacterial endosymbiont genome is characterized by elevated mutation rates and biases resulting from the combined effects of physiological constraint preserving symbiont-critical genes, random genetic drift – driven by frequent population bottlenecks, bacterial asexuality, and lack of genetic recombination – and environmental selection acting to reduce genome size 18 , 28 , 33 – 47 . Endosymbionts have been shown to have higher substitution rates and values of non-synonymous to synonymous substitution rates a result attributed to small N e . 48 , 49 . Acting through Muller’s Ratchet, asexual reproduction can prevent the recovery of wild-type genotypes through recombination 50 . Loss of recombination is a result of lost DNA repair, uptake, and recombination genes; which is a common pattern of all sequenced bacterial endosymbionts [ 51 – 56 reviewed in ref. 57 ]. However, selection in the form of physiological constraint acts to maintain genes important to the bacterial-insect symbiosis, although its role in the continued erosion of non-essential genes in the bacterial genome is largely unknown 45 , 58 , 59 . Described genomes from endosymbionts suggest that physiological constraint acts to maintain a gene set that retains its functionality for the host; though selection might also be driving the erosion of bacterial endosymbiont genomes. Certain metabolites ordinarily produced by the bacteria itself may now be obtained directly from the host; under such circumstances, these genes become superfluous and are necessary for neither bacterial survival nor continued host fecundity. As such, a smaller genome results in a cell that is faster and more efficient to reproduce. Particularly at the beginning of endosymbiosis, rapid loss of unnecessary genes 32 may be advantageous [Reviewed in ref. 60 ]. Thus, while random genetic drift does act to reduce the bacterial endosymbiont genome through Muller’s Ratchet, physiological constraint acts to preserve genes crucial to symbiosis while environmental selection favors a reduced, more energy-efficient genome. When compared to free-living bacteria, endosymbionts exhibit increased levels of mutation at synonymous and non-synonymous sites, as well as higher d N /d S ratios, indicating an increase in positive selective pressures and rapid protein evolution 33 , 61 – 63 . Thus, we may conclude that the endosymbiont genome is the result of interplay between random genetic drift and the reduction of genes through relaxed selection within large portions of the genome, and physiological constraint acting to preserve those genes vital to host survival and fecundity. Genome evolution in insect endosymbionts has been the topic of a number of studies. Full genomes from several endosymbionts have been published, including Buchnera aphidicola 64 from aphids, Wigglesworthia 65 , 66 from the tsetse fly, Blochmannia 67 from carpenter ants, and Blattabacterium 19 , from cockroaches. However, comparatively few of these genomes have been examined for signals of positive selection. Eight fully sequenced Blattabacterium genomes, in addition to the five fully-annotated free-living Flavobacterium genomes for comparison, offers a unique opportunity to investigate the patterns and processes that drive endosymbiotic genome evolution. We estimated the positive and negative selection events in the genomes of all sequenced Blattabacterium strains, and compared them to those present within the closely-related 68 but free-living Flavobacterium species ( F. indicum , F. johnsoniae , and F. psychrophilim ), to examine the similarities and differences between these two evolutionarily related, but divergent, groups. We hypothesized that patterns of selection acting upon the Blattabacterium genome will manifest as an elevation in both non-synonymous and synonymous mutation events, as well as a higher d N /d S ratio at sites under significant levels of selection than the free-living Flavobacterium - indicating increased positive selection pressures and an elevated rate of protein evolution 63 , 69 , 70 . Additionally, we sought to determine whether patterns of selection observed in previous studies across limited numbers of genes are effective at predicting patterns of selection across an entire endosymbiont genome.",
"discussion": "Results and Discussion Selection by Gene Length and COG Groups COG analysis indicates an uneven distribution of functional groups within the 304 genes selected for this analysis (Fig. 1 ). This figure illustrates the functional ‘core’ genes shared by all thirteen genomes analyzed. The majority of these genes are ribosomal in function. Perhaps unsurprisingly, the number of positive selection events (F-value: 40.872, Df: 1, p-value: 6.16e-10, adjusted R-squared: 0.12) as well as negative selection events (F-value: 189.15, Df: 1, p-value: 2.2e-16, adjusted R-squared: 0.38) both showed strong positive correlation with gene length (Fig. 2a,c , respectively). This finding is consistent with the conclusions of previous studies, where natural selection is also correlated with gene length 83 . Building upon this on a functional level, however, we also noted that signatures of both positive and negative selection (response variable) correlated strongly with the total number of nucleotides assigned to a specific COG (explanatory variable) across the Blattabacterium genome (Positive selection events: Chi-square p-value: 2.2e-16; Negative selection events: Chi-square p-value: 2.2e-16; Fig. 2b,d , respectively). Figure 1 Distribution of functional COGs (Clusters of Orthologous Groups) for the 304 ‘core’ genes analyzed here. Letters refer to COG functional categories as follows. C - Energy production and conversion; D - Cell division and chromosome partitioning; E - Amino acid transport and metabolism; F - Nucleotide transport and metabolism; G - Carbohydrate transport and metabolism; H - Coenzyme metabolism; I - Lipid metabolism; J - Translation, ribosomal structure and biogenesis; K - Transcription; L - DNA replication, recombination and repair; M - Cell envelope biogenesis, outer membrane; N – Cell motility; O - Posttranslational modification, protein turnover, chaperones; P - Inorganic ion transport and metabolism; Q - Secondary metabolites biosynthesis, transport, and catabolism; R - General function prediction only; S - COG of unknown function; T - Signal transduction mechanisms. Figure 2 Plots outlining the relationships between number of selection events and gene length or COG size. ( a ) Relationship between gene length and number of nucleotides under positive selection within Blattabacterium (F-value: 40.872, Df: 1, p-value: 6.16e-10, adjusted R-squared: 0.12). ( b ) Relationship between total COG size and number of nucleotides under positive selection within Blattabacterium (Chi-square p-value: 2.2e-16). ( c ) Relationship between gene length and number of nucleotides under negative selection within Blattabacterium (F-value: 189.15, Df: 1, p-value: 2.2e-16, adjusted R-squared: 0.38). ( d ) Relationship between total COG size and number of nucleotides under negative selection within Blattabacterium (Chi-square p-value: 2.2e-16). Blattabacterium Selection Analysis Initial analysis of Blattabacterium homolog sets was carried out across all eight of the fully sequenced strains, using a significance level of p ≤ 0.05 for homology. At this significance level, Blattabacterium displays a strong negative mutational bias, with a ratio of sites under negative selection to sites under positive selection of 11:1 across 304 genes. While most loci within Blattabacterium displayed a bias towards negative selection, a few did exhibit signatures of positive selection (Table 2a ). That the vast majority of genes within the Blattabacterium genome are experiencing neutral (Table 2b ) or negative (not shown in table) selection suggests conserved selective pressures and genome architectures within established endosymbiont lineages 28 , 69 . Accordingly, only a small number of loci were found to show no signs of selection at all (Table 2c ). Table 2 (a) Loci within Blattabacterium displaying a positive selection bias. Position Locus Avg. Length (n) COG No. of Pos. sites No. of Neg. sites \n a \n 61 \n gyrB \n 2513 L 18 11 96 \n tatC \n 1045 U 2 0 171 \n purB \n 1843 F 5 4 191 \n marC \n 742 U 2 1 239 \n folE \n 861 H 2 1 265 \n gmk \n 751 F 1 0 266 \n rpiB \n 606 G 1 0 308 \n rplX \n 324 J 2 1 312 \n rplP \n 541 J 5 4 319 \n rplC \n 821 J 3 2 359 \n entC \n 1374 Q 3 1 365 \n recQ \n 2201 LKJ 3 0 387 \n accA \n 1232 I 5 3 388 \n sdhB \n 985 C 7 0 392 \n phospho \n 1985 R 5 3 405 \n hinT \n 536 FGR 3 2 457 \n pdxA \n 1362 H 2 1 \n b \n 38 \n trmE \n 1821 R 4 4 161 \n accD \n 1098 I 1 1 167 \n purF \n 1997 F 1 1 196 \n accB \n 626 I 1 1 208 \n evoX \n 1009 L 2 2 241 \n m22 \n 843 O 1 1 304 \n rplF \n 712 J 1 1 318 \n rplD \n 821 J 2 2 430 \n rpoD \n 1127 K 7 7 438 \n integral \n 995 P 1 1 478 \n glyS \n 1910 J 3 3 483 \n pth \n 772 J 1 1 \n c \n 12 \n truA \n 1001 J 0 0 103 \n rpsT \n 317 J 0 0 117 \n aroK \n 674 E 0 0 148 \n rpmI \n 249 N/A 0 0 149 \n rplT \n 456 J 0 0 204 \n rpmG \n 239 J 0 0 240 \n nadE \n 1028 H 0 0 301 \n rplO \n 601 J 0 0 329 \n rplM \n 587 J 0 0 332 \n cdsA \n 1035 R 0 0 372 \n rplL \n 487 J 0 0 422 \n sufE \n 567 R 0 0 428 \n rpsO \n 347 J 0 0 491 \n nfsA \n 442 N/A 0 0 \n d \n 93 \n dapF \n 1043 E 112 \n gcvH \n 523 E 148 \n rpmI \n 249 N/A 323 \n rpsL \n 499 J 398 \n rpsU \n 267 N/A Positive selection is defined as those loci that display a greater number of sites under positive selection than under negative selection. (b) Loci within Blattabacterium displaying a neutral selection bias. Neutral selection is defined as those loci which display an equal number of sites under positive selection as negative selection. (c) Loci within Blattabacterium displaying no selection. These genes experience neither positive nor negative selection events. (d) Loci within Blattabacterium and Flavobacterium that display identical selection profiles. These genes display no selection events within either Blattabacterium or Flavobacterium genomes. ‘Position’ indicates that genes starting position within the Mastotermes darwineinsis genome, the model Blattabacterium genome used here. Letters refer to COG functional categories as follows. C - Energy production and conversion; D - Cell division and chromosome partitioning; E - Amino acid transport and metabolism; F - Nucleotide transport and metabolism; G - Carbohydrate transport and metabolism; H - Coenzyme metabolism; I - Lipid metabolism; J - Translation, ribosomal structure and biogenesis; K - Transcription; L - DNA replication, recombination and repair; M - Cell envelope biogenesis, outer membrane; N – Cell motility; O - Posttranslational modification, protein turnover, chaperones; P - Inorganic ion transport and metabolism; Q - Secondary metabolites biosynthesis, transport, and catabolism; R - General function prediction only; S - COG of unknown function; T - Signal transduction mechanisms. In recent years, a growing body of work seeks to place an increased emphasis on the role of selection in molecular evolution 84 – 86 . While no predominant explanatory theory for molecular evolution has yet emerged to replace the largely disproven neutral theory, a re-evaluation of the classic, primarily neutral/drift-centric hypotheses for genome evolution in Blattabacterium is necessitated. With the data presented here - and in the light of previous studies into the genome evolution of Blattabacteria - we suggest that the Blattabacterium genome is shaped by a combination of random genetic drift, environmental selection, and physiological constraint on genetic variation. The Blattabacterium lifestyle is characterized by significant and repeated population bottlenecks with each host generation as bacterial cells are transmitted vertically from mother to offspring 17 , 18 , a drastically reduced genome 25 – 30 , 39 , and elevated rates of mutation. Previous studies into obligate bacterial endosymbiont evolution suggest that the reduction in effective population size through generational bottlenecks and lack of genetic recombination resulting from Muller’s Ratchet elevates the rate of fixation of slightly deleterious mutations through random genetic drift 33 – 35 , 41 – 44 . However, populations that experience a population bottleneck recover much of the lost genetic variation through rapid population growth 45 . While it seems likely that this is the case for free-living and endosymbiotic bacteria as well, the strength of the bottleneck affects the loss of genetic variability much more so than subsequent rates of population growth 45 . Within Blattabacterium and many other bacterial endosymbionts, these bottlenecks are not trivial, and are frequently recurring throughout the insect host’s lifespan 1 – 3 , 7 , 8 , 16 ; an environment that is completely atypical for most free-living populations. Thus, examined alone, population bottlenecks strongly reduce the genetic variation of Blattabacterium . Additionally, Blattabacterium – like other intracellular bacterial endosymbionts - reproduces asexually and lacks genetic recombination [ 51 – 56 reviewed in ref. 57 ]; two mechanisms otherwise crucial for the recovery of genetic variation. This combination of factors – lack of genetic recombination and repeated population bottlenecks – does seem to suggest that Blattabacterium and other obligate symbionts are less capable of recovering lost genetic variance after population bottlenecks than free-living bacteria. However, bacterial endosymbionts also experience much higher mutation rates than their free-living relatives 63 , 69 , 70 . Indeed, mutations are synonymous with increased genetic variation, and we show here that Blattabacterium experiences highly elevated rates of mutation compared to free-living bacterial populations. It is highly unlikely that the elevated mutation rates seen in Blattabacterium are adaptive or somehow function in recovering lost genetic variation, as mutations in Blattabacterium show a strong bias towards deletions rather than insertions; a pattern that is in agreement with previous studies as well as with Muller’s Ratchet 44 , 50 . Additionally, it is suggested that reduced strength of selection on many genes in the endosymbiont genome increases the number of nucleotide sites that may be altered without consequences in fitness, strengthening the impact of deletion biases 44 . Bacterial genomes are primarily functional DNA, and the drastic genome reduction observed within Blattabacterium and has come at the cost of physiological functionality. Intriguingly however, this drastic loss in functionality does not yet appear to have strong negative impacts on Blattabacterium survival or host fitness. Indeed, many physiological tasks are now taken over by the cockroach host, rendering many Blattabacterium genes superfluous within the relatively safe and predictable symbiotic environment 25 – 30 , 36 . As in other obligate endosymbionts, many if not most of these genes come under relaxed selection, as their function is critical to neither Blattabacterium’s survival nor the symbiotic physiological requirements of its cockroach host 32 , 35 . Whether or not elevated mutation rates in physiologically-important genes functions to actively reduce genome size and thus streamline bacterial reproduction, or are the result of random genetic drift is unknown; though that many genes lost by Blattabacterium since transitioning to an intracellular lifestyle coded for otherwise critical functionality - including the loss of many genes involved in DNA maintenance and repair [ 19 , 28 , 30 , 59 , reviewed in ref. 57 ] – suggests that many losses are either only mildly non-adaptive or compensated for by the host and thus do not result in immediate impairment of symbiont or host. However, genome reduction is accompanied by a reduction and cell size and a substantial reduction in energy and nutrients requirements, providing an adaptive payoff for the active removal of non-essential genes. Indeed, a number of prokaryotic Prochlorococcus species display adaptive and rapid genome shrinkage, with genomic patterns similar to those observed in obligate symbionts including reduced G + C content, elevated rates of mutation, and the loss of DNA-repair genes 87 . However, despite these similarities, genome reduction in Prochlorococcus is characterized by largely neutral selection, as large population sizes impose low genetic drift and strong purifying selection 87 . Naturally, if genome reduction in Blattabacterium and other bacterial endosymbionts was being driven by adaptive forces and not random genetic drift, then we might expect patterns of selection similar to those in the free-living Prochlorococcus . Instead, we find here that the overwhelming majority of mutations in the Blattabacterium genome are negative in direction, strongly suggesting that genome reduction is not driven by selective processes, but rather by random genetic drift; as has been suggested for numerous other obligate bacterial endosymbionts 32 , 35 . Specific genes within the endosymbiont genome are expected to vary among endosymbiont lineages as a function of the metabolic and physiological requirements of the host species. As such, these species-specific genes vital to bacterial survival and/or host fecundity experienced elevated selective pressures for their persistence within the Blattabacterium genome. We suspect that many genes in Blattabacterium involved in functions critical to this bacterial-host symbiosis display neutral or positive signatures of selection. Thus, while random genetic drift appears to play a strong role in shaping the Blattabacterium genome, physiological constraint acts to maintain Blattabacterium ’s functionality as a primary nutritional endosymbiont across the cockroach lineage. Accordingly, the Blattabacterium genome architecture and composition is the result of the interplay between random genetic drift and the fixation of slightly deleterious mutations on one hand and physiological constraint promoting maintenance of cockroach-required metabolic functionality on the other. When compared to the signatures of selection and patterns of evolution noted within other obligate bacterial symbionts, Blattabacterium shows striking similarity. While the ratio of negative to positive selection sites of 11:1 is specific to Blattabacterium-Flavobacterium comparisons, similar patterns of strong negative selection have been observed in other insect endosymbiont genomes 69 . Unsurprisingly then, our results conform to the findings of Brynnel et al . 33 , who also measured that the tuf gene of Buchnera is evolving more than 10 times as quickly than the same gene in the free living E. coli and S. typhimurium . Additionally, Blattabacterium - like Wigglesworthia and Buchnera – does show some evidence for maintaining those functions that are highly important to its insect host 33 , 63 , 66 , 70 . Indeed, the combined effects of Muller’s Ratchet appears to be ubiquitous within obligate insect bacterial symbionts: the Buchnera chaperonin groEL displays a 5-fold increase in non-synonymous mutations, and a 10-fold increase in synonymous mutations, when compared to E. coli 63 . Mutational pressure alone likely does not account for the magnitude of these d N /d S rate elevations. Within Buchnera , it has been suggested that this elevation of fixation occurs through random genetic drift resulting from the continual reduction of effective endosymbiont population size with each transmission from host parent to host offspring 33 , 34 , 88 . Given that this same elevation of polymorphisms is observed within Blattabacterium - and that Blattabacterium also undergoes similar population bottlenecks with each host generation - it is likely that similar mechanisms are shaping these two independent lineages. This also parallels the findings of Brynnel et al . 33 , whom suggested that the rate of synonymous codon substitution within Buchnera can be as much as 40 times higher than its free-living relatives. Blattabacterium - Flavobacterium Selection Comparison Blattabacterium displays elevated levels of both positive and negative selection events at a significance level of p ≤ 0.05 when compared to free-living Flavobacterium , indicating a genome-wide increase in mutation rates across the examined genes. In order to ensure that these patterns are not the result of sequences displaying radically different divergence times, we performed a phylogenetic analysis (Fig. 3 ) to elucidate the sequence similarity within each examined group. Phylogenetic analysis of both the Blattabacterium group (Table 3 ) and Flavobacterium group (Table 4 ) indicate similar levels of phylogenetic divergence between the individuals of each 22 , 89 , 90 . Figure 3 Phylogenetic reconstruction of the evolutionary relationship between all bacteria sampled for this project. ( A ) Maximum likelihood phylogram based on whole genomes from Flavobacteria and Blattabacterium lineages, with E. coli strains as outgroup. Numbers below nodes represent percentage bootstrap support. ( B ) ASTRAL cladogram representing the species tree inferred from 200 nuclear gene trees. Numbers below nodes represent multi-locus bootstrapping support (100 replicates). Table 3 Absolute sequence divergence in the 16S rRNA gene of Blattabacterium . BPLAN BCpu BBge BGIGA MADAR BNCIN BBor BPane BPLAN 0.048 0.037 0.044 0.056 0.04 0.015 0.04 BCpu 0.048 0.038 0.043 0.048 0.043 0.043 0.045 BBge 0.037 0.038 0.024 0.045 0.026 0.038 0.021 BGIGA 0.044 0.043 0.024 0.059 0.026 0.044 0.021 MADAR 0.056 0.048 0.045 0.059 0.049 0.051 0.056 BNCIN 0.04 0.043 0.026 0.026 0.049 0.039 0.021 BBor 0.015 0.043 0.038 0.044 0.051 0.039 0.036 BPane 0.04 0.045 0.021 0.021 0.056 0.021 0.036 A phylogenetic tree was created using the 16S rRNA gene from each sequenced Blattabacterium species. From this tree, phylogenetic distances were calculated in order to estimate sequence similarity and divergence. Host species abbreviations are as follows: BNCIN , N. cinerea ; BGIGA, B. giganteus; BBge, B. germanica ; BPLAN, P. americana; BCpu, C. punctulatus; MADAR, M. darwiniensis , BBor, B. orientalis ; BPane, P. angustipennis spadica . Table 4 Absolute sequence divergence in the 16S rRNA gene of Flavobacterium . Fpsych Fbranch Fjohn Findic Fcolum Fpsych 0.041 0.053 0.081 0.064 Fbranch 0.041 0.056 0.084 0.068 Fjohn 0.053 0.056 0.086 0.069 Findic 0.081 0.084 0.086 0.063 Fcolum 0.064 0.068 0.069 0.063 A phylogenetic tree was created using the 16S rRNA gene from each Flavobacterium species used in this study. From this tree, phylogenetic distances was estimated. Species abbreviations: Findic, Flavobacterium indicum ; Fjohn, Flavobacterium johnsoniae ; Fpsych, Flavobacterium psychrophilim ; Fbranch, Flavobacterium branchiophilum ; Fcolum, Flavobacterium columnare . In addition, each group displays comparable percentages of identical sites ( Blattabacterium : 89.4%, Flavobacterium : 87.8%) as well as similar pairwise percent identities ( Blattabacterium : 95.7%, Flavobacterium : 93.3%) when aligning the ribosomal 16S rRNA gene. Thus, extant Blattabacterium display signs of elevated rates of genome evolution in the form of increased levels of selection events. The increase in the number of sites experiencing negative or positive selection when compared to the free-living Flavobacterium suggests elevated levels of functional protein evolution in the endosymbionts. Only a limited number of loci display similar selection profiles between Blattabacterium and Flavobacterium (Table 2d ). Results of MEME selection analysis indicate that all genes analyzed show at least some evidence of negative selection. Sites under negative selection comprise approximately 86 percent of examined loci. However, four loci, at one site each, show evidence for positive selection (Table 5 ). Three of the four loci showing evidence for positive selection are involved in DNA or RNA modification: 2-methylthioadenine synthetase, Holliday Junction resolvase, and 50S ribosomal protein L25 subunit. Within E. coli and Salmonella typhimurium , variations of the protein 2-methylthioadenosine have been shown to stabilize codon-anticodon interactions through the restriction of first codon position wobble during tRNA aminoacylation 91 , 92 . This functionality prevents the misreading of the genetic code, thus reducing the likelihood of mutation. Additionally, Holliday Junction resolvase-like proteins have been shown to play key roles in DNA recombination and repair 93 – 95 . Finally, genes responsible for the production of ribosomes within a cell are crucial for the proper translation of proteins from mRNA 96 , 97 . Modifications to genes responsible for the production of ribosomal proteins will likely impact the efficiency and/or accuracy of protein translation and assembly. Given the broad reduction in functionality of the Blattabacterium genome, and the loss of many ancestral DNA and RNA maintenance and repair genes (Fig. 1 ) 8 , 98 , 99 , it is in some ways not surprising that all currently-described Blattabacterium strains display similar selection pressures on those remaining genes responsible for the maintenance of genetic material. However, of notable absence from our list of genes showing signatures of positive selection is the molecular chaperone and maintenance gene GroEL . These sequences are part of the larger GroL locus in modern Blattabacterium genomes, regions of which were found previously in Blattabacterium to be under positive selection 99 . This inconsistency likely arises from the outgroups used in each study. We utilized Blattabacteria’s closest free-living relative, Flavobacterium 16 , 17 , as an outgroup while Fares et al . utilized relatively distantly-related free-living Gammaproteobacteria 100 . Based on this methodological distinction, we can conclude that the selective pressure noted by Fares et al . was exerted prior to the split between Blattabacterium and Flavobacterium . Table 5 Loci containing sites that display evidence for positive selection, according to MEME episodic selection analysis. First column denotes the locus of interest. Second column contains the names of the proteins coded by these loci; and the third column contains proposed functional information about these proteins, gathered from the UniProt gene database. Loci Protein Name Putative Function miaB 2-methylthioadenine synthetase B family tRNA modification enzyme RNA modification Holliday Junction Holliday junction resolvase-like protein hydrolase, nucleic acid binding, DNA recombination, transcription antitermination rplY 50S ribosomal protein L25 rRNA binding, negative regulation of translation, translation atpG ATP synthase F1 subunit gamma ATP binding, plasma membrane ATP synthesis coupled proton transport In contrast to the previous genes, however, which are involved in the maintenance of genetic material, the remaining locus found to show signatures of positive selection, atpG , codes for ATP synthase F1 subunit gamma. ATP synthase-family subunit proteins typically combine to form an ATP synthase complex, which is responsible for energy production in the form of ATP within the cell 101 , 102 . One of the primary functions of Blattabacterium within its host is amino acid synthesis. Amino acid production is a very endergonic process, requiring large amounts of energy in the form of ATP in order to effectively carry out biosynthesis 103 . Therefore, beneficial modifications to genes coding for an ATP synthase subunit that result in the more efficient functioning of ATP synthase as a complete complex are more likely to be favored within the Blattabacterium genome. In keeping with the previous findings that all Blattabacterium strains examined to date are alike in their function to provide essential and nonessential amino acids to their cockroach hosts, here we demonstrate that Blattabacterium also share signatures of positive selection within genes responsible for the production of the ATP synthase F1 subunit."
} | 9,269 |
29188181 | PMC5699531 | pmc | 7,884 | {
"abstract": "Carbon catabolite repression refers to the preference of microbes to metabolize certain growth substrates over others in response to a variety of regulatory mechanisms. Such preferences are important for the fitness of organisms in their natural environments, but may hinder their performance as domesticated microbial cell factories. In a Pseudomonas putida KT2440 strain engineered to convert lignin-derived aromatic monomers such as p -coumarate and ferulate to muconate, a precursor to bio-based nylon and other chemicals, metabolic intermediates including 4-hydroxybenzoate and vanillate accumulate and subsequently reduce productivity. We hypothesized that these metabolic bottlenecks may be, at least in part, the effect of carbon catabolite repression caused by glucose or acetate, more preferred substrates that must be provided to the strain for supplementary energy and cell growth. Using mass spectrometry-based proteomics, we have identified the 4-hydroxybenzoate hydroxylase, PobA, and the vanillate demethylase, VanAB, as targets of the Catabolite Repression Control (Crc) protein, a global regulator of carbon catabolite repression. By deleting the gene encoding Crc from this strain, the accumulation of 4-hydroxybenzoate and vanillate are reduced and, as a result, muconate production is enhanced. In cultures grown on glucose, the yield of muconate produced from p -coumarate after 36 h was increased nearly 70% with deletion of the gene encoding Crc (94.6 ± 0.6% vs. 56.0 ± 3.0% (mol/mol)) while the yield from ferulate after 72 h was more than doubled (28.3 ± 3.3% vs. 12.0 ± 2.3% (mol/mol)). The effect of eliminating Crc was similar in cultures grown on acetate, with the yield from p -coumarate just slightly higher in the Crc deletion strain after 24 h (47.7 ± 0.6% vs. 40.7 ± 3.6% (mol/mol)) and the yield from ferulate increased more than 60% after 72 h (16.9 ± 1.4% vs. 10.3 ± 0.1% (mol/mol)). These results are an example of the benefit that reducing carbon catabolite repression can have on conversion of complex feedstocks by microbial cell factories, a concept we posit could be broadly considered as a strategy in metabolic engineering for conversion of renewable feedstocks to value-added chemicals.",
"introduction": "1 Introduction To successfully compete in the environmental niches they occupy, most microorganisms have developed innate preferences for certain growth substrates over others. This phenomenon has been singularly termed c arbon c atabolite r epression (CCR), but the mechanisms that govern these preferences are as diverse as the organisms in which they have evolved (Reviewed in Görke and Stülke (2008) ). In pseudomonads, a preference for organic acids and amino acids over glucose, which is generally preferred over hydrocarbons and aromatic compounds, is imparted by a complex combination of global and operon-specific mechanisms (Reviewed in Rojo (2010) ). Perhaps the most important of these mechanisms is the action of the Crc ( c atabolite r epression c ontrol) protein, a global regulator that inhibits translation of targeted mRNAs by binding near ribosome binding sites. This binding occurs in association with another protein, Hfq, at catabolite activity (CA) sequence motifs that contain a AANAANAA core and the presence of Crc, Hfq, and the CA motif is essential for Crc regulation ( Moreno et al., 2014 , Moreno et al., 2009b , Sonnleitner et al., 2009 , Sonnleitner et al., 2012 ). Crc has been shown to target catabolic pathways directly, by inhibiting translation of mRNAs encoding the enzymes themselves, and indirectly, by inhibiting translation of mRNAs encoding transcriptional regulators that drive expression of genes encoding catabolic enzymes as well as transporters required for substrates to enter the cell ( Hernández-Arranz et al., 2013 ). CCR is undoubtedly important for the fitness of saprophytic soil bacteria like Pseudomonas putida KT2440, which is well-suited for its native environment because of its ability to judiciously degrade a wide range of natural and xenobiotic substrates, including those derived from the three major fractions of plant biomass, namely cellulose, hemicellulose, and lignin. Lignin is a heterogeneous polymer of aromatic monomers that is an important component of the plant cell wall and can account for up to 40% of the carbon in terrestrial biomass ( Ragauskas et al., 2014 , Zakzeski et al., 2010 ). We have recently reported the development of P. putida KT2440-based biocatalysts for conversion of lignin monomers such as p -coumarate and ferulate to muconate ( Fig. 1 ) (Johnson et al., 2016; Vardon et al., 2016, 2015), which can be converted to adipic acid ( Vardon et al., 2015 , Vardon et al., 2016 ) and diethyl terephthalate ( Lu et al., 2015 ), precursors to the commodity plastics nylon and polyethylene terephthalate, respectively. Muconate can also be utilized directly or partially hydrogenated to produce novel materials ( Rorrer et al., 2016 , Rorrer et al., 2017 ). p -Coumarate and ferulate are common, ester-linked hydroxycinnamic acids in lignin ( Vanholme et al., 2010 ) and represent the predominant monomeric, lignin-derived species generated by alkaline treatment of herbaceous feedstocks, such as corn stover and switchgrass, accounting for up to 40%, on a mass basis, of the solubilized lignin ( Karp et al., 2014 , Karp et al., 2016 , Karp et al., 2015 , Munson et al., 2016 ). The resulting liquor also contains organic acids including sugar degradation products and acetate from acetyl groups in hemicellulose. In the conversion of lignin to a product such as muconate, we envision that lignin monomers would be converted to the targeted product while other substrates, such as residual organic acids or carbohydrate-derived products, could be utilized to generate energy and carbon required for growth ( Beckham et al., 2016 ). However, such a strategy could be problematic in light of the preference of P. putida KT2440 for organic acids and sugars over aromatic molecules such as those derived from lignin. Fig. 1 Crc regulation of metabolic pathways for production of muconate in the engineered strain, P. putida KT2440-CJ102 ( Vardon et al., 2016 ). The lignin monomers p -coumarate and ferulate are metabolized via protocatechuate, which is redirected to catechol by deletion of pcaHG and integration of aroY, which encodes a protocatechuate decarboxylase from Enterobacter cloacae ( Vardon et al., 2016 , Vardon et al., 2015 ). Catechol then undergoes intradiol ring cleavage by action of the CatA and CatA2 dioxygenases to form cis,cis -muconate, which accumulates due to deletion of catB and catC . Putative targets of Crc regulation are shown ( Browne et al., 2010 , Hernández-Arranz et al., 2013 , Morales et al., 2004 ). Fig. 1 During production of muconate from p -coumarate or ferulate, cultures of our engineered P. putida KT2440 strain, KT2440-CJ102, growing on glucose or acetate exhibit a marked accumulation, respectively, of 4-hydroxybenzoate (4-HBA) and vanillate. These intermediates are converted to protocatechuate (PCA), which is further metabolized, in this engineered strain, to produce muconate via catechol ( Fig. 1 ) ( Vardon et al., 2015 , Vardon et al., 2016 ). Interestingly, both the 4-HBA hydroxylase, PobA, and the vanillate demethylase, VanAB, that convert these molecules to PCA have been identified as putative targets of Crc regulation ( Browne et al., 2010 , Hernández-Arranz et al., 2013 , Morales et al., 2004 ). Here we have confirmed these predictions using mass spectrometry (MS)-based proteomics and demonstrated that deletion of the gene encoding Crc enhances metabolism of both 4-HBA and vanillate, leading to enhanced muconate production from p -coumarate or ferulate when either glucose or acetate are supplied as a source of carbon and energy.",
"discussion": "4 Discussion and conclusions The impact that microbial production processes have had on humanity is a testament to the power and diversity of microbial metabolism and physiology. While many traits that have enabled the biological success of microbes in their native environments are undoubtedly central to such processes, the elimination of superfluous cellular processes could be beneficial to modern microbial cell factories, but must be weighed carefully. To this end, one approach has been to reduce the size of the genome. The recent report of the design and synthesis of a dramatically minimized Mycoplasma mycoides genome found that there was ultimately a tradeoff between genome size and growth rate ( Hutchison et al., 2016 ) and similar effects have been observed upon genome reduction in E. coli ( Kurokawa et al., 2016 ). Genome reduction may also eliminate other traits that are beneficial both in nature and in biotechnological applications. Successfully enhancing microbial cell factories by eliminating unnecessary features of a genome without introducing deleterious effects has been demonstrated, though. Most notably with regard to P. putida KT2440, it was shown that deletion of 4.3% of the genome encoding 300 genes, including those required for the flagellar machinery, reduced growth lag and sensitivity to oxidative stress while increasing growth rate, biomass yield ( Martínez-García et al., 2014 ), and heterologous gene expression ( Lieder et al., 2015 ). To maximize yield and productivity, the ideal biocatalyst would rapidly and simultaneously metabolize any and all carbon sources with which it is provided. CCR is generally inconsistent with this ideal and is of special concern if the aim is to utilize complex, heterogeneous feedstocks such as those often derived from biomass. With this in mind, here we have demonstrated that reducing CCR in P. putida KT2440 is beneficial for the conversion of less preferred substrates in the presence of substrates that are more preferred. The metabolism of both p -coumarate and ferulate were enhanced, leading to a subsequent increase in muconate production in the presence of glucose and acetate, co-feeds that are necessary for growth but, based on the data presented here, clearly induce Crc-mediated regulation that is detrimental to the conversion of these lignin monomers. Based on the presence of the AANAANAA Crc-binding motif near their translation initiating ATGs, both PobA and VanA, as well as several transcriptional regulators (VanR) and transporters (VanP, PcaK, PcaP) that could affect expression and activity of PobA and VanAB, have been proposed as putative targets of Crc repression ( Browne et al., 2010 , Hernández-Arranz et al., 2013 ). pobA has also been demonstrated to be transcriptionally regulated by Crc ( Morales et al., 2004 ). In the present study, we have confirmed the regulation of PobA and VanAB by Crc using MS-based proteomics ( Fig. 2 ). We were unable to identify the transcriptional regulators or transporters involved in the metabolism of 4-HBA or vanillate with the exception of the transcriptional regulator VanR, the expression of which did not obviously reflect regulation by Crc ( Table S1 , Fig. S1 ). However, we believe PobA and VanAB are likely subject to multi-tier regulation by Crc as proposed by Hernández-Arranz et al. (2013 ), the details of which could not be elucidated in the present study based on the conditions examined and methods employed. Similarly, we are unable to confirm or refute other proposed targets of Crc in the relevant pathways ( Fig. 1 ) based on our data ( Table S1 , Fig. S1 ). In that light, it is striking that of proteins involved in these pathways that we were able to quantify in our proteomics data ( Table S1 , Fig. S1 ), PobA, VanA, and VanB exhibit some of the clearest and most substantial regulation by Crc, which could suggest that these proteins represent the primary targets by which Crc regulates the pathways for catabolism of p -coumarate and ferulate. Interestingly, PobA and VanAB are the only enzymes in these pathways that require reducing equivalents for their activity. Because the purpose of carbon catabolite repression is to metabolize substrates judiciously based on the carbon and energy they yield, it might be beneficial to regulate these enzymes more tightly. It should also be noted that while we believe the increase in muconate demonstrated here is attributed to de-repression of 4-HBA and vanillate metabolism, it is likely that other enzymes in these pathways may have also been de-repressed. The effect of de-repressing other enzymes, however, could be masked in the presence of larger bottlenecks such as those caused by PobA and VanAB, but may become relevant if further engineering is performed to remove these bottlenecks, an endeavor we are currently pursuing. CatB, CatC, and PcaHG have been demonstrated to be targets of Crc by RT-PCR ( Morales et al., 2004 ), but the genes that encode these enzymes are deleted in the strains examined here to enable muconate production, so any impact of this regulation can not be evaluated using these strains. Thus, the full effect of global de-repression may be difficult to fully appreciate in any single experiment or condition, including those examined here. While the enhancement in muconate production with the deletion of crc is relatively modest in the shake flask experiments demonstrated here, such effects could be substantial in bioreactor cultivations. We believe it is worthwhile to consider whether the elimination of CCR might be beneficial in other biocatalytic systems. Indeed, many attempts to eliminate or rewire CCR in other systems have been described (Reviewed in Vinuselvi et al. (2012) ). While some of these have been successful, impaired growth resulting from metabolic de-optimization and/or the crosstalk between CCR and other cellular physiology is a pervasive complication. One of the unique aspects of Crc in pseudomonads is its apparent lack of involvement in other cellular processes critical for growth; as we have observed here and others have shown previously, growth on substrates such as glucose, succinate, acetate or benzoate is only marginally affected in mutants lacking Crc ( Hernández-Arranz et al., 2013 , La Rosa et al., 2015 , Moreno et al., 2014 , Sonnleitner et al., 2012 ). In addition, deletion of crc was shown to double the amount of ATP and NADPH in strains grown on glucose and succinate by redirecting flux through central metabolism ( La Rosa et al., 2015 ). These findings and those described here make the deletion of crc an attractive target for those interested in engineering pseudomonads for efficient biocatalysis."
} | 3,653 |
21059789 | PMC3021791 | pmc | 7,886 | {
"abstract": "Lateral gene transfer (LGT) is an important mechanism of natural variation among prokaryotes. Over the full course of evolution, most or all of the genes resident in a given prokaryotic genome have been affected by LGT, yet the frequency of LGT can vary greatly across genes and across prokaryotic groups. The proteobacteria are among the most diverse of prokaryotic taxa. The prevalence of LGT in their genome evolution calls for the application of network-based methods instead of tree-based methods to investigate the relationships among these species. Here, we report networks that capture both vertical and horizontal components of evolutionary history among 1,207,272 proteins distributed across 329 sequenced proteobacterial genomes. The network of shared proteins reveals modularity structure that does not correspond to current classification schemes. On the basis of shared protein-coding genes, the five classes of proteobacteria fall into two main modules, one including the alpha-, delta-, and epsilonproteobacteria and the other including beta- and gammaproteobacteria. The first module is stable over different protein identity thresholds. The second shows more plasticity with regard to the sequence conservation of proteins sampled, with the gammaproteobacteria showing the most chameleon-like evolutionary characteristics within the present sample. Using a minimal lateral network approach, we compared LGT rates at different phylogenetic depths. In general, gene evolution by LGT within proteobacteria is very common. At least one LGT event was inferred to have occurred in at least 75% of the protein families. The average LGT rate at the species and class depth is about one LGT event per protein family, the rate doubling at the phylum level to an average of two LGT events per protein family. Hence, our results indicate that the rate of gene acquisition per protein family is similar at the level of species (by recombination) and at the level of classes (by LGT). The frequency of LGT per genome strongly depends on the species lifestyle, with endosymbionts showing far lower LGT frequencies than free-living species. Moreover, the nature of the transferred genes suggests that gene transfer in proteobacteria is frequently mediated by conjugation.",
"conclusion": "Conclusions Network analyses of proteobacterial genomes reveal frequent LGT among members of the phylum. The main trends in proteobacterial gene sharing are observed among species from different taxa inhabiting the same habitat. Together with the high content of plasmid proteins in laterally shared protein families, this suggests that most of the LGT in proteobacteria occurs over short physical distances, where donor and recipient are proximate. Our analysis shows that higher LGT rates are inferred within the phylum level than the species level; yet, LGT is more probable among similar species from the same class, so that modules of shared protein families are similar to traditional proteobacterial classification schemes but lacking the traditional hierarchy.",
"introduction": "Introduction Lateral gene transfer (LGT or horizontal gene transfer) is the process by which prokaryotes acquire DNA and incorporate it into their genome. Mechanisms for LGT entail transformation, transduction, conjugation, and gene transfer agents ( Thomas and Nielsen 2005 ; Lang and Beatty 2007 ). LGT has a major role in shaping the distribution of genes across genomes during prokaryote evolution ( Doolittle and Bapteste 2007 ) with only few genes that are resistant to it in the laboratory ( McInerney and Pisani 2007 ; Sorek et al. 2007 ) and probably none that are resistant to it over the full course of evolutionary time ( Bapteste et al. 2009 ). The fate of the DNA acquired by the different transfer mechanisms can vary in the laboratory. For example, DNA transferred by conjugation in Escherichia coli is recombined into the genome and can survive there for a few generations or longer ( Babic et al. 2008 ), whereas DNA transferred by phage during transduction may be stably integrated into the genome or degraded by bacterial antiviral defense mechanisms, CRISPRs ( Marraffini and Sontheimer 2008 ; Horvath and Barrangou 2010 ). Phylogenetic inference of LGT frequency during prokaryote evolution—that is, estimating LGT by looking for discordant branching patterns in trees—provides a wide range of estimates that anywhere from about 20% of all genes are affected by LGT ( Snel et al. 2002 ; Beiko et al. 2005 ), to perhaps 40% ( Kunin et al. 2005 ) or up to 90% or more of all genes have been affected at some point in their past ( Mirkin et al. 2003 ). This large range of estimates stems to no small extent from inherent difficulties of sequence alignment and phylogenetic inference using highly divergent and/or poorly conserved sequences ( Roettger et al. 2009 ), which comprise the vast majority of data from sequenced genomes. Estimates of the proportion of recently acquired genes per genome using nucleotide patterns or codon bias deliver similar results, showing that on average about 14% of the genes in each genome are recently acquired by lateral transfer ( Ochman et al. 2000 ; Nakamura et al. 2004 ). Once adapted within the genome, acquired functional genes can then be inherited vertically from generation to generation ( Babic et al. 2008 ) or donated once again at a later time. The modest quantity of 14% recent acquisitions at a given point in time thus accumulates over geological timescales, such that minimum estimates based on network approaches indicate that on average 81 ± 15% of the genes in each prokaryotic genome have been affected by LGT at some stage during evolution ( Dagan et al. 2008 ). Prokaryotic genome content and size reflect prokaryotic lifestyle ( Moran and Wernegreen 2000 ; Podar et al. 2008 ), and the frequency of acquired genes is positively correlated with genome size ( Nakamura et al. 2004 ; Cordero and Hogeweg 2009 ). Yet differences between different bacterial taxonomic groups hint that this is not the only factor affecting the amount of acquired genes within a genome. Recent LGT within the genome of E. coli , having about 4,500 open reading frames (ORFs), was estimated by aberrant codon usage to affect 18% of the gene families ( Lawrence and Ochman 1998 ). In cyanobacteria, having an average of 2,500 ORFs, about 50% of the protein families were inferred to evolve by LGT ( Zhaxybayeva et al. 2006 ; Shi and Falkowski 2008 ), the high frequency of LGT in cyanobacteria possibly relates to their specific ecological niche and the need to adapt quickly to a dynamic environment ( Dufresne et al. 2008 ; Shi and Falkowski 2008 ). Proteobacteria comprise the largest phylum-level group of prokaryotes, including 56,948 currently identified species (44% of all eubacterial species according to NCBI Taxonomy in August 2009). The phylum was named after the Greek god Proteus, who can assume many different shapes, to reflect the enormous biochemical and phenotypic diversity within this group ( Stackebrandt et al. 1988 ). The majority of known proteobacteria are mesophilic, with some exception of thermophilic species (e.g., Thimonas thermosulfata ) and psychrophilic (e.g., Polaromonas hydrogenivorans ). Most of the known proteobacteria are free living, and some can dominate in certain marine environments, such as members of the Roseobacter clade ( Brinkhoff et al. 2008 ). Some are found in symbiotic association, either mutualistic like the Bradyrhizobium japonicum (a symbiont of rice) or aggressive parasites, such as the Rickettsiae. Others are predatory proteobacteria that feed upon other prokaryotes ( Davidov and Jurkevitch 2009 ). Energy metabolism in proteobacteria is extremely diverse, including chemoorganotrophs (e.g., E. coli ), chemolithotrophs (e.g., the sulfur-oxidizing bacteria Thiobacillus aquaesulis and the ammonia-oxidizing bacteria Nitrosomonas europaea ), or phototrophs (e.g., Rhodospirillum rubrum ) ( Kersters et al. 2006 ). Based on phylogenetic reconstruction of 23S ribosomal RNA (rRNA) and other genetic markers, the phylum was originally divided into four classes: alpha-, beta-, gamma-, and deltaproteobacteria ( Stackebrandt et al. 1988 ), the epsilonproteobacteria (Campylobacterales in some schemes) being a later addition ( Gupta 2006 ). Reflecting their diversity, proteobacteria currently comprise almost half (826 of 1,749 as of January 2010) of all completely sequenced genomes ( Markowitz et al. 2010 ). Acquisition of new and adaptatively suitable genes from distantly related species by LGT is an evolutionarily quick alternative to modifying preexisting genes via point mutations. For example, the genome of the eubacterium Salinibacter ruber that resides in the extremely halophilic habitat of saltern crystallizer ponds, harbors many genes shared with haloarchaeal species, probably as a result of niche-specific acquisitions ( Mongodin et al. 2005 ). Shared gene content following LGT is found also between species having similar symbiotic relation with similar host, as in the case of the genus Bradyrizobium (alphaproteobacteria) and Ralstonia solanacearum (betaproteobacteria), both of which are soil bacteria undergoing symbiosis, either mutualistic or parasitic, with plants ( Kunin et al. 2005 ). Networks of shared genes are a useful tool to recover common gene content across many bacterial genomes ( Beiko et al. 2005 ; Kunin et al. 2005 ; Fukami-Kobayashi et al. 2007 ; Dagan et al. 2008 ; Halary et al. 2010 ). Among the proteobacteria, phylogeny for specific groups has been examined using tree-based methods, for example, in the gamma- ( Lerat et al. 2005 ), the epsilon- ( Gupta 2006 ), and the alphaproteobacteria ( Wu et al. 2004 ; Ettema and Andersson 2009 ). However, phylogenies only depict the evolutionary history of one or few genes within a genome, not for the whole genome. Network approaches to study genome evolution within the proteobacteria, where genome sizes can range from under 160 kb ( Nakabachi et al. 2006 ) to over 9 Mb ( Kaneko et al. 2002 ), have not been reported to date. Here, we investigate genome evolution within proteobacteria using a network approach.",
"discussion": "Results and Discussion Clustering of the 1,207,272 proteins within the 329 proteobacterial genomes using amino acid identity threshold of 30% ( T 30 ) resulted in 74,667 protein families of size ≥2 proteins. Only 14 of these families are universally present in all proteobacteria. These include mostly ribosomal proteins together with proteins involved in information processes, such as chaperonin GroEL ( supplementary table S1 , Supplementary Material online). A recent investigation into the quality of genome annotation in NCBI data set revealed frequent misannotation of core genes in gammaproteobacteria ( Poptsova and Gogarten 2010 ), hence the number of universal genes reported here using the standard annotation might be underestimated. Using the T 30 threshold results also in 140,333 (12% of the total) unclustered proteins. Singleton proteins—frequently named also ORFans ( Fischer and Eisenberg 1999 )—are genes for which no reciprocal BBH above T 30 was found within the current genomes sample. These may be either novel genes that are specific to the genome or genes that are shared with genomes not present in our sample. To test the latter possibility, we first searched for homologs to these singletons within 97 proteobacterial genomes that were added to the NCBI database between April 2008 (our version) and January 2009, increasing by 30% the proteobacterial genome sample size. Of the 140,333 singletons, 10,880 (8%) proteins had reciprocal BBH within the larger sample at T 30 . This averages to a removal of 112 singletons with each additional proteobacterial genome that is sampled. The remaining 129,453 singletons were then searched for homologs within 335 nonproteobacterial prokaryote genomes in NCBI genomic database (April 2008 version). For 18,692 proteins, we found nonproteobacterial reciprocal BBH at T 30 . Hence, on average, each nonproteobacterial genome includes 55 homologs to proteobacterial singletons at that protein sequence identity threshold. The remainder of 110,491 (9%) singletons remains as such. The search for homologs to the singletons in our sample supplies two observations. First, increasing the sample of searched genomes by 230% (761 genomes in total) reduced the percent of singletons by only a very modest proportion (from 12% to 9% of the proteobacterial gene repertoire). Second, the ratio of singletons found in newly sequenced proteobacterial genomes and nonproteobacterial genomes is roughly 2:1. The Distribution of Shared Proteins among Proteobacteria Shared gene content among prokaryotes may be the result of either common ancestry or LGT. Notwithstanding various factors affecting protein evolutionary rates ( Graur and Li 2000 ), protein sequence identity among orthologs within protein families that evolve by vertical inheritance alone is expected to be roughly proportional to the divergence time of the compared species ( Novichkov et al. 2004 ; Dagan et al. 2010 ). Protein-coding genes acquired by LGT are expected to have higher sequence identity among donor and acceptor groups than the expected for an average gene reflecting the reference sequence tree, assuming that the transfer event occurred after the divergence of the reference operon sequences. If all proteins were evolving by vertical inheritance alone (i.e., if they were all strictly coevolving, physically linked to the same rRNA operon in their current chromosome), then using ascending amino acid identity thresholds for the reconstruction of protein families would result in a strictly hierarchical genome (taxon) clustering of increasingly narrow taxon sample. Thus, low identity thresholds are expected to yield kingdom- or phylum-specific families, for example, whereas increasing identity thresholds will yield protein families that are specific to lower taxonomic ranks, such as class, order, genus, and finally species-specific protein families, etc. Exceptions to this rule (i.e., anomalously high sequence similarity) can indicate the workings LGT in the data. To study gene distribution patterns over ascending protein similarities in proteobacteria, we repeated the clustering into protein families using ascending thresholds for the sequence similarity between reciprocal BBHs. Increased protein sequence identity thresholds resulted in larger numbers of protein families, each spanning fewer genomes. The number of protein families at T 30 is 74,667 with 41,255 (55%) small protein families spanning ≤4 genomes. No universal families are recovered using T 70 , which results in 139,564 protein families and a larger number of smaller families 96,717 (69%) spanning ≤4 genomes. The frequency of universal protein families decreases with protein sequence identity threshold, leaving a single family at T 55 (ATP-dependent Clp protease) and no universal families found above that threshold ( table 1 and supplementary table S1 , Supplementary Material online). Table 1. Number of Protein Families in the Various Thresholds and Characteristics of the Result Shared Protein Network. Protein Similarity Threshhold No. of Families Singletons No. of Proteins No. of Families ≤4 Species No. of Universal Families No. of Edges Mean Edge Weight Median Edge Weight No. of Modules T 30 74,667 140,333 (12%) 1,066,939 41,255 (55%) 14 53,956 854 ± 527 762 4 T 35 83,740 165,256 (14%) 1,042,016 47,670 (57%) 10 53,956 743 ± 521 631 3 T 40 93,806 194,360 (16%) 1,012,912 54,835 (58%) 6 53,956 624 ± 515 497 3 T 45 104,420 228,996 (19%) 978,276 62,957 (60%) 4 53,956 505 ± 503 362 3 T 50 114,155 266,022 (22%) 941,250 70,817 (62%) 2 53,956 400 ± 489 251 3 T 55 123,386 307,825 (25%) 899,447 79,076 (64%) 1 53,956 304 ± 473 158 5 T 60 130,651 351,589 (29%) 855,683 86,077 (66%) 0 53,956 225 ± 453 92 6 T 65 136,199 398,264 (33%) 809,008 92,094 (68%) 0 53,956 164 ± 431 42 9 T 70 139,564 446,640 (37%) 760,632 96,717 (69%) 0 53,869 118 ± 407 17 11 To summarize shared gene distribution patterns among proteobacteria in various protein sequence identity thresholds ( T i ), we reconstructed an NSP for 30–70% protein sequence identity thresholds. The network includes 329 vertices (genomes) and a maximum of 53,956 edges (number of shared protein families). Edge weights in this network are calculated as the number of shared protein families between two connected genomes. The use of increasing protein sequence identity thresholds results in gradual decrease in common families among distantly related species and leads to a different network for each threshold. Using T 30 – T 65 , the NSP among proteobacteria is a clique where all genomes are connected with each other. Increasing protein sequence identity thresholds of T 70 eliminates 87 edges from the NSP ( table 1 ). A comparison of NSP at T 65 and T 70 shows that edges connected at one end at least to species having small genome size (below 1,500 genes) are the first to be disconnected from the network. Such species include the Rickettsiales (alphaproteobacteria), Zymomonas mobilis ZM4 (alphaproteobacteria), and Helicobacter pylori HPAG1 (epsilonproteobacteria; supplementary table S2 , Supplementary Material online). Although the connectivity distribution in the NSP is almost identical over different protein sequence identity thresholds, edge weights among the connected species changes considerably ( fig. 1 ). The NSP at T 30 reveals a clear taxonomic structure within gene distribution patterns across proteobacterial species. Closely related species within the same taxonomic class are connected by edges of higher weights (they share more protein families) in comparison with species from different classes. Clusters of highly connected species may be observed among different strains of the same genus, such as the Burkholderiales (betaproteobacteria), Enterobacteriales (gammaproteobacteria), and Pseudomonadales (gammaproteobacteria). Genera of small genome size are connected to other species with edges having lower weights. Such are the Rickettsiales (alphaproteobacteria) and Campylobacterales (epsilonproteobacteria). The background frequency of shared protein families at T 30 has a median of 427 shared protein families between any pair of species. F IG . 1. The NSP families. A matrix representation of the NSPs in T 30 (below the diagonal) and T 70 (above the diagonal). The species are sorted by an alphabetical order of the order and genus. The color scale of cell a ij in the matrix indicates the number of shared protein families between genomes i and j . An arrow at the upper diagonal points to genome pairs that are disconnected at T 70 . At T 70 , several highly connected genera clusters within the NSP are clearly observed ( fig. 1 ), and the median of shared protein families between any pair of species is 17. Edges of weight >2,000 are found almost exclusively among species from the same genus or class. However, even at the high identity threshold of 70%, the NSP is almost a clique, with 298 (90.6%) of the species still completely connected within the network. In total, 3,637 protein families are present in species from two classes or more; hence, they are distributed across wide taxonomic scale. These protein families are relatively small, 2,331 (64%) of them are present in ≤10 species. Such patchy protein families comprising orthologs from different classes, at the protein identity threshold where only strains are still highly connected, can be the result of vertical inheritance and widespread differential loss or LGT. If the former, then these are highly conserved proteins that originated in the proteobacterial LUCA and were lost during evolution in most of the species, except for the ones where they are still present. This argument is very problematic. First, because there are no proteobacterial universal proteins at T 70 ( supplementary table S3 , Supplementary Material online) so that proteins of proteobacterial LUCA origin are more diverged than T 70 . Second, protein conservation and the propensity to be lost are negatively correlated ( Krylov et al. 2003 ) so that such an abundant loss during evolution of those protein families would be highly improbable. Hence, orthologs in the highly patchy protein families are candidates for LGT among proteobacterial species. To test the characteristics of these LGT-candidate protein families, we investigated the functional annotation of extra patchy protein families that are present in ≤10 genomes from two proteobacterial classes or more at T 70 (2,430 families). Many of these proteins (729; 31%) are annotated as hypothetical proteins, mostly common to betaproteobacteria and gammaproteobacteria (214). Only one hypothetical protein is common to genomes from four different classes, found in Acidovorax JS42 (betaproteobacteria), Aeromonas hydrophila ATCC 7966 (gammaproteobacteria), Aeromonas salmonicida A449 (gammaproteobacteria), Bdellovibrio bacteriovorus (deltaproteobacteria), Herminiimonas arsenicoxydans (betaproteobacteria), Mesorhizobium loti (alphaproteobacteria), and Sorangium cellulosum str. So ce 56 (deltaproteobacteria). A Blast search in NCBI showed that this protein is annotated in other bacterial genomes as glyoxalase protein family. Proteins of this family are important for the detoxification of methylglyoxal ( Sukdeo and Honek 2008 ). Most of the annotated proteins are involved in metabolic and cellular processes, whereas the minority are informational genes. We find that 101 (4%) of these proteins are plasmid-related proteins, such as IS-elements transposase, integrases, and stabilization proteins. In contrast, we find that 44 (2%) protein families are phage-related proteins, such as phage tail proteins, prophage CP4-57 regulatory protein, and phage integrase. These frequencies may be used for inference about relative contribution of LGT by plasmids (conjugation) versus transduction in the present genome sample. These two modes of LGT are very different from each other in the distance that is required between donor and recipient. Conjugation may be viewed as a personal delivery, whereas transduction is more like long distance mail. The 2-fold higher frequency of plasmid-related genes in comparison with phage related in the very patchy gene distribution patterns suggests that much of the LGT in proteobacteria in this sample is mediated by conjugation, where donor and recipient cells are in close proximity ( Halary et al. 2010 ). The set of highly similar patchy protein families includes three ribosomal proteins and one tRNA synthetase. The T 70 protein family of 50S ribosomal protein L31 groups six betaproteobacteria with Methylococcus capsulatus str. Bath, a gammaproteobacterium. A phylogenetic tree of this protein including all species clustered at T 60 reveals that the same seven species are grouped together in a clade, indicating that M. capsulatus str. Bath, has acquired its ribosomal protein L31 from a betaproteobacterium ( supplementary fig. S1 , Supplementary Material online). The T 70 cluster of asparaginyl-tRNA synthetase groups nine gammaproteobacteria with Myxococcus xanthus str. DK1622, a deltaproteobacterium. A phylogenetic tree of this protein as T 55 results in a clade of the same nine species, indicating that M. xanthus str. DK1622 acquired its asparaginyl-tRNA synthetase from a gammaproteobacterium ( supplementary fig. S2 , Supplementary Material online). Both 50S ribosomal protein L31 and asparaginyl-tRNA synthetase are single-copy genes. Single-copy genes have been recently found to be more resistant to transfer into E. coli than multicopy genes ( Sorek et al. 2007 ). But these two examples show that single-copy informational genes can be replaced via LGT, consistent with other reports in the literature ( Chan et al. 2009 ). Modules within the NSP Using a modularity function that classifies the genomes into modules, we identified connectivity patterns across the NSP. These modules are groups of genomes more densely connected among themselves than with genomes outside the group ( Newman 2006 ; Dagan et al. 2008 ). Across different identity thresholds ( T 30 – T 70 ), the modularity function applied to the NSP reveals a structure of genetic connectivity (shared genes) that does not strictly overlap with the proteobacteria classes as defined by traditional means, that is, their rRNA sequence ( fig. 2 ). At T 30 , the NSP comprises four modules. The first module (purple) includes the majority of alphaproteobacteria and two deltaproteobacteria ( M. xanthus str. DK1622, S. cellulosum str. So ce 56). The second module (green) includes alphaproteobacterial endosymbionts ( Anaplasma , Ehrlichia , Rickettsia , and Wolbachia ), the majority of deltaproteobacteria, all epsilonproteobacteria, two betaproteobacterial human pathogens ( Neisseria meningitides and N. gonorrhoeae ), and several gammaproteobacterial endosymbionts ( Coxiella , Legionella , Francisella , and Xylella ). The third module (blue) includes the majority of betaproteobacteria and few soil bacteria from the gammaproteobacteria, including Pseudomonas and Xanthomonas . The last module (cyan) is specific to gammaproteobacteria. F IG . 2. Modules in the NSP in the different protein sequence identity thresholds. Modules are shown as colored boxes within columns for thresholds from T 30 to T 70 . Proteobacterial orders are indicated in rows for comparison. An expanded table of the panel containing all species names is given in supplementary table S4 ( Supplementary Material online). Reconstruction of modules from the NSP at T 35 – T 50 results in only three modules. One module includes all alphaproteobacteria, epsilonproteobacteria, and deltaproteobacteria together with seven strains of Francisella tularensis and one F. philomiragia (gammaproteobacteria). Another module includes all betaproteobacteria together with many soil gammaproteobacteria, including Acinetobacter baumannii , seven Pseudomonas species, three species of Psychrobacter , and four species of Xanthomonas . The third module is unique to gammaproteobacteria. At T 55 – T 60 , the betaproteobacteria and gammaproteobacteria fall into three class-specific modules, epsilonproteobacteria and deltaproteobacteria are joined with Francisella (gammaproteobacteria), and all alphaproteobacteria are joined with S. cellulosum str. So ce 56 (deltaproteobacteria). At T 65 – T 70 , the alphaproteobacterial endosymbionts fall apart, with several modules that are common to alphaproteobacteria, deltaproteobacteria, and epsilonproteobacteria. The betaproteobacteria appear as a unique module, whereas gammaproteobacteria disarticulate into seven modules ( supplementary table S4 , Supplementary Material online). A hefty debate is currently ablaze about the utility and meaning of the “tree of life” (see Doolittle and Bapteste 2007 vs. Galtier and Daubin 2008 cf. Bapteste et al. 2009 ), particularly in the context of the overall evolutionary history of prokaryotes. One could argue that the debate boils down to the difference between attempts to reconstruct the whole of the evolutionary process and attempts at organismal classification ( Doolittle 1999 ). Proponents of the tree of life are arguing that one or a few genes serve as a useful and valid proxy for the evolution of the whole chromosome ( Ciccarelli et al. 2006 ; Galtier and Daubin 2008 ). Dissidents are arguing that since only about 30 genes are demonstrably present across many genomes (but very often sharing less than 20% amino acid identity in most comparisons) the “tree of life” constructed by such means speaks for only about 1% of the data in genomes ( Dagan and Martin 2006 ), which typically harbor about 3,000 genes. The modules of the present study point to issues concerning the concept of phylogeny within proteobacteria. Phylogeny usually refers to a hierarchical branching pattern, as in a phylogenetic tree. If we look at the modules that are identified here on the basis of shared genes ( fig. 2 ), the classification of proteobacteria into alpha, beta, gamma, delta, and epsilon groups is not recovered for any threshold. Indeed, the only of the five classes that is recovered as a distinct module at any of the nine thresholds is the betaproteobacteria class at thresholds T 55 , T 65 , and T 70 ( fig. 2 ). The modules of shared genes detected here do not reflect a hierarchical “phylogeny” of the proteobacterial classes as suggested by “tree of life” schemes based on a few concatenated genes. For example, Ciccarelli et al. (2006) reported a branching order of ((((( γ , β ), α ), ϵ ), δ ),outgroup) for the proteobacterial classes. No such phylogenetic hierarchy is suggested by the modules of shared genes ( fig. 2 ). This reinforces an earlier criticism that the phylogeny of a sample representing 1% of the genome is a poor proxy for what is to be found in the rest of the genome. We do observe, however, a module at T 35 – T 50 grouping the ( α , ϵ , δ ) classes together with some γ -representatives, most notably the Thiotrichales, represented here by the deep-sea vent chemoautotroph Thiomicrospira ( Scott et al. 2006 ) and strains of the intracellular pathogen F. tularensis ( Oyston 2008 ) plus Magnetococcus . Modules within the gammaproteobacteria correspond to some extent to family-level classifications of this class, which are also poorly resolved with concatenated sequences ( Gao et al. 2009 ). Species included in the NSP modules at all protein sequence identity thresholds differ significantly in their genome size ( P < 0.05 using the Kruskal–Wallis test; Zar 1999 ), hence genome size is not the prime determinant of module structure. Nonetheless, endosymbionts that are all characterized by very small genomes are grouped into common modules across taxonomic class boundaries, but this is because they tend to relinquish the same sets of genes ( Pal et al. 2006 ; Moran 2007 ) not because the genomes are small per se. Moreover, betaproteobacteria and gammaproteobacteria whose habitat is mainly within the soil are clearly grouped together in varying protein sequence identity thresholds ( figs. 1 and 2 ). This finding is in line with the observation that cooccurring microbes have similar genomes regardless (sometimes) of their phylogenetic relatedness ( Chaffron et al. 2010 ) and the view that transfer might be more frequent between genomes of prokaryotes sharing similar habitats ( Jain et al. 2003 ). Overall, community structure within the NSP appears to have a phylogenetic backbone but is also influenced by bacterial lifestyle and habitat. Minimal Lateral Networks Gene sharing patterns found here indicate that LGT is common among proteobacteria. But how frequent is frequent? To quantify the lower bound frequency of LGT at three phylogenetic depths within proteobacteria—phylum, class, and species—we constructed MLNs ( Dagan et al. 2008 ). In that approach, LGT frequency is inferred against the criterion of ancestral genome size. An evolutionary model that entails no LGT during evolution results in untenably large ancestral genomes ( Doolittle et al. 2003 ). Yet, if genome evolution in the past was not fundamentally different from todays, then ancestral genomes should have similar sizes to those of contemporary genomes. The approach is thus based on applying evolutionary models that allow increasing frequencies of LGT, until the distributions of ancestral and contemporary genome sizes are statistically reconciled ( Dagan and Martin 2007 ). Phylogenetic inference by the MLN approach yields estimated ancestral genome sizes together with an inference of laterally shared gene families among species or groups of species, the gene distributions of which are better explained by LGT than the phylogenetic tree. These two outcomes can be graphically represented by a network in which the vertices are the nodes of the reference tree, and the edges are either vertical tree branches or inferred lateral gene sharing edges ( Dagan et al. 2008 ). Our results suggest that LGT is more frequent at the phylum level than in the class or species level. For the data of all proteobacteria species, a model that allows up to seven LGTs per protein family ( LGT 7 ) was accepted ( P = 0.44, using Wilcoxon test ( Zar 1999 ; fig. 3 ). Although seven LGTs per family are allowed in this model, only a minority of the gene occurrence patterns require that amount. In most (28%) of the protein families whose evolution includes LGT, it occurred only once, whereas protein families whose evolution includes seven LGTs are very rare (0.78%; table 2A ). The weighted mean LGT frequency within proteobacteria phylum is thus 1.9 LGTs per protein family. Table 2. Statistically Accepted LGT Allowance Models Using T 30 Protein Families for the Different Data Sets with (A) rRNA and (B) Gene Content Reference Trees. Data Set No. of Species No. of Families LGT Model P Value Mean LGT Frequency 1 Origin 2 Origin 3 Origin 4 Origin 5 Origin 6Origin 7 Origin 8 Origin A Proteobacteria 329 74,667 LGT 1 0.44 1.9 18,763 (25%) 21,366 (29%) 9,048 (13%) 11,760 (16%) 6,535 (9%) 3,520 (5%) 2,707 (4%) 582 (1%) Alphaproteobacteria 82 27,810 LGT 1 0.25 0.6 6,018 (25%) 17,760 (75%) LGT 3 0.43* 1.1 6,018 (25%) 6,792 (29%) 8,329 (35%) 2,639 (11%) Betaproteobacteria 52 25,199 LGT 1 0.26* 0.7 3,830 (19%) 16,492 (81%) LGT 3 0.14 1.1 3,830 (19%) 5,816 (29%) 9,014 (44%) 1,662 (8%) Gammaproteobacteria 157 40,327 LGT 3 0.46 1.2 9,179 (25%) 10,253 (28%) 13,089 (36%) 3,669 (10%) Escherichia coli 12 7,879 LGT 1 0.48 0.7 653 (10%) 5,589 (90%) Francisella tularensis 7 1,840 Origin 0.11 LGT 1 0.47* 0.3 1,255 (73%) 462 (27%) Yersinia pestis 7 4,439 LGT 1 0.59 0.9 122 (3%) 4,080 (97%) B Proteobacteria 329 74,667 LGT 1 0.98 1.7 21,782 (29%) 20,824 (28%) 9,030 (12%) 10,393 (14%) 6,318 (8%) 3,454 (5%) 2,368 (3%) 472 (1%) Alphaproteobacteria 82 27,810 LGT 1 0.37* 0.6 6,397 (27%) 17,381 (73%) LGT 3 0.32 1.1 6,397 (27%) 6,743 (28%) 8,390 (35%) 2.248 (9%) Betaproteobacteria 52 25,199 LGT 1 0.1 0.6 4,707 (23%) 15,615 (77%) LGT 3 0.14* 11 4,707 (23%) 6,116 (30%) 7,117 (35%) 2,382 (12%) Gammaproteobacteria 157 40,327 LGT 3 0.47* 1.5 10,751 (30%) 9,627 (27%) 4,241 (12%) 6,532 (18%) 3,093 (9%) 1,436 (4%) 482 (1%) 28 (0%) E. coli 12 7,879 LGT 1 0.19 0.3 3,714 (60%) 2,453 (40%) F. tularensis 7 1,840 Origin 0.23 LGT 1 0.73* 0.2 1,311 (76%) 406 (24%) LGT 3 0.14 0.3 1,238 (72%) 479 (28%) Y. pestis 7 4,439 Origin 0.1 LGT 1 0.95* 0.2 3,400 (81%) 802 (19%) LGT 3 0.07 0 3 3,400 (81%) 280 (7%) 522 (12%) LGT 1 0.05 0.3 3,400 (81%) 280 (7%) 314 (7%) 208 (5%) * For data sets where more than one model was statistically accepted, the most probable model is marked by an asterisks. F IG . 3. Distribution of contemporary and ancestral genome sizes in phylum depth under the different LGT allowance models (left) and schematic representation of the evolutionary scenario implicated by the models (right). The models ( A ) loss only, ( B ), single origin, ( C ) LGT 1 , and ( D ) LGT 3 result in significantly larger ancestral genome sizes in comparison to contemporary genome sizes ( α = 0.05, using Kolmogorov–Smirnov test). The LGT 7 model ( E ) results in similar distributions of ancestral and contemporary genome size ( P = 0.44, using Wilcoxon test). The LGT 15 model ( F ) results in significantly smaller ancestral genome sizes in comparison to contemporary genome sizes ( α = 0.05, using Kolmogorov–Smirnov test). Within the classes of proteobacteria, the LGT 3 model was accepted for the alphaproteobacteria and gammaproteobacteria, with a weighted LGT frequency of 1.3 per protein family in both groups. The frequency of LGT events per protein family follows a similar distribution in alphaproteobacteria and gammaproteobacteria as well ( table 2A ). Within the betaproteobacteria, the LGT 1 model was accepted, with a weighted mean LGT frequency of 0.8 per protein family ( table 2A ). None of the models was accepted for the deltaproteobacteria and epsilonproteobacteria. However, in both groups, resulting ancestral genome sizes from the origin-only model are significantly larger than contemporary genome sizes ( P < 0.01, using Kolmogorov–Smirnov test [ Zar 1999 ]; supplementary fig. S3 , Supplementary Material online). Moreover, ancestral genome sizes resulting from the LGT 1 model are significantly smaller than contemporary genome sizes ( P < 0.01 using Kolmogorov–Smirnov test; supplementary fig. S3 , Supplementary Material online). This suggests that deltaproteobacteria and epsilonproteobacterial gene distribution patterns, in combination with the rRNA reference tree topology, require an evolutionary model that is somewhere between origin-only and LGT 1 models, allowing probably a single LGT event to only part of the protein families. However, our current MLN reconstruction approach applies uniform model choice to all protein families. A more complicated approach in which each protein family is fitted its own model would require an a priori assumption of gene origin to loss ratios (e.g., Kunin et al. 2005 ), these are regarded in the MLN approach ( Dagan and Martin 2007 ) as a variable whose value is to be inferred rather than a user-defined parameter. For the LGT frequency estimation at the species level, we selected three gammaproteobacterial species whose genome sample of sequenced strains is large enough to conduct our analysis. These include E. coli (12 genomes), F. tularensis (7 genomes), and Yersinia pestis (7 genomes). For these three data sets, the LGT 1 model was accepted ( P = 0.48, 0.47, and 0.49 respectively, using Wilcoxon test; supplementary fig. S3 , Supplementary Material online) with a weighted mean LGT frequency of 0.7 LGTs per protein family in E. coli , 0.3 LGTs per protein family in F. tularensis , and 0.9 LGTs per protein family in Y. pestis . LGT at the species level is recombination. Hence, the LGT rates calculated here for the species data sets may be regarded a lower bound estimate for recombination rates. Because in our approach we analyze the presence/absence patterns of genes and not their sequences, our inference yields an estimate for the gene spread by recombination but largely underestimates overall recombination rates. LGT Inference against a Gene Content Reference Tree LGT frequencies inferred using the MLN approach are robust to different reference phylogenetic trees reconstructed from various protein families, yet they may be affected by the patchiness of the gene distribution patterns across the reference tree ( Dagan and Martin 2007 ). This is because when the LGT allowance in the MLN approach is increased from none to two and then more gene origins (by gene birth or LGT), these are distributed in pairs to descendants of an ancestor for which a gene origin was reconstructed using the previous model. Moreover, by using a species tree reconstructed from the rRNA sequences, we assume that the phylogeny of a single operon truly represents the evolutionary history of proteobacteria. This assumption may be problematic for the evolution of prokaryotes that is reticulated by nature ( Bapteste et al. 2009 ). Here, we test the robustness of the MLN approach to the patchiness of gene distribution patterns and the rRNA phylogenetic tree by using a gene content tree ( Snel et al. 1999 ) as the reference tree. Such a reference tree is expected to minimize the patchy gene distribution patterns and thereby provide more conservative estimates of LGT among proteobacteria. Gene content trees were reconstructed from the complete presence/absence data at T 30 using Wagner parsimony approach ( Felsenstein 1983 ). The gene content tree including all proteobacteria was rooted on the branch separating ( α , δ , ϵ ) from ( β , γ ). The resulting tree supports the monophyly of alphaproteobacteria, deltaproteobacteria, and epsilonproteobacteria (but not the position of the root, obviously). The betaproteobacteria branch with gammaproteobacteria in two groups, one includes N. meningitides (betaproteobacteria) and Polynucleobacter (gammaproteobacteria), whereas the other includes the rest of the species divided into two class-specific clades ( supplementary fig. S4 , Supplementary Material online). Reconstruction of the MLN for all proteobacteria using the gene content tree as the reference tree yielded the LGT 7 model as the best fit between ancestral and contemporary genome sizes, whereas all other LGT allowance models were rejected ( supplementary figs. S5 and S6 , Supplementary Material online). Although it is the same LGT allowance model that was inferred using the rRNA reference tree, the mean LGT rate is lower—“but only slightly so” — using the gene content reference tree, with a weighted mean of 1.7 LGTs per protein family, in comparison to 1.9 with the rRNA tree. This somewhat lower rate is the result of reduced patchiness in gene distribution patterns using the gene content tree, leading to 29% monophyletic families (in comparison to 25% using the rRNA tree) whose distribution on the tree requires no LGT ( table 2B ). The small increment of average LGT rate from 1.9 to 1.7 using the gene content tree, where the patchiness criterion is used to cluster the genomes, simply reflects the patchiness of the data in total. In other words, the present data require a substantial amount of LGT to account for the observed gene distributions, any way one cuts the cake. We repeated the same inference procedure for the class- and species-level data sets. At the class level, the best-fitting model using the gene content tree resulted in an inference of a lower LGT allowance for alphaproteobacteria ( LGT 1 ) and higher LGT allowance in betaproteobacteria ( LGT 3 ) and gammaproteobacteria ( LGT 7 ). As with the rRNA reference tree, no model was accepted for the deltaproteobacteria and epsilonproteobacteria, where the distribution of ancestral genome sizes shows that an allowance model between origin only and LGT 1 , had it existed in our approach, might be the most fitting for these classes ( supplementary fig. S5 , Supplementary Material online). In the three species-level data sets ( E. coli , F. tularensis , and Y. pestis ), the same LGT allowance model was accepted using the gene content reference tree, with slightly lower LGT rates ( table 2B ). Hence, our attempt to minimize LGT rate inference by reducing the patchiness of gene distribution patterns across the reference tree using the gene content tree resulted in more monophyletic protein families, yet the inferred LGT allowance models and average LGT rate were hardly changed and sometimes were even increased. MLN Properties The MLN reconstructed for all proteobacteria using T 30 protein families, with the RNA reference tree, and the LGT 7 model contains in total 657 nodes, with 329 external nodes (—operational taxonomic units [OTUs]) and 328 internal nodes (hypothetical taxonomic units [HTUs]), connected by 51,762 lateral edges ( fig. 4 ). For protein families that have undergone more than one LGT, the number of lateral edges in the MLN exceeds the minimum number of LGTs required to account for the gene distribution. Hence, to address LGT network properties for the MLN, 1,000 rMLN were generated in which the number of lateral edges and the minimum number of LGTs for genes transferred more than once correspond exactly. Lateral edge frequency and edge weight distribution are similar among the rMLN networks. The number of lateral edges in the rMLNs is 3,345 ± 73 (coefficient of variation = 2%) on average. The connectivity (number of lateral edges per node) ranges between 0 and (344–384) with a mean between 100 and 102 and median between 85 and 91 ( table 3 ). The connectivity distribution is semi-exponential with very few nodes that are highly connected ( fig. 5 A ). Bigger genomes are generally more highly connected than smaller genomes, yet genome size explains only 16% of the variation in connectivity ( P < 0.01, using Spearman correlation; Zar 1999 ). Table 3. Statistical Properties of MLN/rMLNs in Phylum and Class Level. Prateobacteria Alphaproteohacteria Betaproteobacteria Gammaproteobacteria Escherichia coli Francisella tularensis Yersinia pestis No. edges 33,457 ± 73 3,606 ± 17 1,595 8,447 ± 31 108 24 37 Mean connectivity 100–102 43–44 31 53–54 6 4 5 Median connectivity 85–91 46–52 33 47–52 5 5 6 No. of edges > 20 982 ± 10 (3 ± 0.03%) 313 ± 6 (9 ± 0.2%) 145 (9%) 409 ± 6 (5 ± 0.l%) 29 (27%) 4 (17%) 7 (19%) No. of edges ≥ 5 587 ± 30 (18 ± 0.l%) 1,395 ± 12 (39 ± 0.4%) 588 (37%) 1,855 ± 15 (22 ± 0.2) 60 (56%) 15 (63%) 18 (49%) No. edges = 1 17,018 ± 91 (51 ± 0.2%) l,017 ± 21 (28±0.5%) 487 (30%) 3,902 ± 42 (46 ± 0.4%) 20 (19%) 1 (4%) 6 (16%) No. of edges = 2 592 ± 61 (18 ± 0,2%) 541 ± 18 (15 ± 0.5%) 263 (16%) 1,476 ± 29 (17 ± 0.3%) 11 (10%) 0 (0%) 7 (19%) No. of OTU–OTU edges 1,188 ± 42(35 ± 0.l%) l,206 ± 9+(33 ± 0.2%) 555 (35%) 2,832 ± 18 (34 ± 0.2%) 41 (38%) 9 (37%) 18 (49%) No. of HTU–HTU edges 604 ± 39 (18 ± 0.l%) 686 ± 9 (19 ± 0.2%) 293 (18%) 1,696 ± 16 (20 ± 0.2%) 17 (16%) 4 (17%) 4 (11%) No. of OTU–HTU edges 1,553 ± 55 (46 ± 0.2%) 1,713 ± 12 (47 ± 0.3%) 747 (47%) 3,919 ± 23 (46 ± 0.2%) 50 (46%) 11 (46%) 15 (41%) OTU–OTU edges connect between two external nodes (contemporary species). HTU–HTU edges connect between two internal nodes (ancestral species). OTU–HTU edges connect between an external and an internal nodes. F IG . 4. A minimal LGT network for 329 proteobacteria. ( A ) The reference tree used to ascribe vertical inheritance for inference of the MLN. ( B ) The MLN showing all 51,762 edges of weight ≥1 gene in the MLN. Vertical edges are indicated in gray, with both the width and the shading of the edge shown proportional to the number of inferred vertically inherited genes along the edge (see scale on the left). The lateral network is indicated by edges that do not map onto the vertical component, with number of genes per edge indicated in color (see scale on the right). ( C ) The MLN showing only the 13,632 edges of weight ≥5 genes. ( D ) The network showing only the 3,007 edges of weight ≥20 genes. F IG . 5. Properties of the minimal LGT networks in phylum and class scales. Properties are shown for a randomly selected replicate. The coefficient of variation for the whole data was ∼2% ( table 3 ). ( A – D ) Distribution of connectivity, the number of one-edge-distanced neighbors for each vertex, in the MLN. ( E – H ) Probability density function (PDF) of edge weight in the lateral component of the MLN. The MLN reconstructed at the proteobacterial class level shows the distribution of laterally shared genes in higher resolution. Network properties for the alphaproteobacteria and gammaproteobacteria were calculated from 1,000 rMLN networks, the statistics of which show uniformity of lateral edge frequency and edge weight distribution ( table 3 ). Data for the betaproteobacteria were extracted from the MLN directly because the best-fitting model was LGT 1 , which results in an MLN where the number of edges corresponds the minimum number of LGT events per protein family. The connectivity distribution in the alphaproteobacterial MLN is bimodal, suggesting two groups of species that are either weakly or strongly connected within the lateral network ( fig. 5 B ). The graphical representation of the MLN for that class reveals that the Rickettsiales comprise the weakly connected group ( fig. 6 A ). In our data set, the order Rickettsiales includes 21 endosymbiotic parasites from the genera Anaplasma , Ehrlichia , and Rickettsia . The host-associated lifestyle of these species is a barrier to LGT in many cases and probably the reason for their low connectivity in the MLN. The connectivity distribution in the betaproteobacterial MLN is almost uniform ( fig. 5 C ) with similar frequencies of nodes across the connectivity range (0–50 edges per node) and five more nodes whose connectivity is above this range. Clades of symbionts within the betaproteobacterial MLN, the Neisseriales and Nitrodomonadales, are weakly connected ( fig. 6 B ). The Burkholderiales in our sample include 31 species of diverse lifestyles that account for the majority (60%) of betaproteobacterial species in the data. The overall gene distribution patterns are quite uniform across that order ( fig. 1 ), yet the parasites among them ( Ralstonia species) having lower connectivity than the free-living species ( Burkholderia species; fig. 6 B ). F IG . 6. A minimal LGT network for proteobacterial classes alpha ( A ), beta ( B ), and gamma ( C ). Vertical edges are indicated in gray, with both the width and the shading of the edge shown proportional to the number of inferred vertically inherited genes along the edge (see scale bar). The lateral network is indicated by edges that do not map onto the vertical component, with number of genes per edge indicated in color (see scale bar). The MLN showing only edges of weight ≥5 genes. The connectivity distribution in the gammaproteobacterial MLN is semi-exponential ( fig. 5 D ). The graphical representation of the gammaproteobacterial MLN shows that symbionts, such as Pasteurellales, are weakly connected within the lateral network. The Enterobacteriales, comprising about third of the gammaproteobacteria in our sample (51 species) include four main genera, Escherichia , Shigella , Salmonella , and Yersinia . The MLN contains 1,326 (16%) lateral edges connecting among the nodes (internal and external) in this clade, suggesting abundant LGT among species in this group, with the exception of Yersinia , that like other pathogenic and symbiotic strains in our data set are relatively disconnected from the network ( fig. 6 C ). The distribution of lateral edge weight in the proteobacterial MLN is linear in log–log scale ( fig. 5 E ), with a majority of single gene edges (51 ± 0.2%) and a minority of heavy edges weighing 20 genes or more (3 ± 0.03%). Similar edge weight distributions are observed within the alphaproteobacteria, betaproteobacteria, and gammaproteobacteria MLNs ( table 3 and fig. 5 F – H ). This means that most of the LGT events among proteobacteria entail single genes rather than bulk transfers. The MLN reconstruction for all species-level data sets, using both reference trees, prefers the LGT 1 model with an average LGT frequency of about one LGT per protein family ( table 2 ). The MLN reconstruction for the species level typically results in a heavy lateral edge that is found close to the root, between the first two nodes that diverge from it ( supplementary fig. S7 A–C , Supplementary Material online). Such a lateral edge means that many gene families of patchy distribution are shared between the two immediate descendants of the root node, the distribution of which cannot be explained by vertical inheritance alone. Reducing the patchiness of the gene distribution patterns by using a gene content reference tree resulted in similar MLNs ( supplementary fig. S7 D–F , Supplementary Material online). In species-level MLNs, the majority of nodes are connected by a lateral edge, exept two to four nodes ( Supplementary fig. S8 , Supplementary Material online). The distribution of edge weights is semi-linear in a log–log scale, hence most of the LGT events are of single genes, whereas individual transfer events involving many genes are rare. The distribution of lateral edges within the proteobacteria rMLN shows that the probability for an intraclass lateral edge (9.4 ± 0.03) is similar to the probability for an interclass lateral edge (6.9 ± 0.02). However, the median edge weight of intraclass edges, which is two genes per edge in all rMLNs, is significantly larger ( P < 0.05) than that of interclass edges, a single gene per edge in all rMLNs. This means that the probability for an LGT event within and outside the class is similar, yet more genes are transferred per LGT event between species from the same class. The probability for a lateral edge between the different classes reveals that LGT between alphaproteobacteria, deltaproteobacteria, and betaproteobacteria is similar, but LGT between epsilonproteobacteria or gammaproteobacteria and other classes is lower ( table 4 ). Table 4. Frequency and Weight of Lateral Edges in Intraclass and Interclass Subsets. Alphaproteobacteria Deltaproteobacteria Epsilonproteobacteria Betaproteobacteria Gammaproteobacteria Alphaproteobacteria 12.9 ± 0.07 (3–3) Deltaproteobacteria 14 ± 1.5 (2–2) 29 ± 0.3 (7–9) Epsilonproteobacteria 3.8 ± 0.1 (1–1) 19 ± 0.3 (2–2) 20.2 ± 0.03 (3–5) Betaproteobacteria 11 ± 0.08 (1–2) 13.8 ± 0.1 (1–2) 3.68 ± 0.1 (1–1) 15.8 ± 0.1 (2–3) Gammaproteobacteria 5.1 ± 0.04 (1–1) 8.1 ± 0.09 (1–1) 3.5 ± 0.06 (1–1) 7.1 ± 0.05 (1–1) 7.4 ± 0.04 (1–2) Numbers in parenthesis denote edge weight range. Edge probability is calculated as the frequency of edges divided by the number of nodes in the group. Proportion of Recent Gene Acquisition and Cumulative Impact of LGT Most of the edges in the proteobacterial MLN (46 ± 0.2% of edges in the rMLN) connect between OTU nodes (contemporary genomes) and HTU nodes (ancestral genomes). Such edges are inferred for protein families that are shared among a group of species where all except one are grouped into one monophyletic clade. The reconstructed lateral edge connects the common ancestor of that clade and the OTU of the outsider species. Lateral edges connecting two OTU nodes are slightly less frequent (35 ± 0.1%), whereas edges connecting two HTU nodes are the minority (18 ± 0.1%; fig. 7 ). Similar ratios of lateral edge types were inferred for the classes and species data sets ( table 3 ). F IG . 7. A three-dimensional projection of the MLN. Edges in the vertical component are shown in the same gray scale as in figure 3 . Vertices inferred as gene origin in the same protein family are connected by a lateral edge signifying a laterally shared gene. Lateral edges are classified into three groups according to the types of vertices they connect within the vertical component (see table 3 for details): 11,941 OTU–OTU edges (magenta), 15,425 HTU–OTU edges (yellow), and 6,066 HTU–HTU edges (cyan). Lateral edges connecting between two OTUs reflect recent LGT events. The proportion of protein families connected by an OTU–OTU edge per genome may serve as a lower bound estimate for the proportion of recently acquired genes within the genome. The average proportion of recent acquisitions per genome inferred from the MLN in phylum depth with the rRNA reference tree is 9.6% recently acquired genes per genome. Moreover, the frequency of recently acquired genes positively correlated with genome size ( r s = 0.6, P < 0.01). Similar mean proportions of recently acquired genes are estimated for the three classes (7–9%; table 5 ). The estimated proportions in the species level are about 4% recent acquisitions ( table 5 ). To test how our estimates are affected by the sample of species included in the MLN, we compared them for the same group of species, from the MLNs reconstructed in class and phylum phylogenetic depth. We find that larger sample size results in slightly higher proportions of recently acquired genes (0.1–4.2% difference; table 5 ). Hence, the phylogenetic depth (i.e., sample size) has little influence on the inferred proportions of recently acquired genes using the MLN approach. Table 5. Recently Acquired Genes and Cumulative Impact of LGT. Phvloqenetic depth % Recent LGT by MLN % Recent LGT by Nucleotide Pattern Ratio of MLN/Nucleotide Pattern % Cumulative LGT by MLN Phylum Class Species Phylum Phylum Class Species Proteobacteria 9.7 ± 7.7 21.5 ± 8.9 0.5 ± 0.6 73.7 ± 10.9 Alpha 9.6 ± 7.0 9.2 ± 8.5 16.6 ± 7.9 0.6 ± 0.5 69.1 ± 9.9 60.9 ± 12.3 Beta 11.1 ± 5.8 6.9 ± 6.0 26.8 ± 8.4 0.5 ± 0.3 75.2 ± 4.7 60.0 ± 8.3 Gamma 7.3 ± 5.1 7.2 ± 6.9 21.4 ± 7.8 0.4 ± 0.4 78.6 ± 9.4 76.6 ± 10.2 Escherichia coli 5.0 ± 3.8 4.1 ± 3.1 3.3 ± 2.7 28.5 ± 2.1 0.2 ± 0.1 85.2 ± 3.8 84.6 ± 3.8 26.8 ± 3.5 Francisella tularensis 5.0 ± 5.3 4.1 ± 5.2 4.4 ± 6.9 17.5 ± 0.7 0.3 ± 0.3 67.4 ± 2.1 65.1 ± 2.0 17.0 ± 2.0 Yersinia pestis 4.0 ± 3.3 3.1 ± 2.6 3.9 ± 7.3 26.7 ± 0.8 0.2 ± 0.1 86.0 ± 2.0 84.6 ± 2.0 17.5 ± 3.4 The MLN, comprising of both phylogenetic tree for the vertically inherited genes and lateral network for the laterally transferred genes, enables us to estimate the cumulative impact of LGT during microbial evolution. The proportion of protein families within each genome that is connected by a lateral edge reflects the proportion of genes within the genome that was affected by LGT during their history. Within the phylum depth using the rRNA reference tree, we find that, on average, 73% of the genes per genome are affected by LGT at some point during evolution. A similar proportion is observed with the class depth for gammaproteobacteria, whereas in alphaproteobacteria and betaproteobacteria, we find lower cumulative impact of LGT (60%; table 5 ). The same inference in species depth yields significantly lower proportions (17–26%; table 5 ). To test if the cumulative impact of LGT in species depth is indeed lower or rather an outcome of smaller sample size, we compared the inference for the same species using the phylum and class depth data sets. We find that the proportion of genes affected by LGT during evolution inferred in species depth is much lower than the inference using the class or phylum data sets. How Severely Does the MLN Underestimate LGT? The estimated proportion of recently acquired genes per genome using the MLN is 9.7% of each genome in the phylum depth on average, that is, lower than the proportion of recent LGT inferred using aberrant nucleotide patterns that in earlier studies was between 14% and 18% per genome ( Lawrence and Ochman 1998 ; Nakamura et al. 2004 ). The MLN approach is expected to yield lower bound minimum estimates mainly because it relies on gene presence/absence patterns that are uninformative for evolutionary events, such as allele recombination and gene replacement by LGT (e.g., Andam et al. 2010 ), and because it conservatively does not count all LGT events that might be detected by tree comparisons ( Dagan et al. 2008 ). How severely does the MLN underestimate LGT? In order to ascertain this, we compared the proportions of recent LGT per genome using the MLN approach with that determined on the basis of aberrant nucleotide patterns by detecting all genes having significantly different GC content in comparison to their genome. The GC content method preferentially reveals recently acquired genes that exhibit an atypical codon usage indicating their foreign origin ( Lawrence and Ochman 1998 ; Nakamura et al. 2004 ). Across the phylogenetic samples studied, the frequency of genes detected as recently acquired using the two methods is positively correlated ( r s = 0.55, P < 0.01) ( table 5 ). However, the GC method detects an average of 21% recent acquisitions per genome in the proteobacterial phylum sample or roughly twice the value estimated by MLN, whereby the degree to which the MLN approach underestimates recent LGT increases to about a factor of six as the sample approaches the species level ( table 5 ). Both effects—MLN underestimation and its increase toward the species level—are attributable to the circumstance that two kinds of genes are excluded from the MLN approach. First, the GC content approach can identify acquisitions from any donor genome, whereas the MLN only identifies LGTs involving genomes within the sequenced set. Second, the GC content approach identifies LGT among singletons, whereas the MLN does not. Both effects become more severe with smaller and more closely related genome samples. Thus, although the graphical representation of the MLN ( fig. 7 ) might appear quite complex in terms of lateral edges, it still represents a minimum estimate, not an optimal estimate, of gene sharing among these genomes."
