pmid stringlengths 8 8 | pmcid stringlengths 8 11 ⌀ | source stringclasses 2
values | rank int64 1 9.78k | sections unknown | tokens int64 3 46.7k |
|---|---|---|---|---|---|
26109480 | null | s2 | 754 | {
"abstract": "Genome-scale metabolic network reconstructions and constraint-based analyses are powerful methods that have the potential to make functional predictions about microbial communities. Genome-scale metabolic networks are used to characterize the metabolic functions of microbial communities via several techniques including species compartmentalization, separating species-level and community-level objectives, dynamic analysis, the 'enzyme-soup' approach, multiscale modeling, and others. There are many challenges in the field, including a need for tools that accurately assign high-level omics signals to individual community members, the need for improved automated network reconstruction methods, and novel algorithms for integrating omics data and engineering communities. As technologies and modeling frameworks improve, we expect that there will be corresponding advances in the fields of ecology, health science, and microbial community engineering."
} | 238 |
37903278 | PMC10636363 | pmc | 755 | {
"abstract": "Significance In microbial communities, metabolic functions are distributed across interacting members, depending on the current environment. If the external environment changes, microbes remodel their metabolism. How are interactions consequentially adjusted? How is the growth of community members resumed after shifts in the external conditions? To understand this, we studied bacterial consortia with members coupled in their anabolic functions. We found that population growth was dominated by a minority of cells that rapidly established interactions in their microenvironment. Distributed functions can lead to population bottlenecks, constraining microbiomes’ interactions evolution. Our findings are relevant to the evolution of metabolic interactions in the face of environmental fluctuations. Fluctuations can lead to growth delays among obligate cross-feeders potentially contributing to the maintenance of metabolic autonomy.",
"discussion": "Discussion Here, we showed that metabolic dependencies can delay the growth resumption of microbial consortia after an environmental shift. We observed these growth delays in a situation where an environmental shift requires the emergence of metabolic complementation between two strains so that each strain depends on a cellular building block from its partner to resume growth. Analysis of the growth dynamics of individual cells in spatially structured environments showed that cells that were within a few micrometers from their partner resumed growth quicker than cells that were further away. As a consequence, environmental shifts induced population bottlenecks in spatially structured microbial consortia: A small number of cells contributed to most of the offspring population that emerged after an environmental shift. Since metabolic dependencies are common in microbial communities ( 4 – 10 ), we expect that such growth delays, and the resulting population bottlenecks, are a general phenomenon across different types of microbiomes. The distance to cells of the other type is a crucial determinant of regrowth times in metabolically dependent consortia. A critical distance that cannot be exceeded for cells to grow efficiently emerges from the interplay between amino acid leakage, diffusion, and uptake ( 24 ). While previous work quantified this so-called interaction range between strains of an auxotrophic consortium under constant conditions ( 24 ), the current work focuses on the emergence of metabolic coupling after an environmental fluctuation. The environmental fluctuation requires cells to synthesize enzymes for amino acid production, which is expected to dynamically affect single-cell leakage. In addition, the uptake rate is also expected to change in response to the shift in the environmental condition. As a result, the interaction range is expected to dynamically change in the course of the environmental fluctuation. Gaining a deeper understanding of how the interplay between leakage and uptake modulates the interaction range over time will greatly enhance our ability to predict the response of complex microbial communities to fluctuating environments. The focus on environmental fluctuations, which are ubiquitous in natural situations ( 26 – 31 ), highlights the importance of the spatial organization at the microscale. How can microorganisms stay close to a partner on which they metabolically depend? Cellular growth in spatially structured systems inherently leads to cell clusters, moving more and more cells away from the partner whose metabolites they need ( 24 , 45 ) This effect is exacerbated in the face of environmental fluctuations, where periods of metabolic independence, and hence cellular unmixing, are followed by periods where different strains are metabolically dependent and thus need to be close to each other. In addition, physical disturbances can rearrange cells in space and further separate mutualistic partners. How are obligate metabolic dependencies maintained in such situations? One possibility is that cells engaged in obligate cross-feeding interactions have evolved mechanisms to stay physically attached to their partner cell types, either through direct cell–cell adhesion ( 46 , 47 ) or by means of an extracellular matrix ( 48 ). Our results add an ecological dimension to the role of cell-to-cell variation in fluctuating environments. A wealth of studies has analyzed phenotypic variation in growth resumption and growth rates in clonal populations ( 37 , 38 , 40 , 41 , 49 – 51 ). Such variation can emerge from stochastic cellular processes that produce phenotypic differences between genetically identical cells ( 52 – 58 ). Cells that grow slowly or transiently cease growth are phenotypically tolerant to a range of stressors including antibiotics ( 39 , 59 – 63 ). Phenotypic variation in regrowth times and growth rates can therefore allow populations to persist periods of exposure to potentially lethal stressors ( 37 , 40 , 59 , 64 ). A high level of cell-to-cell variation can mitigate a fundamental trade-off between growth and survival in fluctuating environments: Population growth after an environmental shift is mostly driven by the minority of cells that resume growth most rapidly, while persistence in the face of external stressors depends on a minority that resumes growth most slowly ( 40 ). While cell-to-cell variation in growth resumption has so far mainly been studied in clonal populations ( 37 , 38 , 40 , 41 , 49 – 51 ), we expect that it might be even more pronounced in spatially structured microbial communities. Physiological adaptation to environmental shifts is often limited by the time needed to express new enzymatic pathways, and by the cellular resources required to synthesize these enzymes ( 65 – 67 ). If some of these resources originate in another cell type that is also undergoing physiological adaptation, this is expected to increase regrowth times. This expectation is supported by the results presented in Fig. 2 . However, regrowth times in microbial communities are not increased for all cells equally. Rather, as shown in Fig. 3 , these times strongly depend on the cellular neighborhood of each individual cell. This factor will increase variation in growth resumption in metabolically connected communities. The magnitude of the variation in growth resumption and of the resulting population bottleneck in microbial communities is therefore not an intrinsic property of community members—it rather is an emergent property of the community that depends on the metabolic interactions and on the community’s spatial organization at the microscale. This is expected to lead to pronounced history dependence of community processes in varying environments: How quickly metabolic activities of community members, and hence functions at the level of the community, are resumed after environmental shifts depending on the spatial distribution of different species in the community. The spatial distribution in turn depends on the nutrients on which community members grew in the past. Given that many microbial communities perform important ecosystem functions under fluctuating environments ( 30 , 68 – 70 ), understanding such history-dependence in distributed metabolic functions is an important goal. We expect the population bottlenecks that result from variation in offspring production and ultimately from differences in regrowth times to be relevant in spatially structured natural ecosystems. Bottlenecks reduce the genetic variability within each strain, thereby reducing the effective population size and promoting genetic drift ( 40 , 71 , 72 ). Our work suggests that in these communities, the spatial arrangement affects the bottleneck size and thus the effective population size. In other words, the effective population size often used in theoretical studies to predict the evolution of bacterial communities ( 73 , 74 ) is a function of the interactions among community members, their spatial arrangement, and the time course of the environmental conditions. The minority-driven regrowth we observed has potentially important consequences for the evolution of metabolic interactions in microbial communities. Mutants that are metabolically dependent on other species can naturally arise through loss of metabolic functions ( 7 , 15 , 21 , 75 ). The important question then is whether, and under what conditions, such mutants can spread. Our results indicate that the spread of mutants with metabolic dependencies also depends on fluctuations in the external environment. Under constant environmental conditions, one expects that microbes can be selected for taking up metabolites from other organisms rather than producing them themselves ( 15 ). This view is supported by studies on natural microbial communities (reviewed in ref. 34 ) and by experimental (e.g., ref. 21 ) and computational ( 76 ) work. Our results suggest that environmental fluctuations can change this outcome. As reported above, metabolic dependencies can lead to long growth delays after environmental shifts, given the time needed to reestablish metabolic exchange with other organisms in the same microenvironment. Given the exponential nature of microbial growth, growth delays have a strong negative impact on an organism’s the long-term growth rate ( 40 ). Ever-changing environments might therefore limit the extent to which metabolic functions become, in the course of evolution, distributed across different species. This hypothesis can be tested by comparing metabolic dependencies in microbial communities across different magnitudes of environmental fluctuations and by competition experiments between prototrophs and auxotrophs in constant and fluctuating conditions."
} | 2,440 |
39077733 | PMC11284109 | pmc | 757 | {
"abstract": "Bioelectrochemical systems offer unique opportunities to remove recalcitrant environmental pollutants in a net positive energy process, although it remains challenging because of the toxic character of such compounds. In this study, microbial fuel cell (MFC) technology was applied to investigate the benzene degradation process for more than 160 days, where glucose was used as a co-metabolite and a control. We have applied an inoculation strategy that led to the development of 10 individual microbial communities. The electrochemical dynamics of MFC efficiency was observed, along with their 1 H NMR metabolic fingerprints and analysis of the microbial community. The highest power density of 120 mW/m 2 was recorded in the final period of the experiment when benzene/glucose was used as fuel. This is the highest value reported in a benzene/co-substrate system. Metabolite analysis confirmed the full removal of benzene, while the dominance of fermentation products indicated the strong occurrence of non-electrogenic reactions. Based on 16S rRNA gene amplicon sequencing, bacterial community analysis revealed several petroleum-degrading microorganisms, electroactive species and biosurfactant producers. The dominant species were recognised as Citrobacter freundii and Arcobacter faecis . Strong, positive impact of the presence of benzene on the alpha diversity was recorded, underlining the high complexity of the bioelectrochemically supported degradation of petroleum compounds. This study reveals the importance of supporting the bioelectrochemical degradation process with auxiliary substrates and inoculation strategies that allow the communities to reach sufficient diversity to improve the power output and degradation efficiency in MFCs beyond the previously known limits. This study, for the first time, provides an outlook on the syntrophic activity of biosurfactant producers and petroleum degraders towards the efficient removal and conversion of recalcitrant hydrophobic compounds into electricity in MFCs.",
"conclusion": "4 Conclusion In this study, we have demonstrated a significant improvement in the efficiency of bioelectrochemical benzene degradation compared to previous similar studies reported in the literature. The power output reported here was 18 times higher, compared to the other benzene/co-substrate studies, and it ranged from 21 mW/m 2 to 120 mW/m 2 , highlighting the potential of MFCs in benzene remediation. Initial inoculation with pure strains probably affected the formation of anodic biofilms after activated sludge enrichment. Differences in MFC performance with the same substrates resulted from the differences in the abundance of electroactive microorganisms in the biofilm. These results highlight the importance of initial inoculation strategies to optimise MFC performance. This study underscores the critical importance of dynamic adaptation of microbial consortia reaching a high diversity of electroactive microbial communities for efficient degradation of benzene in MFCs. Bacterial community analysis revealed that benzene-degrading communities were distinct and rich in petroleum degraders and biosurfactant producers. Notably, the presence of benzene promoted the growth of specific microbial communities, including Geobacter species and various bacteria families such as Arcobacteraceae, Enterobacteriaceae, Dysgonomonadaceae, Crocinitomicaceae, Oscillospiraceae, Nocardiaceae, and Rikenellaceae , underscoring the niche provided by benzene for specialised benzene-degrading microorganisms. Metabolite analysis revealed that these communities were able to remove all the benzene. This is the first study, where the overlook of biosurfactant producers and petroleum degraders is provided for the bioelectrochemical system and the first study where this parameter is being analysed for the efficient degradation of benzene in the bioelectrochemical system. Furthermore, our results demonstrate the importance of co-metabolites in MFC systems aimed at degrading petroleum compounds such as benzene. Microorganisms exposed to benzene exhibited a relatively prolonged period of adaptation to harsh environmental conditions, suggesting the need for sustained efforts in optimising the operation of the MFC for benzene degradation. Understanding the impact of co-metabolites, inoculation strategy, microbial community structure and diversity is important in terms of the practical use of MFC technology in the bioelectroremediation processes.",
"introduction": "1 Introduction Benzene is a common environmental pollutant that occurs frequently in the soil, air and aquatic environments. It is classified as an aromatic hydrocarbon, and due to its highly stable nature, its toxicity, and low water solubility, it is difficult for microorganisms to degrade ( Varjani et al., 2017 ). The main source of contamination with this compound is the petroleum industry: for example, petroleum refining, fuel combustion in vehicles, polymer production and leaks of petroleum products from storage tanks and pipelines. Its presence in the environment has a negative impact on plants, animals and human health, making this compound one of the main environmental concerns with hundreds of thousands of identified pollution sites only in the EU ( JRC, 2018 ). Several classic biodegradation methods have been developed over the last decades that comprise in situ and ex situ methods such as bioaugmentation or bioremediation in prisms. These techniques generally require the use of additional electron acceptors, such as oxygen or nitrates, to perform and complete the degradation processes ( Liu et al., 2018 ). Furthermore, they are often combined with ventilation, which can cause contaminants released into the air ( Alori et al., 2022 ). Another disadvantage is the electrical energy consumption when using such techniques. Therefore, it is essential to search for new methods for the removal and neutralisation of benzene in the environment. These drawbacks can be overcome through the development of new technologies, such as bioelectrochemical systems (BES) technology. The best-known example of such technology is microbial fuel cell (MFC). The key elements of these systems include membranes, electrodes and biotic elements, which can provide a wide variety of functions ( Edel et al., 2022 ). Electroactive species convert organic compounds into electrical energy through respiratory processes. In recent years, MFCs have gained significant attention due to their possible application in the removal of contaminants with simultaneous energy production ( Suresh et al., 2022 ). When microorganisms have a stable, constant electron sink, it is possible to generate electricity, while omitting the need to supply other, costly electron acceptors. However, there are several challenges in implementing the MFC technology in polluted sites. Because benzene is a recalcitrant compound, it is often required to apply co-metabolites to allow or increase the efficiency of the biodegradation. Such co-metabolic biodegradation usually depends on compounds such as sugars, volatile fatty acids, or alcohols ( Shen et al., 2020 ). These co-metabolites were typically required in most of the reported studies, where benzene was the target for biodegradation ( Table 1 ). Glucose is one of the most commonly used co-substrates in MFCs. Its presence could also increase the bioavailability of hydrophobic substrates for microorganisms and affect the increasing removal efficiency ( Obileke et al., 2021 ). Some of the few exceptions were reported by Liu et al., who investigated benzene degradation during contaminated groundwater treatment in tubular MFCs. They have recorded a maximum power density of 3.9 mW/m 2 ( Liu et al., 2018 ). In another study, Adelaja et al. (2018) recorded a power output of 0.82 mW/m 2 when investigating the biodegradation process supported by the addition of glucose. Table 1 MFC performance and comparison with other studies used aromatic hydrocarbons as substrates. Type of inoculum Carbon source Type of MFC Max power density [mW/m 2 ] Max power density [W/m 3 ] COD removal [%] CE [%] References \n Complex microbial community \n \n 0.6 g/L benzene and 1.2 g/L glucose \n \n sc-MFC \n \n 119.62 \n 5.38 \n 80.43 \n 8.4 \n This study \n Complex microbial community petroleum hydrocarbon acclimated 0.06 g/L benzene/groundwater sc-MFC 3.9 N.A. N.A. N.A. \n Liu et al. (2018) \n Benzene and ammonium-contaminated groundwater 0.05 g/L benzene/groundwater sc-MFC N.A. 0.316 N.A. 14 \n Wei et al. (2015) \n Coculture of S. oneidensis and P. aeruginosa 0.2 g/L benzene, 0.1 g/L glucose or pyruvate dc-MFC 0.39 N.A. 61.76 0.36 \n Adelaja et al. (2018) \n Adapted anaerobic digested sludge 0.82 N.A. 87.25 1.04 Anaerobic digested sludge with coculture of S. oneidensis and P. aeruginosa 0.14 N.A. 70.0 0.2 Mixed microbial community petroleum hydrocarbon acclimated 1.5 g/L benzene, 0.1 g/L phenanthrene sc-MFC 6.75 N.A. 77 N.A. \n Adelaja et al. (2017) \n Mixed bacterial community from the oil-cracking wastewater treatment plant 0.1 g/L benzene dc-MFC 0.0205 N.A. N.A. N.A. \n Wu et al. (2013) \n Mix of aerobic and anaerobic activated sludge 1 g/L phenol dc-MFC N.A. 9.1 N.A. 1.5 \n Luo et al. (2009) \n 0.5 g/L glucose, 1 g/L phenol dc-MFC N.A. 28.3 N.A. 2.7 Anaerobic activated sludge 0.3 g/L phenanthrene, 0.2 g/L benzene dc-MFC 1.06 N.A. 79.1 0.4 \n Adelaja et al. (2015) \n sc-MFC, single-chamber MFC; dc-MFC, dual chamber MFC; N.A., not available. Performing the anaerobic benzene degradation process, coupled with anode respiration pathways, is a challenging task. Most likely, this was the principal reason that none of the reported studies have demonstrated the ability of a single bacterial culture to carry out this process in MFCs. A study by Zhang et al. reported using a pure culture of Geobacter metallireducens in the removal of benzene but in hydrocarbon-contaminated sediments ( Zhang et al., 2010 ). It contains several thousands of other compounds, which affect benzene biodegradation pathways and overall MFC performance. Power performance values reported in literature have never exceeded 6.75 mW/m 2 ( Adelaja et al., 2017 ) and 12.7 mW/m 2 in multi-stacked MFCs (3.9 mW/m 2 for a single unit) ( Liu et al., 2018 ). Therefore, the recorded power outputs were very low compared to simple substrates such as acetate ( de Rosset et al., 2023 ) but also other hydrocarbons such as phenol ( Luo et al., 2009 ). In recent years, there has been an ongoing discussion of the optimal diversity of microbial communities. Several authors have concluded that diversity is strongly related to the overall environmental complexity of the system in which biodegradation is performed. In a recent study, the authors investigated the long-term adaptation of activated sludge, which was under continuous stress from the presence of benzene and naphthalene ( Li et al., 2022 ). The authors reported an ongoing downward trend in microbial diversity during long-term operation, although a higher abundance of petroleum degraders was detected. This effect is related to the fact that most microorganisms are not tolerant to the increased presence of harmful chemicals ( Wang et al., 2023 ). A recent study by Zhuang et al. revealed that soil microbial diversity decreases when exposed to high levels of petroleum compounds C10–C40 ( Zhuang et al., 2023 ). The authors observed increased levels of diversity when higher concentrations of petroleum compounds were present, which is in contrast to the findings of Peng et al. (2022) , who indicated that under elevated levels of oil refining waste, diversity was severely restricted, while few dominant genera still contributed to petroleum degradation. When the above complexity of interactions between microbial communities and environmental parameters, in particular, in bioelectrochemical systems aimed for biodegradation is taken into account, it is therefore crucial to understand the microbial growth, diversity and metabolism of benzene in microbial fuel cells. Understanding the mechanisms of this process is required to fully implement the bioelectrochemical system technology in field applications. The purpose of this study was to investigate the efficiency of individual microbial species isolated from MFC communities, as well as complex communities derived from activated sludge in benzene degradation. We determined the impact of benzene and glucose on the structure and metabolism of the microbial community to understand the influence of these two substrates on the development of efficient consortia abundant in petroleum degraders, biosurfactant producers and electroactive species. Developing various communities and investigating their structure and efficiency are important for the practical implementation of MFC technology in environmental biodegradation.",
"discussion": "3 Results and discussion 3.1 Power performance The power generation was recorded in real time for more than 160 days. At the beginning of the experiment, when glycerol (A) and benzene (B) were used as fuel, only low power density of <5 mW/m 2 was recorded. Subsequently, when the substrates were changed on day 18 to glucose (A) and benzene and glucose (B), the power generated by the MFCs increased several times. However, the introduction of new substrates has resulted in only relatively low power output, up to 12 mW/m 2 for (A) and 2 mW/m 2 for (B). These weak results suggest very limited capabilities of individual strains for power production and a toxic effect of benzene, by inhibiting the metabolic activities of the microorganisms and decreasing their abundances. The addition of activated sludge was carried out after 40 days of the experiment to facilitate benzene degradation. Real-time power performance monitoring ( Figures 1A , B ) revealed a significant increase in power density starting from the 60th day of operation. When individual MFCs are compared, it is clear that the presence of benzene has elongated the adaptation time needed for the microbial communities to reach their maximum power performance. Among all MFCs, the fastest adaptation to the efficient microbial consortium was recorded in MFC 3A, initially inoculated with Epilithonimonas hominis , where the power output increased by 30 mW/m 2 after approximately 40 days of activated sludge supplementation. On the contrary, for MFC 3B, where benzene was used along with glucose, a significant increase in power density was recorded at the end of the experimental period, after 150 days. Figure 1 Real-time temporal power performance (A,B) . Maximum power density values from LSV tests (C,D) . Polarisation curves of the anodes and cathodes of the MFCs (E,F) at maximum performance (1A, 2A, 3A, 5A, 2B, and 5B—120th day; 4A, 1B, 3B, and 4B—150th day); the colours for the anode potentials are less intense. The colours in the legend of panels (C,D) are valid for all panels in the same row. Panels (A,C,E) are series A, and (B,D,F) are series B. Different behaviours were observed for MFC 1A, initially inoculated with Ochrobactrum anthropi strain 1. It revealed a stable increase in power over time, reaching a maximum of 40 mW/m 2 after 140 days. Adapting the same starting consortium with benzene resulted in a higher power output (MFC 1B), equal to 85 mW/m 2 , which was the highest power density among all MFCs, reaching levels similar to those of MFCs that were fed without benzene (series A). MFC 5A and MFC 5B (initially Ochrobactrum anthropi strain 2) achieved similar real-time power behaviour throughout the experiment. However, after 140 days, higher power generation was recorded for the cell without benzene. The maximum power density values for MFC 5A and MFC 5B were approximately 65 and 45 mW/m 2 , respectively. MFC 2A and MFC 2B (initially Rhodococcus qingshengii ) reached similar power until day 150, where MFC 2B supplemented with benzene was more efficient and reached a value of 45 mW/m 2 vs. 15 mW/m 2 for the MFC 2A. Overall, it may be possible that the adaptation time to reach a fully matured biofilm and stable community was higher due to the initial presence of pure cultures on the anodic surface. It is known that anodic biofilm structure is very stable and hostile against external bacterial strains once the biofilm has matured ( Ieropoulos et al., 2019 ). Thus, the new consortium established over the anodes required more time to adaptation and resulted in peak performance after 140 days of operation. The open circuit voltage (OCV) ranged from 0.49 V to 0.59 V for glucose-fed MFCs. The highest value was recorded for MFC 3A, which exhibited the highest power density value in the experiment. For MFCs supplemented with benzene and glucose, OCV ranged from 0.51 V to 0.54 V, with the highest value for MFC 2B. The graphs of the maximum power densities calculated from the polarisation tests ( Figures 1C , D ) show that MFC-fed glucose reached higher power generation values and were more stable over time. Benzene-supplemented MFCs showed an increasing trend throughout the duration of the experiment. At the end of the experimental period, these MFCs have reached similar power density values as MFCs fed with glucose. In the benzene group, the highest power density value of the LSV tests was 120 mW/m 2 (108 μW) for MFC 1B, and the lowest was 21 mW/m 2 (18 μW) for MFC 3B. In the glucose-fed group, the highest power value was 104 mW/m 2 (93 μW) for MFC 3A, and the lowest was MFC 2A with a power density of 24 mW/m 2 (22 μW). Figures 1A , B clearly suggested that the dynamic community evolution that took place in glucose-fed MFCs was more divergent and led to higher variation of the results over time. On the other hand, the presence of benzene resulted in stabilising the electrical signal over time and led to less divergent long-term data, suggesting higher similarities between the compositions of these communities. This was further confirmed using 16S rRNA biodiversity analysis. The results achieved in this study showed that the co-metabolic degradation of benzene was significantly higher than showed by other studies ( Table 1 ). Power output was almost 100 times higher compared to the study reported by Adelaja et al. (2018) , where the maximum power density was 0.82 mW/m 2 . The authors have used a dual chamber MFC design, along with a 3-fold lower benzene concentration and a 12-fold lower glucose concentration. However, in this study, MFCs with pure cultures in the first 18 days of operation achieved power performance of approximately 4.5 mW/m 2 . We also identify these values as negligible power, even for pure cultures MFCs. These results indicate the importance of the presence of co-metabolite in MFC system to degrade petroleum compounds such as benzene. Microorganisms exposed to benzene took a relatively long time to adapt to harsh environmental conditions ( Lovley, 2000 ). Furthermore, when individual MFCs were compared, long-term power performance was similar when benzene was added as a substrate, while much higher variability of power output was observed for glucose as a sole source of carbon. The performance of the individual electrodes was investigated at the end of the experimental period (as indicated in Figures 1E , F ). The polarisation curves obtained for series A and B differ significantly in their characteristics, which mean that different factors were limiting their performance. In series A (glucose), strong ohmic losses were observed for MFC 1A and MFC 2A with dominance of the anodic overpotential. The performance of MFC 3A, MFC 4A, and MFC 5A was mainly limited by anodic concentration losses (mass transfer effect), resulting in a power overshoot phenomenon (MFC 3A and 4A). This could have been caused by the biofilm overgrowth and thus lower substrate transfer into the deeper layers of the biofilm. It is also possible that in the initial period, when negligible performance was observed for isolated species, this resulted in contamination of the electrodes with inactive cells ( Greenman et al., 2021 ). A strong overshoot phenomenon was observed for both the anode and the cathode. Although in the cathode this could be associated with difficulties in oxygen delivery to active electrode sites and disruption of electron flow, due to salt precipitation on the cathode surface ( de Rosset et al., 2023 ), strong underperformance of the anodes could also have affected cathodic behaviour. Analysis of the polarisation curves of MFCs supplemented with benzene ( Figure 1F ) indicates that MFC 1B, MFC 2B, and MFC 5B were limited by ohmic resistance and concentration losses at the anode. A primary reason for this behaviour in benzene-degrading MFCs could be the complexity and bioavailability of this substrate ( Pasternak et al., 2022 ). For MFC 3B and MFC 4B, losses were observed in all regions of the cathodic curves. Apart from the salt accumulation, the cathodes can undergo biofouling, which can negatively impact their durability ( Pasternak et al., 2016 ). 3.2 Coulombic efficiency and COD removal Coulombic efficiency (CE) refers to the efficiency with which electrons produced during microbial processes are transferred to the electrode as electrical current. In our study ( Figure 2 ), CE values ranged from 1 to 8%, indicating that electrochemical reactions were responsible for a minor part of total COD removal. However, CE values were higher when benzene and glucose were used together as carbon sources in MFCs. The highest CE values observed were of 8.4% vs. 5.9% when glucose was used as a single substrate. The higher CE values for the combination of benzene and glucose may be due to the fact that the microbial population that oxidises those substrates differs from each other. This contributes to the release of electrons when attacking the respective substrates and hence to the higher values obtained in comparison with those obtained for MFC using only glucose. This could have resulted from various metabolic fluxes represented by different microbial consortia, in particular the presence of fermentative pathways that compete with anodic respiration ( Torres et al., 2007 ). Figure 2 CE and COD values measured after 3 months of MFCs operation. However, data from other studies ( Table 1 ) indicate that CE achieved in this study, when benzene was used as fuel (8.4%), was the highest reported so far. Adelaja et al. (2017) carried out a similar study, where anaerobic digested sludge was used as an inoculum to degrade benzene with glucose as a co-metabolite, resulting in CE of 1.04%. The use of coculture of Shewanella oneidensis and Pseudomonas aeruginosa resulted in CE of 0.36%. Luo et al. (2009) used phenol as a fuel, leading to a CE of 1.5%, which improved significantly when glucose was added to facilitate degradation (CE = 2.7%). In another study by Adelaja et al. (2015) , phenanthrene and glucose degradation were characterised by CE of 0.4%. Herein, the observed CE was one of the highest values reported in the literature for benzene in MFC. In general, glucose results in lower CE values, compared to low-molecular-weight compounds such as acetate, because of its fermentative nature. In addition to electroactive species, glucose could be used by non-electrogenic bacteria such as methanogens and fermentative species ( Chae et al., 2009 ). Glucose, however, is a common co-substrate that can be used in conventional biodegradation studies ( Liu et al., 2023 ), which was confirmed herein for bioelectrochemical degradation, through a significant improvement in Coulombic efficiency. Nevertheless, the high removal of COD achieved of 80–90% still indicates that a significant portion of glucose and benzene was metabolised in a non-electrogenic manner. Metabolite analysis through NMR studies allowed us to provide insight into the potential metabolic pathways involved and led to the decrease in the current generation. Similarly as with CE and maximum power output, the COD values reported herein were among the highest compared to other studies in which this parameter ranged from 26.3 to 87.3%. 3.3 Metabolites The metabolic analysis for group A fed with glucose ( Figure 3A ) revealed seven distinct exometabolites (released to the electrolyte), while for group B fed with benzene and glucose ( Figure 3B ), eight metabolites were present. Five of the metabolites were common for both groups: ethanol, acetate, succinate, glycerol and formate (see Table 2 ). The presence of these compounds indicates fermentation reactions that include acetogenesis, which occurs under anaerobic conditions. Figure 3 Comparison of 1D 1 H NMR cpmgpr1d spectra of (A) MSM medium with glucose (top) and the representative sample obtained from MFC 1A (bottom) (1: glucose; 2: butyrate; 3: propionate; 4: ethanol; 5: acetate; 6: succinate; 7: glycerol; 8: formate); (B) medium with benzene and glucose (top) and the representative sample obtained from MFC 4B (bottom) (1: glucose; 2: benzene; 3: ethanol; 4: acetate; 5: succinate; 6: 5-aminolevulinate; 7: dimethylamine; 8: methanol; 9: glycerol; 10: formate). Table 2 1 H NMR signal assignments for extracellular metabolites in MFCs. L.p. Metabolites KEGG number Chemical shift of representative signal [ppm] Group A Group B 1 Ethanol C00469 3.65 (m) 1.20 (t) 2 Acetate C00033 1.91 (s) 1.91 (s) 3 5-aminolevulinate C00430 – 2.47 (m) 4 Succinate C00042 2.39 (s) 2.39 (s) 5 Dimethylamine C00534 – 2.72 (s) 6 Methanol C00132 – 3.35 (s) 7 Glycerol C00116 3.55 (dd) 3.65(dd) 8 Formate C00058 8.44 (s) 8.44 (s) 9 Butyrate C00246 1.54 (m) – 10 Propionate C00163 1.04 (t) – Identified metabolites are products that can be delivered from both fermentative glucose reactions and microaerophilic benzene transformation ( Toth et al., 2021 ). The oxygen concentration in the anodic chambers was measured below 1% vs. O 2 in the air, indicating that benzene degradation could occur under microaerophilic and anaerobic conditions. This initial oxygen content may have caused aerobic degradation to predominate at the beginning of the feeding batch cycle and succeeded under anaerobic conditions, once oxygen was depleted. Some of the identified metabolites, such as acetate, formate and propionate, are known to be end products of both aerobic and anaerobic metabolic pathways of benzene degradation ( Firmino et al., 2018 ). The presence of these carboxylic acids may also indicate the appearance of methanogens, which was further confirmed by 16S rRNA analysis. Bioelectrochemical degradation did not lead to the presence of any metabolites, typical for the aerobic and anaerobic degradation of benzene, such as catechol and benzoate, which were not identified. This is due to the fact that the sampling was carried out at the end of the feeding batch cycle, where the final products of the metabolic pathways are the most abundant. Under methanogenic conditions, aromatic hydrocarbons are degraded by syntrophic interactions between fermentative bacteria and methanogenic archaea, rather than by a single microbial species. Numerous studies have explained this phenomenon, in which fermentative bacteria convert aromatic hydrocarbons to intermediate metabolites such as acetate and propionate. These metabolites serve as substrates for methanogenesis by archaea ( Vogt et al., 2011 ). Under aerobic conditions, benzene is converted into catechol via cis-benzene dihydrodiol and then attacked through ortho- or meta-ring cleavage. The resulting product is pyruvate, which can then be converted to formate, acetate, or ethanol ( Liu, 2003 ) as detected here. Acetate could be the final product of benzene anaerobic degradation under sulphate-reducing conditions, which can occur in MFCs, and through methylation, hydroxylation and carboxylation, converted to benzoyl-CoA as the central metabolite that can be further reduced by benzoyl-CoA reductases ( Vogt et al., 2011 ). Low CE values are caused by the competition for the carbon source between electroactive and other microorganisms, mainly methanogenic bacteria, which are known to convert benzene into methane ( Grbic-Galic and Vogel, 1987 ). 3.4 Microbial community analysis Microbial community analysis revealed distinct microbial fingerprints. When comparing individual communities, those subjected to glucose exhibited a higher abundance of Proteobacteria (mainly Gammaproteobacteria ), while MFCs supplemented with benzene demonstrated a higher abundance of Campylobacterota , Bacteriota, and Actinobacteriota ( Figure 4A ). Moreover, MFCs 3B, 4B, and 5B displayed a higher amount of Firmicutes . Figure 4 Microbial community analysis for anodic communities fed with glucose (group A) and glucose and benzene (group B). Plots (A–C) show abundance profiling by the most abundant phyla, genera and species, respectively. At the genus level ( Figure 4B ), microbial communities of group A were dominated mainly by Aeromonas, followed by Pseudomonas and Acidovorax . Figure 5A indicates the species identified as key electroactive and petroleum-degrading species, as well as biosurfactant producers. Aeromonas genus is a group of Gram-negative bacteria that are facultative anaerobes. They exhibit electrochemical activity (e.g., Aeromonas hydrophila ) and the ability to reduce nitrate and sulphate using various electron donors such as glucose, glycerol, and pyruvate ( Pham et al., 2003 ). Furthermore, MFCs 1A, 2A, and 3A were co-dominated by Delftia, which is recognised for its proficiency in electrochemical processes and its ability to thrive in anaerobic conditions ( Jangir et al., 2016 ). The microbiomes exposed to benzene were more diverse, and several genera were co-dominant: Citrobacter , Arcobacter , Dysgonomonas , Myroides, and Macellibacteroides . Figure 5 Microbial community analysis for anodic communities fed with glucose (group A) and glucose and benzene (group B). (A) Heatmap of the most abundant genera; (B) PCA plot based on the most abundant genera; (C) box and whiskers plots of Chao1 index at the genus level. At the species level ( Figure 4C ), the glucose-fed MFCs were dominated by Aeromonas hydrophila . The greatest prevalence occurs in MFCs 3A (56.9%) and 1A (47.5%), which were characterised by the highest power output generated during the experiment. This strain exhibits electroactive properties, suggesting that it plays a key role in the current generation in these MFCs ( Pham et al., 2003 ). The glucose-fed communities were also abundant in the Pseudomonas genus: Pseudomonas japonica and Pseudomonas aeruginosa . The presence of P. japonica is expected in the presence of glucose ( Pungrasmi et al., 2008 ); nevertheless, it was not enriched when glucose and benzene were used in MFC. P. aeruginosa is well known to perform indirect extracellular electron transfer, due to the production of various redox mediators, such as pyocyanin ( Pérez-García et al., 2023 ). Its highest contribution was recorded as 5.9% in the most effective MFC 3A. Similarly, Acidovorax caeni was previously reported in anodic communities ( Li et al., 2023 ). Its highest contribution is 5.8% in MFC 1A. Other prominent species observed in anodic communities subjected to glucose were Comamonas testosteroni (the highest value of 19.9% in MFC 4A), Comamonas nitrativorans (4.4% in MFC 1A), Myroides odoratus (4.9% in MFC 5A), and Alcaligenes faecalis (4.3% in MFC 5A), which occur commonly in activated sludge. The bacterial community was different when benzene was used along with glucose as a fuel. The most dominant strain was Citrobacter freundii , whose contribution ranged from 23.1% for MFC 5B to 51.5% for MFC 2B. It is an exoelectrogenic bacterium ( Huang et al., 2015 ), which could be responsible for the anaerobic biodegradation of benzene ( Gupta et al., 2015 ). The high content of this strain in MFCs may suggest its important role in electron transport to the anode. For MFC 5B and MFC 1B, the most abundant strain was Arcobacter faecis with a contribution of 41.5 and 35.5% contribution, respectively. To the best of our knowledge, there are no data available in the literature on the activity of this species in MFCs. However, many other species of Arcobacter have been defined as electroactive and form biofilms in microaerobic and anaerobic environments ( Hassan et al., 2018 ). In the remaining MFCs, its abundance was below 2%. In further studies, these strains should be verified for their degradation capabilities, along with their possible pathogenicity and drug resistance. Comamonas testosteroni, classified as electroactive strain ( Yu et al., 2015 ), was also present in MFCs operating with benzene. Its abundance was significant for all MFCs except MFC 5B and ranged from 5.3% for MFC 2B to 16.2% for MFC 4B. It belongs to Burkholderiales order, which is known as a dominant phylotype in enrichment cultures of anaerobic benzene-degrading microcosms. The genera of this order use the methylation pathway for anaerobic benzene activation ( Melkonian et al., 2021 ). Another distinctive species in the benzene-degrading consortia was Myroides odoratus (13.7%, MFC 3B), Aeromonas hydrophila (7.2%, MFC 1B), Comamonas nitrativorans (4.3%, MFC 2B), and Anaerotignum propionicum (2.2%, MFC 4B). Furthermore, all benzene-degrading consortia comprised Alcaligenes faecalis , Arcobacter faecis, and Desulfovibrio vulgaris (<1%). According to previous studies, Geobacter was the key genus in benzene-degrading communities ( Rooney-Varga et al., 1999 ). In fact, in our study, Geobacter species were only found in MFCs fed with benzene, and their presence was only below <0.2%, which suggests its supporting role in the entire community. In addition, families such as Arcobacteraceae , Enterobacteriaceae , Dysgonomonadaceae , Crocinitomicaceae , Oscillospirales , Nocardiaceae and Rikenellaceae , known from the degradation of petroleum compounds ( Ejaz et al., 2021 ; Duncan et al., 2024 ), were observed only in community-fed benzene. Some of the representatives of these families were reported as dominant genera in MFCs with benzene supplementation ( Citrobacter and Arcobacter ). Furthermore, we have identified several biosurfactant-producing species from Sphingobacteriaceae , as well as anaerobic members from Paludibacteraceae , Anaerovoracaceae , and Crocinitomicaceae and obligate anaerobes from Oscillospirales . The presence of these species is crucial in anaerobic benzene degradation as biosurfactants play an important role in increasing the bioavailability of petroleum compounds ( Kaczorek et al., 2018 ). The PCA ( Figure 5B ) revealed a clear separation between groups A and B. This indicates that the addition of benzene strongly impacted the bacterial community structure in the anode chambers. In MFCs supplemented with benzene, bacterial communities have a higher proportion of Citrobacter , Comamonas , Dysgonomonas , Arcobacter , Macellibacteroides , and Dechlorobacter . It can also be seen that MFC 2B, MFC 3B, and MFC 4B form a distinct group. In glucose-fed MFCs, the microbial structure was shaped mainly by the Aeromonas , Pseudomonas , Acidovorax , Delftia , Desulfovibrio and Petrimonas genera. Therefore, the PCA unveils key microbial genera responsible for the bioelectrochemical degradation of benzene and suggests that the presence of glucose in both types of metabolism is insufficient to maintain some of the important electroactive groups (such as Pseudomonas ) within the consortia, which may have played an important role in a long adaptation process, where low power outputs were recorded. The alpha-diversity index (Chao1, Figure 5C ) calculated at the genus level was significantly different ( t -test, p < 0.01) for communities adapted to the benzene cometabolism. The highest alpha-diversity index was reached by MFC 5B, and it was 1.5 times higher than the index for MFC 5A, which had the highest diversity in group A. This indicates that the presence of toxic and complex substrates in the anode chamber led to the formation of more diverse communities, compared to those exposed to a simple carbon source. This highlights that communities abundant in glucose-metabolising species possess stronger competition mechanisms that limit their abundance. The metabolism of benzene is more complex and requires more steps to fully degrade it. Therefore, a more diversified microbial community is required to allow efficient biodegradation. Moreover, according to the qPCR results (see Supplementary Table S2 ), the abundance of total bacterial DNA in MFCs fed glucose and benzene was slightly higher compared to MFCs fed only glucose. For Archaea, DNA copy number was an order of magnitude higher for MFCs fed with benzene, indicating that the presence of benzene was more favourable for their growth. These results indicate that the complexity of the system in which the biodegradation is performed is an important factor to take into account when analysing the process. Previous findings led to several contradictory observations, where petroleum compounds led to decrease or increase in alpha diversity, or led to increased diversity ( Li et al., 2022 ; Peng et al., 2022 ; Wang et al., 2023 ; Zhuang et al., 2023 ). The principal reason could be the fact that each study varied in methodology, as well as a variety of alpha-diversity indices have been reported. Therefore, the complexity of the dynamics of microbial consortia, along with their metabolic networks in the presence of recalcitrant compounds in bioelectrochemical systems, may vary compared to other environments. This comes from the fact that such an environment is highly dynamic and self-adapts in real time to the conditions provided by the electroactive consortium itself, for example, by changing the electrochemical parameters of the reactors ( Pasternak et al., 2018 )."
} | 9,493 |
36161690 | PMC9829402 | pmc | 759 | {
"abstract": "Summary Ester‐linked p ‐hydroxybenzoate occurs naturally in poplar lignin as pendent groups that can be released by mild alkaline hydrolysis. These ‘clip‐off’ phenolics can be separated from biomass and upgraded into diverse high‐value bioproducts. We introduced a bacterial chorismate pyruvate lyase gene into transgenic poplar trees with the aim of producing more p ‐hydroxybenzoate from chorismate, itself a metabolic precursor to lignin. By driving heterologous expression specifically in the plastids of cells undergoing secondary wall formation, this strategy achieved a 50% increase in cell‐wall‐bound p ‐hydroxybenzoate in mature wood and nearly 10 times more in developing xylem relative to control trees. Comparable amounts also remained as soluble p ‐hydroxybenzoate‐containing xylem metabolites, pointing to even greater engineering potential. Mass spectrometry imaging showed that the elevated p ‐hydroxybenzoylation was largely restricted to the cell walls of fibres. Finally, transgenic lines outperformed control trees in assays of saccharification potential. This study highlights the biotech potential of cell‐wall‐bound phenolate esters and demonstrates the importance of substrate supply in lignin engineering.",
"introduction": "Introduction Lignin is a complex phenolic biopolymer found primarily in the secondary cell walls of vascular plants. The development of lignin was a milestone in plant evolution, and the lignified plant vasculature is a defining feature of tracheophytes (Weng and Chapple, 2010 ). Lignin is a critical component of water conduction and plant defence systems, and it contributes greatly to the compressive strength of stems and branches (Dixon and Barros, 2019 ). Although the chemical complexity and recalcitrance of lignin presents a major challenge for industrial biomass processing, it also paves the way for diverse biochemicals and bioproducts to be made from lignin (Rinaldi et al ., 2016 ). To that end, there is considerable interest in developing designer biomass feedstocks with engineered lignin polymers (Mottiar et al ., 2016 ). Lignin is assembled primarily from three monolignol precursors— p ‐coumaryl, coniferyl, and sinapyl alcohol (Freudenberg and Neish, 1968 ). Once incorporated into a lignin polymer, these become p‐ hydroxyphenyl (H), guaiacyl (G), and syringyl (S) lignin units, respectively. The monolignols are synthesized in the cytosol prior to their export to the cell wall where laccase and peroxidase enzymes generate monolignol radicals that undergo coupling and cross‐coupling reactions (Figure 1 ; Ralph et al ., 2004 ). This process of dehydrogenative polymerization proceeds in a combinatorial manner such that the structure of the resulting polymer reflects the complement of monomers delivered to the site of lignification. Figure 1 Metabolic map depicting the shikimate and general phenylpropanoid pathways in the plastid and cytosol, as well as the export of lignin monomers and polymerization in poplar cell walls. The novel route from chorismate catalysed by bacterial chorismate pyruvate lyase (CPL) that leads to the incorporation of more p ‐hydroxybenzoate pendent groups in lignin is shown in blue. Consequently, a wide variety of lignin composition and structure exists in Nature. The flexibility of lignification is perhaps best exemplified by loss/gain‐of‐function experiments. For example, downregulation of p ‐coumaroyl‐CoA 3′‐hydroxylase (C3′H) in transgenic poplar led to dramatically increased levels of H units (Coleman et al ., 2008 ), whereas overexpression of ferulate 5‐hydroxylase (F5H) resulted in a polymer built predominantly with S units (Huntley et al ., 2003 ; Stewart et al ., 2009 ). The inherent plasticity of lignification extends beyond the three primary monolignols, as demonstrated by the growing number of non‐canonical monomers that have been reported (Del Río et al ., 2022 ). Among these, the γ‐linked monolignol conjugates are the most prevalent and best studied. For instance, the lignin found in kenaf bast fibres is naturally acetylated due to the incorporation of pre‐acylated monolignols (Lu and Ralph, 2008 ; Ralph, 1996 ). Similarly, the lignin of commelinid monocots is decorated with p ‐coumarate groups derived from monolignol– p ‐coumarate conjugates produced via p ‐coumaroyl‐CoA:monolignol transferases, members of the BAHD superfamily of acyltransferases (Hatfield et al ., 2008 ; Ralph et al ., 1994 ; Withers et al ., 2012 ). Analogously in poplars and willows, p ‐hydroxybenzoate groups (henceforth denoted as p HB for the ester‐linked pendent form, and p HBA for the free acid) comprise up to 10% of the lignin (Goacher et al ., 2021 ; Smith, 1955 ). These too derive from monolignol conjugates and are bound to lignin via the γ position (Lu et al ., 2004 ; Morreel et al ., 2004 ). Cell‐wall‐bound p HB has also been detected throughout the family Arecaceae (palms), in Japanese spikenard ( Aralia cordata ), in the roots of carrot ( Daucus carota ), in the stems of purple mountain saxifrage ( Saxifraga oppositifolia ), and at especially high levels in the Mediterranean seagrass Posidonia oceanica (Faleva et al ., 2020 ; Hibino et al ., 1994 ; Parr et al ., 1997 ; Pearl et al ., 1959 ; Rencoret et al ., 2020 ). As p HB groups are ester‐linked to lignin polymers, they can be released by mild alkaline hydrolysis. Once separated from the biomass, these ‘clip‐off’ phenolics may be used directly as platform chemicals or upgraded into a variety of other biochemicals and bioproducts (Becker and Wittmann, 2019 ; Rinaldi et al ., 2016 ; Wang, Bilal, et al ., 2018 ). For example, derivatives of p HBA are used in cosmetics (as paraben preservatives), in polyester plastics (as liquid crystal copolymers), and as precursors to a wide range of pharmaceutical compounds (Aalto et al ., 1953 ; Jackson Jr and Kuhfuss, 1976 ; Manuja et al ., 2013 ). Alternatively, engineered microbes have been developed that catabolize and funnel phenolics including p HBA into metabolic pathways leading to high‐value compounds (Beckham et al ., 2016 ; Kamimura et al ., 2017 ). However, to economize the use of clip‐offs for such applications, the innate levels in feedstock plants must be substantially improved (Karlen et al ., 2020 ). Herein, we report a novel strategy to increase the levels of p HB groups by bolstering the supply of p HBA precursors. It has been established that p HBA derives from phenylpropanoid biosynthesis (Figure 1 ; El‐Basyouni et al ., 1964 ; Zenk and Müller, 1964 ); however, the endogenous pathway has not been fully elucidated and remains an inaccessible engineering target. Fortuitously, an alternate route has evolved in bacteria: chorismate pyruvate lyase (CPL) cleaves chorismate, a product of the shikimate pathway, to produce p HBA and pyruvate (Siebert et al ., 1994 ). Previous studies on the heterologous expression of this enzyme in planta using different promoter sequences reported hyperaccumulation of various soluble p HBA‐containing metabolites in leaves and plant cell cultures (Li et al ., 1997 ; Siebert et al ., 1996 ). Although this historical work demonstrated the utility of CPL in metabolic engineering, heterogeneous product streams can be a challenge in industrial processing. As poplar can deploy p HBA to the cell wall in a stable form where it can later be released by alkaline hydrolysis, we asked whether expression of bacterial CPL would lead to increased p ‐hydroxybenzoylation of lignin in poplar wood and thereby provide a readily accessible and homogeneous source of p HBA.",
"discussion": "Results and discussion Transgenic poplars expressing bacterial CPL \n By expressing bacterial CPL in plastids, we sought to divert carbon away from chorismate and the metabolic pathway leading to lignin. In this way, the addition of more pendent groups could simultaneously result in reduced lignin content. Although recent work has identified a cytosolic supply of chorismate (Qian et al ., 2019 ), it is generally thought that chorismate which is destined for lignin predominates in plastids. Moreover, plastidic expression of CPL is known to be more effective (Sommer and Heide, 1998 ; Viitanen et al ., 2004 ). As chorismate is also an essential precursor to the biosynthesis of proteins and a wide range of other metabolites, a cellulose synthase promoter, PtCesA8pro , was selected in order to drive expression only in cells undergoing secondary cell wall formation. Transgenic hybrid poplar lines were generated by Agrobacterium ‐mediated transformation to evaluate the effects of heterologous expression of bacterial CPL (Figure S1 , Table S1 ). Five lines chosen for in‐depth analysis were grown in a randomized greenhouse trial alongside non‐transformed wild‐type (WT) control trees for 18 weeks. Other than a small penalty in growth and the occurrence of sylleptic branching, which is characteristic of fast‐growing hybrid poplar and could be a response to metabolic stress (Ceulemans et al ., 1990 ), no major developmental defects were observed (Figure 2a ). In comparison with control trees, transgenic lines were as much as 10% shorter and 12% thinner in stem diameter (Figure 2b ). RT‐qPCR was used to measure the transgene expression in developing xylem tissue and revealed that line 5 had the highest expression (Figure 2c ). Figure 2 Expression of bacterial chorismate pyruvate lyase in transgenic poplar. (a) Representative photos of transgenic and wild‐type (WT) control trees. (b) Heights and stem diameters measured at the time of harvest are shown in blue and orange on the left and right axes, respectively. Those values marked with an asterisk are significantly different from the WT control, depicted with horizontal dashed lines (one‐way ANOVA with Dunnett's test, n = 5 for each line with technical triplicates, P ‐value < 0.05). (c) Relative expression of the ubiC gene, shown in green, was measured by RT‐qPCR. No expression was detected for the WT control. Five biological replicates were analysed for each line using technical triplicates. Effects on lignin content, composition and structure The xylem cell wall composition was analysed using extractive‐free wood powder. The total Klason lignin content was slightly, but significantly, reduced in all lines compared with the WT control (Figure 3a ). Of these, line 5 had the lowest lignin content with a 5% reduction in acid‐insoluble lignin and a 15% reduction in acid‐soluble lignin. Thioacidolysis was then used to assess whether there was a change in lignin composition. Analysis of liberated monomers revealed a significant albeit small decrease in S lignin units in all transgenic lines except line 1 (Figure 3a ). Figure 3 Lignin content, composition, and structure were altered in transgenic poplars due to the incorporation of p ‐hydroxybenzoate. (a) The Klason lignin content is plotted on the left axis with the acid‐insoluble fraction shown in dark blue and the acid‐soluble fraction in light blue, and the ratio of syringyl to guaiacyl lignin units (S:G) is plotted on the right axis in orange. Those values marked with an asterisk are significantly different from the wild‐type (WT) control, depicted with horizontal dashed lines (one‐way ANOVA with Dunnett's test, n = 5 for each line with technical triplicates, P ‐value < 0.05). (b) Two‐dimensional 1 H– 13 C HSQC NMR spectra for enzyme‐lignin samples of line 5 and the WT control, as labelled, showing the aromatics region. The colour‐coding and peak annotations for p ‐hydroxyphenyl (H, light purple), guaiacyl (G, blue), and syringyl units (S and S′, dark and light purple), as well as p ‐hydroxybenzoate pendent groups (yellow, p HB) are elaborated with the structures shown below. Proportions based on integrated peak volumes are provided as the mean ± standard error for three biological replicates. (c) The average molecular weight is plotted on a weight‐average basis ( M \n W ) in blue and on a number‐average basis ( M \n N ) in orange. Two‐dimensional 1 H– 13 C heteronuclear single‐quantum coherence (HSQC) NMR spectroscopy was used to further examine changes in the composition and structure of lignin. Spectra of enzyme‐lignin preparations confirmed the small shift in monomer composition (Figure 3b ). Moreover, this analysis revealed an increase in cell‐wall‐bound p HB groups. Volume integrations showed a 50% increase in the amount of p HB in the mature xylem of line 5 compared with WT. Furthermore, the proportion of β‐aryl ethers was slightly increased, primarily at the expense of resinols (Figure S2 ). Resinol structures arise from the dimerisation of monolignol radicals when the γ‐OH traps the quinone methide intermediate (Ralph et al ., 2004 ). However, the occurrence of γ‐linked p HB groups precludes this mechanism such that other structures prevail in highly γ‐acylated lignin (Lu and Ralph, 2008 ). Next, gel‐permeation chromatography was used to ascertain whether increases in p ‐hydroxybenzoylation affected lignin molecular weight (Figure 3c ). This analysis showed that there were no statistically significant differences on either a weight‐ (M W ) or number‐average basis (M N ). To corroborate the NMR results, the amount of p HBA released from extractive‐free cell wall material was quantified following mild alkaline hydrolysis (Figure 4a ). As p HB groups are ester linked, they can be liberated by saponification reactions. The transgenic poplars all contained significantly more p HB than WT control trees, with line 5 again being the most altered. In developing xylem, line 5 had at least nine times greater levels of p HB groups than the control trees. However, the differences were less pronounced in mature xylem which is comprised mostly of cells that have undergone programmed cell death. In mature xylem, line 5 had 50% more p HB than WT, a value that conforms with the NMR spectra of the same samples. Figure 4 Transgenic poplars hyperaccumulate ester‐linked monolignol conjugates of p ‐hydroxybenzoate. (a) The amount of p ‐hydroxybenzoate released by mild alkaline hydrolysis is shown for developing xylem (blue) and mature xylem (orange). (b) The amount of sinapyl p ‐hydroxybenzoate conjugates released by DFRC (sum of cis and trans isomers of 4‐acetoxysinapyl p ‐acetoxybenzoate, S– p HB) is shown in green for mature xylem. No conjugates of coniferyl p ‐hydroxybenzoate were detected. Those values marked with an asterisk are significantly different from the wild‐type (WT) control (one‐way ANOVA with Dunnett's test, n = 5 for each line with technical triplicates, P ‐value < 0.05). Next, the derivatization followed by reductive cleavage (DFRC) method was used to validate the findings from alkaline hydrolysis and to confirm that p HB groups originated from monolignol– p HB conjugates. As reductive cleavage releases monomers bound via β‐aryl ether linkages but does not cleave ester bonds, DFRC yields a diagnostic product of S– p HB conjugates, namely 4‐acetoxysinapyl p ‐acetoxybenzoate. Analysis of mature xylem showed significantly higher levels of S– p HB conjugates in all transgenic lines except line 1, with line 5 having 70% more than the WT control (Figure 4b ). Accumulation of p ‐hydroxybenzoate‐containing metabolites As CPL produces p HBA, the key precursor to p HB groups, it was reasonable to anticipate elevated levels in transgenic trees. However, to be deployed to the cell wall, p HBA must be first converted into p ‐hydroxybenzoyl‐CoA, then conjugated with a monolignol, and finally exported to the site of lignification in the apoplast (see Figure 1 ). Accordingly, the levels of free p HBA were measured next. The developing xylem of all transgenic lines contained significantly more than control trees (Figure 5a ). At the same time, far more p HBA occurred as conjugates—in both alkali‐labile and acid‐labile forms. Compared with control trees, line 5 had 20 times more total p HBA as metabolites in developing xylem. Figure 5 Metabolites of p ‐hydroxybenzoate hyperaccumulate in the developing xylem of transgenic poplars. (a) The amounts of free p ‐hydroxybenzoate (orange) as well as p ‐hydroxybenzoate released by alkali treatment (light blue) and p ‐hydroxybenzoate released by alkali and acid treatments (dark blue) are shown for methanolic extracts of developing xylem. Those values marked with an asterisk are significantly different from the wild‐type (WT) control (one‐way ANOVA with Dunnett's test, n = 5 for each line, P ‐value < 0.05). (b) Volcano plot showing xylem metabolites in methanolic extracts with an abundance significantly different in line 5 compared to the WT control. The metabolites in the shaded areas have a two‐fold change in signal intensity with a P ‐value < 0.01 (Student's t ‐test, n = 12 for each line). Parent molecular ions that fragment to m / z 137.0246 ( e.g. p HBA or salicylate) are coloured red, those that fragment to m / z 123.0445 ( e.g . salicyl alcohol) are coloured purple, and those that fragment to both m / z 137.0246 and m / z 123.0445 (possibly p HBA‐containing homologues of salicinoids) are coloured magenta. Key metabolites described in the text are indicated by compound identification numbers (1–7). (c) Chemical structures of the phenolic and acid glucosides of p HBA (1, 4) and salicylic acid (5, 2), the phenolic glucoside of vanillic acid (3), p HBA (6), and methyl paraben (7), which may be an artefact arising from esterification of p HBA with methanol during the extraction. (d) UHPLC‐PDA traces for methanolic extracts of line 5 and the WT control at 255 and 290 nm with the peaks corresponding to confirmed compounds marked with compound identification numbers (1–7) and peaks with parent ions that fragment to both m / z 137.0246 and m / z 123.0445 and which may be p HBA‐containing homologues of salicinoids marked with magenta circles. UHPLC coupled with high‐resolution mass spectrometry (HRMS) was then applied to methanolic extracts of line 5 and WT control samples to separate and identify the various p HBA‐containing metabolites. This analysis revealed numerous compounds that were over‐represented in line 5 compared with control trees (Figure 5b , Table S2 ). Many of these metabolites appeared to contain p ‐hydroxybenzoate moieties, as shown by the presence of a fragment ion at m / z 137.0246. Although a m / z 137.0246 fragment can also be derived from salicylate ( ortho ‐hydroxybenzoate, o HBA), the greater abundance of glycosylated p HB (peaks 1 and 4) compared with glycosylated salicylic acid (peaks 2 and 5), combined with the presence of peaks generating the m / z 137.0246 fragment in CPL poplar that are greatly reduced or entirely absent in the WT supports their designation as p HB‐containing metabolites. The most abundant peak in line 5 samples (peak 1) was identified as the phenolic glucoside of p HBA, whereas the second‐most abundant peak (peak 4) was determined to be the p HB acid glucoside (Figure 5c,d ). Much weaker signals were identified by MS as glucosides of salicylic acid ( m / z 299.0787, peak 2 and 5) and vanillic acid ( m / z 329.0895, peak 3). These assignments were validated by matching retention times and MS fragmentation patterns with authentic standards. Furthermore, these glycosides of benzoic acid derivatives have previously been detected in the extracts of a variety of plant taxa (Herrmann, 1978 ; Klick and Herrmann, 1988 ). For comparative purposes, methanolic extracts of line 5 and WT leaf tissues were also analysed (Figure S3 , Table S3 ). The two glucosides of p HBA and various other peaks that generate m /z 137.0246 fragmentation ions were important features in the chromatograms, indicating that the effects of CPL expression were not merely restricted to xylem tissue. Diverting carbon flux with CPL \n This study is the first to exploit the inherent plasticity of lignification in order to deliberately engineer p HB levels in poplar. Previously, it has been shown that p HBA is synthesized via p ‐coumarate and a metabolic branch of the phenylpropanoid pathway, and recent evidence favours a β‐oxidation route (Terashima et al ., 1975 ; Widhalm and Dudareva, 2015 ; Yazaki et al ., 1991 ). In this context, it is not surprising that metabolic flux through the lignin pathway can affect p HB content. For example, elevated p HB levels were observed in transgenic poplar deficient in the lignin biosynthetic enzymes caffeoyl‐CoA O ‐methyltransferase, cinnamoyl‐CoA reductase, C3′H, p ‐hydroxycinnamoyl‐CoA:shikimate p ‐hydroxycinnamoyl transferase, cinnamate 4‐hydroxylase (C4H), and 4‐coumarate:CoA ligase (Coleman et al ., 2008 ; Leplé et al ., 2007 ; Meyermans et al ., 2000 ; Peng et al ., 2014 ; Tsai et al ., 2020 ; Wang, Matthews, et al ., 2018 ; Zhong et al ., 2000 ). Conversely, p HB was reduced in F5H‐overexpressing poplar (Stewart et al ., 2009 ). Despite the increases in p ‐hydroxybenzoylation observed in CPL poplar, a large pool of p HBA remained unbound to the cell wall and accumulated as soluble xylem metabolites. Plants possess endogenous glycosyltransferase enzymes that can act on free p HBA to produce both the acid glucoside and the phenolic glucoside (Katsumata et al ., 1989 ; Lim et al ., 2002 ; Nishizaki et al ., 2014 ). These same compounds accumulated when CPL was expressed in tobacco, potato and sugarcane (Köhle et al ., 2003 ; McQualter et al ., 2005 ; Siebert et al ., 1996 ; Viitanen et al ., 2004 ). Furthermore, it has been reported that glucosylation of p HBA occurs cytosolically in Lithospermum and that these glucosides are sequestered in the vacuole (Yazaki et al ., 1995 ). As conjugation of p HBA with monolignols is also believed to occur in the cytosol, the implication is that glucosylation potentially competes with monolignol acylation. This could be problematic because metabolites that have been transported into the vacuole are presumably inaccessible to β‐glucosidase enzymes which could release the phenolic aglycones. However, when radiolabelled p HBA was provided to tobacco cell cultures, both the phenolic glucoside and the acid glucoside initially accumulated before a portion of the latter was subsequently incorporated into the cell wall (Li et al ., 1997 ). Although a fulsome cell wall analysis was not provided in that study nor in any of the earlier work with heterologous CPL expression in plants, this observation suggests that the acid glucoside of p HBA could potentially be cleaved and deployed to the apoplast, perhaps as cells mature, undergo cell death, and release their vacuolar contents. Glucosylation is not the only fate for surplus p HBA in the developing xylem of poplar. Various derivatives of the glucoside salicin occur naturally in poplar and willow (Boeckler et al ., 2011 ). These metabolites contain phenolics, typically benzoate or cinnamate, that acylate either the sugar, as in populin and tremuloidin, or the salicyl alcohol group, as in populoside (Braconnot, 1830 ; Erickson et al ., 1970 ; Pearl and Darling, 1959 ). Although p HBA‐containing versions have been synthesized (Stepanova et al ., 2014 ), these apparently have not been reported as naturally occurring. By examining the mass fragmentation patterns of metabolites that were differentially abundant in CPL poplar xylem, we have detected peaks that may correspond to p HBA‐containing derivatives of salicin in the developing xylem of CPL poplar (Figure 5b,d ; parent ions producing m / z 137.0246 and m / z 123.0445 product ions are coloured in magenta). Six of the metabolites with an apparent parent ion of m / z 405.1191 elute throughout the chromatogram (at 10.1, 24.1, 24.5, 25.0, 25.1, and 26.4 min), and these could be p HBA‐containing homologues of populin, tremuloidin, populoside, salicyloylsalicin, deltoidin, chaenomeloidin, trichocarposide or a fragment of larger metabolites such as p HBA‐containing homologues vigaureoside A, and symconoside A. Glucosides of gentisyl alcohol also occur in poplar, and these too can be acylated with phenolic acids as in salireposide and nigracin (Thieme and Benecke, 1967 ; Wattiez, 1931 ). One compound in the pool of line 5 xylem metabolites ( m / z 421.1091 at 21.8 min) might be a p ‐hydroxybenozylated homologue of these. Another one of the metabolites ( m / z 451.1246 at 26.5 min) could be vanilloylcalleryanin. Finally, two of the p HBA‐containing metabolites ( m / z 397.1503 at 8.4 and 23.9 min) could be homologues of grandidentatin or purpurein, a glucoside of 1,2‐cyclohexanediol acylated with p ‐coumarate which has been identified in the bark of Populus grandidentata and in the xylem of RNAi‐suppressed C3′H transgenic poplar (Coleman et al ., 2008 ; Pearl and Darling, 1962 ). The occurrence of p HB groups in these metabolites in place of other phenolics underscores the effectiveness of bacterial CPL in planta , and reveals substantial plasticity in the biosynthesis of these compounds in poplar xylem. It should be emphasized that, other than the glucosides of p HBA, salicylic acid, and vanillic acid presented in Figure 5c that were validated using reference standards, the remaining chemical designations are merely putative and provided here only for reference. Definitive identification of these metabolites would require isolation of each compound followed by structural determination by NMR and ideally comparison with chemical standards. Distribution of cell‐wall‐bound p ‐hydroxybenzoate Microscopic analysis of stem cross‐sections with phloroglucinol‐HCl and toluidine blue showed no obvious differences in xylem cell morphology nor lignin distribution (Figure S4 ). Time‐of‐flight secondary‐ion mass spectrometry (ToF‐SIMS) was then used to evaluate the distribution of p HB groups in xylem cross‐sections. The peak at m / z 121.03, which has previously been attributed to cell‐wall‐bound p HB (Goacher et al ., 2021 ), was used for mapping purposes and to prepare false‐colour overlay images (Figure 6 ). This analysis revealed that p HB groups were predominantly limited to the cell walls of fibres in the xylem of both line 5 and WT. An index calculated for p HB ion counts normalized to those of S‐lignin structures confirmed that the fibres of line 5 contained significantly more p HB groups compared with the WT control. When calculated for vessel cell walls, the index was again higher for line 5, but the values were lower due to the weak signal intensity of p HB in vessels. Figure 6 Distribution of cell‐wall‐bound p ‐hydroxybenzoate groups examined by positive‐ion ToF‐SIMS imaging. False‐colour ion images with p ‐hydroxybenzoate groups shown in magenta ( m / z 121.03) and lignin shown in green (sum of m / z 77.04 and m / z 91.05) for extractive‐free transversal xylem cross‐sections of line 5 and the wild‐type (WT) control, as labelled. The gradient scales on the right show ion intensities, and the white scale bars represent 50 μm. The p HB/S index was calculated as a ratio of ion counts for p ‐hydroxybenzoate at m / z 121.03 divided by the sum of S‐lignin at m / z 167.07 and m / z 181.05, and the values shown are the mean ± standard error ( n = 5 for line 5, n = 3 for WT). We have previously shown that p HB groups occur predominantly in the cell walls of xylem fibres in poplar wood (Goacher et al ., 2021 ). Although there was a small but proportionate increase in cell‐wall‐bound p HB groups in the vessels of CPL poplar, the levels remained much lower than in fibres. The selection of the cellulose synthase promoter likely ensured an ample supply of p HBA for all xylem cell types, but monolignol– p HB conjugates were apparently still not efficiently produced in vessels. This could be the result of limited expression of p ‐hydroxybenzoyl‐CoA:monolignol transferases, or perhaps related to the poor availability of sinapyl alcohol in vessels as these acyltransferases are known to prefer sinapyl alcohol as an acyl acceptor (de Vries et al ., 2022 ; Zhao et al ., 2021a ). In order to achieve p ‐hydroxybenzoylation in the G‐rich vessel cell walls of poplar, future engineering work may need to contemplate sinapyl alcohol availability or use acyltransferases that accept coniferyl alcohol as a substrate. Strategies to maximize p ‐hydroxybenzoylation of poplar lignin Realizing even greater p HB levels may ultimately require a multipronged approach. Recently, the p ‐hydroxybenzoyl‐CoA:monolignol transferase responsible for p ‐hydroxybenzoylation of poplar lignin was identified (de Vries et al ., 2022 ; Zhao et al ., 2021a ). A priority for future work will be to overexpress this gene in CPL poplar, but additional strategies should also be contemplated since monolignol acyltransferases alone may not achieve the desired titres of p HBA. For example, heterologous expression of p ‐coumaroyl‐CoA:monolignol transferases in poplar and Arabidopsis achieved at most 0.35 and 1.3% w/w cell‐wall‐bound p ‐coumarate, respectively (Sibout et al ., 2016 ; Smith et al ., 2015 ). One of the most impactful interventions would be blocking the formation of p HBA‐containing metabolites. UDP‐glycosyltransferases play a central role in detoxifying and modulating the accumulation of phenolics (Le Roy et al ., 2016 ). Unfortunately, the specific enzymes responsible for glucosylation of p HBA in poplar are currently unknown. Although the biosynthesis of salicinoids also remains largely obscured (Babst et al ., 2010 ), recent work has uncovered some of the steps involved in the formation of phenolic glycosides in poplar so these may soon be accessible engineering targets (Fellenberg et al ., 2020 ). Phenolic metabolites in the xylem of poplar could also represent an untapped fount of valuable compounds (Devappa et al ., 2015 ). Many salicinoid metabolites are active as defence compounds in poplar (Boeckler et al ., 2011 ), and some of these may also have pharmaceutical value. Furthermore, as many of the p HBA‐containing compounds in the xylem of CPL poplar were ester‐linked conjugates, treatment with alkali could be used to liberate the p HBA aglycones. In many ways, the accumulation of readily extractable p HBA as xylem metabolites in CPL poplar could be highly advantageous. As chorismate is an important precursor to a wide range of biosynthetic pathways in plants, the introduction of CPL could conceivably have pleiotropic effects. Chorismate lies upstream of protein biosynthesis, via the aromatic amino acid and shikimate pathways, and upstream of several plant hormones and signalling molecules such as indoles and salicylic acid (Wang et al ., 2022 ). In this study, the selection of a secondary cell wall‐specific promoter should have mitigated the impacts on the metabolome. However, to preclude any potential adverse effects, it could be worth trialling other promoter sequences, including those specific to lignin biosynthesis and those restricted to specific tissues and developmental stages. Impacts of ester‐linked monolignol conjugates Although the biological role of ester‐linked pendent groups remains unclear, it has been suggested that one result may be increased polymerization rates for S‐rich lignin (Takahama et al ., 1996 ). However, p HB groups are evidently negatively correlated with S units in poplar (Mottiar and Mansfield, 2022 ). There is also some evidence that p HB groups may promote plant defence as carrot cells respond to a fungal elicitor by incorporating cell‐wall‐bound p HB (Schnitzler and Seitz, 1989 ). Although plant growth was slightly impaired in CPL poplar, no irregular xylem phenotypes were observed and there was no indication that changes in lignin structure had other discernible effects on development. A recent study showed that gravitropism is altered in transgenic poplars with reduced p HB levels (Zhao et al ., 2021b ). Consequently, it is possible that the minor effects on growth observed in CPL poplar could perhaps be directly related to increased p ‐hydroxybenzoylation. Aside from any potential biological implications, elevated p HB levels led to improved biomass digestibility in assays of saccharification potential following various pretreatments. Transgenic poplar yielded greater amounts of glucose and xylose after 72 h of enzymatic hydrolysis following either acidic or alkaline pretreatments (Figure 7 ), commensurate with the observed increases in p ‐hydroxybenzoylation. For both glucose and xylose release, alkaline pretreatments performed the best. Compared to the WT control, line 5 released 10% more glucose and 50% more xylose following alkaline pretreatment. As the total amount of cell wall polysaccharides was actually slightly lower in line 5 compared with the control samples (Table S4 ), this increase in sugar release represents a bone fide improvement in saccharification. Previous lignin engineering efforts have achieved improved saccharification by significantly altering the lignin content, composition, and/or molecular weight (Mottiar et al ., 2016 ). In one of the few reports on the effects of pendent groups, overexpression of a p ‐coumaroyl‐CoA:monolignol transferase in Arabidopsis led to novel p ‐coumarate groups and significantly improved saccharification, although this was also accompanied by reductions in lignin content (Sibout et al ., 2016 ). Figure 7 Saccharification potential of mature wood was improved in transgenic lines. The release of glucose (shades of blue) and xylose (shades of green) is shown following various pretreatments (alkaline, acidic, and no pretreatment, as labelled) and 72 h of enzymatic hydrolysis. Those values marked with an asterisk are significantly different from the wild‐type (WT) control (one‐way ANOVA with Dunnett's test, n = 3 for each line, P ‐value <0.05). By contrast, CPL‐expressing poplar had only modest reductions in lignin (<5%), a small increase in β‐aryl ether units, and no change in average molecular weight. These observations suggest that the gains realized in saccharification could be specifically related to elevated p HB levels. It may be that these pendent groups alter the interactions between cell wall components or otherwise improve substrate access during biomass pretreatment and/or enzymatic hydrolysis. It has also been suggested that released p HBA acts as a blocking agent to prevent lignin condensation reactions and effectively enhance depolymerisation rates (Chua and Wayman, 1979 ). Beyond improvements in biomass processing, cell‐wall‐bound p HB groups are themselves potentially valuable as phenolic compounds. However, the industrial use of such clip‐offs will only be economical if the recovery processes are cost effective (Karlen et al ., 2020 ). In this light, CPL poplar trees with even higher titres of p HB could be an ideal feedstock candidate for biorefineries as alkaline hydrolysis would provide a pure stream of phenolics without the need for further separations unlike, for example, bioenergy grasses which would release heterogenous mixtures of p ‐coumarate and ferulate. In general terms, the lignin engineering strategy promoted herein could be applied to the manipulation of other clip‐offs as well. For example, increased p ‐coumaroylation and ferulolyation could be facilitated by enriching the pools of those respective phenolic acyl groups. As support for this approach, the levels of cell‐wall‐bound benzoate, which was only recently found to be compatible with lignification in poplar, were elevated in plants deficient in C4H and C3′H (Kim et al ., 2020 ). In summary, expression of bacterial CPL in transgenic poplar led to increased p ‐hydroxybenzoylation of lignin and enhanced saccharification potential of the wood, but no change in the distribution of p HB groups in the xylem. This work demonstrates the importance of substrate supply in engineering cell‐wall‐bound phenolics. However, abundant p HBA‐containing metabolites in these transgenic poplars also represent untapped potential. Accordingly, p HB groups warrant further research consideration in the context of lignin valorisation as these ester‐linked pendent groups can be clipped‐off for use in the production of a wide variety of biochemicals, biomaterials, and bioproducts."
} | 9,122 |
38204088 | PMC10780097 | pmc | 760 | {
"abstract": "Ice formation and accumulation on surfaces has a negative impact in many different sectors and can even represent a potential danger. In this review, the latest advances and trends in icephobic coatings focusing on the importance of their durability are discussed, in an attempt to pave the roadmap from the lab to engineering applications. An icephobic material is expected to lower the ice adhesion strength, delay freezing time or temperature, promote the bouncing of a supercooled drop at subzero temperatures and/or reduce the ice accretion rate. To better understand what is more important for specific icing conditions, the different types of ice that can be formed in nature are summarized. Similarly, the alternative methods to evaluate the durability are reviewed, as this is key to properly selecting the method and parameters to ensure the coating is durable enough for a given application. Finally, the different types of icephobic surfaces available to date are considered, highlighting the strategies to enhance their durability, as this is the factor limiting the commercial applicability of icephobic coatings.",
"conclusion": "5. Conclusions and Prospects The durability of an icephobic coating is the main factor slowing down the transfer from the lab to the real applications. It is rarely well addressed as many times the coatings are tested under mild conditions only, or because no icephobic performance is evaluated after the durability test. Here, we have reviewed the different durability methods and conclude that it is key to carefully select the appropriate parameters to ensure the coating will be durable enough for a given application. Often, the best solution for a specific application is not the one with the best icephobic performance but the one with the best balance between the icephobic performance and durability, considering as well other aspects such as cost effectiveness, ease of application, environmental sustainability and even the aesthetics of the coating. Comparisons between different systems might be difficult as there is a need for more standardization of the procedures and parameters in order to enable faster screening and material selection. Even when the same method is used in different studies, important parameters can vary from one to another, which hinders a direct comparison. Currently, these include key parameters such as the subzero temperature in the determination of the IAS or the sandpaper coarseness and pressure in an abrasiveness test. The paper also reviews the strategies to enhance the durability of the different types of icephobic coatings in order to broaden their potential applications. Traditional low-surface-energy approaches such as the use of SHSs or lubricated surfaces have been extensively investigated. An excellent icephobic performance has been already achieved (e.g., low IAS), so the efforts are now mostly focused on enhancing their durability as they typically suffer from mechanical weakness and lubricant depletion, respectively. There are other paths towards icephobicity, such as icephobicity through interfacial cavitation, which can be addressed by the use of low-shear-modulus materials; however, the low mechanical resilience limits their applications. Following the same path, there are other current approaches such as the use of PCMs or surface crack initiators, but more research is required in order to explore their potential and durability limits. On the other hand, there are recent studies focusing on the achievement of highly durable icephobic surfaces through materials with low interfacial toughness, which is very promising. The use of smart icephobic materials, mostly with PCMs or photothermal materials, have a lot of potential since they can activate their icephobic nature when required with no external energy requirement. The use of icephobic materials with self-healing capability is an appealing strategy to deal with harsh conditions, but their practical applications are mostly limited to those environments where the conditions to heal might naturally occur (e.g., self-healing at low temperature, achieved under sunlight). Finally, when durability is crucial under harsh conditions, the use mechanically stable nanocomposite icephobic coatings is a promising option. In such a case, the properties can be tuned to behave as durable low-interfacial-toughness materials and this even allows us to integrate other desirable properties in the development of advanced multifunctional materials, where active and passive icephobic approaches can be combined to meet the requirements of the most demanding icephobic conditions.",
"introduction": "1. Introduction Ice formation and accretion on surfaces, when not controlled or alleviated, are a great concern for many different industrial sectors, spanning from transport to clean energy or civil engineering. For instance, anti-icing/deicing (AI/DI) systems are of the utmost importance on aviation [ 1 ], as ice accreted on a wing increases drag and decreases lift [ 2 ] or can disturb the performance of key instruments such as the pilot tubes used to determine the airspeed [ 3 ]. Beyond the aerospace sector, other transport industries are also affected by icing events, such as the railway sector [ 4 , 5 ], where iced overhead catenaries can cause equipment malfunction and train delays. In the maritime sector [ 6 , 7 , 8 ] (where the offshore platforms can be also included), ice formation and accretion compromise the safety of marine operations in many different ways. Ice on the superstructure, deck, or walkways may cause accidents due to slippage or falling ice, or even capsize a vessel due to the loss of stability by lateral ice weight. Ice may jam mechanisms such as the anchor or block doors frozen shut. In the same way, ice accretion can drastically affect the performance of different systems related with the energy sector. Wind power systems’ efficiency can be reduced by ice accreted on the wind turbine blades [ 9 , 10 ]. Power line networks can be disrupted due to excessive weight on the overhead cables caused by ice accretion [ 11 ] and the sunbeams that reach solar cells can be scattered and reflected by an ice/snow layer [ 12 ], to name a few undesired scenarios. In order to prevent ice formation and accretion on a surface, a number of engineered solutions have been developed. The strategies to enhance the icephobic performance of a surface can be classified as single to a few icing event solutions, such as the application of salts for highways [ 13 ], glycols for aircrafts [ 14 ] or long term solutions such as icephobic coatings [ 7 , 15 , 16 ]. The application of chemicals is not only less effective but also implies concerns about its environmental impact [ 17 ], even though it is still the preferred solution in some specific cases due to its low cost. In order to determine the most suitable solution for a given application, several aspects are commonly considered. For example, the application of sodium chloride is by far the most widely used deicing strategy for road infrastructure due to its abundance and low cost [ 13 ], but it has important drawbacks to consider, such as the pollution of the soil and corrosion of the drainage systems and other infrastructures. Not in vain, asphalt concrete with improved icephobic performance is an active research area [ 18 ]. Likewise, there are efforts focused on the development of new AI/DI techniques for the aerospace sector, also triggered by the steady increment of carbon-fiber-reinforced polymers (CFRPs) replacing conventional aluminum-based aircrafts. The traditional solutions such as the application of freezing point depressants [ 17 ], the use of thermal-based deicers [ 5 , 19 ] or vibration/surface deformation-based systems [ 20 , 21 , 22 , 23 ] are usually combined with surfaces having a low ice adhesion strength (IAS) by using commercial icephobic coatings [ 24 ]. It is also very common to classify the icephobic approaches as active or passive, depending on the necessity of an external energy input or action. There are many automated mechanical methods, such as the use of pneumatic boots for aircrafts and wind turbines [ 25 ], the use of vibrational methods, either combined with ultrasonic guided waves [ 21 ] or with piezoelectric actuators [ 23 , 26 ], and even manual ones, such as the use of wooden baseball bats and similar tools, commonly used in offshore platforms and vessels operating in polar waters [ 27 ]. There are also thermal systems spanning from the heat pipes [ 28 ] traditionally used in aluminum-based aircrafts to electro-thermal systems [ 29 ] embedded in the laminated CFRP composites used in the last generation of aircrafts. According to this classification, passive/active approaches are considered AI/DI solutions, respectively. It is important to note that typical active deicing methods involve high energy consumption and the corresponding equipment is usually difficult to manufacture and requires considerable cost maintenance, and thus its use should be avoided or minimized as much as possible. On the contrary, a passive solution (icephobic surface) is always highly desired regardless of the presence/absence of an active deicing system, and in general terms, AI/DI solutions can be combined [ 30 ]. There are a countless number of applications where the ice formation and accretion should be avoided or at least minimized. In order to satisfy the requirements for a specific practical application, there might be other properties that an icephobic coating should meet, e.g., high transparency to UV/Vis light for solar cells or windows in the clean energy and transport sectors, respectively. To analyze the suitability of an icephobic coating for a specific application, the first aspect to consider will be the icephobic performance of the coating itself, and right after it, the longevity or durability, cost effectiveness and ease of application should be addressed, not necessarily in this order [ 31 ]. In addition, there are other aspects that might be important to consider, such as the environmental sustainability or even the aesthetics applications. Regardless of the application, the durability of an icephobic coating is a universal requirement given the fact that they are meant to remain effective under very harsh environments, usually outdoors, with very few specific exceptions such as freezers. In addition to the importance of the resilience of an icephobic coating on its long-term performance, there is a lack of consensus on how to measure and compare the durability of a coating. This review is intended to shed light on several important questions of icephobic coatings, bearing in mind the importance of their durability. The subject addressed starts with the primary basic question: what is an icephobic surface? The most common description of an icephobic surface is focused on its ability to prevent ice formation or facilitate its detachment. Thereafter, the following questions are tackled: How is ice formed and accreted? How can the icephobicity and durability of an icephobic surface be measured? How can the durability of different icephobic materials be enhanced? Finally, once these questions are discussed, some conclusions and prospects are drawn."
} | 2,821 |
24324575 | PMC3851020 | pmc | 761 | {
"abstract": "Coral bleaching is a significant contributor to the worldwide degradation of coral reefs and is indicative of the termination of symbiosis between the coral host and its symbiotic algae (dinoflagellate; Symbiodinium sp. complex), usually by expulsion or xenophagy (symbiophagy) of its dinoflagellates. Herein, we provide evidence that during the earliest stages of environmentally induced bleaching, heat stress and light stress generate distinctly different pathomorphological changes in the chloroplasts, while a combined heat- and light-stress exposure induces both pathomorphologies; suggesting that these stressors act on the dinoflagellate by different mechanisms. Within the first 48 hours of a heat stress (32°C) under low-light conditions, heat stress induced decomposition of thylakoid structures before observation of extensive oxidative damage; thus it is the disorganization of the thylakoids that creates the conditions allowing photo-oxidative-stress. Conversely, during the first 48 hours of a light stress (2007 µmoles m −2 s −1 PAR) at 25°C, condensation or fusion of multiple thylakoid lamellae occurred coincidently with levels of oxidative damage products, implying that photo-oxidative stress causes the structural membrane damage within the chloroplasts. Exposure to combined heat- and light-stresses induced both pathomorphologies, confirming that these stressors acted on the dinoflagellate via different mechanisms. Within 72 hours of exposure to heat and/or light stresses, homeostatic processes (e.g., heat-shock protein and anti-oxidant enzyme response) were evident in the remaining intact dinoflagellates, regardless of the initiating stressor. Understanding the sequence of events during bleaching when triggered by different environmental stressors is important for predicting both severity and consequences of coral bleaching.",
"introduction": "Introduction Coral bleaching is a physiological phenomenon in which the symbiosis between the coral host and its symbiotic dinoflagellate terminates [1] . As a result of environmental stressors, bleaching events can increase coral susceptibility to infectious diseases, reduction in reproductive fitness, and can eventually lead to the collapse of coral reef ecosystems [2] – [4] . Field observations of coral bleaching were first described in 1914 [5] , but it was not until 1925 that Boschma [6] provided evidence that the coral's symbiotic dinoflagellates were digested by the host animal. Yonge [7] and Yonge and Nicholls [8] challenged this theory by arguing that the symbiotic dinoflagellates were expelled from the endoderm of the cnidarian, and not digested. Their expulsion theory was corroborated to occur in sea anemones by Smith [9] , and went unchallenged until the work of Steele and Goreau [10] , who reasserted that dinoflagellates were digested. In subsequent years, strong evidence for in situ degradation of dinoflagellates was demonstrated by a number of workers, both as a function of normal physiology and bleaching [1] , [11] – [14] . Recent evidence substantiates expulsion as a mechanism of bleaching [15] – [17] . Thus, coral bleaching may result from a number of non-exclusive mechanisms, including host-cell detachment, Vibrio infection, viral-induced lysis of zooxanthellae, and zooxanthella programmed-cell-death [18] – [21] , though the trigger(s) for the initiating these processes, as well as the processes themselves, remain elusive. In recent decades, studies of bleaching predominantly focused on what happens to the symbiotic dinoflagellate (aka, zooxanthella) during a bleaching event, and whether dissociation of the symbiosis initiates by the dinoflagellate symbiont or by its host. For example, in hospite degradation of the dinoflagellate occurs in several coral species during natural (field) high-temperature or high-light events, either via self-induced dinoflagellate degradation or host xenophagy [11] – [13] . Dunn and co-workers [20] argued that in sea anemones, algal programmed-cell-death may be a prominent mechanism by which symbiotic dinoflagellates degrade, but their methodology did not sufficiently distinguish between a general necrotic response and programmed-cell-death. In corals, symbiotic dinoflagellates can induce several cellular acclimatory defenses that correlate with increased tolerance to bleaching-associated stress. These defenses include induction of mycosporine-like amino acids, heat-shock proteins, anti-oxidant enzymes and compatible solutes, and changes in photosynthetic accessory pigments [22] – [25] . Induction of reactive oxygen species, accumulation of oxidative damage products, and degradation of Photosystem II also have been correlated with many environmental inducers of bleaching, such as heat stress and light stress [23] , [24] , [26] , [27] . To date the specific role each stressor plays during bleaching is unclear, and their exact mechanism(s) of action and time-sequence of occurrence remains unknown [24] , [28] , [29] . Field observations of sudden-onset solar bleaching (high-light-induced bleaching) in the coral Goniastrea aspera indicated that dinoflagellate loss resulted from in hospite algal degradation; gastrodermal cells hosting dinoflagellates exhibited progressive degradation of the dinoflagellates; and perplexing behaviors of chlorophyll a and c concentrations occurred during the bleaching process [13] . Subsequent investigation examining the west-east bleaching behavior of G. aspera inhabiting tidal-flats in Phuket, Thailand, showed the importance of the host's physiological processes as a factor in bleaching [23] . Shallow-water colonies exposed to high light during seasonally low tides had higher concentrations of host antioxidant enzymes and heat shock proteins than polyps on less-exposed sides, and therefore lost fewer symbiotic algae [23] . However, the investigators were unable to deduce the role the symbiotic dinoflagellates played in either initiating bleaching or possibly tolerance to bleaching [23] , adding to the controversy concerning the dinoflagellate as a determinant of bleaching [1] , [14] , [28] , [29] . Recent work from our laboratories indicates that during a bleaching event, symbiotic dinoflagellates are digested by the coral host using an autophagy-associated pathway termed symbiophagy [30] . It remains unclear if the dinoflagellate is affected during the initial phase of bleaching, before symbiophagy is morphologically evident in the holobiont. To address this question, the morphological and cellular responses occurring in the in hospite dinoflagellate symbiont of the coral Pocillopora damicornis , were characterized by subjecting them to acute exposures of heat-stress and light-stress and combined stress bleaching conditions over a period of five days. Changes in the dinoflagellates' cellular structural-integrity, accumulation of oxidative damage-products, heat-shock protein and anti-oxidant protein responses, as well as photosynthetic pigments were examined in depth, in the early stages of the experimentally induced stress.",
"discussion": "Discussion Bleaching is a host-driven response that acts on the symbiont via either symbiophagy or expulsion [30] . One hypothesis is that bleaching is initiated by host-specific mechanisms independent of any influence from the dinoflagellate (e.g., heat-stress induced xenophagy; [30] , [40] – [41] ). An alternative hypothesis is that altered algal physiology initiates changes in the symbiotic equilibrium [20] , [27] – [29] . Numerous studies correlate reduced Photosystem II efficiency and oxidative stress with heat- and light-induced coral bleaching [23] – [24] , [26] . Some studies have correlated oxidative stress and heat- and light-induced bleaching, and proposed that deficient Photosystem II activity may be the primary source of oxidative stress, both in the dinoflagellate and host [23] – [24] , [26] , [42] . Despite advances in understanding the mechanisms of coral bleaching, difficulties remain in assessing causation, particularly in (1) determining the role and time-frame in which temperature and light act during a bleaching event, and (2) their relationship to other environmental stressors that ultimately induce bleaching [23] , [29] , [43] . Here, we provide evidence that temperature and light induce different pathologies in the symbiotic dinoflagellate during the early onset of bleaching, indicating that these stressors act on the algae by different mechanisms. Furthermore, we provide evidence that the sources of oxidative stress and the final receptors of its damage differ when bleaching is induced by heat-stress compared to light-stress. Heat stress, whether applied on corals maintained under low intensity light or prolonged darkness, induced dispersion of thylakoid membranes within the chloroplasts. Chloroplasts of vascular plants display few observable effects of thylakoid disorganization at 35°C, and structural deformities only appear around 45°C [44] – [46] . Plastid deformities in higher plants include swelling of the thylakoid lumen, phase separation of non-bilayer lipids and formation of galactolipid-enriched inverted micelles; none of which were observed with heat-stress in the low-light or darkness treatments [44] – [46] . Dinoflagellate chloroplasts lack grana, which in vascular plants have different lipid, protein, and isoprenoid compositions than unstacked thylakoids [47] . The long rows of lamellated thylakoids, as seen in symbiotic dinoflagellates, may arise from their unusual composition, and result in rather unique behaviors at temperatures considered non-stressful to thylakoids in higher-order algae and plants [47] – [48] . Our study indicates that thylakoid dispersion in these dinoflagellates directly resulted from increased temperature rather than oxidative stress because the same pathomorphologies appeared in the prolonged darkness/32°C treatment, where no substantive oxidative stress was apparent. One interpretation is that the oxidative stress observed in the low light/32°C treatment resulted from physical disruption of photosynthetic electron transport, which then induced photo-oxidative stress. In higher plants, heat stress (e.g., 32°–38°C) destabilizes the oxygen-evolving-complex of Photosystem II, thereby leading to acceptor-side photoinhibition [49] – [50] , but it also causes physical dissociation of the light-harvesting complexes from the Photosystem II core complex [51] – [53] . Determining which pathway is the more thermolabile in symbiotic dinoflagellates may provide insight into the role of thylakoid membrane integrity in photoinhibition. Light stress affected algae differently than heat stress, by condensing thylakoid lamellae and excluding the lumen. These changes could result from lipid auto-oxidation/fixation, where the by-products of lipid autoxidation, such as hydroxynonenals and aldehydes (e.g., malondialdhyde and formaldehyde) crosslink proteins, isoprenoids, and lipids, resulting in aggregate inclusions [54] – [55] . This in turn, adversely affects photosynthetic electron transport, and could explain why bleaching resulting from high-light stress correlates with decreased photosynthetic efficiency and increased damage to Photosystem II [23] – [24] , [27] , [55] – [56] . Oxidative stress was more pronounced in light-stress treatments compared to heat-stress treatments, likely because of photo-oxidative generation of a damaged photosynthetic electron transport chain [57] – [59] . Significantly greater accumulation of chloroplast sHsp in light-stressed samples compared to the low-light/32°C samples suggests increased stress on Photosystem II function [60] – [61] , which is independent of heat stress. Significantly lower levels of mitochondrial MnSOD in the high-light/25°C treatment compared to the low-light/32°C treatment indicates that oxidative stress is not predominately in mitochondria. Greater accumulation of mitochondrial MnSOD in the low-light/32°C and dark/32°C treatments denotes that heat stress is causing a significant oxidative stress in the mitochondria, and perhaps in the cytosol, as indicated by elevated glutathione peroxidase levels. Differing patterns of stress protein levels suggest that elevated temperature and light affect dinoflagellates via different mechanisms. For example, concentrations of Hsp70 were significantly lower at 25°C under both high and low light compared to 32°C, and concentrations of glutathione peroxidase were significantly higher in light-stressed corals regardless of the temperature. To differentiate between the effects of light and temperature on the symbiotic zooxanthellae, corals were exposed to prolonged darkness at both ambient and high temperatures. Electron microscopy did not reveal major pathomorphologies in the dinoflagellates of corals maintained in prolonged darkness at 25°C. In these samples, only slight vacuolization of the host tissue around the zooxanthellae was observed, while those in the 32°C treatment showed major changes. Furthermore, oxidative damage markers were significantly lower in dark-maintained corals from higher temperatures compared with the reference. These results were not unexpected because oxidative stress loads are predominantly driven by a photo-oxidative mechanism [62] . Our results strengthen the argument that heat and light stress differ in their modes of action: heat stress damages thylakoid structures prior to extensive oxidative damage, while in light-stress conditions, photo-oxidative stress is initially induced, resulting in structural damage to the chloroplasts. From the tissue to the sub-cellular level, the sequence of events and the time scale over which they occur have been largely ignored in studies of coral bleaching. The best time-course of bleaching (natural or laboratory induced) reported to date is the annual solar bleaching event of the Goniastrea aspera intertidal reef in Ko Phuket, Thailand where initiation of bleaching to the recovery from bleaching was observed to occur over a 15 day period [1] , [13] . Le Tissier and Brown [13] examined changes in zooxanthellae density, histology, photosynthetic efficiency, and photosynthetic pigments of corals before bleaching, during bleaching and through the recovery phase of a bleaching event. The onset of bleaching in the present study was much more acute than in the time course from Ko Phuket, but results of the two studies are consistent. Le Tissier and Brown [13] saw phenomena similar to that reported in this study; the symbiotic dinoflagellates were ultimately digested, as well as the peculiar behavior in chlorophyll a concentrations. However, they could not discern the nature of the dinoflagellate damage or the source of the damage, nor could they explain the unusual behavior of chlorophyll a . Our experiments demonstrate that even before morphological signs of plasma membrane and thecal plate degradation (occurring from symbiophagy) [30] , the zooxanthellae incur tremendous internal damage as a direct result of either temperature or light stress. Electron micrographs of corals in the high-light/25°C treatment showed zooxanthellae exhibiting minor presentations of thylakoid and chloroplast membrane pathomorphologies without significant symbiophagic vacuolization. This pattern suggests that zooxanthella damage occurs before the onset of symbiophagy. Chlorophyll a levels did not differ significantly among any treatments over 4 days ( Fig. S3 ) although the ratio of chlorophyll a -like compounds∶total chlorophyll a indicated that the large variance in total chlorophyll a levels resulted from accumulation of chlorophyll a -like products ( Figs. S1 and S2 ) [63] and that over the bleaching event, zooxanthellae that were retained in the host must have expressed lower levels of these catabolites. Our study provides evidence for a specific sequence of events that occurs during bleaching, regardless of whether the initiating stressor is heat, light, or the combination of both. Damage to zooxanthellae is a critical first step. This is characterized by ultra-structural damage to organelles, accumulation of biochemical lesions, and changes to metabolic efficiencies [13] , [23] . One of the first homeostatic responses to cellular injury reported here and found across all taxa of life, is induction of stress proteins [64] . These proteins mitigate damage by stabilizing metabolic pathways and cellular structures [60] , [65] . As time progresses beyond the initial shock, other homeostatic defenses come into play, particularly changes in the cell's redox capacity, i.e. , anti-oxidant enzymes, solutes, and isoprenoids. Changes in membrane composition and protein isoforms also can occur, enhancing the inherent stability of cellular structures and metabolic processes in the face of a persistent stress [66] – [68] . Recognizing the sequential nature of this process and characterizing these steps allows us to identify thresholds that determine whether a coral succumbs to environmental conditions conducive to bleaching or has sufficient homeostatic capacity to tolerate the stress event. For example, symbiotic dinoflagellate clades with an inherently greater tolerance to heat stress are known to have higher concentrations of specific unsaturated lipids, which provide increased lipid stability and fluidity [69] . Elucidating sub-cellular and cellular processes, such as molecular switches that activate symbiophagy or exocytosis, are necessary to differentiate between proximate mechanisms involved in physiological acclimation and those responsible for evolutionary adaptation. Understanding the role each of these processes play in a given bleaching response provides insight for predicting its severity, ultimate population effects, as well as the feasibility of potential mitigation options."
} | 4,482 |
31097757 | PMC6522478 | pmc | 762 | {
"abstract": "Self-healing materials with the ability to partially or completely restore their mechanical properties by healing the damage inflicted on them have great potential for applications where there is no or only limited access available to conduct a repair. Here, we demonstrate a bio-inspired new design for self-healing materials, where unit cells embedded in the structure are filled with a UV-curable resin and act as reservoirs for the self-healing agent. This design makes the repeated healing of mechanical damage possible. When a crack propagates and reaches one of these embedded reservoirs, the healing agent is released into the crack plane through the capillary action, and after polymerization through UV light exposure, bonds the crack faces. The structures here were fabricated using a stereolithography technique by a layer-by-layer deposition of the material. “Resin trapping” as a unique integration technique is developed for the first time to expand the capability of additive manufacturing technique for creating components with broader functionalities. The self-healing materials were manufactured in one step without any needs for any sequential stages, i.e. filling the reservoir with the healing agent, in contrast with the previously reported self-healing materials. Multiscale mechanical tests such as nanoindentation and three-point bending confirm the efficiency of our method.",
"conclusion": "Conclusion In summary, we demonstrated a structure capable of self-healing thanks to its biomimetic architecture. These novel structures were created through a stereolithography AM technique as it facilitates the fabrication of complex structures. Healing is accomplished by incorporating reservoirs of UV-curable resin that act as a self-healing agent within the structure. The reservoirs are able to provide the healing agent that is trapped inside them during the fabrication for multiple damage sites or sequential damages. The damage-induced triggering mechanism offers site-specific healing. When a crack forms, the healing agent leaks out to the crack location due to the capillary force. The small amount of UV-curable resin becomes cured when is exposed to the UV light for a short period of time. After a sufficient exposure time, the cracks are repaired, and the structural integrity of the specimen is restored. As the crack reopens under subsequent loading or new cracks form, the healing cycle is repeated. The average healing efficiency of the structures introduced here was ~52%. This value compares favorably with that for the previously reported self-healing materials. Self-healing of the microcracks not only alleviate the effects of environmentally- and functionally-promoted crack initiation such as the stress corrosion cracking, thermal expansion cracking, and fatigue-related cracking, but also mitigate the need for inspection methods. The design reported here can enhance the reliability and working life of many polymer-based additively-manufactured materials. This platform enables not only new avenues for continuous delivery of healing agents for self-repair of a damage that can occur multiple times, but also offers a new way of embedding other active species for additional functionalities.",
"introduction": "Introduction Drawing inspiration from nature for the purpose of solving complex engineering problems has been a technique employed by scientists and engineers for many decades. Many biological materials are hierarchical structures with sophisticated architectures, capable of self-healing and regenerating their functionality after the infliction of damage by external mechanical loads. While studies of biological systems continue to suggest new opportunities, fabrication challenges remain. However, research into applications of their analogs in real devices is becoming more prominent thanks to new developments in advanced manufacturing techniques. For instance, microvascular networks with complex and delicate architectures such as leaf venation 1 – 5 and blood vascularization 6 – 9 that are widely observed in biological systems for targeted delivery of nutrients for growth and healing, may also be replicated in transformative synthetic materials via various techniques, including soft lithography 10 – 12 , laser ablation 13 , 14 , and direct-write assembly 15 , to enable repetitive healing of damage. Similar to the way that human skin undergoes repeated healing of isolated damage, self-healing materials are an artificially-synthesized class of materials capable of repairing damage once or even multiple times, countering degradation of the material and expanding its lifetime, reliability, and efficiency. They are of interest particularly for applications where long-term reliability is required in poorly accessible areas or the structure is subjected to repeated damage. Although most reports of self-healing materials are on polymers or elastomers, self-healing refers to any materials where mechanical damages are healed automatically or with minimum external intervention, resulting in full or partial restoration of the material’s mechanical properties 16 . This definition covers all classes of materials, including metals, ceramics, and cementitious materials that take advantage of the same underlying principles. It needs to be noted that after the material is amended, it is the functionality that is restored rather than the exact external shape or internal micro-structure 16 . White et al . 17 introduced self-healing polymer composites that incorporate microencapsulated healing agents along with an embedded catalyst. Upon introducing a microcrack to the microcapsule, the healing agent is released to fill the crack plane via capillary action. Healing was achieved when the healing agent is polymerized in contact with the catalyst. While including unterminated chain-end polymerization catalyst enables multiple healing events, when the capsules are depleted from healing agent, no subsequent healing was possible if a new crack were to occur. Also, the capsules need to have the right thickness since capsules with thin walls break while processing and capsules with thick walls do not break as the crack propagates. Toohey et al . 18 fabricated a polymer-based structure with an embedded 3D microvascular network using a direct-wire assembly of a fugitive ink. The network was filled with a healing agent and upon crack formation in the coating, the agent, dicyclopentadiene (DCPD), is supplied by the network to the crack planes. In their design, the presence of the microchannel weakens the mechanical strength of the structure. In addition, several fabrication stages are needed to finalize the network. Limitation in the amount of the self-healing agents, as well as the requirement for the injection of the self-healing agent in the network and effectively sealing it are other downsides of this design. While this design enables the healing process for up to seven repairs, the self-healing is limited for any subsequent damages due to the depletion of the healing agent. Even though some of these issues were addressed in other work by the same group 19 , major challenges, such as the need for the two parts of the healing agents to be mixed diffusively within the crack plane and complicated fabrication procedure, still remain. In addition, the difficulty associated with the injection of not just one, but a few components of the healing agent in the network in these structures is a challenge. This limitation makes the need for an integration of micro valves and pumps inevitable for an efficient system 18 . It remains an opportunity to explore the self-healing process in different materials for various applications and novel technological implementations. For instance, a combined microcapsule–microvascular system was recently reported to amend multiscale damages from an impact puncture of vascularized polymeric sheets 20 . Kang et al . 21 reported a new class of self-healing stretchable polymeric material, crosslinked through precisely designed multistrength hydrogen bonding interactions, for potential applications in electronic (e-) skin. Kim et al . 22 reported thermoplastic polyurethane (TPU) for possible implementation in the wearable electronics industry. The polymer-based structure was able to reobtain more than 75% of the mechanical properties of the virgin sample within 2 hours after splitting and bringing the two pieces in contact. Moreover, significant research efforts have been invested in self-healing cementitious materials 23 – 25 , composite materials 26 – 28 , and metals 29 . Additive-manufacturing (AM) and the complexities enabled by it are revolutionizing manufacturing, further than just the geometry of the part, but also the chemistry and microstructure within the part with site-specific properties. Many polymer-based structures that are designed and fabricated by AM, however, tend to undergo damage that manifest commonly in the form of cracks, compromising the integrity of the structure and eventually limited functionality. The formation of these cracks, particularly deep within the structure, makes certification and qualification of additively manufactured polymer-based structures challenging, specifically since their detection and repair procedure are excessively prolonged and costly, if not impossible. Thus, while the AM technique is promising for various purposes such as low production-quantity parts with complex geometries, weight reduction of a system through part consolidation or topology optimization, design customization, rapid design iteration, product development, etc., the reliability issue remains to be addressed. Herein, we introduce self-healing AM structures with remarkable mechanical performance and regenerative ability through the careful design and fabrication of the structure’s geometry. The structures reported here are capable of repair of sequential damages. The repair mechanism is triggered as soon as the crack propagates beneath the surface of the structure and reaches the reservoir of the healing agent. The self-healing process is based on the generation of a mobile phase healing agent, which is “trapped” during the fabrication in the reservoirs, thus no secondary stages are required. The healing agent leaks into the crack planes as a result of the capillary force, and the crack is closed upon exposure to the UV light, completing the healing process.",
"discussion": "Results and Discussion To explore the novel biomimetic self-healing structures, a design that mimics the architecture of the one seen in puntia, also called prickly pear (a member of the cactus family 30 , 31 ) was implemented and fabricated. The self-healing structures were produced from a UV-curable resin via a stereolithography (SLA) AM method (Section I and Movie S1 , Supplementary Information,). The structures were designed with unit cells comprising vertical ribs (Fig. 1a ). The unit cells act as reservoirs that contain the healing agent. Due to the layer-by-layer nature of this 3D printing method and the design of the reservoirs, some resin is “trapped” inside of each unit cell without becoming cured as illustrated in Fig. 1b (Movie S2 , Supplementary Information). This uncured resin acts as a healing agent inside the self-healing structures, right beneath the top face plate. Similar to almost all opuntias with flattened pads (cladodes) that when cut, the mucilage cells ooze out, closing the opened area 30 , as soon as a crack forms and propagates to the reservoir of the self-healing agent, the agent wicks into the crack plane due to the capillary forces. The UV-photocurable resin has a viscosity of 850–1000 cps at 25 °C that promotes flow into the crack opening. This small amount of resin becomes cured upon being exposed to the UV light for a short amount of time (<120 sec) from a UV-light source. Figure 1c shows the healing agent leaking out of a crack and Fig. 1d depicts the curing procedure. Figure 1 ( a ) A CAD model of the self-healing structure (front view). The inset shows the geometry of the straight notch, ( b ) an actual 3D printed specimen with hollow reservoirs and trapped resin, ( c ) a schematic of the structure with an optical image inset, showing the leaked healing agent from a crack, ( d ) curing the leaked UV-photocurable resin, using multi-directional UV LEDs. The short curing time and consequently quick healing process is advantageous for these self-healing structures with embedded healing agent reservoirs compared to many other previously reported self-healing materials. For instance, self-healing materials with microvascular networks reported by Toohey et al . 18 needed to be kept at room temperature for a period of 12 hours in order to become healed. Self-healing materials with interpenetrating microvascular networks reported by Hansen et al . 32 had to go under cyclic bending (50 cycles at 100 μ m displacement) to enhance the mixing of the healing agents at the location of the crack, and after that required to be subjected to 48 hours of curing at 30 °C. Various mechanical tests were conducted to investigate the healing capability of the structures. For each test, three identical samples with an overall dimension of 5 × 10 × 77 mm (H × W × L) were printed using a top-down SLA-based 3D printer. A notch with equal width and depth of 200 µm was incorporated in the middle of the CAD (computer aided design) model of two out of three specimens (Fig. 1a ). This notch enhances the repeatability of the experiments and encourages the initiation of a straight crack under flexural (3-point bending) tests 32 . When a crack forms, propagates, and reaches the reservoir, the resin wicks into the crack planes as a result of capillary forces and closes the crack when it becomes cured under exposure to UV light. These forces are not high enough to drain out and deplete the large amount of healing agent in the reservoirs. The agent’s relatively high viscosity, approximately 850–1000 cps at 25 °C, further aids in limiting its flow out of the damaged area. After healing, the specimen is tested again, and a new crack is formed under a new critical load and the aforementioned process is repeated. The small amount of leaked healing agent in the self-healing samples becomes cured relatively quickly under the UV radiation with a wavelength of 405 nm. A UV-light source was employed to cure the leaked-healing agent for 3 min. at 50 °C. To compare the effectiveness of the capillary forces for filling the crack, the notch of the second sample was manually filled before the tests. The last unnotched specimen (virgin) remained unfilled and was tested to provide a reference. It is worth noting that for simplicity’s sake, the structures were placed into a UV oven for curing; however, other types of UV sources can initiate and complete the healing process. In the case of a difficult to access part, on-site repair can be easily implemented using a remote UV source. Additionally, unloading the structure is not necessary to the healing process. A damaged structure is able to cure under loading as the healing mechanism is not affected. The samples underwent tensile tests following the ASTM D638 standard and their force-displacement curves were recorded at a constant crosshead speed of 13 mm min −1 . Figure 2 shows the force-displacement curves for each specimen type. There was a difference of 22% between the tensile fracture load of the virgin specimen without a notch and the sample that was manually repaired. By comparing the fracture force of the healed sample before and after healing (182 N for Capillary – Cycle 1 and 199 N for Capillary – Cycle 2), it can be seen that the fracture force increases by around 17 N after the sample was repaired. There is a significant difference between the fracture force of the manually repaired sample and the sample after healing (Capillary – Cycle 2). This indicates that the self-healing process is effective and is reviving the original mechanical performance of the structure. Figure 2 Force-displacement curves of three different types of specimens under the tensile loads. In order to evaluate how the properties of the healed location differs from the rest of the structure, the site-specific nanomechanical properties of the healed crack and its vicinity were evaluated and compared with the rest of the structure, using a nanoindentation technique (Method and Section II, Supplementary Information). As can be seen in Fig. 3a , the cured healing agent at the crack location has an elastic modulus in the range of that for the rest of the structure. The effect of the exposure time on the elastic modulus and hardness of the sample is shown in Fig. 3b . As can be seen here, with an increase in the exposure time, both the elastic modulus and hardness increase until the sample becomes fully cured. Figure 3 Nanoindentation test results. ( a ) The elastic modulus of different locations on a cured specimen after 9 min exposure to a 405 nm UV light, ( b ) elastic modulus and hardness versus the exposure time for a cured specimen under the UV light. Figure 4a compares 3-point bending force-displacement responses of the self-healing structure before (Capillary – Cycle 1) and after a crack was formed, healed, and cured (Capillary – Cycle 2) under the UV light, with those for virgin, and manually repaired specimens. Although incorporating a notch into the virgin specimen lowers its bending load capacity for about 44% (compare Virgin and Capillary – Cycle 1), the notched specimen can be strengthened through either a manually self-repair or self-healing mechanism, for up to 22% and 32%, respectively. Again, the higher increase in the fracture load for self-repaired samples in comparison with that for the manually repaired ones under 3-point bending indicates the efficiency of our method. Figure 4 ( a ) Force-displacement curves from the 3-point bending of the specimens before and after healing process (Cycle 1, Cycle 2, respectively) as well as that for manually repaired and virgin samples, ( b ) a schematic of the experimental procedure for the crack detection using an acoustic sensor, ( c ) force-displacement curves for one specimen exhibiting four continuous healing cycles before fracture, ( d ) a schematic and an optical image of the self-healing structure after a crack is formed and the healing agent leaked out to the surface. The inset shows a zoomed-in OM image of the first and second crack. One of the main issues of self-healing materials is their limited number of self-repairs, commonly attributed to the depletion of the healing agent in the vicinity of the crack 18 , 32 . To assess the healing capability of the structure under sequential damages, cyclic 3-point bending tests were conducted on a notched specimen at a monotonic crosshead speed of 13 mm min −1 , following the ASTM D5045 standard. In order to accurately detect the occurrence of the crack, a high-impedance piezo electric transducer (AD-35 by Adeline) along with a recording interface (iRig Pre) were used (Fig. 4b ). The transducer senses audio vibrations through direct contact with the specimen under the test as soon as the crack initiates, while filtering the air vibrations (Figure S5 , Section III, Supplementary Information). An audio recording program was used to record the output for further analysis and the detection of the crack initiation time. A correlation was made between the collected data from the sensor and the recorded force-time data (Figure S6 , Section III, Supplementary Information). Upon the initiation of the crack, the load at which the crack was reopened was recorded and the specimen was exposed to the UV light and cured for 3 min at 50 °C. By doing this, the crack was successfully healed, and the test was repeated for further cycles, under the same testing conditions, until the sample fractured completely. As can be seen in Fig. 4c , with more cycles, the sample’s stiffness degrades and the crack initiates at a smaller load since with additional loadings microcracks form and propagate through the specimen, while might not even reach the reservoirs. These microcracks affect the integrity and as a result the strength of the structure. The fracture location was carefully evaluated in self-healed specimens. It was noticed that the self-healing structures failed at a new location rather than where the initial repair was conducted (Fig. 4d ), indicating that the specimens, were able to restore their homogeneity after the repair. It is worth noting that the ability for the structure to heal multiple times is attributed to the availability of a large amount of healing agent. Two factors control the flow of the agent from the damaged region. The material’s relatively high viscosity of approximately 850–1000 cps at 25 °C aids in introducing only a small amount of agent into the crack. Additionally, the curing of the agent upon exposure to UV light also slows the flow of the healing material, ultimately halting it once fully cured. While theoretically the healing can continue as long as there is healing agent left in the reservoirs, the structure might fail due to a catastrophic fatigue fracture. The healing efficiency, η, is commonly stated as the ratio of fracture toughness of the material after healing to the original fracture toughness. In this work, the healing efficiency was calculated from the ratio of the fracture force of the structure after healing to the fracture force of the as-printed structure. A similar approach was implemented in previous reports 18 , 32 . The time at which the crack initiated was used and the corresponding load was recorded to calculate the healing efficiency for each loading cycle. The healing efficiency for the first cycle was measured to be approximately 60% and an average healing efficiency of 52% was obtained for all four cycles. Figure 5 compares the healing efficiency of the structures reported here with the ones reported before. Our results surpass the ones reported previously for self-healing structures with microcapsules 17 as well as those for an interpenetrating single-network microvascular system 18 . The healing efficiency of our design is less than that of dual-network microvascular systems 32 , however, significantly shorter preparation and healing time make the design reported here excel when compared to other reported self-healing structures. Figure 5 The healing efficiency of the specimens for each cycle of loading, compared with reported data from microcapsule (black) 17 , single network (blue) 18 , and dual network (green) 32 ."