} | 15,090 |
36016062 | PMC9416718 | pmc | 7,888 | {
"abstract": "Reinforcement Learning (RL) comes with the promise of automating network management. However, due to its trial-and-error learning approach, model-based RL (MBRL) is not applicable in some network management scenarios. This paper explores the potential of using Automated Planning (AP) to achieve this MBRL in the functional areas of network management. In addition, a comparison of several integration strategies of AP and RL is depicted. We also describe an architecture that realizes a cognitive management control loop by combining AP and RL. Our experiments evaluate on a simulated environment evidence that the combination proposed improves model-free RL but demonstrates lower performance than Deep RL regarding the reward and convergence time metrics. Nonetheless, AP-based MBRL is useful when the prediction model needs to be understood and when the high computational complexity of Deep RL can not be used.",
"conclusion": "7. Conclusions This paper showed the use of AP for improving RL in the context of functional network management areas and proposed an architecture to realize a cognitive management control loop by combining AP and RL. We evaluated the proposed solution by using a prototype that combined AP and RL framed in MBRL using Q-learning and HTN. Evaluation results of the MBRL-based prototype in a simulated environment evidenced that the combination proposed improves RL but demonstrates lower performance than DRL regarding the reward and convergence time metrics. Consequently, we encourage the use of AP-based RL in scenarios with small datasets, where the expert’s knowledge can be represented in AP domains. When compared with Q-Learning, our MBRL-based approach achieves very good results in only one episode, while Q-Learning achieves similar results in eight episodes. Each episode implies a set of trial-and-error iterations, which may be prohibitive in production scenarios. Conversely, the performance of neural networks is much higher than the MBRL-based approach; this is due to the fact that the recent algorithms can achieve models with high precision. However, these models cannot be understood internally, and the data and computational effort can make it difficult to apply this approach in several scenarios. The main impact of our research is the reduction in the learning time of the reinforcement learning algorithms in the context of network management. This is achieved without the need of big datasets and high processing time. In addition, this method allows modeling the environment in HTN models that may be more human-understandable than complex neural networks. These benefits make RL closer to being used in real scenarios. There are a number of compelling future directions to this work. First, we have explored the feasibility of combining AP with Q-Learning. Our future work will focus on using other RL approaches and exploring the integration with DRL [ 67 ]. Second, as AP requires significant human expertise before it can be applied to new problems and domains. Our future work will focus on exploring the automatic generation of AP problems and domains from natural language [ 30 ]. Third, we also plan to implement and evaluate our AP-based MBRL method in other network management scenarios such as traffic control and classification [ 68 ] and intrusion detection [ 69 ]. We will compare our results with datasets from these new scenarios. We leave these problems as future work.",
"introduction": "1. Introduction Reinforcement Learning (RL) has been used recently in network management for automating diverse tasks such as traffic monitoring and routing [ 1 , 2 , 3 , 4 ]. However, most of theses applications require a huge amount of samples from the environment. This trial and error process implies that some non-optimal or wrong actions can be performed during the learning process, which is unacceptable in real scenarios with strong Service-Level Agreements (SLA) [ 5 ]. Deep Reinforcement Learning (DRL) aims at addressing some of these shortcomings while offering unique advantages in terms of accuracy. However, DRL has its own drawbacks, first the success of DRL depends on the the availability of data [ 6 ]. Second, DRL is vulnerable to adversarial examples [ 7 ]. Third, the tuning of the hyper-parameters may be very complex and typically requires a higher convergence time than RL [ 8 ]. Finally, DRL results in black boxes that are hard to interpret [ 9 ]. Therefore, in some real network management scenarios, alternative solutions are preferred even at the expense of accuracy. Existing AI-based approaches for network management automation focus on simple tasks (i.e., selecting a route). Thus, many complex tasks in network management are still manual. However, to achieve self-driving networks, it is required to automatically perform a broad set of tasks. For example, in the Radio Access Network, some tasks such as the adjustment of tilt, frequency, or radiation pattern need to be carried out automatically to achieve adaptability in cellular networks [ 10 ]. In network slicing, some of these tasks may include on-demand provisioning, migration, or scaling up/down of virtual network functions (VNFs). These tasks must be performed in order and achieve a predefined goal (maintain QoE, optimize costs). In addition, networks architectures are evolving continually; therefore, it is necessary to perform diverse management tasks for every architecture and technology. This process requires a technique capable of finding the best sequence of tasks (plans) necessary to achieve a desired solution (i.e., maximizing a long-term reward). In addition, in these diverse architectures, sometimes there are not enough data available for training neural networks. An approach for these types of Markov Decision Process (MDP) optimization problems [ 11 ] is automated planning (AP). AP builds plans to achieve a particular goal status. To come up with these plans, AP uses a model of the environment called a domain. AP offers various advantages such as expressiveness and simplicity to represent the environment [ 12 ]. Due to their similarities, AP and RL has been combined to model-based reinforcement learning [ 13 ]. Unlike model-free RL methods that learn just by interacting with the environment, MBRL methods model the environment, which reduces the number of interactions needed to learn [ 13 , 14 ]. MBRL approaches have evidenced similar performance to model-free algorithms in diverse domains but require significantly fewer interactions with the environment [ 15 , 16 , 17 ]. This paper explores the use of this concept in network management. Notably, we analyze how AP models (domains) can be used to represent the behavior and status on the network, and RL can exploit such representation to reduce the number of interactions with the environment [ 18 ]. In addition, we propose an architecture that realizes a cognitive management control loop by combining AP and RL. We evaluated our approach in a prototype that automates the network slicing admission control. Most of the existing approaches focus on using RL for punctual task such as network traffic detection [ 19 ] and load balancing [ 20 ]. However, to the best of our knowledge, a proposal describing how AP an RL can be integrated in practice for network management has not been presented. Evaluation results in simulated environment evidence that the combination proposed improves RL but demonstrates lower performance than DRL regarding the reward and convergence time metrics. Consequently, this paper encourages the use of AP-based RL in scenarios where the complexity of neural networks cannot be faced. The remainder of this paper is organized as follows. Section 3 presents a brief background about AP and RL in network management. Section 4 shows how some functional areas of network management can be automated using MBRL. Section 5 presents an architecture for cognitive management based on AP and RL for MBRL. Section 6 introduces a case study that shows how our approach can be used jointly for network slicing admission control. Section 7 concludes and explores the future work."
} | 2,030 |
35862821 | PMC9426549 | pmc | 7,889 | {
"abstract": "ABSTRACT Advances in DNA sequencing technologies have drastically changed our perception of the structure and complexity of the plant microbiome. By comparison, our ability to accurately identify the metabolically active fraction of soil microbiota and its specific functional role in augmenting plant health is relatively limited. Important ecological interactions being performed by microbes can be investigated by analyzing the extracellular protein fraction. Here, we combined a unique protein extraction method and an iterative bioinformatics pipeline to capture and identify extracellular proteins (metaexoproteomics) synthesized in the rhizosphere of Brassica spp. We first validated our method in the laboratory by successfully identifying proteins related to a host plant ( Brassica rapa ) and its bacterial inoculant, Pseudomonas putida BIRD-1. This identified numerous rhizosphere specific proteins linked to the acquisition of plant-derived nutrients in P. putida . Next, we analyzed natural field-soil microbial communities associated with Brassica napus L. (oilseed rape). By combining metagenomics with metaexoproteomics, 1,885 plant, insect, and microbial proteins were identified across bulk and rhizosphere samples. Metaexoproteomics identified a significant shift in the metabolically active fraction of the soil microbiota responding to the presence of B. napus roots that was not apparent in the composition of the total microbial community (metagenome). This included stimulation of rhizosphere-specialized bacteria, such as Gammaproteobacteria , Betaproteobacteria , and Flavobacteriia , and the upregulation of plant beneficial functions related to phosphorus and nitrogen mineralization. Our metaproteomic assessment of the “active” plant microbiome at the field-scale demonstrates the importance of moving beyond metagenomics to determine ecologically important plant-microbe interactions underpinning plant health. IMPORTANCE Plant-microbe interactions are critical to ecosystem function and crop production. While significant advances have been made toward understanding the structure of the plant microbiome, learning about its full functional role is still in its infancy. This is primarily due to an incomplete ability to determine in situ plant-microbe interactions actively operating under field conditions. Proteins are the functional entities of the cell. Therefore, their identification and relative quantification within a microbial community provide the best proxy for which microbes are the most metabolically active and which are driving important plant-microbe interactions. Here, we provide the first metaexoproteomics assessment of the plant microbiome using field-grown oilseed rape as the model crop species, identifying key taxa responsible for specific ecological interactions. Gaining a mechanistic understanding of the plant microbiome is central to developing engineered plant microbiomes to improve sustainable agricultural approaches and reduce our reliance on nonrenewable resources.",
"conclusion": "Conclusions. Here, we present the first metaexoproteomic assessment of the plant microbiome sampled from a field-grown agricultural crop. Our new technique enabled us to identify highly active taxa in the rhizosphere and the key nutrients they target. Crop production heavily relies on the unsustainable use of inorganic N and P fertilizers, and modern agricultural initiatives are moving toward the use of more sustainable organic sources of either N or P. The success of this strategy is dependent on having a deep understanding of the key microbial players involved in N and P cycling and the biotic and abiotic factors that control this. In this regard MEP can greatly advance our understanding of the spatiotemporal dynamics of functionally important taxa and allow us to better engineer the plant microbiome through environmental and plant genotypic selection.",
"introduction": "INTRODUCTION The plant microbiome is integral to plant health as it delivers several life support functions ( 1 , 2 ). This includes enhancing the plant’s ability to acquire both macro- and micronutrients, such as nitrogen, phosphorus, and iron, as well as enhancing plant innate immunity against a range of plant pathogens ( 2 – 4 ). Since the green revolution, intensive agricultural practices have resulted in a decoupling between microbes and their host plants ( 5 ). The breakdown of rhizobia-legume symbiosis in heavily fertilized cropping systems is perhaps the most well-known example ( 6 ). Others, such as the apparent reduction in the relative abundance of Bacteroidetes in domesticated crops relative to their wild cultivars, are more cryptic ( 7 ). Agriculture is now facing a significant global crisis: a rapidly changing climate, an ever-growing human population, and depletion of our natural resources used to fuel crop production has identified severe vulnerabilities in ensuring future food security ( 2 , 8 ). While the industrial production of nitrogen fertilizers is a highly energetic process, the production of inorganic phosphorus fertilizers is reliant on the continued supply of mined rock phosphate ( 9 ). The latter of these fertilizer production regimes is set to cause various socioeconomic and political tensions as global stocks of rock phosphate are depleted ( 9 , 10 ). Thus, there is an urgent need to develop a holistic understanding of the plant microbiome function and its numerous components ( 11 ). Through the release of signaling molecules, exudation of organic nutrients, and the decoration of plant cell walls with specific attachment molecules, plants can actively select for a subset of specialized soil microorganisms ( 12 , 13 ). This frequently involves a reduction in microbial diversity as one moves from the bulk soil > rhizosphere > root tissue ( 1 , 14 ). While bulk soil is considered a relatively carbon poor environment favoring an oligotrophic lifestyle, the rhizosphere and root system is typified by a high turnover of organic matter driven through rhizodeposition, an environment favoring a copiotrophic lifestyle. Indeed, copiotrophic bacteria related to Proteobacteria , Bacteroidetes , and Actinobacteria often dominate plant-associated microbial communities ( 15 , 16 ). While our understanding of the diversity, structure, and functional potential of microbial communities has drastically improved, there is still considerable uncertainty about how this translates into specific plant-microbe interactions, especially carbon for nutrient exchange ( 2 ). Therefore, we still lack understanding of the functional components involved in delivering beneficial plant activities within the root microbiome. Proteins are the functional entities of the cell whose regulation is controlled by surrounding biotic and abiotic conditions. Metaproteomics, the study of the entire protein content of a given environmental sample, holds enormous potential to improve our understanding regarding the function of soil microbial communities ( 17 ). Unlike its application in seawater ( 18 , 19 ), anaerobic digestors ( 20 , 21 ), or the human or animal gut ( 22 , 23 ), soil metaproteomics has been relatively underexploited ( 24 ). This is partly due to conventional soil extraction methods coextracting contaminating substances, such as organic carbon and humic acids ( 24 ). Furthermore, microbial complexity in soils is usually greater than any other environment ( 1 , 11 , 24 ), leading to considerable problems in metagenome sequencing and assembly, which are critical for quality metaproteome measurements. However, its application is increasing due to improved bioinformatics pipelines to correctly identify peptides from mass-spec data sets ( 25 ). Furthermore, the majority of expressed proteins are related to cytoplasmic housekeeping and core metabolic functions, which can often result in poor detection of more ecologically important but less abundant noncytoplasmic proteins ( 26 ). This observation is evident in our previous laboratory-based studies investigating individual bacterial responses to phosphate limitation ( 27 , 28 ). One alternative is to focus on the extracellular (exo) fractions of proteins found outside the cell using metaexoproteomics (MEP), a method that adapts extraction protocols for detecting soil extracellular enzyme activity ( 29 ). MEP has been successfully utilized to determine the active chitin-degrading community of a tropical soil in response to chitin amendment ( 29 ). While this extraction method is applicable to bulk soil analysis (requiring 50 to 100 g soil material), sampling the rhizosphere (typically 1 to 2 g material) is much more challenging. Furthermore, the method currently requires specialized equipment and is relatively low throughput. These technical limitations have likely reduced the take up of this approach, despite its enormous potential. Our recent work has successfully characterized the in vitro exoproteomes of single strain cultures related to Pseudomonas spp. and Flavobacterium spp. in response to phosphate-limiting growth conditions ( 27 , 28 ). These rhizobacteria produce numerous hydrolytic and transport proteins targeting organic phosphorus components in response to phosphate limitation. Thus, exoproteomics can generate significant insights into the mechanisms utilized by microbes to compete for growth limiting nutrients and their contribution to environmental nutrient cycling ( 26 , 30 , 31 ). In this study, we adapted our previous extraction method to efficiently capture the extracellular proteins (metaexoproteome [MEP]) found in agricultural soils to identify the most active microbial taxa in the rhizosphere of Brassica napus L. (oilseed rape) and the major metabolic interactions operating. We hypothesized that (i) the rhizosphere would contain a distinct set of metabolically active microbes relative to the surrounding bulk soil and (ii) microbes would express proteins for the mineralization of N and P as a response to elevated C. In addition to capturing extracellular plant and aphid-pest proteins in the rhizosphere, we observed a distinct shift in active fraction of the microbial community in this compartment relative to the bulk soil with several Pseudomonas spp. dominating the MEP.",
"discussion": "DISCUSSION Building a holistic understanding of plant-microbe interactions relies on the development of suitable tools to investigate complex and simultaneously occurring metabolic processes in situ . Here, first using P. putida BIRD-1 as a model and then analyzing natural field-soil microbial communities, we demonstrate that metaexoproteomic assessment of the rhizosphere is achievable and can significantly refine our understanding of the establishment and function of the plant microbiome. Specifically, these data generate further testable hypotheses surrounding the genomic basis of rhizosphere competence and microbial nutrient cycling, which will ultimately guide our ability to engineer plant microbiomes and better determine abiotic and biotic causes of plant disease. Likewise, this method successfully captured extracellular plant and corresponding pathogenic aphid proteins demonstrating the efficacy of this method to also understand plant host-pathogen interactions. Plants differ in their ability to manipulate soil communities. For example, while a strong “rhizosphere effect,” i.e., enrichment of rhizosphere-specific bacteria recruited from the surrounding bulk soil, can be observed for barley ( 16 ), other plants elicit much more subtle differences ( 15 ). A significant limitation of metagenomics is its inability to clearly identify the most active microbes and metabolic processes occurring in a specific environment, which is further compromised with the inclusion of subtle spatiotemporal parameters. In addition, while metatranscriptomics can provide information on activity, it suffers from neglecting posttranscriptional and posttranslational regulation of protein synthesis, as well as an inability to spatially resolve protein compartmentalization, i.e., intra- versus extracellular location. While our metagenomic data were consistent with the observation that oilseed rape elicits a weak rhizosphere effect on soil microbial communities ( 14 ), by utilizing MEP we observed a clear difference between the active microbial community present in the rhizosphere compared to the surrounding bulk soil. This agrees with metatranscriptomics studies investigating the active plant microbiota in various crops ( 35 ). This difference in activity can be largely attributed to an increase in the quantity of microbial and plant protein captured in the rhizosphere and elegantly demonstrates the rhizosphere as a hot spot for plant-induced microbial activity ( 1 , 2 , 36 – 38 ). Partitioning the MEP between compartments was not only achieved through the capture of significantly more bacterial protein in the rhizosphere ( Fig. 2B ) but also a shift in the taxonomic groups producing these proteins. Rhizosphere-specialized bacterial taxa can also be thought of as copiotrophs, responding to elevated labile and complex organic carbon deposition. On the other hand, oligotrophs are relatively more active in carbon-depleted bulk soil ( 34 ). The large increase in Pseudomonas activity as well other bacterial groups such as Flavobacterium ( Bacteroidetes ), and various Burkholderiales ( Betaproteobacteria ) in the rhizosphere is consistent with their predicted life history strategies ( 34 , 39 ). Likewise, our MEP data also confirmed that “bulk soil-specialized” or oligotrophic bacteria, such as those related to Verrucomicrobia , Actinobacteria , and Acidobacteria ( 39 ), were relatively more active in the surrounding soil. While our study only captured a single time point, our data clearly revealed that various distinct strains of Pseudomonas are highly active in the rhizosphere of oilseed rape and represent a major and ecologically important component of this crop’s microbiome ( 14 , 40 , 41 ). Pseudomonas represents a relatively small fraction of the seed microbiome. Thus, an increase in their relative abundance, especially several strains, in both the rhizosphere and root during the early stage of oilseed rape growth indicates active selection from the surrounding bulk soil ( 40 , 41 ). Further investigation should now focus on determining how the active fraction of the community changes over time, particularly at different growth stages. This would help determine the stability of field-grown plant microbiomes, which could ultimately be used to determine early signs of pathogen-induced dysbiosis. In addition to gaining functional insight into the plant microbiome through identification of active taxonomic groups, we also identified numerous proteins related to various beneficial functions ( 1 , 36 ). Plant root exudates shape microbial communities and amino acids can become the major group of exudates released by plants over time and, in addition to quaternary amines, can represent a significant fraction of the dissolved organic N pool ( 42 – 44 ). Furthermore, the turnover of the soluble amino acid pool in soil may be orders of magnitude greater than that of ammonium or nitrate ( 45 ). Based on our MP data collected from both our inoculated pot experiment and field trial, we discovered Pseudomonas metabolism shifts toward the turnover of amino acids and other N-containing compounds when growing in the oilseed rape rhizosphere. Given that most heterotrophic soil microbes are carbon limited, the high expression of uptake and catabolic proteins targeting amino acids and other nitrogenous carbon sources (predominantly amines) observed here suggests that microbial-mediated mineralization of ammonium may be a key process in the rhizosphere, as observed in marine systems ( 31 , 46 ). Thus, release of nitrogenous organic carbon exudates may represent a mechanism that allows plants to get an immediate return on their metabolic investment in the form of labile ammonium. This aligns with the idea that plants “prime soils” for microbial N mineralization through the exudation of organic C, stimulating expression of peptidases and proteases ( 45 ). Plant-available phosphate is often a small fraction of the total soil P content. The slow diffusion of P i in soil means that plant uptake during growth creates a zone of P i depletion around the roots (1 to 3 mm) ( 47 – 50 ), which is only intensified by increases in microbial growth on plant-derived labile organic carbon ( 1 ). In almost all bacteria, including Pseudomonas , synthesis of phosphatases and the high affinity phosphate transporter PstSABC is negatively regulated by exogenous levels of inorganic orthophosphate. Thus, these proteins serve as excellent markers to assay for phosphate depletion ( 51 – 53 ). Furthermore, elevated soil phosphatase activity has recently been shown to co-occur with the severity of P i depletion ( 54 ). While our pot experiments showed no evidence of localized P i depletion in the rhizosphere, in our field experiment the identification of five and three distinct Pseudomonas PhoX and PstS homologs, respectively, suggests rhizosphere-dwelling Pseudomonas spp. experience phosphate-limiting growth conditions, despite saturation of the soil with inorganic fertilizers. PhoD is commonly used as the major gene marker for microbial phosphatase activity ( 55 – 57 ). However, despite this family being the most abundant phosphatase in the MG for both the total community and Pseudomonas population, only PhoX was detected in the MEP, consistent with its role as the major phosphatase in plant-associated Pseudomonas ( 28 , 53 , 58 ) and other environmental Proteobacteria ( 59 , 60 ). Furthermore, stimulation of Flavobacteriia likely has importance consequences for remineralization of organic P ( 27 , 61 ). Conclusions. Here, we present the first metaexoproteomic assessment of the plant microbiome sampled from a field-grown agricultural crop. Our new technique enabled us to identify highly active taxa in the rhizosphere and the key nutrients they target. Crop production heavily relies on the unsustainable use of inorganic N and P fertilizers, and modern agricultural initiatives are moving toward the use of more sustainable organic sources of either N or P. The success of this strategy is dependent on having a deep understanding of the key microbial players involved in N and P cycling and the biotic and abiotic factors that control this. In this regard MEP can greatly advance our understanding of the spatiotemporal dynamics of functionally important taxa and allow us to better engineer the plant microbiome through environmental and plant genotypic selection."
} | 4,696 |
39563004 | PMC11850973 | pmc | 7,890 | {
"abstract": "Abstract The extensive use of nitrogen fertilizers has detrimental environmental consequences, and it is essential for society to explore sustainable alternatives. One promising avenue is engineering root nodule symbiosis, a naturally occurring process in certain plant species within the nitrogen-fixing clade, into non-leguminous crops. Advancements in single-cell transcriptomics provide unprecedented opportunities to dissect the molecular mechanisms underlying root nodule symbiosis at the cellular level. This review summarizes key findings from single-cell studies in Medicago truncatula , Lotus japonicus , and Glycine max . We highlight how these studies address fundamental questions about the development of root nodule symbiosis, including the following findings: (i) single-cell transcriptomics has revealed a conserved transcriptional program in root hair and cortical cells during rhizobial infection, suggesting a common infection pathway across legume species; (ii) characterization of determinate and indeterminate nodules using single-cell technologies supports the compartmentalization of nitrogen fixation, assimilation, and transport into distinct cell populations; (iii) single-cell transcriptomics data have enabled the identification of novel root nodule symbiosis genes and provided new approaches for prioritizing candidate genes for functional characterization; and (iv) trajectory inference and RNA velocity analyses of single-cell transcriptomics data have allowed the reconstruction of cellular lineages and dynamic transcriptional states during root nodule symbiosis.",
"conclusion": "Conclusion RNS is a complex process involving the interaction of multiple cell types and the activation of different signaling pathways. Single-cell genomics has enabled the identification of the various cell types making up a sample or tissue, the characterization of their gene expression profiles, and the reconstruction of developmental lineages for many plant species. The last few years have seen an enthusiastic embrace of single-cell genomics by the RNS research community. In this review, we covered the main themes emerging from these investigations, such as an apparent common expression program for the intracellular rhizobia infection of both root hair- and cortical-derivate cells and the compartmentalization into different cell populations of the steps required for nitrogen fixation, assimilation, and transport in both determinate and indeterminate nodules. We also reviewed how innovative approaches supported by the cell-level transcriptional information have revealed new genes involved in RNS. Those studies represent an exciting and essential step towards fully understanding the molecular mechanisms enabling RNS in species in the nitrogen-fixing clade. However, the technical challenges and costs of applying this emerging approach, in conjunction with the limited transcriptome information derived from each cell, means that we still capture a shallow representation of the gene expression changes that cells undergo as they acquire new functions in RNS, or as they proliferate to form a nodule. Technological limitations still prevent the simultaneous profiling of plant and symbiont transcripts, preventing the investigation of the intricate interplay between the two partners during symbiosis. Most urgently lacking are comparative studies of lateral root and nodule development, which could uncover the regulatory uniqueness of nodulation. Similarly, inter-specific comparisons of the transcription program associated with cellular lineages that lead to the formation of nodules with contrasting complexity could point towards the ‘easiest path’ to engineer nodules into crops. Despite these challenges, the application of single-cell genomics has undeniably revolutionized RNS research, offering an unprecedented level of detail and resolution in dissecting its molecular mechanisms. The datasets produced by the reviewed research established the foundations for more comprehensive analysis in the future. Efforts are being dedicated to the development of robust computational approaches to allow the integration of multiple samples, and those will enable the construction of an RNS single-cell atlas encompassing various developmental stages and numerous species. Similarly, spatial transcriptomics and other single-cell omics (e.g. single-cell ATAC-seq and single-cell DNA methylation profiling) are starting to be applied to RNS species. Combining those multiple layers of information will generate a better understanding of regulatory mechanisms enabling RNS. This knowledge may eventually allow the engineering of RNS in non-fixing species. The prospect of transferring nitrogen-fixing capabilities to non-leguminous crops such as cereals can revolutionize agriculture and reduce our reliance on synthetic fertilizers. This would be a significant step towards a more sustainable and environmentally friendly agriculture. To achieve this goal, it is fundamental that investments be made for the experimental validation of identified candidate genes and their predictions, including testing their effects in species outside the nitrogen-fixing clade.",
"introduction": "Introduction Nitrogen fertilization plays a crucial role in ensuring high productivity in agricultural systems. However, the extensive use of synthetic nitrogen fertilizers, stemming from the Haber–Bosch process, has disrupted the natural nitrogen cycle ( Leip et al. , 2021 ). Approximately 50–75% of the nitrogen applied to agricultural lands is not used by plants. Excess nitrogen applications lead to environmental degradation, making our current management practices unsustainable ( Sutton et al. , 2011 ; Leip et al. , 2015 ). Given the growing global population and changing climate patterns, it is imperative to address our reliance on nitrogen fertilizers to sustain food production worldwide. Despite nitrogen abundance in the atmosphere as dinitrogen (N 2 ) , plants mostly absorb available nitrogen from the soil as nitrate (NO 3 – ), ammonium (NH 4 + ), or amino acids. Certain bacteria and archaea can convert N 2 to NH 4 + in a process known as biological nitrogen fixation ( Mahmud et al. , 2020 ). These prokaryotes use a nitrogenase enzyme complex to catalyze this conversion ( Wilson and Burris, 1947 ). In general, nitrogen fixation via nitrogenase is extremely energy demanding, and the enzyme can only function when oxygen is limited. Fortunately, certain flowering plant species have evolved the ability to form a symbiotic relationship with nitrogen-fixing bacteria, enabling them to extract nitrogen from the atmosphere. Those species are all part of a monophyletic clade of angiosperms that includes four orders: Fabales , Fagales , Cucurbitales , and Rosales (FaFaCuRo), also known as the ‘nitrogen-fixing clade’ ( Soltis et al. , 1995 ; Kistner and Parniske, 2002 ; Werner et al. , 2014 ; Kates et al. , 2024 ). Nitrogen-fixing plants develop specialized root organs, known as root nodules, that are infected intracellularly by rhizobia and provide a suitable environment for efficient nitrogen fixation. Understanding nodule development and intracellular infection mechanisms in these species could offer insights into the evolutionary innovations necessary for root nodule symbiosis (RNS). The existence of RNS has inspired scientists to attempt to introduce this mechanism in crops to deliver free and sustainable nitrogen to fuel food production. If successful, engineering RNS from legumes to crops would be a route to solve the dependence on inorganic fertilizers in agriculture, and this is a highly active area of current research ( Jhu and Oldroyd, 2023 ). In the last few decades, genetics studies in species capable of RNS have led to significant advancements in our understanding of RNS, as reviewed by Roy et al. (2020) . In addition, many genomics studies have been conducted to identify genes involved in this process. In particular, RNA sequencing (RNA-seq) has revealed thousands of genes whose expression in plant roots is altered in response to infection ( Mun et al. , 2016 ; Carrere et al. , 2021 ; Almeida-Silva et al. , 2023 ). The application of RNA-seq to investigate different stages of the root response to infection has revealed critical regulators of RNS, including uncovering the role of LOB-DOMAIN PROTEIN 16 ( MtLBD16 ) in the promotion of local auxin accumulation via the regulation of STYLISH and YUCCA genes ( Schiessl et al. , 2019 ), and the function of members of the LIGHT-SENSITIVE SHORT HYPOCOTYL ( MtLSH1 and MtLSH2 ) family in establishing nodule identity ( Lee et al. , 2024 ). A limitation of RNA-seq is that the transcriptional signal is obtained in bulk, measuring the pool of mRNA from all cells in the sampled tissue. This averaged expression signal masks subtle or cell type-specific transcriptional variations that may be biologically relevant but undetectable by this technique. This is particularly problematic for RNS, where different cell types respond to infection differently and may undergo specific transcriptional reprogramming. To capture the cell type-specific signal during RNS, techniques such as laser capture microdissection (LCM) have been deployed to isolate the cell types of interest before measuring their expression profile via RNA-seq ( Roux et al. , 2014 ; Jardinaud et al. , 2016 ; Schnabel et al. , 2023 ). However, the same cell types can be in different transcriptional states that are not recognizable visually. Furthermore, when investigating RNS, only a fraction of cells from a specific cell type (i.e. epidermal root hair) respond to the infection. Therefore, the specific transcriptional regulation that enables RNS still needs to be understood, and many questions still need to be investigated. For instance, the precise requirements for a successful intracellular rhizobial infection are unknown. How the plant determines which root hairs will respond to rhizobia and allow infection threads to form is still poorly understood. In addition, it remains to be shown if those requirements are the same for the infection of the epidermal root hair cells as for the infection of cortical-derivate cells in the nodule. Other processes yet to be comprehended are the mechanisms that trigger cell division in the pericycle and cortex, which starts nodule development. Furthermore, how each cell type of the root transitions towards the cell types present in the mature nodule and what are the primary regulators of this transition are significant questions that still need to be answered if one hopes to engineer this capacity in species outside the nitrogen-fixating clade. In recent years, high-throughput single-cell platforms have enabled capture of the transcriptome state of thousands of individual cells in a single assay, resulting in single-cell RNA sequencing (scRNA-seq) rapidly becoming a favored method for investigating gene expression. However, scRNA-seq analysis of plants creates additional challenges compared with animal tissues due to the presence of cell walls, the prevalence of chloroplasts in green tissues, and secondary metabolites that may interfere with molecular biology reactions. Even so, scRNA-seq has now been successfully deployed for many plant species ( Box 1 ). In the last couple of years, single-cell transcriptomics has also increasingly been used to study the transcriptional changes required to successfully establish RNS at a cellular level ( Table 1 ). Box 1. Single-cell technologies to investigate gene expression in plants Single-cell genomics was first applied to mammalian samples for the purpose of transcription profiling heterogeneous cell populations and better understanding their progression across developmental lineages ( Tang et al. , 2009 ). Interest in using this technology in plant species emerged in 2019, propelled by the increase in throughput allowed by newly released technologies ( Fig. 1 ). Most single-cell transcriptomics studies performed in plants addressed the problem of individual cell profiling through nanoliter droplet encapsulation with techniques such as Drop-seq or 10X Genomics ( Macosko et al. , 2015 ; Cuperus, 2022 ). Fig. 1. The number of publications mentioning single-cell transcriptomics in humans (red) and in plants (blue) per year. Data obtained from Dimensions ( https://www.dimensions.ai/ ) using the following research terms for plants: (‘single-cell RNA-seq’ AND plants) OR (‘single-cell RNA sequencing’ AND plants) OR (‘scRNA-seq’ AND plants) OR (‘single-nuclei RNA-seq’ AND plants) OR (‘single-nuclei RNA sequencing’ AND plants) OR (‘snRNA-seq’ AND plants) OR (‘single-cell transcriptome’ AND plants). Results were filtered to keep only research articles. For humans, the same terms and filter were used, replacing ‘plants’ by ‘humans’. Initially, single-cell RNA-seq (scRNA-seq) and single-nucleus RNA-seq (snRNA-seq) were used to characterize cell populations in plant tissues and identify the genes that distinguish the different cell types. Many seminal studies have focused on generating cell atlases for various tissues in model organisms, including Arabidopsis thaliana ( Farmer et al. , 2021 ; Kao et al. , 2021 ; Picard et al. , 2021 ; W. Liu et al. , 2022 ; Neumann et al. , 2022 ; Procko et al. , 2022 ; Shahan et al. , 2022 ; Lee et al. , 2023 , Preprint; Nolan et al. , 2023 ), Nicotiana tabacum ( Kim et al. , 2023 ; Jin et al. , 2024 ), and Lotus japonicus ( Frank et al. , 2023 ; Z. Sun et al ., 2023 ). The same approach was applied to commercially relevant crop species, such as maize ( Bezrutczyk et al. , 2021 ; Marand et al. , 2021 ; Xu et al. , 2021 ; Li et al. , 2022 ; Sun et al. , 2022 ; Yuan et al. , 2024 ), rice ( Wang et al. , 2021 ; Zhang et al. , 2021 ; C. Li et al. , 2023 ; Zhou et al. , 2023 ; Yan et al. , 2024 , Preprint), tomato ( Tian et al. , 2020 , Preprint), and poplar ( Chen et al. , 2021 ; Li et al. , 2021 ; Conde et al. , 2022 ; Xie et al. , 2022 ; R. Li et al ., 2023 ; Tung et al. , 2023 ) or a combination of grass species (maize, sorghum, setaria) ( Ortiz-Ramírez et al. , 2021 ; Guillotin et al. , 2023 ). These studies have enabled the characterization of the transcriptome program of individual cell types. For example, the recent unveiling of the first whole-plant cell atlas offers unprecedented insight into the life cycle of A. thaliana ( Lee et al. , 2023 , Preprint). Encompassing 10 developmental stages, ranging from seed to silique, and including gene expression data from >800 000 nuclei, the study set a new standard for plant samples, approximating the scale often reported in studies of humans and other mammals. Although single-cell transcriptomics has been gradually applied to plants, the adoption of this technology is still limited compared with its use in humans ( Fig. 1 ) and animal models. Cell trajectory inference, pseudotime estimation, and RNA velocity analysis tools have added a new dimension to single-cell research by facilitating the reconstruction of cell lineages and cell dynamics ( Trapnell et al. , 2014 ; La Manno et al. , 2018 ; Saelens et al. , 2019 ; Bergen et al. , 2020 ). Recent work in this domain falls into two categories. The first includes studies using cell pseudotime inference to map developmental trajectories, such as the differentiation of stomata ( Liu et al. , 2020 ), mesophyll ( Tao et al. , 2022 ), and root ( Shahan et al. , 2022 ) cell lineages in A. thaliana . Such an approach enables identification of key cellular differentiation and development regulators. The second category comprises studies using pseudotime inference tools to unravel dynamic cellular processes in stress tolerance. For instance, pseudotime analyses were recently employed to identify cell type-specific responses to drought in A. thaliana ( Liu et al. , 2024 , Preprint), and shed light on the conversion of C 3 mesophyll cells to CAM photosynthesis under drought conditions in Mesembryanthemum crystallinum ( Perron et al. , 2024 , Preprint). Table 1. Overview of articles published using single-cell genomic analysis of root nodule symbiosis Authors Technology Species Condition Number of cells Dataset Web application or portal \n Cervantes-Pérez et al . (2022) \n snRNA-seq \n M. truncatula \n Control and 48 hpi rhizobium-inoculated roots 15 854 Ensifer meliloti -inoculated; 12 521 control GSE210881 \n https://shinycell.legumeinfo.org/medtr.A17.gnm5.ann1_6.expr.Cervantes-Perez_Thibivilliers_2022/ \n \n Ye et al . (2022) \n scRNA-seq \n M. truncatula \n 14 dpi nodules 9756 Ensifer meliloti -inoculated PRJCA012129 \n http://www.medicagowang.com/scrna/ \n \n Liu et al . (2023b) \n snRNA-seq \n M. truncatula \n Control and 30 min, 6 h, and 24 h after Nod factor treatment 25 276 total. The authors did not report the number of nuclei per time point. PRJCA011245 \n http://119.45.35.29:3571/ \n \n Pereira et al . (2024) \n scRNA-seq \n M. truncatula \n Control and 24 hpi, 48 hpi, and 96 hpi rhizobium-inoculated roots 16 211 total. The authors did not report the number of nuclei per time point. GSE224539 \n https://kirstlab.shinyapps.io/scrnaseq_medicago_truncatula/ \n \n Wang et al . (2022) \n scRNA-seq (manually separated) \n L. japonicus \n 4-week-old rhizobium-inoculated nodules 50–100 cells for each of the two cell types GSE188748 Not available. \n Frank et al . (2023) \n scRNA-seq \n L. japonicus \n Control, 5 dpi, and 10 dpi rhizobium-inoculated nodules 11 783 Mesorhizobium loti -inoculated; 13 241 control PRJEB57790 \n https://lotussinglecell.shinyapps.io/lotus_japonicus_single-cell/ \n \n Liu et al . (2023a) \n snRNA-seq \n G. max \n Control, 12 dpi, and 21 dpi rhizobium-inoculated nodules 26 712 total. The authors did not report the number of nuclei per time point. CRA007122 OMIX002290 \n https://zhailab.bio.sustech.edu.cn/single_cell_soybean \n \n B. Sun et al . (2023) \n snRNA-seq \n G. max \n Control and 14 dpi rhizobium-inoculated nodules 19 712 Sinorhizobium fredii -inoculated; 23 063 control PRJCA015369 Not available. \n Cervantes-Pérez et al . (2024) \n snRNA-seq \n G. max \n Control and 28 dpi rhizobium-inoculated nodules 14 369 control; 7830 Bradyrhizobium diazoefficiens -inoculated GSE226149 \n http://soybeancellatlas.org \n These advancements hold the potential to reveal the regulatory mechanisms governing the formation of root nodules, representing a pivotal step toward engineering this trait into crops outside of the nitrogen-fixing clade. In this review, we examine the progress enabled by the use of single-cell transcriptomics analysis in studying RNS using the model legumes Medicago truncatula and Lotus japonicus , as well as the major crop soybean ( Glycine max )."
} | 4,702 |
26477940 | PMC4634325 | pmc | 7,891 | {
"abstract": "The demand for highly scalable, low-power devices for data storage and logic operations is strongly stimulating research into resistive switching as a novel concept for future non-volatile memory devices. To meet technological requirements, it is imperative to have a set of material design rules based on fundamental material physics, but deriving such rules is proving challenging. Here, we elucidate both switching mechanism and failure mechanism in the valence-change model material SrTiO 3 , and on this basis we derive a design rule for failure-resistant devices. Spectromicroscopy reveals that the resistance change during device operation and failure is indeed caused by nanoscale oxygen migration resulting in localized valence changes between Ti 4+ and Ti 3+ . While fast reoxidation typically results in retention failure in SrTiO 3 , local phase separation within the switching filament stabilizes the retention. Mimicking this phase separation by intentionally introducing retention-stabilization layers with slow oxygen transport improves retention times considerably."
} | 270 |
28489332 | null | s2 | 7,893 | {
"abstract": "A deeper understanding of biological materials and the design principles that govern them, combined with the enabling technology of 3D printing, has given rise to the idea of \"building with biology.\" Using these materials and tools, bio-hybrid robots or bio-bots, which adaptively sense and respond to their environment, can be manufactured. Skeletal muscle bioactuators are developed to power these bio-bots, and an approach is presented to make them dynamically responsive to changing environmental loads and robustly resilient to induced damage. Specifically, since the predominant cause of skeletal muscle loss of function is mechanical damage, the underlying mechanisms of damage are investigated in vitro, and an in vivo inspired healing strategy is developed to counteract this damage. The protocol that is developed yields complete recovery of healthy tissue functionality within two days of damage, setting the stage for a more robust, resilient, and adaptive bioactuator technology than previously demonstrated. Understanding and exploiting the adaptive response behaviors inherent within biological systems in this manner is a crucial step forward in designing bio-hybrid machines that are broadly applicable to grand engineering challenges."
} | 313 |
16507154 | PMC1431735 | pmc | 7,895 | {
"abstract": "A method that combines local structure of a metabolic network with phylogenetic profiles is described and used to assign genes to orphan metabolic activities in yeast and Escherichia coli .",
"conclusion": "Conclusion We demonstrate in this work that genes encoding orphan metabolic activities can be effectively identified by integrating phylogenetic profiles with a partially known network. The reported approach is significantly more accurate in comparison to a similar method based on mRNA co-expression [ 31 ]. We are able to predict five times more correct genes as the top candidates and two times more within the top 50 candidates out of about 6,000 unrelated yeast genes. It is likely that the improvement in performance reflects larger functional coverage of the available phylogenetic profiles over mRNA co-expression data. Indeed, the performances of the algorithms based on mRNA co-expression and phylogenetic profiles are similar when only well-perturbed network neighborhoods, the neighborhoods with large changes in gene expression, are considered. The larger functional coverage of phylogenetic profiles allows our approach to be extended to organisms with no or little expression data. As we demonstrate, the optimized parameters are likely to be directly transferable between organisms. Importantly, the incompleteness of the currently available metabolic networks is not a major hindrance to the application of our algorithm. The performance of our algorithm significantly improves if the specificity of the connections established by different metabolites is taken into consideration. To account for the connection specificity, the algorithm assigns smaller cost function weights to connections established by widely used (that is, non-specific) metabolites. Similar specificity corrections should be useful for calculations based on other context-based descriptors, such as mRNA expression. Ultimately, to achieve maximal performance it will be necessary to combine various sequence-based and context-based descriptors. In Figure 6 we show how different context-based associations change as a function of the network distance between the metabolic genes. Four different context-based associations are shown: gene co-expression, gene fusions (Rosetta Stone), phylogenetic profiles, and chromosomal gene clustering (similar relationships for E. coli are shown in Additional data file 7). The figures demonstrate that different context-based associations can contribute to 'focusing' a hypothetical gene to its proper location in the network. We are currently building a combined method (P. Kharchenko, L.C., Y. Freund, D.V., G.M. Church, unpublished data) that will integrate different associations in order to predict genes responsible for orphan metabolic activities. We also plan to apply similar gap-filling methods to other cellular networks.",
"discussion": "Results and discussion The main approach As was demonstrated by us previously [ 31 , 32 ], the closer genes are in a metabolic network the more similar are the genes' evolutionary histories. It is important to know whether this relationship is strong enough to determine the exact network location of a hypothetical gene. The established distance metrics (see Materials and methods) allows us to quantify the relationship between the gene distance in the network and the average gene co-evolution (Figure 1 ). In Figure 1 we show Pearson's correlations of phylogenetic profiles between a target gene and all other network genes separated from the target by distances one, two, three, and so on. The background correlation (0.11) was estimated by averaging correlation coefficients between all non-metabolic and metabolic genes. The average correlation between metabolic genes decreases monotonically with their separation in the metabolic network, ranging between 0.29 for metabolic distance 1 and 0.13 for metabolic distance 8. This relationship suggests that we can use gene phylogenetic profiles and their location in the metabolic network to predict sequences for orphan activities. The idea behind our method is similar to that used by us previously in the context of mRNA co-expression networks [ 31 ]. We used a heuristic cost function to determine how a test gene 'fits' into a network gap. The 'fit' of a test gene in a network gap is determined by its phylogenetic correlations with network genes close to the gap. The parameters of the cost function were optimized to achieve the best predictive ability by minimizing the log sum of the ranks for all correct metabolic enzymes. Several functional forms of the cost function were tested (see Equations 1 to 3 below). Equation 1 represents a cost function similar to the one used previously [ 31 ], where x is the candidate gene, n is a gene from the network neighborhood of the gap, c ( x , n ) is the phylogenetic correlation between genes x and n , is the vector of layer weights, and p1 is the power factor for the phylogenetic correlations. The summation in Equation 1 is, first, over all genes in a given layer N i around the gap and, second, over all layers up to the layer R . Only three layers around the network gaps were used in all calculations in the paper. | N | is the total number of genes in all three layers. Equation 2 represents a cost function that takes into account the specificity of connections established by metabolites. The idea behind the connection specificity is the following: if a metabolite participates in establishing few connections (that is, the metabolite participates in a small number of reactions), the corresponding connections are given more weight in the cost function compared to connections established by widely used metabolites. The connection specificity was taken into account by an additional weight parameter ( g , n ), determined by an inverse power function of the total number of connections established by the metabolite linking the gap gene g and its neighboring gene n . If more than one metabolite establishes the connection between g and n , the most specific one (the metabolite with the fewest connections) was used. Equation 3 represents an exponential cost function, which is used to increase the sensitivity to differences between phylogenetic correlations. A set of new parameters ( β i ) was introduced to account for different weighting of the exponent in different layers. We found that the functions with connection specificity adjustment (Equations 2 and 3) significantly outperform the function without specificity adjustment (Equation 1). However, we found no difference in predictive power between Equation 2 and 3 (Additional data file 4). In the text below, unless otherwise specified, we present results obtained using Equation 2. Self-consistent test and parameter optimization To optimize the cost function parameters and assess the performance of our method we carried out a self-consistent test illustrated in Figure 2 . The test consists of: removing a known gene from its position in the network (leading to a network gap); adding the gene to a collection of 6,093 non-metabolic yeast genes; and ranking all candidate genes in terms of their 'fit' in the network gap according to the cost function. As the correct gene occupying the gap is known, we can accurately measure the performance of the method based on the obtained ranking. The overall performance of the method was quantified by calculating the fraction of correct genes that are ranked as the top, within the top 10 and within the top 50 out of all non-metabolic yeast genes. These performance measures are directly related to the main goal of our method: to suggest candidates for orphan activities to be tested experimentally. Even if our method is not always able to rank the correct gene as the top candidate, it may be useful, for example, to rank it within the top 10 candidates. These top 10 candidates can then be tested experimentally to find out the exact gene responsible for the orphan activity. The optimal values for the cost function parameters were determined by minimizing the log sum of the ranks of all known metabolic enzymes in their correct network positions (see Materials and methods). Two types of parameter optimization algorithm were used: a deterministic Nelder-Mead simplex algorithm [ 33 ] and a stochastic global optimization by simulated annealing (SA) [ 34 ]. The best performance was obtained from the SA optimizations and is reported below. The optimized prediction algorithm identifies 22.8%, 37.3% and 46.2% of the correct genes as the top candidates, within the top 10 candidates, and within the top 50 candidates out of 6,094 genes, respectively (Figure 3a ). In comparison, under random ranking, the fraction of correct genes as the top candidate, within the top 10 candidates, and within the top 50 candidates is only 0.016%, 0.16% and 0.8%, respectively. For Equation 2, optimal performance was observed with the correlation power p1 = 1.81 (95% confidence interval (CI): 1.40-2.21) and the connection specificity power p2 = 0.79 (95% CI: 0.68-0.90). As the ratio of the number of the cost function adjustable parameters to observations is around 1:100, our method does not suffer from overfitting. We achieved almost identical prediction accuracies using the training and test sets in ten-fold cross-validation (Additional data file 5). The functional information present in the currently available phylogenetic profiles allows us to significantly improve the performance in comparison to a similar method based on gene co-expression. Using mRNA co-expression, we predicted 4.1%, 12.7% and 23.8% of the correct enzyme-encoding genes to be top ranked, within the top 10, and within the top 50, respectively [ 31 ]. The improved performance reflects larger coverage of the available phylogenetic profiles, which can be calculated for many sequences in various genomes; in contrast, mRNA co-expression data are mostly available for model organisms and genes with significant mRNA expression changes. Another important improvement of the current approach is the use of the connection specificity adjustment. The specificity adjusted cost functions (Equations 2 and 3) predict 5% to 18% more correct genes within the top ranks compared to functions without specificity adjustment (Equation 1; Figure 3b ). It is interesting to investigate the relative contribution of different layers around a network gap to the cost function. As only the relative difference in layer weights impact the algorithm performance, the weight of the first layer was always set to 1. The best performance of the algorithm based on Equation 2 was achieved with the following weights for the second and third layers around the gap: w2 = 0.0085 (95% CI: 0.0051-0.0120) and w3 = 0.0024 (95% CI: 0.0011-0.0037). Smaller values for the weights w2 and w3 indicate that the phylogenetic correlations at the distances 2 and 3 from the gap are not as informative as the correlations of the first layer neighbors. But, as there are 5 and 13 times more genes in the second and third layers, respectively, their contribution to the cost function values is around 5% to 10% for the highly ranked genes and more than 10% for enzymes ranked between 200 and 600. As we show below, the contribution of the second and third layers roughly doubles for predictions on partially known networks. Performance based on phylogenetic profiles generated using COG As described in Materials and methods, BLAST searches were used in this work to calculate phylogenetic profiles. In contrast, a number of previous studies [ 27 , 35 ] relied on the Cluster of Orthologous Groups (COG) database [ 36 ] to obtain phylogenetic profiles. We investigated the performance of our algorithm on COG-based phylogenetic profiles. Using the same algorithm and the COG-based profiles, we predicted 34.1%, 56.2% and 69.0% of the correct yeast metabolic genes to be the top ranked, within the top 10 and within the top 50, respectively. This indicates an improvement of about 50% over the results based on the BLAST searches; however, this result is unlikely to indicate superior performance. First, the current coverage of the COG database is significantly biased towards genes encoding known metabolic enzymes. For example, 72% (443 out of 615) of known metabolic genes have COG profiles while only 19% (1,148 out of 6,093) of non-metabolic genes have COG profiles. This bias leads to a significant overestimation of the 'real-world' performance of the COG-based profiles. Second, the COG database has a very limited set of hypothetical proteins, making it impractical to predict hypothetical genes responsible for orphan activities using COG. Performance using hypotheticals as candidate genes In practice, it is logical to test only hypothetical genes for orphan metabolic activities in a given organism. To simulate this for the yeast metabolic network, we repeated our self-consistent test procedure using only hypothetical yeast genes as gap candidates. We identified 1,514 hypothetical yeast open reading frames (ORFs) for this analysis. As the number of hypothetical genes is smaller than the total number of genes (usually 30% to 70% smaller), the performance of our method should improve. Indeed, testing only hypothetical genes improved the algorithm performance: 30.4%, 48.0% and 57.1% correct enzymes were ranked as the top 1, within the top 10 and within the top 50 among all candidate sequences, respectively (Figure 3c ). We note that the observed 25% improvement in performance is not due to a better discrimination against hypothetical genes. Similar improvement was observed when a candidate set of 1,514 randomly selected genes with known functions was used (Additional data file 6). Performance on the E. coli metabolic network To understand the transferability of our approach to other organisms, we repeated our analysis using the E. coli metabolic network. The same procedures were used to construct the metabolic network for E. coli (see Materials and methods). First, the optimal parameters obtained for the S. cerevisiae metabolic network, without further modifications, were applied to rank E. coli metabolic genes. As a result, the algorithm predicts 13.3%, 30.0%, and 41.3.% of known E. coli metabolic genes to be top ranked, within the top 10 and within the top 50, respectively, out of 3,578 non-metabolic E. coli genes. Second, the simulated annealing optimization was performed to optimize the cost function specifically for the E. coli network. Based on the optimized parameters slightly better results were obtained: 18.0%, 33.8%, and 45.6% of the correct genes were ranked as the top candidate, within the top 10, and within the top 50, respectively (Figure 3d ). The optimal E. coli parameters for the cost function are generally similar to the optimal parameters for the S. cerevisiae metabolic network. This suggests that parameters obtained on several model organisms can be directly used for predictions in other organisms, although an organism-specific optimization will slightly improve the algorithm performance. Performance based on genes without independent homology information Our prediction method is designed primarily for enzymatic activities without good homology information. Above, we validated the approach using all known metabolic enzymes from E. coli and S. cerevisiae . In addition, it is interesting to identify a set of enzymes for which independent homology information is not available (that is, the biochemical experiments have been conducted only in E. coli , for example) and test the performance on this subset. We obtained a subset of E. coli enzymatic EC numbers without representative sequences in other organisms. The subset, identified using the SWISS-PROT database [ 37 ], includes EC numbers with representative sequences exclusively from E. coli . We also included EC numbers with representative sequences in the TrEMBL database (a computer-annotated complement to the SWISS-PROT), but only if these were computationally annotated from E. coli sequences and, consequently, cannot provide independent homology information. Each identified EC number was then manually checked. The identified subset consists of 25 enzymes and is listed in Table 1 . The performance of our method on the subset was comparable to the performance observed for the set of all E. coli enzymes: 16.0%, 24.0% and 44.0% of the correct enzymes were ranked as the top, within the top 10, and within the top 50, respectively, among all E. coli candidate genes. Consequently, the algorithm is effective for sequences that are likely to be missed by homology-based methods. Importance of the neighborhood The performance of our algorithm for a specific network gap should crucially depend on the available evolutionary information for network genes located around the gap. As we optimized our algorithm we found that for about one-third of all gaps the algorithm performance is no better than random. To investigate this further, we calculated the discrimination ratio of the cost function value for the correct gene and the average for all non-metabolic genes. The distribution of the discrimination ratios for all possible gaps in the metabolic network is shown in Figure 4a . Confirming our expectation, about one-third of all gaps did not allow any discrimination between the correct and average genes (bin 0 in Figure 4a represents gaps with discrimination ratios less than 1). On the other hand, about 50% of the gaps have discrimination ratios equal or greater than 7 (bin >= 7 in Figure 4a ). For comparison, the average rank of the correct genes for the gaps in bin 0 is only 1,989, while it is 26 for the gaps in bin >= 7. We found that an important feature that separates the informative and non-informative gaps is the availability of accurate phylogenetic correlations for the neighborhood genes around the gaps. Clearly, if accurate phylogenetic correlations cannot be calculated - because, for example, the corresponding genes exist only in several related genomes - the cost function will not be able to discriminate between correct and incorrect genes. Figure 4b illustrates this point by showing the relationship between the average phylogenetic correlation between the first layer genes and the fraction of well-predicted gaps. For gaps with a first layer correlation of at least 0.5, 95% of the correct genes are ranked within the top 50. In contrast, less than 20% of the correct genes are ranked within the top 50 if the average first layer correlation is below 0.1. In practice, the discrimination ratio can be used to estimate the predictive ability of different gaps. Performance based on a partially known networks Currently available metabolic networks are significantly incomplete. As our algorithm directly relies on the network structure, it is important to understand that the algorithm performance depends on the network completeness. To investigate this we deliberately removed a certain fraction of known genes from the yeast network and retrained our algorithm on the incomplete network. We tried two approaches to simulate incomplete networks. First, we completely deleted a fraction of genes from the network and removed all connections to the deleted genes. Second, we effectively converted a fraction of the metabolic network into orphan activities. In this case the connections established by the orphan activities are preserved, but the genes responsible for these activities are converted into orphan activities. These two deletion approaches gave similar results and we report here only the effects of complete gene deletions. As Figure 5 demonstrates, the performance of our method decreases only gradually when increasing fractions of network genes are deleted. Even when as many as 50% of the network genes are deleted, the algorithm still performs reasonably well, predicting 13.7% as the top candidate (95% CI: 10.5-15.6%), 27.9% to be within the top 10 (95% CI: 24.2-31.5%), and 33.1% within the top 50 (95% CI: 29.2-37.1%). Interestingly, when a high percentage (20% to 50%) of the network was deleted, the relative cost function contributions from genes of the second and third layers around gaps increased approximately twice. This suggests that, for an incomplete network, the second and third layers play a larger role in 'focusing' a correct gene towards the corresponding gap. The relative insensitivity of our method to the network completeness suggests that the algorithm based on phylogenetic profiles will be useful not only for metabolic networks of model organisms, such as S. cerevisiae and E. coli , but also for networks of less studied organisms. Predictions for orphan activities in S. cerevisiae and E. coli As the metabolic networks of E. coli and S. cerevisiae are relatively well studied, it is likely that the developed algorithm will be most useful in less studied species with a larger fraction of orphan metabolic activities. Nevertheless, we investigated in detail several predictions for orphan activities in the E. coli and S. cerevisiae networks. Although considered as gaps in the originally reconstructed E. coli [ 10 ] and S. cerevisiae networks [ 11 ], a number of orphan activities have been recently identified. For example, the yeast enzyme 5-formyltetrahydrofolate cyclo-ligase (EC 6.3.3.2) appears as a gap in the network model by Forster et al . [ 11 ]. However, the gene responsible for this activity, YER183C/FAU1, has been cloned and characterized by Holmes and Appling [ 38 ]. This gene is present in the updated model by Duarte et al . [ 39 ]. In the E. coli iJR904 model, the arabinose-5-phosphate isomerase (API, EC 5.3.1.13) is listed as an orphan activity. However, the yrbH /b3197 gene has been recently characterized as encoding the enzyme responsible for this metabolic reaction [ 40 ]. Significantly, without any sequence homology information, our algorithm was able to rank the S. cerevisiae FAU1 gene and the E. coli yrbH gene as the number 10 and number 1 candidate, respectively, for their corresponding enzymatic activities. More examples for recently identified orphan activities and predictions can be found in Additional file 9. Several orphan activities in S. cerevisiae and E. coli remain unassigned to any gene. We found several interesting predictions for the NAD+ dependent succinate-semialdehyde dehydrogenase (EC 1.2.1.24) in E. coli . E. coli seems to possess two different types of succinate semialdehyde dehydrogenases [ 41 ]: one is NAD(P)+ dependent and is encoded by the b2661/ gabD gene (EC 1.2.1.16); the other is specific for NAD+ only (EC 1.2.1.24). One E. coli gene, b1525/ yneI , was predicted as the top candidate for this orphan activity. We believe yneI is a good candidate for the orphan activity because of the following additional functional clues. It has 32% sequence identity (E-value 5*10 -61 ) to the other E. coli succinate semialdehyde dehydrogenase encoded by gabD and 30% sequence identity to the human enzyme ALDH5A1 (EC 1.2.1.24, E-value 7*10 -59 ). In addition, yneI is adjacent on the bacterial chromosome to the gene yneH / glsA2 /b3512, which encodes glutaminase 2 (EC 3.5.1.2). The gene yneH is involved in the same glutamate metabolism pathway as EC 1.2.1.24. The closeness of yneI and yneH on the chromosome suggests that they are involved in related functions."
} | 5,880 |
38812916 | PMC11133584 | pmc | 7,896 | {
"abstract": "Acetogens are among the key microorganisms involved in the bioproduction of commodity chemicals from diverse carbon resources, such as biomass and waste gas. Thermophilic acetogens are particularly attractive because fermentation at higher temperatures offers multiple advantages. However, the main target product is acetic acid. Therefore, it is necessary to reshape metabolism using genetic engineering to produce the desired chemicals with varied carbon lengths. Although such metabolic engineering has been hampered by the difficulty involved in genetic modification, a model thermophilic acetogen, M. thermoacetica ATCC 39073, is the case with a few successful cases of C2 and C3 compound production, other than acetate. This brief report attempts to expand the product spectrum to include C4 compounds by using strain Y72 of Moorella thermoacetica . Strain Y72 is a strain related to the type strain ATCC 39073 and has been reported to have a less stringent restriction-modification system, which could alleviate the cumbersome transformation process. A simplified procedure successfully introduced a key enzyme for acetoin (a C4 chemical) production, and the resulting strains produced acetoin from sugars and gaseous substrates. The culture profile revealed varied acetoin yields depending on the type of substrate and culture conditions, implying the need for further engineering in the future. Thus, the use of a user-friendly chassis could benefit the genetic engineering of M. thermoacetica .",
"conclusion": "5 Conclusion Acetoin production by a thermophilic acetogen was demonstrated by successful genetic engineering of M. thermoacetica Y72 using a simplified procedure. The engineered strains produced acetoin from diverse carbon sources, and the culture profiles indicated metabolic bottlenecks that warrant further investigation. Notably, the acetoin productivity needs to be far more enhanced for economically viable production. Nevertheless, this study is valued in a concrete progress for the development of a microbial chassis of thermophilic acetogens for the carbon conversion, such as sugars and carbon-rich gases.",
"introduction": "1 Introduction One of the global challenges in this modern era is establishing a sustainable society free from fossil resource-dependent industries. A key factor is the utilization of sustainable resources, such as biomass, and recycling of used materials. In this context, one of the promising approaches will be the biological conversion of these resources to conventional chemicals. Acetogens are a group of microorganisms with unique and versatile metabolic pathways that catabolize both biomass-derived sugars and gaseous substrates to produce mainly acetic acid. Near stoichiometric conversion of hexose sugar to acetic acid (one hexose molecule is converted to three acetic acid molecules) is possible, inferring great metabolic potential ( Drake and Daniel, 2004 ). Moreover, the gaseous substrates include hydrogen (H 2 ), carbon monoxide (CO), and carbon dioxides (CO 2 ), which can be derived not only from waste gases but also from the gasification of organic materials, including recalcitrant resources, such as lignin from biomass. Therefore, acetogens are promising candidates for biological catalysts that produce conventional chemicals by utilizing sustainable resources. In particular, thermophilic acetogens have multiple advantages over others, such as low risk of contamination, lower cooling cost, faster reaction speed, and simplified product recovery ( Taylor et al., 2009 ; Abdel-Banat et al., 2010 ; Redl et al., 2017 ; Kato et al., 2021 ). For the practical application of thermophilic acetogens, it is essential to produce chemicals that are more diverse than acetate. One method is to screen for new strains with the metabolic capacity to produce these chemicals. Another method is genetic engineering of thermophilic acetogens to introduce new metabolic pathways and tune innate pathways. However, acetogens are known for their inaccessibility in genetic modification ( Bourgade et al., 2021 ). To date, the most successful cases of metabolic engineering to produce other chemicals have been achieved using the model thermophilic acetogen Moorella thermoacetica ATCC 39073. These cases are limited to the production of C2 and C3 compounds, such as ethanol (C2), lactate, acetone, and isopropanol (C3) ( Iwasaki et al., 2017 ; Rahayu et al., 2017 ; Kato et al., 2021 ; Kato et al., 2024 ). Metabolic engineering has been accomplished by overcoming genetic barriers via DNA premethylation ( Kita et al., 2013a ). However, the method is cumbersome, enabling only a small set of engineered strains obtained. Previously, another strain of M. thermoacetica , Y72, was reported to be genetically more accessible than ATCC 39073, because of its less strict restriction-modification system ( Kita et al., 2013b ). The genome is closely related to ATCC 39073, according to the draft genome sequence, with a similar size and the same GC content (56%) ( Tsukahara et al., 2014 ). Phylogenetic analysis revealed that Y72 belongs to the same clade as ATCC 39073, and an average nucleotide analysis further supported the similarity of the two genomes ( Redl et al., 2019 ). Despite this similarity, strain Y72 possesses fewer genes for the restriction-modification system than ATCC 39073, which is assumed to be the reason why Y72 shows high transformation efficiency ( Tsukahara et al., 2014 ). Therefore, Y72 may be useful as an alternative strain for the model ATCC 39073 for metabolic engineering in introducing new pathways. Acetoin (3-hydroxy-2-butanone) is an important C4 chemical in the food, medical, and chemical industries and was selected by the U.S. Department of Energy (DOE) as one of the 30 compounds that require prioritized production from biomass sugars and syngas ( Werpy et al., 2004 ). However, genetic engineering of thermophilic acetogens for acetoin production has not yet been reported. Moreover, no studies have reported the production of C4 chemicals using thermophilic acetogens. In this study, metabolic engineering for the C4 chemical production by a thermophilic acetogen was attempted by taking advantage of the strain M. thermoacetica Y72, with acetoin as the target product. Furthermore, engineered strains were evaluated as the place to study for chassis development.\n\n3.2 Free of customized DNA modification for the introduction of bsALDC-encoding gene into M. thermoacetica Y72 To heterologously express bsALDC in M. thermoacetica Y72, the coding nucleotide sequence was codon-optimized and placed under the constitutive glycerol-3-phosphate dehydrogenase (G3PD) promoter. The plasmid DNA construct was designed to introduce the promoter-gene set (P G3PD - bsALDC ), either at the pyrF or pduL2 locus of the chromosome, by homologous recombination. The P G3PD - bsALDC was placed in homologous regions upstream and downstream of pyrF and pduL2 ( Figure 2A ). pyrF gene was used as a selection marker to transform the pyrF -knockout strain, which is uracil auxotroph ( Kita et al., 2013b ). When the P G3PD - bsALDC was introduced into the pyrF locus, the acetate production pathway remained intact. When the P G3PD - bsALDC was introduced into the pduL2 locus, acetate production was lowered due to the disruption of one of the two genes encoding phosphoacetyl transferase. FIGURE 2 DNA construction for introducing bsALDC into the chromosome of Moorella thermoacetica Y72 by homologous recombination. (A) shows the schematic representation of the constructed plasmids and the homologous recombination events in the pyrF or pduL2 region. (B) shows the agarose gel electrophoresis following PCR amplification of the pyrF or pduL2 region. The size shifts of the amplified DNA confirmed the successful introduction of the bsALDC constructs in the pyrF or pduL2 region. 1.9 kb of the pyrF region was shifted to 2.7, and 0.9 kb of the pduL2 region was shifted to 2.3 kb. M, DNA size marker; lane 1, the Y72-Km strain; lane 2, the Y72- pyrF :: bsALDC strain; lane 3, the Y72- pduL2 :: bsALDC strain. The resulting plasmids were introduced into M. thermoacetica Y72 without any customized DNA modifications. Despite skipping the modification step, the target strains were successfully obtained ( Figure 2B ).",
"discussion": "4 Discussion This brief report describes the metabolic conversion of a thermophilic homoacetogen, M. thermoacetica Y72, using genetic engineering. The strain originally produced only acetate ( Kita et al., 2013b ), and the successful introduction of a key gene conferred the acetoin production. This was the first reported case of C4 chemical production by a thermophilic acetogen. Strain Y72 has been reported to be less stringent in the restriction-modification system, providing high transformation efficiency. A simplified procedure without customized DNA premodification may be helpful in accelerating the metabolic engineering of thermophilic acetogens. There are a few reports concerning acetoin production by thermophiles, but not by thermophilic acetogens that utilize gaseous substrates ( Xiao et al., 2012 ; Sheng et al., 2023 ). Two strains for acetoin production were constructed by tuning carbon flux. Based on the design of the metabolic pathways ( Figure 1 ), certain levels of acetate production are needed to maintain metabolism in terms of the metabolic redox state and energy supply. The Y72- pduL2 :: bsALDC strain showed higher acetoin production than the Y72- pyrF :: bsALDC strain, demonstrating that knockout of a gene in the acetate pathway was effective in improving acetoin production. Varying the fermentation temperature and sugar substrate (hexose to pentose) maintained acetoin production itself; however, these changes significantly affected the yield. The differences could be derived from the responsible enzymatic activities, levels of gene expression, or metabolic redox state, in addition to the original metabolic features of M. thermoacetica Y72. For example, both key enzymatic activities, PduL1 for acetate and bsALDC for acetoin, were higher at 60°C than at 55°C in vitro ( Breitkopf et al., 2016 ; Jia et al., 2017 ). Furthermore, PduL1 activity was more enhanced to a greater extent by a temperature shift ( Breitkopf et al., 2016 ). These differences may have contributed to different culture profiles at different temperatures. In addition, it has been reported that the NAD(P)/NAD(P)H ratio affects distribution of the metabolic flux during acetoin and byproducts ( Bao et al., 2015 ). The difference due to sugar type could be derived from this redox state ( Bao et al., 2015 ). Some of these factors may affect the consequences of competing chemical productions. In future studies, the involvement of these multiple factors should be investigated to optimize the acetoin production from sugars. Acetoin production from gaseous substrates has been demonstrated using syngas and CO 2 +H 2 gas. In both cases, small amounts of acetoin were produced. One reason for this low production may be the utilization of pyruvate as a key intermediate. Whereas the conversion of acetyl-CoA to acetate provides ATP, the conversion of acetyl-CoA to pyruvate requires an energy source ( Figure 1B ). Because the energy provision from gaseous substrates is low ( Schuchmann and Muller, 2014 ; Debabov, 2021 ), cellular growth or anabolic metabolism in acetogens is limited; hence, pyruvate production is limited. Energy shortage was evident in the culture profile, especially when H 2 was solely provided as the energy source, with inactive cell growth and excess formate accumulation. A drastic strategy is necessary for higher acetoin production through enhanced pyruvate production accompanied by sufficient energy in gaseous substrates. However, the detectable level of acetoin indicated that the redox state of metabolism was not as heavily impaired as that of the ethanol-producing Mt-Δ pduL2 :: aldh strain ( Takemura et al., 2021 ). The Mt-Δ pduL2 :: aldh strain had a pduL2 knockout and enhanced the expression of aldehyde dehydrogenase for ethanol production. However, it only produced formate and acetate in CO 2 +H 2 gas."
} | 3,057 |
27833653 | PMC5101670 | pmc | 7,899 | {
"abstract": "Background Due to the unsustainable consumption of fossil resources, great efforts have been made to convert lignocellulose into bioethanol and commodity organic compounds through biological methods. The conversion of cellulose is impeded by the compactness of plant cell wall matrix and crystalline structure of the native cellulose. Therefore, appropriate pretreatment and even post-treatment are indispensable to overcome this problem. Additionally, an adequate utilization of coproduct lignin will be important for improving the economic viability of modern biorefinery industries. Results The effectiveness of moderate alkaline ethanol post-treatment on the bioconversion efficiency of cellulose in the acid-steam-exploded corn stover was investigated in this study. Results showed that an increase of the alcoholic sodium hydroxide (NaOH) concentration from 0.05 to 4% led to a decrease in the lignin content in the post-treated samples from 32.8 to 10.7%, while the cellulose digestibility consequently increased. The cellulose conversion of the 4% alcoholic NaOH integrally treated corn stover reached up to 99.3% after 72 h, which was significantly higher than that of the acid steam exploded corn stover without post-treatment (57.3%). In addition to the decrease in lignin content, an expansion of cellulose I lattice induced by the 4% alcoholic NaOH post-treatment played a significant role in promoting the enzymatic hydrolysis of corn stover. More importantly, the lignin fraction (AL) released during the 4% alcoholic NaOH post-treatment and the lignin-rich residue (EHR) remained after the enzymatic hydrolysis of the 4% alcoholic NaOH post-treated acid-steam-exploded corn stover were employed to synthesize lignin-phenol-formaldehyde (LPF) resins. The plywoods prepared with the resins exhibit satisfactory performances. Conclusions An alkaline ethanol system with an appropriate NaOH concentration could improve the removal of lignin and modification of the crystalline structure of cellulose in acid-steam-exploded corn stover, and consequently significantly improve the conversion of cellulose through enzymatic hydrolysis for biofuel production. The lignin fractions obtained as byproducts could be applied in high performance LPF resin preparation. The proposed model for the integral valorization of corn stover in this study is worth of popularization.",
"conclusion": "Conclusions An alkaline ethanol NaOH post-treatment under mild conditions was confirmed to be a very effective delignification method, and could effectively promote the cellulose digestibility of the acid-steam-exploded corn stover. It was found that the cellulose digestibility of the acid-steam-exploded corn stover post-treated with 4% alcoholic NaOH reached a maximum of 99.3%, although 10.7% lignin still remained. Except the lignin content of the acid-steam-exploded corn stover, an expansion of cellulose I lattice caused by the alkaline ethanol post-treatment also remarkably contributed to the enzymatic hydrolysis of cellulose. The lignin fractions (AL and EHR) obtained from 4% alcoholic NaOH post-treatment and subsequent enzymatic hydrolysis processes could be converted to high performance LPF resin adhesives, and the plywoods prepared with both LPF adhesives could meet the strength requirements of exterior plywood and limitation value of formaldehyde emissions for E 0 -grade panels. The proposed model for integral valorization of corn stover in this study has demonstrated a good application prospects and promotion of values.",
"discussion": "Results and discussion Compositional analysis The acid-steam-exploded corn stover was post-treated with alkaline ethanol with different NaOH concentrations, and an acid sodium chlorite post-treated acid-steam-exploded corn stover was also obtained as a comparison (an extensively delignified sample). In Table 1 , all the samples have been identified by the different post-treatment methods used. The compositions of untreated corn stover, acid-steam-exploded corn stover, the post-treated samples, and the solid yields of each post-treated samples are given in Table 1 . As compared with the composition of untreated corn stover (36.5% cellulose, 22.1% hemicelluloses, and 18.8% lignin), the ratio of relative content of cellulose to lignin slightly decreased in the acid-steam-exploded materials (Sample 1) due to the significant degradation of hemicellulose during the steam-explosion process. In addition, this phenomenon also implied that the compact matrix of plant cell wall has been significantly modified and some of the cellulose has been degraded during the ASE treatment. The accessibility of cellulase to cellulose could be significantly improved, and the conversion rate of cellulose should also increase [ 25 ]. However, as shown in Table 1 , lignin, which limits the digestibility of lignocelluloses, remains in the acid-steam-exposed corn stover in a relative content of 32.8%. In the integrated corn stovers treatment, the ratio of cellulose to lignin increases dramatically as the NaOH concentration increases. Generally, the gradual removal of lignin was accompanied with the increase of conversion of cellulose [ 26 ]. At 4% concentration of NaOH (Sample 4), the relative content of cellulose reached up to 84.0% with only 10.7% lignin remained in the post-treated sample. To this effect, alkaline ethanol post-treatment under the mild conditions was proved to be a very effective delignification method. Sample 5, which was post-treated with acid sodium chlorite, had the lowest lignin content (3.0%) and highest cellulose content (89.2%) among the five samples. Such treatment may result in the highest cellulose conversion rate in this study, yet other factors may also contribute to the cellulose conversion result, and they need to be considered. Table 1 Compositional analysis of the untreated, acid-steam-exploded and integrally treated corn stovers Samples Treatment methods Solid yields Cellulose Hemicelluloses Lignin Others Corn stover Untreated – 36.5 (2.1) 22.1 (0.9) 18.8 (1.7) 22.6 (1.3) 1 Acid steam explosion – 55.6 (3.5) a \n 5.7 (0.7) 32.8 (2.6) 5.9 (0.9) 2 Acid steam explosion followed by a 0.05% alcoholic NaOH post-treatment 78.1 71.1 (2.8) 6.4 (0.6) 19.1 (3.4) 3.4 (0.6) 3 Acid steam explosion followed by a 0.5% alcoholic NaOH post-treatment 70.3 78.9 (3.3) 4.2 (0.6) 15.9 (3.2) 1.0 (0.1) 4 Acid steam explosion followed by a 4% alcoholic NaOH post-treatment 65.4 84.0 (4.4) 4.1 (0.7) 10.7 (1.7) 1.2 (0.8) 5 Acid steam explosion followed by a NaClO 2 post-treatment 59.6 89.2 (2.5) 7.5 (1.1) 3.0 (0.4) 0.3 (0.1) \n a The value in parenthesis is standard deviation \n The relative proportions of hemicelluloses of the five samples listed in Table 1 (4.1–7.5%) did not vary much from one another as compared to the changes in cellulose and lignin. Therefore, the influence of hemicelluloses on sugar yield can be ignored in this work. It should be mentioned that only xylose in hemicelluloses was lost during the post-treatment process. A series of alkaline ethanol with different NaOH concentrations were processed to obtain various degrees of delignification, but only three are displayed in Table 1 . It should be noted that lignin content presented a gradient decrease (32.8–10.7%) within the four post-treated samples (Sample 1–4) which may be an ideal series to investigate the effects of lignin content on the enzymatic hydrolysis of cellulose in the subsequent studies. FT-IR spectra The FT-IR spectra of the acid-steam-exploded corn stover and the post-treated samples are shown in Fig. 1 . The bands at 1597, 1510, and 1429 cm −1 are corresponding to aromatic skeletal vibrations, and C–H deformation combined with aromatic ring vibration was found at 1451 cm −1 [ 27 , 28 ]. In Samples 1–4, the four characteristic absorption peaks of lignin became gradually weaker as the NaOH concentration increased. These four peaks almost disappeared in Sample 5, which was post-treated with acid sodium chlorite. Correspondingly, gradual increase of intensities of the bands at 1198 and 1156 cm −1 corresponding to O–H stretching and C–O–C vibrations at β -glucosidic linkages in cellulose were observed. These phenomena altogether indicated that the relative contents of cellulose in Samples 1–5 increased due to the increasing removal of lignin. The FT-IR results obtained above were all in reasonable agreement with the compositional analysis (Table 1 ). Additionally, a new peak at 1731 cm −1 , which is assigned to non-conjugated carbonyl groups [ 29 ], appeared in the spectrum of Sample 5, which was probably that the oxidation of the cellulose occurred during the process of acid sodium chloride delignification. Fig. 1 FT-IR spectra of the acid-steam-exploded and the integrally treated corn stovers \n CP/MAS 13 C-NMR spectra To further investigate the structural characteristics of the acid-steam-exploded corn stover and the post-treated corn stovers, all the samples were subjected to CP/MAS 13 C-NMR spectroscopy analysis. The corresponding spectra are depicted in Fig. 2 . The region between 55 and 110 ppm was dominated by strong signals assigned mostly to various cellulosic carbons, and the sharp signal at 104.7 ppm is attributed to the ordered cellulose C-1. The signals at 68–80 ppm are due to cellulose C-2, C-3, and C-5 [ 30 , 31 ]. The relative intensity of the signal at 56.0 ppm, which corresponds to the methoxyl groups (–OCH 3 ) of aromatic moieties in lignin, decreased as the NaOH concentration increased until nearly disappearing entirely in the spectrum of Sample 5. Analogous results have been reflected in the compositional and FT-IR analyses discussed above. Fig. 2 CP/MAS 13 C-NMR spectra of the acid-steam-exploded and the integrally treated corn stovers \n Obvious signals for amorphous and crystalline carbons of cellulose could be detected. The chemical shifts at 88.2 and 63.9 ppm are ascribed to crystalline cellulose C-4 and C-6 carbons, respectively. The signal at 83.7 ppm is due to amorphous cellulose C-4 carbons, and the signal at 62.2 ppm is attributed to amorphous cellulose C-6 [ 32 ]. In Sample 4, which was delignified with 4% alcoholic NaOH, the intensity of the signal at 88.2 ppm decreased and increased at 83.7 ppm, respectively. The peak at 63.9 ppm seemingly disappeared and shifted to 62.2 ppm in Sample 4, but the peaks at 63.9 and 62.2 ppm coexisted in Samples 1–3 and 5. This indicated that part of the cellulose structure of Sample 4 transformed from the original crystalline cellulose into amorphous cellulose. The amorphous cellulose is more vulnerable to cellulose enzymes and thus favors glucan conversion [ 33 ]. It should be noted that the signal at 63.9 ppm in the spectrum of Sample 4 should not be completely disappeared. The intensity of this signal in Sample 4 was weak and overlapped with other signal due to the poor resolution of the NMR method used. In Sample 5, both of the crystalline signals 88.2 and 63.9 ppm have not shifted. It seems that the crystalline domain of cellulose was not disrupted by acid sodium chlorite post-treatment. XRD analysis Cellulose crystal structure is considered one of the major substrate properties influencing biomass enzymatic digestibility [ 33 ]. In this study, the acid-steam-exploded corn stover and the post-treated samples were examined by XRD to gain insight into the potential structural features affecting cellulose hydrolysis (Fig. 3 ). The crystallinity index ( CrI ) values of Samples 1–5 were calculated to be 47.6, 48.9, 50.7, 42.4, and 52.8%, respectively. The slight increase of CrI in Samples 1–3 was due to the gradual removal of amorphous lignin and hemicelluloses with alkaline ethanol. While for Sample 5, it was due to the removal of lignin with acid sodium chlorite. All these four samples exhibited typical diffraction patterns of cellulose I, the main peak of which lies near 22.5° and the secondary broad peak at ~16.0°. In fact, the broad peak at ~16.0° should be consisted of two small peaks at 15.2° and 16.8°, but the XRD analysis conditions used make it hard to be distinguished as reported by Trache et al. [ 34 ]. In Sample 4, conversely, the shape of diffractogram and trend of CrI differed notably as compared to other samples. The main peak around 22.5° was weakened and shifted slightly to a lower angle, and there was no peak at 12.1° (one of the typical diffraction patterns of cellulose II). These results confirmed that the cellulose in Sample 4 was in a transition status from cellulose I to cellulose II, and that an expansion of cellulose I lattice occurred. It has been reported that this benefited the conversion of cellulose to glucose [ 35 ]. It is important to note, however, that 4% aqueous NaOH post-treatment alone under similar conditions would not lead to the same results [ 35 ]. It may be due to that the solubility of lignin in 4% alcoholic NaOH was higher than that in 4% aqueous NaOH. Thus, the exposed surface area of cellulose in 4% alcoholic NaOH was more than that in 4% aqueous NaOH, and the cellulose I lattice was easy to be expanded under such moderate condition. Fig. 3 X-ray diffractograms of the acid-steam-exploded and the integrally treated corn stovers \n Enzymatic hydrolysis To evaluate the effects of different post-treatment conditions on glucose yield, the enzymatic hydrolyses of the acid-steam-exploded corn stover and the post-treated samples were comparatively studied as shown in Fig. 4 . Cellulose digestibility was 57.3% for the only acid-steam-exploded material (Sample 1) within 72 h enzymatic hydrolysis, and for the post-treated Samples 2, 3 and 5, the glucose yield was 64.3, 76.7, and 92.7%, respectively. The corresponding lignin contents of Samples 1–3 and 5 were 32.8, 19.1, 15.9, and 3.0%, respectively. These results indicated that the efficiency of enzymatic hydrolysis was significantly affected by the remaining lignin content, and it increased as the lignin content decreased. In general, cellulose conversion is known to decrease as the CrI of cellulose increases, and it has been already reported that the CrI is related to cellulose content or alfa-cellulose [ 31 ]. However, this phenomenon was not observed in the data of Samples 1–3 or 5 in this study. This indicated that the content of lignin in the lignocelluloses is also a very important factor affecting the rate of enzymatic hydrolysis except for the crystallinity of the material. As the lignin content decreased in the sample, the rate of the enzymatic hydrolysis increased. In addition, the results of compositional analysis indicated that the cellulose in the acid-steam-exploded corn stovers was not obviously removed during alkaline ethanol post-treatment. Most of the alfa-cellulose remained in the post-treated samples, and its relative content increased as a result of the removal of amorphous components (lignin and hemicelluloses). Only a slight removal of non-crystalline cellulose may be occurred during the process. Fig. 4 Enzymatic hydrolysis of the acid-steam-exploded and the integrally treated corn stovers \n It was interesting that the cellulose digestibility of Sample 4 reached a maximum of 99.3% after 72 h, higher than that (92.7%) of Sample 5 despite its relatively high lignin content (10.7%). XRD analysis showed that the CrI of Sample 4 was 42.4% and the cellulose was in a transition status from cellulose I to cellulose II, which implied that an apparent change in CrI (especially cellulose lattice transformation) had an important impact on the enzymatic hydrolysis of cellulose in addition to the lignin content. It has been reported that along with the transformation of cellulose I to cellulose II [ 36 ], the expansion of cellulose I lattice promotes the enzymatic hydrolysis of cellulose [ 35 , 37 , 38 ]. Besides, it was found that Sample 4 had a higher enzymatic hydrolysis rate than those of other samples after 24 h. It was hypothesized that lignin was the main obstacle to cellulose degradation within the first 24 h when the degraded substrate was mainly amorphous cellulose. This is the reason why the initial rate of Sample 4 was lower than that of Sample 5 although more amorphous fractions were found in Sample 4. In fact, lignin is also a main factor for the increase of the material recalcitrance and decrease the enzymatic hydrolysis rate [ 39 ]. After 24 h of saccharification, the crystalline structure of cellulose in turn became the main obstacle to cellulose hydrolysis. The amorphous structure of the cellulose or cellulose II in Sample 4 made it more accessible to cellulase than those of other samples. Thus, the enzymatic hydrolysis rate of Sample 4 was higher than those of other samples after 24 h, and the glucose yield of Sample 4 reached up to 99.3% in a relatively short period of hydrolysis. Overall, an alkaline ethanol system with appropriate NaOH concentration that is able to effectively remove lignin and change the crystalline structure of cellulose under moderate conditions represents a promising approach to modern biorefinery. Characterization of lignins The characterization of lignins obtained could facilitate the understanding of treatment process and reasonable utilization of the byproducts. It could be found that approximately 60% of lignin in the acid-steam-exploded corn stover was removed during 4% alcoholic NaOH post-treatment. The compositions of the lignin fraction (AL) released during the 4% alcoholic NaOH post-treatment and lignin-rich residue (EHR) remained after the enzymatic hydrolysis of the 4% alcoholic NaOH post-treated acid-steam-exploded corn stover are listed in Table 2 . The relatively high purity (more than 80%) of AL made this isolated lignin a good material for subsequent utilization. The utilization of AL in LPF resin preparation could be regarded as a component part of the integrated biorefinery of corn stover. The realization of this approach could help to overcome the limitation of byproducts dispose on economic benefit and production scale of corn stover biorefinery. Meanwhile, it could indirectly promote the extension of biorefinery industry toward energy and advanced materials. EHR was obtained via centrifugation followed by freeze-drying. It could be observed that the content of remained carbohydrates (3.82%) in EHR was very low. The relatively high content of lignin in EHR (75.43%) made it as an ideal material for LPF resin preparation. The remained cellulase, which was composed by protein, was also ideal material for wood adhesive production like lignin dose [ 40 ]. Thus, the EHR obtained was also potential material for LPF resin synthesis. The application of this hydrolysis residue could help to realize the whole component utilization of corn stover. Table 2 Compositional analysis of the isolated lignin fraction from the 4% alcoholic NaOH treatment producer and the corresponding enzymatic hydrolysis residue Samples Lignin content (%) Carbohydrate content (%) AIL ASL Rha Ara Gal Glc Man Xyl GlcA/GalA AL 78.83 (0.86) b \n 3.32 (0.47) 0.23 (0.06) – a \n – 0.52 (0.11) – 4.62 (0.14) – EHR 71.83 (0.48) 3.60 (0.75) – – – 2.64 (0.14) – 0.86 (0.02) 0.32/– (0.02/–) \n AL alkaline lignin obtained from the 4% alcoholic NaOH post-treatment producer, EHR lignin-rich residue remained after enzymatic hydrolysis of the 4% alcoholic NaOH post-treated acid-steam-exploded corn stover, AIL acid insoluble lignin, ASL acid soluble lignin, Rha rhamnose, Ara arabinose, Gal galactose, Glc glucose, Man mannose, Xyl xylose, GlcA glucuronic acid, GlaA galacturonic acid \n a Not detected \n b The value in parenthesis is standard deviation \n The two dimensional heteronuclear single-quantum correlation (2D HSQC) spectra of AL and purified EHR are given in Fig. 5 and the spectra were annotated with peak assignments based on previous publications [ 41 , 42 ]. The structures of the identified lignin sub-units in the two lignins are also depicted in Fig. 5 . In the side-chain region of the 2D HSQC spectra of AL, cross-signals of methoxyls ( δ \n C / δ \n H 55.9/3.73) and β - O -4′ aryl ether linkages were the most prominent. The C β –H β correlation in β - O -4′ substructures (structure A) were observed at δ \n C / δ \n H 72.2/4.86. The C β –H β correlations corresponding to the Syringyl-type β - O -4′ substructures could be seen at δ \n C / δ \n H 85.8/4.12. It could be found that the intensities of the signals originated from β - O -4′ substructures in the spectrum of AL were significantly lower than those of the native lignin from the untreated corn stover as detected by our previous work [ 43 ]. This indicated that most of the β - O -4′ linkages in lignin macromolecule have been cleaved during the integrated treatment process. The signals of other substructures found in the spectrum of native lignin in untreated corn stover, such as resinol, phenylcoumaran, and spirodienone, were also absent in the spectrum of AL. The results indicated that lignin in acid-steam-exploded corn stover has been partially degraded and the moderate post-treatment conditions used in this study were competent for removing the lignins. However, it should be noted that the linkages between lignin units were hard to cleave during alkaline ethanol post-treatment in this study. The cleavage of β - O -4′ and other side-chain linkages should mainly occur during ASE treatment [ 44 ]. Fig. 5 2D 13 C- 1 H correlation (HSQC) spectra of AL and purified EHR ( a Aliphatic region of AL; b Aromatic region of AL; c Aliphatic region of purified EHR; d Aromatic region of purified EHR. Key structural details of lignin: ( A ) β - O -4′ aryl ether linkages; ( H ) p -hydroxyphenyl units; ( G ) guaiacyl units; ( S ) syringyl units; ( PCE ) p -coumarates; ( FA ) ferulates; ( X ) β - d -Xylp) \n Weak signals of β - O -4′ and other side-chain linkages implied that the AL possessed low molecular weight and relatively high phenolic hydroxyl content. Therefore, AL should be a precious phenolic material for LPF resin synthesis. Furthermore, in the 2D HSQC spectra of AL, the signals arising from β - d -Xylp were evidently noted, with its C 2 –H 2 , C 3 –H 3 , C 4 –H 4 , and C 5 –C 5 correlations at δ \n C / δ \n H 72.7/2.89, 73.6/3.22, 75.5/3.59, and 62.6/3.40 and 3.72, respectively. This implied that a certain amount of xylans still remained in AL, which was well consistent with the result obtained in carbohydrate analysis of AL (Table 2 ). Xylans were mainly covalently bonded with lignin via lignin-carbohydrate complex linkages [ 45 ], and thus they were inevitably removed along with the lignin during the alkaline ethanol post-treatment. The presence of these carbohydrates in AL may affect the performance of LPF resins prepared from the corresponding lignin fractions to some extent [ 46 ]. In the case of the purified EHR, a strong cross-signal of methoxyl was clearly found and other cross-signals of the side-chain linkages disappeared. The main cross-signals in the aromatic region of the 2D-HSQC spectra of AL are assigned to the aromatic rings of the different lignin units ( p -hydroxyphenyl (H), guaiacyl (G), and syringyl (S) units). Signals corresponding to p -coumarates (PCE) and ferulates (FA) were also observed in this spectrum with relatively low intensities. It should be mentioned that the presence of PCE and FA can also provide potential active sites (unsubstituted 3 or 5 positions of phenolic hydroxyl group) for further utilization, such as LPF resin preparation. In the case of the purified EHR, the cross-signals of S units were very weak, while the signals of G and H units were relatively strong. This indicated that the lignin fraction with a high content of S units was removed as AL during the 4% alcoholic NaOH post-treatment. \n 31 P NMR analysis involving phosphorylation of hydroxyl groups followed by quantitative analysis in the presence of an internal standard allows quantification of all –OH groups present in lignin. For this reason, the content of potential active sites of a lignin for application in LPF adhesive preparation can be obtained using 31 P NMR analysis [ 47 ]. The 31 P NMR spectrum of AL and the quantitative data on the distribution of the various –OH groups are shown in Fig. 6 . The data showed that the content of non-condensed G- and H-type phenolic –OH was 0.67 and 0.57 mmol g −1 , respectively. The active number calculated based on these data was 1.81 mmol g −1 , which was slightly higher than that of a technical lignin (1.72 mmol g −1 ) used in our previous study [ 47 ]. Therefore, the reactivity of AL could make it as an ideal material for LPF resin preparation. The poor solubility of purified EHR in solution of anhydrous pyridine and deuterated chloroform (1.6:1, v/v) seriously impeded the 31 P NMR analysis of this sample. Thus, it was not performed in this study. Fig. 6 \n 31 P NMR spectrum of AL \n LPF resin preparation The adhesive preparations of LPF resin using AL and EHR were both dark-brown aqueous solutions, and their specific properties are listed in Table 3 . Both resins had relatively high solid content (above 50%), which was favorable for forming a continuous bond line between the two cementing limiting surfaces. The viscosities of the ALPF and EHRPF obtained were 987.4 and 766.2 mPa s, respectively. It is well known that the high viscosity of LPF resin will obviously bring problem in the application stage. Fortunately, due to the good water solubility of the synthesized resin products, their viscosity could be adjusted to an acceptable level using water without blemishing the performance of the final products. Similar conclusion has been reported in our previous study [ 48 – 50 ]. Table 3 The properties and plywood performances of lignin-phenol-formaldehyde (LPF) resin adhesives Adhesives Adhesive properties Plywood performances pH Viscosity (mPa s) Solid content (%) Bonding strength (MPa) Formaldehyde emission (mg L −1 ) ALPF 11.8 (0.6) c \n 987.4 (1.58) 51.3 (0.44) 1.14 (0.11) 0.14 (0.06) EHRPF 11.3 (0.46) 766.2 (0.90) 53.3 (0.26) 1.01 (0.17) 0.21 (0.02) GB/T 14732-2006 a \n ≥7 ≥60 ≥35 ≥0.7 ≤0.5 b \n \n ALPF lignin-phenol-formaldehyde resin adhesive prepared with AL, EHRPF lignin-phenol-formaldehyde resin adhesive prepared with EHR \n a GB/T 14732-2006: Wood adhesives: urea formaldehyde, phenol formaldehyde and melamine formaldehyde resins \n b This requirement is defined by Chinese National Standard GB/T 9846.3-2004 \n c The value in parenthesis is standard deviation \n The bonding strength and formaldehyde emissions of the plywoods prepared with the two LPF resins are also listed in Table 3 . It could be observed that the plywoods assembled with the two resins performed differently. Interestingly, the bonding strength of both plywoods (1.14 and 1.04 MPa for samples prepared with ALPF and EHRPF, respectively) met the standard for exterior-grade panels (first grade, >0.7 MPa), and the formaldehyde emissions of the corresponding plywoods were all below 0.5 mg L −1 (0.14 and 0.21 mg L −1 for samples prepared with ALPF and EHRPF, respectively), meeting E 0 grade (<0.5 mg L −1 ) plywood requirements under Chinese National Standard GB/T 9846.3-2004 (Plywood-Part 3: General Specification for plywood for general use). It was also found that the performance of the plywoods prepared with either LPF were similar to that of the plywoods prepared with other LPF resins at the same lignin substitution level [ 50 ]. Thus, these two LPF resins could be utilized as low-toxicity wood adhesives to prepare both exterior plywood and interior E 0 -grade panels. The data shown in Table 3 indicated that the LPF resins prepared in this study had good comprehensive performance. However, it should to be noted that a specific performance of the synthesized LPF resin could be adjusted according to the actual requirement. Such adjustment could be realized by changing the mole ratio of phenol to formaldehyde and the content of lignin or lignin-rich residue in the formulation, and consequently it promotes further improvement in the applicability of AL and EHR in LPF resin preparation. In addition, it should also be emphasized that the AL sample used to prepare LPF resins in this study was isolated and purified before resin synthesis. However, neither isolation nor purification of this lignin fraction was necessary for synthesizing LPF resin in a real industrial production scenario. A simple concentration process may be sufficient to obtain lignin solutions with desired solid content. For EHR, desiccation was unnecessary for their utilization in LPF resin preparation since water was also a common component of LPF resin. In fact, the application of EHR in this study inspired the valorization of fermentation residue obtained from bio-ethanol production. The remained yeast in the lignin-rich fermentation residues also contains a large proportion of protein, and may contribute to resin synthesis like cellulase dose. Thus, the lignin-rich fermentation residues can also be appropriate materials for LPF resin preparation."
} | 7,277 |
36375055 | PMC9704736 | pmc | 7,901 | {
"abstract": "Significance Mercury pollution of soil and water worldwide is a major threat to public health, food chains, and agriculture. Bioremediation is an environmentally friendly solution. Here, we report molecular mechanisms underlying mercury tolerance in the plant symbiotic fungus Metarhizium robertsii . In mercury-polluted soil, this fungus, nourished by plant-derived nutrients, demethylates methylmercury via a demethylase and volatilizes divalent mercury using a reductase. Persistently removing mercury from soil in this manner decreases its accumulation in plants and increases plant growth. Metarhizium can also remove mercury from nutrient-free fresh and sea water. Genetic engineering was used to further improve the ability of M. robertsii to bioremediate mercury-polluted soil and water, facilitating its potential use in helping manage a complex set of dangerous environmental trends.",
"discussion": "Discussion In this study, we describe the genetic and biochemical mechanisms underlying mercury tolerance in the commercially important plant symbiotic fungus M. robertsii . MeHg is demethylated by the demethylase MMD into Hg 2+ , which is subsequently reduced to elemental Hg through the Hg 2+ reductase MIR. MIR homologs were found in many fungi, suggesting Hg 2+ resistance conferred by MIR is widespread. However, MMD homologs were rare and patchily distributed among plant associates and soil fungi, and phylogenetic tracks suggest that they could have been acquired through two different evolutionary trajectories. MMD-mediated fungal resistance to MeHg could therefore be the result of convergent evolution by some soil fungi to survive mercury stress in their environment. M. robertsii can develop mutually beneficial relationships with many diverse agriculturally important plants including maize, the world’s most dominant and productive crop, where it is known to promote growth, suppress insect growth, and alter plant defense gene expression ( 17 ). In return, the plant roots provide a long-term habitat and a source of carbohydrates for the fungus ( 15 ). We found no evidence that either Mmd or Mir contributes to symbiotic interactions between M. robertsii and maize in normal soils. However, in MeHg- and Hg 2+ -polluted soil, a fast-increasing threat to agriculture and ecosystems, detoxification of these Hg forms by MMD and MIR protects the fungus and reduces mercury in plants, facilitating their growth. Therefore, MMD and MIR promote a mutually beneficial relationship between plants and M. robertsii under mercury stress, as reflected in the strong correlations between CFUs of the different Metarhizium mutant strains and plant growth. Given that emission of Hg from soil greatly affects the global Hg cycle ( 18 ) and plant symbiotic Metarhizium species are among the most abundant soil fungi ( 15 ), Metarhizium species with their host plants could represent an important branch of the global Hg cycle. This also has important implications for potential bioremediation of Hg-polluted soil. M. robertsii can be applied to seeds before planting ( 19 ). This contrasts favorably with the bacterial sources of MMD and MIR, as these cannot reproduce in soil, and have yet to be used for bioremediation ( 20 ). There is an extra dimension in the quality of the interactions between fungi and plants as, unlike bacteria, fungi can grow and spread through the rhizosphere as hyphal growth. However, bioremediation appears limited to the vicinity of roots, where the fungus is localized, which will also localize and limit the release of volatile elemental mercury. If required, release of volatile mercury could be further contained by applying sorbents such as activated carbon on the surface of the soil overlying the root system. Ease of cultivation and genetic manipulation of M. robertsii means the ability to remove MeHg and Hg 2+ can also be enhanced by the simple expedient of overexpressing MMD and MIR. In countries that are unfavorably disposed to transgenic products, it may be possible to use chemical mutagenesis and/or artificial selection to enhance production of MMD and MIR. Multiple Metarhizium species can efficiently demethylate MeHg and reduce Hg 2+ , and can develop symbiotic relationships with diverse plants including grasses, trees, vegetables, and crops ( 17 , 21 ). However, there is evidence for coevolution with plants in that M. robertsii preferentially associates with the roots of grasses, M. brunneum with shrubs, and M. guizhouense with trees ( 22 ). So, plants and Metarhizium species could be combined in optimal pairs for cleaning up MeHg and Hg 2+ in different types of polluted soil. Furthermore, Metarhizium mycelium efficiently removes MeHg and Hg 2+ in water at concentrations >1,000-fold (MeHg) and 500-fold (Hg 2+ ) higher than the limit (2 μg/L) recommended by the EPA ( 16 ). Unlike bacteria that usually require nutrients to remediate Hg-contaminated water ( 20 ), M. robertsii did not require any additional supplements. Metarhizium fungi have a long history of being used as biocontrol agents against insect pests, and their safety to humans and the environment has been clearly established through several decades of high-quality research ( 23 ). In addition, industrial production of Metarhizium for insect pest control is highly automated and cost-effective ( 24 ). Therefore, Metarhizium fungi seem well-placed to help manage a complex and unprecedented set of dangerous environmental trends."