} | 5,667 |
35542157 | PMC9082455 | pmc | 763 | {
"abstract": "The present work reported a simple and effective approach to fabricate a low-cost, self-cleaning and mechanically durable superhydrophobic coating. The coating was prepared by dip-coating certain substrates in an ethyl acetate suspension of silica nanoparticles (SiO 2 ), hydroxyl acrylic resin, cross-linking agent and polyethylene wax (PEW). Through the control of the cooling and drying process, vapor-induced PEW micro-clusters were formed on the surfaces during the evaporation of ethyl acetate, and uniform carpet-like hierarchical structures were finally obtained by properly adjusting the dosage of PEW. Under the synergistic effect of hydrophobic SiO 2 nanoparticles and PEW micro-clusters, the composite coating exhibited a remarkable superhydrophobicity with a contact angle of 163° ± 5° with 25 wt% content of PEW, as well as preeminent self-cleaning properties against various food liquids. Moreover, the coating still maintained its surface cleanliness when immersed in the cyclohexane or hexadecane, indicating a superior self-cleaning property against solvent-contamination. The mechanical durability test showed that the coating still kept its excellent water repellency after fairly intensive knife-scratching, tape peeling and 25 cycles of sandpaper abrasion under 100 g of loading, indicating a quite admirable mechanical durability. The facile preparation and high-performance of the coating make it quite suitable for manufacture on a large scale, which is favorable for the development of superhydrophobic coatings.",
"conclusion": "Conclusions A self-cleaning and mechanically durable superhydrophobic coatings with hierarchical structures were successfully constructed by a simple vapor-induced method. The immersion suspension composed of ethyl acetate, silica nanoparticles (SiO 2 ), hydroxyl acrylic resin, cross-linking agent and polyethylene wax (PEW). Various substrates were dip-coated in the suspension, subsequently controllably dried, forming hierarchical structure surfaces with PEW whiskers outer layer in microscale and SiO 2 inner layer in nanoscale. The introduced hydroxyl acrylic resin with remarkable wear resistance and adhesion property effectively connected with SiO 2 nanoparticles, ensuring the hydrophobicity and durability of inner layer. Moreover, at a suitable protocol of the dip-coating at 30 °C and drying at 80 °C for 2 min as well as 25 wt% of PEW, uniform PEW whiskers formed in carpet-like shapes and the coating transferred into a superhydrophobic coating with a water contact angle of 163° ± 5°, excellent self-cleaning property even under solvent-contamination. Besides, the superhydrophobic coating could bear fairly intensive knife-scratch, tape peeling and 25 cycles of sandpaper abrasion under 100 g of loading.",
"introduction": "Introduction Superhydrophobic technologies derived from nature, typically like the “lotus effect”, have attracted much attention around the world both in academic research and in industry applications. Dirt particles on lotus leaves can be easily removed by water droplets due to the hierarchical structures of the surface, which are formed out of a characteristic epidermis and covering of waxes. 1 Similar non-wettable properties are also prevalent among animals, like water striders and flying insects. 2,3 Those surfaces show tremendous value in anticorrosion, self-cleaning, drag reduction, oil-water separation and other fields. 4–6 In general, the superhydrophobic property can be obtained using materials with low surface energy and by constructing morphological structures. 7,8 Many research works have been reported on artificial superhydrophobic surfaces by controlling the chemical composition of polymers 9 and metals 10,11 or by fabricating novel structures using various techniques such as phase separation, 12 polymer imprinting, 13 sol–gel processing, 14 chemical etching 15 and others. 16 However, these methods are usually complex and time-consuming. Therefore, it is crucial to develop a simple and effective approach to fabricate a superhydrophobic coating which is suitable for the manufacture in a large scale. Dip-coating is thought to be one of the most promising methods since it is low-cost and less demanding for production facilities. Furthermore, considering that the components applied in superhydrophobic surfaces like perfluoroalkylsilane are usually expensive and ecologically unfriendly, 17–20 it is more provident to develop a superhydrophobic surface using fluorine-free and sustainable materials. 21–24 Fabricating hierarchical structures in micro- and nanoscale is an important concept to create a superhydrophobic coating. Inorganic nanoparticles like SiO 2 and TiO 2 , combined with organic polymers, 25 are usually used to improve the hydrophobicity and surface roughness of the coating. Nevertheless, those hierarchical structures are usually fragile which exhibited poor resistance to mechanical scratch and abrasion since nanoparticles are difficult to cooperate with the organic polymer materials. One effective solution is to acquire a strong bonding of inorganic nanoparticles to the matrix resin by adding a cross-linking agent, 26,27 endowing the coatings with superhydrophobicity and abrasion resistance. Another useful method is to apply an external process like thermal annealing, to form “welding” between polymer matrix and particle interfaces, 28,29 which could greatly enhance the durability of hierarchical structures. Inspired by the epicuticular wax on lotus surface, some efforts have been made to render wax with superhydrophobicity through the fabrication of hierarchical structures, among which a typical method is phase separation. Some researchers used a combination of solvent and nonsolvent to create phase separation of wax. Lu et al. 30 prepared a low-density polyethylene superhydrophobic surface by controlling its crystallization behavior in a xylene–cyclohexanone mixture. However, those solvents were usually toxic and harmful, thus limiting the application of coatings. Other researchers chose mixed wax for the phase separation during drying process. Zhang et al. 31 endowed paper surface with superior superhydrophobicity via the phase separation of green-based waxes. Wang et al. 32 used the beeswax and carnauba wax to fabricate nontoxic superhydrophobic coatings that could be applied on food containers. Though hierarchical bulbiform shapes in microscale could be generated in this way, the components were stacked on the surface randomly thus might be easily buried underneath, which would lead to structure defects. In this work, a vapor-induced method using fluorine-free materials to fabricate superhydrophobic surfaces was reported. Organic components (hydroxyl acrylic resin, cross-linking agent as well as PEW) and inorganic components (SiO 2 ) were mixed in ethyl acetate to make a homogeneous suspension for dip-coating. The coating was endowed with a micro/nano-structure and enhanced roughness simultaneously through a simple control of preparation process, boosting the operation flexibility compared with the complicated methods reported previously. PEW could be dispersed evenly in ethyl acetate with other components due to its low melting point, therefore eliminating the usage of toxic organic solvent. By the virtue of the high reactivity of isocyanate groups with silanol and hydroxyl groups, hydroxyl acrylic resin and SiO 2 were cured by the cross-linking agent as a basic hydrophobic rough layer onto the surface of glass slides. Therefore, a rather reinforced coating was formed, partly offsetting the mechanical weakness of hydrophobic coatings. Moreover, PEW emulsions were crystallized into uniform whiskers under the induction of ethyl acetate vapor, and a carpet-like surface was formed with the increase of PEW content, which greatly increased the surface roughness and endowed the coating with superhydrophobicity and antifouling ability. Additionally, different preparation conditions were compared to explore their effects on surface morphology and superhydrophobicity. The coatings were characterized by scanning electron microscope, FT-IR and contact angle/interface system. Self-cleaning test and mechanical durability test were used to evaluate the performance of the composite coating.",
"discussion": "Results and discussion Molecular structures of nanoscale coatings The FTIR spectra of the components SiO 2 , hydroxyl acrylic resin, HDI trimer as well as the reaction products are shown in Fig. 1 . Based on the relative absorbance intensity of functional groups, it was discovered that hydroxyl groups of acrylic resin 3480 cm −1 and silanol groups of SiO 2 at 3414 cm −1 almost disappeared due to the possible reaction with isocyanate groups in HDI trimer, while C–H group at 2935 cm −1 in the acrylic resin and Si–O–Si groups at 1086 cm −1 in SiO 2 still remained. In the FTIR spectrum of HDI trimer, two intense peaks at 1688 cm −1 and 2277 cm −1 were assigned to stretching vibrations of C \n \n\n<svg xmlns=\"http://www.w3.org/2000/svg\" version=\"1.0\" width=\"13.200000pt\" height=\"16.000000pt\" viewBox=\"0 0 13.200000 16.000000\" preserveAspectRatio=\"xMidYMid meet\"><metadata>\nCreated by potrace 1.16, written by Peter Selinger 2001-2019\n</metadata><g transform=\"translate(1.000000,15.000000) scale(0.017500,-0.017500)\" fill=\"currentColor\" stroke=\"none\"><path d=\"M0 440 l0 -40 320 0 320 0 0 40 0 40 -320 0 -320 0 0 -40z M0 280 l0 -40 320 0 320 0 0 40 0 40 -320 0 -320 0 0 -40z\"/></g></svg>\n\n O bond and –NCO groups, respectively. However, the absorbance intensity ratio of –NCO groups to C O bond dramatically decreased in reaction products compared with that of HDI trimer, which indicated that most of the –NCO groups were consumed during the curing process and transferred into a –O–C(O)–NH– structure. Based on the above analysis, the possible chemical bondings among SiO 2 , hydroxyl acrylic resin, HDI trimer and glass slides are displayed Scheme 1 . The newly formed –O–C(O)–NH between glass slides and HDI trimer helped to form a stable and robust composite coating. Fig. 1 FTIR spectra of HDI trimer, hydroxyl acrylic resin, SiO 2 and reaction products. Scheme 1 Possible bonding structures among SiO 2 , hydroxyl acrylic resin and HDI trimer on glass slides. Micro-morphology of superhydrophobic SPEW composite coatings The SEM images as well as the images of water contact angles of surface of SPEW coatings with varied PEW weight ratio are shown in Fig. 2 . The particular PEW microscale structures were observed with the addition of PEW. When 10 wt% of PEW was added into the mixture, a small amount of PEW whiskers randomly dispersed on the coating surface, displaying a contact angle of 131° ± 2° ( Fig. 2a 1 ). With the increase of PEW content from 15 wt% to 20 wt%, more PEW clusters formed and the surfaces became much rougher ( Fig. 2b 1 and c 1 ), showing contact angles rising from 147° ± 3° to 153° ± 1°. The hydrophobic coating was successfully transferred into a superhydrophobic coating. As the PEW content was continuously increased to 25 wt%, a homogeneous micro- and nanoscale hierarchical structure was fabricated ( Fig. 2d 1 ), and the contact angle further increased to 163° ± 5°. In the SEM expansion images of coatings ( Fig. 2a 2 –d 2 ), the PEW whiskers in Fig. 2d 2 was observed to provide a more efficient air gap compared with those of other coatings, and acted as a full-scale protection to prevent the coating from being contaminated. Fig. 2 SEM images of micro–nanoscale structures of SPEW coatings with various PEW dosages. (a 1 ) 10 wt%, (b 1 ) 15 wt%, (c 1 ) 20 wt%, (d 1 ) 25 wt%; (a 2 –d 2 ) expansion images from (a 1 –d 1 ). Formation mechanism of superhydrophobic surface The formation mechanism of surface morphology is illustrated in Scheme 2 . The mixture was stirred at 70 °C until a uniform suspension was formed. The inorganic hydrophobic nanoparticles were easily suspended in the PEW emulsion due to similar polarities. In the followed cooling process until 30 °C, PEW crystal nucleuses in microscale were gradually deposited out of the mixture. Then glass slides were fastly dip-coated and put immediately in an oven at 80 °C for 2 min. As the solvent evaporated during the drying process, SiO 2 particles slowly precipitated onto the interface between mixture and slide due to a higher density, and formed a loose hydrophobic layer. Meanwhile, the solvent underneath was evaporated in a slower rate through the voids among nanoparticles. It is worth mentioning that by the virtue of a lower density, PEW crystal nucleuses gradually migrated to the upper layer during the evaporation of solvent. Since the drying temperature was 10–20 °C higher than the melting point of PEW, crystal nucleuses were transferred into a near molten state, and then slowly shaped to small whiskers induced by ethyl acetate vapor. The samples were then slowly cured at room temperature for 24 h, which provided a condition for the growth of PEW whiskers. Finally, a micro–nanoscale composite coating with relatively high surface roughness was fabricated. Scheme 2 Schematic illustration for the formation of superhydrophobic surface. In order to further investigate the influence of preparation conditions on surface morphologies, different coatings with 25 wt% of PEW were prepared, as the results are shown in Table 1 and Fig. 3 . Sample E was the same as sample d in Fig. 2 . Compared with sample C, D and E to cool to 30 °C, it took more time for sample A to cool to 25 °C, which resulted in lager crystal nucleuses of PEW. These crystals might be difficult to be transferred into a near molten state at 80 °C in 2 min, affecting the re-shaped process induced by ethyl acetate vapor, thus depositing in a random shape with a diameter of 1–3 μm on the surface, as shown in Fig. 3a . Process parameters of different coatings Samples Process Cooling Drying Contact angles A Cooling/drying To 25 °C, no stirring 80 °C, 2 min 132° ± 2° B Drying — 80 °C, 2 min 141° ± 1° C Cooling/drying To 30 °C, stirring 80 °C, 2 min 130° ± 3° D Cooling/drying To 30 °C, no stirring 80 °C, 5 min 135° ± 2° E Cooling/drying To 30 °C, no stirring 80 °C, 2 min 163° ± 5° Fig. 3 SEM images of coatings prepared in different conditions. (a) Cooled to 25 °C, (b) without cooling process, (c) cooling with stirring, (d) dried at 80 °C for 5 min. In contrast with sample A, many small PEW whiskers in a diameter of 100–300 nm formed on the surface of sample C ( Fig. 3c ). It might be owed to the shearing effect from stirring during the cooling process, which prompted the formation of many small crystal nucleuses. It was noticed that the coating of sample B, which was dried directly without cooling process, was netlike and consisted of entangled pores with a diameter in the range of several micrometers ( Fig. 3b ). It confirmed that PEW suspension under a molten state was thermodynamically unstable when evaporating the solvent. As a result, phase separation would occur to form a PEW rich phase and a PEW poor phase. The rich phase solidified and formed the matrix after drying the solvent, whereas the PEW poor phase formed the pores. In addition, the hierarchical structures fabricated by present method remarkably increased the surface roughness, displaying an increased water contact angle to 141° ± 1°. Since the drying temperature was much higher than the melting point of PEW, a similar effect on the surface morphologies was observed by raising the drying temperature or prolonging the drying time. Sample D was dried for a longer time compared with other coatings. As a result, PEW crystals were remelted into liquid drops during the evaporation of solvent. However, those drops wrapped with nanoparticles could not aggregate together and then were stacked randomly on the surface, as shown in Fig. 3d . Therefore, the detailed surface morphologies could be controlled by changing the cooling and drying conditions during the process, which was also reported by other researches, such as netlike hollow structures resulting from phase separation. 33 Nevertheless, the carpet-like morphology was seldom illustrated, and its performance was rarely investigated. Applications of SPEW coatings The coating of 25 wt% PEW content was chosen to investigate its applications due to its more compact micro structures and superior superhydrophobicity. Various substrates were used for the characterization of superhydrophobicity, including filter paper, cotton, textile, glass slides and tinplate, as shown in Fig. 4 . Dyed water droplets were placed on the white coated surfaces and the uncoated surfaces, respectively. The composite coating on various substrates exhibited an outstanding super-repellency property to water droplets, which still maintained their sphere-like shapes on different points of coated surfaces. For some substrates with certain inherent interspace, such as cotton and textile ( Fig. 4a and c ), water droplets easily penetrated through those uncoated surfaces. However, on all the coated surface droplets maintained their sphere-like shapes and easily rolled off from the surfaces, indicating a perfect adaptability of the composite coating on both hard (glass and tinplate) and soft (filter, paper, cotton and textile) substrates. It should be noted that the coated surfaces were white and non-transparent, which may partly limit their practical applications. Fig. 4 Dyed water droplets on the coated and uncoated surfaces of cotton (a), filter paper (b), textile (c), glass slide (d) and tinplate (e). Self-cleaning property of SPEW coatings In order to further evaluate the performance of composite coating, various liquids were used for the repelling test, such as some food liquids in the reported work. 34 Fig. 5 showed the pictures of these liquids on the composite coating, as well as the corresponding contact angles and sliding angles. Ketchup, orange juice, cola, vinegar and yogurt maintained the spherical shapes with high contact angles and low sliding angles while olive oil and cooking oil spread out on the surface, showing a lipophilic property of the coating. Since PEW and hydrophobic SiO 2 are both lipophilic, it's reasonable that the coating could not repel oil. And further addition of materials with low surface energy could be one of the potential methods to vest the coating with oleophobic property. 35,36 Fig. 5 Various liquid repelling tests on the coated surfaces (left) and their contact angles/sliding angles (right). Rolling processes of various liquids are shown in Fig. 6 , with contact angle images placed on the bottom right corner. Coated surfaces were put in a certain slope before test. For ketchup and yogurt tests, the coated slides were put in a higher slope (45° and 30°) for a smoother rolling process. When those liquids were dropped on the coating, they could completely roll off from the surfaces with no residual trace. Therefore, the coating could maintain its surface cleanness against both light liquids and thick liquids. Fig. 6 Rolling processes of ketchup (a 1 –a 3 ), orange juice (b 1 –b 3 ), cola (c 1 –c 3 ), vinegar (d 1 –d 3 ), yogurt (e 1 –e 3 ) on the coated surfaces. Self-cleaning tests under hydrophobic solvent system was rarely reported probably due to the loss of water repellency when superhydrophobic surfaces are contaminated by hydrophobic solvent. 37 In this work, the uncoated and coated glass slides with dust contamination were immersed in cyclohexane and the processes of removing the dust were recorded when water droplets rolled over the dust to further evaluate the self-cleaning property of the coating, as presented graphically in Fig. 7 . Fig. 7a 1 –a 3 showed that when water droplets rolled off from glass slides, they spread in a random shape, and the dust still remained along the path. In contrast, the composite superhydrophobic coating allowed water droplets to take the dust away from its rolling and bouncing path, as shown in Fig. 7b 1 –b 3 , indicating that the self-cleaning property was still retained even after cyclohexane-contamination. Changing cyclohexane to hexadecane, the same results of dust successfully removed on the coating surface were observed, further indicating an excellent self-cleaning property of the SPEW coating. Fig. 7 Self-cleaning test on uncoated surface (a 1 –a 3 ) and coated surface (b 1 –b 3 ) when surfaces immersed in cyclohexane. Durability of SPEW coatings Durability of superhydrophobic coatings is a crucial factor that limits their widespread applications, since micro- and nanostructures may be destroyed when faced with mechanical friction, leading to the loss of superhydrophobicity. Here, knife-scratch, sandpaper abrasion and tape peeling tests were used to evaluate the mechanical durability of the SPEW coating. As shown in Fig. 8a , sparse (A) and intensive (B) cross-scratches were applied respectively on the coated surfaces to imitate the scratch conditions in practical applications, and water droplets were placed on the surfaces after knife-scratch ( Fig. 8b ). It was noticed that the coating still retained their non-wetting property, regardless of sparse or intensive cross-scratches. The contact angles were 158° ± 2° (A) and 153° ± 2° (B) respectively, indicating that scratching did not induce a significant decrease of the superhydrophobicity on the SPEW coating. Fig. 8 Knife-scratch test. Sparse and intensive cross-scratches on the coated surfaces (a), water droplets on the scratched surfaces (b). In sandpaper abrasion test, water contact angles and sliding angles were recorded after a certain number of abrasion cycles, shown in Fig. 9 , and SEM images before and after 25, 40 abrasion cycles are shown in Fig. 10 . When the PEW whiskers that protruded onto the SPEW surface ( Fig. 10a ) were first removed, the contact angles decreased a little from 163° ± 5° to 159° ± 1°. As the number of abrasion cycles increased, part of PEW micro structures was destroyed, resulting in the decrease of contact angles and increase of sliding angles. However, the coating still kept its superhydrophobicity after 25 cycles of abrasion, with a contact angle of 152° ± 3° and sliding angle of 9° ± 2°. Fig. 10b showed that some PEW whiskers still remained and SiO 2 nanoparticles stacked on the whiskers, which helped to maintain the superhydrophobicity of coating. However, a significant decrease of contact angle was noticed after 30 cycles of abrasion, leading to the loss of superhydrophobicity. It might owe to the destruction of PEW whiskers microscale structures after multiple abrasion cycles, leaving the nanoscale structures consisting of SiO 2 to resist the abrasion from sandpaper, as SEM image shown in Fig. 10c . In the test after 30 cycles, the contact angles and sliding angles had no remarkable change, showing the increasing durability of nanoscale layer, which might relate with the firm bonding of SiO 2 with hydroxyl acrylic resin due to the reaction among SiO 2 , hydroxyl acrylic resin and HDI trimer. Fig. 9 Water contact angles and sliding angles of SPEW coating during 40 abrasion cycles. Fig. 10 SEM images of coatings before (a) and after 25 abrasion cycles (b), 40 abrasion cycles (c). EDX spectra of the coatings before and after 25, 40 abrasion cycles were also recorded to study the changes of surface element content, shown in Fig. 11 . For the surface almost covered with PEW whiskers before abrasion, the atomic signal of Si was relatively weak. Compared with the benchmark peak at 0 keV, the samples after 25 and 40 abrasion cycles showed much higher Si content on their surfaces, due to a raising number of bare SiO 2 nanoparticles during the process of abrasion test. Fig. 11 EDX spectra of coatings before (a) and after 25 abrasion cycles (b), 40 abrasion cycles (c). To further test the tolerance to common damages, the superhydrophobic coating was then treated with tape peeling, shown in Fig. 12 . It could be clearly observed that the PEW whiskers on the surface were removed by tape peeling but the nanoscale layer resisted the external force and remained. Under the combined effect of SiO 2 , hydroxyl acrylic resin and cross-linking agent, the coating could still maintain a favourable hydrophobicity, with a water contact angle of 140° ± 3° on its surface. Fig. 12 SEM images of coatings before (a) and after (b) tape peeling."
} | 6,099 |
33452425 | PMC7810751 | pmc | 765 | {
"abstract": "The brick-and-mortar architecture of biological nacre has inspired the development of synthetic composites with enhanced fracture toughness and multiple functionalities. While the use of metals as the “mortar” phase is an attractive option to maximize fracture toughness of bulk composites, non-mechanical functionalities potentially enabled by the presence of a metal in the structure remain relatively limited and unexplored. Using iron as the mortar phase, we develop and investigate nacre-like composites with high fracture toughness and stiffness combined with unique magnetic, electrical and thermal functionalities. Such metal-ceramic composites are prepared through the sol–gel deposition of iron-based coatings on alumina platelets and the magnetically-driven assembly of the pre-coated platelets into nacre-like architectures, followed by pressure-assisted densification at 1450 °C. With the help of state-of-the-art characterization techniques, we show that this processing route leads to lightweight inorganic structures that display outstanding fracture resistance, show noticeable magnetization and are amenable to fast induction heating. Materials with this set of properties might find use in transport, aerospace and robotic applications that require weight minimization combined with magnetic, electrical or thermal functionalities.",
"conclusion": "Conclusion The magnetically-assisted assembly and hot pressing of platelets pre-coated with a metallic iron layer is an effective approach for the manufacturing of metal-ceramic bulk composites with nacre-like architecture. With this approach, alumina-based composites with iron fractions up to 12.4 vol% were successfully obtained while maintaining the alumina platelet morphology during the high-temperature reduction and sintering steps involved in the process. Reaction of iron oxides with alumina at temperatures above 600 °C leads to the formation of the spinel interphase hercynite. This interphase is expected to improve the wetting of the metallic phase on the surface of the alumina platelets, thus enhancing the mechanical properties of the resulting composite. Crack deflection at the platelet surface and the presence of a more ductile iron phase in between the alumina platelets lead to a nacre-like architecture that exhibits increasing resistance against crack growth, also known as rising R-curve behavior. As a result of these toughening mechanisms, the bulk composites reach higher mechanical strength and fracture toughness levels comparable to the toughest nacre-like materials reported in the literature. In addition to these remarkable mechanical properties, the metallic iron and hercynite phases also impart soft magnetic response and tunable electrical properties to the final bulk composites. These additional functionalities make the metal-ceramic composites attractive in applications requiring lightweight structural materials that are not only stiff and tough but also sensitive to external magnetic and electrical fields. Such properties are potentially useful for the magnetic shielding of spacecraft and satellites exposed to space radiation. Because it relies on versatile sol–gel chemistry and particle assembly protocols, the proposed manufacturing technology should be applicable to a wide range of metal–ceramic compositions. Future research may identify solid-state reduction routes to increase the iron volume fraction of the nacre-like composites and create metallic alloys with enhanced resistance against crack propagation.",
"introduction": "Introduction Artificial materials that mimic the nacreous architecture of mollusk shells are prominent examples of how evolved hierarchical structures made by living organisms can be harnessed to fabricate synthetic counterparts with outstanding properties and new functionalities 1 – 3 . By combining design principles of nacre’s brick-and-mortar structure with the rich variety of chemistries available in synthetic systems, current nacre-like materials exhibit mechanical properties that even surpass those of the natural counterparts 4 – 9 . As opposed to conventional materials, such bio-inspired structures can be designed to showcase antagonistic properties, such as high stiffness and fracture toughness, that are not accessible through the optimization of chemical compositions alone 7 , 10 , 11 . Despite these promising prospects, a broader application of such bio-inspired materials requires the development of manufacturing technologies that ensure up-scalability and robust structural control. Several processing technologies have been developed in the last decade to manufacture synthetic films and bulk materials with nacre-like brick-and-mortar architecture 11 , 12 . Some of the several approaches proposed include the magnetic or shear-induced alignment of pre-formed platelets into layered structures 5 , 6 , 13 , 14 , the sequential deposition of individual layers into stacked composites 15 – 18 and the freeze-casting of suspensions into lamellar architectures 3 , 19 – 21 . The resulting anisotropic structures can be consolidated by sintering and/or infiltrated with a matrix to generate the desired brick-and-mortar architecture. The use of platelets has become particularly attractive due to the very fine microstructures obtained after consolidation and the increasing availability of two-dimensional materials with different chemical compositions 5 , 6 , 18 , 22 . Recent research efforts have been dedicated to increase further the fracture toughness and stiffness of nacre-like bulk materials or to complement these mechanical properties with additional functionalities. By utilizing building blocks with matched optical properties, nacre-like composites combining for instance crack growth resistance with optical transparency have been reported 22 . In another example, nacre-inspired materials featuring a graphene percolating network as continuous phase have been developed to imbue brick-and-mortar structures with self-sensing capabilities 23 . In addition to these functionalities, studies have also been conducted to enhance the mechanical properties of nacreous materials by utilizing stiffer and tougher metals as the mortar phase between stiff ceramic bricks 7 , 20 , 24 – 30 . Because of the plastic deformation of the metallic phase during fracture, composites containing nickel, aluminum or amorphous metal alloys between alumina platelets have been shown to feature remarkable mechanical stiffness and enhanced resistance to crack growth 7 , 20 , 31 . The use of metals as continuous phase also opens the opportunity to incorporate magnetic, electrical and thermal functionalities that have not yet been fully explored in nacre-like composites with mechanically-robust bulk geometries. Here, we report a processing route to create metal-coated oxide platelets that can be assembled into bulk nacre-like composites that combine enhanced fracture toughness with magnetically-responsive functionalities. This is achieved by utilizing metallic iron as the continuous phase between alumina platelets organized in a brick-and-mortar architecture. The processing route used to create this architecture and the properties of the resulting multifunctional composites are thoroughly investigated in this work. First, we describe a sol–gel route and a thermal reduction protocol to controllably coat pre-existing alumina platelets with well-defined fractions of iron. Next, the mechanical properties of nacre-like composites assembled from the pre-coated platelets are systematically assessed using state-of-the-art mechanical testing procedures. Finally, the magnetic properties of the composites are characterized and explored to create exemplary light-weight materials that can be inductively heated in an oscillating magnetic field.",
"discussion": "Results and discussion The formation of metal-coated platelets and their assembly into nacre-like metal-ceramic composites is achieved through a processing route that includes: (i) coating of platelets with a metallic or an oxide layer, (ii) possible reduction of the oxide layer to generate metal-coated platelets, (iii) assembly of the metal-coated platelets into nacre-like architectures, and (iv) pressure-assisted sintering of the nacre-like structure into tough multifunctional composites (Fig. 1 ). Figure 1 Processing route used for the fabrication of metal–ceramic nacre-like composites through the assembly of coated platelets followed by densification at high temperature. BnOH: benzyl alcohol. M: metal. MO x : metal oxide. H rot : rotating magnetic field. MASC: magnetically-assisted slip casting. SPS: Spark Plasma Sintering. Coating of the platelets with an oxide layer is accomplished through a non-aqueous sol–gel reaction using a metal–organic precursor dissolved in benzyl alcohol 32 , 33 . Taking iron acetylacetonate as precursor, a thin homogeneous layer of magnetite (Fe 3 O 4 ) nanoparticles is formed on the surface of the alumina platelets during the sol–gel reaction at 180°C 34 . The reaction is followed by removal of the solvent and burn-out of the organic phase in air at 700 °C (Fig. 2 ). X-Ray diffraction of the coated platelets reveals that the magnetite coating is converted to crystalline hematite (Fe 2 O 3 ) during the burnout process 35 . The bare alumina platelets present flat surfaces (Fig. 2 a), while the hematite forms a uniform percolating network comprised of nanoparticles on the surface of the platelets (Fig. 2 b,c). Bonds between the nanoparticles are likely caused by partial sintering during the burn-out step. Figure 2 Alumina platelets coated with different concentration of hematite. Scanning electron micrographs of ( a ) bare platelets and of platelets coated with ( b ) 10, ( c ) 23 and ( d ) 43 vol% of hematite. The imaged platelets were obtained after the sol–gel coating procedure followed by a burn-out step. Platelets shown in ( b ) and ( c ) were prepared using precursor/platelet mass ratios of 1 and 2, respectively. The high fraction of hematite formed on the platelet displayed in ( d ) results from two successive coating reactions performed with the same powder. ( e ) Volume fraction of hematite formed on the platelet surface as a function of the initial precursor/platelet mass ratio. The dotted line indicates the predicted volume fraction if all the iron added as precursor for the sol–gel reaction is converted into oxide particles on the surface of the platelets. Optical images of the powders synthesized with increasing precursor/platelet ratios. Scale bars: 500 μm. Importantly, the volume fraction of hematite particles on the platelet surface can be controlled by tuning the relative amount of precursor used in the sol–gel reaction (Fig. 2 e). By comparing the volume fraction of hematite measured on the surface with values predicted assuming a 100% yield, we conclude that not all the iron oxide nanoparticles formed during the reaction are eventually adsorbed on the surface. This is further confirmed by the deep black color of the solvent obtained after filtration of the platelets at the end of the coating procedure. Despite the incomplete adsorption, the fraction of iron oxide nanoparticles formed on the surface is not far from the theoretical predictions for 100% yield (Fig. 2 e). This allows us to use the theoretical prediction as a good first estimate of the precursor/platelet ratio needed to reach a given volume fraction of hematite on the platelet surface. In order to achieve hematite concentrations higher than 25%, successive sol–gel reactions can be performed on the same powder. This successive coating yields additional layers of hematite on the top of already-coated platelets (Fig. 2 d) and increases the coating volume fraction up to 43% while minimizing the fraction of nanoparticles formed in solution. The increase in concentration of iron oxide formed on the surface of the platelets prepared with higher precursor contents is readily visible from a change in the color of the filtrated and dried powder from light to dark red (Fig. 2 e). The alumina platelets coated with hematite nanoparticles were further processed in order to transform the oxide layer into a metallic coating. This was performed by heat treating the oxide-coated platelets in a reducing 5% H 2 /N 2 atmosphere at high temperatures. To identify the different phases formed during the reduction process and establish the minimum temperature required to form the metallic iron coating, we conducted ex-situ X-Ray diffraction analysis on platelets coated with 10 vol% hematite and treated for 8 h at distinct temperatures up to 1000 °C. The experiments show that five different crystalline phases may be present in the coating depending on the temperature used for reduction of the initial hematite particles. Rietveld refinement of the obtained diffractograms allows us to quantify the amount of the various phases formed after the reduction step (Fig. 3 ). Figure 3 Crystallography and morphology of the platelet coating after heat treatment at distinct temperatures under reducing atmosphere in a small-batch process. ( a ) Volume fraction of each crystalline phase present in coated platelets subjected to different reduction temperatures for 8 h. The initial coated platelets consist of 89% Al 2 O 3 and 11% Fe 2 O 3 . The lines between experimental data points are guides to the eye. ( b ) SEM images of the platelet coating at selected temperatures, highlighting the morphological changes associated with the phase transformations of the initial hematite particles. Magnification: 85 k ×. Our quantitative analysis shows that the initial hematite phase (Fe 2 O 3 ) is partially and fully reduced to magnetite (Fe 3 O 4 ) after reduction at 300 and 400 °C, respectively. Further reduction of the magnetite phase into either wustite (FeO) or metallic iron is then observed when the platelets are thermally treated between 400 and 600 °C. The volume fraction of metallic iron in the coating after reduction at 600 °C is estimated to be 4.7 vol%. Thermal treatment at the higher temperature of 700 °C leads to the formation of the alumina-iron spinel FeAl 2 O 4 . This mixed oxide phase, called hercynite, is likely formed by the diffusion of iron atoms to the alumina crystalline phase at the platelet/coating interface. When the temperature is increased beyond 700 °C, the fraction of the spinel phase decreases until it is fully converted into metallic Fe at 900 °C. The iron content of 5.2 vol% measured by X-Ray diffraction after full reduction above 900 °C is in good agreement with the nominal value of 5.4 vol% estimated from the initial fraction of hematite present in the platelet coating. This indicates that our reduction protocol provides an effective means to create metal-coated platelets starting from sol–gel precursors. Because the non-aqueous sol–gel route utilized to coat the platelets is applicable to a wide range of chemistries, it can also be applied to tailor the composition of the coating. Indeed, platelets coated with copper, copper oxide, tungsten oxide, molybdenum, molybdenum oxide or nickel could be successfully prepared by simply changing the metal precursor in the initial sol–gel mixture (Figure S1 ). In addition to changes in the crystalline phases, the reduction process was also found to affect the morphology of the platelet coating. We monitored these morphological transformations by examining scanning electron microscopy (SEM) images of platelets treated at different temperatures in a small-batch process (Fig. 3 b). The images reveal that the thermal treatment under reducing atmosphere promotes sintering between the initial particles, thus increasing the characteristic size of the grains in the platelet coating. As expected, such coarsening effect is more pronounced at higher temperatures. Importantly, grains were found to wet the surface of the alumina platelets if the composition of the coating contained hercynite or any of the iron oxides. By contrast, the formation of a fully metallic coating at 900 °C led to de-wetting and the formation of large iron particles on the platelet surface. X-Ray diffraction and SEM imaging of platelets containing higher fractions of hematite showed similar trends in terms of crystalline phases and coating morphology. However, we noticed that alumina platelets coated with the highest hematite fraction of 43 vol% completely lose their morphology and high aspect ratio when reduced at a temperature of 700 °C or higher (Figure S2 ). Although further investigation is needed to elucidate this effect, the observed morphological change is probably related to the full transformation of the alumina into the hercynite phase. Guided by these experimental results, we selected a reduction temperature of 1000 °C for the preparation of iron-coated platelets with initial hematite fractions of 10 and 23 vol% (Fig. 2 b,c). To preserve the platelet morphology while maximizing the fraction of metallic phase in the coating, platelets containing an initial hematite content of 43 vol% (Fig. 2 d) were reduced at the temperature of 600 °C. The wustite phase observed in platelets reduced at this lower temperature was found to eventually transform into hercynite or iron at the high sintering temperature of 1450 °C used later for the consolidation of the nacre-like composites. The knowledge gained in this small-batch study was eventually used to establish an up-scaled process for the preparation of the large quantities of coated platelets necessary for the metal-ceramic composites (see “ Methods ” section). Our ability to reduce the initial oxide layer of the coated platelets into metallic iron opens the possibility to fabricate nacre-like metal-ceramic composites by assembling the platelets into brick-and-mortar architectures. Because of the magnetic nature of the iron coating, brick-and-mortar architectures were readily obtained using the magnetically-assisted slip casting (MASC) approach 6 . In this process, platelets dispersed in a suspension are aligned with the help of an external magnetic field and collected at the walls of a porous mould to generate structures with highly oriented architectures after casting and drying (Fig. 1 ). Sintering of such structures under pressure at high temperatures allows for densification of the aligned architecture while maintaining a fine-grained microstructure. Following this procedure, we created metal-ceramic composites with a nacre-like architecture consisting of alumina “bricks” and metallic iron as the “mortar” phase (Fig. 4 h). X-Ray diffraction of specimens after sintering confirmed the presence of α-iron in fractions ranging from 4.0 to 12.4 vol% as the main constituent of the mortar phase. In addition to iron, compositions made from platelets originally coated with 23 and 43 vol% of hematite were also found to contain, respectively, 0.3 and 8.2 vol% of hercynite after the sintering process. Figure 4 Mechanical properties of metal-ceramic composites with nacre-like architecture. ( a ) Stress–strain curves obtained from three-point bending tests on un-notched specimens containing different iron fractions. ( b ) Effect of the metal fraction on the flexural strength and the elastic modulus of nacre-like composites. ( c ) Ashby plot depicting the specific elastic modulus and specific strength of our metal-ceramic composites in comparison to literature data for conventional materials. ( d ) Force–displacement data obtained for single-edge notched beam samples with varying iron content. ( f ) Effect of the metal fraction on the critical stress intensity factor for crack initiation ( K IC ) and the maximum intensity factor after crack propagation ( K Jmax ). ( e ) Stress intensity factor ( K J ) of the composite as a function of the propagated crack length (R-curve). ( g , h ) Scanning electron microscopy (SEM) images of the fracture surface of a specimen containing 12.4 vol% Fe highlight ( g ) the plastic deformation of the metallic phase and ( h ) the homogeneous distribution of metallic iron within the structure. Magnification: 88 k ×. In image ( h ) the EDX intensity spectra for Fe and O elements across a selected line is displayed on top of a false coloured micrograph to illustrate the distribution of metallic and ceramic phases in the structure. Magnification: 10 k ×. ( i ) Light microscopy image of the fracture generated in a notched specimen after testing. Magnification: 100 ×. Nacre-like composites containing metallic iron as the main mortar phase exhibit high fracture toughness combined with high mechanical strength and elastic modulus. This unusual set of antagonistic properties were assessed by testing notched and un-notched prismatic specimens under three-point bending following standard mechanical characterization protocols 36 – 38 . Results obtained from the characterization of un-notched samples show that the composites display strong linear elastic response with elastic modulus ranging from 341 to 355 GPa and fracture strength lying between 300 and 400 MPa (Fig. 4 a,b). Except for the relatively low strength of the composition with 4.0 vol% Fe, the fraction of the metallic phase was found to have no significant effect on the elastic modulus and fracture strength. Given that the elastic modulus of iron (200 GPa) and alumina (370 GPa) are relatively close, this experimental result is in line with predictions based on a simple rule of mixtures. Materials with such high elastic modulus and mechanical strength typically fracture in a brittle fashion with no resistance against crack propagation. Instead, the nacre-like architecture of our composites provides toughening mechanisms that progressively increase the resistance of the material against crack propagation (Fig. 4 g,i). Force–displacement data obtained for samples with different metal contents indicate that the notched beams exhibit non-catastrophic failure after reaching their maximum load-bearing capacity (Fig. 4 d). We characterized the increasing crack-growth resistance of the nacre-like composites by measuring the fracture toughness against crack initiation ( K IC ) and propagation ( K J ) as a function of the crack length in single-edge notched beam specimens. We find that the fracture toughness of the nacre-like composite increases by 2 to threefold when a crack propagates for a few hundred micrometers into the structure perpendicular to the orientation of platelets (Fig. 4 e). This resistance against crack growth increases the fracture toughness of the composite to up to 15 MPa m 1/2 , which is about 5 times higher than the typical value of pure alumina 39 . The fact that the strength of the composite remains constant in spite of the nearly twofold toughness increase when the iron content changes from 8.9 to 12.4 vol% points out that larger defects might be introduced during processing of the samples with the highest iron concentration. Such defects may result from platelet packing faults in the presence of large iron pockets. While compositions with 4.0 and 8.9 vol% Fe show a comparable toughening effect, composites with 12.4 vol% display a significantly higher K IC and K Jmax values, as well as a steeper increase in fracture toughness as a function of crack length (Fig. 4 f). This suggests that the presence of a minimum fraction of iron is crucial to fully benefit from the toughening arising from the metallic phase. The main contribution of the metallic phase to the toughness of the composite is expected to be plastic deformation ahead of the crack tip during the fracture process. SEM images of fractured surfaces provide evidence of plastically-deformed metal between the alumina platelets, indicating that this appears to be indeed an important toughening mechanism in composites containing the highest Fe content (Fig. 4 g). In addition to the plastic deformation of the metallic phase, crack deflection by the stiff platelets is another important mechanism that contributes to the enhanced toughness of the nacre-like composites (Fig. 4 i). Microstructural analyses of the fracture surface of nacre-like composites indicate that the alumina platelets are fully wetted by the surrounding metallic phase (Fig. 4 h). This contrasts with the poor wetting of the iron particles on the surface of the alumina after complete reduction of the coated platelets at temperatures above 900 °C (Fig. 3 b). The formation of hercynite during the sintering of specimens containing 8.9 and 12.4 vol% Fe is probably crucial to enable wetting of the iron phase on the alumina surface and to thus increase the fracture toughness of the nacre-like composites. The absence of such spinel as an interphase between iron and alumina in samples containing 4.0 vol% Fe might in fact be the reason for the lower fracture strength of this composition (Fig. 4 b). These observations are in line with a recent study on Ni-containing nacre-like composites 7 . In this case, the formation of a nickel oxide interphase was also found to be essential to enhance the fracture toughness of the metal-ceramic composite. Compared to previous metal–ceramic composites with nacre-like architecture, our iron-based composites show 30 to 60% higher bending strength (400 MPa) combined with a fracture toughness level nearly as high as that of the toughest Ni-based systems (16 MPa m 1/2 ) 7 , 27 . This set of mechanical properties makes our composites attractive materials for load-bearing structural applications, particularly those where minimum weight is a key requirement (Fig. 4 c). Indeed, the metal-ceramic composites developed in this work show 3.2-times higher weight-normalized elastic modulus compared to steels, while keeping the high specific strength of this widely used metal (Fig. 4 c). Besides these attractive mechanical properties, the presence of a metallic phase between platelets provides magnetic and electrical functionalities thus far not fully explored in other nacre-like bulk composites. These additional functionalities stem from the electrically-conductive and ferromagnetic properties of the iron phase, which makes the nacre-like composites magnetically-responsive and amenable to inductive heating effects not observed in conventional ceramics (Figs. 5 a and 6 a). We illustrate such functionalities by measuring the magnetic properties of specimens with varying iron fractions and characterizing their temperature evolution when subjected to induction heating. Figure 5 Magnetic properties of the metal-ceramic composites with nacre-like architecture. ( a ) Picture displaying the magnetic attraction between a nacre-like metal-ceramic specimen and a stack of commercial neodymium magnets. ( b ) Magnetization as a function of the applied magnetic field strength of metal-ceramic composites made with increasing volume fractions of iron and hercynite. ( c ) Closer view of the magnetization curves in ( b ) around the origin. ( d ) Saturation magnetization as a function of the volume fraction of metallic iron in the composite. ( e , f ) Volume magnetic susceptibility χ , the remanent magnetization M r and the coercive field H c as function of the content of metallic iron in the specimens. Figure 6 Inductive heating behavior of the metal-ceramic composites upon exposure to oscillating magnetic fields. ( a ) Pictures showing the coil used to generate the magnetic fields and the position of the specimen in the beginning of an experiment (left) and when it reached the peak temperature (right). ( b ) Temperature recorded over time during the inductive heating experiment performed with specimens of different iron fractions. In this experiment, the magnetic field was turned on at time zero and was switched off at the time points indicated with a star symbol inside the plot. ( c ) Average heating rate of composites with varying volume fraction of iron ( b ). The magnetization of ground samples measured as a function of the applied field show that the nacre-like specimens exhibit strong saturation magnetization and relatively low hysteresis losses when subjected to field strengths up to 800 kA/m (Fig. 5 b,c). We note that the magnetization obtained at the maximum applied field is taken here as an approximation for the actual saturation magnetization. Because it depends only on the chemical composition of the material 40 , 41 , the saturation magnetization measured for our composites was compared with the theoretical values expected based on a simple rule of mixture. Such theoretical prediction takes into account the mass fraction and the intrinsic saturation magnetization of the iron and the hercynite present in the composites 42 , 43 . Comparison of our measurements with the theoretical predictions (Fig. 5 d) shows that the saturation magnetizations of our composites lie within the same order of magnitude of the values predicted by the simplified analysis. The stronger discrepancy observed for composites with higher fractions of iron and hercynite might result from the fact that possible amorphous phases contributing to the total magnetic moment are not detected by X-Ray diffraction and thus not considered in the analysis. In contrast to the saturation magnetization, the shape of the hysteresis loop is strongly dependent not only on the material composition but also on the microstructure, orientation of the ferromagnetic phases, sample shape and domain structure 40 , 41 . Overall, the relatively small area of the hysteresis loop reveals that the composites show low losses when exposed to oscillatory magnetic fields, which is a typical feature of soft magnetic materials. The shape of the hysteresis loop is determined by the magnetic susceptibility χ , the remanent magnetization M r and the coercivity H c . To evaluate the effect of the fraction of ferromagnetic phase on these magnetic properties, we plot the values of χ , M r and H c as a function of the volume fraction of α-iron in the composites. Although it neglects the contribution of the hercynite phase, this simplified analysis should provide insights into possible microstructural features controlling the magnetic properties of the composite. Our results show that the magnetic susceptibility ( χ ) and the remanent magnetization ( B r ) increase monotonically with the volume fraction of α-iron, whereas the coercive field ( H c ) decreases with the fraction of this ferromagnetic phase (Fig. 5 e,f). The low volume fraction of iron present in the composites leads to volume susceptibility levels that are two orders of magnitude lower than typical values for pure annealed iron 44 . However, the magnetic susceptibility of our composites is high enough to make them responsive to the low magnetic fields applied by household magnets (Fig. 5 a). We expect that the magnetic susceptibility of these composites may be further improved in future work through optimization of the microstructure and orientation of the ferromagnetic phase. Another interesting property arising from the metallic phase is electrical conductivity. The imparted electrical conductivity renders the composite sensitive to oscillating magnetic fields, which can be intentionally used to inductively heat the material in a very short time. Because it combines the low weight and enhanced toughness of nacre-like ceramics with the magnetically-induced heating capability of metals, this functionality cannot be easily achieved using conventional materials. We demonstrate the inductive heating functionality of our composites by recording the temperature of bulk specimens when subjected to an oscillating magnetic field of amplitude 15 mT and frequency 285 kHz (Fig. 6 a). A temperature increase of 255 °C was obtained in as little as 5.5 s using composites containing 12.4 vol% Fe. Measurements on samples with different iron fractions show that the heating rate increases exponentially with the metal fraction (Fig. 6 b,c). This strong dependence arises from the fact that a higher metal content increases the amount of heat generated per unit time, the thermal conductivity of the composite and the extent of percolation of the metallic phase. This observation suggests that the investigated iron fractions fall within a range where the metallic phase builds a percolating network throughout the sample volume. Electrical characterization of composites with 4.0, 8.9 and 12.4 vol% Fe shows that the electrical resistance decreases from > 300 MΩ to 119 k \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$\\Omega $$\\end{document} Ω and 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}$$\\Omega $$\\end{document} Ω as the iron content is increased (Figure S3 ). In another possible application scenario, the electrically-conductive nature of composites with 12.4 vol% Fe can potentially also be exploited for monitoring crack growth in load-bearing structures, as demonstrated in previous studies 23 ."
} | 8,345 |
23623045 | null | s2 | 768 | {
"abstract": "Next-generation biofuels must be compatible with current transportation infrastructure and be derived from environmentally sustainable resources that do not compete with food crops. Many bacterial species have unique properties advantageous to the production of such next-generation fuels. However, no single species possesses all characteristics necessary to make high quantities of fuels from plant waste or CO2. Species containing a subset of the desired characteristics are used as starting points for engineering organisms with all desired attributes. Metabolic engineering of model organisms has yielded high titer production of advanced fuels, including alcohols, isoprenoids, and fatty acid derivatives. Technical developments now allow engineering of native fuel producers, as well as lignocellulolytic and autotrophic bacteria, for the production of biofuels. Continued research on multiple fronts is required to engineer organisms for truly sustainable and economical biofuel production."
} | 249 |
29873717 | PMC5989612 | pmc | 769 | {
"abstract": "ABSTRACT Microorganisms are the drivers of biogeochemical methane and nitrogen cycles. Essential roles of chemolithoautotrophic microorganisms in these cycles were predicted long before their identification. Dedicated enrichment procedures, metagenomics surveys and single-cell technologies have enabled the identification of several new groups of most-wanted spookmicrobes, including novel methoxydotrophic methanogens that produce methane from methylated coal compounds and acetoclastic ‘ Candidatus Methanothrix paradoxum’, which is active in oxic soils. The resultant energy-rich methane can be oxidized via a suite of electron acceptors. Recently, ‘ Candidatus Methanoperedens nitroreducens’ ANME-2d archaea and ‘ Candidatus Methylomirabilis oxyfera’ bacteria were enriched on nitrate and nitrite under anoxic conditions with methane as an electron donor. Although ‘ Candidatus Methanoperedens nitroreducens’ and other ANME archaea can use iron citrate as an electron acceptor in batch experiments, the quest for anaerobic methane oxidizers that grow via iron reduction continues. In recent years, the nitrogen cycle has been expanded by the discovery of various ammonium-oxidizing prokaryotes, including ammonium-oxidizing archaea, versatile anaerobic ammonium-oxidizing (anammox) bacteria and complete ammonium-oxidizing (comammox) Nitrospira bacteria. Several biogeochemical studies have indicated that ammonium conversion occurs under iron-reducing conditions, but thus far no microorganism has been identified. Ultimately, iron-reducing and sulfate-dependent ammonium-oxidizing microorganisms await discovery.",
"conclusion": "CONCLUSIONS Taken together, these studies and examples emphasize the fascinating diversity of the most-wanted spookmicrobes. Future detailed field studies using state-of-the-art biogeochemical and microbiology methods in selected environments with counter gradients of iron oxides and ammonium and/or methane are needed to identify suitable niches and samples for the discovery of new methane- or ammonium-dependent iron reducers. We expect that such samples will yield many more exciting discoveries of chemolithoautotrophic spookmicrobes when the microbial ecology and interactions are investigated under controlled substrate-limited conditions in bioreactor systems. During the page proof stage two studies (Table 1 ) appeared online. Huang and Jaffé reported the isolation of an Acidimicrobiaceae strain that can convert ammonium to nitrite at pH 4 with ferrihydrite as electron acceptor (Huang and Jaffé 2018 ). Cai et al . described a 1100 day enrichment of ' Candidatus Methanoperedens ferrireducens' that use methane to reduce Fe 3+ possibly using several highly expressed multiheme cytochrome c proteins (Cai et al . 2018 ).",
"introduction": "GENERAL INTRODUCTION During Earth's history, a set of metabolic processes that evolved exclusively in anaerobic microorganisms changed the chemical speciation of all major elements (Falkowski, Fenchel and Delong 2008 ; Stolz 2017 ). Our present-day environment is thus the integrated result of microbial experimentation that has allowed life to develop and persist, despite major environmental changes documented in the geological record. The recent expansion of microbial genome sequence data combined with increasingly detailed geochemical analyses has yielded insights on how microorganisms became the biogeochemical engineers of life on Earth. Among the most urgent scientific questions are which key groups of microorganisms drive the relevant reactions, how do these microorganisms interact with each other and their geochemical environment, and how do they impact the Earth system (Anantharaman et al . 2016 ; Thompson et al . 2017 ). In this context, it is important to understand the microbial and geochemical pathways for the conversion of methane (CH 4 ), hydrogen sulfide (H 2 S) and ammonium (NH 4 + ), products of the anaerobic degradation of organic matter by a complex web of microorganisms (Fig. 1 ). Methanogens are responsible for the terminal step in this anaerobic food web and produce an estimated 583 Tg (range: 458–748) of methane per year from natural and agricultural sources (Saunois et al . 2016 ). Methane is a notorious greenhouse gas, and its atmospheric concentration has more than doubled since the start of the Industrial Revolution (Allen 2016 ). Concentrations of ammonium, a key player in the deterioration of water quality, have increased dramatically worldwide over the past century, and globally the nitrogen cycle in general has long exceeded safe operational boundaries (Rockström et al . 2009 ). H 2 S is extremely toxic to all higher life forms, and its release can greatly alter biogeochemical cycling in aquatic environments (Diaz and Rosenberg 2008 ). Microorganisms, particularly chemolithoautotrophs, play a critical role in modulating the release of methane, hydrogen sulfide and ammonium by driving a range of redox reactions that ultimately transform these detrimental reductants to comparatively less harmful compounds, such as carbon dioxide (CO 2 ), sulfate (SO 4 2− ) and dinitrogen gas (N 2 ) (Fig. 1 ). Figure 1. Overview of microbial diversity and possible reactions in the global methane and nitrogen cycles. AOP: ammonium-oxidizing prokaryotes including Thaumarchaeota and AOB, CMX: comammox bacteria, NOB: nitrite-oxidizing bacteria, MOB: methane-oxidizing bacteria, ANME: anaerobic methane-oxidizing archaea. OMD: organic matter degradation. The ‘?’ indicates as-yet undiscovered processes of iron- and manganese-dependent ammonium and methane oxidation (see Table 1 ). For more than a century, methane and ammonium were thought to be oxidized by microorganisms only in the presence of oxygen. Unequivocal proof of the anaerobic oxidation of these compounds in the presence of sulfate, nitrite (NO 2 − ) and nitrate (NO 3 − ) was obtained only in recent decades (Boetius et al . 2000 ; Raghoebarsing et al . 2006 ; Ettwig et al . 2010 ; Haroon et al . 2013 ). Attempts to enrich these so-called (impossible) anaerobic microorganisms growing on methane or ammonium were initially not successful, mainly due to their slow growth and highly specific substrate requirements (Table 1 ). Selecting samples from ecosystems with counter-gradients of ammonium/nitrate or methane/nitrate (Zhu et al . 2012 ; Vaksmaa et al . 2017a ) and increasing the number of target cells can reduce enrichment times. Bioreactors with effective biomass retention systems (sequencing batch reactor (SBR) or membrane systems) and optimized growth media (e.g. appropriate trace elements like lanthanides, low substrate availability or low nitrite concentrations) can also contribute to successful enrichment (Strous et al . 1997 ; van Kessel et al . 2015 ). Once sufficient cells are available, metagenomics can be combined with single-cell approaches to quickly reveal the genetic blueprint. Together with stable isotope experiments, this blueprint can be used to design crucial experiments to verify the metabolic potential of these ‘impossible’ microorganisms. Table 1. Overview of chemolitho(auto)trophic reactions in the conversion of methane, ammonium and nitrite by the microorganisms highlighted in this review. Electron acceptor ΔE 0 ΄ ΔG 0 ΄ Reaction equation Micro-organism(s) Origin Growth rate Per cell rate Ks [S] Ks [EA] Reference Methane production from various substrates CH 3 OH +360 –103 4 CH 3 OH → CO 2 + 3 CH 4 + 2 H 2 O \n Methanosarcina semesiae \n Brackish sediment <0.2 – <5 – Lyimo, Pol and Op den Camp ( 2000 ); Thauer, Jungermann and Decker ( 1977 ) and Welte ( 2018 ) CH 3 -R +193 –56 (CH 3 ) 2 SH + H 2 O → 0.5 CO 2 + 1.5 CH 4 + H 2 S \n Methanomethylovorans hollandica \n Freshwater sediment <1 – <30 – Lomans et al . ( 1999 ) CH 3 COOH +46 –36 CH 3 COOH → CO 2 + CH 4 \n Methanothrix soehngenii \n WWTP 7–14 – 500 – Huser et al . ( 1982 ) \n 'Candidatus Methanothrix paradoxum’ Wetland soil – – – – Angle et al . ( 2017 ) CH 3 O-R +366 –106 4 CH 3 O-R + 2 H 2 O → 4 R-OH + CO 2 + 3 CH 4 \n Methermicoccus shengliensis \n Oilfield water <5 20 – – Cheng et al . ( 2007 ) CH 3 OH +172 –113 CH 3 OH + H 2 → CH 4 + H 2 O \n Methanomassiliicoccus luminyiensis \n Human feces 2 – – – Dridi et al . ( 2012 ) ‘ Candidatus Methanonatronarchaeia’ Hypersaline lake – – – – Sorokin et al . ( 2017 ) ‘ Candidatus Methanofastidiosa’ WWTP – – – – Nobu et al . ( 2016 ) Methane (CO 2 /CH 4 at ΔE 0 ΄ = −240 mV) as electron donor O 2 /H 2 O +810 –801 CH 4 + 2 O 2 → CO 2 + 2 H 2 O Methane-oxidizing bacteria (MOB) 0.5–2 158–240 0.06–12.6 6–37 Ren, Amaral and Knowles ( 1997 ); Dunfield and Conrad ( 2000 ) and Steenbergh et al . ( 2010 ) Alphaproteobacteria \n Methylocella palustris \n Peat Dedysh et al . ( 2000 ) \n Methylocella tundra \n Peat Dedysh et al . ( 2004 ) Upland soil cluster alpha Soil Pratscher et al . ( 2018 ) Gammaproteobacteria Upland soil cluster gamma Soil Knief et al . ( 2003 ) Verrucomicrobia \n Methylacidiphilum \n Geothermal area Op den Camp et al . ( 2009 ) NO 3 − /NO 2 − +430 –503 CH 4 + 4 NO 3 − → CO 2 + 4 NO 2 − + 2 H 2 O ‘ Candidatus Methanoperedens nitroreducens’ Freshwater sediment, WWTP >14 0.57 >1000 <50 Haroon et al . ( 2013 ) and Vaksmaa et al . ( 2017a ) NO 2 − /N 2 +320 –928 3 CH 4 + 8 NO 2 − + 8 H + → 3 CO 2 + 4 N 2 + 10 H 2 O ‘ Candidatus Methylomirabilis oxyfera’ Freshwater sediment >14 0.4–0.2 <50 <10 Raghoebarsing et al . ( 2006 ) Fe 3+/ Fe 2+ +360 –454 CH 4 + 8 Fe 3+ + 2 H 2 O → CO 2 + 8 Fe 2+ + 8 H + ‘ Candidatus Methanoperedens nitroreducens’ Freshwater sediment, WWTP – – – – Ettwig et al . ( 2009 ); Ettwig et al . ( 2016 ) and Cai et al . ( 2018 ) ANME-2C Marine sediment – – – – Boetius et al .( 2000 ) SO 4 2− /H 2 S –210 –21 CH 4 + SO 4 2− → HCO 3 − + H 2 S + H 2 O Anaerobic methanotrophic archaea (ANME) Marine sediment >50 0.7 >1000 – Nauhaus et al . ( 2005 ); Knittel et al . ( 2005 ) Ammonium (NO 2 − /NH 4 + at ΔE 0 ΄ = 340 mV) as electron donor O 2 /H 2 O +810 –275 NH 4 + + 1.5 O 2 → NO 2 − + H 2 O + 2 H + Ammonium-oxidizing bacteria (AOB) <1 264–552 0.8–112 1–15 Belser and Schmidt ( 1980 ) and Laanbroek and Gerads ( 1993 ) Ammonium-oxidizing archaea (AOA) Seawater aquarium <5 0.5–32.2 5–44 – Kits et al . ( 2017 ) \n ‘Candidatus Nitrosopumilus maritimus’ Hot spring Könneke et al . ( 2005 ) ‘ Candidatus Nitrosocaldus yellowstonii’ Hot spring de la Torre et al . ( 2008 ) ‘ Candidatus Nitrososphaera gargensis’ Garden soil Hatzenpichler et al . ( 2008 ) \n ‘Candidatus Nitrososphaera viennensis’ Agricultural soil Tourna et al . ( 2011 ) ‘ Candidatus Nitrosotalea devanaterra’ Hot spring Lehtovirta-Morley, Stoecker and Vilcinskas ( 2011 ) ‘ Candidatus Nitrosocaldus islandicus’ Daebeler et al . ( 2018 ) Nitrite (NO 3 − /NO 2 − at ΔE 0 ΄ = 420 mV) as electron donor O 2 /H 2 O +810 –74 NO 2 − + 0.5 O 2 → NO 3 − Nitrite-oxidizing bacteria (NOB) <1 0.6–13.1 9–544 22–166 Féray and Montuelle ( 2002 ) and Nowka, Daims and Spieck ( 2015 ) Alphaproteobacteria \n Nitrobacter winogradskyi \n Soil Winslow et al . ( 1917 ) Betaproteobacteria \n ‘Candidatus Nitroga arctica’ Permafrost Alawi et al . ( 2007 ) Gammaproteobacteria \n Nitrococcus mobilis \n Ocean water Watson and Waterbury ( 1971 ) Nitrospirae \n Nitrospira moscoviensis \n Heating system Ehrich et al . ( 1995 ) \n ‘Candidatus Nitrospira defluvii’ WWTP Lücker et al . ( 2010 ) Nitrospinae \n Nitrospina gracilis \n Ocean water Watson and Waterbury ( 1971 ) Chloroflexi \n Nitrolancetus hollandicus \n Nitrifying reactor Sorokin et al . ( 2012 ) Ammonium (NO 3 − /NH 4 + at ΔE 0 ΄ = 360 mV) as electron donor O 2 /H 2 O +810 –349 NH 4 + + 2 O 2 → NO 3 − + H 2 O + 2 H + Comammox Nitrospira <1 – 0.6 – van Kessel et al . ( 2015 ) and Kits et al . ( 2017 ) \n Nitrospira inopinata \n Hot water pipe Daims et al . ( 2015 ) Ammonium (N 2 /NH 4 + at ΔE 0 ΄ = -280 mV) as electron donor NO 2 − /N 2 +320 –358 NH 4 + + NO 2 − → N 2 + 2 H 2 O Anammox bacteria 4–14 2–20 <5 <5 Lotti et al . ( 2015 ) and Zhang et al . ( 2017 ) Brocadiales Jetten et al . ( 2010 ) \n ’Candidatus Kuenenia stuttgartiensis’ WWTP Schmid et al . ( 2000 ) \n ‘Candidatus Brocadia fulgida’ WWTP Kartal et al . ( 2008 ) \n ‘Candidatus Anammoxoglobus propionicus’ WWTP Kartal et al .( 2007 ) \n ‘Candidatus Scalindua profunda’ Marine sediment van de Vossenberg et al . ( 2013 ) \n ‘Candidatus Jettenia caeni’ WWTP Ali et al . ( 2015 ) Fe 3+ /Fe 2+ +360 –303 NH 4 + + 3Fe(OH) 3 + 5 H + → 3 Fe 2+ + 9 H 2 O + 0.5 N 2 ? – – – – Kuypers et al . ( 2018 ) and Huang and Jaffé ( 2018 ) SO 4 2− /H 2 S –210 –22 8 NH 4 + + 3 SO 4 2− → 4 N 2 + 1.5 H 2 S + 12 H 2 O + 2 H + ? – – – – Zhang et al . ( 2009 ) Redox potentials of the half-reactions are given at 25°C and pH 7. ΔE 0 ΄ is displayed in mV, ΔG 0 ’ is displayed in kJ/mol substrate. Growth rates are displayed in days, per cell rates are given in fmol substrate per cell per day and Ks is given in µM substrate [S] and electron acceptor [EA], ‘–’ indicates values have not been determined yet. For MOB, AOB, NOB, and anammox bacteria the range of the ecophysiological parameters is given. WWTP = wastewater treatment plant. Emerging evidence suggests that there are several important but previously unknown microbial pathways for the oxidation of methane and ammonium involving oxides of iron and manganese (Beal, House and Orphan 2009 ; Ettwig et al . 2016 ). Despite rapid and continuing technological improvements, a large part of microbial diversity has yet to be discovered. Many, particularly chemolithoautotrophic processes, have been hypothesized or observed based on nutrient profiles and metagenomic inventories. Species-level detail is often lacking, leaving open the question of whether specific microorganisms are responsible for the biochemical conversions observed in the field. The discovery of multiple ‘impossible’ anaerobic microorganisms has reinforced the idea that a microorganism or combination of microorganisms should exist for each thermodynamically feasible process (Table 1 ). In this review, we provide an overview of the discoveries of several most-wanted chemolithoautotrophic spookmicrobes that may play significant roles in global methane, sulfur and nitrogen cycles and highlight a few processes that still await detection. Methane cycle Methane is a potent greenhouse gas with a warming potential 34 times stronger than that of carbon dioxide over a time period of 100 years (Henry et al . 1970 ; Lacis et al . 1981 ; Myhre et al . 2013 ). Methane is the most reduced one-carbon compound and plays a key role in the global carbon cycle and the greenhouse effect as was stressed by the first IPCC report in 1990 (Watson et al . 1990 ). Many processes in a wide variety of ecosystems control the global methane budget (Heilig 1994 ; Kirschke et al . 2013 ; Dean et al . 2018 ). The majority of methane released into the atmosphere (70%–80%) is of biogenic origin (Conrad 1996 , 2009 ), and most if not all biogenic methane is produced by methanogenic archaea within the phylum Euryarchaeota. Proposed alternative pathways include methane production by iron-only nitrogenases (Zheng et al . 2018 ), methane release from methylphosphonates in marine ecosystems (Daughton, Cook and Alexander 1979 ), and in situ formation of methane in terrestrial plants (Keppler et al . 2006 ). Methanogenic archaea are obligate anaerobes found in anoxic soils, sediments and water bodies. A fraction of the methane produced directly escapes into the atmosphere via ebullition (Schütz, Seiler and Conrad 1989 ; Aben et al . 2017 ). Before dissolved and trapped methane reaches the atmosphere, it can be oxidized by a range of anaerobic and aerobic methanotrophs using a suite of electron acceptors. These methanotrophs include anaerobic methanotrophic (ANME) archaea, and anaerobic and aerobic methanotrophic bacteria. For an extensive overview of methanogenesis and methanotrophy, see Kallistova et al . ( 2017 ). Methanogens Microbial methanogenesis was first described by Omelianski in 1890 and later experimentally confirmed by Söhngen ( 1906 ), who was the first to describe the ‘fat rod’ Methanothrix soehngenii , which produces methane from acetate (Huser, Wuhrmann and Zehnder 1982 ). Methanogens are dependent on fermentative and syntrophic processes that convert organic compounds to methanogenic substrates (Kotsyurbenko, Nozhevnikova and Zavarzin 1993 ; Schink 1997 ). Methanogens that use H 2 /CO 2 and methylated compounds as substrates were subsequently isolated and characterized (for an overview, see Plugge and Stams 2010 ). Acetate usage appears to be limited to the genera Methanosarcina and Methanothrix (Jetten, Stams and Zehnder 1992 ). There are seven methanogenic orders: Methanosarcinales, Methanomicrobiales, Methanobacteriales, Methanococcales, Methanopyrales, Methanocellales, and the recently discovered Methanomassiliicoccales (Garrity, Bell and Lilburn 2004 ; Thauer et al . 2008 ; Dridi et al . 2012 ; Iino et al . 2013 ; Lyu and Lu 2015 ). Methanomassiliicoccales species ( Methanomassiliicoccus luminyiensis ) were first discovered in human feces and use hydrogen as an electron donor to reduce methanol to methane (Dridi et al . 2012 ). In anoxic sediments, the concerted action of acetogens and methanogens can result in the breakdown of methoxylated aromatic compounds like trimethoxybenzoate (Finster, King and Bak 1990 ). The acetogens cleave off the methoxy-groups and produce dimethylsulfide and methanethiol, which can subsequently be used by methylotrophic methanogens ( Methanosarcina semesiae , Methanomethylovorans hollandica ) employing several unique methyltransferases (Finster, Tanimoto and Bak 1992 ; Lomans et al . 1999 ; Lyimo et al . 2000 ). The list of methanogenic substrates was recently expanded to include the direct use of methoxylated aromatic compounds by methoxydotrophic Methermicoccus shengliensis (Methanosarcinales) found in coal beds (Cheng et al . 2007 ; Mayumi et al . 2016 ). Two novel candidate classes, ‘ Candidatus Methanonatronarchaeia’ and ‘ Candidatus Methanofastidiosa’, were also recently discovered (Nobu et al . 2016 ; Sorokin et al . 2017 ). ‘ Candidatus Methanonatronarchaeia’, which are most closely related to Halobacterium , were detected in a metagenomic dataset of hypersaline lakes (Sorokin et al . 2017 ). ‘ Candidatus Methanofastidiosum methylthiophilus’ has the metabolic potential for methanogenesis through methylated thiol reduction using a methylated-thiol:coenzyme M methyltransferase (Nobu et al . 2016 ). These findings indicate that methanogenic archaea might include more extremophilic and metabolically versatile members than those currently known. The recent observation of the aceticlastic ‘ Candidatus Methanothrix paradoxum’ in oxygenated soils (Angle et al . 2017 ) and indications of methanogenesis under oxic conditions (Wagner 2017 ) are striking since methanogens are considered obligate anaerobes. The occurrence of methane production in oxic environments might dramatically alter our view of methanogenic ecosystems. Whether methanogenesis occurs outside Euryarchaeota remains a matter of debate. The discovery of Bathyarchaeota and Verstraetearchaeota genome bins including methyl-coenzyme M reductase (MCR) genes indicates that methanogenesis might be more widespread in the archaeal domain than previously thought (Evans et al . 2015 ; Vanwonterghem et al . 2016 ). Other studies consider Bathyarchaeota anaerobic heterotrophs that assimilate sedimentary organic carbon compounds (Lazar et al . 2016 ; Xiang et al . 2017 ). Verstraetearchaeota also appear to utilize sugars as carbon compounds (Vanwonterghem et al . 2016 ). For an overview and discussion of potential methanogens outside the Euryarchaeota phylum see Welte ( 2018 ). Methane oxidation Before methane produced by methanogens reaches the atmosphere, first anaerobic methanotrophs oxidize methane using a suite of electron acceptors, and the methane that passes this anoxic filter can ultimately be converted by aerobic methane-oxidizing bacteria. Sulfate-dependent anaerobic methane oxidation The anaerobic oxidation of methane (AOM) was long considered impossible due to the high activation energy needed to break the C-H bonds (439 kJ mol −1 ) (reviewed in Thauer and Shima 2008 ). The discovery of counter-gradients of sulfate and methane changed this view and indicated habitats with active AOM (Reeburgh and Heggie 1977 ). The coupling of AOM to sulfate reduction in marine sediments appeared to be mediated by a microbial consortium (Boetius et al . 2000 ). Sulfate-dependent anaerobic oxidation of methane (S-AOM) is particularly intriguing since the reaction has a relatively low Gibbs free energy change of approximately -20 kJ mol −1 (Table 1 ) in most habitats (for a discussion of kinetics and thermodynamics, see Thauer ( 2011 )). In marine ecosystems, S-AOM is carried out by a consortium of ANME archaea in cooperation with sulfate-reducing bacteria or possibly by ANME alone (Knittel and Boetius 2009 ; Milucka et al . 2012 ; Scheller et al . 2016 ). An inverted and modified methanogenesis pathway has been proposed for the catalysis of AOM by ANME (McGlynn et al . 2015 ; Timmers et al . 2017 ). ANMEs are divided into three distinct groups: ANME-1 (Methanosarcinales-related and Methanomicrobiales), ANME-2 (Methanosarcinales) and ANME-3 (Methanococcoides-related) (Knittel et al . 2005 ; Nauhaus et al . 2005 ; Stadnitskaia et al . 2005 ). The 16S rRNA gene phylogeny indicates that ANME groups are not monophyletic with each other, and the phylogenetic distance between subgroups is large, with nucleotide sequence similarities of 75%–92% (Knittel and Boetius 2009 ). Nitrite- & nitrate-dependent methane oxidation After the discovery of S-AOM in marine sediments, the hunt for nitrate- and nitrite-dependent methane oxidation (N-AOM) intensified. Based on redox calculations, both nitrate and nitrite are suitable electron acceptors for methane oxidation and, compared to sulfate, have much higher energy yields per mole of methane (Table 1 ). In 2006, Raghoebarsing et al . ( 2006 ) reported the first enrichment culture coupling AOM to denitrification. The enrichment culture contained archaea (10%–20% of the community) distantly related to ANME-2, and an NC10 phylum bacterium named ‘ Candidatus Methylomirabilis oxyfera’ (70%–80% of the community). The proposed intra-aerobic pathway for coupling of AOM to nitrite reduction by ‘ Candidatus Methylomirabilis oxyfera’ produces oxygen and dinitrogen gas from two molecules of nitric oxide (NO) (Ettwig et al . 2010 , 2012 ). A major implication of this proposed pathway is that aerobic pathways might have been present before oxygenic photosynthesis arose. Despite this proposed intra-aerobic pathway of ‘ Candidatus Methylomirabilis oxyfera’, oxygen exposure as low as 2% has inhibitory effects on methane and nitrite conversion rates (Luesken et al . 2012 ). A recent survey based on primer-based detection of NO dismutase showed that these genes do occur in many anoxic aquifers (Bhattacharjee et al . 2016 ; Zhu et al . 2017 ). Surveys of both 16S rRNA and pmoA genes (which encode the beta subunit of particulate methane monooxygenase) revealed a wide environmental distribution of N-AOM from wetlands to marine sediments and mud volcanos (Welte et al . 2016 ). The role of the ANME-2 archaea in the first enrichment culture was resolved much later. In a bioreactor fed with nitrate, methane and ammonium, a stable co-culture of anaerobic ammonium-oxidizing (anammox) bacteria ( 'Candidatus Kuenenia stuttgartiensis') and ANME-2d archaea was established (Haroon et al . 2013 ). These archaea were identified as ‘ Candidatus Methanoperedens nitroreducens’ (70%–80% of the community), which are capable of coupling nitrate reduction to methane oxidation (Haroon et al . 2013 ). ANME-2d archaea have subsequently been co-enriched a number of times with NC10 phylum and anammox bacteria, which probably scavenge the nitrite and convert it to dinitrogen gas. Analyses of several genomes of ‘ Candidatus Methanoperedens nitroreducens’ have revealed that all genes of the (reverse) methanogenic pathway are present (Haroon et al . 2013 ; Arshad et al . 2015 ; Berger et al . 2017 ; Narrowe et al . 2017 ; Vaksmaa et al . 2017a ). The best-characterized gene for methanogenesis and AOM is mcrA , which encodes for the alpha subunit of Methyl-coenzyme M reductase. An environmental primer-based study based on 16S rRNA and mcrA genes showed that ‘ Candidatus Methanoperedens nitroreducens’ is abundantly present in paddy fields (9% relative abundance of the archaeal community), river sediments and even marine sediments (Vaksmaa et al . 2016 , 2017b ). Terrestrial agriculture-affected ecosystems that receive high concentrations of nitrogen compounds are also facilitating environments for nitrite- and nitrate-dependent methanotrophy. However, little is known about the relevance of N-AOM in terrestrial ecosystems, particularly those with prolonged anoxic conditions, such as natural or restored peatlands. For an extensive overview of N-AOM, see Welte et al . ( 2016 ). Iron- and manganese-dependent methane oxidation In addition to nitrate and nitrite, oxidized iron (Fe 3+ ) and oxidized manganese (Mn 4+ ) should be suitable electron acceptors for AOM based on Gibbs free energy (Table 1 ). Iron is the most abundant metal in the Earth's crust and can serve as both an electron donor and acceptor in microbial metabolism. Iron forms stable minerals in both the divalent and trivalent states depending on geochemical conditions. Fe 3+ is most stable under oxic conditions (Raiswell and Canfield 2012 ). The reduction-oxidation cycle is coupled to other elements, including carbon, nitrogen, oxygen and sulfur. Conversion in the iron cycle can be abiotic or mediated by microorganisms (Weber et al . 2006 ; Melton et al . 2014 ). Iron bioavailability is generally low due to the poor solubility of iron minerals at neutral pH, but microorganisms have developed strategies to mediate electron exchange with insoluble iron forms (Weber, Achenbach and Coates 2006 ). Although a wide variety of organisms are known to reduce iron, the microorganisms responsible for the reduction of metal-oxides coupled to AOM (here abbreviated as Fe-AOM) have remained elusive. Geochemical profiling and stable isotope tracer studies have demonstrated the occurrence of Fe-AOM in lake sediments (Sivan et al . 2011 ; Norði, Thamdrup and Schubert 2013 ; Torres et al . 2014 ), marine sediments (Beal, House and Orphan 2009 ; Wankel et al . 2012 ; Riedinger et al . 2014 ; Egger et al . 2015 ), paddy field sediments (Miura et al . 1992 ; Murase and Kimura 1994 ), lake water (Crowe et al . 2011 ), a terrestrial mud volcano (Chang et al . 2012 ), and in a contaminated aquifer (Amos et al . 2012 ). However, the responsible microorganisms were not identified in these studies. ANME archaea have been implicated in Fe-AOM in marine and volcanic systems (Beal, House and Orphan 2009 ; Chang et al . 2012 ). A recent study demonstrated that ‘ Candidatus Methanoperedens nitroreducens’ can use various electron acceptors, including iron citrate, and thus may be capable of Fe-AOM (Ettwig et al . 2016 ). Fe-AOM by ANME-2C with iron citrate has been shown in mesocosm experiments using deep-sea methane seep sediment (Scheller et al . 2016 ). Wegener et al . ( 2015 ) observed that ANME archaea, under thermophilic AOM conditions, overexpress genes for extracellular cytochrome production and form nanowire-like cell-to-cell connections, suggesting an important role of direct interspecies electron transfer. However, microbial growth on Fe-AOM has yet not been demonstrated. Identifying the responsible microorganism(s) therefore remains a primary interest. Aerobic methane oxidation Methane that is not oxidized by anaerobic methanotrophs can reach the oxic layer of sediment or soil and undergo conversion by aerobic methanotrophs. Aerobic microbial oxidation of methane was first described in 1906 (Söhngen 1906 ). Based on the isolation and description of numerous aerobic methane-oxidizing bacteria (MOB), it was long assumed that microbial methane oxidation was only possible under oxic conditions (Whittenbury, Phillips and Wilkinson 1970 ). MOB belong to Alphaproteobacteria (type II), Gammaproteobacteria (type I) and the phylum Verrucomicrobia (Trotsenko and Murrell 2008 ; Op den Camp et al . 2009 ; Semrau, DiSpirito and Yoon 2010 ). Aerobic methanotrophs are found in virtually all ecosystems, from acidic permafrost-affected peatlands ( Methylocella palustris (Dedysh et al . 2000 ); Methylocella tundrae (Dedysh et al . 2004 )) to volcanic mud pots with temperatures up to 70°C and pH values as low as 1 (Dunfield et al . 2007 ; Pol et al . 2007 ). These volcanic aerobic Methylacidiphilum methanotrophs belong to the phylum Verrucomicrobia (Op den Camp et al . 2009 ; van Teeseling et al . 2014 ). The verrucomicrobial methanotrophs use the Calvin cycle for CO 2 fixation (Khadem et al . 2012 ) and are able to grow as Knallgas bacteria on hydrogen and oxygen (Carere et al . 2017 ; Mohammadi et al . 2017a ). These methanotrophs express hydroxylamine oxidoreductase, nitrite reductase and nitric oxide reductase to counteract the nitrosative stress induced by high ammonium concentrations in mud volcanoes (Mohammadi et al . 2017b ). The growth of verrucomicrobial methanotrophs is dependent on rare earth elements (lanthanides), which are incorporated into the active center of an XoxF-type methanol dehydrogenase (Pol et al . 2014 ). The unique properties of verrucomicrobial MOB are a striking example of the breadth of microbial diversity and physiology that remains to be explored and discovered. Atmospheric methane levels were long considered too low to sustain microbial methanotrophy, but methane oxidation at atmospheric levels has been described in upland soils (Dunfield et al . 1999 ). Culture-independent studies of these soils, which have high-affinity methane oxidation capacity, detected novel methanotrophic bacteria within Alpha- and Gammaproteobacteria named upland soil cluster (USC) α and γ (Knief, Lipski and Dunfield 2003 ; Kolb et al . 2005 ; Ricke et al . 2005 ). Recently, Pratscher et al . ( 2018 ) obtained a 85% complete draft genome of the USCα genus within Beijerinckiaceae using combined metagenomics and targeted cell enrichments with fluorescence in situ hybridization-fluorescence activated cell sorting. In addition, recent studies have indicated that classic MOB can thrive under extremely low oxygen conditions by apparently coupling fermentative metabolism to nitrate reduction (Kits et al . 2015 ; Kits, Klotz and Stein 2015 ; Oswald et al . 2016 ; Gilman et al . 2017 ). Together with the observations of methanogenesis under oxic conditions, these findings may alter our understanding of controls on methane fluxes. Nitrogen cycle Historically, the nitrogen cycle was thought to include only a few processes: (i) the fixation of dinitrogen gas into ammonium by free-living or symbiotic microorganisms (Beijerinck 1888 ); (ii) nitrification, in which ammonium is oxidized via nitrite to nitrate (Winogradsky 1890 ); (iii) denitrification, in which oxidized nitrogen species are reduced to dinitrogen gas by heterotrophic and/or autotrophic bacteria (Gayon and Dupetit 1886 ) and (iv) nitrate/nitrite dissimilation and assimilation, which provides many microorganisms with ammonium (Berks et al . 1995 ). In 1977, Broda calculated that several nitrogen processes could sustain as-yet undiscovered microorganisms (Broda 1977 ). However, it was not until 1995 that one of these processes, anammox (Table 1 ), was observed, and in 1999, the responsible anammox bacteria were identified as novel Planctomycetes (Mulder et al . 1995 ; Strous et al . 1999 ). In 2005, marine Thaumarchaeota (previously named Crenarchaeota) capable of oxidizing ammonium at oceanic concentrations (1 nM to 10 µM) were isolated and characterized (Könneke et al . 2005 ; Lam and Kuypers 2011 ). The complete ammonium-oxidizing (comammox) bacteria predicted by Costa et al . in 2006 were later identified as Nitrospira bacteria (Daims et al . 2015 ; van Kessel et al . 2015 ). Aerobic ammonium oxidation Ammonium oxidation to nitrate via nitrite Since the description and isolation of Nitrosomonas -like aerobic ammonium oxidizers by Winogradsky at the end of the 19th century, this process was attributed to chemolithoautotrophic bacteria (ammonium-oxidizing bacteria, AOB). In marine environments, ammonium oxidation was thought to be limited to the deeper water layers due to light inhibition and ammonium concentrations below the threshold level for AOB activity (Yool et al . 2007 ). This view was challenged by two metagenomics-based studies surveying the microbial diversity of seawater (Venter et al . 2004 ) and soil (Treusch et al . 2005 ), which identified archaeal ammonia monooxygenase ( amoA ) genes phylogenetically affiliated with the phylum Thaumarchaeota. The link between archaea and ammonium oxidation was established by Könneke et al . ( 2005 ) with the isolation of ‘ Candidatus Nitrosopumilus maritimus’, a marine group I.1a representative, from a saltwater aquarium in Seattle, Washington. In recent years, many more ammonium-oxidizing Thaumarchaeota (AOA) representatives have been isolated or enriched, including ‘ Candidatus Nitrososphaera viennensis’ soil group I.1b from soil, ‘ Candidatus Nitrososphaera gargensis’ soil group I.1b from the Garga hot spring, ‘ Candidatus Nitrosocaldus islandicus’ from an Icelandic hot spring, and ‘ Candidatus Nitrosocaldus yellowstonii’ and ‘ Candidatus Nitrosotalea devanaterra’ soil group I.1a-associated enrichments from soil (de la Torre et al . 2008 ; Hatzenpichler et al . 2008 ; Lehtovirta-Morley et al . 2011 , 2014 ; Stieglmeier et al . 2014 ; Daebeler et al . 2018 ). Recently, ‘ Candidatus Nitrosotalea’ species were also enriched from acidic soils with pH values as low as 3.2 (Herbold et al . 2017 ). \n 15 N stable isotope experiments have confirmed that nitrification, most likely by Thaumarchaeota, occurs in the photic zone of marine ecosystems (Clark, Rees and Joint 2008 ). In terrestrial ecosystems, acidiphilic ‘ Candidatus Nitrosotalea devanaterra’ grows optimally between pH 4 and 5 (Zhang et al . 2010 ; Lehtovirta-Morley et al . 2011 ). However, determining the relative contributions of either AOB or AOA in ecosystems is quite challenging due to the large differences in growth rates, K s for ammonia and oxygen, and sensitivity to inhibitors such as 2-phenyl-4,4,5,5,-tetramethylimidazoline-1-oxyl-3-oxide (PTIO) and allylthiourea (ATU) (Geets, Boon and Verstraete 2006 ; Yan et al . 2012 ; Martens-Habbena et al . 2015 ; Beeckman, Motte and Beeckman 2018 ). For a critical view on the importance of bacterial versus archaeal ammonium oxidation, see reviews by Prosser and Nicol ( 2008 ), Pester, Schleper and Wagner ( 2011 ), Hatzenpichler ( 2012 ) and Stahl and de la Torre ( 2012 ). In many ecosystems, the nitrite produced by ammonium-oxidizing prokaryotes (AOP = AOB and AOA) is subsequently oxidized by nitrite-oxidizing bacteria (NOB). Nitrite oxidation is a widespread trait that is found in six phyla: Alpha-, Beta- and Gammaproteobacteria, Nitrospirae, Nitrospinae and Chloroflexi (Nowka, Daims and Spieck 2015 ). Nitrobacter winogradskyi (Winslow et al . 1917 ) was the first nitrite oxidizer to be studied in extensive detail. Nitrobacter species tend to dominate nutrient-rich and oxygen-saturated environments. Nitrospina was discovered together with Nitrococcus in 1971 in marine ecosystems (Watson and Waterbury 1971 ). Nitrospina gracilis appears to be a major marine nitrite-oxidizing species. In general, Nitrospina species are quite well-adapted to low environmental nitrite concentrations (Maixner et al . 2006 ). Nitrococcus mobilis has a much more versatile metabolism, including nitrate reduction and sulfide oxidation (Füssel et al . 2017 ). Nitrospira species (i.e. Nitrospira moscoviensis and ‘ Candidatus Nitrospira defluvii’) generally dominate environments with low substrate availability and hypoxic conditions (Ehrich et al . 1995 ; Schramm et al . 2000 ; Lücker et al . 2010 ). They are more versatile than initially assumed and can use hydrogen and urea as substrates (Koch et al . 2014 , 2015 ). Among betaproteobacterial nitrite oxidizers, the novel Nitrotoga species ‘ Candidatus Nitrotoga arctica’ was highly enriched from permafrost soils (Alawi et al . 2007 ). NOB can also use cyanate for energy and as a nitrogen source. Cyanate-encoding genes clustered with NOB have also been found in Nitrososphaera gargensis and Scalindua anammox bacteria (Palatinszky et al . 2015 ). Sorokin et al . ( 2012 ) isolated a nitrite oxidizer that belongs to the widespread phylum Chloroflexi, Nitrolancetus hollandicus , from a nitrifying reactor. Nitrolancetushollandicus has a broad temperature range (25°C–63°C) but a low affinity for nitrite (K s = 1 mM) and can use formate as a source of energy and fix CO 2 via the Calvin cycle. Thiocapsa species can couple the anaerobic oxidation of nitrite directly to phototrophy (Griffin, Schott and Schink 2007 ). Intriguingly, although the disproportionation of nitrite (into nitrate and nitrous oxide (N 2 O)) would yield sufficient Gibbs free energy to sustain growth, no organisms capable of carrying out this reaction have been identified (Kuypers, Marchant and Kartal 2018 ). Comammox: complete ammonium oxidation to nitrate Ammonium oxidation to nitrate was once assumed to be a two-step reaction carried out by the subsequent action of ammonium and nitrite oxidizers. Despite a lack of biological proof for complete nitrification by a single organism, its existence and potential competitive advantage in biofilms with low substrate concentrations were proposed in 2006, based on modeling of the trade-off between growth rate (short pathways are faster) and growth yield (more complete pathways result in a higher energy yield) (Costa, Pérez and Kreft 2006 ). In 2015, the first comammox Nitrospira species were discovered in two different ecosystems (Daims et al . 2015 ; van Kessel et al . 2015 ). These observations expanded the metabolic potential of the Nitrospira clade, which was thought to contain only strict canonical aerobic NOB (Watson et al . 1986 ; Ehrich et al . 1995 ; Lebedeva et al . 2011 ). Since its discovery, comammox Nitrospira have been detected in several wastewater treatment reactors using metagenomics and primer-based approaches (Chao et al . 2016 ; Gonzalez-Martinez et al . 2016 ), drinking water systems (Pinto et al . 2016 ; Bartelme, McLellan and Newton 2017 ) and a variety of natural systems using a pmoA primer-targeted approach (Pjevac et al . 2017 ). Very recently, Kits et al . ( 2017 ) experimentally determined that the half saturation constant (K s ) for ammonium (0.65–1.1 µM) of Nitrospira inopinata was two orders of magnitude lower than that of any other cultured ammonium oxidizer, suggesting that N. inopinata is very competitive in environments with low ammonium concentrations. Anaerobic ammonium oxidation by anammox bacteria Hamm and Thompson ( 1941 ) reported that much less ammonium accumulated in anoxic water than expected based on stoichiometric calculations, providing the first indications of anammox. Chemical observations by Richards ( 1965 ) indicated the presence of alternative nitrogen loss pathways. In 1977 , Broda famously proposed two types of lithotrophs based on Gibbs free energy calculations of the reactions. The predicted phototrophic anaerobic ammonium oxidizers have yet to be identified. The other hypothesized ‘missing’ process was anaerobic oxidation of ammonium with nitrite/nitrate as the oxidant. Subsequent field observations also indicated higher ammonium losses than expected (Smith, Howes and Duff 1991 ). In the early 1990s, Mulder et al . ( 1995 ) reported on the biological N-loss in an anoxic wastewater treatment plant at the Gist-Brocades yeast factory in Delft, The Netherlands. To prevent hydrogen sulfide production from the high-sulfate wastewaters, copious amounts of calcium nitrate were added to suppress sulfate reduction. Inadvertently, the presence of sufficient ammonium, nitrite and nitrate under anoxic conditions created a suitable niche for anammox bacteria. Recordings of the ammonium concentrations in the influent and effluent revealed that after 8 months, ammonium disappeared under anoxic conditions (Mulder et al . 1995 ). After the manuscript on the study was rejected by numerous journals for not being relevant with respect to applied or environmental aspects of microbiology, the editor of FEMS Microbiology Ecology was brave enough to accept and publish the story (Mulder et al . 1995 ). The microbial nature and initial characterization of the biomass of the process were investigated by Gijs Kuenen and co-workers at TU Delft (Kuenen 2008 ). A few years after its discovery, a highly enriched anammox culture was obtained by continuous cultivation in an SBR system with substrate limitation and effective biomass retention (Strous et al . 1997 ). The anammox cells were further purified by density gradient centrifugation. These purified cells produced dinitrogen gas from ammonium and nitrite while incorporating 14 CO 2 into biomass (Strous et al . 1999 ). 16S rRNA analysis showed that the anammox bacteria belonged to the order Brocadiales within the phylum Planctomycetes (Jetten et al . 2010 ). For reviews on anammox biochemistry, physiology, application and ecosystem relevance, see Kartal, Kuenen and van Loosdrecht ( 2010 ), van Niftrik and Jetten ( 2012 ), and Kuypers, Marchant and Kartal ( 2018 ). In 2006, the first genetic blueprint of anammox bacteria was elucidated, which, together with sophisticated 15 N-nitrogen experiments, revealed that the anammox reaction includes the reactive intermediates nitric oxide (NO) and the powerful reductant and ‘rocket fuel’ hydrazine (N 2 H 4 ) (van de Graaf et al . 1997 ; Schalk et al . 1998 ). The mechanism, structure and biophysical properties of the key metabolic hydrazine synthase enzyme were recently elucidated (Kartal et al . 2011 ; Dietl et al . 2015 ). Anammox bacteria appear to fix carbon through the Wood-Ljungdahl (reductive acetyl-CoA) pathway with electrons derived from the oxidation of nitrite to nitrate (Schouten et al . 2004 ; de Almeida et al . 2011 ). Five genera ( Kuenenia, Brocadia, Anammoxoglobus, Scalindua and Jettenia ) of anammox bacteria are known, and 10 species have been described. For an extensive overview, see van Niftrik and Jetten ( 2012 ). None of these are available as pure culture, and current enrichments using bioreactors with planktonic cells or aggregates/granules reach up to 95% (Kartal et al . 2011 ). Electron microscopic analyses have indicated a unique intracytoplasmic compartment named the ‘anammoxosome’ with a membrane composed of a single layer of ladderane lipids (van Niftrik et al . 2004 ; Neumann et al . 2014 ). Genomic analysis (Strous et al . 2006 ) and subsequent experimental confirmation (van Teeseling et al . 2015 ) revealed that anammox bacteria do possess a peptidoglycan cell wall and thus should be considered Gram-negative bacteria. Nearly 28 thousand anammox-related 16S rRNA gene sequences have been identified thus far (NCBI, NLM, Bethesda, MA, USA, Feburary 2018), indicating that likely only a fraction of anammox diversity is known. Anammox bacteria have been detected in freshwater environments, including anoxic wastewater, sediments and agricultural soils and in marine systems, including coastal and estuarine sediments, anoxic basins, mangrove sediments and oxygen minimum zones (OMZs) (Isobe and Ohte 2014 ). From an ecosystem perspective, anammox bacteria contribute significantly to the oceanic nitrogen cycle (Dalsgaard et al . 2003 ; Kuypers et al . 2005 ; Lam et al . 2009 ; Pitcher et al . 2011 ; Bale et al . 2014 ; Lüke et al . 2016 ). Lüke et al . ( 2016 ) reported the co-occurrence of Scalindua , Nitrospina and novel microorganisms with dissimilatory nitrate reduction to ammonium (DNRA) potential (novel nrfA gene) in the Arabian Sea. The role of anammox bacteria in nitrogen loss has been investigated in global major OMZs, including the Black Sea, the Chilean and Peruvian OMZ, the Namibian OMZ and the Arabian Sea, where they are estimated to contribute to 50% of N loss (Kuypers et al . 2003 , 2005 ; Lam et al . 2009 ; Jensen et al . 2011 ; Kuypers, Marchant and Kartal 2018 ). In continental shelf sediments, their estimated contribution reaches 79% (Thamdrup and Dalsgaard 2002 ; Engström et al . 2005 ). Quantifying the contribution of anammox bacteria and denitrifiers to total oceanic nitrogen loss is an ongoing challenge (Babbin et al . 2008 ). Anammox bacteria have been shown to perform DNRA with formate as an electron donor (Kartal et al . 2007 ). Furthermore, the use of volatile fatty acids in anammox has been shown for ‘ Candidatus Anammoxoglobus propionicus’, which co-oxidizes propionate, acetate and formate with ammonium and ‘ Candidatus Brocadia fulgida’, ‘ Candidatus Jettenia caeni’ and ‘ Candidatus Scalindua profunda’, which co-oxidize acetate and formate with ammonium (Kartal et al . 2007 ; Kartal, Kuenen and van Loosdrecht 2010 ; van de Vossenberg et al . 2013 ; Ali et al . 2015 ). Caution is needed since experimental data on environmental factors and in situ species activity and regulation of metabolism are scarce. For a relevant perspective, see Voss and Montoya ( 2009 ). Iron- and manganese-dependent ammonium oxidation Several anammox species can reduce Fe 3+ at the expense of formate or acetate (Strous et al . 2006 ; van de Vossenberg et al . 2013 ; Zhao et al . 2014 ; Ali et al . 2015). Fe 2+ can be used as an electron donor for nitrate reduction by anammox and several denitrifiers (Strous et al . 2006 ; Oshiki et al . 2013 ). Contradictory reports on nitrification coupled to metal-oxide reduction appeared in the 1990s (Luther et al . 1997 ; Hulth, Aller and Gilbert 1999 ; Thamdrup and Dalsgaard 2000 ). The coupling of iron and/or manganese reduction to anaerobic ammonium oxidation should be feasible at physiologically relevant concentrations based on thermodynamic calculations (Table 1 ). Similar to Fe-AOM, the so-called Feammox process could be important in sediments with relatively low sulfate concentrations (Rooze and Meile 2016 ; Rooze et al . 2016 ). A number of field observations suggest that oxidation of ammonium can be coupled to the reduction of Fe 3+ , with dinitrogen gas, nitrite, or nitrate as the end product. Acidimicrobiaceae may oxidize ammonium under iron-reducing conditions (Gilson, Huang and Jaffé 2015 ; Huang and Jaffé 2015 ). The Feammox process has been observed in riparian wetlands (Clément et al . 2005 ; Shrestha et al . 2009 ; Ding, Li and Qin 2017 ), forested wetlands (Huang and Jaffé 2015 ), tropical forest soils (Yang et al . 2012 ), paddy field soils (Ding et al . 2014 ; Zhou et al . 2016 ), intertidal wetlands (Li et al . 2015 ) and anammox sludge (Li et al . 2018a , b ). During the Feammox process, the generation of dinitrogen gas is more favorable (−245 kJ/mol) than the generation of nitrite (−164 kJ/mol) or nitrate (−207 kJ/mol) (Luther et al . 1997 ; Clément et al . 2005 ; Shrestha et al . 2009 ; Kuypers, Marchant and Kartal 2018 ). Thermodynamic calculations of the Feammox process under natural conditions in Congo lobe sediments (1 µM Fe 2+ , 1 µM NO 2 – and NO 3 – , 100 µM NH 4 + , pN 2 0.718 atm, pN 2 O 1E-9 atm) revealed a Gibbs free energy change of −206.9 kJ/mol (Kiriazis 2015 ). However, significant accumulation of nitrate up to 113 µM was observed in the incubations, indicating possible nitrifying activity. Isotope tracing studies of Yangtze Estuary sediment slurry incubations showed a potential of 0.24–0.36 mg N kg −1 d −1 (Li et al . 2015 ). Li et al . ( 2015 ) suggested that the effects of tidal fluctuations on ferric iron reduction could mediate Feammox activity and nitrogen loss in intertidal wetland ecosystems. These findings imply alternative pathways of N loss from soils and sediments. Potential Feammox rates (i.e. 30 N 2 production rates) in paddy field soils range from 0.17 to 0.59 mg N kg −1 d −1 (Ding et al . 2014 ), comparable to the Feammox rates found for intertidal wetlands (0.24–0.36 mg N kg −1 d −1 ) (Li et al . 2015 ) and tropical forest soils (approximately 0.32 mg N kg −1 d −1 ) (Yang, Weber and Silver 2012 ). The Feammox reaction depends on the availability of ammonium and Fe 3+ . The oxidized form of iron is affected by pH, which regulates the reactivity of iron oxide minerals and iron redox reactions. However, iron-reducing bacteria can affect the Feammox process by controlling Fe 3+ reduction in anoxic environments. The iron-reducing bacteria Geobacteraceae spp. and Shewanella spp. may be directly or indirectly involved in ammonium oxidation (Clément et al . 2005 ; Shrestha et al . 2009 ; Li et al . 2015 ). Although these studies support the occurrence of Feammox in various environments, the key microbial organisms responsible for this process must be convincingly identified. Anoxic microbial fuel cells fed solely with ammonium could be a good model system to investigate the occurrence of Feammox in sediments but have received limited attention (Qu et al . 2014 ; Zhan et al . 2014 ; Jadhav and Ghangrekar 2015 ; Li et al . 2015 ; Reyes et al . 2016 ). Sulfate-dependent ammonium oxidation Sulfate-dependent ammonium oxidation is thermodynamically very challenging under biologically relevant conditions (Table 1 ) and would barely yield sufficient Gibbs free energy even at molar concentrations of ammonium. Very few field observations are available (Schrum et al . 2009 ), and there is no genomic evidence that anammox bacteria can use sulfate instead of nitrite as an electron acceptor. In 2008, the anammox bacterium ‘ Candidatus Anammoxoglobus sulfate’ was presumably enriched from an anammox reactor biomass fed with ammonium sulfate under anoxic conditions (Liu et al . 2008 ). Fdz-Polanco et al . ( 2001 ) proposed a two-stage sulfate-reducing ammonium oxidation (SRAO) in which sulfate is reduced to elemental sulfur. Zhang et al . subsequently proposed an alternative route in which sulfate is reduced to sulfide (Zhang et al . 2009 ). Furthermore, sulfur-driven iron reduction coupled to anaerobic ammonium oxidation was recently described by Bao and Li ( 2017 ). The interfaces of anoxic deep-sea brine pools may represent a possible ecosystem where very high ammonium and sulfate concentrations can be found (Daffonchio et al . 2006 ; Borin et al . 2013 ). Metagenomic surveys indicated a high diversity of microorganisms, including anammox bacteria, at these interfaces (Daffonchio et al . 2006 ; Speth et al . 2017 ). Dedicated high-pressure salt-resistant reactor equipment would be needed to successfully establish enrichment cultures on ammonium and sulfate from these ecosystems. Microbial interactions in the methane, sulfur and nitrogen cycles While the enrichment and characterization of individual ‘impossible’ anaerobic chemolithoautotrophic microorganisms is of great interest to microbiologists, these organisms do not live in isolation. In ecosystems, these microorganisms must collaborate to remove toxic intermediates or compete for limiting resources. Recently, the fate of ammonium, sulfide and methane under nitrate-reducing conditions similar to those in estuarine ecosystems was elegantly investigated in a bioreactor system (Russ et al . 2014 ; Arshad et al . 2017 ). Over time, an enrichment culture developed in which ‘ Candidatus Methanoperedens nitroreducens’, ‘ Candidatus Methylomirabilis oxyfera' and anammox bacteria coexisted with sulfide oxidizers. ‘ Candidatus Methanoperedens nitroreducens’ converted 53% of the supplied methane while reducing 69% of the nitrate to nitrite. Sulfide oxidizers contributed 31% to nitrite production. The nitrite was converted to dinitrogen gas by anammox bacteria (53%) at the expense of ammonium, by ‘ Candidatus Methylomirabilis oxyfera’ (37%) at the expense of methane, and by sulfide oxidizers (10%). Surprisingly, the metagenome of this anaerobic community was dominated by a new Nitrospirae species, ‘ Candidatus Nitrobium versatile’. Based on the retrieved genome, ‘ Candidatus Nitrobium versatile’ might produce ammonium from nitrite by sulfur disproportionation or utilize other one-carbon excretion products. Relatives of these Nitrospirae with similarly versatile potential have since been detected in gypsum-fertilized paddy fields (Zecchin et al . 2017 )."
} | 13,530 |
39050726 | PMC11267239 | pmc | 770 | {
"abstract": "The persistence of reef-building corals is threatened by macroalgal competitors leading to a major demographic bottleneck in coral recruitment. Whether parental effects exist under coral–algal competition and whether they influence offspring performance via microbiome alterations represent major gaps in our understanding of the mechanisms by which macroalgae may hinder coral recovery. We investigated the diversity, variability and composition of the microbiome of adults and larvae of the coral Pocillopora acuta and surrounding benthic substrate on algal-removed and algal-dominated bommies. We then assessed the relative influence of parental and offspring environmental effects on coral recruitment processes by reciprocally exposing coral larvae from two parental origins (algal-removed and algal-dominated bommies) to algal-removed and algal-dominated environmental conditions. Dense macroalgal assemblages impacted the microbiome composition of coral larvae. Larvae produced by parents from algal-dominated bommies were depleted in putative beneficial bacteria and enriched in opportunistic taxa. These larvae had a significantly lower survival compared to larvae from algal-removed bommies regardless of environmental conditions. In contrast, algal-induced parental and offspring environmental effects interacted to reduce the survival of coral recruits. Together our results demonstrate negative algal-induced parental and offspring environmental effects on coral recruitment that could be mediated by alterations in the offspring microbiome.",
"conclusion": "5 . Conclusion As macroalgae proliferate and threaten reef health [ 8 , 12 , 14 ], it is essential to characterize their effects on the microbial dynamics of coral development stages and their habitats to predict future coral community structure. Using 16S rRNA gene sequencing, we show that microbial specificity of two major life stages (i.e. adults and larvae) was maintained whether corals were transplanted on algal-removed or algal-dominated bommies. While our result revealed a remarkable stability of P. acuta adult microbiome, parental exposure to dense macroalgal assemblages influenced the microbiome of their larvae and their subsequent performance throughout the recruitment process, thereby indicating cross-generational effects under coral–macroalgae competition. Our data provide an additional mechanism by which macroalgae perpetuate their dominance on degraded reefs. The cumulative impacts of low larval survival and low survival of the juvenile phenotypes brooded by parents exposed to macroalgae in algal-dominated conditions may dramatically disrupt coral recovery trajectories. These results further suggest that algal-induced parental versus offspring environmental effects can form complex interactions across coral development stages. Deciphering between parental and environmental drivers of offspring success with respect to coral–macroalgal competition confers promising areas to further investigate the cryptic competitive mechanisms of macroalgae. Such knowledge will be valuable for predicting benthic community changes and providing science-based guidelines for reef conservation.",
"introduction": "1 . Introduction Corals live with a striking diversity of microbial symbionts (Symbiodiniaceae, bacteria, archaea, viruses, fungi) that underpin host health through metabolic and defence functions [ 1 ]. To establish and maintain associations with microbes across life stages is therefore essential for the holobiont survival. Microbial symbionts can be acquired through the environment (i.e. horizontal transmission) [ 2 ], the parents (i.e. vertical transmission) [ 3 ] or a combination of both (i.e. mixed strategy) [ 4 , 5 ]. A dynamic microbiome structuring occurs throughout coral ontogeny, with juvenile corals harbouring a more diverse and variable microbiome than adult microbiomes [ 4 , 6 ]. This ‘winnowing process’ during early life stages of marine invertebrates has been suggested to enable the establishment of fine-tuned microbial associations for a given environment [ 7 ]. Consequently, characterizing how environmental factors shape the coral microbiome at different life stages is critical to identify key host–microbe associations and to assess their stability or flexibility under stressful conditions. On coral reefs, the cumulative impact of human-driven stressors (e.g. warming ocean temperatures, overfishing, pollution) has resulted in extensive changes in benthic assemblages leading to phase shifts from coral to macroalgal dominance [ 8 ]. Once established, macroalgae can compromise the survival, growth, recruitment and fecundity of corals [ 9 – 12 ]. Microbially mediated coral–algal competition has been suggested as one of the main drivers of coral demise, but experimental evidence is clearly lacking [ 13 , 14 ]. Suggested mechanisms include contact-mediated effects such as allelopathy [ 15 , 16 ] and transmission of pathogens [ 17 ]. Water-mediated effects may also impact corals with the release of hydrophilic allelochemicals and dissolved organic carbon (DOC) feeding copiotroph and potentially virulent opportunistic microbes on the benthos and in the water column (i.e. the DDAM model: DOC, Disease, Algae and Microbes [ 13 , 18 – 21 ]). Although macroalgal assemblages shape the microbial and chemical composition of reef waterscapes [ 21 ], their effects on the coral microbiome can be inconsistent. For example, the microbiome of Pocillopora damicornis showed no variation in composition between algal- versus coral-dominated reefs in Fiji [ 22 , 23 ]. In another study, coral microbiome diversity and variability increased with increasing site-level algal cover in two islands of French Polynesia, even in the absence of direct contact [ 24 ]. Given the importance of microorganisms on holobiont health and the variable results produced by current studies, a better understanding of algal effects on the coral microbiome is needed to predict the resilience of coral communities as macroalgae proliferate. Coral recruitment is a fundamental process for the replenishment of coral communities [ 11 , 25 , 26 ]. Macroalgal assemblages can create a major demographic bottleneck in coral recruitment [ 26 , 27 ]. In particular, an in situ study comparing recruitment success between algal- versus coral-dominated reefs suggests negative algal effects on larval settlement and post-settlement survival [ 12 ]. These effects have been related to alterations in the microbiome of microbenthic communities [ 26 , 28 ]. However, the links between macroalgal abundance, coral recruitment success and the microbiome of coral early life stages are poorly understood. In Fiji, the composition of the larval microbiome of P. damicornis between algal- versus coral-dominated reefs remained unchanged despite lower larval survival in algal-dominated seawater [ 22 ]. Furthermore, parental effects (i.e. effect of the parental phenotype on the phenotype or performance of the offspring) can influence the survival and growth of coral offspring. The extent of these parental effects varies depending on the reproductive strategy. For instance, in the spawner Acropora tenuis and the brooder Porites astreoides , parental effects can explain up to 17 and 94% of the variance in juvenile survival, respectively [ 29 , 30 ]. In brooding species, that internally produce larvae, maternal effects are expected to be stronger than in broadcast spawning species due to a greater maternal investment [ 31 ] and a vertical transmission of microbial symbionts [ 3 , 4 ]. The environmental stress experienced by the parents can lead to positive and negative cross-generational effects on offspring performance and has been related to altered maternal provisioning [ 32 ]. For example, the transfer of essential metabolites (e.g. lipids, amino acids) is reduced in eggs of heat-stressed parents in soft corals [ 33 ]. In addition, parents could differentially invest in their offspring by vertically transmitting distinct Symbiodiniaceae [ 30 ] and prokaryotic communities [ 5 , 34 ]. So far, negative parental effects in corals and other marine invertebrates, such as urchins, have essentially been demonstrated relative to climate-associated stressors [ 32 – 34 ]. Yet, given that macroalgae can alter both health and fecundity of corals, one might expect that negative parental effects relative to algal competition might jeopardize coral recruitment success. To our knowledge, only one study [ 22 ] has suggested algal-induced parental effects on corals. However, those effects were not associated with changes in the offspring microbiome. Thus, whether parental effects exist under coral–algal competition and whether they could influence offspring performance via effects on the larval microbiome represent major knowledge gaps in our understanding of the mechanisms by which macroalgae may prevent coral recovery. In this work, we investigated the effect of dense macroalgal assemblages on the brooding coral Pocillopora acuta . We hypothesized that macroalgal assemblages may have negative parental and offspring environmental effects on coral recruitment and that these effects may be mediated by alterations in the coral and substrate microbiome. To test these assumptions, we conducted a manipulative field experiment in which we compared coral (i.e. adults and larvae) and microbenthic substrate microbiomes between algal-removed and algal-dominated bommies within a single fringing reef in Mo'orea, French Polynesia, to control for confounding reef-scale effects. Then, we assessed the relative influence of parental and offspring environmental effects on coral recruitment success by reciprocally exposing coral larvae from two parental origins (i.e. colonies originating from algal-removed versus algal-dominated bommies) to algal-removed and algal-dominated environmental conditions in a series of survival and settlement experiments.",
"discussion": "4 . Discussion By impairing coral holobiont health and recruitment success, macroalgae constitute an extensive threat to coral replenishment processes. Here, we demonstrate that dense macroalgal assemblages alter the coral larval microbiome and have negative parental and offspring environmental effects on coral recruitment. Our study is the first to link parental effects under coral–algal competition with alterations in the coral larval microbiome and to reveal interacting parental and offspring environmental effects on coral early life stages. It provides important new knowledge on the response of coral microbiomes to macroalgal dominance and how it relates to offspring performance. Our work showed that dense macroalgal assemblages did not significantly alter the microbiome of adult corals at the scale of coral bommies within a single fringing reef. Despite the large differences in macroalgal cover between the algal-removed and algal-dominated bommies (i.e. 4% versus 68%) and the five-month exposure to each treatment, the diversity, variability and composition of the coral adult microbiome did not significantly vary between the algal treatments. These results disagree with several studies suggesting that macroalgae can alter the coral microbiome [ 10 , 15 , 24 ]. However, they may be explained by the fact that our sampled colonies were initially placed without direct physical contact with macroalgae. Many of the mechanisms by which algae affect corals, including the release of allelochemicals [ 15 , 16 ] and dissolved organic matter [ 18 , 19 ], operate on small spatial scales. For example, the transmission of coral pathogens is likely to require direct contact [ 49 ]. These results could also be related to the identity of the coral host itself. Pocilloporids have a remarkable capacity to maintain their microbiome under macroalgal competition [ 22 ] and abiotic disturbances [ 50 , 51 ]. Ziegler et al. [ 50 ] proposed that microbiome flexibility is host specific with Pocilloporids being ‘microbiome regulators’, capable of some microbial regulations while keeping a constant microbiome [ 50 ]. This property may be related to their opportunist life-history strategies characterized by high recruitment rates and fast growth [ 52 ]. Although a lack of microbiome flexibility may hinder the acclimatization potential of corals [ 53 ], a stable microbiome could enhance coral resilience to rapid environmental changes and microbial pathogens [ 23 , 51 ]. For example, microbiome stability was associated with unaltered anti-pathogen activity against V. coralliilyticus in adult P. damicornis corals from both coral-dominated marine protected areas and macroalgal-dominated fished reefs in Fiji [ 23 ]. In our study, the relative abundance of Vibrionaceae in adult P. acuta corals was not significantly influenced by the algal treatment. In contrast, corals from both algal-removed and algal-dominated bommies were largely dominated (>85% relative abundance) by bacteria from the Endozoicomonadaceae family. This family of bacteria likely plays critical roles in the health and resilience of coral holobionts of different coral species, including Pocilloporids [ 23 , 54 ]. The coral larval microbiome was compositionally distinct and more diverse and variable compared to the adult microbiome. These ontogenetic changes are associated with a ‘winnowing process’ [ 7 ] and match with the previous studies on corals [ 4 – 6 ]. The coral larval microbiome was dominated by the Rhodobacteraceae and Alteromonadaceae families. These families have been detected in high abundance in larvae of P. acuta [ 4 ] and other coral species [ 3 , 6 ]. Importantly, our study shows that the composition of the coral larval microbiome differed significantly between larvae brooded by corals from algal-removed and algal-dominated bommies. Our knowledge of the processes by which bacterial communities are acquired during the larval stage is still very limited [ 2 – 4 ]. However, in the closely related brooding coral P. damicornis , the establishment of bacterial communities in the offspring is driven by both parental and planulation environments, with the majority of uptake occurring horizontally (i.e. from the environment [ 5 ]). Similarly, in P. acuta , larvae shared a minority of their ASVs with their parents [ 4 ]. Since our colonies were maintained in seawater from their respective treatment during larval release, the relative importance of parental versus planulation environments on larval microbes cannot be separated. Larvae were collected less than 8 h upon emergence and rinsed in FSW three times before preservation. In a similar time frame, vertical transmission of bacteria can occur in the form of bacterial aggregates in newly released P. acuta larvae [ 4 ]. Likewise, bacterial cells in the coral P. astreoides were detected in the ectoderm of larvae in less than an hour after release constituting the vertically transmitted microbiome [ 3 ]. Therefore, it is likely that, in our study, both parental and planulation environments drove the differences in coral larval microbiome between algal treatments. ASVs enriched in larvae brooded by parents from algal-dominated bommies were affiliated to the genera Erythrobacter , Thalassolituus and Marinobacterium . These bacteria have previously been associated with coral disease and coral–algal competition [ 55 , 56 ]. In contrast, the microbiome of these larvae was depleted in putative beneficial bacteria including members of the Rhodobacteraceae and Vibrio spp. For example, Sulfitobacter and Vibrio can produce and/or degrade dimethylsulfoniopropionate (DMSP) [ 57 ]. DMSP and its breakdown products (e.g. DMS) play major roles in coral health as antioxidants and antibiotics. These bacteria are also crucial for carbon and nitrogen acquisition, including V. parahaemolyticus [ 6 , 58 , 59 ]. While some Vibrio species like V. coralliilyticus and V. parahaemolyticus might also constitute opportunistic pathogens [ 1 , 59 ], Vibrio spp. can form mutualistic relationship with corals by providing nutrients and preventing bacterial colonization [ 60 , 61 ]. The loss of these beneficial bacteria could compromise larval physiology and defence through a decrease in their nutritional and protective functions. Remarkably, the relative abundance of the Alteromonadaceae family was not impacted by the algal treatment, demonstrating a strong host–microbe association. Some Alteromonas strains provide nitrogen sources [ 6 , 58 ], participate in larval settlement induction [ 25 ] and prevent pathogenetic invasion [ 62 ]. Macroalgal assemblages significantly structured the substrate microbiome, as previously demonstrated [ 26 ]. Microbial communities on algal-dominated bommies likely responded to macroalgal-induced changes in water chemistry. For example, algal-derived dissolved organic matter differs in composition from that of corals influencing the microbial structure of biofilm and bacterioplankton communities [ 20 , 63 ]. These labile and energy-rich algal exudates typically favour the growth of copiotrophic bacteria potentially detrimental to corals [ 20 , 21 ]. On exposed surfaces, certain bacteria enriched on algal-dominated bommies have been associated with coral diseases, including the genera Lewinella and Rivularia and the family Saprospiraceae [ 24 , 56 ]. Similarly, cryptic surface microbiomes from algal-dominated bommies harboured bacterial opportunists of corals, such as the genus Limibaculum [ 64 ], and several bacterial classes enriched on algal-dominated bommies have been negatively correlated with coral settlement, including Thermonanaerobaculia and Planctomycetes [ 28 ]. On algal-dominated bommies, several bacterial families and genera decreased relative to algal-removed bommies, potentially due to changes in competitive dynamics between microorganisms within microbenthic microbiomes [ 65 ]. After 6 days of exposure to seawater from algal-removed or algal-dominated bommies, mean larval survival remained high (>80%) in both algal treatments. Such high survival rates are consistent with the previous studies showing that Pocilloporid larvae can survive in the plankton for over 100 days [ 66 ]. Nevertheless, we found a significant effect of parental origin on larval survival. Larvae produced by parents from algal-dominated bommies had their survival reduced by approx. 6% relative to those produced by parents from algal-removed bommies. Similar and even stronger cross-generational effects of algal dominance on larval survival have been demonstrated when comparing larval survival from a coral-dominated marine protected area and a macroalgal-dominated fished area in Fiji [ 22 ]. Since the microbiome composition of larvae significantly differed between parental origins, these effects could be due to a compromised larval microbiome. As discussed earlier, larvae brooded by parents from algal-dominated bommies harboured a microbiome enriched with opportunistic bacteria and depleted in beneficial bacteria, which could have increased their mortality in the pre-settlement stage. At 6 days old, which corresponds to the duration of larval exposure to the algal treatments, coral larvae also extensively rely on their parental reserves [ 67 ]. Since reproduction is a costly process [ 9 ], algal-stressed parents may have faced a physiological trade-off in which they re-allocated energy for their own metabolism and defence rather than towards larval provisioning, thereby influencing their offspring survival. For example, bleached corals alter the quality and/or quantity of metabolites transferred to their larvae [ 33 ]. We failed to detect any effect of the environment (in this case, water origin) on larval survival. Previous laboratory-based studies found that polar algal compounds [ 15 ] and/or algal-associated microbes [ 18 ] could significantly decrease larval survival, contradicting our results. By using seawater, where algal dissolved compounds and microbes occur under natural concentrations, our result provides limited evidence of water-mediated effects on larval survival (but see [ 22 ]). However, these effects may have been underestimated as seawater was collected in the upper water column (approx. 50 cm) above the bommies and not near algal surfaces. While macroalgae often prevent coral larval settlement in field and laboratory experiments [ 26 , 27 ], we found no parental and offspring environmental effects on coral larval settlement. Larvae settled equally on tiles from both algal treatments, although, by avoiding settling on tiles from algal-dominated bommies, they could have escaped the enhanced post-settlement mortality. This result occurred in spite of significant differences in substrate microbiome between algal treatments, suggesting that the substrate microbiome was not indicative of the negative environment for the larvae. It is possible that this result is due to the presence of positive cues (e.g. CCA) and/or limited abundance of negative cues (e.g. macroalgae) on the settlement tiles. After four months of conditioning, allowing for a mixed benthic community of settlement inducers and inhibitors [ 68 ], cryptic surfaces of settlement tiles were covered by CCA, a well-known coral settlement cue [ 69 ] and not densely covered by macroalgae (electronic supplementary material, figure S9). In addition, Pocilloporids are rather unselective of their microhabitats during settlement [ 70 ]. In contrast, algal-induced parental effects interacted with offspring environmental effects to affect the survival of coral recruits. Post-settlement survival was only reduced on algal-dominated bommies for recruits produced by parents from algal-dominated bommies. Such interaction may occur if algal-induced environmental effects coincide with the vulnerable state of recruits owing to their (i.e. algal-dominated) parental origin and/or if negative algal-induced parental effects are offset by the positive effects of the algal-removed environment. Our results do not allow us to distinguish between these two scenarios. However, the altered microbiome of coral larvae produced by parents from algal-dominated bommies is consistent with an increase in their vulnerability to detrimental (e.g. algal-dominated) environments. By further supporting the negative effects of macroalgae on corals, our study adds further incentives to reduce macroalgal abundance on coral reefs. Because the conventional management actions of herbivore protection and water quality management are often insufficient to prevent coral–macroalgal phase shifts, physical removal of macroalgae has been suggested as a valuable intervention to rehabilitate degraded reefs [ 71 ]. Manually removing macroalgae (i.e. ‘sea-weeding’) can promote local-scale coral recruitment and recovery [ 72 , 73 ]. Our findings suggest that, if algae are removed from the vicinity of sexually mature adults for a few months prior to larval release, these adults could produce a more robust offspring and enhance coral recruitment, even on algal-dominated reefs. Future research should determine whether the positive parental effects of macroalgal removal on coral recruitment persist in the long term and whether similar effects occur in other coral species, in particular broadcast spawners."
} | 5,833 |
39248387 | PMC11558148 | pmc | 771 | {
"abstract": "Abstract Triboelectric nanogenerators (TENGs) combine contact electrification and electrostatic induction effects to convert waste mechanical energy into electrical energy. As conventional devices contribute to electronic waste, TENGs based on ecofriendly and biocompatible materials have been developed for various energy applications. Owing to the abundance, accessibility, low cost, and biodegradability of biowaste (BW), recycling these materials has gained considerable attention as a green approach for fabricating TENGs. This review provides a detailed overview of BW materials, processing techniques for BW‐based TENGs (BW‐TENGs), and potential applications of BW‐TENGs in emerging bioelectronics. In particular, recent progress in material design, fabrication methods, and biomechanical and environmental energy‐harvesting performance is discussed. This review is aimed at promoting the continued development of BW‐TENGs and their adoption for sustainable energy‐harvesting applications in the field of bioelectronics.",
"conclusion": "6 Conclusion This review provides a comprehensive analysis of the current state of BW‐TENG devices. Outstanding advances have been achieved in the development TENGs, and considerable progress has been made in the use of biocompatible materials for the fabrication of environment‐friendly TENGs. Notably, the recycling and upcycling of waste materials to fabricate TENGs contributes significantly to reducing the ecological impact of these devices. Owing to their abundance and inherent triboelectric properties, BW materials have been extensively explored for TENG fabrication. Specifically, BW‐TENGs have been prepared using numerous plant‐ and animal‐based BW materials. A variety approaches has been used to incorporate BW materials into TENGs, including direct attachment, conversion into fine powders for either direct attachment or dispersion in a polymeric medium, chemical treatment to extract refined constituent materials, and carbonization. Despite being a relatively new approach, BW‐TENGs have been implemented in various modern applications. In particular, BW‐TENGs have been used to power small devices, light multiple LEDs, act as self‐powered sensors, develop HMI‐based devices, and monitor physiological signals. Despite being sustainable and innovative approach, BW‐TENGs often suffer from certain limitations. Challenges such as the excessive use of chemicals during refinement, performance degradation due to humidity and other factors, short lifetimes, poor durability, and easy contamination must be addressed to broaden the commercial applicability of BW‐TENGs. In particular, the effects of mixing different BW materials on the stability, durability, and efficiency of TENGs should be explored. Despite these challenges, BW‐TENGs are promising systems with wide applicability, especially for HMI devices, owing to their biocompatibility, portability, wearability, and efficiency. Following further investigation and the implementation of suitable material selection, processing, and device fabrication strategies, BW‐TENGs are expected to have a notable impact on next‐generation emerging bioelectronics while also contributing to environmental sustainability and pollution control.",
"introduction": "1 Introduction Triboelectric nanogenerators (TENGs) harvest mechanical energy from simple body movements through frictional effects. Owing to the convenience, efficiency, and broad applicability of TENGs, the development of these devices has expanded to incorporate other natural phenomena, such as wind, flowing water, and rain. TENGs are typically employed in small electronic and optoelectronic devices, medical equipment, robotics, and self‐powered electronics. [ \n \n 1 \n \n ] However, as these devices commonly consist of materials that are difficult to biodegrade and discharge toxic substances upon degradation, [ \n \n 2 \n \n ] TENGs contribute to electronic waste (e‐waste). Rapid progress in mechanization and technology has sharply increased global e‐waste production. As e‐waste now accounts for a substantial share of solid waste, the numerous non‐biocompatible and hazardous components in e‐waste pose a critical threat to the environment and human health. To limit e‐waste and the associated hazards, biodegradable materials have received increasing interest for the development of environment‐friendly electronics. [ \n \n 3 \n \n ] The incorporation of natural, nontoxic, and biocompatible triboelectric components into smart electronics could contribute to the creation of a next‐generation green society. Natural materials are discarded daily on a large scale. As biowaste (BW) is abundant in common biocompatible materials including cellulose, lignin, chitosan, gelatin, and collagen, which can participate in triboelectric production, [ \n \n 4 \n \n ] such has been considered for TENG development. The recycling of BW as triboelectric layers that harvest mechanical energy does not require complicated methods. Thus, the large‐scale production and commercialization of next‐generation BW‐based TENGs (referred to as BW‐TENGs hereafter) provides an effective approach for utilizing BW while actively reducing e‐waste toward the realization of a pollution‐free society. BW‐TENGs have attracted significant attention in the field of next‐generation energy harvesting technologies for various electronics including sensors and displays toward achieving the goal of net‐zero emissions. [ \n \n 6 \n \n ] Many studies on BW‐TENGs have primarily focused on mechanical and biomechanical energy harvesting. Owing to their nontoxic and biocompatible nature, BW‐TENGs can be applied in real‐time healthcare and biomedical devices to harvest energy from limb movements, pulses, and heartbeats. Advantageously, these devices can be easily absorbed by the body without negative effects, eliminating the need for surgical removal. [ \n \n 5 \n \n ] Since 2021, the applicability of BW‐TENGs in other fields, such as human–machine interfaces (HMIs) and self‐powered bioelectronics, has also been explored. Notably, BW‐TENGs produce a wide range of output voltages, from a few volts to thousands of volts, which facilitates integration with a variety of devices. Moreover, these energy‐harvesting components offer impressive sustainability, addressing environmental issues including the increasing production of BW and e‐waste as well as the exhaustion of nonrenewable energy sources. However, as BW‐TENGs are a relatively new approach, many aspects of these systems, such as output optimization and application scope, require further investigation. Although TENGs based on active materials such as bioresorbable materials, [ \n \n 6 \n \n ] degradable materials, [ \n \n 7 \n \n ] cellulose, [ \n \n 8 \n \n ] textiles, [ \n \n 9 \n \n ] and polymers [ \n \n 10 \n \n ] have been demonstrated, this is the first review to provide a systematic overview of BW‐TENGs. Herein, we discuss recent progress on BW‐TENGs, with a focus on the use of different types of BW, performance characteristics, and applications as triboelectric components for bioenergy materials, biotechnologies, and bioelectronics. Specifically, Section 1 introduces BW‐TENGs and the need for biocompatible and biodegradable materials in current electronic applications. Section 2 discusses biocompatible and biodegradable types of BW as sustainable bioenergy materials, including the origin of their triboelectric properties, and BW processing methods for preparing TENGs. Section 3 reviews the biomechanical and environmental energy‐harvesting performance of various BW‐TENGs. Section 4 outlines emerging bioelectronics based on BW‐TENGs for next‐generation biotechnologies. Finally, Section 5 discusses challenges, limitations, and opportunities for further advancing the development of BW‐TENGs. In addition to highlighting the suitability of BW‐TENGs for commercialization owing to the abundance of their constituent materials and ability to reduce environmental pollution, this review is anticipated to inspire future research on BW‐TENGs and their applicability in various fields."