} | 1,377 |
27877733 | PMC5090392 | pmc | 7,902 | {
"abstract": "Silica microfiber wool was systematically functionalized in order to provide an extremely water repellent and oleophilic material. This was carried out using a two-step functionalization that was shown to be a highly effective method for generating an intense water repulsion and attraction for oil. A demonstration of the silica wools application is shown through the highly efficient separation of oils and hydrophobic solvents from water. Water is confined to the extremities of the material, while oil is absorbed into the voids within the wool. The effect of surface functionalization is monitored though observing the interaction of the material with both oils and water, in addition to scanning electron microscope images, x-ray photoelectron spectroscopy and energy dispersive x-ray analysis. The material can be readily utilized in many applications, including the cleaning of oil spills and filtering during industrial processes, as well as further water purification tasks—while not suffering the losses of efficiency observed in current leading polymeric materials.",
"conclusion": "4. Conclusions The material reported in this article demonstrates highly efficient and robust oil–water separation. The superhydrophobic silica wool was capable of achieving water contact angles of 165° on average. The wool, which is readily produced using affordable methods, strongly absorbs hydrophobic solvents and thick oils (up to 12.5 mL g –1 of wool). The ability to reuse the material was effectively demonstrated, acting as a sponge, able to absorbed the oils and repel water. The wool samples could be rapidly incorporated in a commercial device, either in the use of cleaning oil spills or essentially in other areas where hydrophobic solvents and oils would need to be separated from water.",
"introduction": "1. Introduction Recent reports on oceanic oil spills have highlighted the devastating effect that these events can have on the environment and their subsequent economic consequences [ 1 ]. The dependence that human populations have for oil ensures that its extraction will be a result of deeper drilling, and other forms of hazardous abstraction methods. This heightened risk not only renders large oil spills more likely, but the locations of these accidents will be ever more remote [ 2 ]. This almost inevitable threat to the environment and the finances of nations and individuals has driven the development of technology to tackle this problem. There are many common approaches currently used to separate oil–water mixtures [ 3 – 5 ], however the selective filtration of water–oil mixtures can be successfully achieved when a material has simultaneous repulsion of water and attraction of oil (or vice versa) [ 5 ]. When contemplating the removal of oil from water, important in oil spills into sea water and separation during processing, then a material best suited to filtration is one that greatly repels water (superhydrophobic) but greatly attracts oils (superoleophilic). This ensures that oil is captured within the material while the water is not [ 6 ]. Surface hydrophobicity is maximized through combining a material that has inherently water repelling properties, in conjunction with an extremely high surface roughness. Water repelling species, such as alkyl or fluorinated alkyl groups, act to lower surface energy and will repel water molecules [ 7 ]. The surface roughness acts to magnify these properties, while allowing air to be trapped under surface protrusions; both features act to increase the hydrophobicity of the surface [ 8 ]. Alkylated surface groups also act to strongly attraction oils [ 9 ]. Porous materials that utilize this hydrophobic/oleophilic property would be able to repel water from the bulk of the material while attracting in oil. The main measure of surface hydrophobicity is the water contact angle. This is the angle made between the plain of a surface and tangent made by a water droplet laying on the surface at the water-surface–air interface. A surface is termed superhydrophilic if this angle is below 5°, hydrophilic if it is below 90°, hydrophobic if it is greater than 90° and superhydrophobic above 150° [ 10 ]. Models for the visualization of surface hydrophobicity have been reported; the Cassie–Baxter model rationalizes the trapping of air under water lying on a surface (this is the case in most superhydrophobic surfaces) [ 8 ]. Superhydrophobicity can also be gauged through water bouncing experiments [ 11 ]. The same principle of surface design can apply for hydrophobic solvents, where oleophilic surfaces give contact angles below 90°; additional roughness will magnify the attraction of oils to the surface [ 12 ]. In order to facilitate water–oil separation a surface must have both low contact angles for oils or hydrophobic solvents, and high contact angles with water. Another requirement is that the surface should be highly porous to facilitate the absorption of solvent into the bulk where it can be retained. The design, manufacture and application of superhydrophobic surfaces are all well reported in the literature [ 13 – 15 ]. There are also device components and commercial products currently on the market, which include foams, meshes/filters and dispersants [ 16 – 18 ]. Most of these technologies are based on superhydrophobic/superoleophilic characteristics or vice versa. General routes to making superhydrophobic surfaces include the roughening of already water repellent surfaces, coating/functionalization of already roughened surfaces, in addition to the development of surface roughness using already water repelling material [ 13 – 15 ]. Separating oil–water mixtures requires further developments on the above approaches. The superhydrophobic surfaces will repel water, but must also attract oil, thus there must be somewhere for that oil to go; either into the bulk of the material, or an alternative pathway must be provided [ 19 , 20 ]. The literature reports the use of superhydrophobic meshes which allow oils to drip through the pores, leaving water on the top side of the mesh and the oil to drip into a collector [ 5 ]. Another reported approach is the use of polymeric materials which preferentially absorb oil over water, however the hydrophobic solvents used could not be easily separated from the polymer which was disposed of after use [ 21 ]. This highlights a major failing of some approaches to separating oil from water; removal of the oil from the material after separation has occurred. Recent reports also include the fabrication of water purification devices [ 22 – 24 ]. These are enclosed systems which use a flow of solvent/water mixtures through a superhydrophobic or superoleophobic filter. This method of purification has been shown to be highly effective, however these systems are susceptible to blockages if the denser liquid is not absorbed by the filter. For example using a water/hexane mixture on a hydrophobic filter—the denser water would sink to cover the filter thus not allowing the collection of hexane. These systems would not be readily applicable to large oil spills (relative to uncomplicated oil absorbent materials), as a pumping mechanism or some other alternative would be required. Superhydrophobic/oleophilic sponges have also been reported, these are materials that can absorb oil while repelling water [ 25 – 27 ]. These can show extremely selective absorption with respect to oils and hydrophobic solvents, however the materials used in their fabrication can suffer from degradation and polymer swelling, in addition to high fabrication cost [ 28 – 30 ]. We report the functionalization of silica (SiO 2 ) wool. The surface of the silica wool was activated with piranha solution and then functionalized using hexamethyldisilazane (HMDS), to form trimethylsiloxane (TMS) groups on the surface of the wool (figure 1 ) [ 31 ]. The result was a lowered surface energy, which when combined with the inherent surface roughness of the wool, gave a superhydrophobic material. The surface functionalization also rendered the material superoleophilic, strongly attracting hydrophobic solvents and oils (toluene, hexane, petroleum ether and motor oil). The result was a material that could separate oil and hydrophobic solvents from water with high efficiency. The range of solvents used could be easily removed from the wool by compression after use, enabling the wool to be reused. This study shows the relative absorption efficiencies of each solvent and also the separation experiments with water–solvent mixtures. Figure 1. Surface reaction scheme at silica surface. Surface silanol groups are formed under the action of piranha solution (i), and subsequently functionalized by using HMDS to form surface TMS groups (ii). The surface undergoes a hydrophilic to hydrophobic transformation during reaction ii .",
"discussion": "3. Results and discussion Preliminary experiments were carried out using small portions of wool (∼0.025 g); these samples were imaged via SEM before and after any treatments were applied. There was no observed change in the physical appearance or robustness of the silica wool, and no change in microstructure (figure 2 ); this suggested that any change to the material was surface chemistry focused. The SEM images did show the inherent roughness of the wool, with silica fibres approximately 16 μ m in diameter. The silica wool was calculated to have a surface area of approximately 0.1 m 2 g −1 . Figure 2. SEM images of functionalized silica wool. The fibres measure ∼16 μ m in diameter. The individual fibres have a relatively smooth surface (B), and so roughness comes from the collection of fibres within the wool (A). The inset shows a water droplet (3 μ L) on silica wool functionalized using HMDS. Water contact angles were above 165°. The as-received silica wool was hydrophilic; any water droplets making contact with the surface were immediately absorbed into the voids of the material. This was not altered by the action of piranha solution, as water absorption happened as readily; rendering water contact angle measurements inapt for describing the hydrophobicity of the material. The wool was rendered superhydrophobic upon functionalization via the overnight treatment with HMDS (figure 1 , embedded). Water contact angles of approximately 165° were observed, however exact determination was made difficult as the surface of the wool was uneven. Water droplets would tend to roll off the surface of the wool when tilted away from the horizontal (see supplementary information S2, available at stacks.iop.org/STAM/15/065003/mmedia ), however trapping of water droplets was common where wool strands intersected. In order to gain a further evaluation of the superhydrophobicity of the wool water bouncing experiments were carried out. The observed total number of water bounces averaged 3–4 (using standard conditions) [ 11 ], which correlates with the estimated water contact angle (∼165°). Figure 3 shows the x-ray photoemission spectra for Si (2p) and O (1s) core transitions for untreated/treated silica wool. The Si (2p) spectra are presented in figure 3 (on the left) and show that for untreated and piranha treated silica wool that surface SiOH groups are present in significant quantities. Silica wool pre-treated with piranha solution (figure 3 (c)) shows a shift to lower binding energy that correlates to a more electropositive environment around the Si. It can be seen that the O 1s position is also shifted towards lower binding energy (figure 3 (c)) with piranha treatment indicating a change in the surface oxygen environment. A broad shoulder can also be seen at higher binding energy, indicative of surface hydroxyl species. This feature is more pronounced in the case of silica wool pre-treated with piranha solution. The experiments were repeated in order to determine experimental reproducibility, as the observed shifts are not very large. Infra-red spectroscopy was carried out before and after functionalization, however no change in surface was able to be detected. This was a limitation of the infra-red instrument used. Figure 3. Si (2p) and O (1s) XPS spectra of (a) as-bought silica wool, (b) HMDS functionalized as-bought silica wool, (c) piranha pre-treated silica wool and (d) HMDS functionalized piranha pre-treated silica wool. Dotted lines show the centre of each signal. The absorption of different hydrophobic solvents was tested (table 1 ). Portions of wool of recorded mass were submerged into excess amounts of water, toluene, hexane or petroleum ether, the samples were withdrawn and weighed immediately. It was found that in all cases water absorption into the wool was significantly lower than that of the three hydrophobic solvents used. The efficiency of wool samples to absorb solvents increased with the mass of the wool sample used. The relative amount of water picked up by the wool also greatly decreased with increased mass. The reason for this is that the little water that was collected was mainly trapped at the surface of the wool, as it did not enter the bulk. Furthermore, with increasing size the surface area-volume ratio decreases, thus there is a relatively greater volume to absorb solvent and less surface to trap water. The three hydrophobic solvents all show similar patterns in absorption volume. The best performing solvent was hexane which was absorbed into the wool over 38 times that of water, when using 10 g of superhydrophobic wool. The ability of the oleophilic wool to absorb and retain each solvent is dependent on three main factors. Firstly, the surface attraction for the respective solvent: the stronger the attraction the more solvent will be absorbed and will be held with a stronger force. Secondly, the viscosity of each solvent affects the amount of solvent collected. This is due to the seeping of solvent into the solvent mixture upon the removal of wool samples. Solvents with lower viscosity would demonstrate a faster flow from the soaked silica wool samples. A final factor that influences the volume of solvent collected is the room temperature density of the respective solvents (Toluene 0.87 g cm −3 , hexane 0.68 g cm −3 and petroleum ether 0.64 g cm −3 ). The density of solvents would affect the absorption efficiency, as the voids in the silica wool samples would hold a greater mass of a denser solvent. Table 1. Table showing the absorption of hydrophobic solvents and water into functionalized portions of wool. Measurements were repeated three times for each wool portion. 0.025 1.5 10 Mass of wool (g) Average volume of solvent absorbed per gram of silica wool (mL) Water 3.33 1.37 0.32 Toluene 7.73 10.02 11.61 Hexane 8.08 11.53 12.33 Petroleum ether 5.54 10.03 10.66 The ‘as-received’ and hydroxylated wool samples did not have contrasting interactions with oil and water, as shown for superhydrophobic samples. Equivalent experiments carried out using 0.025 g potions of ‘as-received’ wool showed that it was able to absorb 8.4 mL g −1 of toluene and 11.5 mL g −1 of water; whereas hydroxylated wool absorbed 3.9 mL g −1 of toluene and 5.9 mL g −1 of water. Silica wool samples exposed to only the HMDS functionalization showed absorption of 7.6 mL g −1 and 6.5 mL g −1 of toluene and water respectively. The ‘as-received’ and hydroxylated samples absorbed more water than toluene, rendering those samples inadequate for the purpose of scavenging oil from water. The wool sample which solely underwent HMDS treatment, had similar affinities for both toluene and water. The latter experiment showed the importance of the initial hydroxylation; this is proposed to produce a greater density of surface silanol groups [ 32 ], which then in turn produce a higher density of TMS groups on the surface. The higher density of TMS groups results in a more hydrophobic material, and thus a material with greater selective absorption of oils [ 33 ]. The ability of the wool to separate out hydrophobic solvents from water was also tested. A 50 g portion of wool was used to separate 250 mL of hydrophobic solvent from 2500 mL of water (figure 4 ). The superhydrophobic wool samples were completely submerged in the solvent mixture, withdrawn, and remaining solvent measured. The results were similar to those seen in the individual testing, with very low amounts of water trapped on the wool’s surface (average ∼1 mL g −1 ). The amounts of toluene, hexane and petroleum ether absorbed into the 50 g mesh were 210 mL, 213 mL and 204 mL respectively. The superhydrophobic wool samples did not pick up 100% of the solvent as there was some amount of dripping back into the water-solvent container. Figure 4. A series of photographs showing the removal of toluene from water. (a) A 50 g portion of superhydrophobic silica wool was used; the starting solvent mixture contained 2500 mL of water and 250 mL of toluene. (b) The wool was submerged into the mixture absorbing the toluene (no organic layer present). (c) The wool was removed; the end mixture contained ∼2450 mL of water and 40 mL of toluene. The water was coloured with low concentrations of methylene blue dye to aid visualization (this did not alter the results obtained). Portions of superhydrophobic wool (∼10 g) were used to assess reusability of the wool. The same wool sample was used repeatedly to pick up hydrophobic solvent, removed and compressed by hand between two glass plates. The mass of solvent removed with each of ten sequential submersions was recorded. Each experiment was repeated three times. The results showed that the first submersion of the wool picked up the most solvent, 12.5 mL g −1 was averaged for hexane. This steadily decreased to 7.6, 7.2 and 6.2 mL g −1 in the subsequent submersions, until after five submersions when an average of 5.4 mL g −1 was removed each time. Similar results were observed with the other solvents used, with 5.7 mL g −1 reached for toluene and 4.8 mL g −1 for petroleum ether. This shows the hydrophobic silica wool can be used repeatedly to remove hydrophobic solvents. The wool samples were examined through water contact angle, SEM and EDX analysis after the absorption experiments. No change in microstructure or composition was observed after separation experiments were carried out. Water contact angles remained unchanged and there was no visual modification of the material observed. Surface hydrophobicity measurements (water contact angle and bouncing) were taken after samples were cleaned, and three months after use; no degradation of hydrophobicity was observed. The wool samples were also exposed to thicker motor oil (20 w/50). Similar results were observed; the wool was able to absorb >10 mL of oil per gram of superhydrophobic wool. The wool was also able to be reused in collecting the motor oil: the average collection of a 10 g sample of wool after five consecutive uses was 6.9 mL g −1 . The wool was also tested to separate the motor oil from water (see supplementary information S3, available at stacks.iop.org/STAM/15/065003/mmedia ). The collection proved slightly more efficient as there was substantially less dripping from the wool when pulled from the mixture. The higher efficiency was caused by the reduced tendency of the thicker motor oil to drip (with respect to the faster flowing hydrophobic solvents) once captured in the wool and removed from the mixture. The separation of oil from water using the reported superhydrophobic silica wool operates via preferential absorption of oil into the voids of the wool material. This can be compared to materials such as activated carbon, where organic impurities in water can be absorbed into the voids within the microstructure [ 34 ]. A substantial benefit that superhydrophobic silica wool has over this and related materials is that the removal of oil can be performed by simple and repeatable compression; without the need for rinsing. This extends the potential areas of application beyond static filtration materials and toward usage as boom materials for oil absorption at sea, due to the ease of reusability. Currently in this area; polyurethane foams are amongst the most successful and widely used [ 35 ], however current limitations of these materials are reflected in the swelling of the polymer on extended exposure to solvents and oils [ 29 ]. The result of this is that the absorbed oil is trapped between polymer chains and will not be able to be easily recovered. In addition, the ability of polyurethane to absorb oil will be diminished over time; a factor that does not affect superhydrophobic silica wool. The effectiveness of these materials could also be improved through the incorporation of additional porosity or roughness [ 36 , 37 ]."
} | 5,184 |
28256065 | PMC5413822 | pmc | 7,903 | {
"abstract": "Abstract Shallow‐water coral reef ecosystems, particularly those already impaired by anthropogenic pressures, may be highly sensitive to disturbances from natural catastrophic events, such as volcanic eruptions. Explosive volcanic eruptions expel large quantities of silicate ash particles into the atmosphere, which can disperse across millions of square kilometres and deposit into coral reef ecosystems. Following heavy ash deposition, mass mortality of reef biota is expected, but little is known about the recovery of post‐burial reef ecosystems. Reef regeneration depends partly upon the capacity of the ash deposit to be colonised by waterborne bacterial communities and may be influenced to an unknown extent by the physiochemical properties of the ash substrate itself. To determine the potential for volcanic ash to support pioneer bacterial colonisation, we exposed five well‐characterised volcanic and coral reef substrates to a marine aquarium under low light conditions for 3 months: volcanic ash, synthetic volcanic glass, carbonate reef sand, calcite sand and quartz sand. Multivariate statistical analysis of Automated Ribosomal Intergenic Spacer Analysis ( ARISA ) fingerprinting data demonstrates clear segregation of volcanic substrates from the quartz and coral reef substrates over 3 months of bacterial colonisation. Overall bacterial diversity showed shared and substrate‐specific bacterial communities; however, the volcanic ash substrate supported the most diverse bacterial community. These data suggest a significant influence of substrate properties (composition, granulometry and colour) on bacterial settlement. Our findings provide first insights into physicochemical controls on pioneer bacterial colonisation of volcanic ash and highlight the potential for volcanic ash deposits to support bacterial diversity in the aftermath of reef burial, on timescales that could permit cascading effects on larval settlement.",
"conclusion": "5 Conclusion This study is the first to investigate volcanic ash as a substrate for pioneer bacterial colonisation using a simulated coral reef environment. We show that volcanogenic substrates support a notably diverse bacterial community, exhibiting higher numbers of OTUs over the course of 1–3 months than both terrigenic and biogenic substrates. The observed diversity amongst substrates indicates that the initial community structure is likely dictated by differences in substrate physicochemical properties. We identify greater diversity in substrates with higher specific surface areas compared to those with lower surface areas but are compositionally similar, which, coupled with comparisons to in situ studies, suggests possible controls associated with particle physical properties (e.g., granulometry, surface morphology). Our findings also suggest a significant control of substrate composition (bulk chemistry and mineralogy), which could result from a direct influence of nutrient availability or an indirect influence through substrate colour, whereby compositionally diverse dark‐coloured volcanogenic substrates favoured development of a larger community structure relative to light‐coloured quartz and biogenic substrates. Identification of the bacterial community diversity using next‐generation sequencing and an “omics” approach would provide additional information about the pioneer bacterial communities on the different substrates and help to understand their function. Critically, our findings indicate the potential for volcanic ash to promote bacterial diversity in an immediate post‐burial scenario in ash‐affected coral reefs, on timescales that could permit cascading effects on larval settlement and ultimately reef recovery. Further investigation of coral reef recovery and resilience following large‐scale natural disturbances, such as volcanic ash deposition, may contribute to predictions on ecosystem recovery and hazard management strategies.",
"introduction": "1 Introduction Coral reefs are unique, biodiverse ecosystems of high socio‐economic importance on both global and local scales (Nicholls et al., 2007 ). Anthropogenic disturbances, such as sedimentation and eutrophication, increasingly pressure fragile coral reef ecosystems worldwide (Wilkinson 1999 ). The deterioration of water quality in coastal regions consequently favours macro‐algal dominance (Fabricius, 2005 ; Schaffelke, Mellors, & Duke, 2005 ) and increases the risk of disease for coral reef‐building species, including sponges and corals (Haapkyla et al., 2011 ; Webster, Xavier, Freckelton, Motti, & Cobb, 2008 ), which further exacerbates coral reef vulnerability to catastrophic natural disturbances, such as volcanic ash deposition (Vroom & Zgliczynski, 2011 ). After an explosive volcanic eruption, widespread dispersal and deposition of volcanic ash over areas up to millions of square kilometres, in thickness of up to several centimetres, may be damaging to ash‐affected coral reef ecosystems; both Maniwavie, Rewald, Aitsi, Wagner, and Munday ( 2001 ) and Vroom and Zgliczynski ( 2011 ) have reported the destruction and mass mortality of reef biota following heavy ash deposition. However, the capacity of reef ecosystems to recover after burial by ash remains uncertain. Maniwavie et al. ( 2001 ) reported that 2 years after burial by volcanic ash corals had only re‐colonised the surfaces of protruding or unburied objects (e.g., boulders, tree stumps), while the ash substrate itself remained barren; in contrast, Schils ( 2012 ) noted that a period of frequent ash deposition into a tropical reef ecosystem promoted a change in benthic microbial and macrofloral communities on a similar timescale. These varying responses indicate a clear need to better understand the factors that may dictate the recovery of vulnerable and valuable coral reef ecosystems after ash deposition. In the aftermath of large‐scale burial, recovery of the reef ecosystem may depend on pioneer colonisation of the new substrate by free‐living bacteria from the water column. After attachment to the surface, these bacteria produce an extracellular polymeric matrix that embeds further microbial organisms, forming so‐called biofilms (Costerton, Lewandowski, Caldwell, Korber, & Lappin‐Scott, 1995 ). Biofilm communities are highly abundant in coral reefs and are crucial in biogeochemical nutrient cycling and the degradation of anthropogenic pollutants (reviewed in Davey & O'Toole, 2004 ). Further, they provide an essential settlement surface for larvae of important reef‐building invertebrates (e.g., corals and sponges) and influence larval settlement cues and metamorphosis (Webster et al., 2004 ; Wieczorek & Todd, 1998 ). Accordingly, any changes in bacterial biofilm communities may influence invertebrate larval settlement, coral reef establishment and further development. Therefore, the capacity of a volcanic ash substrate to support bacterial settlement, particularly compared to the marine substrates it overlies, may play a crucial role in shaping the recovery of ash‐affected reef ecosystems. Previous studies on aquatic biofilm formation using an array of natural and artificial substrates, including basaltic glasses and borosilicate (Thorseth, Furnes, & Tumyr, 1995 ), biotite (Ward, 2013 ), granite (Chung et al., 2010 ), coral skeletons and clay tiles (Witt, Wild, & Uthicke, 2011 ), highlight the importance of physicochemical properties (e.g., granulometry, surface morphology, mineralogy and chemistry, colour) in promoting initial substrate colonisation. Crucially, volcanic ash materials are subject to a wide variation in all of these properties, which are the product of magma composition and eruption history (Dingwell, Lavallée, & Kueppers, 2012 ). Ash particles range in size from the millimetre to submicron scale and vary in morphology from smooth, blocky particles, to rough‐textured vesicular clasts (Heiken, 1974 ). They commonly contain crystalline and amorphous silicates of various compositions, and range in colour from light to dark (Ayris & Delmelle, 2012 ). Ash surfaces can be a source of variably extractable elements, some of which (e.g., Al, Ca, Co, Cu, Fe, K, Mg, Mn, Ni, Mo, P, S, Zn; Jones & Gislason, 2008 ) may be important macro‐ or micronutrients for bacteria and phytoplankton (Duggen, Croot, Schacht, & Hoffmann, 2007 ; Munn, 2003 ), while others (e.g., Al, Cu) may be toxic (Duggen et al., 2007 b). Investigating the propensity for volcanic ash to promote pioneer bacterial colonisation in situ is hampered by the difficulties associated with substrate accessibility, geographic location and the dangers associated with sampling near active volcanoes. Laboratory experiments conducted in simulated tropical coral reef aquaria, therefore, offer a viable method to approximate the properties of volcanic ash that may dictate its capacity to act as a colonisable substrate. In this study, we incubate a selection of volcanogenic, terrigenic and biogenic substrates in a coral reef‐like marine aquarium system for 3 months and correlate differences in the bacterial colonising consortia with physicochemical properties (i.e., chemical composition, mineralogy, granulometry, morphology, colour) of particulate substrates.",
"discussion": "4 Discussion No previous in situ studies have investigated bacterial colonisation of volcanic ash as a substrate within a reef environment, and the only in situ studies documenting the aftermath of ash deposition are both focused on macroflora and fauna and are contrasting in their results: Maniwavie et al. ( 2001 ) implied that recolonisation of ash substrates by reef flora did not occur, whereas Schils ( 2012 ) reported a sudden change in benthic microbial and macrofloral communities of an ash‐affected reef. Therefore, while comparisons between the aquarium experiment of the current study and in situ reef settings should be treated with caution, our study makes an important contribution to a highly uncertain subject by demonstrating that bacteria can colonise ash substrates and can establish a significantly different community structure than those on co‐existing substrates commonly found in reef habitats. Importantly, larval settlement is driven by bacterial community composition and settlement success rates increase with the age of the biofilm, and the timescale over which significant differences in bacterial community structure were observed in the current study are similar, and relevant, to the timescales required for the settlement of invertebrate larva (days to weeks; Bao, Satuito, Yang, & Kitamura, 2007 ; Campbell et al., 2011 ). Consequently, the significant differences in pioneer bacterial colonisation between volcanogenic substrates, and the terrigenic and biogenic substrates in the present study, even after 3 months, could impart further differences in invertebrate larval settlement, and so have cascading effects on the subsequent reef formation and succession. This fast response time echoes recent studies of fresh basaltic lava flows in terrestrial environments, where it was suggested that bacterial communities were rapidly established within days or months of lava flow emplacement (Kelly, Cockell, Thorsteinsson, Marteinsson, & Stevenson, 2014 ). 4.1 Differences in bacterial community structure Evaluation of the bacterial communities colonising volcanogenic, biogenic and terrigenic substrates in a coral reef environment mesocosm, as determined by DNA fingerprinting of the ITS region using Automated Ribosomal Intergenic Spacer Analysis (ARISA), showed both substrate‐specific and shared bacterial OTUs. Such differences in the establishment of biofilms on different substrates are compatible with previous investigations on biofilm formation on different silicate and carbonate substrates incubated in tropical marine waters, which also noted the importance of the substrate for biofilm formation (Chung et al., 2010 ; Dobretsov, Abed, & Voolstra, 2013 ; Witt et al., 2011 ). Further, DNA fingerprinting revealed distinct differences between bacterial biofilm communities on the substrates and the communities found in the water column over the 3‐month experiment. Differences in the structure and diversity between the attached and free‐living bacterial communities are in line with in situ observations in different marine coastal systems (Mohit, Archambault, Toupoint, & Lovejoy, 2014 ; Zhang, Liu, Lau, Ki, & Qian, 2007 ) and have also previously been detected in coral reefs in situ (Santavy & Colwell, 1990 ; Schöttner et al., 2009 ). Of the five substrates, we observed that the volcanogenic group (ash, glass) carried a more diverse, and significantly different, bacterial community compared to the communities associated with biogenic (carbonate and calcite sand) and terrigenic (quartz) substrates. The SIMPER analysis revealed substrate‐specific bacterial OTUs in the volcanogenic substrates contributing the most to the observed differences. These were either OTUs found in all substrate communities, which were more prevalent in the volcanogenic substrates, or were unique to the volcanogenic substrates, highlighting the substrate specificity of the colonising bacterial community. Further, statistical tests of the Shannon–Wiener diversity index confirmed significantly higher diversity within the volcanogenic substrates when compared to the other three substrates, while, amongst the biogenic and terrigenic substrates, no significant differences in diversity were detected. However, within the volcanogenic and biogenic substrate groups, the number of total OTUs and diversity indices indicate that the ash and carbonate reef sands carried a higher bacterial diversity than the calcite and glass. The observed differences in bacterial abundance and community differences amongst substrates indicate a substantive control of substrate physicochemical properties on bacterial community settlement and structure. 4.2 Physicochemical controls on bacterial communities Differences in substrate physical properties (surface morphology and granulometry) may account for the higher bacterial diversity on the ash and carbonate reef sand compared to the glass and calcite sands. The former materials exhibit higher specific surface areas, reflecting higher surface roughness, porosity or the presence of smaller particles adhering to larger particle surfaces; these properties are known to affect growth in other bacterial systems (Gerasimenko, Orleanskii, Karpov, & Ushantinskaya, 2013 ; Yamamoto & Lopez, 1985 ). We detected significantly different bacterial communities on white carbonate and calcite substrates, while quartz was the same as both of these biogenic substrates. In contrast, in a comparable study of silicate and carbonate substrates in situ (Red Sea), Schöttner et al. ( 2011 ) found that both sands showed similar bacterial density and diversity, although with differing bacterial community structure over seasonal changes. However, the carbonate sand investigated by Schöttner et al. ( 2011 ) was poorly sorted and significantly coarser (median particle size = 553 μm) than the quartz sand (median particle size = 326 μm). Yet, no evident influence of substrate chemistry between quartz and carbonate materials in Schöttner et al. ( 2011 ) and the current study implies that differing substrate granulometry may have caused bacterial niche‐partitioning. An effect of substrate colour could explain our observation that the two dark volcanogenic substrates carried more diverse bacterial communities than the three light substrates. The colour of a substrate has recently been observed to affect attraction of different microbial communities, whereby black substrates carry a higher bacterial density than white substrates (Dobretsov et al., 2013 ). Notably, diverse invertebrate settlement assays have shown that some larvae and algal spores prefer dark substrates over light‐coloured ones, likely due to diminished light reflection and greater heat retention of dark substrates, conditions that are preferred by negatively photo‐tactic organisms (Svane & Dolmer, 1995 ). Further, invertebrates often prefer dark substrates for colonisation to benefit from better protection from grazers (Swain, Herpe, Ralston, & Tribou, 2006 ). These factors are likely to apply to bacterial colonisation as well. Therefore, Dobretsov et al. ( 2013 ) and the current study emphasise the effects of colour on bacterial settlement, which should be tested further in future settlement assays. The chemistry and/or mineralogy of the substrates could equally govern the observed differences in community structure, as substrate colour is strongly influenced by composition and the coordination of atoms within a material (Nassau, 1978 ). These properties will, in turn, influence the distribution and availability of nutrients at the substrate surfaces. Nutrients may be extracted from the substrate surface by leaching and dissolution by organic compounds and by water within the established biofilm (Brehm, Gorbushina, & Mottershead, 2005 ). Pre‐leaching of the materials in the present study was essential to ensuring that nutrients were substrate‐derived. However, the response to available nutrients is likely to be species‐dependent (Gerasimenko et al., 2013 ; Nies & Silver, 1989 ); thus, differences in substrate chemistry and/or mineralogy may promote or inhibit growth of different bacterial species on the different substrates. Dependences of bacterial community structure on substrate mineralogy have been previously invoked by Kelly et al. ( 2011 , 2009 ), Gleeson et al. ( 2006 ) and Hutchens, Gleeson, McDermott, Miranda‐CasoLuengo, and Clipson ( 2010 ). The first two studies listed above note correlations between bacterial communities on weathered glasses and crystalline rocks, depending on their composition (basaltic to rhyolitic). The last two studies document significant differences in bacterial community structuring in biofilms growing on adjacent (on length scales of cm to m) silicate minerals (quartz, albite, K‐feldspar, muscovite) at the surface of a terrestrial granite. Accordingly, there may therefore be significant differences in bacterial community structure both spatially across a single ash deposit, and in deposits of differing composition produced by different volcanoes. Direct comparison of the fresh volcanic ash substrates used here and the weathered terrestrial rocks should be made carefully as weathering alters nutrient availability and introduces mineral phases (e.g., palagonite; Kelly et al., 2010 ) absent in the fresh material. It may even be difficult to compare fresh ash surfaces and those of fresh lava flows (e.g., Kelly et al., 2014 ), as the surface of the latter is altered by crystallisation (Burkhard, 2002 ) and volatile degassing processes (White & Hochella, 1992 ) during and immediately after emplacement. However, if similar differences in bacterial communities depending on the silicate mineral substrate are found in our substrates as in terrestrial studies, our data would imply a shared mineralogy between the ash and glass substrate. As ash from Sakurajima volcano is predominantly glassy (70%–90%, Miwa, Geshi, & Shinohara, 2013 ), otherwise comprising plagioclase and mafic minerals, this may suggest that the glass component is driving the bacterial community structure of both volcanogenic substrates. As neither of the volcanic materials contain a significant quartz component, their different community structures relative to that of the terrigenic quartz would be consistent with the mineralogical dependence invoked by Gleeson et al. ( 2006 ) and Hutchens et al. ( 2010 ). Volcanic ash is characterised by a wide range of physical (morphology, particle size, surface area, colour) and chemical properties (mineralogy, composition), and volcanic ash deposits can contain a diverse array of particles, including entirely glassy or crystalline particles from fresh magma, remobilised older ejecta, and fragments of weathered lithic rocks. Furthermore, ash can deposit variably according to eruption dynamics as well as ash dispersal and sedimentation patterns with increasing distance from the volcano (Bonadonna, Costa, Folch, & Koyaguchi, 2015 ). Accordingly, there is the potential for significant variation amongst bacterial communities that establish on ash‐buried reefs at different positions relative to a source volcano as well as amongst different volcanoes. Hence, additional studies should be undertaken to disentangle the contribution of variable ash properties to bacterial colonisation as only one ash sample was considered here."
} | 5,162 |
28140394 | PMC5437920 | pmc | 7,909 | {
"abstract": "Marine and lake snow is a continuous shower of mixed organic and inorganic aggregates falling from the upper water where primary production is substantial. These pelagic aggregates provide a niche for microbes that can exploit these physical structures and resources for growth, thus are local hot spots for microbial activity. However, processes underlying their formation remain unknown. Here, we investigated the role of chemical signaling between two co-occurring bacteria that each make up more than 10% of the community in iron-rich lakes aggregates (iron snow). The filamentous iron-oxidizing Acidithrix strain showed increased rates of Fe(II) oxidation when incubated with cell-free supernatant of the heterotrophic iron-reducing Acidiphilium strain. Amendment of Acidithrix supernatant to motile cells of Acidiphilium triggered formation of cell aggregates displaying similar morphology to those of iron snow. Comparative metabolomics enabled the identification of the aggregation-inducing signal, 2-phenethylamine, which also induced faster growth of Acidiphilium . We propose a model that shows rapid iron snow formation, and ultimately energy transfer from the photic zone to deeper water layers, is controlled via a chemically mediated interplay.",
"introduction": "Introduction Pelagic aggregates form in the water column through adsorption of inorganic and organic matter, including bacteria, phytoplankton, feces, detritus and bio-minerals ( Alldredge and Silver, 1988 ; Simon et al. , 2002 ; Thornton, 2002 ). These aggregates, also called marine or lake snow, typically range in size from millimeters to centimeters ( Alldredge and Silver, 1988 ; Passow et al. , 2012 ) and are held together by extracellular polysaccharides ( Thornton, 2002 ; Giani et al. , 2005 ; Passow et al. , 2012 ). Marine and lake snow contribute substantially to the energy transfer from the photic zone to deeper water layers ( Suess, 1980 ). During the passage to the sediments, many motile bacteria and zooplankton colonize these organic-rich particles and remain attached as the aggregates sink through the water column to the sediment ( Fenchel, 2001 ; Kiørboe et al. , 2003 ). Particle-associated bacteria show higher extracellular enzymatic activities than free-living, planktonic microbes, resulting in rapid and localized turnover of organic matter and its subsequent release into the surrounding water ( Grossart and Simon, 1998 ). Pelagic aggregates are local hot spots for microbial interactions by direct cell contact, feeding activity and also the action of diffusible signals. For example, α - Proteobacteria , including Roseobacter strains, growing on the surface of marine snow produce elevated amounts of inhibitory molecules compared with planktonic forms in the surrounding aquatic environment ( Long and Azam, 2001 ). N -acyl homoserine lactones, compounds known to function as potential quorum-sensing mediators, are produced by Roseobacter strains ( Gram et al. , 2002 ; Zan et al. , 2014 ) and Pantoea sp. ( Jatt et al. , 2015 ) in marine snow communities, suggesting that quorum-sensing mechanisms might be at play. Chemical signaling molecules have also been suspected to regulate bacterial colonization, coordinate group behavior, as well as antagonistic activities within pelagic aggregates ( Dang and Lovell, 2016 ). However, the exact mechanisms for chemical communication and the identity of the compounds mediating such interactions in pelagic aggregates are poorly understood. The chemical diversity of exuded metabolites within such aggregates makes the identification and elucidation of infochemicals highly challenging, due to the loss of the diluted signals during the purification process ( Prince and Pohnert, 2010 ). However, comparative metabolomics has emerged as a powerful tool to overcome the limitations of bioassay-guided approaches by allowing the identification of metabolites produced by microorganisms grown in co-culture by comparison with those of single-strain cultures. Notably, upregulated metabolites are prime-candidates for mediators of microbial interactions ( Gillard et al. , 2013 ; Kuhlisch and Pohnert, 2015 ). In this study, we isolated two predominant bacterial key players from iron-rich lake aggregates (iron snow) to investigate chemical communication mechanisms involved in their interactions. Iron snow is characterized by lower microbial and chemical complexity in comparison with typical organic-rich marine and lake snow aggregates, and shows high sinking velocities because of its high relative fraction of iron ( Reiche et al. , 2011 ). We report that an active control by chemical signals shapes the association of the dominant microbial consortia within the iron snow and their behavior. Comparative metabolomics approach and structure elucidation led to the identification of bacterial extracellular exudates that function as allelopathic aggregate-inducing signal. These results clearly show that even during the very limited time of passage through the water column, interspecies chemical interactions between key organisms in iron snow enable these microorganisms to adhere to and colonize the particles.",
"discussion": "Discussion Previous studies involving the interactions of acidophilic microorganisms focused on gene regulation as the foundations for interactions or consumption of inhibitory compounds ( Liu et al. , 2011 ). Our study highlights the functional basis for microbial interactions by directly targeting the metabolites mediating these exchanges within iron snow aggregates. These chemical mediators not only increase the growth of Acidithrix and Acidiphilium in exchange experiments, but might also affect the rates of Fe-cycling in iron snow. We could show that the rate of Fe(II) oxidation by Acidithrix strain C25 was enhanced when grown in the presence of Acidiphilium strain C61 cell-free supernatant. In addition, we observed increases in the amounts of total nucleic acid in Acidithrix strain C25 cultures grown in presence of Acidiphilium strain C61 cell-free supernatant, suggesting that chemicals produced and secreted by Acidiphilium stain C61 positively affected activity and biomass production of Acidithrix strain C25. Similarly, the chemolithoautotrophic Acidithiobacillus ferrooxidans shows increased rates of Fe(II) oxidation and greater cell densities when co-cultured with the heterotroph Acidiphilium acidophilum ( Liu et al. , 2011 ). These effects were attributed to the activation of Fe(II) oxidation-related genes and a set of RuBisCO-encoding genes likely due to consumption of substrates by the heterotroph A. acidophilum that typically function to inhibit Fe(II) oxidation ( Liu et al. , 2011 ). However, this mechanism is not applicable in our supernatant exchange experiment. More likely, chemical mediators produced by the heterotroph Acidiphilium strain C61 and present in the cell-free supernatant stimulated growth of Acidithrix strain C25. PEA as the aggregation signal In most Acidiphilium species, cell motility is controlled by polar flagella ( Harrison, 1981 ; Wichlacz et al. , 1986 ). The flagellar-mediated cell motility typically prevents cell aggregation ( Caldara et al. , 2012 ). However, Acidiphilium stain C61 cells formed distinct macroscopic aggregates following addition of either Acidithrix strain C25, cell-free supernatant of this strain or pure PEA in concentrations comparable to those in cultures. This clearly demonstrates that PEA acts as a chemical mediator triggering cell aggregation by Acidiphilium stain C61 and thus the formation of iron snow. In another gram-negative bacterium, Proteus mirabilis , PEA is involved in the cell transition cycle ( Stevenson and Rather, 2006 ). PEA signals P. mirabilis cells to shift from swarming to vegetative state by inhibiting the assembly or activity of FlhDC, a key regulator of flagellin expression of lateral (peritrichous) flagellum systems and swarmer cell differentiation ( Stevenson and Rather, 2006 ). However, swarming activity of P. mirabilis cultures is inhibited by 50% at 1 m M PEA and complete inhibition was observed under the influence of 4 m M PEA. Inhibition of swarming in P. mirabilis induces cell transition into the vegetative phase, therefore PEA functions as a consolidation signal for the cells that lead to accelerated cell multiplication ( Matsuyama et al. , 2000 ). In our study, cell aggregation of Acidiphilium was induced following the addition of 5–10 μ M of exogenous PEA, which were in the concentration range of PEA detected in the supernatant of Acidithrix cultures. Our study also revealed that PEA induces faster growth of Acidiphilium cells at low in situ relevant concentrations of 10 μ M . The observed aggregation of Acidiphilium could be triggered by inhibition of flagellar-mediated cell motility by the presence of PEA. According to the genome database on Integrated Microbial Genomes ( Markowitz et al. , 2012 ), all Acidiphilium genera within the database lack the flhDC gene cluster, indicating the absence of a peritrichous flagellum system in Acidiphilium . The type strain of Acidiphilium cryptum was reported to possess one polar flagellum or two lateral flagella ( Harrison, 1981 ), which are supposed to be regulated by the sigma factor 54-dependent NtrC family of transcriptional activators for flagellum gene expression ( Arora et al. , 1997 ; Jyot et al. , 2002 ; Soutourina and Bertin, 2003 ). Differences in the flagellum regulatory systems between Acidiphilium strain C61 and P. mirabilis might explain the very low concentration of PEA necessary to inhibit Acidiphilium cell motility. Metabolomics analyses in our study detected the presence of PEA in Acidithrix cultures but not in Acidiphilium cultures even if they were supplemented with PEA containing supernatant of Acidithrix . In addition, PEA was not detected in cell-free controls amended with supernatant of Acidithrix after 10 days incubation. This suggests that PEA produced by Acidithrix and derived from their supernatant did not remain over the 10-day incubation period due to either abiotic degradation or consumption by Acidiphilium . However, this result also indicates that PEA was produced by Acidithrix cells over the entire bioassay. Microscopic images of Acidithrix and Acidiphilium grown in co-culture ( Figure 5 ) show that cell aggregation of Acidiphilium was induced in the presence of Acidithrix . The similarity of aggregate morphology in co-cultures and PEA-treated Acidiphilium , as well as the concentrations of PEA observed in cultures, supports this aromatic amine to be the exclusive active factor. Interestingly, in presence of Acidiphilium cell-free supernatant, PEA production by Acidithrix was lower than the PEA production observed in Acx ( Acx ) incubations. This result implies the presence of a two-way signal transduction system may potentially function to mediate chemical signaling when Acidithrix and Acidiphilium grow in close proximity to one another, thereby regulating the production, detection and utilization of PEA. Previous studies have reported PEA production by algae, plants, fungi and bacteria ( Irsfeld et al. , 2013 ). The biosynthesis of PEA typically occurs via decarboxylation of L -phenylalanine by the aromatic L -amino acid decarboxylase ( Irsfeld et al. , 2013 ). However, L -phenylalanine decarboxylation pathways and mechanisms in bacteria have not been well characterized, and only small groups of food fermentation-related gram-positive bacteria were reported to produce PEA by enzymatic activities also involving the decarboxylation of tyrosine to tyramine ( Pessione et al. , 2009 ; Marcobal et al. , 2012 ). According to the Integrated Microbial Genomes database, the type strain Acidithrix ferrooxidans Py-F3 lacks any homologous gene sequence encoding the tyrosine/phenylalanine decarboxylase genes found in either the genome of P. mirabilis or the genomes of food fermenting bacteria, possibly due to the incomplete genome sequence. The biosynthetic pathway responsible for PEA production in Acidithrix strain C25 has not been elucidated, so the exact mechanisms regulating the chemical signaling pathways in these two partner bacteria require further investigations. Model for the chemical communication between Acidithrix and Acidiphilium in iron snow Acidithrix strain C25 was suggested to be involved in early stages of iron snow formation via Fe(II) oxidation and subsequent aggregate formation associated with iron minerals ( Mori et al. , 2016 ), followed by colonization of Acidiphilium and other heterotrophs, chemolithoautotrophs and photoautotrophs on the aggregates. Apparently, motile Acidiphilium cells rapidly colonize microbe-mineral assemblages in the iron snow to obtain organic carbon sources as well as oxidized Fe, which can be used by the bacteria as an electron acceptor both under oxic and anoxic conditions ( Küsel et al. , 2002 ). According to our results, these microorganisms lose their motility, attach and ultimately form aggregates on the surface of iron snow particles under the influence of PEA produced by Acidithrix . PEA additionally acts as the consolidation signal that induces faster cell growth by the Acidiphilium cells associated with the iron snow aggregates ( Figure 6 ). Colonization of Acidiphilium benefits Acidithrix by inducing enhanced growth and Fe(II) oxidation rates. In summary, we show that PEA functions as the allelochemical between Acidithrix and Acidiphilium , which promotes rapid co-colonization onto the surface of the iron snow particles. Undoubtedly, the short residence time of iron snow in the water column of these shallow lakes ( Reiche et al. , 2011 ) does not allow all of the potentially complex aspects of biological activities and trophic interactions to occur, which is reflected by the community structure of the iron snow dominated by some groups of Fe(II) oxidizers and Fe(III) reducers ( Lu et al. , 2013 ). The interspecies cell signaling between the model isolates of this study may have evolved to allow both bacterial strains to capitalize on the strengths of the partner and to access the polycrystalline schwertmannite ( Miot et al. , 2016 ), the main Fe mineral of the iron snow aggregates. The interaction we observed can be attributed to the theory of microbial chemical communication, such that both the emitter and the receiver gain benefits through production, secretion and utilization of chemical signals mediating a beneficial partnership that can remain stable over evolutionary time ( Keller and Surette, 2006 ). The stability of the dominant bacterial community associated with iron snow particles ( Lu et al. , 2013 ) suggests the observed interactions between these bacteria have evolved specifically such that the Fe(II) oxidizer and Fe(III) reducer can thrive in acidic lakes. Future studies are expected to reveal the biosynthetic pathway and functional mechanisms of PEA, as well as its effect on other dominant bacterial species that comprise the iron snow surface-associated microbial community, such as Ferrovum sp. ( Lu et al. , 2013 ). These studies will provide further insight into interspecies cell–cell chemical communication in acidic, Fe-rich aquatic environments."
} | 3,852 |
26731732 | PMC4712138 | pmc | 7,919 | {
"abstract": "Though the use of metagenomic methods to sample below-ground fungal communities is common, the use of similar methods to sample plants from their underground structures is not. In this study we use high throughput sequencing of the ribulose-bisphosphate carboxylase large subunit (rbcL) plastid marker to study the plant community as well as the internal transcribed spacer and large subunit ribosomal DNA (rDNA) markers to investigate the fungal community from two wetland sites. Observed community richness and composition varied by marker. The two rDNA markers detected complementary sets of fungal taxa and total fungal composition clustered according to primer rather than by site. The composition of the most abundant plants, however, clustered according to sites as expected. We suggest that future studies consider using multiple genetic markers, ideally generated from different primer sets, to detect a more taxonomically diverse suite of taxa compared with what can be detected by any single marker alone. Conclusions drawn from the presence of even the most frequently observed taxa should be made with caution without corroborating lines of evidence.",
"introduction": "Introduction Fungi are important members of ecosystem functioning and play critical roles in nutrient cycling as symbionts, saprotrophs, and pathogens [ 1 ]. Below-ground mycorrhizal fungi in particular, may physically link the roots of different plant species and help to regulate plant diversity [ 2 – 3 ]. When monitoring fungal and plant communities from bulk soil using DNA-based methods, actively growing fungal mycelia and plant roots are detected as well as inactive propagules such as fungal sclerotia, plant rhizomes, spores, and seeds. However, even inactive portions of the below-ground community may have important future impacts. For example, fungal pathogens can affect the composition of the plant seed bank and subsequent plant recruitment [ 4 – 5 ]. Additionally, fungal mutualists and saprophytes in the fungal spore bank contribute to the rapid turnover of the microbial community in soils in response to disturbance or a change in seasons [ 6 – 9 ]. Due to the recalcitrance of many fungi towards cultivation using standard methods, and an abundance of vegetatively growing fungi with a paucity of characters for morphology-based identification, mycologists were early adopters of PCR-based detection and DNA-based identification methods [ 10 – 12 ]. Many fungal metagenomic studies using standard Sanger sequencing, and now high throughput sequencing, have been conducted in a variety of environments including bulk soil such as [ 7 , 13 – 15 ]. In contrast, PCR-based studies to monitor underground plant parts are rare [ 16 – 18 ]. Since plants and fungi co-exist in the same soil matrix these taxa can be studied in tandem to gain a more holistic understanding of below-ground communities in general and plant-fungal interactions in particular. The internal transcribed spacer (ITS) region of nuclear encoded ribosomal DNA (rDNA) has been proposed as a suitable fungal barcode [ 19 ]. The ITS region is comprised of the internal transcribed spacer 1 (ITS1), 5.8S rRNA gene, and the internal transcribed spacer 2 (ITS2) with the greatest sequence variation in the ITS1 and ITS2 regions. Several studies have examined the implications of using ITS for species identification using high throughput sequencing and have found that numerous methodological biases exist [ 20 – 26 ]. Despite these challenges, many ITS rDNA reference sequences are available in the AFTOL (Assembling the Fungal Tree of Life), UNITE, and GenBank sequence databases and tools have been developed to facilitate the use of ITS for fungal metagenomic studies [ 27 – 31 ]. Large subunit (LSU) rDNA contains variable domains at the 5’ end as well as highly conserved regions at the 3’ end suitable for taxonomically diverse phylogenetic analyses as well as species- to family-level classifications. LSU rDNA reference sequences are also available through the AFTOL, UNITE, and GenBank databases. LSU rDNA is particularly heavily sampled for mushroom-forming fungi [ 32 – 33 ] and has been used as a 'barcoding' marker for yeasts [ 34 – 35 ]. Previous fungal metagenomic studies of various soils have also used this region [ 36 – 38 ]. Similar to studies with ITS, methodological biases also exist with the use of LSU rDNA in metagenomic studies [ 39 ]. The ribulose-bisphosphate carboxylase large subunit (rbcL) plastid gene is one of two proposed plant barcoding markers [ 40 ]. This multi-copy protein-coding gene is relatively conserved and suitable for phylogenetic studies [ 41 ] and it has been shown to resolve species in 85% or more of cases when using BLAST against GenBank sequences [ 42 – 43 ]. Though the rbcL marker may not be able to identify all plants to the species level on its own, it was one of the first plant barcoding markers to be used in a multigene identification approach [ 43 ]. Because the diversity of plants was expected to be quite tractable compared to fungal diversity, we only used a single marker, rbcL, to survey plant diversity. The rbcL marker is well represented in the NCBI GenBank nucleotide database. Most metagenomic studies focusing on soil fungal communities involve the use of a single DNA marker. Because we knew that fungal diversity would likely be orders of magnitude higher than plant diversity in soil, we chose to use two fungal markers to increase our chances of detecting as much of this diversity as possible. To the best of our knowledge this is the first study to use two DNA markers (ITS + LSU) with largely fungal-specific primers as well as a plant-specific marker (rbcL) to monitor both the fungal and plant communities from the same soil samples simultaneously. We hypothesized that the fungal community detected by ITS and LSU rDNA would be largely similar, and that the use of the ITS + LSU + rbcL markers would together detect a richer assortment of organisms than any single marker. This study characterizes the reproducibility and taxonomic breath detected by these various markers and highlights areas of potential concern for future metagenomic and biomonitoring studies.",
"discussion": "Discussion Marker specificity Whole genome shotgun metagenomic approaches can utilize data from an array of markers selected a posteriori to track taxonomic groups of taxa. Using this approach previous work in the literature was able to track genus- to phylum-level Bacterial groups using six markers [ 65 ]. Though this method avoids the use of potentially biased primer-based amplification, it generates data from many loci that lack reference databases to allow species level identification. The alternative approach is to select markers a priori based on the availability of existing reference databases. Although the ability to link data from multiple markers to specific individuals is often lost using metagenomic methods, the data can be used to provide corroborating evidence for species presence and prevalence as we have done here. We hypothesized that the ITS and LSU rDNA markers would recover similar sets of fungal taxa and that the ITS + LSU + rbcL markers together would recover a richer assortment of taxa than any marker on its own. We produced thousands of OTUs from the ITS, LSU, and rbcL markers that we directly compared showing a significant “rare biosphere” [ 66 ]. We observed some similarity between the ITS and LSU datasets when taxa were compared at the most inclusive taxonomic levels, however, this similarity breaks down at more specific taxonomic levels even among the most frequently observed taxa. Each marker detected a taxonomically distinct community that varied more by primer than by site, particularly for the rDNA markers. To date, only a single fungal study that we are aware of has used more than one rDNA region, SSU and ITS, to survey hundreds of fungal sequence types from bulk soil using Sanger sequencing [ 13 ]. A previous fungal study also showed that using alternative primers can affect the recovered richness and community composition of root tips that were sequenced both individually and from a pooled sample [ 52 ]. Our study supports their assertion and shows how community richness, overall taxonomic composition, and even the presence of the most frequently encountered taxa may differ according to the primer and marker used for monitoring. Recent studies in arthropods have shown support for multiple primer and multiple gene frameworks [ 67 – 68 ]. Classification complexities How can we explain our inability to detect differences among sites using the ITS and LSU markers? First, fungi are significantly more diverse than plants and our fungal sampling was not exhaustive. Despite sequencing three soil sample replicates and producing saturated rarefaction curves, the use of additional primer sets for each marker would likely recover additional taxa [ 69 ]. Second, previous work has shown that partial sequences from the 5’ and 3’ ends of the ITS region may BLAST to different species despite coming from the same full length sequence. This type of BLAST result is often used to diagnose putative chimeras in full length ITS sequences [ 70 , 71 ]. Using a dataset of fungal environmental sequences previous work in the literature showed that 40% of partial ITS1 and ITS2 sequences from the same full length query may BLAST to different species [ 22 ]. Using a well-annotated fungal ITS dataset generated from individual PCRs, it was shown that partial sequences from the 5’ and 3’ ends of the same parent sequence had best BLAST matches to the correct species as well as to an incorrect species in 6% of cases for 400 bp fragments and in 15% of cases for 50 bp fragments [ 23 ]. These BLAST results may be best explained by lack of resolution among partial length ITS fragments, insufficient database coverage, or incorrectly annotated database sequences. The consequence of these observations is that taxonomic diversity recovered by the short fragments using different primers in our study may be inflated. Third, intragenomic variation among multicopy rDNA regions means that relaxed concerted evolution may result in sequences that are divergent from the consensus or barcode sequence for a species [ 72 – 74 ]. This type of variation can be detected from individuals by cloning and sequencing or from bulk soil DNA amplified with mixed-template PCR [ 75 ]. As a consequence, there is poor database representation for these rare alleles, and this may result in spurious BLAST matches to incorrect taxa. Fourth, the number of named fungal ITS sequences in GenBank available as references to identify new environmental sequences is greatly exceeded by the number of unnamed environmental sequences [ 76 ]. To improve the utility of reference databases, there has been a plea for increasing the sequencing of type cultures and specimens as well as for the formal classification of environmental sequences [ 76 – 78 ]. Progress towards automated sequence-based identification of fungal ITS sequences has been made [ 79 – 80 ]. As the representation in reference databases increases, so too will our ability to correctly classify taxa. Suggestions for future biomonitoring efforts It is possible that next-generation sequencing platforms producing longer paired-end reads up to 600 bp may be able to produce full length ITS sequences for most fungi to circumvent the problem of working with partial ITS reads. The use of paired-end approaches would also allow forward and reverse reads to be assembled, providing an additional level of quality assurance [ 81 ]. In contrast to the rDNA markers, the rbcL marker did not show strong clustering by primer, though species level identifications using MEGAN were not always possible due to the conserved nature of this gene region. As such, it may be appropriate to use a second marker such as matK to track below-ground plant structures and to corroborate rbcL results. The general rules for setting up mixed-template PCRs that detect the greatest sample diversity, particularly with 16S rDNA, have been known for some time and include using a low PCR cycle number, longer elongation times, and pooling multiple PCR reactions [ 82 – 84 ]. It is clear now that the use of multiple markers and even multiple amplicons for each marker, generated using different primers, may also be a good way to address the issue of primer bias and detect the broadest range of taxa from an environmental sample [ 69 ]. We suggest that future studies consider these parameters carefully since the high throughput nature of next-generation sequencing exaggerates these effects and even brute force sequencing will not detect maximum diversity if the primers and PCR conditions do not facilitate this. In conclusion, high throughput sequencing with multiple markers to study fungal and plant communities will be important for biomonitoring efforts such as in the Alberta oil sands."
} | 3,243 |
24348094 | PMC3855935 | pmc | 7,921 | {
"abstract": "It is often assumed that eukarya originated from archaea. This view has been recently supported by phylogenetic analyses in which eukarya are nested within archaea. Here, I argue that these analyses are not reliable, and I critically discuss archaeal ancestor scenarios, as well as fusion scenarios for the origin of eukaryotes. Based on recognized evolutionary trends toward reduction in archaea and toward complexity in eukarya, I suggest that their last common ancestor was more complex than modern archaea but simpler than modern eukaryotes (the bug in-between scenario). I propose that the ancestors of archaea (and bacteria) escaped protoeukaryotic predators by invading high temperature biotopes, triggering their reductive evolution toward the “prokaryotic” phenotype (the thermoreduction hypothesis). Intriguingly, whereas archaea and eukarya share many basic features at the molecular level, the archaeal mobilome resembles more the bacterial than the eukaryotic one. I suggest that selection of different parts of the ancestral virosphere at the onset of the three domains played a critical role in shaping their respective biology. Eukarya probably evolved toward complexity with the help of retroviruses and large DNA viruses, whereas similar selection pressure (thermoreduction) could explain why the archaeal and bacterial mobilomes somehow resemble each other.",
"conclusion": "12. Conclusion The Scenario I favoured in this paper for the origin and evolution of archaea is at odds with the traditional view that “prokaryotes” gave rise to “eukaryotes”. This traditional paradigm is so entrenched in our minds that it is not surprising that so many scientists endorse now “ fusion scenarios” or “ archaeal ancestor's scenarios” despite their many weaknesses. The confusing view that prokaryotes (assimilated to archaea and bacteria) predated eukaryotes (assimilated to modern eukaryotes) is inherent to the nomenclature “prokaryotes”, meaning “ before the nucleus” . This is only one of the drawbacks of using the term prokaryote. I agree on this point with Pace who has strongly advocated to completely repel the term prokaryote [ 149 ]. However, despite the work of Woese and his followers, the unfortunate term prokaryote is still widely used for its convenience and I use it myself in this paper (although between “brackets”). In some cases, indeed, it is useful to refer to archaea and bacteria as two groups sharing similar traits (the coupling of transcription and translation, for instance) that are characteristic of the “prokaryotic” phenotype. In the future, I will try to replace the term prokaryote by the neutral term “akaryote”, meaning without nucleus, that I proposed twenty years ago [ 150 ], a term that was reused recently by Harish and colleagues [ 91 ]. I proposed in the same 1992 paper to rename in parallel eukaryotes by the neutral term synkaryotes (with a nucleus). Indeed, I think today that it will be very difficult to get rid of the term “prokaryote” as long as we will use the term “eukaryote”. However, synkaryote, referring to a phenotypic trait, is not really adequate to name a domain, defined instead by genotypic traits [ 53 ]. At the moment, I am favouring the name Splicea , instead of eukarya, since possession of the spliceosome is a unique common trait to all “eukaryotes” derived from LECA. With this name, LECA becomes the LSCA and LAECA the LASCA, why not? Indeed, the origin (and fate) of the spliceosome(s) is, in my opinion, one of the more important questions in the history of life. If you like this novel nomenclature, you can change eukarya by Splicea and eukaryotes by spliceotes in this text, LECA by LSCA, and LAECA by LASCA and read it again with a fresh mind. The proposed hypotheses will possibly then seem less unorthodox to you.",
"introduction": "1. Introduction Archaea have been confused with bacteria, under the term prokaryotes, until their originality was finally recognized by 16S rRNA cataloguing [ 1 ]. Archaea were previously “hidden before our eyes”, strikingly resembling bacteria under the light and electron microscopes. Archaea and bacteria are also quite similar at the genomic level, with small circular genomes, compact gene organization, and functionally related genes organized into operons. At the same time, archaea, unlike bacteria, exhibit a lot of “eukaryotic features” at the molecular level [ 2 – 6 ]. It is often assumed that archaea resemble eukarya when their informational systems (DNA replication, transcription, and translation) are considered but resemble bacteria in terms of their operational systems. This is clearly not the case, since many archaeal operational systems (such as ATP production, protein secretion, cell division and vesicles formation, and protein modification machinery) also use proteins that have only eukaryotic homologues or that are more similar to their eukaryotic rather than to their bacterial homologues [ 7 – 14 ]. The bacterial-like features of some archaeal metabolic pathways could be mostly due to lateral gene transfer (LGT) of bacterial genes into Archaea, driven by their cohabitation in various biotopes [ 15 ]. Indeed, beside bacterial-like genes possibly recruited by LGT, metabolic pathways in archaea—such as the coenzyme A or the isoprenoid biosynthetic pathways—also involve a mixture of archaea-specific and eukaryotic-like enzymes [ 16 – 18 ]. Archaea and eukarya share so many features in all aspects of their cellular physiology and molecular fabric that eukaryotes cannot be simply envisioned as a mosaic of archaeal and bacterial features. Archaea and eukarya clearly share a more complex evolutionary relationship that remains to be understood. Whereas many eukaryotic traits of archaea are ubiquitous or widely distributed in that domain, recent discoveries have identified several new eukaryotic traits that are only present in one phylum, one order, or even in one species of archaea [ 6 , 11 , 14 ]. Phylogenetic analyses suggest that these traits were already present in the last archaeal common ancestor (LACA) since, in most cases, archaeal and eukaryal sequences form two well separated monophyletic groups [ 12 , 13 , 19 ]. This indicates that these traits have not been sporadically acquired from eukarya by lateral gene transfer but were lost in most members of the archaeal domain after their divergence from LACA [ 6 ]. Considering this loss of eukaryotic traits and the gain of bacterial traits by LGT, LACA was probably even more “eukaryotic-like” than modern archaea. However, despite their eukaryotic affinity, archaea lack many eukaryote-specific features (ESFs) at the cellular and/or molecular levels. These, for example, include the spliceosome, mRNA capping, and extensive polyadenylation as well as huge transcriptional machineries with unique components, such as the mediator, endoplasmic reticulum, and derived structures such as lysosomes, the Golgi apparatus and the nuclear membrane, an elaborated cytoskeleton and associated vesicle trafficking system with endosomes and ectosomes, nuclear pores, nucleolus and other nucleus-specific structures, linear chromosomes with centromeres and telomeres, mitosis and associated chromosome segregation system linked to the cytoskeleton, complex and great sex with meiosis derived from mitosis, an incredible machinery for cell division apparatus with synaptonemal complex for meiosis, centrioles and midbodies for cell division, and I probably miss some of them. Archaea not only lack all these ESFs but also lack homologues of most proteins (a few hundreds) that are involved in building and operating them [ 20 ]. This remains true even if a few ex-ESPs (e.g., actin, tubulin, and DNA topoisomerase IB) have recently lost this status following the discovery of archaeal homologues [ 12 , 13 , 19 ]. The number, diversity, and complexity of ESFs are impressive and their origin remains a major evolutionary puzzle that should not be underestimated. The puzzle became of even greater magnitude when it was realized during the last decade from phylogenomic analyses that all ESFs (and associated ESPs) were most likely already present in the last common ancestor of all modern eukaryotes, ( the last eukaryotic common ancestor ) (LECA) [ 20 ]. In a recent review, Martijn and Ettema called the period that experienced the emergence of ESFs (so before LECA): “the evolutionary dark ages of eukaryotic cells ” [ 21 ]. This denomination well illustrates the complexity of the “complexity problem” in eukaryotic evolution. Besides lacking (by definition) all ESFs, archaea also fundamentally differ from eukarya in the nature of their membranes (with a unique type of lipids in archaea), and the type of viruses infecting them. The problems raised by the evolution of membranes have been nicely reviewed recently by Lombard et al. and I will refer to their work later on to discuss different models for the origin of archaea [ 22 ]. In contrast, the problem raised by the drastic differences between archaeal and eukaryotic viruses has never been really discussed. For instance, Martijn and Ettema never mentioned the word virus in their review on the origin of eukaryotes [ 21 ]. Viruses are also completely absent from the papers of Cavalier-Smith or Carl Woese himself. This is probably because, as recently stated by Koonin and Wolf, “ viruses are no part of the traditional narrative of evolutionary biology ” [ 23 ]."
} | 2,355 |
34584090 | PMC8478921 | pmc | 7,922 | {
"abstract": "Plant-soil feedbacks are shaped by microbial legacies that plants leave in the soil. We tested the persistence of these legacies after subsequent colonization by the same or other plant species using 6 typical grassland plant species. Soil fungal legacies were detectable for months, but the current plant effect on fungi amplified in time. By contrast, in bacterial communities, legacies faded away rapidly and bacteria communities were influenced strongly by the current plant. However, both fungal and bacterial legacies were conserved inside the roots of the current plant species and their composition significantly correlated with plant growth. Hence, microbial soil legacies present at the time of plant establishment play a vital role in shaping plant growth even when these legacies have faded away in the soil due the growth of the current plant species. We conclude that soil microbiome legacies are reversible and versatile, but that they can create plant-soil feedbacks via altering the endophytic community acquired during early ontogeny.",
"introduction": "Introduction Soil microbes are widely acknowledged to be major drivers of plant growth and plant community assembly 1 . Plants affect soil microbes via the quantity and quality of rhizodeposits 2 , 3 , and litter 4 . These plant-mediated changes in the soil microbiome can influence the growth of other plant species that grow later in the same soil (plant-soil feedbacks 5 – 7 ). Plants can negatively influence succeeding plant species through accumulation of pathogenic microbes in the soil 7 – 10 , or positively through the build-up of beneficial or mutualistic microbes 10 – 12 . These microbiome-mediated plant-soil feedbacks may be general among functional groups of plants (e.g., grasses and forbs) as these groups markedly differ in their effects on - and sensitivity to - soils. While microbial soil legacies can have major impacts on plant growth 2 , we are still far from understanding and predicting these legacy effects. Specifically, we do not know how persistent soil legacies are (i.e., how long they last after the removal of the plant), and if these microbial legacy effects vary between different plant species that both shape and respond to soil legacies 13 , 14 . While a ‘current plant’ grows in soil conditioned by a ‘previous plant’, it will respond to biotic and abiotic soil conditions, but simultaneously change the microbial legacy in the soil. How these temporal changes contribute to the overall outcome of plant-soil feedbacks is not well understood. The specific influence of the previous plant on the soil community is a widely held assumption behind plant-soil feedback theories and experiments, whereas the effect of the newly created soil legacies influenced by the current plant and the combination of the two types of legacies, is often overlooked and lacks rigorous empirical testing. The sensitivity of a plant to the soil microbial community may vary depending on the age of the plant 15 – 17 and generally, seedlings are considered to be more sensitive to for example pathogen effects than adult plants 18 . Besides having an inherited seed microbiome from its parental plants 19 , freshly germinated seedlings can experience only the soil legacy of the previous plant, whereas older plants will experience soils that bear a legacy of previous plants, but which may have also been modified by themselves. Interestingly, a recent study proposed that the soil microbial community present at the plant germination stage may be a stronger determinant of plant growth than the soil microbial communities that are present at later ontogenetic plant stages 20 . Moreover, studies suggest that seedlings are more susceptible to endophytes (i.e. microbes living inside plant roots) colonizing the roots than adult plants 21 , which may be due to the low levels of chemical defenses in younger plants 19 or their greater need for symbiotic partners to survive. Endophytic microbes living inside the roots are in closer contact with the plant than the microbes in the soil 22 . Multicellular fungi may simultaneously grow hyphae in rhizosphere soil and in the endosphere, whereas unicellular bacteria cannot, which may lead to differences between the two microbial kingdoms 23 , 24 . Endophytes can be beneficial for plant growth through their effects on plant nutrient status, through the protection they provide against pathogens and pests, and via increasing stress tolerance and modulation of plant development 23 – 27 . Plants inherit endophytic microbes through transfer of microbes from parental plants in seeds 19 but also select their own endophytic microbes from the pool available in the soil 24 , 27 and as such, the community structure of endophytes within a plant species is known to differ between soils with a different history 28 , 29 . The plant-mediated legacy of a soil may thus affect the endophytes the plant acquires which, in turn, can affect plant growth and performance 30 . Yet, it is largely unclear how the identity of the previous plant and plant traits affect the composition of endophytes across plant species. As many endophytes are acquired at early growth stages and often remain in the plant throughout its growth, this suggests that exposure to soil legacies of a previous plant early in life can have long-lasting effects on plant growth, even when these legacies are no longer detectable in the soil surrounding the plant root. To examine the persistence of plant-specific soil microbial legacies during the next generation of plant growth, we set up a long-term mesocosm common garden experiment with six plant species, belonging to grasses and forbs, that are commonly found in former agricultural grasslands, all known to form a symbiosis with arbuscular mycorrhizal fungi (AMF) and of which plant-soil feedbacks have been well-studied over the past two decades 31 , 32 . As grass species all belong to the same family, but forbs do not, we selected forb species from one family as well to make our design phylogenetically balanced. The three selected grasses were Alopecurus pratensis , Festuca ovina , and Holcus lanatus (all Poaceae) and the three selected forbs were Hypochaeris radicata , Jacobaea vulgaris , and Taraxacum officinale (all Asteraceae). We first created six distinct soil microbial legacies by growing the plants as monocultures in 200-L soil mesocosms for 12 months 33 . We then divided each mesocosm into six physically separated sections (by placing soil in buckets), in which we planted all the six responding plant species (see Fig. 1 for set-up). We monitored the soil microbiome in each section by non-destructive repeated sampling for five months and examined changes in the microbiome (bacteria and fungi) caused by the previous and the current plant over time. After 5 months of plant growth, we destructively harvested the plants to examine their responses to the soil legacies and analyzed the root endophytic microbiome 22 , 29 . Fig. 1 Set-up of the experiment. In short, monocultures of six plant species were maintained for 12 months in 30 (6 species × 5 replicates) mesososms. Then, the top soil was divided into six smaller containers and placed back into the original containers (in mid-May) See Supplementary Movie 1 for how this was done. The existing plants were removed and after three weeks (in June) four seedlings of one of the same six plant species were planted in each of the six smaller containers reciprocally. This equaled to (30 mesocosms × 6 plants) 180 smaller mesocosms. Soil from each of these smaller mesocosms were sampled one month (July), three months (September) and five months (November) after planting the plants. Soil was sampled so that four soil cores were taken per mesocosm, one sample next to each plant and combined into one composite sample per small mesocosm. After five months, the containers were destructively harvested, plant biomass was measured from dried and washed root and shoot material, and endophytic microbial communities were surveyed from sterilized root samples. Statistical models used for the evaluation of current and previous plant family and the effect of growing in own soil (home-away) effects. Same models were used for both plant and microbial data. See text for details. Plants create directional changes in soil microbiomes that differ between plant species and their functional group 33 – 35 . Previous work has shown that bacterial soil legacies have a faster turnover time than fungal legacies 33 , 36 due to differences in turnover rates and traits related to microbial growth strategies 37 . Following these general assumptions, we tested the following hypotheses: (i) The soil microbial legacy of the previous plant will diminish over time, while the effects of the current plant on the soil microbial community will concomitantly increase with time (Fig. 2 ). Furthermore, we expect that the legacy effects will be detected for longer time periods in the fungal than in the bacterial community. Fig. 2 Theoretical change in microbiomes over time. Theoretical framework of the effects of conditioning (previous) and responding (current) plants on soil microbial community composition. A we expect that the microbiome will change in time under a new plant community in either a directional, constant, or fluctuating way. B Furthermore, if this change is directional, we expect that the change will lead to convergence in microbiomes and that over time the microbiome will develop into a species-specific microbiome type but that this will depend on the initial microbiome composition and hence on the legacy of the plant that grew in the soil before. (ii) As the endophytes are acquired by plants in early growth stages, the legacy effect of the previous plant, even though it may not be detected anymore in the soil, will still be visible in the endophytic root microbiomes of the current plant. (iii) Endophytic microbes will have stronger relationships with plant growth than rhizosphere microbes. (iv) Conspecific feedbacks are due to an accumulation of specific communities, microbial groups, functional guilds (such as mutualists or pathogens), or microbial species in the endosphere and rhizosphere To test these hypotheses, we analyzed both bacterial and fungal communities in the soils in the beginning of the experiment and at three time points during plant growth and inside the roots at final harvest, and relate the community composition of microbes in soils conditioned by different previous plant species and families to the plant biomass of the current plant at the final harvest. We show that the legacy of the previous plant stays lingering in the soil fungal community while in soil bacterial community the memory fades away quickly. Yet, both legacies are stored inside the plant roots and affect the growth of the following plant.",
"discussion": "Discussion We show that plant-mediated effects on the soil microbiome are reversible, but also that soil legacies from previous plants at species and plant family level can be detected in the soil fungal community for at least 5 months after removal of the previous plant and subsequent colonization of the same soil by different plants. The effect of the previous plant on soil fungal community structure appears to be larger than the effect of the current plant species. This is important, as soils are dynamic and every soil arguably has a pre-existing legacy caused by previous plants or plant communities. Our results indicate that when a plant arrives or is planted into the soil, even months after growing in this soil, it may still experience the microbial legacy effects created by the plants that grew previously in that soil. We can only speculate how long it will take before the legacy of the previous plant in the soil fungal community has disappeared entirely, as the 5 months of ‘current’ plant growth following one year of ‘previous’ plant conditioning was not enough for these soil legacy effects to fade away. This finding has important consequences for plant-growth experiments using field collected soils with previous legacies, but also for understanding plant community dynamics in natural and anthropogenic ecosystems, and suggests that the legacy of previous plants or plant communities on the soil microbiome lasts for months after new plants colonize the soil, something that has been previously overlooked. For soil fungi, the effects of the previous plant on the soil community outweighed the effects of the current plant, while for soil bacteria, each plant quickly modified its own rhizosphere microbiome although the influence of the previous plant was still detectable. These findings are in line with the conceptual idea that fungal growth rates are slower than those of bacteria 36 , 37 and that because of this, fungi are more stable and less affected by, for instance, temporal variability in the habitat or environment 33 . Importantly, the more persistent effects of plants on the soil fungal communities than on bacterial communities may explain why most correlative studies that link plant responses and changes in the soil microbiome have shown that fungal communities drive plant community dynamics, while soil bacteria do not seem to strongly influence plant-soil-feedbacks 7 , 8 despite their known importance in rhizosphere processes 1 , 9 . This is potentially due to faster turnover times of soil bacterial communities 36 . Furthermore, due to their hyphal growth form, many fungi can simultaneously grow inside the roots and in the rhizosphere environment 38 , and soils could therefore potentially predict intimate active fungal-plant relationships better than bacterial-plant relationships. Another possible option is that not all organisms detected with DNA-based methods are active and thus part of the signal we are detecting originates from dead cells or inactive organisms 39 . Alternatively, bacterial DNA could potentially be recycled quicker than fungal DNA which is protected within the hyphae, but this needs further testing. However, recently it has been shown that 80% of fungi detected in the rhizosphere in a similar grassland system were actively participating in recycling plant-derived carbon 40 . Interestingly, the effects of the previous and current plant, especially on fungi, were conserved between the two plant families. Here, we show that even after 5 months, the fungal communities in soils in which currently Poaceae grow with a legacy of previous Poaceae growth differ from the fungal communities in soils in which Poaceae grow but with a legacy of Asteraceae. Similarly, ‘current Asteraceae’ soils with a legacy of Poaceae had very different fungal communities than ‘current Asteraceae’ soils with a legacy of Asteraceae. The differences between the two plant families, which represent two distinct growth forms confirm and build on previous findings on the role of plant functional groups in plant-soil feedbacks. For instance, meta-analyses show that the dichotomy between grasses and forbs generally creates robust soil legacy effects e.g. 41 . Furthermore, other recent work has shown that plant family and functional group can explain a large portion of the variation in fungal community structure 33 , 42 , 43 . Functional traits as well as growth and nutrient acquisition strategies 44 – 46 and chemical defenses 47 , 48 of the selected plant species likely have more similarities within than between groups. This may explain why the two groups have a driving effect on plant-soil feedbacks in plant communities, in greenhouses, and field studies 7 , 35 . An important outcome of our study is that soil legacies are taken up in early life stages of the plant 49 and remain present inside the roots of the plants when they grow in the soil and by this change the soil microbiome. This is supported by the observation that both the bacterial and the fungal endophyte community sampled after 5 months of plant growth reflect the legacy of the previous plant. Hence, besides the seed-microbiome the plant inherits from its parental plant 19 , also the legacies in the soils where it lands shape its endophytic microbiome. This is important, as endophytic bacterial communities are generally more tightly linked to plant performance than soil and rhizosphere bacterial communities 26 , while for fungi both rhizosphere soil and endophytic communities influence plant performance to a similar extent due to their ability to bridge endophytic and soil environments 24 , 50 , 51 . Here we confirm that the composition of the endophytic microbes is specific to the plant species carrying them partly due to inherited seed microbiome 19 , 27 , 52 but also show that endosphere microbiome of the current plant depends on the previous plant that grew in the soil. Bacterial endophytes, and especially Actinobacteria and Patescibacteria , were modulated by the legacy of the previous plant, and influenced current plant growth. The role of Patescibacteria in plant health is still unclear, but they have been recently found inside the tissues of different plants 29 , 53 . However, Actinobacteria , and especially Streptomycetes , are often detected inside plant roots and can act both as pathogens and can be beneficial to the plant 54 , 55 . Here, we show that root biomass increased across plant species when the relative abundance of Actinobacteria inside the roots increased suggesting a generally positive role of these microbes in influencing plant growth. Interestingly, we also show that all plant species had fewer Actinobacteria , and especially Streptomycetes inside their roots when grown in soils with a legacy of their own species. Endophytic Streptomycetes have been found to be strongly selected by their host plant, they play a role in the magnitude of plant-soil feedbacks and help the plant-host cope with drought 25 . We speculate that plant species-specific selection on soil microbes, such as Streptomycetes , may provide an interesting avenue for further investigation on the role of these endophytic bacteria on conspecific plant-soil feedbacks. We observed a trade-off between the relative abundance of AMF inside plant roots and the biomass of the respective root system, but not shoot biomass. Plants with higher relative abundance of AMF in their roots had a lower root biomass, probably due to a decreased necessity to scavenge for nutrients and especially phosphorus in the presence of AMF 56 . We detected a reduction in the relative abundance of AMF when growing in their own soil only for one plant species and one time point and the previous plant identity or family only had a minor effect on the community structure of AMF within the roots. As we measured relative abundance we cannot draw conclusions from these results on the role in plant-soil feedbacks. Other studies have shown that AMF explain plant-soil feedbacks, through mutualistic relationships 13 , 57 , and that plant soil feedbacks distinctly differ between plants that form arbuscular mycorrhizal interactions and plants with other microbially mediated nutrient acquisition strategies 58 . The species that were used in this study, generally have negative conspecific plant-soil feedbacks 15 and also all form arbuscular mycorrhizae; they all are equally benefitting from association with AMF and the identity of the fungal partners seems not to be important. Alternatively, it is possibly that a potential links between AMF and plant-soil feedbacks, can be detected when focusing on AMF colonization rather than focusing on community composition as was done in the current study. While we here focused on changes in community structure, further studies should also measure absolute abundances and activity of other microbes. All six plant species exhibited strong negative conspecific feedbacks. The variance in microbes explained by growing in conspecific soils was highest at 3 months of growth, and declined substantially after, highlighting a temporal dimension in plant-soil feedback effects, which we speculate to be due to dynamics in the soil, or decreased plant sensitivity with older age 18 . We observed only in Poaceae that feedbacks were due to accumulation of potential plant pathogens in the soil and inside roots when they were grown in soils with a legacy of their own species. The soil fungal community as a whole, and specific plant pathogenic fungi, have recently been shown to modulate plant community dynamics in grasslands 7 , 8 , 59 . Here we show that Poaceae increase the relative abundance of plant pathogenic fungi in the soil when grown in monocultures. More importantly, we show that different Poaceae species accumulate specific fungal pathogens in their roots that can, in turn, cause negative plant-soil feedbacks when these Poaceae are grown in conspecific soils. In all six soil legacies, the soil fungal communities differed when a plant was grown in its own soil, from the communities formed when grown in a soil with a legacy of another plant, and when this effect was largest, depended on plant species. This is in line with work on invasive plant species and their endophytes showing species-specific effects and acquisition of endophytes 30 . Some of the plants in this experiment showed a strong relationship with soil microbial communities at the onset of the experiment, while for others the growth was related more strongly to the current microbiome in the soils, which still contained a detectable legacy of the previous plant’s soil microbiome. The variability in the magnitude of plant-soil feedbacks between plant life stages has been noted earlier 16 , 18 and here we offer a microbial background to this phenomenon. A potential caveat of our study is that we did not measure soil chemistry-mediated legacy effects, which have been shown to play a role mediating plant legacy effects in some studies 13 . However, our recent work in similar systems and conditions revealed no significant role of soil chemistry in driving legacy effects in plant communities 7 . Therefore, in the current study we focused on relationships between microbial composition of soil and root endophytic microbiomes and plant growth. Yet, we acknowledge that plant-soil feedbacks are driven by both soil abiotic and biotic factors, and that these often act in synchrony 10 , 11 . On the basis of our results, we propose a rethinking of how soil microbiome-mediated legacy effects work. First, no soil with plants growing in it is naive and without a legacy. Therefore, in order to evaluate what are the main factors predicting plant and especially crop growth, we need to look into the history of the soil and importantly also evaluate the plant ‘holobiome’ 26 , 60 . We show that part of the microbial legacy effect is plant-species specific while another part of this effect is plant family-specific. Especially, Poaceae generally have negative effects on other Poaceae growing in the same soils while the effects on Asteraceae are mainly plant species-specific. In a wider perspective, our results show that soil and root microbiomes are important for plant growth and that plants can be used to directionally change the microbiomes and hence steer plant growth. In conclusion, our study shows that soil legacies wrought by previous plants can remain present in the soil for months, even when subsequent plants colonize (and condition) the soil. Bacterial communities change quicker than fungal communities and our findings suggest that plants take up microbes and especially bacteria from these pre-existing soil legacies in their endophytic compartments at a very early seedling stage only, and that these endophytes may play a more prominent role in driving plant performance than the microbiome present around the roots of the older plants. Our study also highlights microbial taxa that consistently drive negative conspecific plant-soil feedbacks across plant-species, and characterizes the role of these species in plant-soil feedbacks thereby providing an exciting venue for further research."
} | 6,033 |
36446702 | null | s2 | 7,924 | {
"abstract": "A dynamic field of study has emerged involving long-range electron transport by extracellular filaments in anaerobic bacteria, with Geobacter sulfurreducens being used as a model system. The interest in this topic stems from the potential uses of such systems in bioremediation, energy generation, and new bio-based nanotechnology for electronic devices. These conductive extracellular filaments were originally thought, based upon low-resolution observations of dried samples, to be type IV pili (T4P). However, the recently published atomic structure for the T4P from G. sulfurreducens, obtained by cryo-electron microscopy (cryo-EM), is incompatible with the numerous models that have been put forward for electron conduction. As with all high-resolution structures of T4P, the G. sulfurreducens T4P structure shows a partial melting of the α-helix that substantially impacts the aromatic residue positions such that they are incompatible with conductivity. Furthermore, new work using high-resolution cryo-EM shows that conductive filaments thought to be T4P are actually polymerized cytochromes, with stacked heme groups forming a continuous conductive wire, or extracellular DNA. Recent atomic structures of three different cytochrome filaments from G. sulfurreducens suggest that such polymers evolved independently on multiple occasions. The expectation is that such polymerized cytochromes may be found emanating from other anaerobic organisms."
} | 363 |
36840135 | PMC9965935 | pmc | 7,928 | {
"abstract": "The stability and harmony of ecological niches rely on intricate interactions between their members. During evolution, organisms have developed the ability to thrive in different environments, taking advantage of each other. Among these organisms, microalgae are a highly diverse and widely distributed group of major primary producers whose interactions with other organisms play essential roles in their habitats. Understanding the basis of these interactions is crucial to control and exploit these communities for ecological and biotechnological applications. The green microalga Chlamydomonas reinhardtii , a well-established model, is emerging as a model organism for studying a wide variety of microbial interactions with ecological and economic significance. In this review, we unite and discuss current knowledge that points to C. reinhardtii as a model organism for studying microbial interactions."
} | 227 |
27261087 | null | s2 | 7,929 | {
"abstract": "Recapturing atmospheric CO2 is key to reducing global warming and increasing biological carbon availability. Ralstonia eutropha is a biotechnologically useful aerobic bacterium that uses the Calvin-Benson-Bassham (CBB) cycle and the enzyme ribulose 1,5-bisphosphate carboxylase/oxygenase (RubisCO) for CO2 utilization, suggesting that it may be a useful host to bioselect RubisCO molecules with improved CO2 -capture capabilities. A host strain of R. eutropha was constructed for this purpose after deleting endogenous genes encoding two related RubisCOs. This strain could be complemented for CO2 -dependent growth by introducing native or heterologous RubisCO genes. Mutagenesis and suppressor selection identified amino acid substitutions in a hydrophobic region that specifically influences RubisCO's interaction with its substrates, particularly O2 , which competes with CO2 at the active site. Unlike most RubisCOs, the R. eutropha enzyme has evolved to retain optimal CO2 -fixation rates in a fast-growing host, despite the presence of high levels of competing O2 . Yet its structure-function properties resemble those of several commonly found RubisCOs, including the higher plant enzymes, allowing strategies to engineer analogous enzymes. Because R. eutropha can be cultured rapidly under harsh environmental conditions (e.g., with toxic industrial flue gas), in the presence of near saturation levels of oxygen, artificial selection and directed evolution studies in this organism could potentially impact efforts toward improving RubisCO-dependent biological CO2 utilization in aerobic environments. d-ribulose 1,5-bisphosphate carboxylase/oxygenase, EC 4.1.1.39; phosphoribulokinase, EC 2.7.1.19."
} | 427 |
31127161 | PMC6534613 | pmc | 7,930 | {
"abstract": "Wetting phenomena, i.e. the spreading of a liquid over a dry solid surface, are important for understanding how plants and insects imbibe water and moisture and for miniaturization in chemistry and biotechnology, among other examples. They pose fundamental challenges and possibilities, especially in dynamic situations. The surface chemistry and micro-scale roughness may determine the macroscopic spreading flow. The question here is how dynamic wetting depends on the topography of the substrate, i.e. the actual geometry of the roughness elements. To this end, we have formulated a toy model that accounts for the roughness shape, which is tested against a series of spreading experiments made on asymmetric sawtooth surface structures. The spreading speed in different directions relative to the surface pattern is found to be well described by the toy model. The toy model also shows the mechanism by which the shape of the roughness together with the line friction determines the observed slowing down of the spreading.",
"introduction": "Introduction How do the detailed properties of a dry surface influence the speed at which a liquid spreads over it? From splashing around solid objects in water, to the way plants and insects imbibe water and moisture, surface tension and wetting of solid surfaces are essential 1 – 3 . In technology, miniaturization in chemistry and biotechnology has generated strong interest in wetting phenomena, as a means to control and manipulate small samples and droplets 4 , 5 . Inspired by these, droplets impacting or moving over dry surfaces have recently received considerable attention 6 – 10 . Such motions may be controlled by the microscopic surface structure, both in terms of geometric features and variations in surface energy. An additional complexity is introduced when the microscopic surface structure is highly anisotropic. In nature this is found for instance on the feet of water striders, which are covered with hydrophobic hairs, the orientation of which is important for the locomotion of the insect 11 , 12 . In technology there are many situations where it is desirable to control the motion of droplets, and anisotropic surfaces can be prepared to achieve this, see for instance 13 , 14 . The understanding of the physics of a moving contact line, the intersection of a fluid interface and a solid, is still evolving (See the recent reviews 15 – 17 ). It is a classical problem in fluid dynamics that the equations for viscous flow produce a non-integrable singularity of the viscous stress at the wall 18 . The root of this problem is that micro- and nano-scale processes at the contact line may have macroscopic effects and determine the spreading speed. For the late spreading of a droplet towards equilibrium, the classical Tanner’s law 19 introduces an adhoc cutoff length and assumes that viscous dissipation in the bulk liquid near the contact line provides the dominant resistance. In the generic situation of a droplet that is placed on a dry surface, spreading due to surface tension and the wetting properties of the surface, it has in many situations been found experimentally 20 – 23 that the fast early spreading is dominated by inertia and can be surprisingly insensitive to the substrate properties. Very recently, experiments have also shown that the early spreading may have a first viscous regime, also for not very viscous liquids 24 . The final stages of wetting, when the droplet dynamics has become quasi-static and the contact angle is close to its static value, are generally described well by Cox-Voinov or Tanner’s law, indicating that the viscous flow in the wedge shaped region near the contact line determines the speed 25 , 26 . See for instance Kant et al . 27 for a study of spreading over a complex substrate geometry in this parameter range. Here we will be concerned with the rapid initial motion which is further from equilibrium. Approaching wetting from the molecular point of view, another line of research discusses the resistance to wetting in terms of a non-hydrodynamic energy dissipation at the contact line 28 – 33 , that can be empirically quantified in terms of the contact line friction parameter μ f (in Pascal seconds, Pa s) 28 , 1 \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$\\begin{array}{c}w \\sim \\,{\\mu }_{f}{U}^{2}\\end{array}$$\\end{document} w ~ μ f U 2 where U is the contact line speed, and w the energy dissipation rate associated with molecular processes at the contact line, per unit length. In the Molecular Kinetic Theory (MKT) 28 – 30 , dynamic wetting is described as an activated process on the molecular scale, and the line friction μ f is given a phenomenological interpretation on the molecular scale. On a macroscopic scale, the consequence of the presence of contact line friction is that there is a relation between the dynamic contact angle and the contact line speed. Viewed as a macroscopic phenomenological parameter, μ f may be connected to the ‘Hocking parameter’ 34 , 35 , which is the assumed constant of proportionality between the deviation of the dynamic contact angle from its equilibrium value and the contact line speed. Values of μ f can be obtained from straightforward spreading experiments 36 in conjunction with numerical simulations. The recent studies present independent measurements of line friction values 33 , 37 . Vo and Tran 33 clarified the dependency of the contact line friction on the viscosity both of the droplets and the surrounding oils. Xia and Steen 37 report measurements using water – glycerol mixtures on silicon substrates treated to be partially wetting and to have low contact angle hysteresis. The mobility in their case is the constant of proportionality between the deviation of dynamic contact angle from the static value, and the contact line speed. This can readily be translated into the line friction μ f used here. We have not performed tests on the exact same surfaces as theirs, but note that the numbers they report give values of μ f in the range 0.17–0.53 Pa s, increasing with increasing glycerol content, similar to our values. Recently contact line friction was also measured for very viscous polymer melts 38 , which reveals a line friction proportional to the bulk viscosity, with a constant of proportionality greater than one so that the contact line friction is always greater than the bulk viscosity. We will not be concerned here with the molecular origin of μ f , but rather consider it as a material parameter, characteristic of the combination of the particular liquid and the solid surface. We are considering however the role of line friction in the fluid dynamics of wetting, and in particular how the influence of the details of the substrate properties and geometry on dynamic wetting can be understood in terms of this. In spreading over rough surfaces, it was found that the effective line friction coefficient µ f that characterizes the macroscopic spreading can be written as 2 \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$\\begin{array}{c}\\,{\\mu }_{f}=\\,{\\mu }_{f,flat}S\\end{array}$$\\end{document} μ f = μ f , f l a t S where S denotes the ratio of the total wet surface area of the structure to the projected footprint area, and μ f , flat is the line friction obtained on a (smooth) perfectly flat surface of the same material 36 . The total line friction could thus be separated into one surface-chemistry factor μ f , flat and one purely geometric characterization of the roughness, S . In experiments on droplets impacting structured surfaces 39 , 40 , it was found that the maximum radius of the droplet during impact depended on the surface structures with S calculated for the respective substrate topography. As noted above, line friction, fluid inertia or bulk liquid viscosity can each be the limiting factor for spreading 21 , 25 , 26 , 36 , 41 , depending on the process and the material properties. In order to determine the relative importance of inertia and line friction 41 , the rate of change of kinetic energy can be estimated as \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$\\rho {U}^{3}{R}^{2}$$\\end{document} ρ U 3 R 2 , the line friction dissipation as \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${\\mu }_{f}{U}^{2}R$$\\end{document} μ f U 2 R , and the rate of work done by the surface tension as \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$\\gamma UR$$\\end{document} γ U R ( ρ is density, γ is liquid-gas surface tension, R is length scale taken as droplet radius in droplet spreading, and U is the spreading speed). It is found 41 that for a line friction Ohnesorge number greater than unity, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$O{h}_{f}={\\mu }_{f}/\\sqrt{\\rho \\gamma R}\\gg 1$$\\end{document} O h f = μ f / ρ γ R ≫ 1 , line friction will determine the spreading, inertia can be neglected, and the properties of the substrate will enter through μ f . In the opposite case, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$O{h}_{f}\\ll 1$$\\end{document} O h f ≪ 1 , line friction can be neglected, the spreading will follow an inertial timescale, and the early spreading will be insensitive to the surface properties. In a similar fashion it can be deduced that bulk viscosity can be neglected in early spreading when \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${\\mu }_{f} > \\mu $$\\end{document} μ f > μ . Even though S , the ratio of wet area to projected area, has been successful in quantitatively describing the influence of the substrate roughness 36 , 40 , 42 , the mechanism by which the hindering happens has not been made clear. Also, for substrate structures that are highly asymmetric, as for instance a sawtooth shape, we would expect a direction dependence on the spreading, which cannot be captured by the simple area ratio. In this paper we will describe a series of experiments of droplet spreading on sawtooth-like surface structures with well-defined pitches and angles. A qualitative ‘toy model’ is formulated and applied to the geometries of the substrate structures, and predictions are compared to experiments. This allows us to describe the actual mechanism by which the structures hinder spreading, and how features of the structure will affect the spreading speed.",
"discussion": "Discussion Throughout the results, we see that rescaling time according to the toy model consistently reduces the spread of the data, and we conclude that the geometric expressions for S in Fig. 2 do capture the essential influence of the geometry on the spreading speed for cases where Oh f > 1, and still partially for Oh f \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$\\lesssim $$\\end{document} ≲ 1. The reason for the slowing down is the long time that the contact line has to spend on those faces where the local dynamic contact angle becomes closest to the static contact angle θ e , thus reducing the driving force. It should be noted that the slowing down that we see in our experiments is not due to momentary pinning on edges and ridges. All through the advancement that local angle is greater than θ e , and the speed is determined by the local μ f , flat via Eq. ( 4 ). This is thus primarily a question of kinematics: the contact line will spend more of its time on the slow parts of the structure, and even on a symmetric structure where the contact line passes equal distances over fast and slow faces, the quick passage over the fast parts will not make up for the slow parts. The observation that the factor S could be interpreted as the ratio of wet area to projected footprint area for symmetric substrate topographies, we at this point regard as fortuitous. It is the case also for symmetric structures that not all faces are equally important; the most adverse faces will be responsible for the slowing down. Still, the wet to footprint area ratio seems to capture this well for symmetric shapes. It should also be noted that in cases where the total line friction, including the geometric factors described here, is not large enough to dominate the spreading, i.e., when Oh f ≪ 1, the spreading will be determined by the total inertia of the drop or the viscous dissipation in the bulk, and the substrate roughness will not influence the spreading. In conclusion, it is shown that the speed is significantly reduced for spreading in the directions across the ridges of the asymmetrically sawtooth shaped microstructured surfaces, with the greatest reduction when the contact line traverses the more gently sloping face of the ridge going up, and the steepest going down. The toy model that was constructed estimates the local speed of the contact line from the time required for the contact line to travel across the different parts of the structure, limited by a local flat surface line friction and the local departure of the contact angle from its static value. The reasonable agreement between the predictions from the toy model and the experimental results suggests that this is valid as a qualitative description."