} | 2,017 |
30833729 | PMC6453112 | pmc | 772 | {
"abstract": "Methanogenesis is an ancient metabolism of key ecological relevance, with direct impact on the evolution of Earth’s climate. Recent results suggest that the diversity of methane metabolisms and their derivations have probably been vastly underestimated. Here, by probing thousands of publicly available metagenomes for homologues of methyl-coenzyme M reductase complex (MCR), we have obtained ten metagenome-assembled genomes (MAGs) belonging to potential methanogenic, anaerobic methanotrophic and short-chain alkane oxidizing archaea. Five of these MAGs represent under-sampled (e.g., Verstraetearchaeota, Methanonatronarchaeia, ANME-1) or previously genomically undescribed (ANME-2c) archaeal lineages. The remaining five MAGs correspond to lineages that are only distantly related to previously known methanogens and span the entire archaeal phylogeny. Comprehensive comparative annotation significantly expands the metabolic diversity and energy conservation systems of MCR-bearing archaea. It also suggests the potential existence of a yet uncharacterized type of methanogenesis linked to short-chain alkane/fatty acid oxidation in a previously undescribed class of archaea (‘ Ca . Methanoliparia’). We redefine a common core of marker genes specific to methanogenic, anaerobic methanotrophic and short-chain alkane-oxidizing archaea, and propose a possible scenario for the evolutionary and functional transitions that led to the emergence of such metabolic diversity.",
"discussion": "Results and Discussion Additional lineages of archaea with an MCR or MCR-like complex To identify previously undescribed lineages of potential methanogens, anaerobic methanotrophs and short-chain alkane oxidizers, we probed available metagenomes from the JGI/IMG database for McrA homologues and identified sequences distantly related to well characterized lineages ( Methods ). Ten MAGs were reconstructed from the corresponding metagenomes sourced from a wide range of anoxic environments including an inland petroleum reservoir from Brazil, oil seeps from USA 25 , soda lake sediments from Russia, and hot-springs from China and USA ( Table 1 ). Nine of the ten MAGs had an estimated completeness ranging from 78.4 to 94.4%, and one was only 51.5% complete. Estimated contamination (without strain heterogeneity) ranged from 0 to 3.3%. Four MAGs represent three previously undescribed lineages only distantly related to known methanogenic/methanotrophic archaea ( Fig. 1 ): NM1 (NM1a and NM1b MAGs) branches within the Methanotecta superclass 15 , between Archaeoglobales and the clade formed by ‘ Ca . Syntrophoarchaeales’ and ‘ Ca. Methanophagales’ 15 (ANME-1); NM3 branches within the Acherontia superclass 15 , at the base of the clade formed by the non-methanogenic Theionarchaea 26 and ‘ Ca. Methanofastidiosa’ (former WSA2/Arc1); NM4 branches within the TACK superphylum and is related to Korarchaeum cryptofilum . In the NM4 MAG, the markers for methane metabolism are present on two contigs with a lower coverage than the other contigs and contain few genes related to Korarchaeum. However, an independent study supports that these contigs belong to the same organism, provisionally named “ Candidatus Methanodesulfokores washburnensis” for its methane- and sulfur-cycling capacities (McKay et al., submitted). One additional MAG (NM2) was too partial to assess its phylogenetic placement. Finally, five MAGs correspond to currently under-sampled archaeal lineages: a deep-branching Verstraetearchaeota, a third Methanonatronarchaeia, a second ‘ Ca . Methanophagales’ (ANME-1), a second GoM-Arc1 (showing a close relationship with the methanotrophic Methanoperedens), and the first representative of ANME-2c ( Fig. 1 ). Phylogeny and functional inference of MCR/MCR-like complexes To investigate the evolutionary relationships and characteristics of the MCR complexes identified in this study, we built a phylogeny based on a concatenated alignment of McrABG subunits. This phylogeny is in overall agreement with recently published ones 16 , 18 . Three MAGs (NM1a, NM1b and GoM-Arc1-GOS) encode alternative McrABG-like complexes that cluster with those of ‘ Ca . Syntrophoarchaeum’ and Bathyarchaeota ( Fig. 2A , in blue). The presence of an MCR-like complex and the absence of a canonical MCR in GoM-Arc1-GOS are consistent with the recent description of a MAG of this lineage 23 , and represent so far a unique feature within the Methanosarcinales. The remaining MAGs harbour canonical MCR complexes ( Fig. 2A ), branching next to their closest MCR-bearing neighbors in the reference archaeal phylogeny ( Fig. 1 ), suggesting no recent horizontal gene transfers (HGT). The clustering of NM3 with Methanofastidiosa supports an early presence of methanogenesis in the Acherontia. The clustering of NM4 with Verstraetearchaeota, support that it is a genuine methanogenic/methanotrophic representative of the TACK. The separate branching of ANME-2c from the other ANME-2 lineages suggests that anaerobic methane oxidation in Methanosarcinales emerged multiple times independently from methanogenic ancestors. Interestingly, both NM1a and NM1b encode, in addition to the McrABG-like complex, a canonical MCR complex branching at the base of Class II methanogens, consistently with the reference phylogeny. The coexistence of MCR and MCR-like complexes in the same archaeon has never been observed before and brings into question the metabolism of this lineage (see below). It is striking to observe that most of the predicted or experimentally proven methyl-dependent hydrogenotrophic methanogens are closely related in the MCR tree ( Fig. 2A , in red), irrespective of their placement in the reference phylogeny ( Fig. 1 ). This might be the consequence of ancient exchanges of the MCR complex among these lineages, whose direction is hard to define. Nevertheless, some more recent transfers may be identified. For example, ’ Ca . Methanophagales’ (ANME-1) MCRs branch far from their Methanotecta relatives, and might have acquired their MCR complex from a methanogenic member of the Acherontia. The clustering of MCR-like homologues belonging to distantly related lineages ( Fig. 2A , in blue) is also puzzling. This might be due to HGT and/or tree reconstruction artefacts linked to their high sequence divergence with respect to canonical MCRs, exemplified by their longer-than-average branches. Such divergence is probably related to a change in function, as MCR-like complexes are involved in activating short-chain alkanes (butane and propane) in ‘ Ca . Syntrophoarchaeum’ 21 . Accordingly, several residues playing an important role in canonical MCR, either by interacting with cofactors, forming the catalytic site cavity wall or being post-translationally modified, are not conserved in ‘ Ca . Syntrophoarchaeum’ sequences ( Fig 2B ). The replacement of large aromatic residues (e.g. Phe330, Tyr333, Tyr444, Tyr446) present in the cavity wall of canonical MCR 27 by smaller ones in ‘ Ca . Syntrophoarchaeum’ MCR-like complexes could have occurred to accommodate butane/propane (larger substrate than methane) in the catalytic site ( Fig. 2B ). The presence of smaller amino acids at these positions in NM1 and Bathyarchaeota MCR-like complex suggest a similar function in short-chain alkane oxidation. Finally, the MCR-like sequences of GoM-Arc1 show fewer modifications at these sites, suggesting the utilization of a smaller alkane, possibly ethane or methane. Expanded diversity of methyl-dependent hydrogenotrophic methanogens The NM3 and NM4 MAGs share several similarities with the recently discovered order-level lineages of methanogens that were proposed or experimentally proven to perform methyl-dependent hydrogenotrophic methanogenesis 10 , 17 – 19 ( Fig. 3 , Supplementary Table 1 ). First, these relatively complete MAGs (85,5% completeness) lack at least 24 genes coding for the MTR complex, H 4 MPT biosynthesis, and the H 4 MPT methyl-branch of the WL pathway, otherwise present in all Class I/II methanogens ( Supplementary Table 1 ). Second, they encode [Ni-Fe] hydrogenases and methyltransferases with the potential to support methanogenesis from methanol (MtaABC) in NM3 and NM4 and methanethiol (MtsAB) in NM3 ( Fig. 3 ; Supplementary Table 1 ). Interestingly, energy conservation complexes of NM3 are mostly similar to Methanofastidiosales 17 ( Supplementary Fig. 1 ), their closest related methanogens in the reference phylogeny ( Fig. 1 ). Altogether, this suggests that NM3 and NM4 rely on methyl-dependent hydrogenotrophic methanogenesis ( Fig. 3 ; Supplementary Discussion for details on energy conservation in NM3 and NM4). The predicted methanogenesis pathway in Verst-YHS (Verstraetearchaeota) and Mnatro-ASL (Methanonatronarchaeia) MAGs also supports methyl-dependent hydrogenotrophic methanogenesis ( Fig. 3 , Supplementary Table 1 ), as described in the first genomic assemblies for these lineages 18 , 19 . However, comparison of the energy conservation enzymes in the seven currently available Verstraetearchaeota (order Methanomethyliales) suggests an alternative model than previously described 19 ( Fig. 3 , Supplementary Table 1 ). Indeed, we found that all Methanomethyliales MAGs (95% average completeness) lack the HdrA/MvhD and possibly MvhAG subunits of the electron-bifurcating complex HdrABC/MvhADG, suggesting that this complex is absent in these archaea. In contrast, we identified in these genomes a gene cluster encoding a potential complex composed of a membrane-bound hydrogenase and of HdrBC (tentatively named Energy-converting Hydrogenase D or Ehd; Supplementary Fig. 2 ). We propose that this complex could be involved in a previously unreported mode of energy conservation associated with methanogenesis ( Fig. 3 ; Supplementary Discussion ). Insights into methane and short-chain alkane oxidizers GoM-Arc1-GOS, ANME-1-THS and ANME-2c MAGs possess a WL pathway and lack the methyltransferases and [Ni-Fe] hydrogenases required for methylotrophic and hydrogenotrophic methanogenesis, respectively ( Fig. 3 ; Supplementary Table 1 ), similar to all available MAGs of methanotrophs and short-chain alkane oxidizer ( Supplementary Fig. 3 ). Although they encode an AMP-producing acetyl-CoA synthetase (Acs) which is used for aceticlastic methanogenesis in Methanosaeta spp., they could rather use it for acetate assimilation 11 . Comparison with methanotrophs and short-chain alkane oxidizers also reveals a common core of enzymes for energy conservation, comprising the F 420 H 2 :quinone (or phenazine) oxidoreductase (Fqo/Fpo) and a potential electron confurcating complex (HdrABC/MvhD/FdhB 28 ) coded by a conserved gene cluster ( Supplementary Fig. 4 ). ANME-2c and GoM-Arc1-GOS encode 17 and 10 multiheme c-type cytochromes respectively, supporting the importance of direct electron transfer to syntrophic partners in anaerobic methane 29 , 30 and short-chain alkane oxidation 21 metabolisms ( Supplementary Fig. 3 ; Supplementary Table 1 ). ANME-1-THS MAG is the first sequenced representative of a “Land clade” within the ‘ Ca . Methanophagales’ ( Supplementary Fig. 5 ), suggesting different adaptations to environmental conditions than members of the ANME-1b clade, which are mainly from marine methane seeps. ANME-1-THS differs from the ANME-1b MAG 31 by the presence of a bacterial-like Rnf complex that could couple the NAD:ferredoxin oxidoreduction with chemiosmotic gradient generation/utilisation ( Fig. 3 ; Supplementary Fig. 6; Supplementary Discussion ). If these genes are not in the missing region of this MAG, ANME-1-THS might also differ from the other ANMEs by the lack of multiheme c-type cytochromes to transfer electrons from methane oxidation to a syntrophic partner ( Fig. 3 ; Supplementary Fig. 3 ). Alternatively, two PsrABC-like complexes, including a molybdenum/selenocysteine-containing dehydrogenase subunit, could be involved in the reduction of inorganic compounds such as polysulfide/elemental sulfur 32 , 33 ( Fig. 3 ). This contrasts with ANME-1b MAG which misses the membrane integral (PsrC-like) subunit needed to transfer electrons from membrane-associated electron transporters ( Supplementary Fig. 3 ). These characteristics might indicate growth of ANME-1-THS without bacterial syntrophs. The gene content of GoM-Arc1-GOS is consistent with the recent description of the first member of the GoM-Arc1 lineage 23 . While GoM-Arc1 members encode an MCR-like complex possibly involved in short-chain alkane oxidation ( Fig. 3 ), they lack the beta-oxidation pathway proposed to be involved in butane/propane utilization in ‘ Ca . Syntrophoarchaeales’ 21 ( Supplementary Fig. 3 ). If GoM-Arc1 members are capable of oxidizing ethane (CH 3 CH 3 ), as suggested by the fewer modifications observed in the catalytic site of its MCR-like complex relative to canonical MCRs ( Fig. 2 ), the oxidation of the ethyl-group would lead to an acetyl-group that could directly enter the oxidative WL pathway, making the beta-oxidation pathway unnecessary ( Fig. 3 ; Supplementary Discussion ). With the presence of Fqo, HdrABC/MvhD/FdhB, multiheme c-type cytochromes, and HdrDE ( Supplementary Table 1 ), the energy conservation system associated with this potential ethane-oxidation metabolism in GoM-Arc1 would mostly resemble that associated with methanotrophy in their closely related ANME-2 lineages ( Supplementary Fig. 3 ). The question remains whether the MCR-like homologs of GoM-Arc1 could also be capable of methane oxidation. A previously uncharacterised type of methanogenesis? The two NM1 MAGs represent the first archaea predicted to encode both an MCR and an MCR-like complex ( Fig. 2 ), suggesting that they might be potentially capable of both methane and short-chain alkane metabolisms ( Fig. 4 ; Supplementary Table 1 ). Interestingly, while both NM1 MAGs encode the MTR and the m-WL pathway similarly to Class I/II methanogens, they lack the [Ni-Fe] hydrogenases (MvhA and FrhA) and methyltransferases needed for hydrogenotrophic and methylotrophic methanogenesis, respectively. They also diverge from Class I/II methanogens by the replacement of the F 420 dependent methylene-tetrahydromethanopterin dehydrogenase (Mtd) by MtdB, which relies on NAD(P) redox cofactor in Methylobacterium extorquens 34 . Beyond the presence of an MCR-like complex, the potential ability of NM1 for short-chain alkane oxidation is also suggested by the presence of a complete beta-oxidation pathway with several gene copies per step, and a complete WL pathway (including CODH/ACS) as in ‘ Ca . Syntrophoarchaeales’ 21 . In addition, NM1 encode multiple long-chain fatty acid acyl-CoA synthases (FadD-like), not present in ‘ Ca . Syntrophoarchaeales’. Long chain fatty acids (LCFA) activated with these enzymes can enter the beta-oxidation pathway. NM1a and NM1b also encode multiple AMP-forming acetyl-CoA synthetase (Acd) to generate ATP from LCFA degradation. These enzymatic redundancies suggest a versatility toward substrates, as previously proposed for Syntrophus aciditrophicus 35 and Archaeoglobus fulgidus 36 . Consistently, analysis of the environmental distribution of NM1 ( Supplementary Fig. 7 ) reveals their common association with anoxic hydrocarbon-rich environments including methane seeps and oil-rich environments, where short-chain alkanes and long-chain carboxylic acids can be present in substantial concentrations 37 , 38 . In particular, NM1a and NM1b originate from an enrichment culture based on petroleum fluids and from a natural oil seep. In addition to this potential wide substrate range, NM1 also contrast with ‘ Ca . Syntrophoarchaeales’ in terms of energy conservation by lacking homologues of the NADH/F 420 H 2 :quinone oxidoreductase (Nuo/Fqo) and multiheme c-type cytochromes ( Fig. 4 ; Supplementary Table 1 ). Also, NM1 contain an Rnf complex potentially using NAD instead of menaquinone for ferredoxin oxidoreduction, similarly to ANME-1-THS ( Supplementary Fig. 6; Supplementary Discussion ). In the absence of membrane-bound enzymes involved in oxidoreduction of lipid-soluble electron carriers, of multiheme c-type cytochromes for direct interspecies electron transfer, of confurcating [Fe]-hydrogenase for interspecies H 2 transfer 39 , and of enzymes involved in dissimilatory reduction of inorganic compounds, the nature of the terminal electron acceptor coupled to alkane/LCFA oxidation remains elusive. Although both MAGs are mostly complete (~90%), it cannot be excluded that some of these enzymes are coded in their missing regions, or that an alternative way to transfer electrons to a terminal acceptor exists (e.g. utilisation of the assimilatory-type sulfite reductase present in both MAGs for dissimilatory reduction of sulfite, direct electron transfer not relying on cytochromes, or utilization of cytochromes produced by a syntrophic partner). Alternatively, we speculate that in NM1, methanogenesis involving the canonical MCR complex could act as a sink for the electrons produced during alkane and LCFA oxidation. Several electron-bifurcating/confurcating complexes encoded in the two NM1 MAGs ( Supplementary Fig. 8 ) together with the Rnf complex could be involved in this metabolism. The conversion of alkane and LCFA into CH 4 and acetate is thermodynamically feasible but was only reported to occur through syntrophic partnerships between a bacterium (performing the beta-oxidation) and a H 2 -consuming methanogen 40 , 41 , and it thus remains to be proven experimentally whether this can occur in a single organism. Based on the presence of methane and short-chain alkane/fatty acid-related enzymes and the preferential association with hydrocarbon-rich environments, we propose the provisional class ‘ Candidatus Methanoliparia’, with ‘ Candidatus Methanoliparum thermophilum’ for NM1a and ‘ Candidatus Methanolliviera hydrocarbonicum’ for NM1b (see Supplementary Discussion for full taxonomy and nomenclature). A core of markers related to methane and short chain-alkane metabolisms A group of 38 genes present in most methanogens and absent from most other organisms, generally referred to as “methanogenesis core markers”, was previously defined from Class I/II methanogen genomes 42 , 43 ( Supplementary Table 2 ). Half of them have an unknown function. The others correspond to MCR and MTR subunits, enzymes for biosynthesis and activation of the F 430 prosthetic group of MCR, and post-translational modifications in the McrA catalytic site 44 , 45 . We reassessed the occurrence of these markers in the ten assembled MAGs as well as reference genomes covering all recently discovered lineages of methanogens, methanotrophs and short-chain alkane oxidizers ( Table 2 ). Our analysis shows that some markers are no longer universal in Class I/II methanogens (e.g. m37, 38). Also, many marker genes shared by all or most Class I/II methanogens were predicted to be nonessential in Methanococcus maripaludis S2 46 ( Table 2 ). These non-universal and nonessential genes could possibly be involved in fine-tuning of methanogenesis (e.g. post-translational modification of MCR 47 ) or in its regulation under specific environmental conditions that are not encountered by all methanogens. For example, m21 and m24 are missing in several methanogens from nutrient-rich environments, such as Methanobrevibacter , Methanosphaera and Methanocorpusculum , and could be involved in regulatory processes related to changes in substrate/nutrient availability. All the lineages of predicted and experimentally proven methyl-dependent hydrogenotrophic methanogens 17 – 19 lack numerous markers ( Table 2 ), similarly to what was previously noted in Methanomassiliicoccales 48 . These markers correspond to MTR complex subunits (m27-31), an McrA post-translational modification enzyme (m33) 45 and several uncharacterized markers that are mostly nonessential in M. maripaludis 46 ( Table 2 ). The existence of the same pattern in NM3 and NM4 supports our inference of a potential methyl-dependent hydrogenotrophic methanogenesis. Finally, Bathyarchaeota BA1 and BA2 22 which were described as methyl-dependent hydrogenotrophic methanogens 22 but possess an MCR-like complex instead of the canonical MCR ( Fig. 2 ), lack almost all methanogenesis markers ( Table 2 ), questioning their actual metabolism. Several homologues of the methanogenesis markers are also known to be present in non-methanogenic archaea. This is the case of the MCR/MCR-like (m1-3) and MTR (m27-31) complexes in archaeal methanotrophs 7 and GoM-Arc1 23 , as well as the MCR-like complex in ‘ Ca . Syntrophoarchaeales’ 21 . Based on our analysis, archaeal methanotrophs and short-chain alkane oxidizers also appear to possess numerous markers previously exclusively associated with methanogenesis ( Table 2 ), supporting the common origin and functional links of these metabolisms. In addition to the MCR/MCR-like complex subunits, the most specific and conserved markers in all lineages of methanogens, methanotrophs and short-chain alkane oxidizers appear to be the genes involved in the biosynthesis ( nflD / cfbD , murD / cfbE and possibly mcrD 49 ) and activation ( atwA and possibly mcrC 50 ) of the F 430 prosthetic group of MCR, along with six genes encoding uncharacterized proteins (m4 to m9) ( Table 2 ). These six genes are co-localized in most genomes ( Supplementary Fig. 9 ) and are among those that were predicted to be co-transcribed in Methanolobus psychrophilus R15 51 , suggesting they operate in a common process. These six marker enzymes do not co-purify with MCR 50 . However, their phylogeny ( Supplementary Fig. 10 ) and their restriction to archaea having MCR or MCR-like complexes strongly suggest they are involved in essential aspects of the regulation, folding and/or function of the respective holoenzymes ( Supplementary Discussion ). Finally, several markers are present in archaeal lineages without MCR/MCR-like complexes ( Supplementary Table 3 ) and are possibly remnants of an ancestral methane-metabolism ( Supplementary Fig. 11-13; Supplementary Discussion ). Taken together, these observations indicate that none of the previously defined methanogenesis markers are unique to methanogens but are rather more generally indicative of metabolisms involving MCR or MRC-like complexes, including methanogenesis, methanotrophy, and short-chain alkane oxidation. Elucidating the roles of these markers (MCR-Associated Markers or MAM) will be essential not only for understanding methanogenesis, but also anaerobic methanotrophy and short-chain alkane oxidation in archaea. Evolution of methane and short-chain alkane metabolisms Our results significantly extend recent data by highlighting the overwhelming presence of lineages with an MCR or MCR-like complex in the Archaea ( Fig. 1 ). This supports an early origin of methanogenesis in this domain of life, and multiple losses of this metabolism during archaeal diversification 4 , 14 , 18 , 52 . The sharing of a common set of genes ( Table 2 ) clearly indicates that methanogens, anaerobic methanotrophs and short-chain alkane oxidizers are evolutionarily linked. However, it remains unclear which type of metabolism is the most ancient, and what evolutionary and functional transitions led to such diversity 14 . The antiquity of the WL pathway 53 , 54 , and the recent proposal that the root of the archaeal tree might lie in between Class I and II methanogens 52 , would suggest that CO 2 -dependent hydrogenotrophic methanogenesis is the ancestral type of methanogenesis. Nevertheless, the growing diversity of methyl-dependent hydrogenotrophic methanogens, including this work ( Fig. 1 in red), indicates that this metabolism has been largely overlooked. Its origin and evolutionary relationship with CO 2 -dependent hydrogenotrophic methanogenesis remain unclear. The fact that it is a simpler metabolism, requiring fewer genes than CO 2 -dependent hydrogenotrophic methanogenesis might suggest its earliest origin. However, it may also signify that it could have emerged later through loss of the WL pathway and/or HGT, as suggested by the grouping of most archaea sharing this metabolism in the phylogenies of MCR ( Fig. 2A ) and of m4 to m9 markers ( Supplementary Fig. 10 ). Also, the clustering of NM4 with Verstraetearchaeota on a separate and well-supported clade in the MCR tree ( Fig. 2A ) is compatible with a possible inheritance of this metabolism from the last archaeal common ancestor, even under the classical root in between Euryarchaeota and the TACK. However, the possibility of an acquisition through ancient HGT cannot be excluded at present. More insights into the ancestral type of methanogenesis might also be gained from re-examination of the root of the archaeal tree 52 including all recently discovered archaeal lineages. The phylogenetic placement of the ANME lineages ( Fig. 1 ), strongly suggests that the capabilities for anaerobic methanotrophy emerged multiple times independently during archaeal diversification. In the Methanosarcinales this could have occurred relatively recently and repeatedly by reversal of methanogenesis, possibly through switch of function of a resident canonical MCR, leading to the different ANME-2 ( Fig. 2A ) and possibly ANME-3 lineages. The pool of genes associated with energy conservation in methanogenic and methanotrophic Methanosarcinales is in fact relatively similar 55 ( Fig. 3 and Supplementary Fig. 3 ) and some methanogenic Methanosarcinales encode c-type multiheme cytochromes 11 providing the necessary background for electron transfer in AOM archaea. The identification and experimental demonstration of the capacity for oxidation of short-chain alkanes (butane, propane) by a divergent MCR-like complex in the Synthrophoarchaeales 21 is among the most interesting findings of the recent years in the field of environmental microbiology. Our results extend the distribution of these MCR-like complexes in the archaea ( Fig. 2A ), and therefore also of potential short-chain alkane oxidation capabilities ( Figs. 3 and 4 ). The rapid evolutionary rates of MCR-like homologues coupled to the change of key residues ( Fig. 2B ) suggest that these complexes might have arisen from canonical MCRs through modifications in the catalytic site to accommodate larger hydrocarbons than methane. Transitions between anaerobic methanotrophy and short-chain alkane utilisation could have occurred in both directions as suggested by i) the close phylogenetic relationships between ‘ Ca. Methanophagales’ and ‘ Ca. Syntrophoarchaeales’ and the position of GoM-Arc1 within a clade comprising ANME-2a/ANME-2d ( Fig. 1 ), ii) the proposed mechanism of alkane activation in their MCR/MCR-like complexes 21 , iii) their very similar modes of energy conservation ( Supplementary Fig. 3 ), and iv) their numerous shared markers ( Table 2 ). If GoM-Arc1 is a short-chain alkane oxidizer, as suggested by its MCR-like complex, this capacity could have emerged from methanotrophy. Conversely, ‘ Ca . Methanophagales’ (ANME-1) might have shifted from short-chain alkane oxidation to methanotrophy after acquisition of their MCR through HGT ( Fig. 2 ). Finally, the first report of co-existence of an MCR and an MCR-like complex in members of the “ Ca. Methanoliparia” class opens up the possibility of an additional type of methanogenesis associated with alkane and/or LCFA oxidation. Further exploration of archaeal lineages with an MCR/MCR-like complex and their experimental characterization will lead to a more complete understanding of methane metabolisms and their derivations, as well as their environmental impact."
} | 6,939 |
23673639 | PMC3878185 | pmc | 774 | {
"abstract": "OP9 is a yet-uncultivated bacterial lineage found in geothermal systems, petroleum reservoirs, anaerobic digesters, and wastewater treatment facilities. Here we use single-cell and metagenome sequencing to obtain two distinct, nearly-complete OP9 genomes, one constructed from single cells sorted from hot spring sediments and the other derived from binned metagenomic contigs from an in situ-enriched cellulolytic, thermophilic community. Phylogenomic analyses support the designation of OP9 as a candidate phylum for which we propose the name ‘Atribacteria’. Although a plurality of predicted proteins is most similar to those from Firmicutes, the presence of key genes suggests a diderm cell envelope. Metabolic reconstruction from the core genome suggests an anaerobic lifestyle based on sugar fermentation by Embden-Meyerhof glycolysis with production of hydrogen, acetate, and ethanol. Putative glycohydrolases and an endoglucanase may enable catabolism of (hemi)cellulose in thermal environments. This study lays a foundation for understanding the physiology and ecological role of the ‘Atribacteria’.",
"introduction": "Introduction Over the last ~20 years, cultivation-independent approaches in microbial ecology have dramatically expanded our view of the microbial world 1 , 2 and have revealed that our ability to isolate novel organisms out of the milieu for study remains limited. While this problem is evident at all taxonomic levels, it is most glaring at the phylum level. Currently, <50% of phylum-level lineages of Bacteria and Archaea have been cultivated and studied in the laboratory 2 , 3 . The vast diversity of these uncultured groups represents an enormous genetic reservoir that has been described as ‘biological dark matter’ to call attention to our profound ignorance of these groups 4 . It can be difficult to discover even the most basic facts about the biology of candidate microbial phyla because they are often outnumbered in nature. Metagenomics and single-cell genomics are powerful cultivation-independent approaches for probing the nature of so-called ‘dark matter’ organisms by facilitating access to their genomes 5 , 6 . While metagenomics has allowed the recovery of partial or nearly-complete genomes from several candidate phyla in habitats where they are naturally abundant 7 – 9 , access to less abundant ‘dark matter’ groups has been enhanced by the advent of single-cell genomics, where individual cells are isolated by fluorescence activated cell sorting (FACS) or micromanipulation techniques including microfluidic sorting, optical trapping, and micropipetting; isolated cells are then lysed and the femtogram quantities of DNA released are amplified and sequenced 5 , 6 . Through containment of the sorting and amplification steps within nanoliter reaction volumes, microfluidic approaches ameliorate amplification bias, minimize the amplification of trace DNA contaminants in the sample, laboratory environment, and reagents, and allow for detailed observation of cell morphology 10 – 12 . Complete or partial genomes can be obtained from single cells 10 , 13 , 14 and single-cell genomics has shed light on the possible functions of several candidate phyla 4 , 15 – 18 . Here, we take advantage of the complementarity of metagenomics and single-cell genomics to assemble nearly-complete genomes of two members of candidate bacterial phylum OP9. Since its discovery in Obsidian Pool in Yellowstone National Park (YNP) 1 , OP9 has been found in other geothermal springs, petroleum reservoirs, thermal bioreactors and digesters, and wastewater sludge treatment plants 19 – 23 . This distribution demonstrates an affinity for thermal, anaerobic environments, but no genetic data pertaining to OP9 other than PCR-amplified 16S rRNA gene sequences have been reported. We compare single-cell genomic data from 15 OP9 cells isolated from hot spring sediments near Little Hot Creek, CA, (LHC) 19 with a metagenomic dataset from an in situ , ~77 °C cellulolytic enrichment in Great Boiling Spring, NV (GBS) 20 to obtain distinct draft genomes of the OP9 population from each environment. Analysis of the shared features of these two genomes offers the first insights into the cell structure and metabolic capabilities of OP9 lineages present in geothermal systems and suggests that they play a role in biomass degradation in these environments.",
"discussion": "Discussion The OP9-cSCG and OP9-77CS genomes described here offer the first significant insights into the metabolic capabilities and ecology of this ‘dark matter’ group. Our analysis of the core genome of hot spring OP9 lineages, which focus on basic cell structure and central metabolism, indicate an anaerobic, fermentative, saccharolytic lifestyle with the potential for degradation of (hemi)cellulose. This is consistent with the observed enrichment of OP9 on cellulosic biomass incubated in the hot spring GBS 25 and supports the hypothesis that OP9 plays a role in thermophilic, cellulolytic microbial consortia. It is tempting to extend this prediction to other members of the OP9 lineage, which tend to be found in thermal, anaerobic environments with large amounts of biomass. The potential for utilization of xyloglucan may carve out a niche for OP9, and could explain its low-level but consistent presence in these environments. The apparent reliance of OP9 on exogenous vitamins suggests a dependence on other organisms in the in situ enrichments, corroborated by the presence of COGs involved in de novo vitamin synthesis elsewhere in the 77CS metagenome. These results facilitate enrichment strategies targeted at cultivation of OP9 and pave the way to understanding the role of this candidate phylum. This study also illustrates the utility of combining single-cell and metagenomics approaches, even in cases where datasets originate in distinct species or environments. Based on data presented here, we propose the taxonomic epithets ‘ Candidatus Caldatribacterium californiense’ and ‘ Ca. Caldatribacterium saccharofermentans’ to refer to the OP9-cSCG and OP9-77CS phylotypes, respectively. The descriptions of the taxa are as follows: ‘Caldatribacterium’ (Cald.atri.bac.te’ri.um. L. adj. caldus, hot; L. adj. ater -tra -trum, black; L. neut. n. bacterium, rod or staff. N.L. neut. n. Caldatribacterium refers to a rod-shaped bacterium from a hot environment, where ‘black’ or ‘dark’ references both microbial ‘dark matter’ and the dark, anaerobic environments where the lineage is found), ‘saccharofermentans’ (sac.cha.ro. fer’men.tans. Gr. n. sakchâr, sugar; L. v. fermento, to ferment; N.L. part. adj. saccharofermentans, sugar-fermenting), and ‘californiense’ (ca.li.for.ni.en’se. N.L. neut. adj. californiense, of or belonging to California). 16S rRNA gene phylogenetics firmly places ‘ Ca. Caldatribacterium saccharofermentans’ and ‘ Ca. Caldatribacterium californiense’ within candidate bacterial phylum OP9 along with 16S rRNA gene phylotypes recovered from environments including geothermal systems, petroleum reservoirs, anaerobic digesters, and wastewater treatment facilities. In addition, phylogenetic analyses support the proposal for the candidate phylum ‘ Atribacteria ’ (A.tri.bac.te’ri.a. N.L. n. Atribacteria, the ‘dark’ bacterial phylum) inclusive of members of the OP9 lineage."
} | 1,824 |
23673639 | PMC3878185 | pmc | 774 | {
"abstract": "OP9 is a yet-uncultivated bacterial lineage found in geothermal systems, petroleum reservoirs, anaerobic digesters, and wastewater treatment facilities. Here we use single-cell and metagenome sequencing to obtain two distinct, nearly-complete OP9 genomes, one constructed from single cells sorted from hot spring sediments and the other derived from binned metagenomic contigs from an in situ-enriched cellulolytic, thermophilic community. Phylogenomic analyses support the designation of OP9 as a candidate phylum for which we propose the name ‘Atribacteria’. Although a plurality of predicted proteins is most similar to those from Firmicutes, the presence of key genes suggests a diderm cell envelope. Metabolic reconstruction from the core genome suggests an anaerobic lifestyle based on sugar fermentation by Embden-Meyerhof glycolysis with production of hydrogen, acetate, and ethanol. Putative glycohydrolases and an endoglucanase may enable catabolism of (hemi)cellulose in thermal environments. This study lays a foundation for understanding the physiology and ecological role of the ‘Atribacteria’.",
"introduction": "Introduction Over the last ~20 years, cultivation-independent approaches in microbial ecology have dramatically expanded our view of the microbial world 1 , 2 and have revealed that our ability to isolate novel organisms out of the milieu for study remains limited. While this problem is evident at all taxonomic levels, it is most glaring at the phylum level. Currently, <50% of phylum-level lineages of Bacteria and Archaea have been cultivated and studied in the laboratory 2 , 3 . The vast diversity of these uncultured groups represents an enormous genetic reservoir that has been described as ‘biological dark matter’ to call attention to our profound ignorance of these groups 4 . It can be difficult to discover even the most basic facts about the biology of candidate microbial phyla because they are often outnumbered in nature. Metagenomics and single-cell genomics are powerful cultivation-independent approaches for probing the nature of so-called ‘dark matter’ organisms by facilitating access to their genomes 5 , 6 . While metagenomics has allowed the recovery of partial or nearly-complete genomes from several candidate phyla in habitats where they are naturally abundant 7 – 9 , access to less abundant ‘dark matter’ groups has been enhanced by the advent of single-cell genomics, where individual cells are isolated by fluorescence activated cell sorting (FACS) or micromanipulation techniques including microfluidic sorting, optical trapping, and micropipetting; isolated cells are then lysed and the femtogram quantities of DNA released are amplified and sequenced 5 , 6 . Through containment of the sorting and amplification steps within nanoliter reaction volumes, microfluidic approaches ameliorate amplification bias, minimize the amplification of trace DNA contaminants in the sample, laboratory environment, and reagents, and allow for detailed observation of cell morphology 10 – 12 . Complete or partial genomes can be obtained from single cells 10 , 13 , 14 and single-cell genomics has shed light on the possible functions of several candidate phyla 4 , 15 – 18 . Here, we take advantage of the complementarity of metagenomics and single-cell genomics to assemble nearly-complete genomes of two members of candidate bacterial phylum OP9. Since its discovery in Obsidian Pool in Yellowstone National Park (YNP) 1 , OP9 has been found in other geothermal springs, petroleum reservoirs, thermal bioreactors and digesters, and wastewater sludge treatment plants 19 – 23 . This distribution demonstrates an affinity for thermal, anaerobic environments, but no genetic data pertaining to OP9 other than PCR-amplified 16S rRNA gene sequences have been reported. We compare single-cell genomic data from 15 OP9 cells isolated from hot spring sediments near Little Hot Creek, CA, (LHC) 19 with a metagenomic dataset from an in situ , ~77 °C cellulolytic enrichment in Great Boiling Spring, NV (GBS) 20 to obtain distinct draft genomes of the OP9 population from each environment. Analysis of the shared features of these two genomes offers the first insights into the cell structure and metabolic capabilities of OP9 lineages present in geothermal systems and suggests that they play a role in biomass degradation in these environments.",
"discussion": "Discussion The OP9-cSCG and OP9-77CS genomes described here offer the first significant insights into the metabolic capabilities and ecology of this ‘dark matter’ group. Our analysis of the core genome of hot spring OP9 lineages, which focus on basic cell structure and central metabolism, indicate an anaerobic, fermentative, saccharolytic lifestyle with the potential for degradation of (hemi)cellulose. This is consistent with the observed enrichment of OP9 on cellulosic biomass incubated in the hot spring GBS 25 and supports the hypothesis that OP9 plays a role in thermophilic, cellulolytic microbial consortia. It is tempting to extend this prediction to other members of the OP9 lineage, which tend to be found in thermal, anaerobic environments with large amounts of biomass. The potential for utilization of xyloglucan may carve out a niche for OP9, and could explain its low-level but consistent presence in these environments. The apparent reliance of OP9 on exogenous vitamins suggests a dependence on other organisms in the in situ enrichments, corroborated by the presence of COGs involved in de novo vitamin synthesis elsewhere in the 77CS metagenome. These results facilitate enrichment strategies targeted at cultivation of OP9 and pave the way to understanding the role of this candidate phylum. This study also illustrates the utility of combining single-cell and metagenomics approaches, even in cases where datasets originate in distinct species or environments. Based on data presented here, we propose the taxonomic epithets ‘ Candidatus Caldatribacterium californiense’ and ‘ Ca. Caldatribacterium saccharofermentans’ to refer to the OP9-cSCG and OP9-77CS phylotypes, respectively. The descriptions of the taxa are as follows: ‘Caldatribacterium’ (Cald.atri.bac.te’ri.um. L. adj. caldus, hot; L. adj. ater -tra -trum, black; L. neut. n. bacterium, rod or staff. N.L. neut. n. Caldatribacterium refers to a rod-shaped bacterium from a hot environment, where ‘black’ or ‘dark’ references both microbial ‘dark matter’ and the dark, anaerobic environments where the lineage is found), ‘saccharofermentans’ (sac.cha.ro. fer’men.tans. Gr. n. sakchâr, sugar; L. v. fermento, to ferment; N.L. part. adj. saccharofermentans, sugar-fermenting), and ‘californiense’ (ca.li.for.ni.en’se. N.L. neut. adj. californiense, of or belonging to California). 16S rRNA gene phylogenetics firmly places ‘ Ca. Caldatribacterium saccharofermentans’ and ‘ Ca. Caldatribacterium californiense’ within candidate bacterial phylum OP9 along with 16S rRNA gene phylotypes recovered from environments including geothermal systems, petroleum reservoirs, anaerobic digesters, and wastewater treatment facilities. In addition, phylogenetic analyses support the proposal for the candidate phylum ‘ Atribacteria ’ (A.tri.bac.te’ri.a. N.L. n. Atribacteria, the ‘dark’ bacterial phylum) inclusive of members of the OP9 lineage."
} | 1,824 |
23673639 | PMC3878185 | pmc | 775 | {
"abstract": "OP9 is a yet-uncultivated bacterial lineage found in geothermal systems, petroleum reservoirs, anaerobic digesters, and wastewater treatment facilities. Here we use single-cell and metagenome sequencing to obtain two distinct, nearly-complete OP9 genomes, one constructed from single cells sorted from hot spring sediments and the other derived from binned metagenomic contigs from an in situ-enriched cellulolytic, thermophilic community. Phylogenomic analyses support the designation of OP9 as a candidate phylum for which we propose the name ‘Atribacteria’. Although a plurality of predicted proteins is most similar to those from Firmicutes, the presence of key genes suggests a diderm cell envelope. Metabolic reconstruction from the core genome suggests an anaerobic lifestyle based on sugar fermentation by Embden-Meyerhof glycolysis with production of hydrogen, acetate, and ethanol. Putative glycohydrolases and an endoglucanase may enable catabolism of (hemi)cellulose in thermal environments. This study lays a foundation for understanding the physiology and ecological role of the ‘Atribacteria’.",
"introduction": "Introduction Over the last ~20 years, cultivation-independent approaches in microbial ecology have dramatically expanded our view of the microbial world 1 , 2 and have revealed that our ability to isolate novel organisms out of the milieu for study remains limited. While this problem is evident at all taxonomic levels, it is most glaring at the phylum level. Currently, <50% of phylum-level lineages of Bacteria and Archaea have been cultivated and studied in the laboratory 2 , 3 . The vast diversity of these uncultured groups represents an enormous genetic reservoir that has been described as ‘biological dark matter’ to call attention to our profound ignorance of these groups 4 . It can be difficult to discover even the most basic facts about the biology of candidate microbial phyla because they are often outnumbered in nature. Metagenomics and single-cell genomics are powerful cultivation-independent approaches for probing the nature of so-called ‘dark matter’ organisms by facilitating access to their genomes 5 , 6 . While metagenomics has allowed the recovery of partial or nearly-complete genomes from several candidate phyla in habitats where they are naturally abundant 7 – 9 , access to less abundant ‘dark matter’ groups has been enhanced by the advent of single-cell genomics, where individual cells are isolated by fluorescence activated cell sorting (FACS) or micromanipulation techniques including microfluidic sorting, optical trapping, and micropipetting; isolated cells are then lysed and the femtogram quantities of DNA released are amplified and sequenced 5 , 6 . Through containment of the sorting and amplification steps within nanoliter reaction volumes, microfluidic approaches ameliorate amplification bias, minimize the amplification of trace DNA contaminants in the sample, laboratory environment, and reagents, and allow for detailed observation of cell morphology 10 – 12 . Complete or partial genomes can be obtained from single cells 10 , 13 , 14 and single-cell genomics has shed light on the possible functions of several candidate phyla 4 , 15 – 18 . Here, we take advantage of the complementarity of metagenomics and single-cell genomics to assemble nearly-complete genomes of two members of candidate bacterial phylum OP9. Since its discovery in Obsidian Pool in Yellowstone National Park (YNP) 1 , OP9 has been found in other geothermal springs, petroleum reservoirs, thermal bioreactors and digesters, and wastewater sludge treatment plants 19 – 23 . This distribution demonstrates an affinity for thermal, anaerobic environments, but no genetic data pertaining to OP9 other than PCR-amplified 16S rRNA gene sequences have been reported. We compare single-cell genomic data from 15 OP9 cells isolated from hot spring sediments near Little Hot Creek, CA, (LHC) 19 with a metagenomic dataset from an in situ , ~77 °C cellulolytic enrichment in Great Boiling Spring, NV (GBS) 20 to obtain distinct draft genomes of the OP9 population from each environment. Analysis of the shared features of these two genomes offers the first insights into the cell structure and metabolic capabilities of OP9 lineages present in geothermal systems and suggests that they play a role in biomass degradation in these environments.",
"discussion": "Discussion The OP9-cSCG and OP9-77CS genomes described here offer the first significant insights into the metabolic capabilities and ecology of this ‘dark matter’ group. Our analysis of the core genome of hot spring OP9 lineages, which focus on basic cell structure and central metabolism, indicate an anaerobic, fermentative, saccharolytic lifestyle with the potential for degradation of (hemi)cellulose. This is consistent with the observed enrichment of OP9 on cellulosic biomass incubated in the hot spring GBS 25 and supports the hypothesis that OP9 plays a role in thermophilic, cellulolytic microbial consortia. It is tempting to extend this prediction to other members of the OP9 lineage, which tend to be found in thermal, anaerobic environments with large amounts of biomass. The potential for utilization of xyloglucan may carve out a niche for OP9, and could explain its low-level but consistent presence in these environments. The apparent reliance of OP9 on exogenous vitamins suggests a dependence on other organisms in the in situ enrichments, corroborated by the presence of COGs involved in de novo vitamin synthesis elsewhere in the 77CS metagenome. These results facilitate enrichment strategies targeted at cultivation of OP9 and pave the way to understanding the role of this candidate phylum. This study also illustrates the utility of combining single-cell and metagenomics approaches, even in cases where datasets originate in distinct species or environments. Based on data presented here, we propose the taxonomic epithets ‘ Candidatus Caldatribacterium californiense’ and ‘ Ca. Caldatribacterium saccharofermentans’ to refer to the OP9-cSCG and OP9-77CS phylotypes, respectively. The descriptions of the taxa are as follows: ‘Caldatribacterium’ (Cald.atri.bac.te’ri.um. L. adj. caldus, hot; L. adj. ater -tra -trum, black; L. neut. n. bacterium, rod or staff. N.L. neut. n. Caldatribacterium refers to a rod-shaped bacterium from a hot environment, where ‘black’ or ‘dark’ references both microbial ‘dark matter’ and the dark, anaerobic environments where the lineage is found), ‘saccharofermentans’ (sac.cha.ro. fer’men.tans. Gr. n. sakchâr, sugar; L. v. fermento, to ferment; N.L. part. adj. saccharofermentans, sugar-fermenting), and ‘californiense’ (ca.li.for.ni.en’se. N.L. neut. adj. californiense, of or belonging to California). 16S rRNA gene phylogenetics firmly places ‘ Ca. Caldatribacterium saccharofermentans’ and ‘ Ca. Caldatribacterium californiense’ within candidate bacterial phylum OP9 along with 16S rRNA gene phylotypes recovered from environments including geothermal systems, petroleum reservoirs, anaerobic digesters, and wastewater treatment facilities. In addition, phylogenetic analyses support the proposal for the candidate phylum ‘ Atribacteria ’ (A.tri.bac.te’ri.a. N.L. n. Atribacteria, the ‘dark’ bacterial phylum) inclusive of members of the OP9 lineage."
} | 1,824 |
23673639 | PMC3878185 | pmc | 775 | {
"abstract": "OP9 is a yet-uncultivated bacterial lineage found in geothermal systems, petroleum reservoirs, anaerobic digesters, and wastewater treatment facilities. Here we use single-cell and metagenome sequencing to obtain two distinct, nearly-complete OP9 genomes, one constructed from single cells sorted from hot spring sediments and the other derived from binned metagenomic contigs from an in situ-enriched cellulolytic, thermophilic community. Phylogenomic analyses support the designation of OP9 as a candidate phylum for which we propose the name ‘Atribacteria’. Although a plurality of predicted proteins is most similar to those from Firmicutes, the presence of key genes suggests a diderm cell envelope. Metabolic reconstruction from the core genome suggests an anaerobic lifestyle based on sugar fermentation by Embden-Meyerhof glycolysis with production of hydrogen, acetate, and ethanol. Putative glycohydrolases and an endoglucanase may enable catabolism of (hemi)cellulose in thermal environments. This study lays a foundation for understanding the physiology and ecological role of the ‘Atribacteria’.",
"introduction": "Introduction Over the last ~20 years, cultivation-independent approaches in microbial ecology have dramatically expanded our view of the microbial world 1 , 2 and have revealed that our ability to isolate novel organisms out of the milieu for study remains limited. While this problem is evident at all taxonomic levels, it is most glaring at the phylum level. Currently, <50% of phylum-level lineages of Bacteria and Archaea have been cultivated and studied in the laboratory 2 , 3 . The vast diversity of these uncultured groups represents an enormous genetic reservoir that has been described as ‘biological dark matter’ to call attention to our profound ignorance of these groups 4 . It can be difficult to discover even the most basic facts about the biology of candidate microbial phyla because they are often outnumbered in nature. Metagenomics and single-cell genomics are powerful cultivation-independent approaches for probing the nature of so-called ‘dark matter’ organisms by facilitating access to their genomes 5 , 6 . While metagenomics has allowed the recovery of partial or nearly-complete genomes from several candidate phyla in habitats where they are naturally abundant 7 – 9 , access to less abundant ‘dark matter’ groups has been enhanced by the advent of single-cell genomics, where individual cells are isolated by fluorescence activated cell sorting (FACS) or micromanipulation techniques including microfluidic sorting, optical trapping, and micropipetting; isolated cells are then lysed and the femtogram quantities of DNA released are amplified and sequenced 5 , 6 . Through containment of the sorting and amplification steps within nanoliter reaction volumes, microfluidic approaches ameliorate amplification bias, minimize the amplification of trace DNA contaminants in the sample, laboratory environment, and reagents, and allow for detailed observation of cell morphology 10 – 12 . Complete or partial genomes can be obtained from single cells 10 , 13 , 14 and single-cell genomics has shed light on the possible functions of several candidate phyla 4 , 15 – 18 . Here, we take advantage of the complementarity of metagenomics and single-cell genomics to assemble nearly-complete genomes of two members of candidate bacterial phylum OP9. Since its discovery in Obsidian Pool in Yellowstone National Park (YNP) 1 , OP9 has been found in other geothermal springs, petroleum reservoirs, thermal bioreactors and digesters, and wastewater sludge treatment plants 19 – 23 . This distribution demonstrates an affinity for thermal, anaerobic environments, but no genetic data pertaining to OP9 other than PCR-amplified 16S rRNA gene sequences have been reported. We compare single-cell genomic data from 15 OP9 cells isolated from hot spring sediments near Little Hot Creek, CA, (LHC) 19 with a metagenomic dataset from an in situ , ~77 °C cellulolytic enrichment in Great Boiling Spring, NV (GBS) 20 to obtain distinct draft genomes of the OP9 population from each environment. Analysis of the shared features of these two genomes offers the first insights into the cell structure and metabolic capabilities of OP9 lineages present in geothermal systems and suggests that they play a role in biomass degradation in these environments.",
"discussion": "Discussion The OP9-cSCG and OP9-77CS genomes described here offer the first significant insights into the metabolic capabilities and ecology of this ‘dark matter’ group. Our analysis of the core genome of hot spring OP9 lineages, which focus on basic cell structure and central metabolism, indicate an anaerobic, fermentative, saccharolytic lifestyle with the potential for degradation of (hemi)cellulose. This is consistent with the observed enrichment of OP9 on cellulosic biomass incubated in the hot spring GBS 25 and supports the hypothesis that OP9 plays a role in thermophilic, cellulolytic microbial consortia. It is tempting to extend this prediction to other members of the OP9 lineage, which tend to be found in thermal, anaerobic environments with large amounts of biomass. The potential for utilization of xyloglucan may carve out a niche for OP9, and could explain its low-level but consistent presence in these environments. The apparent reliance of OP9 on exogenous vitamins suggests a dependence on other organisms in the in situ enrichments, corroborated by the presence of COGs involved in de novo vitamin synthesis elsewhere in the 77CS metagenome. These results facilitate enrichment strategies targeted at cultivation of OP9 and pave the way to understanding the role of this candidate phylum. This study also illustrates the utility of combining single-cell and metagenomics approaches, even in cases where datasets originate in distinct species or environments. Based on data presented here, we propose the taxonomic epithets ‘ Candidatus Caldatribacterium californiense’ and ‘ Ca. Caldatribacterium saccharofermentans’ to refer to the OP9-cSCG and OP9-77CS phylotypes, respectively. The descriptions of the taxa are as follows: ‘Caldatribacterium’ (Cald.atri.bac.te’ri.um. L. adj. caldus, hot; L. adj. ater -tra -trum, black; L. neut. n. bacterium, rod or staff. N.L. neut. n. Caldatribacterium refers to a rod-shaped bacterium from a hot environment, where ‘black’ or ‘dark’ references both microbial ‘dark matter’ and the dark, anaerobic environments where the lineage is found), ‘saccharofermentans’ (sac.cha.ro. fer’men.tans. Gr. n. sakchâr, sugar; L. v. fermento, to ferment; N.L. part. adj. saccharofermentans, sugar-fermenting), and ‘californiense’ (ca.li.for.ni.en’se. N.L. neut. adj. californiense, of or belonging to California). 16S rRNA gene phylogenetics firmly places ‘ Ca. Caldatribacterium saccharofermentans’ and ‘ Ca. Caldatribacterium californiense’ within candidate bacterial phylum OP9 along with 16S rRNA gene phylotypes recovered from environments including geothermal systems, petroleum reservoirs, anaerobic digesters, and wastewater treatment facilities. In addition, phylogenetic analyses support the proposal for the candidate phylum ‘ Atribacteria ’ (A.tri.bac.te’ri.a. N.L. n. Atribacteria, the ‘dark’ bacterial phylum) inclusive of members of the OP9 lineage."
} | 1,824 |
29500185 | PMC5866589 | pmc | 776 | {
"abstract": "Significance Lignin valorization is critical for economic viability of future biorefineries but is hindered due to the challenges of engineered bio-chassis such as the slow kinetics of substrate uptake, aromatics toxicity, and cost. Here, an autoregulatory system involving a vanillin autoinducible promoter is demonstrated with an aromatics transporter in Escherichia coli that is induced by lignin-derived aromatics and simultaneously converted to value-added platform chemical with diverse applications. In addition to alleviating toxicity, the engineered E. coli strain eliminates the need for an external inducer such as isopropyl β- d -1-thiogalactopyranoside during fermentation, thereby significantly reducing the process cost. This study demonstrates an autoregulatory system for aromatics bioconversion and serves as a platform for future strain development for lignin valorization.",
"discussion": "Discussion The drivers behind the urgent need to develop economically viable, sustainable, biorefining technologies for renewable fuels and chemicals production include energy security, economic development, and environmental concerns. Lignin, the second-most-abundant plant polymer on Earth after cellulose, is underutilized by its direct combustion for energy ( 4 ). Moreover, lignin is the only renewable feedstock that is composed of aromatic building blocks ( 4 ). Because of its vast supply and aromatic-rich content, lignin has great potential to be valorized to produce value-added chemicals. Being phenolic in nature, the depolymerization of lignin results in a multitude of aromatic compounds such as vanillin, syringaldehyde, ferulic acid, guaiacol, phenol, syringol, allyl guaiacol, and so on ( 5 , 30 – 32 ). Developing microbial factories for the efficient bioconversion of these lignin-derived aromatics is of crucial importance for economically viable lignin upgrading strategies and enabling a bio-based economy and sustainability. Catechol is one of the valuable platform chemicals that can potentially be produced from lignin-derived aromatics. Catechol and its further upgraded products such as muconic acid have been produced in engineered yeast from glucose through the shikimate pathway ( 22 , 33 ). In a recent study, a bacterial coculture incorporating two engineered E. coli strains has been developed for the production of muconic acid from glucose and xylose mixture ( 34 ). Wang et al. ( 35 ) also reported the production of catechol from the aromatic compound benzoate by a benzoate-utilizing Pseudomonas strain. Until recently, only limited work has been done using lignin-derived aromatics as the substrates for the production of muconic acid and adipic acid. Vardon et al. ( 7 ) engineered Pseudomonas putida KT2440, a native muconic acid producer, to convert lignin-derived monomers such as coniferyl acohol, ferulate, vanillin, and p -coumarate to catechol and muconic acid through both the catechol and protocatechuate branches of the β-ketoadipate pathway. In their study, P. putida KT2440 was engineered to replace the PcaHG gene encoding a protocatechuate 3,4 dioxygenase with AroY, which diverted the metabolism of protocatechuate to the synthesis of catechol. The CatBC gene encoding dioxygenases was deleted to eliminate further metabolism of muconic acid and a phenol monooxygenase gene was inserted to enable the conversion of phenol to catechol. In this study, through the expression of a catechol biosynthesis pathway under a vanillin-inducible promoter, ADH7, we have successfully demonstrated an autoregulatory microbial cell factory for the valorization of lignin-derived aromatics using vanillin as an example into valuable chemicals such as vanillic acid and catechol that further serve as platform chemicals for the production of a variety of high-value chemicals and polymer precursors. Furthermore, the coexpression of an associated aromatics transporter CouP with catechol biosynthesis pathway under pTrc or ADH7 promoter improved the catechol yield by 30% and 40%, respectively ( Fig. 5 C ). This self-inducible system eliminates the need for costly IPTG inducer since the substrate vanillin serves as the inducer as well. Another important potential application of our autoregulatory system is that in addition to the ability to convert vanillin to value-added products, it can also serve to detoxify the aromatic compounds when they are copresent with other microbially fermentable substrates such as biomass-derived sugars. For example, although thermochemical depolymerization is an effective route to valorize lignin ( 1 , 5 , 30 ), these lignin-derived aromatics, such as vanillin ( 36 , 37 ), are reported as potent fermentation inhibitors in the microbial production of fuels and other value-added chemicals ( 14 – 17 ). Bioconversion of these heterogeneous organic molecules to simplified product streams with lesser or no toxicity can potentially benefit the subsequent bioconversion of other substrates of interest. In our system, the product of our autoregulatory system is catechol, which has fewer inhibitory effects on the growth of E. coli than vanillin ( Fig. 2 ) and is an intermediate compound that can be further converted to other value-added compounds, such as muconic acid and adipic acid ( 7 , 8 ). Moreover, the resulting microbial cell factory is a promising platform whose function is inducible by lignin-derived aromatics, and it offers a promising path forward for the production of fuels and chemicals from lignocellulosic biomass. For example, by integrating this autoregulatory system with the biosynthesis pathway from sugars, one-pot approaches could be developed to coprocess and convert pretreated lignocellulosic liquor composed of sugars and lignin-derived aromatics into value-added chemicals. This envisioned one-pot approach minimizes separation challenges before fermentation with the additional benefit of reducing the toxicity of the vanillin to the sugar-conversion strain. This study is a demonstration of developing an autoregulatory system for the bioconversion of lignin-derived aromatics. Lignin valorization is being explored in many research laboratories across the globe to improve the economics of biofuel production from cellulosic sugars. To provide a sense of yields, however, compared with the work on catechol or muconic acid production from sugar as a substrate, the catechol titer produced from lignin-derived vanillin in this system is still low. We observed substantial accumulation of the intermediate vanillic acid during the conversion by our engineered E. coli strain ( Figs. 4 and 5 ), which is in agreement with the study in P. putida ( 7 ) and is likely due to the transcriptional or translational regulation of the enzymes in the pathways. To further increase titer, the currently introduced pathway in E. coli needs to be optimized to minimize vanillic acid accumulation and maximize the catechol production. In addition, the strength of the vanillin-inducible promoter ADH7 could be enhanced by further strain engineering. A detailed screening of engineered promoter with varying strength could lead to a tunable system for promoter strength and maximization of protein expression according to variables present in the fermentation liquor. Further metabolic engineering such as removing any bottleneck of the pathway as well as fermentation conditions optimization will likely lead to enhancement in the productivity of this engineered strain with an autoregulatory switch. In conclusion, we believe that the autoregulatory system we engineered in this study will aid in developing a low-cost and reliable platform for the valorization of lignin-derived aromatics and provide fertile ground for future studies."
} | 1,940 |
22347151 | PMC3270576 | pmc | 779 | {
"abstract": "We demonstrate bistable attractor dynamics in a spiking neural network implemented with neuromorphic VLSI hardware. The on-chip network consists of three interacting populations (two excitatory, one inhibitory) of leaky integrate-and-fire (LIF) neurons. One excitatory population is distinguished by strong synaptic self-excitation, which sustains meta-stable states of “high” and “low”-firing activity. Depending on the overall excitability, transitions to the “high” state may be evoked by external stimulation, or may occur spontaneously due to random activity fluctuations. In the former case, the “high” state retains a “working memory” of a stimulus until well after its release. In the latter case, “high” states remain stable for seconds, three orders of magnitude longer than the largest time-scale implemented in the circuitry. Evoked and spontaneous transitions form a continuum and may exhibit a wide range of latencies, depending on the strength of external stimulation and of recurrent synaptic excitation. In addition, we investigated “corrupted” “high” states comprising neurons of both excitatory populations. Within a “basin of attraction,” the network dynamics “corrects” such states and re-establishes the prototypical “high” state. We conclude that, with effective theoretical guidance, full-fledged attractor dynamics can be realized with comparatively small populations of neuromorphic hardware neurons.",
"conclusion": "Conclusion We demonstrate, with a network of leaky integrate-and-fire neurons realized in neuromorphic VLSI technology, that two distinct meta-stable states of asynchronous activity constitute attractors of the collective dynamics. We describe how the dynamics of these meta-stable states – an unselective state of low activity and a selective state of high activity – can be shaped to render transitions either quasi-deterministic or stochastic, and how the characteristic time-scale of such transitions can be tuned far beyond the time-scale of single-neuron dynamics. This constitutes an important step toward the flexible and robust classification of natural stimuli with neuromorphic systems.",
"introduction": "Introduction Neuromorphic VLSI copies in silicon the equivalent circuits of biological neurons and synapses (Mead, 1989 ). The aim is to emulate as closely as possible the computations performed by living neural tissues exploiting the analog characteristics of the silicon substrate. Neuromorphic designs seem a reasonable option to build biomimetic devices that could be directly interfaced to the natural nervous tissues. Here we focus on neuromorphic recurrent neural networks exhibiting reverberating activity states (“attractor states”) due to massive feedback. Our main motivation is the belief that neural activity in mammalian cortex is characterized by “attractor states” at multiple spatial and temporal scales (Grinvald et al., 2003 ; Shu et al., 2003 ; Holcman and Tsodyks, 2006 ; Fox and Raichle, 2007 ; Ringach, 2009 ) and that “attractor dynamics” is a key principle of numerous cognitive functions, including working memory (Amit and Brunel, 1997 ; Del Giudice et al., 2003 ; Mongillo et al., 2003 ), attentional selection (Deco and Rolls, 2005 ), sensory inference (Gigante et al., 2009 ; Braun and Mattia, 2010 ), choice behavior (Wang, 2002 ; Wong et al., 2007 ; Furman and Wang, 2008 ; Marti et al., 2008 ), motor planning (Lukashin et al., 1996 ; Mattia et al., 2010 ), and others. To date, surprisingly few studies have sought to tap the computational potential of attractor dynamics for neuromorphic devices (Camilleri et al., 2010 ; Neftci et al., 2010 ; Massoud and Horiuchi, 2011 ). The more elementary forms of attractor dynamics in networks of biologically realistic (spiking) neurons and synapses are theoretically well understood (Amit, 1989 , 1995 ; Fusi and Mattia, 1999 ; Renart et al., 2004 ). This is particularly true for bistable attractor dynamics with two distinct steady-states (“point attractors”). Attractor networks can store and retrieve prescribed patterns of collective activation as “memories.” They operate as an “associative memory” which retrieves a prototypical “memorized” state in response to an external stimulus, provided the external perturbation does not push the state outside the “basin of attraction.” The “attractor state” is self-correcting and self-sustaining even in the absence of external stimulation, thus preserving a “working memory” of past sensory events. Besides being evoked by stimulation, transitions between attractor states may also occur spontaneously, driven by intrinsic activity fluctuations. Spontaneous activity fluctuations ensure that the energetically accessible parts of state space are exhaustively explored. Thanks to this theoretical understanding, the system designer has considerable latitude in quantitatively shaping bistable attractor dynamics. By sculpting the effective energy landscape and by adjusting the amount of noise (spontaneous activity fluctuations), he can control how the network explores its phase space and how it responds to external stimulation (Mattia and Del Giudice, 2004 ; Marti et al., 2008 ). The kinetics of network dynamics, including response latencies to external stimulation, can be adjusted over several orders of magnitude over and above the intrinsic time-scale of neurons and synapses (Braun and Mattia, 2010 ). Below, we demonstrate step-by-step how to build a small neuromorphic network with a desired dynamics. We show that the characteristics of bistable attractor dynamics (“asynchronous activity,” “point attractors,” “working memory,” “basin of attraction”) are robust, in spite of the small network size and the considerable inhomogeneity of neuromorphic components. Moreover, we demonstrate tunable response kinetics up to three orders of magnitude slower (1 s vs. 1 ms) than the time-constants that are expressly implemented in the neuromorphic circuits. Some of these results have been presented in a preliminary form (Camilleri et al., 2010 ).",
"discussion": "Discussion We show how bistable attractor dynamics can be realized in silicon with a small network of spiking neurons in neuromorphic VLSI hardware. Step-by-step, we describe how various emergent behaviors can be “designed into” the collective activity dynamics. The demonstrated emergent properties include: asynchronous irregular activity, distinct steady-states (with “low” and “high” activity) in a sub-population of neurons, evoked state transitions that retain transient external input (“working memory”), self-correction of corrupted activity states (“basin of attraction”), tunable latency of evoked transitions, spontaneous state transitions driven by internal activity fluctuations, tunable rate of spontaneous transitions. Standard theoretical techniques predict the single-neuron response function , which in turn determines the equilibrium states of the collective dynamics under the “mean-field” approximation (see Section Mean-field theory). In the case of a multi-population network, an Effective Response Function (ERF) for one or more populations of interest can be extracted with the help of further approximations. The ERF provides the central hinge between network architecture and various aspects of the collective dynamics. It predicts quantitatively the number and location of steady-states in the activity of selected populations, both for spontaneous and for input-driven activity regimes. In addition, within the scope of the relevant approximations, the ERF describes qualitatively the energy landscape that governs the activity dynamics. Accordingly, it also gives an indication about kinetic characteristics such as transition latencies. Establishing a proper correspondence between theoretical parameters and their empirical counterparts in an analog, neuromorphic chip is fraught with difficulties and uncertainties. For this reason, we do not stretch theory to the point of directly predicting the network’s behavior in silicon (in contrast to the route taken by Neftci et al., 2011 ). Instead, we implement the theoretical construction of the ERF in hardware and establish this important function empirically. The ERF so obtained encapsulates all relevant details of the physical network, including effects due to mismatch, violation of the diffusion limit, etc. Thus, we rely on an effective description of the physical network, not on a tenuous correspondence to an idealized network. Equipped with these tools (mean-field theory, characterization scripts, empirical ERF), our neuromorphic hardware becomes an easily controllable and reliable system on which we show how the concepts of mean-field theory may be used to shape various aspects of the network’s collective dynamics. Wider objectives The computational possibilities of neural activity dynamics are gradually becoming better understood. Our wider objective is to translate neuroscientific advances in this area to neuromorphic hardware platforms. In doing so, we hope to build step-by-step the technological and theoretical foundations for biomimetic hardware devices that, in the fullness of time, could be integrated seamlessly with natural nervous tissues. Reverberating states of neocortical activity, also called “attractor states,” are thought to underlie various cognitive processes and functions. These include working memory (Amit and Brunel, 1997 ; Del Giudice et al., 2003 ; Mongillo et al., 2003 ), recall of long-term memory (Hopfield, 1982 ; Amit, 1995 ; Hasselmo and McClelland, 1999 ; Wang, 2008 ), attentional selection (Deco and Rolls, 2005 ), rule-based choice behavior (Fusi et al., 2007 ; Vasilaki et al., 2009 ), sensory integration in decision making (Wang, 2002 ; Wong et al., 2007 ; Furman and Wang, 2008 ; Marti et al., 2008 ; Braun and Mattia, 2010 ), and working memory in combination with delayed sensory decision making (Laing and Chow, 2002 ; Machens et al., 2005 ), among others. Dynamical representations involving attractor states are not restricted to the “point attractors” we have considered here (Destexhe and Contreras, 2006 ; Durstewitz and Deco, 2008 ). For example, there is evidence to suggest that “line attractors” may underlie some forms of working memory and path integration (Machens et al., 2005 ; Trappenberg, 2005 ; Chow et al., 2009 ). Chaotic attractors have long been proposed to subserve perceptual classification in certain sensory functions (Skarda and Freeman, 1987 ). More generally, both spontaneous and evoked activity in mammalian cortex may well be characterized by “attractor hopping” at multiple spatial and temporal scales (Grinvald et al., 2003 ; Shu et al., 2003 ; Fox and Raichle, 2007 ; Durstewitz and Deco, 2008 ; Ringach, 2009 ). Thus, the stochastic dynamics of a multi-attractor system offer both a comparatively stereotyped, low-dimensional representation of high-dimensional inputs, and a statistical distribution of possible responses. This motivates the emphasis that we have placed on the stochastic aspects of the collective dynamics of our hardware network. The classification of sensory events at multiple spatial and temporal scales might require “nested attractor” dynamics in a neuromorphic VLSI device. In a “nested” scenario, reverberating activity patterns spanning multiple spatial and temporal scales are generated by many individually bistable attractor modules interacting in a hierarchical network architecture (Gigante et al., 2009 ; Braun and Mattia, 2010 ). The energy landscape of a “nested” system would be considerably more complex than the one described here (Braun and Mattia, 2010 ). It should be imagined with multiple high-dimensional valleys within valleys, ridges, and saddles permitting state transitions. To match the sensory time-scales of interest, the dynamics of such a system could be tuned in much the same way as the simplistic attractor system of the present work (i.e., by adjusting ERFs and noise levels). Yet another challenging perspective is to build attractor representations in an adaptive manner, by means of activity-driven plasticity. Even at the level of theory, surprisingly few studies have addressed this important issue (Amit and Mongillo, 2003 ; Del Giudice et al., 2003 ). The 16,384 synapses of the FLANN chip exhibit a bistable, spike-driven plasticity (Fusi et al., 2000 ) that, in principle, would be well suited for this purpose (Del Giudice et al., 2003 ). Although the present study did not make use of this feature, we consider it imperative to face this challenge with neuromorphic hardware and have taken some initial steps in this direction (Corradi, 2011 ). State of the field Neuromorphic engineering is a broad and active field seeking to emulate natural neural processes with CMOS hardware technology for robotic, computational, and/or medical applications. Recently, two groups have implemented “continuous attractor” dynamics in neuromorphic VLSI (Neftci et al., 2010 ; Massoud and Horiuchi, 2011 ). The two networks in question (comprising 32 and 124 neurons, respectively) realized a continuous-valued memory of past sensory input by means of excitatory-inhibitory interactions between nearest neighbors. The resulting winner-take-all dynamics permitted the authors to represent and update a sensory state with incremental input (Trappenberg, 2005 ). The hardware used by these groups is comparable to ours in that it combines fixed synapses with the neuronal circuit of Indiveri et al. ( 2006 ). The main difference to our study concerns the handling of noise and mismatch. To minimize drift in the continuous attractor dynamics (Massoud and Horiuchi 2011 ), suppress finite-size noise with a synchronous and regular firing regime, while authors of Neftci et al. ( 2010 ) propose an initial precise calibration phase to reduce the mismatch that greatly affects the performance of their system. In contrast, we take advantage of both mismatch and finite-size noise to create a stochastic dynamics. As we have shown, the time-scale of this dynamics can be finely controlled by setting the balance between deterministic forces (energy landscape) and stochastic factors (finite-size noise). To our knowledge, there have been no further demonstrations of self-sustained activity and working memory with neuromorphic VLSI hardware. Other neuromorphic applications concern biomimetic sensors such as “silicon cochleas” (Chan et al., 2007 ; Hamilton et al., 2008 ; Wen and Boahen, 2009 ) or “silicon retinas” (Boahen, 2005 ; Zaghloul and Boahen, 2006 ; Lichtsteiner et al., 2008 ; Kim et al., 2009 ; Liu and Delbruck, 2010 ), implementations of linear filter banks (Serrano-Gotarredona et al., 2006 ), receptive field formation (Choi et al., 2005 ; Bamford et al., 2010 ), echo-localization (Shi and Horiuchi, 2007 ; Chan et al., 2010 ), or selective attention (Indiveri, 2008 ; Serrano-Gotarredona et al., 2009 ). Scaling up As mentioned, our wider objectives include spiking neural networks that operate in real-time and that can be interfaced with living neural tissues. At present, it is not evident which technological path will lead to the network sizes and architectures that will eventually be required for interesting computational capabilities. However, neuromorphic VLSI is a plausible candidate technology that offers considerable scope for further improvement in terms of circuits, layout, autonomy, and silicon area. Multi-chip architectures with a few thousand spiking neurons and plastic synapses may come within reach in the near future (Federici, 2011 ). Such networks could accommodate multiple attractor representations and complex energy landscapes. We note here in passing that moving to larger networks would imply softer constraints on the choice of the synaptic connectivity (see Methods ), thereby allowing more biologically plausible firing rates for the higher meta-stable states. Several consortia are building special-purpose platforms that in principle could able to host large, attractor-based networks. These include the neuromorphic Neurogrid (Boahen, 2007 ) system, which aims to simulate up to one million neurons in real-time, the BrainScaleS project (Meier, 2011 ), which relies on wafer scale technology and promises 160,000 neurons with 40 million plastic synapses operating several thousand times faster than natural networks. In addition, the SpiNNaker project (Furber and Brown, 2009 ) proposes a fully digital, ARM-based simulation of approximately 20,000 Izhikevich neurons and spike-time-dependent synapses and the EU SCANDLE project, which uses a single, off-the-shelf FPGA to accommodate one million neurons (Cassidy et al., 2011 ). Finally, a fully digital VLSI chip has recently been presented by the DARPA-funded SyNAPSE project. Designed with 45 nm technology, it comprises 256 neurons and 65,000 plastic synapses (Merolla et al., 2011 ; Seo et al., 2011 ). Of course, a fully digital implementation would quietly abandon the original vision of a “synthesis of form and function” in neuromorphic devices (Mahowald, 1992 ). Nevertheless, in view of the rapid progress in digital tools and fabrication processes, this may well be the most appropriate route for most applications. However, for applications requiring an implantable device operating in real-time, a mixed-signal approach founded on analog CMOS circuits seems likely to remain a viable alternative."
} | 4,363 |
30453032 | null | s2 | 780 | {
"abstract": "Human functional Magnetic Resonance Imaging (fMRI) data are acquired while participants engage in diverse perceptual, motor, cognitive, and emotional tasks. Although data are acquired temporally, they are most often treated in a quasi-static manner. Yet, a fuller understanding of the mechanisms that support mental functions necessitates the characterization of dynamic properties. Here, we describe an approach employing a class of recurrent neural networks called reservoir computing, and show the feasibility and potential of using it for the analysis of temporal properties of brain data. We show that reservoirs can be used effectively both for condition classification and for characterizing lower-dimensional \"trajectories\" of temporal data. Classification accuracy was approximately 90% for short clips of \"social interactions\" and around 70% for clips extracted from movie segments. Data representations with 12 or fewer dimensions (from an original space with over 300) attained classification accuracy within 5% of the full data. We hypothesize that such low-dimensional trajectories may provide \"signatures\" that can be associated with tasks and/or mental states. The approach was applied across participants (that is, training in one set of participants, and testing in a separate group), showing that representations generalized well to unseen participants. Taken together, we believe the present approach provides a promising framework to characterize dynamic fMRI information during both tasks and naturalistic conditions."
} | 384 |
24939247 | PMC4061545 | pmc | 781 | {
"abstract": "This report discusses the electrical characteristics of two-terminal synaptic memory devices capable of demonstrating an analog change in conductance in response to the varying amplitude and pulse-width of the applied signal. The devices are based on Mn doped HfO 2 material. The mechanism behind reconfiguration was studied and a unified model is presented to explain the underlying device physics. The model was then utilized to show the application of these devices in speech recognition. A comparison between a 20 nm × 20 nm sized synaptic memory device with that of a state-of-the-art VLSI SRAM synapse showed ~10× reduction in area and >10 6 times reduction in the power consumption per learning cycle.",
"discussion": "Discussion From the results in the previous section, it was clear that the hysteresis in I-Vs is caused by the increase of n 0 during positive sweep and decrease in n 0 during negative sweep. Therefore, in order to model the hysteresis, it was necessary to obtain the transient current increase and decrease as a function of applied bias. During a voltage sweep, each bias point is applied to the device for some time before a measurement is done. This stress during each bias point increases/decreases the conductance of the device and hence the current increases/decreases depending on the polarity of the bias. CVS was applied to the device in increasing amplitudes of bias and the SILC increase and the current decrease parameters were extracted as a function of bias. Figure 5(a) shows positive CVS on a device with increasing voltages ranging from 1.25 V to 2.5 V. No significant change of current was observed below 1.75 V constant stress which indicates that the activation of SILC and hence hysteresis requires a minimum electric field. The parameters for SILC were extracted by fitting the I-t curves with equation (2) and their values are provided in the inset of Figure 5(a) . Similarly, the device was stressed using negative bias ranging from −1 V to −1.75 V. Figure 5(b) shows the decrease of current during negative stressing. The parameters were again extracted from fits of the decrease to equation (3) and are presented in the inset. It is observed that for negative bias, the parameters for the stress induced reduction in current have a weak dependency on the applied bias. From the extracted parameters, the change in current during stressing can be estimated as a function of applied bias. During a positive sweep, the excess current generated due to SILC can be obtained by incorporating the field dependency of the extracted parameters in equation (2) . Likewise, the reduction in current during negative sweep can be obtained by incorporating the parameters extracted in Figure 5(b) into equation (3) . Figure 6(a) and 6(b) shows this increase and decrease in current respectively as a function of bias. The time-stamp for each bias point is varied from 20 ms, 50 ms, 100 ms and 200 ms. It is evident that for larger time stamps or in effect slower sweep rates, the change in current due to stress is higher. This field dependent change in current could be added to or subtracted from the F-P equation to model the I–V hysteresis in positive or negative bias. However, using the Φ B from equation (1) the current was under-estimated for the positive bias while the shape of the curve did not fit well. Therefore, it was apparent that along with the increase/decrease in the density of traps in the dielectric, the Φ B was also changing during voltage sweeps. In fact it was observed that the Φ B decreases during positive voltage sweep. This can be explained by assuming that the traps generated during positive voltage sweeps occupy a higher energy in the dielectric than the native traps, thus lowering the average Φ B . In an analogous manner, during negative sweeps, the Φ B would increase as the extra defects generated get annihilated. The hypothesis was confirmed by obtaining the relation of Φ B with E-field. To obtain the variation of Φ B with E-field, the following procedure was applied. A time-stamp of 100 ms was used for each bias point. For positive hysteresis, the current due to SILC was obtained for voltages equal to and above 1.75 V. This current was subtracted from the experimental hysteresis I–V to obtain the unstressed current level of the device, when there is no trap generation. The unstressed current is then used to extract Φ B as a function of E-field using the F-P equation (1) . Similarly, for the negative hysteresis, the unstressed device current refers to the condition when there is no decrease of current due to negative stress. Therefore, the unstressed current was the sum of the experimental current and the current decrease due to stressing. For positive bias, as explained above, Φ B was found to decrease with E-field. The best fit relation was found out as: where α and β are constants and κ is the power for the E-field dependency. Similarly, as explained earlier, for the negative bias Φ B was found to increase with E-field, the relation given as: The fittings for the extracted Φ B are shown in the insets of figure 6(c) and 6(d) for positive and negative sweeps, respectively. Based on this extracted trap depth, henceforth referred to as Φ B (E), the carrier density generated or annihilated during positive or negative stress could be extracted as a function of field. Therefore the overall F-P equation needs to be modified to reflect the variations in n 0 and Φ B (E) as: for positive bias and: for negative bias. Here n 0 denotes the carrier concentration of an unstressed device, while n 0 ' refers to some carrier concentration after the device was positively stressed. Δn + and Δn − are the changes in n 0 and n 0 ' respectively due to stressing, and are functions of both E-field and stress time. Based on these equations, the hysteretic behaviour of the device was modelled as shown in figures 6(c) and 6(d) for positive and negative sweeps respectively. The overlap between subsequent loops was modelled using the relaxation effect between applications of voltage sweeps. Equation (3) was used to include a slight decay of carrier concentration when no bias was applied to the device. The kink in the negative hysteresis is due to the combined effect of increase in Φ B (E) and reduction of current. Such a kink is also observed in the experimental data as shown in Figure 1(b) . Hence, a unified model was obtained to explain the synaptic behaviour of the Mn:HfO 2 synaptic devices. To examine the repeatability in the reconfiguration of these devices, an endurance testing on the device was performed as shown in Figure 7 . A potentiating pulse of 2.5 V and the given pulse width was applied, followed by measurement of the device conductance. A depressing pulse of the same width was then applied and the conductance was again measured at 0.5 V. Clearly, a repeatable reconfiguration in conductance as a function of pulse width is evident for multiple cycles without any obvious signs of failure. STDP and Speech Recognition STDP is a biologically inspired learning algorithm that is typically followed in unsupervised neuromorphic learning 19 . In a neural-synapse aggregate, pre-synaptic action potentials (AP) are incident on the synapses that are connected to the dendrite. The dendrite sums the contribution of synaptic weights to the incoming APs and fires a post-synaptic AP once the membrane reaches a certain threshold potential. Based on the relative timing of pre- and post-synaptic AP (Δt), the corresponding synapse is either potentiated or depressed 20 . In biological systems, STDP usually occurs in the spike timing window of ±40 ms with the highest change in synaptic plasticity occurring in the ±10 ms range. Therefore, when a pre-synaptic AP arrives before postsynaptic depolarization (positive Δt), long-term potentiation (LTP) or the enhancement of synaptic strength occurs, whereas if the postsynaptic firing precedes the pre-synaptic arrival of AP (negative Δt), long-term depression (LTD) or weakening of synaptic strength occurs 31 . The synaptic strength due to STDP in biological systems is usually fitted to an exponentially decaying function 21 : for LTP and for LTD. The plot for STDP using these equations is shown in Figure 8(a) . To demonstrate the possibility of implementing STDP using the proposed synaptic devices, a 2.5 V pulse for potentiation and −2 V pulse for depression were applied while the pulse-width (ω) was modulated based on Δt. When a pre-synaptic spike preceded the post-synaptic firing, a potentiating pulse would be applied, while a depressing pulse would be applied when post-synaptic firing precedes the pre-synaptic AP. Once a neuron fired, the relative timing of spike arrival (Δt) was recorded directly into the applied pulse width and polarity by the following mapping procedure. To be compatible with biological systems, a spike timing window of Δt = ±40 ms was chosen, where the highest change of conductance was intended when Δt = ±10 ms. Since these devices needed much longer ω to show appreciable changes in conductance, a relation between Δt and ω was defined such that a Δt = ±10 ms corresponded to ω = 200 ms, Δt = ±20 ms corresponded to ω = 100 ms, Δt = ±30 ms corresponded to ω = 50 ms, and finally Δt = ±40 ms corresponded to ω = 20 ms. This relation between Δt and ω can be conveniently represented by the equation (11) . where λ and κ are fitting parameters. This simple test to demonstrate STDP is schematically shown in figure 8(b) . The observed potentiation (LTP) and depression (LTD) characteristic of the device, measured using this technique is shown in figure 8(c) . The LTP was calculated by measuring the change in the conductance of the device in response to the applied potentiating pulse. LTD was calculated by measuring the change in the conductance of the potentiated device in response to the applied depressing pulse. Therefore, the timing information is stored in the device as a change of conductance. Next, to model the LTP and LTD behaviour of the device, it was necessary to obtain the change of n 0 (Δn) when potentiating or depressing pulses were applied. Figure 8(d) shows Δn of the device as a function of pulse width for potentiating (2.5 V) and depressing (−2 V) pulse. The increase or decrease of current due to the applied pulse width was evaluated first using equations (2) and (3). Δn was then extracted from that current using an average Φ B (E) = 0.19 eV to account for bias dependent change in the trap depth. Once Δn as a function of applied pulse width was obtained, it was expressed as a percentage change to get the theoretical LTP and LTD values. The pulse width was mapped to Δt using equation (11) . This data is plotted in Fig. 8(c) . To demonstrate the application of these devices in neuromorphic learning, a 16 × 16 crossbar array of Mn:HfO 2 synaptic memory devices was simulated in a neuron-synapse framework to demonstrate STDP algorithm based speech recording. The schematic is shown in Fig. 9(a) . A voice was sampled at 11.025 KHz and recorded for duration of 1 s. To emulate the filtering process of the cochlea, the data was passed through a band-pass filter bank. In this simulation, a central frequency (f c ) ranging from 200 Hz to 4 KHz was used to include most of the audible human range. The frequencies were distributed in a log scale and each channel corresponded to one frequency level. The filter bandwidths were chosen following the work of Moore and Glassberg and were defined by 22 23 : where, f c is in KHz. The filtered signals through each channel were rectified. The rectified speech was processed through three integrate and fire (IF) neurons with three different thresholds levels. Each of the IF neurons would fire a spike once the summation crossed their respective thresholds. These spikes were then directly input to the crossbar array with each cross-point consisting of Mn:HfO 2 synaptic devices connected between input and output neurons. Training of the synaptic array in the simulation was performed as follows. The devices were all initialized to their unexcited conductance levels. When input neurons spike, it sends out 1 V pulses of 1 ms width as the AP. A counter was used to keep track of the time of arrival of each AP. The incoming currents from different rows were summed along a column of the crossbar and fed to the output neuron. The total current through a given column j is given by: where I j is the current summation of the j th column, V i is the magnitude of incoming spike AP for the i th row, and g ij is the synaptic conductance of Mn:HfO 2 device at the cross-point of i th row and j th column. The summed current was used to charge a capacitor of 20 pF for the postsynaptic IF neurom circuit in j th column. Once the potential of the j th column reached a threshold voltage of 2.5 V, a post-synaptic spike was fired. The time of postsynaptic firing was noted and the capacitor was reset to 0 V. The arrival of pre-synaptic spikes was paused temporarily and the Δt was obtained from the pre- and post-synaptic firing instants for each of the rows. Figure 9(b) shows an example of this implementation across one column. The conductance of each synapse in that column was then modified based on the relation shown in figure 8(c) . It is worth noting that the pre-synaptic spikes were chosen to be 1 V since it ensured the devices are not affected by the incoming pulses due to the fact that any significant change in device conductance occurs only when the applied bias is >1.75 V. At the same time the speech was sampled at 11.025 KHz, which meant that each sample of the 1 s recording was of 90 μs duration. Therefore it must be ensured that when the feedback pulses are applied during STDP, the incoming spikes are paused temporarily until the end of feedback. Hence for practical implementation of such a system, a timer based on a global clock is required which can help keep track of the pre- and postsynaptic firing instants. Once the post-synapse fires, the input pulses are paused by activating delay circuits at the input. The stored instances of pre- and postsynaptic firings are used to estimate the feedback pulse width and magnitude for each of the synaptic devices from equation (11) . The above implementation is based on a synchronous learning scheme. In this scheme, the requirement for keeping track of the precise firing times of neurons can add significant overhead in terms of circuit requirement, which needs to be further studied. To alleviate the requirement of additional circuitry, implementation of asynchronous STDP based on the back-propagation of post-synaptic spikes has been proposed 24 25 . Since the device conductance can be modulated using both pulse width and amplitude, such asynchronous STDP can also be implemented by capitalizing on the appropriate overlap between pre-and post-synaptic spikes. However, additional circuitry may be needed in this scheme for the desired spike-profile design, which is currently been studied 24 25 . The initial synaptic weights before training are shown in Figure 10(a) . The current levels at 0.5 V read are shown on the adjacent colour map. A unified colour refers to a constant conductance level for all the devices in the array. Figures 10(b) and (c) show the weight distribution of the synaptic array at the end of the simulation for words “apple” and “hello”, respectively. A clear distinction can be observed in the pattern of conductance levels for these two words. It is interesting to note that the current level along each row of the crossbar is equal as inferred from the colour map. Such a pattern was expected as each of the synaptic elements along a particular row undergoes the same STDP learning since the conductance change and conductance initialization for the synaptic devices were fixed. The impacts of device to device variability and statistical variation in STDP have been examined in a previous work which can be incorporated leading to a more diversified map 26 . The energy requirements of the synaptic device in the face of biological synapses were also evaluated. The energy consumption for transmission of 1 bit of information across a biological synapse is around 1 fJ 27 . The excitatory post synaptic current is less than 1 nA while the postsynaptic potential is <100 mV. The size of these devices is 100 × 100 μm 2 . Scaling down to a 400 nm 2 node would require a current density < (1 nA/400 nm 2 = 250 A/cm 2 ) for matching a biological synapse. It is apparent from the proposed model that the current due to SILC mechanism is dependent on field and pulse width and independent of trap depth and initial trap density. For scaling down the operating voltages it must be ensured that the electric field is still in the same range. Hence, for an operating voltage of 1 V, the dielectric needs to be scaled to 4 nm. Therefore, the optimum parameters for a 20 nm × 20 nm sized synaptic device operating at 1 V are given in Table I . Here the trap density and trap depth have been assumed to be 5 × 10 11 cm −3 and 0.19 eV respectively. It must be noted that there is a limit to dielectric thickness scaling since for very thin films F-P emission would cease to be the dominant conduction mechanism and direct tunnelling would tend to take over. Hence the lower limit of thickness was kept to be 4 nm. A comparison can also be made for the energy and area requirements of the proposed synaptic device with a 22 nm node VLSI synapse based on existing technology as shown in Table II . A ~10× improvement in area is obtained if the synapse circuitry using SRAM cell is replaced by the proposed synaptic devices. The power requirement for programming the device is also significantly low as a reduction of 10 6 times is obtained, while for a switching time of 10 ns for the SRAM cell, the energy requirements are comparable with the device. However, since the devices are slow, the overall energy consumption can be lowered further by designing better materials where defect generation and annihilation is a faster process and can occur at much lower fields. The future work in this area would include hardware implementation of the proposed approach and benchmarking against other technologies."
} | 4,591 |
27058503 | PMC5030696 | pmc | 782 | {
"abstract": "Members of the candidate phylum Hyd24-12 are globally distributed, but no genomic information or knowledge about their morphology, physiology or ecology is available. In this study, members of the Hyd24-12 lineage were shown to be present and abundant in full-scale mesophilic anaerobic digesters at Danish wastewater treatment facilities. In some samples, a member of the Hyd24-12 lineage was one of the most abundant genus-level bacterial taxa, accounting for up to 8% of the bacterial biomass. Three closely related and near-complete genomes were retrieved using metagenome sequencing of full-scale anaerobic digesters. Genome annotation and metabolic reconstruction showed that they are Gram-negative bacteria likely involved in acidogenesis, producing acetate and hydrogen from fermentation of sugars, and may play a role in the cycling of sulphur in the digesters. Fluorescence in situ hybridization revealed single rod-shaped cells dispersed within the flocs. The genomic information forms a foundation for a more detailed understanding of their role in anaerobic digestion and provides the first insight into a hitherto undescribed branch in the tree of life.",
"conclusion": "Ecological significance and concluding remarks This study applied metagenomic sequencing to obtain genomes from the candidate phylum Hyd24-12 and provides the first morphological and physiological information for the lineage. Members of the phylum were shown to be very abundant and stably present in mesophilic anaerobic digesters, occasionally accounting for the most abundant OTU in the samples, but absent in thermophilic reactors. This indicates that they are likely to play a substantial role in the ecology of mesophilic AD systems at wastewater treatment plants fed with primary sludge and surplus activated sludge. Metabolic reconstruction based on the genomic information showed that members of Hyd24-12 are likely to be fermenters relying on simple sugars. In addition, they may also use elemental sulphur as an electron acceptor, thus forming part of the microbial cycling of sulphur in anaerobic systems and partly responsible for production of hydrogen sulphide. Sulphide is unwanted in the biogas due to toxicity and corrosion ( Syed et al. , 2006 ), but will also provide more elemental sulphur by reacting with incoming Fe(III). In that case, members of Hyd24-12 may compete with the methanogens for organics. An in silico investigation of environmental 16S rRNA gene surveys suggests that members of the phylum are present in anaerobic environments, often associated with sulphurous compounds and methane production, such as sediment mats and anaerobic bioreactors. The fact that the genomes are auxotrophic for several amino acids and lacking putative secreted glycoside hydrolases also indicates a strict reliance on other organisms for nutrients. The genomes generated in this study provide the foundation for future detailed analyses of members of the phylum, such as metatranscriptomics and metaproteomics. The design of FISH probes for the phylum also revealed their morphology and spatial arrangement in anaerobic digesters and will also facilitate future in situ investigations of the phylum in digesters and other environments. Phylogenetic and genomic analyses of the three Hyd24-12 genomes classified them as a single species within a novel phylum located within the Fibrobacteres-Chlorobi-Bacteroidetes superphylum. We propose the following taxonomic names for the novel genus and species of Hyd24-12:\n ‘ Candidatus Fermentibacter ' gen. nov. ‘ Candidatus Fermentibacter daniensis ' gen. et sp. nov. Based on this, we propose the following names for the phylum, class, order, and family:\n ‘ Candidatus Fermentibacteria' phyl. nov. ‘ Candidatus Fermentibacteria' classis nov. ‘ Candidatus Fermentibacterales' ord. nov. ‘ Candidatus Fermentibacteraceae' fam. nov. Etymology Fermentibacter (Fer.men.ti.bac'ter. M.L. n. ferment -um to ferment, Gr. dim. n. bakterion a small rod, M.L. neut. n. Fermentibacter a small fermenting rod-shaped bacterium). Fermentibacter daniensis (da.ni.ensis. M.L. fem. adj. daniensis, pertaining to Dania, the Medieval Latin name for the country of Denmark, where the species was first discovered).",
"introduction": "Introduction Production of methane by anaerobic digestion (AD) is widely used to convert organic waste into biogas and forms an important part of the transition from fossil fuel to sustainable energy production. The AD process is divided into four sequential steps that are performed by specialized microbes: hydrolysis, fermentation (acidogenesis), acetogenesis (dehydrogenation) and methanogenesis (acetoclastic or hydrogenotrophic) ( Angenent et al. , 2004 ). Hence, the overall function, stability and efficiency of the AD process are dependent on tightly coupled synergistic activities of the complex microbial communities ( Schink, 1997 ; Weiland, 2010 ). However, the microbial communities in AD are still poorly understood, and relatively little is known about their diversity and function ( Chouari et al. , 2005 ; Werner et al. , 2011 ; Sundberg et al. , 2013 ; De Vrieze et al. , 2015 ). In addition, most of the microorganisms have no pure culture representatives, and, given the synergistic interactions of members of the community, a reductionist approach to understand the ecology of the system is not possible ( Kaeberlein et al. , 2002 ; Fuhrman et al. , 2015 ). The AD environment also harbours extensive diversity of previously uncharacterized bacterial phyla, often known only by their 16S rRNA gene sequence, making it an ideal environment for the study of novel bacterial lineages ( Guermazi et al. , 2008 ; Pelletier et al. , 2008 ; Limam et al. , 2014 ; Sekiguchi et al. , 2015 ). New developments in single-cell genomics and metagenomics have in recent years provided a glimpse into the ecology and evolution of many novel candidate phyla ( Dinis et al. , 2011 ; Albertsen et al. , 2013 ; Rinke et al. , 2013 ; Brown et al. , 2015 ; Nobu et al. , 2015 ; Sekiguchi et al. , 2015 ). The genomes have enabled construction of metabolic models that attempt to explain the physiology of these organisms in detail. The genome-based models form the basis of more extensive investigations, such as in situ single-cell characterization, metatranscriptomics and proteomics ( Koch et al. , 2014 ). In this study, extensive 16S rRNA gene amplicon sequencing was used to screen anaerobic digesters for the presence of members of the Hyd24-12 lineage, which remains one of the few known candidate phyla for which no genomic information is available with nothing known about their morphology, physiology or ecology ( Rinke et al. , 2013 ). Selected samples were subjected to metagenome sequencing and used for retrieval of three near-complete genomes of Hyd24-12 through differential coverage binning. The genomes were used for detailed metabolic reconstruction and design of oligonucleotide probes for the first in situ visualization of these hitherto unrecognized players in AD.",
"discussion": "Results and discussion Survey of 16S rRNA genes of Hyd24-12 in anaerobic digesters The survey of 22 full-scale mesophilic and 7 thermophilic anaerobic digesters from 17 Danish wastewater treatment plants over 3 years revealed that members of the Hyd24-12 lineage were stably present in most mesophilic but no thermophilic anaerobic digesters ( Figure 1 ). In most mesophilic digesters, they were among the five most abundant bacterial OTUs and constituted around 1–3% and, in some cases, up to 8.2% of all sequenced bacterial reads (see Supplementary Figure S1 ). No 16S rRNA gene sequences from Hyd24-12 were detected in the incoming surplus sludge from the activated sludge treatment plants, which demonstrates that these bacteria were actively growing in the digesters. The other abundant bacterial phyla in the mesophilic digesters were Actinobacteria, Firmicutes, Chloroflexi, Synergistetes and Bacteroidetes ( Figure 1 ). The best (LCA) classification is shown in Figure 1 , but the lack of closely related organisms in the databases and a curated taxonomy hampers taxonomic classification for a number of the most abundant OTUs. In general, the abundance stability of these top genera was high, and that may be due to relatively similar growth conditions for all digesters: feed was primary sludge and surplus activated sludge, temperature in the interval 34–37 °C, pH 7.1–8.2 and total ammonium 0.57–1.1 g N/l (see Supplementary Table S1 ). In silico analysis of 16S rRNA gene sequences within Hyd24-12 of SILVA ( Quast et al. , 2013 ) from other surveys confirmed that members of Hyd24-12 are widespread in anaerobic environments. The sequences originate from 48 separate studies, with engineered systems such as anaerobic bioreactors accounting for 10 studies, and natural systems such as marine sediments, microbial mats in hydrogen, methane-rich waters and mud volcanoes accounting for 38 studies (see Supplementary References ). Furthermore, the 48 studies show that members of Hyd24-12 are globally dispersed ( Supplementary Figure S3 and Supplementary Table S2 ) and are potentially important in many microbial ecosystems besides ADs ( Mills et al. , 2005 ; Harris et al. , 2012 ). Some of the surveys of full-scale anaerobic digesters detected some Hyd24-12 sequences (e.g., De Vrieze et al. , 2015 ), while others did not ( Sundberg et al. , 2013 ). This was likely because they used the RDP database, where Hyd24-12 sequences are classified as ‘unclassified bacteria'. Recovering genomic information from Hyd24-12 Three full-scale anaerobic digesters were sampled for metagenomic analyses. To ensure differential abundance of microorganisms needed to bin genomes based on coverage profiles ( Albertsen et al. , 2013 ), biomass samples were either taken from the sludge and foam layer of reactors or from the same reactor weeks apart. More than 50 gigabases of metagenomic data were generated, and population genomes were recovered by differential coverage binning ( Albertsen et al. , 2013 ) from each of the three plants ( Table 2 ). The three population genomes were ~2.2 Mbp with a GC content of ~64%, and the completeness of the genomes were estimated by CheckM ( Parks et al. , 2015 ) to be between 86% and 91% with less than 2.2% estimated contamination ( Table 2 ). However, the level of completeness may be underestimated, given that members of the Hyd24-12 are distantly related to other characterized organisms, and the genes used in the marker sets might be too divergent or simply not present ( Rinke et al. , 2013 ; Brown et al. , 2015 ; Sekiguchi et al. , 2015 ). The three genomes each contained a single rRNA operon and shared identical 16S rRNA gene sequences, which suggests that they belong to the same species ( Yarza et al. , 2014 ). The JSpecies program determined that these three genomes shared between 99.8% and 99.9% average nucleotide identity (ANIb), supporting the close taxonomic relationship observed from the 16S rRNA gene analysis ( Kim et al. , 2014 ). In order to further evaluate the similarity between the strains, the raw metagenome reads from each digester were mapped to the assembled Hyd24-12 genomes obtained from the other two digesters. Complete coverage of all genomes with the metagenome reads from the other digester revealed that the Hyd24-12 genomes were almost identical. This also indicates, along with the high ANIb, that the genomes are more complete than estimated in Table 2 by CheckM. Indeed, the data suggested that the three strains might actually be variants of the same strain with single-nucleotide polymorphisms only. This is very interesting as the digesters were from different parts of Denmark without any exchange of sludge or feed. This could indicate that they are highly adapted to the specific AD environment in this type of mesophilic digesters. Hyd24-12 phylogeny, FISH probe design and morphology The 16S rRNA genes obtained have a sequence identity of 86% with the original clone Hyd24-12 sequence (AJ535232) ( Knittel et al. , 2003 ) and classify to the Hyd24-12-lineage ( Figure 2a ). Additional phylogenetic analyses, based on the genome sequence, placed the Hyd24-12 genomes within the Fibrobacteres-Chlorobi-Bacteroidetes superphylum ( Figure 3 ). The Hyd24-12 genomes are distantly related to all currently available genomes, supporting its status as a novel phylum. Several FISH probes were designed to target different clades within the phylum. In the MiDAS taxonomy (v. 1.21) ( McIlroy et al. , 2015 ), a version of the SILVA taxonomy ( Quast et al. , 2013 ) that is curated for activated sludge-related organisms, the Hyd24-12 lineage is delineated into four clades, designated B-1AC, zEL51, Hyd-32 and B9.18. The Hyd24-12_468 and Hyd24-12_659 probes were designed to cover the B-1AC clade, which includes the Hyd24-12 genome sequences obtained in this study ( Figure 2 ). The former probe covers almost all the B-1AC sequences, with the closest non-target sequence match having three internal base mismatches. The Hyd24-12_659 probe is less specific, having one perfectly matched non-target sequence and several with mismatches not covered by the competitor probes. Overlap in the coverage of these two probes, labelled with different fluorochromes, allows greater confidence in their specificity. A suitable probe to cover the entire Hyd24-12 lineage was not found. However, the Hyd24-12_731 and Hyd24-12_842 probes provide good coverage of the other sequences in the phylum (see Table 1 ). As sequences covered by these additional probes were not detected in the full-scale anaerobic digesters studied here, optimization and assessment of these probes were not pursued. When applied to several full-scale anaerobic digester sludge samples, the Hyd24-12_468 and Hyd24-12_659 probes hybridized to small rods, approx. 2 × 0.4 μm in size, dispersed through the flocs (see Figure 2b ). Good overlap was observed for these probes, supporting their specificity. Of the two probes, a much higher signal was observed for the Hyd24-12_659 probe. There was no observed overlap between the signal of two Hyd24-12 probes and the universal bacterial EUBmix probe set (see Figure 2b ), which is supported by the absence of the target site for the probes of the latter in the Hyd24-12 sequences. Quantitative FISH was very difficult to carry out in the digesters due to high levels of background fluorescence. Instead, abundance estimates were carried out for the domains Bacteria, Archaea, Eukarya, and the Hyd24-12 lineage, based on read mapping from the PCR free metagenomes to the 16S rRNA genes of the MiDAS database. It showed that Archaea constituted 4–9% of the reads in sludge samples and 7–13% in foam samples. Reads from the Hyd24-12 lineage constituted 0.4–3.5% in the different samples ( Supplementary Table S3 ). Morphology and motility The rod shape morphology of B-1AC clade organisms observed by FISH is supported by mreBCD and mrdAB operons in the Hyd24-12 genomes (see Supplementary Data 2 ). These operons encode proteins involved in the formation of membrane-bound actin filaments, which are essential for the biogenesis of rod-shape stabilizing peptidoglycans along the lateral cell wall of rod-shaped bacteria ( Kruse et al. , 2003 , 2005 ; Osborn and Rothfield, 2007 ; Bendezú and de Boer, 2008 ). The cell envelope characteristics of genome-sequenced bacteria can be determined based on PFAM protein families that are substantially enriched or depleted in archetypical monoderm lineages relative to archetypical diderm lineages ( Albertsen et al. , 2013 ). A search for such protein families in the Hyd24-12 genomes revealed an archetypical diderm cell envelope with lipopolysaccharides (see Supplementary Figure S2 ). None of the Hyd24-12 genomes encode any flagella-related proteins, suggesting limited motility. However, genes associated with type IV pili were identified using the PilFind algorithm (see Supplementary Data 2 ) ( Imam et al. , 2011 ). These pili enable the bacteria to generate surface-associated twitching motility. This allows them to move effectively through environments that contain shear-thinning viscoelastic fluids, such as the extracellular polymeric substances of biofilms ( Conrad et al. , 2011 ; Jin et al. , 2011 ). In addition to motility, type IV pili play a role in the attachment to living and non-living surfaces, including those of other bacteria ( Giltner et al. , 2012 ). No genes associated with spore formation were detected in the Hyd24-12 genomes. This suggests that the Hyd24-12 genomes investigated represent non-sporulating bacteria. Energy metabolism The three genomes do not contain any genes for respiration with oxygen, nitrate/nitrite or Fe(III) and seem primarily to have a fermentative metabolism. However, the genomes indicate that the organisms may be able to use elemental sulphur as an electron acceptor, see below. The Hyd24-12 genomes encode a complete glycolysis pathway, along with the non-oxidative branch of the pentose phosphate pathway ( Figure 4 and see Supplementary Data 2 ). This allows Hyd24-12 to potentially catabolize a wide range of hexoses and pentoses to pyruvate, thereby providing the cell with energy in the form of ATP and reducing equivalents in the form of NADH ( Stincone et al. , 2014 ). The sugars are probably obtained from the environment through a major facilitator superfamily transporter at the expense of the proton motive force ( Madej, 2014 ; Wisedchaisri et al. , 2014 ). The transporter does not share similarity (>30%) with any experimentally validated transporters, and it is therefore impossible to infer a specific substrate preference. It is known that primary sludge and activated sludge fed into the digesters contain many different polysaccharides ( Raunkjaer et al. , 1994 ; Frølund et al. , 1996 ). No genes encoding for extracellular glycosylases were identified, which might indicate that Hyd24-12 is reliant on the hydrolytic action of other organisms present within the anaerobic digesters. Hyd24-12 encodes for the complete pathway for glycogen biosynthesis and catabolism ( Figure 4 and see Supplementary Data 2 ) ( Preiss et al. , 1983 ; Wilson et al. , 2010 ). Hence, glycogen may serve as a carbon and energy storage which can be utilized to mitigate fluctuations in substrate availability. The Hyd24-12 genomes did not encode for pathways for other storage compounds such as trehalose or polyhydroxyalkanoates. There are limited catabolic options for the pyruvate formed, for example, by glycolysis. The tri-carboxylic acid cycle of Hyd24-12 is incomplete (8 of 10 key enzymes are missing) and probably non-functional. However, pyruvate can be converted into acetyl-CoA by a pyruvate ferredoxin oxidoreductase, providing additional reducing equivalents in the form of reduced ferredoxin ( Figure 4 and see Supplementary Data 2 ) ( Menon and Ragsdale, 1997 ). Acetyl-CoA can then be converted into acetate by the action of phosphate acetyltransferase and acetate kinase, thus providing the bacterium with additional ATP ( Latimer and Ferry, 1993 ; Mai and Adams, 1996 ). All three Hyd24-12 genomes also encode for two aldehyde ferredoxin oxidoreductases ( Figure 4 and see Supplementary Data 2 ). These may be used to oxidize formaldehyde and acetaldehyde to formate and acetate, respectively, providing the cell with energy in the form of additional reduced ferredoxin ( Mukund and Adams, 1991 ). However, the enzyme may also be used in the reverse reaction to regenerate oxidized ferredoxin. The presence of a membrane-embedded, energy-conserving hydrogenase allows the cell to establish a proton motive force, based on the energy-rich reduced ferredoxin, which reduces H + to H 2 in the process ( Strittmatter et al. , 2009 ). The energy stored in the proton motive force may then be harvested through an ATP synthase to yield ATP. High concentrations of H 2 inhibit glycolysis and acidogenesis due to thermodynamic considerations ( Huang et al. , 2015 ). Hyd24-12 therefore needs a way to remove excess H 2 . This can be achieved by syntrophic association with other microorganisms, or internally by the action of a cytosolic hydrogenase, which couples the oxidation of H 2 with the reduction of NAD + ( Figure 4 and see Supplementary Data 2 ). Alternatively, Hyd24-12 may employ a sulfhydrogenase to couple the oxidation of H 2 to H + with the reduction of elemental sulphur (S 0 ) or polysulphide to hydrogen sulphide (H 2 S) as is seen for Pyrococcus furiosus ( Mukund and Adams, 1991 ). The genomes do not indicate a potential for sulphate reduction. Elemental sulphur is continuously produced in the digesters because activated sludge fed into the digesters contains oxidized iron (Fe(III)), which in the presence of sulphide produces S 0 and black iron sulphide (FeS) ( Rasmussen and Nielsen, 1996 ; Nielsen et al. , 2005 ; Omri et al. , 2011 ). Sulphide is a normal compound in digesters and is produced from amino acids and reduction of sulphate. Notably, other studies have also detected members of the Hyd24-12 phylum in sulphur-rich environments such as hydrothermal vents, sulphur-rich springs and sediments ( Elshahed et al. , 2003 ; Schauer et al. , 2011 ; Pjevac et al. , 2014 ). Thus, Hyd24-12 related organisms potentially play a role in sulphur transformations in digesters and other environments. Such a role requires further investigation. The Hyd24-12 genomes do not contain the genes required for fatty acid β-oxidation or for the catabolism of amino acids. Sugars are therefore considered the primary energy source of the Hyd24-12 in anaerobic digesters. Whereas Hyd24-12 is able to take up carbon in the form of amino acids, carbohydrates, etc., it is unable to carry out fixation of CO 2 as such genes are missing. Amino acid and nitrogen metabolism Based on the genome annotations, Hyd24-12 is only predicted to be able to synthesize few amino acids (glycine, serine, cysteine, threonine, asparagine, aspartate, glutamate and glutamine). Accordingly, Hyd24-12 might rely on amino acids present within the environment. As most amino acids are found as proteins, which cannot be taken up by the bacterium, Hyd24-12 needs a way to degrade these polymers, and this is achieved by the action of multiple extracellular proteases encoded in the genome, which are likely secreted in a Sec- or Tat-dependent mechanism ( Natale et al. , 2008 ) (see Supplementary Data 2 ). The cells may subsequently import the amino acids using ABC-transporters encoded in the genome. Owing to the lack of experimentally validated homologues from closely related species, it is not possible to predict the substrate specificity of these transporters. A reduced capacity of microorganisms for synthesizing amino acids is known from strict symbionts and, recently, also from a number of candidate phyla with very small genomes (<1 Mbp) ( Brown et al. , 2015 ). However, the relatively large size of the Hyd24-12 genomes (~2.2 Mbp) and their dispersed growth in the anaerobic sludge suggest that they are not strict symbionts. Hyd24-12 does not have the necessary pathways for fixation of nitrogen. The nitrogen metabolism of Hyd24-12 is generally limited. Amino acids may also represent a source of nitrogen. However, nitrogen can also be obtained from ammonium assimilation via the glutamine synthetase/glutamate synthase pathways ( Bravo and Mora, 1988 ). Oxidative stress protection The three Hyd24-12 genomes each contains a gene cluster encoding for a superoxide reductase, nitric oxide reductase and ferroxidase. These genes are probably involved in resistance against oxidative stress, and may allow the bacteria to survive in the presence of oxygen. However, 16S rRNA gene sequences from Hyd24-12 have only been observed in oxygen-depleted environments. Ecological significance and concluding remarks This study applied metagenomic sequencing to obtain genomes from the candidate phylum Hyd24-12 and provides the first morphological and physiological information for the lineage. Members of the phylum were shown to be very abundant and stably present in mesophilic anaerobic digesters, occasionally accounting for the most abundant OTU in the samples, but absent in thermophilic reactors. This indicates that they are likely to play a substantial role in the ecology of mesophilic AD systems at wastewater treatment plants fed with primary sludge and surplus activated sludge. Metabolic reconstruction based on the genomic information showed that members of Hyd24-12 are likely to be fermenters relying on simple sugars. In addition, they may also use elemental sulphur as an electron acceptor, thus forming part of the microbial cycling of sulphur in anaerobic systems and partly responsible for production of hydrogen sulphide. Sulphide is unwanted in the biogas due to toxicity and corrosion ( Syed et al. , 2006 ), but will also provide more elemental sulphur by reacting with incoming Fe(III). In that case, members of Hyd24-12 may compete with the methanogens for organics. An in silico investigation of environmental 16S rRNA gene surveys suggests that members of the phylum are present in anaerobic environments, often associated with sulphurous compounds and methane production, such as sediment mats and anaerobic bioreactors. The fact that the genomes are auxotrophic for several amino acids and lacking putative secreted glycoside hydrolases also indicates a strict reliance on other organisms for nutrients. The genomes generated in this study provide the foundation for future detailed analyses of members of the phylum, such as metatranscriptomics and metaproteomics. The design of FISH probes for the phylum also revealed their morphology and spatial arrangement in anaerobic digesters and will also facilitate future in situ investigations of the phylum in digesters and other environments. Phylogenetic and genomic analyses of the three Hyd24-12 genomes classified them as a single species within a novel phylum located within the Fibrobacteres-Chlorobi-Bacteroidetes superphylum. We propose the following taxonomic names for the novel genus and species of Hyd24-12:\n ‘ Candidatus Fermentibacter ' gen. nov. ‘ Candidatus Fermentibacter daniensis ' gen. et sp. nov. Based on this, we propose the following names for the phylum, class, order, and family:\n ‘ Candidatus Fermentibacteria' phyl. nov. ‘ Candidatus Fermentibacteria' classis nov. ‘ Candidatus Fermentibacterales' ord. nov. ‘ Candidatus Fermentibacteraceae' fam. nov. Etymology Fermentibacter (Fer.men.ti.bac'ter. M.L. n. ferment -um to ferment, Gr. dim. n. bakterion a small rod, M.L. neut. n. Fermentibacter a small fermenting rod-shaped bacterium). Fermentibacter daniensis (da.ni.ensis. M.L. fem. adj. daniensis, pertaining to Dania, the Medieval Latin name for the country of Denmark, where the species was first discovered)."
} | 6,752 |
39882623 | PMC11784650 | pmc | 783 | {
"abstract": "ABSTRACT Polyhydroxyalkanoates (PHA) are bioplastics produced by few bacteria as intracellular lipid inclusions under excess carbon source and nutrient-deprived conditions. These polymers are biodegradable and resemble petroleum-based plastics. The rising environmental concerns have increased the demand for PHA, but the low yield in wild-type bacterial strains limits large-scale production. An improvement in the PHA production can be achieved by genetically engineering the wild-type bacterial strains by removing competitive pathways that divert the metabolites away from PHA biosynthesis, cloning strong promotors to overexpress the genes involved in PHA biosynthesis and constructing non-native metabolic pathways that feed the metabolites for PHA production. The desired monomers in the PHA polymers were obtained by elimination of genes involved in PHA biosynthetic pathway. The chain length degradation specific-gene deletion of β-oxidation pathway resulted in the accumulation of PHA monomers having high carbon chain length. A controlled accumulation of monomers in the PHA polymer was achieved by constructing novel pathways in the bacteria and deleting native genes of competitive pathways from the genome of non-PHA producers. The present review attempts to showcase the novel genetic modification approaches conducted so far to enhance the PHA production with a special focus on metabolic pathway gene deletion in various bacteria.",
"conclusion": "7. Conclusion The Industrial-scale PHA production faces challenges due to low yield from wild-type bacterial strains and limited control over the polymer’s characteristics. Modern genetic engineering can significantly enhance PHA production. It is achieved by eliminating competitive metabolic pathways that divert metabolites away from PHA biosynthesis, overexpressing PHA biosynthesis genes using strong exogenous promoters, and constructing new metabolic pathways that allow bacteria to utilize diverse carbon substrates for PHA production. PHA depolymerase, encoded by the PhaZ gene, degrades PHA after reaching its maximum cellular level to mobilize monomers for energy. Deleting the PhaZ gene in bacteria prevents its degradation, significantly improving PHA yield. The selective deletion of PHA biosynthetic genes like PhaA, PhaB, and PhaC leads to the accumulation of specific monomers, enhancing the polymer’s desired properties. Similarly, knocking out fatty acid β-oxidation genes FadA and FadB prevents fatty acid degradation, causing synthesis of longer-chain PHA monomers. The non-PHA-producing bacteria, such as Escherichia coli , can be engineered to produce PHA by introducing desired PhaC genes and deleting genes involved in competitive pathways. Additionally, constructing a reversed fatty acid β-oxidation pathway in E. coli allows for the production of PHA with desired monomers from unrelated carbon substrates. These genetic engineering strategies offer robust methods to optimize PHA production for industrial applications.",
"introduction": "1. Introduction The excessive use of petroleum-based plastics in everyday life has raised severe environmental concerns over the past few decades due to their non-biodegradable and toxic nature. The persistence of these plastics for a long time will adversely affect the environment and hinder various biological processes by traversing across different trophic levels. Petroleum-based plastics undergo deterioration to form microplastics (<5 mm particle size), which are known to infiltrate organisms at various trophic levels, and the absence of enzymes to metabolize them leads to their retention and travels across the food chain, causing biomagnification. Even though microbial whole-cell or enzyme-catalyzed degradation techniques exist to curb the problem, the high crystallinity, hydrophobicity, additives, multilayer nature and cost-intensive pre-processing of synthetic plastics limits the process of degradation [ 1 ]. Therefore, scientists worldwide are searching for a biodegradable alternative to conventional petroleum-based plastics to avoid such a menace. Polyhydroxyalkanoates (PHA) are one of the few biodegradable plastics with properties similar to petroleum-based plastics. PHAs are polymers of ( R )-3-hydroxycarboxylic acid produced by a few bacteria inside the cell as lipid inclusions during nutrient starvation and physiological stress conditions [ 2 ]. The bacteria produce these polyesters as an energy storage molecule in a medium with an excess carbon source and limited nutrients such as nitrogen, oxygen, phosphorous, and magnesium [ 3 , 4 ]. PHAs have characteristics such as high durability, high tensile strength, water insolubility, resistance to hydrolytic damage, nontoxic, biocompatible, recyclable and resistance to ultraviolet degradation. They are soluble in many of the chlorinated solvents [ 5 ]. The ecofriendly nature of bioplastics has triggered an increase in global production capacities. It has increased the global market value for PHA to approximately 73.6 million USD in 2021 and is predicted to rise further to 167 million USD by 2027. Besides the ease of growing bacterial cultures in media to get large amounts of biomass, the production of PHA is low due to less PHA productivity of a few wild-type bacterial strains [ 6 ]. This is a severe setback for the industrial production of PHA. PHA are classified based on the number of carbon atoms in monomer. They are small chain length (scl) with 3–5 carbons, medium chain length (mcl) with 6–14 carbons, and large chain length (lcl) with more than 14 carbons [ 5 ]. Bacteria can either produce homopolymers or copolymers from these monomers [ 7 ]. The PHA properties differ by the composition of alkyl side chain bound at the β-carbon atom of the monomers [ 8 ]. Scl-PHA is crystalline due to their small alkyl groups with a 40–80% crystallinity [ 9 ]. To decrease the crystallinity of scl-PHA and improve flexibility, monomers with high chain lengths are synthesized as copolymers with scl-PHA [ 10 ]. The bacteria such as Cupriavidus sp., Azotobacter sp., Halomonas sp. and Bacillus sp. are notable scl-PHA producers. Mcl-PHA have elastomeric properties with crystallinity below 40% [ 11 ]. It is characterized by high elongation-to-break ratio, tensile strength, and low melting and glass transition temperatures. Bacteria such as Pseudomonas sp. and Burkholderia sp. are notable mcl-PHA producers [ 12 ]. Lcl-PHA are less produced, usually as copolymers with mcl-PHA causing enhanced amorphousness and thermal stability [ 13 , 14 ]. The PHA-producing conditions in media also support the co-production of various value-added products such as biosurfactants, polyglutamic acid, exopolysaccharides, and biofuels. Hence, bacterial PHA production can be used as an integrative approach toward the synthesis of other sustainable products [ 15 ]. The PHA find various industrial and biomedical applications based on their material properties. The biocompatible nature of PHA polymers allows them to prepare matrices for controlled drug delivery systems and body implants [ 16 ]. The drugs can be encapsulated into microspheres made of PHA polymers and used for targeted drug delivery [ 17 ]. It can be used as a wound dressing material [ 16 ]. It finds application in tissue engineering, where the PHA can be used to prepare scaffolds for the cell growth and regenerate the injured tissue [ 18 ]. The scl-PHA with high crystallinity can be used to pack hot substances due to their high-temperature resistance and heat deflection temperature [ 19 ]. Similarly, the mcl-PHA with elastomeric properties can be used to prepare biological latex and rubber-like materials. The mcl-PHA have applications in preparing adhesives, composites, and surface coating materials [ 17 ]. Due to superior mechanical properties, the PHA polymer and its various blends can be used as a packaging material [ 20 ]( Figure 1 ).\n Figure 1. A schematic representation of the broad application of PHA. Apart from regular sugar and fatty acid metabolism, the PHA biosynthetic pathway incorporates metabolites from 14 different pathways [ 21 ]. The bacteria grown using sugars as a carbon source utilize a glycolytic pathway for PHA biosynthesis. The PHA synthesized through this pathway consist of 3-hydroxybutyrate (3-HB) as monomers. The growth in short-chain length fatty acids such as propionic acid leads to the incorporation of 3-hydroxyvalerate (3-HV) as monomers along with 3-HB. The synthesis of mcl-PHA and lcl-PHA occurs through fatty acid β-oxidation. Mcl-PHA and lcl-PHA monomers accumulate when cultured in high-chain-length fatty acids and related substrates [ 16 ]. Enhancing PHA involves diverting substrates toward PHA biosynthesis by deleting genes in competing pathways, increasing substrate flow to PHA synthesis. Creating such mutants requires a thorough understanding of the interconnected metabolic pathways. The bacterial strain improvement using genetic engineering for enhanced PHA production has been explored less. Hence, the present review highlights various strategies in genetic engineering, such as the deletion of genes involved in competitive pathways, overexpression of PHA biosynthetic genes and construction of non-native metabolic pathways in bacteria to improve the strain, thereby enhancing the quantity of the PHA polymer. Most of the previous studies on PHA production involved the improvement of PHA yield by using various carbon substrates, optimizing physico-chemical conditions and employing different substrate-feeding strategies. They focused less on the quality of PHA polymer for industrial and commercial applications. Hence, the review not only deals with enhancing PHA production in bacteria but also gives a PHA quality perspective by emphasizing few studies conducted on the process involved in selective accumulation of desired monomers, production of PHA monomers with high chain length and produce polymers of known molecular weight in bacteria as a strategy to uphold the quality of the PHA polymer for industrial production. Additionally, the review provides valuable insight into the economics of PHA production and cost-cutting strategies involved in substrate choice, substrate pre-treatment and bacterial cultivation conditions, a significant aspect of PHA production."
} | 2,578 |
30532651 | PMC6225455 | pmc | 785 | {
"abstract": "A major biological challenge in the postgenomic era has been untangling the composition and functions of microbes that inhabit complex communities or microbiomes. Multi-omics and modern bioinformatics have provided the tools to assay molecules across different cellular and community scales; however, mechanistic knowledge over microbial interactions often remains elusive. This is due to the immense diversity and the essentially undiminished volume of not-yet-cultured microbes. Simplified model communities hold some promise in enabling researchers to manage complexity so that they can mechanistically understand the emergent properties of microbial community interactions. In this review, we surveyed several approaches that have effectively used tractable model consortia to elucidate the complex behavior of microbial communities. We go further to provide some perspectives on the limitations and new opportunities with these approaches and highlight where these efforts are likely to lead as advances are made in molecular ecology and systems biology.",
"conclusion": "CONCLUSION A variety of ecological theories have long been proposed for macro- and micro-ecosystems. Recent mathematical modeling studies have provided ecological rules and principles ( e.g., [ 84 , 148 , 149 ]). If we could obtain the appropriate microbes, synthetic microbial communities are the best tools to test the ecological theory, with the advantages of easy handling and tracking and an appropriate run time. Electricity-driven and photosynthesis-driven artificial microbial communities could provide desirable systems for testing the theory and rules, since the energy supply to the systems is easily regulated by electricity or illumination [ 150 , 151 ]. Designed microbial communities could be a potential model to experimentally characterize not only ecosystems but also various complex systems such as economy and human society.",
"introduction": "1 INTRODUCTION The sheer complexity of most ecosystems has long attracted and baffled scientists. For instance, Charles Darwin represented the complexity of ecosystems as a “tangled bank” [ 1 ], and Thomas Brock once described the clarification of microbial diversity in nature as “mumbo jumbo” [ 2 ]. In the present post-genomic era of molecular ecology, we have been confronted with the realization that the immense taxonomic and functional diversity of most natural microbial communities render them intractable for comprehensive mechanistic studies. For example, natural soil microbiomes are critical for the functioning of terrestrial ecosystems [ 3 ] and are extremely diverse from both the taxonomic and functional aspects; for instance, 1 g of soil may harbor 10 9 microbial cells with representatives from 10 3 -10 6 species [ 4 , 5 ]. Furthermore, we know that microbes respond to each other to achieve emergent, higher order metabolic functioning and genetic adaptation [ 6 - 8 ] making it essentially impossible to translate the knowledge gained from isolates to complex natural systems. At the turn of the century, synthetic ecology emerged as a nascent offshoot of synthetic biology [ 9 ]. Many consortial systems are being designed and/or adapted for biotechnological applications such as biofuel/bioproduct synthesis and management of greenhouse gases [ 10 - 12 ]. In addition, this movement has revealed a new paradigm for studying the natural microbial ecosystems. Rather than attempting to decompose the so-called “tangled bank”, the complexity is managed in model microbial consortia which are designed and built [ 13 - 15 ] (Fig. 1 ). This practice is now common, and the number of experimental systems continues to grow for testing and learning ecological theories and validating hypothesis. We should choose the appropriate model consortia that are useful to answer questions that we want to solve. Fig. ( 2 ) summarizes the microbial ecological methods applicable to microbial consortia, depending on complexity. In this review, we introduce several approaches that have effectively used model communities (two-species co-cultures, three-species and higher member complex co-cultures, and enriched model systems) to elucidate the microbial community functions and behaviors from wide viewpoints, i.e., gene regulatory networks, metabolic interactions, and ecological theory (Fig. 2 ). In addition, we provide some perspectives on the limitations and emerging opportunities with these approaches as well as highlight the possible consequences of applying these approaches."
} | 1,127 |
25530008 | PMC4335971 | pmc | 787 | {
"abstract": "The anaerobic oxidation of methane (AOM) is an important sink of methane that plays a significant role in global warming. AOM was first found to be coupled with sulfate reduction and mediated by anaerobic methanotrophic archaea (ANME) and sulfate-reducing bacteria (SRB). ANME, often forming consortia with SRB, are phylogenetically related to methanogenic archaea. ANME-1 is even able to produce methane. Subsequently, it has been found that AOM can also be coupled with denitrification. The known microbes responsible for this process are Candidatus Methylomirabilis oxyfera ( M. oxyfera ) and Candidatus Methanoperedens nitroreducens ( M. nitroreducens ). Candidatus Methylomirabilis oxyfera belongs to the NC10 bacteria, can catalyze nitrite reduction through an “intra-aerobic” pathway, and may catalyze AOM through an aerobic methane oxidation pathway. However, M. nitroreducens , which is affiliated with ANME-2d archaea, may be able to catalyze AOM through the reverse methanogenesis pathway. Moreover, manganese (Mn 4+ ) and iron (Fe 3+ ) can also be used as electron acceptors of AOM. This review summarizes the mechanisms and associated microbes of AOM. It also discusses recent progress in some unclear key issues about AOM, including ANME-1 in hypersaline environments, the effect of oxygen on M. oxyfera , and the relationship of M. nitroreducens with ANME.",
"conclusion": "Conclusion The microbes responsible for AOM are difficult to cultivate because of their low growth rates (Jagersma et al. 2009 ). AOM is an “active” process in microbial studies that contributes significantly to the global methane cycle. Currently, three different processes (Table 3 ) are thought to be responsible for AOM, with sulfate, nitrite/nitrate, and metal ions (Mn 4+ and Fe 3+ ) serving as electron acceptors. However, the specific mechanism of AOM is not fully known, and the exact features of the responsible microbes require further study. Table 3 Comparisons between the three processes of AOM: S-DAMO, N-DAMO, and M-DAMO Features S-DAMO N-DAMO M-DAMO Habitat Marine environments and freshwater environments Freshwater environments Marine environments Mechanism hypothesis Reverse methanogenesis, acetogenesis, and methylogenesis Aerobic methane oxidation and reverse methanogenesis ND Electron acceptor Mn 4+ and Fe 3+ Responsible microbes ANME M. oxyfera and M. nitroreducens MBGD (possible) Reaction (AOM) (eq. 3 ) (eq. 12 ) and (eq. 14 ) and (eq. 13 ) (eq. 15 ) S-DAMO, sulfate-dependent anaerobic methane oxidation; N-DAMO, nitrate/nitrite-dependent anaerobic methane oxidation; M-DAMO, metal ion (Mn 4+ and Fe 3+ )-dependent anaerobic methane oxidation; ANME, anaerobic methanotrophic archaea; M. oxyfera , Candidatus Methylomirabilis oxyfera; M. nitroreducens , Candidatus Methanoperedens nitroreducens; MBGD, marine benthic group D; ND, not determined; AOM, anaerobic oxidation of methane.",
"introduction": "Introduction Methane is the second most abundant greenhouse gas after carbon dioxide (CO 2 ), which accounts for 14% of global greenhouse gas emissions (EPA 2006 ). The concentration of methane in the atmosphere has increased ∼2.5 times than the preindustrial level, rising from 720 ppb in 1750 to 1803 ppb in 2011 (Hartmann et al. 2013 ). Although the methane concentration in the atmosphere is lower than the CO 2 concentration (391 ppm), methane is 25-fold more effective in trapping heat in the atmosphere than CO 2 on a per-molecule basis (IPCC 2007 ). Methane contributes to ∼30% of the anthropogenic warming, with the radiative forcing of 0.48 Wm −2 in 2011 (Myhre et al. 2013 ). After maintaining a relatively stable level for approximately a decade in the 1990s, the atmospheric methane concentration began to grow in 2007 (Hartmann et al. 2013 ). The concentration of methane in the atmosphere is determined by the balance of sources and sinks. The anaerobic oxidation of methane (AOM) is an important sink of the atmospheric methane concentration (Conrad 2009 ), which significantly impacts global warming. In marine sediments, the total amount of gas hydrates is up to 150–3000 times the atmospheric methane concentration (500,000–10,000,000 Tg CH 4 ) (Reeburgh 2007 ). Fortunately, most of the mobilized CH 4 is consumed through anaerobic methane oxidation, with a consumption rate of approximately 70–300 Tg CH 4 year −1 . Without this process, there would be an additional 10–60% of CH 4 in the atmosphere (Conrad 2009 ). AOM was first discovered in 1976 and is coupled with sulfate reduction in marine sediments (Reeburgh 1976 , 1980 ). However, the responsible microorganisms for this process were actually identified ∼20 years later (Hinrichs et al. 1999 ; Boetius et al. 2000 ; Bian et al. 2001 ). In 2006, a new AOM process named nitrite-dependent anaerobic methane oxidation (N-DAMO) was reported; this process is coupled with denitrification (Raghoebarsing et al. 2006 ). It was shown that nitrate could also be an electron acceptor of AOM in addition to nitrite (Haroon et al. 2013 ). Likewise, Beal et al. ( 2009 ) suggested that AOM is coupled with the reduction of manganese (Mn 4+ ) and iron (Fe 3+ ) in marine sediments. Overall, there are three different processes of AOM depending on the different electron acceptors: sulfate-dependent anaerobic methane oxidation (S-DAMO) (Fig. 1A ), nitrate/nitrite-dependent anaerobic methane oxidation (N-DAMO) (Fig. 1C and D), and metal ion (Mn 4+ and Fe 3+ )-dependent anaerobic methane oxidation (M-DAMO) (Fig. 1B ). This review summarizes the biochemistry and microbiology of these three AOM processes, including the mechanisms and distribution of AOM processes, the responsible microbes, and their peculiar properties. Moreover, this review also discusses several key issues about the recent progress of AOM that are still unclear. Figure 1 Three different models of anaerobic methane oxidation (AOM) depending on the different electron acceptors: (A) sulfate-dependent anaerobic methane oxidation (S-DAMO); (B) metal ion (Mn 4+ and Fe 3+ )-dependent anaerobic methane oxidation (M-DAMO); and (C, D) nitrate/nitrite-dependent anaerobic methane oxidation (N-DAMO). ANME, an anaerobic methanotrophic archaea; SRB, sulfate-reducing bacteria; M. oxyfera , Candidatus Methylomirabilis oxyfera; M. nitroreducens , Candidatus Methanoperedens nitroreducens; MBGD, marine benthic group D.",
"discussion": "Discussion of Key Issues In the light of recent progress regarding AOM, several unclear issues need to be further elucidated. These issues are discussed below. Is ANME-1 a hypersaline anaerobic methanotroph ecotype? The ANME population consisting only of ANME-1 was first found in a natural sedimentary that was high in salt (Lloyd et al. 2006 ). Then, ANME-1 was reported in other hypersaline environments (Yakimov et al. 2007 ; Cono et al. 2011 ; Maignien et al. 2013 ). Therefore, ANME-1 may be a hypersaline anaerobic methanotroph ecotype, which was suggested to be related to the comparatively low effect of ionic strength on ANME-1 (Maignien et al. 2013 ). ANME-1 cell membranes contain high contents of glycerol dialkyl glycerol tetraethers (GDGTs) (Niemann and Elvert 2008 ), which are characterized by a lower permeability compared with typical membrane lipids (Yamauchi et al. 1993 ; Valentine 2007 ). ANME-2 and ANME-3 cell membranes contain less or no GDGTs; the dominant component of ANME-2 and ANME-3 cell membranes is diethers exhibiting a higher permeability (Rossel et al. 2011 ). In addition, the ANME-1 genome was shown to contain genes coding for mannosylglycerate and di-myo-inositol-phosphate synthesis pathways (Meyerdierks et al. 2010 ), which are widely used to increase the turgor pressure by halophilic microorganisms (Roberts 2004 ; Empadinhas and da Costa 2008 ). Recently, proteins involved in gas vesicle synthesis have been identified in the proteome of ANME-1 (Stokke et al. 2012 ). Gas vesicles have also been observed in halophilic archaea (Walsby 1994 ), which might function in a salt stress response (Hechler and Pfeifer 2009 ). The above may contribute to the domination of ANME-1 in hypersaline environments. More needs to be investigated on this topic in the future. The effect of oxygen on M. oxyfera Candidatus Methylomirabilis oxyfera has the ability to conduct methane oxidation through a strictly O 2 -dependent reaction catalyzed by particulate methane monooxygenase (pMMO) (Ettwig et al. 2010 ). However, it was found that the addition of either 2% or 8% of O 2 had an overall detrimental effect on M. oxyfera , and the ability of this bacterial species did not resume the original level (Luesken et al. 2012 ). These observations suggest that the O 2 production and consumption of M. oxyfera is tightly controlled process, and the detrimental effect of O 2 on M. oxyfera may be unrecoverable. However, most M. oxyfera and M. oxyfera -like bacteria have been observed in the oxic/anoxic interface of freshwater habitats (Raghoebarsing et al. 2006 ; Ettwig et al. 2008 ; Hu et al. 2011 ; Luesken et al. 2011a , b ). Additionally, it is possible that the applied oxygen concentration was too high. In consideration of the above information, it is controversial whether M. oxyfera could use external O 2 to oxidize methane. The effect of oxygen on M. oxyfera still remains unclear. The relationship of M. nitroreducens with ANME Candidatus Methanoperedens nitroreducens, a new species responsible for N-DAMO (Haroon et al. 2013 ), which is affiliated with ANME-2d, is the fourth subgroup of ANME-2 for S-DAMO (Mills et al. 2003 ). In addition, a full reductive acetyl-CoA (carbon fixation) pathway and acetyl-CoA synthetase have been identified in M. nitroreducens . It was predicted that M. nitroreducens might be able to produce acetate (Haroon et al. 2013 ), as can ANME-1 (Meyerdierks et al. 2010 ). The reported relationship of M. nitroreducens with ANME suggests that N-DAMO may be associated with S-DAMO, which warrants further investigation."
} | 2,509 |
37469628 | null | s2 | 788 | {
"abstract": "The study of quorum sensing, bacterial cell-to-cell communication mediated by the production and detection of small molecule signals, has skyrocketed since its discovery in the last third of the 20th century. Building from early investigations of bacterial bioluminescence, the process has been characterized to control a numerous and growing number of group behaviors, including virulence and biofilm formation. Bonnie Bassler has made key contributions to the understanding of quorum sensing, leading interdisciplinary efforts to characterize key signaling pathway components and their respective signaling molecules across a range of gram-negative bacteria. This review highlights her work in the field, with a particular emphasis on the chemical contributions of her work."
} | 194 |
34199202 | PMC8231790 | pmc | 789 | {
"abstract": "We performed a comparative study on the Gaussian noise and memristance variation tolerance of three crossbar architectures, namely the complementary crossbar architecture, the twin crossbar architecture, and the single crossbar architecture, for neuromorphic image recognition and conducted an experiment to determine the performance of the single crossbar architecture for simple pattern recognition. Ten grayscale images with the size of 32 × 32 pixels were used for testing and comparing the recognition rates of the three architectures. The recognition rates of the three memristor crossbar architectures were compared to each other when the noise level of images was varied from −10 to 4 dB and the percentage of memristance variation was varied from 0% to 40%. The simulation results showed that the single crossbar architecture had the best Gaussian noise input and memristance variation tolerance in terms of recognition rate. At the signal-to-noise ratio of −10 dB, the single crossbar architecture produced a recognition rate of 91%, which was 2% and 87% higher than those of the twin crossbar architecture and the complementary crossbar architecture, respectively. When the memristance variation percentage reached 40%, the single crossbar architecture had a recognition rate as high as 67.8%, which was 1.8% and 9.8% higher than the recognition rates of the twin crossbar architecture and the complementary crossbar architecture, respectively. Finally, we carried out an experiment to determine the performance of the single crossbar architecture with a fabricated 3 × 3 memristor crossbar based on carbon fiber and aluminum film. The experiment proved successful implementation of pattern recognition with the single crossbar architecture.",
"conclusion": "5. Conclusions A comparative study was performed on the Gaussian noise and memristance variation tolerance of the complementary crossbar architecture, the twin crossbar architecture, and the single crossbar architecture. To make the comparison, we used 10 grayscale images as input images for recognition with the three crossbar architectures. Gaussian noise was added to the input images before using the crossbar architectures for recognition. The three architectures were also tested for pattern recognition under conditions of memristance variations. The SNR value was varied from −10 to 4 dB and the percentage of memristance variation was changed from 0% to 40% to record the average recognition rates. Finally, we conducted an experiment to determine the performance of the single crossbar array architecture for pattern recognition in which a 3 × 3 memristor crossbar was fabricated and used for recognizing three specific patterns. Based on the simulation results, we conclude that the single crossbar architecture is the best architecture among the three architectures for image recognition under the effect of Gaussian noise and memristance variation in terms of recognition rate.",
"introduction": "1. Introduction The memristor, the new fourth basic circuit element, was mathematically proposed by L. O. Chua in 1971 [ 1 ] and experimentally demonstrated by the HP lab in 2009 [ 2 ]. Since then, memristors have been crucially used to demonstrate neuromorphic computing systems, which were conceptually proposed in 1990 by C. Mead [ 3 ]. The nonlinear charge–flux relationship of the memristor, which can be used to simulate the behavior of human synapses [ 4 , 5 ], makes it a promising candidate for neuromorphic systems. Furthermore, the conductance of memristors could be modified and saved by applying programming pulse [ 4 , 6 ], which is the key characteristic of memristors for supporting neuromorphic system implementation. Interestingly, memristors can be formed as a crossbar array, which is a fully connected mesh of crossing wires [ 7 , 8 , 9 ]. Two crossing wires in the crossbar are connected by a memristor acting as a switch [ 7 , 9 ]. Memristor crossbars have opened opportunities to implement artificial neural networks on chips where the synaptic weights of network are stored in crossbar array [ 10 , 11 , 12 , 13 ]. These potential applications, however, require huge computational tasks and training processes. Recently, other approaches have been proposed where memristor arrays were used for neuromorphic pattern recognition, including speech recognition and image recognition [ 14 , 15 ]. The complementary architecture, in which one memristor crossbar is the inversion of the other, is used for the application of speech recognition [ 14 ]. It is based on a logical Exclusive-NOR (XNOR) operation, which measures the similarity of two binary arrays [ 14 ]. The twin crossbar architecture employing two identical crossbar arrays has been proven capable of measuring the similarity between an input pattern and the stored patterns as well [ 15 ]. The twin crossbar architecture consumes less power than the complementary crossbar architecture for the application of image recognition. In complementary crossbar architecture, the number of ‘1′ bits is always equal to the number of ‘0′ bits, irrespective of the sparsity density of images stored in the crossbars, because the two crossbars are complementary to each other. By contrast, the number of ‘1′ bits in the twin crossbar architecture is dependent on the data density of the images. For this reason, the twin crossbar architecture consumes less power than the complementary crossbar architecture if and only if the images stored in the crossbar array have the number of ‘1′ bits less than the number of ‘0′ bits, for instance, in DCT compressed images [ 15 ]. An up-to-date architecture, the single crossbar architecture, obtained by simplifying the Exclusive-NOR operation, needs only one memristor array for implementing the Exclusive-NOR function in pattern recognition tasks [ 16 ]. The complementary and twin crossbar architectures accept unipolar inputs, but the single crossbar array accepts bipolar inputs instead. In term of power consumption and area occupation, each type of crossbar architecture has significant advantage as they are applied to the specific application. In particular, the power consumption can be saved in the twin crossbar architecture with DCT compressed images, in which the number of ‘1′ bits is much less than the number of ‘0′ bits [ 15 ]. To save area, we can consider the single crossbar architecture, but the unipolar to bipolar circuit must be used in this case [ 16 ]. All the above crossbar architectures require memristors to operate at a desired memristance value, which is either low resistance state (LRS) or high resistance state (HRS). However, the memristance value varies from device to device due to manufacturing variation or being programmed into an undesired state [ 17 , 18 , 19 , 20 , 21 ]. Memristance variation is one of the factors that degrade the performance of the memristor crossbar circuit [ 17 , 18 , 19 ]. All of the above crossbar architectures have been tested with clean images. However, the recognition rate of these crossbar architectures may be reduced with noisy images. In this work, we performed a comparative study on the Gaussian noise and memristance variation tolerance of the complementary crossbar architecture, the twin crossbar architecture, and the single crossbar architecture. Based on the results, we determined that the single crossbar architecture produced the best recognition rate among the three architectures for image recognition under the effect of Gaussian noise and memristance variation. We also performed an experiment on the single crossbar architecture with fabricated 3 × 3 memristor crossbar based on carbon fiber and aluminum film for storing and recognizing three simple patterns.",
"discussion": "4. Discussion The simulation results showed that, overall, the single crossbar architecture produced the highest recognition rate under conditions of Gaussian noise inputs and memristance variations. When input images with Gaussian noise at the SNR of −10 dB was applied to three memristor architectures, the single crossbar architecture had a recognition rate of 91%, which was 2% and 87% higher than the recognition rates of the twin crossbar and the complementary crossbar architecture, respectively. Under the condition of 40% memristance variation, the single crossbar architecture produced a recognition rate as high as 67.8%, which was 1.8% and 9.8% higher than the rates of the twin crossbar and the complementary crossbar architectures, respectively. Our experimental demonstration with a fabricated 3 × 3 memristor crossbar also proved the successful implementation of pattern recognition with the single crossbar architecture based on the XNOR function as presented in Equation (7)."
} | 2,177 |
30679712 | PMC6345747 | pmc | 790 | {
"abstract": "Marine cleaning interactions have been useful model systems for exploring evolutionary game theory and explaining the stability of mutualism. In the Indo-Pacific, cleaner organisms will occasionally “cheat” and remove live tissue, clients use partner control mechanisms to maintain cleaner honesty, and cleaners strategically increase service quality for predatory clients that can “punish” more severely. The extent to which reef communities in the Caribbean have evolved similar strategies for maintaining the stability of these symbioses is less clear. Here we study the strategic service provisioning in Pederson’s cleaner shrimp ( Ancylomenes pedersoni ) on Caribbean coral reefs. In the Gulf of Honduras, we use video observations to analyze >1000 cleaning interactions and record >850 incidents of cheating. We demonstrate that A. pedersoni cheat frequently and do not vary their service quality based on client trophic position or cleaner shrimp group size. As a direct analog to the cleaner shrimp A. longicarpus in the Indo-Pacific, our study highlights that although cleaning interactions in both ocean basins are ecologically analogous and result in parasite removal, the strategic behaviors that mediate these interactions have evolved independently in cleaner shrimps.",
"introduction": "Introduction The evolution and stability of intraspecific cooperation and interspecific mutualism (terms we use as defined by 1 – 4 ) have long fascinated evolutionary biologists, and are generally explained by one of two classic scenarios: (i) helping is the by-product of a self-serving act, or (ii) there is concerted investment by both partners that yields future fitness benefits (reviewed by 4 ). The latter creates an inherent incentive to cheat, or defect, as the immediate benefits of cheating outweigh the immediate benefits of investing. Under certain game theoretical models (e.g. the iterated prisoner’s dilemma), selection should drive the evolution of partner control mechanisms to regulate interactions, at which point helping behaviors emerge as stable evolutionary outcomes 2 , 3 , 5 , 6 . In some settings (e.g. symbioses), large intraspecific groups provide services simultaneously to the same client as part of a larger interspecific mutualism; a situation analogous to the iterated prisoner’s dilemma. Here, complex decisions may be required by each individual regarding the provision of honest and dishonest behaviors, and their standing in the larger intraspecific group and interspecific interactions 4 , 6 – 8 . Bshary et al . 6 developed a model that predicts two stable evolutionary strategies for this scenario: (i) service providers should cheat immediately to gain maximal fitness benefits leading to a complete breakdown in service quality, or (ii) service quality should be greater when provided in pairs, in an iterated game, than by single service providers. These predictions rely on the assumption that each client mediates the interaction in a similar manner (i.e. cheating is always met with the same level of punishment). Variation in a client’s ability to punish defectors may lead to scenarios where service providers, either singly or in pairs, may strategically alter service quality based on perceived risk of expulsion from an intraspecific social group, alongside the severity of interspecific punishment. Under these scenarios, clients with the ability to punish cheating more severely would be predicted to receive high quality services from both single and paired providers, while greater variation in service quality should manifest itself in clients that are unable to punish as severely 7 , 9 , 10 . In marine systems cleaning symbioses have been classic model systems for investigating game theory and the stability of intra- and inter-specific cooperation and mutualism 6 – 9 , 11 . Cleaning interactions are predicated on mutualistic behaviors between one or more cleaners (the service providers) and client reef fish. Cleaners, typically fish or shrimps, remove potentially harmful ectoparasites from clients that pose motionless at cleaning stations, and clients do not prey on small, vulnerable cleaners. Cleaning is done singly or in intraspecific groups and cleaner presence has been shown to positively impact fish health and diversity on coral reefs 12 – 16 . However, not every cleaning interaction is mutualistic. Some cleaner organisms have shown preference for nutritive sources other than client ectoparasites such as fish mucus or live tissue, and thus, an inherent incentive to cheat exists for some cleaners 17 . Partner control mechanisms (e.g. partner switching, aggressive chasing) imposed by the client on the cleaners have evolved to mediate incidences of cheating, and cleaner service quality varies depending on cleaner group size and client trophic position 9 , 10 , 18 . In the tropical Indo-Pacific, both cleaner wrasse ( Labroides dimidiatus ) and cleaner shrimp ( Ancylomenes longicarpus ) have been shown to provide greater service quality to predatory than herbivorous clients 6 , 10 . Cleaning in intraspecific groups of two or more has been shown to reduce incidences of cheating in cleaner wrasse, and while its role remains less clear for cleaner shrimp, larger group sizes have shown a reduction in cheating 6 , 10 . Partner control mechanisms appear to be well developed on Indo-Pacific reefs, and antagonistic displays of aggression by clients have been documented towards both cleaner types 6 , 10 . However, not all cleaning symbioses may so clearly meet the expectations established by game theory. Caribbean cleaner gobies (genus Elacatinus ) appear to show no strategic adjustment of service quality based on client trophic position, and the symbiosis appears to be a system without punishment or partner control 8 , 19 – 21 . At most, clients leave immediately after a cheating incident, but no physical punishment has been documented. Some evidence exists that male cleaner gobies reduce cheating when cleaning in intraspecific pairs, but service quality remains high regardless of group size, in contrast to Indo-Pacific cleaner wrasse 8 . Whether cleaner gobies are directly comparable to cleaner wrasses and broadly representative of how stable cleaning strategies have evolved in both ocean basins is unclear. Wrasses and gobies are not taxonomic analogs and there are notable ecological differences between cleaner types 17 , 21 . Further, there is a lack of research on strategic service behaviors from non-fish cleaners in the Caribbean. This perspective is needed to broaden the taxonomic scope of the cleaning literature and begin to synthesize behavioral patterns that have evolved across ocean basins. On Caribbean coral reefs, cleaner shrimps are important cleaners, and although they have been less studied than cleaner fishes, recent research has shown that they are visited frequently, have broadly overlapping client diversity with cleaner gobies, and are effective at reducing parasite loads on reef fishes 22 – 28 . Among the most well studied and ecologically important Caribbean cleaner shrimps is Pederson’s cleaner shrimp, Ancylomenes pedersoni ; a dedicated cleaner (i.e. a species that cleans for all of its non-larval ontogeny) 29 that associates with sea anemones and that cleans over 20 families of client reef fishes 23 – 26 . Whereas behavioral comparisons between Caribbean cleaner gobies Elacatinus spp. and the Indo-Pacific cleaner wrasse L. dimidiatus are not direct analogs, as they do not share a recent common ancestor and are known to prefer different food sources (i.e. fish mucus vs. ectoparasites), A. pedersoni is a direct analog to A. longicarpus . They belong to the same genus, are both symbiotic with sea anemones, perform cleaning services singly or with multiple conspecifics, and both are known to cheat client reef fish 10 , 23 , 27 , 30 , 31 . Here we investigate the effect of cleaner group size and client trophic position on strategic service provisioning of A. pedersoni in the Gulf of Honduras to fill an important gap in our knowledge on the evolution of cleaning behaviors on tropical coral reefs. Namely, we aim to provide critical data to explore whether strategic cleaning behaviors have converged on similar or dissimilar strategies in separate ocean basins. With these data, we ask whether A. pedersoni , like A. longicarpus , strategically adjusts service quality based on client trophic position and intraspecific group size, or whether, like the co-occurring cleaner gobies in the Caribbean, there is no adjustment of service quality for any client or group size.",
"discussion": "Discussion We demonstrate that Pederson’s cleaner shrimp, Ancylomenes pedersoni , do not vary their service quality in the Gulf of Honduras based on cleaner group size, the trophic position of the client reef fish, reef site, or time of day in the hypothesized directions (i.e. more cheating on non-predatory than predatory fish, and more cheating by solitary shrimp than groups of cleaners). Some significant interaction effects were recovered between group size and client trophic position while attempting to account for biologically realistic behaviors in two of our datasets. These effects suggest that the shrimp may be able to discriminate and respond to some aspects of the interaction. What might be causing these remains unclear, and while we cannot rule out some effects of cleaner shrimp group size or client type on cheating rate, the patterns we have recovered are not consistent with our predictions. Additionally, clients of A. pedersoni did not leave immediately following a putative cheating event, and did not display any aggressive behavior towards cleaners as has been seen in other studies of this cleaner shrimp. In light of our results, when considered in conjunction with studies on cleaner gobies 19 – 21 , it appears that the cleaning system in the Caribbean (cleaners and clients), although functionally and ecologically analogous to those in the Indo-Pacific, have not evolved the same strategic service behaviors or partner control mechanisms. Soares et al . 19 previously noted differences in strategic service behaviors and partner control mechanisms between Caribbean cleaner gobies and L. dimidiatus , and we come to similar conclusions regarding the differences in strategic service provisioning between A. pedersoni in the Caribbean and A. longicarpus in the Indo-Pacific. Soares et al . 19 presented three non-mutually exclusive hypotheses to explain the differences in the evolution of behavioral strategies between cleaner gobies in the Caribbean and cleaner wrasse in the Indo-Pacific that, considering the implications of our findings, may also apply to regional differences between Ancylomenes spp. These are (1) the constraint hypothesis, (2) the low cost of cheating hypothesis, and (3) the foraging preference hypothesis. The constraint hypothesis presents two potential constraints that may explain why Caribbean cleaner systems have not evolved the behavioral complexity of Indo-Pacific systems: (i) Caribbean cleaners are less dependent on a cleaning lifestyle than Indo-Pacific cleaners. Caribbean cleaner gobies, although considered a dedicated cleaner species, have been shown to gain up to 85% of their dietary needs through cleaning while the cleaner wrasse L. dimidiatus acquires 99% of their diet through cleaning. Comparable data for cleaner shrimps A. pedersoni and A. longicarpus do not exist, which are needed to evaluate this hypothesis more broadly. (ii) Differences in cleaning strategy and partner control between ocean basins could arise if cleaners are constrained in their cognitive abilities to remember the interaction and the identity of the client fish that inflicts punishment for cheating. Interestingly, unlike our data on A. pedersoni , Chapuis and Bshary 10 show that A. longicarpus can discriminate between predatory and non-predatory clients, although they have no specific hypotheses to explain this ability. Recent research, however, suggests that differentiating between clients visually would not be likely, as two cleaner shrimps from the family Palaemonidae (including A. pedersoni ) have been shown to have poor spectral sensitivity (i.e. monochromatic vision), and low spatial resolution 32 . Caves et al . 32 conclude that cleaner shrimp vision is sufficient to detect large stimuli, but likely cannot detect sharp outlines, color, or patterns of either potential clients or of conspecific shrimp inhabiting the same cleaning station. Wicksten 33 suggests that chemosensory setae on shrimp antennae and dactyls could aid in client recognition, and that compounds found in fish mucus could have a pheromonal function 34 . Regardless, the findings from Chapuis and Bshary 10 demonstrate that complex cognitive abilities can evolve in a closely related shrimp species. Cognitive comparisons between shrimp and fish species within and between ocean basins would be needed to properly test the cognitive constraint hypothesis. The low cost of cheating hypothesis states that variation exists in the level of cost that a cleaner inflicts on a client when it cheats and removes scales or live tissue rather than parasites. If the cost of being cheated is great enough, selection should drive client strategies to enforce honest behaviors. Soares et al . 19 note that the size differences between the bluestreak cleaner wrasse L. dimidiatus (12 cm total length) vs Caribbean cleaner gobies (4 cm total length) may be great enough so that cheating by L. dimidiatus could inflict greater injuries on client fish, and thus, inflict a greater cost of cheating. This is yet to be measured, but comparative data from A. longicarpus should falsify this hypothesis in terms of its ability to explain differences between ocean basins. Cheating by A. longicarpus , a small and delicate palaemonid shrimp certainly cannot inflict the same types of cost on clients as the vertebrate jaws of Caribbean cleaner gobies. Finally, the foraging preference hypothesis states that differences in the behavioral complexity in cleaning symbioses between ocean basins could be driven by the food preference of the cleaner service providers. Here, the differences between Caribbean and Indo-Pacific cleaner fishes are easy to explain: Caribbean cleaner gobies prefer ectoparasites to client mucus whereas Indo-Pacific cleaner wrasse prefer client mucus to ectoparasites 17 , 21 . Thus, an inherent incentive to cheat exists for cleaner wrasse but does not for Caribbean cleaner gobies. The differences in food preference between clients means that all interactions with cleaner gobies should begin cooperatively with gobies foraging for parasites. Gobies would only resort to eating scales or fish tissue if parasite abundance is low. The explanation for the behavioral differences between the shrimps A. pedersoni in the Caribbean and A. longicarpus in the Indo-Pacific is less clear. The foraging preference hypothesis presents the simplest and most consistent explanation for the evolution of different behavioral cleaning strategies in both ocean basins. To our knowledge no studies have evaluated food preference in cleaner shrimp, nor conducted a gut content analysis of A. pedersoni . To summarize, research focused on cleaner shrimp food preference, gut content analyses, and the strategic behaviors of other cleaner species in the Caribbean and Indo-Pacific are needed, but at face value, it appears that cleaning strategies have evolved independently in both ocean basins and that the stability of these mutualisms is maintained in different ways. Advanced signaling, cheating behaviors, body and chelae morphology, hosting with sea anemones, and conspicuous dorsal patterning and coloration have all been noted as convergent properties in species of the genus Ancylomenes that have evolved dedicated cleaning lifestyles independently on both Indo-Pacific and Western Atlantic coral reefs 31 . Our data demonstrate that the behaviors that maintain these symbioses are decidedly different. Although on the whole the Caribbean appears to be a system where the two most prominent cleaner species do not vary their service qualities and clients do not aggressively punish cheating, there are some important differences in the responses of clients to incidents of cheating by gobies and by A. pedersoni . Soares et al . 19 studied the frequency of cheating in cleaner gobies in Barbados and found that cheating occurred in ~40% of all cleaning interactions, a value similar to our findings. However, more than 90% of all incidents of cheating by cleaner gobies resulted in the immediate termination of the interaction by the client 19 . Our data show that incidents of cheating by A. pedersoni rarely resulted in the client terminating a clean. Even making the most liberal interpretation of our data, only ~27% of all recorded jolts could have resulted in the termination of the interaction. In total, we recorded 163 cleaning interactions with at least 2 incidents of cheating, and 72 cleaning interactions with 3 or more incidents of cheating. Our data clearly demonstrate that clients repeatedly tolerate cheating (or behaviors that resemble cheating) by A. pedersoni . While the “low cost of cheating” hypothesis appears to fail to explain differences between ocean basins, it may provide a valid explanation for why clients tolerate cheating more at A. pedersoni stations than cleaner goby stations. An additional hypothesis for increased tolerance to cheating by A. pedersoni than by cleaner gobies is their provision of default “massages” during cleaning interactions 10 . Tactile stimulation has been shown to be important for the cleaner wrasse L. dimidiatus in manipulating clients into prolonging cleans, reconciling with clients after cheating, and as a pre-conflict strategy towards predators (e.g. 35 , 36 ), but cleaner shrimps may provide tactile stimulation inadvertently during an interaction 10 . Cleaner shrimps are in physical contact with their clients at all times, and shrimps walking across human hands provides a tickling sensation (e.g. 10 ; pers. obs.). Providing a similar sensation to clients could then prolong cleans. Further, it may be possible that tactile stimulation may make clients less bothered by cheats, or that these cheats are less noticeable because of the physical sensation of contact with cleaner shrimp. Cleaning interactions at A. pedersoni stations have been previously shown to be longer than those at cleaner goby stations, even when servicing the same clients (e.g. 26 ). Default massages may reduce stress hormones in clients (e.g. 37 ) providing alternative fitness benefits that may outweigh the cost of being cheated, and could partially explain why we see comparatively fewer clients terminate cleaning interactions immediately following a cheating occurrence at shrimp stations. One aspect of understanding the evolution of strategic service provisioning in cleaner shrimps that deserves more attention is the meaning of client fish jolts and whether it can be confirmed to be a true proxy of cheating. While jolts are largely accepted as a proxy for cheating in the cleaning literature, these studies have come exclusively from cleaner fish [e.g. 6 , 20 ]. In their study of A. longicarpus , Chapuis and Bshary 10 note that they cannot exclude the possibility that fish jolts in response to cleaning by shrimp do not correlate with the removal of live tissue or fish mucus, and that client jolts during cleaner shrimp interactions generated fewer responses than jolts caused by cleaner wrasse. A recent study by Vaughan et al . 38 further casts doubt that the correlation between jolts and cheating is as strong in cleaner shrimps as it is in cleaner fish. Using the Indo-Pacific cleaner shrimp Lysmata amboinensis and the reef fish Pseudanthias squammipinnis in a manipulative laboratory experiment, Vaughan et al . 38 showed that individual fish clients were responsible for 20x the variance in jolt rate than cleaner shrimp. Fish were moved to different aquaria for each treatment, and the same cleaner-client combination was never re-used. They found that only 10 of 54 P. squammipinnis individuals displayed repeated jolting and aggressive behavior towards cleaner shrimp. Although the study by Vaughan et al . 38 uses a different cleaner shrimp species in a non-natural setting, along with only one client species, it’s possible that individual client fish sensitivities are partially responsible for the jolting behaviors we have observed in our study. A targeted study analogous to those that have been conducted with cleaner fish is needed to determine whether non-parasitized clients jolt more frequently in the presence of cleaner shrimps than parasitized clients. In conclusion, our data present the first evidence that A. pedersoni , one of the most ecologically important cleaner species on Caribbean coral reefs, displays no evidence of strategic service behaviors. Because our study is directly comparable to those conducted for cleaner fishes and cleaner shrimp, our work enhances our knowledge of the stability of mutualisms and cleaning behavior on coral reefs, and demonstrates that ecologically analogous cleaning systems in separate ocean basins have evolved independent strategies to regulate and govern these complex behavioral interactions. Like prior research on Caribbean cleaner gobies, our study demonstrates that the cleaning symbiosis involving A. pedersoni do not meet all of the expectations set forth by game theoretical models. The fundamentally different ways that cleaning symbioses in the Caribbean have evolved have important implications for future research testing game theoretical models using cleaning systems. Namely, that there is no “standard” strategic service behavior common across cleaning mutualisms. Dozens of fish and crustacean species are classified as cleaner species. Each may have evolved unique behaviors that ultimately result in ecologically important parasite removal from reef fishes. Our understanding of these systems stands to benefit from future studies that explore cleaner shrimp food preference, the cognitive abilities of cleaner shrimps in both ocean basins, and the hormonal responses of client reef fish during cleaning interactions with cleaner shrimps."
} | 5,612 |
30872585 | PMC6418196 | pmc | 791 | {
"abstract": "Materials with in situ reversible wettability have attractive properties but remain a challenge to use since the inverse process of liquid spreading is normally energetically unfavorable. Here, we propose a general electrochemical strategy that enables the in situ reversible superwetting transition between underwater superoleophilicity and superoleophobicity by constructing a binary textured surface. Taking the copper/tin system as an example, the surface energy of the copper electrode can be lowered significantly by electrodeposited tin, and be brought back to the initial high-energy state as a result of dissolving tin by removing the potential. Tin atoms with the water depletion layer inhibit the formation of a hydrogen-bonding network, causing oil droplets to spread over the surface, while copper atoms, with a high affinity for hydroxyl groups, facilitate replacing the oil layer with the aqueous electrolyte. The concept is applicable to other systems, such as copper/lead, copper/antimony, gold/tin, gold/lead and gold/antimony, for both polar and nonpolar oils, representing a potentially useful class of switchable surfaces.",
"introduction": "Introduction Superwettable surfaces provide effective solutions to many academic and industrial problems 1 – 7 , e.g. controllability in printing or patterning 8 , 9 ; robusticity in antifogging, antifouling, or self-cleaning materials 10 , 11 ; and environmental protection in the petroleum industry 12 , 13 . Although many attempts have recently been made to trigger the wettability transition by applying light illumination 14 , 15 , temperature 16 , solvents 17 , electrical potential 18 , or pH stimulus 19 , the wettability transition between superphilicity and superphobicity usually requires an offline treatment, causing most of these approaches to be ex situ. Other technologies that allow an in situ wettability alternation, such as electrowetting 20 and electrochemical reduction/oxidation 21 , only work within a limited scale with a contact angle (CA) variation from ~50° to ~110° or from <10° to ~90° and remain unable to enjoy the aforementioned superwetting advantages. To date, in situ reversible superwetting transitions (i.e., CA changes between ~0° and ~180°) have been far from effective. Here, by using an electrochemical approach, we successfully realize the in situ reversible superwetting transition between a perfectly spherical-shaped droplet and a completely spread liquid film state by switching off and on a tiny potential.",
"discussion": "Discussion In summary, by constructing a binary Cu/Sn composite surface with a certain texture, we successfully realized the in situ reversible superwetting transition between a perfectly spherically shaped droplet (superphobicity) and a completely spread liquid film (superphilicity) upon the switching off and on of a potential. The electrodeposited Sn layer induces a water depletion layer by inhibiting the hydrogen-bonding network at the water/electrode interface, leading to the spreading of both polar and nonpolar oil droplets underwater. After removing the potential, the original Cu-enriched surface is gradually recovered as a result of dissolving Sn into the electrolyte as Sn 2+ and shows a high affinity to hydroxyls. The energy calculation demonstrates that all processes involved are energetically favorable. In particular, the as-prepared binary composite metal surface that enables the in situ reversible superwetting conversion is also applicable to other systems, such as Cu/Pb, Cu/Sb, Au/Sn, Au/ Pb, and Au/Sb. Here we propose a general electrochemical strategy for realizing the in situ reversible superwetting conversion, which might open an avenue for designing and fabricating various smart materials or devices for manipulating liquids in various applications, such as microfluidics; oil recovery in the petroleum industry; drop control in printing and patterning; and smart antifogging, antifouling; or self-cleaning materials."
} | 986 |
36594031 | PMC9804144 | pmc | 792 | {
"abstract": "Summary Biofilms are surface-attached multicellular microbial communities. Their genetics have been extensively studied, but the cell-scale morphogenetic events of their formation are largely unknown. Here, we recorded the entirety of morphogenesis in Escherichia coli , and discovered a previously unknown multicellular self-assembly process. Unattached, single-cells formed 4-cell rosettes which grew into constant-width chains. After ∼10 cell generations, these multicellular chains attached to surfaces and stopped growing. Chains remained clonal throughout morphogenesis. We showed that this process generates biofilms, which we found are composed of attached clonal chains, aligned in parallel. We investigated genetics of chain morphogenesis: Ag43 facilitates rosette formation and clonality; type-1 fimbriae and curli promote stability and configuration; and extracellular polysaccharide production facilitates attachment. Our study establishes that E. coli , a unicellular organism, can follow a multistage, clonal, genetically-regulated, rosette-initiated multicellular life cycle. These findings have implications for synthetic biology, multicellular development, and the treatment and prevention of bacterial diseases.",
"introduction": "Introduction Bacteria form surface-attached multicellular communities known as biofilms. 1 Biofilm cells resist toxic environments, escape immune responses, tolerate antibiotics, and contribute to chronic infections. 2 , 3 , 4 , 5 Their structure, cellular differentiation, and coordinated behavior suggest they could represent a bacterial equivalent of development in higher organisms. 6 However, the cell-scale morphogenetic events generating biofilms are not fully understood and are under ongoing investigation. 7 , 8 , 9 , 10 , 11 , 12 , 13 In Escherichia coli , genes required for biofilm formation have been identified, many of which encode adhesins that are understood in molecular detail ( Ag43 , type-1 fimbriae, curli, polysaccharides, and pili). 14 , 15 , 16 , 17 , 18 , 19 , 20 , 21 , 22 , 23 , 24 , 25 , 26 , 27 , 28 , 29 Type-1 fimbriae are rigid protein polymers that extend 2 μm from the cell surface. 30 , 31 Their adhesive tips bind to mannose groups and play a role in host-cell interactions, though their essentiality to biofilm formation on mannose-lacking surfaces suggests they perform cell-cell interactions as well. 32 Curli are extracellular fibers that facilitate cell-cell and cell-surface interactions and are a major component of the biofilm extracellular matrix. 16 , 33 Their function is thought to overlap with fimbriae, though their rigidity and adhesion differ. 34 Poly-β-1,6- N -acetyl-D-glucosamine (polyglucosamine) is a sugar polymer and the primary component of the extracellular matrix. 18 , 19 It is required for surface attachment during biofilm formation in E. coli , Staphylococcus epidermis , Staphylococcus aureus , and Pseudomonas fluorescens 18 , 19 and its production has a significant metabolic cost, 18 , 20 suggesting growth and attachment might be mutually exclusive behaviors. Antigen-43 (Ag43), encoded by flu , is a self-recognizing adhesion protein that coordinates cell-cell aggregation and has been implicated in E. coli biofilms. 22 , 23 , 24 , 25 , 26 It acts over short distances (∼10 nm from the cell membrane) and can cause auto aggregation when overexpressed. 22 , 23 , 24 , 25 , 26 Ag43 has also been shown to localize to the poles and allow sister-cells to remain adhered after cell division. 35 However, its expression and precise role in biofilm formation are debated. 22 , 23 , 24 , 25 , 26 Though all these adhesins are understood at the molecular-scale, their precise contributions to morphogenesis at the cell-scale are unclear. Deciphering the specific roles of biofilm genes will require tracking of the entirety of morphogenesis at the cell-scale, beginning from individual planktonic cells. Hydrodynamic conditions (fluid flow) are often used to study biofilms at the cellular scale as they allow continuous supply of nutrients and enhance biofilm formation in many species. 1 However, biofilm formation in commensal and lab K-12 E coli strains occurs in hydrostatic conditions, and is inhibited by fluid flow. 36 Moreover, genetic studies of biofilm formation have relied on standardized bulk-scale hydrostatic experiments. 15 , 18 , 19 , 21 , 24 , 37 , 38 , 39 Although robust, these methods cannot track cell-scale behavior.",
"discussion": "Discussion E. coli is one of the best studied unicellular organisms, though our understanding of its natural existence, in hosts and the environment, is incomplete. Here, we have uncovered a previously unknown multicellular morphogenesis in E. coli and demonstrated that it has reproducible, well-regulated growth and dynamic community shape properties ( Figure 1 ); is initiated by the formation of rosettes and is clonal ( Figures 1 and 2 ) similar to multicellular organisms; can self-propagate and reproduce new chains by the dissociation of cell clusters ( Videos S7 and S8 ); ends with surface attachment and growth arrest ( Figure 1 ); and has multiple stages ( Figure 1 ), each of which is temporally and dependently organized and is regulated by a specific genetic mechanism ( Figures 4 , 5 , and 6 ). Together, these findings suggest that E. coli can follow a multicellular life cycle that occurs at the microscopic scale. We show that this life cycle can generate the macroscopic communities known as biofilms ( Figures 2 , 3 , 4 , 5 , 6 , and 7 ) in E. coli K-12 (a model commensal strain), and in a “wild” strain isolated from a urinary tract infection ( Figure 7 ). The similar morphogenesis uncovered in these two strains, which were isolated 100 years apart, could suggest the intriguing possibility of a general multicellular life cycle in E. coli . Based on our findings, we propose a preliminary model for a multicellular life cycle in E. coli ( Figure 7 E). Single cells divide and remain adhered to create 4-cell rosettes. Communities grow longitudinally into chains hundreds-of-μm long while maintaining rosette configuration for ∼8 additional cell generations. Chains then attach to surfaces or to each other and stop elongating. Though attached chains do not elongate, they are not completely dormant, given the metabolic potentiation of antibiotics against biofilms. 53 , 54 , 55 Each morphogenetic stage relies on specific adhesins ( Figure 7 E): Ag43 facilitates sister-cell adhesion, rosette formation, and clonality; type-1-fimbriae and curli organize chain stability and structure after rosette formation, and polyglucosamine production attaches chains to surfaces. The temporal organization of these adhesins has a potential logic ( Figure 7 E): as morphogenesis progress, adhesin interaction distance increases and specificity decreases. Cells initially only adhere to sister cells ( Ag43 ), then they generate clonal communities (fimbriae and curli), and ultimately they attach to surfaces or other cellular communities (polyglucosamine). The insights we have uncovered into E. coli ’s morphogenesis could serve as a blueprint to engineer biofilms, design “living matter,” and create “synthetic morphogenesis”. 56 , 57 , 58 , 59 , 60 , 61 , 62 , 63 , 64 , 65 Rosettes are not generally observed in bacteria, with a few exceptions 66 , 67 , 68 including E. coli invasion of the host epithelial cells in UTI pathogenesis. Invasion of host cells is critical to UTIs, which are the most common type of infection and are caused by E. coli in 80% of cases. 69 The host cell cytoplasm is a hydrostatic environment and the reported rosette geometry 67 is equivalent to the 4-bacterial cell rosettes we have described. Within host cells, E. coli creates biofilm-like intracellular bacterial communities (IBCs) which express Ag43 and type-1 fimbriae, and produce polyglucosamine (and possibly curli) 69 despite lacking surface attachment and containing only hundreds of bacterial cells. IBCs are clonal 70 and their requirement for type-1 fimbriae is independent of host-cell surface attachment 32 , 71 indicating fimbriae organize the bacterial communities themselves. These parallels may indicate that E. coli IBC formation, chain morphogenesis, and biofilm formation result from the same underlying processes. Rosette formation is often considered a distinguishing feature between multicellular organisms and their unicellular relatives. 72 Hence, rosette-initiated morphogenesis in E. coli may represent an example of convergent evolution and an alternative model system for studying the principles of multicellular development. The significance of E. coli’ s multicellular life cycle is not fully clear, and many questions remain unanswered. For example, does E. coli follow this life cycle in its natural habitats: the mucus layer coating the large intestine (where it lives as a commensal organism) or on plants and soil (where it lives as a saprophyte)? Does this life cycle have advantages in such environments, which can be exposed to antibiotic treatment, phage attack, interspecies competition, 73 bacterial predators like amoeba and nematodes, 74 and dramatic fluctuations in hydration, temperature, and ultraviolet light? 75 There are also more immediate questions to resolve, such as: how do rosettes form, how is stage-specific gene expression regulated, and how does this life cycle relate to strains that form biofilms under fluid flow? 23 , 36 , 76 , 77 Addressing these questions will enhance our understanding of E. coli which, in addition to being the best-studied unicellular organism, is critical to human health and disease. Limitations of the study This study does not address biofilm formation in hydrodynamic conditions by adherent E. coli strains; does not directly measure the adherent force among cells in clonal chains; does not quantify the “breaking-away” phenomena during morphogenesis which could indicate the “reproductive rate” of multicellular chains; and does not directly observe the behavior of unattached chains within a growing biofilm or host environment itself."
} | 2,538 |
32680450 | null | s2 | 793 | {
"abstract": "Quorum sensing is a process in which bacteria secrete and sense a diffusible molecule, thereby enabling bacterial groups to coordinate their behavior in a density-dependent manner. Quorum sensing has evolved multiple times independently, utilizing different molecular pathways and signaling molecules. A common theme among many quorum-sensing families is their wide range of signaling diversity-different variants within a family code for different signal molecules with a cognate receptor specific to each variant. This pattern of vast allelic polymorphism raises several questions-How do different signaling variants interact with one another? How is this diversity maintained? And how did it come to exist in the first place? Here we argue that social interactions between signaling variants can explain the emergence and persistence of signaling diversity throughout evolution. Finally, we extend the discussion to include cases where multiple diverse systems work in concert in a single bacterium."
} | 250 |
36950108 | PMC10025973 | pmc | 794 | {
"abstract": "Summary Neuromorphic computing is a promising computing paradigm toward building next-generation artificial intelligence machines, in which diverse types of synaptic plasticity play an active role in information processing. Compared to long-term plasticity (LTP) forming the foundation of learning and memory, short-term plasticity (STP) is essential for critical computational functions. So far, the practical applications of LTP have been widely investigated, whereas the implementation of STP in hardware is still elusive. Here, we review the development of STP by bridging the physics in emerging devices and biological behaviors. We explore the computational functions of various STP in biology and review their recent progress. Finally, we discuss the main challenges of introducing STP into synaptic devices and offer the potential approaches to utilize STP to enrich systems’ capabilities. This review is expected to provide prospective ideas for implementing STP in emerging devices and may promote the construction of high-level neuromorphic machines.",
"introduction": "Introduction Building intelligent machines have long been a vision, especially with the advent of the intelligent era. The emergence of edge computing, big data, and other new technologies has significantly promoted the development of the intelligent era, but the growing volume of data and the demand for real-time processing of edge tasks pose a great challenge to conventional computers. The classical computing architecture separates the data storage from the central processing unit, generating “walls” that prevent further optimization of power consumption and computational speed, particularly in data-intensive tasks. The human brain is compact and unique for cognitive tasks and computing in memory with power dissipation as low as 20 W, whereas the Fugaku supercomputer requires 28 MW. Therefore, simulating the information processing of the human brain is an effective way to achieve high-speed and power-efficient data handling. The highly efficient biological nervous system originates from the huge connection network between neurons enabling the highly parallel processing ability of brain. 1 , 2 As the fundamental unit for signal transmission and regulation in the neural network, 3 synapses are considered to play essential roles in the realization of brain features, which are more than analog weights. 4 It is noted that synapses are leaky memories which possess different timescales and state parameters ruling their modifications. 5 , 6 On the other hand, synapses are also highly stochastic since the received spikes are transmitted in an uncertain manner. 7 Basically, the functional connections in neural networks are determined by the dynamic regulation of synaptic plasticity. 8 As a rule of thumb, synaptic plasticity can be simply divided into long-term plasticity (LTP) 9 , 10 , 11 and short-term plasticity (STP) 6 , 12 , 13 , 14 according to the timescale. LTP refers to the \"permanent\" change in synaptic connection strength, which is generally related to the formation of long-term memory underpinning the learning and memory functions. 15 On the contrary, STP denotes the temporary efficacy change (milliseconds to seconds) during calculation, possessing the time window in good agreement with many computing tasks, such as speech recognition, 16 information filter, 17 working memory, 18 and spatial orientation perception. 19 Therefore, STP provides a neural basis to achieve higher cognitive functions in the brain. LTP has been widely emulated both at device and system levels. 20 , 21 , 22 , 23 , 24 In detail, the synaptic devices with LTP characteristics are usually programmed into various weights to conduct vector-matrix multiplication in an array structure, resulting in the significant acceleration of the computing process in artificial neural networks (ANNs). In contrast, the implementation of STP remains limited in behavioral emulation at the device level, whereas the prototype at the system level is still unexplored. One of the key challenges is that the STP-correlated mature computational model and network algorithms are lacking, which severely hinders the system design. Complementary metal-oxide-semiconductor (CMOS) devices were initially used to achieve the temporary dynamic features in STP by exploiting their subthreshold characteristics. 25 , 26 , 27 For example, C. Bartolozzi et al. utilized diff-pair integrator synapse to reproduce the dynamics in STP, where the V thr is introduced to achieve the dynamic tuning of STP timescale. 28 Although STP emulation can be realized with CMOS devices, the artificial synapses suffer from severe challenges in energy- and power-efficiency since they generally consist of several transistors and bulky capacitors. Moreover, to further control the temporal dynamics in such synaptic unit flexibly, complex bias circuits are required to adjust the working state of the subthreshold circuit. 29 Compared to the traditional CMOS devices, the emerging devices with novel physical mechanisms, e.g. ion-migration, electron trapping, ferroelectric polarization, and magnetic skyrmion motion, show intriguing potential to emulate the dynamic behaviors of synapses in a single unit with less area and energy consumption. Figure 1 schematically illustrates the correspondence between emerging artificial synapses and biological synapses from the fundamental mechanisms to system applications. In biology, synaptic efficacy can be modulated by the interactions between Ca 2+ and neurotransmitters, kinases, and so forth. In contrast, the conductance in artificial synapses can be controlled by tuning the movement of ions or other physics, which simulates the modulation of synaptic efficacy. The hysteresis in the I−V curves together with the conductance increment under successive excitation indicates the memory effect, 30 while the spontaneous decay of conductance is consistent with the feature of short-term property. Attributing to the intrinsic dynamics processes, such as the migration of metal ions, the changed conductance simplifies the emulation of ion-induced STP in biological synapses. Devices based on different physical mechanisms can be used as artificial synapses with distinctive features, forming the basis for mimicking various advanced short-term synaptic behaviors, such as pair-pulse facilitation (PPF), pair-pulse depression (PPD), and post-tetanic potentiation (PTP). These features could in turn enable higher-order brain functions such as sound localization, 31 associative learning, 32 and working memory. 33 Thus, implementing STP with the emerging mechanisms is the foundation for pursuing neuromorphic computing with neuroscience principles, which needs further investigation. Figure 1 Functions of synaptic plasticity Top panel: schematic diagram of the STP correspondence between biological and emerging artificial synapses, from the basic mechanisms to system implementation. Bottom panel: the behavioral and functional differences between LTP and STP in biological synapses. This article reviews the current advances in the field of short-term artificial devices and their applications in neuromorphic computing. First, an overview of the synaptic bio-kinetic process and physics for implementing artificial synapses are presented. After that, recent advances in STP implementation using physical devices are summarized from the perspective of emulating synaptic behaviors. Next, we overview the realization of the STP system by using the emerging synaptic devices. Lastly, we discuss the challenges in developing short-term devices and propose potential routes. We expect that this review could provide fundamental guidance and inspiration for implementing STP in physics and system applications, thus toward a high-order neuromorphic machine.",
"discussion": "Discussion and outlook STP has an important computational role far beyond the current perception, which is promising to overcome the challenges in nowadays computing systems. The rich dynamics in emerging devices make it possible to implement STP in hardware with high efficiency. Nevertheless, the computing capabilities of the STP have not been fully developed. The above-mentioned emerging devices also need further optimization. For example, two-terminal memristors are favorable for high-density integration arising from their high area efficiency. The unit area can be as low as the ideal 4F 2 or 4F 2 /N in a 3D structure. Nevertheless, the sneak pass issue in the crossbar array requires extra selective devices, such as transistors, which hinder the expected integration density. Thus, the investigation of area-efficient two-terminal selectors is of great significance for the implementation of high-density memristor arrays with superior performance. On the other hand, the intrinsic switching stochasticity in two-terminal memristors poses a great challenge in precisely programming the synaptic weights in artificial synapses, which severely degrades the performance of the neural networks. Therefore, the development of bio-inspired algorithms utilizing stochasticity is in high demand for practical applications. Compared to two-terminal memristors, three-terminal EGTs combine the architecture superiority of separated control and transduction terminals, making it possible for precise control and multiple-terminal operation. 59 Moreover, the separated read/write operations of EGTs contribute to the emulation of the delayed release of neurotransmitters in biological synapses, which is important in synaptic computation. However, EGTs suffer from immature fabrication techniques and low integration density, which could be addressed through joint efforts of the material and electrical engineers. The time constant of the STP in magnetic devices is in nanoseconds, however, the devices suffer from physical implementation and controllability. Additionally, the comprehensive investigation of the synaptic dynamics in different devices is also emergent, since it is still difficult to fully apply the STP features in neuromorphic computing. We propose that the current challenges of STP exist in both the biological and physical aspects, which are discussed later in discussion. i) Brain is a very powerful information-processing center composed of huge population of neurons and synapses. Information encoding and decoding in the brain is achieved in a statistical manner, which can be adjusted in real-time according to environmental conditions for self-learning and adaptation, therefore contributing to the strong fault tolerance. 147 The continuous perception of external information and the integrated processing of self-stimuli are realized by impulse transmission in neural networks. However, so far people still do not have a comprehensive understanding of the spatiotemporal encoding of impulse signals, which is a major obstacle to the development of artificial intelligence. ii) Many behaviors and perceptions of organisms cannot be explained solely by short-range plasticity. The typical biological features including temporal-order learning, 148 , 149 , 150 , 151 information compression, and working memory are closely related to long-range plasticities such as STDP and neuronal population behavior., However, the coupling between short-term and long-term plasticity is complicated with mutual modulation between each other, which should be further explored. After the first demonstration of the LTP device, 152 numerous works have been reported to achieve single plasticity at the device level. The realization of LTP arrays 141 has facilitated the computation of convolution and accelerated neural networks. On the other hand, the development of STP computing is much slower. It is recently proposed that the mixed-timescale devices 100 are capable of improving the conventional network on temporal information processing. Nevertheless, the implementation of a single device to achieve tunable plasticity at multi-timescale so as to design integrated brain-like computational chips is our ultimate goal. The computing system will act like a brain with in-sense and in-memory computing capabilities in the future. The current issues that we need to address are. i) The coupling and decoupling of long- and short-term plasticity in neural networks is a complex process. Supports from dedicated algorithm engineers are in high demand. ii) Biological synapses are capable of performing multi-timescale plasticity depending on the environment. The design of the specific single device possessing both STP and LTP is expected to achieve more efficient in-memory computing. iii) STP computing lacks targeted application scenarios and algorithmic networks, which could be addressed through the close collaboration between the electronic and artificial intelligence engineers. Many of the underlying neuronal mechanisms are still obscure, which requires further investigation in an interdisciplinary approach through researchers’ cooperation in different fields such as brain science, electronics, physics, materials science, and computing science. Novel computing paradigms based on volatile devices should be further explored to accelerate existing AI computing systems. Integrating the unique properties of each device to build computational systems is one of the future paths for system designers. It is expected that the implementation of STP in artificial synaptic devices can significantly advance the development of neuromorphic computing."
} | 3,386 |
29713790 | PMC5959977 | pmc | 796 | {
"abstract": "The production of biogas by anaerobic digestion (AD) of agricultural residues, organic wastes, animal excrements, municipal sludge, and energy crops has a firm place in sustainable energy production and bio-economy strategies. Focusing on the microbial community involved in biomass conversion offers the opportunity to control and engineer the biogas process with the objective to optimize its efficiency. Taxonomic profiling of biogas producing communities by means of high-throughput 16S rRNA gene amplicon sequencing provided high-resolution insights into bacterial and archaeal structures of AD assemblages and their linkages to fed substrates and process parameters. Commonly, the bacterial phyla Firmicutes and Bacteroidetes appeared to dominate biogas communities in varying abundances depending on the apparent process conditions. Regarding the community of methanogenic Archaea , their diversity was mainly affected by the nature and composition of the substrates, availability of nutrients and ammonium/ammonia contents, but not by the temperature. It also appeared that a high proportion of 16S rRNA sequences can only be classified on higher taxonomic ranks indicating that many community members and their participation in AD within functional networks are still unknown. Although cultivation-based approaches to isolate microorganisms from biogas fermentation samples yielded hundreds of novel species and strains, this approach intrinsically is limited to the cultivable fraction of the community. To obtain genome sequence information of non-cultivable biogas community members, metagenome sequencing including assembly and binning strategies was highly valuable. Corresponding research has led to the compilation of hundreds of metagenome-assembled genomes (MAGs) frequently representing novel taxa whose metabolism and lifestyle could be reconstructed based on nucleotide sequence information. In contrast to metagenome analyses revealing the genetic potential of microbial communities, metatranscriptome sequencing provided insights into the metabolically active community. Taking advantage of genome sequence information, transcriptional activities were evaluated considering the microorganism’s genetic background. Metaproteome studies uncovered enzyme profiles expressed by biogas community members. Enzymes involved in cellulose and hemicellulose decomposition and utilization of other complex biopolymers were identified. Future studies on biogas functional microbial networks will increasingly involve integrated multi-omics analyses evaluating metagenome, transcriptome, proteome, and metabolome datasets. Electronic supplementary material The online version of this article (10.1007/s00253-018-8976-7) contains supplementary material, which is available to authorized users.",
"introduction": "Introduction In Europe, flammable swamp gas is presumably known since the roman period. Microbial anaerobic digestion (AD) of plant material as origin of burnable biogas was scientifically recognized and analyzed in detail since mid of the twentieth century. Complex microbial consortia are responsible for successive degradation of organic biomass to biogas consisting of methane (CH 4 ) and carbon dioxide (CO 2 ) and, in smaller proportions, of other gases. In industrial-scale biogas plants (BGPs), biogas is produced by AD from agriculturally produced renewable resources such as maize, grass, and sugar beet, and even biodegradable organic wastes can be used as substrates (Weiland 2010 ; Zhang et al. 2016 ). Today, biogas production considerably contributes to the recovery of energy from renewable resources thereby also positively affecting the balance of the climate-relevant gas carbon dioxide. In Germany, nearly 9000 biogas plants (including about 450 bio-waste biogas plants), with 4.5 Giga-Watt installed electric power and about 1 Giga-Watt thermal energy usage, are operated (FNR 2017 ). The biomethanation process is formally subdivided into four phases, i.e., (i) hydrolysis/cellulolysis of complex organic compounds, namely carbohydrates, proteins and lipids, towards corresponding oligomers and monomers, (ii) acidogenesis (fermentation) of the latter metabolites to the intermediates propionate, butyrate, other short-chain volatile fatty acids (VFA) and alcohols, (iii) acetogenesis of primary fermentation products to acetic acid, CO 2 , and H 2 , and (iv) methanogenesis resulting in CH 4 and CO 2 (Angenent et al. 2004 ). The first three phases are solely performed by fermentative Bacteria. Only certain methanogenic Archaea are able to synthesize CH 4 from the end-products of bacterial fermentation. Despite the technical improvements of anaerobic wastewater treatment in the beginning and mid of the twentieth century, for a long period, anaerobic and biomass degrading microorganisms were regarded as a “relatively unimportant group of organisms” (McBee 1950 ). However, their potential for the development of economically valuable bioconversion processes utilizing cellulosic or organic waste materials for production of valuable end-products was recognized. First attempts to study microbial biogas communities relied on cultivation-based approaches and started in the beginning of the twentieth century (e.g., Schnellen 1947 ; McBee 1954 ) resulting in over 150 newly discovered species of microorganisms (Söhngen et al. 2016 ). Technical improvements concerning anaerobic cultivation of microorganisms are still leading to the isolation and characterization of new type strains for hydrolytic/acidogenic Bacteria such as Clostridium bornimense , Herbinix hemicellulosilytica , Herbinix luporum , Herbivorax saccincola, Proteiniphilum saccharofermentans , Petrimonas mucosa , Fermentimonas caenicola, and Proteiniborus indolifex (Hahnke et al. 2014 ; Koeck et al. 2015 ; Hahnke et al. 2016 ; Koeck et al. 2016a ; Koeck et al. 2016b ; Hahnke et al. 2018 ). Likewise, new species for methanogenic Archaea such as Methanobacterium aggregans or Methanosarcina flavescens were described (Kern et al. 2015 , 2016 ). However, the options of cultivation-based approaches to uncover all members of biogas communities are intrinsically highly limited. Thus, cultivation-independent methods are indispensable to tackle the whole complexity of biogas communities. Metagenomics gained in importance for the dissection of microbial assemblages in the same way as the performance and efficiency of next-generation sequencing (NGS) technologies were advanced. Inevitable, anaerobic digestion communities were elucidated by applying methods of metagenome, genome and post-genome research taking advantage of high-throughput sequencing of environmental whole community DNA and RNA. In the present review, state-of-the-art metagenomics approaches are presented to illustrate their usefulness in analyses tackling microbial communities residing in BGPs. To elucidate the metabolically active biogas community, metatranscriptome, metaproteome, and metabolome studies are addressed in an integrated manner. It is commonly accepted that biogas producing microbial communities are the key for process shaping and development of optimization strategies since they provide opportunities for their management and engineering (Carballa et al. 2015 ; Koch et al. 2014 ). Accordingly, this article reviews current knowledge on structure and performance of microbial communities residing in full-scale biogas-producing reactors considering microbiome management and monitoring options. Previous reviews addressing microbial communities of anaerobic digestion systems were mainly focused on laboratory-scaled reactors and classical molecular methods to uncover community compositions and functions (e.g., Demirel and Scherer 2008 ; Nasir et al. 2012 ; Čater et al. 2013 ; Venkiteshwaran et al. 2015 ; Schnürer 2016 ; Demirel 2014 ; Vanwonterghem et al. 2014 ). In the present review, microbial communities of BGPs converting agriculturally produced renewable primary products, organic residues, and manure were considered. Laboratory systems were only included if they feature pilot character exemplifying fundamental methodological approaches or important insights in microbial community structure and functionality."
} | 2,067 |
33614094 | PMC7890512 | pmc | 797 | {
"abstract": "Social insects are one of the best examples of complex self-organized systems exhibiting task allocation. How task allocation is achieved is the most fascinating question in behavioural ecology and complex systems science. However, it is difficult to comprehensively characterize task allocation patterns due to behavioural complexity, such as the individual variation, context dependency and chronological variation. Thus, it is imperative to quantify individual behaviours and integrate them into colony levels. Here, we applied bipartite network analyses to characterize individual-behaviour relationships. We recorded the behaviours of all individuals with verified age in ant colonies and analysed the individual-behaviour relationship at the individual, module and network levels. Bipartite network analysis successfully detected the module structures, illustrating that certain individuals performed a subset of behaviours (i.e. task groups). We confirmed age polyethism by comparing age between modules. Additionally, to test the daily rhythm of the executed tasks, the data were partitioned between daytime and nighttime, and a bipartite network was re-constructed. This analysis supported that there was no daily rhythm in the tasks performed. These findings suggested that bipartite network analyses could untangle complex task allocation patterns and provide insights into understanding the division of labour.",
"introduction": "1. Introduction How the simple elements at lower levels can evolve to higher-level systems—e.g. the evolution of solitary individuals into social animal groups—through natural selection has been a central question in evolutionary biology [ 1 , 2 ]. Such systems often exhibit the division of labour through efficient task allocation, in which different elements perform different tasks [ 3 ]. It has been considered that the division of labour can be a primary advantage for social evolution [ 4 – 6 ]. Therefore, it is crucial to understand how the division of labour is achieved. Social insects are one of the most sophisticated examples of division of labour as individuals exhibit reproductive division of labour between the queen(s) and workers, and various tasks are allocated among workers [ 5 , 7 – 9 ]. Individuals of many insect colonies are often specialized, so that workers engage in nursing, foraging, nest defence or cleaning. Task allocation is accomplished without any central control or hierarchal control [ 3 , 6 , 10 , 11 ]. Instead, local cues, such as interactions with others, play a key role [ 9 , 12 – 14 ]. Although an understanding of the proximate causes and ultimate consequences of task allocation is crucial, it is difficult to describe individual behaviours and quantify the task allocation patterns. This difficulty arises from the fact that there are a large number of individuals with different characteristics, such as morphology, age and spatial position in a colony [ 15 – 18 ]. Also, individual workers generally show flexibility in task performance [ 19 – 22 ]. Additionally, some typical behaviours can be decomposed into smaller components. For instance, brood care can be divided into nursing eggs, larvae and pupae. Therefore, methods to extract useful information from such complexity (i.e. many elements and high variation) and to integrate them to understand the complex social systems are strongly required. Over the past two decades, network analysis has developed as a useful tool for analysing complex systems in various fields, including ecology, social science and animal behaviour [ 23 – 33 ]. The network perspective assumes that systems comprise nodes (i.e. elements) and links (i.e. the relationship between elements). The quantification of networks makes it possible to provide information on the structure of systems or characteristics of each individual within the network. Previous studies have revealed the structures of ant interaction networks [ 30 , 34 , 35 ], and there are some functional units within a colony that are based on tasks, such as foraging and brood care [ 17 ]. Additionally, it has been shown that these networks underlie the spreading dynamics of information, food or disease in a colony [ 25 , 31 – 34 , 36 , 37 ]. The described examples suggest that network analyses can provide information about the characteristics at various scales, from the individual to the colony level, and may provide new insights into understanding social insects. Most network analyses in animal societies show relationships among the individuals within a group [ 17 , 35 ], although they can be applied to any elements and relationships. Previous studies have suggested that the task allocation patterns in social insects can be described as a bipartite network ( figure 1 ), which has two node classes, and links are established only between nodes in different classes [ 38 , 39 ]. Workers are connected to the tasks they are engaged in, and tasks are linked to the workers who perform them [ 38 , 39 ]. Pasquaretta & Jeanson [ 39 ] quantified the degree of division of labour at the individual and colony levels by using an information theory approach. Additionally, they demonstrated that a community detection method for bipartite networks is an effective approach to determine the clusters of individuals that are engaged in similar subsets of tasks. However, bipartite network analyses have greater potential to quantify the task allocation patterns in terms of various levels (i) network level, (ii) cluster level and (iii) individual level. Moreover, the analysis also makes it possible (iv) to combine the properties in the network structure with biological features (e.g. diurnal activity, age). Here, we applied the following four new methods to characterize individual-behaviour relationships.\n Figure 1. Individual-behaviour relationship as a bipartite network. An adjacency matrix for individual behaviour ( a ) can be described as a bipartite network with weights and information on day ( b ). The thickness of the links represents the weight of the interactions. Module structures can be detected by maximizing modularity ( c ). There were specialized individual-behaviour relationships in each module. The links between the modules represent the extent to which individuals within a module exhibited behaviour in the other modules. First, we applied nestedness to quantify the overlapping structure of the bipartite network as the index of network level [ 40 , 41 ]. In theoretical studies, the nested structure is generally related to the stability and persistence of the complex system [ 42 – 44 ]. In individual-behaviour networks, we might observe the overlapping structure of executed tasks because the overlapping structure reflects that generalists, which can emerge in small groups in particular [ 4 ], complement tasks in a colony [ 45 , 46 ]. When specialists that perform a certain task are lost unexpectedly in a colony with high nestedness, the generalists can supplement the task, leading to avoidance of critical function loss at the colony level. Therefore, nestedness could be a useful index for describing the robustness of colonies. Second, we evaluated a module-module network (hereafter module network) to quantify the relationships between task groups. Although this idea of task networks was proposed approximately two decades ago [ 9 , 21 , 47 ], bipartite network analysis would offer a refined approach to construct task networks. As mentioned earlier, some behaviours can be decomposed into smaller components. Therefore, it is imperative to integrate such behavioural components into task groups. Once we identify task groups, we can consider the groups as new nodes at the module network level. This approach can quantify the relationship between task groups as a module network, that is, how task groups (modules) connect with each other ( figure 1 c ). For example, workers' tasks could change from the order of brood-care to forage [ 17 , 48 ]. Thus, workers rarely execute both brood care and forage. In this case, the module of brood care may be less connected with that of forage. Third, we quantified the degree of specialization of each individual with consideration of module structures based on the idea that nodes with the same role should have similar topological properties [ 49 , 50 ]. Although previous studies on the division of labour have analysed the degree of specialization based only on the ratio of executed behaviours (e.g. Shannon entropy, d' , DOL ) [ 39 , 51 ], the group structures were not incorporated into these indices. To consider the topological property for the degree of specialization, we applied the c -score, which was defined as among-module connectivity, (i.e. the level to which the individuals were linked to other modules) to evaluate the degree of specialization at the individual level. A low c -score indicates a high degree of specialization in the task group, and vice versa. Finally, we proposed how to integrate other parameters with the network metrics. For example, individual characteristics, such as age or body size, are crucial for understanding the factors underlying task allocation [ 5 ]. Therefore, we investigated whether similarly aged individuals belonged to the same task groups. Additionally, chronological information can be incorporated in the analyses (e.g. year, season and time of day). Here, to investigate the daily rhythms of tasks, daytime (9:00–18:00) and night-time (21:00–6:00) data were separately analysed, and the network structures were compared between daytime and nighttime. In the present study, we observed all colonial individuals with verified age in a ponerine ant ( Diacamma sp.) and analysed individual-behaviour networks. We addressed three specific questions that are central to the study of colony organization in social insects: (i) What is the structure of individual-behaviour networks? (ii) How are the task groups distributed among workers? (iii) Does the age of the individual or the time of day affect the task group pattern? This study further applied network analysis to individual-behaviour networks, leading to a comprehensive approach to understanding task allocation in complex social systems.",
"discussion": "4. Discussion To understand the complex task allocation patterns in insect societies, network analysis is helpful for quantifying the characteristics of task allocation. In this study, we recorded the behaviours of all individuals in five ant colonies of Diacamma sp. and investigated the individual-behaviour networks using the bipartite network approach. We found a non-nested structure and determined the characteristics of module networks, including the inactive groups. Moreover, we detected consistent module structures in classifying task groups throughout a day (i.e. no daily cycle) and three consecutive days (i.e. no daily changes). Network analysis can summarize the task allocation patterns as various values of a network-level index. The result of a significantly large modularity suggested the presence of the module structures, which were comprised of some individuals and a few behaviours. However, the normalized modularity largely deviated from 1, suggesting that there were no complete specializations and that individuals were loosely linked to each behaviour. We found no significant nested structure in most colonies ( table 1 ), suggesting that a specific individual did not cover the tasks that other individuals executed. Interestingly, the smallest colony had a significant nested structure ( table 1 ). This structure may reflect the notion that generalists can emerge to accomplish all tasks when colony size is small [ 4 ]. It is possible that young (small) colonies are vulnerable to disturbance. If a specialized individual is dead, the task cannot be executed, which can lead to the critical function loss. Thus, a high nestedness structure might be a reasonable strategy for maintaining a colony's productivity. Conversely, the benefits of specialization may offset the decline in robustness in large colonies. Further studies are needed in order to investigate the relationship between the network structure (i.e. nestedness) and the colony size [ 18 ]. The number of tasks within a colony is one of the fundamental factors in task allocation research. While many studies have focused on only a few prominent tasks, such as foraging, building and brood care (e.g. [ 11 , 65 , 66 ]), high specialization has been observed in other tasks, such as guarding [ 48 , 67 ], grooming [ 68 , 69 ] and inactive [ 70 ]. At this point, how many tasks exist in insect societies is still not completely understood. This seemingly simple question is not easy to answer because the demand for tasks can change depending on the context. A community detection method can be valid for the classification of task groups (i.e. modules) [ 39 ]. In this study, we found that the number of modules differed among colonies ( figure 3 ). We also found not only the module of forage and brood care but also the module of walk and inactive. It suggested that the inactive state could be categorized as one of the task groups, which was consistent with the previous study [ 70 ]. In our definition, walking included several behaviours, such as grooming, trophallaxis and dominance behaviours. Therefore, it is possible that there were several task groups in the ‘walk' module. A more detailed observation should be incorporated in future studies. Other clustering methods (e.g. hierarchical clustering analysis) can also classify workers into separate groups [ 70 ]. However, there are at least two advantages of maximizing modularity instead of other methods. First, we could determine the number of modules based on the modularity criteria. Second, comparing empirical networks with null models, which are constructed from a randomly shuffled adjacency matrix, made it possible to infer the behavioural rules underlying the empirical task allocation patterns. In our analysis, we used the null model in which the behaviour of each individual was executed as a proportion of the total number of executed behaviours (see method 2.5). In other words, each individual chooses a behaviour depending only on the work demands. The results showed that there were significant differences between the empirical and null networks ( table 1 , modularity). Therefore, we confirmed that individual ants determined their behaviour depending not only on the task demands but also on other factors, including internal states, experiences or interactions with other individuals. We quantified the c -score, which was the degree of specialization based on module structures obtained through community detection. Using this approach, we evaluated individual roles with respect to their position within and among modules. Previous studies have quantified the degree of specialization from the original weighted adjacency matrix (e.g. d ' [ 59 ] and DOL [ 51 ]). These metrics do not take the relationships that connect across modules into account. As modules are considered as functional task groups if modules are not taken into account, important properties could be overlooked. Some complete specialists with a c -score = 0 have d ' indiv < 1 (electronic supplementary material, figure S2). When individuals in a module engage in only tasks in the module, the c -score becomes zero. Hence, the c -score reflects the degree of specialization in a functional module. For example, the c -score of an individual in a nursing module represents how much the individual specializes in a function, including nursing eggs, larva and pupa. By contrast, the value d ' represents the degree of specialization for each behaviour. These metrics provided different information about the level of specialization. Here, we would like to emphasize that the combination of several metrics is needed to understand the whole aspect of task allocation at the individual level, as pointed out by a previous study [ 39 ]. Once we obtained individual-level metrics, it was possible to answer further questions, such as what kind of individuals belonged to the module by comparing the scores between modules ( figure 4 ) and what the relationship was with other individual traits, such as age. Additionally, this method enabled us to perform inter-species as well as inter-colony comparisons. As shown in figure 2 , since task allocation patterns were complex, it would be difficult to compare them without such metrics. Behaviour can change with time, such as daily rhythms. We analysed the daily changes by dividing the data into daytime and nighttime (electronic supplementary material, figure S3). We found that there was no daytime and nighttime module structure, suggesting that the task allocation pattern was consistent throughout the day. Interestingly, solitary ants show circadian activity rhythms even under constant light conditions [ 61 – 63 ]. Previous studies have shown that nurse ants take care of the brood throughout the day [ 61 , 62 ]. Moreover, Diacamma sp. is known to occasionally forage at night according to field observations [ 71 ]. Foragers might be active during the night as long as the temperature is suitable for foragers. Our results supported these previous findings by providing evidence of a non-circadian pattern in task execution. In addition to the daily cycle, temporal information, such as their experiences and previous state, might affect decision making in individuals. This assumption might be addressed in future studies, exploring the temporal network analyses that have ordered temporal data. For social insect colonies, an increasing number of studies have indicated that worker inactivity is common (reviewed in [ 72 ]). In our focal ant, some individuals specialized in inactivity ( figure 3 ). Although the existence of inactive individuals has been documented across many social insect species, we know very little about the variations between individuals. The simplest hypothesis for inactive (lazy) ants is immaturity or senescence (proposed by [ 73 , 74 ]). Young workers may be less active due to their still-developing physiology; on the other hand, older workers may be less active due to degraded physiology. Here, we tested this hypothesis. Our data showed that young workers tended to be inactive; however, there were differences between colonies. Although age could be related to inactivity, our results suggested that immaturity was related to inactivity; however, ant inactivity was also colony dependent. Assessing how the quantified patterns emerged based on decision making in individuals is our next question for understanding task allocation. Previous studies have shown that the interactions between individuals and division of labour can have substantial interplay [ 14 , 75 ]. Thus, it is imperative to integrate individual-behaviour networks revealed here into the interaction networks between individuals. Our results on the module networks implied that not only the interactions between individuals within a module but also the interactions between individuals in different modules with significant links were important. Further studies are needed to examine how interactions affect individual decision making and subsequent task allocation patterns. One possible way is to analyse mathematical models to construct the patterns of the networks from rules at individual levels [ 3 , 75 , 76 ]. The network perspective can capture complex phenomena composed of many elements and extract important information at multiple levels. The individual-task relationships in mammal and even human societies [ 77 ] are crucial for understanding the division of labour in complex societies. The methodology we introduced here may be applied to a wide range of individual-behaviour networks."
} | 4,966 |
37242908 | PMC10223960 | pmc | 798 | {
"abstract": "The extensive utilization of traditional petroleum-based plastics has resulted in significant damage to the natural environment and ecological systems, highlighting the urgent need for sustainable alternatives. Polyhydroxyalkanoates (PHAs) have emerged as promising bioplastics that can compete with petroleum-based plastics. However, their production technology currently faces several challenges, primarily focused on high costs. Cell-free biotechnologies have shown significant potential for PHA production; however, despite recent progress, several challenges still need to be overcome. In this review, we focus on the status of cell-free PHA synthesis and compare it with microbial cell-based PHA synthesis in terms of advantages and drawbacks. Finally, we present prospects for the development of cell-free PHA synthesis.",
"introduction": "1. Introduction Traditional petroleum-based plastics possess excellent material properties, including lightweight, stability, durability, economic feasibility, and other desirable material characteristics. These features enable them to be synthesized and processed into various strength materials and molded into different shapes. These properties have made them an attractive option for utilization in construction, packaging materials, computer equipment, automotive components, medical devices, and other fields [ 1 , 2 , 3 , 4 , 5 , 6 , 7 ]. However, the extensive use of petroleum-based plastics has resulted in serious pollution to land and sea over the past few decades as large-scale production and inappropriate waste management strategies have inevitably negatively impacted the environment [ 8 ]. Thus, there is an urgent need for wise alternatives to using petroleum derivatives to reduce dependency on finite fossil resources while limiting pollution and CO 2 emissions [ 9 ]. One attractive solution to this issue is scaling up the production and application of microbial-synthesized bioplastic polyhydroxyalkanoates (PHAs). PHAs are biobased materials that can be produced by approximately 40% of prokaryotic strains that are capable of accumulating PHA biopolymers [ 10 , 11 ]. The bacterial stains primarily responsible for PHA production, including Ralstonia eutropha [ 12 ] and Pseudomonas sp. [ 13 ], have demonstrated the ability to synthesize PHAs during the nongrowth stage with minimal nutrient requirements [ 14 ]. However, it is worth noting that natural PHA producers typically exhibit lower yields [ 15 ]. Over the past decades, developments in industrial biotechnology and metabolic engineering have demonstrated tremendous potential in exploiting different PHA biosynthetic host strains via natural or heterologous metabolic pathways to enhance intracellular PHA content. While the numbers of successful cases continues to grow [ 13 , 16 ], only a small proportion of the resulting products are economically viable for large-scale production. As such, considerable effort is still required to bring them to market. The complexity of the metabolic system in the cell factory poses the greatest challenge to PHA synthesis. It is very hard to balance the intracellular flux in the best way to satisfy the target synthetic pathway while sustaining the host’s needs for growth and reproduction [ 17 ]. This leads to a series of challenges, including low yield and conversion, high costs of raw materials and energy, and poor material properties. The crucial problem lies in the fact that the current state-of-the-art technology is unable to cater to the demands of both mass production and tailored material properties. Additionally, the separation or postprocessing of PHA products is constrained by the properties of the PHA product itself and the bacterial cell wall. The scaling up of PHA synthesis from laboratory to industrial levels requires careful consideration of carbon source cost issues and the feasibility of fermentation conditions [ 18 ]. The intracellular synthesis of PHA with cofactors (e.g., ATP, NAD + , and CoA) requires a substantial amount of energy, leading to an imbalance in resource allocation (e.g., electron, CoA, and ATP fluxes) and limiting the feasibility of achieving high PHA yields with the complete synthesis of the target product. Consequently, low volumetric productivity may be observed. In addition, the intricate nature of cellular regulation can impede the implementation of high-yield pathways designed for intracellular PHA synthesis, making it difficult to assign metabolic flux according to the design [ 19 , 20 ]. To address the challenges faced in microbial cell-based polyhydroxyalkanoate (PHA) synthesis, several strategies have been vigorously developed. These include the development of newly engineered strains [ 21 , 22 ], the utilization of inexpensive substrates [ 14 ], and improved extraction methods [ 23 ]. Despite these efforts, the high-cost bottleneck in PHA synthesis has not yet been substantially overcome. As an alternative approach, cell-free systems have rapidly emerged in the synthesis of PHAs and other products. By disconnecting PHA synthesis from the generation of bacterial biomass, in vitro systems can overcome the challenges associated with microbial cell-based PHA synthesis. Through the reconstruction of the PHA synthesis pathway using cell lysates or purified enzymes, in vitro systems allow for precise detection of metabolite concentration, regulation of enzyme activity, and the control of flux distribution [ 24 , 25 ]. In recent years, cell-free synthesis has also emerged in combination with energetic and cofactor regeneration techniques such as electric or light energy-based methods. In addition, a range of non-natural pathways for PHA and its precursors have been shown to be compatible with in vitro systems. In this review, we focus on the current status of in vitro PHA synthesis, comparing its advantages and drawbacks to microbial cell-based PHA synthesis, and offer perspectives and outlooks on cell-free PHA synthesis."
} | 1,493 |
21541019 | PMC3081813 | pmc | 799 | {
"abstract": "Division of labour (DoL) is a fundamental organisational principle in human\nsocieties, within virtual and robotic swarms and at all levels of biological\norganisation. DoL reaches a pinnacle in the insect societies where the most\nwidely used model is based on variation in response thresholds among\nindividuals, and the assumption that individuals and stimuli are well-mixed.\nHere, we present a spatially explicit model of DoL. Our model is inspired by\nPierre de Gennes' 'Ant in a Labyrinth' which laid the foundations\nof an entire new field in statistical mechanics. We demonstrate the emergence,\neven in a simplified one-dimensional model, of a spatial patterning of\nindividuals and a right-skewed activity distribution, both of which are\ncharacteristics of division of labour in animal societies. We then show using a\ntwo-dimensional model that the work done by an individual within an activity\nbout is a sigmoidal function of its response threshold. Furthermore, there is an\ninverse relationship between the overall stimulus level and the skewness of the\nactivity distribution. Therefore, the difference in the amount of work done by\ntwo individuals with different thresholds increases as the overall stimulus\nlevel decreases. Indeed, spatial fluctuations of task stimuli are minimised at\nthese low stimulus levels. Hence, the more unequally labour is divided amongst\nindividuals, the greater the ability of the colony to maintain homeostasis.\nFinally, we show that the non-random spatial distribution of individuals within\nbiological and social systems could be caused by indirect (stigmergic)\ninteractions, rather than direct agent-to-agent interactions. Our model links\nthe principle of DoL with principles in the statistical mechanics and provides\ntestable hypotheses for future experiments.",
"introduction": "Introduction Both human and animal societies display a division of labour, in which there may be\nan unequal distribution of effort between or within particular tasks, according to\nage or experience [1] , [2] , sex [3] , physiology [4] or morphology [5] . Such\nspecialisation has long been known to improve collective productivity [6] because learning\nallows individuals that focus on a subset of tasks to perform more efficiently than\ngeneralists (note however the exception to the rule provided by Dornhaus, 2008).\nDivision of labour is most advanced in the societies of insects such as ants, bees,\nwasps and termites [7] . Within an insect society, there is typically considerable\nindividual variation in the sensitivity to stimuli associated with particular tasks.\nOne of the simplest models of Division of Labour (DoL), the fixed-threshold model\n(FTM), invokes this individual variation in sensitivity to such task-related stimuli\n [8] , [9] . There is good\nevidence for the existence of such response thresholds in ants [10] , [11] , [12] , bumblebees [13] , the honey\nbee [14] , [15] , [16] , wasps [17] and termites\n [18] .\nExperiments also provide strong support for the role of response thresholds for the\nmaintenance of colony homeostasis [13] , [19] , [20] . Individual variation in thresholds has genetic [21] , [22] ,\nmorphological [11] , hormonal [23] and developmental [24] \ncomponents. Although direct evidence for a positive relationship between colony\nfitness and wide threshold distributions is lacking, there is evidence in the honey\nbee that genetic variation (the number of patrilines within the colony) positively\ninfluences colony fitness [25] . In the FTM, the decision of an individual whether or not to undertake a particular\ntask, such as foraging or brood care, is determined by two parameters; the\nsensitivity of the individual to stimuli associated with the task (its response\nthreshold), and the level of demand for that task (the stimulus value). When an\nindividual senses that the stimulus exceeds its threshold value, it becomes\nactivated, and performs some work. Through such activity, sensitive (low threshold)\nindividuals reduce the stimulus level such that it often does not reach the\nthreshold of their less sensitive nestmates. This negative feedback loop\nhomeostatically maintains the stimulus level (the task demand) at a steady state,\naround which it fluctuates. A further consequence of this mechanism, and one that\nmatches the pattern observed in nature, is that the activity distribution becomes\nright-skewed; a small minority of sensitive individuals perform the majority of the\nwork [13] , [26] , [27] , [28] , [29] . Here, we extend the FTM by explicitly including space. This modification induces a\nspatial 'percolation' effect [30] , [31] , [32] in which small differences\namongst agents in their response thresholds, are related to large differences in\ntheir probabilities of performing work. The extension of the original\nfixed-threshold models of DoL to include space removes the assumptions that\nindividuals and task-associated stimuli are well-mixed. The\nmovement and activity of the individuals in the spatial fixed-threshold model (SFTM)\nmay then be analysed as a case of diffusion in disordered media - a well studied\nbranch of statistical mechanics [33] , [34] . The FTM assumes that individuals and stimuli are well-mixed and that each individual\nexperiences the same global stimulus level equally. This is a simplifying\nassumption. However, it is realistic only for a minority of cases when the stimulus\nis spatially uniform. For example, honey bees homeostatically maintain the nest\ntemperature and CO 2 levels within certain acceptable ranges [13] , [20] . When it\ngets too hot inside the nest, the bees with the lowest threshold to temperature\nbegin to fan their wings, thereby increasing the airflow and reducing the\ntemperature such that it never reaches the thresholds of their less sensitive\nnestmates. So because temperature and CO 2 levels can be expected to be\nfairly uniform within the nest, the assumption of perfect mixing of stimulus and\nbees is justified. Therefore modelling this process as a non-spatial process is\nreasonable. However, when the stimulus in question is heterogeneous over space,\nperfect mixing can no longer be assumed. To appreciate the importance of modelling\nDoL without the assumption of perfect mixing, consider the honey bee comb,\norganised- or rather compartmentalised- into different zones in which the cells\ncontain either brood, pollen or honey [35] . Therefore, tasks are not\nuniformly distributed in space [36] . Furthermore, individuals themselves are not well mixed.\nDespite their high potential mobility, individual ants [37] , honey bees [38] , bumble bees\n [39] and wasps\n [40] tend to\nbe faithful to particular parts of the nest and this spatial fidelity persists even\nwhen many tasks are removed [41] . In the FTM, the distribution of the individual response thresholds\nwithin the colony- the Colony Threshold Distribution (CTD)- will bear directly upon\nthe proportion of individuals that are mobilised to respond to a given stimulus\nlevel. Indeed, the precise form of the CTD will have significant adaptive\nconsequences [42] . Abrupt discontinuities in the CTD would affect the ability\nof the colony to produce an appropriate response to small changes in the demand for\nlabour. For example, consider the scenario in which the colony is evenly split\nbetween two types of individual; half of the ants have low thresholds and the other\nhalf have high thresholds. In that case, the colony will be unable to produce a\ngraded response to fluctuating stimulus levels, because only 0% (both\nthresholds above stimulus level), 50% (stimulus level above the low threshold\nbut below the high threshold level) or 100% (both thresholds below stimulus\nlevel) of the individuals may be active at any one time. If, on the other hand, the\nCTD has a continuous distribution, the colony will produce a more finely graded\nresponse that is proportionate to the stimulus levels. Experimental data on the form of the CTD is rather limited. To our knowledge, only in\nthe honey bee, Apis melifera is there a quantitative description of\nthe CTD, which is approximately Gaussian [20] . Indeed, several previous\nsimulation studies of the FTM have assumed a Gaussian distributed CTD [19] , [43] . For\nsimplicity, we first consider the case of the uniform CTD. Uniform distributions\nlack any central tendency (they are not humped) and so have a variance, defined by\nthe range of the distribution. For completeness we also investigate the influence of a Gaussian CTD upon the ability\nof the colony to minimise both the total task demand and the spatial variation\nthereof. As well as the aforementioned uniform and Gaussian CTDs, we also explore\nthe consequences of completely removing individual threshold variation.",
"discussion": "Discussion As in earlier non-spatial Division of Labour models based on thresholds, the SFTM\nreproduces the massively right-skewed activity distribution observed in real social\ninsect colonies [13] , [26] , [27] , [70] , [71] as the vast majority of the labour is performed by a\nhighly active minority ( Figure\n3 ). In both the one- and two-dimensional models imperfect mixing of heterogeneous\nindividuals and stimuli generated a non-random spatial structure of both the\nindividuals ( Figure 1b &\n Figure 11 ) and the stimuli\n( Figure 5 , Figure 8 & Figure 10 ). Non-random\ndistributions of individuals or task-associated stimuli are ubiquitous within social\ninsect colonies. Complex spatial structuring of the distribution of individuals or\ntask associated stimuli within social insect colonies includes the aggregation of\nindividuals by role or caste [72] , [73] , [74] , clustering of nest-building material [75] , [76] , [77] dead\nindividuals [78] \nand brood [79] ,\n [80] . Whilst it is known that cells, individuals and societies can achieve some degree of\nhomeostasis by minimising the temporal fluctuations of relevant stimuli [19] , [24] , [49] , [81] , it is\nimportant to emphasise that homeostasis may also be achieved by minimising these\nfluctuations across space. In many cases this capability will be highly adaptive.\nFor example, it might be advantageous to minimise the spatial variation of stimuli\nassociated with brood hunger, as the brood may die when a critical hunger is\nexceeded. In our model for all CTD's the spatial fluctuation amplitude of the\nstimulus is minimised when the stimulus drive is relatively low (i.e., when the\ndrive is less than the average threshold, , Figure 10 ), which is actually when the activity\nskew is greatest (9b). So spatial homeostasis is maximised when the division of\nlabour is greatest. Let us now turn to the issue of interactions and competition for work between\nindividuals. Circumstances in which many individuals ‘graze’ a stimulus\nsurface are ubiquitous in biology. For example, the removal of parasitic fungal\nspecies from the fungus gardens in fungus growing ants [82] , brood sorting and tending\n [79] , and the\ngeneral activity of honey bee inside-nest workers [36] all involve multiple\nindividuals moving across and performing work upon a spatially and temporally\nvariable stimulus landscape. Clearly an individual performing work on the stimulus\nlandscape causally influences the subsequent activity, or often the lack thereof, of\nits nestmates. On the rare occasions when a high threshold ant is active, it tends\nnot to move very near its low threshold nestmates ( Figure 11 ). This is because those nestmates have\nreduced the stimulus in the surrounding sites to such a low level that they appear\nto contain no work, so those sites act as barriers to movement. Conversely, when the\nactive ant has a lower threshold, it is not ‘repelled’ by a\nhigher-threshold neighbour ant, because that neighbour only reduced the stimulus in\nthe area to its own threshold level, and no lower, hence the active ant detects that\nthose sites contain work. Thus apparent aversion between behavioural or\nmorphological castes [8] , [10] resulting in spatial segregation of individuals might\narise from indirect spatial exclusion rather than direct repulsion. In ecology, explicit con-specific attraction or repulsion is often invoked to explain\nthe observation of non-random spatial patterns, such as over or under-dispersion.\nThe clustering of agents (under-dispersal) is associated with attraction, whereas\nregularly spaced agents (over-dispersal) is related to repulsion. In ants, spatial\nclustering of individuals has been explained by invoking direct \ninter-individual attraction and repulsion based on physical or behavioural\ndifferences between individuals [83] , [84] . Similarly, spatial DoL through the segregation of\nphysical castes has been explained by invoking explicit between-caste aversion [10] . At the colony\nlevel a high degree of regularity in the spacing of ant nests is ubiquitous and is\nconventionally understood in terms of competition for space [85] , [86] , [87] . One of the main results of\nthis paper is that the spatial patterns previously ascribed to individuals that\n‘pay attention’ to the proximity of their nestmates, can also be\nproduced when individuals do not directly account for the proximity of nestmates.\nThis conclusion is concordant with the concept of self-organisation through\nstigmergic processes [53] . Finally, we wish to highlight the scale-free structure of the stimulus landscape\n( Figure 5 & 6 ) and the similarity of the\nsigmoid threshold-activity functions ( Figure 9 ) to phase-transition curves. Such phenomena are typical of\ncomplex systems at critical points [52] , [88] and suggest that\nthreshold-based DoL can self-organise towards a critical point. Division of labour characterises all levels of biological organisation as well as\nhuman and artificial social systems. Our spatial fixed-threshold model links this\norganisational principle with the statistical mechanics approach to complex systems\nand provides testable hypotheses for future experiments."
} | 3,490 |
37599920 | PMC10433335 | pmc | 800 | {
"abstract": "Slippery liquid-infused porous surfaces (SLIPS) are self-healing\nprotective coatings that can be made by infiltration of a porous scaffold\nwith a chemically resistant oil. A popular method to apply a SLIPS\ncoating is using electrospinning to deposit a nanofiber mesh onto\nthe intended substrate. However, electrospinning only lightly deposits\nthe nanofibers onto the intended substrate, so the coating detaches\neasily even when unintended. We report a simple, yet effective, solution\nto the adhesion problem. Electrospun nanofiber meshes are typically\nwell entangled and cohesive, so they can be detached from the electrospinning\ntarget and transferred onto the final target. Using a thin layer of\nadhesive on the intended surface, the electrospinning mesh can be\nsecurely attached and infiltrated with protective oil to yield a more\nstable SLIPS coating. An adhered coating can be submerged under corrosive\nsolution and repeatedly self-heal from damage to the same spot. With\nthe electrospun nanofiber meshes’ flexibility and stretchability,\nthe meshes can be fitted around a wide range of targets including\nones that are otherwise difficult to apply a nanofiber mesh on. The\nuse of an adhesive interlayer between the nanofiber mesh and substrate\nis a simple solution to improve coating stability, and the solution\nfacilitates application of SLIPS onto a broader range of substrates.",
"conclusion": "4 Conclusions Nanofiber-based SLIPS coatings\nexhibit similar barrier properties\nas SLIPS coatings made with other porous scaffolds. Prior to this\nreport, SLIPS made with electrospun nanofibers were directly deposited\nonto the intended substrates. However, the weak adhesion between the\ndeposited nanofiber mesh and substrate can lead to unintended detachment\nof the coatings. Requirements of electrospinning on the conductivity\nand geometry of the electrospinning target also limit the potential\nof SLIPS made with electrospun nanofibers. This report presents a\npossible solution using a thin layer of adhesives to improve coating\nstability. The applied adhesive allows the nanofiber meshes to be\ntransferred onto arbitrary substrates, and the use of adhesive does\nnot eliminate the self-healing properties or weaken the barrier properties.\nUnlike typical SLIPS made with static scaffolds, nanofiber-based SLIPS\nhave a flexible scaffold. Upon damage, flexible nanofibers can move\nout of the way to minimize damage. Once the damage event passes, protective\noil flows back into the damaged area and brings the surrounding nanofibers\nalong to re-entangle into a scaffold. Meshes of electrospinning nanofibers\nare convenient scaffolds for SLIPS coatings, so the improvement of\nnanofiber mesh adherence onto the substrate will improve the performance\nand stability of nanofiber-based SLIPS coatings.",
"introduction": "1 Introduction Corrosion of metals is\nan ongoing significant problem for modern\nsociety because it can lead to the reduction of service life for materials.\nCorrosion is the process by which an unwanted process occurs between\nmaterials and corrosive species, such as oxygen, acidic protons, basic\nions, and water. 1 − 3 In particular, corrosion of metals can lead to depreciation\nof valuable parts, loss of function, reduced efficiency of catalytic\nprocesses, and even loss of life. 4 Metals\nare mechanically robust, so they are often used for stress-bearing\nparts of automobile, ships, planes, and buildings. 5 However, exposed metals readily react with corrosive species,\nwhich can damage the part. A method to prevent corrosion-led problems\nis to use protective coatings to separate metals from their surroundings.\nAt present, various thermoplastic polymers are used as protective\ncoatings, but these polymers are susceptible to physical damages that\ncan expose the metal underneath. 6 − 8 Resin-based coatings tend to be\nmore crosslinked and brittle, so they are similarly susceptible to\nmicroscopic damage. 9 − 11 Unfortunately, corrosion occurs if the coating is\ndamaged, and the substrate underneath is exposed to corrosive species.\nEven pinhole-sized holes, hidden in plain sight, can develop into\na significant corrosion problem over time. 12 , 13 When coatings receive microscopic damage, corrosive species can\nenter through the pinhole and begin to damage the material underneath.\nIf the damage is not repaired in a timely fashion, then corrosion\ncan propagate from exposed parts and eventually cause a catastrophic\nfailure. Hence, periodic repair and inspection of coated metal is\nnecessary even if the coating appears intact. Preventive inspections\nand replacements of parts cost resources, so there is interest to\ndevelop more consistent protection against corrosion. A coating\nthat repairs on its own can prevent microscopic damage\nfrom progressing undetected. There are several reports of self-healing\ncoatings, 14 − 19 such as inclusion of microcapsules that include self-healing components 20 − 23 or coatings that had self-healing agents dissolved within. 24 , 25 If damage occurs repeatedly at the same location, reactive agents\nin the microcapsules would be exhausted and embedded self-healing\ncomponents would be depleted. 26 Hence,\nrepeated self-healing at the same site requires a large reservoir\nof reactive components. Slippery liquid-infused porous surfaces (SLIPS)\nfirst reported by Aizenberg et al. are a class of self-healing coatings\nthat can heal from repeated damages. SLIPS coating consists of protective\noil embedded in a porous mesh over the substrate. Variations of SLIPS\nhave utilized different methods for their applications. These include\nusing direct growth of porous material onto the targeted substrates, 27 − 29 deposition of nanomaterials during hydrothermal synthesis, 30 , 31 laser ablation to produce a porous surface, 32 dip coating, 33 spin coating, 34 spray-on, 35 , 36 and electrospinning\nnanofibers onto the subject. 37 − 40 Of these options, electrospinning is a convenient\nway to directly deposit continuous nanofibers onto a substrate surface.\nAfterward, infusion of protective oil into the mesh of electrospinning\nnanofibers can yield a coating with SLIPS properties. These continuous\nnanofibers are flexible and entangled, so they can generate conformal\ncoatings that are more bendable and durable compared to other SLIPS\nmade from porous structures or short fibers. However, electrospun\nnanofibers directly deposited onto substrates are only weakly attached\nvia van der Waals interactions, and these nanofibers can detach unexpectedly\nduring service. Due to nanofibers’ potential benefits as SLIPS\nscaffolds, there is a need for a better method for attaching electrospinning\nnanofibers onto arbitrary substrates for fabricating SLIPS. In recent years, Lim et al. published several articles on a type\nof easily applied, self-healing coating to take advantage of liquids’\ntendency to conform to any container. 41 , 42 These coatings\nwere made with lightweight graphene microcapsules as temporary scaffolds\nthat maintain a protective oil layer on the surface of metal and allow\noil to flow upon being damaged. The self-healing mechanism is like\nthat of SLIPS coatings. Yet, the coating is easily applicable as a\npaste-like mixture onto any arbitrary surface. Instead, the coating\nretains its shape by forming temporary scaffolds from the microcapsules.\nSLIPS and the microcapsule-filled oil coatings have similarities;\nthe microcapsule-filled oil is simultaneously easier to apply onto\nother substrates and less mechanically robust. SLIPS coatings\nwith electrospinning nanofiber scaffolds use the\nintended coating target as the electrospinning target, but the deposited\nnanofiber mesh is only weakly attached onto the substrate. 38 − 40 A one-time convenience that quickly attaches a porous scaffold can\nlead to unintended detachment of the SLIPS coating during service.\nDuring service, even mild agitation can cause the electrospinning\nnanofibers to detach and expose the surface underneath ( Figure S1 ). Furthermore, limitations of electrospinning\nalso apply during the coating application process. If the substrate\nis not conductive, then the electrospinning process will be off target.\nThe electrospinning process also has trouble coating sharply concave\nsurfaces or targets with detailed features as this would cause nanofibers\nto bridge between opposing edges. An improved coating will combine\nproperties of SLIPS, ease of application\nof microcapsule-filled oil, and mechanical robustness of long electrospinning\nnanofibers. A potential solution may be quite simple and effective.\nElectrospinning nanofibers can be attached onto substrates using adhesives\nwhile still allowing nanofibers to dangle off the substrate surface.\nThe adhesive will reinforce contact and the dangling nanofibers form\nthe scaffold to hold onto protective oil. Because the nanofibers are\nfree to move about, they retain the self-healing capabilities of SLIPS\ncoatings. As a proof of concept, a coating was fabricated from polystyrene\n(PS) nanofibers mixed with silicone oil. The PS nanofiber mesh is\ncohesive enough to transfer from the electrospinning substrate onto\na variety of arbitrary surfaces. The obtained PS nanofiber-based SLIPS\nbears the expected protective properties and self-healing capabilities.\nThe coating can protect aluminum against concentrated hydrochloric\nacid. Upon being exposed to damage, the coating rapidly self-heals\nto prevent corrosion. SLIPS coatings have been demonstrated for a\nwide variety of materials combinations, so using adhesives to stabilize\nelectrospinning-based SLIPS coatings eases the application process\nand will aid in the discovery of more effective designs in the future.",
"discussion": "3 Results and Discussion A SLIPS coating\nhas two key components, a porous scaffold and a\nprotective liquid. For electrospinning nanofiber-based SLIPS, the\nnanofiber mesh functions as the porous scaffold. The electrospinning\nprocess ejects polymer solution toward a grounded target and deposits\nan entangled mesh of polymer nanofibers. Because electrospinning nanofibers\nsolidify during flight, the deposited nanofiber meshes are only lightly\nattached onto the substrate ( Figure 1 a). The weakly attached coating may fall off under\ngravity’s pull even under slight agitation ( Figure S1 ). The weak interaction is instead helpful in detaching\nthe whole nanofiber mesh from the electrospinning target. Adhesives\ncan be applied to the intended target for a SLIPS coating, so the\nelectrospinning nanofiber mesh can be transferred and adhered ( Figure 1 a). Without additional\nadhesive, the coating does not have significant adherence to the substrate\n( Figure S2a,b ). Once an adhesive is added\nunderneath the coating, the coating would require significant strength\nto peel and lead to more stable coatings ( Figure S2c,d ). Because the nanofibers are long and continuous, the\nmesh remains cohesive during transfer ( Figure 1 b). Use of adhesives resolves several application\nchallenges facing electrospinning nanofiber-based SLIPS. The coating\ncan be adhered onto non-conductive 3D printed poly(lactic acid) (PLA)\nobjects such as a tensile bar ( Figure 1 c) or a glass cup ( Figure 1 d). 3D printed objects without post processing\ntypically have rough surfaces that are susceptible to staining ( Figure 1 c inset). A well-coated\nnanofiber-based SLIPS helps the tensile bar repel the dyed water solution\nwithout being dyed (the process is shown in Supporting Information Video S2 ). Nanofiber-based SLIPS coatings on\nthe insides of a glass cup help the complete transfer of a water solution.\nThis minor difference can significantly improve the transfer of liquids\nfor culinary, manufacturing, or synthetic applications. Furthermore,\nthe adhesive-assisted method can be used for applying a SLIPS coating\nonto other porous substrates or flexible material. A coated piece\nof paper can protect its writing from discoloring while underwater\n( Figure 1 e). Video S3 demonstrating the effects are provided\nand described in Supporting Information Section S2. PLA and paper are not conductive, so they are not suitable\nas electrospinning substrates. PLA and paper are also less stable\nthan metals or ceramics, so porous scaffolds cannot be deposited over\nthese materials using techniques such as chemical vapor deposition,\nhydrothermal growth, or brush-on techniques. Figure 1 PS nanofibers were processed\nusing electrospinning. (a) A typical\nexperiment ejects PS solution through a high-voltage charged needle\nto force the viscous polymer solution to fly toward a grounded substrate.\n(b) An SEM image shows the deposited nanofibers are interweaved and\ncontinuous. (c) A cohesive nanofiber mesh can be attached onto an\narbitrary shape to prevent staining as opposed to the uncoated sample\nshown in the inset. (d) The inside of a cup was coated with nanofiber-based\nSLIPS to prevent any liquid residue from remaining upon pouring. (e)\nNanofiber-based SLIPS can also be coated onto paper and placed under\nwater to prevent water-based ink from leeching out. For a nanofiber-based SLIPS coating to display\nself-healing properties,\nthe nanofibers need to entangle into scaffolds and display viscoelastic\ntransitions for self-healing. Nanofiber scaffolds must allow protective\noil to infuse, and the nanofibers must reform into scaffolds after\ndamage events. To do so, the coating can undergo a transition between\nsolid-like and liquid-like viscoelastic behaviors. While at rest,\nthe coating will stay still without flowing. When experiencing deformation,\na viscoelastic solid will turn liquid-like and flow. Nanofibers adhered\nto the substrate retain long segments that are freely dangling, and\nthey can entangle into scaffolds before and after damage events. To\nemulate the environment of free-flowing nanofibers, nanofiber meshes\nare chopped and dispersed in liquids. Nanofibers made by electrospinning\nare effectively continuous, so they can be systematically shortened\ninto desired lengths. To study various entanglement states, mixtures\nof varying nanofiber concentrations were prepared. Through frequency\nscans, all the nanofiber mixtures show crossovers between loss modulus\nand storage modulus ( Figure 2 ), showing a change between solid-like and liquid-like behaviors.\nThe samples have a higher storage modulus than loss modulus at low\nfrequencies, which represents the relaxation of the entangling nanofibers.\nWhen the frequency is increased, the mixture has a greater loss modulus\nthan storage modulus to indicate its dissipation of mechanical energy.\nThis can be understood as the protective oil relaxing independent\nof the nanofibers. While the dispersion of nanofibers imparts viscoelastic\nproperties, the oil can still flow between the nanofibers to generate\nthe self-healing activities of the nanofiber SLIPS coating. Figure 2 Frequency scans\nfor several PS nanofibers and silicone oil mixtures\nat different concentrations. (a) 0.1 wt %, (b) 0.4 wt %, (c) 0.7 wt\n%, (d) 1.0 wt %, and (e) 1.5 wt %. The nanofibers and oil mixtures’ viscoelastic\nbehaviors\ncan also be examined by the strain scan to detect variations in behavior\nwhen at different fiber lengths and concentrations. While viscoelasticity\nof a mixture is indicated by the frequency scans, the strain scan\nhelp to study how the nanofiber may behave to different scales of\ndamage and if there is qualitative change to the self-healing activity\ndue to repeated damages shortening the nanofibers. Nanofibers with\ndifferent lengths were obtained by cutting mats of continuous nanofibers\ninto different-sized squares. PS particles were used to represent\nparticles with a low aspect ratio ( Figure S3 ). As shown in Figure 3 , mixtures with longer nanofibers generally have greater moduli compared\nto those made with shorter nanofibers. Nanofiber-based SLIPS have\ncrossovers between their loss modulus and storage moduli ( Figure 3 b–e), but\nparticles’ samples do not have the crossover ( Figure 3 a). The transition is a sign\nof nanofibers’ differing behavior to large damage events where\nsmall damages will allow the SLIPS coating to behave like a solid\ncoating and large damage events will see the SLIPS coating change\ninto a liquid-like behavior to avoid nanofibers from being decimated.\nCombined with the frequency scans, the importance of nanofiber entanglement\nbecomes apparent. This is to form a tortuous network to keep the coating\nin place and also to allow self-healing via oil flow. As entanglement\nis necessary for nanofibers to form scaffolds, nanofiber-based coatings\nwill retain self-healing activity if they are not completely decimated\ninto low-aspect-ratio particles. Figure 3 Strain scans for several PS nanomaterials\nand silicone oil mixtures\nat 1.5 wt %. (a) PS particles were used as the low-aspect-ratio control.\n(b) A PS nanofiber mesh cut into 5 × 5 mm 2 was used.\n(c) 10 × 10 mm 2 PS nanofiber mesh. (d) 12 × 12\nmm 2 PS nanofiber mesh. (e) 15 × 15 mm 2 PS\nnanofiber mesh. Another challenge for nanofiber-based SLIPS coatings\nis keeping\nthe oil and nanofibers in place. These coatings are part solid and\npart liquid, held in place by the porous scaffold. For nanofiber-filled\noils, their mechanical stability and capability to hold the coating\nin place can be described with their storage modulus under low shearing\nconditions, such as 0.01% shear on a rheometer. Figure 4 shows that the coating mixtures reinforced\nwith PS nanofibers have storage modulus greater than their loss modulus\nat 0.01% strain. The differences ensure that the coatings behave like\nsolids at rest. To have a stable coating and significant self-healing\nbehavior, effective coatings will maximize the storage modulus and\nhave a large difference between loss and storage moduli. Figure 4 Storage and\nloss modulus for several PS nanomaterials and silicone\noil mixtures at 0.01% strain obtained through strain scans with a\nparallel plate rheometer. (a) PS particles were used as the low-aspect-ratio\ncontrol. (b) A PS nanofiber mesh cut into 5 × 5 mm 2 was used. (c) 10 × 10 mm 2 PS nanofiber mesh. (d)\n12 × 12 mm 2 PS nanofiber mesh. (e) 15 × 15 mm 2 PS nanofiber mesh. When the storage modulus is plotted over a range\nof fiber lengths,\nmixtures with different fiber concentrations show significant differences\nin their viscoelastic behaviors. Under 0.01% strain, increasing nanofiber\nconcentration increases the storage modulus as more nanofibers are\navailable to entangle and obstruct each other. Increasing fiber length\nincreases storage modulus, but the effect depreciates significantly\nafter reaching sufficient entanglement. This can be observed in the\nplateauing of storage moduli of samples made with nanofiber mesh squares\ngreater than 10 mm 2 ( Figure 5 a). When SLIPS mixtures undergo large shearing strains\n( Figure 5 b), nanofiber\nconcentrations increase in storage moduli just like the scenario with\nlow strain ( Figure 5 a). Increasing average nanofiber lengths results in minor changes\nin the storage modulus for nanofiber meshes of 5 to 10 mm 2 ( Figure 5 b). At low\nnanofiber concentrations, the storage modulus is seemingly unaffected\nby average nanofiber length. For nanofiber mesh squares beyond 12\nmm 2 in size, the storage modulus increased dramatically\nfor nanofiber concentrations above 0.7 wt % ( Figure 5 b). The differences in behaviors are due\nto high concentration mixtures having sufficient nanofibers to entangle. 43 Nanofibers in low concentration mixtures do\nnot have enough nanofibers to remain entangled during high-shear motion.\nWith respect to the formation of SLIPS, scaffolds based on electrospinning\nnanofibers need to be above a minimum length. The observed behavior\ncan be combined with prior reports that much shorter and thinner nanowires\nare also capable of forming SLIPS. 27 − 31 , 44 , 45 The combined activities suggest that materials with sufficient aspect\nratios may be suitable as scaffolds for SLIPS. Figure 5 Storage modulus of PS\nnanofibers of different lengths mixed with\nsilicone oil at various concentrations as indicated by line color.\n(a) When the storage modulus is examined at 0.01% strain simulating\nrest and (b) at 10% strain simulating a large damage event. Nanofiber-based SLIPS coatings have great barrier\nproperties against\nvarious environments. SLIPS demonstrate great stability against corrosive\nenvironments that can quickly corrode a substrate. When a coated substrate\nis floated over an acid bath, ineffective coatings will quickly allow\nthe substrate to dissolve. An effective coating will protect the coating\nfor many weeks. By applying different types of coatings onto the bottom\nof an aluminum boat, the experimental boat can be floated over a solution\nof 1 M hydrochloric acid ( Figure 6 a,b). An uncoated aluminum boat sank below the solution\nsurface after just 25 min, and the aluminum boat completely dissolved\nafter 46 min ( Figure 6 c). A separate aluminum boat was coated with commercial paint and\nfloated over a similar 1 M HCl solution. Like the unprotected boat,\nthe commercial paint boat sank after 25 min, and the aluminum portion\nof the boat completely dissolved after 46 min to leave just the paint\nbehind ( Figure 6 d).\nWhile commercial paint is corrosion-resistant, the painting process\ninevitably creates pinholes that allow corrosion to start, propagate,\nand destroy the boat. As pinholes can exist even for simple surfaces\nsuch as the flat aluminum surface, more complex shapes will be even\nmore susceptible due to difficulty in application of the coating.\nAn aluminum boat protected by the nanofiber-based SLIPS was able to\nprevent corrosion for at least 1 week without showing any sign of\ndamage ( Figure 6 e).\nIf the coating consists of only the adhesive layer, then the acid\nresistance would be poor in comparison and cause catastrophic corrosion\nwithin 5 h ( Figure S4 ). Despite having\nlimited resistance to acid solution, the PS nanofibers are able to\nmaintain their shape and function as scaffolds. A similarly processed\nsample using phenolic resin as the adhesive to hold onto the PS nanofiber\nfilm also showed superb anti-corrosion activities ( Figure S5a ) because the surrounding protective oil protected\nthe nanofibers as well as the substrate. 46 Another sample using a poly(methyl methacrylate) nanofiber SLIPS\ncoating also showed good acid solution resistance when combined with\nsilicone ( Figure S5b ). When combined, the\nsynergistic coating is chemically stable and resilient to pinhole\nformation. Figure 6 Anti-corrosion experiments were conducted using commercial aluminum\npans floating over 1 M HCl. (a) A typical experiment coats the underside\nof an aluminum pan with various coatings, while it floats over a body\nof acid solution. (b) Each experiment follows the time aluminum stays\nafloat over the body of acid. (c) Without any coating, the aluminum\npan sinks after 25 min and completely disintegrates after 46 min.\n(d) With a commercial anti-corrosion coating, the aluminum pan also\nsinks after 25 min and aluminum dissolved after 46 min. (e) When coated\nwith the SLIPS coating, the aluminum pan stayed afloat for over 168\nh (1 week) without any visible damage to the coating. Nanofiber-based SLIPS coatings are capable of self-healing\nthat\ngive them resilience against pinhole formation. However, SLIPS made\nwith rigid porous scaffolds cannot rebuild their scaffold upon damage.\nNanofiber-made scaffolds can partially recover from damage because\nmost nanofibers can maneuver out of the way during damaging events\nand return to original positions afterward ( Figure 7 a). To demonstrate the self-healing effects\nunder challenging conditions, aluminum foils coated with nanofiber-based\nSLIPS were sliced under 1 M HCL solution. Cuts were made consecutively\nat the same spot to demonstrate the repetitive self-healing capabilities\nand nanofiber scaffold’s recovery after damage ( Figure 7 b). As a control, an aluminum\nsample coated with commercial paint bubbled vigorously immediately\nafter the cuts were made, and the aluminum underneath dissolved away\nafter 18 h ( Figure 7 c). After corrosion is completed, commercial paint peels away to\nreveal the aluminum foil wrapped on the back side of the glass slides.\nOn the other hand, the nanofiber-based coating sustained submersion\nfor 18 h after being consecutively cut 5 times at the same spot ( Figure 7 d). The coating is\nsemi-transparent, which helps one observe that the initial gash did\nnot worsen after several hours. The gas bubbles are above the coating\nand not in the gash. From these signs, we can see that the nanofiber\ncoating is capable of self-healing and is effective in protecting\nthe aluminum surface from corrosion. Figure 7 SLIPS coatings can self-heal and the nanofiber\nscaffolds can recover\nfrom damage. (a) A cut will sever many nanofibers, but loose strands\nwill return and the oil will reinfiltrate. (b) A typical test of self-healing\nactivity makes 5 cuts at the SLIPS coating while it is submerged under\n1 M HCl. Each cut is purposefully made at the same location and deep\nenough to cut the aluminum substrate underneath. (c) When the self-healing\ntest was done on a commercial coating, the exposed aluminum reacted\nto the acid solution to generate bubbles. (d) The SLIPS coating self-healed\nfrom the cuts, so bubble was limited to the instant after the cut\nwas made. When the nanofiber scaffold is exposed to damage,\nflexible nanofibers\ncan reel back to buffer against the intrusion ( Figure 7 a). Once the foreign object exits, the protective\noil can reinfiltrate the damaged area to reduce surface energy. Despite\nthe dense network of nanofibers, the protective oil will behave like\na liquid, and the oil flow after damage will bring nanofibers into\nthe damaged area from the surrounding. Once nanofibers return to the\ndamaged area, they can entangle to form a new porous scaffold. In\nless than 1 s, the nanofiber-based SLIPS coating recovers from the\ndamage and stops corrosion. Just like SLIPS coatings with porous scaffolds,\nthese nanofiber coatings can self-heal multiple times as long as there\nis oil embedded in the coating. Severed nanofibers retain much of\ntheir length, so they remain capable of entangling and forming scaffolds.\nHence, a nanofiber-based SLIPS coating can self-heal multiple times\nin the same spot. Corrosion protection and self-healing processes\nof SLIPS coatings are not limited to specific materials, so adhesives\nfor transferrable electrospinning nanofiber meshes are also applicable\nfor chemically resistant natural oils that do not degrade from interactions\nwith HCl ( Figure S4 ). Aluminum pans coated\nwith canola oil or soybean oil also show great corrosion-resistant\nproperties ( Figure S8 ). In addition to\nbeing easily applicable with the aid of a thin adhesive layer, nanofiber\nSLIPS is just as capable as coatings made with other porous material\nand protective oil."