} | 3,758 |
36466694 | PMC9716627 | pmc | 7,938 | {
"abstract": "Background The soil fungal community is one of the most important drivers of the soil nutrient cycling that sustains plant growth. However, little research has been done on the effects of different land uses on soil fungal communities in northeast China. Methods In this study, we conducted a field experiment to investigate the effects of continuous cropping of grass, maize, and alfalfa on their respective fungal communities and co-occurrence networks. Results We showed that the physicochemical properties of the soil, such as nitrate (NO 3 - N), available phosphorus, and soil pH, were the most important driving factors affecting the structure of the soil fungal community in different cropping systems. In addition, compared to the cultivation of grass and maize, the continuous cropping of alfalfa increased the abundance of several beneficial as well as pathogenic species, such as Mortierella and Gaiellales. In addition, the networks differed among plant species and according to the number of years of continuous cultivation. Conclusion This suggests that the continuous cropping of alfalfa results in greater cooperation among fungi, which may be beneficial to the soil as well as to the development of the alfalfa.",
"introduction": "Introduction Alfalfa ( Medicago sativa L .) is a leguminous, perennial plant of great importance for livestock and agriculture and is therefore widely grown in many countries and regions (Han et al., 2005 ; Raiesi, 2007 ). The arid regions of northeast China are the main areas where alfalfa is grown. Due to the climatic specificity of the long winters in northeast China, livestock in the region rely heavily on summer pasture storage for forage (Chen et al., 2013 ). Alfalfa has a high yield and a comprehensive range of nutrients and can therefore reduce forage shortages for herbivores in winter (Su, 2007 ; Chen et al., 2013 ). As a result, perennial alfalfa is grown year after year in the region to meet the winter demand for fodder and to increase livestock productivity (Dong et al., 2003 ). However, this type of agricultural intensification has led to a loss of biodiversity (Sala et al., 2000 ; Romdhane et al., 2022 ). Moreover, the number of pathogenic microorganisms increases with the continuous planting of alfalfa, eventually leading to a decrease in yield, a phenomenon that is closely linked to soil microorganisms (Yan et al., 2012 ; Yao et al., 2019 ; Liu et al., 2020 ). The number of years that alfalfa is grown is related to its productivity. Generally, alfalfa yields increase with the number of years planted; however, yields begin to decline when they reach a certain critical year, usually considered to be the 9 th year (Jiang et al., 2007 ; Li and Huang, 2008 ). In addition, continuous alfalfa cultivation significantly alters the physicochemical properties of the soil, which is significantly associated with yields (Ren et al., 2011 ). Previous studies have shown that planting alfalfa increases the organic matter and nitrogen content of the soil compared to virgin sandy soils. In addition, soil nutrients such as organic matter, nitrogen, and phosphorus increase with the number of continuous planting years. However, previous research has shown that soil nutrients tend to decrease after 10 continuous years of alfalfa cultivation (Jiang et al., 2007 ; Dong et al., 2016 ; Luo et al., 2018 ). Soil microorganisms serve a crucial function in the maintenance of plant health by interacting with plants, participating in nutrient uptake and resisting stress, and responding rapidly to changes in the physical and chemical characteristics of the soil (Song et al., 2021 ). Differences in tillage systems, soil types, crop species, and cropping systems greatly influence the structure of the soil's microbial community (Zhou et al., 2018 ; Yao et al., 2019 ; Yuan et al., 2021 ). For example, one study observed that soil microbial biomass declined in the short term, but ultimately increased over the long term, in a context of continuous alfalfa cultivation (Jiang et al., 2007 ). Another study reported that continuous cultivation of alfalfa changed the microbial diversity by altering the physicochemical properties of the soil (Luo et al., 2018 ). Some studies have shown that continuous alfalfa cultivation can increase the relative abundance of Paecilomyces phaeomycocentrospora and Fusarium sp. and decrease the relative abundance of Penicillium sp. (Xu et al., 1995 ; Yao et al., 2019 ). However, other studies have found no effect of continuous alfalfa cultivation on soil microbiota structure (Hu and Wang, 1996 ). These different results might be attributed to heterogeneity among the soil types, sample collection times, and tillage systems used in these studies. Therefore, there is a need for more in-depth studies to investigate the barriers to continuous alfalfa cultivation under intensive tillage patterns. Co-occurrence network analysis is a useful tool for exploring microbial associations and obtaining key information on microbial co-abundance communities associated with soil functions (Banerjee et al., 2018 ; Fan et al., 2021 ). One recent study used symbiotic networks to demonstrate that members of network modules were significantly associated with the genes involved in nutrient cycling after long-term fertilization, and that the number of members (operational taxonomic units, OTUs) in each module, rather than overall microbial diversity, influenced soil function (Fan et al., 2021 ). This raises the question of whether different land use practices alter the topological structures of networks, and what potential effects an altered network structure may have on soil function. Here, we examined the influence of continuously cultivating grass, maize, and alfalfa (for 6, 10, 14, 20, and 30 years) on soil microbes and soil characteristics. Because grass grows without human intervention, we hypothesized that alfalfa and maize would have higher microbial diversity than grass, and that continuous planting of alfalfa would increase the complexity of the co-occurrence. The aims of this study were to explore the various changes in the structures of soil fungal communities across different cropping systems, and to assess the association between the physical characteristics of soil and the characteristics of its fungal community.",
"discussion": "Discussion In the present study, fungal diversity was higher in the AC treatment than in the Me or Ma treatments, and the fungal diversity of alfalfa soils increased with subsequent years of planting. This result is in line with our first hypothesis, which suggests that alfalfa has more microbial species than grass and maize, and that continuous alfalfa planting is more beneficial to fungal diversity maintenance and soil sustainability, at least in terms of fungal diversity. Previous studies have found that continuously planted soybean has less soil microbial diversity than corn–soybean rotation systems (Liu et al., 2020 ). The number of years of continuous planting is also related to the microbial diversity of the soil (Liu et al., 2020 ). However, other studies have found no difference in microbial diversity between soils farmed continuously with soybean and soils farmed continuously with soybean–corn rotations (Li et al., 2010 ). These inconsistent results might be based on the soil types and the number of years of repeated harvests. Alterations in plant genotype may also be responsible for this result, given that microbial diversity has been shown to exhibit diverging trends in the context of successive cultivation of resistant and vulnerable cultivars (Yuan et al., 2021 ). Organic acids, phenols, and other compounds found in plant root exudates have an influence on microbial diversity in a range of agricultural situations (Tan et al., 2017 ; Lian et al., 2019 ; Liu et al., 2020 ; Shi et al., 2020 ). Furthermore, soil pH influences other soil characteristics that can directly or indirectly affect microbial diversity (Lian et al., 2019 ). Regarding the beta diversity, the results from the principal coordinate analysis showed that crop types and continuous tillage time were the two most important factors affecting the structure of the soil fungal community ( p < 0.05). Every species of plant releases a specific set of metabolites during growth, and this in turn allows its root system to provide a unique habitat for, and host different types of, soil fungal microorganisms. These microbes may also help plants absorb nutrients and resist stresses (Lian et al., 2019 ). Moreover, our results were also consistent with previous studies that found that continuous crop planting also affects soil fungal community structure (Zhu et al., 2017 ; Yao et al., 2019 ; Yuan et al., 2021 ). This is mainly due to the effect of root exudates on soil microorganisms. For example, long-term continuous cultivation of soybean can lead to the accumulation of organic acids in the soil and ultimately cause soil acidification. This provides an ideal environment for pathogenic fungi to survive and alters the community structure of soil fungi, ultimately leading to reduced soil quality and crop yields (Zhu et al., 2017 ; Lian et al., 2019 ; Yuan et al., 2021 ). Furthermore, according to the CCA results, the physicochemical characteristics of our soil samples, such as NO 3 - N, available phosphorus, and soil pH, were the main factors that altered the structure of the soil fungal community in the different treatments ( Figure 6 ). This is similar to the findings of many previous studies, in which changes in tillage practices were shown to affect the microbial environment by altering the soil characteristics (Lian et al., 2019 ; Yao et al., 2019 ). Over all, our results also suggest that some important soil parameters changed significantly through continuous cropping over time, leading to changes in the fungal community. However, this change was not unidirectional, and longer periods of continuous cropping will perhaps lead to more positive developments for the soil microorganisms, such as increased diversity and significant enrichment of the beneficial fungi. Figure 6 Relationship between soil properties and fungal community structure based on canonical correspondence analysis (CCA). In AC soils, the relative abundance of Ascomycota was substantially higher compared to that of soils from the Ma and Mc systems ( Figure 4 ). Many ascomycetes are plant-pathogenic, such as rice blast, black knot, the ergot fungi, and the powdery mildews, which suggests that continuous planting of alfalfa may have increased the abundance of potential pathogens, and influence the growth of alfalfa (Yuan et al., 2021 ). The relative abundance of Mortierella, Gibberella, Solicoccozyma, Metarhizium , and Phaeomycocentrospora was increased in AC fields compared to Me and Ma fields. It has been reported that Mortierella can survive under very unfavorable environmental conditions and make efficient use of carbon sources contained in polymers such as cellulose, hemicellulose, and chitin, and that it can synthesize phytohormones and 1-aminocyclopropane-1-carboxylic acid deaminase through improved access to bioavailable forms of phosphorus and iron in the soil, thereby protecting agricultural plants from pathogens (Ozimek and Hanaka, 2020 ). However, some pathogenic microbial species, such as Gaiellales , with high relative abundance in the AC treatment, can cause sear rot, suggesting that these fungi may suppress soil diseases, suggesting that these fungi may suppress soil diseases (Gómez Expósito et al., 2015 ). Therefore, it is likely that the changes in these fungi caused by the different treatments are related to the soil nutrient structure and the antagonistic activity of the plant pathogens. In the present study, co-occurrence networks have helped us explore the complex relationships between fungi in different treatments in greater depth (Xue et al., 2018 ; Xiong et al., 2021 ). Our results show that both negative network correlations and modularity were significantly higher for AC than for Ma and Mc, which is consistent with our second hypothesis. This suggests that successive plantings of alfalfa promoted cooperation between fungi, which may be beneficial for alfalfa survival (Yao et al., 2019 ; Liu et al., 2020 ). Moreover, the difference in network topology of these treatmentsmay be due to the fact that certain microbial species are enriched to help the host increase nutrient uptake or resist stress, causing the structure of the microbial community to deviate from its original equilibrium (Lian et al., 2019 ). However, the results for the fungal networks alone are one-sided, as functional bacteria are also present in the soil. Therefore, in future studies, combined bacterial and fungal network analysis may yield more comprehensive results and a fuller assessment of the effects of different tillage practices on soil microbes. In conclusion, alfalfa crop cultivation increased the alpha-diversity of soil fungi compared to grass and maize cultivation, and alpha-diversity increased further in continuous cropping systems, which is of great interest to maintain soil microbial diversity. The physicochemical properties of the soil, such as NO 3 - -N, soil Ph, and available phosphorus, were the most important driving factors affecting the soil fungal community structure across the different cropping systems. Compared to the cultivation of grass and maize, the continuous cropping of alfalfa increased the abundance of several beneficial as well as pathogenic fungal species, such as Mortierella and Gaiellales . In addition, the networks differed among plant species and also among different lengths of time of continuous alfalfa cultivation. This suggests that the continuous cropping of alfalfa results in greater cooperation among fungi, which may be beneficial to the soil as well as to the development of the alfalfa."
} | 3,499 |
30834185 | PMC6397634 | pmc | 7,940 | {
"abstract": "The importance of soil microbial flora in agro-ecosystems is well known, but there is limited understanding of the effects of long-term fertilization on soil microbial community succession in different farming management practices. Here, we report the responses of soil microbial community structure, abundance and activity to chemical (CF) and organic fertilization (OF) treatments in a sandy agricultural system of wheat-maize rotation over a 17-year period. Illumina MiSeq sequencing showed that the microbial community diversity and richness showed no significant changes in bacteria but decreased in fungi under both CF and OF treatments. The dominant species showing significant differences between fertilization regimes were Actinobacteria, Acidobacteria and Ascomycota at the phylum level, as well as some unclassified genera of other phyla at the genus level. As expected, soil organic matter content, nutrient element concentrations and bacterial abundance were enhanced by both types of fertilization, especially in OF, but fungal abundance was inhibited by OF. Redundancy analysis revealed that soil enzyme activities were closely related to both bacterial and fungal communities, and the soil nutrient, texture and pH value together determined the community structures. Bacterial abundance might be the primary driver of crop yield, and soil enzyme activities may reflect crop yield. Our results suggest a relatively permanent response of soil microbial communities to the long-term fertilization regimes in a reclaimed sandy agro-ecosystem from a mobile dune, and indicate that the appropriate dosage of chemical fertilizers is beneficial to sandy soil sustainability.",
"conclusion": "Conclusions Succession of microbial community characteristics and soil biogeochemical properties from an arid sandy system showed that both long-term chemical and organic fertilization enhanced the soil nutrient content available to plants, changed the microbial community structure, promoted soil enzyme activity, and increased crop plant yield. These results were compatible with our first hypothesis. However, contrary to our second hypothesis, no significant differences were observed in prokaryotic microbial community diversity in relation to fertilization regimes, and FYM inhibited the development of the fungal community. In contrast, chemical fertilization was more beneficial to bacterial propagation and crop yield, and bacterial abundance might be the main determinant of crop yield. Thus, changes in soil microbial activities and crop yield under long-term fertilization practices may be achieved mainly by bacteria rather than fungi. This field experiment indicates a relatively permanent response of soil bacterial and fungal communities in a sandy agro-ecosystem to long-term fertilization, and provides a comprehensive understanding of the effects of fertilization regimes on sustainable soil development. Overall, the results obtained from a sandy agricultural system managed by chemical and organic fertilization over a 17-year period indicated that it is possible to perform reasonable chemical fertilization management while maintaining soil health during farming practices.",
"introduction": "Introduction A growing global demand for agricultural crops is one of the main challenges in the 21st century. The pursuit of high productivity, long-term sustainability and optimal resource use efficiency without negative effects in the restricted land available for agricultural cultivation has led to the emergence of a variety of management practices ( Xin & Li, 2018 ). Because of the low level of soil nutrients, the productivity of sandy land agro-ecological systems in arid areas is mainly dependent on intensive agricultural management. Generally, controlling watering and fertilization is a basic farming practice in which chemical or organic fertilizers with a certain amount of water are used primarily to improve soil nutrients and hence crop productivity ( Mayer et al., 2015 ). However, determining how to evaluate the soil quality and sustainability in such systems is a major issue in farming management. Many variables have been studied in relation to long-term system sustainability, including available nutrients ( Oehl et al., 2004 ), soil carbon (C) or nitrogen (N) stock development ( Bosshard et al., 2009 ), and soil microorganism abundance and biodiversity ( Rillig, 2004 ). Soil functioning and sustaining soil fertility is largely governed by the decomposition activity of the microflora ( Anderson, 2003 ). In recent years, soil microbial community structure, diversity and activity have been used as indicators of overall soil health and productivity potential ( Rasmussen et al., 1998 ; Böhme, Langer & Böhme, 2005 ; Kaschuk, Alberton & Hungria, 2010 ; Mbuthia et al., 2015 ) because the development of microbial metagenomic technology facilitates comprehensive analysis of culturable and unculturable microorganisms ( Lentendu et al., 2014 ; Eo & Park, 2016 ). In addition, shifts in microbial community structure and the abundance of various plant-beneficial and detrimental soil microorganisms have been shown to influence the productivity and stability of the agroecosystem ( Francioli et al., 2016 ). Because microorganisms play a leading role in soil development and preservation ( Schloter et al., 2018 ), there is a demand for quick and reproducible microbial-based indicators. Such indicators should ideally describe organisms with key functions in the system or with key regulatory/connecting roles (so-called keystone species). Therefore, in the search for suitable soil quality indicators it is important to consider parameters in which the biotic–abiotic interlinkages would find their expression ( Vestergaard et al., 2017 ). However, in light of the huge functional redundancy in most soil microbiomes and the poor understanding of the relationship between microbial community structure and soil function ( Pereira e Silva et al., 2012 ), finding specific keystone markers is not a trivial task and there has been a long debate on the best method for their selection ( Anderson, 2003 ; Sharma et al., 2010 ; Schloter et al., 2018 ). Furthermore, there is currently no consensus regarding what would constitute a reasonable set of proxies that together constitute a good framework at the microbial community level for soil quality assessments ( Schloter et al., 2018 ). The effects of farming management practices on changes in the soil microbial community (i.e., bacteria and fungi) have been extensively studied in different soils, but the results have differed. For example, most studies of long-term chemical fertilization have revealed that a significant increase in soil microbial biomass ( Geisseler & Scow, 2014 ) and long-term organic and chemical fertilization result in a clear shift in bacterial and fungal community composition ( Lentendu et al., 2014 ; Hartmann et al., 2015 ; Francioli et al., 2016 ). However, some field studies based on short-term application of N amendments have revealed no significant changes or opposite results ( Lazcano et al., 2013 ), and both positive and negative effects of chemical fertilizers on soil microbial activities have been reported ( Guo et al., 2011 ; Nannipieri et al., 2012 ). Although higher microbial community diversity and biomass have been observed in most of these studies during short-term organic fertilization ( Marschner, Crowley & Yang, 2004 ; Lentendu et al., 2014 ), these observations differ from those in long-term experiments ( Eo & Park, 2016 ) and cannot be used to evaluate the effects of different farming systems on soil quality and sustainability ( Rasmussen et al., 1998 ). To provide a comprehensive understanding of whether sandy fields will be healthy and sustainable after application of integrated farming management, changes in soil microbial community structure and activity and their relationships with soil characteristics and crop yield were analyzed in a 17-year fertilization treatment experiment. Illumina MiSeq sequencing was used to analyze the prokaryotic 16S rDNA and fungal 18S rDNA in different fertilization regimes (chemical fertilization, organic fertilization and no fertilization), as well as in uncultivated mobile sand as a control. Soil physicochemical properties, bacterial and fungal abundance, and the activity of microbial intracellular oxidizing enzymes and extracellular hydrolases involved in nitrogen and phosphorus cycles in the soil were also determined. We hypothesized that (i) the composition, abundance and activity of the soil microbiome are driven by different fertilization regimes in the primary successional process; (ii) soil nutrient content is the main determinant for bacterial and fungal structure, whereas organic fertilizers (farmyard manure, FYM) are more helpful for the development of microbial community; and (iii) shifts in the structure and abundance of soil microorganisms affect the agro-ecosystem crop yield.",
"discussion": "Discussion Transformation from mobile sand to arable land in arid areas is good not only for economic returns but also for the future sustainable development of land in China. Therefore, reasonable management strategies without negative effects on soil quality and environmental systems are the main objectives of land development in agroecosystems. This study provides an overview of the responses of soil microbial communities to long-term fertilization from a traditional farming practice, and it could be useful in the development of methods of sustainable land management for newly developed land. Responses of the soil microbiome to long-term fertilization Our first hypothesis that the composition, abundance and activity of the soil microbiome are driven by different fertilization regimes was verified by investigation of shifts in prokaryotic microbial and fungal community composition, abundance and soil enzyme activities between CF and OF. In the present study, long-term chemical and organic fertilization had significantly different effects on prokaryotic microbial and fungal community structures. There were no substantial changes in the prokaryotic microbial community diversity and richness; however, there were clear shifts in the proportion of prokaryotic microbial community compositions in relation to fertilization treatments. Fungal communities were much more sensitive to fertilization, showing a significant decrease in community diversity in response to organic fertilizers. These different responses in bacterial and fungal community structures have been reported in most previous studies of long-term fertilization treatments ( Lentendu et al., 2014 ; Hartmann et al., 2015 ; Eo & Park, 2016 ) and even in a 113-year fertilization experiment ( Francioli et al., 2016 ). Contrary to our expectations, the fungal community diversity was not enhanced by long-term organic fertilization. From the perspective of soil microbial community succession, PCoA ordination plots revealed significant effects on the microbial community structure after cultivation, and different fertilization management methods greatly affected the fungal community structure ( Fig. 2 ). Further study of the different proportions of dominant prokaryotic microbial and fungal groups responding to fertilization treatments at the phylum and genus levels ( Figs. 4 and 5 ) revealed significant differences in all dominant phyla except Proteobacteria, Chloroflexi and Firmicutes in the prokaryotic communities between treatments, but Ascomycota were the only fungi affected by chemical and organic fertilization. Proteobacteria and Firmicutes, which are generally seen as copiotrophic bacteria ( Lienhard et al., 2014 ) that prefer carbon-rich environments, are more abundant in FYM treatments than in systems subjected to long-term chemical fertilization ( Francioli et al., 2016 ). However, our results showed that the OM and TN content were comparable between OF and CF, thus potentially leading to comparable growth proportions of these two phyla. Chloroflexi, which mainly consist of the filamentous anoxygenic phototrophic bacteria, is a thermophilic phylum that is well adapted to drought conditions ( Hugenholtz and Stackebrandt, 2004 ) and is insensitive to fertilization treatments. Among bacterial phyla that differed slightly ( p < 0.05) between treatments, Acidobacteria and Actinobacteria showed the most significant changes ( p < 0.001). Actinobacteria, the predominant phylum in all soils investigated in this study, play a major role in agricultural soil quality promotion via organic matter decomposition ( Strap, 2011 ) and these organisms were more abundant in the mobile sand and unfertilized soil. This phylum may consist of oligotrophs and may have a positive correlation with pH. In contrast, Acidobacteria were significantly lower in OF, a result that may be explained by the many reports of a negative correlation between this phylum and pH ( Rousk et al., 2010 ; Fierer et al., 2012 ). Ascomycota, the predominant fungal phylum in agro-ecosystems and sandy soil ( Lienhard et al., 2014 ), are important decomposers of organic substrates such as wood, leaf litter and dung, and thus showed the greatest diversity when applied to the FYM. At the genus level, the universal primers of 16S rDNA revealed that the genera present in the highest proportion was the soil crenarchaeotic group, which consists of ammonia-oxidizing archaea and plays a major role in C fixation and nitrification of N metabolism ( Zhang et al., 2012 ). Two unclassified genera in Actinobacteria and Gemmatimonadaceae were the most abundant bacteria after cultivation ( Fig. 5A ), whereas two unclassified genera in Pezizales and Sordariales of fungi accumulated in OF ( Fig. 5B ). These observations demonstrated that unknown specific functions should be considered in these species that differed significantly between treatments. In this light, knowledge of how fertilization strategies enhance the proliferation of taxa from high throughput sequencing is invaluable. Consequently, wet lab experiments are needed to increase understanding of the effects of specific management strategies on soil microorganisms and their interaction processes, to increase crop yields under sustainable agricultural systems. Moreover, there were clear responses of microbial abundance and soil enzyme activities to long-term fertilization. The bacterial abundance was highest in OF, which is consistent with the soil organic matter and available potassium levels and indicates that the bacterial growth rate was closely related to soil fertility. Soil enzyme activity, a potential indicator of organic matter decomposition, is often affected by specific management strategies and plays important roles in sustainable agroecosystems. Soil dehydrogenase activity occurs intracellularly in all living microbial cells and is therefore commonly used as an indicator of overall soil microbial activity ( Kaczynski et al., 2016 ). Catalase activity has been used as an indicator of soil fertility, because it is related to the metabolic activity of aerobic organisms ( Trasar-Cepeda, Gil-Sotres & Leiros, 2007 ). The activity of both of these enzymes was highest in CF, indicating that CF promoted soil microbial activity and metabolic potential. Extracellular enzyme activities can be used as indicators of microbial nutrient demand because they are the proximate agents of organic matter decomposition ( Sinsabaugh et al., 2010 ). UE and AKP, which are involved in N acquisition and P mineralization, respectively, had the highest activity in OF. This result can be interpreted as indicating that all extracellular enzyme activities are significantly related to soil pH and nutrient content, and that hydrolytic activities are more variable than oxidative activities in response to soil nutrients ( Sinsabaugh et al., 2010 ). Correlation between soil microbiome and soil and crop characteristics Our second hypothesis that soil nutrient content is the main determinant of microbial structure, and FYM is more beneficial for development of the microbial community, was partly verified by our analysis of prokaryotic microbial and fungal community structures according to soil pH, soil nutrients and texture. No significant differences were detected in the prokaryotic community structure between CF and OF, whereas FYM inhibited the fungal community diversity and richness. Soil pH is an important factor that drives prokaryotic microbial community composition ( Lauber et al., 2009 ). The application of mineral fertilizer significantly decreases the soil pH ( Luo et al., 2015 ). Our analysis confirmed this result after 17 years of continuous application of mineral fertilizers and indicated that this effect may have been due to the release of hydrogen (H + ) ions by oxidation of ammonium ( \\documentclass[12pt]{minimal}\n\\usepackage{amsmath}\n\\usepackage{wasysym} \n\\usepackage{amsfonts} \n\\usepackage{amssymb} \n\\usepackage{amsbsy}\n\\usepackage{upgreek}\n\\usepackage{mathrsfs}\n\\setlength{\\oddsidemargin}{-69pt}\n\\begin{document}\n}{}${\\mathrm{NH}}_{4}^{+}$\\end{document} NH 4 + ), which is commonly used as a nitrogenous fertilizer in paddy cultivation ( Kumar et al., 2017 ). However, the pH after FYM application was higher than that after chemical fertilization, probably because of enrichment of cations ( Murugan & Kumar, 2013 ). Variations in pH between CF and OF affected the species that were closely related to soil pH, such as Acidobacteria and Actinobacteria. Application of FYM promoted a higher level of soil nutrient and texture with high OM, silt, clay and WHC than that in CF, possibly because of the amount of organic carbon and microbial biomass carbon in FYM added to the soil ( Joergensen, Mäder & Fließbach, 2010 ). These changes in turn promoted the rapid propagation of bacteria and inhibited fungal community abundance and diversity in OF. In recent years, some studies have shown that soil nutrient elements such as OC and TN are more closely correlated to the bacterial community structure than pH, whereas no correlations were observed between the fungal community and these elements ( Zhong, Yan & Shangguan, 2015 ). Extensive studies of the effects of different agricultural management practices on the bacterial community have revealed that these communities are greatly influenced by soil physicochemical properties, whereas fungal communities have been shown to be unchanged or negatively affected by these practices ( Luo et al., 2015 ). Moreover, short-term application of chemical and organic fertilizers have revealed no significant changes in microbial community ( Lazcano et al., 2013 ), and most studies of long-term fertilization have revealed a clear shift in bacterial and fungal community composition ( Lentendu et al., 2014 ; Francioli et al., 2016 ). Both positive and negative effects of organic and chemical fertilizers on soil microbial activity have been reported ( Guo et al., 2011 ; Nannipieri et al., 2012 ) because of differences among crop plants and fertilization treatment times. The third hypothesis, in which shifts in the structure and abundance of soil microorganisms affect the crop yield of the agro-ecosystem, was verified by our observation that bacterial abundance and fungal diversity indexes were closely related to crop yield. Many studies have attempted to link crop productivity to below-ground microorganisms and have addressed the potentially important role of bacterial taxa in soil OM accumulation, increases in soil enzyme activity, plant growth promotion and disease suppression in different fertilization regimes ( Marschner, Crowley & Yang, 2004 ); however, other reports have shown reduced crop yield and biomass after continuous application of N and FYM ( Kumar et al., 2017 ). Because of the limited knowledge of bacterial ecological function ( Hartmann et al., 2015 ), it is difficult to use metagenomics to determine which beneficial or pathogenic taxa are promoted or suppressed in crop plants as a result of different fertilization regimes. In the present study, RDA analysis showed that bacterial abundance may be the most important positive factor influencing crop yield, and fungal diversity was negatively correlated to crop yield, indicating that specific types of fertilization can promote beneficial rather than detrimental groups of beneficial soil microorganisms. However, soil enzyme activities were closely correlated to prokaryotic microbial and fungal community structures in our study. Generally, soil enzyme activity may be a useful indicator of soil functional diversity, and it has been reported to provide a unique integrative biochemical assessment of soil function and condition ( Chaparro et al., 2012 ). In this light, soil enzyme activity is very important for evaluating the crop productivity and sustainability of agricultural systems. Evaluation of soil sustainability The living soil system, which provides several ecosystem services including nutrient cycling, water regulation and controlling pests and diseases ( Sharma et al., 2010 ; Kumar et al., 2017 ), is of primary importance in sustainable agricultural production. Microorganisms are one of the originators of soil and play an important role in determining soil functions including decomposition of above- and below-ground plant material ( Biswas & Kole, 2017 ). In addition, microbes provide an integrated measure of soil quality that cannot always be obtained from physical and chemical measures and/or analysis of higher organisms. Shifts in the structure and composition of the distinct microbial community are strong indicators of soil biological activity and crop productivity of terrestrial agro-ecosystems ( Sharma et al., 2010 ); consequently, microbial analyses can discriminate soil quality status and utilization of shifts in microbial populations and activity and can be used as indicators of changes in soil quality. The use of microbial community structure and diversity as an indicator to monitor soil quality is challenging, because understanding of the relationship between community structure and soil function is lacking. Although the relationship between soil quality and microbial diversity is not completely understood, a moderate improvement in the soil environment accompanied by increasing microbial biomass and diversity in agricultural soil is generally considered to indicate ‘good’ soil quality. Our study demonstrated that soil microbial communities and crop agronomic traits were significantly altered by long-term fertilization, and that these shifts promoted the quality of agricultural systems. Many studies have shown that organic fertilization stimulates the release of plant available nutrients in soil and enhances soil biological activities, thereby establishing a sustainable agricultural ecosystem ( Carpenter-Boggs, Kennedy & Reganold, 2000 ). However, farming practices sometimes adversely affect soil biological properties, resulting in low soil quality ( Böhme, Langer & Böhme, 2005 ). For example, fungal community structure and abundance play important roles in soil stability through a spatial hyphal network, which develops throughout the soil and can be strongly affected by fertilization and other agricultural management practices ( Frey, Elliott & Paustian, 1999 ). As a result, the ecological functions of arbuscular mycorrhizal fungi as they relate to soil quality have garnered increased scientific attention in recent years ( Rillig, 2004 ). In the present study, the soil microbial properties in CF showed a pattern of similar bacterial diversity and abundance, better fungal diversity and abundance, and higher microbial activities to that of OF; accordingly, the crop yield and biomass were also higher in CF. Thus, long-term sustainable soil management can be achieved in a fertilization system with chemical fertilizers, not just organic manure. The present experiment clearly shows the practicality of converting arid mobile sand to an arable system with a reasonable chemical fertilization level. The environmental and agronomic long-term optimum might lie between chemical and organic fertilization in an irrigable arid sandy system. Our data provide a good basis for avoiding the blind reduction of chemical fertilizers by considering the long-term sustainable development of arable land."
} | 6,097 |
27054762 | PMC4824447 | pmc | 7,941 | {
"abstract": "A ball-on-plate wear test was employed to investigate the effectiveness of graphene (GP) nanoparticles dispersed in a synthetic-oil-based lubricant in reducing wear. The effect by area ratio of elliptically shaped dimple textures and elevated temperatures were also explored. Pure PAO4 based oil and a mixture of this oil with 0.01 wt% GP were compared as lubricants. At pit area ratio of 5%, GP-base oil effectively reduced friction and wear, especially at 60 and 100°C. Under pure PAO4 oil lubrication, the untextured surfaces gained low friction coefficients (COFs) and wear rates under 60 and 100°C. With increasing laser—texture area ratio, the COF and wear rate decreased at 25 and 150°C but increased at 60 and 100°C. Under the GP-based oil lubrication, the textured surface with 5% area ratio achieved the lowest COF among those of the area ratios tested at all test temperatures. Meanwhile, the textured surface with 20% area ratio obtained the highest COF among those of the area ratios. With the joint action of GP and texture, the textured surface with 10% area ratio exhibited the best anti-wear performance among all of the textured surfaces at all test temperatures.",
"conclusion": "Conclusion Graphene(GP) has attracted much interest because of its excellent properties, such as mechanical strength and good conductivity. Its lamellar structure affords the material its potential for tribological applications. From the series of tests conducted, the following conclusions were drawn: Span-80 generates a stable GP dispersion in PAO4 base oil by inhibiting the occurrence of the agglomeration phenomenon. GP can effectively improve the anti-wear properties of contact surfaces. This improvement is most apparent at 60 and 100°C. The maximum COF can be reduced by 78%, whereas the maximum wear rate can be decreased by 90%. Without GP additive, textured surfaces can exhibit increased wear losses at temperatures below 100°C. With the joint action of added GP and texture, the textured surface with an area ratio of 10% exhibits the best anti-wear performance among the different ratios under all the temperatures tested.",
"introduction": "Introduction Graphene (GP) is a monolayer of graphite (2D) and is the basic building block of all graphitic forms, such as fullerenes (0D), carbon nanotubes (1D), and graphite (3D) [ 1 ]. GP possesses numerous excellent properties, such as extreme thinness, high mechanical strength, high electrical conductivity, high surface area, and thermal mechanical properties. Thus, GP has been attracting worldwide interest since its discovery [ 2 – 4 ]. These unique properties render GP a promising candidate for different applications, such as in transistor—transparent electrodes, chemical and biological sensors, and energy-storage materials. However, only a few studies on the tribological applications of GP have been reported, particularly, the addition of nano-GP particles into lubricating oil [ 5 ]. GP easily aggregates in solution and difficultly forms a stable dispersion, thereby limiting the application in the field of lubrication. GP exhibits good chemical stability and possesses massive inter-layer van der Waals forces. Hence, forming a stable dispersion solution with GP is difficult to achieve, especially on account of the agglomeration phenomenon [ 6 ]. Two kinds of methods are often used to improve the dispersibility of GP in oils. One is adding a dispersant, which leads to a uniform dispersion of GP in oil. Another is modifying the surface appropriately to enhance the lipophilic property of GP. Tadmor [ 7 ] and N'guessan’s [ 8 ] used the centrifugal adhesion balance method to measurements of the lateral adhesion forces at a solid-liquid interface in GP. Varrla et al. [ 9 ] reported that the coefficient of friction (COF) and the wear scar diameter are reduced by 80% and 33%, respectively, when the concentration of GP is 0.025 mg/mL in base oil. Lin et al. [ 10 ] found the significant reduction in frictional coefficient and wear using GP modified by stearic/oleic acid as lubricant additive. To improve the effective functioning of two friction surfaces, lubricants with additives are required. However, surface texturing is another efficient approach to improve the interface [ 11 – 13 ]. Through this approach, hydrodynamic lubrication is facilitated, lubricants are stored, and the interface contains the debris [ 14 – 17 ]. To date, most of the research on this aspect has been conducted on smooth surfaces. The combined effect of surface textures and additives on friction interfaces is worth studying and exploring. In the present work, we prepared oil fluid containing GP and texturized the surface under study by laser. Tribology tests were carried out under different test conditions, and the interaction between the textured surface and additive was investigated.",
"discussion": "Results and Discussion Tribological properties The friction coefficients obtained from the tests on the textured and untextured plates differed significantly from each other ( Fig 5 ). GP addition effectively reduced friction, and the coefficient of friction (COF) decreased by nearly 75%, especially at 60 and 100°C, relative to that of the oil. At room temperature, the COF was relatively stable because of sufficient oil thickness and mild oxidation reaction. In the initial stage at 150°C, GP effectively reduced COF, but the effect became less noticeable with the progress of time. The COF curves of the PAO4 oil present a rise and fall in the initial stage at temperatures 60, 100, and 150°C. This finding might be due to the mechanical removal action in the running-in period. The COF at 150°C was lower than those at 60 and 100°C, which may be attributed to the debris formed at high temperatures that can reach the interface and reduce wear. 10.1371/journal.pone.0152143.g005 Fig 5 Friction coefficients at different temperatures. The surface texture significantly influenced the COFs under pure PAO lubrication ( Fig 5 ). Below 100°C, the COFs obtained from testing the dimpled plate exhibited a sharp increase at the beginning of the test, followed by a steady decrease after reaching a peak value. Compared with the PAO4 base oil, the COF curves after GP oil addition were smaller and more stable under all the test temperatures. This finding may be due to the GP adsorbed on the tribo-surface and formed a protective film that separated the samples under contact preventing the occurrence of oxidation reaction [ 19 – 22 ]. The wear mechanism will be analyzed and discussed in the subsequent section. The images of the wear scars are displayed in Fig 6 . Lubricated by pure PAO4, the worn dimensions of the balls and plates dramatically varied with changing temperatures. The scar sizes of the balls and widths of the plates were smaller at 25 and 150°C and larger at 60 and 100°C. Especially, the wear scars of the balls were divided into two parts at 60 and 100°C, namely, the worn part I (area inside the white line) and the oxidized part II (intermediate area between the white and yellow line), respectively. Under GP-based oil lubrication, the wear scar was reduced relative to that under pure oil lubrication ( Fig 6e–6h ). Energy dispersive spectroscopy (EDS) analyses were carried out to estimate the chemical changes induced by friction. As shown in Fig 6i , except for the substantial oxygen content, many elements in the plate can be found in the worn ball. These elements include tin, zinc, and lead (correspond to the worn surface in Fig 6c ). When the GP was added to the oil, the worn surface attained a bright color unlike that of the dark scar lubricated by pure PAO oil. The EDS result indicates the elemental transfer and that oxidation did not occur on the worn ball ( Fig 6j ). Interestingly, this phenomenon proves that the additive can inhibit the oxidation reaction effectively and modify the friction interface. 10.1371/journal.pone.0152143.g006 Fig 6 Morphology of the wear scars on the steel balls and 2# textured plates (a–h), as well as the EDS patterns of the wear scar on the ball (i, j). The SEM morphologies and EDS results of the worn plate surface are shown in Fig 7 . Under the pure PAO4 oil, ploughing and wear debris could be observed ( Fig 7 ). However, the wear scar was relatively smooth and shallow under the addition of GP in oil ( Fig 7b ). The EDS data revealed that the oxygen of the worn surface was higher under pure PAO4 oil, indicating the presence of oxidative wear phenomenon. The oxygen content of the worn surface was lower in the GP-based oil, revealing that GP can effectively inhibit the oxidation reaction to improve the tribological environment of the interfaces. Fig 7d displays the Raman spectra of the GP and the wear scar under the GP-containing oil. As observed, the D and G peaks of the worn surface were consistent with the peaks of GP. Therefore, GP possibly adsorbed on the bronze surface and formed a film during the wear process. 10.1371/journal.pone.0152143.g007 Fig 7 SEM and EDS of the worn surfaces. SEM images of the worn surfaces with (a) PAO4 oil and (b) PAO4 oil with GP; (c)EDS patterns and (d)Raman spectrum of the wear scar. Fig 8 illustrates the cross-sectional profile of the wear scar. The wear scar profile lubricated by GP-containing oil was significantly shallower than that lubricated by pure PAO4. The distinction was evident at 60 and 100°C. After adding GP, the maximum depth of the wear scar reduced from 3.8 μm prior to addition to 2.6 μm after addition at 25°C ( Fig 8a ) and even from 14.2 μm to 2.7 μm at 60°C ( Fig 8b ). 10.1371/journal.pone.0152143.g008 Fig 8 Cross-sectional profiles of the wear scars (samples 0# and 2#). Effect of texture and wear mechanism The hardness of the wear scars are presented in Fig 9 . The hardness of the wear scar lubricated by the GP-containing oil was lower than that of pure PAO4 in all textured and non-textured samples, especially at 60 and 100°C. By comparing the hardness data, we conclude that the hardness remained constant regardless of temperature. 10.1371/journal.pone.0152143.g009 Fig 9 Hardness of the wear scar surface. Fig 10 reveals the average COF values under different textured samples. Under pure PAO4 oil, the COFs exhibited similar tendencies with increasing temperature; the COFs rose at 60–100°C and then declined at 150°C. When the ratio of the surface texture increased, the high textured ratio reduced the high COF values at 60–100°C. The emergence of this phenomenon might be due to the presence of the dimples collecting debris at 25 and 150°C that becomes saturated at 60 and 100°C because of excessive amounts of debris that cannot be collected efficiently. By contrast, under GP-based oil lubrication, the COFs appeared to remain constant regardless of the textural changes. By detailed analysis, we determined that the COF of the 20% textured surface is higher than that of the 5% textured surface. These findings may be explained by the ability of the dimples to collect debris during the friction process and the magnification of surface roughness by excessive macroscopic dimples. We determined the optimal ratio between the two components to be 5%, which also achieved the lowest COF in the test. 10.1371/journal.pone.0152143.g010 Fig 10 Comparison of the average COFs among different textured surfaces. The depths and widths of the worn plate samples are depicted in Fig 11 . Under pure PAO4 lubrication, the depths and widths of the wear scars under different textures exhibited the same trend with the change in temperature. In particular, the depths and widths increased initially and then decreased. The depths and widths were approximately the same at 25 and 150°C. At 60 and 150°C, the depths and widths increased with increasing texture ratio, whereas an opposite trend was observed at 25 and 100°C. Under the GP-based oil, the widths and depths of the wear scars of the samples exhibited an upward trend at texture ratios of 0%, 5%, and 20%, and a downward trend at 10%. 10.1371/journal.pone.0152143.g011 Fig 11 Widths and depths of the wear scars under different temperatures. The wear rates are shown in Fig 12 . The four different textured surfaces exhibited similar trends after GP addition. At 60 and 100°C, wear rates dropped significantly, even up to 95% reduction, under GP-based oil lubrication. Furthermore, under this lubrication, the wear rate of the 10% textured surface is lower than those of the others at any of the temperatures tested. The temperature can affected the tribology situation during the tribology [ 23 ], J. Taha-Tijerina reported the Temperature-dependent viscosity variation of nanofluids- two dimensional (2D) atomic sheets, such as hexagonal boron nitride (h-BN) and graphene (GP)[ 24 ]. 10.1371/journal.pone.0152143.g012 Fig 12 Wear rates under different temperatures. As shown by the tests on tribological performance, GP-containing oil exhibited excellent anti-friction and anti-wear effects. GP possesses a lamellar structure and involves sliding between layers. This structure can then contribute to the anti-wear and anti-friction properties during tribology processing Understanding the lubrication mechanisms and anti-wear functions of additives is highly important. Fig 13 shows the wear mechanism of the GP additive on the textured plate. Forming stable GP dispersion in solution is difficult to achieve. A base oil with well-dispersed additives plays an important role in friction reduction and anti-wear action. In this research, a slow sliding speed of 5 mm was employed and hence corresponds to a boundary lubrication situation. Additionally, the point of contact in the ball-against-plate contact model involves high contact stress. In this case, the spherical cap can be easily cut off at the base oil (Figs 8 and 13a ). The contact zone is small (<1 mm diameter) in the untextured plate. After GP addition, the particle hardly exists at the smooth tribological interface during the sliding process ( Fig 13b ). In the textured surface, the holes can store several particles and release and improve the contact state of the friction interface (bearing or formation of protective film) ( Fig 13c–13e ). However, exceedingly dense textures will cause the rubbing surface to be excessively rough, resulting in increased wear loss ( Fig 13e ). The area ratio of the texture should neither be exceedingly large nor exceedingly small. Raman spectroscopy determined the presence of GP on the wear scars after ultrasonic washing ( Fig 8d ). Two scenarios may explain this finding First, (a) GP may have adsorbed on the wear surface and formed a tribofilm. Second, (b) GP and lose materials may have collectively formed the debris and protected the interface. The interlayer friction behavior is affected by the GP structure, such as stacking form, relative sliding direction, size, defect, and number of layers [ 25 – 26 ]. However, many other forms of mechanisms exist and further research is needed to understand these processes better. 10.1371/journal.pone.0152143.g013 Fig 13 Schematic of the mechanism of the role of the GP additive in the texture plate."
} | 3,783 |
39856709 | PMC11761781 | pmc | 7,942 | {
"abstract": "Background\n Seed-associated microorganisms play crucial roles in maintaining plant health by providing nutrients and resistance to biotic and abiotic stresses. However, their functions in seed germination and disease resistance remain poorly understood. In this study, we investigated the microbial community assembly features and functional profiles of the spermosphere and endosphere microbiomes related to germinated and ungerminated seeds of Astragalus mongholicus by using amplicon and shotgun metagenome sequencing techniques. Additionally, we aimed to elucidate the relationship between beneficial microorganisms and seed germination through both in vitro and in vivo pot experiments. Results Our findings revealed that germination significantly enhances the diversity of microbial communities associated with seeds. This increase in diversity is driven through environmental ecological niche differentiation, leading to the enrichment of potentially beneficial probiotic bacteria such as Pseudomonas and Pantoea. Conversely, Fusarium was consistently enriched in ungerminated seeds. The co-occurrence network patterns revealed that the microbial communities within germinated and ungerminated seeds presented distinct structures. Notably, germinated seeds exhibit more complex and interconnected networks, particularly for bacterial communities and their interactions with fungi. Metagenome analysis showed that germinated seed spermosphere soil had more functions related to pathogen inhibition and cellulose degradation. Through a combination of culture-dependent and germination experiments, we identified Fusarium solani as the pathogen. Consistent with the metagenome analysis, germination experiments further demonstrated that bacteria associated with pathogen inhibition and cellulose degradation could promote seed germination and vigor. Specifically, Paenibacillus sp. significantly enhanced A. mongholicus seed germination and plant growth. Conclusions Our study revealed the dynamics of seed-associated microorganisms during seed germination and confirmed their ecological role in promoting A. mongholicus seed germination by suppressing pathogens and degrading cellulose. This study offers a mechanistic understanding of how seed microorganisms facilitate successful seed germination, highlighting the potential for leveraging these microbial communities to increase plant health. \n Video Abstract Supplementary Information The online version contains supplementary material available at 10.1186/s40168-024-02014-5.",
"conclusion": "Conclusion In summary, our study provides compelling evidence for the structural and functional contributions of seed-associated microorganisms in facilitating seed germination and early plant growth. We demonstrated their crucial role in pathogen suppression and cellulose degradation, thereby enhancing seed vigor. The observed shift from fungal dominance to bacterial dominance may represent an adaptive strategy, allowing plants to benefit from different microbial partners at various stages of their early development. Our findings suggest that seeds may actively recruit beneficial microorganisms during germination to combat pathogenic fungi, and add to our understanding of how plants “cry for help” during early life stages, but the mechanism needs to be explored in more detail. Importantly, inoculation with Paenibacillus sp. effectively enhanced the germination and stimulated the growth of Astragalus seeds. These findings highlight the immense potential of seed-associated microorganisms in improving seed germination and plant growth, underscoring the importance of these inconspicuous microbial resources within seeds.",
"introduction": "Introduction Plant-associated microbial communities profoundly influence various plant traits, including growth promotion [ 1 ], drought tolerance [ 2 ], pathogen resistance, and overall plant health maintenance [ 3 ]. Emerging studies have demonstrated that the microbiomes residing in the rhizosphere and phyllosphere play pivotal roles in enhancing plant adaptations, particularly in bolstering resistance against pathogens [ 4 , 5 ]. For example, rhizosphere microbes such as Trichoderma and Bacillus could help plants resist a range of diseases, including tomato wilt [ 6 , 7 ]. Despite their crucial role as a primary source of the plant microbiome, particularly in shaping early rhizosphere microbial communities, plant seed-associated microorganisms remain largely underexplored regarding their beneficial functions for plant health and adaptability [ 8 ].\n Seeds represent the beginning and end of the plant life cycle [ 9 ], providing a unique habitat (ecological niche) for numerous microorganisms. Bacterial isolates from plant seeds perform diverse functions, including nitrogen fixation, phosphorus solubilization, phytohormone synthesis, and antimicrobial compound production, potentially contributing to plant fitness [ 10 – 12 ]. Importantly, seed microbiomes can be vertically transmitted between generations via seeds [ 13 , 14 ], suggesting that the conserved traits of seed microorganisms offer new opportunities for agricultural production. The spermosphere, the immediate interface for seed-microbe interactions, represents the most proximal ecological niche to the seed [ 15 ]. Microbes in the spermosphere community are derived primarily from the surrounding soil and are recruited through seed exudation during germination [ 16 ], and their assembly is dependent on the soil type and varying seed exudation characteristics. Although germination-driven dynamics of seed endophytic microorganisms have been demonstrated [ 17 – 19 ], the relationship between germination and the assembly of seed-associated microbial communities, particularly regarding the structural and functional characteristics of spermosphere microorganisms, remains inadequately understood. Successful seed germination is influenced by both beneficial endophytic microorganisms within the seeds and environmental factors, as well as seed-borne pathogens. Seed-borne pathogens, which reside within seeds and are difficult to detect and identify, can spread globally due to animal and human activities, posing a significant threat to plant health. Some pathogens, such as Fusarium , can produce toxins that pose serious health risks to animals and humans [ 20 ]. Studies have documented the transgenerational transmission of various pathogens through seeds, such as Fusarium [ 21 ] and Burkholderia [ 12 ]. These pathogens can lead to seedling wilt or germination failure, potentially causing substantial crop losses. However, natural soil microorganisms can assist plants against seed-borne Fusarium [ 21 ]. For example, microbial communities recruited by seeds can inhibit pathogenic fungi by preventing their colonization of the seed surface [ 22 ]. Additionally, seed-borne bacteria are crucial for seed germination and growth. Studies have shown that rice seeds treated with bacterial antibiotics significantly reduced seed germination rates, while reinoculation with Enterobacter asburiae and Pantoea dispersa restored seedling growth and development [ 23 ]. Moreover, Sphingomonas melonis can suppress the seed-borne pathogen Burkholderia plantarii by producing metabolites and promoting the growth of seedlings [ 12 ]. These findings suggest that seed-associated microorganisms are involved in important processes related to seed germination and plant development. Astragalus membranaceus Bge. var . mongholicus (Bge.) Hsiao (hereafter, Astragalus ), a perennial legume with high economic and medicinal value in traditional Chinese medicine, faces challenges due to its low germination rate under natural conditions [ 24 ]. The factors that affect seed germination of Astragalus include insufficient seed vitality and external environmental conditions. Specifically, physical dormancy, resulting from mechanical obstruction of the seed coat, and physiological dormancy, induced by chemical signals, are significant factors that impede the germination of Astragalus seeds. Traditional mechanical treatments and chemical agents exert adverse effects on the growth and health of Astragalus in subsequent developmental stages. Harnessing microbial engineering to promote Astragalus seed germination and protect plant health may be another alternative and effective strategy for the future [ 25 ]. Although studies have shown that rhizosphere microorganisms can assist Astragalus in resisting Fusarium wilt disease [ 26 , 27 ], the assembly, interactions and functions of seed-associated microbial communities during Astragalus seed germination and their impact on plant health remain to be elucidated. In this study, we employed amplicon sequencing and shotgun metagenomic sequencing to explore the microbial community assembly and functional adaptation of seed in germinated and ungerminated Astragalus seeds. We further validated the link between seed microorganisms, seed germination, and disease resistance through culture-dependent methods. Specifically, we aimed to address the following questions: (1) How does seed germination affect the spermosphere and endophytic microbiome signatures of seeds? (2) Which beneficial microorganisms have been harnessed by the seed? and (3) How do these microorganisms facilitate seed germination and protect plant health?",
"discussion": "Discussion The intricate mutualistic relationship between seed microorganisms and plants is a fascinating area in the field of plant-soil-microbe interactions, particularly in the early stages of plant life [ 28 ]. During seed germination, beneficial microorganisms protect against successful seed germination by producing phytohormones and alleviating biotic and abiotic stresses [ 9 ]. However, it remains unclear whether germination failure is attributed to inherent seed vitality issues or interactions with microbes. Using imbibed seeds as research material allowed us to more effectively demonstrate the process of seed microbiome-mediated seed germination by partially excluding the influence of insufficient seed vitality. Furthermore, the dynamics of the overall composition and functional adaptations of the microbial community during seed germination remain elusive. Our investigation revealed that the diversity of endophytes and spermosphere soil microbial communities associated with germinated seeds was significantly greater than that associated with ungerminated seeds. The bacterial communities associated with germinated seeds presented a relatively high abundance of taxa with potential growth-promoting characteristics. Conversely, the enrichment of pathogens serves as a significant limiting factor for seed germination in Astragalus . Analysis of the metagenomic data revealed that the microbiome present in the spermosphere of germinated seeds performs functions involving pathogen inhibition and cellulose degradation. Further experiments with inoculated strains revealed that seed-associated microorganisms promoted seed germination and enhanced seed vitality by inhibiting pathogenic fungi and degrading cellulose (Fig. 7 ). These findings provide fundamental evidence for the potential of manipulating seed-associated microbial communities to increase seed germination and protect plant health. Fig. 7 Conceptual diagram of microbiome-mediated germination of Astragalus seeds Germination leads to a shift in seed microbial communities from fungal to bacterial dominance The germination process profoundly reshapes the assembly of seed-associated microbial communities. Previous studies conducted under sterile soil conditions indicated a reduction in the diversity of seed endophytic bacteria during germination [ 18 , 19 ]. However, our findings in natural soil reveal a contrasting pattern. We observed a significant increase in microbial diversity associated with germinating seeds. This heightened diversity is likely driven by the secretion of a complex blend of exudates from the germinating seeds, creating a nutrient-rich spermosphere that attracts and facilitates the colonization of a wider range of microorganisms [ 29 ]. The microbial communities colonizing germinating seeds presented greater complexity, which was particularly evident in the early stages of seed germination. This high complexity may be due to the diverse exudates offering multiple carbon sources and promoting microbial interactions [ 30 ]. Our observations revealed a succession of microbial communities during seed germination, with fungi initially dominating the microbial network. This early fungal dominance is likely due to their rapid colonization of available niches and ability to utilize these recalcitrant carbon sources [ 31 ]. Interestingly, we found that the potential pathogen Fusarium was enriched in early ungerminated seeds, whereas in the later stages of germination, beneficial Firmicutes and Basidiomycota became increasingly abundant. For example, some Basidiomycota are well known for forming mycorrhizal fungi with plants, increasing nutrient uptake [ 32 ]. Similarly, Firmicutes , particularly Bacillu s species, are recognized for promoting plant growth and disease resistance [ 21 ]. This succession from a pathogenic fungal-dominated community to one rich in beneficial bacteria and symbiotic fungi highlights the intricate ecological choreography unfolding during seed germination. Our findings suggest that as seeds transition to seedlings, they actively shape their microbial environment, potentially selecting for microorganisms that contribute to their growth and health. Seeds recruit beneficial microorganisms to assist in seed germination Many Fusarium species are globally distributed seed-borne pathogens that hinder seed germination and threaten plant health, often leading to severe crop yield losses in various species, including soybean, maize, and wheat [ 21 , 33 , 34 ]. However, the impact of seed-borne Fusarium species on seed germination in Astragalus remains unknown. Our results revealed an enrichment of Fusarium in ungerminated Astragalus seeds, with this genus emerging as a keystone species in the co-occurrence network of the ungerminated seed microbiome, indicating its dominant role in shaping the microbial community within ungerminated seeds. Pathogenicity experiments demonstrated that Fusarium solani significantly inhibited Astragalus seed germination. The mechanism underlying the detrimental effects of Fusarium is attributed to its production of toxins and enzymes that disrupt crucial cellular processes. For example, Fusarium graminearum produces deoxynivalenol, a mycotoxin that disrupts normal cellular functions by inhibiting protein synthesis [ 35 ]. Similarly, Fusarium solani secretes cell wall-degrading enzymes, including cutinase, which contribute to plant tissue decay and pathogenicity [ 33 , 36 ]. In addition, a study has shown that Fusarium solani infection can increase alanine and fatty acids during seed germination, thereby reducing the seed germination rate [ 37 ]. In contrast to ungerminated seeds, germination activates metabolic processes that recruit and enrich a greater number of microorganisms in the endosphere and spermosphere through the release of seed exudates [ 15 , 38 ]. We observed an enrichment of potentially beneficial microorganisms, including Pseudomonas and Pantoea , which are known to antagonize pathogenic fungi and promote plant growth [ 39 , 40 ]. Our experiments confirmed the ability of Pseudomonas and Pantoea to promote seed germination. The presence of pathogens, such as tomato Fusarium wilt [ 41 ], wheat Fusarium head blight [ 42 ] and Diaporthe citri melanose pathogens [ 4 ], alters the composition of plant-associated microbial communities. The symbiotic relationship between the microbiome and the plant host, which functions as a “holobiont,” serves as a defence mechanism against pathogen stress [ 43 ]. Plants exhibit a remarkable ability to recruit beneficial microbes in response to pathogen-induced stress [ 4 , 44 , 45 ]. Intriguingly, our findings revealed that ungerminated seeds in the early stages harbored diverse potentially beneficial microorganisms, including Paenibacillus and Bacillus . However, not all isolated Bacillus species have a positive effect on seed germination. Thus, considering the variations in their metabolic diversity and functions [ 46 ], the diversity of Bacillus species and strains may play a significant role in seed germination and plant growth. Subsequent disease suppression assays demonstrated the efficacy of these bacteria in inhibiting the growth of pathogenic fungi and increasing Astragalus seed germination rates (Fig. 5 c; Tale S11). Furthermore, under pathogen stress, plants may mitigate this stress by recruiting beneficial microbes [ 3 ]. This “cry-for-help” strategy may partially elucidate the observed enrichment of Paenibacillus and Bacillus in ungerminated seed. Nevertheless, the exact mechanism remains unclear, which is a limitation of our study. Specific functions of seed microorganisms stimulating Astragalus seed germination The microbial communities associated with germinated seeds present diverse functions crucial for promoting germination, particularly pathogen inhibition and cellulose degradation. Our analysis revealed fungal cell wall degradation in the spermosphere, including an increased abundance of genes encoding CBM50 (chitin or peptidoglycan binding) and CH23 (chitin degradation). This finding is consistent with previous research linking chitinase activity to disease suppression in sugar beet infected with Rhizoctonia solani [ 47 ]. Furthermore, we observed enrichment of rhamnosidase synthesis-related functions (GT2 and GT41 glycosyltransferase families), which are often associated with Paenibacillus species known to mediate plant–microbe interactions and confer pathogen resistance [ 48 ]. Notably, the microbial cell motility functional pathway was enriched in the germinated seeds, potentially enabling Pseudomonas and Bacillus to effectively colonize the seed environment [ 49 ]. This targeted recruitment of beneficial microbes with specific functions underscores the active role of plants in shaping their microbiome to promote germination success. Another key mechanism by which microbes support seed germination is microbial cellulose degradation, which releases nutrients that directly nourish germinating seedlings [ 50 ]. In our study, we identified numerous cellulose-degrading enzymes, such as GH94 and AA3, within the spermosphere, emphasizing the importance of this process. Moreover, the positive correlation between Pantoea and Paenibacillus and GH94 was confirmed in cellulose degradation experiments, supporting the view that seed-associated microorganisms promote germination via degrading cellulose. This aligns with previous studies showing that Pantoea , Bacillus , and Paenibacillus , which are common seed endophytes, have robust cellulose-degrading abilities, effectively converting cellulose into readily available carbon sources [ 51 – 53 ]. The presence of these specific bacteria and their associated cellulose degradation enzymes further highlights their critical role in enhancing seed vigor and germination success."
} | 4,838 |
39980378 | PMC11852197 | pmc | 7,943 | {
"abstract": "Engineered living\nmaterials (ELMs) integrate aspects of material\nscience and biology into a unique platform, leading to materials and\ndevices with features of life. Among those, ELMs containing microalgae\nhave received increased attention due to the many benefits photosynthetic\norganisms provide. Due to their relatively recent occurrence, photosynthetic\nELMs still face many challenges related to reliability, lifetime,\nscalability, and more, often based on the complicated crosstalk of\ncellular, material-based, and environmental variables in time. This\nViewpoint aims to summarize potential avenues for improving ELMs,\nbeginning with an emphasis on understanding the cell’s perspective\nand the potential stresses imposed on them due to recurring flaws\nin many current ELMs. Potential solutions and their ease of implementation\nwill be discussed, ranging from choice of organism, adjustments to\nthe ELM design, to various genetic modification tools, so as to achieve\nELMs with longer lifetime and improved functionality.",
"introduction": "1 Introduction to Microalgae-Based Engineered\nLiving Materials In so-called engineered living materials (ELMs),\nliving cells constitute an intrinsic component of the material and\nimpart a wide variety of life-like functions difficult to achieve\nin more traditional materials (e.g., sense-and-respond; self-repair;\nself-cleaning; or photo-, chemo-, thermo-, and mechanosensing functionalities).\nThe term “engineered” herein applies to both the material\nand the organism. The cells within ELMs are typically immobilized\nin a matrix of biological or synthetic polymers. Functional living\nmaterials have been proposed for usage in a wide range of potential\napplications, including smart textiles and wearable devices, 1 , 2 soft robots and actuators, 3 and methods\nto control materials’ growth and mechanical properties. 4 − 7 Microalgae are a highly diverse polyphyletic group of photosynthetic\npro- and eukaryotes. They are important primary producers for many\necosystems, 8 − 12 and jointly with plants they are responsible for the oxidizing atmosphere\nmany current life forms depend on. With the urgent need to face the\nmanifold environmental problems caused by anthropogenic activities,\nmicroalgae are investigated as a sustainable alternative to produce\nfuels, food products, or high-value metabolites. 13 − 15 Of particular\ninterest are ELMs containing microalgae, which endow\nphotosynthetic function to the material ( Figure 1 ). 16 − 19 Based on their ability to capture and produce gases\n(CO 2 , NO X , SO X and O 2 ,\nH 2 , respectively), photosynthetic ELMs are explored to\ntreat wastewater, 20 air, or soil 21 and to deliver O 2 to engineered tissues,\norganoids, or wounds. 22 Inspired by liquid\nculture systems, they are also explored as a simple production platform\nfor fatty acids, 23 nutritional additives\n(e.g., vitamins, antioxidants), or carbonate minerals. 5 , 24 Moreover, the electrons generated in the photosynthetic light reaction\nmay be utilized in biophotovoltaic devices. 25 Figure 1 Representative\nfunctions of photosynthetic ELMs. When designing\nphotosynthetic ELMs, the ability to fix CO 2 often represents\nthe primary motivation. This can also be combined with other functions\nsuch as electricity generation, growth platforms, biosensing, or the\ncreation of novel regenerative materials. 1.1 Advantages of ELMs as Growth Platform In most microalgal\nELMs, cells are captured in flexible hydrogels,\nwhich present several advantages as a simple growth platform that\ncan hold water and nutrients. To some degree, hydrogels mimic biofilms\nfound in nature: Both form a physical barrier, which shields the cells\nfrom predatory organisms, slows down diffusion processes of toxins,\nand may absorb mechanical stresses through their viscoelastic properties.\nImmobilized microalgae are more tolerant to harmful conditions such\nas high light, grazers, 26 high temperature, 19 heavy metals, 27 or\npH, 28 and often show higher cell numbers\nper volume compared to free growth conditions. 19 , 29 − 31 However, while hydrogels include only a few main\ningredients, natural biofilms additionally contain diverse EPS (extracellular\npolymeric substances), extracellular DNA, and enzymes (e.g., proteases)\nthought to intercept and defuse harmful agents such as reactive oxygen\nspecies, 32 antibiotics, toxins, or enzymes\n(e.g., lysozyme). 33 It remains to be explored\nin which way cells in ELMs “customize” their ELM environment\nin time, for instance through the secretion of proteins or EPS. Growth in hydrogels also comes with practical benefits, such as reduced\nwater use compared to liquid culture, 29 , 34 improved handling\n(transport, harvest), 35 and potentially\nmore efficient light management. 18 , 36 Some hydrogels\nalso allow control over 3D shapes, 17 , 37 , 38 further improving space and light management. Despite their advantages, microalgal ELMs face many challenges.\nIn this Viewpoint, we discuss variables that affect the functionality\nof ELMs ( Figure 2 ),\ni.e., environmental variables like hydration or illumination; materials’\ndesign variables like porosity or mechanical properties; and phenotypic\nvariables (also linked to genotype) like lipid content, reactive oxygen\nspecies (ROS) production, or growth. Finally, we briefly discuss the\npotential avenues to address these challenges, including various genetic\nmodification tools. Figure 2 Challenges of ELMs: a complex communication of variables.\nWithin\nan ELM, the phenotype of a cell (top left; lipid content, growth rate,\netc.) is dependent on the environment into which an ELM is placed\n(top right; temperature, light, humidity, etc.). The perception and\nimpact of environmental variables is moderated by design choices of\nthe material (top middle; surface to volume ratio, water retention,\netc.). The longer the duration, the more feedback occurs between cells,\nenvironment, and material (dashed arrows), potentially leading to\nnew phenotypes. The genotype of the employed species with its inherent\ncapacities and tolerance levels defines the type, onset, and scope\nof a phenotype (bottom left; lipid content, salinity tolerance, bioluminescence,\netc.). The genotypes may also be expanded through genetic modification\ntools (bottom right; mutagenesis, genetic engineering, etc.). All\nthese variables will influence the efficiency, reliability, lifetime,\nand thus the functionality of an ELM."
} | 1,605 |
35112356 | PMC9285558 | pmc | 7,944 | {
"abstract": "Abstract Synchronous dynamics (fluctuations that occur in unison) are universal phenomena with widespread implications for ecological stability. Synchronous dynamics can amplify the destabilizing effect of environmental variability on ecosystem functions such as productivity, whereas the inverse, compensatory dynamics, can stabilize function. Here we combine simulation and empirical analyses to elucidate mechanisms that underlie patterns of synchronous versus compensatory dynamics. In both simulated and empirical communities, we show that synchronous and compensatory dynamics are not mutually exclusive but instead can vary by timescale. Our simulations identify multiple mechanisms that can generate timescale‐specific patterns, including different environmental drivers, diverse life histories, dispersal, and non‐stationary dynamics. We find that traditional metrics for quantifying synchronous dynamics are often biased toward long‐term drivers and may miss the importance of short‐term drivers. Our findings indicate key mechanisms to consider when assessing synchronous versus compensatory dynamics and our approach provides a pathway for disentangling these dynamics in natural systems.",
"conclusion": "CONCLUSIONS Understanding patterns of synchronous versus compensatory dynamics remains an ongoing challenge in community ecology. Our results demonstrate how multiple mechanisms, including environmental drivers, species demography, and dispersal can shape the timescale of synchronous versus compensatory dynamics. To date, most empirical assessments of community synchrony, particularly in terrestrial systems, have not accounted for timescale specificity. Building from recent methodological advances that allow timescale specificity to be determined with shorter timeseries of abundances (Zhao et al., 2020 ), our work points to specific mechanisms of community dynamics that, if characterized, can help us better understand synchrony and stability patterns across timescales.",
"introduction": "INTRODUCTION The extent to which communities of interspecific competitors exhibit synchronous versus compensatory temporal fluctuations and the underlying mechanisms driving these fluctuations have been of fundamental interest in ecology for decades (MacArthur, 1955 ). Community synchrony and its alternative pattern, compensatory dynamics, describe how the dynamics of species aggregate to influence community stability through time. Synchronous fluctuations of species’ abundances reduce stability and often arise when species respond similarly to environmental fluctuations (Ives, 1995 ; Loreau & de Mazancourt, 2013 ), or through facilitative interactions between species. In contrast, compensatory dynamics stabilize overall community properties, as species fluctuate in a negatively correlated manner (Peterson, 1975 ) often due to competitive interactions or opposing responses to environmental drivers (Gonzalez & Loreau, 2009 ; Ives, 1995 ; Loreau & de Mazancourt, 2013 ). Quantifying the degree of synchronous versus compensatory dynamics has emerged as a key component of several recent advances in community ecology, such as how functional diversity influences resilience and how environmental change may impact coexistence (Hallett et al., 2019 ; Lindegren et al., 2016 ). Synchronous and compensatory dynamics have often been considered mutually exclusive, as they reflect opposite correlations among abundances (e.g., Houlahan et al., 2007 ); however, there is a growing recognition that synchronous and compensatory dynamics can instead be timescale and spatial scale dependent (Downing et al., 2008 ; Lasky et al., 2016 ; Vasseur et al., 2014 ). For instance, species may be synchronous at one timescale and compensatory at other timescales (Downing et al., 2008 ; Vasseur et al., 2014 ), they may be synchronous in certain life history stages but asynchronous in others (Lasky et al., 2016 ), and they may be synchronous under some environmental conditions and asynchronous in others (Xu et al., 2015 ). A wide range of processes can influence species dynamics and correlations in species fluctuations, including environmental variation (Allstadt et al., 2015 ; Tredennick et al., 2017 ), biotic interactions (Pedersen et al., 2016 ), variability in species demographic rates (Jucker et al., 2014 ), and dispersal (Wang et al., 2019 ). While all of these processes may affect synchronous versus compensatory dynamics, many have not been explored in a timescale‐specific manner. Linking patterns of timescale‐specificity to mechanisms is essential for predicting stability patterns under global change. For example, multiple environmental drivers operating at different timescales are one potential mechanism explaining timescale‐specificity (Frost et al., 1995 ; Sheppard et al., 2016 ). If this is a primary mechanism, then shifts in the timescale of dynamics would reflect shifts in the timescale of each driver. Alternatively, different life history strategies, in which some species respond quickly to the environment while others exhibit a lagged response, are another mechanism that could drive timescale‐specific dynamics (Loreau & de Mazancourt, 2013 ). If this is the primary mechanism, then shifts in the timescale of environmental drivers may have a nonlinear effect on synchronous versus compensatory dynamics, depending on whether lagged species have sufficient time to recover (Benton et al., 2001 ). These primary effects may be further mediated by species interactions, causing emergent fluctuations to depend not only on underlying environmental drivers or species demographic rates, but also on the abundances of other species in the community (Gonzalez & Loreau, 2009 ; Loreau & de Mazancourt, 2013 ). While there is a long history in population ecology of assessing the timescale of oscillations for single‐species abundance patterns (Sheppard et al., 2016 ), and even how species interactions may modulate these oscillations (Ives, 1995 ; Ripa et al., 1998 ), at the community level many fundamental studies of synchronous versus compensatory dynamics have used simple covariance and correlation methods that aggregate across timescale (e.g., de Mazancourt et al., 2013 ; Grman et al., 2010 ; Gross et al., 2014 ; Hallett et al., 2014 ; Houlahan et al., 2007 ). However, when examining community dynamics using a timescale‐specific methodology, multiple signals of differing periodicity can be identified in a single timeseries (Downing et al., 2008 ; Vasseur et al., 2014 ). Advances in scale‐specific metrics allow us to gain a new understanding of synchronous versus compensatory dynamics (Brown et al., 2016 ; Downing et al., 2008 ; Keitt & Fischer, 2006 ; Vasseur et al., 2014 ; Vasseur & Gaedke, 2007 ), and new and less data‐intensive methodologies are opening up our ability to empirically characterize timescale‐specificity in terrestrial and aquatic systems (Zhao et al., 2020 ). These methodologies, coupled with growing open‐access and long‐term monitoring data, have the potential to expand our understanding of temporal fluctuations and their drivers across a wide range of ecosystems, with implications for connecting patterns of synchrony and stability to underlying mechanisms. Here we use simulations and empirical analyses to examine four mechanisms that can underlie timescale‐specificity of synchronous and compensatory dynamics. We first examine timescale‐specific dynamics with multiple underlying environmental drivers of species’ abundances, where environmental fluctuations occur at different timescales. Second, we assess a biotic mechanism in which species share a response to drivers but differ in their demographic rates. Third, we assess a spatial mechanism in which different timescale dynamics occur in local patches connected via dispersal, and we examine how they aggregate to affect larger‐scale metacommunity dynamics. Finally, we consider a nonstationarity mechanism in which global change may alter the timescale‐specific signature of synchronous and compensatory dynamics coupled with species‐specific threshold responses. We focus our analyses on pairwise interactions to mechanistically and directly examine how differences in species’ environmental responses and demography manifest across timescales. Our approach identifies key ecological factors that may cause synchronous and compensatory dynamics to operate at different timescales, while providing a path forward to parsing these dynamics and understanding their importance for stability in natural systems.",
"discussion": "Results and discussion Our simulated community exhibited highly synchronous dynamics at short timescale and highly compensatory dynamics at long timescales, and this expected pattern was easily discernable using the timescale‐specific variance ratio (Figure 1f,g ). In comparison, the effect of the short‐timescale driver was largely masked with the classic variance ratio (Figure 1g ). Our empirical case study at Jasper Ridge mirrored our theoretical results, such that species were synchronous on short timescales and compensatory on long timescales. Further, the classic variance ratio primarily captured the longer timescale dynamics (Figure 2 ). FIGURE 2 Applying the timescale‐specific synchrony metric to a case study at Jasper Ridge, California, USA. (a) Averaged timeseries ( ± SE) of two native annual forbs Plantago erecta and Microseris douglasii before and after gopher disturbance (disturbance occurred during the gray bar at time 1). (b) Short timescale, long timescale, and classic variance ratio for P. erecta and M. douglasii communities (average value of the metric after it was calculated on individual timeseries ± SE). (c) Averaged timeseries ( ± SE) of a native annual forb P. erecta and native perennial grass Elymus glaucus before and after gopher disturbance (gray band). (d) Short timescale, long timescale, and classic variance ratio for P. erecta and E. glaucus communities ( ± SE) There are both statistical and ecological explanations for the striking importance of long‐timescale dynamics for the classic variance ratio. Statistically, the relationship between the classic variance ratio and the timescale‐specific decomposition depends on the length of the timeseries and on differences in the amplitude of short‐term versus long‐term environmental fluctuations ( a e ) and species’ sensitivities ( ϵ ). As such, the contribution of long‐term dynamics to the variance ratio increases with both timeseries length and the amplitude of long‐term drivers. As all timeseries here are relatively long (i.e., ≥17 years), the classic variance ratio mirrors the long timescale signal. Ecologically, long‐timescale drivers may most strongly influence patterns of community synchrony for several reason. First, long‐timescale climate drivers, such as the Pacific Decadal Oscillation and the North Atlantic Oscillation, are more regular than short interannual variation in weather patterns and may therefore be more detectable in their effect on community structure (Chiba et al., 2012 ; Downing et al., 2008 ). Long‐term climate drivers like the Pacific Decadal Oscillation may underlie the pattern of long‐timescale compensatory dynamics we observed at Jasper Ridge (Figure 2a,b ), where annual species often rebound quickly from frequent disturbance (such as gophers), leaving only a fleeting signature on patterns of community synchrony (Figure 2a,b ). Second, long‐timescale fluctuations may reflect the differing role that rare versus common events have on populations. Daily temperature fluctuations and summer heat waves, for example, can both influence population dynamics. The effect of daily temperature fluctuations on populations are typically felt over short timescales, whereas high mortality due to a rare and extreme heat wave may have a long‐lasting signal on population dynamics (Lindström et al., 2012 ). While the effects of the timescales of environmental fluctuation and disturbance have been explored in the context of population synchrony and extinction risk (Heino, 1998 ; Schwager et al., 2006 ), if some species can tolerate extreme events while others cannot, extreme events may leave a long‐lasting signature of compenasatory dynamics in communities (Till et al., 2019 ).\n\nResults and discussion The presence of species with different environmental response rates can reduce synchrony and even generate compensatory dynamics, even when all species share the same directional response to the environmental driver, as shown in our simulations (Figure 3b–d ). The compensatory effect of lagged responses was strongest when the timescale of the lag matched the timescale of the environmental driver. For example, the presence of a species with a slow growth rate generated compensatory dynamics across all timescales (Figure 3c,g ). Across systems, differences in the timescale of species responses versus recovery times in poor environmental conditions may drive timescale‐dependent patterns. We observed this at Jasper Ridge, in which both the annual and perennial species shared an initial, negative response to disturbance that enhanced short‐timescale synchrony (Figure 2c,d ). However, the perennial species had a slower recovery rate (i.e., a lagged response) that promoted compensatory dynamics at longer timescales by delaying its recovery relative to the annual species (Figure 2c,d ). These difference in recovery rate caused timescale‐specific dynamics, even when species responded similarly to underlying abiotic drivers, such as disturbance. Endogenous population cycles, often induced by fast growth rates, may decouple some species’ fluctuations from the environment (Haynes et al., 2019 ). As such, species cycling at different rates will be less synchronous than predicted based solely on environmental response, although this effect is timescale dependent (Figure 3d,h ). In our simulation, a species with a fast growth rate more closely tracked short‐timescale environment fluctuations, reflecting the fact that environmental conditions changed before the species internal dynamics dominated its population cycles. As a result, the fast‐growing species was synchronous with a species whose growth tracked the environment at short timescales (Figure 3d,h ). At long timescales, however, the population cycles of the fast‐growing species became increasingly decoupled from the environmental driver, and correspondingly, the other species (Figure 3d,h ). Consequently, internal dynamics generated by fast growth rates may also promote increased compensatory dynamics, particularly in communities structured by long‐timescale drivers.\n\nResults and discussion Applying the timescale‐specific variance ratio at multiple spatial scales elucidated the interplay between local and regional processes in landscapes with spatial heterogeneity (Figure 4a,d ). For example, in the absence of dispersal, synchrony in abundances was driven solely by within‐patch dynamics, as expected (Figure 4b,e ). Here, the timescale‐specific variance ratios recover the classic variance ratio. However, at the larger landscape scale (Figure 4g ), the signatures of each patch's environmental fluctuations were detected with the timescale‐specific variance ratio, while the classic variance ratio was again biased toward the long‐term drivers occurring in patch 2 (Figure 4i ). Dispersal between patches was detectable in a heterogeneous landscape using the timescale‐specific metric, highlighting how spatial processes can impact our interpretation of temporal synchrony (Figure 4c,f,h ). In this case, synchronous dynamics from short‐term drivers (in patch 1) and compensatory dynamics from long‐term drivers (in patch 2) were evident in both patches (Figure 4i vs. j). Within patches, comparing the timescale‐specific variance ratio with the classic ratio elucidated the focal patch; the classic variance ratio was synchronous in patch 1 but compensatory in patch 2 (Figure 4j ). This shows how a temporally‐focused method can detect spatial heterogeneity and dispersal effects when applied at different levels of spatial aggregation. In more complex scenarios, we expect these spatiotemporal patterns to also yield signatures in the timescale‐specific variance ratio. For example, the order in which species arrive can alter long‐term community composition and patterns in synchrony (Fukami et al., 2016 ). Furthermore, arrival itself can depend on fluctuations in underlying environmental conditions that alter species’ dispersal kernals and propagule density, yielding complex spatiotemporal dynamics (Sullivan et al., 2018 ). At larger spatial scales, spatial patterning and interdependence between patches plays a key role in determining stability. Here we compare patches with different underlying environmental drivers. However, previous work highlights how overall landscape stability depends not only on trade‐offs and compensatory dynamics of species within patches, but also on trade‐offs among patches (Wang et al., 2019 ; Wilcox et al., 2017 ). These among patch trade‐offs can create compensatory flucutuations among patches, stabilizing overall landscape fluctuations. Our results provide additional insight into the role of dispersal and how connectivity between plots manifest as different synchrony and stability patterns depending on the scale of spatial aggregation.\n\nResults and discussion Under baseline historic conditions, the combination of competition and species‐specific environmental thresholds yielded strong compensatory dynamics on long timescales and weaker compensatory dynamics with the classic variance ratio (Figure 5b,e,h ). This occurred even though species responded in the same manner to environmental fluctuations. Compensatory dynamics driven by competition overshadowed synchronizing effects of a shared environmental driver, as species 2 responded to the environment only relatively rarely. In contrast, global change increased synchrony at all timescales, with dynamics intermediate between those observed under the historic versus new environment (Figure 5c,f,i ). Once the environment settled on a new equilibrium, our model yielded increased synchrony. This increase in synchrony occurred because the environment was more often above the threshold where both species responded to the driver. Environmental fluctuations therefore were more important under e new , while competition dominated dynamics under e historic conditions (Figure 5d,g,j ). Our model results hint that increased environmental forcing from more extreme climate conditions in the future may overshadow competitive effects, yielding an increase in community synchrony and a loss of compensatory dynamics. To date, the effects of climate change on synchrony have been examined primarily in a single‐species context or in relation to phenological synchrony between pairs of species. For example, increased spatial synchrony between populations has been observed among populations of damselfish in the Great Barrier Reef (Cheal et al., 2007 ), North American wintering birds (Koenig & Liebhold, 2016 ), and Greenland caribou (Post & Forchhammer, 2004 ). In contrast, however, climate‐induced shifts in phenology can disrupt synchrony in plant–herbivore (Hunter & Elkinton, 2000 ; Tikkanen & Julkunen‐Tiitto, 2003 ), predator–prey (Logan et al., 2006 ; Sanford, 1999 ), and host–parasitoid interactions (Hance et al., 2007 ; Klapwijk et al., 2010 ; Visser & Holleman, 2001 ), causing increased extinction risk for codependent species. Our model suggests that, as for single‐species populations, synchrony within communities may increase with climate change. The application of timescale‐specific methods in empirical communities, including Fourier transformations as employed here, and wavelet analyses when longer timeseries are available, provides a pathway for assessing whether natural communities match theoretical expectations."
} | 5,006 |
38621291 | PMC11072716 | pmc | 7,946 | {
"abstract": "When a droplet interacts with a water-repellent surface,\nits triple-phase\ncontact line typically exhibits varying contact angles, which can\nvary from point-to-point across the surface. Consequently, measuring\nthe contact angles along the contact line would provide a better representation\nof the wetting properties of the surface than a single average contact\nangle. However, an effective method for estimating the local contact\nangle along the contact line on opaque hydrophobic surfaces is currently\nlacking. Here we present a method that combines through-drop imaging\nof the wetting interface during a sliding experiment with Finite Element\nModeling of the droplet to estimate contact angle values along the\ncontact line. Using this method, the mean advancing and receding contact\nangles were measured on four types of hydrophobic samples with contact\nangles between 99 and 178.9°. The method was further used to\nproduce detailed advancing and receding contact angle maps of surfaces\nwith wetting patterns with an unprecedented resolution of 3 μm.",
"conclusion": "Conclusions The method presented estimates advancing\nand receding contact angle\nvalues along the contact line of a droplet probe sliding on hydrophobic\nsurfaces with unprecedented resolution. The approach combines through-droplet\nimaging with FEM modeling of the shape of the droplet to obtain detailed\ninformation about the wetting properties of hydrophobic surfaces.\nThe technique showcases how a millimeter-sized droplet can be used\nto obtain wetting information on the micrometer scale, the same scale\nas the surface features. The ability to measure the contact angles\nalong the CL of sliding droplets provides direct point-by-point maps\nof the wetting properties of inhomogeneous surfaces, where previous\ntechniques provided only average values. This ability is critical\nin both the development of novel wetting surface treatments and the\nstudy of natural surfaces. The method was used to characterize\nthree types of nanograss superhydrophobic\nsurfaces and a hydrophobic SAM sample. The measurements revealed the\ngreat homogeneity of the surfaces. The mean CL shape and the mean\ncontact angle values along the CL were calculated, showing great homogeneity\nof these surfaces. The data show that during the sliding experiments,\nthe advancing contact angle profile along the CL is observed throughout\nmost of the front of the CL while the receding contact angle is observed\non a large extent of the back of the CL, with transition zones in\nbetween. Moreover, the method reveals important adaptation phenomena\nthat may directly affect wetting measurement methods in general, where\nthe longer the liquid interacts with the test surface, the lower the\ncontact angle tends to become. To complement the results, the\nability to map the contact angle\nvalues along the contact line was used to produce detailed wetting\nmaps of the patterned surfaces with a resolution of 3 μm. On\nthese maps, a pair of advancing and receding contact angle values\nare obtained at each point of the sample. The maps are able to distinguish\ncontact angles between zones with very similar wetting properties,\ndespite the values being very similar between zones. The boundary\nbetween the modified and nonmodified areas forms locations where pinning\nof the CL is likely to occur. In turn, such pinning may lead to abrupt\nCL movements during sliding. The insights provided by this technique\nhave the potential to contribute\nto the growing body of knowledge on surface wetting characterization\nand may inform future experimental design and interpretation of the\ndevelopment of highly sensitive hydrophobic and superhydrophobic surfaces.",
"introduction": "Introduction A multitude of natural surfaces are water-repellent,\nsuch as lotus\nleaf, 1 rose petals, 2 butterfly wings, 3 and bird feathers. 4 Artificial surfaces with similar properties have\nbeen created with applications in fog-collection, 5 self-cleaning, 6 and micro- and\nnanoassembly. 7 , 8 The performance of these surfaces\ndepends on their wetting characteristics, which can vary spatially\nacross their surface. In particular, the wetting state of a droplet\nis defined by the contact angle formed with such surfaces. 9 Moreover, the state of the droplet is also dependent\non the history of the interaction, 10 leading\nto contact line (CL) shapes beyond the ideal circular shape and contact\nangle values that can vary along the CL at any given moment. 11 For this reason, measuring the contact angles\nalong the CL would better describe the wetting properties of such\nsurfaces than a single average contact angle value. Currently,\nthere is no effective method for estimating the local\ncontact angle along the CL on opaque hydrophobic surfaces. For hydrophilic\nsamples the CL shape can be directly observed from the top-view, 12 for example, using the tilting plate method\nwhere the sample is slowly tilted and gravity pulls the droplet downhill. 13 Alternatively, the CL shape can be controlled\nto a prescribed shape. 14 However, for hydrophobic\nsurfaces, direct top-view observation is challenging because the CL\nis covered by the body of the droplet. Side-view contact angle goniometry\nis unable to accurately determine the shape of the CL and is limited\nto measuring the contact angle at only two points located on opposite\nsides of the droplet. In the superhydrophobic regime, the problem\nof an obscured CL is particularly noticeable, making measurements\nincreasingly inaccurate at higher contact angle values. 15 , 16 Although there are many techniques for measuring the CL shape in\nthe hydrophobic regime, they either require specialized transparent\nsurfaces, necessitate a stationary droplet during measurement, or\ncannot measure CL progression. Transparent samples can be imaged from\nunderneath, 17 , 18 using, e.g., inverted scanning\nlaser confocal microscopy that can measure the three-dimensional (3D)\nshape of the droplet-sample-air system in real time. 19 Alternatively, the CL shape can be estimated from the apparent\ndiameter observed by rotating the side-view camera around the droplet-sample\nsystem, 20 but the technique is limited\nto CLs with convex shape and the droplet must remain static during\nthe measurement process. Many of these methods lack control over both\nthe shape and the mobility of the droplet, complicating modeling of\nthe interaction and subsequent contact angle estimation. For instance,\nin the tilting plate method, the droplet velocity is not controlled.\nOr, methods that use a micropipette to dispense and hold the liquid\nresult in a droplet with variable and irregular shape. Recently, we\ndemonstrated that a through-drop imaging method allows visualizing\nthe shape of the wetting interface on opaque surfaces. 21 The method can also estimate the mean contact\nangle with a precision down to 0.2°, where the results were verified\nusing a Digital Holography Microscope (DHM) for contact angles above\n178°. We also reported a method combining the through-drop imaging\nwith force sensing that can separately measure the wetting forces\ndue to surface tension from the forces due to the Laplace pressure. 22 However, both methods assume the CL to be circular\nand do not estimate the local contact angles along the contact lines. Here we report a method for estimating the contact angle values\nalong the CL from the direct observation of the wetting interface\nduring sliding experiments on hydrophobic and superhydrophobic samples.\nThe method combines the CL shape obtained from through-drop images\nwith precise droplet volume and position control. Based on this information,\nwe calculate the shape of the droplet via Finite Element Modeling\n(FEM), from which the contact angles along the CL can be estimated.\nWe measured average CL contact angle profiles on three types of superhydrophobic\nnanograss surfaces and a hydrophobic self-assembled monolayer surface,\nwhere the surface contact angle was found to vary by as little as\n0.4° across the 1 × 0.35 mm 2 measurement area.\nThe segments of the CL exhibiting the advancing contact angle, θ a , receding contact angle, θ r , and values between them are also identified. Furthermore,\nwe can generate θ a and θ r maps for\npatterned surfaces with an unprecedented spatial resolution of 3 μm,\nat the same scale as the surface features of ≤1 μm."