} | 6,714 |
25365180 | PMC4264147 | pmc | 801 | {
"abstract": "In order to use cyanobacteria for the biological production of hydrogen, it is important to thoroughly study the function and the regulation of the hydrogen-production machine in order to better understand its role in the global cell metabolism and identify bottlenecks limiting H 2 production. Most of the recent advances in our understanding of the bidirectional [Ni-Fe] hydrogenase (Hox) came from investigations performed in the widely-used model cyanobacterium Synechocystis PCC6803 where Hox is the sole enzyme capable of combining electrons with protons to produce H 2 under specific conditions. Recent findings suggested that the Hox enzyme can receive electrons from not only NAD(P)H as usually shown, but also, or even preferentially, from ferredoxin. Furthermore, plasmid-encoded functions and glutathionylation (the formation of a mixed-disulfide between the cysteines residues of a protein and the cysteine residue of glutathione) are proposed as possible new players in the function and regulation of hydrogen production.",
"conclusion": "6. Conclusions Though significant progress has been made recently in our understanding of the function and regulation of the complex hydrogen Ni-Fe hydrogenase of the model cyanobacterium Synechocystis PCC6803, important questions remain. Recent in vitro data suggested that the Hox enzyme can receive electrons from not only NAD(P)H as usually shown, but also, or even preferentially, from ferredoxin. This proposal will certainly stimulate the investigation of the selectivity/redundancy of the nine ferredoxins of Synechocystis , which are conserved in cyanobacteria. Using gene deletion and over-expression, it has been shown recently that the hoxEFUYH operon operates in the defense against redox stresses triggered by H 2 O 2 or reduced carbon or nitrogen metabolites. These findings strengthen the proposal that hydrogenase operates as an electron valve to prevent the supernumerary electrons to recombine with O 2 to generate toxic reactive oxygen species. Hence, the hydrogenase complex can be viewed as an important enzyme in cyanobacteria like Synechocystis , which can grow in biofilm, a thick network of auto-aggregated cells where they are inevitably exposed to H 2 O 2 and reduced metabolites released by their neighbors (living or dying and lysing). This view is strengthened by the absence of hydrogenase enzyme in most planktonic cyanobacteria living in open oceans. It will be interesting in the future to investigate the influence of the Synechocystis hydrogenase enzyme in the growth and response to environmental stresses of cells incubated not only in agitated liquid cultures but also in static conditions or on solid media to favor the formation of biofilm. The recently achieved simultaneous overproduction of the HoxEFUYH and HypABCDEF proteins within the same cells led to a 20-fold increase in active hydrogenase. These sophisticated mutants with a higher hydrogenase content and a healthy growth will be very useful cell factories for the purification of large hydrogenase quantities for structural analyses, the data of which should enable the design of a meaningful mutational strategy to increase the low natural O 2 tolerance of the hydrogenase enzyme to increase hydrogen production. All hydrogenase-overproducing mutants displayed higher levels of expression of the hoxEFUYH and hypABCDEF genes than that of active hydrogenase, indicating that limiting post-transcriptional factors should be dealt with. Thus, it will be important to pay particular attention to glutathionylation (the formation of a mixed-disulfide between the cysteines residues of a protein and the cysteine residue of glutathione) because it was recently reported that the AbrB1 and AbrB2 hydrogen regulator and the HoxH (diaphorase) and HoxF (hydrogenase) protein-subunits can be glutathionylated. These findings will certainly stimulate the analysis of the redox crosstalk between hydrogen production and the oxidative stress-responsive glutathionylation-and-deglutathionylation process.",
"introduction": "1. Introduction Energy is crucial to modern industry. In 2012, 570 ExaJoules (1 EJ = 10 18 Joules) of energy was consumed worldwide, of which approximately 80% was generated from burning fossil fuels thereby liberating into the atmosphere about 6 gigatons (6 × 10 9 tons) of the green-house-generating gas CO 2 . Even hydrogen (H 2 ), which has a higher energy content than oil (142 MJ/kg for H 2 \n vs. 44.2 MJ/kg for oil) and burns cleanly, producing only water as its by-product, is not yet a clean biofuel because it is mostly produced from burning oil [ 1 ]. Hence, the pollution problem and the fossil fuel shortfall anticipated to occur during the 21st century make it important to develop new energy sources that are plentiful, renewable and environmentally friendly. Sunlight is naturally attractive as it is the most readily available and inexpensive source of energy on Earth. For instance, the annual solar flux received by Earth—approximately 5 YottaJoules (1 YJ = 10 24 Joules)—is in huge excess of the 570 ExaJoules used by our society. Consequently, there is a growing interest in processes that could couple the solar energy-powered capture of CO 2 to the production of energies through the use of photosynthetic organisms, such as microalgae [ 2 , 3 ]. Indeed, production by microalgae of renewable biofuels from nature’s most plentiful resources, solar light, water, mineral salts and CO 2 , is of great interest in recycling CO 2 and saving arable soils, fertilizers, pesticides and fresh waters for crop production. Cyanobacteria (formerly termed blue-green algae) have the potential for that. First, cyanobacteria are the most abundant photosynthetic organisms of our planet. They colonize most ecological niches, fresh and salt waters, terrestrial and extreme environments (pH, temperature) that counter-select most competing organisms, a primary concern of mass cultivation [ 4 ]. Second, cyanobacteria convert captured solar energy into biomass at high efficiencies (3%–9%) [ 2 , 5 ] to produce a large part of the atmospheric oxygen and organic assimilates for the food chain [ 6 ]. On a global scale, cyanobacteria fix an estimated 25 Giga tons of carbon from CO 2 per year into energy dense biomass [ 7 ]. To perform this huge CO 2 fixation, cyanobacteria use 0.2%–0.3% of the solar energy, 178,000 terawatts (1 TW = 10 12 watts), reaching the Earth surface [ 8 ]. Thus, the amount of energy passing through cyanobacteria exceeds by more than 25 times the energy demand of our society; roughly 1000-fold the total nuclear energy produced on Earth. Third, as cyanobacteria tolerate high CO 2 content in gas streams and they can grow in a variety of locations, they can be used as “low-cost” microbial cell factories for the capture and storage of industrial CO 2 gas near the sites of industrial productions, thereby reducing transportation costs. Fourth, many cyanobacteria as the unicellular model strain Synechocystis PCC6803 have a small sequenced genome amenable to genetic manipulations including with versatile plasmid vectors [ 9 , 10 , 11 ]. A powerful genetics is welcome as natural (wild-type) cyanobacteria are not suitable biofuel producers because some of the required metabolic pathways are partially lacking or need optimization [ 3 , 5 ]. In cyanobacteria, two enzymes can produce hydrogen, the [Ni-Fe] bidirectional hydrogenase (Hox for h ydrogenase oxidation) and the nitrogenase. Both enzymes are sensitive to oxygen and do not occur in all strains, but for H 2 production the [Ni-Fe] hydrogenase, which does not use ATP [ 12 ], is energetically favored over the nitrogenase enzyme that consumes 16 ATP per molecules of H 2 produced [ 13 ]. The bidirectional Hox enzyme has been mostly studied in the model cyanobacterium Synechocystis PCC6803 (hereafter Synechocystis ), where it is the sole enzyme capable of combining electrons with protons to produce H 2 under specific conditions [ 12 , 14 ]. The Hox enzyme comprise five protein-subunits, HoxEFUYH, and a [Ni-Fe] and several [Fe-S] redox clusters ( Figure 1 ). Figure 1 Schematic representation of hydrogen production machine in Synechocystis adapted from [ 12 , 15 ]. The genes are represented by arrows, which point in the direction of their transcription, and are colored similarly to their protein products. The green numbers indicate the spacing distance (in kilobases) between the scattered genes. The hoxEFUHY operon is weakly transcribed [ 16 ] as the polycistronic mRNA (bent blue arrow), which encodes (i) the hydrogenase sub-complex (made by the HoxY protein and the HoxW-matured HoxH subunit); (ii) the HoxEFU diaphorase sub-complex; and (iii) the three proteins of unknown function (white forms). The electron transfer FMN cofactor, Fe-Ni center, and [4Fe-4S] and [2Fe-2S] clusters of the Hox proteins, are represented by the blue squares, the pink star, dark-red squares and light-red diamonds, respectively. The zinc-bound to HypA1 and HypB1 proteins is shown as the blue form. CP designates carbamoyl phosphate. The brown lines stand for the reversible inactivation of Hox activity mediated by oxygen. The photosynthetic membrane is represented in green. The HoxEFU subunits make up the diaphorase sub-complex that transfers to the [Ni-Fe] hydrogenase sub-complex HoxHY the NAD(P)H-transported electrons produced by photosynthesis and/or sugar catabolism [ 17 ]. Recently, based on an in vitro analysis, it was also proposed that the Hox enzyme can be reduced by ferredoxin or flavodoxin [ 18 ], similarly to the [Fe-Fe] hydrogenase of eukaryotic algae [ 19 ]. During HoxHY assembly, the HoxH subunit is processed by the HoxW protease [ 14 ], and subsequently the [Ni-Fe] HoxEFUYH complex is assembled by the six subunits HypABCDEF complex [ 12 , 14 ]. The Hox enzyme is dispensable to the photoautotrophic growth in standard laboratory conditions [ 20 ]; it has a bias towards H 2 production [ 21 ] and it is reversibly inactivated by oxygen [ 17 ]. Hence, investigating the photobiological production of H 2 by cyanobacteria has both an evident biotechnological interest, and a basic research interest, in addressing the paradox of the antagonistic production of hydrogen and oxygen (O 2 inhibits H 2 production)."
} | 2,578 |
30906420 | PMC6426328 | pmc | 802 | {
"abstract": "Biofilms are communities of bacteria adhered to surfaces. Recently, biofilms of rod-shaped bacteria were observed at single-cell resolution and shown to develop from a disordered, two-dimensional layer of founder cells into a three-dimensional structure with a vertically-aligned core. Here, we elucidate the physical mechanism underpinning this transition using a combination of agent-based and continuum modeling. We find that verticalization proceeds through a series of localized mechanical instabilities on the cellular scale. For short cells, these instabilities are primarily triggered by cell division, whereas long cells are more likely to be peeled off the surface by nearby vertical cells, creating an “inverse domino effect”. The interplay between cell growth and cell verticalization gives rise to an exotic mechanical state in which the effective surface pressure becomes constant throughout the growing core of the biofilm surface layer. This dynamical isobaricity determines the expansion speed of a biofilm cluster and thereby governs how cells access the third dimension. In particular, theory predicts that a longer average cell length yields more rapidly expanding, flatter biofilms. We experimentally show that such changes in biofilm development occur by exploiting chemicals that modulate cell length.",
"discussion": "Discussion Our results suggest that bacteria have harnessed the physics of mechanical instabilities to enable the generation of spatially and orientationally patterned architectures. Here, we showed that cell verticalization begins to occur when the local effective surface pressures that arise from cell growth become large enough to overcome the cell-to-surface adhesion that otherwise favors a horizontal orientation. Subsequently, the reduction in cell footprint that occurs upon cell verticalization, which acts to reduce the effective surface pressure, provides a mechanical feedback that controls the rate of biofilm expansion. We expect that individual cell parameters have evolved in response to selective pressures on global biofilm morphology, e.g. during resource competition 6 , 35 – 38 . Since optimal morphology may be condition dependent, cells may also have evolved adaptive strategies that alter biofilm architecture, which could be investigated experimentally by screening for environmental influences on cell size, shape, and surface adhesion 39 . For simplicity, we focused on flat surfaces, nutrient-rich conditions, and V. cholerae strains that have been engineered to have simpler interactions than those in wild type biofilms (Methods). Moreover, our agent-based model does not explicitly incorporate the VPS matrix secreted by cells 2 , 27 , 40 . Understanding the modifying effects of the VPS matrix, cell and surface curvature ( Supplementary Fig. 15 ), cell-to-cell adhesion ( Supplementary Fig. 16 ), and chemical feedback 41 will be important directions for future studies. More broadly, we must develop a systematic method to account for the diversity of architectures that can be produced by local mechanical interactions ( Supplementary Discussion ). Our study of a two-fluid model for verticalizing biofilms led us to discover a novel type of front propagation. Interestingly, in the biofilm surface layer, the front profile of cell density is precisely uniform starting at some finite distance from the edge, whereas previous models of front propagation saturate asymptotically toward uniformity 42 – 45 . The self-organized nature of this process yields a universal dependence of the expansion speed on the cell geometrical and mechanical parameters that is robust to details of the mechanical feedback. We have focused on the mean-field behavior of biofilms, but an open question is to understand the role of fluctuations in the “pressure” acting on cells, e.g. either from a jamming perspective 46 , a fluctuating hydrodynamical perspective 47 , 48 , or a combination of approaches. In summary, we have elucidated the physical mechanism underlying a complex developmental program observed at the cellular scale in bacterial biofilms. The relative biochemical and biophysical simplicity of this prokaryotic system allowed us to quantitatively understand the developmental pathway from the scale of a single cell to the scale of a large community assembly. Going forward, we expect bacterial biofilms will take on increasingly important roles as tractable models that can be used to understand how living systems generate and maintain their structures."
} | 1,129 |
40335917 | PMC12060303 | pmc | 804 | {
"abstract": "Background Low salinity is a crucial environmental stressor that affects estuarine coral ecosystems considerably. However, few studies have focused on the effects of low-salinity conditions on coral-associated microorganisms and the adaptability of coral holobionts. Methods We explored the community structure of coral symbiotic Symbiodiniaceae and associated bacteria in low-salinity conditions using samples of six coral species from the Pearl River Estuary and analyzed the adaptability of coral holobionts in estuaries. Results The symbiotic Symbiodiniaceae of all six studied coral species were dominated by Cladocopium, but, the Symbiodiniaceae subclades differed among these coral species. Some coral species (e.g., Acropora solitaryensis ) had a high diversity of symbiotic Symbiodiniaceae but low Symbiodiniaceae density, with different adaptability to low-salinity stress in the Pearl River Estuary. Other coral species (e.g., Plesiastrea versipora ) potentially increased their resistance by associating with specific Symbiodiniaceae subclades and with high Symbiodiniaceae density under low-salinity stress. The microbiome associated with the coral species were dominated by Proteobacteria , Chloroflexi , and Bacteroidetes ; however, its diversity and composition varied among coral species. Some coral species (e.g., Acropora solitaryensis ) had a high diversity of associated bacteria, with different adaptability owing to low-salinity stress. Other coral species (e.g., Plesiastrea versipora ) potentially increased their resistance by having minority bacterial dominance under low-salinity stress. Conclusions High Symbiodiniaceae density and high bacterial diversity may be conducive to increase the tolerance of coral holobiont to low-salinity environments. Different coral species have distinct ways of adapting to low-salinity stress, and this difference is mainly through the dynamic regulation of the coral microbiome by corals. Supplementary Information The online version contains supplementary material available at 10.1186/s12866-025-04013-4.",
"conclusion": "Conclusions Different coral species have different adaptations to low-salinity stress by regulating the diversity and composition of Symbiodiniaceae and bacteria. Some coral species improved their adaptation to low-salinity stress in the Pearl River Estuary based on a high diversity of symbiotic Symbiodiniaceae and associated bacteria. In contrast, some other coral species may increase their resistance by associating with specific Symbiodiniaceae subclades, with high Symbiodiniaceae density, or minority bacterial dominance under low-salinity stress.",
"introduction": "Introduction Coral reefs, known as the “rainforests of the ocean,” are distributed in 110 countries and regions worldwide, accounting for approximately 0.17–0.5% of the global ocean and providing habitat for approximately 30% of marine organisms [ 1 , 2 ]. The association between coral hosts and their microbiome is the basis for coral growth and reef development [ 3 ]. Coral holobionts include cnidarian polyps, bacteria, Symbiodiniaceae, archaea, fungi, and other microorganisms [ 4 , 5 ]. Symbiodiniaceae are critical microorganisms in coral holobionts that use light energy for photosynthesis and provide fixed organic carbon to the coral host [ 6 ]. Symbiodiniaceae can utilize the metabolic waste produced by the coral host to replenish its nitrogen, phosphorus, and other vital nutrients [ 7 ]. Community-associated bacterial communities play key roles in coral holobionts, including nitrogen [ 8 , 9 ], carbon [ 10 ], sulfur [ 11 , 12 ], phosphorus fixation [ 13 ], intrametallic homeostasis [ 13 ], tissue repair [ 13 ], and antibiotic production [ 14 ]. Many coral-associated bacteria also protect corals from invasion by pathogens and other exogenous bacteria by secreting antibiotics [ 14 , 15 ]. Corals are generally considered stenohaline, with limited ability to regulate osmotic pressure and adapt to or survive salinity changes [ 16 ]. Salinity is a critical environmental factor affecting their growth and distribution, and the suitable salinity range for growth is generally 32–40‰ [ 17 ]. Changes in salinity due to climate change may affect the physiology and metabolism of corals [ 18 – 20 ]. The corals have poor ability to regulate cellular osmotic pressure [ 21 ]. A short-term decline in salinity leads to significant changes in cellular respiration and photosynthesis in the coral symbiont Symbiodiniaceae [ 22 ]. Freshwater runoff is a major cause of coral mortality in estuarine coral reefs, particularly in those close to major river systems [ 22 ]. Environmental changes caused by global warming and destructive anthropogenic activity exacerbate damage to coral reef ecosystems and predispose coral holobionts to dysbiosis [ 23 , 24 ], leading to high susceptibility to infections by opportunistic pathogens and coral mortality. Estuarine ecosystems are characterized by complex and variable environments, high biodiversity and productivity, and a high degree of human disturbance [ 25 ]. Global climate change has led to an unusual increase in extreme rainfall events in multiple regions [ 26 – 28 ], resulting in marine organisms frequently experiencing low-salinity stress in estuarine regions. Low salinity is one of the causes of coral bleaching. Many studies have explored the effects of low salinity on the physiological performance of the early stages of coral; for example, low salinity decreases coral larval recruitment and growth rates [ 29 – 31 ]. Low salinity may affect the physiological and biochemical processes, cellular changes, microbiome alterations, as well as the reproductive and survival capabilities of corals [ 32 – 34 ]. Low salinity stress can disrupt the symbiotic relationship within the coral holobiont, influencing the adaptability and stability of the coral holobiont [ 22 ]. In conclusion, Low-salinity stress can severely impact the growth, reproduction, photosynthesis, and respiration of corals, impair the normal functions of cells, and pose a serious threat to the health of corals [ 22 , 35 ]. The sea section of the Pearl River Estuary in Guangdong represents the estuary of the Pearl River, the second largest river in China in annual runoff. Facing the South China Sea, the Pearl River Estuary is a typhoon-prone area and is susceptible to storm surges and flooding outbreaks, resulting in a large influx of freshwater into the estuary [ 36 ]. Corals in the Pearl River Estuary are affected by the year-round influx of freshwater. In the summer of 2022, record-breaking pre-flood rainfall in South China was lasted from May to June, resulting in a severe coral bleaching event in the coral communities of the Wanshan Islands in the Pearl River Estuary [ 37 ]. However, the effects of low-salinity stress on coral holobionts in the Wanshan Islands, and corals’ adaptability to low-salinity environments, are not well studied. In this study, we aimed to answer two fundamental questions: are there differences in coral microorganisms in the coral communities of the Pearl River Estuary, and how well are the coral holobionts adapted to low salinity. We tested two hypotheses: (1) the compositions of coral microorganisms in the coral communities of the Pearl River Estuary vary in low-salinity conditions when compared with high salinity conditions, and (2) the physiological parameters of corals influence coral adaptability to low-salinity. To answer the two questions, we collected local water quality parameters and the coral symbiotic Symbiodiniaceae density, chlorophyll a (Chl a) content, and microbial diversity throughout the Pearl River Estuary. The findings of this study enhance the understanding of changes in estuarine coral reef communities and facilitate the conservation of estuarine corals and further research.",
"discussion": "Discussion Adaptation of coral-Symbiodiniaceae holobionts to the low-salinity in the Pearl River Estuary Owing to the influences of short-term rainstorms and massive freshwater input, low-salinity conditions endure in the Pearl River Estuary. Low-salinity conditions have significant negative impacts on coral-Symbiodiniaceae holobionts, and coral symbiotic Symbiodiniaceae density decreases in low salinity [ 22 , 60 ]. The symbiotic Symbiodiniaceae density reflects differences in the adaptability of coral species in low-salinity conditions. For example, G. fascicularis can normally survive more than 60 d when exposed to 20‰ salinity [ 61 ]; however, the growth rate of Platygyra acuta slows when the salinity is < 26‰ and stops at salinities < 22‰ [ 30 ]. In low-salinity conditions, Liu et al. [ 60 ] found that the Symbiodiniaceae density and Chl a content of Porites lutea decreased significantly when the surrounding seawater salinity was < 30‰, and the Symbiodiniaceae density and Chl a content decreased dramatically when the salinity decreased from 30‰ to 20‰ until the corals bleached. Similarly, in this study, the density and Chl a content of coral symbiotic Symbiodiniaceae differed significantly in the low-salinity conditions of the estuary, with some corals showing high Symbiodiniaceae densities and Chl a content, such as G. fascicularis ; other corals showed low Symbiodiniaceae densities, such as E. aspera . The Symbiodiniaceae density of P. versipora was 6.8 × 10 6 cells/cm 2 [ 39 ], and that of M. peltiformis was 4 × 10 6 cells/cm 2 at normal salinity [ 40 ]. Both corals exhibited a decrease in Symbiodiniaceae density in low-salinity conditions. The Chl a content of A. solitaryensis was 15.21 μg/cm 2 , and that of M. peltiformis was approximately 11 μg/cm 2 at normal salinity [ 41 , 42 ]. In this study, the Chl a content of A. solitaryensis was 2.43 μg/cm 2 , and that of M. peltiformis was 8.09 μg/cm 2 in low-salinity conditions. Both corals showed decreases in Chl a content. These findings suggest that low-salinity conditions affect Symbiodiniaceae density and Chl a content, and differences in Symbiodiniaceae density and Chl a content could be responsible for the different adaptations of corals to low-salinity conditions. Environmental factors drive various types of and different environmental influences in Symbiodiniaceae coral holobionts [ 62 , 63 ]. Studies have found that the dominant symbiotic Symbiodiniaceae is closely related to the tolerance of coral-Symbiodiniaceae holobionts to environmental stress [ 64 – 66 ]. Symbiodiniaceae subclades confer different physiological characteristics and environmental tolerances to coral holobionts [ 65 , 67 , 68 ]. In this study, the dominant symbiotic Symbiodiniaceae subclades in all corals were Cladocopium C1, excluding M. peltiformis and E. aspera . Some corals with Cladocopium C1-dominant species had a lower percentage of Cladocopium C1 than corals in normal salinity conditions, for example, P. versipora with 75% in normal salinity and 41.5% in this experiment [ 43 ], and some coral Cladocopium C1 percentage gaps do not vary much, for example, G. fascicularis , with approximately 80% in normal and low-salinity conditions [ 43 ]. Cladocopium C1 is a subclade of Symbiodiniaceae with high photosynthetic efficiency [ 69 ]. Under the low-salinity stress, corals may establish a symbiotic relationship with highly photosynthetically efficient Symbiodiniaceae to maintain the functioning of the calcium carbonate secretion-storage system [ 70 ]. However, some of the dominant symbiotic Symbiodiniaceae subclades change from corals between the normal salinity and low-salinity conditions. In normal conditions, for example, the dominant symbiotic Symbiodiniaceae subclades of the M. peltiformis were Cladocopium C3d, followed by Cladocopium C2r [ 44 ]. In low-salinity conditions, the dominant symbiotic Symbiodiniaceae subclades were Cladocopium C3d, followed by Cladocopium Cspc. Notably, some corals were symbiotic with different symbiotic Symbiodiniaceae subclades when experienced low-salinity stress. For example, Cladocopium Cspc and C3d were detected from M. peltiformis in low salinity conditions, but Cladocopium C3d and C2r were detected in normal salinity conditions. Research has observed that M. peltiformis has high susceptibility to environmental influences [ 45 ]. The flexible changes in the composition of symbiotic Symbiodiniaceae subclades may be conducive M. peltiformis to survive and adapt to low-salinity stress. Coral symbiotic Symbiodiniaceae types are typically determined by environmental selection and co-evolution [ 71 ]. The results of this study showed that among the six coral species, some of them differed significantly in their Symbiodiniaceae diversity and composition. Coral holobionts can regulate physiological and ecological characteristics (e.g., coral host stress, Symbiodiniaceae densities, Symbiodiniaceae subclades, and bacterial communities) to adapt to external environmental changes [ 72 , 73 ]. The results suggest that different corals adapt to low salinity by regulating their symbiotic relationship with Symbiodiniaceae. Based on the phylogenetic tree of Symbiodiniaceae, two typical symbiotic Symbiodiniaceae species selected by corals survived in low-salinity conditions. For example, the Symbiodiniaceae subclades of M. peltiformis belong to similar branches of the phylogenetic tree, and those of A. solitaryensis are distributed throughout the phylogenetic tree. This suggests that symbiotic Symbiodiniaceae may be regulated by corals in low-salinity conditions. Some coral species establish symbiotic relationships with specific Symbiodiniaceae subclades (e.g., Cladocopium C1) to increase their photosynthetic rate, and other coral species establish symbiotic relationships with multiple Symbiodiniaceae subclades to enhance their Symbiodiniaceae diversity and tolerance to low-salinity conditions. Adaptation of coral-bacterial holobionts to low-salinity Pearl River Estuary condition Salinity is a key factor influencing the diversity and composition of bacterial communities in coastal corals [ 74 , 75 ]. Microbiome dynamics are linked to coral environmental tolerance [ 76 ]. And a high diversity of the bacterial community may contribute to niche complementation and/or functional redundancy [ 23 , 77 ]. In our study, six studied coral species suggested that the bacterial communities associated with each of them showed significant differences in diversity and composition. Specifically, relatively high bacterial diversity was detected in the P. versipora , while relatively low bacterial diversity in the A. solitaryensis . Compared with normal salinity environment, bacterial composition of different corals widely changed in low-salinity environment [ 45 , 46 ]. For example, M. peltiformis showed distinct changes in bacterial composition. At the phylum level, the bacteria associated with M. peltiformis were dominated by Cyanobacteria, with relative abundances of 92.15% in normal salinity environment. At the genus level, the bacteria associated with M. peltiformis were dominated by Synechococcus CC9902 (89.16%) [ 45 ]. However, in low-salinity conditions, the dominant ASVs were Proteobacteria at the phylum level. At the genus level, u_Rhodobacteraceae were the dominant bacteria. Synechococcus spp. is conducive coral holobiont to nitrogen fixation [ 78 ] and photosynthesis [ 79 ], indicating their potential roles in the hologenomic nutrient cycling of corals and the health state of coral hosts. Rhodobacteraceae may also enhance the resilience of corals by absorbing DMSP from Symbiodiniaceae and producing antibacterial compounds against pathogens [ 80 ]. Evidently, low salinity can affect changes in the bacterial composition of corals, which further impacts the health of coral holobionts [ 45 ]. In contrast, the bacterial composition of G. fascicularis was relatively in stable. Under normal-salinity conditions, the relative abundances of γ-Proteobacteria, α-Proteobacteria, and Cyanobacteria of G. fascicularis were 20%, 12%, and 39% respectively, remaining at a relatively high level [ 46 ]. Under low-salinity environment, Proteobacteria becomes were also the dominant bacteria, which is similar to the bacterial composition under normal conditions. This indicates a potential variation between the stability of the microbiome and the susceptibility to bleaching among coral species. The bacterial diversity might be important factors in coral adaptability to low-salinity conditions. Some coral species may survive better than others in low-salinity conditions by maintaining a high level of bacterial diversity. Coral microbiome with high diversity may have high physiological and ecological acclimatization to low-salinity conditions [ 81 ]. Community-associated bacterial communities have high flexibility and diverse functions and are closely related to the health of coral symbiotic functionaries. For example, Proteobacteria and Bacteroidetes have a high tolerance to salinity stress [ 82 ]. In this study, bacteria from Proteobacteria and Bacteroidetes tended to dominate the coral-associated communities, and their insensitivity to salinity allowed these bacteria to survive better than other bacteria in low-salinity conditions. Rhodobacteraceae were highly abundant in all six coral species, and their ability to grow through photosynthesis, metabolize CO 2 , and fix nitrogen plays an important role in the carbon and nitrogen cycles of marine ecosystems [ 83 ]. In this study, additional differences were observed in the relative abundances of bacterial phyla and genera, reflecting varying adaptability among coral species in low-salinity conditions. Coral-associated bacteria may be related to coral growth, nutrient metabolism, the immune system, antioxidant capacity, resilience, and tolerance [ 84 , 85 ]. In our study, the coral microbial communities in low-salinity areas revealed that Pseudoalteromonas was extremely abundant in some corals, such as P. daedalea . Pseudoalteromonas has been shown to have significant antagonistic effects on various pathogens [ 86 ]. In this study, Pseudoalteromonas and Vibrio were detected in the composition of coral-associated bacteria. Pseudoalteromonas may be conducive to inhibit the growth and reproduction of Vibrio via antagonism, potentially decreasing the risk of disease in corals [ 87 ]. Furthermore, a high abundance of Ruegeria was found in P. daedalea and M. peltiformis . Ruegeria is a potential probiotic coral that produces antibiotics that can inhibit the growth of Vibrio [ 88 ]. High abundance of Vibrio in low-salinity conditions may interfere with the associated relationship between corals and bacteria [ 89 , 90 ]; the antagonistic effect of Pseudoalteromonas and the antibiotics produced by Ruegeria could be conducive to inhibiting the growth of Vibrio , maintaining the stability of the relationship and the mutual cooperation between corals and bacteria. Under low-salinity stress, the increased abundance of specific bacterial taxa may contribute to the health and survival of corals [ 91 , 92 ]."
} | 4,785 |
39915510 | PMC11802840 | pmc | 805 | {
"abstract": "Coral-associated microbes have essential roles in promoting and regulating host function and health. As climate change advances and other environmental perturbations increasingly impact corals, it is becoming ever more important that we understand the composition of the microbial communities hosted. Without this baseline it is impossible to assess the magnitude and direction of any future changes in microbial community structure. Here, we characterised both the bacterial and Symbiodiniaceae communities in four coral species ( Diploastrea heliopora, Porites lutea, Pachyseris speciosa , and Pocillopora acuta ) collected from Sabah, Malaysia. Our findings reveal distinct microbial communities associated with different coral species tending to reflect the varied life history strategies of their hosts. Microbial communities could be differentiated by collection site, with shifts in Symbiodiniaceae communities towards more stress tolerant types seen in samples collected on the shallow Sunda Shelf. Additionally, we identified a core microbiome within species and a more discrete core between all species. We show bacterial and Symbiodiniaceae communities are structured by host species and appear to be influenced by host life history characteristics. Furthermore, we identified a core microbiome for each species finding that several amplicon sequence variants were shared between hosts, this suggests a key role in coral health regardless of species identity. Given the paucity of work performed in megadiverse regions such as the Coral Triangle, this research takes on increased importance in our efforts to understand how the coral holobiont functions and how it could be altered as climate change advances.",
"introduction": "Introduction Corals are keystone species in reef habitats, serving a multitude of roles in maintaining a healthy ecosystem 1 . In performing these roles, they are reliant on the symbioses that exist between the host and its microbial constituents, including, but not limited to bacteria and Symbiodiniaceae 2 , 3 . These symbiotic relationships are crucial for the healthy functioning of the coral holobiont, with bacteria involved in nutrient cycling, metabolism and immune system functioning 4 , 5 , while Symbiodiniaceae provide a significant proportion of their hosts nutritional requirements through photoautotrophy 6 . Importantly, these symbionts can also help promote adaptation to environmental change 7 – 11 . Given this, an understanding of the symbiotic microbial communities and how they may change is an important consideration in efforts to understand how corals are expected to adapt as climate change advances. However, without baseline work describing the current community structure and microbial compositions, determining the magnitude and direction of any changes is impossible. Corals display a variety of life-history strategies that can influence microbial community composition 12 , with coral species, life-history and traits (e.g., competitive, weedy, stress-tolerant, opportunistic or generalist) all influencing microbial community assemblage 13 – 15 . Competitive species have fast growth rates and reproduce via broadcast spawning, whereas species classified as weedy tend to be brooders with smaller colony sizes and shorter generation times. Stress tolerant species have slow growth rates tending to reproduce by broadcast spawning and are highly fecund, with species described as generalist or opportunistic having moderate growth rates and a degree of overlap with competitive, weedy and stress-tolerant life histories, for a complete description of coral life history traits see Darling et al. 2012 16 . Additionally, in terms of how corals shape their microbial community, they can be divided into two groups (i.e., microbial conformers or regulators) 12 . Conformers have a microbiome that changes in response to the environment, while regulators are able to maintain stable microbiomes across environments. Similarly, some microbial taxa show specialized versus generalist behaviors. Members of the Symbiodiniaceae genus, Durusdinium, specifically D. trenchii are considered more heat-tolerant and can confer an advantage to their host in elevated temperatures 17 – 19 , and some bacteria are implicated in the suppression of bacterial pathogens and play a role in maintaining a healthy microbiome, while others are involved in sulphur and nutrient cycling processes 20 . Here, we investigate the bacterial and Symbiombiodiniaceae communities of four coral species, Diploastrea heliopora , Porites lutea , Pachyseris speciosa and Pocillopora acuta , collected around the Malaysian state of Sabah (Borneo). Specifically, we determine whether (1) these communities show local spatial patterns and host-specificity, and (2) a ‘core microbiome’ is present within and amongst species. The concept of a ‘core microbiome’ was initially explored in humans where it was defined as a group of microbial taxa shared by all, or most humans 21 . The shared nature of these microbes led to suggestions of an important role in the maintenance of host functioning 22 , similar ideas have been applied in natural ecosystems where it is hypothesised that this core has roles in the maintenance of ecological functioning 9 , 23 , 24 . Here we apply this concept in an effort to determine whether a ‘core microbiome’ exists between the four coral host species collected in this work. We hypothesise that the bacterial and Symbiodiniaceae communities will differ by host species, a consequence of their different life history strategies e.g., D. heliopora and P. lutea demonstrate stress tolerant strategies and traits, while P. speciosa is a generalist 16 , 25 and P. acuta is classed as an opportunistic species 26 . Additionally, but to a lesser extent in comparison to host species, we expect that site specific environmental conditions will influence microbial community composition 27 . This study seeks to characterise the microbial diversity associated with reef-building corals in an area that remains understudied but contains highly diverse and rich coral reef communities. Much of the world’s coral reef biodiversity resides in the global south, yet much of the research capacity is focused on the much less diverse and arguably less healthy reefs of the global north 28 . If coral reefs are to persist into the future it is critical that we work on understanding these megadiverse marine ecosystems and their essential microbial constituents, especially those at the epicenter of global marine biodiversity, for example the Coral Triangle and its surrounding regions. It is here that the evolutionary novelty required to adapt to a rapidly changing climate likely resides.",
"discussion": "Discussion This is the first examination and characterization of coral associated microbes to be performed around the Malaysian state of Sabah, on the island of Borneo. Collections spanned a distance of nearly 1000 km and include part of the Coral Triangle diversity hotspot as well as the South China Sea. Unsurprisingly, and similar to other studies performed throughout the Southeast Asian region 26 , 27 , 51 we show that bacterial and Symbiodiniaceae communities from the four collected host species of coral are structured by, and can be differentiated by geographic location and species. Bacterial community patterns Life history characteristics of the coral host can be useful when describing the relationship between host and its associated microbial communities, for example, Diploastrea heliopora and Porites lutea are considered stress tolerant species 52 , Pocillopora acuta is described as an opportunistic colonizer, while Pachyseris speciosa is considered a generalist 53 , 54 . Broadly in line with these characteristics, and as previously reported 26 we find that the more stress tolerant species tend to have less diverse microbial communities, whereas the more opportunistic and generalist species have a more diverse microbial communities which should allow them to persist in a wider range of conditions. This pattern is evident in bacterial communities with diversity highest in the generalist, P. speciosa. But, contrary to expectations the opportunistic colonizer P. acuta has the lowest bacterial diversity. Further work is needed to confirm this, but it is possible that P. acuta is a microbiome regulator and as such is able to select its microbial constituents. Adding weight to this suggestion, the low microbial diversity is indicative of a microbiome that has been tailored to host specific requirements through microbial winnowing 55 with similar, low bacterial diversity reported in this species previously 56 , 57 . These species-specific differences are supported by our ordinations that show well defined clusters indicating differences in the hosted bacterial communities between coral species. Diploastrea heliopora shows the lowest degree of differentiation in bacterial community structure, this is consistent with work performed in Peninsular Malaysia 48 and could be a consequence of stress tolerant species requiring fewer symbiotic partners for survival 26 , 58 . A similar, but less pronounced pattern is seen in P. lutea, another species described as stress tolerant. Bacterial community structure is more defined in Pocillopora acuta and Pachyseris speciosa , described as opportunistic or as a generalist respectively. It is likely that the life history characteristics of P. acuta and P. speciosa allows these species to take advantage of disturbance or exist in a wide range of habitats, to do this, it is probable they acquire bacteria that are prevalent in the local environment, hence we see clear differentiation in community structure between sampling locations. In fact, a number of studies exploring the microbiome of P. acuta and how it responds to new environments when transplanted indicate that it is strongly influenced by and quickly conforms to local conditions 59 – 61 . The bacterial communities in all four coral species are dominated by Proteobacteria, and to a lesser extent Bacteroidota, Cyanobacteria, and Firmicutes. These general associations are well documented in corals throughout the world 11 , 14 , 62 – 65 . The bacterial family Colwelliaceae observed in P. speciosa and P. lutea has been implicated in buffering the negative effects of photooxidative stress linked to bleaching 66 , this family would be a good candidate for future research attempting to determine how microbes can contribute to coral stress adaptation. This association fits well with the generalist and stress tolerant life histories these corals have, and may in part be responsible for their ability to persist in stressful conditions. Symbiodiniaceae community patterns Coral-associated symbidiniaceae communities typically vary by species and geographic location 26 , 67 . The Porites lutea colonies examined here are dominated by Symbiodiniaceae type C15 and its variants, this C15 association is found throughout Southeast Asia 52 , 67 – 69 and more generally throughout the world 70 . Similar to other work performed in the South China Sea 71 we find the Diploastrea heliopora host is dominated by type C3u/C3/C115 at three of the four sites, whereas this type is absent from Mataking and of the four sites type C40/C3 is only present here. At all sites, and as previously reported from the Malay Peninsula 26 , 58 , type C27 is most prevalent in Pachyseris speciosa, with type C21 only found at Mataking. Pocillopora acuta is dominated by types D1 and D1b in three of the sites, whereas Mataking is dominated by types C1d and C1d/C1. The Symbiodiniaceae community composition of P. lutea shows little variation across all four sites, while the other three species have similar Symbiodiniaceae communities at Kota Kinabalu, Labuan and Lankayan, whereas Symbiodiniaceae communities differ considerably in P. speciosa , D. heliopora and P. acuta collected from Mataking in comparison to the other sample sites (Fig. 5 b). This difference is particularly striking in P. acuta where a large shift from Durusdinium to Cladocopium is seen. Further work is needed to confirm this and with the data we have, we can only speculate why the samples collected from Mataking are frequently different. Unlike the other collection sites, Mataking is in close proximity to the deep waters of the Celebes Sea, whereas the other three sites are in shallower coastal water associated with the Sunda Shelf. Shallower water does heat up faster and corals found here could experience warmer waters and have shifted their Symbiodiniaceae communities in response to the more stress tolerant Durusdinium , members of this genus, particularly D. trenchii have been shown to confer an advantage in warmer waters 17 , 18 , 71 . Similar phenomena has previously been reported in Pocillopora spp from the eastern tropical pacific and has been suggested as a mechanism that may confer increased resilience to ocean warming 11 . Whereas the deeper water of the Celebes Sea could buffer and prevent heat stress in corals from Mataking, consequently we see a different Symbiodiniaceae community, one that is adapted to the different conditions experienced here. This site is also the only one within the Coral Triangle biodiversity hotspot, an area where coral biodiversity reaches its maxima 72 . Work shows that the composition of the coral community in the immediate area can influence the endosymbionts found in a host 73 . Given the elevated coral diversity of the Coral Triangle region, this could, in part also be responsible for the different Symbiodiniaceae communities seen at Mataking and could play a role in structuring Symbiodiniaceae communities in hosts that acquire their symbionts from the environment, rather than through vertical transmission. However, it should be noted that despite having significantly lower coral host diversity, previous work suggests the Caribbean actually has more Symbiodiniaceae species than the Indo-Pacific 74 . Although, this may change as high throughput DNA sequencing approaches are increasingly applied to examine Symbiodiniaceae communities and diversity throughout the Coral Triangle and other hotspots of marine biodiversity. Both D. heliopora and P. speciosa have more mixed Symbiodiniaceae profiles at Kota Kinabalu in comparison to P. acuata and P. lutea, mixed communities have been implicated as advantageous in certain environmental conditions such as elevated temperatures or low light environments 75 , 76 . As the capital of Sabah, it is not unreasonable to suggest that the corals found at Kota Kinabalu are subject to the increased anthropogenic stresses associated with a human population that is larger than that at other locations in Sabah (e.g., terrestrial runoff). Further work is needed to confirm this, but, this mixed profile maybe a response to the increased stresses unique to Kota Kinabulu. Core microbiome Six core microbial ASVs are shared between three of the four corals—ASV2, ASV67, ASV1, ASV3, ASV6, and ASV7. These ASVs have been identified in other work examining coral microbial communities, this ubiquity suggests a role in promoting host health. ASV2 ( Alteromonas spp ) and ASV67 ( Synechococcus CC9902 spp ) are shared between P. speciosa , P. acuta , and P. lutea , while the remaining four are shared among D. heliopora , P. speciosa , and P. lutea . Pseudoalteromonas (ASV 7) and Vibrio (ASVs 1 & 3) have been observed associated with numerous coral hosts, suggesting important roles of these bacteria in maintaining microbiome functioning 48 , 77 , 78 . ASVs 6 ( Algicola bacteriolytica ) & 7 ( Pseudoalteromonas spp ) are involved in dimethylsulfoniopropionate (DMSP) metabolism, DMSP metabolizers are part of the Beneficial Microorganisms for Corals (BMC) group and have been identified as having important roles in nutrient cycling and preventing bacterial pathogen outbreak, helping to maintain the hosts microbiome 10 . ASV2 has been implicated as having potential roles in nitrogen fixation and could offer pathways of alternative fixation during periods of stress (e.g., thermal bleaching) and thus increase coral host resilience 79 . ASV67 has been recovered from coral tissue where it is thought to have roles in nutrient uptake 80 – 82 . Studies such as this are increasingly common place throughout the world, however, they remain comparatively rare in the global epicenter of marine biodiversity. Given this, studies such as the one performed here are valuable as they provide an initial foundation upon which we can build and develop more hypothesis driven research. For example, future work will more specifically examine differences in environmental conditions and coral community composition between sampling sites around the island of Borneo to determine why differences in microbial community composition exist. Additionally, as coral reef restoration activities become increasingly popular, it will become more important that we understand the diversity contained within an ecosystem, for example, if we do not know what is there already, how will we determine if restoration is successful? Additionally, a more complete understanding of the microbial constituents that are present in marine ecosystems could help restoration practitioners determine and prioritize the most suitable candidate sites for restoration."
} | 4,368 |
39835901 | PMC7617349 | pmc | 806 | {
"abstract": "Significance Experiments have shown that when one plant is attacked by a herbivore, this can lead to other plants connected to the same mycorrhizal network up-regulating their defense mechanisms. It has been hypothesized that this represents signaling, with attacked plants producing a signal to warn other plants of impending harm. We found theoretically that plant warning signals are rarely evolutionarily stable. Instead, we identify two viable alternatives that could explain the empirical data: 1) being attacked leads to a cue (information about the attack) which is too costly for the attacked plant to suppress; 2) mycorrhizal fungi monitor their host plants, detect when they are attacked, and then the fungi signal this information to warn other plants in their network.",
"discussion": "Discussion We have applied a body of signaling theory that has been well developed to explore animal behavior to a different context, information transfer between plants. Our results show that plants: 1) are unlikely to be selected to signal to their neighbors about the presence of herbivores ( Fig. 2 ); 2) can be favored to produce dishonest signals, where they signal an attack when none has occurred, but that this will select for other plants to ignore signals; and 3) can be favored to pay resources to suppress any information (cue) to neighbors about attack ( Fig. 3 A ). In contrast, mycorrhizal fungi can be selected to monitor their host plants, detect when they are attacked, and then signal (warn) other plants in their network ( Fig. 3 B ). Our results do not support the hypothesis of warning signals by plants, passed via any route, including common mycorrhizal networks or the air ( 2 , 11 – 24 , 27 , 81 ). This is because these warnings would benefit neighboring competitors, to the cost of the signaling individual. Furthermore, we found that not only are plants not expected to signal but that they can be selected to signal dishonestly or to actively suppress any cues of being attacked. Dishonest signaling or suppression of cues is favored to harm or avoid helping neighboring competitors. Empirically, there is little evidence for plant–plant honest signaling (intraspecific), though plants can be favored to signal honestly to their pollinators and seed dispersers (interspecific), who they are not in direct competition with ( 34 , 82 – 86 ). For a helping behavior to be favored, such as signaling a warning of herbivore or pathogen attack, we showed that this would require that helping and competition occur at different scales (economic neighborhoods) or some method of kin recognition / discrimination ( 30 , 31 , 46 , 47 , 55 , 87 – 89 ). Helping and competition occurring at different scales is relatively unlikely for plants because they are immobile, meaning local interactions involve both cooperation and competition for resources, disfavoring warning signaling ( 90 ). This is not always the case in mobile organisms such as animals and bacteria. The same problem of local competition has been demonstrated empirically in other organisms, such as when fig wasps compete for mates in the closed environment of a fig fruit, or when bacteria secrete “public goods” ( 91 – 93 ). However, the problem of local competition can be overcome in mobile organisms, if helping occurs between relatives before they disperse to compete with nonrelatives ( 94 – 96 ). In animals, warnings about the presence of predators have also been argued to be favored because they also reduce predation on the individual making the warning call—this is different from plants, where the warning arises after attack ( 97 – 99 ). In contrast, local competition favors harming behaviors, because harming neighbors can decrease competition for resources ( 31 , 100 – 102 ). Alternative modeling approaches to examine these issues could include explicit spatial structures, such as on a graph or lattice, but these have been shown to lead to analogous results ( 87 , 103 – 105 ). Information about herbivore attack could potentially occur through mycorrhizal networks or volatiles in the air ( 33 , 34 , 74 ). The production of herbivore-induced plant volatiles may be unavoidable and does not seem to confer a fitness benefit on the producer ( 34 ), leading to the suggestion that herbivore-induced plant volatiles are likely to represent cues rather than signals ( 62 – 75 ). This is consistent with our plant signaling and cue-suppression models, which did not make any assumptions about how information about herbivore attack is transferred. More generally, there may be many ways for information about herbivore attack to be transferred, suppressed, directed toward kin, etc., with some mechanisms more biologically plausible than others ( 34 ). Our intention has been to examine the evolutionary stability of different forms of information transfer, in a way that could be applied to a diversity of proximate mechanisms ( 106 ). In contrast to the situation for plants, we found that mycorrhizal fungi can be favored to monitor their host plants, detect a cue of when they are attacked, and then signal this to (warn) other plants in their network ( 13 ). Fungi are selected to monitor and signal because defended plants will maintain better condition and hence become better trade partners. Previous theory has shown that selection for fungi and plants to trade resources with each other is increased when multiple plants and fungi interact in the same network, because this stabilizes efficient trading ( 77 ). To conclude, we examined hypotheses explaining the empirical result that, when one plant in a mycorrhizal network is attacked, this leads to other plants in the network up-regulating their defense mechanisms ( 1 – 5 , 8 ). Our modeling suggests that this is more likely to represent either a cue produced by plants that is too costly to suppress, or fungi monitoring plants, and then signaling to other plants. Further experiments could test between these possibilities, by examining the underlying mechanism in networks or experimental multiple root systems. How is information conveyed? Where does that information arise from? What are the fitness consequences for all the individuals involved? A greater understanding of these mechanisms could potentially also be exploited in an agricultural context, by facilitating plant defense against herbivores."
} | 1,593 |
39763704 | PMC11702357 | pmc | 807 | {
"abstract": "Marine heatwaves are starting to occur several times a decade, yet we do not understand the effect this has on corals across biological scales. This study combines tissue-, organism-, and community-level analyses to investigate the effects of a marine heatwave on reef-building corals. Adjacent conspecific pairs of coral colonies of Montipora capitata and Porites compressa that showed contrasting phenotypic responses ( i.e. , bleached vs . not bleached) were first identified during a marine heatwave that occurred in 2015 in Kāne’ohe Bay, Hawai‘ i. These conspecific pairs of bleaching-resistant and bleaching-susceptible colonies were sampled for histology and photographed before, during, and after a subsequent marine heatwave that occurred in 2019. Histology samples were quantified for: (i) abundance of mesenterial filaments, (ii) tissue structural integrity, (iii) clarity of epidermis, and (iv) cellular integrity (lack of necrosis/granulation) on a 1–5 scale and averaged for an overall tissue integrity score. Tissue integrity scores revealed a significant decline in overall tissue health during the 2019 heatwave relative to the months prior to the heatwave for individuals of both species, regardless of past bleaching history in 2015 or bleaching severity during the 2019 heatwave. Coral tissue integrity scores were then compared to concurrent colony bleaching severity, which revealed that tissue integrity was significantly correlated with colony bleaching severity and suggests that the stability of the symbiosis is related to host tissue health. Colony partial mortality was also quantified as the cumulative proportion of each colony that appeared dead 2.5 years following the 2019 bleaching event, and tissue integrity during the heatwave was found to be strongly predictive of the extent of partial mortality following the heatwave for M. capitata but not P. compressa , the latter of which suffered little to no mortality. Surprisingly, bleaching severity and partial mortality were not significantly correlated for either species, suggesting that tissue integrity was a better predictor of mortality than bleaching severity in M. capitata . Despite negative effects of heat stress at the tissue- and colony-level, no significant changes in coral cover were detected, indicating resilience at the community level. However, declines in tissue integrity in response to heat stress that are not accompanied by a visible bleaching response may still have long-term consequences for fitness, and this is an important area of future investigation as heat stress is commonly associated with long-term decreases in coral fecundity and growth. Our results suggest that histology is a valuable tool for revealing the harmful effects of marine heatwaves on corals before they are visually evident as bleaching, and may thus improve the predictability of ecosystem changes following climate change-driven heat stress by providing a more comprehensive assessment of coral health.",
"conclusion": "Conclusions and future directions Histological analysis of coral tissues revealed signs of stress in the absence of visual symptoms of bleaching and onset of tissue stress prior to the accumulation of heat stress on the reef (as measured by degree heating weeks) in two distantly related reef-building coral species. These results indicate that histology is a valuable method for detecting coral stress before or in the absence of a visible stress response, and could be a useful tool for predicting coral health and mortality following heat stress. Marine heatwaves pose a major threat to the conservation of coral reefs, yet there is hope that the acclimatization of corals, in conjunction with proper management strategies, continued research across biological scales, and global policy to limit greenhouse gas emissions, can ensure a future for these unique ecosystems.",
"introduction": "Introduction Coral bleaching occurs when abnormally high ocean temperatures cause corals to lose their symbiotic dinoflagellate algae (Symbiodiniaceae). These instances of elevated ocean temperatures, known as marine heatwaves, are increasing in both frequency and intensity due to increased atmospheric carbon dioxide ( Oliver et al., 2018 ). The increasing occurrence of marine heatwaves has led to multiple global coral bleaching events within the last decade ( Heron et al., 2016 ). Coral bleaching events have major impacts on the coral individual as well as the ecosystem, as coral symbionts provide energy to coral colonies, the building blocks of coral reefs. These impacts range from the changes in community composition ( Bellwood et al., 2006 ), reductions in habitat complexity ( Hughes et al., 2018b ), and alterations to ecological function such as nutrient and energy cycling ( Graham et al., 2006 ). Economic services of coral reefs are also threatened by coral bleaching, and include fisheries production, coastal protection, tourism, pharmaceutical potential, and more ( Costanza et al., 2014 ). To preserve these valuable ecosystems under a rapidly changing climate, more research is needed to better understand the full effects of recurring marine heatwaves on corals and how coral reefs may respond to anthropogenic stressors in the future. Many questions remain surrounding the long-term health effects of heat stress on coral colonies and reef-wide changes following multiple bleaching events. Coral bleaching does not always translate to coral mortality, as coral colonies can recover and regain their symbionts once heat stress subsides ( Hughes et al., 2018a ; Matsuda et al., 2020 ). Kāne’ohe Bay, Hawai‘ i provides a useful location to address these questions, as the reef system has now experienced four marine heatwaves resulting in mass coral bleaching since 1996 ( Bahr, Rodgers & Jokiel, 2017 ; Brown et al., 2023 ). In the summers of 2014 and 2015, two successive marine heatwaves occurred, resulting in a range of coral bleaching responses within and between species. Specifically, some coral colonies of the dominant reef-building species Montipora capitata and Porites compressa severely bleached while others remained fully pigmented, despite being directly adjacent to one another and seemingly experiencing the same environmental conditions ( Cunning, Ritson-Williams & Gates, 2016 ; Matsuda et al., 2020 ; Ritson-Williams & Gates, 2020 ). Impacts of the 2015 marine heatwave were also apparent in M. capitata several years later, as colonies that had remained pigmented ( i.e., bleaching-resistant) and those that had bleached ( i.e., bleaching-susceptible) exhibited distinct metabolomic signatures from each other ( Roach et al., 2021 ). In 2019, the corals in Kāne’ohe Bay experienced a third marine heatwave in under a decade ( Innis et al., 2021 ; Yadav et al., 2023 ; Brown et al., 2023 ), which resulted in 19% of P. compressa and 23% of M. capitata experiencing moderate to severe bleaching across the bay ( Yadav et al., 2023 ). Furthermore, the 2019 heatwave led to metabolic depression in M. capitata and P. compressa regardless of bleaching phenotype or prior bleaching history, although greater declines in metabolism were observed in the bleaching-susceptible individuals ( Innis et al., 2021 ). The absence of visible bleaching in pigmented corals therefore does not indicate the absence of stress, and more needs to be learned about the consequences of marine heatwaves on both bleaching-susceptible and bleaching-resistant individuals. Histological examination is an informative method for observing the cellular effects of stress on corals that are not readily apparent at the organismal level. Hematoxylin and eosin (H&E) staining is commonly used in observing diseased coral tissue in both M. capitata ( Aeby et al., 2016 ; Burns & Takabayashi, 2011 ; Work & Meteyer, 2014 ) and P. compressa ( Domart-Coulon et al., 2006 ; Sudek et al., 2012 ). Yet, few studies have observed the impacts of heat stress at the tissue-level for these species. Among related genera, tissue analysis during heat stress revealed cellular changes within the host tissue even before a decline in symbiont abundance ( Ainsworth et al., 2008 ). In addition to symbiont loss, a common observation in heat stressed corals is the loss of energetically-costly structures, including mesenterial filaments and epidermal thickness ( Brown, Le Tissier & Bythell, 1995 ; Hayes & Bush, 1990 ; Szmant & Gassman, 1990 ; Traylor-Knowles, 2019 ). In addition, overall poor staining uptake is associated with host cell necrosis and lack of tissue integrity due to heat stress ( Traylor-Knowles, Rose & Palumbi, 2017 ). Further, H&E staining can be used to observe reproductive features and determine potential trade-offs during environmental stress ( Henley et al., 2022 ; Sudek et al., 2012 ). As such, histological examination of bleaching-susceptible and bleaching-resistant corals throughout a marine heatwave may provide important information about the health and integrity of corals with contrasting bleaching phenotypes at the tissue and cellular levels. Here, we used histology to determine the impact of the 2019 marine heatwave on coral health for two reef-building species from Kāne’ohe Bay, M. capitata and P. compressa . We examined the responses of individuals from each species with contrasting bleaching phenotypes during the 2015 marine heatwave ( i.e., bleached versus pigmented; Matsuda et al., 2020 ) to the 2019 marine heatwave and the following recovery period. In addition, we compared the tissue-level responses with colony-level bleaching severity and partial mortality as well as reef-wide changes in bleaching prevalence and live coral cover. This study thus compares the effects of the 2019 marine heatwave across multiple biological scales, taking a holistic approach to reef monitoring to better understand the future of coral reefs under global change.",
"discussion": "Discussion Coral tissue integrity declined during a marine heatwave regardless of bleaching phenotype All colonies of M. capitata and P. compressa showed decreases in tissue integrity during the 2019 marine heatwave, despite a lack of visible coral bleaching in bleaching-resistant colonies. These results complement physiological and metabolic measurements on these same individuals during the peak of the 2019 marine heatwave, where metabolic depression and declines in photochemical capacity occurred regardless of visually-observed bleaching ( Innis et al., 2021 ). In our study, both bleaching-resistant and bleaching-susceptible M. capitata and P. compressa showed declines in tissue integrity that manifested as poor stain uptake, declines in mesenterial filaments, and reduced clarity of the epidermis. These results align with several earlier studies that found abnormal tissue architecture and poor stain uptake in coral tissue during heat stress ( Brown, Le Tissier & Bythell, 1995 ; Traylor-Knowles, Rose & Palumbi, 2017 ; Traylor-Knowles, 2019 ). Both bleaching-susceptible and bleaching-resistant M. capitata were previously found to lose tissue biomass during heat stress, while P. compressa colonies showed the opposite pattern ( Innis et al., 2021 ), yet we did not find species-specific differences in tissue integrity during the heatwave. Further, tissue integrity was negatively correlated with colony-level bleaching across the time series in both species, suggesting that heat stress not only resulted in severe loss of tissue integrity, but was also related to declines in symbiont density. During the 2019 heatwave, only bleaching-susceptible M. capitata showed significant declines in symbiont density ( Innis et al., 2021 ). Significant symbiont loss accompanied by decreased tissue integrity and cell necrosis may explain a switch in trophic strategies from autotrophically produced energy to the catabolism of tissue biomass, which would be expected to manifest as reduced levels of proteins, lipids, and carbohydrates during marine heatwaves ( Grottoli, Rodrigues & Palardy, 2006 ; Rädecker et al., 2021 ; Schoepf et al., 2015 ). Interestingly, the effects of heat stress were evident at the tissue-level in M. capitata before the accumulation of heat stress ( i.e., degree heating weeks), similar to a study on Acropora aspera ( Ainsworth et al., 2008 ), suggesting that even short-term incursions of temperature stress above the local maximum monthly mean (MMM) negatively influences coral health. Tissue samples collected approximately two years after the 2019 marine heatwave (March 2022) were the most intact, showing little to no signs of tissue stress and the appearance of distinct reproductive features. Whether the observed patterns were a reflection of seasonality in tissue biomass ( e.g. , Scheufen, Iglesias-Prieto & Enríquez, 2017 ) cannot be determined; however, we would not expect seasonal declines in biomass in the summer to be accompanied by the severe declines in tissue integrity that were observed during the heatwave. Reproductive declines may also result from heat stress observable at the tissue-level ( Johnston et al., 2020 ; Rodrigues & Padilla-Gamiño, 2022 ), although our data were insufficient to quantify these responses. Specifically, oocyte abundance and size did not significantly differ between bleaching-susceptible and bleaching-resistant individuals within either species; however, this may be due to a limited sample size and calls for higher-frequency sampling across gametogenesis. Species-specific differences in oocyte abundance and size were most likely a result of divergent reproductive strategies, as P. compressa is gonochoric (individuals are different sexes) and M. capitata is hermaphroditic, and both species have different reproductive timelines ( Henley et al., 2022 ; Neves, 2000 ). Sampling across seasons would help tease apart the effects of heat stress from seasonality, and represents an important avenue of future study. Decreased tissue integrity suggests a host-specific response to marine heatwaves Although decreased tissue integrity correlated with bleaching severity, there were no significant differences in tissue health between bleaching-susceptible and bleaching-resistant corals. Interestingly, coral tissue integrity of all individuals showed complete recovery two years following the 2019 marine heatwave, indicating repair of tissue damage and recovery from heat stress for all colonies regardless of historical bleaching and symbiont loss during the heatwave. Differences in bleaching susceptibility during marine heatwaves may be explained by the different species of symbionts hosted by the corals, as M. capitata can host symbionts of two genera: Durisdinium and Cladocopium ( Cunning, Ritson-Williams & Gates, 2016 ). Bleaching-resistant M. capitata have been shown to host a mixed community of Cladocopium and Durisdinium with a higher concentration of the heat tolerant Durisdinium , whereas bleaching-susceptible M. capitata hosted only Cladocopium ( Drury et al., 2022a ). Additionally, following artificial and natural heat stress, corals with mixed communities of symbionts decreased in Cladocopium abundance and shifted towards Durisdinium -only communities ( Dilworth et al., 2021 ). Such shifts indicate a mode of heat tolerance in relation to the symbiont community in addition to the host genotype ( Drury et al., 2022b ). However, bleaching-resistant corals still exhibited decreases in tissue integrity during the marine heatwave despite hosting more heat tolerant symbionts and appearing to bleach less. Such results indicate a mechanism of heat stress only present in the host that occurs regardless of symbiont community abundance or composition. A similar pattern was observed for P. compressa, which despite hosting only a single species of Cladocopium (C15 ITS2 classification; Putnam et al., 2012 ), exhibited significant tissue damage during heat stress for individuals of both bleaching phenotypes. Bleaching-susceptible P. compressa showed less severe bleaching during the 2019 marine heatwave compared to bleaching-susceptible M. capitata in 2019 ( Fig. 4C ; Innis et al., 2021 ; Brown et al., 2023 ), even though both groups had bleached severely during the 2015 marine heatwave ( Matsuda et al., 2020 ). Given that all colonies of P. compressa remained relatively bleaching-resistant in the 2019 heatwave regardless of bleaching history, this may indicate acclimatization to increased temperatures in which individuals have gained increased resistance to bleaching during heat stress ( Brown et al., 2023 ). In 2019, bleaching-susceptible P. compressa showed mild bleaching and significant metabolic depression but was not sustained long enough to result in decreases in tissue biomass, as was observed in M. capitata ( Innis et al., 2021 ). However, both resistant and susceptible colonies of P. compressa , all of which showed mild to no pigmentation loss, displayed significant declines in tissue integrity. These results indicate that this species underwent significant tissue stress during the heatwave despite the absence of visible bleaching. Optimistically, histology revealed full recovery in tissue integrity among P. compressa two years after the 2019 heatwave, similar to M. capitata tissue recovery. Regardless of symbiont presence and historical bleaching, these coral colonies displayed acute heat stress at the tissue-level, and were able to recover two years following a marine heatwave. Coral mortality and reef-wide changes in benthic community composition Despite significant differences in visually assessed bleaching severity between resistant and susceptible phenotypes of P. compressa during the 2019 heatwave, bleaching severity was mild in susceptible individuals ( Brown et al., 2023 ; Innis et al., 2021 ). This corresponded with low levels of partial mortality, which were not significantly different between bleaching-resistant and bleaching-susceptible phenotypes (−2% with coral growth vs. 0%, respectively). However, all corals exhibited signs of heat stress at the tissue-level, which may explain the partial mortality that occurred during the years following the heatwave even in the absence of visual signs of bleaching. This response is in contrast with the response of these same individuals to the previous 2015 heatwave, where the susceptible corals bleached severely and had higher partial mortality in the two years following the 2015 heatwave (22%) than they did after the 2019 heatwave (0%) or the resistant corals following the 2015 heatwave (4%; Matsuda et al., 2020 ). The lower bleaching severity and rates of mortality in bleaching-susceptible P. compressa following the 2019 heatwave suggest these corals may have undergone beneficial acclimatization or experienced less heat stress in the subsequent 2019 event relative to the 2015 event. In contrast, the response to repeated heatwaves differed for M. capitata. In the two years following the 2015 heatwave, these same individuals of M. capitata exhibited 14% and 10% partial mortality among bleaching-susceptible and bleaching-resistant individuals, respectively ( Matsuda et al., 2020 ). Following the 2019 heatwave, M. capitata exhibited greater partial mortality among both phenotypes after the 2019 event (37% for bleaching-susceptible and 36% for bleaching-resistant). This increase in mortality following the second event for M. capitata indicates that these individuals may be accumulating the negative effects of heat stress ( i.e., incomplete recovery), and thus performing worse upon repeat exposure to heat stress. Across both species, decreased tissue integrity was correlated with increasing bleaching severity, supporting the hypothesis that heat stress can be seen at the tissue-level in addition to visual signs of bleaching. Interestingly, partial mortality was not significantly correlated with increasing colony bleaching severity for either species. Instead, partial mortality correlated with decreased tissue integrity in M. capitata, although they did not correlate in P. compressa . This may indicate that for some species, such as M. capitata , tissue integrity may be a better predictor of partial mortality than colony bleaching severity. These results support the importance of combining multiple physiological metrics across biological scales to better predict coral mortality rates following marine heatwaves. The 2019 marine heatwave peaked in October 2019 with 5.1 °C-weeks −1 at Patch Reef 13 ( Brown et al., 2023 ), leading to significant loss in tissue integrity, moderate bleaching responses, and partial mortality among all individuals of bleaching-susceptible M. capitata . In comparison to other marine heatwaves observed in Kāne’ohe Bay, such as the bleaching event in 2015 in which some areas of the bay recorded 14 °C-weeks −1 ( Brown et al., 2023 ) and more severe bleaching responses (>40% of all colonies; Bahr, Rodgers & Jokiel, 2017 ), the 2019 marine heatwave was relatively moderate. This is consistent with our findings, where no measurable changes in coral cover were observed, and only significant bleaching in sensitive individuals of the less thermally-tolerant species, M. capitata . However, partial mortality among the same colonies were observed after the 2015 bleaching event ( Matsuda et al., 2020 ), indicating that the cumulative effects of multiple bleaching events may result in significant mortality over longer periods of time. In its most extreme case, one bleaching-susceptible and one bleaching-resistant M. capitata colony showed complete mortality (100%) from November 2015–September 2023 ( Brown et al., 2023 ). These patterns were also qualitatively observed, where entire sections of reef consisting of M. capitata appeared dead, indicating that significant species-specific mortality may be occurring following bleaching events, even though declines in live coral cover were not captured in our surveys. However, another study using more sensitive sampling methods detected declines in live coral cover of 19% for P. compressa and 23% for M. capitata in Kāne’ohe Bay after the 2019 heatwave ( Yadav et al., 2023 ), corroborating our observations that M. capitata mortality following heatwaves can lead to significant loss of coral cover. These observations call for further analysis of potential ecosystem-wide changes due to a loss in thermally-sensitive coral species such as M. capitata , and accompanying changes in benthic community composition. Conclusions and future directions Histological analysis of coral tissues revealed signs of stress in the absence of visual symptoms of bleaching and onset of tissue stress prior to the accumulation of heat stress on the reef (as measured by degree heating weeks) in two distantly related reef-building coral species. These results indicate that histology is a valuable method for detecting coral stress before or in the absence of a visible stress response, and could be a useful tool for predicting coral health and mortality following heat stress. Marine heatwaves pose a major threat to the conservation of coral reefs, yet there is hope that the acclimatization of corals, in conjunction with proper management strategies, continued research across biological scales, and global policy to limit greenhouse gas emissions, can ensure a future for these unique ecosystems."
} | 5,866 |
39915603 | PMC11803100 | pmc | 808 | {
"abstract": "Mining activities produce a significant amount of gold mine tailings (GMT) rich in rare earth elements. Despite this, little information is available regarding the effective extraction of these precious elements from GMT. We describe a novel strategy for removing Pr, Ce, and Eu from GMT using Acidithiobacillus thiooxidans combined with a pretreatment step. For the pretreatment, GMT was exposed to 2 M oxalic acid for 6 h at 90 °C and 500 rpm at a liquid-to-solid ratio of 60 mL/g, which selectively removed iron and enhanced rare earth element bioavailability. As a result, GMT powder was introduced during the logarithmic growth phase of A. thiooxidans to maximize the bacterial activity and acid production during bioleaching. Surface modifications revealed by structural analysis validated the progress of the bioleaching procedure. Based on kinetic modeling, it was determined that chemical reactions determined the rate-limiting stage. The results showed a notable improvement in the recovery of rare earth elements after iron removal, with recovery of Pr, Ce, and Eu increasing by 24.4%, 14.4%, and 9.1%, respectively. A promising approach to resource recovery in mining operations could be achieved by integrating pretreatment and bioleaching for enhancing the recovery of rare earth elements from GMT.",
"conclusion": "Conclusion A. thiooxidans acidic bacterium was used in conjunction with a robust pretreatment protocol to extract REEs from GMT powder. Pretreatment involved washing for 6 h at 90 °C while maintaining a 60 mL/g liquid–solid ratio with a 2 M oxalic acid solution. As A. thiooxidans exhibited strong logarithmic phase initiation by day 12, this time point was selected for initiation of the two-step bioleaching process with GMT powder. Pr, Ce, and Eu recovery efficiency improved significantly following oxalic acid pretreatment, which reduced iron interference effectively. In order to achieve high recovery rates, bioleaching was combined with chemical processes. Despite the kinetic analysis showing that chemical reactions control the rate, bioleaching enabled these reactions to take place under acidic conditions. As a result of the metabolic activity of A. thiooxidans , biogenic sulfuric acid was generated, which facilitated the dissolution of metals from GMT powder. Due to GMT’s buffering properties, chemical reactions alone would not have produced similar results. Bioleaching plays an important role in metal recovery workflows because of the synergy between biological and chemical processes. The combined approach has also been confirmed to be effective by structural analysis. GMT powder’s surface and compositional changes were detected by XRD, SEM–EDX, and FTIR, indicating its enhanced dissolution after pretreatment and bioleaching. Further elucidation of bioleaching mechanisms and optimization of recovery efficiency will be achieved by incorporating advanced mineralogical and spectroscopic techniques. For recovering rare earth elements from secondary resources like GMT, bioleaching in conjunction with pretreatment strategies has been shown to be a sustainable and effective approach.",
"introduction": "Introduction Crushed rocks and processing fluids from mills, washings, or concentrators make up mine tailings 1 . Most of the ore is processed into tailings in many operations. For instance, 90–98% of the ore in porphyry copper ores is disposed of as tailings. Tailings to concentrate ratios are often very high, averaging 200:1 1 . Due to the dire economic conditions, most of the mine tailings are disposed of in the surrounding environment without any treatment 2 . The accumulation of mine tailings on land poses a serious environmental challenge. The mobilization and migration of toxic metals and metalloids from mine tailings to aquatic systems through natural pollution processes jeopardize ecological systems and human health 3 . In addition, a wide range of minerals contain rare earth elements. But their ongoing disappearance from original sources is now a serious worry 4 . As a result, the recovery of these elements from secondary sources such as mine tailings has recently attracted global attention 5 . When the benefits to the environment are taken into account, processing wastes can be justifiable even though they may contain fewer rare elements than primary sources 6 . The rare earth elements (REEs) are 17 in the periodic table, including 15 lanthanide elements and two elements, yttrium (Y) and scandium (Sc) 7 . Due to their intense inclination to oxygen, these elements are mainly found in craniates, oxides, phosphates, and silicates in the form of three-capacity cations in the earth’s shell 8 , 9 . These elements are increasingly used in modern technology every day 10 . For example, they are used in permanent magnets, electronic energy storage systems, and superconducting metal alloys. It is predicted that by 2030, demand for rare earth elements will grow by 41%. Due to the increasing demand for rare earth elements in recent years, the prospect of traditional extraction of these elements has been re-evaluated 11 . Mining extraction has advanced using two traditional methods: pyrometallurgy and hydrometallurgy. For low-grade ore and mining tailings, these processes require high costs and lead to severe air, water, and soil pollution, prompting a reconsideration of traditional methods 12 . Compared to conventional approaches, bioleaching is thought to be a more environmentally benign method of recovering metals by utilizing the activity of microorganisms 13 . Bioleaching can be divided into three methods: single-step, two-step, and spent culture medium. The culture medium is supplemented with waste powder and bacterial inoculum in the one-step procedure 14 . When using the two-step approach, the powder is added during the logarithmic growth phase, which is when the bacteria attain their maximum growth, and in the spent medium, the powder is added to bacterial metabolites 15 . Various bacterial strains have the potential to be employed in performing bioleaching processes. Microorganisms used in bioleaching share common characteristics that make them suitable for mineral dissolution 16 . These microorganisms act as catalysts to enhance and expedite the dissolution of specific metals from rocks and mining waste. One of the autotrophic acidophilic bacteria employed in bioleaching is Acidithiobacillus thiooxidans (A. thiooxidans) , which consumes sulfur and produces sulfuric acid. The produced acid aids in the extraction of rare earth elements 17 . Microorganisms capable of producing sulfuric acid are commonly used for the biological leaching of minerals 18 . Among them A. thiooxidans is one of the most common and best bacteria studied for bioleaching 3 . The research on bioleaching from GMT has focused on extracting heavy metals and environmental aspects. At the same time, very few studies have been on extracting rare earth elements. The extraction of metals such as cerium (Ce), praseodymium (Pr), and europium (Eu) from GMT and the removal of iron (Fe) as an interfering factor from the tailings are prominent features of these studies, which highlight the novelty of the research. In this study, an attempt was made to remove Fe from GMT powder using oxalic acid pretreatment. Subsequently, a two-step bioleaching process was employed to recover Ce, Pr, and Eu. Additionally, the analyses of X-Ray diffraction (XRD), Transform Infrared Spectroscopy (FTIR), Field Emission Scanning Electron Microscope (FE-SEM), and energy dispersive X-ray (EDX) were conducted on the untreated powder, pretreated powder, and powder After bioleaching to assess the advancement of the bioleaching. The next step was comparing the metal recovery from bioleaching from untreated powder to that from bioleaching from powder that had been pretreated.",
"discussion": "Results and discussion Pretreatment with oxalic acid Although pretreatment with oxalic acid can improve iron removal, iron dissolution in this method was at most 20%. This was due to the presence of iron in the structure of silicate minerals detected by XRD analysis. In these structures, due to the presence of complex mineral and organic complexes, metals are stable and their release is challenging, requiring very acidic conditions and extended periods for solubilization 25 . However, when silica is present in an acidic solution, the bulk becomes gelatinized 26 . Another reason for the low dissolution of iron may be due to the precipitation of oxalates around mineral layers, reducing the leaching rate 23 . The mineral phase is also crucial in determining the reaction rate. Quartz, for instance, dissolves slowly, whereas iron hydroxides and hydroxy oxides such as goethite and lepidocrocite dissolve readily 27 . The dissolution process of iron depends on the surface interactions that take place on the mineral surfaces. Through this process, the dissociation of oxalic acid according to reaction ( 2 ) and ( 3 ) results in the dissolution of Fe site from the mineral surface 28 : 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}$${\\text{H}}_{2} {\\text{C}}_{{2}} {\\text{O}}_{{4}} \\leftrightarrow {\\text{H}}^{ + } + {\\text{HC}}_{{2}} {\\text{O}}_{{4}}^{ - }$$\\end{document} 3 \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${\\text{HC}}_{{2}} {\\text{O}}_{{4}}^{ - } \\leftrightarrow {\\text{H}}^{ + } + {\\text{C}}_{{2}} {\\text{O}}_{{4}}^{{2 - }}$$\\end{document} Dissolution of iron from the surface of minerals can done through reactions ( 4 ) - ( 6 ): 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}$${\\text{Fe}}^{{3 + }} + {\\text{C}}_{{2}} {\\text{O}}_{{4}}^{{{2} - }} \\leftrightarrow {\\text{Fe}}\\left( {{\\text{C}}_{{2}} {\\text{O}}_{{4}} } \\right)^{ + }$$\\end{document} 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}$${\\text{Fe}}\\left( {{\\text{C}}_{{2}} {\\text{O}}_{{4}} } \\right)^{ + } + {\\text{C}}_{{2}} {\\text{O}}_{{4}}^{{{2} - }} \\leftrightarrow {\\text{Fe}}\\left( {{\\text{C}}_{{2}} {\\text{O}}_{{4}} } \\right)_{2}^{ - }$$\\end{document} 6 \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${\\text{Fe}}\\left( {{\\text{C}}_{{2}} {\\text{O}}_{{4}} } \\right)_{2}^{ - } + {\\text{C}}_{{2}} {\\text{O}}_{{4}}^{{{2} - }} \\leftrightarrow {\\text{Fe}}\\left( {{\\text{C}}_{{2}} {\\text{O}}_{{4}} } \\right)_{{3}}^{{{3} - }}$$\\end{document} In the dissolution process of iron, the formation of Fe \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${({\\text{C}}_{2}{{\\text{O}}}_{4})}_{3}^{{3}-}$$\\end{document} ot only promotes the ionization of oxalic acid to the right but also lowers 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}$${\\text{Fe}}^{3+}$$\\end{document} concentration to enhance the iron removal action. Moreover, in the oxalic acid leachate concentrations of rare earth elements (Pr, Ce and Eu) were below the detection limit (< 0.01 mg/L), which shows that there is a large content of rare earth elements in the oxalic acid leaching residue. The growth of A. thiooxidans Figure 1 illustrates how the A. thiooxidans strain grew in terms of cell count, pH, Eh, and sulfate content. After incubation for 12 days, the bacterial population increased sharply, and the maximum cell count was reached. The growth parameters of A. thiooxidans suggest that the logarithmic phase of growth commenced following a 12-day incubation period. Fig. 1 Changes during the two-step bioleaching process by A. thiooxidans in a cell count and sulfate concentration and b pH and Eh. Furthermore, after 10 days of bacterial incubation, the pH (a measurement of H + in the medium) dropped from 3.5 to 0.7 as a result of bacterial activity and acid generation. According to Eq. ( 7 ), A. thiooxidans obtains its energy during growth by converting elemental sulfur to sulfate ions 29 : 7 \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${\\text{S}}^{{0}} \\left( {\\text{s}} \\right) + {1}.{\\text{5O}}_{{2}} \\left( {{\\text{aq}}} \\right) + {\\text{H}}_{{2}} {\\text{O }}\\left( {{\\text{aq}}} \\right) \\to {\\text{2H}}^{ + } \\left( {{\\text{aq}}} \\right) + {\\text{SO}}_{{4}}^{{{2} - }} \\left( {{\\text{aq}}} \\right)$$\\end{document} Two-step bioleaching of GMT The entry of A. thiooxidans into the logarithmic phase of growth led to the selection of day 12 for the addition of GMT powder to the medium in the two-stage bioleaching procedure, based on the results of section \" The growth of A. thiooxidans \". The growth properties of A. thiooxidans were measured in the presence of GMT powder, including changes in pH, Eh, sulfate content, and cell count. Figure 1 a shows the sulfate concentration and cell count changes. The initial cell counts at the start of bioleaching was \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${4} \\times {10}^{{6}}$$\\end{document} cell/mL. Cell counts drastically dropped following the addition of GMT powder, possibly as a result of bacteria adhering to the powder’s surface 30 . Over the next few days, the number of cells remained almost unchanged before starting to decline until the bacteria eventually died. At the moment of adding the GMT powder, the concentration sulfate suddenly decreased from 7000 to 6200 and the pH increased from 0.6 to 0.8. The reaction between metal compounds and protons produced by the dissociation of sulfuric acid is mostly responsible for the waste’s acid consumption since it allows the metals to become soluble in the solution (Eq. ( 8 )) 29 : 8 \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${\\text{H}}_{2} {\\text{SO}}_{4} + {\\text{ Materials}} - {\\text{Me}} \\to {\\text{ materials}} - {\\text{2H }} + {\\text{ MeSO}}_{4}$$\\end{document} where Me is a bivalent metal. Then, the pH decreases sightly, which can be a reason for the growth of bacteria and the production of sulfuric acid. Stabilization of sulfate concentration and pH shows the bacteria are not producing acid 31 . Figure 1 b show the pH and Eh change during the bioleaching process. When the powder was added to the medium, the Eh increased significantly from 440 to 493 and was not affected by the pH. This increase may be due to the release 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}$${\\text{Fe}}^{+{3}}$$\\end{document} in the medium. After a day, Eh decreases, which shows that 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}$${\\text{Fe}}^{+{3}}$$\\end{document} 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}$${\\text{Fe}}^{+{2}}$$\\end{document} has decreased (Eq. ( 9 )) 32 : 9 \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$6Fe^{3+}+S+4H_{2}O\\rightarrow6Fe^{2+}+HSO_4^-+7H^{+}$$\\end{document} Extraction of metals The two-step bioleaching of REEs and Fe from the pretreated GMT powder was conducted and the results obtained on the 3, 5 and 7 days were analyzed. Also, in order to investigate the effect of pretreatment on the extraction of rare earth elements, bioleaching was performed from untreated powder after 7 days. Figure 2 shows the bioleaching of REEs from pretreated powder on days 3, 5, 7 and the bioleaching of elements from untreated powder on day 7. The iron content in the pretreated powder decreased to 1% after 7 days of leaching. It is evident that the bioleaching of Pr, Ce and Eu increased on day 7 compared to days 5 and 3. The leaching Pr was 57% on the 3 days, which increased to 76% on the 7 day. Similarly, the leaching of Ce increased from 31% on the 3 day to 46% on the 7 day. However, the leaching of Eu was 5% on the 3 days, which increased to 14% on the 7 days. Also, by comparing the bioleaching of elements from pretreated and untreated powder, it is found that the recovery of Pr, Ce and Eu increased by 24, 14 and 9% respectively after 7 days. The results show that removing iron increases the recovery of RREs. Fig. 2 Bioleaching of REEs from the pretreated and untreated powders. Control experiments were performed under the same conditions as the bioleaching tests but without bacterial inoculum. The results showed that the recovery of Pr, Ce, and Eu was near zero in the absence of bacteria, highlighting the essential role of A. thiooxidans in metal solubilization. Due to the negligible recovery in the control tests, these data were not included in Fig. 2 to maintain clarity and focus on the bioleaching process’s effectiveness. However, they further emphasize the critical contribution of bacterial activity to the overall recovery process. A. thiooxidans has the ability to dissolve metals in the media and change elemental sulfur into sulfuric acid. Acid consumption by REEs shown in Eqs. ( 10 – 12 ) 33 : 10 \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${\\text{H}}_{{2}} {\\text{SO}}_{{4}} { } \\leftrightarrow {\\text{HSO}}_{{4}}^{ - } + {\\text{ H}}^{ + }$$\\end{document} 11 \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${\\text{REE}}^{{3 + }} + {\\text{HSO}}_{4}^{ - } \\leftrightarrow {\\text{(REEHSO}}_{{4}} {)}^{{{2} + }}$$\\end{document} 12 \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${\\text{REE}}^{{3 + }} + 2{\\text{HSO}}_{4}^{ - } \\leftrightarrow {\\text{REE(HSO}}_{{4}} {)}_{{2}}^{{^{ + } }}$$\\end{document} Most of the sulfuric acid produced is consume by metals such as iron with a concentration of 322,743 (mg/L). By removing iron and other interfering metals with a high concentration, the acid produced can form a complex with REEs and the dissolution of metals increases. Important parameters such as specific surface area, particle size, crystal network and chemical composition have an effect on the dissolution of metals. In addition to these parameters, the dissolution of REEs strongly depends on their bonding in the powder 34 . As a result of chemical bond disruption during the oxalic acid pretreatment, rare REEs previously trapped in the mineral matrix were exposed, increasing their availability for leaching and leading to higher concentrations of REEs in the solution. Structural analysis of the GMT powder Various structural analyses, including XRD, FTIR, SEM, and EDX, were conducted on GMT powder before and after pretreatment and bioleaching in order to track the development of the bioleaching procedure and the impact of pretreatment. XRD analysis After the pretreatment, the clinochlore phase and gypsum were removed. GMT powder XRD analysis is shown in Fig. 3 , both before and after pretreatment and after bioleaching. Based on the XRD analysis of the original GMT particles (before bioleaching), quartz, clinochlore, muscovite, albite, and gypsum dominate the mineral phase composition. Peak intensities associated with clinochlore and gypsum were reduced after pretreatment, indicating partial removal. However, the data presented cannot conclusively prove that all contaminants have been removed. Based on these changes, it is likely that oxalic acid pretreatment altered the mineralogical structure of GMT powder, which facilitated bioleaching. Muscovite and albite peaks also became shorter after bioleaching, further demonstrating bioleaching’s impact on these phases. Fig. 3 XRD analysis of GMT powder before pretreatment, after pretreatment and bioleaching. FTIR analysis Figure 4 showed the results of FTIR analysis before pretreatment after pretreatment and bioleaching. The broad peaks in 3539, 3635, 3095 and 3133 cm −1 are related to O–H and N–H bonds. These vibrations are indicative of the presence of hydroxyl and amino groups in the sample structures 36 , 37 . The peak at 1408, 1510, 1458 and 1384 cm −1 corresponds to the flexural vibrations of C–H bonds and peaks located at 2952 and 2873 cm −1 correspond to the asymmetric and symmetric stretching vibrations of C–H bonds present in CH 3 and CH 2 structures 38 , 39 . Obvious Peaks observed at 1736 and 1701 cm −1 are related to stretching vibrations of C=O bonds in carboxylic groups and peaks at 1607 and 1615 cm −1 are related to bending vibrations of O–H bonds or stretching vibrations of C=C bonds 40 , 41 . In addition, the C =O bond may be associated with organic matter produced by bacterial metabolism 42 . Peaks at 1260 and 1369 cm −1 are attributed to the stretching vibrations of C–N bonds and the bending vibrations of C–H bonds present in various compound structures 43 , 44 . Peaks observed at 1237 and 1182 cm −1 are related to the stretching vibrations of C–N and S–O bonds, respectively, in amino and sulfate structures. An obvious peak appears at 1042 cm −1 for the experimental sample, which is ascribed to the Si–O asymmetric stretching vibration 39 , 45 , 46 . Multiple peaks leading to the formation of a broad and intense peak in the range of 900–1400 cm −1 are due to various stretching vibrations of Si–O bonds in quartz and silicate structures such as albite and muscovite 24 , 38 , 45 . The peaks related to the stretching vibrations of Al–O–Si in albite and muscovite structures are located at 778, 797, 724 and 694 cm −1 38 , 45 . When various spectrums were compared, it was possible to see that some peaks became less intense in bioleached residue, indicating that metals were being leached off the powder surface and transferred into the solution 42 . The peaks related to the flexural and stretching vibrations Al–O–Si/Si–O–Si are located at 30, 531 ,and 533 cm −1 47 , 48 . Also, peaks related to the stretching vibrations of Fe–O bonds in specific structures, are located at 561, 467 and 470 cm −1 . One of the reasons for the narrowing of peaks after bioleaching is the reduction of particle size 49 . Therefore, the FTIR results confirm the existence of quartz and other silicate phases. In addition, it shows the existence of amino compounds, methyl aliphatic and methylene in the sample. Fig. 4 FTIR analysis of GMT powder before pretreatment, after pretreatment and bioleaching. SEM–EDX analysis Figure 5 shows the FE-SEM photo micrographs of before powder, after pretreatment and bioleaching at × 5k, × 10k and × 25k magnification. As is seen in Fig. 5 a, the original GMT particles before pretreatment had a non-porous surface and were smooth, while after pretreatment and bioleaching disintegrated into smaller ones. The finer structure of GMT is thus the result of the progressive breakdown of GMT following bacterial leaching. The surface morphology of GMT particles shows a considerable variation before and after the bioleaching process at a greater magnification, as depicted in Fig. 5 b and c. In addition, the surface coatings of the solid sample become less after leaching by oxalic acid, the maximum dissolution efficiency of iron is only %20. This indicates that the residual iron impurities are trapped in the silica lattice and are very challenging to extract 23 . The EDX results shows after pretreatment, the iron decreased from 15.65% to 5.12% (v/v). After bioleaching, the iron content decreased from 5.12 to 1.71%. The results from EDX confirm the removal of iron from the powder. Figure 6 indicates the EDX and mapping of original powder, after pretreatment and bioleaching. Fig. 5 SEM analysis of GMT powder at different magnifications: \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$\\times$$\\end{document} 5.00k, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$\\times$$\\end{document} 10.0k 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}$$\\times$$\\end{document} 25.0k and morphological characterization: a before pretreatment, b after pretreatment and c after bioleaching. Fig. 6 EDX analysis of GMT powder a before pretreatment, b after pretreatment and c after bioleaching. Evaluation of the rate-controlling step Research findings indicate that understanding the kinetics of bioleaching is crucial for enhancing metal extraction efficiency 50 . Metals were bioleached from GMT by means of heterogeneous solid–fluid processes, which happened in at least three stages: (a) Reactant agents diffusing through the shell, (b) reactants diffusing from the liquid boundary layer encircling the solid, and (c) chemical reaction 51 . In this study, GMT was prepared using a single batch to maintain consistency and minimize variability in the sample’s mineralogical and chemical properties. Variations in ore composition from batch to batch can, however, affect pretreatment and bioleaching efficiency. Mineral content, particle size distribution, or geochemical properties of GMT may account for such variations.To address this limitation, future experiments will evaluate the robustness and reproducibility of the proposed methodology using multiple batches of GMT. By applying the process to a broader range of mining tailings, we will gain a deeper understanding of its applicability. Because bioleaching occurs under vigorous shaking conditions, it can be assumed that the first stage is not the rate-limiting step. If the heterogeneous reaction is controlled by the diffusion, shrinking core model theory as the following Eq. 52 : 13 \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${\\text{kt = 1}} - {2}/{\\text{3X}} - {\\text{(1 {-} X)}}^{{2/3}}$$\\end{document} where t is the bioleaching time, X is the conversion of leached metals from the GMT powder, and is the parabolic rate constant. The following formula can be used to describe the kinetics of bioleaching when the reaction rate is regulated by a heterogeneous chemical reaction on the particle surface: 14 \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${\\text{kt = 1}} - {\\text{(1 {-} X)}}^{{1/3}}$$\\end{document} It is assumed that the concentration of the leaching agent, sulfate, remains constant in kinetic investigations of the bioleaching process. In reality, Due to the activity of bacteria and the conversion of sulfur element into sulfuric acid, the concentration of sulfate ions varies during the process 51 . Therefore, assuming a constant leaching agent concentration may lead to incorrect results. To account for these variations, the following modified equations should be used for leaching control Eq. ( 15 ) and chemical reaction Eq. ( 16 ) 53 , 54 : 15 \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${\\text{P }}\\left( {\\text{X}} \\right){ = 1}{-}{2}/{\\text{3X }}\\left( {\\text{t}} \\right) - {\\text{(1 {-} X (t))}}^{{2/3}} = \\frac{{{\\text{2bD}}_{{\\text{e}}} }}{{\\rho_{{\\text{x}}} {\\text{R}}^{{2}} { (1 } - { }\\varepsilon {)}}}\\mathop \\smallint \\limits_{{0}}^{{\\text{t}}} {\\text{C}}_{{{\\text{sulfate}}}} {\\text{dt = }}\\lambda {\\text{F(t)}}$$\\end{document} 16 \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$G(X) = 1 - {\\text{(1 {-} X (t))}}^{{1/3}} = \\frac{{{\\text{bk}}^{*} }}{{\\rho_{{\\text{x}}} {\\text{R (1 }} - \\varepsilon {)}}}\\mathop \\smallint \\limits_{{0}}^{{\\text{t}}} {\\text{C}}_{{{\\text{sulfate}}}} {\\text{dt = \\pounds F(t)}}$$\\end{document} where 17 \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$\\frac{{{\\text{2bD}}_{{\\text{e}}} }}{{\\rho_{{\\text{x}}} {\\text{R}}^{{2}} { (1 } - \\varepsilon {)}}} = \\lambda$$\\end{document} 18 \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$\\frac{{{\\text{bk}}^{*} }}{{\\rho_{{\\text{x}}} {\\text{R (1}} - \\varepsilon { )}}} = \\pounds$$\\end{document} where \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$\\rho_{{\\text{x}}}$$\\end{document} is the molar density of metals in the GMT powder, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${\\text{D}}_{\\text{e}}$$\\end{document} is the diffusion coefficient of the sulfate, R is the particle radius, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$\\varepsilon$$\\end{document} is the porosity of the GMT particle, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${\\text{C}}_{\\text{sulfate}}$$\\end{document} is the sulfate concentration in the bioleaching solution, λ is the diffusion rate constant 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}$$\\pounds$$\\end{document} is the chemical reaction rate constant. Using the data of sulfate concentration changes, metals recovery and modified equations of this section, conventional and modified kinetic models for Pr, Ce and Eu metals were plotted, which are shown in Fig. 7 . In order to better compare the results, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${\\text{R}}^{2}$$\\end{document} values for each bioleaching kinetic model without sulfate change and considering sulfate changes are given in Table 1 . Fig. 7 Comparison between measurements and correlations expressing controlled diffusion and chemical reaction mechanisms for bioleaching kinetic models: a conventional Pr model, b modified Pr model, c conventional Ce model, d modified Ce model, e model conventional Eu, and f modified Eu model. Table 1 Comparison of correlation coefficient values for Pr, Ce and Eu bioleaching kinetic models. Metal \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$$R^{2}$$\\end{document} Time-independent sulfate variations Time-dependent sulfate variations Diffusion P(X) Chemical reaction G(X) Diffusion P(X) Chemical reaction G(X) Pr 0.9354 0.9747 0.6392 0.85 Ce 0.901 0.9922 0.2013 0.4012 Eu 0.9168 0.9611 0.1894 0.2617 According to the results, when sulfate concentration changes are neglected, the chemical reaction model (higher \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${\\text{R}}^{2}$$\\end{document} ) fits better with the results ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${\\text{R}}_{\\text{(Pr)}}^{2}$$\\end{document} = 0.9747, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${\\text{R}}_{\\text{(Ce)}}^{2}$$\\end{document} =0.9922, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${\\text{ R}}_{{\\text{(Eu)}}}^{{2}}$$\\end{document} = 0.9168) than the diffusion model. Also, when changes in sulfate concentration are considered, the chemical reaction model showed a better prediction of the results ( \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${\\text{R}}_{{\\text{(Pr)}}}^{{2}}$$\\end{document} = 0.85, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${\\text{ R}}_{{\\text{(Ce)}}}^{{2}}$$\\end{document} = 0.4012, \\documentclass[12pt]{minimal}\n\t\t\t\t\\usepackage{amsmath}\n\t\t\t\t\\usepackage{wasysym} \n\t\t\t\t\\usepackage{amsfonts} \n\t\t\t\t\\usepackage{amssymb} \n\t\t\t\t\\usepackage{amsbsy}\n\t\t\t\t\\usepackage{mathrsfs}\n\t\t\t\t\\usepackage{upgreek}\n\t\t\t\t\\setlength{\\oddsidemargin}{-69pt}\n\t\t\t\t\\begin{document}$${\\text{R}}_{{\\text{(Eu)}}}^{{2}}$$\\end{document} = 0.2617) that the diffusion model. The results suggested that the leaching Pr, Ce and Eu followed the chemical reaction model. also, it can be concluded that the application of sulfate changes in the conventional equations leads to a better prediction of the results."
} | 9,708 |
38817912 | PMC11137280 | pmc | 809 | {
"abstract": "In this study, we explore spintronic synapses composed of several Magnetic Tunnel Junctions (MTJs), leveraging their attractive characteristics such as endurance, nonvolatility, stochasticity, and energy efficiency for hardware implementation of unsupervised neuromorphic systems. Spiking Neural Networks (SNNs) running on dedicated hardware are suitable for edge computing and IoT devices where continuous online learning and energy efficiency are important characteristics. We focus in this work on synaptic plasticity by conducting comprehensive electrical simulations to optimize the MTJ-based synapse design and find the accurate neuronal pulses that are responsible for the Spike Timing Dependent Plasticity (STDP) behavior. Most proposals in the literature are based on hardware-independent algorithms that require the network to store the spiking history to be able to update the weights accordingly. In this work, we developed a new learning rule, the Bi-Sigmoid STDP (B2STDP), which originates from the physical properties of MTJs. This rule enables immediate synaptic plasticity based on neuronal activity, leveraging in-memory computing. Finally, the integration of this learning approach within an SNN framework leads to a 91.71% accuracy in unsupervised image classification, demonstrating the potential of MTJ-based synapses for effective online learning in hardware-implemented SNNs.",
"conclusion": "5 Conclusion In this study, we explored the potential of MTJs to create efficient spintronic synapses for SNNs, utilizing multiple MTJs in parallel to form a proposed synapse. By operating the synapse at low voltages and exploiting the stochastic nature of MTJs, we enable the synapse to achieve multiple conductance levels. These levels are attained through training, guided by the novel Bi-sigmoid STDP learning rule. This rule, facilitated by engineered linearly decaying presynaptic and bi-rectangular postsynaptic pulses, translates the delay between the two pulses into voltage modulation, effectively updating the synapse state. The resultant rule operating in the proposed synapse will enable energy-efficient neuromorphic systems capable of supporting unsupervised and online learning, eliminating the need for labor-intensive data labeling and enabling continuous learning after deployment. Through detailed electrical SPICE simulations, we optimized the MTJ-based synapse design, demonstrating how different pulse widths and synapse configurations influence the Bi-sigmoid rule. Finally, the integration of this rule in an SNN led to a notable 91.71% accuracy in unsupervised image classification. While this work has centered on synaptic mechanisms, addressing future challenges such as leakage currents, reliability concerns including MTJ variability and defects, and further tuning of neuron and network parameters, alongside exploring datasets suited for spiking data, presents promising pathways for enhancing system performance and advancing neuromorphic computing development.",
"introduction": "1 Introduction The current landscape of computing, dominated by traditional Von Neumann (VN) architectures, faces significant challenges when it deals with Artificial Intelligence (AI) applications (Ye et al., 2021 ; Momose et al., 2020 ). VN architectures which are based on the separation between processing and memory, suffer from substantial energy consumption and computational latency due to the data transfer overhead between the memory and the processor unit (Ma et al., 2020 ; Petrenko and Petrenko, 2018 ). On top of that, VN architectures are not the best candidates for IoT and edge-computing intelligent devices because they don't allow online and unsupervised learning Syed et al. ( 2024 ). These two characteristics though are important for systems that are intended to learn continuously and adapt themselves in real-time, like autonomous vehicles. In contrast to this architecture, Neuromorphic Engineering, a concept introduced by Carver Mead (Mead, 2020 ) in the early nineties, has emerged as a promising alternative. This approach, inspired by the biological brain's structure and function, offers a distributed processing model. The brain's computational model operates through a vast network of neurons interconnected by synapses, each capable of processing and storing information. This decentralized approach allows for efficient parallel processing. Neurons communicate via electrical impulses or 'spikes'. The strength of connections, or the synaptic weight, changes in response to the patterns and timings of neuronal activity. This dynamic adaptability, known as synaptic plasticity, is fundamental to learning and memory in the brain. Unlike VN architectures, the huge number of neurons and synapses where processing and memory are colocalized leads to highly efficient computation with minimal energy consumption. Translating the brain's computational principles into artificial systems has led to the development of various systems that all aspire to reproduce the synaptic plasticity of the brain. These systems range from nanowires-based networks (Caravelli et al., 2023 ; Loeffler et al., 2023 ; Milano et al., 2022 ) to Spiking Neural Networks (SNNs), which are considered the third generation of Artificial Neural Network (ANN) models (Maass, 1997 ; Ghosh-Dastidar and Adeli, 2009 ). The SNN encodes information in the timing of spikes, and utilizes a dedicated learning rule: Spike-Timing-Dependent Plasticity (STDP) (Caporale and Dan, 2008 ), that modulates synaptic strengths, either strengthening or weakening, based on the relative timing between spikes. While the SNN model with STDP promises a more energy-efficient solution for AI applications and enables online and unsupervised learning, its practical implementation extends beyond the model and the algorithm itself. The effectiveness and energy efficiency of SNNs largely depend on the appropriate hardware implementation. It is this hardware, especially when designed with in-memory computing, that unlocks the full potential of SNNs, ensuring energy efficiency and brain-like computation. The state-of-the-art SNN hardware implementations can be split into three categories: first, systems like CPUs, GPUs, and TPUs focused on computational complexity with high accuracy but high power use (Baji, 2017 ; Wang et al., 2020 , 2019 ); second, power-efficient CMOS-based engines such as TrueNorth (Merolla et al., 2014 ) and Loihi (Davies et al., 2018 ), which are constrained by memory bottlenecks; and finally, biologically plausible in-memory computing using non-volatile technology, yet lacking online learning capabilities (Maranhão and Guimarães, 2021 ; Lone et al., 2022 ). These approaches struggle to balance energy efficiency with the capability for online unsupervised learning. Our work aims to address this challenge with an innovative design of synapses using Magnetic Tunnel Junctions (MTJs) Ikeda et al. ( 2010 ). These non-volatile and energy-efficient devices have the potential for unsupervised learning through dynamic conductance adjustments, guided by an innovative learning rule. Although there are a few works that proposed the use of MTJs for plasticity dynamics in SNN (Shreya et al., 2020 ; Jang et al., 2021 ; Leonard et al., 2022 ), there is a need to demonstrate this in a full network trained with a device-specific learning rule. The key contribution of this paper focuses on the synaptic design and a compatible learning rule . We particularly explore the use of a compound synapse made of multiple parallel-connected MTJ devices, these two-state devices lead to a multi-state synapse thanks to their inherent stochasticity. We carry an extensive design space exploration of the MTJ-based synapse to find the optimal parameters that allow the unsupervised adjustment of the synaptic conductance through a learning rule that is rooted in the physics of the MTJ, we labeled it Bi-sigmoid Spike Timing Dependent Plasticity (B2STDP). This rule gave a good accuracy (>90%) in an image classification task. The remainder of this paper is structured as follows: Section 2 reviews current SNN hardware implementations and their training approaches. Section 3 details the MTJ-based synapse design, and various design choices that enable the Bi-sigmoid STDP learning rule. Section 4 presents our results and discussions on the implemented network and its performance. The paper concludes with a recap and perspectives in Section 5.",
"discussion": "4 Results and discussion 4.1 Derivation of Bi-Sigmoid STDP The comprehensive design space exploration shown in the previous section led to the development of a novel learning rule for the synapse. This rule has an STDP-like behavior, where a short delay of V post relative to V pre leads to potentiation, and a longer delay results in depression. By analyzing the conductance changes in response to the relative time between pre- and post-synaptic spikes, we identified a pattern best represented by two sigmoid functions. This behavior was captured in a rule that we labeled Bi-sigmoid STDP (B2STDP) learning rule, formulated from electrical simulations data. Figure 7A presents the results of our Spice simulations, which were conducted to determine the synaptic weight updates relative to the timing differences between pre- and post-synaptic spikes. In these simulations,we utilized a synapse model with 12 MTJs, with the parameters detailed in Table 1 , and initialized the synapse to an intermediate conductance state. By varying the arrival time of V post with respect to V pre , we record the resultant changes in synaptic conductance. After each sweeping step of V post , the synapse was reset to its initial state. This process was repeated 10 times, and the conductance updates were then averaged and normalized. The fitting of this data to a Bi-sigmoid function, as depicted in the figure, accurately represents the updates in synaptic conductance. The fitting function is given by: \n Δ w ( Δ t ) = - A 1 + e - k 0 ( Δ t - t 0 ) - A 1 + e - k 1 ( Δ t - t 1 ) + A A \n Figure 7 (A) Normalized synaptic weight updates relative to the timing differences between pre- and post-synaptic spikes. Spice simulation data fitted with the Bi-sigmoid function. (B) Label distribution map for each excitatory neuron after training, indicating learned representations. (C) Incremental improvement in classification performance as a function of the number of training samples. (D) Visualization of synaptic weights between the input layer and 100 excitatory neurons, reshaped into 28x28 matrices, each representing a learned digit from the MNIST dataset. where Δ w is the normalized synaptic update and Δ t is the delay between pre- and post-synaptic spikings, and the following are fitting constants: \n A = 7 . 95 × 1 0 5 ; k 0 = 0 . 474723045 ; t 0 = 20 . 77893753 ; k 1 = 0 . 757072031 ; t 1 = 48 . 93860322 \n The B2STDP is distinguished from classical STDP by being intrinsically tied to the physical properties of MTJs, enabling on-the-fly synaptic updates based on spiking activity, thus promoting local learning with in-memory computing without the need to store the spiking history of neurons (traces) in an external memory. More details on the derivation of the Bi-sigmoid rule can be found in the Supplementary material . 4.2 Integration in a functional network To evaluate our B2STDP learning rule in an image classification task, we integrated it into Bindsnet Hazan et al. ( 2018 ); an SNN simulation framework ( Supplementary material provides detailed steps on integrating our learning rule in the SNN framework). We used the SNN architecture introduced in Diehl and Cook ( 2015 ). This network utilizes a two-layer structure: the first layer consists of 784 input neurons, corresponding to the 28 × 28 pixels of the MNIST images (LeCun et al., 1989 ), and the second layer comprises 100 excitatory and an equal number of inhibitory neurons. The network employs leaky integrate-and-fire neuron models and conductance-based synapses. Inputs are presented as Poisson spike trains, with firing rates proportional to pixel intensities, converting the intensity values of MNIST images into spikes. The synapses between input and excitatory neurons learn according to B2STDP, which allows not only unsupervised learning but also neuromorphic efficiency through in-memory computing thanks to MTJ-based synapses. 4.3 Network performance During training, each input image is converted into spikes proportional to the pixel intensities and shown to the network for 250 ms . For each output neuron, the network keeps track of how many spikes it produces in response to each class of input. After exposing the network to the complete training dataset, we calculate the average firing rate for each neuron for every class. The class that causes the highest average firing rate for a particular neuron becomes that neuron's assigned label. Figure 7B displays the assigned label for each excitatory neuron after training. This label distribution map showcases the learned representations across the excitatory layer. Notably, the assignment of labels is based on the predominance of neuron firing in response to specific input classes throughout the training phase. During evaluation, the accuracy is calculated by counting all the spikes from excitatory neurons which were all assigned a label during training, and seeing which class gets the most spikes for each input. Each neuron 'votes' for its assigned class every time it fires. The class with the most votes across all neurons is the predicted class for that input. The network's accuracy is measured by how often the predicted class matches the actual class of the inputs. Figure 7C , demonstrates the incremental improvement in classification performance as the network processes a greater number of training samples. Finally, Figure 7D presents a visualization of the synaptic weights between the input layer and the excitatory neurons in the network. Each weight vector, originally 784-dimensional corresponding to the flattened 28 × 28 pixel MNIST images, is reshaped back into a 28 × 28 matrix. The figure illustrates 100 such matrices, each corresponding to a different excitatory neuron in the network. These matrices serve as a snapshot of what each neuron has learned to recognize, with each matrix visually resembling a digit from the dataset. This underscores the network's ability to extract key features from the training data. The results presented in Table 2 demonstrate the effectiveness of our B2STDP learning rule within SNNs, showcasing a competitive performance compared to other notable works. In our experiments, we trained networks with different numbers of output neurons - specifically 100, 400, and 1,600–over 1, 3, and 3 epochs, respectively. The networks achieved testing accuracies of 85.15%, 90.28%, and 91.71% for each neuron count. Remarkably, with 100 output neurons, our network surpasses the classification accuracies of both Diehl and Cook ( 2015 ) and Tang and Gao ( 2023 ) implementations. Although increasing the number of output neurons does improve accuracy, the enhancement is not dramatic, likely due to the need for additional epochs to achieve convergence in accuracy. Nonetheless, our model still outperforms (Tang and Gao, 2023 ) and attains comparable results to Diehl and Cook ( 2015 ), while requiring fewer epochs. For instance, where Tang and Gao ( 2023 ) and Diehl and Cook ( 2015 ) employed 4 and 7 epochs to train the SNN with 1,600 neurons, our network needed only 3 epochs to reach similar levels of accuracy. Two adjustments may have contributed to the performance of our network: the duration for which an image is presented to the network, and the conversion coefficient for translating pixel intensity into firing rates. By fine-tuning these parameters to 250ms for image presentation and setting a maximum firing rate at 60 Hz, our network demonstrates enhanced efficiency and accuracy. While traditional ANNs employing backpropagation may achieve higher accuracies, our approach, centered on an unsupervised learning paradigm, offers significant advantages for neuromorphic hardware implementations, particularly in terms of energy efficiency, online and unsupervised learning which are important for IoT devices. Central to our approach is the adoption of a novel learning rule that is intrinsically tied to the physical properties of spintronic synapses. Table 2 Comparison of classification accuracy across different works and network sizes. \n Neurons \n \n 100 \n \n 400 \n \n 1,600 \n Diehl and Cook ( 2015 ) 82.9 % 87.0 % 91.9 % Tang and Gao ( 2023 ) 75 % 83 % 88 % \n This work \n 85.15 % 90.28 % 91.71 % Our study primarily focused on optimizing synapse behavior in accordance with MTJ dynamics, revealing several areas for potential future exploration. Enhancing accuracy is an ongoing objective, with improvements potentially arising from adjustments in network parameters such as neuron counts, spiking thresholds, and the balance between excitatory and inhibitory neurons. It is crucial to note that our approach with SNNs is not about outperforming DNNs in accuracy; rather, it emphasizes the unique advantages of SNNs, especially empowered by unsupervised and online learning, while being known for their energy efficiency, making them well-suited for IoT devices. Issues like dealing with leakage currents in the crossbar, significant in design and optimization phases, should be further investigated to optimize power efficiency. Reliability concerns related to MTJ variability and defects, such as pinhole defects and dielectric breakdown, are active research topics when MTJs are used as MRAMs. Their use as synapses warrants further examination to understand how these issues affect the learning rule and, subsequently, the overall performance of the SNN. Additionally, developing a comprehensive network that includes digital neuron designs integrated with crossbar arrays may facilitate an accurate evaluation of power consumption in SNNs employing the bisigmoid learning rule. The choice of dataset is another important aspect, where data originally obtained in a spiking manner, like images from event-based DVS cameras, are better suited for SNNs than images converted to spikes. These potential refinements represent promising avenues to elevate the capabilities of our neuromorphic computing model, potentially achieving higher accuracies and more efficient learning processes."
} | 4,628 |
31720211 | PMC6838517 | pmc | 810 | {
"abstract": "Recent advances in metabolic engineering enable the production of high-value chemicals via expressing complex biosynthetic pathways in a single microbial host. However, many engineered strains suffer from poor product yields due to redox imbalance and excess metabolic burden, and require compartmentalization of the pathway for optimal function. To address this problem, significant developments have been made towards co-cultivation of more than one engineered microbial strains to distribute metabolic burden between the co-cultivation partners and improve the product yield. In this emerging approach, metabolic pathway modules can be optimized separately in suitable hosts that will then be combined to enable optimal functionality of the complete pathway. This modular approach broadens the possibilities to fine tune sophisticated production platforms and thus achieve the biosynthesis of very complex compounds. Here, we review the different applications and the overall potential of natural and artificial co-cultivation systems in metabolic engineering in order to improve bioproduction/bioconversion. In addition to the several advantages over monocultures, major challenges and opportunities associated with co-cultivation are also discussed in this review.",
"conclusion": "6 Conclusion and future perspectives Microbial biosynthesis via co-cultivation engineering provides a paradigm shift in metabolic pathway balancing. It broadens the possibilities to tune complex metabolic pathways and can be customized for efficient production of a variety of bioproducts. Co-cultivation engineering has several advantages over mono-cultivation systems such as robustness, modularity, higher tolerance (toxic intermediate/waste produced from one partner get consumed/degraded by the other partner) and higher productivity. It utilizes the metabolic power and resources of each co-cultivation partner to meet the demand of specific co-factors and precursors and thus improves the conversion yield of the modularized biosynthetic pathway. This approach allows to produce more complex compounds with improved productivity by distributing the metabolic pathway between each consortium member. Co-cultivation fermentations may lead to enhanced production performance, and allow the utilization of cheaper substrates. Moreover, artificial consortia open the door to address the issues of functional expression of complex biosynthetic pathway enzymes without compromising the yield and product quality. They can also reduce the effort of reconstitution of recombinant biosynthetic pathways. As an emerging research area in the field of metabolic engineering, co-cultivation engineering is still in its infancy. Most of the recent reports on co-cultivation engineering that are based on employment of microbial consortia have only two constituent strains/species in order to achieve their engineering goals. Co-cultivation of multiple populations is more complicated, as co-cultivation behaviour of individual strains using common cultivation methods is still unknown and potentially more challenging to control when increasing the number of the constituent strains/species. However, recent development in co-cultivation engineering has greatly expanded our understanding of microbial behaviour in communities ( Wang et al., 2016 ; Zuñiga et al., 2017 ). While the potential of synthetic microbial consortia holds great promise, there are inherent challenges that need to be addressed with the help of synthetic biology approaches. It is anticipated that co-cultivations comprising multiple specialized members, or polycultures, will be developed and utilized for meeting the demand of more complicated biosynthetic pathways in the near future.",
"introduction": "1 Introduction Metabolic engineering of microorganisms enables production of chemicals via construction and optimization of different metabolic pathways. The functionality and overall conversion efficiency of the biosynthetic pathway depends on various factors including precursors, cofactor demand and optimal expression of the pathway enzymes. Problems arise however when, due to the complexity and length of the recombinant pathway, a single strain cannot cope with the pathway demand, a phenomenon commonly referred to as metabolic burden ( Wu et al., 2016 ). To overcome the limitations posed by metabolic burden, significant developments have been made towards rationally designed microbial co-cultures to distribute metabolic burden of complex and long biosynthetic pathways into different strains/species in order to improve bioproduction performance ( Jones et al., 2017 , 2016 ; Liu et al., 2018 ; Saini et al., 2015 ; Tsoi et al., 2018 ; Zhang and Wang, 2016 ) ( Fig. 1 ). This approach has been inspired by microbial natural consortia, which carry out complex chemical reactions to provide favourable environment for survival of the community. Fig. 1 Schematic representation of artificial consortium for bioproductions. (A) Co-cultivation comprising strains of the same species, (B) Co-cultivation comprising strains from different species and (C) Co-cultivation comprising mixed strains i.e. polyculture. Fig. 1 The modularity in co-cultivation allows rapid optimization of the strains carrying each pathway module and latter assembly of the engineered strains into a synthetic consortium that enables optimal conversion of a substrate or precursor metabolite to the desired final product. It provides a platform to optimize each segregated pathway under optimal cellular environments for functional expression of different pathway genes. It also provides balancing of the complex pathway by optimizing the ratio of the consortia members to improve overall yield ( Zhou et al., 2015 ). Compartmentalization between co-cultivation partners reduces the possibility of undesired cross-reactions between the pathway modules and thus enables efficient bioproduction ( Martínez et al., 2016 ; Shong et al., 2012 ). Co-cultivation methodologies are widely used in animal tissue engineering. Such approaches provide fine control of the target cells through paracrine signalling to make functional tissues ( Cittadella Vigodarzere and Mantero, 2014 ; Paschos et al., 2015 ). Co-cultivations are very beneficial for testing drug efficacy during drug development as they provide more realistic in vivo–like conditions than mono-cultures. They allow high-throughput screening and in-depth monitoring of drug effects on cell–cell interactions ( Fang and Eglen, 2017 ; Goers et al., 2014 ). Co-cultivation strategies have been also applied for efficient degradation of different organic contaminants ( Benner et al., 2015 ; Mekuto et al., 2018 ; Zhang et al., 2013 ; Zhao et al., 2016 ). Despite extensive work on engineering microbial consortia for chemical biosynthesis, very few co-cultivation strategies have been applied in industrial biotechnology. Such industrial applications include wastewater treatment, biogas production, and the production of traditional foods. In the case of food industry, synthetic consortia are used for making dairy products such as cheese, yoghurt and kefir; bakery products like sauerkraut and sourdough; and meat products like salami ( Bader et al., 2010 ). Liquor industry widely uses different microbial consortia for making whisky, beer and wine ( Benkerroum et al., 2005 ). Finally, a co-cultivation approach has extensively been used for the production of vitamin C ( Guleria et al., 2016 ). This review describes the recent successful implementation and applications of co-cultivation methods for microbial biosynthesis using metabolic engineering approaches. It also highlights the challenges and limitations in existing co-cultivation systems and discusses how it can be improved to reach their full potential for industrial applications."
} | 1,953 |
31921047 | PMC6929412 | pmc | 811 | {
"abstract": "Quorum sensing (QS) is a cell–cell communication mechanism among bacterial populations that is regulated through gene expression in response to cell density. The pathogenicity of Xanthomonas campestris pv. campestris ( Xcc ) is modulated by the diffusible signal factor (DSF)-mediated QS system. DSF is widely conserved in a variety of gram-negative bacterial pathogens. In this study, DSF-degrading bacteria and their enzymes were thoroughly explored as a biocontrol agent against Xcc . The results indicated that a novel DSF-degrading bacterium, Acinetobacter lactucae QL-1, effectively attenuated Xcc virulence through quorum quenching. Lab-based experiments indicated that plants inoculated with QL-1 and Xcc had less tissue decay than those inoculated with Xcc alone. Co-inoculation of strains Xcc and QL-1 significantly reduced the incidence and severity of disease in plants. Similarly, the application of crude enzymes of strain QL-1 substantially reduced the disease severity caused by Xcc . The results showed that strain QL-1 and its enzymes possess promising potential, which could be further investigated to better protect plants from DSF-dependent pathogens.",
"conclusion": "Conclusion In this study, a simple and efficient method was developed for the screening of highly active DSF-degrading microorganisms. A novel bacterial isolate, A. lactucae QL-1, possessing excellent DSF-degrading activity, was identified. Strain QL-1 utilized DSF as the sole carbon source for its growth and rapidly degraded DSF. Experiments confirmed that DSF-degrading bacterium QL-1 and its crude enzymes substantially reduced the disease severity of Xcc and hence could also be used in the pre-infection period to control and prevent DSF-mediated bacterial infections. These findings establish strain QL-1 and its crude enzymes as potential biocontrol agents against infectious diseases caused by DSF-dependent bacterial pathogens. The application of efficient biocontrol agents will minimize the use of chemical pesticides and reduce their harmful effects. An in-depth study of DSF degradation-related genes and enzymes will provide a clear understanding of the pathway of DSF molecules in the natural microenvironment. Recently developed omics tools can facilitate the complete analysis of the mechanisms of QQ strains, which will help in the reduction of economic losses due to crop diseases.",
"introduction": "Introduction Cruciferous vegetables are of global economic importance and include a wide range of unique members such as Bok Choy, cabbage, and broccoli ( Rakow, 2004 ; Somec and Sondi, 2019 ). Black rot caused by Xanthomonas campestris pv. campestris ( Xcc ) becomes severely destructive under warm and humid conditions ( Gupta et al., 2013 ). This leads to serious economic losses by reducing the shelf life and market value of infected crop plants ( Massomo, 2002 ; Massomo et al., 2004 ). To overcome the effects of black rot, chemical control ( Mew and Natural, 1993 ; Mishra and Arora, 2012 ; Vicente and Holub, 2013 ), cultivation of certified disease-free transplants and seeds, cultivation of resistant cultivars, hot water treatment ( Nega et al., 2003 ), and drip irrigation are generally recommended ( Luiz et al., 2016 ). Over the past few decades, pesticides have been the main preventative applied against black rot. Pesticide overuse results in increased production cost, pesticide-resistant strains, human health issues, and environmental concerns ( Sexton et al., 2007 ). Therefore, biological control by quorum quenching (QQ) has been focused on in recent years as a promising alternative strategy against plant disease ( Bzdrenga et al., 2017 ). QQ disrupts quorum sensing (QS) either by the degradation of QS signals or interference with signal generation or perception ( He and Zhang, 2008 ). Quorum sensing is a common density-dependent mode of communication between bacteria. The expression of QS-dependent genes is regulated by synthesizing and secreting signal molecules and auto-inducers (AIs) ( Miller and Bassler, 2000 ). Signal molecules accumulate in the extracellular environment and are recognized by bacteria to detect their own population density. When the population density reaches a certain threshold, the signal molecule binds to the corresponding receptor protein and activates the expression of downstream-related genes ( Fuqua and Winans, 1994 ). Many gram-negative bacteria have QS systems to synchronize their behavior and monitor population density-dependent growth ( Nealson and Hastings, 1979 ). QS-regulated pathways play a central part in biofilm formation and the secretion of virulence factors for pathogenicity, toxicity, surface-attachment, and antimicrobial agent resistance ( Kaye and Pogue, 2015 ). Quorum quenching is a reverse technique that targets QS to affect the growth of pathogenic bacteria. QQ using enzymes and QS-inhibitors can be conducted without affecting bacterial growth. Effective reduction of bacterial infections without inducing resistance makes QQ an attractive strategy to control plant disease ( Kalia, 2015 ). QQ enzymes are especially promising because of their extracellular AI degradation and because they can be applied in large quantities. QQ enzymes can be applied in various sectors including human health, aquaculture, agriculture, water supply and drainage, and biofouling ( Bzdrenga et al., 2017 ). Currently, the most commonly discovered enzymes are N -acyl homoserine lactone (AHL)-degrading enzymes, including AHL-lactonase, AHL-acylase, and oxidoreductases ( Dong and Zhang, 2005 ; Waters and Bassler, 2005 ; Krysciak et al., 2011 ). Bacillus sp. has been reported to naturally produce lactonases or acylases to degrade AHL signals ( Dong et al., 2000 ; Zhang et al., 2002 ; Carlier et al., 2003 ). Degradation of AHLs by Bacillus thuringiensis lactonase (AiiA) reduced Pectobacterium carotovorum pathogenicity on potato slices ( Dong et al., 2004 ). Quenching signaling molecules to disrupt the QS regulatory function of bacteria is a feasible strategy for controlling signaling-mediated diseases. The quorum-sensing signal of Xcc is known as DSF and has been characterized as an α and β unsaturated fatty acid, cis -11-methyl-2-dodecenoic acid ( Supplementary Figure S1 ; Wang et al., 2004 ; He and Zhang, 2008 ). DSF represents a family of widely conserved QS signals involved in the regulation of virulence factor production in a variety of gram-negative bacteria ( Zhou et al., 2017 ). DSF plays a key role in regulating virulence, motility, antibiotic resistance, biofilm formation, and interspecies and inter-kingdom communications ( Boon et al., 2008 ; Deng et al., 2011 ). In this study, a highly efficient DSF-degrading Acinetobacter lactucae strain QL-1 was identified. The DSF-degrading ability, degradation products, and degradation mechanism of QL-1 were studied. Furthermore, the properties of QL-1 and its crude enzymes were investigated to develop pre-infection preventive measures against DSF-dependent bacterial pathogens. This preliminary study of QL-1 as a biological control agent provides new insight for signaling-mediated plant disease control.",
"discussion": "Discussion Plant diseases are the outcome of the synchronized activity of regulation networks, virulence factors, and infection processes of bacterial pathogens ( Bzdrenga et al., 2017 ). Along with sensing plant signals and nutrient availability, QS also plays a vital role in the pathogenic cycle. QQ disrupts QS by degrading and interfering with signal generation or perception. This technique has evolved as a promising novel strategy for the prevention and control of QS-mediated bacterial infections ( Mole et al., 2007 ). The plant pathogenic bacteria Pseudomonas spp., Burkholderia spp., Dickeya spp., Pectobacterium spp., Erwinia spp., and Pantoea spp. produce AHLs; Pantoea spp., Erwinia spp., and Pectobacterium spp. produce autoinducer-2 (AI-2), a furanosyl-borate diester; Legionella spp. and Vibrio spp. produce alpha-hydroxy-ketones; and Xylella fastidiosa and Xanthomonas spp. produce a wide variety of diffusible signal factors (DSFs) ( Flavier et al., 1997 ; Chen et al., 2002 ; Kendall and Sperandio, 2007 ; Zhang et al., 2007 ; Mei et al., 2010 ; Churchill and Chen, 2011 ; Winans, 2011 ; Lee and Zhang, 2015 ; Ryan et al., 2015 ; Liu et al., 2017 ; Zhan et al., 2018 ). DSF is widely conserved in gram-negative bacterial pathogens ( Deng et al., 2011 ; Zhou et al., 2017 ). Interestingly, QQ enzymes can degrade most of these signaling molecules ( Dong et al., 2000 , 2004 ; Zhang et al., 2002 ; Carlier et al., 2003 ; Uroz et al., 2008 ; Krysciak et al., 2011 ). To date, only a few studies have reported the application of DSF-degrading bacteria against Xcc , but the production of DSF-degrading enzyme CarAB (a carbamoyl phosphate synthetase) has been reported in several Pseudomonas spp. ( Shinohara et al., 2007 ; Newman et al., 2008 ). These reports verified QQ as an efficient disease prevention method in plants under laboratory conditions. During this study, a DSF-degrading bacterium, QL-1, was isolated from soil, which completely degraded (100%) DSF and produced a major intermediate product ( trans -2-dodecenoic acid). Alpha-oxidation is important in the catabolism of branched-chain fatty acids ( Casteels et al., 2010 ) for the synthesis of hydroxy fatty acids. Alpha-hydroxy fatty acid can be further oxidized and decarboxylated to a fatty acid one carbon shorter than the original. Therefore, if an odd-chain-length compound is initially used, an even-chain-length acid is produced that can be further oxidized by β-oxidation ( Quant and Eaton, 1999 ). Similarly, we speculated that hydrogen of the branched carbon atom in DSF was oxidized to hydroxyl fatty acid. Hydroxyl fatty acid was further oxidized to remove the carboxyl group, and fatty acids with one less carbon atom were formed. To facilitate beta-oxidation, the shortened carbon chain converted cis double bonds into trans double bonds, and, eventually, DSF was degraded to carbon dioxide and water without any persistent accumulative product ( Figure 5 ). Therefore, we deduced that QL-1 may possess a complete metabolic pathway for DSF degradation and metabolism. Morphological features, 16S rRNA gene analysis, and Biolog tests revealed the identity of the QL-1 strain to be A. lactucae , which is a novel species recently reported by Rooney et al. (2016) . No literature is available about the potential biocontrol activity of A. lactucae against Xcc . In addition to its DSF-degrading ability, this study also investigated the biocontrol potential of strain QL-1. Garge and Nerurkar (2017) reported an AHL-degrading isolate, Pls17, that caused tissue maceration in potato and therefore was not considered as a biocontrol agent. Not all bacterial species are capable of degrading signaling molecules, and a successful biocontrol agent should not produce detrimental effects on the host. During the current study, strain QL-1 had 100% biocontrol efficiency against Xcc without any side-effects. Similar to with a mixture of STR and XC1, tissue maceration was not observed after treatment with a mixture of QL-1 and XC1. The results indicated a significant reduction in black rot disease indices after treatment with strain QL-1, and preventive treatment completely inhibited disease symptoms. A slight decay on radish slices was observed after treatment with extracellular and intracellular enzymes of strain QL-1 and XC1. The results of the present study suggest that strain QL-1 and its crude enzymes are potent biological control agents. They can also be applied as a pre-infection preventive measure to control/attenuate black rot and other infectious diseases caused by DSF-dependent bacterial pathogens."
} | 2,963 |
35208432 | PMC8879938 | pmc | 812 | {
"abstract": "In embedded neuromorphic Internet of Things (IoT) systems, it is critical to improve the efficiency of neural network (NN) edge devices in inferring a pretrained NN. Meanwhile, in the paradigm of edge computing, device integration, data retention characteristics and power consumption are particularly important. In this paper, the self-selected device (SSD), which is the base cell for building the densest three-dimensional (3D) architecture, is used to store non-volatile weights in binary neural networks (BNN) for embedded NN applications. Considering that the prevailing issues in written data retention on the device can affect the energy efficiency of the system’s operation, the data loss mechanism of the self-selected cell is elucidated. On this basis, we introduce an optimized method to retain oxygen ions and prevent their diffusion toward the switching layer by introducing a titanium interfacial layer. By using this optimization, the recombination probability of Vo and oxygen ions is reduced, effectively improving the retention characteristics of the device. The optimization effect is verified using a simulation after mapping the BNN weights to the 3D VRRAM array constructed by the SSD before and after optimization. The simulation results showed that the long-term recognition accuracy (greater than 10 5 s) of the pre-trained BNN was improved by 24% and that the energy consumption of the system during training can be reduced 25,000-fold while ensuring the same accuracy. This work provides high storage density and a non-volatile solution to meet the low power consumption and miniaturization requirements of embedded neuromorphic applications.",
"conclusion": "5. Conclusions Due to its high-density storage and non-volatile properties, 3D VRRAM is suitable for neural network hardware implementation in embedded neuromorphic IoT systems. In order to improve the long-term accuracy and reduce the refresh rate of BNNs after training, an optimized 3D VRRAM with a TiN/TiO x /HfO x /Ti/Ru structure is proposed to implement a BNN. Since the failure of the DOT is a process of electron de-trapping near the interface, the retention of the DWT is improved by introducing a deep-level trap through the Ti interface layer. By using this DWT-based 3D RRAM, BNNs can be constructed to represent long-term inference with a high accuracy and ultra-low training power consumption. This work provides a solution for high-density neural network implementation for edge computing.",
"introduction": "1. Introduction Energy is a crucial resource for smart devices in the Internet of Things (IoT), as most applications are powered by batteries or use energy-harvesting techniques [ 1 , 2 , 3 , 4 ]. Because of this, energy-efficient artificial intelligence technologies are becoming increasingly important for the IoT. Since Deep Neural Networks (DNNs) require a high bandwidth, large memory capacity, and large power consumption, running DNNs on target embedded systems and mobile devices has become a challenge [ 5 , 6 , 7 , 8 , 9 , 10 ]. In comparison, Binarized Neural Networks (BNN) can significantly reduce computational complexity and memory consumption while having satisfactory accuracy on various image datasets [ 11 ]. In embedded IoT systems, neural networks must be able to perform pre-trained cognitive tasks in an efficient way. In this case, the weights of the trained neural network should remain unchanged and only limited in-field updates should be performed. Currently, the resource consumption by add-ons has become a limitation in memristor-based analog computing in memory systems. Analog designs require additional circuits, such as analog-to-digital and digital-to-analog converters, to fight against undesirable device properties. In contrast, binary networks offer obvious advantages in terms of speed, energy consumption, memory occupation and other aspects. However, more storage units are required as storage for the weights in BNNs. Although several software algorithms, such as sparse mapping schemes, have been proposed to address the large number of weights [ 12 , 13 ], neuromorphic architectures still demand a high amount of storage. Against this background, a 3D memristor array would be the most effective hardware scheme for maximizing the area’s efficiency [ 14 , 15 , 16 ]. However, the retention of written data on the device could affect the energy efficiency of the system’s operation [ 15 , 16 ]. The 3D architecture for a memristor includes both stacked and vertical arrays. The latter, also known as vertical RRAMs (VRRAMs), are patterned with fewer lithography steps and therefore less costly. The self-selected device (SSD) is the only way to overcome sneak current in 3D vertical integration. The currently reported SSDs, which are based on interface barrier modulation, possess good uniformity but poor retention performance [ 17 , 18 , 19 , 20 ]. Retention failures have an adverse impact on the overall accuracy of the implemented neural network, which worsens when the device operates at high temperatures during edge computing [ 11 ]. The undesirable properties of devices have also hindered the practical application of emerging memristors. In order to restore the recognition accuracy, the network needs to be retrained when the accuracy drops to a certain level. The additional power consumption of the refresh process is in direct contrast to the low power requirements of edge computing. Therefore, the retention issue in 3D memristor arrays is highly non-trivial, especially for long-term accuracy, and needs to be considered and addressed. In this work, firstly, we fabricated SSDs with the TiN/TiO x /HfO x /Ru and TiN/TiO x /HfO x /Ti/Ru structures, and then tested the electrical properties including operation voltages, nonlinearity and retention characteristics. We proposed that the underlying mechanisms of LCSs (low-conductance states) and HCSs (high-conductance states) are due to Poole–Frenkel (PF) emission and trap-assisted tunneling (TAT). Then, we brought the mechanism of retention degradation forward through a series of comparative experiments. By introducing a deep-level trap in this structure with a Ti interfacial layer, the oxygen ions are firmly trapped, and retention is highly developed. Thirdly, we simulated the long-term accuracy of the BNN through fashion-MNIST tasks by mapping binarized weights to the non-volatile 3D memristor arrays constructed by these SSDs. Due to the improved retention characteristics of the device, the trained network can guarantee a good recognition accuracy in 10 5 s, which is estimated as up to 3 years. At the same time, this scheme can reduce the energy consumption of training the network by a factor of 25,000, since the network needs to be retrained when the recognition accuracy falls below 80%. This optimized 3D vertical RRAM offers an option to provide high storage density and a non-volatile means to meet the requirements of embedded neuromorphic applications for low power consumption and miniaturization.",
"discussion": "4. Discussion As discussed above, the schematic diagram of the retention loss of the DOT is demonstrated in Figure 2 b. During the measurement of retention, electrons trapped at sites near the interface are released through tunneling, while the electrons in the middle are emptied by thermal emission, resulting in data degradation in the HCS. According to this mechanism, the retention characteristics are closely related to the energy level of the trap sites, which can be enhanced reasonably by introducing deep-level traps. The Ti layer in the proposed device is an effective approach to implement a deep-level trap [ 22 ]. Figure 3 a demonstrates the retention test results on the DWT and the DOT, where the DWT showed almost no degradation within 5 × 10 4 s, much better than that of the DOT (2000 s). Titanium has a strong ability to absorb oxygen and introduce oxygen vacancies. During the retention test for the DWT, the titanium interfacial layer retained oxygen ions and prevented their diffusion toward the switching layer. In this case, the recombination probability of Vo and oxygen ions was reduced, effectively improving the retention characteristics of the HCS of the SSD. In addition, titanium is a CMOS-compatible material and can be widely used in mass production. Therefore, a Ti interface layer insertion is a practical and effective optimization method for 3D VRRAM arrays. The statistical results of the retention testing for the DWT and the DOT are shown in Figure 3 b,c. For the DWT, the distributions of the LCS and the HCS were fully overlapped after 1000 s at room temperature (RT). In contrast, the retention evaluation of the DWT was performed at a high temperature (85 °C). Even after 30 h of baking, there was still a 10-fold difference in conductance between the LCS and the HCS. To evaluate the lifetime of the DWT, three high temperatures were adopted to accelerate aging. By extrapolating the Arrhenius plot, the lifetime of the DWT at RT was calculated to be 3 years, as shown in Figure 3 d, indicating that the introduction of the Ti layer greatly improves data retention properties. To evaluate the effectiveness of this device structure optimization on the long-term accuracy of the trained network, a multi-layer perceptron BNN was implemented for the Fashion-MNIST classification task using a 3D VRRAM array, as shown in Figure 2 b. The weights of the BNN were constrained to +1 or −1 by the activation function and were then mapped to the 3D VRRAM array. When the conductance of the device decayed over time to a pre-set threshold, the weights stored in the device became invalid. During inference, the fading weights reduced the systematic accuracy of the trained BNN. Figure 4 a shows that the recognition rate of the BNN decreased with inference time. For the network implemented by the DOT, the recognition rate dropped rapidly to below 60% after 10 4 s. For the DWT, the recognition rate remained at 84% when the inference time exceeded 10 5 s, achieving an improvement of 24%. According to the DWT lifetime evaluated in Figure 3 d, the BNN implemented with a DWT can maintain a satisfactory accuracy of more than 80% over 3 years without any refresh operations, thus achieving an ultra-low power consumption. In this scenario, we propose a hypothesis that the network needs to be retrained when the recognition accuracy drops below 80%. According to this premise, we calculated the comparison of the training consumption for the BNN implemented with the DWT and the DOT. Figure 4 b shows the energy consumed by BNN inference in Fashion-MNIST classification tasks. The HCS of the DWT was larger than that of the DOT, resulting in more energy being needed for a one-step identification operation. However, the BNN network constructed by the DOT needed to be constantly refreshed due to the poor retention characteristics of the DOT. A refresh operation corresponds to the programming operation of the device, which consumes much more power than a read operation. For picture inference, the DWT saved 25,000 times more energy than the DOT. Therefore, the high-density 3D VRRAM constructed by DWT is suitable for power-sensitive edge computing applications."
} | 2,805 |
38360683 | PMC10870720 | pmc | 813 | {
"abstract": "Background Bioconversion of plant biomass into biofuels and bio-products produces large amounts of lignin. The aromatic biopolymers need to be degraded before being converted into value-added bio-products. Microbes can be environment-friendly and efficiently degrade lignin. Compared to fungi, bacteria have some advantages in lignin degradation, including broad tolerance to pH, temperature, and oxygen and the toolkit for genetic manipulation. Results Our previous study isolated a novel ligninolytic bacterial strain Erwinia billingiae QL-Z3. Under optimized conditions, its rate of lignin degradation was 25.24% at 1.5 g/L lignin as the sole carbon source. Whole genome sequencing revealed 4556 genes in the genome of QL-Z3. Among 4428 protein-coding genes are 139 CAZyme genes, including 54 glycoside hydrolase (GH) and 16 auxiliary activity (AA) genes. In addition, 74 genes encoding extracellular enzymes are potentially involved in lignin degradation. Real-time PCR quantification demonstrated that the expression of potential ligninolytic genes were significantly induced by lignin. 8 knock-out mutants and complementary strains were constructed. Disruption of the gene for ELAC_205 (laccase) as well as EDYP_48 (Dyp-type peroxidase), ESOD_1236 (superoxide dismutase), EDIO_858 (dioxygenase), EMON_3330 (monooxygenase), or EMCAT_3587 (manganese catalase) significantly reduced the lignin-degrading activity of QL-Z3 by 47–69%. Heterologously expressed and purified enzymes further confirmed their role in lignin degradation. Fourier transform infrared spectroscopy (FTIR) results indicated that the lignin structure was damaged, the benzene ring structure and groups of macromolecules were opened, and the chemical bond was broken under the action of six enzymes encoded by genes. The abundant enzymatic metabolic products by EDYP_48, ELAC_205 and ESOD_1236 were systematically analyzed via liquid chromatography–mass spectrometry (LC–MS) analysis, and then provide a speculative pathway for lignin biodegradation. Finally, The activities of ligninolytic enzymes from fermentation supernatant, namely, LiP, MnP and Lac were 367.50 U/L, 839.50 U/L, and 219.00 U/L by orthogonal optimization. Conclusions Our findings provide that QL-Z3 and its enzymes have the potential for industrial application and hold great promise for the bioconversion of lignin into bioproducts in lignin valorization. Supplementary Information The online version contains supplementary material available at 10.1186/s13068-024-02470-z.",
"conclusion": "Conclusions In this study, we reported lignin degrader Erwinia billingiae QL-Z3, of which degradation rate was 25.24% at 1.5 g/L lignin. Disruption of potential ligninolytic genes significantly reduced the lignin-degrading activity of QL-Z3 by 47–69%. The potential metabolic pathways were deduced by FTIR and LC–MS analysis according varies degradation products. Enzyme activities of LiP, MnP and Lac were 367.50 U/L, 839.50 U/L and 219.00 U/L by orthogonal optimization. In conclusion, our study provides new insights for the biological valorization of lignin into high-value bioproducts.",
"introduction": "Introduction Lignin is the second most abundant renewable carbon source on earth, next to cellulose. It is a crosslinked phenolic polymer mainly comprised of three constituent monomers, 4-hydroxyphenyl (H), guaiacyl (4-hydroxy-3-methoxyphenyl, G), and syringyl (4-hydroxy-3,5-dimethoxyphenyl, S). The complex of these components is cross-linked together through carbon–carbon, ester, and ether linkages [ 1 ]. Recent genetic evidence indicates that cross-linking between lignin and polysaccharides has a substantial impact on plant cell wall recalcitrance [ 2 ]. Lignin inhibits plant biomass digestion in several ways, mainly by preventing microbes and enzymes from gaining access to cellulose, but also by binding digestive enzymes or releasing inhibitory breakdown products [ 3 ]. In agricultural production, a large amount of lignin exists in the waste straw, which is difficult to degrade and damage the soil environment [ 4 ]; lignin is also a by-product of wood hydrolysis and paper industry application [ 5 ]. Statistically, pulp and paper facilities produce 50–70 million tons of lignin per year, more than 695.7 million cubic meters in the global papermaking wastewater discharge, of which lignin accounts for 600–1000 mg/L in the black liquor of papermaking, accounting for 47.4% and 59.4% of chemical oxygen demand and chroma of papermaking wastewater [ 6 , 7 ]. Compared with chemical and physical methods, biological process for the degradation of lignin is regarded as eco-friendly, cost-effective and sustainable [ 6 ]. Therefore, finding efficient lignin-degrading strains is increasingly being considered an effective method for the biological pretreatment of lignocellulosic biomass. As an alternative raw material, lignin has enormous potential to replace diminishing fossil-based resources for the sustainable production of many chemicals and materials, including valuable aromatic and non-aromatic chemicals, such as vanillin [ 8 ], aromatic carboxylic acids including cis-muconic acid, adipic acid, polyols, lipids, and polyhydroxyalkanoates (PHAs). In the past few decades, researches on the microbial degradation of lignin mainly focused on the degradation of lignin by fungi [ 9 ]. Compared with fungi, bacteria can adapt to extreme environments such as strong acid, strong alkali, and abundant temperatures because they have better environmental adaptability and abundant biodiversity. In addition, bacteria provide a vast and diverse toolbox for mining and selecting ligninolytic bacterial strains and enzymes and for the metabolic engineering of bacterial cell factories for effective biomass conversion processes. Therefore, bacteria are currently in the spotlight as promising candidates for novel biomass conversion strategies due to their wide functional diversity and versatility [ 10 ]. Recently several bacterial strains, such as Streptomyces [ 11 ], Burkholderia [ 12 ], Bacillus [ 13 ], Pseudomonas [ 14 ], and Microbacterium [ 15 ], have demonstrated lignin degradation abilities. Studies on Sphingomonas paucimobilis SYK-6 have shown that this strain has a wide assimilation ability of lignin-derived biaryl groups [ 16 ]. Bacterial ligninolytic enzymes can be divided into two categories: lignin-modifying enzymes (LME) and lignin depolymerizing auxiliary enzymes (LDA). LME includes lignin peroxidase (LiP, EC 1.11.1.14), manganese peroxidase (MnP, EC 1.11.1.13), versatile peroxidase (EC 1.11.1.16), laccase (Lac, EC 1.10.3.2), and Dyp-decolorizing peroxidase (EC 1.11.1.19) [ 17 ]. All members of this group belong to class II of the catalase superfamily and participated in the depolymerization of large lignin polymers to produce phenoxy intermediates [ 18 , 19 ]. LDA enzyme includes several enzymes, such as cellobiose dehydrogenase (EC 1.1.99.18), glyoxal oxidase (EC 1.2.3.5), superoxide dismutase (EC.1.15.1.1) and glucose dehydrogenase (EC 1.1.99.10), which are incapable of degrading lignin on their own but are required for ligninolytic enzymes to degrade lignin. Moreover, other enzymes, such as oxygenase, that have not been classified, was also found to play a role in lignin degradation [ 20 ]. Here, we report the complete genome sequence of a novel lignin degrader, Erwinia billingiae QL-Z3. Based on the whole-genome annotation, eight knock-out mutants and complementary strains were constructed to confirm the function of lignin degradation genes. The heterogonous expression of six selected genes derived from QL-Z3 was performed to clarify its enzymatic properties. Then the structure and composition of the products of oxidative degrading enzymes were detected by FTIR and LC–MS. Enzymatic hydrolysis products on lignin degradation and its mechanism were analyzed to provide a speculative pathway for lignin biodegradation. Finally, the enzyme activities (LiP, MnP, Lac) from the fermentation medium of stain QL-Z3 were optimized. Taken together, the results suggest that diverse oxidative enzymes are involved in lignin degradation by QL-Z3, making this strain an ideal source of ligninolytic and lignocellulolytic enzymes and a good candidate for plant cell wall/biomass hydrolysis. Strain QL-Z3 has a significant potential for biological valorization of lignin bioconversion.",
"discussion": "Results and discussion Strain characteristics Lignin, cellulose, and hemicellulose constitute the whole biomass. At present, cellulose and hemicellulose can be converted into value-added useful products such as fuel ethanol and soluble sugars by enzymatic and/or fermentation pathways. To solve the difficulty of lignin degradation, reducing the cross-linking with other polymers in biomass is an important consideration in lignin biorefinery . Erwinia billingiae QL-Z3 was previously isolated from soil samples and stored in 25% glycerol at − 80 °C. It is known that the strain QL-Z3 can degrade lignin at 30 °C. [ 21 ]. Now we use scanning electron microscopy (SEM) imaging to reveal that QL-Z3 was a typical bacilliform with a rough surface and irregular folding of the cell surface (Additional file 1 : Fig. S1). Erwinia was reported to be related to bacterial soft rot when it was first discovered and named [ 29 ], most studies focus on the pathogenic mechanism of plants at present. Erwinia billingiae QL-Z3 is the only Erwinia strain that had been confirmed to degrade lignin, while more researchers pay attention to Bacillus , Pseudomonas and other lignin-degrading strains. Influence of culture conditions on lignin degradation While ligninolytic microbial species have been isolated and identified in recent years, there were also reports on the effects of culture conditions on lignin degradation [ 30 ]. The newly isolated E. billingiae QL-Z3 was capable of using alkali lignin, a common kraft lignin that constitutes about 85% of the total world lignin production [ 31 ], as a sole carbon source. When pH and nitrogen source were fixed, varying alkali lignin concentrations affected lignin degradation (Additional file 1 : Fig. S2). Increasing lignin concentration from 0.5 to 1.5 g/L enhanced lignin degradation from 14.5% to 18.3%, respectively, while further increasing lignin concentration to 2.0 g/L reduced lignin degradation to 11.8%. In addition to carbon source, nitrogen source has also been shown to play a role in lignin biodegradation. Supplementing nitrogen increased lignin degradation by the bacterium Rhodococcus opacus PD630 by about 10% [ 32 ], but did not significantly influence lignin degradation by the fungus Phanerochaete chrysosporium [ 33 ]. In this study, we found that ammonium sulphate ((NH 4 ) 2 SO 4 ) supplemented medium resulted in onefold higher lignin weight loss (~ 20%) than ammonium nitrate (NH 4 NO 3 ), sodium nitrate (NaNO 3 ), or tryptone supplemented medium (~ 10%). pH was another culture condition that showed significant influence on lignin degradation by QL-Z3, which was over 50% higher at acidic or neutral pH than basic pH (Additional file 1 : Fig. S2). Based on the lignin degradation rates at different single factor optimization of each gradient, the top three conditions (lignin concentration, pH, nitrogen source) and three gradients were chosen for further orthogonal experiments. The optimal results are shown in Fig. 1 , the orthogonal experimental listed in Additional file 1 : Table S1-1 and Additional file 1 : Table S1-2, respectively. Compared with the initial medium (medium 0 in Fig. 1 ), when the lignin concentration was fixed at 1.5 g/L, pH = 5, nitrogen source was (NH 4 ) 2 SO 4 , the optimized medium enhanced the degradation rate from 14.23% to 24.25%, which equates to an increase of 41.32%. Compared with other lignin-degrading strains (Additional file 1 : Table S2), QL-Z3 has the advantages of short culture cycle, simple fermentation conditions and higher degradation ability. Fig. 1 Effect of three factors on lignin degradation rate of E. billingiae QL-Z3. Medium Number: 0, 2.0 g/L lignin, pH = 7, tryptone (initial medium); 1, 1.0 g/L lignin, pH = 7, (NH 4 ) 2 SO 4 ; 2, 1.0 g/L lignin, PH = 5, NaNO 3 ; 3, 1.0 g/L lignin, pH = 9, tryptone; 4, 1.5 g/L lignin, pH = 9, NaNO 3 ;5, 1.5 g/L lignin, pH = 5, (NH 4 ) 2 SO 4 ; 6, 1.5 g/L lignin, pH = 7, tryptone; 7, 2.0 g/L lignin, pH = 5, tryptone; 8, 2.0 g/L lignin, pH = 7, NaNO 3 ; 9, 2.0 g/L lignin, pH = 9, (NH 4 ) 2 SO 4 . Average values of three replicates are shown with the standard errors of the mean shown as error bars. Each experimental group was conducted in triplicate Genomic analysis and functional verification Genomic identification of candidate ligninolytic genes and expression analysis To further dissect functional genes for lignin degradation in QL-Z3 via genome sequencing. As shown in Additional file 1 : Fig. S3, the complete genome of QL-Z3 consisted of a circular chromosome (4286,943 bp) and a circular plasmid (526,446 bp), whose GenBank accession numbers were CP037949 and CP037950, respectively. The G + C content of the chromosome and the plasmid were 56.25% and 54.67%, respectively (Additional file 1 : Table S3). The chromosome contained 4556 predicted coding sequences (CDSs), 22 rRNA genes (8 5S rRNA, 7 16S rRNA, 7 23S rRNA), and 78 tRNA genes, while the plasmid contained 166 predicted CDSs. Among these CDSs, 4428 chromosome genes and 140 plasmid genes were predicted to encode proteins through the Non-Redundant Protein Sequence (NR) Database. 3445 genes can be classified into 25 functional categories based on clusters of orthologous groups (COG) designations, including those of intracellular trafficking, secretion and vesicle transport, signal transduction mechanisms, carbohydrate transport and metabolism (Additional file 1 : Fig. S4). Genome analysis based on KEGG database identified about 287, 146, and 432 genes that are potentially involved in membrane transport, signal transduction, and carbohydrate metabolism, respectively; 74 genes belong to Xenobiotics biodegradation and Metabolism class, including glutathione S-transferase (GST), acylphosphatase which are related to Aminobenzoate degradation, Metabolism of xenobiotics by cytochrome P450 (Additional file 1 : Fig. S5). Genome analysis based on Carbohydrate-Active enZYmes Database (CAZy database) further identified the CAZyme genes, including 44 glycosyltransferases (GTs), 54 glycoside hydrolases (GHs), 24 carbohydrate esterases (CEs), 16 auxiliary activities (AAs), 4 carbohydrate-binding modules (CBMs), and 1 polysaccharide lyase (GL) (Additional file 1 : Fig. S6). The abundant GH, CBM and AA in QL-Z3 indicated that the strain also had potential cellulase, hemicellulase and pectinase, among which the AA9 family gene encoding Lytic polysaccharide monooxygenases had been confirmed to be involved in the degradation of lignocellulose [ 34 ]. To identify these enzymes related to lignin degradation by QL-Z3, the encoding genes known as lignin enzymes of lignin-degrading model strains Sphingobium sp. SYK-6, Pseudomonas putida KT2440 and other reported strains were selected as query sequences, and a BLASTp (Basic Local Alignment Search Tool) search of the QL-Z3 genome was carried out [ 16 , 35 , 36 ]. Then the reported enzymes in SYK-6 and KT2440 were compared with candidate lignin-degrading enzymes of QL-Z3 by sequence similarity and protein domain analysis. 16 enzymes were found to belong to the same protein family with their similar proteins (Table 1 ), and 14 of them had the same protein domains, indicating that these genes encoding proteins may have similar functions and lignin products. These oxidoreductases including laccases, Dyp-type peroxidases, catalases, dioxygenases, monooxygenases, glutathione synthases, superoxide dismutases etc. Laccase can degrade lignin on its own through the coordinated transfer of electrons by four copper ions [ 37 ], and there are fewer examples of bacterial laccases than of fungal laccases, therefore, it is very necessary to find more bacterial laccases. Some peroxidase in lignin-degrading white-rot fungi belong to the plant peroxidase superfamily (Class II), but LiP, MnP seem to be limited to fungi, which is rarely found in bacteria. Bacteria are relatively rich in other types of peroxidases, including Dyp peroxidase and Mn-catalase [ 38 ]. Superoxide dismutase is an antioxidant enzyme that can degrade different lignin model substrates into several compounds and play a critical role in the polymerization of lignin monomers [ 39 , 40 ]. In addition, enzymes such as dioxygenase, monooxygenase, and glutathione synthase may also degrade lignin through REDOX action. Xu et al. [ 35 ] have already identified these enzymes might be involved in the lignin catabolism pathway. Table 1 Similarity analysis of protein domain and lignin products comparison in candidate lignin degrading genes Gene ID Gene name Encode protein Reference organism Same domain % Identity/aa Accession no Lignin products Chr _ 48 / Dyp-type peroxidase Pseudomonas putida KT2440 Streptomyces pharetrae CZA14 IPR048328 31/298 25/298 WP _ 010954130 OSZ61955 Vanillic acid Chr _ 205 pgeF polyphenol oxidase/Laccase Pseudomonas putida KT2440 Sphingobium sp. SYK-6 – 56/242 35/242 WP _ 010951886 WP _ 014078201 / Chr _ 410 gshB Glutathione synthase Pseudomonas putida KT2440 Sphingobium sp. SYK-6 IPR004215 IPR011761 IPR004218 66/314 42/314 WP _ 010955561 BAK68397 / Chr _ 468 / SDR family oxidoreductase Pseudomonas putida KT2440 Sphingobium sp. SYK-6 / 33/250 38/250 WP _ 003249597 WP _ 014077645 / Chr _ 858 tauD Taurine dioxygenase Sphingobium sp. SYK-6 IPR003819 45/284 WP _ 014075385 / Chr _ 996 / NADP dependent oxidoreductase Pseudomonas putida KT2440 Sphingobium sp. SYK-6 IPR006115 IPR029154 31/304 31/304 WP _ 010955321 WP _ 014076334 / Chr _ 1016 gorA Glutathione-disulfide reductase Pseudomonas putida KT2440 Sphingobium sp. SYK-6 IPR023753 IPR004099 29/450 39/450 WP _ 010953139 WP _ 014076796 / Chr _ 1236 sodA Superoxide dismutase Pseudomonas putida KT2440 Sphingobacterium sp. T2 IPR019831 IPR019832 57/205 52/205 AAN66571 WP _ 231561313 Benzoic acid /vanillin/catechin Chr _ 1594 / Quinone oxidoreductase Pseudomonas putida KT2440 Sphingobium sp. SYK-6 IPR013154 IPR013149 IPR020843 30/326 31/326 WP _ 010951478 WP _ 014077905 / Chr _ 2803 ssuD Alkanesulfonate monooxygenase Pseudomonas putida KT2440 IPR011251 80/381 WP _ 003255833 / Chr _ 3132 katE Catalase Pseudomonas putida KT2440 Streptomyces sp. UNC401CLCol IPR010582 IPR011614 41/754 42/754 WP _ 010951777 WP _ 028959332 / Chr _ 3182 gloA Lactoylglutathione lyase Pseudomonas putida KT2440 IPR004360 IPR037523 34/135 WP _ 010954597 / Chr _ 3330 ssuD Alkanesulfonate monooxygenase Pseudomonas putida KT2440 IPR011251 62/388 WP _ 003255833 / Chr _ 3467 / Catalase Pseudomonas putida KT2440 Streptomyces sp. UNC401CLCol IPR010582 IPR011614 53/489 52/489 WP _ 010951777 WP _ 028959332 / Chr _ 3587 / Manganese catalase Sphingobium sp. SYK-6 IPR039377 66/305 WP _ 014075027 / Chr _ 4287 / Catalase Pseudomonas putida KT2440 Streptomyces sp. UNC401CLCol IPR010582 IPR011614 60/479 55/479 WP _ 010951777 WP _ 028959332 / “–” means protein domain is not same, “/” means domain or lignin products were unknown [ 16 , 35 , 36 ] To confirm the expression of the 16 candidate ligninolytic genes are subjected to induction by lignin or carbon starvation, RT-qPCR was applied to the samples grown in the media supplemented with or without 1.5 g/L lignin for 6 h and 72 h. The results showed that the transcription levels of the 3 genes ( Chr_468, Chr_2803, Chr_3132 ) were not regulated by lignin (Additional file 1 : Table S6), 13 genes were identified to be induced by lignin, including Dyp-type peroxidase ( EDYP_48 ), laccase ( ELAC_205 ), glutathione synthase ( EGSS_410 ), dioxygenase ( EDIO_858 ), oxidoreductase ( EOXI_996 , EOXI_1594 ), glutathione reductase ( EGRE_1016 ), superoxide dismutase ( ESOD_1236 ), glutathione lyase ( EGLY_3182 ), monooxygenase ( EMON_3330 ), catalase ( ECAT_3467 , ECAT_4287 ), and manganese catalase ( EMCAT_3587 ) (Fig. 2 ). The results revealed that ELAC_205 , EOXI_996 , ESOD_1236 , EGLY_3182 , and ECAT_3467 , EMCAT_3587 , ECAT_4287 were involved in transcription at 6 h, which may have the function of the LMEs to convert lignin into smaller aromatics that can be imported into the cell for aromatic catabolism [ 17 , 41 ]. Interestingly, ELAC_205, ECAT_3467 and EMCAT_3587 had high responses in both lignin and starvation medium at the first 6 h, indicating that the three genes were also induced by starvation response. Such starvation response gradually decreased and have more regulated by lignin after 72 h. We notice that ELAC_205 has no domain predicted results, but it belongs to the same family in Pseudomonas putida KT2440 and Sphingobium sp. SYK-6, which contains four classical copper ion binding sites, may catalyze the oxidation of various phenolic substrates via the reduction of oxygen to water. The highly abundant catalase and manganese catalase in QL-Z3 are catabolic genes that may be involved in lignin depolymerization or oxidative stress reaction [ 34 ]. We speculate that ECAT_3467 , EMCAT_3587 and ECAT_4287 can not only provide “H” and “O” atoms for the decomposition reaction as a substrate, but also play a vital role in aromatic catabolic reactions. ECAT_3587 contains an extraordinary manganese catalase type domain, that is, the active site containing manganese ions, which can oxidize Mn 2+ to Mn 3+ , and Mn 3+ can oxidize a large number of phenolic substrates. It has been reported that superoxide dismutase has lignin-degrading activity in Sphingobacterium sp. T2 [ 39 ], Lin L found through proteomic analysis that when lignin is used as a carbon substrate, the encoded superoxide dismutase is expressed extracellular, indicating its potential role in lignin decomposition [ 42 ]. Similar to their results, ESOD_1236 have highly expression level at 6 h, while it decreased at 72 h. Moreover, EGLY_3182 only participated in the regulation at 6 h, glutathione reductases from Sphingobium sp. SYK-6 (LigE and LigF) were reported they have the ability to cleave the intermediate with the attachment of glutathione at the C position [ 20 , 43 ], but EGLY_3182 has no similar domain with them. Fig. 2 Gene real-time expression levels at 6 h and 72 h under lignin and starvation Collected Ct values of lignin, glucose and starvation group of different genes at 6 h and 72 h, then analyzed and calculated the relative expression levels of corresponding target genes in different samples. Finally, the expression level of genes in the glucose group was set as the control value 1, the relative expression levels of the lignin and starvation group were calculated, respectively, when glucose was used as the control, and then Log 2 E logarithm was used to draw the heat map. Each experimental group was conducted in triplicate EDYP_48 , EDIO_858 , EOXI_996 , EOXI_1594 , EMON_3330 were strongly induced by lignin at 72 h, indicating that they also played an important role in lignin degradation. Compared with them, EGSS_410 and EGRE_1016 expression levels were relatively lower. Dyp is a peroxidase involved in extracellular lignin depolymerization, Ahmad M [ 44 ] found that knockout of Dyp gene in jostii Rhodococcus RHA1 significantly reduced the ability of lignin depolymerization compared with the wild type. Dyp peroxidase can oxidize the β-O-4 bond of lignin in the presence of manganese ions to obtain the lignin product vanillic acid [ 35 ]. After the lignin macromolecule depolymerization, LDA enzymes assist the core lignin-degrading enzymes to degrade the lignin monomer. Monooxygenase ( EMON_3330 ) and dioxygenase ( EDIO_858 ) are involved in lignin degradation in other studies [ 45 , 46 ], we believe that they should be considered as LDA enzymes. In addition, qPCR results identified the highly expression of oxidoreductases (especially EOXI_996) indicating its potential role in lignin decomposition, but the mechanisms of EOXI_996 and EOXI_1594 , oxidoreductase proteins, in the oxidative decomposition of lignin are unknown. In summary, we select eight different types of genes who have expression at 72 h ( EDYP_48 , ELAC_205 , EDIO_858 , EOXI_996 , ESOD_1236 , EMON_3330 , ECAT_3467 , EMCAT_3587 ) to verify the function via gene knockout and heterologous expression in the next step. Gene knock-out and phenotypic validation Deletion of EDYP_48 , ELAC_205 , EDIO_858 , EOXI_996 , ESOD_1236 , EMON_3330 , ECAT_3467 and EMCAT_3587 were confirmed using PCR amplification of the 8 target genes (Additional file 1 : Fig. S7). Complementation of the deleted genes was confirmed using the replenishment plasmid pKT100 with universal primer F/R (Additional file 1 : Fig. S8). Then, the degradation rate and enzyme activity of the knock-out mutants and the complementary strains were verified. As shown in Fig. 3 , After 72 h cultivation, lignin degradation by the wild-type strain reached about 25%. Among the mutants with gene deletions ΔEDYP_48 , ΔELAC_205 , ΔEDIO_858 , ΔEOXI_996 , ΔESOD_1236 , ΔEMON_3330 , ΔECAT_3467 and ΔECAT_3587 reduced lignin degradation rates of 11.65%, 16.73%, 13.71%, 3.67%, 17.30%, 15.55%, 5.95% and 8.94% were observed. Genetic complementation analysis demonstrated that complementary strains with six corresponding genes fully or partially restored the lignin degradation activities of the mutant strains (Fig. 3 ). The degradation rates of the two genes ( EOXI_996 and ECAT_3467 ) knock-out strains reduced less and did not differ significantly from those of their complementary strains (Fig. 3 ), that’s may because QL-Z3 contained many catalases and oxidoreductase which have the same function, while have generated other compensation pathways to complete the complex degradation process. Then we measured enzyme activities of six knock-out mutants and complementary strains, six enzyme activities assay further demonstrated that deletion of the ligninolytic genes resulted in a significant decrease in the activities of secreted ligninolytic enzymes, while expression of complementary genes fully or partially restored the decreased activities of extracellular ligninolytic enzymes (Fig. 4 ). Taken together, the results corroborate multiple enzymes in lignin biodegradation and synergism of different ligninolytic enzymes from QL-Z3. Fig. 3 Determination of lignin degradation rate of knock-out mutant mutants and complementary strains in strain QL-Z3. Blue: wild type (WT), yellow: knock-out mutants, red: complementary strains. a \n EDYP_48 wild type, knock-out mutant and complementary strain; b \n ELAC_205 wild type, knock-out mutant and complementary strain; c \n EDIO_858 wild type, knock-out mutant and complementary strain; d \n EOXI_996 wild type, knock-out mutant and complementary strain; e \n ESOD_1236 wild type, knock-out mutant and complementary strain; f \n EMON_3330 wild type, knock-out mutant and complementary strain; g ECAT_3467 wild type, knock-out mutant and complementary strain; h \n EMCAT_3587 wild type, knock-out mutant and complementary strain. a, b, c: significant difference (* p ≤ 0.05) (ANOVA analysis). Average values of three replicates are shown with the standard errors of the mean shown as error bars. Each experimental group was conducted in triplicate Fig. 4 Enzyme activity of knock-out mutant mutants and complementary strains in strain QL-Z3. Blue: wild type (WT), yellow: knock-out mutants, red: complementary strains. a \n EDYP_48 wild type, knock-out mutant and complementary strains; b \n ELAC_205 wild type, knock-out mutant and complementary strains; c \n EDIO_858 wild type, knock-out mutant and complementary strains; d \n ESOD_1236 wild type, knock-out mutant and complementary strains; e \n EMON_3330 wild type, knock-out mutant and complementary strains; f \n EMCAT_3587 wild type, knock-out mutant and complementary strains. a, b, c: significant difference (* p ≤ 0.05) (ANOVA analysis). Average values of three replicates are shown with the standard errors of the mean shown as error bars. Each experimental group was conducted in triplicate FTIR analysis The peaks in the FTIR spectra in lignin were identified by reference to the literature [ 47 , 48 ], and the infrared spectra before and after degradation of alkali lignin by wild type, mutant type, and compensatory type (using non-inoculated strains as control) were analyzed. It was observed that the infrared spectral intensity of alkaline lignin changed significantly at 2800 cm −1 , 1700 cm −1 –1000 cm −1 (lignin fingerprint), and 1000 cm −1 –500 cm −1 bands with QL-Z3 treatment (Fig. 5 ). The absorption peak at 2800 cm −1 represents the antisymmetric stretching vibration of methyl group C–H bond and methoxy O–CH 3 bond in lignin, and the strength of this peak decreases, indicating that lignin was degraded. In the wild type, the C–H deformation of the representative methyl group and the 1464 cm −1 –1462 cm −1 cm vibration of the aromatic ring are reduced, which means that the content of the methoxy group is reduced due to demethylation [ 49 ]. Interestingly, with the exception of EDIO_858 and EMON_3330 , there was no significant difference between the complement and mutants of other genes and the control group, indicating that these genes were involved in the C–H bending of methyl groups, which caused strong shaking of the lignin skeleton. The intensity of the peak corresponding to the C–O bond (1147 cm −1 ) in lignin and the guaiac-based C–H plane (989 cm −1 –846 cm −1 ) were changed, and the peaks of the wild type were obviously stronger, while the peak signal of the mutant type was weakened and disappeared, indicating that these genes played an important role in C–O and C–H fracture. In addition, the changes in the peaks of 623 cm −1 and 549 cm −1 also indicate that the C–H bond in the lignin skeleton has a violent oscillation, which is higher than that of mutant and complement type. Fig. 5 Fourier transform infrared spectra of alkali lignin degradation genes. a \n EDYP_48 , b \n ELAC_205 , c \n EDIO_858 , d \n ESOD_1236 , e \n EMON_3330 , f \n EMCAT_3587 in knock-out mutants (blue), complementary strains (green), wild type (WT, red) and alkali lignin without strains (black). Each experimental group was conducted in triplicate Heterologous expression and characterization of ligninolytic enzymes To further characterize the function of the ligninolytic enzymes encoded by genes from E. billingiae QL-Z3, recombinant plasmids expressing these proteins were constructed and transformed into E. coli BL21. Single colony grown on the culture dish was verified by colony PCR. Recombinant enzymes were induced by isopropyl-β-D-thiogalactopyranoside (IPTG) and purified by Ni–NTA affinity chromatography (Additional file 1 : Fig. S9). The activities of these proteins were performed as described in Materials and Methods by using different substrates, such as H 2 O 2 and veratryl alcohol etc. Enzyme activity assays of purified proteins demonstrated that as predicted by genomic and genetic analyses, recombinant proteins of EDYP_48, ELAC_205, EDIO_858, ESOD_1236, EMON_3330, and EMCAT_3587 possessed the activities of dye decolorization peroxidase, laccase, dioxygenase, superoxide dismutase activity, monooxygenase, and manganese catalase, respectively (Additional file 1 : Table S7). After comprehensive analyses, including bioinformatics analysis, RT-qPCR, genetics, phenotyping, biochemistry, and FTIR, we believed that enzymes encoded by EDYP_48 , ELAC_205 , and ESOD_1236 have an important role in the first step of lignin depolymerization process. Therefore, we investigated their enzymatic characteristics. The degradation efficiency of 2, 2’-azino-bis (3-ethylbenzothiazoline-6-sulfonic acid) (ABTS, lignin structure analogs) by these three enzymes were tested. After the assays, the resulting visible ESOD_1236 had weak oxidation activity on ABTS, which was consistent with the results reported that SOD treatment did not produce precipitation in acid-precipitable polymeric lignin [ 42 ]. These results suggested that SOD was an LDA enzyme and had a poor ability to degrade lignin alone. The activity conditions (pH, temperature) of enzymes encoded by EDYP_48 , ELAC_205 were optimized (Additional file 1 : Fig. S10), and the activity of ELAC_205 was measured in sodium acetate buffer at pH 0.5 to 8.0. The results showed that ELAC_205 had the highest enzyme activity in sodium acetate buffer at pH and temperature was 1.0 and 50 °C, respectively, which had excellent tolerance in strong acids. Because of the partial denaturation of EDYP_48 at pH < 4.0, we measured EDYP_48 activity only in sodium acetate buffers at pH 4.0 to 9.0. The results showed that EDYP_48 had the highest enzyme activity at 40 °C and pH 5.0. On this basis, EDYP_48 (Km = 0.73 mM) and ELAC_205 (Km = 0.36 mM) both showed catalytic activity with substrate ABTS. Compared with reported proteins, it was observed that EDYP_48 performed higher catalytic efficiency (Table 2 ). Table 2 Kinetics parameters comparison of EDYP_48 and ELAC_205 Name Encoding protein Km (mM) Kcat (s −1 ) Kcat/Km (s −1 mM −1 ) References EDYP_48 Dyp-type peroxidase 0.730 ± 0.021 5.87 8.04 × 10 3 This study PputA514_2985 Dyp-type peroxidase 0.585 ± 0.012 4.85 8.29 × 10 3 [ 42 ] PputA514_3152 Dyp-type peroxidase 0.217 ± 0.022 1.42 6.53 × 10 3 [ 42 ] ELAC_205 Laccase 0.360 ± 0.07 0.60 1.67 × 10 3 This study LacZ1 Laccase 0.35 / / [ 37 ] Lacc Laccase 1.44 1378.09 9.54 × 10 5 [ 50 ] Enzymatic hydrolysis and product analysis of lignin by ligninolytic enzymes Enzymatic hydrolysis rate of alkali lignin EDYP_48 and ELAC_205 are the core LME enzymes in the process of lignin degradation. It is necessary to elucidate the degradation products of a single enzyme and the synergism of them. As an LDA enzyme, ESOD_1236 can also assist the LME enzyme in producing various centrosomes of lignin [ 39 ]. Then we calculated the enzymolysis efficiency of alkaline lignin under different combinations of three pure enzymes. When three enzymes were used alone (Fig. 6 ), the degradation rates of ELAC_205 (6.16%) and EDYP_48 (5.81%) were significantly higher than those of ESOD_1236 (2.32%). In the case of double enzyme combination, the efficiency of LD (ELAC_205 and EDYP_48) combination was the highest (9.40%), and there was no significant difference between LD combination and triple enzyme combination. Fig. 6 Enzymatic hydrolysis rate of alkali lignin in different combinations of the three enzymes. L: ELAC_205; D: EDYP_48; S: ESOD_1236; LD: ELAC_205 and EDYP_48; DS: EDYP_48 and ESOD_1236; LS: ELAC_205 and ESOD_1236; LDS: three enzymes mixed. a, b, c: significant difference (* p ≤ 0.05) (ANOVA analysis). Average values of three replicates are shown with the standard errors of the mean shown as error bars. Each experimental group was conducted in triplicate LC–MS determination and product analysis To further know how three enzymes encoded by genes of QL-Z3 depolymerize large lignin polymers to produce phenoxy radical intermediates and then the conversion of heterogeneous lignin derivatives to central intermediates. The metabolites of three enzymes, in different combinations were detected via LC–MS. The metabolites were identified by accuracy mass and MS/MS data which were matched with the metabolite database. The changes of each degradation product were obtained by quantitative comparison with the peak area of the control group, the significant difference analysis and screening were conducted from the list of primary and secondary metabolites (Fig. 7 ). PCA score plot showed obvious separation of different treatments, especially the control group and Dyp group, which indicated significant changes of metabolites after enzymolysis for 24 h, and the amount of “up” “down” metabolites varies with different treatments. Fig. 7 Three enzymes degrade alkali lignin in different combinations. a PCA scores with positive ion modes, b PCA scores with negative ion modes and c differential metabolite statistics with positive ion modes, d differential metabolite statistics with negative ion modes. Each experimental group was conducted in sextuplicate According to LC–MS analysis, the degradation products produced by enzymatic hydrolysis of alkali lignin were mainly acids, alcohols, ketones, phenols and aromatics. Compared with the control group, polyaromatic acids were listed in Additional file 1 : Table S8, which significantly up-regulated and down-regulated as degradation products. Single enzyme hydrolysis of alkali lignin by Lac produced the most hydrolysis products, a total of 36 up regulation and 16 down regulation, the contents of the most including phenylethanol, protocatechuic acid, catechin, and benzoic acid, because laccase is generally considered to play an important role in the ring opening of benzene, and studies have shown that it can bind to Cu 2+ ions and catalyze oxidation reactions including polymer degradation and oxidative coupling of phenolic compounds to generate more degradation products [ 37 ]. Dyp-peroxidase recently received attention due to its special primary and tertiary structure and broad substrate specificity, which has a higher REDOX potential without additional mediator to promote lignin degradation [ 51 ]. Rahmanpour et al. [ 52 ] found that overexpression of DyP1B in E. coli has the activity of oxidizing Mn 2+ and degrading powdery wheat grass lignocellulose. It has been reported that the DyP peroxidase of Basidiomycete Auricularia auriculajudae has the potential of cracking lignin substructural bonds [ 53 ], EDYP_48 seems to have the same potential. Lignin degradation by EDYP_48 alone accumulated 32 main products and consumed 12 products. Similar to LiP mechanism, EDYP_48 utilize free radical reaction to destroy the C α –C β bond on the monomer side chain of lignin [ 35 ], thus producing a large amount of methoxyphenol and its derivatives. In addition, synergy of EDYP_48 and ELAC_205 produced the largest variety of “up” products, such as aromatic acids and quinones (Additional file 1 : Table S8). The treatment of organic solutes lignin by MnSOD1&2 would result in aryl-Cα oxidation bond cleavage, as well as C α –C β oxidation cleavage, decarboxylation and O-demethylation, and authors speculated that the reaction cycle consisted of two half reactions, that is, Mn (III) oxidized superoxide to dioxygen, and Mn (II) reduced superoxide to hydrogen peroxide [ 39 ]. In this study, the products of ESOD_1236 were the least in several combinations, but increased after combining with the two enzymes. In the aromatic degradation products obtained by SOD alone, products such as 1-Naphthol, 2-Ketobutyric acid, o-Xylene, alpha-Ketoisovaleric acid, Aminomalonic acid, (+)-Syringaresinol, and 3-Furoic acid intermediates, however, these intermediates were not detected in L (ELAC_205) and D (EDYP_48) alone or the combined of the two and three enzymes, indicating that further degradation of these intermediates by SOD is difficult. We speculated that the large lignin polymer was depolymerized under the first step of hydrolysis/LME. Lac generates phenoxy free radicals through the synergistic action of Dyp peroxidase under aerobic conditions, and then SOD and LME enzyme were co-oxidized and decomposed or demethylated to transform into more intermediates, forming an unstable aromatic ring structure and promoting the degradation of lignin. Main degradation pathways of lignin by three enzymes Lignin is formed from the guaiacyl units (G, precursor coniferol alcohol), the syringyl units (S, precursor sinapyl alcohol) and p-hydroxyphenyl units (H, precursor p-coumarol alcohol) by the polymerization of β-O-4, β-5 and β–β types. In this study, a comparison of the degradation products obtained from different combinations of EDYP_48, ELAC_205 and ESOD_1236, we found that sinapyl alcohol and coniferyl alcohol are the main monomer forms that exist after the first damage of alkaline lignin skeleton (Fig. 8 ). Next, under enzymatic catalysis, the phenylpropane monomer structure is converted into 3-Methoxyphenylacetic acid, 2-methoxyhydroquinone, syringic acid, 2, 4-dihydroxyacetophenone and other complex intermediates, which are often produced by the pyrolysis of C α –C β phenols. Under aerobic conditions, 3-Methoxyphenylacetic acid reacts to form 2-Methylbenzoic acid and 4-Methylbenzaldehyde, syringic acid is demethylated to produce acetosyringone, which is reoxidized to produce eugenol and 3, 4-dimethoxyacetophenone. Ferulic acid is another phenolic component of lignocellulose, which was a significant product of coniferol monomer. In Sphingomonas paucimobilis SYK-6, ferulic acid coenzyme a converts ferulic acid to vanillin; Decarboxylase of Bacillus BP-7, which can be converted into corresponding 4-vinyl aromatic compounds [ 54 ]. In this study, the addition of ESOD_1236 down-regulated isoferulic acid and produced a variety of benzaldehyde and vanillic acid products. Vanillic acid is converted into protocatechuic acid by demethylation of three combined enzymes. Literature indicated that protocatechuic acid is mainly degraded by protocatechuic 3, 4-dioxygenase through oxidative ring cleavage, and then metabolized via the β-keto-adipic pathway or the orthogonal cleavage pathway [ 55 ], but LDS (three enzymes mixed) lacks this capability, which becoming a rate-limiting step and results in the accumulation of protocatechuic acid more than 1000 times. Finally, under the action of multi-enzyme oxidation, these intermediate aromatic rings break to form aliphatic compounds and enter the TCA cycle. Fig. 8 Proposed lignin degradation pathways with three enzymes. The enzymes final degradation products in the pathway map are identified via LC–MS and the reaction mechanisms are cross-validated by previous literature [ 39 , 45 – 50 , 52 ]. Red indicates that the products is produced in different combinations of the three enzymes. L: ELAC_205; D: EDYP_48; S: ESOD_1236; LD: ELAC_205 and EDYP_48; DS: EDYP_48 and ESOD_1236; LS: ELAC_205 and ESOD_1236; LDS: three enzymes mixed. Detailed enzyme information can be found in Additional file 1 : Table S8 Optimization of enzyme activities Three extracellular enzyme activities were determined by UV–visible spectrophotometry at fixed absorbance. As is evident in Table 3 , 2.0 g/L lignin concentration, pH = 9.5, nitrogen Source was NH 4 NO 3 , supported the highest enzyme activity of all. At 30 °C, Lac and MnP maximum activity reached 265 U/L and 202 U/L, followed by LiP was 110 U/L at 25 °C. Table 3 Effect of different single factor on ligin degradation enzyme activity of QL-Z3 Condition Gradient Lip (U/L) Lac (U/L) Mnp (U/L) Lignin concentration 0.5 g/L 141 38 150 1.0 g/L 195 56 149 1.5 g/L 226 58 225 2.0 g/L 293 78 326 2.5 g/L 240 64 271 pH 5 80 113 161 6.5 112 129 163 8 115 214 220 9.5 130 226 224 11 109 146 219 Nitrogen source NaNO 3 151 138 120 NH 4 NO 3 245 152 198 NH 4 SO 4 128 113 110 Tryptone 110 53 75 yeast extract 87 181 76 Temperature 20 81 150 108 25 110 177 175 30 99 265 202 35 56 145 154 40 87 196 165 Based on the above results, the top three significant factors for supporting enzyme activity were lignin concentration, pH, nitrogen source. Next, we performed an L9 (3 3 ), three-factor, three-level orthogonal experiment to test the effects of different combinations of these conditions. The optimal results are shown in Fig. 9 , the orthogonal experimental listed from Additional file 1 : Table S9-1 to Additional file 1 : Table S9-6. The result validated that the optimal conditions of LiP production were 3 g/L alkaline lignin, pH = 8, nitrogen source was (NH 4 ) 2 SO 4 , enzyme activity increase 3.57 time; Under the condition of 3 g/L alkaline lignin, pH = 8 and nitrogen source NH 4 NO 3 , the MnP activity of strain QL-Z3 was 839.5 U/L and Lac activity was 219.0 U/L, which were 3.18 and 2.84 times of those before optimization, respectively (Table 4 ). Fig. 9 Effect of three factors on enzyme activity of QL-Z3. Medium Number: A 2.0 g/L lignin, pH = 11, NaNO 3 ; B 2.0 g/L lignin, pH = 9.5, (NH 4 ) 2 SO 4 ; C 2.0 g/L lignin, pH = 8, NH 4 NO 3 ; D 2.5 g/L lignin, pH = 11, (NH 4 ) 2 SO 4 ; E 2.5 g/L lignin, pH = 8, NaNO 3 ; F 2.5 g/L lignin, pH = 9.5, NH 4 NO 3 ; G 3.0 g/L lignin, pH = 8, (NH 4 ) 2 SO 4 ; H 3.0 g/L lignin, pH = 11, NH 4 NO 3 ; I 3.0 g/L lignin, pH = 9.5, NaNO 3 . Each experimental group was conducted in triplicate Table 4 Comparison of the initial medium and optimized medium in enzyme activity Enzyme Initial medium (U/L) Optimized medium (U/L) Increase (%) LiP 104.1 367.5 253 Lac 77.0 219.0 184 MnP 263.8 839.5 218 Compared with other strains, LiP, MnP and Lac have the advantages of short fermentation time, multiple and higher enzyme activities (Additional file 1 : Table S10). The LiP, Lac and MnP activities in the crude enzyme solution of QL-Z3 fermentation were optimized to improve their enzyme activities, which applied for the production of biopharmaceuticals and industrial enzymes, such as laccase has great application potential as biocatalyst for papermaking, in environmental remediation, drug detection and biosensors in the clothing industry [ 56 ]; Peroxidase is widely used in wastewater treatment technology in paper production and clothing production [ 41 ]; These enzymes were discovered that can transform lignin-derived aromatic compounds have important theoretical and practical significance in alleviating environmental pollution caused by lignin, sustainable agricultural development and diversification of earth resources utilization."
} | 11,479 |
39309514 | PMC11414745 | pmc | 817 | {
"abstract": "Superhydrophobic polyurethanes offer robust hydrophobicity and corrosion resistance. However, it is essential to consider the durability and environmental constraints associated with these materials. This study prepared a bio-based superhydrophobic polyurethane coating film using epoxidized soybean oil, superhydrophobically modified silica nanoparticles, and OH–PDMS–OH as surface modifiers. The coating film exhibited sustained super-hydrophobicity and an excellent antifouling effect for pu-erh tea and edible oils, among other substances, after 14 days of immersion in solutions with different pH values, 28 days of exposure to air, and 2000 abrasion cycles. This finding can be applied to the development of daily indoor and outdoor antifouling protective coatings and provides a new method for the preparation of green and durable superhydrophobic antifouling coating films.",
"conclusion": "4. Conclusion Bio-based polyols with highly reactive groups, containing a hydroxyl value of 1.89 mm g −1 , were synthesized from epoxidized soybean oil and mono-ethyl itaconic acid as raw materials with monothioglycerol in a two-step process. The synthesized polyurethane is transparent and has good flexibility in 18 MPa and WCA = 80°. Polyurethanes with binary soft segments were prepared using OH–PDMS–OH as an antifouling agent. Subsequently, during the synthesis of the polyurethane, a superhydrophobic antifouling coating film was obtained by blending superhydrophobically modified nano-SiO 2 to form a rough microstructure inside and on the surface of the coating film and WCA max = 156.7°. The coating film retained its superhydrophobic and oil- and dirt-resistant properties after 2000 friction cycles, immersion in acid and alkali solutions, and outdoor placement for 28 days. The bio-based polyurethane coating film exhibited long-lasting super-hydrophobicity and oil- and dirt-resistant properties, which provide a promising solution for future indoor and outdoor antifouling protective coating films.",
"introduction": "1. Introduction Polyurethane, as one of the most widely used and highly researched protective products in the world today, is extensively applied in various products in our daily lives. These include coatings for leather, 1 high-strength adhesives for concrete, 2 artificial blood vessels, 3 controlled drug release 4 and more. Haibo Wang et al. reported a PPG-based corrosion-resistant and hydrophobic self-healing polyurethane coating with mesoporous silica loaded with CeO 2 and Phen, and the coating has a maximum WCA = 108°. 5 This type of polyurethane, which consists of petroleum-based polyols and isocyanates, is easier to synthesize but is usually only incinerated or landfilled after use, resulting in soil and ecosystem contamination. 6 In light of the increasing depletion of petroleum resources and global climate warming, this type of polyurethane does not align with the concept of green and sustainable development. 7 To address such issues, substituting petroleum-based polyols with bio-based polyols for synthesizing polyurethane has gained widespread favor among researchers. 8 In a study by Zhou et al. , they utilized castor oil and long-chain hydrophobic chain extenders to create water-resistant polyurethane with strong hydrophobic properties and a corrosion-resistant water contact angle (WCA) of 87.5°. 9 Vegetable oil, as one of the most common bio-based sources, is characterized by its low toxicity, biodegradability, and long alkyl chains that enhance the hydrophobic properties of materials. Although the low molecular weight of vegetable oil can ultimately affect the tensile strength and glass transition temperature of polyurethane, thus limiting its applications, the hydroxyl groups, triglycerides, and carbon–carbon double bonds in its structure provide different reactive sites. Therefore, vegetable oil-based polyurethanes have great potential for functionalization. 10 The lotus leaf effect in nature, butterfly wings and insect epidermis have inspired researchers to study superhydrophobic surfaces. 11 Superhydrophobic surfaces have a very high contact angle (CA ≥ 150°). Several superhydrophobic polyurethane materials (films, 12 sponges 13 and coatings 14 ) have been studied in recent years. The current research on superhydrophobic coatings is primarily focused on applications such as acid and alkali resistant and corrosion protection coatings, 15,16 highly efficient separators of oil–water emulsion, 17,18 mechanically robust and anti-icing application coatings, 19,20 self-healing superhydrophobic specialty coatings, 21 and UV-resistant, self-cleaning coatings. 22 Fluorine atoms possess strong hydrophobic and oleophobic properties; the traditional method for preparing superhydrophobic polyurethane involves using fluorides chemically bonded to the polyurethane structure, or embedding or spraying fluoride-modified nanoparticles onto the polyurethane. 23,24 Despite the advantages of fluorinated reagents, their potentially long-lasting effects on the environment, the accumulation of fluoride, which can pose a health hazard (endocrine, fertility), and the relatively high cost of fluorinated reagents have greatly limited their use in everyday applications. 25 During the curing of polyurethane, the Si–O–Si chains in PDMS migrate from the interior of the molecular segments to the surface of the coating driven by low surface energy, 26,27 so that PDMS has the same hydrophobic properties as fluorinated reagents and is considered to be an alternative to fluorinated reagents due to their inherent good biocompatibility, the low burden on the environment during production and relatively low costs. 28 In recent years, in the field of hydrophobicity, researchers have chemically bonded PDMS with different end groups onto polyurethane substrates to produce a protective coating film with excellent hydrophobicity, oleophobic, and anti-icing properties. 29–31 In this study, a high hydroxyl value polyol (with a hydroxyl value of 1.89 mm g −1 ) with multiple reaction sites was prepared from renewable epoxidized soybean oil (ESO), mono-ethyl itaconic acid (MEI), and monothioglycerol. A biobased superhydrophobic material was produced by combining octadecylsilane-modified fumed silica nanoparticles and mono-terminal dihydroxy PDMS, which serve as a low-surface-energy modifier and a biobased superhydrophobic composite material, respectively. The coating film was subjected to comprehensive evaluation to ascertain its resilience to super-hydrophobicity under diverse application scenarios, including abrasion, acid and alkali corrosion, and outdoor exposure. Additionally, its antifouling and self-cleaning properties were assessed under varying life use conditions, such as methylene blue solution, methyl orange solution, pu-erh tea, edible oil, and more. The film is anticipated to serve as a protective coating on a range of materials, both indoors and outdoors. The synthesis process is illustrated in Fig. 1 . Fig. 1 The preparation process of the composite coating films.",
"discussion": "3. Results and discussion 3.1 Material synthesis and characterization Acetone-d6 was chosen as the solvent. Fig. 2d shows the 1 H NMR spectra of epoxy soybean oil-based polyols and soybean oil. The peaks at 6.32 and 5.82 ppm belong to the protons c on the carbon–carbon double bond structure from mono-ethyl itaconic acid, 34 and peaks at 3.42 and 4.1 ppm come from the protons e and d on the mono-ethyl itaconic acid. The peaks at 3.62 and 3.85 ppm come from the protons g and h of monothioglycerol. The peaks at 4.85, 0.89, and 2.38 ppm belong to the protons a, b, and f of the soybean oil structure. The integrated area of the peak belongs to proton c on the 6.32 ppm carbon–carbon double bond and the integral area of the peak at 0.89 ppm corresponds to the protons b on terminal –CH 3 of soybean oil. The grafting rate of itaconic acid mono-ethyl ester can be determined using the equation G C \n \n\n<svg xmlns=\"http://www.w3.org/2000/svg\" version=\"1.0\" width=\"13.200000pt\" height=\"16.000000pt\" viewBox=\"0 0 13.200000 16.000000\" preserveAspectRatio=\"xMidYMid meet\"><metadata>\nCreated by potrace 1.16, written by Peter Selinger 2001-2019\n</metadata><g transform=\"translate(1.000000,15.000000) scale(0.017500,-0.017500)\" fill=\"currentColor\" stroke=\"none\"><path d=\"M0 440 l0 -40 320 0 320 0 0 40 0 40 -320 0 -320 0 0 -40z M0 280 l0 -40 320 0 320 0 0 40 0 40 -320 0 -320 0 0 -40z\"/></g></svg>\n\n C = 9 × I double bond / I CH 3 , G C C = 2.65. It can be inferred from this that an average of 2.5 mono-ethyl itaconic acid is integrated into the structure of epoxy soybean oil. FTIR can determine the occurrence and termination of the reaction. Fig. 2e shows the FTIR spectra of poxy soybean oil-based polyols and soybean oil. The success and completion of ring opening and click reactions can be demonstrated by the peak of carbon–carbon double bonds that first appears and then disappears at around 1640 cm −1 . The disappearance of the peak of the epoxy group at approximately 827 cm −1 indicates the success of the impurity removal step. And the increase in the hydroxyl value of polyols can be estimated by the change in peak at 3300–3500 cm −1 . Fig. 2f shows the FTIR spectra of fumed nano-SiO 2 and superhydrophobically modified SiO 2 . At 2840 cm −1 and 2920 cm −1 , corresponding to the stretching vibrations of CH 3 and CH 2 , the absorption bands are attributed to the long alkyl chains brought by OTMS. However, the self-condensation of OTMS may occur during the hydrolysis process, making it difficult to rule out the possibility that CH 2 may also come from self-condensation. Fig. 2g shows the XPS of OTMS modified the SiO 2. It is evident that SiO 2 is composed of the elements Si and O. Superhydrophobically modified SiO 2 comprises the elements Si, O, and C. The presence of up to 81.14% carbon in superhydrophobically modified silica confirms the successful integration of the hydrophobic modifier OTMS into the silica. 3.2 Effect of OH–PDMS–OH and OTMS/SiO 2 loading on coating film physical properties OH–PDMS–OH and silica are used together as hydrophobic modifiers. Therefore, it is necessary to consider the influence of different loading amounts on the physical properties of the coating film. It was heated from room temperature to 800 °C at a heating rate of 20 °C min −1 under an inert environment (N 2 ) using a thermogravimetric analyzer. Fig. 3a and b show the TGA-DTG results. It can be seen that the thermal decomposition process of polyurethane mainly consists of two stages. It is located around 250 °C and 380 °C respectively. The initial stage, spanning a temperature range of 20 °C to 250 °C, is characterized by a loss in weight due to the breakage of residual small molecules and fatty acid chains. This phenomenon is attributed to the evaporation of physically adsorbed water and the breaking of small molecule fatty acid chains during synthesis. The second stage, spanning 250 °C to 380 °C, exhibits the most rapid rate of thermal decomposition. This stage encompasses the degradation of carbon and sulfur single bonds and urethane carbamate bonds within the hard segments of the urethanes. 33 The temperature range from 380 °C to 500 °C is primarily associated with the degradation of the soybean oil-based soft segments of the urethanes and the siloxane chains of OH–PDMS–OH. As illustrated in Fig. 3a and b the thermal decomposition rate of polyurethane exhibits a decline with the incorporation of both OH–PDMS–OH and superhydrophobically modified SiO 2 into the system. This phenomenon can be attributed to the chemical bonding of the OH–PDMS–OH moiety added between the chains of polyurethane macromolecules, which results in the formation of a cross-linked network structure. This process markedly enhances intermolecular interactions. Furthermore, flame-retardant silicone serves to impede the transfer of heat. 35 Fig. 3 The TG results of PU (a), the DTG results of PU (b), the stress–strain curve of PU (c), and the visible light transmission degree of PU (d). \n Fig. 3c shows the mechanical properties of cured samples by tensile testing. The tensile strength of the polyurethane without the modifier was 18 MPa and the elongation at break was 29%. The incorporation of OH–PDMS–OH resulted in an enhancement of the tensile strength of PU up to 40 MPa. However, both the tensile strength and elongation at break of PU exhibited a decline with the increase in the addition amount. Furthermore, the incorporation of superhydrophobically modified silica resulted in a further reduction in both the tensile strength and elongation at break of PU. The incorporation of OH–PDMS–OH into a polyurethane system gives rise to the formation of binary polyurethane soft segments, which in turn leads to an enhancement in the cross-linking density of the system. This is accompanied by the formation of intermolecular hydrogen bonds that impede the relative sliding of the molecules, thereby increasing the tensile strength. 36 However, the tensile strength decreased with the addition of OH–PDMS–OH, probably because too much molecular weight of OH–PDMS–OH affected the density of effective crosslinking points in the PU, resulting in a decrease in the overall structural strength of the material. The tensile strength of the coating film decreased with the addition of superhydrophobically modified silica, which was attributed to the uneven dispersion of modified silica particles in the composite coating film to form aggregates, and these microscopic defects were highly susceptible to crack initiation and propagation when the material was subjected to stretching, ultimately leading to a decrease in mechanical properties. 37 In a region from 300 to 800 nm in LAMBDA, the wavelength indicating transparency of the coating film is 500 nm. Fig. 3d shows that the polyurethane-coated film with no modifier added and the polyurethane-coated film with only OH–PDMS–OH added are transparent. The transparency of the coated film is observed to decrease with the addition of OH–PDMS–OH, yet it remains transparent. However, the incorporation of superhydrophobically modified SiO 2 results in the film becoming opaque. 3.3 Superhydrophobic surface tests The hydrophobicity and antifouling properties of the polyurethane were improved by the addition of OH–PDMS–OH, and finally the polyurethane was brought to a superhydrophobic level by the addition of superhydrophobically modified meteorological nano-silica. Based on the superhydrophobic polyurethane composite coating film, three kinds of lossy properties that are easily encountered in daily life were investigated. First, contact angle tests were conducted on polyurethane-coated films prepared using all experimental protocols. Fig. 4a shows that the contact angle of the polyurethane coating film synthesized from the original soybean oil-based polyol was 86°. This angle was observed to increase to 122.9° when OH–PDMS–OH was introduced to the synthesis. To minimize the impact of petroleum-based polymers on the biobased content of the coating film, an experimental group of 20% P/PU coating films was selected for further investigation. Upon the addition of superhydrophobically modified SiO 2 up to 20% wt., the interfacial contact angle of the coating film reached 154.8°, indicating a transition from hydrophobicity to super-hydrophobicity. A video of the superhydrophobic coating can be seen in Video S1. † Fig. 4 The contact angle test graphs of the different experimental groups (a). The contact angle graphs after two weeks of immersion in various acid and alkali solutions (b). The contact angle graphs after different friction cycles (c). The contact angle graphs after different days of outdoor exposure (d). The coating film is capable of maintaining its physical and chemical properties in an acidic or alkaline environment, preventing degradation and ensuring its long-term hydrophobic properties. However, if these properties are not maintained, the coating film's applications in acidic or alkaline environments are significantly limited. Fig. 4b illustrates the deionized water control group. Solutions with varying pH values were prepared, and the samples were fully submerged in them. After 2 weeks, the samples were removed for contact angle testing. The results demonstrate that following immersion in disparate pH solutions, the contact angle of the coating film diminished, yet it retained its super-hydrophobicity at WCA = 150.1°. This evidence substantiates the assertion that the synthesized polyurethane coating film is well-suited for application in daily life environments. The ability of superhydrophobic coatings to withstand abrasion is a critical factor that contributes to their extensive range of applications. A coating film that is both abrasion-resistant and stable is less susceptible to damage from external environmental factors and is capable of maintaining stable hydrophobicity over an extended period, thereby extending the service life of the coating. Fig. 4c shows the testing methodology employed for the superhydrophobic sample, which was placed face down on 2000 SiC sandpaper with a load weight of 100 g. The sample was then pulled horizontally by 10 cm at a constant speed, and the contact angles were measured after 500, 1000, 1500, and 2000 cycles, respectively. Following 2000 cycles of friction, the contact angle of the coating film exhibited a slight decrease. However, it remained superhydrophobic at WCA = 150.3°, indicating that the coating film has the potential to be utilized in abrasive environments. As illustrated in Fig. S1, † the maximum contact angle in this work is better than that of most of the fluorine-free superhydrophobic polyurethanes reported in the existing literature, and the superhydrophobic retention after sandpaper friction cycle testing is better than that of most existing superhydrophobic polyurethanes in the literature. 38–45 As a protective coating film that can be applied outdoors, it is necessary to evaluate its performance under the influence of various environmental factors, including ultraviolet radiation, humidity, precipitation, and dust contamination. Fig. 4c shows the testing of superhydrophobic samples which were placed outside on the roof of an outdoor building for different periods and then retrieved and tested for the contact angle. The contact angle of the coating film exhibited a slight decrease. However, it remained superhydrophobic at WCA = 153.3°, indicating that the coating film has potential for use in outdoor protective coatings. 3.4 Morphological studies of the hydrophobic coating film Scanning electron microscopy enables the observation of thin coating film surfaces, providing information on surface roughness, texture, and pore structure. Fig. 5a displays the surface morphology of the unmodified polyurethane coating film, revealing its smooth texture and low adhesion to water droplets, which indicates its relatively good hydrophobicity. Fig. 5 The SEM testing of PU, 20% P/PU, 20% O/Si–20% P/PU, O/Si (a), the EDS results of 20% P/PU (b), the EDS results of 20% O/Si–20% P/PU (c), and the EDS results of O/Si (d). Following the addition of OH–PDMS–OH, the microscopic morphology of the PU coating film exhibited a wrinkle-shaped alteration. This can be discerned through the surface elemental analysis depicted in EDS in Fig. 5b . The wrinkled region is characterized by an enrichment of silicon and oxygen elements, thereby indicating that the silicone cover layer is capable of forming a microscopic concave-convex structure on the surface of the coating film, which serves to increase the surface roughness and consequently enhances the contact angle between the coating film and the liquid, thereby resulting in a more hydrophobic surface. When superhydrophobically modified silica is added, it creates numerous tiny particles that adhere to the surface of the coated film. This is in addition to the fold-shaped bumps created by OH–PDMS–OH, which significantly enhances the surface roughness of the coated film, making it superhydrophobic. \n Fig. 5c shows the modified silica addition to the PU coating film; the surface of the PU coating film showed a rough microscopic nanoscale structure. This is because the modified silica can increase the roughness and surface area of the surface, and this structure allows the water droplets to form a larger contact angle on the surface of the coating film, making it easier for the water droplets to slide on the surface and exhibit super-hydrophobicity. According to the micro-zone compositional analysis of the EDS in Fig. 5d , it is evident that the agglomerated round spheres, sized at 10 nm, are rich in silicon, oxygen, and carbon elements. This analysis confirms that these microspheres are hydrophobically modified silica. The SEM and EDS analyses show that as the amount of OH–PDMS–OH and hydrophobically modified silica added increases, the microscopic surface of the membrane becomes progressively rougher. This change can account for the increase in the contact angle of the PU membrane from 86° to 156.7°. The detailed results of the SEM-EDS tests for all experimental groups are shown in Fig. S2 and S3. † 3.5 Antifouling application test The ultimate goal of the superhydrophobic antifouling coating film is to achieve excellent liquid repellency and easy-to-clean self-cleaning properties. Fig. 6 shows that the objective of superhydrophobic antifouling coating films is to attain optimal liquid-repellent properties and straightforward self-cleaning capabilities. Polyurethane samples were vertically immersed in a series of common liquids, including methylene blue solution, methyl orange solution, pu-erh tea, and cooking oil, to study the coating films' antifouling ability. Upon removal of the coated films from water-soluble pollutants, it was observed that no liquid adherence occurred on the surface of all coated films, thereby demonstrating that the coated films possess excellent anti-fouling capabilities against water-soluble pollutants. However, in the pure polyurethane and the experimental group with the addition of OH–PDMS–OH, it can be observed that there are notable residual deposits of oily pollutants on the surface of the coating film. Following the addition of superhydrophobically modified SiO 2 , it was observed that oily pollutants gradually converged to the bottom and could be easily wiped away with a paper towel. This demonstrates that the coating film exhibits dual characteristics of super-hydrophobicity and oil repellency after adding superhydrophobically modified SiO 2 . The results of the antifouling tests for all experimental groups are presented in detail in Fig. S4. † Fig. 6 The diagram of the antifouling application test process."
} | 5,719 |
35253476 | PMC9045482 | pmc | 818 | {
"abstract": "ABSTRACT The success of tropical scleractinian corals depends on their ability to establish symbioses with microbial partners. Host phylogeny and traits are known to shape the coral microbiome, but to what extent they affect its composition remains unclear. Here, by using 12 coral species representing the complex and robust clades, we explored the influence of host phylogeny, skeletal architecture, and reproductive mode on the microbiome composition, and further investigated the structure of the tissue and skeleton bacterial communities. Our results show that host phylogeny and traits explained 14% of the tissue and 13% of the skeletal microbiome composition, providing evidence that these predictors contributed to shaping the holobiont in terms of presence and relative abundance of bacterial symbionts. Based on our data, we conclude that host phylogeny affects the presence of specific microbial lineages, reproductive mode predictably influences the microbiome composition, and skeletal architecture works like a filter that affects bacterial relative abundance. We show that the β-diversity of coral tissue and skeleton microbiomes differed, but we found that a large overlapping fraction of bacterial sequences were recovered from both anatomical compartments, supporting the hypothesis that the skeleton can function as a microbial reservoir. Additionally, our analysis of the microbiome structure shows that 99.6% of tissue and 99.7% of skeletal amplicon sequence variants (ASVs) were not consistently present in at least 30% of the samples, suggesting that the coral tissue and skeleton are dominated by rare bacteria. Together, these results provide novel insights into the processes driving coral-bacterial symbioses, along with an improved understanding of the scleractinian microbiome. IMPORTANCE The rapid decline of coral reefs, driven by climate changes, calls for manipulative interventions such as the use of probiotics, which can assist the resilience of these ecosystems. However, many knowledge gaps still exist in our understanding of coral-bacterial symbioses that need to be addressed before effectively applying interventions like probiotics. Here, by investigating the microbiomes of 12 common coral species we show that the associations with bacterial symbionts, thought to be critical to coral health, were influenced to some extent by host phylogeny, skeletal architecture, reproduction, and anatomical compartments. We therefore propose that fundamental and applied functional exploration of coral-associated microbes will help inform successful reef management measures.",
"conclusion": "Conclusions. Through a combination of a homogeneous experimental design that minimizes external biases affecting the microbiome; the use of innovative technologies, including micro-CT scanning to evaluate host traits; and the application of a range of statistical analyses, our study allowed us to unravel the structure of the coral microbiome and quantify how it is influenced by host phylogeny, skeletal architecture, and reproductive mode. We show that host phylogeny and traits explained 14% of the tissue and 13% of the skeletal microbiome composition and were associated with a range of microbial partners thought to affect holobiont health and functioning. Based on our results, we hypothesize that reproductive mode may influence the microbiome composition in a predictable manner, while skeletal architecture works like a filter affecting bacterial relative abundance. Although our analysis accounted for some of the most influential processes known to affect the microbiome composition, these could only marginally explain the microbiome variation of tissue and skeleton. A holistic view of the mechanisms determining the holobiont composition will be gained by assessing the influence of the physicochemical and dynamic biochemical environment of the coral colony on the microbiome composition and by assessing whether the presence of some bacteria (whether dominant or rare) may affect the overall structure of the microbiome. In this study, we provided substantial evidence that coral tissue and skeleton microbiomes are dominated by rare taxa and differ in bacterial abundance, but a consortium of bacteria can colonize both compartments and the skeleton can function as a microbial reservoir. While our study answers several unsolved questions about the bacterial community structure of scleractinian corals and the mechanisms driving its composition, it also exposes knowledge gaps. Despite our focus on commonly studied coral species, we identified three bacterial groups that are understudied or were not previously reported in the coral literature. This highlights that a full characterization of the taxonomic composition of the coral microbiome has still not been achieved. Substantial further work will be needed to fully understand its functions in the coral holobiont, its fine-scale distribution in relation to ecological microniches, and the metabolic hand-offs that happen among microbiome members and with the host. The use of putative beneficial microorganisms has been proposed as a tool to mitigate the increasing pressure of anthropogenic activities on coral reefs ( 75 ); therefore, we hope that the detailed knowledge gained from our study about persistent association between specific coral lineages and bacterial taxa can form the basis for further advances in probiotic strategies to improve coral resilience in future climate scenarios.",
"introduction": "INTRODUCTION Ecology and evolution shape trait variation across species and populations, influencing host-microbiome associations ( 1 , 2 ). Some of the interactions between the host and its associated microbial symbionts affect the fitness of the holobiont ( 3 ), leading to extraordinary evolutionary outcomes that have shaped life on Earth. Corals form a holobiont with unicellular microalgae (Symbiodiniaceae) and a diverse range of microbes, which include bacteria, fungi, and viruses ( 4 ). The coral holobiont is considered to be an independent level of selection ( 5 ), but our understanding of the key mechanisms driving host-symbiont assemblages is limited. It has been observed that host evolutionary processes ( 6 ), skeletal architecture ( 7 , 8 ), and mode of reproduction ( 9 – 12 ) contribute to microbiome composition; alongside these factors, a range of host ecological and morphological traits take part in the establishment and development of the coral microbiome. The environment can be considered the main microbial reservoir for corals ( 9 , 13 , 14 ), and, for instance, the substrate underlying the colony is known to influence the holobiont structure ( 14 ). Host health status can be associated with shifts in the symbiotic microbial community that can disrupt the integrity of the holobiont ( 15 , 16 ). The developmental stages of the host are related to successional processes of the microbiome that occur over time and influence the microbial richness ( 17 ). Disentangling the individual and combined effects of these factors is paramount to understanding microbial community assembly in corals. Closely related species can harbor microbiomes that resemble each other, a pattern known as phylosymbiosis ( 18 ), which has been shown for several terrestrial and marine host-microbe systems ( 18 , 19 ), including those of corals ( 6 , 19 ). Phylosymbiosis can be driven by a range of mechanisms, including microbial filtering moderated by evolving host traits, ecological interactions with the host, or cophylogenetic relationship between the host and microbes ( 20 , 21 ). However, the fraction of the coral microbiome influenced by these mechanisms remains unquantified. Several coral traits can more directly affect the microbiome. Coral skeletal architecture differs among species ( 22 ), affecting the physicochemical properties of the colony ( 23 ) and the microbial biomass ( 7 ). For instance, light scattering is a function of skeletal density ( 8 ), resulting in different spatial gradients of light intensity and spectral composition that affect the microbiome ( 8 , 24 – 26 ). Coral reproductive traits can influence the early microbiome, which plays a key role during coral ontogeny and affects the fitness of the host ( 10 ). Broadcast-spawning corals depend mostly on the acquisition of their symbionts from the environment (horizontal transmission) ( 9 , 27 , 28 ), as their gametes, embryos, and 3-day-old larvae are devoid of bacteria ( 29 – 31 ). Some brooding species exhibit a certain degree of vertical transmission ( 12 , 27 , 31 ), as bacteria have been found associated with the ectoderm layer in newly released planulae ( 12 ). However, whether coral bacterial communities can be predicted based on host traits is still to be investigated. The coral microbiome is a dynamic system and provides an excellent case study of a highly diverse biological community contributing to holobiont success. Unravelling the composition of coral microbiomes is a necessary step toward understanding the interspecies relationships and functioning of the holobiont. To reduce the variability introduced by a range of factors known to affect the coral holobiont, such as colony age ( 17 ), spatial-temporal variability ( 32 ), and health status ( 33 ), we only sampled visibly healthy adult colonies from a small geographical location over 3 weeks. Using this approach, our study aimed to disentangle the influence of drivers affecting the microbiome composition, with a particular focus on host phylogeny, skeletal architecture, and reproductive mode. We also aimed to unravel the bacterial community structure and composition of the coral tissue and skeleton, including rare bacterial taxa.",
"discussion": "RESULTS AND DISCUSSION Experimental design and sequencing statistics. We investigated and compared the structure and composition of the microbial communities of the coral tissue and skeleton, as well as those of surrounding seawater and sediment, and assessed differences among groups using multivariate analyses and ordination ( Fig. 1 ). Then, to quantify the relative contribution of host phylogeny, skeletal architecture, and reproductive mode to the bacteria residing in the tissue and skeleton ( Fig. 2 ), we performed variation partitioning analysis. Finally, to identify the influence of these predictors on the presence and abundance of specific bacterial lineages, we used canonical correspondence analysis (CCA). Our sampling design also aimed to reduce factors known to affect the coral holobiont. Thus, we sampled only healthy adult colonies from a small geographic area, within 3 weeks, and we identified their tissue and skeletal prokaryotic communities using 16S rRNA gene amplicon sequencing because it provided the most comprehensive characterization of the bacterial community composition, including that of the rare bacterial taxa. FIG 1 Vignette of the experimental design aiming to answer the following questions: what are microbiome structures of the tissue and the skeleton of scleractinian corals, and how do they differ from the seawater and sediment microbiomes? FIG 2 Vignette of the experimental design aiming to answer the question: to what extent host phylogeny, skeletal architecture and reproductive mode affect the microbiome composition of tissue and skeleton of scleractinian corals? After denoising, amplicon sequence variant (ASV) filtering and contaminant removal, the 16S rRNA gene data set consisted of 6,977,071 reads (see Table S1 in the supplemental material) with an average length of 255 bp. The data set did not include unclassified reads ( Table S1 ). The sediment samples contained 91,738 reads (minimum, 11,203; mean, 13,508; maximum, 17,475) resulting in 1,047 ASVs; the seawater samples contained 67,540 reads (minimum, 15,946; mean, 18,348; maximum, 23,430) resulting in 1,026 ASVs; and the coral samples contained 6,817,793 reads (minimum 1,993; mean, 47,872; maximum, 115,669) resulting in 14,910 ASVs. 10.1128/msystems.00044-22.1 TABLE S1 Amplicon sequence variant (ASV) table showing reads count and taxonomical identification after denoising, ASV filtering, and contaminant removal. The table also shows which samples belong to each coral species. Download Table S1, XLSX file, 8.8 MB . © Crown copyright 2022. 2022 Crown https://creativecommons.org/licenses/by/4.0/ This content is distributed under the terms of the Creative Commons Attribution 4.0 International license . Coral tissue and skeletal microbiomes overlap but differ in their relative abundances. By comparing the coral tissue and skeleton microbiome ( Fig. 1 ) our work shows that a large fraction of bacteria can colonize both anatomical compartments (i.e., 86% of tissue ASVs were found in at least one skeleton sample, and 56% of skeletal ASVs were found in at least one tissue sample). Thus, our results support the hypothesis formulated by Marcelino et al. ( 34 ) that the skeleton can serve as a reservoir for coral tissue microbes. On the basis of this, it can be hypothesized that after a period of dysbiosis, beneficial bacteria could quickly repopulate the tissue from the skeleton ( 34 ). Despite this, the β-diversity of coral tissue and skeletal microbiomes differed significantly in 8 out of 12 coral species ( P values in the range of 0.001 to 0.016; see Fig. S1 ), with the exceptions of Isopora palifera ( P = 0.053), Montipora digitata ( P = 0.075), Goniopora tenuidens ( P = 0.061), and Platygyra daedalea ( P = 0.21). Therefore, our results indicate that while many bacteria are not selective in colonizing the coral tissue or skeleton, the complex array of biotic and abiotic interactions characteristic of each compartment shapes the microbiome composition differently ( Fig. S1 ). 10.1128/msystems.00044-22.7 FIG S1 Biplot principal-component analyses (PCAs) represent β-diversity metrics of each coral species. Red dots and areas represent skeletal samples, while blue dots and areas represent tissue samples. Within each biplot are also reported the results of nonparametric multivariate analysis of variance (NPMANOVA) analyses assessing differences in composition between the bacterial communities of the skeleton and the tissue of each coral species. Download FIG S1, TIF file, 2.2 MB . © Crown copyright 2022. 2022 Crown https://creativecommons.org/licenses/by/4.0/ This content is distributed under the terms of the Creative Commons Attribution 4.0 International license . Coral tissue and skeleton are also known to differ in their physicochemical environment, with the skeleton offering a wide array of microniches ( 23 ), and one study reported that the skeletal communities are more diverse than their tissue counterparts ( 6 ). Our data showed comparable ɑ-diversities of tissue and skeleton microbiomes for most coral species ( Fig. S2 ). This conflicts with the findings of Pollock et al. ( 6 ), who found differences in the ɑ-diversities of the two anatomical compartments; while it may represent an underlying biological cause, methodological differences between the studies may also contribute to the observed differences (e.g., using ASVs versus operational taxonomic units [OTUs] for taxonomic resolution). 10.1128/msystems.00044-22.8 FIG S2 Box plots showing observed ɑ-diversity metrics of the tissue (blue) and skeleton (red) of each coral species. Within each plot are reported the P value calculated through Welch’s t test or the Mann-Whitney U test. Download FIG S2, TIF file, 1.8 MB . © Crown copyright 2022. 2022 Crown https://creativecommons.org/licenses/by/4.0/ This content is distributed under the terms of the Creative Commons Attribution 4.0 International license . Host traits and phylogeny shape the bacterial community composition. We quantified the influence of some of the most significant processes known to influence the establishment and development of the coral microbiome ( Fig. 2 ), such as host phylogeny ( 6 ), skeletal architecture ( 7 ), and reproductive mode ( 9 , 10 ), by applying variation partitioning analysis to the tissue and skeletal bacterial communities. Our models explained 14% of the tissue and 13% of the skeletal microbiome variation, leaving a high proportion of the variation (>85%) unexplained by the predictors ( Fig. 3a and d ). This is not entirely surprising, since past reports have found that the coral microbiome is variable within a single coral colony ( 35 ) and among host genotypes ( 36 ), species, and reef habitats ( 37 – 39 ). We hypothesize that a combination of unmeasured environmental variables, functional redundancy (same functions putatively conveyed by different bacterial taxa), community assembly processes, and physicochemical properties specific to each coral colony take part in determining the coral holobiont assemblage and limiting its broader consistency. FIG 3 Partitioning of variation in the bacterial community composition of coral tissue (a) and skeleton (d) explained by host skeletal architecture, phylogeny, and reproductive mode. Adjusted R 2 values are given. Canonical correspondence analysis (CCA) biplots represent the tissue (b) and skeletal (c) microbiome structures according to each explanatory variable. Arrows represent the quantitative explanatory variables skeletal architecture and phylogeny (PV1, PV2, and PV3), with arrowheads indicating the direction of increase. The categorical explanatory variables broadcast spawners, brooders, and mixed mode are positioned on the biplot according to their coordinates. All of the explanatory variables in both CCA biplots were highly significant ( P < 0.001). In variation partitioning analysis (a, d), the total explained and unexplained variance can exceed 100%. Host phylogeny drives key coral-bacterial associations. Employing linear discriminant analysis effect size (LeFSe) analysis, we identified members of the coral microbiome that were present at higher abundance in the coral tissue or skeleton ( Fig. 4 and Table S2 ); then, through CCA analysis, we assessed their association with host phylogeny and coral traits. In the CCA analysis, the three coral phylogenetic variables (PV1, PV2, and PV3; see “Host phylogeny” in Materials and Methods) showed a strong association with some ASVs ( Fig. 2b and c ). FIG 4 Heatmap resulting from the linear discriminant analysis effect size (LEfSe) analysis showing differentially abundant bacterial amplicon sequence variants (ASVs) between coral tissue and skeleton. Specifically, ASVs with a log(LDA) of >3 (Kruskal-Wallis test: P < 0.05) and present in at least 10 samples (inclusive of tissue and skeleton) were considered differentially abundant, and their abundance was Z-score transformed (legend on the top right; darker red indicates more abundant bacteria). On the top are reported the tissue (Tis.) and skeleton (Ske.) of each coral species analyzed in this study, their clade (robust or complex), and their reproductive mode (spawner, brooder, or mixed). On the left are reported the taxonomic classification of the differentially abundant bacterial ASVs at the genus, family (f), order (o), or class (c) level. Taxa indicated in boldface represent the bacterial groups discussed in this study. Tissue and skeleton microbiomes differed in terms of bacterial abundances. Bacteria that preferentially colonized the tissue included Alteromonas , Endozoicomonas , Halomonas , Pseudoalteromonas , Vibrio , and Woesia , while among those that preferentially colonized the skeleton were “Candidatus Amoebophilus,” Kiloniellaceae , Myxococcales , Rhizobiales , and Spirochaeta . The heatmap also shows that the microbiome of each coral species is characterized by a few abundant and many rare bacterial taxa. The coral species A. aspera and G. retiformis are not shown on the heatmap because LeFSe analysis did not detect differences between their respective tissue and skeletal microbiomes. 10.1128/msystems.00044-22.2 TABLE S2 Linear discriminant analysis effect size (LeFSe) analysis of ASV output list. The table shows in which coral anatomical compartment and species the ASVs were differentially abundant, along with ASV identifier and taxonomical identification. Download Table S2, XLSX file, 0.02 MB . © Crown copyright 2022. 2022 Crown https://creativecommons.org/licenses/by/4.0/ This content is distributed under the terms of the Creative Commons Attribution 4.0 International license . Endozoicomonas species were found at higher relative abundance in the coral tissue (33.30% of the tissue ASVs and 7.43% of the skeleton ASVs across the whole data set; see Fig. 4 and Table S2 ), and individual ASVs belonging to this genus were more abundant in corals of the robust clade ( Fig. 3b ; positive correlation with PV1) and the families Poritidae ( Fig. 3b ; positive correlation with PV2) and Pocilloporidae ( Fig. 3b ; positive correlation with PV3). Our results confirmed Endozoicomonas being a highly prevalent genus of coral symbionts across coral species ( 40 , 41 ); given their putative beneficial roles, which include nutrient provision and host homeostasis support ( 41 , 42 ), it seems possible that these bacteria developed a mutualistic symbiosis with corals over time. Alteromonas and Pseudoalteromonas species were found at higher relative abundances in the tissue ( Fig. 4 and Table S2 ) of robust corals (0.74% and 0.37% of the tissue ASVs and 0.05% and 0.08% of the skeleton ASVs; Fig. 3b ) and in the family Pocilloporidae (0.14% and 0.25% of the tissue ASVs and 0.05% and 0.08% of the skeleton ASVs; Fig. 3b ). Additionally, Pseudoalteromonas species were associated with the family Poritidae ( Fig. 3b ). These bacteria are thought to take part in nitrogen cycling and antibacterial activity ( 43 , 44 ), and because of these functional features, they could play pivotal roles in the holobiont of these coral lineages. We hypothesize that the absence of correlation between Alteromonas species and Poritidae, which suggests a low relative abundance of these bacteria in this coral lineage, could be driven by competition for similar resources between these and other bacteria. Bacteria of the genera Spirochaeta and “ Candidatus Amoebophilus” were found at higher relative abundances in the coral skeleton (1.29% and 14.60% of the skeleton ASVs and 0.19% and 0.97% of the tissue ASVs; Fig. 4 and Table S2 ), and individual ASVs belonging to these genera correlated with corals in the robust clade and in the family Poritidae ( Fig. 3c ). Spirochaeta species usually thrive in oxygen-deprived environments ( 45 ) such as the coral skeleton and, given their ability to fix nitrogen and carbon ( 46 ), could be key members of the community of the skeletal environment. Despite “ Candidatus Amoebophilus” having been flagged as a member of the coral core microbiome ( 24 , 38 ), its role in the holobiont is still unclear. Available data on “ Candidatus Amoebophilus” ( 47 ) and “ Candidatus Amoebophilus asiaticus” ( 48 ) show reduced genomes with limited metabolic capabilities, suggesting they may rely on the host for survival. Interestingly, the “ Candidatus Amoebophilus asiaticus” genome harbors a high count of eukaryotic domain-like proteins, which include Ankyrin repeats and WD40 repeat domain proteins. These proteins are used by intracellular pathogens, as well as by symbiotic bacteria, to interact with hosts and modulate host response via a multitude of protein-protein interactions ( 48 ). A wide arsenal of eukaryote-like proteins has been reported in many coral-associated bacterial groups ( 49 ). Bacteria in the order Myxococcales were found at higher relative abundance in the coral skeleton (5.82% of the skeleton ASVs and 0.63% of the tissue ASVs across the whole data set; Fig. 4 and Table S2 ) and were correlated with corals in the robust clade ( Fig. 3c ). Moreover, recent studies have shown that these microbes might have codiversified with corals ( 6 ). Myxococcales species are known to play beneficial roles in other systems, such as agricultural settings ( 50 ), where they keep pathogen populations under control by releasing large quantities of antibiotics. The skeletons of many corals analyzed in this study concurrently harbored a high relative abundance of sequences affiliated with Myxococcales and potentially pathogenic bacteria (i.e., Vibrio and Serratia ). Thus, as reported for agricultural settings, and given that all the sampled colonies were visibly healthy, Myxococcales species may play similar roles in the coral holobiont by controlling pathogen populations ( 51 ). Skeletal architecture affects microbial abundance. Micro-computed tomography (micro-CT) analysis allowed us to characterize the skeletal architecture of each coral species and showed the variation in porosity within and across coral species ( Fig. 5g ). The characteristic skeletal architecture of each coral species explained a portion of the tissue and skeletal microbiomes variations ( Fig. 3a and d ). The skeleton affects the physicochemical properties of the whole coral colony, including the tissue, and Marcelino et al. ( 8 ) found that, by refracting light back, the skeleton transports and redistributes light across the colony and alters the light environment in the tissue, affecting the Symbiodiniaceae. One could propose that this also causes downstream effects on other members of the tissue microbiome. Accordingly, we found that skeletal architecture was associated with several key members of the tissue microbiome, including Endozoicomonas , Alteromonas , and Pseudoalteromonas ( Fig. 3b ). Our results also suggest that skeletal architecture could alter the relative abundance of bacterial species rather than filtering them out entirely. For instance, we found that despite the bacterial genera Bacillus , Halomonas , and Vibrio being mainly associated with coral skeletons with larger pore sizes, such as M. digitata and Acropora aspera ( Fig. 5g and Table S3 ), they were also present in species with more dense skeletons at low relative abundances (e.g., Pocillopora damicornis , Isopora palifera , and Stylophora pistillata ; see Fig. 5g and Table S3 ). Similarly, the bacterial genera Pseudohaliea and Rhodoplanes , which we found at higher relative abundances in species with more dense skeletons (e.g., Goniastrea retiformis , I. palifera , and Porites lutea ; see Fig. 5g and Table S3 ), were also present in coral skeletons with larger pore sizes (e.g., those of Platygyra sinensis and M. digitata ; see Fig. 5g and Table S3 ) at low relative abundances. While our data do not allow us to disentangle the causative factors determining the differences in relative abundance of bacteria across coral species with porous or dense skeletons, it is possible that the skeletal architecture could affect the microniches of the coral colony and determine a more suitable or hostile environment for specific bacteria. FIG 5 (a) Representative slice of micro-computed tomography (micro-CT) data through a sample of Porites lutea (FRH49) showing denser structure as brighter gray to white values and less dense structure, such as air, as dark gray to black values surrounding the specimen. Black scale bar, 10 mm. (b) The same slice segmented into three phases, as follows: skeletal material (shaded yellow); organic matrix (shaded purple); and air/bubble wrap within and surrounding the specimen (unshaded black). (c) Grayscale histogram of 16-bit micro-CT data showing the points of segmentation between the different phases. The two peaks within the lowest density phase represent the air within and surrounding the specimen (peak around 12,000) and the bubble wrap and plastic specimen holder (peak around 17,000). (d) Difference between the 20-μm and 10-μm data trends highlighting the very small variability within the porosity trends, with no consistent trend toward an under- or overestimate of porosity at one given resolution. (e) Plots of skeletal volume trends within the x - y cross-section of a 12 mm by 12 mm by 12 mm region of interest (ROI) as a function of distance along the z axis of the ROI for both 20-μm and 10-μm micro-CT data. (f) Representative slice (at 4.3 mm) from a P. lutea sample comparing the 20-μm to 10-μm micro-CT data (left subpanels) and segmented skeleton in both (right subpanels). White scale bar, 6 mm. (g) Box plots showing the variability of porosity within and across coral species. 10.1128/msystems.00044-22.3 TABLE S3 Some bacterial taxa that were more abundant in coral species with a denser or more porous skeleton and their relative abundances. Download Table S3, XLSX file, 0.02 MB . © Crown copyright 2022. 2022 Crown https://creativecommons.org/licenses/by/4.0/ This content is distributed under the terms of the Creative Commons Attribution 4.0 International license . Reproductive mode influences the microbiome composition in a predictable manner. Variation partitioning analysis showed that reproductive mode explained a portion of the microbiome variation (tissue, 3%; skeleton, 4%; see Fig. 3a and d ). Additionally, CCA analysis revealed that ASVs associated with the bacterial taxa Acinetobacter spp., Bacillus spp., Cryomorphaceae , Endozoicomonadaceae , Pseudomonas spp., and Rhodobacteraceae correlated with the reproductive mode variables ( Fig. 3b ). These bacterial taxa were reported in studies whose focus was the coral microbiome establishment ( 9 – 11 , 52 ), and some of them are thought to play important roles, including in nutrient provision and support of host homeostasis ( 41 – 44 ). Although our analysis was not conceived to investigate the microbiome establishment, the correspondence between our findings and those of past reports suggests that coral reproductive mode could predictably influence the microbiome composition and that some early host-symbiont associations may persist across a coral’s lifetime. Our data show that tissue and skeletal bacteria correlated with the reproductive mode variables (broadcast spawners, brooders, and mixed mode; see Fig. 3b and c ) and were associated with key holobiont members, including Endozoicomonas , Alteromonas , Pseudoalteromonas , and Myxococcales ( Fig. 3b and c ). These bacteria are all known for their putative beneficial roles and could help the host by being involved in processes such as nutrient cycling and support of homeostasis ( 42 – 44 , 51 ), which could also facilitate the early developmental stages of the coral. In the tissue, the reproductive mode was also associated with bacterial taxa such as Serratia and Vibrio ( Fig. 2b ), which are known as potential pathogens in some coral species ( 11 , 53 , 54 ). As suggested in previous studies, under normal conditions, these putative pathogens are commensal members of the holobiont, while their detrimental potential could emerge during dysbiotic states ( 55 ). Coral microbiomes are dominated by rare bacterial species. Except for a few bacteria that were consistently present across several coral species (e.g., Endozoicomonas , “ Candidatus Amoebophilus,” Fulvivirga , Vibrio , Alteromonas , and Pseudoalteromonas ), ASVs affiliated with rare bacterial taxa dominated the coral microbiomes. The great majority of bacterial ASVs (99.6% for tissue and 99.7% for the skeleton) were not consistently present in at least 30% of the samples of our data set. The core microbiome, i.e., the ASVs occurring in at least 30% of samples, consisted of only 16 ASVs in the tissue (12 Gammaproteobacteria ASVs, 1 Alphaproteobacteria ASV, 1 Deltaproteobacteria ASV, 1 Bacilli ASV, and 1 Bacteroidia ASV; see Table S4 ) and 9 in the skeleton (2 Gammaproteobacteria ASVs, 2 Alphaproteobacteria ASVs, 2 Bacteroidia ASVs, 2 Bacilli ASVs, and 1 Spirochaeta ASV; Table S4 ). Even fewer ASVs were present when we analyzed the core microbiome at higher thresholds, i.e., 5 ASVs occurring in at least 50% of tissue samples (4 Gammaproteobacteria ASVs and 1 Bacilli ASV) and 1 single ASV in the skeleton ( Bacteroidia ; see Table S4 ). Finally, only 1 ASV in the tissue ( Gammaproteobacteria ) and 1 in the skeleton ( Bacteroidia ; see Table S4 ) occurred in at least 60% of samples, and none were present at higher thresholds. The prevalence of few abundant and many rare bacterial lineages is clearly apparent when the composition of the coral microbiomes is displayed as a heatmap ( Fig. 4 ). 10.1128/msystems.00044-22.4 TABLE S4 List of core taxa retrieved from the tissue and skeleton of the 12 coral species confounded. The core microbiome was identified as ASVs occurring in at least 30%, 50%, and 60% of the samples. Download Table S4, XLSX file, 0.01 MB . © Crown copyright 2022. 2022 Crown https://creativecommons.org/licenses/by/4.0/ This content is distributed under the terms of the Creative Commons Attribution 4.0 International license . A closer look at microbiome variability within species showed that even within closely related hosts, most bacterial ASVs were not present across individuals of the same species ( Table S5 ), leading to wide variability between samples. The skeletal microbiome showed a higher proportion of core members than the tissue microbiome in 9 out of 12 coral species ( Table 1 and Table S5 ). Since the microbial communities of tissue and skeleton were dominated by rare bacterial ASVs ( Table 1 and Table S5 ), we analyzed their Pielou’s evenness, a measure indicating how similar the abundances of different species in the microbiome are ( 56 ). Despite finding comparable evenness of the tissue and skeletal bacterial communities for all of the species confounded ( Fig. S3a ), when we compared across coral species, we found high variability in the evenness of both tissue and skeletal microbiome ( Fig. S3 b and S3 c). For instance, in G. retiformis , both tissue (Pielou’s evenness mean, 0.87) and skeletal (Pielou’s evenness mean, 0.74) microbiomes showed evenly distributed bacterial populations, while in S. pistillata , the tissue microbiome (Pielou’s evenness mean, 0.39) showed variability across individuals, and the genera Endozoicomonas and “ Candidatus Amoebophilus” dominated the skeletal microbiome (Pielou’s evenness mean, 0.20). In line with our results, it has been previously reported that the microbiome composition of each coral species can show various degrees of diversity ( 57 ), and in some cases, one or a few bacterial taxa can be dominant, like in Pocillopora verrucosa , whose bacterial community is dominated by the genus Endozoicomonas ( 58 ). A range of processes that includes host evolution ( 6 ), traits ( 7 , 8 , 10 ), microniche partitioning ( 23 ), priority effects, and functional redundancy ( 59 , 60 ) may synergistically affect the microbiome assembly and ultimately determine the variability of corals’ bacterial communities. TABLE 1 Percentages of rare and core ASVs present in the tissue and skeleton of each coral species Coral species and compartment Rare ASVs (%) Core ASVs (%) Porites lutea tissue 94.7 5.3 Porites lutea skeleton 72.2 27.8 Paragoniastrea australensis tissue 68.5 9.2 Paragoniastrea australensis skeleton 71.6 28.4 Acropora aspera tissue 77.2 22.8 Acropora aspera skeleton 72.6 27.4 Montipora digitata tissue 80.0 20.0 Montipora digitata skeleton 79.9 20.1 Pocillopora damicornis tissue 80.8 19.2 Pocillopora damicornis skeleton 83.1 16.9 Porites annae tissue 74.3 25.7 Porites annae skeleton 82.2 17.8 Goniastrea retiformis tissue 80.7 19.3 Goniastrea retiformis skeleton 71.2 28.8 Stylophora pistillata tissue 82.4 17.6 Stylophora pistillata skeleton 79.3 20.7 Goniopora tenuidens tissue 84.4 15.6 Goniopora tenuidens skeleton 79.0 21.0 Platygyra sinensis tissue 85.8 14.2 Platygyra sinensis skeleton 65.5 34.5 Platygyra daedalea tissue 80.1 19.9 Platygyra daedalea skeleton 81.5 18.5 Isopora palifera tissue 91.5 8.5 Isopora palifera skeleton 83.8 16.2 10.1128/msystems.00044-22.5 TABLE S5 Rare and core microbiome of the tissue and skeleton of each coral species. The table shows the rare and core ASVs, calculated as ASVs that were not and were consistently present in at least 30% of the tissue and skeletal samples of each coral species. The table also shows the number of sequences affiliated with each rare and core ASV. Download Table S5, XLSX file, 1.0 MB . © Crown copyright 2022. 2022 Crown https://creativecommons.org/licenses/by/4.0/ This content is distributed under the terms of the Creative Commons Attribution 4.0 International license . 10.1128/msystems.00044-22.9 FIG S3 Box plots showing Pielou’s evenness index α-diversity metrics of the bacterial communities of the skeleton (red) and tissue (blue) of scleractinian corals (a) and of each coral species (b and c). Within each plot are reported the P values calculated through the Mann-Whitney U test (a), analysis of variance (b), and the Kruskal-Wallis test (c). Download FIG S3, TIF file, 2.1 MB . © Crown copyright 2022. 2022 Crown https://creativecommons.org/licenses/by/4.0/ This content is distributed under the terms of the Creative Commons Attribution 4.0 International license . Seawater and sediment as potential microbial reservoirs for the coral microbiome. We compared the coral tissue and skeleton microbiomes with those of seawater and sediment ( Fig. 1 ). Although our results suggest that the β-diversity of coral tissue and skeletal microbiomes differed from those of seawater and sediment ( Fig. S4 ), a component of the tissue and skeletal microbiome overlapped with that of the environment ( Table 2 ). Our comparison of the coral microbiota with those of the surrounding seawater and sediment showed that bacterial species shared between corals and their environment at the time of sampling accounted for more than 30% of the ASVs in A. aspera and P. damicornis , and the values for the other species were in the range of 10.5 to 27.5% ( Table 2 ). Processes such as microbial dispersal across space and environmental heterogeneity can influence the structure of host-microbial systems ( 61 ), including those of corals ( 32 ). Accordingly, our results suggest that the bacteria present in the surrounding environment may function as potential environmental reservoirs for corals and, given that our seawater and sediment samples were taken over a period of 1 month, it seems likely that a more prolonged sampling would recover a higher fraction of the coral microbiome. The seasonally variable microbiome present in the environment ( 62 , 63 ) can offer a large pool of important bacterial symbionts to be sourced by the coral holobiont through processes like selection and winnowing ( 9 , 64 ). TABLE 2 Percentages of ASVs retrieved from seawater and sediment that were concurrently present in each coral species Coral species Seawater ASVs (%) Sediment ASVs (%) \n Acropora aspera \n 19.5 11.9 \n Goniastrea retiformis \n 6.1 8.4 \n Goniopora tenuidens \n 8.2 9.6 \n Isopora palifera \n 14.4 10.9 \n Montipora digitata \n 15.7 10.3 \n Paragoniastrea australensis \n 4.3 6.2 \n Platygyra daedalea \n 5.5 8.3 \n Platygyra sinensis \n 4.3 6.2 \n Pocillopora damicornis \n 19.1 13.7 \n Porites annae \n 12.1 9.8 \n Porites lutea \n 8.1 10.5 \n Stylophora pistillata \n 17.0 10.5 10.1128/msystems.00044-22.10 FIG S4 Biplot PCAs represent β-diversity metrics of each coral species, seawater, and sediment. Within each biplot are also reported the results of nonparametric multivariate analysis of variance (NPMANOVA) analyses assessing differences in composition between the bacterial communities of skeleton, tissue, seawater, and sediment. Download FIG S4, TIF file, 2.4 MB . © Crown copyright 2022. 2022 Crown https://creativecommons.org/licenses/by/4.0/ This content is distributed under the terms of the Creative Commons Attribution 4.0 International license . Unusual suspects show persistent associations with corals. Our work identified several bacteria that were consistently associated with corals but are either new to the coral microbiome field or understudied. Cyclobacteriaceae taxa accounted for 7.30% of the ASVs of the data set and were present in several samples of every coral species ( Table S6 ). Currently, Cyclobacteriaceae have only been reported in fire coral colonies ( 36 ), in 7-month-old Acropora recruits ( 65 ), and in intracolony changes in abundance after bleaching of Acropora spp. ( 66 ). Our work shows that this family is much more widespread across the complex and robust clades. From a functional point of view, members of this family could benefit the coral through carbohydrate metabolism, carotenoid biosynthesis, antibiotic resistance, and quorum-sensing regulation ( 67 ). 10.1128/msystems.00044-22.6 TABLE S6 Relative abundance table of bacteria identified as understudied or unknown to the coral reef microbial ecology. The table also shows in which coral species and sample the ASVs associated with these bacteria were retrieved. Download Table S6, XLSX file, 0.01 MB . © Crown copyright 2022. 2022 Crown https://creativecommons.org/licenses/by/4.0/ This content is distributed under the terms of the Creative Commons Attribution 4.0 International license . Bacteria in the genus Paramaledivibacter accounted for 2.92% of the data set ASVs and were found in multiple coral species ( Table S6 ). This genus of strictly anaerobic bacteria was only reported recently in the coral literature ( 68 ). Information about these bacteria is still scarce, but given their ability to degrade amino acids and peptides, they could exploit the resources of the host ( 69 ). As mentioned above, we hypothesize that detrimental effects by pathogens could be counterbalanced by beneficial holobiont members, including Myxococcales and Pseudoalteromonas . Despite Paramaledivibacter being described as strictly anaerobic, we did find these bacteria in the coral tissue. This finding may seem unusual, but this is not the first study reporting obligate anaerobes in coral compartments known to be oxic ( 70 , 71 ). Roseospira has also not been previously reported in the coral literature, but ASVs affiliated with these bacteria accounted for 0.29% of the data set and were found in the species P. sinensis , P. daedalea , P. lutea , Paragoniastrea australensis , G. retiformis , and Goniopora tenuidens ( Table S6 ). These purple nonsulfur bacteria seem to be able to colonize a diverse range of environments and grow optimally under photoheterotrophic conditions ( 72 ). Thus, given their ability to utilize substrates known to be present in corals, such as acetate ( 73 ) and glutamate ( 74 ), and to use near-infrared wavelengths not absorbed by Symbiodiniaceae species, the coral colony could offer an array of microenvironments where these bacterium’s niche preferences are met. Conclusions. Through a combination of a homogeneous experimental design that minimizes external biases affecting the microbiome; the use of innovative technologies, including micro-CT scanning to evaluate host traits; and the application of a range of statistical analyses, our study allowed us to unravel the structure of the coral microbiome and quantify how it is influenced by host phylogeny, skeletal architecture, and reproductive mode. We show that host phylogeny and traits explained 14% of the tissue and 13% of the skeletal microbiome composition and were associated with a range of microbial partners thought to affect holobiont health and functioning. Based on our results, we hypothesize that reproductive mode may influence the microbiome composition in a predictable manner, while skeletal architecture works like a filter affecting bacterial relative abundance. Although our analysis accounted for some of the most influential processes known to affect the microbiome composition, these could only marginally explain the microbiome variation of tissue and skeleton. A holistic view of the mechanisms determining the holobiont composition will be gained by assessing the influence of the physicochemical and dynamic biochemical environment of the coral colony on the microbiome composition and by assessing whether the presence of some bacteria (whether dominant or rare) may affect the overall structure of the microbiome. In this study, we provided substantial evidence that coral tissue and skeleton microbiomes are dominated by rare taxa and differ in bacterial abundance, but a consortium of bacteria can colonize both compartments and the skeleton can function as a microbial reservoir. While our study answers several unsolved questions about the bacterial community structure of scleractinian corals and the mechanisms driving its composition, it also exposes knowledge gaps. Despite our focus on commonly studied coral species, we identified three bacterial groups that are understudied or were not previously reported in the coral literature. This highlights that a full characterization of the taxonomic composition of the coral microbiome has still not been achieved. Substantial further work will be needed to fully understand its functions in the coral holobiont, its fine-scale distribution in relation to ecological microniches, and the metabolic hand-offs that happen among microbiome members and with the host. The use of putative beneficial microorganisms has been proposed as a tool to mitigate the increasing pressure of anthropogenic activities on coral reefs ( 75 ); therefore, we hope that the detailed knowledge gained from our study about persistent association between specific coral lineages and bacterial taxa can form the basis for further advances in probiotic strategies to improve coral resilience in future climate scenarios."