} | 2,076 |
38412971 | PMC10898968 | pmc | 7,947 | {
"abstract": "Cooperation is prevalent across bacteria, but risks being exploited by non-cooperative cheats. Horizontal gene transfer, particularly via plasmids, has been suggested as a mechanism to stabilize cooperation. A key prediction of this hypothesis is that genes which are more likely to be transferred, such as those on plasmids, should be more likely to code for cooperative traits. Testing this prediction requires identifying all genes for cooperation in bacterial genomes. However, previous studies used a method which likely misses some of these genes for cooperation. To solve this, we used a new genomics tool, SOCfinder, which uses three distinct modules to identify all kinds of genes for cooperation. We compared where these genes were located across 4648 genomes from 146 bacterial species. In contrast to the prediction of the hypothesis, we found no evidence that plasmid genes are more likely to code for cooperative traits. Instead, we found the opposite—that genes for cooperation were more likely to be carried on chromosomes. Overall, the vast majority of genes for cooperation are not located on plasmids, suggesting that the more general mechanism of kin selection is sufficient to explain the prevalence of cooperation across bacteria.",
"conclusion": "5 . Conclusion Combining our findings with both previous analyses and recent theoretical modelling, there is now convincing evidence that plasmid transfer does not specifically favour cooperation in bacteria [ 26 , 42 ]. We are not saying that horizontal gene transfer has no influence on the evolution of genes for cooperative traits in bacteria. Horizontal gene transfer is highly prevalent in bacteria, and therefore will likely influence many aspects of how genes for cooperative traits are maintained in and spread though populations. However, it will also influence the spread of many genes for non-cooperative traits in the same or analogous ways. Many kinds of genes in bacteria are transferred via horizontal gene transfer, allowing for the rapid spread of traits such as antibiotic resistance and virulence factors, irrespective of whether they are cooperative. Similarly, there are many other factors, such as gene complexity, that can influence where genes are more likely to be carried, irrespective of whether they are for cooperative traits [ 38 ].",
"introduction": "1 . Introduction Cooperation appears to play a key role in the growth and success of many bacteria [ 1 – 3 ]. Bacteria produce and secrete a range of molecules that provide benefits to the local population of cells, and therefore act as cooperative ‘public goods’. Examples include iron-scavenging siderophores and enzymes that can break down host defences [ 4 – 7 ]. The problem with such cooperation is that the benefit of producing public goods is potentially shared with ‘cheat’ cells that do not produce the public good, which could lead to cooperation being unstable [ 8 ]. A likely solution is that the clonal growth of bacteria means that public goods are shared mainly with relatives (clonemates) that also carry the gene for cooperation, an example of kin selection [ 9 ]. Horizontal gene transfer has been suggested as another mechanism to stabilize cooperation in bacteria [ 10 – 24 ]. Horizontal transfer of genes for cooperation could increase relatedness at those loci and prevent invasion by non-cooperative cheats. Horizontal gene transfer could even lead to high relatedness at the loci for cooperation between cells that are not genetically related across the rest of the genome. This possibility has been explored theoretically in particular for plasmids, which are extra-chromosomal sequences found across bacteria, and which are often capable of transferring, along with all the genes they carry, to other cells [ 10 – 14 ]. A key prediction of this hypothesis is that genes for cooperation should be overrepresented on more mobile parts of the genome, such as plasmids compared with chromosomes [ 19 , 25 ]. If cooperation is favoured by horizontal gene transfer, then genes for cooperative traits are more likely to be maintained if they are on plasmids, or can be preferentially moved onto plasmids. Plasmids usually carry far fewer genes than chromosomes, suggesting there is likely to be some constraints associated with moving extra genes onto a genome's plasmids. Therefore, to control for the difference in chromosome and plasmid size, the prediction of the hypothesis is expressed as a relative proportion of genes: all else being equal, if cooperation benefits from horizontal gene transfer such as via plasmids, a higher proportion of plasmid genes should code for cooperative traits compared with the chromosome (i.e. overrepresented on plasmids). Our recent comparative genomics analysis of 51 species of bacteria did not find support for the horizontal gene transfer hypothesis [ 26 ]. We tested the hypothesis by examining the genomic location of genes which coded for proteins that are secreted by bacteria into the extracellular space (genes coding for extracellular proteins). These proteins are likely to act as public goods because they will often diffuse away from the producing cell, and so any benefit of their function is shared by neighbouring cells. In contrast to the predictions of the horizontal gene transfer hypothesis, we found that: (i) genes coding for extracellular proteins were not overrepresented on plasmids compared with chromosomes, (ii) plasmids with a higher mobility did not carry more genes coding for extracellular proteins [ 26 ]. However, there are potential problems with using genes coding for extracellular proteins as a method for identifying genes for cooperation. Extracellular proteins that act as public goods are among the simplest kind of cooperative behaviour in bacteria, because one gene codes for one protein which is secreted out of the cell to act as a cooperative public good. However, other genes code for cooperative traits in more complex ways, such as by coding for a protein which combines with other proteins and molecules inside the cell before being secreted, or by catalysing a reaction that helps make the cooperative molecule. For example, iron-scavenging siderophores are secondary metabolites of a large gene cassette, with each gene coding for intracellular proteins which work together to produce the secreted siderophore molecules [ 27 ]. Genes coding for siderophores would therefore not be counted as cooperative, despite siderophores being one of the most studied cooperative traits in bacteria [ 28 ]. Consequently, analyses considering only genes for extracellular proteins are likely to miss a number of genes involved in cooperation. Additionally, our previous analysis was based on the genomic data available in 2019, which was only 51 species, and biased towards human pathogens [ 26 ]. Since then, the number of complete prokaryotic genomes in the RefSeq database has more than doubled, meaning there is now potential to examine a much wider and more representative range of bacterial species [ 29 ]. We addressed these problems by conducting a comparative genomics analysis using a new tool for identifying a broad range of genes for cooperation (SOCfinder), not just those coding for extracellular proteins [ 30 ]. SOCfinder comprises three modules which identify genes for extracellular proteins, genes with a cooperative functional annotation and genes which are part of a cooperative secondary metabolite cluster. In addition, we were able to expand the dataset to 146 species, almost three times as many species as in the previous study [ 26 ]. We used phylogeny-based statistical methods to control for non-independence of species, and also for any unevenness in the taxonomic distribution of studied species. We tested whether genes for cooperation were more likely to be carried on plasmids compared with chromosomes, and also examined whether this differed for each of three broad kinds of genes for cooperative traits.",
"discussion": "4 . Discussion We have shown, across 146 bacterial species, that genes for cooperation are not more likely to be on plasmids. Instead, we found evidence that plasmid genes are actually less likely to code for cooperative traits compared with chromosome genes ( figure 1 ). An average bacterial genome carries only 2% of its genes for cooperation on its plasmid(s), with the remaining 98% of its genes for cooperation on the chromosome(s) ( figure 2 ). (a) Where are genes for cooperation located within bacterial genomes? Contrary to the key prediction of the horizontal gene transfer hypothesis, we found the opposite – plasmids had a significantly lower proportion of genes for cooperation than chromosomes. This result was driven by genes found by the two additional modules of SOCfinder: genes with a cooperative functional annotation and genes which were part of a cooperative secondary metabolite cluster were both more likely to be found on chromosomes ( figure 3 ). The genes found by these two additional modules generally code for more complex cooperative traits compared with genes for extracellular proteins. We recently found that plasmid genes have a consistently lower complexity compared with chromosomal genes, where complexity was measured by the number of connections each gene had within the genome's protein–protein interaction network [ 38 ]. This lower complexity on plasmids could explain our result that chromosomes carry proportionally more of these more complex cooperative traits. Carriage on chromosomes could be more likely to be favoured for relatively more complex traits, because carriage on a plasmid could risk the breakup of the gene cassette or be non-functional and/or metabolically disruptive if the plasmid was transferred into a new recipient or over-expressed due to high copy number [ 39 , 40 ]. The carriage of certain genes on plasmids could be favoured for reasons other than horizontal gene transfer [ 26 , 41 ]. One potential benefit is that plasmids usually exist as multiple copies in the cell, meaning plasmid genes will, on average, have a higher expression than genes on the single copy chromosome(s). This could be important for traits like secreted virulence factors and antibiotic resistance mechanisms, where the strength of the phenotype will be directly related to the quantity of the effector molecule produced. For example, bacteria carrying a gene coding for a secreted beta-lactamase on a multi-copy plasmid had a higher level of resistance to beta-lactam antibiotics compared with those carrying the same gene on a chromosome [ 22 ]. These benefits could explain why Dewar et al. found that species which were broad host-range pathogens were most likely to have genes for extracellular proteins overrepresented on their plasmids [ 26 ]. Taken together, even when genes for cooperative traits are carried on plasmids, it could be for a reason other than the plasmid's ability to transfer. Methods to find genes for cooperative traits could disproportionately miss those carried on plasmids, because plasmid genes tend to be less well annotated. However, we have several reasons why we think this is unlikely to be driving our results. First, when we looked at only genes for extracellular proteins, which are identified by the presence of a highly conserved signal peptide sequence and so should be unaffected by gene annotation, we found no difference in plasmid and chromosome proportion of genes for cooperation ( figure 3 a ). Second, most of the focus in the literature has been on the proportion of genes for cooperation on plasmid(s) and chromosome(s), because plasmids usually carry far fewer genes than the chromosome(s) [ 19 , 25 , 26 ]. In our dataset, the average bacterial genome carries only 4.6% of its genes on plasmid(s). However, comparing the proportion to control for this imbalance in number of genes ignores something often overlooked: the imbalance is itself evidence against a major role of horizontal gene transfer via plasmids in the maintenance of cooperation. Third, the previous comparative genomics study across 51 species found that plasmid mobility, defined as the extent to which a plasmid was able to mobilize via conjugation, did not correlate overall with the proportion of a plasmid's genes which coded for extracellular proteins [ 26 ]. We have not considered plasmid mobility in this study, and it offers an alternate method for examining why genes are carried on plasmids. In terms of absolute number, instead of proportion, we found that the vast majority (98%) of genes for cooperative traits are carried on the chromosome(s) ( figure 3 ). If we were missing a few plasmid genes for cooperation due to a lower annotation quality, we would have needed to miss an average of 81 genes for cooperation on plasmids for there to be a higher absolute number of genes for cooperation on plasmids compared with chromosomes. However, for this to be the case, more than 41% (87/208) of all plasmid genes across bacterial genomes would have to code for a cooperative trait, compared with only 2.8% of the genome as a whole. We think this is very unlikely. Instead, what this imbalance in number of genes suggests is that very few genes for cooperation in bacteria ever benefit from plasmid transfer, and yet cooperation is highly prevalent and stable across bacteria. Something else other than horizontal gene transfer must maintain cooperation for the approximately 98% of genes for cooperative traits carried on chromosomes. (b) Why is cooperation not favoured by horizontal gene transfer? Our results support recent theory which suggested that horizontal gene transfer does not appreciably favour or stabilize cooperation [ 26 , 42 ]. Older theory had focused on the invasion of cooperation, and found that this could be facilitated by plasmids [ 10 , 12 , 13 , 19 – 21 ]. However, plasmids can facilitate the invasion of any gene, and not just cooperation. In addition, when the potential for cheating as well as cooperative plasmids was allowed for, it was found that horizontal gene transfer did not appreciably help maintain cooperation [ 26 , 42 ]. Cooperation tends to only be favoured on plasmids in the same conditions where it is favoured on chromosomes. An exception, where cooperation can be preferentially favoured on plasmids, is if the rate of plasmid transfer is high and the rate of plasmid loss is intermediate [ 26 , 42 ]. However, these conditions also lead to a low rate of plasmid carriage, and so plasmids can only have a small effect on the level of cooperation. Put simply, plasmids either evolve to be rare and more cooperative, or common and not more cooperative. Consequently, theory predicts that the overall influence of plasmid transfer on the level of cooperation in bacterial populations will be low or negligible [ 42 ]. More generally, theory and empirical work examining this hypothesis has tended to assume that there are no fundamental differences between plasmids and chromosomes, other than their ability to transfer. However, it is now clear that there are many features of plasmids which could reduce the suitability of directly comparing their gene content to bacterial chromosomes [ 41 ]. First, plasmids can exist in many copies per cell, which could lead to genetic dominance effects and potentially impact dynamics of plasmid-carried cooperative loci within cells [ 43 ]. Second, plasmids carry ‘backbone’ genes which allow them to replicate and in the case of mobilizable and conjugative plasmids, transfer via bacterial conjugation [ 15 ]. The presence of these genes will constrain the maximum proportion of plasmid genes that can code for traits such as cooperation. While the hundreds of essential genes that bacterial chromosomes must carry will also impose a similar constraint on the maximum proportion of chromosomal genes coding for cooperative traits, it is unclear how similar the proportion of these core plasmid and chromosome genes are. Third, plasmids will be under selection to increase their own transmission, which could include reducing costs to their hosts. Consequently, plasmids could be constrained by the number of genes they carry and the metabolic cost associated with those genes. This could lead to conflict between selection for plasmid carriage of a gene for cooperation compared with selection at the plasmid level to reduce gene number [ 39 , 44 ]. (c) What favours cooperation in bacteria? Kin selection provides a simple and widely applicable explanation for cooperation in bacteria, without needing to invoke a special role of horizontal gene transfer [ 9 ]. The clonal growth of bacteria means that individuals are more likely to be near relatives (kin), who would also carry the genes for cooperative traits. Consequently, any benefits of a secreted public good molecule would be shared with relatives who are also producing that molecule [ 45 ]. By contrast, non-producers will be growing with other non-producers, and so will be less able to benefit from (cheat) the public goods produced by other cells. The kin selection hypothesis has been supported by experimental evolution, population genetics and across-species comparative studies. Experiments on a number of bacterial species have shown that the production of public goods is maintained when species are cultured at a high relatedness, but lost when cultured at a low relatedness [ 4 , 28 , 46 – 48 ]. Population genetic studies on Pseudomonas aeruginosa and Bacillus subtilis have shown signatures (footprints) of selection at the genomic level that are expected from kin selection for cooperation [ 49 , 50 ]. Specifically, genes controlling cooperative traits showed higher polymorphism, greater divergence and were more likely to harbour deleterious mutations, relative to genes for non-cooperative private traits. Comparative studies have found that species which form groups where relatedness is likely to be higher show higher levels of cooperation, as measured by the occurrence of altruistic helping cells, frequency of genes for cooperative traits or aid provided to insect hosts [ 51 – 53 ]."
} | 4,528 |
26513664 | PMC4626383 | pmc | 7,948 | {
"abstract": "Understanding the factors determining the spatial and temporal variation of ecological networks is fundamental to the knowledge of their dynamics and functioning. In this study, we evaluate the effect of elevation and time on the structure of plant-flower-visitor networks in a Colombian mountain forest. We examine the level of generalization of plant and animal species and the identity of interactions in 44 bipartite matrices obtained from eight altitudinal levels, from 2200 to 2900 m during eight consecutive months. The contribution of altitude and time to the overall variation in the number of plant ( P ) and pollinator ( A ) species, network size ( M ), number of interactions ( I ), connectance ( C ), and nestedness was evaluated. In general, networks were small, showed high connectance values and non-nested patterns of organization. Variation in P , M , I and C was better accounted by time than elevation, seemingly related to temporal variation in precipitation. Most plant and insect species were specialists and the identity of links showed a high turnover over months and at every 100 m elevation. The partition of the whole system into smaller network units allowed us to detect small-scale patterns of interaction that contrasted with patterns commonly described in cumulative networks. The specialized but erratic pattern of network organization observed in this tropical mountain suggests that high connectance coupled with opportunistic attachment may confer robustness to plant-flower-visitor networks occurring at small spatial and temporal units.",
"conclusion": "Conclusions Our results contribute to understanding the way plant—flower-visitor network structure and function are influenced by elevation and seasonal precipitation in tropical systems. Our sampling design allowed us to detect variation in the majority of metrics that were analyzed. Two general patterns emerged: (1) Although plant and flower-visitors are active throughout the year, plant—flower-visitor systems tend to be larger and more complex (i.e., have more links) at lower altitudes and high precipitation months; and (2) the specialized but erratic pattern of network organization suggest that high connectance coupled with dynamic and opportunistic interactions may represent an alternative pattern of interaction that confers robustness to perturbations [ 60 , 61 ]. Increasing evidence suggests that network structure may arise from multiple factors, including interaction neutrality, trait matching, spatio-temporal distribution of species, and sampling effects [ 48 , 52 , 62 ]. Our study adds to this body of evidence through the observation that highly dynamic networks occur in mountain ecosystems where turnover of interactions within individual networks is mainly driven by the replacement of flowering plant species in space and time, which are highly influenced by precipitation. The extent to which opportunistic and erratic species attachment provides structure and influences the persistence of plant—pollinator systems needs to be addressed in future studies. Especially useful will be studies that estimate extinction probabilities of plant and pollinator species in networks with high vs. low partner fidelity.",
"introduction": "Introduction Understanding the factors that determine the structure and dynamics of pollination networks is one of the unsolved questions in network ecology. Description of ecological networks as bipartite sets of interacting species allows the quantification of diverse descriptors of community complexity, which can reveal important properties of ecosystem structure and function [ 1 , 2 , 3 , 4 ]. Examination of tropical, temperate and arctic systems has shown consistent broad patterns of network structure [ 1 , 5 ]. However, the high environmental heterogeneity and sensitivity of species to small-scale variation in climatic conditions [ 6 , 7 ] often makes it difficult to predict the fine-scale patterns of occurrence and persistence of plant-animal interactions based on broad descriptors. In this regard, recent studies in plant—pollinator networks have examined the extent to which network structure varies across time and space. Results have revealed mixed patterns. For example, while some studies have shown slight variation in connectance estimates [ 8 , 9 , 10 , 11 , 12 ] and high and consistent nestedness values through time [ 8 , 9 ], other studies carried out at detailed temporal scales have found high and very variable connectances[ 13 , 14 ]. Complex temporal dynamics concerning the number and identity of species involved in pollination relationships has also been observed [ 8 , 9 , 13 , 14 ], and related to their phenology and abundance among other factors [ 10 , 15 , 16 ]. Furthermore, this complexity can occur even within a season, which suggests that to understand the effects of environmental variation on the structure of mutualistic networks, studies with appropriate temporal and spatial community resolution are needed [ 17 ]. To date, almost all ecological network studies are snapshots of limited temporal extent (i.e., a few years) that lack any finer temporal resolution [ 18 ]. At the same time, the influence of space, elevation, or precipitation gradients on network structure is by far less known (but see [ 17 , 19 , 20 ]), and studies analyzing simultaneously the temporal and spatial variation of network structure at small scale are almost nonexistent in the literature (but see [ 21 ]). Studies assessing the temporal dynamics of plant—pollinator networks have been conducted mostly on temperate and arctic regions [ 8 , 9 , 10 , 11 , 13 , 14 ]. In these ecosystems, flowering plants have a well-defined flowering season [ 11 ] that contrasts with the lack of seasonality shown by tropical ecosystems. Even though plants and pollinators can in principle be active all year long, most species inhabiting tropical ecosystems have very specific habitat requirements and asynchronous phenologies that prevent their simultaneous occurrence in space or time(i.e., the ‘forbidden links’) [ 22 ]. In such cases, spatial and temporal data aggregation from cumulative samples (i.e., cumulative networks) may not necessarily represent adequately the real set of interactions and more realistic samplings should rely on records that capture species with truly coincident phenologies [ 13 , 14 , 17 , 23 ]. Consequently, the lack of data at finer temporal and spatial resolutions hinders the possibility to assess the role of the ecological factors underlying network structure and its variation in space and time [ 18 ]. In this work, we used a community-level approach to examine the influence of space and time on a tropical plant-flower-visitor network within a non-disturbed and continuous Andean mountain forest in Colombia. This mountain forest is characterized by having pronounced slopes and a bimodal precipitation regime; both factors that condition the observed set of interactions among species [ 6 , 7 , 24 ]. Additionally, a conspicuous change in the thermal band from mountain to alti-mountain at 2700 masl has been described for tropical Andean forests, with subsequent changes in the composition and structure of tree vegetation and microclimate, specifically in mean and minimum annual temperature [ 25 ]. Even though plants and animals can be active throughout the year in mountain forests, a strong altitudinal and temporal segregation in the distribution of species has been frequently observed, leading to a marked pattern of species turnover in space and time [ 25 , 26 , 27 , 28 , 29 ], including those involved in plant—pollinator networks (e.g., [ 9 , 14 ]). Thus, in general, we can expect that variation in environmental conditions associated with altitude and time create a highly heterogeneous mosaic of interactions due to changes in the composition of the interacting guilds. Specifically, in this study we asked the following questions: (1) Does network topology, measured as numbers of species, numbers of links, connectance, and nestedness vary through time or with elevation?; (2) Does network topology differ above and below 2700 masl where the thermal band changes?; (3) How consistent is the level of ecological specialization across altitude and time?; (4) Does the composition of plant and flower-visitor assemblages and therefore the identity of species interactions vary through time or along the altitude gradient?; and (5) Does network topology differ between individual vs. cumulative networks? To answer these questions, we analyzed a series of network metrics on eight plant- flower-visitor communities encompassing an elevation gradient of 800 m, during eight consecutive months in a Columbian mountain forest. We examined the pattern of change of network metrics across space and time, and assessed whether elevation or time is the main source of variation for network metrics in the overall system.",
"discussion": "Discussion Network structure Our analysis of plant—flower-visitor networks along eight altitudinal levels over a period of eight months revealed ample variation in network topology, in the identity of interacting species, and in the composition of plant and flower-visitor assemblages. These results are in agreement with recent studies indicating high turnover of species and interactions among and within seasons and along environmental gradients [ 8 , 9 , 11 , 45 ]. In spite of variation, however, some clear patterns emerged for this this mountain forest. First, individual networks are small and the proportion of realized links relative to all potential interactions is high. A characteristic network size in this system consists of four plant and eight insect species. These small networks lie in the low side of network size distribution for other interaction systems in the world. Only 10% of the reviewed networks are of comparable size to the mean size of our networks [ 1 , 5 ]. Second, most networks exhibited a non-nested pattern of organization. The high interconnected pattern of interactions was not related with high nestedness. Nestedness has been described to occur consistently in large networks ( S > 50) [ 5 ]. Although our matrices were smaller than 50 cells, the metric used for nestedness estimation (NODF) is largely insensitive to matrix size, even for those with 25–100 cells [ 33 ]. At the same time, nestedness is a robust meausure of network structure which is only slightly influenced by insufficient sampling [ 46 , 47 ]. As matrices in our system have a mean size of 35 cells, our estimates provide reliable nestedness values. Moreover, NODF consistently rejects a nested pattern for different types of random matrices, which is congruent with the mixed random-checkerboard distribution of links in the analyzed matrices [ 33 , 48 ]. However, it is interesting to note that networks become relatively more structured (i.e., nested) as their complexity (number of links) increases, which is consistent with the observed trend in pollination networks showing wide variation in size [ 5 ]. The observed pattern of network structure differed from those produced by random networks indicating that the system-specific patterns here described are not produced by random processes. Effects of altitude and time on network structure Variation in network topology was explained in a different way by altitude and time. In the analyzed webs, the number of plant species, network size, number of interactions, and connectance were best explained by time and precipitation. The last three variables have shown to be highly correlated in several pollination networks around the world [ 1 ] and also in our system (Pearson r \n M-I = 0.93, r \n M-C = -0.74, r \n C-I = -0.63), and in consequence, it is not surprising they change in a similar way along time. The effect of time is likely associated to the expansion and contraction of networks, probably reflecting the influence of plant phenology on the overall system: an increase in the number of flowering plant species during the transition from rainy to dry season, results in large-sized networks with more interactions, decreasing C values. Conversely, during the other months, network size tends to be smaller with a consequent increase in connectance values. These findings may have important consequences for the functioning of this and other systems. For instance, similar pulses in network size and connectance have been previously described during summer and spring months in a temperate forest [ 14 ]. Unexpectedly, the altitude gradient did not influence network metrics as described in other studies [ 1 , 17 , 19 ]. However, network structure and composition changed conspicuously at 2700 m. Networks from altitudes higher than 2700 m were smaller, due to a reduction in the number of insect species. As a consequence, the number of interactions declined at the network and species level above such altitude. It is likely that the well-defined altitudinal divide observed at 2700 masl in this study, associates with changes in the composition and structure of tree vegetation and microclimate described for mountain forests of the Andes [ 25 ]. The decline in flower-visitor diversity at high altitude is consistent with findings in other high mountain ecosystems [ 49 , 50 , 51 ]. Flowering was more prevalent during rainy months, which translated into a higher variation in the number of flowering plants through time than across elevation. Flower-visitor activity and abundance, in turn, were influenced by both altitude and precipitation, indicating that insect dynamics depends strongly on small-scale abiotic conditions [ 6 ]. Composition, interactions and specialization in plant—flower-visitor assemblages The whole system consists of a high proportion of unique combinations of interacting species with limited periods of activity and a small number of redundant interactions across months or altitudes. Similarly, analyses of small network units revealed that most species within networks interacted with only one species. Insects had a maximum of four interactions. Only three species, Cyclanthura spp., Frankliniella spp. and Nitulidae, visited four flowering plant species within one month. Plants, in turn, were visited by a larger number of insect-visitor species, i.e., assymetrical networks. One plant species, Sphaeradenia spp., was visited by more than 15 insect species during one month, followed by Anthurium panduriforme and Xanthosoma undipes that were visited by nine and seven insect species, respectively. A high proportion of insect and plant species were replaced every month and at every 100 m elevation. In consequence, networks were very different not only in the composition of species but in detailed aspects related to the identity of species involved in interactions, suggesting that turnover of interactions ( β -diversity of interactions) is mainly driven by the replacement of species in space and time [ 52 ]. Distance decay patterns (i.e., the increased species dissimilarity with geographical distance) have been observed in mutualistic networks [ 52 ]. In contrast, a high and consistent turnover of species and interactions over the whole system occurred in our study system without reference to changes in elevation or precipitation [ 44 ]. This pattern may result from the presence of a tiny group of permanent species (i.e., core species) and a high proportion of sporadic species. Only few species were present in all elevations and months. This result contrasts with the large core groups observed in other mutualistic pollination systems [ 8 , 10 , 53 ]. But what factors underly the enormous variation in the composition of species in our system? It is likely that variation in the proportion of different taxa across altitude and months accounts in part for such variability. In our system, Coleoptera and Diptera were involved in most interactions. The dominance of Coleoptera shifted along the altitudinal gradient and with time; in general, there was a replacement of Coleoptera by Thysanoptera at increasing altitude and during rainy months. This result contrasts with other mountain ecosystems in temperate regions where Diptera and Hymenoptera are the most important orders (e.g., [ 17 , 49 , 51 , 54 ]) and their shift is often associated with a humidity gradient, with Diptera being dominant at more humid sites and Hymenoptera at dryer sites. Since our study site has a mean annual precipitation of 2000 mm and a relative humidity near 90%, it is likely that abiotic conditions restrict the abundance and diversity of Hymenoptera in this tropical mountain forest [ 55 ]. In general, insect species seem to interact in a somewhat erratic pattern with flowering plants. Most insects are minute, often flying short distances and probably taking advantage of local and immediate floral availability in their surroundings. This behavior can keep insects attached with the network even when plant species change across altitude or time [ 56 ]. Unlike nested networks, where specialists interact with generalists and the generalist core interacts with generalist species [ 5 ], specialized interactions dominated the pattern in this forest mountain system when examined at small scale. This pattern of labile plant-animal interactions may represent an alternative mechanism of interaction in species-poor and highly dynamic networks (i.e., short periods of activity or flowering for the majority of insects and plants). It is likely that species ensure growth and survival under the commonest environmental conditions (i.e., available floral resources), regardless of how many species converge functionally (e.g., exploiting same floral resource) as suggested by Hubbell [ 57 ]. However, this does not imply that networks are randomly structured. For instance, in our system, the duration of plant flowering and insect activity followed an exponential and logarithmic frequency distribution, respectively, but contrary to expectations based on random interactions among coexisting species [ 16 ], nestedness values were low, suggesting that a random model of interactions was not appropriate to represent our system. In our study system, plant and animal species seem to interact freely, replacing each other according to enviromental conditions, without any apparent species-specific constraint [ 58 ]. Even though we do not have information on the effectiveness of flower-visitors at present, our results suggest that a functional equivalence between very different insects may drive the pollination network dynamics in this system (see also [ 9 ]). Topology of individual vs. cumulative networks In this system, cumulative networks were medium-sized, showed low connectance, and lack of nestedness, revealing a relatively unconventional pattern of interaction [ 1 , 5 ]. However, when the whole system was divided into smaller spatial and temporal units, networks showed higher and variable connectances, similar to studies that partitioned networks along time [ 13 , 14 ]. Monthly cumulative networks (i.e., time networks) showed higher variation for number of species, matrix size and number of interactions than spatial cumulative networks (i.e., altitude networks), which is consistent with the trends just described for individual matrices. Also, as expected, connectance decreased in cumulative matrices as they are bigger in size. This reduction has been also observed in plant—pollinator networks from temperate and artic regions. In general, patterns of interaction from pollination networks from around the world exhibit low connectance values in temperate (mean = 9.6, n = 20), artic (mean = 11.5, n = 5), and tropical grassland/shrub vegetation and low elevation forest communities (mean = 14, n = 4) [ 1 ], compared to the estimated connectance values for the analyzed networks in this study. In contrast, estimated nestedness for most networks is higher than those observed in our system and showed similar values in temperate (mean = 0.86, n = 13), artic (mean = 0.86, n = 5), and tropical grassland/shrub vegetation and low elevation forest networks (mean = 0.83, n = 6) [ 5 ]. Regarding biotic specialization it has been reported that specialization is lower at tropical than temperate latitudes and that it decreases with increasing local and regional plant diversity: specialization is a response of pollinators to low plant diversity [ 59 ]. In our system, plant richness was higher at low elevation [ 44 ] and the percentage of specialist insect species visiting flowers was low. The opposite trend was observed at high altitude suggesting that the relationship between plant diversity and specialization may be consistent for mutualistic networks along environmental gradients (i.e, latitude and altitude). However, network properties as connectance and nestedness may exhibit wide variation. In consequence, the organization of network cohesiveness around a central core of interactions may be a contingent property of specific ecosystems rather than an invariant property of pollinator networks."
} | 5,273 |
33811231 | PMC8018972 | pmc | 7,952 | {
"abstract": "Algal biofuel research aims to make a renewable, carbon–neutral biofuel by using oil-producing microalgae. The freshwater microalga Botryococcus braunii has received much attention due to its ability to accumulate large amounts of petroleum-like hydrocarbons but suffers from slow growth. We performed a large-scale screening of fast-growing strains with 180 strains isolated from 22 ponds located in a wide geographic range from the tropics to cool-temperate. A fast-growing strain, Showa, which recorded the highest productivities of algal hydrocarbons to date, was used as a benchmark. The initial screening was performed by monitoring optical densities in glass tubes and identified 9 wild strains with faster or equivalent growth rates to Showa. The biomass-based assessments showed that biomass and hydrocarbon productivities of these strains were 12–37% and 11–88% higher than that of Showa, respectively. One strain, OIT-678 established a new record of the fastest growth rate in the race B strains with a doubling time of 1.2 days. The OIT-678 had 36% higher biomass productivity, 34% higher hydrocarbon productivity, and 20% higher biomass density than Showa at the same cultivation conditions, suggesting the potential of the new strain to break the record for the highest productivities of hydrocarbons.",
"conclusion": "Conclusions This study performed a large-scale screening of the natural genetic resource of Botryococcus braunii on an unprecedented scale with 180 strains isolated from tropical to temperate climates and identified 9 fast-growing strains that have growth rates faster or similar to Showa , a standard fast-growing strain. Their biomass productivities were 12–37% higher than that of Showa. One strain, OIT-678, established a new record doubling time (1.2 days) as the fastest race B strain. Further studies are important to test whether the newly-isolated fast-growing strains outperform the highest productivities for hydrocarbons recorded by the formerly fastest strain Showa.",
"introduction": "Introduction Algal biofuel research aims to make a renewable, carbon–neutral biofuel by using oil-producing microalgae. It initially prospered at 1980th, triggered by oil crisis, and has been reactivated at 2000th by global warming issues 1 – 3 . At the onset of algal biofuel research in the 1980s, there were large projects to screen natural bioresources, such as the Aquatic Species Program in the USA 4 , and lists of useful oil-producing microalgae were prepared. Subsequently, improved cultivation and harvesting techniques 5 , 6 , and strain selection and genetic engineering of these algae 7 – 10 have been pursued. A high-throughput screening method for lipid-rich microalgae from environmental samples by fluorescence-activated cell sorting is now available 11 , 12 . A lower-cost direct method of isolation for lipid-rich microalgae using a fluorescence microscope and manipulator has also been reported 13 . This technical progress provides new opportunities to identify cryptic bioresources in nature. The freshwater, colonial green microalga Botryococcus braunii accumulates large amounts of petroleum-like hydrocarbons in the colony 14 , 15 and has received much attention in algal biofuel research since most oleaginous microalgae accumulate neutral lipids, which are energetically inferior to hydrocarbons 16 , 17 . Furthermore, B. braunii accumulates the hydrocarbon oils in an extracellular matrix of the colony, which enables “milking” of the algal oils without killing the algae 18 , 19 . Despite these remarkable characteristics, B. braunii has not occupied a leading position in algal biofuel research because of its slow growth rate: typical doubling times are between 3 and 7 days 14 , 15 . The typical doubling time of Chlorophyta is 24 h and that of Cyanobacteria is 17 h 20 . The top 20% of Chlorophyta , Cyanobacteria and other taxa with respect to growth rate have doubling times in the range of 7 to 8 h 20 . Among oleaginous microalgae, Chlorella vulgaris , Neochloris oleoabundans and Scenedesmus obliquus have short doubling times of 8–9 h 21 . To overcome the slow growth of B. braunii , we searched for novel fast-growing strains from natural genetic resources. The Showa strain, known as a fast-growing strain of B. braunii , holds the fastest growth record with a doubling time of 1.4 days 22 and the highest record of hydrocarbon productivity (340 mg L −1 d −1 ) to date 23 , 24 . This strain is a wild strain isolated from the pond of a greenhouse in California in the 1980s and was not selected by screening or artificially modified by mutagenesis 25 . There should be wild strains that grow faster and are more productive than the Showa strain, but, to the best of our knowledge, there has been no large-scale study seeking these strains, a potential of natural genetic resource. A few wild strains and collections of B. braunii have been investigated 26 – 30 , but researchers failed to find significantly faster and more productive strains than Showa, probably due to the limited number of strains tested (< 10) and the limited geographic range of locations searched. This study performed a large-scale screening of fast-growing strains from a total of 180 strains isolated from 22 ponds located in a wide geographic range, from the tropics to a cool-temperate climate. Although B. braunii is widely distributed in freshwater and brackish lakes, reservoirs, or ponds from temperate to tropical environments 15 , their natural densities are commonly quite low (10–10 2 colonies per L) 31 , which makes it difficult to find the alga in natural environments. We have developed a simple method for isolating B. braunii from the natural environments 32 and have newly isolated 70 wild strains from natural ponds in tropical Indonesia and temperate Japan 31 . This study has isolated an additional 110 strains for the large-scale screening. The objective of this study is to evaluate the potential of the natural genetic resource of B. braunii on an unprecedented scale to answer the following questions with the following approaches: (i) In what kind of natural environment can we expect to find fast-growing strains? We analyzed whether the variations in growth rate between strains are attributable to the differences in the ponds and climate regions where the strains originate. (ii) Are there any wild strains that grow faster than the Showa? We measured biomass and hydrocarbon productivities of nine fast-growing wild strains selected by the screening and compared them to those of Showa and the highest recorded values reported by previous studies for B. braunii .",
"discussion": "Results and discussion Screening of fast-growing wild strains We have investigated a total of 112 natural ponds, and a total of 180 wild strains were successfully isolated from 22 ponds located in various climate regions (Supplementary Table 1S ). The screening of 180 wild strains was performed based on the increasing rate of optical density (OD 660 ) in a 10 mL glass tube. The daily increases in OD 660 in glass tubes were generally well fitted by an exponential function, and the coefficient of determination ( R 2 ) of the curve fitting was > 0.9 for 70% of the data ( n = 590). After removing the 30% of the data with low R 2 values, we calculated the doubling time ( D t ) from the exponential growth curve in a total of 163 wild strains. Figure 1 shows the histogram of D t for 163 wild strains: there was a large variation in D t with the median at 6.0 days from a minimum of 2.7 days to a maximum of 23.4 days. The Showa strain had a D t of 4.7 days, and 34 wild strains (20%) had a shorter D t than Showa. We thus successfully identified several wild strains that are potentially faster-growing than Showa. Of these 34 wild strains, 22 strains (65%) are originated from the tropics (Indonesia). When the chemical races of these fast-growing wild strains were estimated from a molecular phylogenetic analysis of 18S rRNA, 83% were estimated as B race, which produces triterpene hydrocarbons. Figure 1 Histogram of doubling time for 163 wild Botryococcus braunii strains. Numbers above bars indicate the number of strains. The shaded portion shows wild strains with doubling times shorter than Showa (4.7 days) under the same cultivation conditions. To answer the question (i) In what kind of natural environment we can expect to find fast-growing strains, we analyzed the relative importance of strain, chemical race, the pond of origin, and climate where the pond is located as determinants of D t of a strain using a mixed model analysis. The model analysis showed that 57% of the total variance of D t data was attributable to the variance between strains (σ 2 S ; Table 1 ). The σ 2 S was significantly larger than zero ( P < 0.05), indicating a significant genetic variation in D t between strains. In contrast, the variance component of the pond effect (σ 2 P ) was not significantly different from zero ( P > 0.05). This reflects substantial variations in D t between strains isolated from the same pond. There were no specific ponds where fast-growing strains originated. The least-square means of D t calculated from the model for each pond had relatively larger error bars compared to the differences of the least-square means between ponds (Fig. 2 ). Furthermore, the effects of race and climate on D t were not significant ( P > 0.05; Table 2 ). Table 1 Summary of the fit of a mixed model for doubling-time data of Botryococcus braunii wild strains. Random Effect a Variance ratio Variance component SE 95% confidence interval b % of total variance Strain [Race] 2.62 4.89 0.95 3.03–6.75 57.0 Pond [Climate] 0.97 1.82 1.59 − 1.29 to 4.93 21.2 Residual 1.87 0.16 1.59–2.23 21.8 Total 8.58 1.54 6.21–12.6 100 Model R 2 = 0.79; n = 380; AIC = 1570; BIC = 1609. REML Variance Component Estimates. a Strain was nested in Race, and Pond was nested in Climate. b The lower–upper 95% confidence limits for the variance component. Intervals including zero indicate non-significant variance components ( P > 0.05). Table 2 Summary of the fit of a mixed model for doubling-time data of Botryococcus braunii wild strains. Fixed effect No. parameter b DF (DF Den ) c F Ratio P value Race a 3 3 (39.9) 0.96 0.42 Climate 3 3 (11.6) 1.50 0.27 Model R 2 = 0.79; n = 380; AIC = 1570; BIC = 1609. Fixed Effects Tests. a The Race (A, B, L, S) was estimated based on the molecular phylogeny of 18S rRNA sequences (See Hirano et al. 31 ). b Number of parameters associated with the effect. c Degrees of freedom associated with the effect (Denominator degrees of freedom in parenthesis). Figure 2 Least square means of the doubling time ( D t ) of wild strains of Botryococcus braunii in relation to chemical races and location of origin. Least square means were computed from a mixed model with race and climate as fixed effects and strain and pond as random effects. Bars indicate 95% confidence intervals. There were no significant differences ( P > 0.05) between ( a ) Pond, ( b ) Race, or ( c ) Climate. See Supplementary Tables 1S for details of the abbreviation and locations of the ponds. Our analyses did not provide any clear answers to the question (i). Fast-growing strains were found in both tropical and temperate climates, and in every pond, there were substantial variations in intrinsic growth rates between strains (Fig. 2 ). This implies that natural selection does not operate on the growth characteristics measured under laboratory conditions. Because the culture conditions of the screening are far from the conditions in natural environments, the growth rate measured in the laboratory might be a ‘hidden characteristic’ in nature and is unlikely to be exposed to the process of natural selection. Frequent gene flow may also damp the power of natural selection. Hirano et al. reported no genetic differentiation between tropical and temperate strains 31 , probably due to dispersal by birds and wind across a large geographic scale. Biomass and hydrocarbon productivities of fast-growing wild strains We selected 9 fast-growing strains based on the screening results and determined their biomass and hydrocarbon productivities in two separate tests. Test 1 cultured 4 wild strains plus the Showa strain for 30 days, and Test 2 cultured the other 5 strains plus the Showa strain for 40 days. Figure 3 shows the changes in algal biomass density in the fed-batch cultures. Biomass densities increased over the first two weeks, reached a plateau, and then gradually decreased in some culture bottles. The specific biomass growth rate ( μ max ) was the highest, and D t was the shortest during the first 4 days of cultivation, and then the μ max decreased with increasing biomass density. In contrast, biomass productivity ( P x ) increased with an increase in algal density despite the decrease in the growth rate and reached the maximum at 2–7 days before the peak of algal density. Figure 3 Biomass growth curves of fast-growing wild strains of Botrycoccus braunii in a fed-batch culture system. Every 2–3 days, 100 mL (20%) of culture was sampled to estimate algal biomass density, and the same amount of new medium was added. ( a )–( e ): Test 1 for 4 wild strains with a standard strain, Showa. ( f )–( k ): Test 2 for five wild strains with Showa. There were 3 repetitions (■, ▲, ●) of culture bottles for each strain. The arrow indicates the period of minimum doubling time, the horizontal bar indicates the period of maximum biomass productivity. Figure 4 shows the minimum D t and the maximum P x of 9 wild strains compared to the Showa strain. In each culture bottle, the two minimum D t and the two maximum P x values were used to calculate the averages for each strain. In Test 1, the D t of Showa was 1.97 days, and all four wild strains had shorter D t than Showa. The D t of OIT-678 (1.48 days) was significantly shorter than that of Showa ( P < 0.01). In Test 2, another five wild strains had D t (2.22–2.65 days) similar to that of Showa (2.23 days), and there were no significant differences ( P > 0.05). These results demonstrate that the first screening by OD measurements in glass tubes successfully identified fast-growing strains with similar or faster biomass growth rates than the Showa strain. The P x was also significantly higher in fast-growing wild strains than that of Showa ( P < 0.05, Fig. 4 ). Three wild strains (OIT-678, OIT-805b, OIT-685) had an approximately 36% higher P x than Showa ( P < 0.05). Thus, the fast-growing strains also had high biomass productivities. Figure 4 Doubling times and biomass productivity of 9 wild strains and Showa of Botryococcus braunii . ( a ) The bar indicates average with standard error (n = 6). Numerical values are shown in the bar. The values significantly different from that of Showa are indicated by asterisks: *, P < 0.05; **, P < 0.01 (Dunnett’s post hoc test). ( b ) Microscopic view of a wild strain, OIT-678 and the Showa strain. The colony was flattened by cover glass, and hydrocarbons (HCs) exuded from extracellular matrices (EMs). Hydrocarbon content, productivity, and constituent were analyzed for 6 productive wild strains (Table 3 ). The hydrocarbon content of the Showa strain was 26%, and the wild strains had similar or somewhat higher hydrocarbon contents (26–38%). Hydrocarbon productivities of the wild strains ranged from 40 to 58 mg L −1 d −1 , which is 1.1–1.9 times higher than that of Showa. The major constituent hydrocarbons were botryococcenes, C 30 H 50 for OIT-678, C 33 H 56 for OIT-805b, C 34 H 58 for OIT-685, and OIT-775, respectively. Therefore, all these strains were identified as B race. Table 3 Hydrocarbon content, productivity, and major constituent of wild fast-growing Botryococcus braunii strains. Test Strain Hydrocarbon content T HC (%) Biomass productivity P x (mg L −1 d −1 ) Hydrocarbon productivity P HC (mg L −1 d −1 ) Major constituent hydrocarbon Mean a (SE) Ratio Mean b (SE) Ratio a × b Ratio 1 Showa 26.6 (1.05) 1 132 (6.6) 1 35 1 NA OIT-678 26.0 (0.06) 1.0 182 (8.7) 1.4 47 1.3 C 30 H 50 OIT-805b 29.4 (1.55) 1.1 182 (13.3) 1.4 53 1.5 C 33 H 56 OIT-686 35.3 (1.45) 1.3 163 (11.3) 1.2 58 1.6 NA OIT-808 26.2 (5.11) 1.0 148 (2.3) 1.1 39 1.1 NA 2 Showa 26.4 (0.73) 1 100 (6.7) 1 26 1 NA OIT-685 29.2 (3.03) 1.1 136 (7.4) 1.4 40 1.5 C 34 H 58 OIT-775 38.9 (4.98) 1.5 125 (4.6) 1.3 49 1.9 C 34 H 58 Results of two independent experiments, each of which uses Showa strain as a standard, are shown. Ratios of mean values of wild strains to that of the standard strain Showa are also indicated. SE = Standard Error ( n = 6), NA = Not Analyzed. A novel fast-growing strain, OIT-678 The OIT-678 is the only wild strain with a significantly faster growth rate and higher biomass productivity than the Showa strain (Fig. 4 ). Therefore, we assessed the potential of this strain with additional experiments. To assess the maximal growth rate of the strain, we performed a fed-batch cultivation with a high flow rate (40% replacement every two days). Even in this high-flow rate condition, both OIT-678 and Showa increased biomass density for 10 days. The OIT-678 showed a significantly lower doubling time (t-test: ∣t∣ = 2.1, df = 38, P < 0.05) and a higher specific growth rate (∣t∣ = 2.06, df = 38, P < 0.05) than Showa. The D t of OIT-678 and Showa were 1.23 days and 1.37 days, respectively (Fig. 5 a). Although the difference in D t between the two strains appears small, the outcome in biomass production is quite large. When starting cultivation with the same initial amount of biomass, the biomass of OIT-678 will be twice as much as that of Showa after 12 days. The biomass productivity of OIT-678 was 1.4 times larger than that of Showa, as mentioned earlier (Fig. 4 ). Figure 5 Growth rate, biomass density, and colony size of a novel fast-growing strain, Botryococcus braunii OIT-678, compared with the Showa strain. ( a , b ) Fast-growing potential was assessed by a 500 mL fed-batch culture (4 replicates for each strain), in which 40% of the culture volume is replaced by new culture medium every 2 days. Until the 10th day of cultivation, biomass growth rates of two consecutive sampling points were calculated, and their averages were compared between strains. ( c , d ) Potential for a high-density culture was assessed by using a nutrient-rich medium, WFAM. First, 1 g L −1 of algae were inoculated and cultivated in two 500 mL bottles for each strain with AF-6 culture media for 24 days (Phase I). Algal density was monitored by OD. At the end of Phase I, 83 mL of 6 × WFAM were prepared in new 500 mL bottles, and 415 mL of algal culture was transferred to the bottle and cultivated (Phase II). From the 60th to the 75th day, 5% of the algal culture was sampled four times (red arrows), and the same amount of 6 × WFAM was added. This gradually increased the concentration of the media up to 2 × WFAM (Phase III). Cultivation was continued until the 87th day of cultivation (Phase IV). ( e ) Colony size measured on the 75th day (star). The box plot indicates the range of 10–90% of the data with bars and the range of 25–75% of the data with a box. Wilcoxon signed-rank test results are shown by letters in the boxes, and significant differences between the samples are indicated by different letters ( P < 0.001). ( f , g ) Appearance of typical colonies: microscopic view (left) and macroscopic view (right). We also assessed the ability for long-term, high-density cultivation of the strain by using a nutrient-rich media, WFAM (Supplementary Table 2S ). WFAM contains higher amounts of nitrate (841 mg L −1 NO 3 − ) and phosphate (94 mg L −1 PO 4 3− ) than AF-6 (309 mg L −1 NO 3 − and 9.7 mg L −1 PO 4 3− ). We started the cultivation by using AF-6 as the culture media and used two 500 mL bottles per strain in batch culture mode (Phase I, Fig. 5 c,d). The initial biomass density was adjusted to 1 g L −1 . After the algal density plateaued on the 24th day of cultivation, the culture media was replaced with WFAM. After replacement, algal density increased again for both strains (Phase II, Fig. 5 c,d) and surpassed the plateau of Phase I, suggesting that the increases in algal biomass density were limited by the amounts of nutrients available in Phase I. A second plateau in the algal density appeared between the 50th and 60th days of cultivation. We then sampled 5% of the algal culture four times (red arrows, Phase III in Fig. 5 c,d) and added the same amount of 6 × WFAM. This gradually increased the concentration of the media up to 2 × WFAM until the 75th day of cultivation (Phase III). However, the algal density did not increase for 2 weeks following 75 days of cultivation (Phase IV in Fig. 5 c,d), suggesting that the increases in algal biomass density in this phase were light-limited. The maximal algal biomass density was 6.3 g L −1 for OIT-678 and 5.2 g L −1 for Showa. OIT-678 had a sixfold larger colony size than Showa (Fig. 5 e–g). The increase in colony size induces an aggregated distribution of biomass in the culture and decreases the optical density at a given biomass density 23 . In agreement with this expectation, the ratio of OD to biomass density was smaller in OIT-678 (= 0.38) than in Showa (= 0.52; Fig. 5 c,d), indicating a lower light attenuation in the culture of OIT-678. Thus, the OIT-678 can make a higher-density culture than Showa , which is at least partly due to the ability to reduce light attenuation in the culture by forming larger colonies. Top records of fast growth rate and high productivity We performed literature reviews on the growth rate and biomass productivity of B. braunii to find the fastest growth rates and the highest productivities (Table 4 ). The OIT-678 strain established a new record of D t = 1.23 days in B. braunii race B, where the previous record was 1.4 days of Showa 22 . The D t of OIT-678 falls short of the 1.11 days reported in the strain CCAP807/1, race A 27 , and the 0.52 days of an unknown strain 33 . The strain CCAP807/1 showed the D t of 1.11 days under a continuous light condition ( Php = 24:0, Table 4 ), which should be better than our culture condition ( Php = 14:10). In addition, although the race A and race B strains are classified as the same species, their hydrocarbon structures and biosynthetic processes are largely different 14 , and they showed a high divergence value of 18S rRNA sequences at almost the species level 34 . Therefore, our finding of a novel fast-growing strain in race B is a significant advance. Table 4 Record-holders for the fastest growth rate and highest productivity of biomass and hydrocarbons in Botycococcus braunii . Strain (Race) Culture conditions Growth rate Biomass Hydrocarbon Ref System ℃ PAR Php CO 2 μ max D t X max P x T HC P HC OIT-678 (B) 500-mL flask, semi-continuous 28 100 14:10 2 0.58 1.23 6.2 182 26 47 TS Showa (B) 30-mL tube, semi-continuous 30 850 14:10 1 0.5 1.40 NA NA NA NA 22 Algobank761 (B) 30-mL bubble column, batch 25 100 24:0 4 0.48 1.44 6.4 281 2.0 6 27 Showa (B) 0.32-m 2 trickle-film, continuous 25 282 16:8 5 NA NA 20 1500 23 340 23 AC761 (B) 400-mL bubble column, batch 23 150 18:6 2 0.11 6.3 NA 150 45 68 26 CCAP807/1 (A) 30-mL bubble column, batch 25 100 24:0 4 0.624 1.11 8 408 4 15 27 UTEX2441 (A) 2-L tubular column, continuous 25 71 NA NA 0.378 1.83 5.0 223 37 82 61 UTEX572 (A) 3.2-L flat PBR, batch 24 300 12:12 NA NA NA 7.8 1300 9 110 35 CCAP807/1(A) 0.8-L Air-lift cylinder, batch 25 120 a 14:10 1 NA NA 2.2 599 44 265 62 LB572 (A) 400-mL bubble column, batch 22 50 24:0 2 NA NA 4.6 296 65 b 190 b 63 NA 28.57-L flat panel, batch 27 800 NA 1 1.344 0.52 1.8 358 22 b 80 b 33 TRG (NA) 1.5-L stirred tank, continuous 25 49.5 16:8 NA 0.366 1.89 5.2 368 32 119 64 ℃, temperature; PAR , photosynthetic active radiation (μmols of photons m −2 s −1 ); Php , photoperiod (light:dark hours); CO 2 , % v/v; μ max , specific cell growth rate (d −1 ); D t , doubling time (d); X max , maximum biomass concentration (g L −1 ); P x , biomass productivity (mg L −1 d −1 ); T HC , total hydrocarbons (% dry weight); P HC , hydrocarbon productivity (mg L −1 d −1 ); NA , no information available; TS , this study. a Estimated from the value of 26 W m −2 of a cold-white fluorescent tube. b Values of total lipids instead of hydrocarbons. Khichi et al. reported D t = 0.52 days in a flat panel photobioreactor with an unknown strain of B. braunii 33 . This D t record is extraordinarily fast compared to previous studies. However, they estimated the D t based on indirect measurements of algal biomass using the OD. The OD of an algal culture can change greatly with the growth of contaminant bacterium as well as changes in the colony size of B. braunii 23 . Further studies, including race identification and direct measurements of biomass growth and productivity, are needed to confirm the record as the fastest B. braunii strain. In terms of biomass and hydrocarbon productivity, our wild strains ( P x = 182 mg L −1 d −1 and P HC = 58 mg L −1 d −1 ; Fig. 4 and Table 3 ) are not record-breaking. This is because productivity largely changed due to cultivation methods, and our method may not be optimal. Khatri et al. reported an exceptionally large productivity values ( P x = 1500 mg L −1 d −1 , P HC = 340 mg L −1 d −1 ) by an ultra-high density cultivation ( X max = 20 g L −1 ) of the Showa strain (Table 4 ) 23 , which are some of the highest productivities for algal oil ever reported. In strains belonging to race A, Song et al. also reported a record value of P x = 1300 mg L −1 d −1 with a high-density culture (7.8 g L −1 ) 35 . Thus, the high-density culturing appears to be a promising method to improve the volumetric productivities of biomass and hydrocarbons of B. braunii . We determined the biomass and hydrocarbon productivities of our wild strains in relatively low densities (0.5–2.0 g L −1 ; Fig. 3 ). Therefore, there is a great potential for improvement of the productivity of our strains by adopting the high-density cultivation methods. For high-density cultivations, strains with large-sized colonies might be suitable due to the increased permeability of light. Since light tends to be the most limiting resource for algal growth in high-density cultures, strains with efficient light capture characteristics should be useful. The increase in average colony size in culture is expected to increase the average amount of light received by the surface of a colony because of the reduction of self-shading among colonies 23 . Although the increase in colony size should decrease the amounts of light transmitted into the inner parts of a colony, cells placed in the inner parts of a colony are old and may have a limited physiological capacity to utilize strong light 36 . The OIT-678 strain formed a higher-density culture with a larger-sized colony than Showa (Fig. 5 ) and is therefore expected to have the potential to achieve higher productivities than Showa in dense-culturing methods. Comparisons to other oleaginous microalgae The growth rates of the Showa strain and OIT-678 are still much slower than those of other fast-growing microalgae. The oleaginous microalgae Chlorella vulgaris , Neochloris oleoabundans , and Scenedesmus obliquus have short doubling times of 8–9 h 21 . Because of its colony-forming habit, B. braunii invests resources to construct and maintain the extracellular matrix of the colony, and the self-shading among cells within the colony appears to be unavoidable. The synthesis of energetically expensive hydrocarbons may also restrict the potential for fast growth 14 . Despite of the slow growth, the biomass and oil productivities of B. braunii are comparable to other fast-growing microalgae. The oil productivities of 30 microalgal species, which were assessed in 250 mL flasks under continuous illumination and bubbled with CO 2 -enriched air, showed that the averages (maximums) of the biomass productivity, the lipid content, and the lipid productivity are 190 mg L −1 d −1 (370 mg L −1 d −1 ), 23% (40%), and 40 mg L −1 d −1 (61 mg L −1 d −1 ), respectively 37 . Our novel wild strains of B. braunii had comparable productivities (Table 3 ). Furthermore, a high-density and continuous cultivation of the Showa strain 23 achieved a hydrocarbon productivity of 340 mg L −1 d −1 , and OIT-678 has the potential to exceed that productivity. This record is one of the highest productivities of algal oil to date. According to a comprehensive review 38 published in 2011, the highest lipid productivities of microalgae under phototrophic conditions were 142 mg L −1 d −1 by Nannochloropsis oculate 39 and 133 mg L −1 d −1 by Neochloris oleoabundans 40 . Subsequently, Ho et al. reported a high lipid productivity of 140 mg L −1 d −1 for Scenedesmus obliquus 41 . Further high lipid productivities were then reported for Chlorella vulgaris (1425 mg L −1 d −1 ) 42 and C. protothecoides (590 mg L −1 d −1 ) 43 . Thus, the Chlorella species may hold the current record for highest records of lipid productivity of microalgae under phototrophic cultivation. However, as these records of the Chlorella species are instantaneous values measured in batch cultivations, the average productivities under continuous cultivations should be lower than this value. Most oleaginous microalgae, including Chlorella and Scenedesmus species, require nutrient depletion to initiate lipid accumulation 41 – 43 . Because the nutrient depletion restricts further biomass growth in the culture, an initial period of cultivation for biomass accumulation is required prior to lipid production. Consequently, even if high lipid productivity is noted during the lipid accumulation phase, the average productivity over the total period, including the period for biomass growth, should decrease 44 . In contrast, B. braunii accumulates hydrocarbons mainly during the exponential and early linear growth stages 45 – 47 , 62 (i.e., this alga can produce hydrocarbons and grow biomass simultaneously). In addition, the hydrocarbons produced by B. braunii have superior fuel properties to the lipids (triacylglycerol) produced by the oleaginous microalgae such as the Chlorella and Scenedesmus species in terms of high thermal values and compatibility with the existing petroleum infrastructure. Therefore, B. braunii can be regarded as one of the most promising species for the production of algal biofuel, and our novel fast-growing strains are expected to increase the feasibility of biofuel production. The promises and challenges of creating a biorefinery from Botryococcus braunii Improvements to cultivation methods such as mixotrophic 48 and attached cultivation 49 , as well as the high-density cultivation 23 , have great potential to increase the biomass and hydrocarbon productivity of B. braunii. Furthermore, biological research is being developed on the genetic transformation 50 , genome sequencing 51 , and bacterial symbionts 52 , 53 of B. braunii . Although many biotechnological and engineering advances have been made in the course of biomass production and biomass processing techniques to yield biofuels from B. braunii , the high production costs are still one of the major issues for the commercialization of algal biofuel production 54 . In order to reduce the production costs, other potential applications such as wastewater treatment, CO 2 mitigation, and the manufacture of high-value products should be coupled with biofuel production 2 . Wastewater treatment by B. braunii removes nutrients 55 and heavy metals 56 and may also be effective for pharmaceutical products remediation as recently reported in other microalgae 57 . Substantial amounts of high-value chemicals aside from hydrocarbons are also identified in B. braunii 24 , 58 . These potential applications can be combined with hydrocarbon production in a biorefinery from B. braunii . When the biorefinery is scaled up to industrial levels in the future, the environmental sustainability concerns of the system cannot be ignored. In particular, due to the irreversibility of energy systems, exergy-based measures should be incorporated into traditional life cycle assessments to analyze the sustainability of the biorefinery from thermodynamic, economic, and environmental perspectives 59 . Conclusions This study performed a large-scale screening of the natural genetic resource of Botryococcus braunii on an unprecedented scale with 180 strains isolated from tropical to temperate climates and identified 9 fast-growing strains that have growth rates faster or similar to Showa , a standard fast-growing strain. Their biomass productivities were 12–37% higher than that of Showa. One strain, OIT-678, established a new record doubling time (1.2 days) as the fastest race B strain. Further studies are important to test whether the newly-isolated fast-growing strains outperform the highest productivities for hydrocarbons recorded by the formerly fastest strain Showa."
} | 8,316 |
27422734 | null | s2 | 7,953 | {
"abstract": "Modern microbialites are often used as analogs of Precambrian stromatolites; therefore, studying the metabolic interplay within their associated microbial communities can help formulating hypotheses on their formation and long-term preservation within the fossil record. We performed a comparative metagenomic analysis of microbialite samples collected at two sites and along a depth gradient in Lake Alchichica (Mexico). The community structure inferred from single-copy gene family identification and long-contig (>10 kb) assignation, consistently with previous rRNA gene surveys, showed a wide prokaryotic diversity dominated by Alphaproteobacteria, Gammaproteobacteria, Cyanobacteria, and Bacteroidetes, while eukaryotes were largely dominated by green algae or diatoms. Functional analyses based on RefSeq, COG and SEED assignations revealed the importance of housekeeping functions, with an overrepresentation of genes involved in carbohydrate metabolism, as compared with other metabolic capacities. The search for genes diagnostic of specific metabolic functions revealed the important involvement of Alphaproteobacteria in anoxygenic photosynthesis and sulfide oxidation, and Cyanobacteria in oxygenic photosynthesis and nitrogen fixation. Surprisingly, sulfate reduction appeared negligible. Comparative analyses suggested functional similarities among various microbial mat and microbialite metagenomes as compared with soil or oceans, but showed differences in microbial processes among microbialite types linked to local environmental conditions."
} | 389 |
34180595 | PMC8123916 | pmc | 7,955 | {
"abstract": "Abstract Aquatic ecosystems are often stratified, with cyanobacteria in oxic layers and phototrophic sulfur bacteria in anoxic zones. Changes in stratification caused by the global environmental change are an ongoing concern. Increasing understanding of how such aerobic and anaerobic microbial communities, and associated abiotic conditions, respond to multifarious environmental changes is an important endeavor in microbial ecology. Insights can come from observational and experimental studies of naturally occurring stratified aquatic ecosystems, theoretical models of ecological processes, and experimental studies of replicated microbial communities in the laboratory. Here, we demonstrate a laboratory‐based approach with small, replicated, and liquid‐dominated Winogradsky columns, with distinct oxic/anoxic strata in a highly replicable manner. Our objective was to apply simultaneous global change scenarios (temperature, nutrient addition) on this micro‐ecosystem to report how the microbial communities (full‐length 16S rRNA gene seq.) and the abiotic conditions (O 2 , H 2 S, TOC) of the oxic/anoxic layer responded to these environmental changes. The composition of the strongly stratified microbial communities was greatly affected by temperature and by the interaction of temperature and nutrient addition, demonstrating the need of investigating global change treatments simultaneously. Especially phototrophic sulfur bacteria dominated the water column at higher temperatures and may indicate the presence of alternative stable states. We show that the establishment of such a micro‐ecosystem has the potential to test global change scenarios in stratified eutrophic limnic systems.",
"conclusion": "5 CONCLUSION This study reveals that two environmental change factors (i.e., temperature and nutrient availability) caused large and non‐additive variation in the composition of aquatic microbial communities and the abiotic conditions of the ecosystem. The advantages of this system include high parallelization and replication, easy and non‐destructive sampling, the versatility of testable conditions and manipulations, and are of additional interest besides in situ ecosystem studies. We believe that these insights are only the tip of the iceberg of what can be learned from such a model micro‐ecosystems with strong and even stratified spatial environmental gradients. Even in the described study, we could have included many more elements, such as characterization of the conditions at the sampling site, measurements of organic compounds composition in the water column, and control without cellulose. Further research with this new experimental system could take many paths, including studying the stability of the communities to press and pulse perturbations, and how this stability may depend on aspects of community composition, such as functional composition and intraspecific diversity; the extent and significance of evolutionary processes such as mutation and selection for mediating effects of environmental change; and observation of community composition via metagenomic methods, to capture not just the bacterial component of the micro‐ecosystems, but also to research the functional significance of other likely inhabitants, such as viruses. Finally, one could research why these micro‐ecosystems did not become entirely oxic or entirely anoxic, and why there was little evidence of the discontinuous responses to environmental change that are predicted for systems with strong positive feedbacks, such as this one.",
"introduction": "1 INTRODUCTION Micro‐organisms are key players in nearly all ecosystems, ranging from the human gut to marine and freshwater habitats. The functioning of microbial communities is critical for many ecosystem services since micro‐organisms are the driving force behind biogeochemical cycles (Falkowski et al., 2008 ; Kertesz, 2000 ; Stein & Klotz, 2016 ; Weiss et al., 2003 ). Nevertheless, microbes are highly dependent on environmental conditions; even slight changes can lead to taxonomic and functional community shifts (Allison et al., 2013 ; Evans & Wallenstein, 2014 ). Consequently, microbial communities and their functions are significantly affected by global change (Cavicchioli et al., 2019 ), which has recently led to the emergence of the research field termed “Global Change Microbiology” (Boetius, 2019 ). Stratification of lakes is a common seasonal phenomenon of the temperate zone, which is dependent on the depth and the surface area of the lake (Gorham & Boyce, 1989 ). Stratified lakes show warm and oxygen‐rich upper water layers, dominated by cyanobacteria and algae, and colder (anoxic) deeper layers, that harbor heterotrophic biomass‐degraders and sulfate‐reducing bacteria (Diao et al., 2017 , 2018 ; Guggenheim et al., 2020 ; Vigneron et al., 2021 ). In between these layers, the metalimnion can be found, which is formed by decreasing light‐intensities and oxygen concentrations (microaerophilic), and an increasing pool of reduced sulfur compounds, creating ideal niches for phototrophic sulfur bacteria and chemolithoautotrophic micro‐organisms (Bush et al., 2017 ; Diao et al., 2017 , 2018 ; Jorgensen et al., 1979 ; Morrison et al., 2017 ; Nyirabuhoro et al., 2020 ; Savvichev et al., 2018 ; Vavourakis et al., 2019 ; Vigneron et al., 2021 ; Wörner & Pester, 2019 ; Wu et al., 2019 ). Besides chemical and physical parameters that influence the presence of oxic and anoxic layers, it was recently postulated that mutual inhibition between cyanobacteria and anaerobic sulfur‐dependent bacteria creates and maintains the distinct oxic and anoxic zones of such ecosystems (Bush et al., 2017 ). The mathematical model used in that study predicts abrupt transitions between aerobic and anaerobic microorganisms and the occurrence of oxic–anoxic regime shifts. Moreover, studies have demonstrated that global change consequences could have strong impacts on this sensitive ecosystem, including dominant growth of harmful cyanobacteria due to warming (Posch et al., 2012 ), or increasing primary production and creation of anoxic layers due to increased nutrient input (Luek et al., 2017 ). The mentioned studies show that important insights come from analyses of observations of naturally occurring aquatic ecosystems (Vigneron et al., 2021 ), and theoretical models of the plausible and relevant ecological and biochemical processes. In addition, in the field of global change biology, targeted experiments that employ a limited number of standardized synthetic model ecosystems have been described as an “entirely different approach” (Hutchins et al., 2019 ) and are considered vital alongside studies of naturally occurring ecosystems (Lahti et al., 2014 ) and hybrids of both approaches (De Vos et al., 2017 ). Key challenges in global change microbiology that are amenable to research with synthetic ecosystems include the role of evolutionary processes, the role of historical contingency (Widder et al., 2016 ) and interactions among micro‐organisms (Overmann & van Gemerden, 2000 ), and the integration of data and theory (Widder et al., 2016 ). Experiments about the role of environmental, organismal interactions, and biochemical feedbacks in stratified aquatic micro‐ecosystems appear absent, however. Establishing a suitable such standardized model ecosystem for analyzing potential responses of a broad range of micro‐organisms remains crucial. Being able to monitor global change responses of diverse functional groups of potentially interacting microbes is a desirable feature of such an experimental system, which points toward systems with strong abiotic gradients in space and or time. In this work, we get inspiration from an “old” but highly valuable approach, the Winogradsky column (Dworkin, 2012 ; Pagaling et al., 2017 ; Zavarzin, 2006 ). We constructed a modified and smaller version that allows a highly replicable development of a broad range of complex and dynamic microbial groups in one experimental unit. In contrast to classical Winogradsky columns, our micro‐ecosystems are mostly liquid, with only a small sediment layer (~ 6% v/v of the column). Thus, we created a highly replicable liquid oxic–anoxic interphase and self‐developing model systems. By applying this approach, we were able to analyze microbial responses to multifarious environmental change (temperature and nutrient addition manipulated factorially), including the responses of community composition and abiotic environmental conditions. As well as showing how microbial community composition, dissolved oxygen, pH, and hydrogen sulfide respond to temperature and nutrient addition, we highlight the potential for global change microbiology research of this new approach coupled with state‐of‐the‐art sequencing technologies. We hypothesize that higher temperature will increase the dominance of anaerobic microbes due to an increase in the extent of the anoxic zone, due to the lower solubility of oxygen in warmer water. We hypothesize that nutrient addition (addition of ammonium phosphate, highly used fertilizer (Cao et al., 2018 )) will alter community composition, but we cannot a priori say how, as this would depend on features of the organisms, such as competitive and facilitative interactions among them, which we do not have sufficient information about. We did not have any a priori expectation of whether the combined effect of temperature and nutrient addition would be additive (no interaction), more than additive (positive interaction), or less than additive (negative interaction). We anticipated the potential for micro‐ecosystems to become entirely oxic or entirely anoxic, and for changes in composition to be non‐linear and to potentially include discontinuous responses that are predicted by theory.",
"discussion": "4 DISCUSSION Understanding and predicting the consequences of multifarious environmental change is very important in a broad range of research disciplines, and investigations should be made from global‐ to micro‐scales (Balser et al., 2006 ). In contrast to existing work on higher‐developed organisms (González‐Varo et al., 2013 ; Pennekamp et al., 2018 ; Tylianakis et al., 2008 ), less is known about how bacterial ecosystems react to multifarious global change scenarios (Cavicchioli et al., 2019 ; Coyle et al., 2017 ; Hutchins & Fu, 2017 ; Rillig et al., 2019 ). Besides, there is a need for experimental studies (in addition to natural systems and mathematical models) of complex microbial model ecosystems that can potentially display diverse responses to changing environmental conditions for various microbial groups in one experimental unit. With inspiration from Winogradsky columns (Dworkin & Gutnick, 2012 ; Pagaling et al., 2017 ), we introduce a modified experimental micro‐ecosystem, that includes the development of aerobic and anaerobic microbial communities in one experimental ecosystem in a replicable manner. In this study, we demonstrate that such an approach represents an appropriate model system for analyzing responses of complex and stratified microbial communities to global change scenarios. Some of our results, for example, the stratification and temperature effects, may be considered unsurprising and “textbook knowledge.” This is good evidence that these small and liquid‐dominated microbial ecosystems are appropriate analogs of naturally occurring communities. Furthermore, we used this model system to shed light on the consequences of multifarious environmental change, namely simultaneous warming and nutrient deposition—a relatively novel area of ecology, evolution, and microbiology. We hypothesized a simultaneous effect on oxygen dynamics and microbial community compositions. While temperature is considered to be the main driver for shaping soil/sediment communities (Cole et al., 2013 ; Deng et al., 2018 ; Garcia‐Pichel et al., 2013 ), we furthermore observed non‐additive effects of warming and nutrient addition on the aquatic microbial community (Figure 4c , Table 1 ). This highlights the importance of studying potential interactions of multiple environmental change parameters simultaneously (Niiranen et al., 2013 ), given that they can differ in effect compared to the sum of individual disturbances. Environmental change cannot be reduced to the sum of individual factors. It is multifarious, providing the opportunity for “risk multiplication” and for greater challenges when attempting to understand and predict ecosystem responses to environmental change (Schulte to Bühne et al., 2020 ). Besides the effect on overall microbial community composition, our results additionally indicated that multiple drivers affect the oxic and anoxic layers differently. The oxic layer of the system was affected by temperature and nutrient addition additively (Figure 5a,c , Table 1 ). In contrast, the anoxic layers were affected by non‐additive effects of temperature and nutrient addition (Figure 5b,c ). These observations might be due to the different biogeochemistry processes in the two strata, such as the interplay of specific microbial communities, and their interplay with available nutrients, trace elements, as well as abiotic factors, and the presence of specific oxygen‐reducing sulfur compounds (Bush et al., 2017 ; Chapra & Canale, 1991 ; Garcia et al., 2013 ; Jorgensen et al., 1979 ; Luther et al., 2003 ; Yu et al., 2014 ). The microbial community composition and the oxygen dynamics are tightly coupled in a feedback loop. Key players in this feedback loop are oxygen‐producing cyanobacteria, sulfate‐reducing bacteria, and phototrophic sulfur bacteria (Bush et al., 2017 ; Lee et al., 2014 ). While cyanobacteria and especially green and purple sulfur bacteria had high relative abundance in the micro‐ecosystems, the relative abundance of sulfate‐reducing micro‐organisms was comparably low. This observation could be explained by the sampling schedule: detection of high H 2 S concentration within the micro‐ecosystems indicates a nutrient‐rich habitat for phototrophic sulfur bacteria (Guerrero et al., 1985 ; Hamilton et al., 2014 ), which suggests the previous high activity of sulfate‐reducing micro‐organisms (Pimenov et al., 2014 ; Sass et al., 1997 ). Earlier or later sampling may have revealed more oxidized sulfur compounds and a higher abundance of sulfate‐reducing micro‐organisms. Further explanation of this unbalanced ratio between these two functional groups could probably be found in the light‐dark‐ and the sediment‐water ratio of our model system. With increasing temperature, micro‐organisms changed from aerobic representatives ( Giesbergeria , Uliginosibacterium ) to anaerobic ones ( Chlorobaculum , Chlorobium ), which is also confirmed by the oxygen measurements. The higher turnover rate of organic substrates of the microbes in the sediment and higher H 2 S production supports the dominant growth of phototrophic sulfur bacteria at moderate and higher temperatures (Findlay & Kamyshny, 2017 ; Nedwell et al., 1994 ; Velthuis et al., 2018 ). The dominant blooms of Chlorobium with increasing temperature is an important finding and its ecological consequences should be investigated for future global change scenarios. Some results hint at the presence of alternate stable states (Beisner & Cuddington, 2010 ) in the phototrophic sulfur bacterial communities. There was considerable variation among some of the replicates of the same treatment combination, despite them having very similar initial conditions. For example, some contained either the green phototrophs Chlorobium or Chlorobaculum , or the purple phototroph Allochromatium (based on sequencing and macro/microscopical observation of the respective columns). Besides these alternative compositions within the same functional group, also the oxygen dynamics indicated interesting patterns, particularly at 28°C, where two of three replicates of the nutrient addition‐treated micro‐ecosystems became oxic at the bottom sensor. Since no significant differences in the microbial community can be observed in these incubations, vertical movements of the oxic–anoxic interphase may have occurred, which was also visible by the different heights of the white microbial community, consisting of Rhodospirillaceae representatives. Additionally, the sinking of cyanobacteria into the anoxic layer could also result in temporary changes in oxygen dynamics in our dynamic system. An interesting aspect could be observed regarding the pH gradient of the water column, which showed a slightly alkaline pH on the water surface in the controls, probably due to photosynthesis reactions of cyanobacteria/algae and HCO 3 \n − in the unbuffered system, and a slightly acidic pH at the bottom, probably due to organic matter degradation. This gradient was removed by the addition of NH 4 H 2 PO 4 , which decreases the pH due to its acidic character and buffering activity on the surface to pH ~6."
} | 4,251 |
32810759 | PMC8021483 | pmc | 7,956 | {
"abstract": "When engineering microbes to overproduce a target molecule, engineers face multiple layers of trade-offs to allocate limited cellular resources between the target pathway and native cellular systems. These trade-offs arise from limited free ribosomes during translation, competition for metabolic precursors, as well as the negative relationship between production and growth rate. To achieve high production performance, microbes need to spontaneously make decisions in the dynamic and heterogeneous fermentation environment. In this review, we discuss recent advances in microbial control strategies that are used to manage these trade-offs and to improve microbial production. This review focuses on design principles and compares different implementations, with the hope to provide guidelines to future microbial engineering.",
"conclusion": "Conclusions Considerable progress on various control strategies has been made to improve microbial production in recent years. Different control strategies were developed to solve different problems in bioproduction. Feedback control is a versatile tool to provide robust production of burdensome proteins and metabolites, avoid the accumulation of toxic intermediates, and shorten the rise-time of metabolites. Metabolic switch can replace the traditional inducible system used in two-stage batch fermentation to avoid resource competition between cell growth and engineered metabolic pathways. Population quality control can effectively prevent low-producing cells from dominating cell cultures, thus improving overall yields and titers, especially during large-scale production. Understanding the strengths and weaknesses of each control strategy can help obtain the most effective control for optimal performance enhancements. Different types of control strategies can also be potentially combined in one engineered strain to fulfill distinct functions for bioproduction. As demonstrated in a recent study on naringenin production, a metabolic feedback control was used to prevent overflow from malonyl-CoA to lipid biosynthesis, while a sensor-selector was used to couple naringenin production with cell growth [ 71•• ]. Combination of these two control strategies has increased both naringenin titer and strain stability. With the development of more sophisticated control technologies, the era of intelligent manufacturing in biology is coming.",
"introduction": "Introduction Microbial fermentation has provided an environment-friendly and versatile platform for manufacturing various bio-based products. Typical fermentation products include alcohols, organic acids, amino acids, vitamins, commodity chemicals, antibiotics, antibodies, and industrial enzymes [ 1 – 4 ]. Recent advances in metabolic engineering and synthetic biology have added an increasing number of products to this list, including fragrances, pharmaceuticals, nutraceuticals, and advanced materials [ 5 – 10 ]. The ability to biologically produce diverse products could have profound societal impacts on multiple industries, only if cost-effective bioproduction can be achieved. This challenge demands the development of effective strategies to improve the production performance (i.e. titers, yields, productivities, and robustness) of engineered microbes. When engineering microbes to overproduce a specific molecule at high productivity and yield, one needs to consider multiple layers of trade-offs rather than simply overproduce pathway molecules to the highest level. The first layer stems from ribosomal cost of translating target proteins ( Figure 1a ). Sequestered ribosomes by mRNAs of target proteins reduce a cell’s ribosomal budget to make native proteins for biomass generation and energy synthesis [ 11 – 14 ]. Furthermore, the allocation of limited translational power between multiple modules within a target pathway affects the overall catalytic efficiency of the pathway [ 15 , 16 ]. The second layer is metabolic trade-offs that involve both carbon cost and energy cost ( Figure 1b ). Conversion of precursor metabolites (e.g. acetyl-CoA) to target products can lead to insufficient material and/or energy supply for the synthesis of cellular structures [ 17 ]. Protein synthesis and enzymatic reactions from engineered pathways also consume energy molecules (i.e. ATP and NAD(P) H), which can be otherwise used to support cell growth. Besides, molecules in a target pathway can be toxic to cells, particularly when they are accumulated to high concentrations. Thus a balanced allocation of metabolites and energy molecules is required for optimizing microbial production [ 18 ]. The third layer of trade-offs comes between growth rate and product yield. High producers usually have a slower growth rate than low producers. The difference in single-cell growth rate caused by mutations or molecular noise allows low producers to accumulate, lowering overall yields ( Figure 1c ). Over the past few years, many genetically encoded control strategies have been developed to improve microbial production by managing these trade-offs [ 19 – 23 ]. These control strategies help engineered cells to adjust their metabolism to combat dynamic and heterogeneous environments in large fermenters as well as stochastic cellular processes. In this review, we discuss recent research advances in control strategies with the focus on design principles. Here, we divide these control strategies into three categories based on their mode of operation: feedback control, two-stage metabolic switch, and population quality control."
} | 1,379 |
29552656 | PMC5851919 | pmc | 7,959 | {
"abstract": "Microbial secondary metabolites represent a rich source of valuable compounds with a variety of applications in medicine or agriculture. Effective exploitation of this wealth of chemicals requires the functional expression of the respective biosynthetic genes in amenable heterologous hosts. We have previously established the TREX system which facilitates the transfer, integration and expression of biosynthetic gene clusters in various bacterial hosts. Here, we describe the yTREX system, a new tool adapted for one-step yeast recombinational cloning of gene clusters. We show that with yTREX, Pseudomonas putida secondary metabolite production strains can rapidly be constructed by random targeting of chromosomal promoters by Tn5 transposition. Feasibility of this approach was corroborated by prodigiosin production after yTREX cloning, transfer and expression of the respective biosynthesis genes from Serratia marcescens . Furthermore, the applicability of the system for effective pathway rerouting by gene cluster adaptation was demonstrated using the violacein biosynthesis gene cluster from Chromobacterium violaceum , producing pathway metabolites violacein, deoxyviolacein, prodeoxyviolacein, and deoxychromoviridans. Clones producing both prodigiosin and violaceins could be readily identified among clones obtained after random chromosomal integration by their strong color-phenotype. Finally, the addition of a promoter-less reporter gene enabled facile detection also of phenazine-producing clones after transfer of the respective phenazine-1-carboxylic acid biosynthesis genes from Pseudomonas aeruginosa . All compounds accumulated to substantial titers in the mg range. We thus corroborate here the suitability of P. putida for the biosynthesis of diverse natural products, and demonstrate that the yTREX system effectively enables the rapid generation of secondary metabolite producing bacteria by activation of heterologous gene clusters, applicable for natural compound discovery and combinatorial biosynthesis.",
"conclusion": "6 Conclusions We show here that the yTREX system combines vector assembly via homologous recombination in yeast, enabling fast and easy one-step gene cluster cloning, advantageous e.g. in pathway engineering or reporter integration, with the features of the TREX system facilitating transfer, integration and expression. As demonstrated in this study, effective gene cluster expression can be achieved in P. putida simply by random genomic integration of unidirectional gene clusters, a strategy allowing a highly straightforward workflow and fast assessment of the host's suitability for the production of a certain compound. In summary, the presented synthetic biology tool effectively enables the rapid generation of secondary metabolite producing bacteria by activation of heterologous gene clusters, applicable for natural compound discovery and combinatorial biosynthesis.",
"introduction": "1 Introduction Microorganisms exhibit an immense biosynthetic capability for the production of valuable compounds offering versatile bioactivities, applicable in sectors like human medicine or agriculture [1] . A vast multitude of gene sequences has become available, in which more and more gene clusters are identified that encode secondary metabolite biosynthetic pathways [2] . One key technology enabling effective exploration of the encoded chemical wealth is the functional expression in amenable heterologous hosts [3] . Therefore, increasing efforts are put in the development of diverse genetic systems for accessing natural compounds by heterologous expression of biosynthetic genes and gene clusters [4] . Here, the critical determinants for successful heterologous compound production currently represent (i) the efficient gene cluster cloning and (ii) the functional expression of all pathway genes requiring an appropriate host strain which offers a genetic codon usage compatible with the genes to be expressed, can provide metabolic precursors and is tolerant against putative toxicity of heterologous biosynthetic products [5] . Regarding cloning, restriction-independent methods have proven to be a key enabling technology in natural product research [6] . Phage enzyme-dependent recombination in E. coli and in vitro homology-based methods have been developed and successfully applied for gene cluster cloning and engineering [7] , [8] , [9] , [10] , and recently, increasing use of yeast-based recombination cloning highlights the value of such approaches [6] , [11] . Regarding heterologous expression, the number of sophisticated tools refined for the use in different hosts increases likewise. Here, especially P. putida KT2440 represents one promising host for heterologous secondary metabolite biosynthesis [12] , [13] . Valuable tools include diverse vector and promoter systems enabling calibrated gene expression [14] , [15] . Furthermore, we have previously established the pathway tr ansfer and ex pression (TREX) system which allows the straight-forward generation of stable expression strains in different species, employing random chromosomal integration of the heterologous gene cluster in the host by transposition and bidirectional expression of all biosynthetic genes by T7 RNA polymerase [16] . Moreover, we recently applied the tool for random integration of a unidirectional gene cluster into the chromosome of P. putida which resulted in strains exhibiting effective heterologous expression via a chromosomal promoter [17] . Nonetheless, the lack of suitable advanced cloning and expression systems for gene clusters was identified as one drawback hampering the broad utilization of this bacterium [5] . Thus, novel easy to apply tools for the fast activation of heterologous pathways in the host are needed. Here, we describe the yTREX system, a new tool which like TREX enables the tr ansfer, chromosomal integration and ex pression of gene clusters, but is enhanced by the key feature of fast one-step y east recombinational cloning. As an application example, we moreover present the rapid generation of P. putida secondary metabolite production strains based on yTREX-mediated random chromosomal integration of biosynthetic genes. Employing the biosynthetic gene clusters of prodigiosin from Serratia marcescens , of violacein from Chromobacterium violaceum , and of phenazines from Pseudomonas aeruginosa , we demonstrate the system's applicability not only for i) the rapid transfer of metabolic pathways to the host, but also for ii) straightforward pathway engineering via targeted gene cluster re-design, and iii) the implementation of reporter systems for indication of biosynthetic gene expression.",
"discussion": "4 Discussion We demonstrate here a straightforward approach for the rapid generation of bacterial secondary metabolite producers, using the yTREX system to facilitate yeast recombinational cloning and gene cluster transposition into the chromosome of P. putida KT2440 as heterologous host. This strategy yielded production strains for six different bioactive secondary metabolites of the prodiginine, violacein and phenazine family, achieved by transfer and engineering of three gene cluster-encoded biosynthetic pathways from different original hosts. The yTREX system is uniquely easy in application. It is based on only one vector that is used for gene cluster cloning and transfer of biosynthetic genes. The employment of yeast recombinational cloning not only enabled one-step reconstitution of whole gene clusters from several PCR fragments but also allowed the introduction of targeted manipulations with ease, as here demonstrated by adaption of the vio gene cluster and thus re-routing of the respective violacein pathway. Therefore, the DNA assembly method is a highly useful means for gene cluster cloning and rearrangement. At the same time, the technique is well-described [34] and easily established in molecular biology laboratories. Recent application examples include yeast recombinational cloning of the 36 kb grecocycline biosynthetic gene cluster for functional expression in Streptomyces albus \n [49] , or cloning and refactoring of silent gene clusters encoding biosynthesis of lazarimides A, B and C (22.5 kb), and taromycin A (67 kb) for activation in S. albus and Streptomyces coelicolor , respectively [50] , [51] . Assembly of an artificial bacterial chromosome in yeast documents the potential of this method [52] , and its increasingly widespread use, including the protocols presented in this study, shall catalyze advances in natural products research. We demonstrate here the implementation of yeast recombinational cloning for the assembly of plasmids carrying native or engineered gene clusters of interest, together with the yTREX cassettes. The genetic elements of the yTREX cassettes enable conjugational plasmid transfer and random chromosomal integration via Tn5 transposition. Remarkably, random transposition as a key feature of the procedure not only leads to the effective generation of strains stably carrying the gene cluster of interest in their genome, but also to the regular emergence of individual P. putida clones exhibiting immediate expression of biosynthetic genes by chromosomal promoters. The principle of exploiting randomly targeted chromosomal locations in P. putida with promoters suitable for gene cluster expression was previously established with the prodigiosin gene cluster [17] and is in the present study verified as broadly applicable for diverse gene clusters. Note that success of this approach can only be expected if the gene cluster consists of unidirectionally organized genes. Consequently, complex gene clusters with multiple transcription units arranged in different orientations need to be re-constructed into unidirectional organization in the cloning step to follow this strategy. Using the here presented tool, the respective changes in cluster architecture can be introduced easily via one-step yeast recombinational cloning. Effective metabolite production after introduction of biosynthetic genes at a random position in the bacterial host chromosome requires straightforward identification of expressing clones. Here, in the case of red prodiginines and purple or green violaceins, clones expressing inserted genes were readily identified by their color. Moreover, as demonstrated by addition of promoter-less lacZ at the 3′-end of the phenazine gene cluster, a reporter for gene cluster transcription can be included, enabling facile identification of clones exhibiting gene cluster expression from a chromosomal promoter. The implementation of other reporter genes, e.g. if an additional enzyme activity is undesirable, is likewise feasible due to the versatile adaptability of the yTREX tool. The here presented strategies allow for rapid generation and identification of recombinant P. putida expression strains in under two weeks ( Table S4 ). Especially for novel biosynthetic pathways, the approach of random chromosomal integration of biosynthesis genes together with a transcription reporter should enable fast pathway investigation, and additionally provide insights if the host and novel pathway are compatible for metabolite production. P. putida represents an especially promising host for heterologous secondary metabolite production [12] , [13] , [53] . In agreement with previous reports, we corroborate the bacterium's versatile applicability for the biosynthesis of diverse natural products. Maximal titers obtained here for prodigiosin (150 mg/L) and PCA (424 mg/L) are in similar ranges and higher as in previous studies reporting on heterologous metabolite production in the host (94 mg/L prodigiosin in P. putida KT2440 [17] , 27 mg/L PCA in P. putida KT2440 [48] ). Maximal violacein titers found here (105 mg/L) are tenfold higher than those obtained previously by vio gene expression in P. putida KT2440 (10 mg/L [54] ), whereas those of deoxyviolacein (21 mg/L) are two orders of magnitude below previously reported possible outcomes (1.5 g/L deoxyviolacein in P. putida mt-2 [55] ). In addition, we describe here for the first time the biosynthesis of prodeoxyviolacein together with deoxychromoviridans in P. putida KT2440. In all cases, very different product levels were found in the individual strains tested, presumably resulting from individual transposon integration sites with differentially active promoters upstream. Notably, here described product accumulation represents the result from chromosomal gene cluster integration and standard cultivation without any further optimization approaches regarding the strains or production processes. With other host systems, such optimization has been undertaken successfully, producing metabolite titers at gram scale using e.g. native prodigiosin producer Serratia marcescens \n [56] , and Corynebacterium glutamicum \n [57] or E. coli \n [58] as heterologous hosts for production of violaceins, as well as titers of 660 mg/L PCA using native producer Pseudomonas chlororaphis \n [59] . Further studies are necessary to elucidate the potential of P. putida -based metabolite production in comparison to these platforms. Our findings show that the yTREX tool contributes to overcoming the lack of advanced expression strategies for biosynthetic gene clusters which is currently accepted as a major bottleneck limiting the versatile applicability of P. putida \n [5] . This tool, applied with the here presented strategy of exploiting chromosomal promoters for gene cluster expression may thus in the future serve to identify further compounds for which this bacterium represents an ideal production host. In principle, utilization of conjugational transfer and Tn5 transposition should allow the application of here presented yTREX strategies likewise with other host organisms."
} | 3,478 |
34841056 | PMC8614103 | pmc | 7,961 | {
"abstract": "Lignin has long been\nrecognized as a potential feedstock for aromatic\nmolecules; however, most lignin depolymerization methods create\na complex mixture of products. The present study describes an alkaline\naerobic oxidation method that converts lignin extracted from poplar\ninto a collection of oxygenated aromatics, including valuable commercial\ncompounds such as vanillin and p -hydroxybenzoic acid.\nCentrifugal partition chromatography (CPC) is shown to be an\neffective method to isolate the individual compounds from the complex\nproduct mixture. The liquid–liquid extraction method proceeds\nin two stages. The crude depolymerization mixture is first subjected\nto ascending-mode extraction with the Arizona solvent system L (pentane/ethyl\nacetate/methanol/water 2:3:2:3), enabling isolation of vanillin, syringic\nacid, and oligomers. The remaining components, syringaldehyde, vanillic\nacid, and p -hydroxybenzoic acid ( p HBA), were resolved by using ascending-mode extraction with solvent\nmixture comprising dichloromethane/methanol/water (10:6:4) separation.\nThese results showcase CPC as an effective technology that could provide\nscalable access to valuable chemicals from lignin and other biomass-derived\nfeedstocks.",
"conclusion": "Conclusion The\nresults described herein demonstrate the use of a two-stage\nCPC strategy for isolation of enriched or purified aromatic monomers\nfrom complex product mixtures obtained following oxidative alkaline\ndepolymerization of lignin. Access to isolated oligomers from the\nsame process presents opportunities to use these materials in other\napplications, such as biological funneling or the preparation of polyurethanes.\nThe similarity between the product mixtures here to those of other\nlignin depolymerization methods (e.g., using different plant sources,\nbiomass fractionation methods, and/or depolymerization conditions)\nsuggest that these results have broad implications. Moreover, precedents\nfor large-scale application of liquid–liquid chromatography 52 − 54 suggest this approach warrants serious attention among contemporary\nlignin valorization efforts.",
"introduction": "Introduction Lignocellulosic\nbiomass represents a valuable resource that is\ncapable of supplementing or substituting petroleum-based chemical\nfeedstocks and raw materials. 1 , 2 Many biomass valorization\nefforts prioritize the recovery and conversion of (hemi)cellulose\nsugar streams, 3 while extracting little\nvalue from lignin. 4 , 5 The economic viability of biorefineries,\nhowever, will greatly benefit from utilization of all three major\ncomponents: cellulose, hemicellulose, and lignin ( Figure 1 a). Lignin is a biopolymer\nderived from radical polymerization of monolignols and offers significant\npromise as a source of aromatic chemical feedstocks and products. 6 − 12 Conventional methods for biomass processing lead to modification\nor degradation of the lignin, making it notoriously recalcitrant. 3 , 13 Extensive efforts in recent years have led to new biomass fractionation\nmethods that improve separation and recovery of the different components,\nand processes that retain native sugar and lignin structures typically\nlead to improved conversion yields. 10 , 14 − 18 Figure 1 Lignin\nis a substantial component of lignocellulosic biomass and\npotentially an abundant renewable resource of aromatics pending the\ndevelopment of economical valorization methods. (a) Typical ranges\nof lignin, hemicellulose, and cellulose composition in trees. Representative\nchemical structures of each biopolymer are shown. (b) Oxidative depolymerizations\nof lignin yield oligomers and bifunctional monoaromatics, some of\nwhich are commercially relevant such as vanillin and p HBA. The present study builds on the\n“Cu-AHP” method for\nfractionation of poplar-derived biomass. 19 , 20 This method employs a homogeneous Cu catalyst in combination with\nalkaline hydrogen peroxide to separate lignin from sugars. The cellulose\nderived in this manner affords high yields of glucose following enzymatic\nhydrolysis, 21 and the various studies indicate\nthat the lignin stream is also of high quality. 22 − 24 Depolymerization\nof lignin into aromatic monomers is a prominent\ntarget among lignin valorization efforts, and numerous chemical and\ncatalytic methods have been developed to achieve this goal. The product\ncompositions and yields depend on the depolymerization method employed,\nin addition to the biomass plant source and fractionation method. 6 − 12 Oxidative methods are particularly appealing, as they yield valuable\nbifunctional aromatic products 25 ( Figure 1 b) and offer advantages\nfor biological funneling. 26 Lignin oxidative\nalkaline depolymerization (LOAD) methods have been the focus of extensive\nstudy and application, 27 including commercial\nuse for the production of vanillin. 28 , 29 Whereas softwoods\ngenerate vanillin as the major product under these conditions, hardwood\nlignins (e.g., from poplar) generate multiple aromatic chemicals in\nsignificant quantities, including para -hydroxybenzoic\nacid ( p HBA), vanillin, vanillic acid, syringaldehyde,\nand syringic acid. 30 Isolation of\nthe individual aromatic products from complex mixtures\nfollowing lignin depolymerizations is a major, largely unaddressed\nchallenge. The separation and purification of multiple monomers from\ncomplex product streams is challenging, 31 and most previous efforts have prioritized isolation of a single\nvaluable product, such as vanillin. 32 − 34 Methods targeting isolation\nof multiple products will inevitably require numerous unit operations. 35 An ideal isolation method would feature few\nunit operations, require no chemical additives, and be compatible\nwith large-scale application. Liquid–liquid chromatographic\nmethods have the potential to meet these criteria 36 and offer advantages over the use of membranes or solid\nstationary phases, which are costly and susceptible to fouling. 37 − 39 Liquid–liquid chromatography of lignin depolymerization\nmixtures has been used as an analytical tool to gain insight into\ndepolymerization methods (e.g., following pyrolysis), 40 , 41 demonstrated on model sample mixtures, 42 and analyzed computationally. 43 Here,\nwe present the first preparative scale application of liquid–liquid\nchromatography to separate aromatic monomers following lignin depolymerization.\nA two-stage centrifugal partition chromatography (CPC) strategy is\nemployed to isolate monomers from alkaline oxidative depolymerization\nof Cu-AHP lignin. The methodology and results outlined herein establish\nan important foundation for future efforts to isolate valuable aromatic\nproducts derived from lignin.",
"discussion": "Results and Discussion Oxidative Alkaline Depolymerization\nof Cu-AHP Lignin Lignin used to conduct depolymerization\nexperiments was obtained\nfrom debarked NE-19 poplar wood chips subjected to previously reported\nCu-AHP conditions. 19 Solid Cu-AHP lignin\nsamples were dissolved in aqueous NaOH solutions and subjected to\nLOAD conditions to promote depolymerization into aromatic products. 30 Optimization efforts were conducted as parallel\nbatch experiments, using a 1 L Parr pressure reactor equipped with\neight PTFE reaction vessels containing 10 mL of solution with 50 mg\nlignin, variable quantities of a CuSO 4 as a catalyst, and\na stir bar. The Parr vessel was pressurized with 25 bar of air and\nheated to 160 °C over 45 min. Upon reaching the set temperature,\nthe heating mantle was removed and the vessel was cooled with an ice\nbath to quench the reaction. Each reaction mixture was then acidified\nwith HCl (conc.), the aqueous mixture was extracted with EtOAc, and\nthe mixture of organic products was analyzed by HPLC. Products of\nthe reaction include p HBA, vanillin, vanillic acid,\nacetovanillone, syringaldehyde, syringic acid, and acetosyringone\n( Figure 2 a). Figure 2 Aromatic monomers\nobtained from oxidative alkaline depolymerization\nof Cu-AHP lignin (a) and the effects of catalyst concentration and\nreaction alkalinity on the yield (b). Standard conditions: 50 mg lignin,\n10 mL 2 M NaOH, 1.5 mM CuSO 4 , and 25 bar air. Reaction\nwas heated from r.t. to 160 °C over 45 min, and then cooled in\nan ice bath. Representative screening data,\ndepicted in Figure 2 b, highlight the effects of [NaOH] and [CuSO 4 ] and demonstrate\nthat the seven major aromatic products could\ngenerated up to ∼30% total yield. The syringyl (S):guaiacyl\n(G) product ratios closely resemble the composition of the S:G monomer\nratios in the original biomass lignin. 44 Determination of Partition Coefficients in CPC Solvents Measurement of partition coefficients ( K P ) for each solute in biphasic solvent systems provides a means to\nguide solvent selection for CPC separations. Efforts were initiated\nwith the widely used “Arizona” (AZ) solvent system, 45 which features a series of 23 biphasic mixtures\ncomposed of alkane (e.g., pentane), ethyl acetate, methanol, and/or\nwater. The mixtures exhibit systematic variations in polarity, ranging\nfrom 1:1 ethyl acetate/water (system A, most polar) to 1:1 alkane/methanol\n(system Z, least polar), and the middle point consists of equal volumes\nof all four solvents (system N). Ideal solvents for CPC will show\nvalues of log( K P ) between −0.4\nand +0.4 for each of the solutes, in addition to maximizing differences\nin the individual log( K P ) values. 46 When these criteria are met, the solutes will\nequilibrate between the mobile and stationary solvent phases and display\neffective separation with practical retention times. K P values were measured for p HBA, vanillin, vanillic acid, syringaldehyde, and syringic acid with\nAZ solvent mixtures J–N using shake-flask experiments and HPLC\nquantitation ( Figure 3 a; see Figure S3 in the Supporting Information\nfor details). Acetovanillone and acetosyringone were also tested;\nhowever, they show similar K P values to\nvanillin and syringaldehyde, respectively (see Figure S4 in the Supporting Information for details), and\nwe did not attempt resolution of the vanillin/acetovanillone, syringaldehyde/acetosyringone\npairs. Each of the five compounds displayed a log( K P ) value between −0.4 and +0.4 with solvent system\nL, consisting of 2:3:2:3 pentane/ethyl acetate/methanol/water. Vanillin\nand syringic acid show log( K P ) values\nthat are sufficiently different from the others (Δlog( K P ) > 0.2) to facilitate their separation.\nThe\nother compounds, p HBA, vanillic acid, and syringaldehyde,\nhave similar log( K P ) values that complicate\ntheir separation using AZ L as the solvent system. 47 Literature precedent suggested that aqueous/halogenated\nsolvent mixtures could be used to separate these three compounds. 48 Figure 3 Partition coefficient data and CPC traces for separation\nof aromatic\nmonomers obtained from oxidative alkaline depolymerization of Cu-AHP\nlignin. Log( K P ) values for all five aromatic\ncompounds in AZ solvents J–N (pentane:ethyl acetate:methanol:water)\n(a) and for p HBA, vanillic acid, and syringaldehyde\nin halogenated solvent systems (b). Stage one CPC separation of a\nmodel sample mixture of five aromatics using AZ L Asc mode operation\nat 1400 rpm and 30 mL·min –1 flow rate (c),\nand stage two CPC separation of p HBA, vanillic acid,\nand syringaldehyde using DCM/MeOH/H 2 O (10:6:4) Asc operation\nat 1100 rpm and 25 mL·min –1 flow rate (d).\nCPC traces were generated by monitoring CPC elutions at λ =\n200–600 nm. Measurement of K P values in four solvent\nmixtures inspired by this precedent showed a Δlog( K P ) > 0.2 among all the three components in each case\n( Figure 3 b). A 10:6:4\nmixture\nof dichloromethane (DCM)/methanol/water (DiMW) was selected for this\nseparation, owing to the reduced solvent toxicity and lower density\nof DCM versus CHCl 3 , the lower cost of MeOH versus MeCN,\nand the proximity of the log( K P ) values\nto the desired range of −0.4 to +0.4. CPC Separation of a Model\nProduct Mixture The K P data provide\nthe basis for a two-stage separation\nscheme, initiated with AZ L to isolate vanillin and syringic acid,\nfollowed by DCM/MeOH/H 2 O to obtain p HBA,\nvanillic acid, and syringaldehyde. Initial CPC testing was conducted\nwith a model mixture containing small quantities (50 mg) of each of\nthe five compounds. The CPC rotor was equilibrated with AZ L solvents\nin ascending (Asc) mode, 49 and the five\ncompounds dissolved in the AZ L lower-layer solvent (predominantly\nMeOH/H 2 O) was injected into the CPC. The mobile phase was\nthen eluted at 30 mL·min –1 for 70 min, followed\nby column extrusion for 15 min ( Figure 3 c). The AZ L separation showed a resolved peak for\nvanillin, and pure fractions of syringic acid were obtained during\ncolumn extrusion. (Note: the complex peak shape for syringic acid\nreflects emulsions in the effluent that scatter light from the detector\nduring extrusion.) Syringaldehyde, p HBA, and\nvanillic acid eluted in the order expected from their K P values (cf. Figure 3 a) but were not resolved using the AZ L solvent system.\nThese compounds were therefore transferred to the second stage process.\nThe CPC was prepared for 10:6:4 DiMW Asc mode elution. The mixture\nof p HBA, vanillic acid, and syringaldehyde was dissolved\nin the DiMW upper-layer solvent (again, predominantly MeOH/H 2 O), injected, and the mobile phase was eluted for 45 min at 25 mL·min –1 followed by extrusion for 13 min ( Figure 3 d). Good separation of the\nmixture of three compounds using the DiMW solvent system ( Figure 3 d) completed the\ntwo-stage protocol the leads to effective separation of all five solutes. 50 This successful demonstration was then\napplied to larger-scale\nseparation, using a 2 g sample of the five compounds (400 mg each; Figure 4 a–e). HPLC\nwas used for quantitative analysis of the compounds following CPC\nseparation. Nearly quantitative recovery and purity (>95%) was\nobtained\nfor each compound ( Figure 4 c), consistent with the good resolution of the peaks in the\ntwo traces ( Figure 4 d and e). Figure 4 Two-stage separation sequence enabling isolation of aromatic compounds\nin a model sample and authentic lignin oxidative alkaline depolymerization\nmixture. (a) The first-stage AZ L Asc separation isolates vanillin\n(collected peak 1) and syringic acid (collected peak 3) and a mixture\nof p HBA, vanillic acid, and syringaldehyde (collected\npeak 2). (b) The second-stage DiMW Asc separation resolves the mixture\nof p HBA (collected peak 4), vanillic acid (collected\npeak 5), and syringaldehyde (collected peak 6). (c) Recoveries and\npurities attained from CPC separation of the model sample mixture.\n(d) AZ L Asc CPC trace from a model sample mixture. (e) DiMW (10:6:4)\nAsc CPC trace from a model sample mixture of the compounds present\nin peak 2 in trace d. (f) Recoveries and purities attained from CPC\nseparation of the authentic lignin depolymerization mixture. (g) AZ\nL Asc CPC trace from the authentic lignin depolymerization mixture.\nPeak #1 includes both vanillin and acetovanillone. (h) DiMW (10:6:4)\nAsc CPC trace from the authentic lignin depolymerization mixture of\nthe compounds present in peak 2 in trace g. Peak #6 includes both\nsyringaldehyde and acetosyringone. Note: The large absorption features\nin traces e and h after initiating the extrusion phase (indicated\nby the gold dashed line) correspond to the dichloromethane-rich stationary\nphase being displaced from the column. CPC Separation of Lignin Depolymerization Products The above\nresults provided the basis for CPC separation of products\nobtained from Cu-AHP lignin depolymerization. A crude 1.6 g mixture,\nobtained by pooling products of multiple depolymerization reactions,\nwas analyzed by HPLC and shown to consist of 25.5 wt % aromatic monomers,\nincluding 4.0 wt % pHBA, 5.0 wt % vanillin, 1.2 wt % vanillic acid,\n0.6 wt % acetovanillone, 11.2 wt % syringaldehyde, 2.1 wt % syringic\nacid, and 1.4 wt % acetosyringone. The HPLC trace of the crude mixture\n( Figure 4 f) shows the\npresence of numerous small peaks corresponding to unidentified low-molecular-weight\nproducts, in addition to a larger peak at long retention times corresponding\nto oligomers. This crude depolymerization mixture was subjected\nto the same CPC workflow used with the model mixture described above\n( Figures 4 a,b,g,h).\nUse of AZ L Asc conditions in the first stage separated vanillin/acetovanillone 51 and syringic acid in 91% and 79% recoveries,\nrespectively, based on HPLC analysis of the collected fractions. Oligomers\nwere also recovered from this stage (see Figure S11 in the Supporting Information for details) but were not\nanalyzed further. The large quantity of impurities in the crude sample\n(nearly 3/4 of the material consists of undesired products) limits\nthe purity of the obtained compounds to 65 and 50 wt %, respectively.\nNonetheless, this outcome reflects major enrichment of vanillin/acetovanillone\nand syringic acid relative to their composition in the original sample\n(5.6 and 2.1 wt %, respectively). The mixture of pHBA, vanillic\nacid, and syringaldehyde/acetosyringone,\nwhich coelute during the first stage, were then subjected to DiMW\nAsc CPC conditions in the second stage. The fractions obtained from\nthis run enabled isolation of the three compounds in recoveries of\n82%, 86%, and 91%, and purities of 72%, 81%, and 87%, respectively.\nHigher purities of compounds were obtained from this stage due to\nthe removal of most of the impurities during the first-stage separation.\nIn both stages, recovery values could be increased; however, dilute\nfractions before and after each of the main peaks were rejected to\nenhance compound purities."
} | 4,397 |
36207526 | PMC9546865 | pmc | 7,962 | {
"abstract": "Soil microbiota, including arbuscular mycorrhizal fungi (AMF), are critical for plant nutrition in non-agricultural ecosystems. A new study by Edlinger et al. shows that agricultural soils are negatively impacted by fungicide use and generally have lower AMF diversity and abundance."
} | 70 |
32193327 | null | s2 | 7,964 | {
"abstract": "The structural and functional complexity of multicellular biological systems, such as the brain, are beyond the reach of human design or assembly capabilities. Cells in living organisms may be recruited to construct synthetic materials or structures if treated as anatomically defined compartments for specific chemistry, harnessing biology for the assembly of complex functional structures. By integrating engineered-enzyme targeting and polymer chemistry, we genetically instructed specific living neurons to guide chemical synthesis of electrically functional (conductive or insulating) polymers at the plasma membrane. Electrophysiological and behavioral analyses confirmed that rationally designed, genetically targeted assembly of functional polymers not only preserved neuronal viability but also achieved remodeling of membrane properties and modulated cell type-specific behaviors in freely moving animals. This approach may enable the creation of diverse, complex, and functional structures and materials within living systems."
} | 259 |
36174970 | PMC9853505 | pmc | 7,966 | {
"abstract": "Over the past 40\nyears, structural and dynamic DNA nanotechnologies\nhave undoubtedly demonstrated to be effective means for organizing\nmatter at the nanoscale and reconfiguring equilibrium structures,\nin a predictable fashion and with an accuracy of a few nanometers.\nRecently, novel concepts and methodologies have been developed to\nintegrate nonequilibrium dynamics into DNA nanostructures, opening\nthe way to the construction of synthetic materials that can adapt\nto environmental changes and thus acquire new properties. In this\nReview, we summarize the strategies currently applied for the construction\nof synthetic DNA filaments and conclude by reporting some recent and\nmost relevant examples of DNA filaments that can emulate typical structural\nand dynamic features of the cytoskeleton, such as compartmentalization\nin cell-like vesicles, support for active transport of cargos, sustained\nor transient growth, and responsiveness to external stimuli.",
"conclusion": "5 Conclusions\nand Perspectives Synthetic DNA filaments may find interesting\napplications as components\nof nanoelectronic devices, biomimetic materials for tissue engineering,\nor alternative tools for biophysical characterizations. Early studies\non DNA nanotubes showed the use of these structures as templates for\nthe growth of conductive nanowires. 76 In\nother reports, DNA filaments were employed to mediate the weak alignment\nof membrane proteins and facilitate their structural elucidation by\nNMR spectroscopy. 77 Further developments\nin the field allowed the application of DNA filaments for the super-resolution\nimaging of DNA nanostructures. 78 The emulation of the structure and function of the cytoskeleton\nis probably one of the most fascinating and instructive fields of\napplications of DNA-based filaments. In the past few years, ingenious\nconstructs and advanced technologies have contributed to a deeper\nunderstanding of the mechanisms that regulate the dynamics of protein\nfilament growth and shrinkage. Moreover, biochemical methods have\nbeen developed to reproduce synthetic analogues of protein scaffolds\nfor the anchorage or transport of molecular cargos, as well as for\ncellular propulsion and motility. For example, hierarchical\nself-assembly strategies were applied\nto link small DNA rodlike units into complex two- and three-dimensional\nmeshlike architectures, 7 , 38 which can be used as artificial\nmodels of cell matrices. Other examples report about the use of DNA\nfilaments as tracks for programmed long-range molecular motion. 79 , 80 Finally, striking examples of stimuli-responsive DNA filaments have\nshown that the potential of these structures can go much beyond their\nrole as scaffolding static elements. These filaments, indeed, can\nbe designed to sense the surrounding environment and actuate a change\nin response to it, thus emulating the active functioning and adaptiveness\nof their natural analogues. The path is therefore open to the more\nambitious goal of constructing artificial cells from the bottom. An illustrative example of this endeavor was reported by the Liedl\ngroup a few years ago. 81 In this work,\nmagnetic nanoparticles were decorated with supertwisted DNA nanotubes\nvia hybridization with complementary strands. Upon the application\nof a rotating and homogeneous magnetic field, the DNA filaments formed\na bundle that spontaneously aligned on one side of the particles,\nthus resembling the shape of bacterial flagella. Even more remarkably,\nthe bundle acquired a coordinated clockwise (or counterclockwise) rotation that moved the particles forward\nalong the rotation axis, thus mimicking the propulsion movement of\ntheir natural analogues. Despite still being simple prototypes of\nprokaryotic flagella, such artificial DNA swimmers undoubtedly demonstrate\nthat DNA self-assembly procedures can be employed to engineer biomimetic\nmaterials, otherwise extremely challenging to realize with other fabrication\nmethods. Major progress in this field has been recently shown\nby Schulman\nand co-workers. 82 In their work, DNA nanotubes\nwere designed to nucleate and grow on a surface at defined locations.\nOne end of the filament was attached to the surface, while the opposite\nend was free to diffuse in solution. Pairs of tubular structures at\npredefined distances (between 1 and 10 μm) and relative orientations\nwere then joined at their growing tips. Unpaired filaments were selectively\nmelted away by exchanging the buffer with a solution that contained\nno free tiles, thus effectively reducing the concentration of the\nmonomer in solution below the critical value necessary for polymer\ngrowth. In a recent report, the same group organized micrometer-scale\nDNA nanotubes at specific receptors on living cells, 83 proposing a model system to mimic the structural and functional\nrole of cell membrane protrusions. For this purpose, the authors engineered\na quite complex construct to link DNA nanotubes to epidermal growth\nfactor receptors (EGFR) overexpressed on the surface of HeLa cells\n( Figure 6 c, top panel).\nThe construct was sequentially composed of EGFR primary antibodies,\nbiotinylated secondary antibodies, streptavidin, and a biotin-modified\nstrand that hybridizes to complementary DNA sequences on the DNA origami\nseed, from which the nanotube grows. The authors demonstrated that\nsuch artificial DNA filaments specifically recognize receptors on\nthe surface of live cells, grow on the membrane, and sense an external\nfluid flow ( Figure 6 c, bottom panel). Recently, DNA origami-coiled filaments have been\nused to control cell motion by targeting clusters of integrin receptors\non the surface of HeLa cells. 84 The filaments\nwere modified with RGD ligands domains, the spacing of which (and\nconsequently the extent of integrin-receptors clustering) was regulated\nby pH through the insertion of i-motif structures between neighboring\nstruts of the DNA origami device. The internalization of DNA\nfilaments within artificial compartments\nand the sculpting of membrane shape, from either the outside or inside\nof lipid vesicles, are other crucial features of cells that have been\nemulated in several synthetic biology approaches. 85 − 89 For example, Göpfrich and coauthors succeeded\nin reconstituting a DNA cytoskeleton inside giant unilamellar vesicles\n(GUV) and showed that, by suitably modifying the design of the DNA-tile\nunit, various types of filament responses can be visualized. 86 In this way, they demonstrated the light-induced\nassembly and disassembly of the filaments, the formation of bundles\nand rings with high persistence lengths, and the appearance of DNA\ncortices deforming the vesicle from the interior. The Schwille group\nintensively investigated the impact of DNA filaments on the topological\ntransformation of membranes, revealing that the curvature, membrane\naffinity, and surface density of the filaments are crucial for the\ninduction of tubular membrane deformations. 89 These and similar studies will surely advance the understanding\nof the physical–chemical mechanisms that control membrane deformation\nand will boost the further development of exciting applications, in\nwhich more elaborate DNA origami assemblies can be envisaged, for\nexample, as drug delivery vehicles for targeting biological membrane\nbarriers. A final representative example of the level of sophistication\nthat can be achieved by the smart combination of microfluidic technologies,\nDNA nanotechnology, and enzyme chemistry has been recently reported\nby the Göpfrich group. 90 In their\nwork, the authors engineered a system that embodies some of the most\npeculiar features of natural cytoskeletons, namely, compartmentalization,\nATP-driven polymerization, dynamic growth/collapse, and intracellular\ncargo transport. Inorganic gold nanoparticles and lipid vesicles were\nmodified with DNA sequences for hybridization to complementary RNA\nhandles extending from the DNA filaments. Molecular transport along\nthe track was finally powered by the RNase H-mediated hydrolysis of\nthe RNA handles, thus causing the detachment of the cargo and its\nforward movement according to a burnt-bridge mechanism ( Figure 6 d). In conclusion, synthetic\nDNA filaments present several structural\nand dynamic aspects that are typical of their protein counterparts.\nThe structural properties of DNA filaments are critical for the support\nand transport of molecular cargos as well as membrane sculpturing\nand essentially rely on the fact that the mechanical features of these\nstructures (i.e., the persistence length and curvature) can be rationally\ndesigned, approaching the performance and morphological diversity\nexhibited by their natural protein analogues. Nevertheless, more sophisticated\nhierarchical assembly strategies are required to improve the mechanical\nresilience of DNA filaments, ideally up to the micrometer scale. This\nwould allow, for example, to better mimic the structural role played\nby microtubules and actin bundles or even to reproduce the complex\norganization of the axoneme, the fundamental element of cilia and\neukaryotic flagella. 91 When the physical\nproperties of the filaments are coupled to mechanisms of molecular\nrecognition and transient consumption/release of chemical energy,\nmore complex and intriguing phenomena emerge, such as sensing, motility,\nand dynamic instability. Design and assembly procedures have meanwhile\nevolved to enable the engineering of such “active” DNA-based\nmaterials, which can fulfill defined tasks only when needed by adapting\ntheir shape in response to environmental inputs and, most importantly,\nin a predictable fashion. Nowadays, synthetic DNA filaments can impressively\nresemble the adaptive and evolutionary behavior of natural systems,\nand, in this sense, they are one of the most illustrative examples\nof truly biomimetic materials. Remarkably, the programmability of\nDNA with nanometer-scale accuracy provides these structures with the\nright level of architectural detail, which is necessary to establish\na robust structure–function relationship and master the physical\nproperties and dynamic behavior of the filaments at the macroscopic\nscale. What are the next challenges? How can the gap between\nartificial\nand natural filaments be reduced? For the long term, a very ambitious\ngoal would be to mimic the spatial organization and temporal evolution\nof larger filament assemblies, enabling for example to emulate, although\nprobably only in a minimal form, the extremely complicated and fascinating\ncoordination of mechanical forces occurring during chromosome segregation.\nDespite arduous tasks that still must be overcome to reach this and\nother ambitious goals, the achievements obtained up to now and mainly\nin the past few years undoubtedly demonstrate that the future of man-made\nDNA filaments for the bottom-up construction of artificial cells is\nbright and promises to be extremely stimulating and instructive.",
"introduction": "1 Introduction Filaments, with one dimension\nof the structure being much larger\nthan the other two, are ubiquitous in nature. Emblematic examples\nare genomic DNA 1 and protein filaments\nof the cytoskeleton. 2 − 4 Independent of their structural composition, a common\nfeature of many natural filaments is the periodicity of their pattern,\ni.e., the recurrence of identical or very similar building components\nalong the entire polymer chain. These units are linked together according\nto precise interunit association rules, and the resulting linear structure\nis often further organized into hierarchical architectures of higher\nstructural order. Despite being simple, this self-assembly principle\nis extremely powerful for the generation of materials with superior\nmechanical properties, meaning that the global features of the final\npolymer are more than the sum of the features of its single components.\nIn the cell, filamentous protein structures ensure structural rigidity,\ncell motility, cargo transport, as well as growth and division. Hence,\nthe advancement of methods for the synthesis of man-made filaments\nwith programmable energetic and kinetic features is very appealing,\nnot only for a better understanding of the functioning of many biological\nbeams but also for the creation of novel bioinspired materials that\ncan adapt, respond, and evolve autonomously, once a sufficient energy\nsource is provided. A possible way to achieve this ambitious\ngoal relies on the programmability\nof the DNA molecule. 5 , 6 In the past few decades, both\nstructural and dynamic DNA nanotechnologies 7 − 9 have amply demonstrated\nthat DNA sequences can be designed to achieve a desired structure\nat equilibrium and that not only can this structure be predictably\nreconfigured in a postassembly process, but also the assembly itself\ncan be even controlled during its occurrence. In\nother words, almost every aspect of the energy landscape of nanostructure\nformation and transformation can be affected in a rational manner\nand reshaped by suitable means. Moreover, by combining DNA strand-displacement\nreaction networks with smart chemical or enzymatic systems, as well\nas crystallization methodologies and microfluidic techniques, synthetic\nDNA filaments have reached a level of complexity that allows the structural,\nmechanical, and dynamic properties of their natural protein analogues\nto be mimicked in many ways. In this Review, we will briefly\nsurvey three aspects of synthetic\nDNA filaments, with each aspect reported in a dedicated section. The\nfirst section will focus on the design strategies used so far to engineer\nDNA filaments. Specifically, we will describe how base hybridization\nand base stacking can be rationally mastered to achieve the desired\nshape of the building unit and how several units can be programmed\nto associate into linear structures with predictable features. In\nthe second part, we will explain how these approaches have been implemented\nto control the elastic properties of DNA filaments at equilibrium,\nsuch as their persistence length, bending degree, or twisting extent.\nFinally, in the third section, we will describe some of the newly\nemerging methods that allow the control of the polymerization state\nof DNA filaments, either through equilibrium-switching mechanisms\nor through more complex out-of-equilibrium (i.e., truly dynamic) processes.\nThis topic has been deeply treated in recent authoritative reviews\nand references therein. 10 − 14"
} | 3,593 |
32934307 | PMC7492276 | pmc | 7,969 | {
"abstract": "Engineering bacteria to clean-up oil spills is rapidly advancing but faces regulatory hurdles and environmental concerns. Here, we develop a new technology to harness indigenous soil microbial communities for bioremediation by flooding local populations with catabolic genes for petroleum hydrocarbon degradation. Overexpressing three enzymes (almA, xylE, p450cam) in Escherichia coli led to degradation of 60–99% of target hydrocarbon substrates. Mating experiments, fluorescence microscopy and TEM revealed indigenous bacteria could obtain these vectors from E. coli through several mechanisms of horizontal gene transfer (HGT), including conjugation and cytoplasmic exchange through nanotubes. Inoculating petroleum-polluted sediments with E. coli carrying the vector pSF-OXB15-p450camfusion showed that the E. coli cells died after five days but a variety of bacteria received and carried the vector for over 60 days after inoculation. Within 60 days, the total petroleum hydrocarbon content of the polluted soil was reduced by 46%. Pilot experiments show that vectors only persist in indigenous populations when under selection pressure, disappearing when this carbon source is removed. This approach to remediation could prime indigenous bacteria for degrading pollutants while providing minimal ecosystem disturbance.",
"conclusion": "Conclusion Cleaning up environmental contamination from human activities is one of the greatest un-met challenges of the twenty-first century. Our pilot research has shown that transferring catabolic genes involved in petroleum degradation from E. coli DH5α to indigenous bacteria may be a viable solution. This system could be adapted to exploit genes from local microbial populations which are already primed for degradation. This could be achieved by isolation and identification of native strains which degrade petroleum and proteomic identification and screening of candidate enzymes for over-expression. Future research is needed to determine (1) how long these plasmids are maintained under field conditions, (2) whether genetic mutations accumulate over time that might impact enzyme functioning, and (3) how vector-based gene drives harnessing natural processes of conjugation may affect local microbial community composition and soil metabolic functions. Concentrated efforts among microbiologists, ecologists, synthetic biologists and policy makers in this new area of research may usher in a new era of how we respond to environmental disasters and toxic waste management in the Anthropocene.",
"introduction": "Introduction Oil spills in recent decades have left a long-term mark on the environment, ecosystem functioning, and human health 1 – 3 . In the Niger Delta alone, the roughly 12,000 spills since the 1970s have left wells contaminated with benzene levels 1,000× greater than the safe limit established by the World Health organization and have irreparably damaged native mangrove ecosystems 4 , 5 . Continued economic reliance on crude oil and legislation supporting the oil industry mean that the threat of spills is unlikely to go away in the near future 6 .\n At present, there are few solutions to cleaning up oil spills. Current approaches to removing crude oil from the environment include chemical oxidation, soil removal, soil capping, incineration, and oil skimming (in marine contexts) 7 , 8 . While potentially a ‘quick fix,’ none of these solutions are ideal. Soil removal can be costly and simply moves toxic waste from one site to another 9 . Chemical oxidants can alter soil microbial community composition and pollute groundwater 10 . Incineration can increase the level of pollutants and carbon dioxide in the air and adversely affect human health 11 . Practices such as skimming only remove the surface fraction of the oil while the water-soluble portion cannot be recovered, negatively affecting marine ecosystems 12 , 13 . Synthetic biology has now given us the tools to tackle grand environmental challenges like industrial pollution and could usher in a new era of ecological engineering based on the coupling of synthetic organisms with natural ecosystem processes 14 – 18 . Consequently, using bacteria specially engineered to degrade petroleum could present a viable solution to cleaning up oil spills in the near future. Previous studies have identified which bacterial enzymes are involved in petroleum hydrocarbon degradation (reviewed in references 9 , 19 ) and have engineered bacterial enzymes like p450cam for optimal in vivo and in vitro degradation of single-substrate hydrocarbons under lab conditions 20 , 21 . However, there are several critical gaps in or knowledge of engineering bacteria for oil-spill bioremediation. First, we know little about how the performance of these enzymes compare and which enzyme would present an ideal target for over-expression in engineered organisms. Second, it is unclear how well engineered organisms can degrade petroleum hydrocarbons compared to native wild-type bacteria which naturally degrade alkanes, such as Pseudomonas putida . Third, the environmental effects of engineered bacteria on native soil populations are unclear. For example, do these bacteria persist over time in contaminated soils? Although the use of genetically modified bacteria in bioremediation is attractive, this solution faces significant regulatory hurdles which prohibit the release of genetically modified organisms in the environment 22 . Here, we propose a new bioremediation strategy which combines synthetic biology and microbial ecology and harnesses natural processes of horizontal gene transfer in soil ecosystems. We screened five enzymes involved in petroleum degradation in E. coli DH5α (alkB, almA, xylE, ndo and p450cam) to identify (1) where these enzymes localize and their effect on crude oil using advanced microscopy and (2) to asses each enzyme’s ability to degrade three petroleum hydrocarbon substrates (crude oil, dodecane, and benzo(a)pyrene) compared to two wild type bacteria ( Pseudomonas putida and Cupriavidus sp. OPK) using bioassays and SPME GC/MS. Based on these results, we selected one vector (pSF-OXB15-p450camfusion) to determine whether small, synthetic vectors carrying catabolic genes could be transferred to indigenous bacteria found in petroleum-polluted sediments and whether this shift in community metabolism could increase rates of pollutant degradation.",
"discussion": "Results and discussion Overexpression of petroleum hydrocarbon-degrading enzymes in E. coli To compare the localization and activity of known petroleum hydrocarbon-degrading enzymes, we inserted five enzymes (alkB, almA, xylE, ndo, and p450cam) and required electron donors into the vector backbone pSF-OXB15 using Gibson Assembly 23 (SI Fig. 1 ). To identify where each enzyme localized within E. coli DH5α, we tagged each enzyme with a fluorophore (gfp or mcherry). Fluorescence microscopy revealed that alkB was localized to bacterial cell membranes and almA was found throughout the cytoplasm. The camphor-5-monooxygenase camC from the p450cam operon was expressed throughout the cell cytoplasm while another enzyme in the operon, the 5-exo-hydroxycamphor dehydrogenase camD, was expressed within a small compartment at one end of the cell (Fig. 1 A). The dioxygenase ndoC from the ndo operon was also localized to a small compartment at one end of the cell. The dioxygenase xylE was found in small amounts in the bacterial cell membrane and larger amounts in a microcompartment at one end of the cell. In all cases, these compartments were 115–130 nm wide and could be seen in young, mature and dividing cells. The presence of microcompartments in E. coli expressing p450cam, ndo, and xylE could reflect the known use of protein-based microcompartments by bacteria to concentrate highly reactive metabolic processes 24 . Figure 1 Expression and localization of bacterial monooxygenases and dioxygenases involved in petroleum degradation in E. coli DH5α. ( A ) Structured Illumination Microscopy (SIM) image of E. coli DH5α expressing proteins involved in petroleum degradation (cam A, B, C and D) from the CAM plasmid in E. coli . camC (fused to mcherry) was found throughout the cell while camD (fused to gfp) localized to a microcompartment at one end of the cell. The scale bar is 5 μm. ( B ) E. coli DH5α expressing alkB fuse to gfp were found clinging to spheres containing crude oil, mimicking a behavior seen in wild-type oil-degrading bacteria. ( C ) EPS from E. coli DH5α expressing xylE. Gfp-tagged xylE were found around small pores (ca. 500 nm) within the EPS matrix. ( B ) and ( C ) were taken using the GFP filter on a Zeiss AxioImager M1. Over-expression of all five enzymes imbued E. coli DH5α with metabolism-dependent chemotactic behavior, where cell movement is driven towards substrates affecting cellular energy levels 25 . E. coli DH5α do not have flagella, but rather, moved towards petroleum hydrocarbon substrates via twitching 26 – 28 . Fluorescence microscopy showed E. coli DH5α expressing alkB ‘clinging’ to oil droplets (Fig. 1 B) and those expressing xylE seemed to use the compartment-bound enzyme as a ‘guide’ towards crude oil (SI Fig. 2 ). Both behaviors mimic the interactions of wild-type oil-degrading bacteria 7 . Figure 2 Growth of wild-type and engineered bacteria on dodecane ( A ), benzo(a)pyrene ( B ), and crude oil ( C ). Wild type strains are denoted as ‘ Cupriavidus ’ and ‘ P. putida .’ Synthetic strains are denoted according to what enzyme they are engineered to express (e.g. alkB, almA). The two negative controls are a control with the carbon substrate but no cells and E. coli DH5α transformed with the vector backbone used in the experiment (pSFOXB15) but without genes inserted for hydrocarbon degradation. Optical density measurements were taken at OD 600 nm. Scale bars show standard error of the mean. Fluorescence microscopy also revealed for the first time the key role of extracellular enzymes in degradation of petroleum hydrocarbons. Three enzymes, alkB, almA, and p450cam were found in extracellular vesicles ranging in size from 0.68 μm to 1.67 μm (SI Figs. 3 and 4 ). These vesicles were only seen when E. coli DH5α was exposed to petroleum hydrocarbons. They are larger than minicells (which range from 200–400 nm in diameter) 29 and seem to serve some other function. Confocal images suggest that these vesicles may come into contact with oil droplets, potentially attaching to (or merging with) their surface (SI Fig. 4 ). Figure 3 Expression of pSF-OXB15-p450camfusion in the marine bacteria Planococcus citreus . ( A ) DIC image of wild-type P. citreus (no exposure to E. coli ) . These bacteria are coccoid-shaped with cells found individually or in groups of 1–4. ( B ) Image of P. citreus and E. coli expressing p450cam after 48 h of co-cultivation (described in methods). ( B ) was taken using the Texas Red filter on a Zeiss AxioImager M1 (excitation/emission 561/615). Figure 4 TEM confirmation of horizontal gene transfer among E. coli expressing pSF-OXB15-p450camfusion and P. putida . ( A ) E. coli DH5α (E) connected to P. putida (P) via conjugative pili. ( B ) Mating-pair bridge between P. putida and E. coli DH5α. ( C ) E. coli DH5α with conjugative pili. ( D ) Nanotube connecting P. putida and E. coli DH5α. We also found three enzymes, alkB, xylE, and p450cam, within the E. coli exopolysaccharide (EPS) matrix. AlkB and xylE were concentrated around the 500 nm pores within the EPS and found dispersed in smaller amounts throughout the exopolysaccharide (Fig. 1 C; SI Fig. 5 ). In contrast, p450cam was distributed in high amounts throughout the EPS. Cryotome sectioning of the EPS from bacteria expressing p450cam indicates that the monooxygenase camA from the p450cam operon co-localized with a second enzyme involved in hydrocarbon degradation, the dehydrogenase camD (SI Fig. 6 ). Protein levels of EPS varied significantly among the different strains of wild-type and engineered bacteria (ANOVA test, F 7,16 = 11.3, p < 0.001, adjusted R 2 = 0.76). Bacteria expressing alkB (0.84 ± 0.04 mg/ml), xylE (0.85 ± 0.13 mg/ml), and p450cam (0.97 ± 0.06 mg/ml) had higher levels of EPS protein than E. coli DH5α expressing the empty vector pSF-OXB15 (0.44 ± 0.02 mg/ml) (SI Fig. 7 7 and were comparable to the EPS protein levels of Cupriavidus sp. OPK (1.2 ± 0.15 mg/ml), a bacteria known to use biofilms to degrade crude oil 7 . Although previous studies suggest that EPS may be involved in the extra-cellular metabolism of environmental pollutants 7 , 30 – 34 , this is the first study to identify several enzymes which may play a role in this process. Figure 5 Expression of vector pSF-OXB15-p450cammcherry in indigenous microbial communities in sediment from a former Shell Oil refinery (Shell Pond, Bay Point, CA). ( A ) Example of native soil bacteria expressing the vector. These bacteria construct spheres made of polymers in a honey-comb pattern. ( B ) Expression of p450cam tagged with mcherry in biofilm matrices within sediment from Shell Pond. The biofilms form thin nets over large soil particles; small soil particles can be seen as black ‘dots’ sticking to the biofilm surface. Bacteria carrying the vector can be seen embedded in the EPS. In all microscopy images, mcherry is false-colored yellow. The image was taken using the Texas Red filter on a Zeiss AxioImager M1 (excitation/emission 561/615). Figure 6 Overview of horizontal “gene-drive” system. The diagram depicts two approaches to flooding indigenous bacterial populations with catabolic genes of interest (1) through controlled mating and re-release or (2) direct application of E. coli DH5α with the catabolic genes of interest. Step 4 is there as a check to ensure the native bacteria take up the plasmid while in a soil matrix and can be shortened/extended as needed. The final step 5 is for monitoring purposes and can be shortened/extended as needed. This protocol works with water/marine samples (substituting water for soil). Figure created with BioRender.com. Finally, extracellular enzyme expression also influenced the size of the oil droplets within the cell culture media. For example, E. coli DH5α expressing xylE produced very small oil droplets (primarily > 1 μm in diameter) of crude oil while those expressing alkB and almA produced droplets ranging in size from 1 μm-120 μm in diameter (SI Fig. 8 ). Although xylE was not seen in vesicles, confocal suggests that crude oil droplets can flow through pores in biofilms and may become coated in EPS in the process. Potentially, attachment of enzymes to oil droplets (through fusion with vesicles or contact with EPS) may influence how fast a droplet is degraded over time. Previous studies have shown that vesicles embedded with enzymes can catalyze chemical reactions 35 , 36 . In addition, Dmitriev et al. have shown that two bacteria, P. putida BS3701 and Rhodococcus sp. S67, use vesicular structures containing oxidative enzymes which attach to and play a role in degradation of crude oil droplets 37 . Our results thus suggest that bacterial monooxygenases and dioxygenases involved in petroleum hydrocarbon degradation may be involved in multiple, complex inter and intra-cellular processes that lead to the degradation of crude oil. Comparison of enzyme activity To determine which enzymes were most useful for degrading long-chain hydrocarbons, polyaromatic hydrocarbons (PAHs), and crude oil, we conducted 96-well plate assays exposing wild-type and genetically engineered bacteria to 1% (v/v) of dodecane, benzo(a)pyrene or crude oil. We found that bacteria engineered to over-express specific enzymes in petroleum degradation were able to degrade single-carbon substrates better than the wild-type bacteria P. putida and Cupriavidus sp. OPK. One way analysis of variance (ANOVA) of the assay data showed that there was significant variation in bacterial growth when exposed to dodecane (F 8,31 = 33.4, p = < 0.001), benzo(a) pyrene (F 8,31 = 73.03, p = < 0.001) and crude oil (F 8,31 = 240.6, p = < 0.001). When exposed to dodecane, E. coli DH5α expressing p450cam increased in biomass the most (139.4%), followed by E. coli DH5α expressing xylE (136.3%), alkB (120.8%), and almA (97.6%) (Fig. 2 A). Expressing p450cam and xylE led to significantly greater conversion of dodecane to biomass compared to P. putida ( t = 4.71, df = 3.17, p < 0.01 and t = 4.41, df = 3.17, p < 0.01 respectively). Solid-Phase Micro-Extraction (SPME) GC/MS analysis of these cultures revealed that all three bacteria degraded 99% of dodecane in 10 days. When exposed to benzo(a)pyrene, P. putida had the greatest increase in biomass (119.2%) followed by E. coli DH5α expressing almA (117.1%), xylE (94.8%), and p450cam (90.8%) (Fig. 2 B). T-tests showed there was no significant difference in the biomass of P. putida and these three strains ( p > 0.10). SPME GC/MS showed that E. coli expressing P450cam, almA and xylE degraded 90%, 97% and 98% of the benzo(a)pyrene respectively while P. putida degraded 86%. In contrast, when engineered and wild-type bacteria were exposed to crude oil, P. putida converted the oil to biomass more efficiently, increasing in biomass by 110.9% (Fig. 2 C). Only two genetically engineered bacteria, E. coli DH5α expressing p450cam and almA, had comparable increases in biomass to Cupriavidus sp. OPK (61.93%, 52%, and 48.7% respectively). The assay was repeated with crude oil stained with Nile Red and rates of degradation were determined according to French and Terry 7 . P. putida degraded 79% of crude oil while E. coli DH5α expressing p450cam and almA degraded 64% and 60% respectively. E. coli expressing alkB, xylE, and ndo only grew ~ 25% and degraded 35–40% of crude oil. The high performance of p450cam when exposed to crude oil likely reflects the enzyme’s known substrate promiscuity 38 , 39 which makes it a better catalyst for degrading crude oil, a complex substrate made of over 1,000 compounds 40 . Vector-exchange between E. coli DH5α and indigenous bacteria To determine whether our engineered bacteria could transfer non-conjugative, synthetic vectors containing petroleum-degrading genes to indigenous soil and marine bacteria, we conducted a series of mating experiments. We found that wild-type bacteria readily received the vector pSF-OXB15-p450camfusion through horizontal gene transfer (HGT) (Fig. 3 ; SI Fig. 9 ). Frequencies of transformation ranged from 19 to 84% in 48 h depending on the recipient species (SI Table 1) and were > 90% after seven days of incubation for all tested species. Plasmid expression was stable for over three months in the absence of antibiotic pressure. Although our vectors carried a ColE1 origin of replication, this did not seem to present a barrier to HGT. This agrees with previous studies which suggest ColE1 plasmids can be found in wild bacteria 41 and wild-type bacteria can receive ColE1 plasmids from E. coli 42 . HGT can occur through transformation, transduction, conjugation, transposable elements, and the fusing of outer membrane vesicles (OMVs) from one species to another 43 , 44 . To test whether wild-type bacteria could take up naked plasmids from cell culture, we adding 1 μl of purified plasmid (at a concentration of 1 ng/μl and 10 ng/μl) to LB cultures containing wild-type bacteria and by spreading diluted vectors onto agar plates. We saw no transformed cells. In addition, neither fluorescence microscopy or TEM showed OMV production or release by the E. coli DH5α strains created in this study. OMVs are 50–250 nm in diameter 45 , much smaller than any of the vesicles produced by our strains. To determine whether HGT of the synthetic vectors to wild-type bacteria was achieved through mating, we conducted TEM of wild-type bacteria after exposure to E. coli DH5α carrying pSF-OXB15-p450camfusion. TEM suggests several mechanisms of HGT through direct cell-to-cell contact may explain how vectors were transferred between transgenic E. coli DH5α and wild-type bacteria 46 . In our study, we found E. coli DH5α and wild-type bacteria engaging in DNA transfer through conjugation and cell merging and in cytoplasmic transfer via nanotube networks. TEM showed E. coli DH5α expressing p450cam tethered to wild-type cells by conjugative pili over long distances (Fig. 4 A), the formation of mating pair bridges between wild-type cells and E. coli DH5α (Fig. 4 B), E. coli with conjugative pili (Fig. 4 C), and E. coli and wild-type cells connected via nanotubes (Fig. 4 D) (see also SI Fig. 10 ). Although the plasmids used in this study were non-conjugative, such plasmids can be mobilized and transmitted via conjugation in the presence of other conjugative plasmids or by merging with conjugative plasmids 47 – 49 . Previous studies have also shown that plasmids (and chromosomal DNA) can be transferred through cell-contact dependent transfer without the use of conjugative pilii. This mechanism was first observed in 1968 in Bacillus subtilis and subsequently in other species (e.g. Vibrio , Pseudomonas , Escherichia ) (reviewed in 49 ). For example, Paul et al. 50 found that lab strains of E. coli could transfer plasmid DNA to Vibrio through this process. Nonconjugal plasmids can also be transferred from between bacteria through nanotubes 51 . Dubey and Ben-Yehuda 52 show in their classic paper that gfp molecules, calcein, and plasmids could be transferred between B. subtilis cells. They also show that a non-integrative vector carrying a resistance marker from B. subtilis could be transferred to Staphylococcus aureus and E. coli (with recipient cells expressing antibiotic resistance). This transfer was rapid and could happen in as little as 30 min. In our study, it is impossible to say definitively by which mechanism our vectors were transferred from E. coli DH5α to the wild-type bacteria and in reality multiple mechanisms of transfer may be possible. We conducted an additional experiment to determine the survival rate of engineered bacteria in petroleum polluted sediment from a former Shell Oil refinery in Bay Point, CA and whether these genes could be transferred to native, complex soil microbial communities. This sediment is contaminated with high levels of petroleum hydrocarbons, arsenic, heavy metals, and carbon black. At D 0 , E. coli DH5α containing the plasmid pSF-OXB15-p450camfusion were seen in aliquots of contaminated sediment and there were no autofluorescent bacteria visible in the media. After D 5 , the population of E. coli DH5α had declined. Instead, a number of diverse native soil bacteria now contained the plasmid, a trend which continued over the course of the experiment (Fig. 5 ). Based on morphological analysis of over 100 microscopy images and pairwise mating between bacteria isolated and identified (via sequencing) from Shell Pond, these bacteria belonged to the Pseudomonas , Flavobacteria , and Actinomycete genera among others. Plating out aliquots of soil at regular time points confirmed the data gathered by microscopy: the number and diversity of bacteria expressing the plasmid increased 50-fold over the first 30 days of the experiment (from 2.6 × 10 –4 CFU at D 0 to 1.25 × 10 –6 CFU at D 30 ) (SI Fig. 11 ). The spider-silk-like biofilms formed by native soil microbiota present in the soil were also fluorescent (Fig. 5 ), suggesting the p450cam enzymes also play a role in extracellular degradation of petroleum hydrocarbons under real-world conditions. GC/MS analysis showed that the amount of total petroleum hydrocarbons within the contaminated sediments decreased by 46% within 60 days compared to untreated soil samples. We left the experiment running and after 120 days bacteria carrying the vector were still prolific (SI Fig. 12 ). A pilot experiment using artificially contaminated water samples suggest that genes are transferred from E. coli to indigenous bacteria only when oil is present. In the control samples without oil, we saw no bacteria carrying the vector over the course of the 30-day experiment (the experiment will be published fully in a later publication). HGT is thought to play a role in the degradation of environmental toxins 53 and several studies have shown that wild-type bacteria carrying large plasmids with degradative genes can pass these genes on to a limited number of bacteria 54 . However, this is the first study to provide evidence for HGT between E. coli DH5α carrying a small, non-conjugative vector and wild soil microbiota. Our results show that adding engineered E. coli DH5α carrying small synthetic plasmids to polluted environmental samples may be even more effective than adding a wild-type bacteria with a larger catabolic vector. Previous studies show that frequencies of HGT between the donor and recipient bacteria in soil are low (e.g. 3 × 10 –3 CFU transconjugants per gram of sterile soil), recipient cells come from only a few genera, and the spread of the catabolic vector through the microbial community does not always lead to enhanced degradation 55 – 57 . The high transfer frequency of synthetic plasmids like the ones used in this study could be due to several reasons, including the small size of the vector (natural plasmids carrying degradative genes, e.g. OCT from Pseudomonas putida , are > 30 k bp), the inability of E. coli DH5α to avoid conjugation/cytoplasmic exchange with wild-type bacteria, and selective pressure (acquiring genes related to hydrocarbon degradation would increase host fitness) 58 , 59 . Our results also suggest native soil and marine microbial communities will continue to carry synthetic vectors with useful catabolic genes, likely until the metabolic cost of replicating the vector no longer presents an advantage. This agrees with previous studies which show that HGT transfer events that alter organism metabolism can increase organism fitness by allowing them to colonize new ecological niches 57 , 60 , 61 . It also suggests that this approach to engineering polluted ecosystems is self-limiting: selective pressure from the existing ecosystem could remove introduced genes from local microbial populations when no longer needed. Our results suggest transgenic bacteria can easily transfer genes to wild-type soil bacteria. Potentially, engineered bacteria could be used in soil and marine-based ‘gene drives.’ This drive could be achieved two ways: by adding the engineered E. coli DH5α carrying catabolic genes of interest directly to oil-contaminated soil/water or by adding pre-mated indigenous bacteria to the contaminated site (Fig. 6 ). Previous studies have proposed this form of ecological engineering 62 , 63 , but the best of our knowledge no studies have shown that such drives would be successful. E. coli DH5α engineered to carry plasmids containing genes involved in degradation of environmental toxins could be used to augment the capacity of native soil microbial communities to degrade pollutants of interest. Replacing antibiotic selection markers with chromoprotein ones 64 would eliminate the release of antibiotic resistance genes into the environment. The primary barriers to implementing this approach on current polluted industrial sites are (1) lack of standardized procedures to test and ultimately allow the use of GM organisms for environmental applications and (2) the willingness of site managers to adopt this approach to remediation. The first barrier can be overcome by developing a comprehensive checklist system for assessing the impact of GM organisms on the environment which would involve a standardized field-trial with appropriate biocontainment measures in place. For example, soil cores could be taken, encased in plastic tubing, and returned to the site for long-term monitoring on the effects of the GM organism on microbial community structure, pollutant degradation, and other factors such as vegetation cover. The EPA does have a system in place for assessing the risk posed by GM bacteria to human and environmental health 65 ; yet the assessment procedures used aren’t clear from publicly available documentation and the frequency of researchers seeking out EPA approval for field trial of bacteria engineered for remediation is unknown. Publicly available documentation and potentially even de-centralized approval of GM field trials (e.g. through university Environmental Health and Safety offices) could make field trials of GM bacteria more achievable in the near future. The second barrier can be overcome through public engagement with those working in the remediation sector (industry, site managers, and remediation consulting firms) and a shift in our approach to how we conduct remediation (favoring slower biological-based solutions that harness local ecological and chemical processes over faster processes such as oxidation and soil removal)."
} | 7,259 |
40268969 | PMC12019246 | pmc | 7,970 | {
"abstract": "Stretchable organic electrochemical transistors (OECTs) are promising for flexible electronics. However, the balance between stretchability and electrical properties is a great challenge for OECTs. Here, high-performance stretchable all-gel OECTs based on semiconducting polymer gel active layers and poly(ionic liquid) ionogel electrolytes are developed. The all-gel network structures effectively promote ion penetration/transport and endows the OECTs with high stretchability. The resulting OECTs exhibit an excellent combination of ultra-high transconductance of 86.4 mS, on/off ratio of 1.2 × 10 5 , stretchability up to 50%, and high stretching stability up to 10000 cycles under 30% strain. We demonstrate that the all-gel OECTs can be used as stretchable pressure-sensitive electronic skins with a low detection limit for tactile perception of robotic hands. In addition, the all-gel OECTs can be applied as stretchable artificial synapses for neuromorphic simulation and highly sensitive stretchable gas sensors for simulating olfactory perception process and monitoring food quality. This work provides a general all-gel strategy toward high-performance flexible electronics.",
"introduction": "Introduction With the continuous development of electronic technology, stretchable and wearable electronics are urgently needed to meet the increasing demand for electronic skins, soft robots, human-machine interfaces, and biomedical monitoring 1 – 3 . Stretchable organic electrochemical transistors (OECTs), as an emerging electronic component, has attracted extensive attention due to their unique properties and promising applications in flexible electronics, biosensors, chemical sensors, and neuromorphic systems 4 , 5 . The stretchable nature of the stretchable OECTs allows them to operate under extreme deformation conditions, making them suitable for wearable and conformal devices that require large deformations. The emergence of this technology not only promotes the innovation of electronic device design, but also offers possibilities for the development of flexible electronics and biosensors. However, current OECTs face the following challenges in practical applications. On one hand, the reported stretchable OECTs show limited transconductance, resulting in relatively low amplification capability and limited electrochemical performances. On the other hand, the reported OECTs that demonstrate high transconductance often suffer from poor stretchability and low fatigue resistance against stretching. It is a great challenge to achieve OECTs combining both high stretchability and excellent electrical properties such as high transconductance. Many efforts have been made to enhance the stretchability of OECTs 6 – 17 . For example, the reported OECT with a folded PDMS substrate can withstand 10% tensile strain and 400 stretch-release cycles 7 . The reported OECT based on a three-dimensional poly(3-hexylthiophene) (P3HT)/styrene-ethylene-butylene-styrene (SEBS) blend porous film exhibits a stretchability of 30% and a peak transconductance ( g m,max ) below 6 mS 8 . Facchetti et al. reported stretchable (30‒140% strain) OECTs with a g m,max in the range of 1.5‒3 mS based on a semiconducting polymer film with a honeycomb porous microstructure achieved by a pre-stretching method combined with a breath figure tachnique 10 . Besides, stretchable (∼30%) OECTs with a g m,max of 22.5 mS based on poly(3,4‑ethylenedioxythiophene): poly(styrene sulfonate) (PEDOT:PSS) can be achieved by an inkjet printing method, but the cyclic stability was not mentioned 11 . While there are many reports on stretchable OECTs, they usually sacrifice the electrical performances such as transconductance. The transconductance of the previously reported state-of-the-art stretchable OECTs is usually lower than 5‒20 mS. Therefore, the development of stretchable OECTs with superior electrical properties is urgently needed and represents one of the cutting-edge research areas of OECTs. In recent years, OECTs based on ionic gel electrolyte are emerging as a promising research direction in the field of transistors due to their superior ionic conductivity, good biocompatibility, and high flexibility of ionic gels 18 – 26 . The high stretchability of ionic gels can meet the mechanical demands of stretchable OECTs and the three-dimensional network structure of these gels can accommodate a large amount of water and solvent, which can lead to high ion transport efficiency, thus improving the electrochemical performance of OECTs 20 . Furthermore, the use of ionic gel electrolyte not only avoids the leakage issues associated with traditional liquid electrolytes but also improves the stability and usability of the devices 25 . Besides, the interface layers based on ionic gels composed of poly(vinyliflon-hexafluoropropylene) and ionic liquid can improve the ion transport properties of hydrophobic conjugated polymers, which effectively enhances the electrochemical performance of OECTs 27 . The electrochemical performance of OECTs can be enhanced by constructing semiconducting polymer gel active layers with both high ion and charge transport properties. There is one report on rigid OECTs with PEDOT:PSS-based gels as active layers and aqueous solution containing ions as electrolyte. The resulting OECTs exhibit a g m,max of ∼40 mS and an on/off ratio of 10 3 28 . Besides, Lei et al. developed single- and multi-network n-type semiconductor polymer hydrogels with good electron mobility and OECTs based on the hydrogels with a g m,max of ∼3 mS 23 . But the reported OECTs based on gel active layers in the above two works are not stretchable. Recently, Wang et al. reported stretchable semiconducting hydrogel-based OECTs with high conformability and immune compatibility 29 . The obtained OECTs exhibit a charge carrier mobility ( μ ) of 1.4 cm −2 V −1 s −1 and a g m,max of ∼4 mS. Inspired by these works, the electrochemical performance of stretchable OECTs may be further enhanced by constructing all-gel OECTs with flexible semiconducting polymer gel and ionic gel as active layer and electrolyte, respectively. However, stretchable all-gel OECTs with gels as both the electrolyte and active layer are rarely reported. OECTs with good amplification capability show promising applications in pressure sensing 24 , 30 – 32 and gas sensing 33 – 35 . For example, the OECTs with a microstructured gate can regulate the ion transport by changing the contact area between gate and active layer, which can afford pressure sensors with a detection limit of ∼10 Pa 31 . Similarly, OECT-based pressure sensors with a detection limit of 1.1 Pa can be obtained by using a pyramid-microstructured gel as the electrolyte layer and a P3HT film as the active layer 32 . The detection limit of the OECT-based pressure sensors may be further decreased by enhancing the transconductance of OECTs and optimizing the structures of the devices. Additionally, OECTs with high transconductance are suitable for highly sensitive gas sensing applications. OECTs with ionic liquids as electrolytes can detect gases such as NH 3 , NO 2 , H 2 S, and SO 2 because ionic liquid can adsorb these gases and ion transport will be changed after gas adsorption 33 – 37 . Stretchable gas sensors show promising applications in wearable electronics and flexible bionic olfactory system. However, highly sensitive stretchable OECT-based gas sensors are rarely reported. Herein, we report highly stretchable OECTs—all-gel OECTs in which both the active layer and electrolyte are gels. A double-network semiconducting polymer gel consisting of PEDOT:PSS and polyacrylamide (PAM) and a poly(ionic liquid) (PIL) ionogel are used as the active layer and electrolyte of the OECTs, respectively. The flexible network structures of the semiconducting polymer gel active layers and ionogel electrolytes allow the resulting OECTs to reach a good stretchability up to 50%. More importantly, the all-gel network structures effectively promote ion penetration and transport, which endows the OECTs with a high transconductance of 86.4 mS, on/off ratio of 1.2 × 10 5 , μC s ( C s is areal capacitance) of 7118.6 μF V −1 s −1 , and μ of 5.7 cm −2 V −1 s −1 . Benefiting from their unique structures and excellent mechanical/electrical properties, the all-gel OECTs can be used as pressure-sensitive stretchable electronic skins, stretchable artificial synapses, and highly sensitive stretchable olfactory-inspired gas sensors for NH 3 detection. This work provides a versatile strategy toward high-performance stretchable all-gel OECTs promising for next-generation high-performance flexible electronics.",
"discussion": "Discussion In summary, we have developed high-performance and stretchable all-gel OECTs utilizing PIL ionogels as electrolytes and semiconducting polymer gels as active layers. The all-gel structure of electrolyte and active layer of OECTs effectively promotes ion penetration and transport, enabling efficient doping of the semiconductor, while the flexible gel electrolyte and active layer endows the all-gel OECT with good stretchability. The all-gel OECTs combine high transconductance of 86.4 mS, on/off ratio of 1.2 × 10 5 , μC s of 7118.6 μF V −1 s −1 , μ of 5.7 cm −2 V −1 s −1 , stretchability up to 50%, and excellent stretching stability (10,000 cycles under 30% strain). The resulting all-gel OECT pressure-sensitive electronic skins combine a good stretchability and a low detection limit of 0.1 Pa. Furthermore, the all-gel OECTs can be applied as stretchable artificial synapses capable of emulating biological synaptic behaviors such as EPSC, PPF, and learning-forgetting-relearning process. Additionally, the all-gel OECTs can simulate the perception process of human olfactory system, making them high-performance gas sensors combining low detection limit, high selectivity, good stretching stability, and the capability of food quality monitoring. This work provides an idea for the design of high-performance stretchable transistors promising for flexible electronics, sensors, logic circuits, healthcare, human-machine interfaces, etc."
} | 2,536 |
39298470 | PMC11459196 | pmc | 7,974 | {
"abstract": "Significance The establishment of naturalized plant species often leads to invasions that change ecosystem functioning and associated ecosystem services. The traits of native and naturalized species differ, but how these differences shift the functional composition of whole plant communities remains unknown. Our research shows that across deserts, grasslands, and forests, plant communities with higher abundance of naturalized species are more acquisitive above and belowground, shorter, more shallowly rooted, and less dependent on mycorrhizal symbionts for resource acquisition. These functional shifts likely drive observed changes in carbon storage, litter decomposition, and nutrient and water cycling in invaded ecosystems. This mechanistic understanding of functional community change is a crucial step toward predicting and mitigating impacts of naturalized and invasive species.",
"discussion": "Discussion Our results demonstrate that differences between traits of native and naturalized species lead to predictable and consistent functional trait shifts in plant communities. Changes in environmental conditions caused by disturbances likely set the stage for altered community composition ( 52 ), but two key findings indicate that naturalized species drive changes in community-level functional traits: 1) substantial differences in abundance-weighted trait means of co-occurring naturalized and native species, and 2) unique changes in abundance-weighted trait means of native and naturalized species along gradients of naturalized species abundance. These shifts may underpin many of the impacts naturalized species have on ecosystem functions ( 3 ). Like patterns found at individual sites ( 40 – 44 ), we show that the establishment and proliferation of naturalized species with acquisitive leaf and root traits results in communities that align with the fast end of the aboveground and belowground conservation gradient. These results fit with previous observations of acquisitive traits among naturalized species ( 9 – 11 , 20 ) and may arise from biases in the types of species that people tend to introduce, such as those associated with agriculture or other human-modified landscapes ( 53 ). The replacement of native species that have long-lived conservative leaves by naturalized species that have short-lived acquisitive leaves likely contributes to observed trends of increased nutrient cycling, litter decomposition, and aboveground net primary production in communities dominated by naturalized plant species ( 29 , 33 , 34 , 39 ). Although research on root traits is more scarce, similar dynamics may occur belowground: naturalized plants with high root nitrogen concentrations and low root tissue densities likely promote rapid nutrient cycling and decomposition ( 54 ). Our work supports emerging evidence that naturalized species, and the communities they dominate, are positioned toward the do-it-yourself end of the root collaboration gradient. Trait values associated with this independent strategy of resource acquisition have been documented for naturalized species established in desert ( 19 , 22 ), grassland ( 23 , 55 , 56 ), and forest ( 16 , 17 , 57 ) communities across the continental United States. Where community-level functional composition has been assessed, researchers have found increased CWM SRL and reduced CWM root diameter with increasing abundance of naturalized species ( 58 ). Further, experimental and synthesis research has found that invasive species often benefit less from mycorrhizal symbionts than natives ( 57 , 59 ). Our findings provide additional support for the notion that differences in root collaboration traits between native and naturalized species scale to the community level and may result in plant assemblages with reduced dependency on mycorrhizal symbionts. These community-level shifts may help explain the myriad effects naturalized species have on the diversity of soil biota ( 60 ) and associated soil processes ( 54 ). Higher proportions of herbaceous species likely contribute to the trend of highly invaded communities being shorter and more shallowly rooted than communities with low abundance of naturalized species (see additional discussion below). Short, shallowly rooted invasive species commonly invade deserts and shrubland ecosystems ( 26 , 27 , 57 ) and understory grasses, forbs, and shrubs that are shorter and more shallowly rooted than dominant trees, are common invaders in eastern forests ( 28 ). Given that nutrient, water, and carbon cycling are influenced by numerous belowground traits ( 54 ), shifts in belowground functional composition may lead to myriad widespread impacts. For instance, increased establishment of short, shallow-rooted naturalized species likely contributes to reductions in carbon storage documented in invaded systems throughout western US rangelands ( 35 ) and eastern US forests ( 61 ) and may drive observed trends of increased water utilization in upper soil layers ( 36 ) and reduced nutrient utilization in deep soil layers ( 32 ) in grassland systems. Trait differences between native and naturalized species likely account for much of the shift in community-level traits, but environmental gradients that select for specific functional strategies in both native and naturalized species pools likely also contribute to community-level changes in functional composition. For example, acquisitive leaf traits underpin abundance increases of both native and naturalized species responding to nutrient enrichment ( 49 ). Our results suggest that naturalized species can both drive and respond to ecological change ( 47 ). Naturalized species appear to drive shifts in community-level leaf traits including SLA, LDMC, leaf N, and leaf P ( SI Appendix , Fig. S3 A– D ) but underlying environmental gradients appear to select for similar root N and RTD in both native and naturalized species ( SI Appendix , Fig. S3 E– F ). As in other continental scale analyses ( 50 , 62 ), we could not fully account for the role of local disturbances and environmental conditions. Data on human disturbance ( 63 ) and changing environmental conditions ( 64 ) at local scales are essential to more accurately quantify the interactive effects of invasive species and environmental conditions on changing plant communities and ecosystems. Trends in functional composition also appear to be influenced by two mechanisms: 1) a general shift toward herbaceous-dominated communities ( 52 ) and 2) trait shifts within woody and herbaceous functional groups. Many of the naturalized species documented in SPCIS are herbaceous ( 46 ). For example, nonnative annual grasses are ubiquitous invaders in arid and semiarid shrublands of the western US ( 26 ). Other common herbaceous species in the database include Alliaria petiolata (i.e., garlic mustard) and Microstegium vimineum (stiltgrass), which are highly abundant in forests in the eastern US ( 65 ). These and other key invaders drive general shifts toward herbaceous-dominated communities, which contributes to changes in functional community composition and shifts the structure of the ecosystems toward lower-stature vegetation. However, trait shifts within woody and herbaceous functional groups also contributed to functional composition at the community level. For example, increases in root N were observed within both herbaceous and woody species, suggesting that community-level shifts can be attributed primarily to trait shifts within functional groups. Our results indicate that both mechanisms drive community-level functional change but their importance varies by trait. Clarifying the consequences of species introductions on community-level traits is a critical first step toward linking changes in community composition to changes in ecosystem function. Our results demonstrate that shifts in functional composition stemming from trait differences between native and naturalized species are widespread and consistent across several ecosystem types. As a consequence of these trait shifts, changes to ecosystem functions are likely also widespread. Being able to identify such shifts in community composition may provide new and efficient means for anticipating whether invasion by specific species will lead to associated changes in ecosystem function. Detecting these shifts in community functional composition from local to regional scales would advance our ability to conserve and manage communities and ecosystem functions threatened by naturalized and invasive species."
} | 2,139 |
37714853 | PMC10504285 | pmc | 7,975 | {
"abstract": "Multisensory integration is a salient feature of the brain which enables better and faster responses in comparison to unisensory integration, especially when the unisensory cues are weak. Specialized neurons that receive convergent input from two or more sensory modalities are responsible for such multisensory integration. Solid-state devices that can emulate the response of these multisensory neurons can advance neuromorphic computing and bridge the gap between artificial and natural intelligence. Here, we introduce an artificial visuotactile neuron based on the integration of a photosensitive monolayer MoS 2 memtransistor and a triboelectric tactile sensor which minutely captures the three essential features of multisensory integration, namely, super-additive response, inverse effectiveness effect, and temporal congruency. We have also realized a circuit which can encode visuotactile information into digital spiking events, with probability of spiking determined by the strength of the visual and tactile cues. We believe that our comprehensive demonstration of bio-inspired and multisensory visuotactile neuron and spike encoding circuitry will advance the field of neuromorphic computing, which has thus far primarily focused on unisensory intelligence and information processing.",
"introduction": "Introduction Relying on visual senses for navigation in complete darkness is not useful; instead, tactile senses can be more effective. While a hard touch can reveal more information about an object or an obstacle owing to large neural responses, a soft touch may be inadequate in evoking neural feedback. However, hard touches can also lead to undesired consequences such as damage to an object, e.g., during locomotion inside a dark room with valuable artwork or injury to the body due to the presence of dangerous entities. In such situations, even a short-lived flash of light can significantly enhance the chance of successful locomotion. This is because visual memory can subsequently influence and aid the tactile responses for navigation. This would not be possible if our visual and tactile cortex were to respond to their respective unimodal cues alone. Integration of cross-modal cues is, therefore, one of the essential features of how the brain functions. In the brain, each sense functions optimally under different circumstances, but collectively they can enhance the likelihood of detecting and identifying objects and events. Commonly, it is believed that there are dedicated areas in the brain, such as the visual, auditory, somatosensory, gustatory, and olfactory cortices, that process sensory input from one modality, whereas cross-modal integrations occur in higher cortical areas. However, recent findings show that multisensory integration can take place in primary sensory areas via specialized neurons that receive convergent inputs from two or more sensory modalities 1 . For example, S1 neurons found in the primary somatosensory cortex of trained monkeys respond to visual and auditory stimuli in addition to somatosensory inputs 2 , 3 . Similarly, A1 neurons in the auditory cortex respond to both auditory and somatosensory cues 4 . The advantage of multisensory integration is that the multisensory response is super-additive, i.e., it not only exceeds the individual unisensory responses but also their arithmetic sum. Another key feature of multisensory integration is that multisensory enhancement is typically inversely related to the strength of the individual cues that are being combined 5 . This is referred to as the inverse effectiveness effect and makes intuitive sense, as highly salient unimodal stimuli will evoke vigorous responses in corresponding unisensory neurons, which can be easily detected. In contrast, weak cues are comparatively difficult to detect via unisensory neurons; in such cases, multisensory integration can substantially enhance neural activity and positively impact animal behavior by increasing the speed and likelihood of detecting and locating an event 6 – 8 . In other words, multisensory amplification is the greatest when responses evoked by individual stimuli are the weakest. Finally, multisensory integration demonstrates temporal congruency, i.e., the magnitude of the integrated response is sensitive to the temporal correlation between the responses that are initiated by each sensory input 9 . In other words, the response is maximal when the peak periods of activity coincide. Examples of multisensory information processing are abundant in nature. Dolphins, for instance, combine auditory cues derived from echoes with their visual system, enabling them to develop a comprehensive understanding of objects, distances, and shapes present in their environment. Honeybees communicate the whereabouts of food sources to their hive mates through intricate dances called “waggle dances.” These dances incorporate visual cues, such as the angle and duration of the waggle, along with odor cues obtained from the nectar, effectively conveying information about the food source’s distance and direction. Electric fish integrate sensory inputs from their electric sense, vision, and mechanosensation to form a comprehensive perception of their surroundings. While multisensory integration has been widely studied in neuroscience, particularly in the context of cognition and behavior, its benefits are yet to be fully utilized in the fields of robotics, artificial intelligence, and neuromorphic computing. Note that there are some recent demonstrations of neuromorphic devices that can respond to more than one external stimulus. For example, Liu et al. 10 demonstrated a stretchable and photoresponsive nanowire transistor that can perceive both tactile and visual information, You et al. 11 demonstrated visuotactile integration using piezoresistors and MoS 2 field-effect transistors (FETs), Jiang et al. 12 used commercial sensors and spike encoding circuits to encode bimodal motion signals such as acceleration and angular speed and subsequently integrated the two using a dual gated MoS 2 FET, Wang et al. 13 demonstrated gesture recognition by integrating visual data with somatosensory data from stretchable sensors, Yu et al. 14 realized a mechano-optic artificial synapse based on a graphene/MoS 2 heterostructure and an integrated triboelectric nanogenerator, and Sun et al. 15 reported an artificial reflex arc that senses/processes visual and tactile information using a self-powered optoelectronic perovskite (PSK) artificial synapse and controls artificial muscular actions in response to environmental stimuli. The visual and somatosensory information was also encoded as impulse spikes. Similarly, Chen et al. 16 reported a CsPbBr 3 /TiO 2 -based floating-gate transistor that can respond to both light and temperature, and Han et al. 17 proposed a fingerprint recognition system based on a single-transistor neuron (1T-neuron) that can integrate visual and thermal stimuli. Finally, Yuan et al. 18 demonstrated VO 2 -based artificial neurons that can encode illuminance, temperature, pressure, and curvature signals into spikes, and Liu et al. 19 reported an artificial autonomic nervous system to emulate the joint action of sympathetic and parasympathetic nerves on organs to control the contraction and relaxation of artificial pupils and visually simulate normal and abnormal heart rates. However, none of the neuromorphic devices mentioned above embrace the true characteristic features of multisensory integration, i.e., super-additivity, inverse effectiveness effect, and temporal congruency. Furthermore, except for the study by Sun et al. 15 , none of the above studies demonstrated spike encoding of multisensory information. Here, we introduce a neuromorphic visuotactile device by integrating a triboelectric tactile sensor with a photosensitive monolayer MoS 2 memtransistor that can mimic the characteristic features and functionalities of a multisensory neuron (MN). A benchmarking table highlighting the advances made in this work over previous demonstrations on multisensory integration is shown in Supplementary Information 1 . Bio-inspired MN Figure 1a shows a schematic representation of the multisensory integration of visual and tactile information within the biological nervous system, and Fig. 1b shows our bio-inspired visuotactile MN comprising a tactile sensor connected to the gate terminal of a monolayer MoS 2 photo-memtransistor as well as the associated spike encoding circuit. The tactile sensor exploits the triboelectric effect to encode touch stimuli into electrical impulses, which are subsequently transcribed into source-to-drain output current spikes ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${I}_{{{{{{\\rm{DS}}}}}}}$$\\end{document} I DS ) in the MoS 2 photo-memtransistor. Similarly, visual stimuli are encoded into threshold voltage shifts by exploiting the photogating effect in monolayer MoS 2 photo-memtransistors. The encoding circuit is also built using MoS 2 memtransistors. The entire experimental setup is shown in Supplementary Fig. 1 . Figure 1c summarizes the three characteristic features of multisensory integration, i.e., super-additive response to cross-modal cues, inverse effective effect, and temporal congruency. In other words, our artificial visuotactile neuron and encoding circuitry can mimic the essential attributes of multisensory integration. Fig. 1 Multisensory integration. a Schematic representation of multisensory integration of visual and tactile information within the biological nervous system. b A bio-inspired visuotactile multisensory neuron (MN) comprising a triboelectric tactile sensor connected to the gate terminal of a monolayer MoS 2 photo-memtransistor along with the associated spike encoding circuit. Electrical impulses generated by the tactile sensor are transcribed into source-to-drain output current spikes ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${I}_{{{{{{\\rm{DS}}}}}}}$$\\end{document} I DS ) by the MoS 2 photo-memtransistor. Similarly, visual stimuli are encoded into threshold voltage shifts by exploiting the photogating effect in monolayer MoS 2 photo-memtransistors. The encoding circuit is also built using MoS 2 memtransistors to convert analog \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${I}_{{{{{{\\rm{DS}}}}}}}$$\\end{document} I DS spikes to digital voltage spikes ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{{{{{\\rm{out}}}}}}}$$\\end{document} V out ). c The three characteristic features of multisensory integration, i.e., super-additive response to cross-modal cues, inverse effective effect, and temporal congruency, are demonstrated by our bio-inspired visuotactile MN. In this study, we have used monolayer MoS 2 grown via metal–organic chemical vapor deposition (MOCVD) on an epitaxial sapphire substrate at 1000 °C. Details on material synthesis, film transfer, and device fabrication can be found in the Methods section and in our previous works 20 – 27 . Preliminary material characterization, which includes Raman and photoluminescence spectroscopic analysis of monolayer MoS 2 , and electrical characterization, which includes the transfer and output characteristics of MoS 2 photo-memtransistors measured in the dark, are shown in Supplementary Fig. 2a–d , respectively. All MoS 2 devices used in this study have channel length \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${L}_{{{{{{\\rm{CH}}}}}}}$$\\end{document} L CH = 1 µm and channel width \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${W}_{{{{{{\\rm{CH}}}}}}}$$\\end{document} W CH = 5 µm, as shown using the plan-view optical micrograph in Supplementary Fig. 3 . For the cross-sectional transmission electron microscopy (TEM) image and energy-dispersive X-ray spectroscopy (EDS) demonstrating the elemental distribution, please refer to our recent study in which we employed an identical device stack 28 . Note that while any photo-memtransistor can be used for this demonstration, the use of monolayer MoS 2 is motivated by the fact that beyond visual 27 and tactile 29 sensations, MoS 2 -based transistors can be used as chemical sensors 30 , gas sensors 31 , temperature sensors 32 , and acoustic sensors 33 , which greatly expands the scope for multisensory integration to gustatory, olfactory, thermal, and auditory sensations as well. In addition, MoS 2 -based devices have enabled various neuromorphic and bio-inspired applications through the integration of sensing, computing, and storage capabilities 34 – 40 . Finally, MoS 2 is among the most mature two-dimensional (2D) materials and can be grown at the wafer scale using chemical vapor deposition techniques 20 ; at the same time, aggressively-scaled MoS 2 -based transistors 41 with near Ohmic contacts 42 have achieved high performance that meets the IRDS standards for advanced technology nodes 43 – 45 . Unisensory tactile and visual response of the MN First, we study the response of our MN to tactile and visual stimuli alone. The tactile response is obtained using the triboelectric effect, where electrical impulses are generated due to charge transfer when two dissimilar materials come into contact. For our demonstration, the tactile sensor is composed of a stack of commercially available Kapton and aluminum foil separated by an air gap. PDMS stamps with different surface areas were prepared to serve as the touch stimuli (see Supplementary Fig. 4 ). Note that the magnitude of the electrical impulse generated by our triboelectric tactile sensor is directly proportional to the surface charge, which is strongly dependent on the surface contact area ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$T$$\\end{document} T ). Since the tactile sensor is connected to the gate terminal of the MoS 2 photo-memtransistor, the electrical impulse ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{{{{{\\rm{spike}}}}}}}$$\\end{document} V spike ) generated by the touch gets encoded as source-to-drain current ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${I}_{{{{{{\\rm{DS}}}}}}}$$\\end{document} I DS ) spikes at the output of the MN. Figure 2a shows the MN response ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${I}_{{{{{{\\rm{DS}}}}}}}$$\\end{document} I DS spikes) for different tactile stimuli under dark conditions ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${I}_{{{{{{\\rm{LED}}}}}}}$$\\end{document} I LED = 0 A) with the touch inputs having contact surface areas of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$T1$$\\end{document} T 1 = 25 mm 2 , \n \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$T2$$\\end{document} T 2 = 49 mm 2 , \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$T3$$\\end{document} T 3 = 100 mm 2 , and \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$T4$$\\end{document} T 4 = 400 mm 2 , respectively. A source-to-drain supply voltage ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{{{{{\\rm{DS}}}}}}}$$\\end{document} V DS ) of 1 V was applied across the MoS 2 photo-memtransistor. Supplementary Fig. 5a shows the histogram of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${I}_{{{{{{\\rm{DS}}}}}}}$$\\end{document} I DS spikes. All touch inputs given to the tactile sensor started from a height of approximately 1 cm with a tapping frequency of ~1 Hz. For any given touch stimulus, there is some inherent variation in the magnitude of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${I}_{{{{{{\\rm{DS}}}}}}}$$\\end{document} I DS spikes, which is typical of the triboelectric effect. However, with the increasing size of the touch input, the magnitude of the \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${I}_{{{{{{\\rm{DS}}}}}}}$$\\end{document} I DS spikes also increases, indicating that more tactile information is required to navigate in the dark. Furthermore, the additive tactile response of the MN was investigated by applying two identical touch inputs simultaneously, with each input having the same area as mentioned above. For better visualization of the tactile simulation, we have included Supplementary Videos 1 and 2 for single and dual touches, respectively. Figure 2b shows the corresponding \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${I}_{{{{{{\\rm{DS}}}}}}}$$\\end{document} I DS spikes and Supplementary Fig. 5b shows the corresponding \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${I}_{{{{{{\\rm{DS}}}}}}}$$\\end{document} I DS histograms. Figure 2c and Supplementary Fig. 5c show the bar plot of median spike magnitude ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${I}_{{{{{{\\rm{DS}}}}}}-{{{{{\\rm{m}}}}}}}$$\\end{document} I DS − m ) as a function of the strength of the tactile stimuli, i.e., touch contact area for both single and dual touches, in linear and logarithmic scale, respectively. Clearly, the MN’s response is enhanced for double touch inputs ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${TT}$$\\end{document} T T ) compared to a single touch ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$T$$\\end{document} T ) input, demonstrating the unisensory integration capability of our MN. Figure 2d shows the unisensory integration factor for tactile stimuli ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{{\\rm{UIF}}}}}}}_{{{{{{\\rm{T}}}}}}}$$\\end{document} UIF T ), which we define as the ratio of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${I}_{{{{\\rm{DS}}}}-{{{\\rm{m}}}}}$$\\end{document} I DS − m for dual to single-touch responses from the MN, as a function of strength of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$T$$\\end{document} T . Note that the unisensory integration is super-additive for the weakest tactile stimulus ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$T1$$\\end{document} T 1 ), as \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{UIF}}}}}}$$\\end{document} UIF ≈ 4.5, whereas, for the strongest tactile stimulus ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$T4$$\\end{document} T 4 ), the unisensory integration becomes near additive with \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{UIF}}}}}}$$\\end{document} UIF ≈ 2. Fig. 2 Tactile and visual response of MN and unisensory integration. a Tactile response from MN, i.e., source-to-drain current ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${I}_{{{{{{\\rm{DS}}}}}}}$$\\end{document} I DS ) spikes obtained from the MoS 2 photo-memtransistor under dark condition with increasing strength of touch size, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$T1$$\\end{document} T 1 = 25 mm 2 , \n \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$T2$$\\end{document} T 2 = 49 mm 2 , \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$T3$$\\end{document} T 3 = 100 mm 2 , and \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$T4$$\\end{document} T 4 = 400 mm 2 , respectively. b Response of the MN to two identical touch inputs applied simultaneously, with each input having the same area as mentioned above. c Bar plot of extracted median values for the source-to-drain current spike magnitude ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${I}_{{{{{{\\rm{DS}}}}}}-{{{{{\\rm{m}}}}}}}$$\\end{document} I DS − m ) as a function of the strength of the touch size for both single ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$T$$\\end{document} T ) and dual ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${TT}$$\\end{document} T T ) touches in linear scale. d Unisensory integration factor for tactile stimuli ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{{\\rm{UIF}}}_{T}}}}}$$\\end{document} UIF T ), defined as the ratio of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${I}_{{{{\\rm{DS}}}}-{{{\\rm{m}}}}}$$\\end{document} I DS − m for dual to single touch responses from the MN, as a function of strength of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$T$$\\end{document} T . e Pre-illumination dark current and post-illumination persistent photocurrent response of the MN to different visual stimuli ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$V$$\\end{document} V ) obtained from a light emitting diode (LED) with constant input current ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${I}_{{{{{{\\rm{LED}}}}}}}$$\\end{document} I LED = 100 mA) and varying illumination time ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${t}_{{{{{{\\rm{LED}}}}}}}$$\\end{document} t LED ); \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$V1$$\\end{document} V 1 = 1 s, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$V2$$\\end{document} V 2 = 2 s, and \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$V3$$\\end{document} V 3 = 10 s, respectively. Both single ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$V$$\\end{document} V ) and dual ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${VV}$$\\end{document} V V ) illuminations were used to study the unisensory visual integration of the MN. f Bar plot of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${I}_{{{{{{\\rm{DS}}}}}}-{{{{{\\rm{m}}}}}}}$$\\end{document} I DS − m as a function of the strength of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$V$$\\end{document} V for both single and dual illuminations. To explain the unisensory integration response of our bio-inspired MN to tactile stimuli, we have developed an empirical model. We have used the virtual source (VS) model to describe the MoS 2 photo-memtransistor using parameters extracted from the experimental transfer characteristics 34 , 39 , 46 . In the VS model, both the subthreshold and the above threshold characteristics of the MoS 2 photo-memtransistor are captured through a single semi-empirical relationship described in Eq. 1 . 1 \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${I}_{{{{{{\\rm{DS}}}}}}}=\\frac{{V}_{{{{{{\\rm{DS}}}}}}}}{{R}_{{{{{{\\rm{CH}}}}}}}}{{{{{\\rm{;}}}}}}\\,{R}_{{{{{{\\rm{CH}}}}}}}=\\frac{{L}_{{{{{{\\rm{CH}}}}}}}}{{W}_{{{{{{\\rm{CH}}}}}}}{\\mu }_{{{{{{\\rm{N}}}}}}}{Q}_{{{{{{\\rm{CH}}}}}}}}{{{{{\\rm{;}}}}}}\\,{Q}_{{{{{{\\rm{CH}}}}}}}={C}_{{{{{{\\rm{G}}}}}}}m\\frac{{k}_{{{{{{\\rm{B}}}}}}}T_{{{\\rm{a}}}}}{q}{{\\log }}\\left[1+{{\\exp }}\\left(\\frac{{V}_{{{{{{\\rm{spike}}}}}}}-{V}_{{{{{{\\rm{TH}}}}}}}}{m{k}_{{{{{{\\rm{B}}}}}}}{T}_{{{{{{\\rm{a}}}}}}}/q}\\right)\\right]$$\\end{document} I DS = V DS R CH ; R CH = L CH W CH μ N Q CH ; Q CH = C G m k B T a q log 1 + exp V spike − V TH m k B T a / q In Eq. 1 , \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${R}_{{{{{{\\rm{CH}}}}}}}$$\\end{document} R CH is the channel resistance, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${L}_{{{{{{\\rm{CH}}}}}}}=1$$\\end{document} L CH = 1 \n \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{\\mu }}}}}}{{{{{\\rm{m}}}}}}$$\\end{document} μ m is the channel length, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${W}_{{{{{{\\rm{CH}}}}}}}=5$$\\end{document} W CH = 5 \n \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{\\mu }}}}}}{{{{{\\rm{m}}}}}}$$\\end{document} μ m is the channel width, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${\\mu }_{{{{{{\\rm{N}}}}}}}$$\\end{document} μ N is the carrier mobility for electrons in MoS 2 , \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${Q}_{{{{{{\\rm{CH}}}}}}}$$\\end{document} Q CH is the inversion charge, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$C_{{{\\mathrm{G}}}}$$\\end{document} C G is the gate capacitance, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{{{{{\\rm{spike}}}}}}}$$\\end{document} V spike is the applied gate voltage, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{{{{{\\rm{TH}}}}}}}$$\\end{document} V TH is the threshold voltage, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$m$$\\end{document} m is the band movement factor, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${k}_{{{{{{\\rm{B}}}}}}}$$\\end{document} k B is the Boltzmann constant, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${T}_{{{{{{\\rm{a}}}}}}}$$\\end{document} T a is the temperature, and \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$q$$\\end{document} q is the electron charge. Note that \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${\\mu }_{{{{{{\\rm{N}}}}}}}$$\\end{document} μ N can be extracted from the peak transconductance and was found to be ~8 cm 2 /V-s. Similarly, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{{{{{\\rm{TH}}}}}}}$$\\end{document} V TH was extracted using the iso-current method at 10 nA and was found to be ~0.85 V, and \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$m$$\\end{document} m was extracted from the subthreshold slope ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{SS}}}}}}=m{k}_{{{{{{\\rm{B}}}}}}}T_{{{\\rm{a}}}}{{{{\\mathrm{ln}}}}}10$$\\end{document} SS = m k B T a ln 10 ) and was found to be ~4.5. Supplementary Fig. 6 shows the VS model fitting of the experimental transfer characteristics. Using this VS model, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${I}_{{{{{{\\rm{DS}}}}}}}$$\\end{document} I DS spikes were mapped to their corresponding \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{{{{{\\rm{spike}}}}}}}$$\\end{document} V spike values. Supplementary Fig. 7a, b shows the histograms for \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{{{{{\\rm{spike}}}}}}}$$\\end{document} V spike corresponding to single and dual touches of various strengths. The distributions for \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{{{{{\\rm{spike}}}}}}}$$\\end{document} V spike were described using Gaussian functions with \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${\\mu }_{{{{{{\\rm{T}}}}}}}$$\\end{document} μ T and \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${\\sigma }_{{{{{{\\rm{T}}}}}}}$$\\end{document} σ T as the mean and standard deviation, respectively. Note that the strength of the tactile stimulus ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$T$$\\end{document} T ) is captured through \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${\\mu }_{{{{{{\\rm{T}}}}}}}$$\\end{document} μ T , whereas the uncertainty associated with any triboelectric response is captured through \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${\\sigma }_{{{{{{\\rm{T}}}}}}}$$\\end{document} σ T . Supplementary Fig. 7c, d , respectively, show \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${\\mu }_{{{{{{\\rm{T}}}}}}}$$\\end{document} μ T and \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${\\sigma }_{{{{{{\\rm{T}}}}}}}/{\\mu }_{{{{{{\\rm{T}}}}}}}$$\\end{document} σ T / μ T as a function of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$T$$\\end{document} T . As expected, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${\\mu }_{{{{{{\\rm{T}}}}}}}$$\\end{document} μ T increases with increasing strength of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$T$$\\end{document} T , which we model using the empirical relationship described in Eq. 2a with \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${\\mu }_{01}$$\\end{document} μ 01 = 0.75 V, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${T}_{01}$$\\end{document} T 01 = 25 mm 2 , \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${\\mu }_{02}$$\\end{document} μ 02 = 5 V, and \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${T}_{02}$$\\end{document} T 02 = 10000 mm 2 as the fitting parameters; \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${\\sigma }_{{{{{{\\rm{T}}}}}}}/{\\mu }_{{{{{{\\rm{T}}}}}}}$$\\end{document} σ T / μ T = 0.25 was assumed constant. The tactile response was subsequently modeled using Eq. 2b . 2a \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${\\mu }_{{{{{{\\rm{T}}}}}}}={\\mu }_{01}\\left[1-{{\\exp }}\\left(-\\frac{T}{{T}_{01}}\\right)\\right]+{\\mu }_{02}\\left[1-{{\\exp }}\\left(-\\frac{T}{{T}_{02}}\\right)\\right]$$\\end{document} μ T = μ 01 1 − exp − T T 01 + μ 02 1 − exp − T T 02 2b \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{{{{{\\rm{spike}}}}}},{{{{{\\rm{T}}}}}}}={{{{{\\rm{rand}}}}}}\\left({{{{{\\rm{Gaussian}}}}}},\\,{\\mu }_{{{{{{\\rm{T}}}}}}},\\,{\\sigma }_{{{{{{\\rm{T}}}}}}}\\right)$$\\end{document} V spike , T = rand Gaussian , μ T , σ T The super-additive response of the MN to weaker tactile stimuli can be explained from the fact that when \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{{{{{\\rm{spike}}}}}}}$$\\end{document} V spike < \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{{{{{\\rm{TH}}}}}}}$$\\end{document} V TH , the magnitude of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${I}_{{{{{{\\rm{DS}}}}}}}$$\\end{document} I DS spikes increases exponentially with the strength of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$T$$\\end{document} T , whereas for \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{{{{{\\rm{spike}}}}}}}$$\\end{document} V spike > \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{{{{{\\rm{TH}}}}}}}$$\\end{document} V TH , the MN operates in the linear regime, leading to additive unisensory integration. Figure 2d shows the model-derived \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{UIF}}}}}}$$\\end{document} UIF , which captures the experimental findings. Next, we evaluate the unisensory visual response of our artificial MN. The visual response is obtained by exposing the MN to illumination pulses of constant amplitude ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${I}_{{{{{{\\rm{LED}}}}}}}$$\\end{document} I LED = 100 mA) of varying durations ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${t}_{{{{{{\\rm{LED}}}}}}}$$\\end{document} t LED ) from a light-emitting diode (LED). During illumination, photocarriers generated in the monolayer MoS 2 channel are trapped at the channel/dielectric interface, which leads to a negative shift in the transfer characteristics of the photo-memtransistor that persists even beyond the illumination. Supplementary Fig. 8a shows the transfer characteristics of the MoS 2 photo-memtransistor after being exposed to different visual stimuli ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$V$$\\end{document} V ), i.e., \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$V1$$\\end{document} V 1 = 1 s, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$V2$$\\end{document} V 2 = 2 s, and \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$V3$$\\end{document} V 3 = 10 s, and Supplementary Fig. 8b shows the extracted \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{{{{{\\rm{TH}}}}}}}$$\\end{document} V TH as a function of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$V.$$\\end{document} V . The phenomenon of a persistent shift in the \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{{{{{\\rm{TH}}}}}}}$$\\end{document} V TH is known as the photogating effect and is exploited in many neuromorphic devices and vision sensors 22 , 27 , 47 – 52 (See Supplementary Information 2 for more discussion on the photogating effect). For our demonstration, this emulates the role of visual memory in enhancing the tactile response through multisensory integration. Figure 2e shows the pre-illumination dark current and post-illumination persistent photocurrent response of the MN for different \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$V$$\\end{document} V . As expected, with increasing strength of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$V$$\\end{document} V , the MN’s response ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${I}_{{{{{{\\rm{DS}}}}}}}$$\\end{document} I DS ) increases monotonically. Figure 2e also shows the response of the MN when two identical visual stimuli ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${VV}$$\\end{document} V V ) are applied. Figure 2f shows the bar plot for \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${I}_{{{{{{\\rm{DS}}}}}}-{{{{{\\rm{m}}}}}}}$$\\end{document} I DS − m as a function of the strength of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$V$$\\end{document} V for both single and dual illuminations. Supplementary Fig. 9 shows the unisensory integration factor for visual stimuli ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{{\\rm{UIF}}}}}}}_{{{{{{\\rm{V}}}}}}}$$\\end{document} UIF V ), which we define as the ratio of the MN’s response to dual and single illuminations, as a function of strength of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$V$$\\end{document} V . \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{{\\rm{UIF}}}}}}}_{{{{{{\\rm{V}}}}}}}$$\\end{document} UIF V ranges from ~1.5 to 2, which confirms the sub-additive/additive nature of visual cues and highlights the unisensory integration capability of our artificial MN to visual stimuli. Multisensory visuotactile integration In this section, we will evaluate the response of the MN to cross-modal cues and compare the results with corresponding unisensory responses. Figure 3a shows the \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${I}_{{{{{{\\rm{DS}}}}}}}$$\\end{document} I DS spikes obtained from the MN for different combinations of tactile ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$T$$\\end{document} T ) and visual ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$V$$\\end{document} V ) cues, and Fig. 3b shows a bar plot of extracted \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${I}_{{{{{{\\rm{DS}}}}}}-{{{{{\\rm{m}}}}}}}$$\\end{document} I DS − m as a function of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$T$$\\end{document} T and \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$V$$\\end{document} V (see Supplementary Fig. 10 for the \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${I}_{{{{{{\\rm{DS}}}}}}}$$\\end{document} I DS histograms). As expected, the response of the MN monotonically increases with increasing strength of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$T$$\\end{document} T for any given \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$V$$\\end{document} V . However, the response of the MN also shows a monotonic decrease with increasing strength of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$V$$\\end{document} V . This is the so-called inverse effectiveness effect. The physical origin of this effect lies in the screening of the triboelectric gate voltage, obtained from the touch stimuli, by the trapped charges at the interface induced by the visual stimuli. With increasing strength of the visual stimuli, more photo-generated carriers become trapped at the interface, leading to more screening of the triboelectric voltage. Interestingly, this effect resonates remarkably well with its biological counterpart, i.e., a clear visual memory naturally diminishes the sensitivity to tactile stimuli. Our empirical model can capture this effect by introducing a visual-memory-induced screening factor ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${\\alpha }_{{{{{{\\rm{V}}}}}}}$$\\end{document} α V ) for \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{{{{{\\rm{spike}}}}}}}$$\\end{document} V spike . The MN’s response to cross-modal visual and tactile stimuli can be described by Eqs. 3a – 3c . 3a \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${I}_{{{{{{\\rm{DS}}}}}}}=\\frac{{W}_{{{{{{\\rm{CH}}}}}}}}{{L}_{{{{{{\\rm{CH}}}}}}}}{C}_{{{{{{\\rm{G}}}}}}}{\\mu }_{{{{{{\\rm{N}}}}}}}m\\frac{{k}_{{{{{{\\rm{B}}}}}}}T_{{{\\rm{a}}}}}{q}{{\\log }}\\left[1+{{\\exp }}\\left(\\frac{{\\alpha }_{{{{{{\\rm{V}}}}}}}{V}_{{{{{{\\rm{spike}}}}}},{{{{{\\rm{T}}}}}}}-{V}_{{{{{{\\rm{TH}}}}}},{{{{{\\rm{V}}}}}}}}{m{k}_{{{{{{\\rm{B}}}}}}}{T}_{{{{{{\\rm{a}}}}}}}/q}\\right)\\right]$$\\end{document} I DS = W CH L CH C G μ N m k B T a q log 1 + exp α V V spike , T − V TH , V m k B T a / q 3b \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{{{{{\\rm{TH}}}}}},{{{{{\\rm{V}}}}}}}={V}_{{{{{{\\rm{TH}}}}}},0}+{V}_{0}{{\\exp }}\\left(-\\frac{{t}_{{{{{{\\rm{LED}}}}}}}}{{\\tau }_{{{{{{\\rm{LED}}}}}}}}\\right){{{{{\\rm{;}}}}}}\\,{V}_{{{{{{\\rm{TH}}}}}},0}=0.2V,\\,{V}_{0}=0.66V,\\,{\\tau }_{{{{{{\\rm{LED}}}}}}}=4.1{{{{{\\rm{s}}}}}}$$\\end{document} V TH , V = V TH , 0 + V 0 exp − t LED τ LED ; V TH , 0 = 0.2 V , V 0 = 0.66 V , τ LED = 4.1 s 3c \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${\\alpha }_{{{{{{\\rm{V}}}}}}}={\\left(\\frac{{t}_{{{{{{\\rm{LED}}}}}}}}{{\\tau }_{0}}\\right)}^{\\gamma }{{{{{\\rm{;}}}}}}\\,\\gamma=-\\!0.6,\\,{\\tau }_{0}=2.2{{{{{\\rm{s}}}}}}$$\\end{document} α V = t LED τ 0 γ ; γ = − 0.6 , τ 0 = 2.2 s Fig. 3 Observation of inverse effectiveness effect and super-additive response from the visuotactile integration by the MN. a \n \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${I}_{{{{{{\\rm{DS}}}}}}}$$\\end{document} I DS spikes obtained from the MN for different combinations of tactile ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$T$$\\end{document} T ) and visual ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$V$$\\end{document} V ) cues. b Bar plots of extracted \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${I}_{{{{{{\\rm{DS}}}}}}-{{{{{\\rm{m}}}}}}}$$\\end{document} I DS − m as a function of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$T$$\\end{document} T and \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$V$$\\end{document} V . While the response of the MN increases with increasing strength of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$T$$\\end{document} T for any given \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$V$$\\end{document} V , as expected, a monotonic decrease in the response of MN with increasing strength of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$V$$\\end{document} V confirms the inverse effectiveness effect. c Results obtained for \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${I}_{{{{{{\\rm{DS}}}}}}-{{{{{\\rm{m}}}}}}}$$\\end{document} I DS − m from an empirical model developed for describing the response of the MN to tactile and visual stimuli also exhibit the inverse effectiveness effect. d Comparison of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${I}_{{{{{{\\rm{DS}}}}}}-{{{{{\\rm{m}}}}}}}$$\\end{document} I DS − m obtained from the MN in the presence of multimodal ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${VT}$$\\end{document} V T ) and corresponding unimodal cues. Each graph represents the results corresponding to different \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$V$$\\end{document} V and each group of bars within a graph represents results for different \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$T$$\\end{document} T ; each bar within a group represents \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${I}_{{{{{{\\rm{DS}}}}}}-{{{{{\\rm{m}}}}}}}$$\\end{document} I DS − m for \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$V$$\\end{document} V , \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$T$$\\end{document} T , and \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${VT}$$\\end{document} V T from left to right. e Multisensory integration factor ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{MIF}}}}}}$$\\end{document} MIF ), defined as the ratio of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${I}_{{{{{{\\rm{DS}}}}}}-{{{{{\\rm{m}}}}}}}$$\\end{document} I DS − m for the multisensory response to the sum of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${I}_{{{{{{\\rm{DS}}}}}}-{{{{{\\rm{m}}}}}}}$$\\end{document} I DS − m for the individual unisensory responses, as a function of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$T$$\\end{document} T and \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$V$$\\end{document} V . Note that \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{MIF}}}}}}$$\\end{document} MIF >> 1 for all combinations of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$T$$\\end{document} T and \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$V$$\\end{document} V , confirming the super-additive nature of multisensory integration by our artificial MN. Supplementary Fig. 11a, b , respectively, show the experimentally obtained and model-fitted \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{{{{{\\rm{TH}}}}}},{{{{{\\rm{V}}}}}}}$$\\end{document} V TH , V and \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${\\alpha }_{{{{{{\\rm{V}}}}}}}$$\\end{document} α V . Figure 3c shows the bar plot for \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${I}_{{{{{{\\rm{DS}}}}}}-{{{{{\\rm{m}}}}}}}$$\\end{document} I DS − m obtained using the empirical model as a function of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$T$$\\end{document} T and \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$V$$\\end{document} V , which clearly exhibits the inverse effectiveness effect. Next, we assess the benefits of multisensory integration over unisensory responses. Figure 3d shows the comparison of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${I}_{{{{{{\\rm{DS}}}}}}-{{{{{\\rm{m}}}}}}}$$\\end{document} I DS − m obtained from the MN in the presence of multimodal ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${VT}$$\\end{document} V T ) and corresponding unimodal cues. Clearly, the multisensory response exceeds the unisensory responses, as well as their sums, irrespective of the strength of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$T$$\\end{document} T and \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$V$$\\end{document} V . In order to evaluate the effectiveness of multisensory integration, we define a quantity called the multisensory integration factor ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{MIF}}}}}}$$\\end{document} MIF ) as the ratio of the multisensory response ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${I}_{{{{{{\\rm{DS}}}}}}-{{{{{\\rm{m}}}}}},{{{{{\\rm{VT}}}}}}}$$\\end{document} I DS − m , VT ) to the sum of individual unisensory responses given by Eq. 4 . 4 \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{MIF}}}}}}=\\frac{{I}_{{{{{{\\rm{DS}}}}}}-{{{{{\\rm{m}}}}}},{{{{{\\rm{VT}}}}}}}}{{I}_{{{{{{\\rm{DS}}}}}}-{{{{{\\rm{m}}}}}},{{{{{\\rm{V}}}}}}}+{I}_{{{{{{\\rm{DS}}}}}}-{{{{{\\rm{m}}}}}},{{{{{\\rm{T}}}}}}}}$$\\end{document} MIF = I DS − m , VT I DS − m , V + I DS − m , T Figure 3e shows the \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{MIF}}}}}}$$\\end{document} MIF as a function of various \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$T$$\\end{document} T and \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$V$$\\end{document} V stimuli. Note that \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{MIF}}}}}}$$\\end{document} MIF » 1 for all combinations of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$T$$\\end{document} T and \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$V$$\\end{document} V , thus confirming the super-additive nature of multisensory integration by our artificial MN. We found that the \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{MIF}}}}}}$$\\end{document} MIF can be as high as ~26.4 when the illumination period is the shortest ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$V1$$\\end{document} V 1 = 1 s), and the touch area is the smallest ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$T1$$\\end{document} T 1 = 25 mm 2 ); conversely, the \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{MIF}}}}}}$$\\end{document} MIF reduces to ~3.3 when the illumination period is the longest ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$V3$$\\end{document} V 3 = 10 s), and the touch area is the largest ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$T4$$\\end{document} T 4 = 400 mm 2 ). In other words, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{MIF}}}}}}$$\\end{document} MIF decreases monotonically as the strength of the individual cues increases. This makes intuitive sense; offering the greatest multisensory enhancement for the weakest cues can be critical for the survival of the species, whereas diminishing the response when the cues are stronger ensures that the nervous system is not overwhelmed with the multisensory response, highlighting the importance of the inverse effectiveness effect. The inverse effectiveness effect found in the experimentally extracted \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{MIF}}}}}}$$\\end{document} MIF can also be obtained from the empirical model, as shown in Supplementary Fig. 12 . Next, we compare the effectiveness of multisensory integration against unisensory integration. Figure 4a shows the comparison of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${I}_{{{{{{\\rm{DS}}}}}}-{{{{{\\rm{m}}}}}}}$$\\end{document} I DS − m obtained through multisensory integration ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${VT}$$\\end{document} V T ) against unisensory integration for different strengths of visual ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${VV}$$\\end{document} V V ) and tactile ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${TT}$$\\end{document} T T ) cues. Clearly, the response due to multisensory integration exceeds the response obtained through unisensory integration, irrespective of the strengths of the individual sensory cues. To assess the effectiveness of multisensory integration over unisensory integration, we define \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{{\\rm{MIF}}}}}}}_{{{{{{\\rm{T}}}}}}}$$\\end{document} MIF T and \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{{\\rm{MIF}}}}}}}_{{{{{{\\rm{V}}}}}}}$$\\end{document} MIF V as the ratio of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${I}_{{{{{{\\rm{DS}}}}}}-{{{{{\\rm{m}}}}}}}$$\\end{document} I DS − m obtained through multisensory integration ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${VT}$$\\end{document} V T ) to unisensory integration, i.e., \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${TT}$$\\end{document} T T and \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${VV}$$\\end{document} V V , respectively. Figure 4b, c shows the \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{{\\rm{MIF}}}}}}}_{{{{{{\\rm{T}}}}}}}$$\\end{document} MIF T and \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{{\\rm{MIF}}}}}}}_{{{{{{\\rm{V}}}}}}}$$\\end{document} MIF V , respectively. Interestingly, both \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{{\\rm{MIF}}}}}}}_{{{{{{\\rm{T}}}}}}}$$\\end{document} MIF T and \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{{\\rm{MIF}}}}}}}_{{{{{{\\rm{V}}}}}}}$$\\end{document} MIF V exceed 1 irrespective of the strength of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$T$$\\end{document} T and \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$V$$\\end{document} V stimuli, which reinforces the super-additive nature of multisensory integration. For example, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{{\\rm{MIF}}}}}}}_{{{{{{\\rm{T}}}}}}}$$\\end{document} MIF T can be as high as ~7 when \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$T$$\\end{document} T and \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$V$$\\end{document} V cues are both weak and decrease to ~1.1 with a stronger tactile input. In other words, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{{\\rm{MIF}}}}}}}_{{{{{{\\rm{T}}}}}}}$$\\end{document} MIF T also demonstrates the inverse effectiveness effect with the strength of the tactile stimulus. Similarly, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{{\\rm{MIF}}}}}}}_{{{{{{\\rm{V}}}}}}}$$\\end{document} MIF V can be as high as ~66 when both \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$T$$\\end{document} T and \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$V$$\\end{document} V cues are weak. While \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{{\\rm{MIF}}}}}}}_{{{{{{\\rm{V}}}}}}}$$\\end{document} MIF V , as expected, increases with increasing strength of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$T$$\\end{document} T for any given \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$V$$\\end{document} V , it shows a monotonic decrease with increasing strength of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$V$$\\end{document} V for all \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$T$$\\end{document} T . In other words, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{{\\rm{MIF}}}}}}}_{{{{{{\\rm{V}}}}}}}$$\\end{document} MIF V demonstrates the inverse effectiveness effect with the strength of the visual stimulus. Fig. 4 Comparison between unisensory and multisensory integration and demonstration of temporal congruency. a Bar plots of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${I}_{{{{{{\\rm{DS}}}}}}-{{{{{\\rm{m}}}}}}}$$\\end{document} I DS − m obtained through multisensory integration ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${VT}$$\\end{document} V T ) and unisensory integration for different strengths of visual ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${VV}$$\\end{document} V V ) and tactile ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${TT}$$\\end{document} T T ) cues. Each graph represents the results corresponding to different \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$V$$\\end{document} V and each group of bars within a graph represents results for different T ; from left to right, each bar within a group represents \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${I}_{{{{{{\\rm{DS}}}}}}-{{{{{\\rm{m}}}}}}}$$\\end{document} I DS − m for \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${VV}$$\\end{document} V V , \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${TT}$$\\end{document} T T , and \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${VT}$$\\end{document} V T . Multisensory integration factor for b tactile integration ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{{\\rm{MIF}}}}}}}_{{{{{{\\rm{T}}}}}}}$$\\end{document} MIF T ) and c visual integration ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{{\\rm{MIF}}}}}}}_{{{{{{\\rm{V}}}}}}}$$\\end{document} MIF V ). Both \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{{\\rm{MIF}}}}}}}_{{{{{{\\rm{T}}}}}}}$$\\end{document} MIF T and \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{{\\rm{MIF}}}}}}}_{{{{{{\\rm{V}}}}}}}$$\\end{document} MIF V exceed 1 irrespective of the strength of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$T$$\\end{document} T and \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$V$$\\end{document} V stimuli, confirming the advantage of multisensory integration over unisensory integration. d Bar plot of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${I}_{{{{{{\\rm{DS}}}}}}-{{{{{\\rm{m}}}}}}}$$\\end{document} I DS − m as a function of temporal lag ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$\\triangle \\tau$$\\end{document} △ τ ) between \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$V$$\\end{document} V and \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$T$$\\end{document} T for different \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$T$$\\end{document} T . A monotonic decrease can be observed for \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${I}_{{{{{{\\rm{DS}}}}}}-{{{{{\\rm{m}}}}}}}$$\\end{document} I DS − m with increasing \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$\\triangle \\tau$$\\end{document} △ τ , confirming that our artificial MN exhibits temporal congruency. e Long-term temporal response of the MN after exposure to visual stimuli. The monotonic decay in persistent photocurrent can be attributed to the gradual detrapping of trapped photocarriers at the dielectric/MoS 2 interface. Next, we investigate the temporal congruency offered by our MN. As mentioned earlier, biological MNs show the highest response when cross-modal cues appear simultaneously, while the response falls off monotonically with increasing lag between the cues. Supplementary Fig. 13 shows the response of the MN to touch stimuli as a function of lag ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$\\triangle \\tau$$\\end{document} △ τ ) between the tactile ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$T$$\\end{document} T ) and visual ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$V$$\\end{document} V ) stimuli, and Fig. 4d shows the corresponding bar plots of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${I}_{{{{{{\\rm{DS}}}}}}-{{{{{\\rm{m}}}}}}}$$\\end{document} I DS − m as a function of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$\\triangle \\tau$$\\end{document} △ τ for different \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$T$$\\end{document} T . A monotonic decrease can be observed for \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${I}_{{{{{{\\rm{DS}}}}}}-{{{{{\\rm{m}}}}}}}$$\\end{document} I DS − m with increasing \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$\\triangle \\tau$$\\end{document} △ τ , confirming that our artificial MN exhibits temporal congruency. The physical origin of temporal congruency can be attributed to the fact that the persistent photocurrent in MoS 2 photo-memtransistors is a direct consequence of photocarrier trapping at the MoS 2 /dielectric interface; with time, the detrapping process gradually resets the device back to its pre-illumination conductance state. Figure 4e shows the long-term temporal response of the MN after exposure to visual stimuli. Clearly, the post-illumination \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${I}_{{{{{{\\rm{DS}}}}}}-{{{{{\\rm{m}}}}}}}$$\\end{document} I DS − m decreases monotonically. This can be regarded as a gradual loss of visual memory. Naturally, tactile cues that appear long after the visual cues are expected to evoke significantly reduced responses. The detrapping process leading to a monotonic decrease in \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${I}_{{{{{{\\rm{DS}}}}}}}$$\\end{document} I DS or, equivalently, a monotonic increase in \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{{{{{\\rm{TH}}}}}}}$$\\end{document} V TH can be described using an exponential decay function with a time constant, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${\\tau }_{{{{{{\\rm{detrap}}}}}}}$$\\end{document} τ detrap = 260 s, given by Eq. 5 . 5 \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${I}_{{{{{{\\rm{DS}}}}}}}\\left(t\\right)={I}_{{{{{{\\rm{DS}}}}}},0}{{\\exp }}\\left(-\\frac{t}{{\\tau }_{{{{{{\\rm{detrap}}}}}}}}\\right){{{{{\\rm{;}}}}}}\\,{I}_{{{{{{\\rm{DS}}}}}},0}=28\\,{{{{{\\rm{nA}}}}}}$$\\end{document} I DS t = I DS , 0 exp − t τ detrap ; I DS , 0 = 28 nA By combining Eqs. 2 , 3 , and Eq. 5 , the phenomenon of temporal congruency can be captured using an empirical model. Note that it is critical to strike a balance between the visual and tactile response for proper functioning of the MoS 2 photo-memtransistor-based MN. This can be accomplished by ensuring that \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{{{{{\\rm{TH}}}}}}-{{{{{\\rm{V}}}}}}}$$\\end{document} V TH − V and \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{{{{{\\rm{spike}}}}}},{{{{{\\rm{T}}}}}}}$$\\end{document} V spike , T are of similar magnitudes. To do so, first, the expected strength of the visual ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${I}_{{{{{{\\rm{LED}}}}}}},{t}_{{{{{{\\rm{LED}}}}}}},{\\lambda }_{{{{{{\\rm{LED}}}}}}}$$\\end{document} I LED , t LED , λ LED ) and tactile ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$T$$\\end{document} T ) stimuli must be determined based on the application requirements and the operating environment. Next, Eq. 1 through Eq. 5 can be self-consistently and iteratively solved to arrive at the required device design dimensions. This is shown schematically in Supplementary Fig. 14 . At the same time, it is also important to understand how various device-related parameters influence multisensory integration. Supplementary Fig. 15a–c , respectively, show the dependence of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{MIF}}}}}}$$\\end{document} MIF on \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{SS}}}}}}$$\\end{document} SS , \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${\\mu }_{{{{{{\\rm{N}}}}}}}$$\\end{document} μ N , and \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{{{{{\\rm{TH}}}}}}}$$\\end{document} V TH for the weakest tactile and visual stimuli. Clearly, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${\\mu }_{{{{{{\\rm{N}}}}}}}$$\\end{document} μ N has the least influence on \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{MIF}}}}}}$$\\end{document} MIF since it is related to the ON-state performance of the photo-memtransistor, whereas visuotactile responses are generated in the OFF-state of the photo-memtransistor. Therefore, as expected, both \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{{{{{\\rm{TH}}}}}}}$$\\end{document} V TH and \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{SS}}}}}}$$\\end{document} SS have a significant impact on \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{MIF}}}}}}$$\\end{document} MIF , with a more positive \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{{{{{\\rm{TH}}}}}}}$$\\end{document} V TH and lower magnitude of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{SS}}}}}}$$\\end{document} SS leading to improved \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{MIF}}}}}}$$\\end{document} MIF . Note that the dependence of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{MIF}}}}}}$$\\end{document} MIF on various device related parameters can become a critical design consideration when an ensemble of multisensory neurons is present. Supplementary Fig. 15d shows the transfer characteristics of 100 multisensory neurons and Supplementary Fig. 15e-g , respectively, show the corresponding neuron-to-neuron variation in \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{SS}}}}}}$$\\end{document} SS , \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{{{{{\\rm{TH}}}}}}}$$\\end{document} V TH , and \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${\\mu }_{{{{{{\\rm{N}}}}}}}$$\\end{document} μ N . The mean values for \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{SS}}}}}}$$\\end{document} SS , \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{{{{{\\rm{TH}}}}}}}$$\\end{document} V TH , and \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${\\mu }_{{{{{{\\rm{N}}}}}}}$$\\end{document} μ N were found to be 255 mV/decade, 0.42 V, and 10.37 cm 2 V −1 s −1 , respectively, with corresponding standard deviation values of 26 mV/decade, 0.15 V, and 6.3 cm 2 V −1 s −1 , respectively. Supplementary Fig. 15h shows the projected neuron-to-neuron variation in \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{MIF}}}}}}$$\\end{document} MIF based on the model discussed earlier. Note that the inherent variation in the tactile response was already built into the model as described in Eq. 2. The mean value for \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{MIF}}}}}}$$\\end{document} MIF was found to be ~18.4 with a standard deviation of ~7.3. It is possible to minimize the neuron-to-neuron variation and improve the device performance through further optimization of synthesis, transfer, and cleanliness of the processes associated with device fabrication (see Supplementary Information 3 for more discussion). Finally, the impact of temperature on the performance of multisensory neurons is discussed in Supplementary Fig. 16 . Spike encoding of visuotactile cross-modal cues The above demonstrations establish the fact that our bio-inspired visuotactile neuron exhibits all characteristic features of multisensory integration. However, unlike the brain, where information is encoded into digital spike trains, the response from our MN is analog. To convert the analog current responses into digital spikes, we use a circuit comprising four monolayer MoS 2 photo-memtransistors, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{MT}}}}}}1$$\\end{document} MT 1 , \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{MT}}}}}}2$$\\end{document} MT 2 , \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{MT}}}}}}3$$\\end{document} MT 3 , and \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{MT}}}}}}4$$\\end{document} MT 4 , as shown in Fig. 5a (optical image shown in Fig. 1b ). Since the gate and drain terminals are shorted for both \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{MT}}}}}}1$$\\end{document} MT 1 and \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{MT}}}}}}3$$\\end{document} MT 3 , these photo-memtransistors are always in saturation and act as depletion loads. Figure 5b shows the voltage, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{{{{{\\rm{N}}}}}}2}$$\\end{document} V N 2 , measured at node \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{N}}}}}}2$$\\end{document} N 2 as a function of the input voltage, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{{{{{\\rm{N}}}}}}3}$$\\end{document} V N 3 , applied to node \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{N}}}}}}3$$\\end{document} N 3 , i.e., the gate terminal of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{MT}}}}}}2$$\\end{document} MT 2 , under dark condition and after exposure to different visual stimuli ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$V$$\\end{document} V ). For \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{{{{{\\rm{N}}}}}}3}$$\\end{document} V N 3 = −2 V, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{MT}}}}}}2$$\\end{document} MT 2 is in the OFF-state (open circuit), pulling up \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{{{{{\\rm{N}}}}}}2}$$\\end{document} V N 2 to \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{{{{{\\rm{DD}}}}}}}$$\\end{document} V DD = 2 V, which is applied to the source terminal of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{MT}}}}}}1$$\\end{document} MT 1 , i.e., node \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{N}}}}}}1$$\\end{document} N 1 . Similarly, for \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{{{{{\\rm{N}}}}}}3}$$\\end{document} V N 3 = 2 V, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{MT}}}}}}2$$\\end{document} MT 2 is in the ON-state (short circuit), pulling \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{{{{{\\rm{N}}}}}}2}$$\\end{document} V N 2 down to \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{{{{{\\rm{GND}}}}}}}$$\\end{document} V GND = 0 V, which is applied to the drain terminal of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{MT}}}}}}2$$\\end{document} MT 2 , i.e., node \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{N}}}}}}4$$\\end{document} N 4 . This explains why \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{{{{{\\rm{N}}}}}}2}$$\\end{document} V N 2 switches from 2 V to 0 V as \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{{{{{\\rm{N}}}}}}3}$$\\end{document} V N 3 is swept from −2 V to 2 V. In other words, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{MT}}}}}}1$$\\end{document} MT 1 and \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{MT}}}}}}2$$\\end{document} MT 2 operate as a depletion mode inverter. In the transfer curve shown in Fig. 5b , the value of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{{\\rm{V}}}}}}}_{{{{{{\\rm{N}}}}}}3}$$\\end{document} V N 3 at which \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{{{{{\\rm{N}}}}}}2}$$\\end{document} V N 2 = \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{{\\rm{V}}}}}}}_{{{{{{\\rm{DD}}}}}}}/2$$\\end{document} V DD / 2 is defined as the inversion threshold ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{{\\rm{V}}}}}}}_{{{{{{\\rm{Inv}}}}}}}$$\\end{document} V Inv ). Note that \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{{\\rm{V}}}}}}}_{{{{{{\\rm{Inv}}}}}}}$$\\end{document} V Inv decreases monotonically with exposure to stronger visual stimuli. This is owing to the photogating effect, which results in a negative shift in \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{{\\rm{V}}}}}}}_{{{{{{\\rm{TH}}}}}}}$$\\end{document} V TH of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{MT}}}}}}2$$\\end{document} MT 2 . Also note that \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{MT}}}}}}3$$\\end{document} MT 3 and \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{MT}}}}}}4$$\\end{document} MT 4 share a similar configuration and hence their role is to invert \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{{{{{\\rm{N}}}}}}2}$$\\end{document} V N 2 . Figure 5c shows the voltage, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{{{{{\\rm{N}}}}}}5}$$\\end{document} V N 5 , measured at node \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{N}}}}}}5$$\\end{document} N 5 as a function of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{{{{{\\rm{N}}}}}}3}$$\\end{document} V N 3 under the same visual stimulus ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$V$$\\end{document} V ). Clearly, the circuit comprising \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{MT}}}}}}1$$\\end{document} MT 1 , \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{MT}}}}}}2$$\\end{document} MT 2 , \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{MT}}}}}}3$$\\end{document} MT 3 , and \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{MT}}}}}}4$$\\end{document} MT 4 operates as a 2-stage cascaded inverter that can convert an analog input voltage, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{{\\rm{V}}}}}}}_{{{{{{\\rm{N}}}}}}3}$$\\end{document} V N 3 , into a digital output, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{{{{{\\rm{N}}}}}}5}$$\\end{document} V N 5 ; at the same time, this circuit offers visual memory, which in turn determines the spiking threshold, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{{\\rm{V}}}}}}}_{{{{{{\\rm{TH}}}}}}-{{{{{\\rm{spike}}}}}}}$$\\end{document} V TH − spike , for the tactile stimulus ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{T}}}}}}$$\\end{document} T ). Fig. 5 Visuotactile spike encoding. a Circuit diagram of the spike encoder comprising four monolayer MoS 2 photo-memtransistors ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{MT}}}}}}1$$\\end{document} MT 1 – \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{MT}}}}}}4$$\\end{document} MT 4 ). The circuit operates as a two-stage cascaded inverter. b Voltage, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{{{{{\\rm{N}}}}}}2}$$\\end{document} V N 2 , measured at node \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{N}}}}}}2$$\\end{document} N 2 , and c voltage, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{{{{{\\rm{N}}}}}}5}$$\\end{document} V N 5 , measured at node \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{N}}}}}}5$$\\end{document} N 5 as a function of the input voltage, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{{{{{\\rm{N}}}}}}3}$$\\end{document} V N 3 , applied to node \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{N}}}}}}3$$\\end{document} N 3 under dark condition and after exposure to different visual stimuli ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$V$$\\end{document} V ). The inversion threshold ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{{{{{\\rm{Inv}}}}}}}$$\\end{document} V Inv ) of the first stage inverter and the spiking threshold ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${V}_{{{{{{\\rm{TH}}}}}}-{{{{{\\rm{spike}}}}}}}$$\\end{document} V TH − spike ) of the second stage inverter are functions of the applied visual stimulus. d Spiking output from the encoding circuit, recorded at node \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{N}}}}}}5$$\\end{document} N 5 in response to different tactile ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$T$$\\end{document} T ) and visual ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$V$$\\end{document} V ) stimuli. e Bar plot of the spiking probability ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${P}_{{{{{{\\rm{spike}}}}}}}$$\\end{document} P spike ) for all combinations of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$T$$\\end{document} T and \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$V$$\\end{document} V . f Corresponding bar plots of the probability enhancement factor ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{PEF}}}}}}$$\\end{document} PEF ) obtained from the ratio of post-illumination \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${P}_{{{{{{\\rm{spike}}}}}}}$$\\end{document} P spike to pre-illumination \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${P}_{{{{{{\\rm{spike}}}}}}}$$\\end{document} P spike . g Comparison of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${P}_{{{{{{\\rm{spike}}}}}}}$$\\end{document} P spike obtained for single ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$T$$\\end{document} T ) and dual ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${TT}$$\\end{document} T T ) touches in the dark for all touch sizes. Inset shows the probability enhancement factor for unisensory tactile integration ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{{\\rm{PEF}}}}}}}_{{{{{{\\rm{T}}}}}}}$$\\end{document} PEF T ). Figure 5d shows the spiking response from the multisensory neural circuit for different combinations of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$V$$\\end{document} V and \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$T$$\\end{document} T . Clearly, the analog current response has now been converted into digital voltage spikes with the probability of spiking ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${P}_{{{{{{\\rm{spike}}}}}}}$$\\end{document} P spike ) encoding the strengths of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$V$$\\end{document} V and \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$T$$\\end{document} T . Figure 5e shows the bar plot of extracted \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${P}_{{{{{{\\rm{spike}}}}}}}$$\\end{document} P spike for different combinations of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$V$$\\end{document} V and \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$T$$\\end{document} T . As expected, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${P}_{{{{{{\\rm{spike}}}}}}}$$\\end{document} P spike shows a monotonic increase with increasing strength of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$V$$\\end{document} V and \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$T$$\\end{document} T before saturating at the maximum value of 1. Figure 5f shows the bar plot of the probability enhancement factor ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{PEF}}}}}}$$\\end{document} PEF ), defined as the ratio of post-illumination \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${P}_{{{{{{\\rm{spike}}}}}}}$$\\end{document} P spike to pre-illumination \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${P}_{{{{{{\\rm{spike}}}}}}}$$\\end{document} P spike , as a function of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$V$$\\end{document} V and \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$T$$\\end{document} T . Note that \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{PEF}}}}}}$$\\end{document} PEF » 1 for all combinations of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$V$$\\end{document} V and \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$T$$\\end{document} T , i.e., the spike encoder preserves the super-additive nature of multisensory integration. Also, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{PEF}}}}}}$$\\end{document} PEF increases with increasing strength of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$V$$\\end{document} V as \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{{\\rm{V}}}}}}}_{{{{{{\\rm{TH}}}}}}-{{{{{\\rm{spike}}}}}}}$$\\end{document} V TH − spike is reduced and can reach as high as ~24 for the weakest tactile cue, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{T}}}}}}1$$\\end{document} T 1 . Moreover, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{PEF}}}}}}$$\\end{document} PEF is the largest for the smallest \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$T$$\\end{document} T and decreases monotonically with increasing strength of \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$T$$\\end{document} T for any given \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$V$$\\end{document} V . In other words, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{\\rm{PEF}}}}}}$$\\end{document} PEF also demonstrates the inverse effectiveness effect with \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$T$$\\end{document} T . Finally, Fig. 5g shows the bar plots for \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${P}_{{{{{{\\rm{spike}}}}}}}$$\\end{document} P spike for single ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$T$$\\end{document} T ) and dual ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${TT}$$\\end{document} T T ) tactile stimuli under dark conditions; the inset shows the probability enhancement factor for unisensory tactile integration ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{{\\rm{PEF}}}}}}}_{{{{{{\\rm{T}}}}}}}$$\\end{document} PEF T ) (see Supplementary Fig. 17 for spiking response from the multisensory neural circuit for dual touches of different strengths, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${TT}$$\\end{document} T T ). Note that \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${{{{{{\\rm{PEF}}}}}}}_{{{{{{\\rm{T}}}}}}}$$\\end{document} PEF T » 1 for all \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$T$$\\end{document} T , confirming that the super-additive nature of unisensory integration is also preserved by the spike encoding circuit. Also, note that the footprint of the visuotactile circuit is primarily determined by the dimensions of the 2D photo-memtransistor. Recently, we have shown ultra-scaled 2D devices with \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${L}_{{{{{{\\rm{CH}}}}}}}$$\\end{document} L CH down to 100 nm along with scaled contacts ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${L}_{{{{{{\\rm{C}}}}}}}$$\\end{document} L C down to 20 nm) 53 . There are several other reports in the literature that confirm the aggressive scalability of 2D devices 20 , 54 , 55 . The footprint of the visuotactile circuit can therefore be made significantly smaller through device dimension scaling. However, the limiting factor is going to be the dimension of the photosensor, which will be determined by the diffraction limit of the light. For the visible spectrum, the dimensions of the photosensors have been stagnant at the micrometer scale or larger. Also note that instead of realizing the readout circuit using 2D photo-memtransistors, integration of 2D sensors with silicon CMOS is a viable alternative, but at the cost of increased processing and fabrication complexity while having no significant performance advantages. In fact, many recent reports highlight the energy and area benefits of using 2D memtransistors for in-sensor and near-sensor processing and storage since these multifunctional devices can be used as logic, memory, and sensing devices 25 – 27 , 34 , 35 , 37 , 51 , 52 , 56 – 62 . The motivation for our current work lies in the integration of these photo-memtransistors with tactile sensors to produce efficient and reliable multisensory integration. Finally, we envision that exploring arrays of multisensory neurons can enable more sophisticated visuotactile information processing.",
"discussion": "Discussion In conclusion, we have realized a visuotactile MN comprising a triboelectric tactile sensor and a monolayer MoS 2 photo-memtransistor that offers all three characteristic features of multisensory integration, i.e., super-additive response, inverse effectiveness effect, and temporal congruency. We have also developed a visuotactile spike encoder circuit that converts the analog current response from the MN into digital spikes. We believe that our demonstration of multisensory integration will advance the field of neuromorphic and bio-inspired computing, which has primarily relied on unimodal sensory information processing to date. We also believe that the impact of multisensory integration can be far-reaching with applications in defense, space exploration, and many robotic and AI systems. Finally, the principles of multisensory integration can be expanded beyond visuotactile information processing to other sensory stimuli, including audio, olfactory, thermal, and gustatory stimuli."