} | 11,448 |
38306426 | PMC10836727 | pmc | 819 | {
"abstract": "Conductive hydrogels have a remarkable potential for applications in soft electronics and robotics, owing to their noteworthy attributes, including electrical conductivity, stretchability, biocompatibility, etc. However, the limited strength and toughness of these hydrogels have traditionally impeded their practical implementation. Inspired by the hierarchical architecture of high-performance biological composites found in nature, we successfully fabricate a robust and sensitive conductive nanocomposite hydrogel through self-assembly–induced bridge cross-linking of MgB 2 nanosheets and polyvinyl alcohol hydrogels. By combining the hierarchical lamellar microstructure with robust molecular B─O─C covalent bonds, the resulting conductive hydrogel exhibits an exceptional strength and toughness. Moreover, the hydrogel demonstrates exceptional sensitivity (response/relaxation time, 20 milliseconds; detection lower limit, ~1 Pascal) under external deformation. Such characteristics enable the conductive hydrogel to exhibit superior performance in soft sensing applications. This study introduces a high-performance conductive hydrogel and opens up exciting possibilities for the development of soft electronics.",
"introduction": "INTRODUCTION Hydrogels have immense potential for various applications in biomedicine ( 1 – 3 ), energy storage ( 4 ), soft robotics ( 5 , 6 ), and flexible electronics ( 7 , 8 ). Nevertheless, a notable obstacle to their practical utilization is their limited strength (less than 1 MPa) ( 9 , 10 ). Moreover, achieving an optimal combination of characteristics, such as high sensitivity, self-healing capability, and mechanical robustness, is crucial for the successful implementation of soft hydrogels in practical applications ( 11 , 12 ). The development of conductive hydrogels typically entails the integration of conductive components into the hydrogel matrix. These conductive phases encompass a variety of types such as nanomaterials [e.g., graphene ( 13 ), carbon nanotube ( 14 ), MXene ( 15 – 17 ), and Ag nanoparticles ( 18 )], conductive ions [e.g., metal ions ( 9 ) and ionic liquid ( 19 )], and conductive polymers [e.g., poly(3,4-ethylenedioxythiophene) polystyrene sulfonate (PEDOT:PSS) ( 20 ), polyaniline (PANI) ( 21 ), and polypyrrole (PPy) ( 22 )]. It has brought to attention that often weak molecular bonds between the conductive phases and the hydrogel matrices, along with disordered structural distribution within the matrix, can adversely affect the mechanical properties of conductive hydrogels. Biological soft tissues, such as tendon and cartilage, exhibit an exceptional combination of strength, flexibility, and message-passing ability, although they are composites based on simple compositions comprising biopolymers and minerals. For instance, the hierarchical organization of tendon endows them with high strength and flexibility, allowing them to endure mechanical stress and execute contractions and relaxations. Moreover, the interplay between message-passing proteins and other constituents within soft tissues contributes to their overall strength and sensing capabilities ( 23 , 24 ). Inspired by the remarkable properties of bio-soft tissues, researchers have explored various structural and molecular engineering technologies to enhance the mechanical properties of hydrogels. Structural approaches, including freeze casting to create hierarchical structures ( 17 , 25 ), stretch training to align hydrogel fibers ( 26 ), and salting out to induce strong chain aggregation ( 27 , 28 ), have been extensively investigated. Molecular engineering approaches, including the formation of covalent bonds ( 29 ), hybrid networks ( 30 ), and nanocrystalline domains through annealing ( 31 ), have shown great potential in enhancing the mechanical strength of hydrogels. However, most of these engineering methods are “top-down” approaches that have limitations in creating strong bonding between hierarchical units after gelation ( 32 ). Biological soft tissues are formed through the intricate process of cell division, which can be characterized as a “bottom-up” approach. A similar approach is observed in self-assembly, where physical driving forces are used to construct materials with ordered structures ( 33 , 34 ). During self-assembly, the basic units gradually aggregate and become bulk materials with strong interface bonding. For instance, during the evaporation process, two-dimensional (2D) nanosheets in a solution effortlessly align under the influence of unidirectional solvent evaporation forces. Simultaneously, dehydration reactions occur between the hydroxyl groups of soluble hydroxyl-containing polymer molecules, leading to layer-by-layer crystallization of polymer molecules ( 35 – 37 ). To combine both structural and molecular engineering strategies, we incorporate 2D MgB 2 nanosheets as the electrical conductive phases within polyvinyl alcohol (PVA) hydrogel matrix and create them into nanocomposite hydrogels through a simple self-assembly–induced boron bridging method. Structurally, the evaporation approach facilitates the layer-by-layer self-assembly of PVA and MgB 2 nanosheet layers into a lamellar microstructure. Meanwhile, MgB 2 nanosheets can serve as high-functionality cross-links, forming a large number of strong B─O─C covalent bonds with PVA nanolayers for mechanical strengthening ( 38 , 39 ). Furthermore, MgB 2 nanosheets could provide conducting ions and electrons for enhanced electrical properties ( 40 ). Through this approach, we have successfully created hierarchical lamellar MgB 2 nanosheet (HLMN)–bridging PVA hydrogels, exhibiting impressive tensile strength (8.58 to 32.7 MPa) and toughness (27.56 to 123.3 MJ/m 3 ). In addition, these nanocomposite hydrogels demonstrate high sensitivity to deformation pressure, characterized by a remarkable response/relaxation time of 20 ms and a detection lower limit as low as 1 Pa. These remarkable properties make them suitable for a wide range of applications across various fields.",
"discussion": "RESULTS AND DISCUSSION The fabrication process of the HLMN-hydrogel begins with the synthesis of 2D MgB 2 nanosheets using a chemical exfoliation method based on chelation reactions. Figure 1A shows the crystal structure of MgB 2 , highlighting its layered characteristics consisting of distinct and alternating Mg and B layers. This layered configuration is further confirmed by the scanning electron microscopy (SEM) image (fig. S1) of a micro-MgB 2 particle, which reveals a plate-like structure with a pronounced length-to-thickness ratio. This layered structure of MgB 2 presents the potential for its exfoliation into nano-sized, 2D nanosheets. However, the crystal bonding within MgB 2 is complex and strong. Specifically, MgB 2 exhibits an AlB 2 -type crystal structure, consisting of closely packed Mg layers and graphene-like B layers bonded by a combination of ionic interlayer bonding and a mixture of covalent and metallic bonding ( 41 ). To overcome the formidable crystal bonding and successfully exfoliate MgB 2 nanosheets, we use chelation reactions, through which a metal ion forms a strong complex with an organic ligand, thereby extracting the metal ion from its original position ( 42 ). EDTA, chosen as the chelating agent, has four acidic protons enabling the formation of strong Mg-EDTA complexes, as displayed in fig. S2. When MgB 2 nanoparticles react with EDTA, EDTA firmly engages with the Mg, extracting them from their native positions within the crystal structure. This interaction leads to a Mg-deficient layered structure, which weakens the interlayer bonding within the MgB 2 crystals and facilitates their cleavage into MgB 2 nanosheets under the influence of ultrasound vibrations. The resultant exfoliated MgB 2 nanosheets exhibit a thin-plate microstructure approximately 500 nm in length and 20 nm thick, as confirmed by the transmission electron microscope (TEM) image and atomic force microscopy (AFM) data in Fig. 1 (B, C, and F, respectively). This result verifies the successful exfoliation into 2D MgB 2 nanosheets. Furthermore, Brunauer-Emmett-Teller (BET) surface area analysis measurements reveal a significant increase in surface area from 6.904 m 2 /g for the MgB 2 nanoparticles to 31.580 m 2 /g for the MgB 2 nanosheets ( Fig. 1E ). Energy-dispersive spectroscopy analysis under TEM (TEM-EDS) confirms the composition of Mg and B elements in the MgB 2 nanosheets ( Fig. 1D ). X-ray photoelectron spectroscopy (XPS) is used to demonstrate the reduction of the Mg element in the MgB 2 nanosheets ( Fig. 1E ). A clear distinction is observed in the B1s XPS spectra between the MgB 2 nanoparticles and nanosheets. In the MgB 2 nanosheets, the B1s peak splits into three distinct peaks at approximately 187, 187.7, and 188.4 eV ( Fig. 1G ). The broader peak observed at 188.4 eV in the MgB 2 nanosheets can be attributed to an increased presence of B─OH species ( 43 ). With this change in dimension and increased boron composition, these MgB 2 nanosheets show great promise for enhancing the mechanical and electrical performances of conductive PVA hydrogels. Fig. 1. Fabrication and characterizations of the MgB 2 nanosheets. ( A ) Schematic illustration of the exfoliation process for MgB 2 nanosheets. ( B and C ) TEM and HRTEM images of exfoliated MgB 2 nanosheets. ( D ) TEM-EDS of the MgB 2 nanosheets. ( E ) Surface area of the MgB 2 nanosheets. ( F ) AFM image of the nanosheets. ( G ) XPS of MgB 2 nanosheets. The HLMN-hydrogel was then prepared by a combination of evaporation-driven self-assembly and bridging–cross-linking approach, as shown in Fig. 2A . This process starts by blending 0.16 wt % MgB 2 nanosheets with 4 wt % PVA aqueous solutions, which is the highest ratio of MgB 2 nanosheets that can be added without affecting the fluidity of the mixture. As shown in fig. S4, when the PVA content is increased to 8 wt % or the MgB 2 nanosheets content is raised to 0.24 wt %, the mixture undergoes a swift transition from a liquid state to a gel-like consistency, thereby hindering the self-assembly process. This change also indicates an effective boronate-ester bonding creation between the B─OH species of the MgB 2 nanosheets and the hydroxyl groups (-OH) of the PVA. This flowable mixture then undergoes an evaporation process in an ambient environment to form either a thin film or a bulk shape, depending on the duration of evaporation. During the evaporation process, dehydration reactions occur among the hydroxyl groups of PVA, leading to the layer-by-layer assembly of PVA micro/nanolayers bonded by hydrogen bonds. Meanwhile, leveraging their high aspect ratio, the MgB 2 nanosheets naturally align with their flat surfaces parallel to the PVA layers, facilitating the formation of boronate-ester bonds in between. This process forms the initial cross-linking of the composite system assisted by the hydrogen bonds and boronate-ester bonds. Subsequently, the sample undergoes a low-temperature thermal annealing, which largely increases the number of PVA nanocrystalline domains within the PVA layers and the formation of boronate-ester bonds bridging the MgB 2 nanosheets with the PVA layers, as illustrated in Fig. 1A . Last, the annealed sample is swelled in water until equilibrium to achieve the HLMN-hydrogel, which consists of ~93% water content, as confirmed in fig. S5. Fig. 2. Fabrication and mechanical performance of the HLMN-hydrogel. ( A ) The hierarchical design strategy ranging from molecular to macro scale via self-assembly and boron bridging. ( B ) Morphology of the HLMN-hydrogel. SEM images revealing the lamellar micro/nanostructure of the HLMN-hydrogel. ( C ) Strength and toughness of the PVA hydrogel, MP-hydrogel, and HLMN-hydrogel. ( D ) Engineering stress-strain curves of the four kinds of hydrogels. ( E ) Cyclic loading-unloading of one HLMN-hydrogel and its hysteresis performance. ( F ) A plot of the toughness versus strength of HLMN-hydrogel and other tough hydrogels. The data are summarized in table S1. The data reported in (B) represent the mean value, with the error given as the standard deviation. The sample size is 5 for all of the cases. For comparison, we also prepared pure PVA hydrogels (PVA-hydrogel) and MgB 2 nanoparticle–doped PVA hydrogels (MP-hydrogel) using similar evaporation-assisted self-assembly and thermal annealing methods. The microstructures of these comparing samples are examined in fig. S6. The PVA hydrogels display a well-ordered lamellar microstructure due to the evaporation-driven self-assembly effect. However, the MP-hydrogels exhibit a less orderly structure due to the spherical shape of the MgB 2 nanoparticles (fig. S7) that disrupts the orderly formation of PVA layers during evaporation. The HLMN-hydrogels, however, showcase a perfectly aligned lamellar microstructure, owing to the 2D platelet–like shape of the MgB 2 nanosheets. These nanosheets, with their high aspect ratio and nanometer thickness, fit perfectly between the PVA layers. The formation of boronate-ester bonds further strengthens the connections between the PVA layers and the MgB 2 nanosheets, promoting a refined layer-by-layer assembly of the materials during the evaporation process. The SEM images ( Fig. 2B ) highlight the hierarchical lamellar microstructure of the HLMN-hydrogel, with nanoscale alternating PVA layers and MgB 2 nanosheet layers. Specifically, the nanocomposite hydrogel consists of the following hierarchical lamellar structures and multiple molecular bonds: (i) Through the evaporation process, PVA chains assemble into nano- and micro-scale lamellar layers with MgB 2 nanosheets embedded in-between; (ii) the PVA layers, interconnected via hydrogen bonds, comprise numerous nanocrystalline domains formed during thermal annealing; (iii) MgB 2 nanosheets serve as high-functionality cross-links, producing an abundance of boronate-ester bonds through the interaction of B─OH species on the MgB 2 nanosheets and the -OH groups of the PVA layers. Uniaxial tensile tests were conducted on all samples, with results presented in Fig. 2C . The HLMN-hydrogel, despite consisting of a high-water content of ~93%, displayed substantially superior mechanical properties compared to the other groups. Its tensile strength (8.58 ± 1.34 MPa) was 11 times higher than that of the PVA-hydrogel (0.78 ± 0.04 MPa) and 2.7 times that of the MP-hydrogel (3.17 ± 0.56 MPa) ( Fig. 2, C and D ). The toughness of the HLMN-hydrogel, measured as the area under the stress-strain curve, further showcased its superiority. The value for the HLMN-hydrogel (27.56 ± 7.48 MJ/m 3 ) was nearly 20 times larger than the PVA-hydrogel (1.37 ± 0.21 MJ/m 3 ) and 3.7 times larger than the MP-hydrogel (7.45 ± 1.88 MJ/m 3 ). By modulating the water contents, the mechanical performance of the HLMN-hydrogels could be further enhanced (fig. S10). The mechanical properties of the HLMN-hydrogel exceeded that of most hydrogels ( Fig. 2F and table S1). Furthermore, the HLMN-hydrogel displayed a great shape recovery capability and decent hysteresis performance ( Fig. 2E ). These results demonstrate the advantages of our facile assembly approach for fabricating strong and tough hydrogels. The remarkable strength and toughness of HLMN-hydrogel, capable of supporting a 2 kg load with a thin strand (0.3 mm by 3 mm), can be attributed to the synergistic energy dissipation mechanisms from their hierarchical lamellar structures and multiple molecular bonds, as illustrated in Fig. 3A . On a micro- and nanoscale, the well aligned MgB 2 nanosheets embedded within the PVA layers ( Fig. 3A ) impede crack propagation via fracture deflection. Simultaneously, the pullout of MgB 2 nanosheets from the PVA layers dissipates additional mechanical energy and enhances the overall deformation capability, allowing the HLMN-hydrogel to stretch more than 10-fold its original length. This is over three times the stretchability of pure PVA-hydrogels, which can only reach thrice their original length before fracturing ( Fig. 2D ). SEM images of cross-sectional fractures corroborate this, revealing a deflected fracture surface across multiple layers and the identifiable pullout of individual MgB 2 nanosheets ( Fig. 3, C and D ). At the nanoscale, the PVA layers were reinforced by thermal annealing to form abundant nanocrystalline domains, which consist of highly aggregated and intertwined PVA chains with strong molecular bonding. The fracture of these highly crystallized PVA layers lead to significantly energy dissipation. Differential scanning calorimetry (DSC) analysis ( Fig. 3F ) reveals a broad curve for annealed HLMN-hydrogels with higher intensity compared to unannealed ones. This suggests the formation of more crystalline domains in the annealed HLMN-hydrogels. The highly crystallized anisotropic lamellar microstructure of the HLMN-hydrogel can also be identified by the flattened circular and elliptical-like pattern from the small-angle x-ray scattering (SAXS) tests ( Fig. 3E ). An enormous number of boronate-ester bonds were formed to bridge the MgB 2 nanosheets and the PVA layers at molecular scale, augmenting the strength and toughness of the hydrogels as high-functionality cross-links. The existence of B─OH species on the surface of the MgB 2 nanosheets was confirmed from the XPS results in Fig. 1G . The appearance of the B─O─C bonds (1125 cm −1 ) and O─B─O bonds (710 cm −1 ) ( 38 , 44 , 45 ) in the Fourier transform infrared spectroscopy (FTIR) results of HLMN-hydrogels confirmed the creation of strong covalent boronate-ester (B─O─C) bonds in between the MgB 2 nanosheets and PVA layers ( Fig. 3G ). During stretching, debonding of these B─O─C covalent bonds consumes a significant amount of energy that leads to high strength and toughness of the hydrogels. In summary, the HLMN-hydrogel’s outstanding mechanical performance is a result of a synergic effect of multiple energy dissipation mechanisms, including crack deflection by the MgB 2 nanosheets, the pulling-out of the MgB 2 nanosheets from the PVA layers, the fracture of the highly crystallized PVA layers, and the debonding of the strong B─O─C covalent bonds in between the MgB 2 nanosheets and PVA layers. Fig. 3. Toughening mechanisms of hierarchical lamellar design and boron bridging cross-linking. ( A ) Hierarchical strengthening mechanism of the HLMN-hydrogel during fracture. ( B ) Bridging mechanism between MgB 2 nanosheets and PVA chains in the molecular scale (green circle, B─O─C bond; blue circle, H bond; yellow circle, PVA bonding). ( C ) Cross-sectional microstructure of the HLMN-hydrogel. Yellow arrows indicated the fracture of hierarchical layers. ( D ) Pull-out of nanosheets (green arrows) and PVA nanolayers (yellow arrows) from the hydrogel. ( E ) SAXS patterns of the HLMN-hydrogel. ( F ) Crystalline domains analysis via DSC. ( G ) FTIR of three kinds of hydrogels. The HLMN-hydrogel not only exhibits outstanding mechanical performance but also exceptional response sensitivity and low detection limit. Specifically, the pressure sensor can respond to a slight pressure of ~1 Pa, equivalent to a small flower weighing 56 mg on an area of 5 cm 2 ( Fig. 4A ). The response/relaxation time of HLMN-hydrogel is remarkably low at ~20 ms, even faster than that of human skin (≈30 to 50 ms) ( 46 ) and electronic skin (table S2). In addition, the gauge factor (GF) of the HLMN-hydrogel increases from 0 to 4.43 as the stretch strain increases from 0 to 800% ( Fig. 4B ). In contrast, control studies of the MP-hydrogel showed a poorer response time, relaxation time, and GF compared to the HLMN-hydrogel (fig. S13). The electrical stability of the HLMN-hydrogel is retained after 1500 tensile cycles, indicating its potential as a robust mechanical sensor ( Fig. 4C ). Fig. 4. Sensing capabilities and applications of the conductive HLMN-hydrogel. ( A ) Photo image and electrical response of a conductive HLMN-hydrogel sensor weighted by a small flower. The pressure applied on the hydrogel by the flower is ≈1 Pa. The response/relaxation time is 20 ms. ( B ) GF of the HLMN-hydrogel. ( C ) Cycle stability of the HLMN-hydrogel. ( D ) Noncontact speaking detection using the HLMN-hydrogel. The detection device showed different and unique signals when they “sounded” different words. ( E ) Schematic illustration of the mechanism of the outstanding detection capability. ( F ) Electronic-glove fabrication using HLMN-hydrogel. Photos showing different hand gestures with E-glove and the associated signal responses from five fingers. The HLMN-hydrogel displays electrical conductivity by facilitating the movement of both electrons (through MgB 2 nanosheets) and ions (such as Mg 2+ and B species). The presence of conducting ions was confirmed through inductively coupled plasma emission spectrometer (ICP) analysis of the hydrogels (fig. S15). In a temperature-resistance experiment (fig. S17), the resistance of the hydrogel decreased with the increase in temperature, indicating an increased movement of ions due to the elevated temperature, and further validating the existence of the ionic conducting mechanism of the HLMN-hydrogel. We further conducted a comprehensive electrochemistry study to analyze the conductive format and calculate conductivity. From the results shown in fig. S18, it is evident that the ion resistance measures at 9.93 ohms, while the electron resistance is notably high. This observation indicates the dominance of ion resistance, which aligns with expectations, considering the coverage of nanosheets by the PVA, affecting their electronic conductivity. Further insight from the Bode plot in fig. S18C reveals an exceptionally low deviation frequency for ionic conduction in the material. These outcomes reaffirm that ionic transport is the predominant factor in the hydrogel’s behavior. Our calculations yielded a material conductivity of 0.13 S/m. We compared this conductivity with that of several other conductive hydrogels and composites in table S3 and observed that its value is not exceptionally high. However, it is essential to emphasize that our objective does not prioritize achieving high conductivity. This is because high conductivity alone does not necessarily correlate with improved sensing performance. The exceptional sensing ability of the HLMN-hydrogel is due to its ordered nanoscale lamellar conductive PVA-MgB 2 layers and extremely low compressive modulus (1.86 ± 0.10 kPa), as shown in fig. S19. The HLMN-hydrogel’s low compressive modulus allows for noticeable nanoscale deformation even under minimal force. This leads to a reduction in layer gaps that accelerates the transportation of Mg 2+ , B species ions, and electrons for enhanced conductivity ( Fig. 4E ). Significantly, the nanoscale lamellar microstructures lead to a significant increase in contact areas between the multiple conductive PVA-MgB 2 nanolayers, creating a more efficient conducting path for electrons and ions. This change in contact area is particularly noticeable and can persist until the pressure is removed ( Fig. 4A ). Notably and interestingly, our HLMN-hydrogel demonstrated remarkable noncontact speaking sensing ability due to its exceptional response sensitivity and low detection limit ( Fig. 4D and movie S1). The HLMN-hydrogel produced stable and distinguishable signals when receiving spoken words such as “NUS,” “mechanical,” and “engineering.” The signal for mechanical engineering represented the combined signal of “mechanical” and “engineering,” demonstrating the accuracy and stability of the HLMN-hydrogel ( Fig. 4D ). In addition, the HLMN-hydrogel also exhibited great underwater speaking sensing capability, as depicted in fig. S21. We spoke the words “Mechanical” and “Engineering” to the hydrogel while it was immersed in water. Each word was repeated twice to confirm its accuracy. The signal corresponding to “Mechanical” is presented in fig. S21B. After applying Fourier transformation, it becomes evident that the signals for the two instances of “Mechanical” exhibit similar characteristic frequencies, which distinctly differ from the signals for “Engineering” (fig. S21, C, D, F, and G). This underscores the hydrogel’s exceptional sensing capacity even when submerged in water. Moreover, by using the five output channels, the HLMN-hydrogel can be fabricated into an electronic glove to distinguish different unique signs, including “stop,” “great,” “victory,” and “ok” ( Fig. 4F ). In addition, the HLMN-hydrogel can act as a handwriting-sensing device, detecting distinguishable and highly repeatable waveforms for the words “ME” and “OK” (fig. S23). These experiments demonstrate the superior sensing capabilities and versatile applications of the HLMN-hydrogel. The HLMN-hydrogel and its device, with outstanding sensing capability, displayed tremendous potential for consumer electronics, including speaking detection, virtual recreational gaming, sign language translation, remote control of the surgical robot, rehabilitation tool for patients with hand disease, etc. For noncontact speaking detection, we displayed a simple application example. The signals of verbal instructions, such as “blue,” “red,” “yellow,” “green,” and “turn off” were received by the HLMN-hydrogel. The collected quantities of signals were used for the training of a neural network algorithm self-built using Python (fig. S24). Afterward, the speaking and commands transfer was achieved by a microcontroller ( Fig. 5A ). Different verbal instructions of “blue,” “red,” “yellow,” “green,” and “turn off” exhibited accurate corresponding commands of turning up the blue, red, yellow, or green light-emitting diode (LED) lights and turning off the lights ( Fig. 5B and movie S2). In addition, Fig. 5C showcased another application formats. A robotic car executed the instructions to pick up and transport the “tomato,” “orange,” “plum,” and “grape,” as the hydrogel received the spoken commands for “tomato,” “orange,” “plum,” and “grape” (movie S3). Accordingly, it is apparent that the fabricated hydrogels with the noncontact speaking sensing capabilities can realize the ability to “listen” and to “understand” by themselves. The results showed that HLMN-hydrogel has high sensitivity, accuracy, stability, and great potential as sensors and consumer electronics. Fig. 5. Speaking and commands recognition of the conductive HLMN-hydrogel. ( A ) Scheme showing the speaking and commands recognition procedure. ( B ) The LED light performed the commands of turning up blue, red, yellow, green light or turning off the light, when hydrogel received the verbal instructions of “blue,” “red,” “yellow,” “green,” and “turn off.” ( C ) Robot performed the commands of pick up and deliver the tomato, orange, plum and grape, when hydrogel received the verbal instructions of “tomato,” “orange,” “plum,” and “grape.” Furthermore, in anticipation of future applications of this sensing hydrogel in the biomedical field, we carried out an extensive assessment of the biocompatibility of the hydrogel, encompassing both in vitro and in vivo studies. To initially assess the biocompatibility of the hydrogel, we performed a hemolysis assay. As depicted in Fig. 6 (A and B) , the suspension of red blood cells (BCs) treated with the hydrogel appeared colorless and clear, with an almost negligible hemolysis ratio when compared to Triton X-100. This observation signifies the excellent blood compatibility and overall biosafety of the hydrogel. Next, we cultured rat skin–derived 1 (RS1) fibroblast cells with the hydrogel for 48 hours, followed by a live/dead staining procedure to assess the cells’ growth status ( Fig. 6C ). Our observations revealed robust cell growth and almost no cell death even after 48 hours of culture with the hydrogel. Furthermore, we performed a quantitative analysis using the cell counting kit-8 (CCK-8) assay on the RS1 cells after 48 and 72 hours of culture with the hydrogel ( Fig. 6D ). The results demonstrated a consistent increase in cell viability over time, and the proliferation capacity of the hydrogel sample closely resembled that of the blank culture plate. These findings underscore the excellent cell biocompatibility of the hydrogel. Last, to assess the in vivo biosafety of the hydrogel, it was surgically implanted into bilateral pockets within the perivertebral fascia lumbodorsalis of Sprague-Dawley (SD) rats under anesthesia. After 2 weeks, the optical images revealed no signs of tissue swelling or inflammation in the implanted area ( Fig. 6E ). Histological analysis was performed on subcutaneous tissue and vital organs, including the heart, liver, spleen, lung, and kidney, using hematoxylin and eosin (H&E) staining ( Fig. 6, E and F ). No discernible differences were observed between the tissues of the blank and hydrogel samples. Furthermore, there were no notable abnormalities in cell morphology and no signs of neutrophil or lymphocyte infiltration or the presence of necrotic cells in these tissues. These findings indicate the good biocompatibility and safety of the HLMN-hydrogel. Fig. 6. Biocompatibility study of the HLMN-hydrogel. ( A ) Hemolytic test to assess the blood compatibility of the hydrogel in vitro. ( B ) Quantitative analysis of the hemolysis ratios ( n = 5, *** P < 0.001). ( C ) Culture of RS1 cells with hydrogel to assess cell biocompatibility using a live/dead dye assay. ( D ) Quantitative analysis the cell growth after 48 and 72 hours of culture ( n ≥ 3, * P < 0.05). ( E ) Implant the hydrogel into bilateral pockets within the perivertebral fascia lumbodorsalis to assess the in vivo biocompatibility of the hydrogel. Optical image and histological analysis of subcutaneous tissue following 2-week hydrogel implantation with H&E staining. ( F ) Histological analyses of the rat organs (heart, liver, spleen, lung, and kidney) with hydrogel implantation and without. If MgB 2 nanosheets were to be replaced with a different type of nanosheet while maintaining the same assembly method, the nanosheets and PVA nanolayers would still be present based on the fabrication mechanism analysis. However, the crucial molecular-level interaction between the nanosheets and the polymer matrix would cease to exist, resulting in a decline in mechanical performance. What sets MgB 2 nanosheets apart from other nanosheets is their ability to establish cross-links with the PVA matrix through the formation of B─O─C bonds, thereby fortifying and toughening the matrix. Typically, such cross-linking processes happen rapidly, resulting in PVA gelation, which can affect the assembly and alignment of nanosheets within the polymer matrix. In our study, we harnessed the self-assembly–induced bridge cross-linking approach, allowing the combination of MgB 2 nanosheets with PVA and the alignment of MgB 2 nanosheets within PVA layers to coincide with water evaporation. This method overcomes the limitations associated with traditional “top-down” engineering techniques, which often face challenges in establishing strong bonds between hierarchical units after the gelation process. Regarding sensitivity, if MgB 2 nanosheets were replaced with other types of nanosheets, it is very likely that the hydrogel’s sensitivity would remain intact. This is because the presence of nanoscale layers, a key factor contributing to high sensitivity to external pressure, would persist. Therefore, fabricating a highly sensitive hydrogel using this method with different nanosheets remains a viable approach. However, achieving a hydrogel with the same level of strength and toughness as demonstrated in this study may not be feasible. Certainly, by modifying the nanosheets with specific functional groups and incorporating these modified nanosheets into the hydrogel system, it is possible to establish cross-links between the inorganic nanosheets and the polymer matrix. This approach can yield a conductive hydrogel with exceptional mechanical performance and sensitivity. This research sets the foundation for developing a series of robust, resilient, and sensitive conductive hydrogels through hierarchical structural design spanning from the micro to molecular scales. In summary, we have introduced an electrically conductive hydrogel, which uses MgB 2 nanosheets to bridge a unique hierarchical lamellar structure of PVA nanocomposite hydrogels. The resulting HLMN-hydrogel exhibits exceptional mechanical properties, including strength (8.58 to 32.7 MPa) and toughness (27.56 to 123.3 MJ/m 3 ), as well as remarkable sensing capabilities with a reaction time of 20 ms and a detection lower limit of ~1 Pa. By incorporating hierarchical lamellar layers and interface cross-linking bridges, the mechanical performance of HLMN-hydrogel exceeds that of recently reported tough hydrogels. Moreover, the HLMN-hydrogel demonstrates exceptional noncontact speaking sensing ability due to its unique nanoscale layered structure, enabling accurate and stable detection of verbal commands. These features make the HLMN-hydrogel an attractive material for flexible electronics, e-skins, soft robotics, energy, and biomedical applications. Overall, our hierarchical design strategy provides an effective approach for developing robust and functional hydrogels with advanced capabilities."
} | 8,366 |
35141504 | PMC8810406 | pmc | 820 | {
"abstract": "Summary The metabolic activity of microbial communities plays a primary role in the flow of essential nutrients throughout the biosphere. Molecular genetics has revealed the metabolic pathways that model organisms utilize to generate energy and biomass, but we understand little about how the metabolism of diverse, natural communities emerges from the collective action of its constituents. We propose that quantifying and mapping metabolic fluxes to sequencing measurements of genomic, taxonomic, or transcriptional variation across an ensemble of diverse communities, either in the laboratory or in the wild, can reveal low-dimensional descriptions of community structure that can explain or predict their emergent metabolic activity. We survey the types of communities for which this approach might be best suited, review the analytical techniques available for quantifying metabolite fluxes in communities, and discuss what types of data analysis approaches might be lucrative for learning the structure-function mapping in communities from these data.",
"introduction": "Introduction The structure-function problem in microbial communities The evolutionary history of the biosphere is inextricably linked to the metabolic activities of microbes. Since life arose on this planet, microbes have lived in consortia that saturated nearly every biochemical niche on the planet, driving global transformations in the chemical composition of the biosphere via metabolic processes from fermentation to photosynthesis to respiration ( Falkowski et al., 2000 ; Canfield et al., 2010 ; Nelson et al., 2016 ; Sunagawa et al., 2015 ; Zakem et al., 2020 ). As such, microbes and the communities in which they reside are the result of an ongoing eco-evolutionary process that couples the transformation of metabolites to the complex dynamics of interacting ecological systems across many spatial and temporal scales. Given the importance of the metabolic activity of microbial communities, we argue that a major goal for the field should be to predict, design, and control the metabolism of microbial communities in complex, natural, and engineered settings. Accomplishing this goal requires understanding how the structure of a community, in terms of the taxa present and its genomic composition, determines its metabolic activity in a given environmental context. The sequencing revolution has revealed the structure of microbial communities at the level of the taxa present, the genes they possess, and the dynamics of gene expression. This means that we now have a detailed and dynamic “parts list” for microbial communities in terms of taxonomic and genomic composition across a range of environments, from anaerobic digesters ( Bocher et al., 2015 ; Toerien and Hattingh, 1969 ; Vanwonterghem et al., 2014 ) to the human gut ( Blanton et al., 2016 ; Raman et al., 2019 ), soils ( Bahram et al., 2018 ), and the ocean ( Sunagawa et al., 2015 ). For some metabolic processes, we can interpret gene content and taxa in terms of the specific metabolic processes that they are capable of. For example, we know the dominant taxa that perform processes such as nitrification ( Bock and Wagner, 2013 ) or polysaccharide degradation ( Sanchez-Gorostiaga et al., 2019 ). Further, by annotating metagenomic data, we can assign specific functional roles for many (but not all) of the genes present in a given community. As a result, we can measure the prevalence of enzymes that perform the reactions necessary for specific metabolic processes. Despite the remarkable scale and breadth of these sequencing data, we still do not have a predictive, quantitative framework for using these data to understand, predict, and design the metabolism of the communities in complex environments. In this perspective, we explore what makes this problem both challenging and important. We propose a specific approach to begin to address this question, and examine what types of communities and associated metabolic processes might be amenable to this approach. We review the techniques that are relevant to implementing the approach with a focus on methods for quantifying metabolites. The significance of finding a solution Microbial communities play an outsized role in driving fluxes of nutrients through the biosphere. Photosynthetic microbes are responsible for nearly half of the carbon fixation on the planet ( Falkowski, 1994 ). These phototrophs work in concert with heterotrophic bacteria that enable primary productivity in terrestrial, marine, and freshwater environments ( Kirchman, 2012 ; Madigan et al., 2018 ). We are only beginning to glimpse the role of the collective in this nearly 100 gigaton annual carbon flux. Bacteria and archaea in anaerobic environments degrade complex carbon sources to methane, playing an important role in carbon recycling and climate change ( Madigan et al., 2018 ). In the nitrogen cycle, microbes play a key role in nitrogen fixation (dinitrogen gas to ammonia), nitrification (ammonia to nitrate), and denitrification (nitrate to dinitrogen gas) ( Stein and Klotz, 2016 ). These processes are key for wastewater treatment ( Cydzik-Kwiatkowska and Zielińska, 2016 ) and human health ( Turnbaugh et al., 2007 ). A critical challenge is to form a quantitative and predictive understanding of how microbial communities drive these fluxes. To give a concrete example, the process of denitrification, performed by bacterial communities in soils, reduces nitrate to dinitrogen gas. An intermediate in the conversion of nitrate to dinitrogen is the potent greenhouse gas and ozone depleting compound nitrous oxide ( Tian et al., 2020 ). Denitrifying communities in some cases (especially in agricultural soils) can leak nitrous oxide, but in other cases fully convert nitrous oxide to harmless dinitrogen gas. The question then becomes: What controls the production of nitrous oxide from denitrifying communities in soils? Can we manipulate these microbial communities to limit nitrous oxide production? To address this, we need to understand how the structure of the community and the environmental context determine the flux of metabolites through the system. Similarly, the essential importance of resident microbiota in host health is now clear ( Turnbaugh et al., 2007 ), but as yet, it is unclear how to rationally manipulate these communities to benefit the host. There exists tantalizing evidence that this can be done, for example, by altering metabolic phenotypes ( Turnbaugh et al., 2006 ) or treating persistent infections ( Lawley et al., 2012 ), but we lack general approaches for developing such strategies. Here we focus on environmental microbiomes, but we emphasize that the strategy proposed here could, and in a few cases has been ( Raman et al., 2019 ), applied to host-associated communities. Defining structure and function Before moving forward, we take a moment to define community structure and function. We define the structure of the community as the taxa present as well as the genomic structure of each taxon, which may include everything from the detailed knowledge of the regulatory architecture of each gene, to the syntenic organization of the genome ( Junier et al., 2018 ), to even the presence of phage. The structure of the community may, if necessary, include transcriptional or proteomic information at the metagenomic or single-taxon level as well. We define the function of a community as the collective metabolic activity of all constituent organisms, which, therefore, operates in the space of metabolites. The dynamic or steady-state flux of metabolites through the consortium defines its metabolic function. Depending on the context, the most important metabolic fluxes may include electron donors (e.g., organic carbon), electron acceptors (e.g., oxygen, nitrate), secondary metabolites, biomass, overall catabolic activity, or byproducts. A note about usage: Some readers may find the term “function” teleological, implying some sort of purpose on microbial communities. We use the term function to mean the activity or action of a community without any implication of purpose. Despite this potential confusion, we find that the term function is a useful shorthand. In particular, we would like to invoke a certain symmetry between the ideas presented here and the problem of sequence, structure, and function at the level of proteins. The structure-function problem for microbial communities is therefore to deduce the mapping from the space of genes, transcripts, proteins, and taxonomic organization to metabolite fluxes, and to understand the environmental dependence and context in which this mapping is relevant. How should we approach the problem? Understanding how the metabolic activity of a microbial community emerges from the taxa present and their metabolic capabilities is a problem of connecting hierarchical scales of biological organization, from genes to phenotypes and interactions in specific environmental contexts. What makes this challenging is the fact that processes at different scales feedback on one another. For example, compounds that mediate interactions between strains can do so by modulating gene expression ( Beliaev et al., 2014 ). Similarly, phenotypic variation in individuals, determined by gene expression and interactions, can modify community interactions ( Mickalide and Kuehn, 2019 ) and the chemical environment, with widespread impacts on other members of the consortium ( Ratzke et al., 2018 ). One way to proceed is via the reductionist mode that has motivated biology over the last century ( Woese, 2004 ). In the context of communities, this would mean dissecting the mechanistic and physiological metabolic properties of each member of the community and understanding how metabolite dynamics emerge at the level of the collective. The challenge is the complexity of these systems, which makes a detailed, mechanistic understanding of the collective metabolism a massive undertaking. For synthetic communities comprising model organisms, where detailed information is available, this type of approach has had some success ( Orth et al., 2010 ; Harcombe et al., 2014 ). For comparatively simple communities of a few strains, such as bacteria cross-feeding amino acids ( Wintermute and Silver, 2010 ) or ciliates consuming bacteria ( Mickalide and Kuehn, 2019 ), detailed models have been constructed and validated. However, in environmental or host-associated contexts, where the number of strains present are enormous and many organisms present are poorly studied or challenging to work with in the laboratory, this approach faces huge challenges. In this scenario, what are we to do? On the one hand, building detailed models as described above is ill-advised. Even when such detailed models can be built, the challenge of distilling simple principles from these models increases with their complexity (see Borges’ “On Exactitude in Science” ( Borges and Hurley, 1998 )). On the other hand, we know from many examples that a huge number of processes from antibiotic warfare ( Vetsigian et al., 2011 ) to competition ( Friedman et al., 2017 ), mutualism ( Hom and Murray, 2014 ), and stress responses ( Amarnath et al., 2021 ) influence interactions and metabolism. So, how can we justify not building models that include these details? In the spirit of Philip Anderson’s influential essay ( Anderson, 1972 ), it may be that understanding communities requires explaining entirely new and potentially simpler properties that are emergent at the level of the collective. In this case, due to the scale and complexity of the community, new and distinct phenomena emerge from the individual parts, and discovering the organizing principles requires an approach that goes beyond dissecting the detailed metabolic phenotypes of each individual member of the community. To be clear, we are not advocating the idea that reductionist approaches are not useful. Their utility is clear from many examples ( Jacob and Monod, 1961 ; Zumft, 1997 ; Alon et al., 1999 ; Basan et al., 2020 ; Amarnath et al., 2021 ). Instead, we are asking whether making headway on the “structure-function problem”, as we have defined it, might require a complementary approach that focuses not on the detailed mechanisms and physiology of each organism, but on patterns that are evident at the level of the collective. Here we discuss such an approach. Learning the right variables: the power of statistics across ensembles We find it useful to cast the structure-function question in terms of a prediction problem. In this framing, the goal is to predict metabolite dynamics or fluxes from community structure for a given environment or set of environmental conditions. The key question then becomes: What structural elements must we quantify to predict function? Equivalently, we could ask, how predictive of metabolic function is community gene content, taxonomy, transcriptional or proteomic data? One approach to this problem, which has found success in both physics and biology, is to look for statistical regularities across replicate systems and allow these patterns to naturally define variables that can be used to make predictions. In many cases, this approach can reveal the salient, emergent features of a system and provide deep insight into their function. Specifically, from proteins ( Halabi et al., 2009 ) to multicellular organisms ( Alba et al., 2021 ), examining statistical variation across many replicates of a system, or an ensemble, has proven powerful. For example, covariation across ensembles of homologous proteins have been used to reveal which amino acids are in contact in the folded structure ( Russ et al., 2005 ), and co-evolving groups of residues that correlate with enzyme function ( Halabi et al., 2009 ). Careful analysis of behavioral variation in large numbers of microbes ( Jordan et al., 2013 ) and flies ( Berman et al., 2014 ) suggests that apparently very complex behaviors can, in fact, be described by a relatively small number of elementary behavioral features ( Berman et al., 2014 ; Katsov et al., 2017 ). Statistical analysis of morphological variation across higher organisms suggests that morphologies adhere to constraints that some believe are associated with specific functional capabilities ( Raup and Michelson, 1965 ; Shoval et al., 2012 ). A recent study of variation in patterning in the fly wing shows that a single mode of variation describes the response of the wing developmental program to genetic and environmental perturbations ( Alba et al., 2021 ). For a recent piece discussing why low-dimensionality might be an inherent property of evolved systems, see Eckmann and Tlusty (2021) . The common feature of all of these examples is that, by judiciously studying the variation across a carefully chosen ensemble of systems, one can often discover simple, relatively low-dimensional features that enable the prediction of the functional properties of the system. Here we advocate a similar approach to communities. An ensemble approach to the structure-function problem In light of these considerations, we propose an ensemble approach to the structure-function problem in microbial communities. We motivate this approach by analogy to a similar approach taken at the level of proteins ( Figure 1 ). For proteins, the structure-function problem is to predict the fold and function of a protein from its amino acid sequence. One way to approach this problem is by performing detailed physico-chemical simulations of a polypeptide chain via molecular dynamics ( Figure 1 A). While much progress has been made in this approach, it has proven a huge technical challenge. However, a statistical approach, ignoring the mechanistic details and instead considering only statistics of a multiple sequence alignment does a remarkably good job of predicting protein folds ( Morcos et al., 2011 ). Similar approaches reveal low-dimensional structure in proteins which is predictive of function ( Halabi et al., 2009 ; Russ et al., 2020 ). These studies suggest that much can be learned by carefully considering variation across a suitably chosen ensemble of systems. Figure 1 Sequence, structure and function in proteins and microbial communities We propose that there exist analogous solutions to the sequence-structure problem in protein folding and the structure-function problem in microbial communities. (A) The mapping from amino acid sequence to 3D protein structure can be accomplished either by a simulation approach (e.g., molecular dynamics) or by a statistical approach (e.g., direct coupling analysis). The former is a computationally intensive strategy to simulate 3D protein structure based on first-principles modeling of atomic interactions. The latter leverages information about residue coevolution from an ensemble of amino acid sequences to infer which residues are in contact, allowing for an elegant and interpretable statistical inference of 3D structure. (B) The mapping from genomic and metagenomic sequences to community metabolic activity can be achieved through community flux balance modeling or, as we propose, a statistical ensembles approach. The former requires genome-level metabolic models of each organism to be built, a labor-intensive iterative process that so far has been successful primarily in a handful of model organisms. The latter leverages the diversity and variation in an ensemble of communities to learn an effective mapping between community sequence content metabolic activity In the context of microbial communities, flux balance models of community metabolism are analogous to molecular dynamics simulations in proteins because they attempt a detailed mechanistic accounting of all of the phenomena within a community ( Figure 1 B). However, in communities, there is comparatively little work pursuing a statistical approach analogous to the one taken at the level of proteins. Taking such an approach is precisely what we are advocating here. We propose quantifying the structure of a collection, or ensemble, of communities using sequencing while simultaneously measuring metabolite dynamics. Given such data, one can then approach the structure-function problem by asking whether variation in community structure (e.g., across metagenomes or metatransciptomes) permits quantitative insights into the functional properties of these communities. The proposal is then to leverage these insights to design, predict, and control community function. Several studies in the past few years have begun to explore statistical approaches ( Gowda et al., 2022 ; Raman et al., 2019 ). However, we suggest that this approach is under-explored and that there is a pressing need to collect new datasets that are explicitly designed to pursue a statistical approach to the structure-function problem in communities. Moreover, while we focus on community metabolic function, the approach we propose could just as easily be applied to other salient features of ecosystems from spatial structure to resilience to stability. Overview of the paper We will focus on three main challenges that must be surmounted to apply the ensemble approach to the structure-function problem: (1) Choosing an ensemble across which one should make comparisons and look for patterns, (2) measuring metabolite dynamics, which requires analytical chemistry techniques that are often not standard practice in microbial ecology labs, and (3) using these data to distill the mapping from structure to function. Our intention is to provide a roadmap for how such an approach might be applied across communities of interest. We recognize that this roadmap is far from complete and that many pitfalls exist that may render this approach challenging in various circumstances. We will not review sequencing technologies, which have been widely and capably recapped elsewhere ( Knight et al., 2018 ). We will focus largely on microbial communities in environmental contexts rather than health-related (human microbiome) contexts, in part because environmental microbiology is where our expertise lies, but also because of the abundance of existing literature on the latter topic. Finally, we will neglect the many recent advances in theoretical ecology, in particular the renaissance in consumer-resource models applied to communities, in service of focusing on questions that can be settled empirically."
} | 5,112 |
26776497 | PMC4716644 | pmc | 821 | {
"abstract": "Background Methanosarcina acetivorans is a model archaeon with renewed interest due to its unique reversible methane production pathways. However, the mechanism and relevant pathways implicated in (co)utilizing novel carbon substrates in this organism are still not fully understood. This paper provides a comprehensive inventory of thermodynamically feasible routes for anaerobic methane oxidation, co-reactant utilization, and maximum carbon yields of major biofuel candidates by M. acetivorans . Results Here, an updated genome-scale metabolic model of M. acetivorans is introduced (iMAC868 containing 868 genes, 845 reactions, and 718 metabolites) by integrating information from two previously reconstructed metabolic models (i.e., iVS941 and iMB745), modifying 17 reactions, adding 24 new reactions, and revising 64 gene-protein-reaction associations based on newly available information. The new model establishes improved predictions of growth yields on native substrates and is capable of correctly predicting the knockout outcomes for 27 out of 28 gene deletion mutants. By tracing a bifurcated electron flow mechanism, the iMAC868 model predicts thermodynamically feasible (co)utilization pathway of methane and bicarbonate using various terminal electron acceptors through the reversal of the aceticlastic pathway. Conclusions This effort paves the way in informing the search for thermodynamically feasible ways of (co)utilizing novel carbon substrates in the domain Archaea . Electronic supplementary material The online version of this article (doi:10.1186/s12934-015-0404-4) contains supplementary material, which is available to authorized users.",
"conclusion": "Conclusions An updated genome-scale metabolic reconstruction (iMAC868) for the archaeon Methanosarcina acetivorans that integrates the latest literature findings and provides complete pathways and electron flow systems for reversing methanogenesis is introduced. Upon improving growth and gene-knockout outcome prediction for M. acetivorans grown on its native substrates, the model was used in a prospective mode for assessing thermodynamically feasible methanotrophic pathways leading to the production of biofuel candidate molecules such as methanol, ethanol, butanol, and isobutanol. We found that anaerobic methanotrophy favored the production of acetate and CO 2 as they provide free energy equivalents to support growth. Co-utilization of CO 2 (in the form of bicarbonate) and methane was feasible for certain ratios leading to improved carbon yields for acetate and biofuel molecules. Re-routing of a fraction of carbon towards CO 2 was found to be a recurring mechanism for driving growth and production within thermodynamically constrained metabolic states. Finally, the availability in excess of inorganic electron acceptors resulted in a switch between substrate-level and chemiosmotic ATP synthesis. Thermodynamic constraints were often the limiting factor in product yields. Unsurprisingly, the inability to completely reverse the aceticlastic pathway in the absence of an external electron acceptor was confirmed. The theoretical limits of external electron acceptor utilization possibilities to drive the reversal of methanogenesis were thoroughly explored. Key challenges that confound the fidelity of model predictions are still unknown sodium gradient requirements, substrate-dependent regulation and the detailed mechanism of electron transport from internal electron carriers to their external counterparts. Shedding light to these questions will require systematic experimental investigations to confirm or refute electron flow paths guided by the rapidly expanding modeling infrastructure.",
"discussion": "Results and discussion Updated genome-scale metabolic model reconstruction of M. acetivorans , iMAC868 iMAC868 contains 868 genes, 845 reactions, and 718 metabolites (Additional file 1 ) and provides better agreement with the observed growth yields on methanol and acetate compared to earlier reconstructions (see Table 1 ). Improved prediction is due to the correction of charge and mass imbalances of the reactions inherited from the previous metabolic models, incorporation of accurate ion exchange stoichiometries for membrane-bound reactions, and optimization of Na + /H + ratio for sodium/proton antiporter (Mrp) and ATP synthase. Among the charge re-balanced reactions, those involving cofactor F 420 in the methylotrophic pathway and ATP synthase also required proton rebalancing to accurately account for proton exchange across the cell membrane. The number of Na + pumped out by ferredoxin-dependent methanophenazine reductase (Rnf) was updated from three Na + in iMB745 [ 36 ] to four Na + per methanophenazine reduced in accordance with experimental findings [ 38 ]. In addition to this, the ATP synthase reaction was modified to co-utilize Na + and H + gradients [ 37 ]. Using the procedure described in Methods section, the optimal Na + /H + ratio recapitulating the growth yields on native substrates was estimated to be 2:1 for Mrp and 3:1 and 1:2 for ATP synthase, respectively. The two identified solutions for ATP synthase are rendered equivalent by the reversible 2:1 antiport of Na + /H + across the cell membrane by Mrp, which makes one intracellular H + equivalent to two extracellular Na + . We also added an F 420 -dependent NADP reductase to the iMAC868 model (personal communications with James G. Ferry), which functions as a source of NADPH for cell biosynthesis. Table 1 Growth yield predictions of iMAC868 model of M. acetivorans compared with predictions of previous models iVS941 [ 35 ] and iMB745 [ 36 ] Substrate Observed growth yield Predicted growth yield iVS941 [ 35 ] iMB745 [ 36 ] iMAC868 [this study] \n Methanol 5.2 [ 69 ] 9.5 4.0 5.26 Acetate 2.4 [ 29 ] 4.0 3.0 2.6 Yield units are gram dry cell weight per mol of substrate Upon correcting 64 GPRs based on updated gene annotations (Additional file 2 ) and implementing proteomics-dependent growth condition-specific R-GPR switches, iMAC868 correctly predicts gene knockout outcomes for 27 out of 28 mutants of M. acetivorans (see Table 2 ; Additional file 1 ). The only false prediction by the iMAC868 model is missing the in vivo essentiality of the mutant lacking methanol-specific methyltransferases (Δ mtaA1 Δ mtaCB1 Δ mtaCB2 Δ mtaCB3 ) growing with acetate due to the unknown role of the enzymes in acetate-grown cells [ 45 ]. The model correctly captures the essentiality of mch [ 46 ] by identifying the role of the methylotrophic pathway as a source of reduced F 420 for NADPH production in acetate-grown cells thereby rendering mch (methenyl-H 4 SPT cyclohydrolase) essential. Comparing with the iVS941 model, the iMAC868 model correctly predicts the essentiality of rnf , mtr , and the membrane-bound hdr due to the updated GPRs and ion transport stoichiometries included in this model. Table 2 Gene deletion lethality predictions by iMAC868 model of M. acetivorans compared with predictions of previous models Gene deletion Acetate Methanol References iVS941 iMB745 iMAC868 iVS941 iMB745 iMAC868 \n ΔackΔpta \n NGNG NGNG NGNG GG GG GG [ 30 ] \n ΔhdrABC \n GG GG GG GG GG GG [ 70 ] \n ΔhdrED \n GNG NGNG NGNG GNG NGNG NGNG [ 70 ] \n Δmch \n GNG GNG NGNG NGNG NGNG NGNG [ 46 ] \n ΔmtaA1 \n GG GG GG NGNG NGNG NGNG [ 45 ] \n ΔmtaB 1C1ΔmtaB 2C2ΔmtaB 3C3 \n GG GG GG NGNG NGNG NGNG [ 45 ] \n ΔmtaA1ΔmtaB 1C1ΔmtaB 2C2ΔmtaB 3C3 \n GNG GNG GNG NGNG NGNG NGNG [ 45 ] \n ΔmtbA \n GG GG GG GG GG GG [ 45 ] \n ΔmtsDΔmtsFΔmtsH \n GG GG GG GG GG GG [ 71 ] \n ΔmtsXΔmtsY, X and Y any two mts genes \n GG GG GG GG GG GG [ 71 ] \n ΔrnfHCDGEABF \n GNG NGNG NGNG GG GG GG [ 70 ] \n ΔlysK \n GG GG GG GG GG GG [ 72 ] \n ΔlysS \n GG GG GG GG GG GG [ 72 ] \n Δmtr \n GNG NGNG NGNG NGNG NGNG NGNG [ 73 ] Total correct 9/14 12/14 13/14 13/14 14/14 14/14 \n GG growth in silico/growth in vivo, GNG growth in silico/no growth in vivo, NGNG no growth in silico/no growth in vivo Model customization to capture methanotrophy by M. acetivorans In order to allow for methanotrophy, the iMAC868 model was customized to enable three new processes: (1) reversal of methyl-coenzyme M reductase (Mcr) reaction, (2) inclusion of a cytosolic methyltransferase (CmtA), and (3) inclusion of a mechanism enabling electron bifurcation and its subsequent discharge to an external electron acceptor. The methyl-coenzyme M reductase of an anaerobic methanotroph (ANME-MCR), capable of oxidizing methane [ 47 ], was appended to the iMAC868 model upon deactivating the native Mcr to prevent methanogenesis based on evidence regarding the reversal of methanogenesis in M. acetivorans [ 23 , 24 , 48 ], reversibility of native Mcr [ 49 ], and the confirmed heterologous expression of ANME-MCR in M. acetivorans demonstrating ferric-dependent methanotrophy [ 44 ]. CmtA [ 50 ] serves as a soluble alternative to membrane-bound Mtr, allowing the conversion of methyl-coenzyme M to methyl-tetrahydrosarcinapterin without drawing on sodium ion gradients across the membrane. Cdh, a key enzyme in the modeled pathway, is dependent on ferredoxin to reduce CO 2 that generates the carbonyl group in acetyl-CoA [ 51 ]. Two flavin-based mechanism are postulated in which an electron pair derived from oxidation of coenzyme B and coenzyme M ( E m = −143 mV) bifurcates yielding a high-potential electron reducing Fe 3+ to Fe 2+ ( E m = +770 mV) and a low-potential electron reducing ferredoxin ( E m = −420 mV). Flavin-based electron bifurcation is common among anaerobic microbes including methanogens [ 52 , 53 ]. Although, both postulated mechanisms depend on delivering electrons to Fe 3+ on the outer aspect of the cytoplasmic membrane, the bifurcation event occurs either at the cytoplasm or the membrane. Bifurcation in the membrane depends on the Rnf complex, abundant in acetate-grown M. acetivorans , which interacts with ferredoxin and contains two FMN-bound subunits that are possible sites for electron bifurcation [ 54 ]. Oxidation of coenzyme B and coenzyme M is catalyzed by the membrane-bound CoMS-SCoB heterodisulfide reductase (HdrDE) [ 54 ]. Methanophenazine (MP) is a quinone-like electron carrier that shuttles electrons between HdrDE and the Rnf complex. Importantly, the reduction of ferredoxin is not dependent on a sodium gradient. Bifurcation in the cytoplasm is postulated to be dependent on the fused HdrA2:MvhD protein shown previously to be present in acetate-grown M. acetivorans [ 34 ]. Oxidation of HS-CoB and HS-CoM is catalyzed by the soluble heterodisulfide reductase (HdrB2) that donates electrons to the flavin-containing HdrA2 component where bifurcation takes place reducing ferredoxin and transferring an electron to the membrane where reduction of Fe 3+ takes place. Finally, an electron transfer reaction is included in the model to transfer the electrons from reduced methanophenazine to an externally supplied electron acceptor based on its reported in vivo essentiality [ 19 , 20 , 55 ]. The essentiality of this reaction was confirmed by the absence of any in silico external electron acceptor-independent thermodynamically feasible metabolic state despite allowing the production of all reported reduced products such as hydrogen gas [ 56 ] and organic acids such as acetate and formate [ 30 ]. These additions complete the pathways for the oxidation of methane to various end products such as acetate, formate and CO 2 . The addition of prospective biofuel molecule production pathways for ethanol [ 57 ], butanol [ 58 ], and isobutanol [ 59 ] to the model allows the exploration of their thermodynamically feasible maximum theoretical yields for different electron acceptors. Products of electron-acceptor-dependent AOM The model supports acetate, CO 2 and biomass as the main products of methanotrophy using all tested electron acceptors. Methane is oxidized to methyl-H 4 SPT by ANME-MCR and Mtr (or CmtA), a part of which is oxidized via the methylotrophic pathway to produce intracellular CO 2 . The remaining methyl-H 4 SPT is used to produce acetyl-CoA, the primary building block for all biomass precursors. Additional carbon fixation occurs via reductive carboxylation by Cdh and Por. Acetyl-CoA is also converted to acetate, generating ATP via substrate-level phosphorylation (Fig. 1 ). The electrons released upon activation of methane by ANME-MCR are transferred to ferredoxin via soluble and membrane-bound electron transport chains involving flavin-based electron bifurcation mechanisms. In addition, further oxidation of methyl-H 4 SPT through methylotrophic pathway generates reducing equivalents in the form of ferredoxin and F 420 . Fpo and Rnf complexes facilitate the transfer of electrons from reduced F 420 and ferredoxin, respectively, to the external electron acceptor via methanophenazine, thereby generating H + and Na + gradients across the membrane for chemiosmotic ATP synthesis (see Fig. 1 ). The primary carbon fixation mechanism via reductive carboxylation prompted a quantitative analysis of the impact of utilizing CO 2 as a co-substrate in the form of bicarbonate on acetate and biomass yields. Fig. 1 Proposed methanogenesis reversal pathway supported by the iMAC868 model of M. acetivorans for co-metabolization of methane and bicarbonate in the presence of Fe 3+ as external electron acceptor. Soluble and membrane-bound electron bifurcation routes are shown as blue and orange , respectively, and enzymes within ovals. In both routes, electrons originate from coenzyme B and coenzyme M. For enzymes with multiple subunits, only the subunits of soluble Hdr and Rnf involved in electron bifurcation are shown in detail. Numbers in italics next to enzyme ovals denote reaction fluxes (in mmol/gDCW-h) calculated under maximization of acetate production at bicarbonate to methane ratio of 0.44. This ratio corresponds to the maximum thermodynamically feasible value ensuring biomass production at 30 % of its theoretical maximum for Fe 3+ as the electron acceptor. The flux towards growth was calculated by assuming that 1 g of biomass contains 36 mmol of carbon. Intracellular proton and water stoichiometries are omitted for the sake of simplicity. Soluble methyltransferase (CmtA) is not present in the network since the minimum possible flux through this reaction is zero. Mcr \n * putative ANME-like Mcr homolog to methyl-coenzyme M reductase, HdrBC:HdrA:MvhD soluble ferredoxin-dependent heterodisulfide reductase, Mtr methyl-THSPT:coenzyme M methyltransferase, Mer methenyl-THSPT reductase, Mtd methenyl-THSPT dehydrogenase, Mch methenyl-THSPT cyclohydrolase, Ftr formylmethanofuran:THSPT formyltransferase, Fmd formylmethylfuran dehydrogenase, Cdh CO dehydrogenase, Pta phosphotransacetylase, Ack acetate kinase, Por pyruvate synthase, Atps ATP synthase, Mrp sodium/proton antiporter, Rnf methanophenazine reductase, Cyt cytochrome c subunit of Rnf complex, Fpo F 420 dehydrogenase, Cam carbonic anhydrase, F4nr F 420 -dependent NADP reductase, THSPT tetrahydrosarcinapterin, MF methanofuran, MP methanophenazine, MPH \n 2 reduced methanophenazine, Fd \n o oxidized ferredoxin, Fd \n r reduced ferredoxin, F \n 420 coenzyme F 420 , F \n 420 \n H \n 2 reduced coenzyme F 420 \n Thermodynamic feasibility of methanotrophy is ensured only when the free energy of reduction (ΔG red ) of the supplied electron acceptor is less than 50.5 kJ/electron-pair (Fig. 2 ), corresponding to the maximum free energy equivalents generated by CO 2 production (see Table 3 ). Using methane as the sole carbon source, maximum biomass yield is constrained by thermodynamic feasibility when ΔG red of the electron acceptor is greater than −20 kJ/electron-pair. Sulfate-dependent methanotrophy falls within this regime, in which thermodynamic coupling with an exergonic pathway such as acetate or CO 2 production (Table 3 ) drives only partial conversion of methane to biomass. In contrast, biomass production is limited only by stoichiometry during ferric-dependent methanotrophy due to the far greater free energy equivalents produced by the reduction of Fe 3+ to Fe 2+ (ΔG = −140.44 kJ/electron-pair) compared to SO 4 2− reduction (ΔG = 44.53 kJ/electron-pair). This thermodynamic advantage for Fe 3+ allows for the co-utilization of bicarbonate up to a maximum HCO 3 − /CH 4 ratio of 0.44 with complete incorporation of all substrate carbons into biomass. However, the endergonic nature of bicarbonate uptake disallows HCO 3 − /CH 4 co-utilization for increasing biomass yield under sulfate-dependent methanotrophy. Methanotrophy using NO 3 − and MnO 2 also allows for co-utilization of bicarbonate as both electron acceptors have ΔG red greater than −20 kJ/electron-pair (see Table 3 ). Fig. 2 Biomass yield (per 10 mmol methane) as a function of the ΔG of external electron acceptor reduction (kJ/electron-pair) predicted by the iMAC868 model of M. acetivorans . Solid line methane as the sole carbon source; dashed line bicarbonate and methane (at a ratio of HCO 3 \n − /CH 4 = 0.44) as carbon sources. Vertical dotted lines show the ΔG (kJ/electron-pair) of reduction for Fe 3+ (−140.44), MnO 2 (−77.65), NO 3 \n − (−68.15), and SO 4 \n 2− (+44.53). A magnified insert shows the maximum biomass yields for sulfate-dependent methanotrophy. All ΔG values were evaluated at pH of 7, 25 °C, and an ionic strength of 0.25 M as described by Alberty [ 64 ] Table 3 Oxidation half reactions of methane to various products and reduction half reactions of various electron acceptors Oxidation half reactions ΔG (kJ/mole product) ΔG (kJ/mole methane) \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}$$CH_{4} + 2 H_{2} O \\to CO_{2} + 8 H^{ + } + 8 e^{ - }$$\\end{document} C H 4 + 2 H 2 O → C O 2 + 8 H + + 8 e - \n −202.79 −202.79 \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}$$2 CH_{4} + 2 H_{2} O \\to CH_{3} COOH + 8 H^{ + } + 8 e^{ - }$$\\end{document} 2 C H 4 + 2 H 2 O → C H 3 C O O H + 8 H + + 8 e - \n −193.44 −96.72 \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}$$4 CH_{4} + H_{2} O \\to CH_{3} CH_{2} CH_{2} CH_{2} OH + 8 H^{ + } + 8 e^{ - }$$\\end{document} 4 C H 4 + H 2 O → C H 3 C H 2 C H 2 C H 2 O H + 8 H + + 8 e - \n −113.48 −28.37 \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}$$4 CH_{4} + H_{2} O \\to CH_{3} CH(CH_{3} )CH_{2} OH + 8 H^{ + } + 8 e^{ - }$$\\end{document} 4 C H 4 + H 2 O → C H 3 C H ( C H 3 ) C H 2 O H + 8 H + + 8 e - \n −104.87 −26.22 \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}$$2 CH_{4} + H_{2} O \\to CH_{3} CH_{2} OH + 4 H^{ + } + 4 e^{ - }$$\\end{document} 2 C H 4 + H 2 O → C H 3 C H 2 O H + 4 H + + 4 e - \n −31.06 −15.53 \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}$$CH_{4} + H_{2} O \\to CH_{3} OH + 2 H^{ + } + 2 e^{ - }$$\\end{document} C H 4 + H 2 O → C H 3 O H + 2 H + + 2 e - \n +14.83 +14.83 Reduction half reactions ΔG (kJ/mole oxidant) ΔG (kJ/electron-pair) \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}$$Fe^{3 + } + e^{ - } \\to Fe^{2 + }$$\\end{document} F e 3 + + e - → F e 2 + \n −70.22 −140.44 \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}$$MnO_{2} + 4 H^{ + } + 2 e^{ - } \\to Mn^{2 + } + 2 H_{2} O$$\\end{document} M n O 2 + 4 H + + 2 e - → M n 2 + + 2 H 2 O \n −77.65 −77.65 \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}$$NO_{3}^{ - } + 10 H^{ + } + 8 e^{ - } \\to NH_{4}^{ + } + 3 H_{2} O$$\\end{document} N O 3 - + 10 H + + 8 e - → N H 4 + + 3 H 2 O \n −272.75 −68.15 \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}$$SO_{4}^{2 - } + 10 H^{ + } + 8 e^{ - } \\to H_{2} S + 4 H_{2} O$$\\end{document} S O 4 2 - + 10 H + + 8 e - → H 2 S + 4 H 2 O \n +178.14 +44.53 Standard transformed ΔGs were calculated at pH of 7, 25 °C, and an ionic strength of 0.25 M as described by Alberty [ 64 ] The model predicts a maximum acetate production (0.5 mol/mol-methane), constrained only by stoichiometry for both Fe 3+ and SO 4 2− during growth on only methane. This yield is further increased to 0.94 mol/mol-methane at an optimal HCO 3 − /CH 4 ratio of 0.88 for ferric-dependent methanotrophy, and 0.68 mol/mol-methane at an optimal HCO 3 − /CH 4 ratio of 0.36 during sulfate-dependent methanotrophy (Fig. 3 a). The improvement in acetate yield arises from the reduction in the fraction of methane oxidized via the methylotrophic pathway from 50 to 6 % and 32 % during ferric- and sulfate-dependent methanotrophy, respectively. However, a complete reversal of the aceticlastic pathway with a co-utilization ratio of one could not be achieved using either electron acceptor due to thermodynamic restrictions during sulfate-dependent methanotrophy and reduced ferredoxin availability during ferric-dependent methanotrophy. Under sulfate-dependent methanotrophy, the minimum essential flux through the methylotrophic pathway enables thermodynamic coupling with sulfate reduction for the generation of free energy equivalents. Mandatory channeling of electrons towards ferric ions by the electron bifurcation mechanism decreases available reduced ferredoxin for acetate synthesis during ferric-dependent methanotrophy. Despite the exergonic nature and the ATP generating capability of the acetate production pathway, it is never essential (minimum acetate production is always zero) at any HCO 3 − /CH 4 ratio due to the fact that there exist other competing products and an electron acceptor-driven chemiosmotic ATP synthesis. Fig. 3 Acetate and carbon dioxide production yields as a function of bicarbonate to methane ratio ( a , b ) and biomass yield ( c , d ) using Fe 3+ ( solid lines ) or SO 4 \n 2− ( dashed lines ) as external electron acceptors. Methane was the sole carbon source for generating the plots shown in panels ( c ) and ( d ). All yields are per 10 mmol of methane CO 2 production remains non-essential during ferric-dependent methanotrophy as revealed by the model (Fig. 3 b) due to the fact that reductive carboxylation of acetyl-CoA allows the production of many different thermodynamically feasible products. In contrast, CO 2 production for sulfate-dependent methanotrophy beyond a HCO 3 − /CH 4 ratio of 0.36 becomes mandatory. Beyond this ratio, increased CO 2 production via the methylotrophic pathway serves to offset the free energy increase associated with uptake of bicarbonate. Carbon channeling towards the methylotrophic pathway leads to increased CO 2 production thus decreasing methane flow towards other major products, thereby adversely affecting acetate and biomass yields at HCO 3 − /CH 4 ratios beyond 0.36. The trade-off plot between the products of AOM and biomass did not reveal any thermodynamic restrictions in the solution space during ferric-dependent methanotrophy (Fig. 3 c, d). However, the model predicts that acetate becomes thermodynamically constrained beyond a biomass yield of 0.018 for sulfate-dependent methanotrophy. Up to this yield value, the minimum required CO 2 production remains zero due to the fact that either acetate or CO 2 production pathways can generate the necessary free energy equivalents, ATP and reducing equivalents for biomass production. At biomass yields above 0.018, CO 2 production becomes mandatory. The production of ethanol, butanol, isobutanol, and methanol is thermodynamically feasible through both ferric-dependent and sulfate-dependent methanotrophy (Fig. 4 ). However, complete carbon conversion of methane to candidate biofuel molecules is thermodynamically feasible only for ferric-dependent methanotrophy with methane as the sole carbon source (Fig. 4 a) due to the favorable thermodynamics of coupling the biofuel production pathways by ferric reduction (see Table 3 ). Upon co-utilization of methane and bicarbonate, electron bifurcation limits the availability of reduced ferredoxin for fixing CO 2 by Cdh to produce acetyl-CoA (biofuel precursor), thereby restricting maximum achievable biofuel yield (Fig. 4 a). Moreover, biofuel production pathways require additional energy in the form of NAD(P)H necessitating elevated amounts of reduced F 420 at increasing bicarbonate to methane ratios which is also controlled by electron bifurcation. Incorporation of bicarbonate into methanol occurred via the CO 2 reduction pathway (reversal of the methylotrophic pathway) as opposed to CO 2 reduction by acetyl-CoA synthesis, causing all electrons to be generated by the ANME-MCR. During sulfate-dependent methanotrophy, none of the products could be produced with the complete carbon conversion efficiency due to the fact that coupling biofuel production with SO 4 2− reduction remains thermodynamically infeasible (see Table 3 ) requiring the co-production of by-products such as acetate or CO 2 . As a consequence of this, co-utilization of bicarbonate and methane is not supported (see Fig. 4 b). Although both oxidation of methane to methanol and reduction of sulfate to sulfide are thermodynamically infeasible, methanol can be still produced with SO 4 2− due to coupling with the concomitant production of CO 2 . The lower bound for the production of all biofuel molecules is zero indicating that their production is not growth-coupled when methane is either the sole carbon source or co-utilized with bicarbonate. Fig. 4 Biofuel yields using methane ( shaded in black ) as the sole carbon source and HCO 3 \n − /CH 4 ( shaded in gray ) with a ratio of 0.44 for ferric-dependent ( a ) and sulfate-dependent ( b ) methanotrophy. The dashed bars denote the excess carbons that could not be incorporated into the product due to thermodynamic restrictions. Ferric-dependent methanotrophy predicted increased yields at higher HCO 3 \n − /CH 4 ratios, but the carbon conversion was less than 100 %. Sulfate-dependent methanotrophy cannot achieve 100 % carbon conversion at any HCO 3 \n − /CH 4 ratio due to mandatory co-production of thermodynamically feasible by-products. Product yields are in mol per mol methane Interplay between electron acceptors and by-products of AOM at no growth The interplay between the external electron acceptor choice and various products of AOM is pictorially illustrated (see Fig. 5 ) using feasible production envelopes for growth-arrested cells. We constrained the model for zero growth, ATP production for only maintenance requirements, and bicarbonate to methane ratio of 0.44. Analysis of the product profiles predicted by the iMAC868 model, based on the imposed constraints, identifies acetate as the main product of co-utilization of methane with bicarbonate along with the possible production of formate, CO 2 , and methyl sulfide (Fig. 5 ). A minimum Fe 3+ uptake (i.e., 0.5 mol/mol-methane) is necessary to maintain thermodynamic feasibility of the observed solution spaces (Fig. 5 a–d) at which methyl sulfide is found to be essential (Fig. 5 a) due to the fact that it is the least oxidized by-product of methanotrophy by M. acetivorans . The maximum methyl sulfide yield at this Fe 3+ uptake exceeds methane uptake indicating that bicarbonate is reduced via the methylotrophic pathway. The reversal of the methylotrophic pathway, however, is limited by the availability of reducing equivalents provided by Mcr, thereby resulting in an incomplete conversion of substrate carbons (methane and bicarbonate) to methyl sulfide. Increasing Fe 3+ uptake allows more flux through the methylotrophic pathway, thereby generating additional intracellular CO 2 for an increased acetate production of up to a maximum of 0.71 mol/mol-methane at an Fe 3+ uptake of 2.2 mol/mol-methane (Fig. 5 b). At this uptake rate, acetate can be produced as the sole product of methanotrophy resulting in non-essentiality of methyl sulfide production. Beyond this Fe 3+ uptake rate, acetate production decreases due to the paucity of methyl-coenzyme M arising from increased flux through the methylotrophic pathway and channeling of electrons towards Fe 3+ reduction via the membrane-bound electron transport chain. This increase in methylotrophic pathway flux also increases the yield of formate, an intermediate of this pathway. Maximum formate yield is found to be 1.44 mol/mol-methane at a Fe 3+ uptake of 5.1 mol/mol-methane (Fig. 5 c) where all taken up carbons are converted to formate. Beyond this Fe 3+ uptake rate, CO 2 production becomes essential so as to generate sufficient electrons for reduction of Fe 3+ (Fig. 5 d). A consequence of the essentiality of CO 2 is the reduction of maximum formate yield. At a maximum Fe 3+ uptake of 8 mol/mol-methane, only CO 2 is produced due to the fact that it is the most oxidized form of carbon that can be produced by M. acetivorans . Fig. 5 Thermodynamically feasible production envelope ( highlighted in grey ) of methyl sulfide ( a ), acetate ( b ), formate ( c ), and carbon dioxide ( d ) as a function of Fe 3+ uptake predicted by the iMAC868 model under no growth. All product yields and Fe 3+ uptake are defined as mol per mol methane"
} | 7,786 |
37367620 | PMC10301184 | pmc | 822 | {
"abstract": "Arbuscular mycorrhizal fungi (AMF) have been shown to assist plants in increasing metal tolerance and accumulation in heavy metal (HM)-contaminated soils. Herein, a greenhouse pot experiment was conducted to assess the interactions of growth substrates (S1, S2, and S3, respectively) with various HM contamination and nutrient status sampling from a typical contaminated soil and tailings in Shuikoushan lead/zinc mining in Hunan province, China, and AMF inoculation obtained from plants in uncontaminated areas ( Glomus mosseae , Glomus intraradices, and uninoculated, respectively) on the biomass and uptake of HMs and phosphorus (P) by the black locust plant ( Robinia pseudoacacia L.). The results indicated that the inoculation with AMF significantly enhanced the mycorrhizal colonization of plant roots compared with the uninoculated treatments, and the colonization rates were found to be higher in S1 and S2 compared with S3, which were characterized with a higher nutrient availability and lead concentration. The biomass and heights of R. pseudoacacia were significantly increased by AMF inoculation in S1 and S2. Furthermore, AMF significantly increased the HM concentrations of the roots in S1 and S2 but decreased the HM concentrations in S3. Shoot HM concentrations varied in response to different AMF species and substrate types. Mycorrhizal colonization was found to be highly correlated with plant P concentrations and biomass in S1 and S2, but not in S3. Moreover, plant biomass was also significantly correlated with plant P concentrations in S1 and S2. Overall, these findings demonstrate the interactions of AMF inoculation and growth substrates on the phytoremediation potential of R. pseudoacacia and highlights the need to select optimal AMF isolates for their use in specific substrates for the remediation of HM-contaminated soil.",
"conclusion": "5. Conclusions In three typical mining soils with different levels of HM pollution (S1, S2, and S3), a symbiotic relationship between plants and AMF was well established. Glomus mosseae and Glomus intraradices significantly promoted growth in S1 and S2, whereas they had no effect in S3, which was characterized with the highest Pb contamination and P availability. Similarly, the P content and accumulation in the shoots and roots of the host plants in S1 and S2 increased significantly after inoculation with AMF, whereas there was no significant change observed in S3. Additionally, AMF exhibited different influences on HM uptake by R. pseudoacacia , which depended on the HM concentrations and P availability. Overall, AMF significantly increased the efficiency of phytoextraction in S1 and S2 but decreased this efficiency in S3. Therefore, further research on the enzymes, proteins, and genes involved in the AMF-assisted improvement of the host plant HM resistance is needed.",
"introduction": "1. Introduction Heavy metals (HMs) have little mobility in soil, and are not easily leached via water or degraded by microorganisms [ 1 ]. Soil pollution by HMs is attracting increasing attention due to the persistence and hazardous effects of HMs on plant growth, such as the disruption of metabolic processes, nutrient homeostasis, and beneficial microbial activities. HMs are also transferred up the food chain via crops and therefore pose a great threat to human health [ 2 ]. HM pollution mainly originates from automobile emissions, industrial waste, and the continuous exploitation and refining of mineral resources [ 3 , 4 ]. Among these sources, metal mining and smelting activities are the most impactful. Although the exploitation and utilization of mineral resources promote China’s economic development, HM contamination is still attracting attention. Therefore, there is an urgent need for the remediation of HM-contaminated soil caused by the metal mining and smelting activities. Various chemical and physical remediation methods have been used in the restoration of HM-contaminated soils, although their high cost has seriously restricted their general utilization [ 5 ]. Phytoremediation is a promising method for the clean-up of HM-contaminated soils as it can be performed in situ, inexpensively, and effectively via the employment of hyperaccumulators that are extremely tolerant to HMs present in the soil environment [ 6 , 7 , 8 ]. HM-hyperaccumulating plant species (such as Cannabis sativa , Sedum alfredii , and Pteris vittata ) can take up and store elevated concentrations of HMs without suffering from metal toxicity or cell damage [ 9 , 10 , 11 ]. Higher growth rates can lead to the production of large amounts of biomass, thus ensuring efficient phytoremediation [ 12 ]. However, most hyperaccumulators grow slowly under high HM concentrations, and thus the phytoextraction process of HMs is limited by the biomass [ 13 ]. Moreover, the viability of hyperaccumulators in different polluted environments has become a major issue in phytoremediation [ 14 , 15 ]. The efficiency of phytoextraction primarily depends on the translocation efficiency of HMs from the soil to the plants as well as the plant biomass. Arbuscular mycorrhizal fungi (AMF), which establish a symbiotic interaction with 90% of terrestrial plant species, are involved in the transport of HMs from the soil to the plants by establishing a direct link between the soil and the roots of the host plant [ 16 ]. The formation of arbuscular mycorrhizas in association with the AMF have shown that they can reduce HM toxicity in plants and promote plant growth. For example, the enhanced mycorrhizal colonization of Alfred stonecrop ( Sedum alfredii Hance ) with both Glomus caledonium and Glomus mosseae was found to be able to reduce the translocation of the HMs to the shoots by binding the HMs to the cell walls of the fungal hyphae [ 10 ]. Therefore, AMF can act as a filtration barrier against the transfer of the HMs to the plant shoots, which is critical to alleviate HM toxicity [ 17 ]. In addition, mycorrhizal association has also been found to improve nutrient uptake—particularly phosphorus (P), a macronutrient that often limits primary productivity in terrestrial ecosystems—indirectly by optimizing phytoremediation via promoting plant growth [ 18 , 19 , 20 ]. Recently, the inoculation of AMF in hyperaccumulators has been suggested to enhance the efficiency of phytoremediation. For instance, inoculation with Funneliformis mosseae significantly improved Solanum nigrum L. growth and phytoremediation efficiency in cadmium (Cd)-contaminated soil [ 21 ]. Black locust ( Robinia pseudoacacia L.), frequently found in HM-polluted areas and commonly colonized by AMF, is a promising candidate for phytoremediation [ 22 ]. Inoculation with AMF can immobilize lead (Pb) in the roots and stems of R. pseudoacacia and alleviate the toxic effects of Pb on root development in Pb-contaminated soils [ 23 ]. Moreover, elevated temperatures and carbon dioxide were found to promote the removal of HMs by R. pseudoacacia seedling roots associated with AMF in Cd-contaminated soils [ 24 , 25 ]. These studies showed that the symbiotic functions of AMF in the remediation of HM-polluted soils are greatly influenced by the environmental conditions. However, AMF-assisted phytoremediation in HM-contaminated soils varies according to the AMF species and growth substrate, and is dependent on plant–fungus–soil combinations [ 26 , 27 , 28 ]. HMs inhibit AMF spore germination, hyphal extension, and colonization, and the inhibition can be intensified with increasing HM levels and is dependent on the HM species [ 29 ]. It has been repeatedly observed that most soil ecosystems in abandoned mining areas are polluted with multiple HM species and have varying physical and chemical properties [ 30 , 31 ]. Following inoculation with AMF, various effects have been observed in phytoremediation. For example, maize ( Zea mays L.) has demonstrated a great variation in HM uptake in response to AMF colonization on different growth substrates; Funneliformis mosseae was found to be the most effective in maize development on recently discharged coal mine spoils and may be the most appropriate for the revegetation of this substrate, while Rhizophagus intraradices was determined to be the most beneficial in weathered and spontaneously combusted coal mine spoils [ 26 ]. To guarantee the feasibility of AMF-assisted phytoremediation, it is important to determine how AMF inoculation reacts to HM stress, and to explore the optimal AMF–plant–soil combinations in order to improve their efficiency. The aims of this study were to assess the effects of AMF on plant growth and HM uptake and accumulation using R. pseudoacacia grown in three substrates contaminated with different HM levels, and to explore the relationship between mycorrhizal colonization, plant biomass, and P and HM uptake by R. pseudoacacia . Here, we hypothesized that the impacts of AMF inoculation on the phytoremediation potential may be mediated by the interactions between the AMF species and the growth substrates. This study will provide experimental evidence for use of the pioneer species R. pseudoacacia in a symbiotic association with the appropriate AMF to remediate HM-contaminated soils.",
"discussion": "4. Discussion Mycorrhizal colonization is an important indicator for evaluating the establishment of a symbiotic relationship between AMF and a plant host [ 21 ]. In this study, symbiotic relationships were successfully established to varying degrees between the AMF species and R. pseudoacacia grown in three substrates with different pollution, physical, and chemical properties ( Table 2 ). Glomus intraradices and Glomus mosseae exhibited strong adaptability to HM-polluted environments, even though they were not isolated from metalliferous soils. However, the two AMF strains acted differently in S1, S2, and S3 ( Table 2 ), indicating the differing ecological adaptabilities of these two strains. In comparison with S3, characterized with the highest available P, the colonization rates of both AMF strains were found to be significantly higher in S1 and S2 ( Table 2 ). This phenomenon was deemed to mainly be due to the differences in P concentrations among the substrates, which may have caused varying plant physiological activities, such as rhizodeposition characteristics, which stimulate or suppress the activity of the AMF community and the potential of mycorrhizal colonization [ 38 ]. Phosphorus is one of the most limiting nutrients for plant growth and terrestrial ecosystem productivity [ 39 ]. For example, high P concentrations in mycorrhizal plants can lead to a higher RNA production to meet protein synthesis needs, resulting in improved plant growth rates [ 26 ]. The formation of arbuscular mycorrhizas in association with AMF is an essential strategy for host plant P acquisition to support growth. For example, AMF hyphae can release insoluble P, thereby contributing to enhanced P uptake [ 40 , 41 ]. However, the strains evaluated in this study showed no effect on the plant dry weights and height in the S3 soil sample ( Figure 1 and Table 2 ). A previous study demonstrated that a variety of factors may influence the beneficial effects of AMF on plant growth, such as AMF tolerance to HMs, drought, and nutrient limitation [ 31 , 42 ]. In this study, plant dry weight increased with the plant P concentration in S1 and S2 ( Figure 4 and Table 2 ), whereas it did not change in S3. This is possibly because the higher P availability in S3 than in S1 and S2 may eliminate plant P limitation as a result ( Table 1 ), and thereby eliminate the influence of mycorrhizal colonization on plant growth ( Figure 1 and Figure 3 ) [ 43 , 44 ]. On the other hand, the increased Pb content observed in S3 may constrain the colonization rates of both AMF strains ( Table 2 ). Thus, a combination of higher Pb and AP concentrations in S3 might be unfavorable for the application of AMF for phytoremediation. It has been widely reported that high HM contents in growth substrates can cause a higher plant phytotoxicity, resulting in a limited phytoremediation efficiency [ 45 ]. The beneficial effects of AMF on phytoremediation have often been related to the regulation of HM acquisition. Understanding such mechanisms is crucial for the optimal application of AMF during phytoremediation activities. AMF inoculation has been shown to both enhance and reduce HM uptake in plant tissues and, in some cases, have no effect on HM uptake which may depend on HM-contaminant levels [ 46 ]. Pb, which is one of the most frequent HMs found in the soil and nonessential for plants [ 47 , 48 ], is relatively stable in the soil [ 49 ]. AMF inoculation increased the root Pb concentration in S1 and S2 but decreased it in S3 ( Figure 5 d). This was probably because AMF alleviated the toxic effects of Pb on root development, improved root biomass in S1 and S2 ( Table 2 ), and further immobilized more Pb in the roots as a result [ 23 ]. Compared with S1 and S2, soil Pb concentrations were 7.44 and 4.91 times higher in S3, respectively ( Table 1 ). However, the uptake of Pb by the roots and shoots was significantly lower in S3 ( Figure 5 d and Figure 6 d). This phenomenon may have resulted from an excessive Pb contamination in S3, which could have induced more oxidative stress in the plant and AMF, thus being unfavorable to symbiosis formation, and resulting in a very low phytoremediation efficiency, regardless of inoculation [ 50 ]. Notably, considering the seriously negative effects of Pb contamination on the phytoremediation of R. pseudoacacia , it is therefore necessary to develop novel strategies, including the selection of AMF or plants which harbor a higher Pb accumulation potential and tolerance performance regarding to excessive Pb ions. In plant shoots, AMF inoculation decreased the concentration of Pb in S1 ( Figure 5 d). This finding was consistent with Sabra et al. [ 51 ], who also observed that the Pb concentration was significantly decreased in the shoots after colonization of plants with Rhizophagus irregularis or Serendipita indica . However, the Pb concentration was increased in AMF-inoculated shoots in S2. The different responses of the shoot Pb concentration to the substrate type may suggest that the AMF-assisted phytoremediation of Pb is substrate-specific. Similarly, the application of AMF markedly increased root Zn concentrations in S1 and S2 rather than S3, and both AMF-inoculated and uninoculated plant roots had higher Zn concentrations in S2 than in S1 ( Figure 5 c). This may be due to the large difference in Zn concentrations between S1 and S2 ( Table 1 ). Zn is easy to transfer during plant metabolism, and can exhibit toxic effects on the plant cells when the concentration exceeds a certain range [ 52 , 53 ]. In contrast to Pb, despite excessive Zn concentrations in S2, the application of AMF markedly increased root Zn concentrations in S2, and both AMF-inoculated and uninoculated plant roots had higher Zn concentrations in S2 than in S1 ( Figure 5 c). Furthermore, uninoculated roots had higher concentrations of HMs, particularly Zn, compared with inoculated roots in S3 ( Figure 5 ). These AMF species may be more suitable for the phytoremediation of Zn-contaminated soils than for the Pb-contaminated soils. In addition, organic matter (OM), a critical factor in controlling the sorption and sequestration of pollutants, was found to be significantly higher in S3, which could thereby buffer the negative effects of excessive Zn ions on AMF inoculation. However, AMF did not affect the shoot Zn concentration in R. pseudoacacia plants ( Figure 6 c). The retention of Zn by the R. pseudoacacia roots may be explained by a self-protection mechanism that prevents excess HMs from entering the stems and leaves, thereby reducing the toxic effects of HMs [ 54 ]. Soil with a Cd concentration exceeding 0.5 mg/kg is considered as contaminated with a high phytotoxicity [ 55 ]. Although the Cd concentration was increased in the AMF-inoculated roots in S1 and S2, the shoot Cd concentration was significantly increased in S1 but decreased in S2, suggesting that AMF-assisted phytoremediation was more suitable in S2 ( Figure 5 a and Figure 6 a). A previous study demonstrated that antioxidant activities in mycorrhizal plants were increased at lower Cd concentrations but decreased at higher Cd concentrations [ 21 ]. Cu, as a component of several enzymes, participates in various physiological metabolic processes, and exhibits an important impact on the growth and development of plants. In the present study, soil Cu concentrations were 319% and 293% higher in S2 and S3 than in S1, respectively ( Table 1 ). However, the application of AMF did not increase the Cu concentration in the roots in S2 or S3. This may be related in that soil Cu contamination will lead to the disordering of the plant metabolic process and interfere with the balance between the ions in plants, which leads to very little effects on the Cu uptake in plant roots in response to the application of AMF [ 56 ]. In this experiment, inoculation with AMF played an important role in facilitating HM accumulation by R. pseudoacacia grown in S1 and S2, especially in roots. Compared with S1 and S2, the root accumulation of HMs, particularly Pb and Zn, was found to be significantly lower in S3 ( Table 3 ). Although R. pseudoacacia grown in S3 had the highest shoot dry weight ( Table 2 ), the shoot HM accumulation was the lowest compared with S1 and S2 ( Table 4 ), suggesting that the difference in substrate greatly influenced the beneficial role of AMF in R. pseudoacacia phytoremediation. Therefore, it is important to select appropriate AMF species for phytoremediation in a specific environment. Furthermore, we must explore the role of AMF in relieving HM-induced phytotoxicity under HM stress to take the full advantage of their potential value in phytoremediation."
} | 4,497 |
26982030 | PMC4794241 | pmc | 823 | {
"abstract": "Pollinators, such as bees, often develop multi-location routes (traplines) to exploit subsets of flower patches within larger plant populations. How individuals establish such foraging areas in the presence of other foragers is poorly explored. Here we investigated the foraging patterns of pairs of bumble bees ( Bombus terrestris ) released sequentially into an 880m 2 outdoor flight cage containing 10 feeding stations (artificial flowers). Using motion-sensitive video cameras mounted on flowers, we mapped the flower visitation networks of both foragers, quantified their interactions and compared their foraging success over an entire day. Overall, bees that were released first (residents) travelled 37% faster and collected 77% more nectar, thereby reaching a net energy intake rate 64% higher than bees released second (newcomers). However, this prior-experience advantage decreased as newcomers became familiar with the spatial configuration of the flower array. When both bees visited the same flower simultaneously, the most frequent outcome was for the resident to evict the newcomer. On the rare occasions when newcomers evicted residents, the two bees increased their frequency of return visits to that flower. These competitive interactions led to a significant (if only partial) spatial overlap between the foraging patterns of pairs of bees. While newcomers may initially use social cues (such as olfactory footprints) to exploit flowers used by residents, either because such cues indicate higher rewards and/or safety from predation, residents may attempt to preserve their monopoly over familiar resources through exploitation and interference. We discuss how these interactions may favour spatial partitioning, thereby maximising the foraging efficiency of individuals and colonies.",
"introduction": "Introduction Understanding how foragers distribute themselves within and among resource patches is a central question in behavioural ecology. Historically it was usually assumed that individuals select and remain in the patches providing them with the highest rewards [ 1 , 2 ]. However, for most animals, foraging decisions are complicated by several additional factors such as an individual’s knowledge of their environment [ 3 ], the nutritional composition of foods [ 4 ], or their interactions with social partners, competitors and predators [ 5 , 6 ]. Pollinators, for instance, often exploit complex foraging areas comprised of multiple flower patches whose nectar rewards replenish over time [ 7 ]. Through repeated visits to familiar places, individuals accumulate knowledge about the location and the profitability of flower patches, enabling them to forage more efficiently than if they explored a novel environment each time [ 8 ]. In many species of bees [ 9 , 10 ], butterflies [ 11 ], hummingbirds [ 12 ] and nectarivorous bats [ 13 ], foraging individuals regularly revisit flower patches in stable, repeatable sequences called ‘traplines’. Recent studies on bumble bees collecting sucrose solution from artificial flowers (equivalent, in terms of nectar profitability, to natural flower patches) have begun to reveal how pollinators develop such movement patterns when foraging alone in highly predictable environments, by prioritizing visits to the most rewarding flowers while minimizing overall travel distances between them [ 14 – 16 ]. Whilst this is an important first step, none of these studies have yet captured the considerable additional variation in nectar rewards provided by flowers in field conditions due to the activity of other foragers competing for the same resources [ 17 – 21 ]. Previous attempts to address this question suggest that foragers avoid extensive spatial overlap so that each specializes on a different subset of flowers within larger plant populations. For instance, bumble bees tend to adjust the size of their foraging area in response to changes in the density of conspecific foragers, either by increasing the number of flower patches they visit following the removal of competing foragers [ 22 – 24 ], or by reducing the number of patches they visit after the introduction of new foragers [ 25 ]. Similar observations were made with nestmates and non-nestmates, suggesting that bumble bees do not discriminate kin from non-kin during foraging interactions [ 23 , 25 ]. In order to fully understand how these complex patterns of spatial resource partitioning develop over time, as bees learn to exploit their foraging environment, it has now become crucial to study the spatial movements of individual foragers, their behavioural interactions and the potential consequences of such interactions on their future foraging decisions over several consecutive foraging events. In principle, the presence of other foragers can have different effects on a bumble bee’s foraging decisions depending on its experience of the environment. Firstly, foragers can use social information, such as visual or olfactory cues inadvertently provided by conspecifics on flowers, to decide whether or not to visit flowers [ 26 , 27 ]. For instance, inexperienced bees discovering a new foraging environment tend to copy the flower choices of other foragers to identify the most rewarding flowers [ 28 ], whereas experienced individuals tend to avoid flowers occupied by conspecifics whose nectar reserves are probably depleted [ 29 ]. Secondly, when forager density is high, experienced bees may also try to preserve their foraging area by increasing their visitation rates to particular flowers (exploitative competition [ 18 , 21 ]) or by chasing potential competitors away (interference competition [ 17 , 19 , 20 ]). Since bumble bees typically enter a foraging environment at different times, depending on their age and the nutritional status of the colony [ 30 ], two foragers are unlikely to have the same knowledge about available foraging opportunities. We therefore hypothesize that these natural knowledge asymmetries among foragers have important consequences for their spatial foraging strategies and, ultimately, on resource partitioning. To explore this possibility, we analysed the foraging patterns of pairs of bumble bees with different levels of experience when exploiting a common array of artificial flowers in a large outdoor flight cage. In each pair, a resident (experienced) forager was allowed to perform 25 foraging bouts before a newcomer (inexperienced) forager was released. Using motion-sensitive cameras mounted on flowers, we mapped the network of flower visitation for each bee, recorded their interactions on flowers and compared their foraging success for another 25 foraging bouts.",
"discussion": "Discussion Bumble bees foraging simultaneously in a common environment adopted different strategies depending on their experience of the flower array. Residents, that had started to establish a foraging area several hours before the arrival of newcomers, continued to exploit familiar feeding sites by increasing their frequency of floral visits and evicting newcomers when they encountered them on flowers. In contrast, newcomers prioritized revisits to flowers from which they had successfully evicted residents and obtained a nectar reward, presumably to establish their own foraging area. Our results highlight significant spatial overlap between bee foraging areas, which may have emerged from this combination of exploitation and interference. In the absence of other foragers, bumble bees given exclusive access to multiple replenishing feeding sites tend to exploit a subset of these resources within larger foraging areas based on their spatial memories [ 10 ]. Over consecutive bouts, bees develop routes (traplines) enabling them to adjust their timing of revisits to feeding sites to enable nectar replenishment [ 14 ], prioritize visits to most rewarding sites [ 15 ] and minimize overall distances travelled between them [ 16 , 31 ]. Consistent with findings from a previous study on a different bumble bee species ( Bombus impatiens ) [ 24 ], we found that foraging experience confers a competitive advantage to bees; enabling experienced residents to visit more feeding sites, travel faster between them, and achieve greater foraging success than less well informed newcomers. Our analyses of individual movement patterns show how this home advantage diminishes with time, due to a sharp drop of the foraging success of residents, that suddenly lose their exclusive access to resources, and a gradual increase of newcomer success, as they discover flowers and start to include them in their developing foraging areas. Several lines of evidence indicate that the behavioural changes of residents are a direct consequence of the introduction of newcomers. Firstly, immediately following the start of the two-forager phase (bouts 26–30), resident bees made 42% longer foraging bouts, 253% more revisits to empty flowers per bout, resulting in 33% lower energy intake rates than when they foraged alone. The sharp drop in resident foraging performance at the start of the two-forager phase, followed by a stabilisation of foraging success shows markedly different dynamics to the typical gradual increase and stabilisation of foraging success observed in bees foraging alone in stable arrays of flowers (as described in the one-forager phase and in numerous other studies using similar experimental approaches [ 7 , 14 – 16 , 24 , 25 , 32 , 33 ]). Secondly, the sudden changes in resident foraging behaviour were accompanied by a significant shift in both the size and geometry of their foraging area. Bees with exclusive access to a stable array of flowers establish durable traplines to exploit selected feeding sites as long as these sites continue to provide enough resources [ 7 , 14 , 16 ]. When the array is perturbed, for instance because some flowers are experimentally moved [ 32 , 33 ], their relative rewards are changed [ 15 ], or competitors are added or removed [ 22 – 25 ], bees search for new feeding sites and modify the their established foraging areas. Therefore, sudden alterations of resident behaviour between the two experimental phases observed in our study cannot be explained by them simply accumulating more foraging experience, but are instead the result of the presence of newcomers. Interestingly, the foraging areas of residents and newcomers showed significant levels of spatial overlap notwithstanding that there were enough resources available for foragers to exploit different subsets of flowers. Spatial overlaps are not the consequence of random movements since flower visitation patterns were more similar between foragers within a pair than those of different pairs. The fact that resident bees continuously revisited the same flowers/areas throughout the two experimental phases also indicates that environmental heterogeneities within the flight cage (e.g. light, temperature, humidity, wind), which may have greatly fluctuated between the start (morning) and the end (evening) of the observations, had no apparent influence on space use by bees. Furthermore, we found no indication that any flower positions were more attractive than others as all pairs of bees tested focused their foraging areas around different subsets of flowers. Spatial overlaps are also unlikely to have emerged from a tendency for bees to follow one other, as foragers followed and joined each other on the same flowers in less than 6% of all recorded visits. Instead, our data show that bees engaged in competitive interactions over access to flowers. Video recording of all flower visits revealed that more than 90% of encounters on flowers involved physical interactions, during which bees pushed each other, resulting in the eviction of one contestant. In contrast to the overt aggressive events observed among bumble bee workers competing over reproduction in colonies that have passed the ‘competition point’ [ 43 ], interactions on flowers did not result in visible injuries or death, suggesting that bees attempted to monopolize nectar rewards rather than directly impair the long-term performance of a potential rival forager. Aggressive interactions on flowers have been previously reported between different bee species competing for limited nectar or pollen resources, for instance in bumble bees [ 17 ] and stingless bees [ 20 ]. However, we are not aware of previous reports about such interactions between conspecifics. Nieh [ 47 ] described aggressive interactions between honeybees ( Apis mellifera ) on abundant food resources that could accommodate up to 40 bees, an experimental context that was probably closer to hive robbing (when stronger colonies attack weaker hives to steal their honey stockpiles) than flower foraging. This explanation is unlikely for our experiments as they involved low forager densities and multiple flowers each providing small amounts of nectar when compared to ad libitum feeding conditions [ 47 ]. Although relatively rare (less than 4% of all flower visits), these encounters on flowers may have had critical consequences for bees’ subsequent foraging decisions, depending on their individual experience. Resident foragers initiated and won most interactions, indicating that they engaged in active defence of their foraging area. Such ‘prior-residence effect’ ( sensu [ 48 ]) on contest outcomes suggests that resident bees had a stronger motivation to contest and escalate competition due to the greater value they placed on these resources that are part of their foraging area, compared to less experienced newcomers [ 49 ]. In the rare cases when newcomers won interactions on flowers, residents increased their visitation rates to contested flowers in the subsequent foraging bout, potentially an attempt to discourage newcomers from revisiting flowers in future bouts by keeping nectar rewards low [ 12 ]. In contrast, newcomers increased their visitation rates to flowers from which they recently evicted residents. Presumably, the nectar rewards obtained after a successful eviction of residents reinforced newcomer motivation to exploit particular flowers and make it more likely they are included in their developing foraging area or trapline through simple associative learning [ 50 ]. Although our study involved interactions between nestmates, we are not aware of any evidence that bumble bees respond differently to nestmates compared to other conspecifics in a foraging context. Several recent studies indicate that bees equally use social information when deciding to either join or avoid conspecifics on flowers [ 26 , 27 , 28 , 29 ], be they closely related nestmates or foragers from different colonies, thus suggesting that similar results would be observed with foragers from different colonies. Moreover, if bees could recognise non-nestmates, we would expect competitive interactions to be even more pronounced in a situation involving multiple colonies. Nonetheless it would be useful to conduct similar experiments with pairs of workers from different colonies to explore whether the levels of interference competition and aggression depend on relatedness between foragers. It is likely that these competitive interactions on flowers were favoured (at least initially) by cues inadvertently provided by both competitors. Although we found no indication that bees visually followed one another to choose the same flowers, they had access to olfactory footprint cues that accumulated on visited flowers throughout the experiment [ 26 ]. Bees learn to associate these scent marks with reward levels experienced on flowers and develop different responses depending on context, so that the same mark can become attractant, neutral or repellent to a forager based on its past foraging success [ 51 ]. Since all foragers tested in our study were pre-trained in groups on a single flower delivering ad libitum nectar rewards before being tested, it is likely that they each associated the presence of scent marks (due to repeated visits by multiple bees from the colony) with flowers containing high rewards in areas relatively safe from predators. During the two-forager phase, newcomers may thus have initially been attracted to flowers already exploited and scent marked by residents, increasing the probability of encounters on flowers. However, after a few foraging bouts each bee may have adjusted its interpretation of scent marks based on individual experience of rewards from flowers. Overall our results suggest that the development of bumble bee foraging areas occurs through a combination of resource depletion and interference. These mechanisms have been proposed to explain resource partitioning among territorial animals competing for divisible spaces, such as large habitat patches [ 52 , 53 ]. Future studies on pollinator space use might be expanded to explore this hypothesis in experimental scenarios involving more foragers and from more colonies in various arrangements of flowers. Specifically, the frequency of the competitive interactions between foragers in the field, where nectar secretion rates of individual flowers are typically lower ([ 54 – 56 ] but see [ 57 ]) and the number of available flowers per bee may be larger than in our experimental conditions, remain to be confirmed. Potential behavioural differences among foragers from different colonies will also have to be examined [ 58 , 59 ]. How, or indeed whether, foraging interactions and their consequences on space use observed in pairs of bees might scale up to the level of colonies, populations and communities, and how they could shape patterns of pollination are exciting questions deserving future attention."
} | 4,417 |
31459509 | PMC6648443 | pmc | 824 | {
"abstract": "A water\ncontact angle greater than 150° together with a sliding\nangle less than 10° is a special surface phenomenon that appears\non superhydrophobic surfaces. In this paper, a brief introduction\nof the development history and present research on superhydrophobic\nsurfaces was given. Polymeric superhydrophobic surfaces with biomimetic\nhierarchical roughness were fabricated by a simple method of hot embossing\nwithout any chemical treatments. Stainless steel meshes with different\nmesh numbers were used as template. Moreover, the influences of processing\nparameters, including mesh number, mold temperature, and pressure,\nwere deeply investigated. Hierarchical microplatforms, microfibers,\nand oriented arrayed nanowrinkles structure on them, which were resembled\nwith the nanowrinkles structure and hierarchical roughness on a red\nrose petal, were observed by a scanning electron microscope. A water\ncontact angle of 154° can be achieved after parameter optimization.\nThe method proposed in this study offered a fine and affordable choice\nfor the fabrication of polymeric superhydrophobic surfaces. With the\nrapid development of functional applications in micro- and nanodevices,\nthis method will show greater superiority in large-area and large-scale\nproduction due to its advantages of low cost, high efficiency, and\nhigh reliability.",
"conclusion": "3 Conclusions In this paper, stainless steel meshes were applied as templates\nfor the preparation of polymeric superhydrophobic surfaces with biomimetic\nhierarchical roughness via isothermal hot embossing. Herein, the hierarchical\nroughness was composed of microplatforms, microfibers, and oriented\narrayed nanowrinkles formed during the demolding step. After parameter\noptimization, the optimal processing parameters for PE/EVA samples\nprepared by the hot embossing method were decided to be 94 °C\nand 6 MPa, while the best templates were stainless steel meshes with\na mesh number of 500. A water contact angle of 154° and an oil\ncontact angle of 97° can be achieved under the aforementioned\noptimal parameters. After further modifying using fluorosilane, the\noil contact angle would increase to 140° under the cooperation\nof hierarchical roughness and low surface energy. In addition, water\ndroplets were found to roll readily on the as-prepared surface tilted\n∼1°, which indicated an extremely low water sliding angle.\nIt was worth mentioning that the whole hot embossing process can be\nfinished within 20 s, and this method was also suitable for extra\nlarge (>feet 2 ) samples. By changing the size of polymer\nsubstrate, mesh template, and hot embossing equipment, superhydrophobic\nsurfaces with biomimetic hierarchical roughness can be even larger.\nTherefore, the method proposed in this paper was a simple and affordable\nway for the mass industrial production of polymeric superhydrophobic\nsurfaces.",
"introduction": "1 Introduction Compared with the pioneer\nresearch work around the turn of the\ncentury, 1 − 3 the fabrication and fundamental of superhydrophobic\nsurfaces have been extensively studied in recent years. 4 − 8 Superhydrophobic surfaces are typically defined as surfaces with\ncontact angles of water droplets greater than 150° and sliding\nangles less than 10°. This rapid expansion of research interest\nis because of their great application potential in the area of rain,\nsnow, or ice adhesion prevention; 9 , 10 drag reduction; 11 antibiofouling; 12 , 13 tunable isotropic\nor anisotropic wettability surfaces; 14 − 16 etc. The research of\nsuperhydrophobic surfaces is often inspired by natural examples, such\nas the water repellency and self-cleaning effect of plant leaves and\ninsect wings. 17 − 21 One of the most well-known examples is the so-called “lotus\neffect”, which can realize self-cleaning using rolling water\ndrops to remove pollutants and dust. 22 , 23 Basing on\nthe research on natural superhydrophobic phenomenon, artificial superhydrophobic\nsurface was more controllable under the combination of low-surface-energy\nmaterials or coatings and designed hierarchical roughness on both\nmicro- and nanoscales. 24 − 26 Numerous methods for artificial superhydrophobic\nsurfaces fabrication, including chemical composition control and fabrication\ntechniques, such as self-assembly, 27 , 28 spin coating, 29 electrospinning, 30 etching, 31 imprint lithography, 32 etc., have been developed to reduce surface\nenergy and generate hierarchical roughness. 2 , 33 − 36 Hierarchical roughness is a catch-all for all types of combined\nstructure ranging from microscale to nanoscale. 37 Numerous kinds of hierarchical structures, both natural\nand artificial, are conducive to the increase of contact angle toward\nsuperhydrophobic surfaces. Among the fabricating techniques listed\nabove, imprint lithography is often used in the preparation of superhydrophobic\nsurfaces with hierarchical structures because of its high precision,\nhigh fidelity, and simplicity. 20 , 38 Liu et al. reported\nan imprint lithography approach to transfer complex micro/nanostructures\ninto polymeric materials with an aluminum oxide mold. 39 Lee’s group prepared overhang structures using reverse\nnanoimprint lithography with poly(vinyl alcohol) transfer template.\nA fluoroalkylsilane monolayer coating was further performed to reduce\nthe surface energy to form superhydrophobic surfaces on silicon substrates. 40 Many biomimetic hierarchical structures, such\nas red rose petal, butterfly wing, gecko foot, and plant leaf structures, 41 , 42 have also been replicated by imprint lithography with either natural\nmaterial or artificially patterned molds. Beyond ordered hierarchical\nstructures, the spontaneous surface\nwrinkling technique 43 , 44 developed in recent years provides\nan alternative way to create textured structures on both micro- and\nnanoscales. Researchers around the world have made many efforts to\nfabricate hierarchical wrinkle and fold structures for the realization\nof controllable wetting characteristics. To date, external stress\nor strain under appropriate conditions is considered to be an effective\nway to fabricate hierarchically wrinkled surfaces. 45 Some other techniques such as plasma treatment, thermally\ninduced shrinkage, and imprint lithography are also applied combining\nwith spontaneous wrinkling for better controllability. Lee et al. 46 transformed flat polystyrene substrates into\nsuperhydrophobic hierarchically wrinkled surfaces by sequential wrinkling\nprocess. Plasma treatments followed by directional strain relief were\nperformed to control the nanowrinkle orientation and wavelength. Zhang\net al. 47 explored the formation mechanism\nof hierarchically wrinkled surfaces, which can be controlled between\nsuperhydrophobicity and superhydrophilicity by mechanical strain.\nThe relationship between different contact states and levels of hierarchical\nroughness was also discussed. Moreover, superhydrophobic surfaces\nwith hierarchical roughness, such as combinations of wrinkles with\nmicropillars, microplatforms, microfibers, etc., have been successfully\nfabricated and comprehensively studied. 48 − 50 Although the aforementioned\ntechniques could generate superhydrophobic\nsurfaces with hierarchical roughness, the preparation of hierarchically\nstructured superhydrophobic products still cannot meet the demands\nof large-scale industrialization (e.g., efficiently, massively, and\ncheaply). Therefore, it is worthwhile to develop efficient, controllable,\nand affordable techniques to fabricate optimized hierarchically structured\nsurfaces and further promote the research depth on their hydrophobic\ncharacteristics. In this paper, we select the hot embossing\nmethod, which can be\nregarded as a special type of imprint lithography, to fabricate polymeric\nsuperhydrophobic surfaces with biomimetic hierarchical structures.\nHierarchical microplatforms, microfibers, and oriented arrayed nanowrinkles\nstructures, which are similar to the wrinkled surfaces of red rose\npetal, are generated on polymer substrates under precisely controlled\ntemperature and pressure. As a kind of top-down approach, the hot\nembossing method has the advantages of easy handling, high efficiency,\nhigh fidelity, and low cost. 51 , 52 Extra large (>feet 2 ) products with superior hydrophobic performance can be obtained\nwithin 1 min with no chemical treatment. The application of commercial\npolymer substrate and stainless steel mesh template makes the fabrication\nof superhydrophobic surfaces really repeatable and affordable. The\nsize of products with biomimetic hierarchical roughness can be even\nlarger when larger polymer substrate, mesh template, and hot embossing\nequipment are used. Therefore, this method is ideal for mass industrial\nproduction of polymeric superhydrophobic surfaces.",
"discussion": "2 Results and Discussion 2.1 Formation and Evolution\nof the Biomimetic\nHierarchical Roughness The surface morphology of final product\nwas determined by the combination of stainless steel mesh and the\nprocessing parameters of hot embossing. The diameter of the wire mesh\nand mesh size, which are influenced by the mesh number, would directly\naffect the size of the microplatform structure in final product. Meanwhile,\nthe processing parameters (e.g., embossing temperature and pressure)\nwould greatly influence the formation and evolution of microfiber\nand nanowrinkle structures. The specifications of stainless steel\nmesh had already been standardized. Here, meshes with mesh numbers\nof 300, 400, 500, 800, 1000, and 1500 were selected for hot embossing\nexperiments. Figure 1 a shows the scanning electron microscopy (SEM) image of the pristine\nstainless steel mesh with a mesh number of 300. Figure 1 (a) SEM image of the\npristine stainless steel mesh with a mesh\nnumber of 300; (b, c) the as-prepared surfaces using same template\nshowing totally different morphologies under varying processing parameters;\n(d) regularly arranged nanowrinkles on microfiber structures; (e)\nSEM image of the nanowrinkles on red rose petal; and (f) the formation\nand evolution process of the special biomimetic hierarchical structures. Two SEM images of the as-prepared\nsurfaces using same template\nare presented in Figure 1 b,c. Although the template of these two samples was the same, different\nprocessing parameters, especially different embossing temperatures,\nled to totally different morphological characteristics. When the embossing\ntemperature was lower than the melting temperature ( T m ) of ethylene vinyl acetate (EVA) and polyethylene (PE),\nthe structure of the surface was mainly microplatforms, which were\nbasically of the same size with the meshes (as shown in Figure 1 b). The microplatform structures\ncan be regarded as notches of the stainless steel meshes on polymer\nsubstrates caused by mechanical compression and embedding. With the\nincrease of embossing temperature, polymer substrate (especially the\nEVA phase) would have stronger deformability and mobility, which led\nto total variation of the microplatform structures. Figure 1 c presents the morphology of\nsample prepared under embossing temperature several degrees higher\nthan in Figure 1 b.\nObviously, almost all of the microplatforms changed into microfibers.\nMore details on the surface of microfibers can be found under higher\nmagnification (30 K). As shown in Figure 1 d, nanowrinkle array with a cycle of ∼100\nnm, which was very similar to the nanowrinkle structures on red rose\npetal ( Figure 1 e),\nwas regularly arranged along the axial direction of microfiber. The\nnanowrinkles together with microplatforms and microfibers formed by\nthe hot embossing process constituted the needed hierarchical roughness\nfor superhydrophobic surfaces. The formation and evolution process\nof this special biomimetic hierarchical structure is revealed in Figure 1 f. First, the stainless\nsteel mesh and PE substrate were prepared under room temperature and\nthen put into the hot embossing device together. Second, they were\nembossed and held for a certain time under predefined embossing temperature\nand pressure. During this step, the mesh would be gradually pressed\ninto PE substrate and the morphology should be similar to the one\npresented in Figure 1 b. After that, the demolding step (peeling the mesh off PE substrate)\nwas performed immediately in the case of incomplete cooling. To get\nperfect microfibers and nanowrinkles on them, the embossing temperature\nshould be set at certain values that are slightly higher than the T m of EVA and lower than the T m of PE. On account of the strong deformability and mobility\nof melting EVA, the gaps of stainless steel mesh could be fully filled,\nand the polymer (both EVA and PE) attached on net wires would be stretched\nand turned into the microfibers in Figure 1 c, while unmelted PE phase held the whole\nsubstrate together and kept it from falling apart. During the stretching\nprocess, the surface layer (outer layer) of the formed microfibers\nwas cooled down rapidly by air; however, the temperature of the inner\nlayer still kept at a high level. Thus, different deformabilities\nbetween the outer and inner layers of microfibers would lead to continuous\nmelt fracture phenomenon and form the axially arranged nanowrinkles\nin Figure 1 d. The final\nmorphology of nanowrinkles was also significantly influenced by the\nnonuniform shrinkage and creep after fracture of microfibers or detaching\nfrom net wires. In this way, superhydrophobic surfaces with biomimetic\nhierarchical roughness were finally obtained. The whole hot embossing\nprocess can be done within 20 s, representing a relatively high efficiency\nfor the preparation of superhydrophobic surfaces. Although many\nresearchers had reported their methods for the preparation\nof biomimetic nanowrinkles and hierarchical roughness, the method\nwe proposed in this paper was undoubtedly one of the most efficient\nand affordable ones. All our method required was a piece of commercialized\nscreen mesh, polymer substrates, and suitable temperature and pressure\nconditions. No chemical treatment and expensive equipment were needed. 2.2 Influence of Different Parameters on Surface\nMorphology and Hydrophobic Performance Different parameters,\nespecially mesh number, embossing temperature, and pressure, had significant\nimpacts on the surface morphology and hydrophobic performance of final\nproducts. In this section, their influences will be investigated and\ndiscussed systematically to find out the optimal processing parameters.\nFurthermore, the relationship between surface morphology and hydrophobic\nperformance will also be explored. Figure 2 shows the influence of templates (stainless\nsteel meshes with different mesh numbers) on surface morphology and\nhydrophobic performance of the as-prepared samples. In this figure,\n(a)–(c) represent the characterization results of water contact\nangle and morphologies on micro- and nanoscales, while 1–6\ndenote the variation of mesh numbers from 300 to 1500. Therefore, Figure 2 a1 here represents\nthe water contact angle of the as-prepared sample using stainless\nsteel mesh with a mesh number of 300 as template. Similarly, the SEM\nimage named Figure 2 b2 represented the sample morphology on microscale (microplatforms\nand microfibers) using stainless steel mesh with a mesh number of\n400 as template, and Figure 2 c3 represents the sample morphology on nanoscale (regularly\narranged nanowrinkles on microfibers) using stainless steel mesh with\na mesh number of 500 as template. The rest of the pictures can be\nnamed and explained in the same manner. Figure 2 (a) Water contact angles\nand morphologies on (b) microscale and\n(c) nanoscale of the as-prepared samples using stainless steel meshes\nwith different mesh numbers as templates. (d) Relationship between\nmesh number and hydrophobic performance of the as-prepared samples.\nThe scale bars in the insets of (b) and (c) are 200 and 1 μm,\nrespectively. The embossing temperature\nand pressure for all of the as-prepared\nsamples presented in Figure 2 were 94 °C and 6 MPa, respectively. According to the\nmeasurement results, the contact angle of untreated PE/EVA composite\nsubstrates was ∼93°, whereas that of the as-prepared samples\nhad been enlarged in different degrees (as shown in Figure 2 d). The contact angle of the\nas-prepared sample using stainless steel mesh with a mesh number of\n300 as templates was 148°. This contact angle would increase\nwith increasing mesh number and reach a maximum of 154° when\nthe mesh number was 500. After that, further increase of mesh number\nwill lead to opposite effects and sharply reduce the contact angle\nto around 109° (mesh number = 1500). The increase of contact\nangle in the first stage was attributed to the size reduction of microstructures\non PE/EVA substrate, which was determined by the decreasing mesh size\nwith larger mesh numbers. Based on the Wenzel–Cassie model,\nthe increase of surface roughness was beneficial to enhance the hydrophobic\ncharacteristic of the as-prepared samples, while higher surface roughness\ncan raise the proportion of air phase in solid–liquid–air\ninterface. Larger mesh number would lead to higher surface roughness\nduring the hot embossing process (as shown in Figure 2 b1–b3) before the mesh number reached\n500. Higher surface roughness would keep more air bubbles (on both\nmicro- and nanoscales) when water droplets dropped onto polymer surfaces.\nThe wetting of areas under air bubbles, especially the wetting of\nbubbles between each nanowrinkle, was extremely difficult owing to\nthe effect of surface tension. Further, the dense air bubbles within\nhierarchical structures would form a layer of air cushion between\nwater droplet and polymer surface, which finally created a polymeric\nsuperhydrophobic surface with durable performance. In Figure 2 b3, the most compact microfiber\nstructure was obtained. Further, oriented arranged nanowrinkle structures\ncould be observed on the surface of microfibers in each sample at\nmuch higher magnification (30 000×). Compact nanowrinkles\nwith higher amplitudes were available when the mesh numbers were 400\nand 500 (as shown in Figure 2 c2, 2 c3, respectively). The combination\nof compact microfibers and convex nanowrinkles on them ultimately\nled to superior hydrophobicity of the as-prepared sample in Figure 2 a2,a3 (150 and 154°,\nrespectively). Then, in the second stage, further increase of the\nmesh number would lead to decline of contact angle. In most cases,\nwhen the mesh number was higher than 500, the whole stainless steel\nmesh would be composed by several layers (with lower mesh number)\nto reduce the relative size of the aperture. The overlap between different\nlayers would increase the entering difficulty of melting polymer and\nshow negative effects on the formation of hierarchical roughness.\nAs shown in Figure 2 b4–b6, the size of microstructures on PE/EVA substrates did\nnot decrease but became more messy and irregular with the increase\nof mesh number above 500. The amplitude and compactness of nanowrinkles\nin Figure 2 c4–c6\nalso presented obvious downtrends. Furthermore, the water adhesion\nof the surface with a contact angle\nof 154° was investigated by droplet rolling test. A 10 μL\nwater droplet was suspended from a microsyringe in air. Figure 3 shows that the water droplet\nreadily rolled on the as-prepared surface tilted ∼1°,\nwhich indicated that the water sliding angle was extremely low. Thus,\ncomprehensive analysis of the results of water contact angle, sliding\nangle, and SEM image was conducted, and the optimal mesh number of\nthe template used for the preparation of superhydrophobic surfaces\nwith biomimetic hierarchical roughness should be around 500. Figure 3 Water droplet\nrolling on the as-prepared PE surface tilted ∼1°. After confirming the optimal mesh\nnumber for templates, further\ninvestigation on the influence of embossing temperature on surface\nmorphology and hydrophobic performance was carried out. The mesh number\nof template and pressure for all of the as-prepared samples presented\nin Figure 4 were 500\nand 6 MPa, respectively. When the embossing temperature was just reached\nor lower than 92 °C, rare microfibers can be found after embossing\n(as shown in Figure 4 b1,b2) according to the poor deformability. Similarly, the formation\nof nanowrinkles was also restricted by the limited deformability,\nwhich led to irregular arrangement and low amplitude of nanowrinkles\n( Figure 4 c1,c2). As\nmentioned above, the biomimetic hierarchical roughness was composed\nof microplatforms, microfibers, and nanowrinkles, and undesirable\ncombinations of the micro- and nanostructures eventually led to poor\nhydrophobicity (116 and 120° in Figure 4 a1,a2, respectively). After the embossing\ntemperature rose up to around or above T m of EVA, normally formed microplatforms, dense microfibers ( Figure 4 b3,b4) with a length\nof ∼100 μm and compact convex nanowrinkles ( Figure 4 c3,c4) with relatively\nhigh amplitude (ca. 100–150 nm) together constituted the perfect\nhierarchical roughness and sharply rose the water contact angle up\nto more than 150°. The contact angles of the as-prepared samples\nwith embossing temperatures of 94 and 96 °C were 154 and 153°\n( Figure 4 a3,a4), respectively.\nThat is, 94–96 °C can be recognized as the optimal range\nof embossing temperature for the preparation of superhydrophobic surfaces\nwith hierarchical roughness. However, the optimal range of embossing\ntemperature was so narrow that further increase would lead to undesirable\nmorphologies ( Figure 4 b5,c5) and sharp reduction of contact angle (122° in Figure 4 a5). Figure 4 (a) Water contact angles,\nand morphologies on (b) microscale and\n(c) nanoscale of samples prepared by isothermal hot embossing with\ndifferent embossing temperatures. (d) Relationship between embossing\ntemperature and hydrophobic performance of the as-prepared samples.\nThe scale bars in the insets of (b) and (c) are 200 and 1 μm,\nrespectively. A similar research process\nwas also applied to find the optimal\npressure. Figure 5 presents\nthe results of water contact angle and morphology of the as-prepared\nsamples. The mesh number of template and embossing temperature applied\nin the hot embossing process were 500 and 94 °C, respectively.\nAs shown in Figure 5 d, the optimal pressure was around 6 MPa and the water contact angle\ncan be higher than 150° ( Figure 5 a2). The most important influence factor in this experimental\ngroup was the forming results of microplatforms and microfibers on\naccount of the same temperature condition and the consequent same\nforming ability of nanowrinkles. Thus, although the forming results\nof nanowrinkles in Figure 5 c1 were good enough, the hydrophobicity of the corresponding\nsamples was not that good (129° in Figure 5 a1) due to the sparse microfibers (as shown\nin Figure 5 b1) and\nthe consequent imperfect hierarchical roughness. This could also explain\nthe undesirable hydrophobicity of the as-prepared samples fabricated\nwith 10 MPa pressure (132° in Figure 5 a4). Majority of the microfibers would be\nsnapped during the demolding step as too much melting polymer went\nthrough the template under such high pressure and formed a “mushroom-like”\nstructure during the embossing step. Figure 5 (a) Water contact angles, and morphologies\non (b) microscale and\n(c) nanoscale of samples prepared by isothermal hot embossing with\ndifferent pressures. (d) Relationship between pressure and hydrophobic\nperformance of the as-prepared samples. The scale bars in the insets\nof (b) and (c) are 100 and 1 μm, respectively. Overall, the optimal processing parameters for\nthe preparation\nof PE/EVA superhydrophobic surfaces with biomimetic hierarchical roughness\nshould be 94 °C and 6 MPa. The best templates were stainless\nsteel meshes with a mesh number of 500. In this way, perfect biomimetic\nhierarchical roughness can be obtained, and the whole preparation\ncycle of isothermal hot embossing can be as short as 20 s. With the\nadvantages of low equipment requirement and easy operation of isothermal\nhot embossing, the preparation method proposed in this paper was doubtlessly\na simple and affordable way to achieve polymeric superhydrophobic\nsurfaces. Furthermore, it also showed greater superiority in large-area\nand large-scale production due to its advantages of low cost, high\nefficiency, and high reliability. Besides hydrophobicity, the\nlipophobicities of the as-prepared\nsamples before and after hot embossing were also measured and compared\nto demonstrate the promoting effect of biomimetic hierarchical roughness\non wettability. Since the surface tension of oil was lower than that\nof water, the promotion of lipophobicity was much harder than that\nof hydrophobicity. As shown in Figure 6 a,b, water contact angles of the as-prepared samples\nbefore and after isothermal hot embossing under the aforementioned\noptimal parameters were 93 and 154°, respectively. However, the\noil contact angles of the same samples were 26 and 97°, respectively\n( Figure 6 c,d). Although\nthe promotion of oil contact angles from 26 to 97° was an obvious\none, it still cannot meet the usage requirements in most situations\n(e.g., self-cleaning and antioil). A much higher oil contact angle\nof 140° can be obtained after modifying the PE/EVA surface with\nfluorosilane (as shown in the inset of Figure 6 d). The hierarchical roughness and low surface\nenergy of modified samples had worked together to create the significant\npromotion of oil contact angle from 26 to 140°. Figure 6 Comparison of (a, b)\nhydrophobicity and (c, d) lipophobicity of\nPE samples before and after hot embossing. The inset presents the\nlipophobicity of PE sample after hot embossing and fluorosilane modification."