} | 40,218 |
36381281 | PMC9618013 | pmc | 7,976 | {
"abstract": "Background: Microalgae have the potential to generate high-value products. The design of photobioreactors (PBRs), in which microalgae are cultured, is crucial because alterations in their configuration and operational conditions can affect the biomass production and productivity. Objective: The objective of this study was to optimize the diameter of the internal tube of an airlift PBR and to characterize the growth of Spirulina maxima in an optimized design. Material and Methods: \n S. maxima was cultured in a mineral medium without an organic carbon source. The PBR consisted of an acrylic cylinder with an operational volume of 7 L. Daily determinations of biomass (by filtration), chlorophyll, N-NO 3 and P-PO 4 (spectrophotometrically) were carried out. Results: The use of a concentric tube with a diameter of 3 inches led to an increased biomass concentration of 1.14 ± 0.136 g.L -1 , allowing a global biomass productivity of 153 mg.L -1 .d -1 . The culture reached a volumetric consumption velocity of 27.34 ± 1.596 and 2.29 ± 0.353 mg.L -1 .d -1 for N and P, respectively. Conclusions: It was concluded that operational conditions must be specifically selected for each cultivated strain and that this configuration of airlift PBR can produce Spirulina biomass under laboratory conditions with a high biomass productivity.",
"conclusion": "6. Conclusion The design and construction of static mixers in an airlift PBR must be done with consideration to the engineering, biotechnological, and biological parameters to ensure optimal growth of the selected strain. For flagellated microalgae and filamentous cyanobacteria, the use of operational conditions in which the mixing time is short (ensuring good mass transfer) is not ideal. This was observed in cultures grown with an insufflation of 0.5 vvm (3.5 L.min -1 ), where S. maxima grew but did not reach higher biomass concentrations that were achieved in cultures with an insufflation of 0.3 vvm (2.1 L.min -1 ). As such, a specific analysis based on the characteristics of the strain to be cultured is necessary to find an equilibrium between the optimal growth and good mass transfer (mixing) conditions. The PBR and the configuration employed here allowed us to obtain high biomass and chlorophyll concentrations, even with moderate light (250 µE.m -2 .s -1 around PBR). However, it is necessary to study the effects of increasing light flux and modified operational conditions (e.g., nutrient concentration, amount of inoculum, etc.) to further optimize the process and achieve higher biomass concentrations.",
"discussion": "5. Discussion There are many obstacles in obtaining microalgal cultures with very high biomass concentrations. One of them is optimizing the light supply – if the biomass concentration is very high the auto-shading effect appears; nevertheless, the growth diminution by the auto-shading phenomenon is significant only when the biomass concentration is higher than 5 g.L -1 or when the light supply is lower than 500 µE.m -2 s -1 ( 10 \n, 14 \n, 25 \n). In this work, the differences observed between the different PBR configurations can be attributed only to the presence of concentric tubes and their diameters, because the biomass concentration did not exceed 2 g.L -1 , regardless of whether the illumination was high and sufficient to reach this concentration ( 10 \n). The dimensions of the brackets in the bottom permitted the liquid to enter from the downcomer to the riser, without any obstacles and significant modifications of the flow pattern. Selecting the diameter of the concentric tube is challenging because many dimensions have been published; however, some authors agree that the most important parameter is not the size, but rather that the ratio between the transversal area of both the sections, raiser and downcomer, which is the limiting factor. ( 5 \n, 11 \n, 26 ). Some authors have proposed employing airlift systems for the cultivation of Spirulina ; however, PBR configurations, experimental conditions, and DW values remain very diverse. Oncel and Sukan (2008) suggested analyzing the ratio between the transversal areas of the riser and the downcomer (A D /A R ratio) to ensure optimal distribution and mixing ( 12 \n). In their investigation, they utilized an airlift PBR with an A D /A R ratio close to 1 and obtained biomass concentrations similar to those obtained in this work. However, other authors have proposed an A D /A R ratio ranging from 1.5–3, and they assured that these values lead to optimal flow patterns and mass transfer, regardless of the cultured microalgal strain ( 11 \n, 27 \n, 28 \n). Considering the diameters of the PBR and ICT, the A D /A R ratio can be estimated. The A D /A R ratios were 8.3, 2.8, and 1.1 for tubes with 2, 3, and 4 inches of diameter, respectively. These values explain the fact that the maximum DW value was achieved in the culture with an ICT of 3 inches, given that it meets the values for the A D /A R ratio suggested in previous studies ( 27 \n). In the selected configuration, S. maxima growth was fast, reaching values higher than those reported for this strain in other PBR configurations, which was due to the good mixing achieved without shear stress in this configuration. Spirulina is a filamentous cyanobacterium, and the hydrodynamic characteristics inside the reactor are very important because an increase in shear stress can cause cellular death and reductions in biomass concentrations ( 11 \n). In addition, the chlorophyll concentration was high, causing an increase in the levels of this pigment, which may be due to both the optimal culture conditions and sufficient light supplied ( 12 \n). Despite the light used (250µE.m -2 .s -1 around the PBR), the maximum concentration of biomass was high, owing to the indoor culturing conditions. The rapid consumption of phosphorus and nitrogen was due to the fact that carbon is the principal macronutrient required by S. maxima for the production of biomass, as is the case for all photosynthetic microorganisms ( 1 \n). The consumption of nitrogen was higher than 25 mg.L -1 .d -1 , which corresponds to the absorption of this nutrient by highly active microalgal cultures. Normally, nitrogen is eliminated from the culturing media at rates ranging from 20–35 mg.L -1 .d -1 ( 14 \n, 23 \n, 29 ). These values indicate accelerated growth under the selected conditions and make it possible to ensure that no light limitation occurred during Spirulina growth. For phosphorus, a special phenomenon is observed; this nutrient is removed from the culturing medium very quickly, regardless of its initial concentration. This is called luxurious consumption. This type of consumption causes an almost constant consumption rate of approximately 1–3 mg.L -1 .d -1 , regardless of the strain, medium, or even the PBR configuration employed ( 23 \n, 30 \n). The phosphorus inside the cell is accumulated in polyphosphate bodies that can be observed by electronic microscopy, which can be used when the concentration in the medium is low, without any need for metabolic changes ( 17 \n, 23 )."
} | 1,782 |
39804831 | PMC11733691 | pmc | 7,977 | {
"abstract": "Biological memory networks are thought to store information by experience-dependent changes in the synaptic connectivity between assemblies of neurons. Recent models suggest that these assemblies contain both excitatory and inhibitory neurons (E/I assemblies), resulting in co-tuning and precise balance of excitation and inhibition. To understand computational consequences of E/I assemblies under biologically realistic constraints we built a spiking network model based on experimental data from telencephalic area Dp of adult zebrafish, a precisely balanced recurrent network homologous to piriform cortex. We found that E/I assemblies stabilized firing rate distributions compared to networks with excitatory assemblies and global inhibition. Unlike classical memory models, networks with E/I assemblies did not show discrete attractor dynamics. Rather, responses to learned inputs were locally constrained onto manifolds that ‘focused’ activity into neuronal subspaces. The covariance structure of these manifolds supported pattern classification when information was retrieved from selected neuronal subsets. Networks with E/I assemblies therefore transformed the geometry of neuronal coding space, resulting in continuous representations that reflected both relatedness of inputs and an individual’s experience. Such continuous representations enable fast pattern classification, can support continual learning, and may provide a basis for higher-order learning and cognitive computations.",
"introduction": "Introduction Autoassociative memory establishes internal representations of specific inputs that may serve as a basis for higher brain functions including classification and prediction. Representation learning in autoassociative memory networks is thought to involve experience-dependent synaptic plasticity and potentially other mechanisms that enhance connectivity among assemblies of excitatory neurons ( Hebb, 1949 ; Ko et al., 2011 ; Miehl et al., 2023 ; Ryan et al., 2015 ). Classical theories proposed that assemblies define discrete attractor states and map related inputs onto a common stable output pattern. Hence, neuronal assemblies are thought to encode internal representations, or memories, that classify inputs relative to previous experience via attractor dynamics ( Amit and Tsodyks, 1991 ; Goldman-Rakic, 1995 ; Hopfield, 1982 ; Kohonen, 1984 ; Lagzi and Rotter, 2015 ; Mazzucato et al., 2015 ). However, brain areas with memory functions such as the hippocampus or neocortex often exhibit dynamics that is atypical of attractor networks including irregular firing patterns, transient responses to inputs, and high trial-to-trial variability ( Iurilli and Datta, 2017 ; Renart et al., 2010 ; Shadlen and Newsome, 1994 ). Irregular, fluctuation-driven firing reminiscent of cortical activity emerges in recurrent networks when neurons receive strong excitatory (E) and inhibitory (I) synaptic input ( Brunel, 2000 ; Shadlen and Newsome, 1994 ; van Vreeswijk and Sompolinsky, 1996 ). In such ‘balanced state’ networks, enhanced connectivity among assemblies of E neurons is prone to generate runaway activity unless matched I connectivity establishes co-tuning of E and I inputs in individual neurons. The resulting state of ‘precise’ synaptic balance stabilizes firing rates because inhomogeneities in excitation across the population or temporal variations in excitation are tracked by correlated inhibition ( Hennequin et al., 2017 ; Hennequin et al., 2014 ; Lagzi and Fairhall, 2024 ; Rost et al., 2018 ; Vogels et al., 2011 ). E/I co-tuning has been observed experimentally in cortical brain areas ( Bhatia et al., 2019 ; Froemke et al., 2007 ; Okun and Lampl, 2008 ; Rupprecht and Friedrich, 2018 ; Wehr and Zador, 2003 ) and emerged in simulations that included spike-timing-dependent plasticity at I synapses ( Lagzi et al., 2021 ; Litwin-Kumar and Doiron, 2014 ; Vogels et al., 2011 ; Zenke et al., 2015 ). In simulations, E/I co-tuning can be established by including I neurons in assemblies, resulting in ‘E/I assemblies’ where I neurons track activity of E neurons ( Barron et al., 2017 ; Eckmann et al., 2024 ; Lagzi and Fairhall, 2024 ; Mackwood et al., 2021 ). Exploring the structural basis of E/I co-tuning in biological networks is challenging because it requires the dense reconstruction of large neuronal circuits at synaptic resolution ( Friedrich and Wanner, 2021 ). Modeling studies started to investigate effects of E/I assemblies on network dynamics ( Chenkov et al., 2017 ; Mackwood et al., 2021 ; Sadeh and Clopath, 2020a ; Schulz et al., 2021 ) but the impact on neuronal computations in the brain remains unclear. Balanced state networks can exhibit a broad range of dynamical behaviors, including chaotic firing, transient responses, and stable states ( Festa et al., 2014 ; Hennequin et al., 2014 ; Litwin-Kumar and Doiron, 2012 ; Murphy and Miller, 2009 ; Roudi and Latham, 2007 ), implying that computational consequences of E/I assemblies depend on network parameters. We therefore examined effects of E/I assemblies on autoassociative memory in a spiking network model that was constrained by experimental data from telencephalic area Dp of adult zebrafish, which is homologous to mammalian piriform cortex ( Mueller et al., 2011 ). Dp and piriform cortex receive direct input from mitral cells in the olfactory bulb (OB) and have been proposed to function as autoassociative memory networks ( Haberly, 2001 ; Wilson and Sullivan, 2011 ). Consistent with this hypothesis, manipulations of neuronal activity in piriform cortex affected olfactory memory ( Meissner-Bernard et al., 2019 ; Sacco and Sacchetti, 2010 ). In both brain areas, odors evoke temporally structured, spatially distributed activity patterns ( Blazing and Franks, 2020 ; Stettler and Axel, 2009 ; Yaksi et al., 2009 ) that are dominated by synaptic inputs from recurrent connections ( Franks et al., 2011 ; Rupprecht and Friedrich, 2018 ) and modified by experience ( Chapuis and Wilson, 2011 ; Frank et al., 2019 ; Jacobson et al., 2018 ; Pashkovski et al., 2020 ). Whole-cell voltage clamp recordings revealed that neurons in posterior Dp (pDp) received large E and I synaptic inputs during odor responses. These inputs were co-tuned in odor space and correlated on fast timescales, demonstrating that pDp enters a transient state of precise synaptic balance during odor stimulation ( Rupprecht and Friedrich, 2018 ). We found that network models of pDp with assemblies but without E/I co-tuning generated persistent attractor dynamics and exhibited a biologically unrealistic broadening of the firing rate distribution. Introducing E/I assemblies established E/I co-tuning, stabilized the firing rate distribution, and abolished persistent attractor states. In networks with E/I assemblies, population activity was locally constrained onto manifolds that represented learned and related inputs by ‘focusing’ activity into neuronal subspaces. The covariance structure of manifolds supported pattern classification when information was retrieved from selected neuronal subsets. Furthermore, the continuity of the olfactory coding space provided a metric representing the similarity of inputs to learned stimuli. These results show that autoassociative memory networks constrained by biological data operate in a balanced regime where information is contained in the geometry of neural manifolds. Predictions derived from these analyses may be tested experimentally by measurements of neuronal population activity in zebrafish.",
"discussion": "Discussion A precisely balanced memory network constrained by pDp Autoassociative memory networks map inputs onto output patterns representing learned information. Classical models proposed this mapping to be accomplished by discrete attractor states that are defined by assemblies of E neurons and stabilized by global homeostatic inhibition. However, as seen in Scaled I networks, global inhibition is insufficient to maintain a stable, biologically plausible firing rate distribution. This problem can be overcome by including I neurons in assemblies, which leads to more precise synaptic balance. To explore network behavior in this regime under biologically relevant conditions we built a spiking network model constrained by experimental data from pDp. The resulting Tuned networks reproduced additional experimental observations that were not used as constraints including irregular firing patterns, lower output than input correlations, and the absence of persistent activity. Hence, pDp sim recapitulated characteristic properties of a biological memory network with precise synaptic balance. Neuronal dynamics and representations in precisely balanced memory networks Simulated networks with global inhibition showed attractor dynamics and pattern completion, consistent with classical attractor memory. However, the distribution of firing rates broadened as connection density within assemblies increased, resulting in unrealistically high (low) rates inside (outside) assemblies and, consequently, in a loss of synaptic balance. Hence, global inhibition was insufficient to stabilize population activity. In networks with E/I assemblies, in contrast, firing rates remained within a realistic range and the balanced state was maintained. Such Tuned networks showed no discrete attractor states but transformed the geometry of the coding space by confining activity to continuous manifolds near representations of learned inputs. This observation is consistent with the hypothesis that E/I co-tuning decreases the probability of multistable attractor states as compared to balanced networks with global inhibition or lateral inhibition between assemblies ( Boerlin et al., 2013 ; Hennequin et al., 2018 ; Wu and Zenke, 2021 ). Geometric transformations in Tuned networks may be considered as intermediate between two extremes: (1) geometry-preserving transformations as, for example, performed by many random networks ( Babadi and Sompolinsky, 2014 ; Marr, 1969 ; Schaffer et al., 2018 ), and (2) discrete maps as, for example, generated by discrete attractor networks ( Freeman and Skarda, 1985 ; Hopfield, 1982 ; Khona and Fiete, 2022 ; Figure 7 ). We found that transformations became more discrete map-like when amplification within assemblies was increased and precision of synaptic balance was reduced. Likewise, decreasing amplification in assemblies of Scaled networks changed transformations toward the intermediate behavior, albeit with broader firing rate distributions than in Tuned networks ( Figure 4—figure supplement 3B ). These observations indicate that a modest amplification of activity within assemblies contributes to geometric modifications of activity manifolds in Tuned networks, but other factors such as structured inhibition may also contribute. Hence, further analyses are required to obtain a deeper mechanistic understanding of manifold geometry in networks with E/I assemblies. Nonetheless, our results suggest that precise synaptic balance may generally favor intermediate over discrete transformations because this regime tends to linearize the relationship between the mean input and output firing rates of neuronal populations ( Baker et al., 2020 ; Denève and Machens, 2016 ). Figure 7. Schematic of geometric transformations. ( A ) Randomly connected networks tend to preserve the geometry of coding space. Such networks can support neuronal computations, for example, by projecting activity patterns in a higher-dimensional coding space for pattern classification. ( B ) We found that balanced networks with E/I assemblies transform the geometry of representations by locally restricting activity onto manifolds. These networks stored information about learned inputs while preserving continuity of the coding space. Such a geometry may support fast classification, continual learning and cognitive computations. Note that the true manifold geometry cannot be visualized appropriately in 2D because activity was ‘focused’ in different subsets of dimensions at different locations of coding space. As a consequence, the dimensionality of activity remained substantial. ( C ) Neuronal assemblies without precise balance established discrete attractor states, as observed in memory networks that store information as discrete items. Networks establishing locally defined activity manifolds ( B ) may thus be considered as intermediates between networks generating continuous representations without memories ( A ) and classical memory networks with discrete attractor dynamics ( C ). E/I assemblies increased variability of population activity along preferred directions of state space and reduced its dimensionality in comparison to rand networks. Nonetheless, dimensionality remained high compared to Scaled networks with discrete attractor states. These observations indicate that geometric transformations in Tuned networks involved (1) a modest amplification of activity in one or a few directions aligned to the assembly, and (2) a modest reduction of activity in other directions. E/I assemblies therefore created a local curvature of coding space that ‘focused’ activity in a subset of dimensions and, thus, stored information in the geometry of coding space. As E/I assemblies were small relative to the total size of the E neuron population, stored information may be represented predominantly by small neuronal subsets. Consistent with this hypothesis, d M was increased and the classification of learned inputs by QDA was enhanced when activity was read out from subsets of assembly neurons as compared to random neuronal subsets. Moreover, signatures of pattern completion were found in the activity of assemblies but not in global pattern correlations. The retrieval of information from networks with small E/I assemblies therefore depends on the selection of informative neurons for readout. Unlike in networks with global attractor states, signatures of memory storage may thus be difficult to detect experimentally without specific knowledge of assembly memberships. Computational functions of networks with E/I assemblies In theory, precisely balanced networks with E/I assemblies may support pattern classification despite high variability of spike trains and the absence of discrete attractor states ( Denève and Machens, 2016 ). Indeed, we found in Tuned E+I networks that input patterns were classified successfully by generic classifiers, particularly relative to learned inputs. Analyses based on the Mahalanobis distance d M indicate that classification of learned inputs was enhanced by two effects: (1) local manifolds representing learned odors became more distant from representations of other odors due to a modest increase in firing rates within E/I assemblies, and (2) the concomitant increase in variability was not isotropic, remaining sufficiently low in directions that separated novel from learned patterns. Hence, information contained in the geometry of coding space can be retrieved by readout mechanisms aligned to activity manifolds. Efficient readout mechanisms may thus integrate activity primarily from assembly neurons, as mimicked in our QDA-based pattern classification. This notion is consistent with the finding that the integrated activity of E/I assemblies can be highly informative despite variable firing of individual neurons ( Boerlin et al., 2013 ; Denève et al., 2017 ; Denève and Machens, 2016 ). It will thus be interesting to explore how the readout of information from local manifolds could be further optimized. Networks with E/I assemblies may also perform computations other than pattern classification. A toy model of continual learning indicates that precise balance can stabilize firing rate distributions when new memories are added, helping to prevent catastrophic failures. Furthermore, we found that continuous manifolds in Tuned networks support metric analyses, for example, to determine the relatedness of a (potentially complex) input to previously learned stimuli. Such tasks are not supported by discrete attractor states because information about gradual differences between inputs is lost by pattern completion. E/I assemblies can thus enhance the quantitative analysis of relevant information through changes in manifold geometry. Conceivably, E/I assemblies may have further consequences for neuronal computation. Unlike discrete attractor networks, Tuned networks do not exhibit persistent activity, suggesting that they mediate fast computations rather than short-term memory functions. Fast classification may, for example, be important to interpret dynamical sensory inputs on a moment-to-moment basis. Moreover, the representation of learned inputs by small neuronal subsets, rather than global activity states, raises the possibility that multiple inputs can be classified simultaneously. Generally, the absence of discrete attractor states indicates that information is not stored in the form of distinct items. Rather, E/I assemblies cause geometric modifications of a continuous coding space that result in an overrepresentation of learned (relevant) inputs at the expense of other stimuli. These pattern transformations may thus contribute to different classification and learning processes by re-formatting and extracting relevant information for further processing by distributed networks. Such a function would be loosely related to computations within layers of artificial neuronal networks and consistent with the notion that piriform cortex is embedded in a larger network comprising multiple telencephalic brain areas ( Haberly, 2001 ). Balanced state networks with E/I assemblies as models for olfactory cortex Piriform cortex and Dp have been proposed to function as attractor-based memory networks for odors. Consistent with this hypothesis, pattern completion and its modulation by learning has been observed in piriform cortex of rodents ( Barnes et al., 2008 ; Chapuis and Wilson, 2011 ). However, odor-evoked firing patterns in piriform cortex and Dp are typically irregular, variable, transient and less reproducible than in the OB even after learning ( Jacobson et al., 2018 ; Pashkovski et al., 2020 ; Schoonover et al., 2021 ; Yaksi et al., 2009 ), indicating that activity does not converge onto stable attractor states. Balanced networks with E/I assemblies, in contrast, are generally consistent with these experimental observations. Alternative models for pattern classification in the balanced state include networks endowed with short-term plasticity, which respond to stimuli with an initial amplification phase followed by a tonic inhibition-stabilized state ( Wu and Zenke, 2021 ), or mechanisms related to ‘balanced amplification’, which typically generate pronounced activity transients ( Ahmadian and Miller, 2021 ; Murphy and Miller, 2009 ). However, it has not been explored whether these models can be adapted to reproduce characteristic features of Dp or piriform cortex. Our results generate predictions to test the hypothesis that E/I assemblies establish local manifolds in Dp: (1) odor-evoked population activity should be constrained onto manifolds, particularly in response to learned odors. (2) Learning should increase the magnitude and asymmetry of d M between odor representations. (3) Activity evoked by learned and related odors should exhibit lower dimensionality and more directional variability than activity evoked by novel odors. (4) Careful manipulations of inhibition may unmask assemblies by increasing amplification. These predictions may be addressed experimentally by large-scale measurements of odor-evoked activity after learning. The direct detection of E/I assemblies will ultimately require dense reconstructions of large neuronal networks at synaptic resolution. Given the small size of Dp, this challenge may be addressed in zebrafish by connectomics approaches based on volume electron microscopy ( Denk et al., 2012 ; Friedrich and Wanner, 2021 ; Kornfeld and Denk, 2018 ). The hypothesis that memory networks contain E/I assemblies and operate in a state of precise synaptic balance can be derived from the basic assumptions that (1) synaptic plasticity establishes assemblies and (2) that firing rate distributions remain stable as network structure is modified by experience ( Barron et al., 2017 ; Hennequin et al., 2017 ). Hence, Tuned networks based on Dp may also reproduce features of other recurrently connected brain areas such as hippocampus and neocortex, which also operate in a balanced state ( Renart et al., 2010 ; Sadeh and Clopath, 2020b ; Shadlen and Newsome, 1994 ; Znamenskiy et al., 2024 ). Future experiments may therefore explore representations of learned information by local manifolds also in cortical brain areas."
} | 5,221 |
31332325 | PMC6679743 | pmc | 7,978 | {
"abstract": "Spatial structuring is important for the maintenance of natural ecological systems 1 , 2 . Many microbial communities, including the gut microbiome, display intricate spatial organization 3 – 9 . Mapping the biogeography of bacteria can shed light on interactions that underlie community functions 10 – 12 , but existing methods cannot accommodate hundreds of species found in natural microbiomes 13 – 17 . Here we describe m et a genomic p lot-sampling by seq uencing (MaP-Seq), a culture-independent method to characterize the spatial organization of a microbiome at micron-scale resolution. Intact microbiome samples are immobilized in a gel matrix and cryo-fractured into particles. Neighboring microbial taxa in the particles are then identified by droplet-based encapsulation, barcoded 16S rRNA amplification and deep sequencing. Analysis of three regions of the mouse intestine revealed heterogeneous microbial distributions with positive and negative co-associations between specific taxa. We identified robust associations between Bacteroidales taxa in all gut compartments and showed that phylogenetically clustered local regions of bacteria were associated with a dietary perturbation. Spatial metagenomics could be used to study microbial biogeography in complex habitats."
} | 321 |
34606725 | null | s2 | 7,979 | {
"abstract": "Many common bacteria use amphiphilic "
} | 9 |
39487991 | PMC11850969 | pmc | 7,982 | {
"abstract": "Abstract \n Frankia cluster-2 strains are diazotrophs that engage in root nodule symbiosis with actinorhizal plants of the Cucurbitales and the Rosales . Previous studies have shown that an assimilated nitrogen source, presumably arginine, is exported to the host in nodules of Datisca glomerata ( Cucurbitales ), while a different metabolite is exported in the nodules of Ceanothus thyrsiflorus ( Rosales ). To investigate if an assimilated nitrogen form is commonly exported to the host by cluster-2 strains, and which metabolite would be exported in Ceanothus , we analysed gene expression levels, metabolite profiles, and enzyme activities in nodules. We conclude that the export of assimilated nitrogen in symbiosis seems to be a common feature for Frankia cluster-2 strains, but the source of nitrogen is host dependent. The export of assimilated ammonium to the host suggests that 2-oxoglutarate is drawn from the tricarboxylic acid (TCA) cycle at a high rate. This specialized metabolism obviates the need for the reductive branch of the TCA cycle. We found that several genes encoding enzymes of central carbon and nitrogen metabolism were lacking in Frankia cluster-2 genomes: the glyoxylate shunt and succinate semialdehyde dehydrogenase. This led to a linearization of the TCA cycle, and we hypothesized that this could explain the low saprotrophic potential of Frankia cluster-2.",
"conclusion": "Conclusions Based on the data presented in this study, the export of an assimilated form of N by Frankia cluster-2 strains in symbiosis is a common feature of the clade. In Cucurbitales host plants, such as D. glomerata and C. myrtifolia , this export form is arginine, while in Rosales , such as C. thyrsiflorus , it is asparagine or glutamine. The assimilation of fixed N for export during symbiosis puts a high demand on 2-OG. The TCA cycle therefore seems to work linearly from the carbon source(s) provided by the host to 2-OG. Due to this special metabolism, the need for the glyoxylate shunt is obviated, as well as the production of succinate from 2-OG. This led to gene losses which can explain the low saprotrophic potential of Frankia cluster-2 strains.",
"introduction": "Introduction Root nodules are formed as the result of a symbiotic relationship between a nitrogen- (N) fixing soil bacterium and its host plant. All root nodule-forming plants belong to a single clade ( Soltis et al. , 1995 ). While legumes ( Fabaceae , Fabales ) and Parasponia ( Cannabaceae , Rosales ) engage with rhizobia, actinorhizal symbiosis is established between a diverse group of plants within the Rosales , Fagales , and Cucurbitales , and their endosymbiont Frankia . Recent phylogenomic studies support the hypothesis that the common ancestor to all nodulating plants was symbiotic, but the symbiosis was subsequently lost in the majority of the lineages ( Griesmann et al. , 2018 ; van Velzen et al. , 2018 ). Unlike rhizobia, Frankia strains do not depend on their host plant for the oxygen protection of the nitrogenase enzyme. Therefore, the polyphyly of the oxygen protection system for nitrogenase in nodules indicates that the original symbiont would have been Frankia ( van Velzen et al. , 2019 ). The bacterial genus Frankia can be split into four phylogenetically distinct clades, which are referred to as cluster-1 to -4 ( Normand et al. , 1996 ; Nguyen et al. , 2016 ). The first three clusters represent symbiotic strains, and the phylogeny roughly coincides with host specificity. Cluster-4 encompasses strains that do not engage in symbiosis ( Normand et al. , 1996 ). With the sole exception of Frankia coriariae ( Gtari et al. , 2015 ; Nouioui et al. , 2017 ; Gueddou et al. , 2019 ), and despite numerous efforts, cluster-2 strains cannot be cultivated in vitro . This implies that most analyses, such as gene expression studies via quantitative reverse transcription–PCR (RT–qPCR), must be conducted on nodules. Studies on all Frankia strains are further complicated by the fact that stable transformation has not been reproduced since 2019 ( Gifford et al. , 2019 ; Pesce et al. , 2019 ) despite numerous attempts ( Kucho et al., 2022 ). Regardless of the challenges involved in the investigation of Frankia cluster-2, this cluster is particularly interesting. Frankia cluster-2 represents the earliest divergent symbiotic clade ( Sen et al. , 2014 ; Gtari et al. , 2015 ; Persson et al. , 2015 ; Berckx et al. , 2022 , 2024 ). Given that the original root nodule symbiont was Frankia ( van Velzen et al. , 2019 ), cluster-2 represents the closest relative to the first N-fixing endobacterium. In addition, while host plants of cluster-1 are solely found within actinorhizal Fagales and most of the host plants of Frankia cluster-3 are found within the Rosales , cluster-2 Frankia has a broad host range. They engage in symbiosis with all actinorhizal Cucurbitales , namely the families Datiscaceae and Coriariaceae . Cluster-2 also engages with actinorhizal Rosales of the family Rosaceae and with Ceanothus spp. ( Rhamnaceae ). Aside from the pure taxonomic diversity of actinorhizal host plants, actinorhizal symbiosis shows metabolic diversity, for instance in terms of the exchange of nutrients. During symbiosis, the host plant is provided with fixed N in exchange for photosynthates. In legumes and most actinorhizal plants, such as Alnus glutinosa ( Betulaceae , Fagales ), the fixed N is exported to the cytosol of infected host cells as ammonium (NH 4 + ) ( Guan et al. , 1996 ; Alloisio et al. , 2010 ; Udvardi and Poole, 2013 ; Hay et al. , 2020 ). Ammonium is assimilated in the plant cytosol, and transported to the xylem in different forms depending on the plant species. In actinorhizal nodules of A. glutinosa , it was shown that ammonium is assimilated via the plant glutamine synthetase/glutamate synthase (GS/GOGAT) cycle. The assimilation by Frankia only occurs at low levels ( Guan et al. , 1996 ; Alloisio et al. , 2010 ). In Datisca glomerata , several studies indicate that an assimilated form of N, presumably arginine, is exported from the endosymbiont to the cytosol of the infected cell and then transported to the surrounding uninfected cells ( Berry et al. , 2004 , 2011 ; Salgado et al. , 2018 ). In the cytosol of the uninfected cells, arginine is broken back down to ammonium and re-assimilated via the GS/GOGAT cycle ( Berry et al. , 2004 ). A similar accumulation of arginine could be found in nodules of Coriaria myrtifolia ( Coriariaceae , Cucurbitales ) ( Wheeler and Bond, 1970 ). In conclusion, the export of assimilated N, specifically arginine, from the bacterial symbiont to the plant seems to be a common feature for actinorhizal Cucurbitales , which are all host plants of Frankia cluster-2. However, no evidence for the export of arginine by Frankia cluster-2 has been found in nodules of Ceanothus thyrsiflorus ( Rhamnaceae , Rosales ) based on plant gene expression levels ( Salgado et al. , 2018 ). It has been previously shown that asparagine accumulates in nodules of Ceanothus velutinus ( Wheeler and Bond, 1970 ). So while arginine can be excluded as the form of exported N by Frankia cluster-2 in Ceanothus spp., the export of assimilated N cannot be excluded as such. We wanted to identify the exported N metabolite to find out whether the specialized metabolism of exporting an assimilated form of N was unique to host plants from the Cucurbitales or to Frankia cluster-2 symbioses in general. Carbon and N metabolism are directly linked by the tricarboxylic acid (TCA) cycle. This cycle can be closed via several distinct reactions. In the classic TCA cycle, 2-oxoglutarate (2-OG) is the substrate to be converted to succinyl-CoA by the activity of 2-OG dehydrogenase. Succinyl-CoA synthase then converts succinyl-CoA to succinate. An alternative TCA cycle has been described for some bacterial species ( Tian et al. , 2005 ), including rhizobia ( Green et al. , 2000 ), which is advantageous under reducing conditions as required during N fixation. Here, succinic semialdehyde (SSA) is produced from 2-OG by 2-OG decarboxylase. SSA is then converted to succinate via SSA dehydrogenase (SSA-DH). A third alternative pathway to produce succinate from 2-OG is the γ-aminobutyrate (GABA) shunt ( Xiong et al. , 2014 ), which again requires the activity of SSA-DH. Lastly, the TCA cycle can be closed by the glyoxylate shunt, where the production of 2-OG would be avoided altogether ( Zhang and Bryant, 2015 ). We wanted to identify which of the reactions take place to close the TCA cycle in Frankia cluster-2. This has not been done before and is important in light of which N metabolite is exported to the host plant. If not used for maintaining the TCA cycle, 2-OG can be pulled out and used for the assimilation of ammonium in the GS/GOGAT cycle, which leads to the production of glutamate. Glutamate can then be used as a substrate in the biosynthesis of glutamine, asparagine, or arginine. In all root nodule symbioses examined thus far, TCA cycle intermediates are supplied to the microsymbiont by the host plant ( Jeong et al. , 2004 ; Udvardi and Poole, 2013 ). The export of an assimilated form of N, such as arginine, by Frankia cluster-2 would require more carbon skeleton input from the host plant than in systems where ammonium is exported directly, such as in Frankia cluster-1 symbiosis ( Guan et al. , 1996 ). Most, but not all, of these carbon skeletons would be returned by the endosymbiont during the export of assimilated N. Thus, the export of assimilated N by the endosymbiont is not energy efficient for the symbiosis as a whole, as it requires more complex transport processes. Our study aims to compare features of N and carbon metabolism in nodules of Frankia cluster-2 host plants from two different orders: D. glomerata representing Cucurbitales , and C. thyrsiflorus representing Rosales . The former was nodulated by Candidatus Frankia californiensis Dg2, while the latter was nodulated by the closely related strain Candidatus F. californiensis Cv1 ( Nguyen et al. , 2016 , 2019 ; Normand et al. , 2017 ). As a comparative system, nodules of A. glutinosa induced by Frankia alni ACN14a ( Normand et al. , 2007 ) were included in some of the analyses. For this symbiosis, it is known that ammonium is exported from the bacterium to the host plant ( Guan et al. , 1996 ). Our study presents analyses of gene expression levels and enzyme activities related to carbon and N metabolism, as well as protein modelling, which were performed to elucidate the metabolite exchange between host and symbiont.",
"discussion": "Results and discussion Expression levels of nitrogenase genes allowed for direct comparison of three types of nodules While the analysis at the transcriptional level has its limitations due to post-transcriptional regulation, it has the advantage of separating plant from bacterial transcription in nodules. Most pathways contain a rate-limiting step catalysed by a key enzyme ( Rognstad, 1979 ). The expression level of the corresponding gene can be the main indicator of the overall activity of the pathway. Nodules of A. glutinosa induced by cluster-1 F. alni ACN14a, C. thyrsiflorus induced by cluster-2 Candidatus Frankia californiensis Cv1, and D.glomerata induced by cluster-2 Candidatus F. californiensis Dg2 show various anatomical differences ( Pawlowski and Demchenko, 2012 ). It was unclear whether the contribution of N-fixing Frankia mRNA in total nodule mRNA was similar enough to allow a direct comparison of bacterial gene expression levels. The expression levels of the structural nitrogenase genes nifDHK were compared and analysed against the bacterial housekeeping gene infC , encoding the translation initiation factor IF-3 ( Alloisio et al. , 2010 ; Nguyen et al. , 2019 ). No significant difference could be found for any of the three genes ( P >0.05; Supplementary Fig. S1 ). We concluded that a direct comparison of the different nodule types was appropriate. The expression of nifD showed the least variation between biological replicates in all treatments. It was used, together with infC , to normalize all further gene expression data. \n Frankia cluster-2 assimilates ammonium at similar levels but glutamate has different fates depending on the host plant To determine if an assimilated N source is exported by Frankia cluster-2 to its host plant, in our case C. thyrsiflorus and D. glomerata , ammonium assimilation activities were investigated by looking at the expression levels of the bacterial genes encoding enzymes of the GS/GOGAT pathway ( Fig. 1A ). For comparison, the expression levels of Frankia genes of the GS/GOGAT pathway were also examined in cluster-1 Frankia in nodules of A. glutinosa , where ammonia and ammonium are known to be exported to the host ( Guan et al. , 1996 ; Alloisio et al. , 2010 ). The gene gltB , coding for one of the subunits of glutamate synthase, was found to be significantly more highly expressed by Frankia in nodules of C. thyrsiflorus or D. glomerata , respectively, compared with nodules of A. glutinosa ( Fig. 1B ). In addition, the glutamine synthetase gene, glnII , was significantly more highly expressed in nodules of C. thyrsiflorus compared with nodules of both D. glomerata and A. glutinosa. On the other hand, the gene gdh , encoding glutamate synthase, was expressed at significantly higher levels in nodules of D. glomerata compared with C. thyrsiflorus and A. glutinosa ( Fig. 1B ). No significant differences could be observed between the expression levels of the genes glnA1 , glnA2 , and gltD across all three nodule types. Fig. 1. Relative expression levels (ΔCt value) of genes involved in the GS/GOGAT pathway. (A) Illustration of the GS/GOGAT pathway connected to the TCA cycle and the arginine biosynthesis pathway, given in grey. (B) The gene expression data. The Ct value is normalized against the gene infC , encoding the translation initiation factor IF-3 ( Alloisio et al. , 2010 ), and the nitrogenase subunit (MoFe protein) gene nifD . An asterisk indicates a significant difference ( P <0.5), based on one-way ANOVA followed by Tukey post-hoc analysis, of gene expression measured in four technical replicates of three biological replicates of nodules from Alnus glutinosa induced by Frankia alni ACN14a (red, left), Ceanothus thyrsiflorus induced by Candidatus Frankia californiensis Cv1 (yellow, centre), and Datisca glomerata induced by Candidatus Frankia californiensis Dg2 (black, right). Individual data points of the biological repeats are presented. Abbreviations: gdh , glutamate dehydrogenase; glnA1/glnA2 , glutamine synthetase I subunits; glnII , glutamine synthetase II; gltB , glutamate synthase, large subunit; gltD , glutamate synthase, small subunit. Based on these data, we propose that 2-OG is pulled out of the TCA cycle at similar levels in Frankia cluster-2 in nodules, presumably to assimilate ammonium. This does not occur in cluster-1 F. alni ACN14a in nodules of A. glutinosa , which is in line with previous studies ( Guan et al. , 1996 ; Alloisio et al. , 2010 ). However, in Frankia cluster-2, the fate of glutamate differs between nodules of C. thyrsiflorus and D. glomerata : in C. thyrsiflorus , more glutamate is converted into glutamine. In nodules of D. glomerata , on the other hand, glutamate can support the biosynthesis of arginine, which has been indicated to be the main N export product to the host plant ( Berry et al. , 2004 , 2011 ; Salgado et al. , 2018 ). We therefore looked at the expression levels of the arginine biosynthesis genes ( Fig. 2 ) in the two different nodule types induced by Frankia cluster-2. This would allow us to see if arginine is synthesized at similar levels in nodules of C. thyrsiflorus compared with D. glomerata , where it is known to be exported to the host plant. We found that the gene argE/argJ , encoding the bifunctional enzyme acetylornithine deacetylase, which catalyses the first step in the biosynthesis pathway ( Fig. 2A ), was expressed at significantly higher levels in Frankia in nodules of D. glomerata compared with those of C. thyrsiflorus ( Fig. 2B ). The genes argB and argD were expressed at significantly higher levels in nodules of C. thyrsiflorus . The remaining biosynthesis genes, namely argC , argF , argG , and argH , were expressed at similar levels in both nodule types. Xu et al. (2020) have demonstrated that the first step, catalysed by ArgE/ArgJ, is rate limiting for arginine biosynthesis. Therefore, our results indicate that more glutamate is used for the production of arginine in Frankia in nodules of D. glomerata compared with C. thryrsiflorus , despite the higher expression of argB and argD in C. thyrsiflorus. This is supported by previous work based on the expression levels of plant genes ( Salgado et al. , 2018 ). Based on the gene expression data of Frankia , we conclude that arginine cannot be the main export source in nodules of C. thyrsiflorus. Fig. 2. Relative expression levels (ΔCt value) of genes involved in the arginine biosynthesis pathway. (A) Biosynthesis pathway and connection to the TCA and GS/GOGAT cycle, illustrated in grey. (B) Gene expression levels. The Ct value is normalized against the gene infC , encoding the translation initiation factor IF-3 ( Alloisio et al. , 2010 ), and the nitrogenase subunit (MoFe protein) gene nifD . An asterisk indicates a significant difference ( P <0.5), based on Student’s t -test (Cv1 and Dg2), of gene expression of four technical repeats of three biological repeats of nodules from Ceanothus thyrsiflorus induced by Candidatus Frankia californiensis Cv1 (yellow, left), and Datisca glomerata induced by Candidatus Frankia californiensis Dg2 (black, right). Individual data points of the biological repeats are presented. Abbreviations: argB , acetylglutamate kinase; argC , N -acetyl-γ-glutamyl-phosphate reductase; argD , acetylornithine/succinyldiaminopimelate aminotransferase; argE/argJ , bifunctional gene acetylornithine deacetylase; argF , ornithine carbamoyltransferase; argG , argininosuccinate synthase; argH : argininosuccinate lyase. Metabolic profiling of Ceanothus roots and nodules, and glutamine synthetase enzyme activity The above findings raise the question: which metabolite is exported from the Frankia to the plant in nodules of C. thyrsiflorus ? To address this question, the levels of different amino acids and N metabolites were compared between nodules, inoculated roots, and uninoculated roots of C. thyrsiflorus ( Fig. 3 ). For a comparison, data from a previous study on D. glomerata and A. glutinosa were included ( Persson et al. , 2016 ). Only the top eight most abundant metabolites were included in the analysis. Fig. 3. Concentrations of nitrogen metabolites in Ceanothus thyrsiflorus compared with previously published data on Datisca glomerata and Alnus glutinosa ( Persson et al. , 2016 ). Data are presented to compare the concentration (log10 nmol g FW –1 ) per tissue: uninoculated roots (grey), inoculated roots (brown), and nodules (mustard). Significant differences, based on one-way ANOVA followed by Tukey post-hoc analysis, are indicated with compact letter display. The results show that, in general, in samples of C. thyrsiflorus , the concentrations of all N metabolites were significantly higher in nodules compared with roots ( P <0.05; Fig. 3 ). Consistent with the results on gene expression, and in contrast with the results for nodules of D. glomerata ( Berry et al. , 2004 ; Persson et al. , 2016 ) and Coriaria myrtifolia ( Coriariaceae , Cucurbitales ; Wheeler and Bond, 1970 ), arginine did not accumulate at high levels in C. thyrsiflorus . Asparagine and glutamate were the most dominant nitrogenous solutes in nodules. Glutamine could not be detected in uninoculated and inoculated roots, but was found to accumulate in nodules of C. thyrsiflorus . This is in agreement with gene expression data for glnII ( Fig. 1 ), where the gene was significantly more highly expressed in Frankia in nodules of C. thyrsiflorus compared with nodules of A. glutinosa or D. glomerata. It is important to consider that no distinction between accumulation in plant and bacterial cells can be made based on the metabolite data ( Fig. 3 ). Glutamate, for instance, is present at significantly higher levels in nodules than in uninoculated or inoculated roots, in all three nodule types. For nodules of D. glomerata and C. thyrsiflorus , this could be because glutamate is an intermediate for the production of arginine or glutamine, respectively. In nodules of A. glutinosa , it was shown that assimilation of ammonium by Frankia occurs only at low levels, but the plant assimilates ammonium via the GS/GOGAT pathway, followed by synthesis of citrulline, which is then transported via the xylem ( Guan et al. , 1996 ; Alloisio et al. , 2010 ). Given the importance of post-transcriptional regulation, we wanted to conduct an enzyme activity assay of GS to verify the results of the expression analysis and the metabolic profiling. As Frankia cluster-2 cannot be cultivated in vitro , and we were interested in activity under symbiotic conditions, enzyme activities needed to be conducted on nodule material. Unlike plants, certain prokaryotes such as rhizobia and Frankia ( Edmands et al. , 1987 ; de Bruijn et al. , 1989 ) have been shown to contain two variants of GS: a heat-stable GSI, encoded by glnA , and a heat-labile GSII, encoded by glnII ( Huss-Danell, 1997 ). Aside from their different heat sensitivity, the two variants also differ in their regulation. GlnA is controlled post-translationally by reversible adenylylation, whereas glnII is under transcriptional control ( de Bruijn et al. , 1989 ). Eukaryotes, such as plants, commonly only contain the heat-labile variant of GS (GSII). This allows for the distinction between plant and bacterial activity of GS. In nodules of A. glutinosa , we found significant activities of GS in the crude extract, but not after exposure to heat treatment at 40, 50, or 60 °C. ( Fig. 4 ). This indicates that, here, only the heat-labile GS is responsible for GS activity, most probably from plant origin. This would be in agreement with our gene expression data ( Fig. 1 ), as well as with the metabolite data ( Fig. 3 ) ( Persson et al. , 2016 ), if we assume most fixed N is exported to, and assimilated by, the plant. These results are supported by previous work, showing that GlnII could not be detected immunologically in nodules of Alnus incana , while it could be detected in N-fixing cultures of the infective Frankia ( Lundquist and Huss-Danell, 1992 ), and gene expression data on nodules of A. glutinosa ( Guan et al. , 1996 ). In nodules of D. glomerata , we found the highest activity in the crude extract. After heat exposure at 60 °C, GS activity was significantly reduced to the levels of the negative control. Activity was not statistically reduced compared with the crude extract after exposure to 40 °C or 50 °C, but there was also no significant difference from the negative control. This would indicate that both a heat-stable and a heat-labile copy of GS are active in nodules of D. glomerata. In nodules of C. thyrsiflorus , we found significant GS activity in the crude extract. However, this activity was not significantly reduced after heat treatment. This would indicate that most, if not all, nodule GS activity is performed by a heat-stable version of GS, namely by Frankia GSI. Fig 4. GS transferase activity assay in nodules of Alnus glutinosa , Datisca glomerata , and Ceanothus thyrsiflorus . Nodules of A. glutinosa were induced by Frankia alni ACN14a, of D. glomerata induced by Candidatus Frankia californiensis Dg2, and of C. thyrsiflorus by Candidatus F. californiensis Cv1. Activity measurements are based on the amount of enzyme required to produce 1 μmol of γ-glutamyl hydroxamate (nKat). The activity was measured in the negative control in crude protein extract without glutamine (red), and otherwise either using crude protein extracts (yellow), or in protein extract exposed for 10 min to 40, 50, or 60 °C (brown, light grey, and dark grey), to distinguish between Frankia GSI and plant GS/ Frankia GSII activity. The absorbance at 530 nm was corrected against the background absorbance of total denatured protein (boiled for 10 min at 95 °C). The assay was conducted on two technical replicates of two biological replicates. The individual data points are given. Different letters indicate significant differences ( P <0.05), based on one-way ANOVA followed by Tukey post-hoc analysis. Another N metabolite which accumulated in nodules of C. thyrsiflorus was asparagine. The biosynthesis of asparagine, as well as aspartate, relies on a high input of oxaloacetate, an intermediate of the TCA cycle ( Fig. 5B ). We investigated the gene expression levels of phosphoenolpyruvate carboxykinase ( pepck ), as well of the genes encoding the bidirectional malate dehydrogenase ( mdh ), and citrate synthase ( gltA1/A2 and citA/citA4 ) ( Fig. 5A ). We found that while pepck was not expressed at significantly higher levels in Frankia in nodules of either D. glomerata or C. thyrsiflorus , the genes mdh , gltA2 , and citA were expressed at significantly higher levels in Frankia in nodules of C. thyrsiflorus. Potentially, this could indicate an accumulation of oxaloacetate, which would be shuttled into asparagine biosynthesis. This would be supported by the metabolite data ( Fig. 3 ), where asparagine accumulated in nodules, and could not be detected in uninoculated roots. Fig. 5. Relative gene expression (ΔCt value) of genes encoding enzymes involved in the TCA cycle. The Ct value is normalized against the gene infC , encoding the translation initiation factor IF-3 ( Alloisio et al. , 2010 ), and the nitrogenase subunit (MoFe protein) gene nifD . An asterisk indicates a significant difference ( P <0.5), based on Student’s t -test (Cv1 and Dg2), of gene expression of four technical repeats of three biological repeats of nodules from Ceanothus thyrsiflorus induced by Candidatus Frankia californiensis Cv1 (yellow, left), and Datisca glomerata induced by Candidatus F. californiensis Dg2 (black, right). Individual data points of each biological repeat are presented. The pathway with enzymes interacting is illustrated below. Abbreviations used: pepck , phosphoenolpyruvate carboxykinase; gltA/gltA2 , citrate synthase; citA/citA4 , citrate synthase; acnA , aconitate hydratase A; icd , isocitrate dehydrogenase; gltB , glutamate synthase, large subunit; gltD , glutamate synthase, small subunit; gdh , glutamate dehydrogenase; sucC/sucD , succinate-CoA ligase subunit alpha/beta; sdhA/shdB/sdhC/sdhD , succinate dehydrogenase complex subunit A/B/C/D; fum , fumarate hydratase; mdh : malate dehydrogenase. The amino acid interconversions of glutamine–glutamate and asparagine–aspartate can occur rapidly and play important roles in central metabolism. Analysis of the full transcriptome might provide some more insights; however, enzyme activity assays would be more compelling. These would have to be conducted under symbiotic conditions and, unlike the heat-liable and heat-stable forms of GS, such assays could not distinguish between plant and bacterial activity. Altogether, our results suggest that asparagine and/or glutamine could be exported by Frankia cluster-2 in nodules of Ceanothus spp. These metabolites might also play a role as N transport forms in the xylem. Taking together our gene expression, metabolite, and enzyme activity data, our results suggest that in C. thyrsiflorus glutamine is the most likely metabolite to be exported by symbiotic Frankia to the host plant. As the Frankia strains under investigation here, Candidatus F. californiensis Cv1 and Candidatus F. californiensis Dg2, belong to the same species, this would indicate that the export of assimilated N is a common feature for cluster-2 Frankia , but the host plant determines which metabolite is exported. Gene loss of isocitrate lyase indicates lack of the glyoxylate shunt in Frankia cluster-2 The export of assimilated N, whether in the form of arginine or asparagine, would require a high input of carbon skeletons. Carboxylates provided by the host plant can be used to provide energy or carbon skeletons for ammonium assimilation, both by being shuttled into the TCA cycle. Genome analyses, facilitated through BLAST, revealed that several genes of TCA cycle-related enzymes were lacking in Frankia cluster-2 (indicated by red crosses in Fig. 5 ). Firstly, the gene icl , encoding isocitrate lyase which is responsible for catalysing the first step of the glyoxylate shunt ( Fig. 5 ), was lacking in all Frankia cluster-2 genomes available ( Supplementary Table S2 ; Supplementary Fig. S2 ). Further analysis of Frankia genomes revealed this gene to be present in most cluster-1 genomes, but lacking in most cluster-3 and cluster-4 genomes. Instead, using the sequence from F. alni ACN14a (Genbank accession CAJ61004.1) led to the identification of methylisocitrate lyase as the closest homologue. The corresponding enzyme has been shown to be unable to catalyse the same reaction ( Dunn et al. , 2009 ). For Candidatus F. meridionalis, no significant hit for the icl sequence could be found. This suggests that the production 2-OG cannot be avoided in most Frankia strains, except for most cluster-1 strains. All Frankia strains exhibit a variant TCA cycle 2-OG can have different fates as a metabolite and signalling molecule, as reviewed by Huergo and Dixon (2015) . Within the TCA cycle, 2-OG will be converted to succinate. This can happen through one of three pathways, as explained before: by 2-OG dehydrogenase, 2-OG decarboxylase, or through the GABA shunt ( Green et al. , 2000 ; Tian et al. , 2005 ; Xiong et al. , 2014 ). The latter two pathways require the activity of SSA-DH. The 2-OG dehydrogenase complex is composed of three enzymes: the E1 2-OG dehydrogenase, the E2 dihydrolipoyl succinyltransferase, and the E3 dihydrolipoyl dehydrogenase. \n Frankia cluster-2 genomes have previously been shown to contain a high abundance of transposable elements ( Nguyen et al. , 2019 ). We therefore continued to utilize BLAST to identify genes present in Frankia genomes. Using the corresponding amino acid sequence of Mycobacteriodes abscessus and Staphylococcus epidermis , the closest homologue of E1 in all Frankia genomes examined was found to be a multifunctional 2-OG decarboxylase/oxoglutarate dehydrogenase with <45% amino acid sequence identity with the query sequences (accession numbers listed in Supplementary Table S2 ). To fully understand if the Frankia gene encodes E1 2-OG dehydrogenase or 2-OG decarboxylase, homology modelling of the amino acid sequence of the Frankia proteins was performed on the SWISS Model portal ( Fig. 6 ; Supplementary Fig. S3 ). It was compared with the solved crystal structures of 2-OG decarboxylase from M. smegmatis and 2-OG dehydrogenase of S. epidermis . The best homology model for all the Frankia models was found to be 2-OG decarboxylase (global model quality estimate QMEANDisCo >0.70) while it showed much less similarity to 2-OG dehydrogenase (QMEANDisCo <0.50). The model was highly conserved for all Frankia genomes examined, which included all declared species of the different clusters ( Fig. 6 ; Supplementary Fig. S3 ). In addition, a phylogenetic tree was built based on the amino acid sequences, which separated the 2-OG dehydrogenase sequences from the 2-OG decarboxylase sequences ( Supplementary Fig. S4 ). Taken together, we conclude that all Frankia genomes available contain a 2-OG decarboxylase gene instead of a 2-OG dehydrogenase. This would imply that the variant TCA cycle is active in which succinate is synthesized from 2-OG via SSA and not succinyl-CoA ( Fig. 5 ), as has been shown for some other bacteria. Interestingly, this pathway has been suggested to be distributed widely among rhizobia as an adaptation to the microaerobic conditions for catabolizing dicarboxylic acids ( Green et al. , 2000 ). We suggest this adaptation to have evolved in Frankia as well. Fig. 6. Protein modelling of 2-oxoglutarate decarboxylase. (A) The solved crystal structure of Mycobacterium smegmatis (red, reference A0R2B1). (B) The model of Candidatus Frankia californiensis Dg2 (yellow, global model quality estimate QMEANDisCo >0.70). (C) The solved crystal structure of 2-oxoglutarate dehydrogenase of Staphylococcus epidermis (rosy brown, reference Q5HPC6). (D) Overlay of three different models, with arrows indicating the major differences between the 2-OG decarboxylase and 2-OG dehydrogenase. The production of succinate is thus dependent on the activity of SSA-DH. However, the gene encoding this enzyme could not be identified in Frankia cluster-2 genomes. The NADP + -dependent variant was found to be common in genomes of strains from other Frankia clusters, whereas the NAD + -dependent variant was found only in some of these genomes ( Supplementary Table S2 ). Using a pBLAST query with the corresponding sequence from F. alni ACN14a, the closest homologue in cluster-2 genomes was found to be a protein belonging to the aldehyde dehydrogenase family (listed in Supplementary Table S2 ). Yet, this dehydrogenase had <45% amino acid sequence identity with the NADP + -dependent SSA-DH in Frankia genomes of other clusters. The SSA-DH could also not be identified in the proteome data available for F. coriariae BMG5.1 ( Ktari et al. , 2017 ), the only Frankia cluster-2 species which has been cultivated to date. The lack of strong homologues of NADP + -dependent or NAD + -dependent SSA-DH alone cannot eliminate the possibility for another enzyme to catalyse the reaction. To determine if any activity was present in Frankia cluster-2 strains, an SSA-DH activity assay as described by Tian et al. (2005) was conducted for F. coriariae BMG5.1. F. alni ACN14a was used as a positive control because its genome contains genes encoding NADP + -dependent as well as NAD + -dependent SSA-DH ( Supplementary Table S2 ). The absorbance significantly increased after the addition of SSA in F. alni ACN14a ( Fig. 7 ) in a buffer containing NADP + , indicating that the NADP + variant was active. However, a successful assay could not be established for the NAD + variant. In extracts of F. coriariae BMG5.1, no activity could be measured after the addition of SSA, in the presence of either NADP + or NAD + , thus confirming the lack of SSA-DH activity ( Fig. 7 ). F. coriariae BMG5.1 grows considerably more slowly than any other cultivable Frankia strain. The medium recommended by the DSMZ (medium 1589) contains sodium succinate. This indicates that external supplementation with an intermediate from the reductive part of the TCA cycle is required for this strain to show sufficient growth. The strain was maintained in medium without external succinate for at least 3 weeks before the activity assay to ensure that SSA-DH activity was not down-regulated. Fig. 7. Enzyme activity assay for succinic semialdehyde dehydrogenase (SSA-DH) in cluster-1 Frankia alni ACN14a compared with cluster-2 Frankia coriariae BMG5.1. Absorbance was measured at 340 nm, to detect the production of NADPH from NADP + ; absorbance was blank corrected. Protein extracts were allowed to acclimate for 30 min in the reaction buffer with NADP + but without substrate, after which SSA or solvent only was injected. The absorbance was measured for an additional 50 min. Time is represented in cycles, with increasing darkness (red for ACN14a and yellow for BMG5.1) indicating that more time has passed. Boxplot indicates the absorbance at the beginning, before injection, after injection, and at the end of the experiment, based on two technical replicates of three biological replicates. The compact letters display indicates significant differences based on three-way ANOVA (sample, treatment, cycle), followed by Tukey post-hoc analysis. Apart from acting as a precursor of succinate, 2-OG can be used for the synthesis of glutamate via the GS/GOGAT cycle, and support ammonium assimilation. As shown above, our results suggest that the export of an assimilated form of N from the host is a common feature in nodules induced by Frankia cluster-2 strains. This would explain the loss of the genes for the glyoxylate shunt and SSA-DH observed in Frankia cluster-2, as in symbiosis ammonium assimilation would require the majority of 2-OG to be used for ammonium assimilation, leading to glutamate biosynthesis. Most reactions of the TCA cycle are reversible, except for the step catalysed by isocitrate dehydrogenase (encoded by icd ). The carboxylate provided by the plant as carbon source, which thus far has only been identified for A. glutinosa where it represents malate ( Jeong et al. , 2004 ), could be converted into any compound on the reductive side of the TCA cycle ( Fig. 5 ), and thus into fumarate and succinate by the reverse reactions of succinic dehydrogenase; and into fumarate, as well as into oxaloacetate, by malate dehydrogenase. Under symbiotic conditions in nodules, the high input of presumably malate to the reductive side of the TCA cycle would allow 2-OG to be drawn out of the oxidative side of the TCA cycle at a high rate. Hence, the TCA cycle would not be required to work as a cycle, but only as a linear pathway. We hypothesized that Frankia cluster-2 strains have this carbon metabolism: the TCA cycle acts as a linear pathway instead of a cycle to keep up with the constant removal of 2-OG for ammonium assimilation ( Fig. 8 ). Gene losses within the TCA cycle are not unique to Frankia , as shown by the symbiotic cyanobacteria UCYN-A and Trichormus azollae (previously Nostoc azollae ), which even lack the entire TCA cycle ( Ran et al. , 2010 ; Tripp et al. , 2010 ), as well as some rhizobial mutants ( Green et al. , 2000 ; Schulte et al. , 2021 ). Frankia cluster-2 strains represent the earliest divergent symbiotic clade within the Frankia genus ( Sen et al. , 2014 ; Persson et al. , 2015 ). Their metabolism might be based on an ancient form of symbiotic metabolite exchange. The export of an assimilated form of N is not energy efficient: it requires a much larger supply of carbon skeletons from the host to the endosymbiont. In contrast to ammonia, which can leak through the bacterial membrane and be converted to ammonium in the acidic perisymbiont space, assimilated forms of N would have to be transported across the bacterial membrane at the expense of energy, which has to be provided by the host ( Fig. 8 ). Fig. 8. Summary of the data presented. The host plant provides Frankia with carbon sources which are shuttled into the TCA cycle. Several gene losses were identified, resulting in a linearization of the cycle. As a result, the produced 2-oxoglutarate is used for ammonium assimilation. The export of an assimilated nitrogen source is dependent on the host plant. A heatmap represents all Frankia gene expression analysed, including genes which were not discussed in the manuscript. Gene expression levels are indicated in light (lower) to dark (higher) green. The divergence between the symbiotic cluster-2 and the non-symbiotic cluster-4 took place early in the evolution of Frankia , followed by the divergence of the precursor of cluster-1 and cluster-3 from cluster-4 ( Sen et al. , 2014 ; Berckx et al. , 2024 ). Frankia cluster-1 strains export ammonia or ammonium to the host, as opposed to assimilated N ( Guan et al. , 1996 ). This indicates that in different symbiotic clusters, the preferred export N form evolved differently. The de facto linear version of the TCA cycle from the provided carbon source to 2-OG might have led to the loss of genes involved in the production of succinate from 2-OG. The loss of the SSA-DH gene in Frankia cluster-2 might have prevented the evolution of a more energy-efficient nutrient exchange system. Given their adaptation of the TCA cycle to the metabolite exchange in symbiosis, it is not surprising that cluster-2 stains have such a low saprotrophic potential and are rarely found in the soil in the absence of a host plant ( Persson et al. , 2015 ; Battenberg et al. , 2017 ). Conclusions Based on the data presented in this study, the export of an assimilated form of N by Frankia cluster-2 strains in symbiosis is a common feature of the clade. In Cucurbitales host plants, such as D. glomerata and C. myrtifolia , this export form is arginine, while in Rosales , such as C. thyrsiflorus , it is asparagine or glutamine. The assimilation of fixed N for export during symbiosis puts a high demand on 2-OG. The TCA cycle therefore seems to work linearly from the carbon source(s) provided by the host to 2-OG. Due to this special metabolism, the need for the glyoxylate shunt is obviated, as well as the production of succinate from 2-OG. This led to gene losses which can explain the low saprotrophic potential of Frankia cluster-2 strains."
} | 10,416 |
36596783 | PMC9810717 | pmc | 7,985 | {
"abstract": "With advances in robotic technology, the complexity of control of robot has been increasing owing to fundamental signal bottlenecks and limited expressible logic state of the von Neumann architecture. Here, we demonstrate coordinated movement by a fully parallel-processable synaptic array with reduced control complexity. The synaptic array was fabricated by connecting eight ion-gel-based synaptic transistors to an ion gel dielectric. Parallel signal processing and multi-actuation control could be achieved by modulating the ionic movement. Through the integration of the synaptic array and a robotic hand, coordinated movement of the fingers was achieved with reduced control complexity by exploiting the advantages of parallel multiplexing and analog logic. The proposed synaptic control system provides considerable scope for the advancement of robotic control systems.",
"introduction": "Introduction In recent decades, advances in robotic control systems have enabled more sophisticated and delicate robot movement control 1 – 5 . However, high-level robotic control has inevitably increased the complexity of the control system. Although efforts have been made to develop efficient control systems, current CMOS-based robotic control systems have inherent limitations—the von Neumann bottleneck and the binary logic structure—resulting in signal delay and poor chip integration 6 – 12 . The advent of synaptic transistors provided a breakthrough in addressing signal delay and chip integration issues 13 – 17 ; these devices are capable of parallel computation and analog signal processing, similar to the human nervous system 18 – 22 . Among the various types of synaptic transistors 6 , 12 , 23 – 27 , ion-based electrochemical synaptic transistors are attracting attention for biomimetic applications since their working mechanism is similar to that of human synapses. Human synapses transmit biological signals by releasing neurotransmitters from presynaptic neurons; the neurotransmitters pass through the synaptic cleft and reach postsynaptic neurons 18 , 20 – 22 , 28 – 30 . In the case of an electrochemical synaptic transistor, signal transmission occurs through the electrical-input-signal-induced penetration of a semiconducting channel by the ionic species. Ionic movement in an electrochemical synaptic transistor depends on the transmission distance. This unique property of synaptic transistors facilitates the control of multiple devices with a single input signal through the selection of an appropriate signal transmission distance and the integration of multiple signals from each device into a unified signal 31 , 32 . The multi-control and signal processing capabilities of ion gel reduce computational efforts when synaptic transistors are used in robotic systems, since they allow parallel control under a reduced input signals and efficient distance-based signal multiplexing. Owing to their analog processing capability, synaptic transistors also help enhance robotic control performance by allowing digital-to-analog circuits (DACs)—an essential circuit component of conventional CMOS-based control systems—to be omitted and thereby reducing the circuit complexity of the control system. The combination of these core characteristics of synaptic transistors is helpful for realizing complex actuation such as the coordinated movement of a human finger 33 – 35 with simplified circuits and fewer computational requirements. In this study, we developed a fully parallel-processable control system capable of signal processing, by promoting free ionic movement in the dielectric of an electrochemical synaptic transistor. The system was constructed by connecting an ion-gel-based parallel-processable synaptic array (PPSA) to a coordinated robotic hand (CRH). These two control system components were fabricated using eight organic artificial synaptic transistors (OASTs) that shared a common ion gel dielectric and assembling three independent NiTi shape memory alloy fiber-based robotic fingers in a 3D-printed body, respectively. First, it was verified that three OASTs could operate simultaneously with a single input and that their synaptic output could be adjusted by changing the distance between the input gate and the transistor. Subsequently, the bending angle of robotic finger for different synaptic outputs was evaluated. Finally, coordinated robotic actuation of grabbing an object with curvature and a complex design was successfully performed for an optimized input gate-transistor distance by using a circuit architecture with reduced complexity. The results of this study are expected to contribute to the improvement of the control efficiency of humanoids and animatronics, for which complex calculations are a requisite for robotic operation.",
"discussion": "Discussion The PPSA-based synaptic control system is capable of coordinated actuation through parallel multiplexing and analog control resulting from the characteristics of signal multiplexing in the shared ion gel of PPSA. The control system comprised the PPSA and CRH. The robotic fingers of the CRH were controlled simultaneously by a single input that could be modulated by the signal transmission distance in the PPSA. The coordinated actuation performance of the control system was verified for grabbing action by the CRH by using a cup and a complex-shaped object. The proposed PPSA-based synaptic control system serves as an efficient signal processing platform and provides a breakthrough for robotic control systems involving massive amounts of computation."
} | 1,385 |
29315312 | PMC5760028 | pmc | 7,986 | {
"abstract": "Worldwide, coral reef ecosystems are experiencing increasing pressure from a variety of anthropogenic perturbations including ocean warming and acidification, increased sedimentation, eutrophication, and overfishing, which could shift reefs to a condition of net calcium carbonate (CaCO 3 ) dissolution and erosion. Herein, we determine the net calcification potential and the relative balance of net organic carbon metabolism (net community production; NCP) and net inorganic carbon metabolism (net community calcification; NCC) within 23 coral reef locations across the globe. In light of these results, we consider the suitability of using these two metrics developed from total alkalinity (TA) and dissolved inorganic carbon (DIC) measurements collected on different spatiotemporal scales to monitor coral reef biogeochemistry under anthropogenic change. All reefs in this study were net calcifying for the majority of observations as inferred from alkalinity depletion relative to offshore, although occasional observations of net dissolution occurred at most locations. However, reefs with lower net calcification potential (i.e., lower TA depletion) could shift towards net dissolution sooner than reefs with a higher potential. The percent influence of organic carbon fluxes on total changes in dissolved inorganic carbon (DIC) (i.e., NCP compared to the sum of NCP and NCC) ranged from 32% to 88% and reflected inherent biogeochemical differences between reefs. Reefs with the largest relative percentage of NCP experienced the largest variability in seawater pH for a given change in DIC, which is directly related to the reefs ability to elevate or suppress local pH relative to the open ocean. This work highlights the value of measuring coral reef carbonate chemistry when evaluating their susceptibility to ongoing global environmental change and offers a baseline from which to guide future conservation efforts aimed at preserving these valuable ecosystems.",
"conclusion": "Concluding remarks The results of this study clearly demonstrate that there is a large amount of variability in the carbonate chemistry overlying coral reef communities and ecosystems throughout the world, which could affect how individual reefs respond to global change. While some reefs experienced instances of net CaCO 3 dissolution, all reefs in this study were net calcifying for the majority of observations. This indicates that, from a chemical perspective, the reefs in this study are currently maintaining positive CaCO 3 production critical for the maintenance of reef structure and function ( Fig 3 ). However, these chemical measurements do not tell us if net calcification is able to offset any mechanical CaCO 3 erosion and off-reef transport. It is important to emphasize that, despite the lack of reef scale estimates of absolute rates of net calcification, ΔTA serves as an important indicator of whether a reef is net calcifying or net dissolving, and can show how relative rates of NCC vary over time. When using ΔTA as an estimate of coral reef net calcification potential it is also important to consider other biogeochemical processes that could alter seawater TA concentrations such as submarine groundwater discharge [ 37 ]. The observed differences in the relative balance of NCP and NCC within these systems reflect inherent differences in net reef metabolism (i.e., net autotrophic, heterotrophic, calcifying, or dissolving) [ 12 , 14 , 49 , 62 , 73 ], the benthic community composition [ 22 , 23 ], organic and inorganic carbon inputs from rivers, groundwater, and/or advection from offshore [ 74 , 75 ], as well as the spatial scale of the study (Figs 2 and 4 ). Similar to ΔTA measurements, TA-DIC slopes and the relative balance of NCP and NCC can provide important metrics of reef metabolism and function over time [ 29 , 32 , 43 ], but it is critical that the spatial scale of the data is carefully considered. Additional efforts integrating TA and DIC measurements with traditional surveying techniques and/or remote sensing at different scales will be important to fully link the chemical seawater signal with coral reef benthic community structure, function, and health. Currently, coral reef community composition and health are mainly monitored visually by SCUBA divers using transects and, more recently, photographic methods which are typically labor intensive, expensive, and partly reliant on subjective interpretation. Recent advances in remote sensing, imaging technologies, and machine learning will greatly improve these visual observations [ 76 , 77 ]. However, these approaches do not provide direct information about the biogeochemical processes and elemental cycling of coral reefs, which are becoming increasingly important to monitor. Ideally, coral reef monitoring programs should include information on both community composition and quantitative biogeochemical data assessing the organic and inorganic carbon cycles. Carbonate chemistry measurements are becoming increasingly automated [ 78 , 79 ], which will improve the feasibility and practicality of monitoring the ‘metabolic pulse’ of coral reefs and other marine ecosystems. Furthermore, the use of autonomous vehicles, satellite products, and perhaps even citizen scientists [ 80 ] will facilitate data collection over larger spatial and longer temporal scales. In conclusion, this study offers an overview of the metabolic function for a range of coral reefs at different scales that warrants a more sophisticated and interdisciplinary approach in future studies. This synthesis of carbonate chemistry within a selection of the world’s coral reefs provides a framework for developing hypotheses aimed at addressing the mechanistic attribution to observed differences in the ‘metabolic pulse’ and serves as a critical baseline for building an understanding of how ocean acidification and other perturbations will impact individual coral reef ecosystems. Results herein confirm, based on global observations, that seawater pH within reef systems will not only be determined by how open ocean CO 2 chemistry is changing, but by complex interactions between benthic metabolism, local hydrodynamics, terrestrial inputs, and any changes in community composition [ 20 , 21 , 31 ]. It is also apparent that the carbonate chemistry within individual reef systems varies greatly across the globe, and even spatiotemporally within individual reefs and habitats. Importantly, knowing the current balance of NCP and NCC within a coral reef already gives us an idea of which reefs experience greater variability and extremes in pH, and how the chemistry within each reef will change as a result of changing community structure, net reef metabolism, and ongoing ocean acidification (e.g., [ 20 ]). This knowledge is critical to accurately determine the biological and geochemical effects of ocean acidification on reef systems. Observed differences in the TA-DIC slope between different reefs, and thus, their ability to exacerbate or alleviate ocean acidification could translate into different susceptibilities and resistance to environmental perturbations. If so, monitoring changes in ΔTA and the TA-DIC slope through time will provide valuable insights into the function and health of coral reefs that may be critical in guiding conservation efforts aimed at maximizing the success of these ecosystems.",
"introduction": "Introduction The health of coral reefs is declining globally due to human induced environmental changes [ 1 – 4 ]. However, individual coral reefs may exhibit different susceptibilities and resistance to environmental change that are dependent on a number of factors such as community composition, reef biogeochemistry, environmental and oceanographic properties, and pressure from local human activities [ 5 – 8 ]. Arguably, one of the most important determinants of overall reef function is the construction and maintenance of calcium carbonate (CaCO 3 ) reef structure, which is vital to the myriad of ecosystem services that coral reefs provide [ 3 , 9 , 10 ]. Ocean acidification is expected to eventually shift reefs from a state of net CaCO 3 precipitation to net dissolution through a reduction in seawater pH and aragonite saturation state (Ω a ) [ 11 , 12 ]. Other threats such as coral bleaching, overfishing, and eutrophication could exacerbate the loss of CaCO 3 structure and habitats via the loss of coral cover and shifts towards algal dominated systems [ 3 , 6 ]. Consequently, resolving whether coral reef ecosystems will maintain net calcification under future environmental change is perhaps one of the most urgent coral reef research questions that needs to be addressed [ 13 ]. While diel and seasonal variations in open ocean surface seawater carbonate chemistry are relatively modest compared to future changes anticipated from ocean acidification, the biogeochemical processes and high metabolic rates of coral reef communities (i.e., the balance of photosynthesis, respiration, calcification, and CaCO 3 dissolution) can drive dramatic changes in the carbonate chemistry of seawater overlying reef ecosystems on these timescales (e.g., [ 14 – 17 ]). Field observations and numerical modeling results have demonstrated that ocean acidification will interact with local reef biogeochemistry, leading to enhanced variability in future seawater pH due to a reduction in seawater buffering capacity [ 18 , 19 ]. It has recently been proposed that these biogeochemical processes could locally counteract or exacerbate ocean acidification through changes in seawater pH and Ω a [ 20 – 24 ]. While it is clear that pH will change differently in reefs across the globe due to ocean acidification [ 18 ], it is not fully understood whether marine organisms and communities will mainly respond to a reduction in average pH, the occurrence of extreme pH values, changes in pH variability, or some combination [ 25 – 28 ]. Therefore, to gain a better understanding of how coral reefs will respond to ocean acidification it is important to understand how seawater carbonate chemistry currently varies within individual coral reefs, largely because this will affect how local chemistry changes in response to global perturbations. A coral reef’s metabolic influence on seawater chemistry is reflected as the rhythmic, diel and seasonal cycles of seawater TA and DIC (i.e., the ‘metabolic pulse’). Because changes in TA and DIC largely reflect coral reef metabolism ( Fig 1 ), they can provide useful metrics that indicate perturbations to coral reefs on organismal, community, and ecosystem scales. One such metric is the TA anomaly between reef and source seawater, which we term the net calcification potential. Comparing reef TA concentrations to surrounding open ocean or source seawater (ΔTA) provides a direct indication of whether a reef is net calcifying (TA depletion) or dissolving (TA repletion) [ 12 , 29 – 31 ], although to determine the exact rate of CaCO 3 precipitation/dissolution also requires estimates of the seawater residence time and volume of water above the reef. However, changes in ΔTA monitored over time at individual reefs could indicate shifts in NCC assuming there are no major concurrent changes to reef hydrodynamics and seawater residence time. Therefore, monitoring the net calcification potential of coral reefs (i.e., ΔTA) through time could be critical in guiding conservation and management efforts by contributing to a better understanding of a reef’s ability to deposit CaCO 3 . 10.1371/journal.pone.0190872.g001 Fig 1 The dominant metabolic processes on coral reefs and their influence on seawater total alkalinity (TA), dissolved inorganic carbon (DIC), and pH. (A) The organic carbon cycle (NCP) is dominated by photosynthesis and respiration, which take up or release 1 mole of DIC for every mole of organic carbon (CH 2 O) produced or decomposed with little influence on seawater TA. In contrast, the inorganic carbon cycle (NCC) is dominated by CaCO 3 precipitation and dissolution, which alter TA and DIC in a ratio of 2:1 for every mole of CaCO 3 precipitated or dissolved. Photo credit: Yuna Zayasu, OIST. (B) Depending on the relative contribution from different metabolic processes, the resulting change in TA and DIC influences seawater pH differently (colored contours). Photosynthesis and CaCO 3 dissolution increase seawater pH while respiration and CaCO 3 precipitation decrease pH. If NCP and NCC are closely balanced (i.e., TA-DIC slope ~1), there is little change in seawater pH owing to net reef metabolism. This is because the slope of pH isolines within the normal oceanic concentration of seawater TA and DIC are close to 1. Therefore, when the slope of the TA-DIC vector is different from 1 the pH isolines are crossed and seawater pH can be altered considerably. The calculations for pH at each TA and DIC value assume constant temperature (25°C) and salinity (35). (C) Conceptual schematic of the biogeochemical and metabolic function of coral reefs. Net CaCO 3 precipitation (+NCC, green area) vs. net dissolution (-NCC, pink/red area); net autotrophy (+NCP) vs. net heterotrophy (-NCP), and different TA-DIC slopes, as well as the resulting changes in reef seawater pH (pH r ) relative to the open ocean (pH o ) under constant salinity and temperature conditions. Another potential metric of reef biogeochemistry derived from TA and DIC measurements is the slope of TA-DIC regressions, which reflects the balance of NCP and NCC, such that in a system dominated by NCP the slope approaches 0 while in a system dominated by NCC the slope approaches 2. Because of these properties, graphical vector analysis of seawater TA and DIC (e.g., the TA-DIC slope or ΔTA:ΔDIC) quantifies the relative balance of CaCO 3 to organic carbon fluxes [ 29 , 32 , 33 ], which can directly relate to properties such as community composition (e.g., calcifying vs. non-calcifying organisms) and net reef metabolic status ( Fig 1 ) [ 34 , 35 ]. The balance of NCP and NCC also reflects how a reef modifies seawater pH, and thus, whether net reef metabolism elevates or suppresses reef pH relative to the open ocean. Graphical vector analysis of TA and DIC [ 33 ] has been successfully used to characterize reefs as sources or sinks of CO 2 to the atmosphere [ 32 ], to characterize the dominant processes responsible for the modification of seawater chemistry across different reef habitats [ 9 , 23 ], and to demonstrate how coral bleaching and outbreaks of predatory starfish resulted in a reef-wide decrease in net community calcification [ 29 ]. Consequently, the TA-DIC vector approach is a powerful tool that can be used to compare the balance of NCP and NCC on different coral reefs, and potentially, a simple and effective monitoring tool of changes to coral reef community metabolism arising from ocean warming, ocean acidification, and other perturbations over time. Herein, we gathered seawater TA and DIC data from 27 temporal or spatial sampling expeditions covering 23 global coral reef locations in all major ocean basins ( S1 Fig ). The studies ranged in scope from one day at a single reef location to multiple years covering extensive spatial scales (>10 km 2 ). Based on these studies we evaluate the current net calcification potential and the relative balance of NCP and NCC across a global selection of coral reefs; we qualitatively assess whether TA-DIC slopes can be a useful metric for monitoring changes in reef biogeochemistry due to anthropogenic perturbations; we detail how seawater carbonate chemistry measurements can better inform our understanding of coral reef susceptibility to future ocean acidification; and, we propose that future studies need to take an interdisciplinary approach combining ecological, biogeochemical, and physical measurements to develop a mechanistic understanding of coral reef ecosystems.",
"discussion": "Results and discussion The current state of net ecosystem calcification Most of the coral reefs in this study showed extensive depletion of TA relative to adjacent open ocean values (negative ΔTA) for the majority of observations (Figs 2 and 3 ). The average ΔTA within individual reef systems ranged from -114 to +68 μmol kg -1 ( S1 Table ). The only reef that had a positive average ΔTA was Muri Lagoon in the Cook Islands, which was influenced by the input of high TA groundwater [ 37 ]. The lowest average ΔTA (-114 μmol kg -1 ) was observed in Western Panama (Pacific), which could be due to high rates of NCC and/or long residence times assuming that no other processes such as upwelling or freshwater input from rain or runoff affected seawater TA [ 44 ]. On average, the observed ΔTA relative to offshore was -36 ± 26 μmol kg -1 for all locations, with the Cook Island dataset excluded because of substantial groundwater input [ 37 ]. This is direct evidence that the reefs in this study, on average, are currently maintaining net calcification. However, most reefs also appeared to undergo net CaCO 3 dissolution at times (mainly at night) as indicated by positive ΔTA values. Therefore, when evaluating ΔTA it is critical to devise a sampling strategy that adequately assesses times when both net CaCO 3 production and dissolution are more likely to affect water column TA measurements (e.g., midday, early morning, or night). 10.1371/journal.pone.0190872.g003 Fig 3 Net calcification potential measured as anomalies between open ocean and coral reef TA concentrations (ΔTA) at each location. Negative values are lower TA concentrations within the reef and represent net CaCO 3 precipitation (+NCC), while positive values are higher TA concentrations within the reef and represent net CaCO 3 dissolution (-NCC). Edges of the box are the 25 th and 75 th percentiles, the line within each box is the median, the whiskers represent the most extreme data points that are not outliers, and the red + symbols are outliers. The proportion of positive ΔTA observations at each reef site (indicative of times of net dissolution or negative NCC) ranged from 0 to 67% ( S1 Table ). Between oceanic regions, the average percent occurrences of net dissolution ranged from 25 ± 22% for reefs in the Atlantic Ocean, 11 ± 16% within the Great Barrier Reef (GBR), 19 ± 22% throughout the rest of the Indo-Pacific, and 6% in Western Panama (we were unable to calculate ΔTA for the Red Sea dataset as the TA of the source water was highly variable) (Figs 2 and 3 ). With the exception of Lady Elliot Island, reefs in the GBR had relatively low occurrences of net dissolution (≤11%). Importantly, reefs in the GBR were all sampled over complete diel cycles and had no temporal sampling bias, which would not be the case for reefs sampled only during the daytime. Given the large variability within each region, it can be concluded that incidents of net dissolution are mostly related to local reef conditions rather than regional differences. Net dissolution mainly occurs at night coincident with net respiration, high CO 2 concentrations, and low rates of gross calcification [ 12 , 14 , 36 , 45 ]. Net dissolution can also occur on a seasonal basis and was observed during winter months in both Bermuda [ 31 , 56 ] and Florida [ 46 ]. In fact, 34% and 58% percent of the observations from Cheeca Rocks and the Florida Keys, respectively, indicated net dissolution, which was hypothesized to be caused by seasonally driven net respiration and low rates of calcification during winter [ 46 ]. Based on results from this and previous studies, monitoring changes in ΔTA over time serves as a relatively easy, highly sensitive, and instantaneous indicator of coral reef net calcification under both natural and anthropogenic environmental change. However, it is important to recognize that ΔTA doesn’t provide a quantitative measurement of NCC without knowing seawater residence times and volume. A number of approaches have been used to estimate seawater residence time for different reefs (e.g., using salinity anomalies [ 48 , 57 , 58 ], radioactive tracers such as 7 Be [ 46 , 57 ], and direct measurements of water movement and currents [ 59 , 60 ]) but currently these estimates remain limited to a few locations and are generally associated with large uncertainty [ 57 ]. However, recent advances in analytical capabilities to measure various isotopes and/or tracers in seawater combined with access to hydrodynamic modeling tools may assist in characterizing the residence time more precisely and at a larger number of reefs in future studies. The relative balance of NCP and NCC The relative influence of NCP on changes in DIC concentrations at the different reef sites ranged from 32% to 88% as indicated by TA-DIC slopes ranging from 0.24 to 1.36 (Figs 2 and 4A , S1 Table ). Overall, reefs in the Atlantic Ocean had relatively high TA-DIC slopes compared to most other reefs, ranging from 0.69 (Cheeca Rocks) to 1.36 (Puerto Rico), while Indo-Pacific reefs (including the GBR) had a larger range from 0.24 (Hawaii) to 1.35 (Maldives). Importantly, the relatively higher TA-DIC slopes in Atlantic compared to Indo-Pacific reefs was strongly influenced by the spatial scale of the studies. In general, studies covering a larger spatial scale (>10km 2 ) had significantly higher TA-DIC slopes compared to studies characterizing seawater carbonate chemistry at only one site (p<0.001; Wilcoxon test; Fig 4B ). At larger spatial scales, carbonate chemistry measurements represent an integrated signal of multiple reef habitats and communities rather than the signal of one localized community. At these scales, the trophic status of coral reefs is most likely closely balanced (i.e., NCP = 0) [ 61 , 62 ], and thus, TA-DIC slopes are more strongly influenced by NCC compared to localized studies that are more strongly influenced by high frequency variations in NCP. Consequently, when comparing the relative balance of NCP and NCC between different reefs it is important to only compare studies of similar spatial coverage. In contrast, the duration of the studies appeared to have no influence on TA-DIC slopes ( Fig 4C ). Based on these considerations, most of the temporally based studies were dominated by NCP with the percent influence of NCP on DIC changes ranging from 53% to 88% ( S1 Table ). Of the temporally based studies, sites in the Red Sea were most strongly dominated by NCP (87% ± 0.9%; average ± SD), while sites in the GBR (80% ± 5%) and the rest of the Indo-Pacific (81% ± 9%) had a relatively lower and similar average percent influence of NCP comprising a larger range. Within the spatially based studies, Cheeca Rocks (65%) and St. John (60%) were most strongly influenced by NCP while the Maldives (32%) and Puerto Rico (32%) were least influenced. Unfortunately, all of the Atlantic based studies were classified as spatial, while the majority of the Pacific based studies were classified as temporal, making any comparisons between the ocean basins tenuous. 10.1371/journal.pone.0190872.g004 Fig 4 The influence of spatial and temporal sampling scales on TA-DIC slopes. (A) The TA-DIC slopes from the different coral reef locations calculated using a Type II linear regression ( see also S1 Table ). Error bars are ±1 SD of the slope. The colors indicate whether the sampling protocol had a significant spatial component (>10 km 2 ; blue) or was predominantly sampled over time at one location (orange). (B) Comparison of the average TA-DIC slope (± standard deviation) of all temporally and spatially sampled reefs. (C) TA-DIC slopes as a function of the duration of each study in days grouped by spatial (blue) and temporal (organge) sampling protocols. It might be hypothesized that reefs with higher TA-DIC slopes (i.e., higher %NCC) would also have a greater depletion of TA due to high rates of NCC. Notably, there was no apparent correlation between average ΔTA or percent occurrences of dissolution and TA-DIC slopes within our dataset ( S4 Fig ). This qualitatively indicates that the relative balance of NCP and NCC (i.e., organic vs. inorganic metabolism) is not directly related to the production of CaCO 3 , recognizing the caveat that ΔTA is only a quasi-quantitative estimate of calcification (i.e., even reefs with a relatively small depletion of TA may produce large amounts of CaCO 3 if the residence time is short). Most importantly, the results show that even reefs and/or reef habitats dominated by NCP still undergo substantial net calcification. This is supported by visual and quantitative observations of reefs with low TA-DIC slopes (e.g., Heron and One Tree Islands) that have extensive and healthy coral populations and high rates of NCC [ 14 , 45 , 63 ]. Differences in TA-DIC slopes, and subsequently NCP:NCC, could be related to differences in benthic community composition, including the percent cover of calcifying vs. non-calcifying organisms [ 9 , 32 , 43 ]. Regrettably, the current TA-DIC dataset was not associated with detailed benthic community composition characterizations or estimates of chemical footprint dimensions for all study sites. Consequently, it was not possible to evaluate how the observed balance of NCP and NCC and variation in carbonate chemistry were related to in situ benthic communities. However, a recent short-term (24 h) mesocosm experiment with diverse benthic communities from Hawaii illustrated that the TA-DIC slope, TA anomaly, and seawater pH varied predictably according to community composition and relative rates of NCP and NCC [ 34 ]. These results demonstrate that different calcifying communities can have similar TA-DIC slopes (e.g., corals, calcifying algae, and sand), but radically different TA anomalies related to differences in the absolute rates of NCC. This adds insight into the lack of correlation between average ΔTA and TA-DIC slopes from the in situ data ( S2 Fig ), and indicates that reef sites with similar TA-DIC slopes may have different benthic community composition with different calcification capacities. Alternatively, recent work from Mo’orea has also shown TA-DIC slopes to be taxa-specific, suggesting that a community’s TA-DIC slope may not only be dependent on the overall percent cover of calcifiers, but also the relative abundance and diversity of these calcifiers [ 35 ]. On an ecosystem scale, TA-DIC slopes reveal the relative balance between NCP and NCC, and observations over time at a single reef location will be useful for detecting local changes in the percent cover of calcifiers and shifts in biogeochemistry due to environmental perturbations. Changes in the TA-DIC slope over time will also be affected by other processes besides coral calcification such as changes in CaCO 3 sediment dissolution [ 11 , 64 ] and chemical bioerosion [ 65 , 66 ], making it a true indicator of ecosystem level response to climate change. A coral reef’s influence on seawater pH The local modification of TA and DIC determines how seawater pH changes within that reef. Reef sites with the lowest TA-DIC slopes and the largest absolute change in DIC experienced the greatest variability in seawater pH (Figs 1 and 2 ). In contrast, reefs with balanced NCP and NCC ratios experienced less variability in pH. The reason for this is that reefs with a TA-DIC slope close to 1 change seawater chemistry along isolines of constant pH while reefs with a slope significantly different from 1 cross isolines of constant pH (Figs 1 and 2 ) [ 9 ]. This is because the slope of pH isolines within the normal oceanic concentration of seawater TA and DIC are close to 1. Consequently, if the net metabolism of coral reefs remains unchanged under future ocean acidification, reefs with a TA-DIC slope substantially different from 1 could exacerbate or counteract the effects of ocean acidification depending on the reef’s overall trophic status (i.e., net autotrophic or heterotrophic) ( Fig 1C ). The modification of pH within individual coral reefs may relate to different susceptibilities and resistance, as variability in seawater pH has been shown to modulate the response of some corals to ocean acidification [ 67 ]. Recent studies have shown that reductions in coral calcification due to decreasing pH may be tempered under oscillating compared to constant carbonate chemistry conditions [ 27 , 28 ], while other studies have showed no enhanced tolerance to ocean acidification by corals in low compared to highly variable pH conditions [ 26 , 68 ]. Furthermore, while the response of coral calcification to a reduction in average pH has been shown to be robust [ 69 ], there are examples of corals thriving in low pH environments [ 70 , 71 ], including deep water corals thriving in seawater with a Ω ar well below 1 [ 72 ]. The complex and nuanced responses of corals to seawater pH make it critical to evaluate how seawater pH changes currently, and will change into the future across a diverse array of coral reef systems. It is important to highlight that seawater pH variability is localized within coral reef ecosystems and dependent on a number of factors such as water column depth, residence time, community composition, etc. For example, the deeper areas of a reef (e.g., reef slopes where a large portion of calcification occurs) will most likely experience conditions similar to the open ocean rather than experiencing localized variability in carbonate chemistry which is most pronounced in shallow areas such as on rim reefs and reef flats. This work offers an initial, baseline examination of local carbonate chemistry in a global sample of coral reefs, knowledge of which is critical to predicting future changes due to ocean acidification within individual reefs. Concluding remarks The results of this study clearly demonstrate that there is a large amount of variability in the carbonate chemistry overlying coral reef communities and ecosystems throughout the world, which could affect how individual reefs respond to global change. While some reefs experienced instances of net CaCO 3 dissolution, all reefs in this study were net calcifying for the majority of observations. This indicates that, from a chemical perspective, the reefs in this study are currently maintaining positive CaCO 3 production critical for the maintenance of reef structure and function ( Fig 3 ). However, these chemical measurements do not tell us if net calcification is able to offset any mechanical CaCO 3 erosion and off-reef transport. It is important to emphasize that, despite the lack of reef scale estimates of absolute rates of net calcification, ΔTA serves as an important indicator of whether a reef is net calcifying or net dissolving, and can show how relative rates of NCC vary over time. When using ΔTA as an estimate of coral reef net calcification potential it is also important to consider other biogeochemical processes that could alter seawater TA concentrations such as submarine groundwater discharge [ 37 ]. The observed differences in the relative balance of NCP and NCC within these systems reflect inherent differences in net reef metabolism (i.e., net autotrophic, heterotrophic, calcifying, or dissolving) [ 12 , 14 , 49 , 62 , 73 ], the benthic community composition [ 22 , 23 ], organic and inorganic carbon inputs from rivers, groundwater, and/or advection from offshore [ 74 , 75 ], as well as the spatial scale of the study (Figs 2 and 4 ). Similar to ΔTA measurements, TA-DIC slopes and the relative balance of NCP and NCC can provide important metrics of reef metabolism and function over time [ 29 , 32 , 43 ], but it is critical that the spatial scale of the data is carefully considered. Additional efforts integrating TA and DIC measurements with traditional surveying techniques and/or remote sensing at different scales will be important to fully link the chemical seawater signal with coral reef benthic community structure, function, and health. Currently, coral reef community composition and health are mainly monitored visually by SCUBA divers using transects and, more recently, photographic methods which are typically labor intensive, expensive, and partly reliant on subjective interpretation. Recent advances in remote sensing, imaging technologies, and machine learning will greatly improve these visual observations [ 76 , 77 ]. However, these approaches do not provide direct information about the biogeochemical processes and elemental cycling of coral reefs, which are becoming increasingly important to monitor. Ideally, coral reef monitoring programs should include information on both community composition and quantitative biogeochemical data assessing the organic and inorganic carbon cycles. Carbonate chemistry measurements are becoming increasingly automated [ 78 , 79 ], which will improve the feasibility and practicality of monitoring the ‘metabolic pulse’ of coral reefs and other marine ecosystems. Furthermore, the use of autonomous vehicles, satellite products, and perhaps even citizen scientists [ 80 ] will facilitate data collection over larger spatial and longer temporal scales. In conclusion, this study offers an overview of the metabolic function for a range of coral reefs at different scales that warrants a more sophisticated and interdisciplinary approach in future studies. This synthesis of carbonate chemistry within a selection of the world’s coral reefs provides a framework for developing hypotheses aimed at addressing the mechanistic attribution to observed differences in the ‘metabolic pulse’ and serves as a critical baseline for building an understanding of how ocean acidification and other perturbations will impact individual coral reef ecosystems. Results herein confirm, based on global observations, that seawater pH within reef systems will not only be determined by how open ocean CO 2 chemistry is changing, but by complex interactions between benthic metabolism, local hydrodynamics, terrestrial inputs, and any changes in community composition [ 20 , 21 , 31 ]. It is also apparent that the carbonate chemistry within individual reef systems varies greatly across the globe, and even spatiotemporally within individual reefs and habitats. Importantly, knowing the current balance of NCP and NCC within a coral reef already gives us an idea of which reefs experience greater variability and extremes in pH, and how the chemistry within each reef will change as a result of changing community structure, net reef metabolism, and ongoing ocean acidification (e.g., [ 20 ]). This knowledge is critical to accurately determine the biological and geochemical effects of ocean acidification on reef systems. Observed differences in the TA-DIC slope between different reefs, and thus, their ability to exacerbate or alleviate ocean acidification could translate into different susceptibilities and resistance to environmental perturbations. If so, monitoring changes in ΔTA and the TA-DIC slope through time will provide valuable insights into the function and health of coral reefs that may be critical in guiding conservation efforts aimed at maximizing the success of these ecosystems."
} | 8,800 |
24859310 | PMC4103531 | pmc | 7,987 | {
"abstract": "The addition of ferrihydrite to methanogenic microbial communities obtained from a thermophilic anaerobic digester suppressed methanogenesis in a dose-dependent manner. The amount of reducing equivalents consumed by the reduction of iron was significantly smaller than that expected from the decrease in the production of CH 4 , which suggested that competition between iron-reducing microorganisms and methanogens was not the most significant cause for the suppression of methanogenesis. Microbial community analyses revealed that the presence of ferrihydrite markedly affected the bacterial composition, but not the archaeal composition. These results indicate that the presence of ferrihydrite directly and indirectly suppresses thermophilic methanogenesis."
} | 190 |
21643704 | null | s2 | 7,988 | {
"abstract": "The predominant strategy for using algae to produce biofuels relies on the overproduction of lipids in microalgae with subsequent conversion to biodiesel (methyl-esters) or green diesel (alkanes). Conditions that both optimize algal growth and lipid accumulation rarely overlap, and differences in growth rates can lead to wild species outcompeting the desired lipid-rich strains. Here, we demonstrate an alternative strategy in which cellulose contained in the cell walls of multicellular algae is used as a feedstock for cultivating biofuel-producing microorganisms. Cellulose was extracted from an environmental sample of Cladophora glomerata-dominated periphyton that was collected from Lake Mendota, WI, USA. The resulting cellulose cake was hydrolyzed by commercial enzymes to release fermentable glucose. The hydrolysis mixture was used to formulate an undefined medium that was able to support the growth, without supplementation, of a free fatty acid (FFA)-overproducing strain of Escherichia coli (Lennen et. al 2010). To maximize free fatty acid production from glucose, an isopropyl β-D-1-thiogalactopyranoside (IPTG)-inducible vector was constructed to express the Umbellularia californica acyl-acyl carrier protein (ACP) thioesterase. Thioesterase expression was optimized by inducing cultures with 50 μM IPTG. Cell density and FFA titers from cultures grown on algae-based media reached 50% of those (∼90 μg/mL FFA) cultures grown on rich Luria-Bertani broth supplemented with 0.2% glucose. In comparison, cultures grown in two media based on AFEX-pretreated corn stover generated tenfold less FFA than cultures grown in algae-based media. This study demonstrates that macroalgal cellulose is a potential carbon source for the production of biofuels or other microbially synthesized compounds."
} | 452 |
39440244 | PMC11493745 | pmc | 7,990 | {
"abstract": "Background Arbuscular mycorrhiza (AM) refers to a symbiotic association between plant roots and fungi that enhances the uptake of mineral nutrients from the soil and enables the plant to tolerate abiotic and biotic stresses. Although previously reported RNA-seq analyses have identified large numbers of AM-responsive genes in model plants, such as Solanum lycopersicum L., further studies are underway to comprehensively understand the complex interactions between plant roots and AM, especially in terms of the short- and long-term responses after inoculation. Results Herein, we used RNA-seq technology to obtain the transcriptomes of tomato roots inoculated with the fungus Rhizophagus irregularis at 7 and 30 days post inoculation (dpi). Of the 1,019 differentially expressed genes (DEGs) in tomato roots, 635 genes showed differential expressions between mycorrhizal and non-mycorrhizal associations at the two time points. The number of upregulated DEGs far exceeded the number of downregulated ones at 7 dpi, and this difference decreased at 30 dpi. Several notable genes were particularly involved in the plant defense, plant growth and development, ion transport, and biological processes, namely, GABAT , AGP , POD , NQO1 , MT4 , MTA , and AROGP3 . In addition, the Kyoto encyclopedia of genes and genomes pathway enrichment analysis revealed that some of the genes were involved in different pathways, including those of ascorbic acid ( AFRR , GME1 , and APX ), metabolism ( CYP , GAPC2 , and CAM2 ), and sterols ( CYC1 and HMGR ), as well as genes related to cell division and cell cycle ( CDKB2 and PCNA ). Conclusion These findings provide valuable new data on AM-responsive genes in tomato roots at both short- and long-term postinoculation stages, enabling the deciphering of biological interactions between tomato roots and symbiotic fungi.",
"conclusion": "5 Conclusion This study reveals the dynamic transcriptomic changes in the roots of S. lycopersicum L. after inoculation with R. irregularis and demonstrates the shifts in gene expressions over the short term (7 days) and long term (30 days). Using RNA-seq technology, it was possible to identify numerous DEGs that play crucial roles in improving nutrient uptake, stress resistance, and overall plant vigor. The results imply significant temporal changes in plant defense, growth, and metabolic processes while highlighting the complex molecular interplay between tomato roots and AMF. These insights into phased genetic responses to AM colonization not only deepen our understanding of plant–microbe symbioses but also pave the path for using these interactions to develop sustainable agricultural practices aimed at improving plant resilience and productivity under changing environmental conditions.",
"introduction": "1 Introduction Profound changes can be observed in the metabolomes of plants, such as tomatoes, from mycorrhizal interactions ( Rivero et al., 2015 ). In such instances, it is also noted that under certain conditions, arbuscular mycorrhizal fungi (AMF) as well as non-cultivated plant genes are expressed in tomato roots under the affected mycorrhiza ( Ruzicka et al., 2013 ). Mycorrhizal interactions are also known to affect tissues that are far away from the roots; in tomato plants, their influence also extends to fruit metabolism ( Zouari et al., 2014 ). This clearly shows the systemic effects of mycorrhizal interactions on the physiological functions of plants. Many of the molecular mechanisms involved in mycorrhizal symbiosis are only now being uncovered slowly, but sucrose transport has been found to be regulated by various genes as well as oxylipin metabolism and gene expression ( Bitterlich et al., 2014 ; León Morcillo et al., 2012 ). Scientists have devoted extensive efforts to determining how mycorrhizal symbiosis works in plants, but this area of research has remained underinvestigated. The dynamic interactions between mycorrhizal symbiosis and other environmental factors, such as nitrogen availability, have been revealed through the linked gene expression profiles of mycorrhizal tomato roots grown in different nutrient environments ( Ruzicka et al., 2010 ; Ruzicka et al., 2012 ). In addition, research indicates that mycorrhizal fungi enhance the defense mechanisms against pathogens, such as nematodes, demonstrating the multifaceted role of AM symbiosis in plant health ( Silva et al., 2022 ; Mahdy et al., 2008 ). Transcriptional profiling techniques such as microarray and RNA sequencing have been used for expression analyses of the interactions between arbuscular mycorrhiza (AM) and tomato plants in the roots, leaves, and fruits during the developmental stages ( Schubert et al., 2020 ; Cervantes-Gámez et al., 2016 ; Zouari et al., 2014 ; Ullah et al., 2023 ). AMF have been shown to significantly alter gene expressions in the roots of various plant species ( Hogekamp and Küster, 2013 ; Groten et al., 2015b ; Casarrubias-Castillo et al., 2020 ). Interestingly, very few studies have reported the differentially expressed genes (DEGs) in AM-inoculated tomato roots using microarray analysis ( Chialva et al., 2019 ; Dermatsev et al., 2010 ; Fiorilli et al., 2009 ) and RNA-seq analysis ( Sugimura and Saito, 2017 ; Vangelisti et al., 2019 ; Zeng et al., 2023 ; Tominaga et al., 2022 ) by focusing specifically on the early and late stages of inoculation. AM have been shown to mediate various aspects of plant physiology, including lipid peroxidation regulation, reactive oxygen species (ROS) level control, and antioxidant enzyme accumulation. These interactions play crucial roles in enhancing plant defense mechanisms against oxidative stresses. Moreover, the key antioxidant enzymes and mechanisms include proanthocyanidins, flavonoids, ascorbic acid, superoxide dismutase (SOD), monodehydroascorbate reductase (MDAR), peroxidase (POX), and total antioxidant capacities ( Hogekamp and Küster, 2013 ; Groten et al., 2015a ; Guether et al., 2011 ). Notably, previous studies have shown that both short- and long-term colonization by AMF can transcriptionally promote a specific retrotransposon in the roots of the sunflower plant ( Vangelisti et al., 2019 ). Additionally, post-transcriptional regulation has been shown to play a significant role in AM regulation in tomato roots ( Zeng et al., 2023 ). By recognizing the interplay in these processes, we aim to provide a comprehensive picture of the tomato root transcriptome under mycorrhizal interactions. Through comparative analyses of the short- and long-term post-infection responses, we expect to unravel some of the complex regulatory networks and molecular pathways that maintain the symbiotic relationships between plants and fungi, thereby shedding light on the vast field of knowledge encompassing plant–mycorrhiza interactions.",
"discussion": "4 Discussion The plant root is the central organ for sensing and recognizing symbiotic soil fungi and signaling responses. In this study, we detected major transcriptomic perturbations in tomato roots treated with R. irregularis over experimental periods of 7 and 30 dpi. Our results indicate that AM symbiosis induces vital effects in the tomato roots, such as nutrient availability as well as abiotic and biotic stresses ( Abarca et al., 2024 ; Singh et al., 2024 ; Pellegrino et al., 2024 ). In this study, RNA-seq experiments were performed on tomato roots after inoculation with R. irregularis for a comprehensive overview of the changes in gene expressions. In addition, the molecular mechanisms and signaling pathways underlying the effects of long-term AM interactions with root systems were investigated to provide valuable insights into the symbiotic relationships between tomato roots and AMF. In recent years, several omics technologies have emerged and gained acceptance in disciplines, such as plant sciences and life sciences. Among these technologies, RNA-seq stands out as it enables more precise and comprehensive understanding of the differences in gene expressions between different conditions or phenotypes ( Panthee et al., 2024 ; Zhao et al., 2018 ). Therefore, a global gene expression study using RNA-seq is essential to better understand the molecular basis of the tomato plant at different developmental stages. Our analysis identified 1,019 DEGs in the roots of the tomato plant; among these, 635 genes were upregulated and 384 were downregulated, suggesting that these genes may play important roles in the symbiotic relationships between tomato and mycorrhiza. To further clarify the functions of these genes in tomato, we subjected the identified DEGs to GO enrichment analysis ( Figure 5 ). The present work identified a set of transcription-associated genes during AMF symbioses in tomato roots colonized at 7 and 30 dpi. These genes are associated with plant defense ( LOC101247808 ( NRC4c ), MTA , LOC101265917 ( AGP5 ), AVT6A (BP), MT4 , POD51 , UBC24 , GABAT , AROGP3 , and NQO1 . \n UBC24 , which was identified as PHO2 and negatively regulates Pi uptake, was downregulated at 7 and 30 dpi upon AM inoculation. This could indicate a role of AM in Pi uptake in plant roots and promote resistance to fungal infections ( Grennan, 2008 ; Val-Torregrosa et al., 2022 ). Although the AVT6A gene plays an important role in abiotic stress tolerance ( Ayyappan et al., 2024 ), its expression was downregulated under AM inoculation at 7 and 30 dpi, suggesting that AM has a negative effect on AVT6A phosphorylation status in tomato roots ( Tahmasebi et al., 2023 ). The positively affected gene was AROGP3 , which encodes a JA-regulated polygalacturonase gene with a non-catalytic subunit; it is believed to introduce AM into the plant upon long-term inoculation (30 dpi) by producing aromatic amino acids such as phenylalanine and tyrosine, which are essential for AMF growth and development ( Bergey et al., 1999 ). In addition, the GABAT gene is a key enzyme involved in the degradation of γ-aminobutyric acid (GABA) to succinic semialdehyde and is strongly expressed in AM-inoculated roots, especially at 30 dpi. This is considered to positively influence ion transport in tomato plants and consequently plant growth ( Ramesh et al., 2015 ; Hu and Chen, 2020 ). Positive regulation of AGP in the AM-inoculated roots is thought to play an important role in short-term (7 dpi) inoculation, especially to recognize and attract root colonization by AM-forming mycorrhizae ( Nguema-Ona et al., 2013 ). The POD gene is an antioxidant responsible for the removal of H 2 O 2 and plays a crucial role in various functions. When plant roots are inoculated with AMF, they induce the roots to produce POD, which catalyzes the formation of lignin and oxidative phenolics; this process is essential for plant defense mechanisms and structural support ( Song et al., 2010 ). We hypothesize that NQO1 plays a role in haustorium development and was positively regulated at 7 and 30 dpi to support the interactions between plant roots and AMF ( Bandaranayake et al., 2010 ). The MTA and MT4 genes were positively expressed under AM inoculation, confirming their roles in the antioxidant functions of the plant in response to exogenous elicitors and antioxidant inducers ( Dabrowska et al., 2021 ). In addition, we hypothesize that the MT4 gene promotes accumulation of essential minerals and phytoprotective genes in tomato root upon inoculation with AMF ( Kısa et al., 2016 ). In the present work, the PPI network analysis of the altered proteins showed the interaction of a particular protein in the AM-inoculated plants at the two stages (7 and 30 dpi) with 12 other proteins ( Figure 7 ). AFRR , GME1 , and APX are involved in the ascorbic acid metabolic pathway, confirming the role of AMF colonization in increasing the ROS levels ( Ruiz-Lozano et al., 2012 ). CYP , GAPC2 , and CAM2 are involved in metabolic pathways required for root and soil colonization ( Mitra et al., 2004 ; Handa et al., 2015 ; Keymer et al., 2017 ). The transcription of genes involved in the sterol pathway, such as CYC1 and HMGR , suggests that AMF enable transfer of lipids from the plant host, which is a hallmark of its obligate biotrophy, as R. irregularis does not possess the genes encoding cytosolic fatty acids ( Keymer et al., 2017 ). The observation of the expression patterns of CDKB2 and PCNA as genes related to cell division and cell cycle confirm the roles of AM in supporting plant roots during plant development and stress tolerance ( Boyno et al., 2023 )."
} | 3,139 |
30304798 | PMC6213289 | pmc | 7,993 | {
"abstract": "In order to prepare parabolic superhydrophobic materials, copper meshes were used as the substrate and ultrasonic etching and oxidative corrosion were carried out with FeCl 3 solution and H 2 O 2 solution, respectively, and then the surface was modified with stearic acid (SA). The topological structure and surface wettability of the prepared mesh were characterized by fluorescence microscope, scanning electron microscopy and contact angle measurement. Finally, the as-prepared copper meshes were applied to oil-water separation. The results showed that the micro-nano-mastoid structure on the surface of the copper mesh was flaky bulges, forming a rough structure similar to a paraboloid. When the oxidative corrosion time of H 2 O 2 was 1 min, it is more beneficial to increase the hydrophobicity of the surface of the copper mesh and increase the contact angle of water droplets on the surface of the membrane. Additionally, based on superhydrophobic materials of the parabolic copper mesh, the static contact angles of the water droplets, engine oil and carbon tetrachloride with the surface were approximately 153.6°, 5° and 0.1°, respectively and the sliding angle of the water droplets with the surface were approximately 4.9°. The parabolic membrane was applied to discuss the separation efficiency of different oils with deionized water and the separation efficiency was obtained as benzene > carbon tetrachloride > oil > machine oil. Therefore, based on the research, the parabolic superhydrophobic material has good efficiency of oil-water separation.",
"conclusion": "4. Conclusions In a summary, by controlling etching and oxidation conditions, the rough micro-nano structure can be obtained on the surface of the copper mesh and finally a parabolic superhydrophobic membrane was obtained by modifying the SA with low surface energy. The as-prepared copper mesh was used for the oil-water separation of different oils and the oil-water separation efficiency was benzene > carbon tetrachloride > cooking oil > engine oil, therefore, the parabolic superhydrophobic membrane has a good oil-water separation effect.",
"introduction": "1. Introduction In recent years, inspired by the hydrophobic phenomenon of plants and insects, scholars all over the world pay attention to superhydrophobic materials because of their wide application prospects [ 1 , 2 , 3 , 4 ]. In order to obtain hydrophobic materials with superior performance, researchers constantly explore the prediction theory of superhydrophobic model and the preparation technology [ 5 , 6 , 7 ]. In terms of model theory prediction, based on the classical Young’s equation [ 8 ], the Wenzel equation [ 9 ] and the Cassie-Baxter model [ 10 ], in order to explore which structure is more conducive to the superhydrophobicity of the surface, Patankar et al. [ 11 ] firstly constructed a cylindrical groove model with micro-nano composite structure and theoretically analyzed the Wenzel equation of the primary structure and the Cassie-Baxter equation under steady state. Yamamoto [ 12 ] et al. generalized the microscopic surface into a column for analysis. For modeling and infiltration analysis, Marmur et al. [ 13 ] divided the rough surface into a cylindrical structure, a truncated cone structure, a parabolic structure and a hemispherical structure, which verified that the parabolic was most advantageous. Zhang et al. [ 14 ] designed the simulated surface microstructure of different models and found that the sinusoidal microstructure similar to the parabolic structure was the most suitable morphology for making superhydrophobic surfaces. It can be seen from the perspective of model analysis that the parabolic structure was the most favorable for forming superhydrophobic materials but there were few reports on how to prepare parabolic superhydrophobic materials. Therefore, the preparation of parabolic superhydrophobic materials was worthy of further exploration. Based on model predictions, the correct choice of substrate materials was the basis for achieving superhydrophobic surface. In the existing research of superhydrophobic substrate materials, copper is one of the most attractive oil-water separation materials due to its ductility, low density, high specific surface area, mechanical strength, recyclability and environmental friendliness [ 15 , 16 , 17 ]. Cao et al. [ 18 ] found that a 1-dodecanethiol film modified with dopamine was prefabricated on a coarse mesh by a simple impregnation process to obtain a superhydrophobic copper mesh film with a micro-nano layered structure. Rong et al. [ 16 ] reported fabrication of a superhydrophobic copper foam with high oil-water separation efficiency, which can serve both as oil absorption material and oil-water separation membrane. In addition, low surface energy is also a key factor in achieving superhydrophobicity, which can be achieved by the introduction of chemical modifiers such as coupling agents and stearic acid. Li et al. [ 19 ] constructed a superhydrophobic PVDF/SA nanofiber membrane that maintained high separation efficiency and corrosive treatment after 10 cycles. Fan et al. [ 20 ] found that CuO film modified with SA became superhydrophobic and had good corrosion resistance. Maryam Khosravi et al. [ 21 ] used a stencil to modify the surface with SA as the substrate and found that it was reused many times during the oil-water separation process without reducing its separation ability. Therefore, the study on the preparation of superhydrophobic materials with copper as the substrate and SA as the modifier have a broad development prospect. The preparation of superhydrophobic materials was not only related to the substrate material and modifier but also to the construction method. Various methods, including electrospinning [ 22 , 23 ], chemical vapor deposition [ 24 , 25 ] and one-step coating [ 26 , 27 ] have been reported for constructing superhydrophobic membranes with a micro–nano structure [ 28 ]. Most of these methods focused on the specific coating technology of hydrophobic materials. Li et al. [ 19 ] reported that a facile electrospinning technology was utilized to prepare polyvinylidenefluoride (PVDF)/(SA) nanofibrous membranes. This method is, however, time-consuming and difficult to control accurately. Wu et al. [ 29 ] studied the preparation of superhydrophobic surfaces by spraying, dipping, painting and so forth, and used nanoparticles (i.e., SiO 2 or TiO 2 ) and epoxy resin as raw materials to simulate ordinary household coatings, which was easy to operate but the surface stability and the durability of the modification layer were often too poor. In order to improve these problems, wet etching [ 15 , 30 , 31 ] technology has been developed in recent years. Cao and co-workers [ 32 ] fabricated super-wettable surfaces on copper mesh and copper foam by etching with H 2 O 2 and HNO 3 and then immersed them into AgNO 3 solution, the Cu@Ag films were formed on copper substrates and exhibited superhydrophilicity and underwater superoleophobicity. This method is more effective for constructing a surface roughness structure, however, the etching solution is often harmful to the environment. Similar research results were [ 33 , 34 ] and so forth, they only paid attention to the roughness of the surface of the material and there was very little analysis on how to etch a specific morphology. Therefore, the technique used to etch a specific shape on the surface and then apply it as a superhydrophobic material was the most interesting issue. In this study, a 200 copper mesh was used as the substrate, firstly, the copper mesh was ultrasonically etched to roughen the surface and then ultrasonic oxidized and etched to realize the parabolic specific morphology. The copper mesh was modified with SA to obtain the surface of the copper mesh similar to the parabolic rough structure. Then, the superhydrophobic surface was characterized by fluorescence microscopy, scanning electron microscopy (SEM) and wettability. Finally, the parabolic copper mesh was applied to oil-water separation and the relationship between superhydrophobic surface contact angle and oil-water separation efficiency was studied for different density oils.",
"discussion": "3. Results and Discussion 3.1. Fluorescence Microscope Characterization The morphology of copper mesh before and after etching and oxidation under fluorescence microscope was shown in Figure 1 . As can be seen that the shape of the copper mesh had a significant change. The size of the untreated copper mesh was uniform and the thickness of the copper mesh was basically the same. However, the copper mesh after etching and oxidation was clearly narrowed, indicating that part of the mesh of the copper mesh was subjected to cavitation impact through the ultrasonic etching of the acidic etchant FeCl 3 solution, the surface of the copper meshes had been deformed and the materials had been eroded. That is, the cavitation erosion occurred. This ultrasonic cavitation and cavitation can enhance the etching effect and form a three-dimensional rough surface on the surface of the copper mesh, which is a necessary condition for oil-water separation [ 35 ]. Therefore, the treatment of copper meshes with FeCl 3 and H 2 O 2 solutions was crucial for the formation of superhydrophobic surfaces. Figure 2 showed that the surface morphology of copper mesh after etching and oxidation without modification and modification of SA. By comparing Figure 2 a and Figure 2 b, it can be seen that after the copper mesh was modified with SA, the pore diameter of the mesh became significantly smaller, which indicated that SA was successfully attached to the surface of the copper meshes and the copper mesh surface became rough, indicating that the surface roughness of the copper mesh was increased by modification with SA [ 19 , 36 ]. However, the meshes were still present, so the separation effect will not be affected when the screen hole was blocked by SA. 3.2. SEM Characterization Owing to the fact that the fluorescence microscope can only roughly see the difference in surface morphology of the copper mesh before and after treatment. Thus, we used a SEM to characterize the sample for further observed the microstructure of the copper mesh surface. 3.2.1. Effect of Oxidation Time of Hydrogen Peroxide After modification of SA by copper mesh with different oxidation time, the surface morphology was characterized by SEM, as shown in Figure 3 . It can be clearly seen from the Figure 3 a that when the oxidation time was 0, the surface of the copper mesh after the SA modification did not show large irregular mastoid structure. However, when the oxidation time of hydrogen peroxide was 1min, as shown in Figure 3 b, the surface of the copper mesh was scattered with irregular sheet mastoid which were basically consistent in height. Meanwhile, Rong et al. [ 1 ] also studied the amplified SEM images in the three-dimensional copper foam, which showed papillae with a flower-like structure. These micro-nano-blocked mastoids of different sizes formed mastoid clusters and there were obviously spacing and gap between the blocky mastoids and the mastoid clusters, which can store air. Therefore, this can form the air cushion on the surface of the copper mesh to prevent contact between the water droplets and the solid surface, which increased the CAs of the water. With the increase of oxidation time, as shown in Figure 3 c,d, it can be seen that the mastoid clusters on the surface of the membrane showed a large area of massive structure and the spacing and gap between the mastoids and the mastoid clusters became smaller. When the oxidation time of hydrogen peroxide was 5 min, the mastoid clusters on the surface of the membrane were more compact. At this point, the space between the membrane surface can be used to store air was small, which was not conducive to improving the hydrophobicity of the membrane surface. According to the study of the hydrophobic surface structure of the lotus leaf and the Cassie-Baxter theory, the air cushion on the surface of the membrane was essential for obtaining a hydrophobic surface [ 10 , 16 , 37 , 38 ]. From the comprehensive analysis of Figure 3 , it was more beneficial to obtain a rough structure similar to the geometry of the lotus leaf surface when the oxidation time of hydrogen peroxide was 1 min. Therefore, the selection of hydrogen peroxide oxidation time was 1 min. 3.2.2. Surface Morphology of Copper Mesh Not Modified by SA Figure 4 a,b were SEM images of untreated and etching and oxidation copper meshes with the 800× without SA modification, respectively. Comparing Figure 4 a with Figure 4 b, it showed that the surface morphology of the etching and oxidation copper mesh had changed significantly. The surface of the untreated copper mesh was smooth, while the etching and oxidation copper mesh was basically unchanged in shape and the surface of the copper mesh became very rough [ 39 ]. In addition, the diameter of copper mesh wire was slightly narrowed and the corresponding mesh size was slightly increased, which was basically consistent with the observation under the fluorescence microscope, indicating that copper mesh was affected by ultrasonic cavitation and cavitation after ultrasonic etching of acidic etching agent FeCl 3 solution and even rough morphology was formed on the surface of copper mesh. 3.2.3. Surface Morphology of Copper Mesh Modified with SA The untreated copper mesh and the morphology of the copper mesh under different treatment processes were shown in Figure 5 . As can be seen from Figure 5 a, the surface of the untreated copper mesh in the SEM at 5000× was smooth. And after ultrasonic etching by FeCl 3 solution, the surface of the copper mesh became rough, showing a microporous structure of different sizes as shown in Figure 5 b. Comparing Figure 5 b and Figure 5 c, it can be concluded that after H 2 O 2 solution oxidation etching, there was a large number of mastoid structures on the surface of the copper mesh and the surface roughness was obviously improved. Figure 5 d showed that the surface of the copper mesh was completely covered with SA after modification with SA, compared with Figure 5 c, the surface mastoid structures were interlaced and clustered to form a large number of mastoid clusters, which further improved the surface roughness of the copper mesh. Figure 6 showed the surface morphology of the etching and oxidation copper mesh after modification of SA. Under 200× electron microscopy, as shown in Figure 6 a. After the copper mesh was modified with SA, the surface of the copper mesh substrate was uniformly adhered by SA and it was apparent that the surface after the modification was a rough surface, moreover, the mastoid height was basically the same. By contrast, a 5000× SEM was shown in Figure 6 b. The micro-nano-mastoid structure on the surface of the copper mesh modified by SA was a sheet-like bulge and adjacent sheet mastoids were interlaced and clustered to form a mastoid cluster similar to a paraboloid. In addition, there were clear spacing and gap between the mastoid clusters and these voids can capture a large amount of air, which had a significant impact on the hydrophobic properties of the surface. Therefore, it can be concluded that after modifying the SA, the copper mesh can form a rough structure similar to a paraboloid. In order to further prove that the surface morphology of the copper mesh was parabolic, the microstructure of the copper mesh surface in the SEM image was generalized, as shown in Figure 7 . It can be seen that the fitting curve obtained from the SEM image was parabolic and the correlation coefficient was greater than 0.80 (as shown in Table 1 ). According to the absolute value of a in Table 1 , the opening of the first parabola was the largest and the fifth was the smallest, it meant that a parabolic micro-nano structure of various sizes was formed on the surface of the copper mesh. Therefore, the copper mesh modified by SA can form a rough structure similar to a parabolic structure. 3.3. Wettability Characterization The CAs of a droplet on a solid surface a direct measure of its wetting performance. For lyophobic surfaces, the droplets were not easily spread and appeared spherical on the surface, however, for lyophilic surfaces, the droplets were easily wet the surface and spread out. The wetting behaviors of water-oils on the as-prepared copper meshes were evaluated by the contact angle measurement. The CA of the water droplets on the untreated copper mesh and the as-prepared copper mesh were shown in Figure 8 a,b. The untreated copper mesh was hydrophilic and the CA of water droplets with the surface was 86.3°, however, the static contact angle of the water droplets with the surface of the as-prepared copper mesh was measured to be 153.6°. That is, the surface of the prepared copper mesh which had a parabolic rough structure was a superhydrophobic surface. Cao et al. reported that the DDT-modified copper mesh showed a superhydrophobicity and the CA of water droplets with the surface was 152° [ 32 ]. Therefore the hydrophobic properties of the copper mesh modified with SA had certain advantages. It is well known that the static contact angle and the dynamic contact angle (sliding angle) are important indicators for measuring the wetting properties. When the static contact angle is greater than 150° and the dynamic contact angle (sliding angle) is less than 10°, it can be proved that the materials have good superhydrophobic properties. The diagram of water droplets sliding on the inclined surface was shown in Figure 9 . It can be seen that when the measuring device is inclined at a certain angle, the water droplet can freely slide down along the inclined surface from the Figure 9 . After five tests, the critical angle (sliding angle) of the water droplets slipped instantly was measured to be approximate 4.9°. Therefore, it can be concluded that the copper mesh with a parabolic structure is superhydrophobic. Analogously, the CAs of the engine oil and carbon tetrachloride on the untreated copper mesh and the as-prepared copper mesh were measured, as shown in Figure 10 a–d. Among them, ab and cd showed the wettability of engine oil and carbon tetrachloride on the surface of the two kinds of membranes, respectively. After taking the average of three measurements, the CA of the engine oil on the surface of the untreated copper mesh was measured to be approximate 60.2°, as shown in Figure 10 a, indicating that the copper mesh membrane was inherently lipophilic [ 1 , 40 ]. Figure 10 b showed that the modified copper mesh surface had a super-lipophilic property and the CA with the engine oil was approximate 5°, which indicated that the copper mesh with the rough parabolic structure was superhydrophobic surfaces. Because carbon tetrachloride had a much lower viscosity than engine oil, carbon tetrachloride was immediately spread on the surface of the prepared copper mesh in the experiments and the CA that was approximate 0.1° was obtained as shown in Figure 10 d. Analogously, the CA of the untreated copper mesh surface was approximate 67.4°, as shown in Figure 10 c. Comparing Figure 10 c with Figure 10 d, it can conclude that the as-prepared copper meshes with the parabolic microstructure was super-lipophilic when the oil droplet was carbon tetrachloride. Comparing Figure 10 b and Figure 10 d, because the oil type was different (i.e., oil/carbon tetrachloride, light oil/heavy oil), the viscosity coefficient was different, resulting in different CAs of oil droplets on the surface of the material. In summary, by measuring the wettability of water droplets, engine oil and carbon tetrachloride on two kinds of membranes, it was shown that the microscopic morphology of the surface of the material directly affected the CAs, moreover, the hydrophobic and lipophilic properties of the material affected by its CAs. So, the copper mesh surface with the microstructure of the paraboloid was the superhydrophobic surface. 3.4. Determination of Oil/Water Separation Performance Figure 11 showed the oil-water separation process after mixing 30 mL of carbon tetrachloride with 30 mL of deionized water. In the oil-water separation process, the density of carbon tetrachloride was greater than that of water, so the blue deionized water firstly entered the separation device through the upper glass tube. As mentioned above, the as-prepared membrane was superhydrophobic, the deionized water was all the time blocked on the membrane, red carbon tetrachloride, meanwhile, through the blue deionized water reached the surface of the membrane and the red carbon tetrachloride penetrated downward through the membrane under the action of gravity and capillary force, then gradually flowed into the beaker below the device, thereby the separation of oil-water was achieved. Analogously, the oil-water separation experiment of benzene was carried out. 3.4.1. Separation Efficiency of Different Oil-Water Mixtures The density of engine oil and cooking oil is less than that of water and the viscosity of engine oil and cooking oil is much higher than the carbon tetrachloride and benzene, thus, it was not effective to separate the engine oil and the cooking oil using the above device. In order to accurately measure the separation efficiency of the copper mesh for different oil-water mixtures, the oil-water separation experiment was carried out by using a self-made simple oil-water separation device. Taking the cooking oil as an example in Figure 12 . As shown in Figure 12 a, the oil-water mixtures prepared from 4 mL of engine oil and 2 mL of deionized water in a syringe and the prepared membrane was placed on top of the small beaker to completely cover the mouth of the beaker and the small beaker was placed in the center of the large culture dish. After shaking the mixture of oil and water in the syringe, then gradually squeezed the syringe and slowly dripped the mixture from the needle to the membrane, as shown in Figure 12 b. Due to gravity and capillary forces, as well as the superlipophilic properties of the copper mesh, the oil droplets gradually flowed through the membrane into the small beaker. However, because of the superhydrophobicity, water was trapped above the membrane, as shown in Figure 12 c,d. The process of oil-water separation in the whole process was approximately 4 min. Figure 12 e,f showed that after finishing separating the oil-water, the water that trapped above the copper mesh was poured into a large petri dish and almost no blue residue was observed on the membrane. According to the above methods, the mixtures of engine oil, benzene, carbon tetrachloride and deionized water were separated by oil-water separation experiment and the oil-water separation efficiency of these four different oils was measured, respectively. The results were shown in Figure 13 . It can be seen from Figure 13 that the separation of benzene, carbon tetrachloride, cooking oil and engine oil by the prepared parabolic morphology copper mesh was more than 91% and even the separation efficiency of benzene and carbon tetrachloride mixture was above 97%. Moreover, the separation efficiency of the oil-water mixture of cooking oil and engine oil was clearly smaller than the benzene and carbon tetrachloride. It can be seen that due to different oil types and different parameters such as viscosity coefficient, thus, the CAs of the oil droplets on the surface of the membrane were different and the separation efficiency was also different [ 21 , 41 , 42 ]. Hereby a comparison of the copper mesh prepared from different modified materials and methods for oil water separation is summarized in Table 2 . And the separation efficiency and oil-water mixtures were shown in Table 2 . It can be seen from the Table 2 that the parabolic superhydrophobic material prepared in this study had certain advantages for oil-water separation and the etching method was more advantageous than other methods. In addition, different modified materials, methods and oils have an effect on the separation efficiency. 3.4.2. Relationship between CAs and Oil-Water Separation Efficiency The oil-water separation efficiency is not only related to the viscosity coefficient of oil droplets but also closely related to the contact angle. For this reason, the separation efficiency of different oils and CA of the copper mesh with the rough shape of the paraboloid were experimentally determined and the separation efficiency was determined by a needle tube experiment, the whole oil-water separation process after mixing 4mL benzene with 2 mL deionized water lasted approximately 15 s and the separation process of the same dose of carbon tetrachloride, cooking oil, engine oil took about 20 s, 4 min 5 s, 5 min 44 s, respectively and the average rate of decline of oil droplets was engine oil < cooking oil < carbon tetrachloride < benzene, the aforementioned oil-water separation efficiency was benzene > carbon tetrachloride > cooking oil > engine oil, therefore, the oil and water separation efficiency was positively correlated with the average flow rate of oil droplets. The contact angle of the oil with the membrane was benzene (0°) < carbon tetrachloride (0.1°) < cooking oil (4.6°) < engine oil (5°), which can be obtained: (a) The smaller the contact angle of the surface of the membrane, the larger the average flow rate of the oil droplets, the conclusion verified the rule that the CAs was positively correlated with the average flow velocity of water droplets and negatively correlated with the average flow velocity of oil droplets by Zhang [ 47 ]. (b) The smaller the CAs of the surface of the membrane, the higher the separation efficiency of oil-water [ 48 ]."
} | 6,473 |
37938684 | PMC9723709 | pmc | 7,994 | {
"abstract": "Microbes associate in nature forming complex communities, but they are often studied in purified form. Here I show that neighbouring species enforce the re‐distribution of carbon and antimicrobial molecules, predictably changing drug efficacy with respect to standard laboratory assays. A simple mathematical model, validated experimentally using pairwise competition assays, suggests that differences in drug sensitivity between the competing species causes the re‐distribution of drug molecules without affecting carbon uptake. The re‐distribution of drug is even when species have similar drug sensitivity, reducing drug efficacy. But when their sensitivities differ the re‐distribution is uneven: The most sensitive species accumulates more drug molecules, increasing efficacy against it. Drug efficacy tests relying on samples with multiple species are considered unreliable and unpredictable, but study demonstrates that efficacy in these cases can be qualitatively predicted. It also suggests that living in communities can be beneficial even when all species compete for a single carbon source, as the relationship between cell density and drug required to inhibit their growth may be more complex than previously thought.",
"introduction": "Introduction The notion of pure culture is fundamental in microbiology. The isolation and growth of individual species is justified by the seemingly unpredictability [ 1 ] of in vitro assays when cultures contain multiple species. For antimicrobial sensitivity tests, this means the same microbe can show different sensitivities to the same drugs, depending on whether the cultures contain one or multiple species [ 2 – 4 ]. But microbes live in communities in nature, even when they are mostly competing for resources [ 5 ]. The question is, therefore, whether co‐existing species can threaten drug efficacy in vivo with respect to standard sensitivity tests. Current data suggest they do [ 4 , 6 – 9 ], notably reducing drug efficacy, and the result is that infections containing multiple species are more difficult to treat [ 10 – 12 ]—requiring alternatives to antibiotics altogether [ 10 , 12 ]—and waters more difficult to treat [ 13 ]. This phenomenon has been associated with biofilm formation [ 14 , 15 ], stochastic phenotypic variations of isogenic bacterial populations [ 14 , 15 ], signalling molecules [ 14 , 15 ] or enzymatic degradation of antimicrobials [ 14 , 15 ]. But questions still remain about its predictability, and why, sometimes, drug efficacy seems to increase [ 6 ]. Below I present a model that provides a passive physical mechanism to explain and, perhaps more importantly, predict this phenomenon. The model relies on Fick’s first diffusion law and aims to understand how microbial growth changes the flow of carbon and antimicrobial molecules, and, thus, influence drug efficacy against all species exposed. Note that, when surrounded by neighbours, species attain lower densities within a community with respect to that in pure culture because carbon is shared between multiple species [ 16 ]. Now, the model predicts that carbon and antimicrobial molecules will distribute evenly among all species exposed if they have similar sensitivity to the antimicrobial. The result is relatively less drug molecules per cell of each species, and therefore lower drug efficacy. If one or more species are not sensitive to the antimicrobial, carbon molecules still distribute evenly as it is an active process [ 17 , 18 ] but drug molecules are not: They flow back through diffusion from non‐sensitive species into the environment, re‐exposing those that are sensitive to the drug. Here, the model predicts relatively more drug per cell of sensitive species resulting thus in higher drug efficacy. These predictions were maintained across a range of parameter values for carbon uptake rate, carbon affinity, and biomass yield—fundamental components of the growth function in microbes [ 19 ]. To validate these predictions, I measured the efficacy of tetracycline against Escherichia coli Wyl in standard sensitivity assays using pure cultures, cultures with equal proportion of another microbe with similar drug‐sensitivity ( Salmonella typhimurium ), and cultures containing equal proportion of another microbe now tolerant to tetracycline ( Escherichia coli GB(c)). Consistently with the theoretical predictions, inhibiting Wyl required more tetracycline—with respect to pure cultures—in the presence of equally sensitive neighbours and less tetracycline in the presence of drug‐tolerant neighbours.",
"discussion": "Discussion Why do microbes form communities? It was recently [ 5 ] suggested that bacteria rarely work together based on the biological interactions between them. But fundamental physics could draw a different picture, and help answer the question. If, as my study suggests, two species growing together can tolerate tetracycline better than they would otherwise do in isolation, it is not unreasonable to hypothesise that forming such communities can help diminish the effect of toxic molecules in nature. For example, to grow closer to an antibiotic‐producing microorganism and benefit from the carbon available or tolerate chemotherapies. Or, indeed, taking over a particular niche in case of drug‐tolerant microbes without producing specialised molecules such as bacteriocins [ 39 ]. All as a by‐product of a physical law. It should be stressed that my theory is not specific to Escherichia coli and tetracycline, so I anticipate that other species and molecules obey the same principle described here. An example can be found in the interactions between tumours and bacteria, and the resulting tolerance to chemotherapies [ 8 ]; or between microbial pathogens [ 6 ]—albeit reliably predicting the outcome is still challenging [ 5 ]. Now, there are limitations in the model some of which are imposed by experimental conditions. For simplicity, the Fick’s term in Eq. 1b and c has no spatial derivative as it would be expected. However, during the experiment I shook the cultures to homogenise the distribution of tetracycline and nutrients. Here the aforementioned derivatives can be approximated as the difference in drug molecules between environment and cytoplasm. The model also assumes reliance on a single carbon source. In reality, however, I used two: Glucose as the main carbon source, and casamino acids. The latter can indeed be used as a carbon source when glucose is scarce [ 40 ]. The concentration I used (0.1% w/v) was low enough to enable the growth of the microorganisms without showing diauxic growth dynamics typical from cultures that are sustained on multiple carbon sources [ 41 ]. While speciation might indeed occur over time to avoid competition [ 42 ], is it really important to my conclusions? The model that I present here is, like all models, wrong in some ways but also useful in others. The goal of my study is to demonstrate that growing alongside other microorganisms can predictably change the efficacy of a drug used against one—or against all—of them. It is, from the model standpoint, indifferent which carbon type is used to grow so long the relative frequency of the microbes surrounding the target is sufficiently high. Figure 1 shows the changes in IC 90 are consistent regardless of the differences in growth parameters, akin to those existing between microorganisms using different carbon sources [ 43 ] or metabolic efficiencies [ 19 ]. Moreover, the model and experiment use two species mixed in the same proportion for the purpose of pinpointing the physical mechanism. In nature, communities are more complex and this phenomenon is likely to be more nuanced, if not even stronger, based on the ratio between the number of cells of the target species and its neighbours’. My contribution with this model is to demonstrate that, contrary to current belief, the change in drug efficacy derived from the growth of other microorganisms is not arbitrary. There is, at least, one simple physical mechanism behind the changes in efficacy and, therefore, the changes can be qualitatively predicted—a necessary step if competition is to be harnessed [ 5 ] and used either in the clinic or treatment of pollutants in the soil."
} | 2,059 |
27018915 | null | s2 | 7,995 | {
"abstract": "Manganese oxides are often highly reactive and easily reduced, both abiotically, by a variety of inorganic chemical species, and biologically during anaerobic respiration by microbes. To evaluate the reaction mechanisms of these different reduction routes and their potential lasting products, we measured the sequence progression of microbial manganese(IV) oxide reduction mediated by chemical species (sulfide and ferrous iron) and the common metal-reducing microbe Shewanella oneidensis MR-1 under several endmember conditions, using synchrotron X-ray spectroscopic measurements complemented by X-ray diffraction and Raman spectroscopy on precipitates collected throughout the reaction. Crystalline or potentially long-lived phases produced in these experiments included manganese(II)-phosphate, manganese(II)-carbonate, and manganese(III)-oxyhydroxides. Major controls on the formation of these discrete phases were alkalinity production and solution conditions such as inorganic carbon and phosphate availability. The formation of a long-lived Mn(III) oxide appears to depend on aqueous Mn(2+) production and the relative proportion of electron donors and electron acceptors in the system. These real-time measurements identify mineralogical products during Mn(IV) oxide reduction, contribute to understanding the mechanism of various Mn(IV) oxide reduction pathways, and assist in interpreting the processes occurring actively in manganese-rich environments and recorded in the geologic record of manganese-rich strata."
} | 381 |
21573199 | PMC3088647 | pmc | 7,996 | {
"abstract": "Bacteria move towards favourable and away from toxic environments by changing their swimming pattern. This response is regulated by the chemotaxis signalling pathway, which has an important feature: it uses feedback to ‘reset’ (adapt) the bacterial sensing ability, which allows the bacteria to sense a range of background environmental changes. The role of this feedback has been studied extensively in the simple chemotaxis pathway of Escherichia coli . However it has been recently found that the majority of bacteria have multiple chemotaxis homologues of the E. coli proteins, resulting in more complex pathways. In this paper we investigate the configuration and role of feedback in Rhodobacter sphaeroides , a bacterium containing multiple homologues of the chemotaxis proteins found in E. coli . Multiple proteins could produce different possible feedback configurations, each having different chemotactic performance qualities and levels of robustness to variations and uncertainties in biological parameters and to intracellular noise. We develop four models corresponding to different feedback configurations. Using a series of carefully designed experiments we discriminate between these models and invalidate three of them. When these models are examined in terms of robustness to noise and parametric uncertainties, we find that the non-invalidated model is superior to the others. Moreover, it has a ‘cascade control’ feedback architecture which is used extensively in engineering to improve system performance, including robustness. Given that the majority of bacteria are known to have multiple chemotaxis pathways, in this paper we show that some feedback architectures allow them to have better performance than others. In particular, cascade control may be an important feature in achieving robust functionality in more complex signalling pathways and in improving their performance.",
"introduction": "Introduction Living organisms respond to changes in their internal and external environment in order to survive. The sensing, signalling and response mechanisms often consist of complicated pathways the dynamical behaviour of which is often difficult to understand without mathematical models [1] . Considering the structure and dynamics of these signalling pathways as integrated dynamical systems can help us understand how the pathway architecture and parameter values result in the performance and robustness in the response dynamics [2] . One extensively studied sensory pathway is bacterial chemotaxis. This pathway controls changes in bacterial motion in response to environmental stimuli, biasing movement towards regions of higher concentration of beneficial or lower concentration of toxic chemicals. The chemotaxis signalling pathway in the bacterium Escherichia coli is a simple network with one feedback loop [3] which has been extensively studied and used as a paradigm for the mechanism of chemotaxis signalling networks [4] . In E. coli , chemical ligands bind to methyl-accepting chemotaxis protein (MCP) receptors that span the cell membrane and alter the activity of a cytoplasmic histidine kinase called CheA. When attractant ligands stimulate the chemotaxis pathway by binding to MCP, there is a decrease in the autophosphorylation rate of CheA; conversely, repellent binding or lack of attractant binding increase CheA autophosphorylation activity. CheA, when phosphorylated, can transfer the phosphoryl group to two possible response regulators: CheY and CheB. CheY-P (where ‘-P’ denotes phosphorylation) interacts with FliM in the multiple E. coli flagellar motors resulting in a change in the direction of rotation of the motor. At the same time, a negative feedback loop allows the system to sense temporal gradients and react to a wide ligand concentration range: the MCP receptors, which are constantly methylated by the action of a methyltransferase CheR, are de-methylated by CheB-P. This negative feedback loop restores the CheA autophosphorylation rate and the flagellar activity to the pre-stimulus equilibrium state [5] , [6] . Describing this pathway mathematically as a dynamical system can be facilitated by using tools from control theory. For example, it has been shown that the adaptation mechanism in the E. coli model [7] , [8] is a particular example of integral control, a feedback system design principle used in control engineering to ensure the elimination of offset errors between a system's desired and actual signals, irrespective of the levels of other signals [9] . Many species have chemotaxis pathways that are much more complicated than that of E. coli \n [10] , [11] , either containing chemotaxis proteins not found in E. coli , e.g. in the case of Bacillus subtilis \n [12] ; or containing multiple homologues of the proteins found in the E. coli pathway, as in the case of Rhodobacter sphaeroides \n [11] , [13] . Furthermore, in R. sphaeroides there are two receptor clusters containing sensory proteins which localize to different parts of the cell, one located at the cell pole and the other in the cytoplasm [14] . Although the purpose of the two clusters is unclear, in vitro phosphotransfer experiments [15] , [16] show that the CheA homologues located at the two clusters can phosphotransfer to different CheY and CheB homologues: at the cell pole CheA 2 -P phosphotransfers to CheY 3 , CheY 4 , CheY 6 , CheB 1 and CheB 2 , while at the cytoplasm CheA 3 A 4 -P phosphotransfers to CheY 6 and CheB 2 . The two methylesterase proteins, CheB 1 and CheB 2 , which are homologues of CheB in E. coli , are responsible for the adaptation mechanism in R. sphaeroides \n [13] , [17] . Past localization studies have shown that CheB 1 and CheB 2 are found diffuse throughout the cytoplasm [14] . This is different to E. coli where the CheB protein is localized at the cell pole, and could potentially mean that the two proteins de-methylate either receptor cluster [14] . As a system featuring an adaptation mechanism similar to that in E. coli , but with multiple homologues of the E. coli chemotaxis proteins, it is useful to examine the R. sphaeroides chemotaxis pathway from a control engineering perspective. In this way, we can suggest structures for the R. sphaeroides chemotaxis pathway that integrate the control mechanisms thought to be responsible for adaptation in E. coli along with the possible feedback architectures that arise from the dual sensory modules present in R. sphaeroides . The relative evolutionary advantages of the different architectures can then be compared from both control engineering and biological points of view. The fact that there are two endogenous ‘measurements’ available to the feedback mechanism (CheB 1 -P and CheB 2 -P) which can be used to regulate two signals (CheA 2 and CheA 3 A 4 ) makes the whole chemotaxis feedback pathway a multi-input, multi-output control system (as opposed to possessing only one CheB and one CheA as in the E. coli models [7] , [18] ). This introduces extra degrees of freedom in the feedback control mechanism of the system and, thus, the potential for better regulation. However, the different conceivable connectivity configurations between the two CheB-P proteins and the two receptor clusters actually correspond to different feedback control architectures, each with different properties. Some of these configurations, as will be demonstrated, could allow the bacterium to integrate information from both internal and external sources and to function more efficiently, e.g., by varying how strongly it reacts to external attractants depending on its internal state. At the same time, the additional receptor cluster not found in E. coli has the potential of introducing extra sources of performance degradation such as noise (both intrinsic and extrinsic) and variations in quantities internal to the cell such as protein copy numbers and phosphorylation rates: the feedback signalling pathway may be required to remedy this, and in this regard, some of these feedback architectures perform better than others. One of the different pathway configurations that is possible in this system has similarities to a feedback architecture commonly found in engineering control systems termed cascade control \n [19] , which is usually employed when the process to be controlled can be split into a slow ‘primary’ sub-process ( in Figure 1 ) and a faster, secondary sub-process ( in Figure 1 ). Without the internal feedback shown dashed in Figure 1 the primary module maintains a set-point for the secondary module to follow and the output of the secondary module is fed back to the primary. A cascade control design places an additional feedback loop around the fast secondary process (shown dashed). This has been known to improve system performance in several ways: it reduces the sensitivity of the output of the secondary module to changes in the parameters (thus improving robustness), it attenuates the effects of disturbance signals, it makes the step response of the control system to inputs and disturbances less oscillatory and, since the secondary process is relatively fast, the effects of unwanted disturbances are corrected before they affect the system output. Including this additional internal feedback also allows the control system designer more flexibility in increasing the feedback gain to achieve higher bandwidth and faster system responses without losing stability. In fact, cascade control is employed as a design principle in several engineering systems such as aircraft pitch control and industrial heat exchangers (see Text S1 for further details). 10.1371/journal.pcbi.1001130.g001 Figure 1 A cascade control system. The subsystem is slow relative to . Cascade control involves placing a negative feedback loop (dashed line) around the fast secondary module. This scheme helps reduce the sensitivity of the system's output to uncertainties in the subsystems and . In our previous work [20] , we used a model invalidation technique to arrive at a possible pathway architecture that allows the R. sphaeroides chemotaxis system to convey, via a signalling cascade, sensed changes in ligand concentration outside the cell to the flagellar motor. In that model, proteins CheY 3 -P and CheY 4 -P act together to promote autophosphorylation of CheA 3 A 4 (schematically illustrated in Figure 2(A) ) whilst CheY 6 -P binds with the FliM rotor switch to increase the frequency of motor switching (and hence reduce the motor rotation frequency). This stimulation of CheA 3 A 4 need not be a direct interaction [20] . 10.1371/journal.pcbi.1001130.g002 Figure 2 Chemotaxis in R. sphaeroides . (A) The chemotaxis pathway in R. sphaeroides as currently understood, including the forward chemotaxis pathway previously proposed [20] . MCP: transmembrane methyl accepting chemotaxis protein, Tlp: cytoplasmic methyl accepting chemotaxis protein, A: CheA histidine protein kinase, W: CheW a linker protein between receptors and CheA, Y: the response regulator CheY, B: the response regulator CheB, R: the methyltransferase CheR. P indicates a phosphoryl group. The number in subscript denotes one of the multiple homologues in R. sphaeroides . The flagella motor is shown at the right of the figure. (B) The possible de-methylation feedback structures for the phosphorylated proteins CheB 1 -P and CheB 2 -P in R. sphaeroides . Each possible connection is denoted by a (red) thick solid, dashed or dotted line. Possible models involve combinations of these four lines. Interactions from the phosphotransfer network are shown in (black) thin dashed arrows, receptor activation/de-activation is denoted by (black) thin solid lines. In this paper, we assume that the chemotaxis pathway has the same forward signalling pathway of [20] and then suggest four plausible interconnection structures for the feedback pathway between the two CheB-P proteins and the two receptor clusters. Following this, we present the results of experiments that are used to invalidate all but one of these structures. We then discuss the results of in silico experiments that highlight the differences in chemotactic performance between the different models with particular focus on the robustness of chemotaxis to parametric variations in the chemotaxis pathway and noise [21] , [22] . Using analytical techniques from control theory, we demonstrate that the model not invalidated by our experiments is structurally similar to the cascade control architecture, and we use the structural properties of this interconnection, which are commonly used to reduce the effects of uncertainty and disturbances in various engineering applications, to explain the robustness features of the suggested model.",
"discussion": "Discussion From the designed experiments performed, it was possible to invalidate all models but Model III. This suggests that the feedback in the chemotaxis system could occur in an asymmetric fashion. That is, CheB 1 -P may only interact with the membrane signalling cluster whilst CheB 2 -P interacts with both clusters. It is likely that the two chemotaxis pathways initially evolved independently and then became part of the same organism by horizontal gene transfer. Thus one would possibly expect either full connectivity or complete isolation of the two pathways until a further mutation occurs. Understanding the outputs of the designed experiments \n R. sphaeroides has a more complex chemotaxis network than E. coli and the multiple receptor clusters and multiple feedback pathways mean that mutants will not always have an intuitive phenotype. For example the Δ cheB1 mutant does not have the loss of response phenotype one would expect from a direct comparison with the E. coli system. We can try to understand why Δ cheB1 has a steady state at −8 Hz by looking at the structure of the model we have been unable to invalidate, and the reason is as follows: CheB 1 , CheB 2 and CheY 6 (along with CheY 3 and CheY 4 ) each compete for phosphoryl groups from CheA 2 -P. CheB 1 is present in relatively large copy numbers and CheB 1 -P has negligible degradation rate (see Table 1 ). When present, CheB 1 ‘stores’ a large proportion of phosphoryl groups. When absent, the competition for phosphoryl groups from CheA 2 -P remains between CheB 2 , CheY 6 , CheY 3 and CheY 4 . The rate of phosphorylation of CheY 6 by CheA 2 -P is relatively small, CheY 6 -P receiving most of its phosphorylation from the CheA 3 A 4 -P complex. Therefore deleting cheB 1 shifts the equilibrium of the system so that a higher proportion of the phosphoryl groups from CheA 2 -P go to CheY 3 , CheY 4 or CheB 2 . The increase in CheY 3 -P and CheY 4 -P results in a stronger negative feedback to the cytoplasmic cluster, and the steady-state amount of active receptors at the cytoplasmic cluster is therefore less in the case of Δ cheB 1 . The consequence of this is that the main source of phosphorylation for CheY 6 -P, which is CheA 3 A 4 -P, is reduced, and hence the level of CheY 6 -P is reduced. The stopping frequency is consequently reduced. Therefore, rather than Δ cheB 1 leading to a loss of response to stimulus, the result of this deletion is a shift in the steady state to a high rotation frequency. 10.1371/journal.pcbi.1001130.t001 Table 1 Model parameters. Reaction Parameter(s) Value(s) (R 1 ) \n \n \n 0.03 s −1 \n (R 2 ) \n \n \n 0.035 (µM s) −1 , 0.01 (µM s) −1 \n (R 3 ) \n \n \n 0.065 (µM s) −1 , 0 (R 4 ) \n \n \n 0.004 (µM s) −1 , 0 (R 5 ) \n \n \n 0.0006 (µM s) −1 , 0 (R 6 ) \n \n \n 0.0035 (µM s) −1 , 0.01(µM s) −1 \n (R 7 ) \n \n \n 0 (R 8 ) \n \n \n 0.08 s −1 \n (R 9 ) \n \n \n 0.02 s −1 \n (R 10 ) \n \n \n 0.1 s −1 \n (R 11 ) \n \n \n 0.015 s −1 \n (R 12 ) \n \n \n 0.1 (µM s) −1 , 0 (R 13 ) \n \n \n 0.006 (µM s) −1 , 0.07 (µM s) −1 \n (R 14 ) \n \n \n 0.02 s −1 \n CheA \n 26000 copies per cell CheY \n 1000 copies per cell CheY \n 4000 copies per cell CheA A \n 12000 copies per cell CheY \n 51500 copies per cell CheB \n 23000 copies per cell CheB \n 3000 copies per cell Relative advantages of the chemotaxis models The performance measure of Figure 6 suggests that in ascending a ligand gradient under ideal conditions the four models behave almost identically, which may be expected as they all exhibit the same output profile under a step ligand addition. At the same time, simulations of the chemotaxis models showed a difference in robustness between Model III and the other models. From an evolutionary point of view, this may suggest that Model III may have advantages in terms of the robustness of chemotactic performance with respect to the other models. These differences in performance and their implications for chemotaxis are discussed next. Sensitivity to parameter variations, noise and ligand inputs It is desirable that the chemotactic performance of the bacterium is unaffected by changes such as noise in gene expression between the expression of CheOp2 and CheOp3 and therefore the ability to filter out any parametric variations from the pathway's output would be an advantageous feature. The pathway's primary output and the main determinant of chemotaxis performance is the flagellar rotation frequency, which, according to the four models presented, is directly controlled by CheY 6 . It was shown that Models I and III (the latter of which was not invalidated) have a slightly lower sensitivity to variations in the copy number of CheY 6 compared to Models II and IV ( Figure 9 ). If Model III is indeed valid, such robustness could serve to better maintain the nominal steady state rotation frequency. Model III also has advantages with respect to Model I due to the CheB 2 -P feedback to the polar cluster. Strengthening this feedback to the polar cluster, which corresponds to increasing the de-methylation rate of the polar cluster by CheB 2 -P, is equivalent to increasing the gain in the linear system (1) – see Figure 10 . For the linear model (1), this reduction in sensitivity is illustrated in the Bode sensitivity plot in Figure 11(A) . From the point of view of control system design, this feedback is typically used to reduce the magnitude of the system's sensitivity function (see Text S1 ). This function is dependent on the frequency at which the system is excited and can be shown to be equal to the relative incremental change in the overall system's transfer function in response to an incremental change in the transfer function of the system's sub-modules and . If the sensitivity of the chemotaxis system is low, then the bacterium would be able to maintain its chemotactic response despite changes in the system's biological parameters. The Bode plots ( Materials and Methods ) in Figure 11(A) illustrate the effect of increasing in reducing the sensitivity function of the system (1) over most excitation frequencies. This effect can observed in the chemotaxis models in Figure 9 and Figure 13 , where it is shown that strengthening the CheB 2 feedback to the polar cluster reduces the sensitivity of the steady state rotation frequency to changes in the copy numbers of CheB 1 and CheA 2 (see Materials and Methods ). 10.1371/journal.pcbi.1001130.g013 Figure 13 Sensitivity to copy number with varying external feedback. Sensitivity of the chemotaxis steady state to random changes in copy numbers of chemotaxis proteins under different CheB 2 feedback strengths to the polar cluster. Sensitivity is measured as the ratio of the standard deviation of the steady state to the nominal steady state. Solid line: Sensitivity of the chemotaxis steady state to changes in the copy number of CheA 2 under different strengths of CheB 2 feedback to the polar cluster. Dashed line: Sensitivity of the chemotaxis steady state to changes in the copy number of CheB 1 under different strengths of CheB 2 feedback to the polar cluster. Simulation results in Figure 7 show that the switching frequency in Model III has a low sensitivity to noisy variations in ligand signals detected at the polar receptor cluster relative to the other models. Figure 8 shows the result of a further set of simulations of the four chemotaxis models in which the gain of each chemotaxis model in response to sinusoidal ligand variation detected at the two clusters is given as a function of ligand fluctuation frequency (see Materials and Methods ). The figure shows that the switching frequency in Models I and III has a relatively low gain with respect to varying ligand signals detected at the polar receptor cluster and a relatively high gain with respect to ligand variations detected at the cytoplasmic cluster. The Bode magnitude plots in Figure 12 show the frequency-dependent gain of the linear system (1) to sinusoidal ligand inputs in the case , which is structurally similar to Models I and III. These plots parallel the results of the frequency response magnitude plots of Figure 8 which, for Models I and III, show low gain in response to high frequency inputs at the polar receptor cluster and high gain in response to high frequency signals at the cytoplasmic receptor cluster. The rejection of high frequency inputs at the cell pole may be advantageous in that the flagellar switching rate is then only varied when the polar cluster senses a relatively significant ligand concentration gradient that is large in spatial extent, and remains relatively unchanged when the receptors are subject to rapid fluctuations in sensed ligand due, for example, to molecular noise at the receptor such as that simulated in Figure 7 . Although the chemotaxis model assumes that the cytoplasmic cluster input depends on the sensed ligand, it is unknown what the cytoplasmic cluster senses. In addition to the possibility that this input is a function of the sensed ligand concentration, this cluster may potentially also integrate information about the metabolic state of the cell. In this case, this signalling may well be important to chemotactic performance and the relatively high gain of Model III to inputs at the cytoplasmic cluster may suggest that this configuration would favour internal signals over external signals in terms of output. However, if chemotaxis is sensitive to such signals, it would be important that: (i) these signals are tightly controlled and relatively free of the influence of noise and (ii) the cytoplasmic cluster be insensitive to variations in its biological parameters, as sensitivity to such variations would diminish the system's ability to correctly respond to inputs to the cytoplasmic cluster. In Model III, the CheB 2 -P feedback loop around the cytoplasmic cluster could offer this reduction in the sensitivity function of this cluster to such parametric variations. This reduction in sensitivity to variations of cytoplasmic cluster parameters is illustrated in Figure 11(B) using the linear model (1) of the chemotaxis system. The figure shows that increasing the feedback gain , which corresponds to the gain of the CheB 2 -P feedback to the cytoplasmic cluster in Model III, achieves a reduction in the sensitivity of the cytoplasmic cluster. In this way, the cytoplasmic cluster remains sensitive to its inputs, as shown by the large gain at high frequency in Figure 12(B) , whilst its sensitivity to parametric variation is reduced due to the internal CheB 2 -P feedback. This effect can be observed in the chemotaxis models in Figure 14 , where it is shown that strengthening the CheB 2 feedback to the cytoplasmic cluster reduces the sensitivity of the steady state rotation frequency to changes in the copy numbers of CheA 3 A 4 and CheY 6 (see Materials and Methods ). 10.1371/journal.pcbi.1001130.g014 Figure 14 Sensitivity to copy number with varying internal feedback. Sensitivity of the chemotaxis steady state to random changes in copy numbers of chemotaxis proteins under different CheB 2 feedback strengths to the cytoplasmic cluster. Sensitivity is measured as the ratio of the standard deviation of the steady state to the nominal steady state. Dashed line: Sensitivity of the chemotaxis steady state to changes in the copy number of CheA 3 A 4 under different strengths of CheB 2 feedback to the cytoplasmic cluster. Solid line: Sensitivity of the chemotaxis steady state to changes in the copy number of CheY 6 under different strengths of CheB 2 feedback to the cytoplasmic cluster. \n Figure 8 also shows that for Model I and III, high frequency variations in the ligand concentration sensed at the polar cluster are largely filtered out before causing flagellar switching. This may suggest that the relatively slow dynamics of the polar receptor cluster enable it to function as a low pass filter, preventing any high-frequency noisy variations in the sensed concentration of ligand from being signalled through to the flagellar motor. Figure 12(A) illustrates this attenuation of high frequency polar cluster ligand inputs for the linear model (1). Chemotaxis as a cascade controlled system When combined with the forward signalling pathway which was not invalidated previously [20] , Model III has a feedback structure that corresponds to a control scheme termed cascade control . This term is used to denote a modular system that includes two feedback loops, one nested within the other. The nested loop is used to regulate a sub-process of the system whilst the ‘external’ negative feedback loop from the system output to the input is used to regulate the entire system. The measured reaction rates of the two clusters [15] , [16] are also such that the cytoplasmic cluster is faster than the polar cluster in responding to inputs, which would be required for the chemotaxis pathway to function as a cascade controlled system [19] . This modularization of the chemotaxis system into fast and slow parts mirrors the division of the cascade controlled system in Figure 1 into the slow and fast subsystems and respectively. The cascade control architecture enables the slow (primary) subsystem to fix a set-point for the fast (secondary) system and for the feedback around the secondary system to quickly regulate the secondary output in response to disturbances and variations in the secondary process [19] . This difference in speed is represented by having , and in the linear model (1). Model III also features both an ‘internal’ feedback loop nested within an ‘external’ one corresponding to the dashed and solid feedbacks in Figure 1 , respectively. These two feedbacks are manifested by the CheB 2 -P feedback that de-methylates the cytoplasmic and the polar clusters respectively. Interestingly this architecture mirrors the ability of the system to phosphotransfer, with the membrane cluster being able to phosphotransfer to and be de-methylated by both CheB proteins and the cytoplasmic cluster only phosphotransferring to CheB 2 , the protein that is able to de-methylate it. It does however raise an interesting question. Whereas CheB in E. coli is localised to the polar signalling cluster, in R. sphaeroides both expressed CheB's are found to be delocalised. Yet, only one of the CheB proteins interacts with both signalling clusters. Thus the advantage of having delocalised CheB 1 is unclear. We have shown that if the R. sphaeroides chemotaxis pathway has a cascade control architecture, this would enable robust chemotaxis in an uncertain, noisy environment, conferring a selective advantage. In E. coli , one feedback loop is used to achieve perfect adaptation and sensing of temporal gradients and because there is only one signalling cluster all signal integration occurs there. Unlike E. coli , the R. sphaeroides chemotaxis pathway with cascade control feedback provides the bacterium with two feedback loops, one embedded within the other, to adapt and to reduce its sensitivity to parameter variations and noise. The other advantage to this architecture is demonstrated by the simulations shown in Figure 12 , which illustrate that with this structure the system would be strongly sensitive to fast-changing inputs to the cytoplasmic cluster, perhaps from the metabolic state of the cell. Understanding how biological networks achieve robust functionality in the face of disturbances and noise in their internal and external environment is a key question in systems biology. Such networks can be seen as control engineering feedback systems and can be analyzed using system engineering tools in order to understand the advantages of particular internal connectivities over others. In line with this methodology, this paper first utilized a network discrimination approach [20] to construct a model of the feedback connectivity within the R. sphaeroides chemotaxis pathway, and then explained the robustness properties of that model by re-interpreting the theoretical advantages of its cascade control structure in a biological framework and comparing it to the other possible models. This suggests a mechanism by which the bacterium can achieve robust chemotactic performance despite biochemical parameter variations and noise. Given that many chemotactic systems have multiple homologues [10] it would appear that using more complex feedback architectures to improve performance may be common in chemotaxis and in other signalling pathways, raising the possibility that this methodology can be used to analyze a wide set of biological systems."
} | 7,367 |
22230956 | null | s2 | 7,997 | {
"abstract": "Many cellular processes are carried out by molecular 'machines'-assemblies of multiple differentiated proteins that physically interact to execute biological functions. Despite much speculation, strong evidence of the mechanisms by which these assemblies evolved is lacking. Here we use ancestral gene resurrection and manipulative genetic experiments to determine how the complexity of an essential molecular machine--the hexameric transmembrane ring of the eukaryotic V-ATPase proton pump--increased hundreds of millions of years ago. We show that the ring of Fungi, which is composed of three paralogous proteins, evolved from a more ancient two-paralogue complex because of a gene duplication that was followed by loss in each daughter copy of specific interfaces by which it interacts with other ring proteins. These losses were complementary, so both copies became obligate components with restricted spatial roles in the complex. Reintroducing a single historical mutation from each paralogue lineage into the resurrected ancestral proteins is sufficient to recapitulate their asymmetric degeneration and trigger the requirement for the more elaborate three-component ring. Our experiments show that increased complexity in an essential molecular machine evolved because of simple, high-probability evolutionary processes, without the apparent evolution of novel functions. They point to a plausible mechanism for the evolution of complexity in other multi-paralogue protein complexes."
} | 373 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.