} | 6,431 |
28666005 | PMC5493405 | pmc | 825 | {
"abstract": "Corals in the Arabian/Persian Gulf (PAG) survive extreme sea temperatures (summer mean: >34°C), and it is unclear whether these corals have genetically adapted or physiologically acclimated to these conditions. In order to elucidate the processes involved in the thermal tolerance of PAG corals, it is essential to understand the connectivity between reefs within and outside of the PAG. To this end, this study set out to investigate the genetic structure of the coral, Platygyra daedalea , and its symbiotic algae in the PAG and neighbouring Gulf of Oman. Using nuclear markers (the ITS region and an intron of the Pax-C gene), this study demonstrates genetic divergence of P . daedalea on reefs within the thermally extreme PAG compared with those in the neighbouring Gulf of Oman. Isolation by distance of P . daedalea was supported by the ITS dataset but not the Pax-C intron. In addition, the symbiont community within the PAG was dominated by C3 symbionts, while the purportedly thermotolerant clade D was extremely rare and was common only at sites outside of the PAG. Analysis of the psbA ncr indicates that the C3 variant hosted by P . daedalea in the PAG belongs to the newly described species, Symbiodinium thermophilum . The structuring of the coral and symbiont populations suggests that both partners of the symbiosis may contribute to the high bleaching thresholds of PAG corals. While limited gene flow has likely played a role in local adaptation within the PAG, it also indicates limited potential for natural export of thermal tolerance traits to reefs elsewhere in the Indian Ocean threatened by climate change.",
"introduction": "Introduction Coral reefs have undergone global decline in recent decades, often as a result of bleaching events, where a breakdown of the symbiosis between the coral hosts and their algal partners is associated with declines in coral health and survival. Mass bleaching typically occurs under elevated sea temperatures, with increases of only 1–2°C above the normal maximum often resulting in widespread mortality of corals throughout the tropics [ 1 ]. Given that tropical sea temperatures are predicted to rise by 0.5–4°C by the end of the century [ 2 ], the maintenance of present-day coral reefs will depend on the ability of the holobiont to acclimate or adapt to future temperature increases. Both the algal symbiont and the coral host have been proposed to play a role in thermal tolerance. Different strains of algal symbionts vary in their sensitivity to thermal stress, with thermally tolerant clades typically common in high-temperature coral habitats [ 3 , 4 ]. With regard to the host, studies have demonstrated that differences in host thermal tolerance are correlated with genetic divergence and differences in gene and protein expression under different thermal regimes [ 5 , 6 ]. In reality, both partners in the symbiosis likely play an interactive role in the thermal tolerance of the holobiont [ 7 ], suggesting that studies on thermal adaptation must consider the symbiont and host genotypes. Corals that exist in contemporary extreme environments represent useful models for understanding the potential for acclimation or adaptation to future temperature increases [ 8 ]. Corals in the Arabian/Persian Gulf (hereafter ‘the PAG’) are exposed to extreme summer sea temperatures compared with other tropical regions, with temperatures exceeding 34°C for several months annually and summer maxima >36°C [ 9 – 11 ], and therefore have the potential to provide important insights into how corals respond to thermal stress. Although numerous studies have demonstrated that the PAG is biogeographically unique in terms of community structure (e.g. [ 12 ]), few studies have explored the molecular mechanisms that allow corals to persist in this environment. Given that PAG reefs are relatively geographically isolated from the Indian Ocean by the narrow (42 km wide) Strait of Hormuz, yet are relatively young (present-day shorelines reached ~6000 years ago [ 13 ]), it is unclear whether there has been sufficient isolation to support genetic adaptation to the local environment. The aim of this study was to assess the genetic structure of both the coral host and the symbiotic algae of the locally abundant and pan-tropically distributed coral, Platygyra daedalea . By comparing the genetic structure of the holobiont collected from sites within the PAG, with those in the neighbouring Gulf of Oman (including the Straits of Hormuz), where temperatures are more benign (mean monthly maximum <31.5°C), this study aims to elucidate whether differences in thermal tolerance are linked to genetic divergence in either the host or symbionts. Such findings would have important implications regarding the potential for natural exchange of thermal tolerance traits with Indian Ocean reefs threatened by future climate change.",
"discussion": "Results & discussion The role of the host in the thermal tolerance of PAG holobionts has, to date, been unexplored. The results presented here from the host ITS marker indicate that there is strong genetic structuring between local populations in the PAG and Gulf of Oman. The haplotype frequencies ( Fig 1 ) and pairwise ɸ ST comparisons ( Table 1 ) from the ITS marker demonstrate differentiation between subpopulations from the southern PAG (Delma and Saadiyat) and the Gulf of Oman (Fujairah and Muscat) and revealed isolation by distance (IBD) occurring in this system (Mantel test, r = 0.60, p = 0.026). As differences between the PAG and Gulf of Oman could arise from the presence of cryptic species, haplowebs were used to visualise “fields for recombination” (FFRs) [ 14 , 15 ] ( S1 and S2 Figs). This analysis did not reveal the presence of cryptic species in our dataset. The reduced gene flow observed between conspecific Platygyra populations in the southern PAG and Gulf of Oman has important implications, particularly as the ITS marker has previously shown panmixia of coral populations over a comparatively much larger scale of thousands of kilometres [ 16 ]. As limited gene flow promotes local adaptation [ 17 ], the strong genetic structuring observed in this marker supports the assertion that these coral hosts have adapted to the PAG’s extreme environment [ 18 ]. Nevertheless, such restricted gene flow suggests that substantial direct export of the acquired thermal tolerance traits of P . daedalea from the southern PAG to the wider Indian Ocean is unlikely. As a consequence, PAG corals may have a limited capacity to naturally enhance the resilience of Indian Ocean reefs threatened by future temperature increases associated with climate change. 10.1371/journal.pone.0180169.g001 Fig 1 Genetic structure of Platygyra daedalea in the PAG and Gulf of Oman. Upper panel : Mean monthly averaged maximum SST, 2004–2014. Lower panel : Host haplotype frequencies. Each colour represents a different haplotype and white sectors indicate private haplotypes. SH = Strait of Hormuz. 10.1371/journal.pone.0180169.t001 Table 1 Pairwise ɸ ST comparisons between sites for the ITS region ( below diagonal ) and PAX-C intron ( above diagonal ). DEL SAD RAK MDM FUJ MCT DEL -0.010 0.051 -0.023 0.016 0.051 SAD 0.051 0.021 -0.014 0.035 0.132 RAK 0.052 0.008 0.052 0.120 0.239 MDM 0.124 0.113 0.026 -0.007 0.087 FUJ 0.297 0.260 0.105 0.071 0.150 MCT 0.230 0.172 0.038 0.034 -0.027 Bold values indicate significance at p<0.05 level. In agreement with the ITS marker, the results from the host Pax-C intron also indicated some structuring between sites inside the PAG and those in the Gulf of Oman, although there are some inconsistencies in the overall pattern. While there was agreement with structuring between the southern-most Gulf of Oman sites (Fujairah and Muscat) and the more proximal PAG sites (Ras al Khaimah and Saadiyat), it was not evident between the two most distant sites in this study (Muscat versus Delma). This was not consistent with the geography and oceanography of the region. Consequently, IBD is not evident from the PAX-C dataset (Mantel test, r = -0.17, p = 0.323). It was postulated that the discrepancies resulted from this type of marker, since nuclear intron markers such as Pax-C have frequently failed to resolve patterns observed in morphological features and other genetic markers, including the ITS region (e.g. [ 19 , 20 ]). Although mitochondrial markers could provide an alternative to nuclear intron markers, they typically lack the resolution for intraspecific studies due to the slow evolution of the mitochondrial genome in anthozoans [ 21 , 22 ]. The incongruence between the markers highlights the limitations of traditional markers as population genetic parameters vary across the genome. However, more representative estimates can now be obtained using approaches such as RADseq, which benefit from genome-wide marker coverage. The symbiont communities associated with P . daedalea showed clear structuring ( Fig 2A ). There were significant differences in the abundances of different ITS2 types hosted by corals (Fisher-Freeman-Halton Exact, p = 0.000) between the PAG and Gulf of Oman. Corals from the Gulf of Oman were largely dominated by clade D derivatives whereas corals from the PAG were associated with C3 symbionts. These results were unusual as it would be expected that clade D symbionts, widely regarded as thermotolerant, would be abundant in the world’s warmest reefs, and since previous work has shown that clade D symbionts become more dominant over C3 types in Platygyra populations as temperature increases along a thermal gradient in Taiwan [ 23 ]. In north-eastern Arabian reefs, Platygyra populations tend to be dominated by either D variants or C3 symbionts. Clade D dominated populations associate with the more benign environments such as the Gulf of Oman, Iranian and Saudi Arabian reefs [ 3 , 4 ], while the more thermally extreme reefs of the southern PAG host C3 symbionts [ 10 , 18 , 24 – 27 ]. The distinct D or C3 dominated communities were clearly exemplified by the sharp switch from D dominated communities in the Musandam peninsula, Strait of Hormuz, to C3 dominated reef (Ras Al Khaimah) that is situated only 35km away. Although Musandam and Ras Al Khaimah are close and experience similar summer thermal maxima, they are situated at the transition between the Gulf of Oman and the PAG and therefore, environmental variables other than thermal maxima, such as salinity [ 28 ] may be influencing the community structure. The clear separation of D and C3 communities contrasts the gradient observed in Porites symbioses [ 18 ], where there is a more gradual transition between the two symbiont types. As Porites spp. are vertical transmitters, there is the potential for transport of symbionts with the host larvae and this may increase the dispersal potential of C3 in Porites spp, relative to the C3 in the horizontally transmitting P . daedalea . The clear boundary between the C3 and D dominated communities in P . daedalea is in contrast to the IBD pattern observed in the host ITS marker and could reflect different processes affecting the distribution of the coral larvae and the algal symbionts. 10.1371/journal.pone.0180169.g002 Fig 2 Characterisation of symbiont communities in the southern PAG and Gulf of Oman. (A) Bar charts indicate the proportion of colonies hosting each ITS2 type or combination of ITS2 types. Vertical separation within the bars indicates colonies where more than one symbiont type was present. (B) Radial phylogenetic tree depicting the relationships between C3 variants found in P . daedalea from PAG reefs, C3 variants from Porites spp . within the region and other C3 variants found elsewhere. The tree was generated by Bayesian inference analysis of the psbA ncr region using the alignment generated in a previous study [ 18 ]. The branches containing samples from this study are shown in red with labelled arrows indicating the location of the samples (D: Delma; S: Saadiyat; R: Ras al Khaimah). The accession numbers for samples used to generate the tree are shown in S3 and S4 Tables. The ITS2 type C3 found in the PAG have recently been described as a new species [ 25 ], Symbiodinium thermophilum , which is genetically distinct from other C3 variants, and explains the unusual occurrence of a purportedly sensitive symbiont in the thermal extreme PAG [ 24 ]. Analysis of the psbA ncr sequences from the P . daedalea C3 symbionts in the Gulf demonstrated they are more closely related to S . thermophilum than other C3 variants, and revealed further diversity within this newly described lineage ( Fig 2B ). The apparent ubiquity of S . thermophilum among species in the southern PAG supports previous assertions that it may be important in the persistence of corals in this region [ 25 , 26 ]. The differences observed in the structuring of coral-symbioses in the PAG and Gulf of Oman are likely to be driven by the oceanographic conditions. While the extremes in salinity and summer temperatures exert strong selective pressures on the corals and their symbionts, the circulation patterns may also impact the population structuring and connectivity. As P . daedalea is a broadcast spawner, dispersal will be strongly driven by hydrodynamic conditions. The water entering the PAG travels as a surface current and proceeds north along the Iranian coast, while the outflow from the PAG travels as a subsurface current [ 29 ]. The reduced direct mixing of Gulf of Oman waters with the southern PAG, subsurface outflow preventing transport of buoyant larvae, and long residence times of PAG waters (>3 years; [ 30 ]) may limit exchange between corals outside and inside the Gulf. Nevertheless, studies on Diadema setosum have shown comparatively lower structuring (Fst = 0.05) between PAG and Gulf of Oman sites than was observed for the P . daedalea ITS marker [ 31 ]. Furthermore, the circulation patterns evident at the Strait of Hormuz may help explain why the Musandam and Ras Al Khaimah symbiont communities are distinct, as the Musandam reef is strongly influenced by the Gulf of Oman whereas Ras Al Khaimah is found outside of the inflow limits [ 29 ]. Considering the additional thermal stress associated with predicted rising sea surface temperatures, there is increasing interest in understanding the processes that lead to survival of corals in existing extreme reefs. The data presented here suggest that survival in the world’s warmest reefs likely involves both partners of the symbiosis. Previous work has documented responses associated largely with only one of the partners [ 5 , 6 , 23 ]. However, the ITS data indicates that P . daedalea in the PAG exhibits structuring of both the host population and the symbiont communities, although further work is required to ascertain the relative contribution of the different environmental factors to the structuring. While temperature and salinity are major factors controlling the coral community composition in the north-eastern Arabian peninsula, other environmental factors such as chlorophyll a concentrations have an influence [ 12 ] and could also impact corals and their symbionts at the population scale. Nevertheless, the data presented here provide a platform on which the genomic basis of thermal tolerance and responses to other environmental factors can be explored in both the host and symbiont."
} | 3,865 |
24988360 | PMC4119280 | pmc | 826 | {
"abstract": "Certain environmental parameters are accessible to cells only indirectly and require an encoding step for cells to retrieve the relevant information. A prominent example is the phenomenon of quorum sensing by microorganisms, where information about cell density is encoded by means of secreted signaling molecules. The mapping of cell density to signal molecule concentration and the corresponding network modules involved have been at least partially characterized in many bacteria, and vary markedly between different systems. In this study, we investigate theoretically how differences in signal transport, signal modification, and site of signal detection shape the encoding function and affect the sensitivity and the noise characteristics of the cell-density-encoding process. We find that different modules are capable of implementing both fairly basic as well as more complex encoding schemes, whose qualitative characteristics vary with cell density and are linked to network architecture, providing the basis for a hierarchical classification scheme. We exploit the tight relationship between encoding behavior and network architecture to constrain the network topology of partially characterized natural systems, and verify one such prediction by showing experimentally that Vibrio harveyi is capable of importing Autoinducer 2. The framework developed in this research can serve not only to guide reverse engineering of natural systems but also to stimulate the design of synthetic systems and generally facilitate a better understanding of the complexities arising in the quorum-sensing process because of variations in the physical organization of the encoder network module.",
"conclusion": "Outlook and conclusions Dedicated experimental studies aimed at elucidating the encoding process are still scarce. By comparison with the quite impressive efforts that have gone into the development of experimental set-ups for quantitatively studying the decoding process, the approaches available for elucidating the encoding process might still be considered rather primitive. Most measurements rely on bioassays obtained from growing cultures, where the molecular concentrations may differ from the steady-state values ( 27,50,53–55 ), WHICH report on the average encoding behavior only and which almost exclusively measure the extracellular concentration only. Developing assays that are capable of quantitatively reporting on the intracellular concentrations of SMs to allow us to probe encoding systems using intracellular receptors, or measuring fluctuations in signal concentrations under steady-state conditions represent important experimental challenges for the future. Measurements of noise levels might be particularly interesting, as noise has become a powerful source of information that facilitates an understanding of the decoding system ( 17,19,56 ). Our model likewise suggests that noise might be an important factor in understanding the limitations of the encoding process. Moreover, high-quality quantitative data on encoder profiles are also expected to enable us to go beyond the topological toward a quantitative network reconstruction. This may in turn permit realistic estimations of parameters and may then require an expansion of our basic model to include more mechanistic details and/or relax certain model assumptions. In conclusion, the classification of encoder modules and their encoding characteristics developed here provides a basis for a deeper understanding of the differences among the diverse signaling architectures used in quorum sensing, as it can help to elucidate their potentials and limitations. Thus it opens up new opportunities for the reverse engineering of natural, and the forward engineering of synthetic quorum-sensing systems. However, to obtain a complete picture of quorum-sensing systems, knowledge of both network modules—the decoder and the encoder—must eventually be productively combined.",
"introduction": "Introduction Cells constantly monitor their surroundings to detect and adapt to salient changes in environmental conditions. Many environmental cues are directly accessible to cells because they activate a signal transduction pathway. However, certain parameters, such as cell density, cannot be sensed directly. To be able to infer information about cell density, an additional step is required: cells must first “encode” this information in the form of a directly perceptible signal. They do this by secreting signaling molecules that accumulate in the external medium in a manner that reflects the population density of the emitting cells. The signal compounds are subsequently detected by cellular signal transduction systems that “decode” the information and shape an appropriate “quorum response” ( Fig. 1 \n A ). Quorum sensing is widely used by cells to regulate gene expression in response to cell density ( 1–3 ). It has been studied most extensively in bacteria but is also present in eukaryotic microorganisms ( 4,5 ). It plays an important role in microbial physiology and ecology and has important repercussions for human, animal, and plant health ( 6 ). In addition, quorum-sensing circuitry has become an integral component of many synthetic systems. Since the earliest experiments in synthetic biology, it has been engineered into diverse cellular backgrounds, including metazoans ( 7 ), starting with relatively simple, synthetic toy systems ( 8,9 ) and extending to more sophisticated applications in advanced biosensors ( 10–12 ), synthetic ecology ( 13 ), and systems engineering of multicellular behaviors ( 14 ), in recent years. In accordance with the two-step nature of quorum sensing, one may partition the underlying quorum-sensing network into two functional modules: an “encoder module” (EM) and a “decoder module” (DM) ( Fig. 1 \n B ). In this study, we define the EM to comprise all network components that are required to convert information about cell density ρ into a corresponding concentration of signaling molecules [SM], e.g., the signal synthase, signal modification enzymes, and signal transporters. The quorum-sensing receptor perceives the signaling molecules (SMs) and provides the interface between the EM and the DM. On the one hand, the specificity of the receptor for a particular SM and the location of the receptor are crucial elements of the encoding process, as they determine which concentration is detected by the cell. For example, when cells detect intracellular (extracellular) concentrations, they employ an intracellular (extracellular) encoding scheme. On the other hand, properties of the receptor that determine how the concentration of SMs is transduced into a cellular response, such as the affinity of the receptor, can be considered to be part of the DM. The DM consists of the signal transduction and gene regulatory components of the network that are required to control target gene expression in accordance with the level of the signal. In some quorum-sensing systems, the encoding and decoding processes appear to be largely decoupled; e.g., for ComX in Bacillus subtilis or AI-2 in Vibrio harveyi there is no feedback on any known component of the EM ( 15,16 ). Other systems make use of feedback on encoder components, e.g., in the LuxIR system of V. fischeri SM production is up-regulated upon activation of the pathway ( 17 ). One may therefore classify the former as feedforward and the latter as feedback encoding systems. This modular view of quorum-sensing systems, although never formally introduced in this way before, has already been quite successfully applied in the past to selectively characterize the decoder module by focusing on the signal transduction process. To this end, the DM is typically isolated (and decoupled) from the EM that is experimentally achieved by studying mutant systems, e.g., systems where the signal synthase has been knocked out. The DM can then be probed by stimulation with varying levels of externally supplied SMs by measuring the activity of the pathway. The systematic analysis of such input-output relationships has begun to reveal intricate relationships between network architecture and response behavior, in both natural ( 18–21 ) and synthetic systems ( 17,22 ). Thus, quorum-sensing systems have been shown to be capable of implementing a rather diverse spectrum of cellular response behaviors that is comparable with those of other sensory systems, and permits graded, threshold, and bistable responses ( 22–24 ). On the other hand, much less is known about the role of the encoder module in shaping the quorum-sensing process. Quantitative determination of the concentration profile of the SMs as a function of the cell density, i.e., [SM]( ρ ), provides insight into the encoding behavior of a quorum-sensing system. Available measurements by means of quantitative bioassays or mass spectrometry suggest that the density encoding implemented by natural EMs comprises quite a diverse set of linear and nonlinear encoding relationships that include linear, saturating, hypersensitive, and even inverted regimes, where the concentration of the SM falls as cell density increases ( 25–27 ). From a systems perspective, the EM could be studied with the same rigor as the DM by focusing on its function instead of that of the decoder. This could be achieved for instance, by using similar strategies, such as decoupling the EM and DM of feedback systems, by studying, e.g., receptor mutants to assess the properties of the open loop, i.e., the feedforward, encoder module. In addition, one naturally expects a tight relationship between system behavior and system structure. Just as the architecture of the decoder network determines the dose-response behavior of the system to SM stimulation, the architecture of the encoder network is expected to leave its imprint on the cell-density-encoding profile [SM]( ρ ) and shape the overall cell density encoding process. Such relationships have not yet been systematically investigated, despite the fact that nature has come up with quite a rich repertoire of encoder architectures. For example, some cells produce SMs that can freely diffuse across the cell membrane, whereas others use transport proteins that pump SMs into or out of the cell ( 28,29 ). SMs may also be subject to modification, either intracellularly, extracellularly, or during transport through the cell membrane ( 30,31 ). Furthermore, cells make use of either extra- or intracellular encoding, as signals may activate either transmembrane receptors or cytoplasmic receptors ( 32,33 ). Fig. 1 \n C schematically illustrates the architectural diversity of the EM. In this paper, we develop a theoretical model to systematically study the encoding profile, and its sensitivity and noise properties, in relation to the underlying network architecture of the encoder module. The model suggests that feedforward encoders show a wide variety of linear and nonlinear encoder functions that are shaped by the presence in the encoder network of certain core architectural elements. These elements can be combined to yield more complex encoding behaviors in systems with signal modification, allowing us to build a hierarchical classification scheme. We take advantage of this knowledge to derive predictions regarding the physical network architecture of partially characterized feedforward encoders, based on published experimental data on encoder functions, and demonstrate the utility of our approach by experimentally verifying one such prediction, namely that V. harveyi cells are indeed capable of AI-2 uptake.",
"discussion": "Discussion and Conclusions Quorum sensing might be considered as an “active” sensing process, since the inferred environmental parameter “cell density” cannot be “passively” detected by the cellular sensing machinery but must first be mapped onto a biochemical or biophysical parameter by a molecular encoding process. This “encoding” process has, however, been less well studied than the “decoding” of quorum-sensing signals. In this study we have focused on the development of a comprehensive theoretical classification scheme for feedforward encoder architectures. The rather high frequency (> 85%) with which one can obtain encoder networks capable of sensitive cell density encoding at least under some cell density regimes by “randomly connecting” different basic functions, such as receptors, transporters, and signal-modifying enzymes, suggests that multiple evolutionary trajectories could have given rise to functionally distinct quorum-sensing systems independently. But perhaps this finding should also warn us that not every network that has the capacity to mediate quorum sensing has necessarily been selected to carry out this function ( 48 ). Different architectures result in basic and more complex cell-density-encoding schemes that go beyond the “more cells—more signal” relationship. Although encoding relationships have not been systematically investigated by experimentalists, the data available indicates that the encoding relationships found in nature are likewise quite diverse. The biological function of these differences is essentially unknown, but it is intriguing to speculate that different systems may have evolved to create distinct “brands” of quorum sensing. Classification of encoding architectures Our model suggests that there is a tight relationship between network architecture and encoding behavior, which can be traced down to the presence of basic “core motifs.” This modular decomposition of the network provides the foundation for the development of an alternative classification scheme for encoder architectures. Importantly, although our analysis was focused on feedforward encoding systems, such as AI-2 in V. harveyi and ComX in B. subtilis , our taxonomic classification is also relevant for studying feedback encoder systems, such as the prominent acylhomoserine lactone (AHL) systems found in Gram-negative bacteria that operate with feedback on the signal synthase ( 17 ). By disrupting the coupling between decoder and encoder modules (e.g., by studying receptor mutants), the encoder function of these systems could be studied experimentally in the open-loop regime (which is captured by the model). Given sufficient insight into the topology of an encoder network, one may make predictions regarding its encoding behavior. For example, AHLs with short acyl-chains are apparently freely diffusive, thereby ensuring import of the SM ( 28 ). Hence, the open-loop behavior of many AHL systems is expected to implement a band-pass or low-pass encoding function, depending on whether they sense the intracellular (such as the Lux system in V. fischeri ) or the extracellular SM concentration (such as the Ain system in V. fischeri ), respectively. In contrast, Gram-positive bacteria often use signaling peptides, which generally cannot diffuse freely across lipid membranes, and are often modified after production of pre-peptides ( 49 ). One expects these systems to show low-pass or ideal encoding behavior, depending on whether they sense intracellular or extracellular, respectively. Furthermore, the posttranslational modification of the SMs potentially gives rise to more complex encoder characteristics, depending on where the modification takes place. Examples include the Phr systems in B. subtilis , where modification may either occur tightly linked to transport or in the extracellular environment ( 50 ). The latter alternative theoretically allows for an inverted sensitivity regime at high-cell densities according to our model. Such inverse encoding relationships might be relevant for terminating signaling, e.g., in biofilms. Feedback encoder systems The role of feedback encoding, although common in nature, is not well understood. Starting from an understanding of the open-loop encoder one may be able to parse out the contributions to the shaping of the encoder function that result from feedback and from the effects of the physical network architecture by comparing the encoding behavior of the open with that of the closed-loop in the wild-type system. Of course, the model can also be expanded to include feedback regulation. In analogy to theoretical studies that focus on architectural features of the DM and therefore treat the EM in a rather implicit way ( 17,24 ), one could adopt the reverse approach to implicitly model the effect of the DM in the model of the EM. For example, one can add up-regulation of SM production that depends in a Michaelis-Menten–like fashion on the relevant SM concentration. Although this feedback would obviously not change the nonfunctional encoder, it will increase the sensitivity of the quorum-sensing architectures in that part of the input regime where the feedback sets in ( Fig. S3 in Supporting Material ). This would, for example, enable members of the basic encoder classes to express ultrasensitive regimes (i.e., ε ( ρ ) > 1). However, the sensitivities in the limits of high- and low-cell densities ( ρ → 0 and ρ → ∞) remain unchanged in all classes. Feedback on SM production is especially well known in LuxIR-type systems ( 17 ). These systems share the architectural features of a band-pass encoder, which is the only functional encoder in our model with a sensitivity smaller than one ( ε ( ρ ) < 1) over the whole input range. Therefore it might make sense for these systems to increase the sensitivity of the encoding process by applying positive feedback on their SM production. Reverse engineering of quorum-sensing encoder modules Motivated by the vision of developing new therapeutics based on the manipulation of quorum sensing ( 51 ), the molecular identification of quorum-sensing network components that may affect the decoding but also the encoding process is an important task in the search for potential drug targets. However, it is clear that, for many quorum-sensing systems, not all network components have been identified. Generally speaking, the reverse engineering of a molecular system remains quite challenging. Although our theory cannot make predictions that pinpoint any particular molecule that might act in the encoder module, it may still serve as a useful tool for guiding network reconstruction by constraining the topology of the physical network. Along these lines, we have successfully used our model to make a prediction regarding the network topology of the AI-2 signaling system in V. harveyi. Surprisingly we found that V. harveyi is capable of importing AI-2. Elucidating the nature of the transport mechanism and studying the functional consequences of this unexpected import are important points for future investigations. Another example for a partially reconstructed encoder network is the ComX system in B. subtilis , in which an isoprenylated peptide SM activates a two-component system. As in the case of AI-2 in V. harveyi , the transport mechanism for ComX has not been experimentally determined. It may be excreted by passive diffusion because of its hydrophobic modification, or it might be transported by the modification enzyme ComQ or alternatively transported by an unknown transporter(s) ( 52 ). As with the AI-2 system in V. harveyi , the B. subtilis ComX signaling system is apparently based on a feedforward encoding module, as there is no feedback on any known component of the encoder network ( 15 ). The sensitivity of extracellular SM concentration profiles for the ComX pheromone in B. subtilis , obtained from bioassays published by Bacon Schneider et al. ( 25 ), suggests the presence of linear encoding over the entire input regime (although we cannot strictly exclude the possibility that saturating levels of ComX were not observed experimentally because of a finite cell-density input range). According to our model, this encoding behavior is suggestive of an ideal encoder core motif. The most likely explanation for the observed encoding behavior is therefore that ComX transport across the cytoplasmic membrane occurs via an active export system (or at least if ComX reimport exists it will be negligible under these conditions). For many quorum-sensing systems, such as the ComX and AI-2 examples, the mode of transport of SM in and out of the cells is still unclear, and our approach could therefore be of use to a broader community. Outlook and conclusions Dedicated experimental studies aimed at elucidating the encoding process are still scarce. By comparison with the quite impressive efforts that have gone into the development of experimental set-ups for quantitatively studying the decoding process, the approaches available for elucidating the encoding process might still be considered rather primitive. Most measurements rely on bioassays obtained from growing cultures, where the molecular concentrations may differ from the steady-state values ( 27,50,53–55 ), WHICH report on the average encoding behavior only and which almost exclusively measure the extracellular concentration only. Developing assays that are capable of quantitatively reporting on the intracellular concentrations of SMs to allow us to probe encoding systems using intracellular receptors, or measuring fluctuations in signal concentrations under steady-state conditions represent important experimental challenges for the future. Measurements of noise levels might be particularly interesting, as noise has become a powerful source of information that facilitates an understanding of the decoding system ( 17,19,56 ). Our model likewise suggests that noise might be an important factor in understanding the limitations of the encoding process. Moreover, high-quality quantitative data on encoder profiles are also expected to enable us to go beyond the topological toward a quantitative network reconstruction. This may in turn permit realistic estimations of parameters and may then require an expansion of our basic model to include more mechanistic details and/or relax certain model assumptions. In conclusion, the classification of encoder modules and their encoding characteristics developed here provides a basis for a deeper understanding of the differences among the diverse signaling architectures used in quorum sensing, as it can help to elucidate their potentials and limitations. Thus it opens up new opportunities for the reverse engineering of natural, and the forward engineering of synthetic quorum-sensing systems. However, to obtain a complete picture of quorum-sensing systems, knowledge of both network modules—the decoder and the encoder—must eventually be productively combined."
} | 5,652 |
29308265 | PMC5750032 | pmc | 827 | {
"abstract": "Extraordinary water-repelling properties of superhydrophobic surfaces make them novel candidates for a great variety of potential applications. A general approach to achieve superhydrophobicity requires low-energy coating on the surface and roughness on nano- and micrometre scale. However, typical construction of superhydrophobic surfaces with micro-nano structure through top-down fabrication is restricted by sophisticated fabrication techniques and limited choices of substrate materials. Micro-nanoscale topographies templated by conventional microparticles through surface coating may produce large variations in roughness and uncontrollable defects, resulting in poorly controlled surface morphology and wettability. In this work, micro-nanoscale hierarchical nanowire network was fabricated to construct self-cleaning coating using one-dimensional TiO 2 nanowires as microscale templates. Hierarchical structure with homogeneous morphology was achieved by branching ZnO nanowires on the TiO 2 nanowire backbones through hydrothermal reaction. The hierarchical nanowire network displayed homogeneous micro/nano-topography, in contrast to hierarchical structure templated by traditional microparticles. This hierarchical nanowire network film exhibited high repellency to both water and cell culture medium after functionalization with fluorinated organic molecules. The hierarchical structure templated by TiO 2 nanowire coating significantly increased the surface superhydrophobicity compared to vertical ZnO nanowires with nanotopography alone. Our results demonstrated a promising strategy of using nanowires as microscale templates for the rational design of hierarchical coatings with desired superhydrophobicity that can also be applied to various substrate materials.",
"conclusion": "3. Conclusion In summary, using one-dimensional TiO 2 NWs as microscale templates, micro-nanoscale hierarchical nanowire network with excellent water-repelling property was successfully developed. TiO 2 NWs of micrometres length not only provided sufficient contact area with the substrate or NWs entangling for producing a robust network film but also served as excellent backbone structure for ZnO NW branches growth. The hierarchical structure exhibited water-repelling properties after surface modification, displaying excellent superhydrophobicity compared to flat substrate or vertical ZnO NW substrate without hierarchical micro-nanoscale structure. This hierarchical nanowire-based superhydrophobic coating possessed homogeneous morphology in contrast to conventional microparticles/powders-based micro-nanoscale structure where significant surface defects or vacancies were present due to highly varied surface morphologies. Our work provides a promising approach of designing hierarchical nanowire film structure that offers great superhydrophobicity, which is a versatile approach for the coating of a wide range of materials to be self-cleaning.",
"introduction": "1. Introduction Self-cleaning surfaces, which possess extraordinary water repellency properties are currently the focus of considerable research [ 1 – 3 ]. They can be applied to the fields including fabric coatings [ 4 , 5 ], anti-biofouling paints [ 6 , 7 ], liquids separation [ 8 , 9 ], self-healing surfaces [ 10 , 11 ] and microfluidic devices [ 12 , 13 ]. One of the most well-known examples of self-cleaning surfaces in nature is the lotus leaf that can effortlessly roll off water drops. The underlying micro-nano hierarchical structure of lotus leaf surface has been revealed to play a key role in the water-repelling properties [ 14 ], and a large number of novel self-cleaning surfaces have been inspired by the micro-nano hierarchical structure of lotus leaf surface [ 1 , 2 , 15 ]. To understand the wetting behaviours of hierarchical surface, the Wenzel and Cassie–Baxter models are generally considered. In the Wenzel state, liquid impregnates the textures on solid surface, and the wetting or non-wetting feature of the solid is amplified due to the increase of the contact area between the liquid and the solid substrate [ 16 ]. While in the Cassie–Baxter state, air is trapped between the solid–liquid interface, which stably supports liquid drops sitting on the top of the substrate surface [ 17 ]. According to the Cassie–Baxter model, the presence of air pockets provided by the micro-nano hierarchical structure is crucial for superhydrophobic property. So far, it is commonly accepted that both hierarchical micro-nano structure and low surface energy coating on the solid surface are significantly important in achieving superhydrophobic effects. While low surface energy is generally prepared through surface functionalization with highly hydrophobic organic compounds, surface topography of micro-nano structure requires sophisticated fabrication or coating techniques. To date, most micro-nanoscale structures have been pre-fabricated on substrates by either top-down or bottom-up approaches [ 13 , 18 – 23 ]. The micro-nanoscale structures were physically attached to the substrate after fabrication, and were further functionalized to be superhydrophobic, forming a robust surface with water-repelling effects. For example, hierarchical Al, Si, ZnO, SiO 2 based surfaces with various surface roughness or surface topographies have been reported with excellent wetting properties [ 19 , 24 – 27 ]. However, the pre-fabrication process of micro-nanoscale structures and the requirement of physical attachment limit the choices of substrate material. Such approach relies heavily on the intrinsic properties of the used substrate, and can become unsuitable especially for the cases when the substrate material is unable to be fabricated with sub-micro features, or when the substrate material is susceptible or vulnerable to the fabrication process. On the other hand, superhydrophobic surface produced by coating or drop-casting micro-nanoscale objects on substrate rather than through direct fabrication may get rid of the restriction of substrate types [ 28 ]. In such approaches, the micro-nanoscale objects were physically absorbed or coated on the substrate, which allows a variety of substrates to be functionalized with superhydrophobicity regardless of the intrinsic material of substrate. For example, superhydrophobic coatings were fabricated by self-assembling micro- and nano-sized silica spheres [ 29 , 30 ]. Superhydrophobic and superamphiphobic structures were also designed using candle soot as the template [ 31 – 33 ]. Generally the micro-to-nanoscale characteristics in these structures were produced by using microspheres or micropowders as underlying micro-scale templates, followed by decorating with nano-scale objects such as nanoparticles or nanowires (NWs) on the micro-scale templates. However, due to the spherical nature, microparticles have limited contact area with substrate, exhibiting low adhesion with substrate so that they are improper templates for constructing a robust superhydrophobic coating. In addition, when microparticles are drop-casted on the substrate, it is difficult to maintain the surface uniformity. The produced rough surface by stacking microparticles usually has large variations in roughness and uncontrollable defects. As a result, the surface morphology and wettability of the resultant micro-nanoscale structures cannot be well controlled. Recently, much research has been focused on investigating the wetting behaviours of metal oxide semiconductors such as TiO 2 and ZnO for desirable functional properties [ 28 , 33 – 37 ]. For instance, Lu et al . created an ethanolic suspension of perfluorosilane-coated dual-scale TiO 2 nanoparticles as an excellent water-repellent paint [ 34 ]. Campbell et al . studied the electrowetting properties of ZnO nanorods with sputtered Teflon. Reversible wettabilities were observed under a threshold voltage of 35 V [ 35 ]. Sun et al . synthesized ZnO hollow microspheres with robust superhydrophobicity [ 28 ]. Although there has been significant progress in fabricating hydrophobic coatings, the possibility of using nanowires as building blocks for superhydrophobic hierarchical structure assembling is worth exploring. In this work, one-dimensional (1D) TiO 2 NWs were employed as microscale templates for the fabrication of micro-nanoscale hierarchical nanowire network to construct superhydrophobic coating. The TiO 2 NWs were coated on substrate by drop-casting method followed by sputtering with a thin ZnO layer. Due to their one-dimensional feature, the TiO 2 NWs of micrometres length formed a robust network film, with each NW sufficiently contacting with the substrate or entangling with other NWs. Secondary ZnO NWs were branched on the TiO 2 nanowire backbones through hydrothermal reaction, which produced nanoscale topography on the TiO 2 NW templates, forming a hierarchical coating of NW network on substrate with homogeneous morphology. The hierarchical surface was further functionalized with fluorinated organic molecules, and exhibited excellent repelling properties against both water and cell culture medium in contrast to flat substrate or vertical ZnO NW substrate. Contrary to the conventional pre-fabrication techniques where substrate materials were limited, our work demonstrated a promising and universal strategy to produce homogeneous micro-nano textured superhydrophobic coating on solid surface that may be applicable for a wide range of substrate materials."
} | 2,368 |
36234496 | PMC9565234 | pmc | 828 | {
"abstract": "As a unique surface wettability, superhydrophobicity has great application value. A variety of preparation methods for superhydrophobic surfaces have been reported, which have the disadvantages of high cost and complicated process. In order to design a method that is easy to operate, low-cost, and suitable for large-scale preparation of superhydrophobic surfaces, in this paper, hydrophobic nano-SiO 2 particles are used as spray fillers, and superhydrophobic surfaces are successfully obtained by the spraying process. According to the classical Cassie and Wenzel theory, the influence of the concentration change of hydrophobic nano-SiO 2 particles on their wettability is explained, and the appropriate spray concentration parameters are obtained. The results show that the proportion of hydrophobic nano-SiO 2 particles is lower than 0.05 g/mL, which will lead to insufficient microstructure on the surface of the coating, and cannot support the droplets to form the air bottom layer. However, an excessively high proportion of hydrophobic nano-SiO 2 particles will reduce the connection effect of the silicone resin and affect the durability of the surface. Through theoretical analysis, there are Wenzel state, tiled Cassie state, and stacked Cassie state in the spraying process. When the substrate surface enters the Cassie state, the lower limit of the contact angle is 149°. This study has far-reaching implications for advancing the practical application of superhydrophobic surfaces.",
"conclusion": "4. Conclusions In this paper, the influence of the concentration of hydrophobic nano-SiO 2 in the coating on the wetting performance of the coating was studied, and the change of the wetting state of the substrate surface with the increase of the concentration of the hydrophobic nano-SiO 2 was analyzed through a theoretical model. When the concentration of hydrophobic nano-SiO 2 in the coating is higher than 0.05 g/mL, the contact angle of the coating surface sprayed on the polished substrate is greater than 150°, but if the concentration is too high, cracks will appear on the surface of the coating and reduce the mechanical properties, so the suitable hydrophobic nano-SiO 2 concentration is 0.05 g/mL. The prepared superhydrophobic surface has low adhesion to droplets, mainly because the microstructure can effectively encapsulate air, reducing the solid-liquid contact area, and thereby reducing the adhesion between the solid-liquid interface. Theoretical analysis shows that there are three wetting states in the spraying process: Wenzel state, tiled Cassie state, and stacked Cassie state. When the substrate surface enters the Cassie state, the lower limit of the contact angle is 149°. This work provided a promising conception for advancing the practical application of superhydrophobic surfaces.",
"introduction": "1. Introduction Superhydrophobicity is an extreme surface property whose static contact angle of the surface liquid is greater than 150° [ 1 , 2 , 3 , 4 , 5 ]. The lotus leaf is the most representative biological model in the field of superhydrophobicity. However, the mechanism of keeping the lotus leaf clean was not known until the development of scanning electron microscopy (SEM) in the mid-1960s. The secret of the lotus leaf is “coming out of the silt without being stained”. In 1977, Barthlott and Neinhuis studied the surface structure of the lotus leaf by scanning it with an electron microscope [ 6 , 7 ]. There is a layer of nano-scale hair-like structures on the surface of lotus leaves, and the air between the hair gaps forms an air cushion between the hair-like structures and the water. The tension of the water surface also contributes. Therefore, when water falls on the lotus leaf, it exists in the form of water droplets. When the leaf surface is tilted slightly, the water droplets roll off the leaf surface, taking away the dust. After testing, the static contact angle of water on the lotus leaf is about 164° [ 8 , 9 ]. The superhydrophobic surface with this “lotus leaf effect” has wide application value in the fields of anti-corrosion, anti-icing, anti-fog, oil-water separation, self-cleaning, and underwater drag reduction [ 10 , 11 , 12 , 13 , 14 , 15 ]. At present, the preparation methods of superhydrophobic surfaces mainly include plasma etching, laser etching, electrochemistry, and vapor deposition [ 16 , 17 , 18 , 19 ]. The plasma etching and the laser etching methods use high energy to generate plasma sputtering or use a laser to etch the substrate [ 20 , 21 ]. The electrochemical method needs to energize the substrate, and use the ionization of the polymer solution to deposit on the substrate to form a micro-nano structure [ 22 ]. Vapor deposition is currently mainly used to prepare carbon nanotubes [ 23 ]. Such methods require special equipment, which is not conducive to large-area processing, and will cause damage to the substrate. Therefore, it is of great practical significance to further study the superhydrophobic surface with a simple preparation process and wide application range. The spray technique sprays nanoparticles and low surface energy solutions directly onto the surface to form a superhydrophobic surface. By spraying a mixed solution of nanoparticles and low-surface-energy substances, the surface has both nano-scale protrusions and low surface energy, so that a superhydrophobic surface can be easily and quickly obtained [ 24 , 25 , 26 ]. This method has a wide range of applications, can be processed in a large area, and has great advantages in the preparation of superhydrophobic surfaces. Gong et al. [ 27 ] obtained a surface with excellent superhydrophobicity on an aluminum alloy substrate by spraying with F-SiO 2 (SiO 2 nanoparticles treated with 1H, 1H, 2H, 2H-Perfluorodecyltriethoxysilane fluorination), epoxy resin adhesive, fluorosilicone varnishes, and white fluorinated polyurethane coatings. Results showed that superhydrophobic coating exhibited excellent water-repellency. The static contact angle of the superhydrophobic coating is 161.35°. Kim et al. [ 28 ] prepared coatings with hydrophobic nano-silica particles and siloxane binders and used them to prepare microstructured surfaces by a simple spraying method, obtaining a superhydrophobic surface with a static contact angle of about 160°. Guo et al. [ 29 ] sprayed the nano-silica dispersion on the semi-cured silicone-modified polyurethane resin by a two-step spraying method and successfully prepared a superhydrophobic coating with a static contact angle of 159°. It can be seen that a superhydrophobic surface with a high static contact angle can be constructed by spraying technology, but the literature does not mention how the concentration of hydrophobic nano-SiO 2 particles in spraying affects the superhydrophobicity of the surface. In this paper, a combination of theory and experiment was used to study the effect of the concentration of hydrophobic nano-SiO 2 particles in the spray coating on the superhydrophobicity of the surface, and a better spray concentration parameter was obtained. The three states in the spraying process were analyzed theoretically, and the lowest static contact angle after the droplet enters the Cassie wetting state was calculated. This study has far-reaching implications for advancing the practical application of superhydrophobic surfaces.",
"discussion": "3. Results and Discussion 3.1. Surface Topography Figure 2 shows the surface topography of the polished plane under the spraying of the hydrophobic nano-SiO 2 coating at different concentrations. Figure 2 A–D are the surface topography of the polished plane under the spraying of the hydrophobic nano-SiO 2 coating with a concentration of 0.025–0.1 g/mL, while Figure 2 E–H are the corresponding enlarged images of the surface topography. The mechanical properties of the coating are determined by the microstructure. When the concentration is 0.025 g/mL, the surface of the sample has a continuous film layer, basically no pores and defects, the larger protrusions are large particles formed by agglomeration of SiO 2 , and there are submicron protrusions on the large particles. The microspheres are in the Wenzel wet state, and the microstructure of the surface is not enough to support the droplets to form a superhydrophobic surface. When the concentration is 0.05 g/mL, the proportion of film-forming resin decreases, and pores begin to appear on the surface of the film layer. After that, the surface in the subsequent concentration range is in a Cassie-wet state and has a certain superhydrophobicity. When the concentration is 0.075 g/mL, the proportion of film-forming resin is further reduced, and cracks are formed in the plane. When the concentration is 0.1 g/mL, there are only rifts formed by the aggregation of nano-SiO 2 particles in the microscopic morphology, as well as sub-micron particles. As the concentration increases, the content of film-forming substances decreases, and the surface morphology tends to be large particles formed by the accumulation of nano-SiO 2 particles. The ability of the plane to store air becomes stronger, the solid-liquid contact area is further reduced, the static contact angle increases macroscopically, and the superhydrophobicity of the surface is enhanced, but the mechanical properties of the connection are weakened due to the lack of film-forming substances. Figure 3 shows the 3D surface profile of the polished coating surface after 15 sprays. It can be found that the rubbed polished surface is completely covered by the sprayed coating. The surface of the coating has a granular protruding microstructure similar to a lotus leaf, with a size of about 5 μm. These microstructures can effectively encapsulate air, thereby reducing the contact area between the liquid and the solid surface, reducing the adhesion between the solid-liquid interface, and finally achieving the effect of enhancing hydrophobicity. 3.2. Surface Wettability Figure 4 is a data graph of wetting performance under spraying of SiO 2 coatings with different concentrations. When the hydrophobic nano-SiO 2 particle concentration of the coating gradually increases to 0.025 g/mL, the contact angle of the sample increases. This is due to the gradual increase in the number of nanoparticles acting as microstructure supports, and the corresponding increase in the gas-liquid contact area, which is in the range of the contact angle described by the theoretical Wenzel state. When the concentration of the sprayed coating on the substrate reaches 0.05 g/mL, the change of the contact angle is large; when it exceeds 0.05 g/mL, the change is small, and the concentration reaches saturation. The adhesion test process of the polished substrate sample sprayed at a concentration of 0.05 g/mL is shown in Figure 5 . Figure 5 A volume of 5 μL of droplets is output from the infusion needle, and the superhydrophobic sample gradually approaches the droplets as the test bench rises. During this process, it can be seen that when the test bed moves up in the direction of the red arrow, it makes contact with the droplet when it reaches the position of Figure 5 B. It continues to move upward as the liquid is squeezed, and when it reaches the position shown in Figure 5 C, the test bed moves downward (as shown in Figure 5 D,E). When the droplet is separated from the sample, there is no residual droplet on the surface of the sample, indicating that the sample has low adhesion to the droplet. 3.3. Wetting Mechanism The microstructure of the hydrophobic nano-SiO 2 particles as the supporting droplets is attached to the surface of the substrate through the connection of film-forming substances. With the increase of the concentration, the wetting performance of the sprayed surface can be divided into three states, as shown in Figure 6 : When the concentration of nano-SiO 2 particles is low, the distance between the microspheres on the polished surface is too large, the linear density of the three-phase contact is too low, and the surface tension provided by the micro-nano structure cannot support the droplets, resulting in the infiltration of the droplets. With the increase of the nano-SiO 2 particle concentration, the contact angle of the droplet increases with the increase of the surface roughness, which belongs to the Wenzel state. At this time, based on Wenzel’s theory [ 30 ], the relationship between the contact angle and the concentration of nano-SiO 2 particles is as follows: (1) cos θ w = S 1 ( T ) cos θ 0 / S 2 \nwhere θ w , θ 0 , S 1 , and S 2 represent the apparent contact angle in the Wenzel state, intrinsic contact angle, actual solid-liquid contact area of the rough surface, and apparent solid-liquid contact area, respectively. The three-phase contact line density is an important index to measure the transition of the wetting state. According to the research conclusions of Extrand [ 31 ], the calculation formula of the critical value of the three-phase contact line density is: (2) Λ c = − ρ g V 1 / 3 { tan ( θ a / 2 ) [ 3 + tan 2 ( θ a / 2 ) ] } 2 / 3 [ ( 36 π ) 1 / 3 γ cos θ a ] \nwhere ρ is the density of the droplet, which is 1000 kg/m 3 ; g is the acceleration of gravity, which is 9.8 N/kg; V is the volume of the droplet, which is 5 μL; θ a is the advancing contact angle, which is 128°; and γ is the gas-liquid surface tension, which is 72 mN/m. The final Λ c is 0.471. Equation (2) Tiled Cassie state: When the critical value of the three-phase contact line density is the most ideal microstructure distribution state, the gas-liquid contact area fraction is the largest, and the contact angle is also largest at this time. However, with the continuous increase of the concentration of hydrophobic nano-SiO 2 particles, the three-phase contact line density on the surface continues to increase, the gas-liquid contact area fraction decreases, and the corresponding solid-liquid contact area fraction increases, resulting in a smaller droplet contact angle. Based on Cassie’s theory [ 32 ], the mathematical relationship is as follows: (3) cos θ c = f 1 ( T ) cos θ 0 + f 1 ( T ) − 1 \nwhere θ c is the apparent contact angle in the Cassie state, θ 0 is the intrinsic contact angle, and f 1 is the solid-liquid contact area fraction of the substrate. Figure 7 shows a schematic diagram of the microstructure-supported droplet. In the three-dimensional topography, it can be found that the radius R of the microspheres formed by the aggregation of the hydrophobic nano-SiO 2 particles sprayed on the surface is about 2.5 μm. From the advancing contact angle data, it can be obtained that the depth h when the droplet infiltrates the microsphere is 0.43 μm and the radius r of the wetting ball section is 1.97 μm. At this time, the area of the microsphere crown S 1 is 6.77 μm 2 , and the three-phase contact wire length c is 12.38 μm. According to the critical three-phase contact line density, the critical point from the Wenzel state to the Cassie state is obtained, and the center distance d between the microspheres is 5.13 μm. At this time, the gas-liquid contact area S 2 is 14.08 μm 2 , and the theoretical maximum contact angle size obtained by substituting the data into Equation (3) is 150°. When a layer of microspheres completely covers the surface, d is 5 μm. At this time, the gas-liquid contact area fraction of the Cassie state is the smallest, forming the minimum contact angle of the superhydrophobic surface Cassie state. Substituting the data into Equation (3) calculates that the droplet contact angle is 149° when it is completely flattened. When the hydrophobic nano-SiO 2 particles have completely tiled the surface and reach saturation, the excess hydrophobic nano-SiO 2 particles will be stacked on the three-dimensional level, and the solid-liquid contact area will be smaller in the process of random stacking, which leads to a larger contact angle of the superhydrophobic surface. Due to the characteristics of the spraying process, the same surface will be sprayed multiple times. The ideal tiled Cassie state will not appear. During the spraying process, it should be in the form of a stacked Cassie state."
} | 4,060 |
39125253 | PMC11314723 | pmc | 830 | {
"abstract": "Biopolymers are highly desirable alternatives to petrochemical-based plastics owing to their biodegradable nature. The production of bioplastics, such as polyhydroxyalkanoates (PHAs), has been widely reported using various bacterial cultures with substrates ranging from pure to biowaste-derived sugars. However, large-scale production and economic feasibility are major limiting factors. Now, using algal biomass for PHA production offers a potential solution to these challenges with a significant environmental benefit. Algae, with their unique ability to utilize carbon dioxide as a greenhouse gas (GHG) and wastewater as feed for growth, can produce value-added products in the process and, thereby, play a crucial role in promoting environmental sustainability. The sugar recovery efficiency from algal biomass is highly variable depending on pretreatment procedures due to inherent compositional variability among their cell walls. Additionally, the yields, composition, and properties of synthesized PHA vary significantly among various microbial PHA producers from algal-derived sugars. Therefore, the microalgal biomass pretreatments and synthesis of PHA copolymers still require considerable investigation to develop an efficient commercial-scale process. This review provides an overview of the microbial potential for PHA production from algal biomass and discusses strategies to enhance PHA production and its properties, focusing on managing GHGs and promoting a sustainable future.",
"conclusion": "8. Perspectives and Concluding Remarks Microbial PHAs are biodegradable polymers synthesized by various microorganisms for storage under nutrient-limiting conditions. However, their production encounters several challenges: (i) the limited availability of suitable carbon sources for microbial growth and PHA production, (ii) the complexity of regulating and controlling the metabolic pathways involved in PHA synthesis, (iii) the low yield and productivity of PHA production in microbial systems, (iv) the challenges associated with scaling up the production of microbial PHAs to an industrial scale, (v) the need for extensive downstream processing and purification of microbial PHAs, (vi) the potential risk for contamination in microbial PHA production processes, and (vii) the challenges of achieving consistent and reproducible product quality. These challenges hinder the widespread adoption and commercialization of microbial PHAs as sustainable alternatives to traditional petroleum-based plastics. PHAs are highly desirable over homopolymers (PHBs) because of their broad biotechnological applications, especially in the biomedical sector. Generally, precursor supplementation is required during microbial fermentation to produce the corresponding PHA copolymers. Therefore, co-substrate strategies, such as the co-digestion of feed, can be employed to produce desirable copolymers of PHAs with desired variations in their properties. Additionally, the successful conversion of algal biomass into valuable sugars, including PHAs, for various biotransformation applications, is highly dependent on effective pretreatment approaches. Genetically engineered algal cultures show promise for an efficient bioremediation of contaminated wastewater, or GHGs, such as CO 2 utilization (flue gas as origin), enhancing cell biomass for use as a feedstock for various biotechnological applications, including PHA production. Utilizing algal biomass as a carbon source to produce microbial PHAs on a pilot scale offers a sustainable solution for addressing the issue of plastic waste and reducing dependence on fossil fuels, thereby promoting a circular economy. Furthermore, using genetically engineered PHA for easy recovery or value-added co-product formation can minimize economic competitiveness compared to plastics. Advances in molecular and genetic engineering have enabled the selection of bacterial strains capable of producing PHAs in high quantities, up to 90% of their total biomass. However, the high cost of production remains a significant challenge for the commercialization of PHAs, limiting their widespread adoption.",
"introduction": "1. Introduction Plastics are polymers widely used in various human activities because of their exceptional physicochemical properties, affordability, and practicality [ 1 , 2 ]. However, plastics pose significant environmental hazards attributed to their slow degradation under normal circumstances, resulting in their accumulation in nature [ 3 , 4 ]. To address this issue, research is underway to develop biodegradable plastics [ 5 , 6 , 7 ]. Among the different biopolymers produced metabolically by various microorganisms, polyhydroxyalkanoates (PHAs) have emerged as promising alternatives to plastics [ 8 , 9 ]. PHAs are biodegradable; however, they are produced in only limited amounts by certain prokaryotes and eukaryotes [ 10 ]. Specific microorganisms can synthesize poly-β-hydroxybutyrates (PHBs), a class of PHAs. However, the fragile nature and poor physicochemical properties of PHBs limit their scope for biotechnological applications and, thus, their commercialization [ 11 , 12 ]. Additionally, the cost of this bioprocess approach is affected by feed costs, which can account for up to 45% of the total cost [ 9 , 10 ]. These restrictions have resulted in a search for biowaste as an alternative feedstock for PHA production [ 12 , 13 ]. Microbial fermentation of biowaste is more challenging than that of pure sugars because of its complex nature and diverse composition. Therefore, pretreatment is required for better accessibility to enzymatic/microbial hydrolysis, facilitating the production of fermentable sugars or bioactives [ 14 , 15 , 16 ]. Biological methods are generally helpful, owing to the clean transformation of biowaste into biofuels, biopolymers, or other value-added products [ 9 , 17 , 18 , 19 ]. PHAs are aliphatic polyesters produced by numerous prokaryotic organisms, which can constitute up to 90% of their dry cell weight (DCW) [ 10 ]. Fundamentally, the synthesis of PHAs occurs under physiological stress conditions, including an abundance of carbon (C) sources in the growth environment and lower concentrations of essential elements, such as magnesium, potassium, nitrogen (N), phosphorus (P), sulfur, or iron, during their development period [ 10 , 20 ]. PHAs can be blended into either homopolymer or copolymer structures. PHB is the most popular homopolymer of PHA and is produced by many microbes. The biosynthesis of PHB involves three crucial steps: (i) the initial step is catalyzed by β-ketothiolase (encoded by phaA , which involves the condensation of two acetyl-CoA molecules to form acetoacetyl-CoA); (ii) the acetoacetyl-CoA is reduced to acetoacetyl-CoA by NADPH-dependent acetoacetyl-CoA dehydrogenase (encoded by phaB ); (iii) finally, the monomers [(R)- 3-hydroxybutyryl-CoA] are polymerized into PHB by PHB synthase (encoded by phaC ) [ 9 , 21 ]. The type of PHB produced by various organisms depends on the phaC present in specific organisms [ 10 , 11 ]. Microorganisms, including Gram-positive bacteria such as Bacillus , Rhodococcus , and Staphylococcus ; Gram-negative bacteria such as Acinetobacter , Azotobactor , Burkholderia , Halomonas , Klebsiella , Pseudomonas , and Ralstonia ; algae such as Arthrospira , Botryococcus , Chlamydomonas , Chlorella , Nostoc , and Spirulina ; and engineered microorganisms such as Escherichia , Halomonas , Bacillus , and Saccharomyces have been reported for PHA production [ 10 , 22 , 23 , 24 ]. Microbial PHA production via fermentation has been widely reported using—(i) pure sugars such as glucose, fructose, maltose, and starch and (ii) biowastes of diverse origins, including agricultural (rice straw, wheat straw, potato, onion, carrot, cauliflower, tomato, pea-shells, orange peels, grape peels, melon, and apple pulp), municipal (vegetable, fruit, and other food waste), industrial (algal biomass, molasses, cheese whey, biodiesel waste, and dairy waste), and synthetic (wastewater) sources [ 9 , 23 , 25 ]. Microbial PHAs are classified into two significant subdivisions based on the C chain length of their monomeric units: short-chain-length (scl) PHAs with 3–5 C atoms and medium-chain-length (mcl) PHAs with 6–18 C atoms [ 26 , 27 ]. Microalgae are photosynthetic microorganisms capable of converting carbon dioxide (CO 2 ) and sunlight into biomass [ 23 , 28 ]. They have faster growth rates and do not compete with food crops for arable land or freshwater resources. Additionally, microalgae can be cultivated using waste streams such as CO 2 derived from flue gas and wastewater, further improving the sustainability of the production process [ 29 , 30 ]. In estuarine environments, benthic macroalgae can account for a significant portion of total primary production, sometimes up to 50% [ 29 ]. The chemical composition of biomass, including carbohydrates (7.8–30.8%), proteins (13.0–65.2%), and lipids (3.2–30.4%), is quite variable in red, brown, and green algae along with various pigments and secondary metabolites [ 23 , 28 ]. The simple structure of microalgae, which contains less lignin than other renewable biomass, can facilitate the bioprocessing of their carbohydrate fraction for producing PHAs. Carbohydrate recovery from algal biomass highly depends on pretreatment methods, such as biological, physical, and chemical methods, which significantly vary among types of algae due to their diverse composition [ 31 , 32 ]. Algae are widely preferred for the bioremediation of toxic compounds from the wastewater of diverse origins and are also effective in mitigating greenhouse gases (GHGs) such as CO 2 . The production of algal biomass using wastewater or CO 2 for biomass production, followed by the further utilization of biomass hydrolysate to produce PHAs, presents a sustainable approach to achieving a circular economy [ 13 , 33 , 34 ]. Moreover, microbes such as Bacillus , Cupriavidus , Halomonas , Haloferax , Paracoccus , and Rhodotorula demonstrate high potential for producing PHAs with yields up to 22.5 g/L from algal-biomass-derived sugars [ 23 , 35 , 36 , 37 ]. The microbial production yield and composition of PHAs from algal biomass are highly dependent on factors such as pH, temperature, incubation period, the type of limitation (such as N- and P-sources), the origin of feed, and the kind of PHA-producing culture [ 31 , 36 , 38 , 39 , 40 ]. Monomeric forms of PHAs, such as PHB, are widely produced by these microbes using algal biomass [ 35 , 36 , 38 , 39 ]. A few studies have reported the microbial production of PHA copolymers as poly-3-hydroxybutyrate- co -3-hydroxyvalerate [P(3HB- co -3HV)] from algal-derived sugars or residues [ 37 , 41 , 42 ]. Generally, copolymer precursors such as valerate are required for the microbial synthesis of P(3HB- co -3HV) using pure or lignocellulosic-biomass-derived sugars [ 4 , 43 ]. In contrast, Halomonas mediterranei DSM 1411 can efficiently produce P(3HB- co -3HV) copolymers lacking any external precursor addition [ 38 ]. Additionally, the properties of the produced PHAs must be suitable for biotechnological applications, particularly in tissue engineering and drug carrier biomedical applications [ 3 , 43 ]. Largely produced PHB by microbes using sugars or algal biomass demonstrate low potential for biomedical applications. Given the broad benefits of PHA copolymers, it is highly recommended that they be easily synthesized by various methods, such as altering the feed sources, production conditions, and the use of engineered microbes [ 23 , 44 , 45 , 46 ]. Therefore, different technologies require the integration of complementary microorganisms and algal biomass as feed to produce various PHA copolymers with diverse properties. The recent pandemic has highlighted the need for waste management approaches and technological advancements for sustainable development. This review aims to evaluate the existing literature on microbial PHA production from algal-based biomass hydrolysates. Furthermore, various strategies and challenges have been discussed to improve PHA production and develop sustainable approaches that offer environmental benefits."
} | 3,061 |
28555880 | null | s2 | 831 | {
"abstract": "Methods to generate fibers from hydrogels, with control over mechanical properties, fiber diameter, and crystallinity, while retaining cytocompatibility and degradability, would expand options for biomaterials. Here, we exploited features of silk fibroin protein for the formation of tunable silk hydrogel fibers. The biological, chemical, and morphological features inherent to silk were combined with elastomeric properties gained through enzymatic crosslinking of the protein. Postprocessing via methanol and autoclaving provided tunable control of fiber features. Mechanical, optical, and chemical analyses demonstrated control of fiber properties by exploiting the physical cross-links, and generating double network hydrogels consisting of chemical and physical cross-links. Structure and chemical analyses revealed crystallinity from 30 to 50%, modulus from 0.5 to 4 MPa, and ultimate strength 1-5 MPa depending on the processing method. Fabrication and postprocessing combined provided fibers with extensibility from 100 to 400% ultimate strain. Fibers strained to 100% exhibited fourth order birefringence, revealing macroscopic orientation driven by chain mobility. The physical cross-links were influenced in part by the drying rate of fabricated materials, where bound water, packing density, and microstructural homogeneity influenced cross-linking efficiency. The ability to generate robust and versatile hydrogel microfibers is desirable for bottom-up assembly of biological tissues and for broader biomaterial applications."
} | 384 |
38739105 | PMC11181286 | pmc | 832 | {
"abstract": "Piezoelectric energy harvesters have gained significant\nattention\nin recent years due to their ability to convert ambient mechanical\nvibrations into electrical energy, which opens up new possibilities\nfor environmental monitoring, asset tracking, portable technologies\nand powering remote “Internet of Things (IoT)” nodes\nand sensors. This review explores various aspects of piezoelectric\nenergy harvesters, discussing the structural designs and fabrication\ntechniques including inorganic-based energy harvesters (i.e., piezoelectric\nceramics and ZnO nanostructures) and organic-based energy harvesters\n(i.e., polyvinylidene difluoride (PVDF) and its copolymers). The factors\naffecting the performance and several strategies to improve the efficiency\nof devices have been also explored. In addition, this review also\ndemonstrated the progress in flexible energy harvesters with integration\nof flexibility and stretchability for next-generation wearable technologies\nused for body motion and health monitoring devices. The applications\nof the above devices to harvest various forms of mechanical energy\nare explored, as well as the discussion on perspectives and challenges\nin this field.",
"conclusion": "8 Conclusion Energy harvesting of ambient\nmechanical sources and human daily\nmovement has gained lots of attention due to the increasing development\nof portable microelectronics. Energy harvesters exhibit great application\npotential in the field of medicine, monitoring, and entertainment\nsuch as smart watches, heart-beat monitoring, and walking-step monitoring.\nIn addition, mechanical energy is an abundant and clean source of\nenergy in our daily life, which can be harvested independent of the\ntime and location compared to solar and in some cases thermal energy.\nTherefore, a range of related research has been reported to investigate\ntechniques to enhance the performance of energy harvesters. In this\nreview, the development progress of piezoelectric energy harvesters\nwas summarized. Various piezoelectric materials can be utilized to\nharvest energy from mechanical forces by fabricating into energy harvesters.\nThe common fabrication and measurement methods of energy harvesters\nwere also introduced. To further enhance the output performance, the\nsurface of the piezoelectric active layer can be modified using a\nvariety of techniques. Device structure and hybrid energy harvesters\ndesign can also be investigated to improve their output performance\nand mechanical properties. However, facing the future of commercial\napplications, there are still some important issues related to flexible\nenergy harvesters specially for wearable piezoelectric energy harvesters: (1) Appropriate electrodes and effective\nencapsulation methods to integrate the whole energy harvester structure\nplay a key role in the efficiency and durability improvement of energy\nharvesters. The electrode should satisfy the ability to form a good\ncontact with the active layer and good stability. For sustainability\nand mass production, the price of the electrodes needs to be considered\nas well. (2) The fabrication\nprocess of energy\nharvesting technology is still in the experimental stage with associated\ncomplicated processes and high cost. For practical applications, large-scale\nproduction of energy harvesters with lower production cost and environmentally\nfriendly procedures and materials should receive more attention in\nfuture studies. (3) The\ndevice size of current energy\nharvesters being developed in research utilizes a small working area\nin the scale of cm 2 . Large-area flexible devices should\nbe considered to harvest energy from human daily movements (walking,\nrunning, etc.). In addition, more research should be focused on the\nwashable performance and durability of the piezoelectric microelectronics,\nwhich should also meet the requirements of comfort and breathability. (4) Suitable storage system\nto store the\npower produced by the energy harvesters also needs to be developed.\nFor wearable applications, the storage system also needs to meet the\nrequirement of comfort and durability of the wearable applications. 280 − 282 (5) While highly challenging\ndue to the\nwide potential range of actuation methods and kinetic energy sources,\nefforts toward standardization of test conditions and parameter reporting\nwould be highly beneficial for the translation of any energy harvester\nresearch into commercial applications. At a minimum, the available\nenergy levels of the sources under consideration should be carefully\nconsidered to ascertain if they could ever provide sufficient power\nlevels to power even low-level IoT devices. Then input energies/forces/power\nshould be better measured and described so that the overall conversion\nefficiency could be quantified and alternative approaches and publications\ncould be compared. Certainly, minimum levels of parameter reporting\nsuch as peak power levels, as called for many years previously, 283 are still lacking in some works and should\ndefinitely always be included. Through this the many highly ambitious\nclaims for the potential for energy harvesters can be tested with\nconcrete data. Flexible electronics have been the subject of much research,\nand\nsignificant advancements have been realized. Particularly wearable\npiezoelectric technology has developed with an enormous and expanding\nrange of uses. Piezoelectric materials can be combined with portable\ndevices, textiles, and other flexible applications which, if proven\neffective through rigorous testing and reporting, have the potential\nto revolutionize how we power the ever increasing range of portable\ndevices in the future.",
"introduction": "1 Introduction The Internet of Things\n(IoT), robotics, artificial intelligence\n(AI), and big data have drawn significant research attention, which\nwill bring us a huge revolution into many aspects of our daily life.\nAs a network of connected computing devices, the IoT is expected to\nhave the ability of real-time location tracking, monitoring our body\nmovement or health condition, such as wearable displays and wireless\nhealth tracking devices. In this regard, piezoelectric energy harvesters\nhave emerged as a promising technology with good potential to convert\nambient mechanical movement to electricity by exploiting the piezoelectric\neffect, presenting an attractive solution for sensors, powering low-power\nIoT devices, and reducing the reliance on conventional power sources. Over the past few years, a large number of piezoelectric materials\nhave been reported for energy harvesting applications in self-powered\nsensors and wearable electronics, such as zinc oxide (ZnO), barium\ntitanate (BaTiO 3 ), and lead zirconate titanate (PZT). Despite\nthat, with increasing development of portable/wearable electronic\ndevices such as smart watches, health, and activity monitors, it is\nparticularly desirable to research a flexible energy harvester that\ncan capture multiple forms of mechanical energy with enhanced energy\nconversion efficiency, which holds great promise in personal smart\ndevices. To meet the requirement of flexibility and comfort, a number\nof flexible substrates with their unique properties of lightweight,\ncomfort, softness and wearable convenience hold great potential to\nbe used as platform to be integrated with piezoelectric materials\nused as portable/wearable electronic devices, which can generate energy\nfrom jumping, joint bending, and running etc. In this regard,\na multitude of scientific papers have been reported\ninvestigating the various range of energy harvesters using different\nstrategies to obtain higher output performance with high flexibility.\nIn this review, the basic working principle and classifications are\ndiscussed. We also cover the recent research into different piezoelectric\nmaterials, material and device fabrication and measurement methods.\nStrategies for improving the energy harvesting performance are also\ninvestigated. The current challenges and future directions in their\ndevelopment are summarized, which can be used as reference and an\nintroduction to the energy harvester field to help the development\nof portable/wearable energy harvesters."
} | 2,032 |
23868689 | PMC3715747 | pmc | 835 | {
"abstract": "Conventional leaching (extraction) methods for gold recovery from electronic waste involve the use of strong acids and pose considerable threat to the environment. The alternative use of bioleaching microbes for gold recovery is non-pollutive and relies on the secretion of a lixiviant or (bio)chemical such as cyanide for extraction of gold from electronic waste. However, widespread industrial use of bioleaching microbes has been constrained by the limited cyanogenic capabilities of lixiviant-producing microorganisms such as Chromobacterium violaceum . Here we show the construction of a metabolically-engineered strain of Chromobacterium violaceum that produces more (70%) cyanide lixiviant and recovers more than twice as much gold from electronic waste compared to wild-type bacteria. Comparative proteome analyses suggested the possibility of further enhancement in cyanogenesis through subsequent metabolic engineering. Our results demonstrated the utility of lixiviant metabolic engineering in the construction of enhanced bioleaching microbes for the bioleaching of precious metals from electronic waste.",
"discussion": "Discussion Here we report the construction of two metabolically-engineered C. violaceum strains (pBAD and pTAC, respectively) that produced significantly more cyanide lixiviant compared to wild-type bacteria. Lixiviant production was effectively decoupled from quorum control through the use of exogenous promoters. In corroboration with the observed increase in cyanide production by the engineered strains, Au recovery from ESM was significantly increased in the engineered strains. Our efforts in modulating the lixiviant metabolism of C. violaceum demonstrate proof-of-concept that enhanced bioleaching microbes can be constructed as sustainable means of recovering precious metals such as Au from electronic waste. Wild-type C. violaceum produces cyanide from glycine for short periods at the early stationary phase in its growth. The reliance of lixiviant production on cell population density is quorum-controlled, and is probably co-evolved as a defense mechanism in its bid for niche colonization 19 . In an effort to obtain suitable strains of C. violaceum for bioleaching of Au from electronic waste, we examined if lixiviant production can be decoupled from quorum control, and if so, whether cyanide production can be increased by modulating the lixiviant metabolic pathways of the organism. Normal cyanide metabolism is maintained by combined actions of the cyanogenic hcnABC operon and the cyanolytic cynTSX operon. We sought to engineer a tightly-regulated and tunable (responsive to varying concentrations of inducer) bioleaching strain by integrating a single copy of the hcnABC operon, under the transcriptional control of the exogenous promoters, into the bacterial genome. This site-specific genomic integration of an inducible cyanogenic operon was performed using Tn7-mediated transposition developed by Schweizer and coworkers 20 . We inserted a duplicate copy of the hcnABC operon (under the transcriptional control of exogenous promoters pBAD or pTAC , respectively) to decouple cyanogenesis from quorum control. By design, genomic insertions can be made at specific Tn7 attachment (attTn7) sites located downstream of the highly conserved glmS gene (encoding for glucosamine 6-phosphate synthetase) in bacteria. The glmS gene of wild-type C. violaceum (GI: 34496132) is located upstream of a functionally unassigned gene (GI: 34496133) with a 45 bp intergenic region that is predicted to contain the attTn7 site. However, in contrast to studies with other bacterial species 21 , genomic integration in C. violaceum did not occur at the intergenic region downstream of the glmS gene; instead, the transposable segment was inserted within the 3′region of the glmS gene (after 1607 bp of the 1830 bp intact glmS gene). Despite disruption of the glmS gene, bacterial fitness in the engineered C. violaceum strains was not compromised (growth curves of wild-type and engineered strains are virtually identical, Fig. S2 ). In attempts to increase the production of the cyanide lixiviant, we used a range of inducer concentrations to obtain dose-responsive profiles of cyanide production. Using L-arabinose as the exogenous inducer, cyanide production was increased above wild-type levels ( Fig. 3 ), with maximal cyanide production (relative to uninduced levels) observed with 0.002% L-arabinose. An analogous dose-responsive profile was obtained with IPTG as the exogenous inducer ( Fig. 4 ), where the addition of 1 mM IPTG resulted in maximal cyanide production, compared to uninduced or wild-type levels. Our observations indicated that the engineered pBAD and pTAC strains exhibited tunable, enhanced cyanide lixiviant production. The observed proto-typical dose-responsive profiles also suggested that lixiviant production in these strains had been decoupled from quorum-control. A comparison of the lixiviant profiles of wild-type versus engineered strains revealed that the pBAD and pTAC strains produced peak concentrations (at 30 hrs after inoculation) of 34.5 mg/L and 31 mg/L of cyanide, respectively (representing significant increases over the wild-type peak concentration of 20 mg/L of cyanide, Fig. 2 ). With the expectation that an increase in cyanide lixiviant production would accompany a corresponding increase in bioleaching of Au from electronic waste, the engineered pBAD and pTAC C. violaceum strains were used to recover precious Au from ESM. Bioleaching studies revealed that there were significant increases in the recovery of Au from the ESM electronic waste for both strains of engineered C. violaceum , as compared to wild-type bacteria ( Fig. 5 ). There have been studies on the use of cyanogenic bacteria in the recovery of Au from solid waste 10 22 23 24 : the reported modest recovery (up to 14.9% of total amount of Au present) corroborate with our observation (11% of total amount of Au present) on the limited utility of wild-type cyanogenic bacteria in bioleaching. In comparison, we were able to achieve Au recovery in excess of 30% of total amount of Au present using engineered C. violaceum . We view these significant increases in Au recovery from electronic waste (over wild-type cyanogenic bacteria) as a heartening possibility of using bioleaching as a sustainable means of recovering precious metals from electronic waste in the future. To facilitate subsequent engineering efforts, we sought insights into the modulations of the lixiviant metabolic network of the pBAD strain through a comparative proteomics study of the cyanogenically-enhanced variant against the wild-type C. violaceum strain ( Fig. 7 ). We examined if the observed levels of cyanide produced in the engineered strain could be further increased with future metabolic engineering. A logical corollary follows that an increase in cyanide production could be met with modulations in the proteome that could have decreased cyanogenesis, and increased cyanolysis, leading to a limiting “cap” in the amount of cyanide production. Conversely, one could expect the observed increase in cyanogenesis to be within the inherent lixiviant capacity of the cyanogenic bacterium, and minimal modulations in the cyanogenic and cyanolytic pathways of the engineered strain would be observed. Cyanide production through the actions of HCN synthase can be decreased by decreasing the availability of the substrate glycine; consequently, glycine flux away from cyanogenesis can be independently achieved through the actions of enzymes such as serine hydroxymethyltransferase 17 (glyA, in the biosynthesis of serine from glycine), glycine decarboxylase 25 (gcvP, in the glycine cleavage system), serine dehydratase 26 (sdaA1, in serine catabolism), threonine aldolase and glycine C-acetyltransferase 27 (CV_4309 and kbl, respectively, in the biosynthesis of threonine). Our comparative proteomics study revealed that upon an increase in cyanide production in the engineered strain, there were no significant changes in the protein levels of these enzymes (directly and indirectly) associated with the cyanolytic pathways ( Fig. 7 ). Cyanolysis can also be directly achieved through the actions of cyanase 28 ; our study did not detect significant changes in the levels of cyanase in the engineered strain, compared to wild-type C. violaceum . In addition, we did not observe any significant changes in the protein levels of enzymes associated with glycine biosynthesis, and hence, cyanogenesis. Cyanogenesis in bacteria and plants is associated with virulence and defense 12 ; in the context of C. violaceum , our comparative proteomics study revealed significant modulations in metabolic pathways associated with metabolic dormancy and energy conservation. Taken together, an increase in nucleotide salvage (represented by a decrease in the enzyme purine nucleoside phosphorylase 29 /pnp), a decrease in deoxyribonucleotide biosynthesis (represented by a decrease in thioredoxin/trxA levels, the essential cofactor for ribonucleotide reductase 30 ), a decrease in the capacity for oxidative stress response 31 (represented by decreases in superoxide dismutase/sodB1 and thioredoxin/trxA levels), and an increase in fatty acid biosynthesis 32 (represented by an increase in enoyl-ACP reductase/CV_3743 levels, an essential enzyme in the biosynthesis of saturated straight-chain fatty acids; and a decrease in 3-hydroxyacyl-CoA dehydrogenase/fadB levels, an enzyme associated with β-oxidation of fatty acids) suggested that the increased levels of cyanide production was within the cyanolytic capacity of the engineered strain. Our hypothesis was further corroborated by the observation that there was no significant change in protein levels of cioA 33 , a cyanide-insensitive terminal cytochrome oxidase in the respiratory electron transport chain of C. violaceum ; if cyanide levels were to exceed the inherent cyanolytic capacity of the bacterium, the expectation would be a significant up-regulation of cioA, so that aerobic respiration (in particular, cytochrome c oxidase) would not be inhibited. We have demonstrated that lixiviant metabolism in C. violaceum can be engineered for enhanced cyanide production; a decoupling of cyanogenesis from quorum control resulted in a significant increase in cyanide production, and correspondingly, an increase in Au recovery from electronic waste. Comparative proteomics analyses suggested that further increase in cyanogenesis is possible, and our results highlight the utility of lixiviant metabolic engineering in the construction of next-generation bioleaching microbes for the recovery of precious metals such as Au from electronic waste."
} | 2,695 |
21633702 | PMC3102110 | pmc | 836 | {
"abstract": "Background Corals worldwide are in decline due to climate change effects (e.g., rising seawater temperatures), pollution, and exploitation. The ability of corals to cope with these stressors in the long run depends on the evolvability of the underlying genetic networks and proteins, which remain largely unknown. A genome-wide scan for positively selected genes between related coral species can help to narrow down the search space considerably. Methodology/Principal Findings We screened a set of 2,604 putative orthologs from EST-based sequence datasets of the coral species Acropora millepora and Acropora palmata to determine the fraction and identity of proteins that may experience adaptive evolution. 7% of the orthologs show elevated rates of evolution. Taxonomically-restricted (i.e. lineage-specific) genes show a positive selection signature more frequently than genes that are found across many animal phyla. The class of proteins that displayed elevated evolutionary rates was significantly enriched for proteins involved in immunity and defense, reproduction, and sensory perception. We also found elevated rates of evolution in several other functional groups such as management of membrane vesicles, transmembrane transport of ions and organic molecules, cell adhesion, and oxidative stress response. Proteins in these processes might be related to the endosymbiotic relationship corals maintain with dinoflagellates in the genus Symbiodinium. \n Conclusion/Relevance This study provides a birds-eye view of the processes potentially underlying coral adaptation, which will serve as a foundation for future work to elucidate the rates, patterns, and mechanisms of corals' evolutionary response to global climate change.",
"introduction": "Introduction Reef-building corals (Cnidaria: Hexacorallia: Scleractinia) are of fundamental ecological significance in tropical and sub-tropical shallow marine environments as they form the most important components of coral reefs. These organisms are sensitive to the current rising global seawater temperatures [1] resulting in increased frequencies of mass coral bleaching events, which in turn have caused severe declines in live coral cover [2] . To this end, much effort has been committed to assessing factors affecting overall vulnerability and resilience of reef corals [3] , [4] , [5] , [6] . Additional work has been devoted to the identification of stress-responsive genes [7] , [8] , [9] , [10] , [11] , [12] . However, few studies have looked into the genetic makeup of corals that might help determining to what extent corals are able to respond to increasing disturbances and stress by means of evolutionary adaptation [13] . Thompson and van Woesik [14] found that corals at sites with a high-frequency of thermal stress displayed less bleaching than at other sites, despite being exposed to a greater level of stress. The authors suggest that bleaching resistance is most likely a consequence of rapid directional selection following an extreme thermal event, i.e. corals are able to respond adaptively from the pool of standing genetic variation. Other studies have shown that multicolored fluorescent proteins display a considerable amount of adaptive, convergent, and parallel evolution in corals [15] , [16] . Schwarz et al. \n [17] characterized a ferritin in Acropora millepora and Acropora palmata that displays signs of positive selection. Hayes et al. detected adaptive evolution in tachylectins [18] . Adaptive evolution, at the molecular level, is characterized by an excess of nonsynonymous nucleotide substitutions ( d N ) in comparison to synonymous ones ( d S ) [19] , [20] , [21] , [22] , [23] . If this is the case, the so-called d N /d S ratio becomes >1, and the gene of interest may be under positive selection. Note that a single important amino acid change may be sufficient to demonstrate positive selection. However, methods for site-specific adaptive evolution analyses require multiple pair-wise comparisons, thus inclusion of sequence data from multiple species. Evolutionary screens are designed in a way that orthologs in a designated group of genes are ‘scanned’ for elevated d N /d S ratios. These screens provide a powerful way to identify, in a single effort, many candidate genes that are potentially subject to positive selection. Circumstantially, there is no a priori requirement to know the function of the protein, a factor that is particularly beneficial in non-model organisms such as corals. However, lack of annotation cannot be considered a difficulty exclusively associated with non-model species as the number of genes without any significant sequence similarity to genes of other species in any eukaryotic genome surveyed so far seems to be about 10–20% [24] , [25] , [26] . It is assumed that these genes represent lineage-specific adaptations of the species under study as they not only lack sequence similarity to genes or proteins in other organisms, but also display a narrow phylogenetic distribution. There is no general agreement or rule, but usually, proteins which do not show any sequence similarity in BLASTp searches with cut-off values of E<10 −5 or E<10 −10 have been denoted as so-called taxonomically-restricted genes (TRGs) [27] , and have been hypothesized to provide one of the sources of phenotypic diversity [28] , [29] , [30] . TRGs are synonymously referred to as lineage-specific genes [29] . A recent screen by Sunagawa et al. \n [31] identified a family of small, cysteine-rich proteins (SCRiPs) that appear to be restricted to Hexacorallia. A study in Hydra identified Periculin-1, a peptide that has strong bactericidal activity and at present no identifiable orthologs in sequence databases [32] . The amount of positive Darwinian selection has not yet been systematically surveyed in any coral. We set out to conduct an evolutionary screen of orthologs in two congeneric acroporid coral species: Acropora millepora from the Indo-Pacific and Acropora palmata from the Caribbean. We identified a set of 2,604 orthologous cDNA sequences for which we calculated pair-wise d N /d S ratios in order to (i) identify the extent of adaptive evolution in scleractinian corals, and (ii) assess the nature of proteins that are potentially subject to positive selection. Our results indicate that a considerable fraction of coral proteins might be under positive selection, and that TRGs display on average significantly higher evolutionary rates. As such, they might represent important mediators of microevolution and lineage-specific adaptations that warrant further examination for assessing the future response of corals to a changing environment.",
"discussion": "Discussion Evolutionary Screen A major factor that comes into play when assessing d N /d S ratios is that with higher evolutionary divergence, d S becomes saturated with multiple substitutions per site on long branches. Hence, neutral evolution is underestimated and, as a consequence, comparisons between different species are only valid within a given divergence range. The genus Acropora (Scleractinia: Acroporidae) is one of the most widespread genera of corals as it spans the Indian and Pacific Oceans and the Caribbean Sea. It is also the largest extant reef-building coral genus with numbers of species estimates ranging from 113 to 180 [48] . In this study, A. millepora is representing a member of the Indo-Pacific Acropora species and A. palmata is representing a member from the Caribbean. Molecular analyses suggest that A. millepora and A. palmata had their latest contact around 12 Myr ago, while Indo-Pacific Acropora species have radiated over the last 10 Myr [48] . If we assume a generation time of 1 to 10 years [49] , [50] and a mutation rate of 10 −8 per nucleotide site per generation for both species [51] , we come up with the following proxy for d S : 10 7 generations (divergence time/generation time) * 10 −8 (mutation rate) = 0.1. Hence, we expect an average divergence at neutrally evolving sites of approximately 10% (given that both species have the same mutation rate). This estimate is the same order of magnitude as the median d S of our set of orthologs (median d S = 0.043), and consequently our approach does not seem to inflate measures of d S . Even for genes that evolve fast, this divergence time frame allows one to identify the respective ortholog in both species. We found that a considerable portion of the orthologs showed d N /d S values exceeding 1 (7% of all orthologs), and that TRGs had significantly higher d N /d S values. This finding might indicate that the group of TRGs plays a vital role in adaptive evolution. These genes did not show homology anywhere, including the sea anemone Nematostella vectensis , which belongs to the same subclass but a different order (subclass Hexacorallia, order Actiniaria). Although many of these genes may be coral-specific (i.e., restricted to stony corals, order Scleractinia), we cannot rule out that they are present in other, currently unsampled, orders of Hexacorallia (e.g. Corallimorpharia and Zoanthidea), or even have a broader pan-Anthozoan distribution but happen to be missing in the sea anemone. Studies in Drosophila have shown that TRGs represent a group of genes that on average display higher d N /d S ratios and are likely to play an important role in lineage-specific adaptations [35] , [52] . Furthermore, a recent study on orthologs from coral symbionts, Symbiodinium spp., identified the highest d N /d S ratio in a Symbiodinium -specific gene, and the authors speculated accordingly that the portion of genes with elevated d N /d S values might be higher in the group of lineage-specific genes in comparison to conserved genes [53] . The authors further hypothesized that a symbiotic lifestyle might affect sequence evolution, as genes might need to coevolve with their symbiotic partners. The ability to differentiate between self and non-self plays a particular role for reef-building corals in light of their mutualistic, intracellular symbioses with dinoflagellate algae as these need to be distinguished from other eukaryotic protists (dinoflagellates are alveolates and a sister group to the apicomplexans – obligate intracellular parasites – that may use the same receptors and signaling pathways to gain access to the host cell). In addition, competition between different symbiont strains might facilitate the evolution of genes involved in recognizing different clades of Symbiodinium , which often can associate with the same coral species [54] . A recent study suggested that transcriptomic states of the Caribbean coral Montastraea faveolata (a coral that can host multiple Symbiodinium genotypes) were correlated with differences in the Symbiodinium genotype hosted [55] . It will be interesting to test whether the percentage of genes under adaptive evolution is higher in corals that are able to host multiple versus only one genotype of Symbiodinium . Given that the generation time of Symbiodinium spp. is orders of magnitude smaller than those of corals, selection in corals might act on being less discriminating between different algal types that in turn evolve to cope with a changing environment. Expession of d N /d S ortholgs By definition, functional inference by homology to known genes is not available for TRGs. However, expression of those genes might indicate and serve as a proxy for functional significance. Given that some of the TRGs are likely a result of lineage-specific adaptations, it seems plausible to assume that they are expressed specifically with regard to life stage as 1) they presumably have specialized functions, and 2) restricted expression tends to minimize pleiotropic interference. Here, we wanted to test whether specific life-stages show an enrichment or depletion of high d N /d S genes in the group of annotated and non-annotated genes. While we were not able to show statistically significant differences, our data indicates that the adult life stage of corals has a similar d N /d S distribution for conserved (i.e. annotated) and coral-specific (i.e. non-annotated) genes ( Table 2 ). By contrast, all other life stages show a higher mean d N /d S in supposedly coral-specific genes in comparison to conserved genes ( Table 2 ). This might indicate that in order to better understand and investigate adaptive evolution of corals, particular attention has to be paid to non-annotated genes, and that expression of these genes are more easily found in life stages other than adult. It is interesting to note that we found lineage-specific genes with low d N /d S values that showed stage-specific expression (e.g., lineage-specific orthologs in the adult stage). Those genes either arose de novo \n [56] , through gene duplication and subsequent diversification [35] , or were retained from a common ancestor but lost elsewhere [57] . They likely represent genes that support coral-specific adaptations, as they are conserved among corals but not found outside this lineage. A promising approach that arises from these considerations is that slowly evolving TRGs are enriched for “coral-specific” processes and are expressed stage-specifically. A combination of in silico and in situ approaches that couples evolutionary analyses with signatures of expression might prove a useful strategy to target such genes for further functional studies. Functional analysis of the variation in protein evolution rates Analysis of all the functional categories represented within our set of orthologs suggested a number of processes that experience accelerated rates of protein evolution ( Figures 2 and 3 , Table 3 ). Although many of these signatures may be due to relaxed purifying selection rather than positive selection, we detected anticipated targets of the latter. For example, proteins involved in immunity and defense, reproduction, and sensory perception (including transmembrane receptors and associated signaling pathways) are under positive selection in a wide variety of animals, from primates [22] and other mammals [58] to fruit flies [52] . “Bioluminescence” category in our case contained GFP-like fluorescent proteins, which have been shown to experience strong positive selection in corals [16] . In addition to these “usual suspects”, we saw elevated rates of evolution in several other functional groups that are not highlighted in studies of other animals and may therefore reflect the specifics of coral evolution. Some of these are ostensibly related to the corals' endosymbiotic relationship with Symbiodinium dinoflagellates, such as management of membrane vesicles, transmembrane transport of ions and organic molecules, and cellular homeostasis (which includes the category “maintenance of cellular location”). The category that was the most enriched with rapidly evolving proteins – cell adhesion – may also be related to symbiosis, as its members are linked to a number of cell surface molecules that may mediate host-symbiont recognition ( Table S1 ). These proteins are expected to evolve under positive selection due to the need for frequent specificity readjustments and potentially due to “arms race” between the coral and cheater (i.e. non-compatible) strains of Symbiodinium . Some of the functions highlighted by our analysis were rather unexpected. Most notably, proteins involved in metabolism of lipids and steroids feature prominently in both biological process and especially molecular function analyses ( Table 3 , Table S1 , Table S2 ), for which we are not yet ready to offer a biological explanation. Some other functional groups may appear as rapidly evolving due to sharing of orthologs with other GO categories (reflected by the dendrogram in Figures 2 and 3 ). For example, “multicellular organismal development” may have become highlighted in our analysis due to substantial sharing of orthologs with the “cell adhesion” category. More generally, the inherent redundancy of the GO database leads to partial overlaps in the outcomes of different categories, so that any result on functional analyses based on GO annotations (irrespective of the methodology) must be viewed as the union of all possible interpretations of the data. Since only one of these interpretations is correct, some false positives are unavoidable. Selecting the correct interpretation would be possible based on additional systems biology data, which is still lacking for corals but may become available in the near future from whole-transcriptome expression profiling studies. Candidate genes with potential relevance to cnidarian-dinoflagellate symbioses that display elevated rates of evolution Since the global GO analysis may not adequately reflect the mechanisms specific for coral biology, we looked at a set of candidate genes that were either directly implicated in cnidarian-dinoflagellate symbiosis by empirical evidence, or functionally interconnected with them in molecular pathways. Within this gene set, proteins that play a role in corals' response to stress and genes related to immunity were the most prominently represented ones. Stress-induced photoinhibition and damage to the algal photosystem II are thought to be responsible for an increased production of reactive oxygen species [59] , [60] , and consequently, diffusion of hydrogen peroxide (H 2 O 2 ) through the membranes into the host cell(s) [61] . H 2 O 2 then activates a cellular cascade, which results in expulsion of symbionts and bleaching [62] . The molecular pathways in the coral host to prevent bleaching (i.e. heat stress and oxidative stress) might therefore be under positive selection in order to mitigate the effects of stress on the coral-algae symbiosis. Consequently, many of the stress genes we identified ( Table 4 ) were identified as differentially expressed in recent microarray studies on heat stress and bleaching in corals [7] , [8] , [10] , [11] . Among the stress-related genes, we detected elevated rates of evolution in Hsp-16.2, Glutaredoxin-1, Glutathione S-transferase omega-1, and a Peroxidasin homolog ( Table 4 ). Genes related to innate immunity gave rise to another partial inventory of rapidly evolving genes. From the gene expression regulation standpoint, coral-algae specificity seems to arise not from the fact that a coral responds to an appropriate symbiont strain, but from active exclusion of other strains through immunity and apoptosis [37] , [63] . Evolution of association with novel algal strains could therefore be enabled by mutations in recognition receptors typically responsible for their exclusion, such as immunity genes. Several genes thus far implicated in the establishment of coral-algae partnerships may indeed be broadly responsible for allorecognition and immune response regulation, such as glycans and lectins [64] , [65] , fasciclin [66] , and MAPK-kinase and NF-kappa-B [67] . The latter two genes regulate antimicrobial response in invertebrates [68] , which is somewhat different from their function in mammals. In this study, Gamma-IFN-inducible lysosomal thiol reductase, lipopolysaccharide-binding protein, and Toll-like receptor 2, all implicated in the innate immune response to bacterial pathogens, displayed elevated rates of evolution. While our evolutionary screen between two coral species allowed for the delineation of fast-evolving functional categories, ultimately one is interested in identifying the specific genes and amino acid sites that are under adaptive evolution. Conducting similar analyses in a multi-species framework will make it possible to investigate this question in a robust statistical framework, allowing for amino acid site- and species-specific identification and characterization of positively selected genes. However, at the moment we have next to no information about the evolutionary mechanisms that brought about morphological, ecological, and physiological diversity of corals. This study provides an initial birds-eye view of genome-wide evolutionary patterns in corals and will serve as a guide for subsequent studies focusing on finer details of adaptation. Some of the genes that we highlighted in this initial screen may be responsible for thermal adaptation and therefore be targets of natural selection driven by increasing seawater temperatures as a consequence of climate change. They therefore represent a meaningful set of genes providing working hypotheses to look for genetic markers of climate change-driven evolution."
} | 5,165 |
35648822 | PMC9191775 | pmc | 837 | {
"abstract": "Significance Enabling distributed neurologic and cognitive functions in soft deformable devices, such as robotics, wearables, skin prosthetics, bioelectronics, etc., represents a massive leap in their development. The results presented here reveal the device characteristics of the building block, i.e., a stretchable elastomeric synaptic transistor, its characteristics under various levels of biaxial strain, and performances of various stretchy distributed neuromorphic devices. The stretchable neuromorphic array of synaptic transistors and the neuromorphic imaging sensory skin enable platforms to create a wide range of soft devices and systems with implemented neuromorphic and cognitive functions, including artificial cognitive skins, wearable neuromorphic computing, artificial organs, neurorobotics, and skin prosthetics.",
"discussion": "Discussion To enable stretchable distributed neuromorphic and cognitive functions in a soft skin device, the building block, i.e., a stretchable elastomeric synaptic transistor, was entirely developed using elastomeric electronic materials. Using elastomeric rubbery materials for all the device components is the key to realizing a biaxially stretchy elastomeric synaptic transistor and its array-based neuromorphic devices. The elastomeric synaptic transistor was biaxially stretched up to 30% and exhibited a full set of synaptic behaviors, including EPSC, PPF, and memory characteristics similar to the biological synapse. Stretchy distributed neuromorphic devices based on biaxially stretchable synaptic transistor arrays, memorization, and cognitive functions were successfully exhibited, including image memorization, long-term memorization, fault tolerance, and programming/erasing function under 30% biaxial mechanical strain. To implement the neuromorphic processing and cognition capability in a stretchy distributed device, a fully stretchy neuromorphic imaging sensory skin device based on a 5 × 4 sensing node array was constructed. Its neuromorphic pattern reinforcement function was validated even when the skin device was deformed nonuniformly. With the recent surge of smart skin devices ( 52 , 53 ), implementing neuromorphic functions into these devices open the door for a future direction toward more powerful biomimetics limited not just to neuromorphic imaging but also to multimodal somatic sensing and many others. By leveraging neuromorphic hardware and computing power with sensing technologies, e.g., a soft elastic system in this work, a broad range of benefits, such as high-accuracy, low-cost sensing and recognition could be easily reached. The methodology of implementing neuromorphic functions into smart skin devices as shown in this work could be exploited into many other areas, such as wearable neuromorphic computing ( 28 ), artificial organs ( 54 ), neurorobotics ( 55 ), and skin prosthetics ( 53 ) toward the next-generation intelligent systems."
} | 730 |
33506264 | PMC7842101 | pmc | 841 | {
"abstract": "Abstract Background Biogas production with anaerobic digestion (AD) is one of the most promising solutions for both renewable energy production and resolving the environmental problem caused by the worldwide increase in organic waste. However, the complex structure of the microbiome in AD is poorly understood. Findings In this study, we constructed a microbial gene catalog of AD (22,840,185 genes) based on 1,817 Gb metagenomic data derived from digestate samples of 56 full-scale biogas plants fed with diverse feedstocks. Among the gene catalog, 73.63% and 2.32% of genes were taxonomically annotated to Bacteria and Archaea, respectively, and 57.07% of genes were functionally annotated with KEGG orthologous groups. Our results confirmed the existence of core microbiome in AD and showed that the type of feedstock (cattle, chicken, and pig manure) has a great influence on carbohydrate hydrolysis and methanogenesis. In addition, 2,426 metagenome-assembled genomes were recovered from all digestate samples, and all genomes were estimated to be ≥80% complete with ≤10% contamination. Conclusions This study deepens our understanding of the microbial composition and function in the AD process and also provides a huge number of reference genome and gene resources for analysis of anaerobic microbiota.",
"conclusion": "Conclusions Here, we present a microbial gene catalog of AD, by using in-depth sequencing of the digestate samples from 56 full-scale BGPs treating diverse feedstocks, and provide >22.8 million taxonomically and functionally annotated genes. Our results confirmed the existence of core microbiome in AD and showed that the type of feedstock (cattle, chicken, and pig manure) has a great influence on carbohydrate hydrolysis, VFAs oxidation, and methanogenesis. Additionally, we also provided 2,426 MAGs derived from full-scale BGPs. Compared to previously published microbial gene catalogs of different ecosystems such as soil, ocean, and animal gut and rumen [ 20 , 32 , 56–59 ], BGPs are man-made extremely anaerobic ecosystems where AD is performed by a complex consortium of anaerobic microbes. Hence, our gene catalog will not only serve as a useful reference database for quick analyses of AD microbiome data but also provide a huge number of microbial gene resources for the study and utilization of anaerobic microbiota."
} | 584 |
34877492 | PMC8633030 | pmc | 842 | {
"abstract": "Summary Maintaining the superhydrophobicity underwater offers drag resistance reduction, antifouling, anti-corrosion, noise reduction, and gas collection for boat hulls and submarine vehicles. However, superhydrophobicity typically does not last long underwater since the Cassie state is metastable. Here, we report a reversible and localized recovery of superhydrophobicity from the fully wetted state via air bubble spreading. Composed of sparse fluorinated chained nanoparticles, the submerged surface shows super-low energy barrier for bubble attachment. Especially the recovered plastron exhibits excellent longevity. Based on a simplified, truncated nanocone model, the dynamic spreading of bubbles is analyzed considering two basic parameters, i.e., surface geometric structure and surface energy (which appeared as intrinsic water contact angle). Numerical simulation results via COMSOL confirms the effect of geometric structure on bubble spreading. This investigation will not only offer new insights for the design of robust recoverable superhydrophobic surfaces but also broaden the applications of superhydrophobic coatings.",
"conclusion": "Conclusions The metastable Cassie state in submerged underwater environments could severely destroy the superhydrophobic surfaces. Many aquatic insects and spiders with sparse, hairy surfaces can survive when submerged by virtue of plastron replenishment from gas dissolved in water. Mimicking Nature, we create aerophilic surfaces as a respiratory skin for submarine vehicles, which facilitates the air bubble pinning and stable spreading on the skin, allowing for underwater recovery of the superhydrophobic surface from the fully wetted state using bubble injection. Through the self-limited assembly of chained NPs on the surface, a super-low repulsive force barrier (0.4 nN) is measured for bubble attachments upon being submerged in water. The durability tests demonstrate that the recovered plastron can suffer from standing for more than 24 h, 120 cycles of ethanol wetting-bubble injection, and high water pressure (<25 kPa). Considering two basic parameters, i.e., surface geometric structure and surface energy (which appeared as the intrinsic water contact angle), the dynamic process of bubbles on a simplified truncated nanoconed surface was analyzed. We further obtained specific geometric criteria for the air bubble pinning and lateral spreading without vertical escaping, which are validated by the numerical simulation. The study here offers new insights for the design of recoverable underwater superhydrophobic surfaces, where the air bubble pinning and stable spreading on such surfaces are essential. In turn, it will broaden the applications of superhydrophobic coatings to underwater environments, where, for example, drag reduction, anti-fouling, anti-corrosion, and underwater breath are important for the operation and lifetime of ship hull, submarine vehicles, and equipment but have remained challenging to achieve so far because of the metastable nature of the nonwetting Cassie state.",
"introduction": "Introduction Surfaces that can maintain superhydrophobicity underwater are of great interest for ship hulls and submarine vehicles for potential applications including underwater breath ( Flynn and Bush, 2008 ), gas collection ( Yong et al., 2018 ), drag reduction ( Srinivasan et al., 2015 ; Wang et al., 2014 ), antifouling ( Wang et al., 2019 ), anti-corrosion ( Vilaró et al., 2017 ; Zhang et al., 2020 ), and acoustic blocking ( Huang et al., 2020 ). Bioinspired superhydrophobic surfaces ( Feng et al., 2020 ; Gurumukhi et al., 2020 ; Li et al., 2019b ) are usually achieved by hierarchical structures with low surface energy that keep a thin layer of air (plastron) trapped under water, named as the Cassie state ( Cassie and Baxter, 1944 ). However, the plastron is metastable ( Duan, 2017 ): large fluctuations of external conditions (such as pressure and liquid flow) can cause the superhydrophobic surface to transform from the Cassie state into the Wenzel state or fully wetted state. Much effort has been devoted to extending the entrapment time of plastron via micro-nano structure design ( Domingues et al., 2018 ; Hu et al., 2017 ; Liu and Kim, 2014 ). Nevertheless, this passive strategy is not able to eliminate the hidden risk of submarine vehicles during practical applications once plastron is disrupted. Therefore, the active replenishment of plastron from the wetted state will be highly desired, e.g., (1) using an external field, (2) in situ gas generation, and (3) gas injection. Despite the success of achieving temporary transition from the Wenzel state to the Cassie state using methods such as vibration ( Boreyko and Chen, 2009 ), electrical ( Manukyan et al., 2011 ), temperature ( Adera et al., 2013 ), water pressure ( Forsberg et al., 2011 ), or gas concentration ( Ling et al., 2017 ; Vakarelski et al., 2013 ) changes, these approaches have not been tested in practical submerged environments. Plastron could be regenerated via in situ chemical reactions ( Aebisher et al., 2013 ; Lee and Kim, 2011 ; Lee and Yong, 2015 ; Lloyd et al., 2017 ; Panchanathan et al., 2018 ); however, this strategy is not repeatable or localizable. In contrast, the gas injection method ( de Maleprade et al., 2016 ; Ma et al., 2015 ), viz. using injected gas bubbles on the aerophilic surfaces to recover plastron by draining liquid film, is simple, reversible, localizable, and instantaneous. Dewetting transition between the Cassie and partial Wenzel states on hierarchically structured surface has been observed ( Boreyko and Collier, 2013 ; Verho et al., 2012 ; Wu et al., 2017 ) owing to the retention of nano-plastron. The effects of geometric parameters on the dewetting on cone/pillar arrays have been theoretically discussed ( Li et al., 2019a ; Zhang et al., 2016 ). However, two key problems remain unresolved: (1) no report is available on reversible recoverable superhydrophobicity from the fully wetted state and (2) no general geometric criteria are provided for air bubble pinning and spreading to recover superhydrophobic surfaces. In order to achieve an underwater recoverable superhydrophobic surface from the fully wetted state via gas injecting, two questions should be addressed. First, is superhydrophobicity in air equivalent to underwater superaerophilicity? Various superhydrophobic surfaces have been proved to be superaerophobic in the fully wetted state but superaerophilic when plastron remains ( Pei et al., 2018 ; Shan et al., 2018 ; Yong et al., 2019 ). As illustrated in Figure 1 A, in air a water droplet cannot repel the gas anchored on the hierarchical structures. Vice versa, in an aqueous environment the air bubble cannot repel the water imbibed in the nanostructures. Video S1 shows a non-sticking bubble on the wetted lotus leaf. Second, is underwater aerophilicity sufficient to lead to the plastron recovery? As presented in Figure 1 B, air bubble can adhere on the wall of the hydrophobic polyethylene terephthalate (PET) bottle, revealing underwater aerophilicity. However, the air bubble could not anchor and will disappear or run up under buoyancy ( Video S1 ). Therefore, a nano/micro-structure that is both superhydrophobic and aerophilic to facilitate the pinning and spreading of air bubbles is required in order to recover the superhydrophobicity from the fully wetted state for submarine applications. Figure 1 Water/air behaviors on different surfaces (A) Optical image and SEM image of lotus leaf. Schematic for the lotus microstructures revealing superhydrophobicity in air but underwater superaerophobicity in the fully wetted state. Images from high-speed camera showing non-sticking bubble on wetted lotus leaf. (B) A PET bottle of water with air bubbles. Schematic illustrations exhibiting the hydrophobicity in air and underwater aerophilicity of PET. Images from high-speed camera showing bubble running up on tilted PET sheet. (C) A picture of water bug Notonecta and its SEM image of setae on the abdominal sternites. Schematic for a promising sparse microstructure with superhydrophobicity, aerophilicity that facilitate the pinning and spreading of air bubbles. \n Video S1. Submerged dynamic bubble behavior on lotus leaf and PET sheet, related to Figure 1 \n To design such a strategy, we have sought inspiration from the sparse hairy surface structure of nature aquatic arthropods of water bug ( Ditsche-Kuru et al., 2011 ) or spider ( Seymour and Hetz, 2011 ). As shown in Figure 1 C, the water bug Notonecta glauca is surrounded by a thin film of air covering most of body parts with a silvery sheen appearance. Expectedly, a sparse nanostructure with low surface energy facilitates the bubble to repel the water in the nanostructure and regenerates plastron. In our previous work ( Pei et al., 2018 ), an ultrathin superhydrophobic coating with two-tiered surface nanotextures assembled by chained nanoparticles shows excellent resistance to high water pressure or impingement of small water droplets. Here we demonstrate, for the first time, a reversible, localized recovery of plastron from the fully wetted state using bubble injection. The unique sparse structure shows ultralow repulsive force of ∼0.4 nN to air bubbles with radius ∼80 μm before the bubble-surface attachment under water. Furthermore, based on a simplified truncated nanocone model, dynamic bubble spreading behavior is studied considering two basic parameters, i.e., surface geometric structure and surface energy (appeared as intrinsic water contact angle). We develop basic criteria of nanotextured parameters for bubble adhesion and then lateral spreading without vertical growth. Modeling of the bubble spreading via COMSOL confirms the geometric effect of the surface structures."
} | 2,457 |
35558060 | PMC9090926 | pmc | 843 | {
"abstract": "To promote the water vapor condensation efficiency in the presence of a non-condensable gas, both high nucleation rate and efficient droplet departure are desired on the condensing surface. Superhydrophobic surfaces with large water contact angles ensure the dropwise condensation mode and efficient droplet departure ability. Alternatively, efficient nucleation requires the surface to be hydrophilic. To combine these two seemingly contradictory factors on a single surface, we presented a copper-based hydrophilic-slippery surface in this study by depositing a lubricant (trimethoxysilane) on the microstructured copper substrate. The water droplet had both low contact angles and sliding angles on the surface, and stable dropwise condensation could be realized with and without non-condensable gas. The present hydrophilic-slippery surface demonstrated promising potential to enhance condensation heat transfer, particularly for cases with non-condensable gas. Improved droplet mobility was observed as compared to a superhydrophobic surface, hydrophobic surface, and hydrophobic-slippery surface. The most attractive feature lies in the enhanced nucleation process due to hydrophilicity, which is more favorable as it requires small subcooling degree and large non-condensable gas content. By revealing that a sliding angle could be accompanied by a small contact angle, this hydrophilic-slippery surface could improve our understanding in designing new functional surfaces for phase change, anti-icing, self-cleaning, and anti-fouling applications.",
"conclusion": "5. Conclusions In this work, a copper-based HPI-SLIPS with both low water contact angle and small SA was fabricated. Stable DWC was observed and enhanced heat transfer coefficient was tested for the first time on HPI-SLIPS with NCG. When the nitrogen volumetric content was 9.2%, the heat transfer coefficient for HPI-SLIPS was enhanced by 113.8–120.3% and 27.4–44.8% as compared to those of HPO and HPO-SLIPS, respectively, within the subcooling degree ranging from 33.6 K to 10 K. The present HPI-SLIPS demonstrated an effective enhancement of heat transfer, which can be attributed to the efficient nucleation ability and high droplet mobility, particularly highlighted for a small subcooling degree or large NCG content. The present HPI-SLIPS reveals that the SA can be accompanied by a small CA, which could improve our understanding in designing new functional surfaces for phase change, anti-icing, self-cleaning, and anti-fouling applications.",
"introduction": "1. Introduction The vapor condensation heat transfer process in the presence of non-condensable gas (NCG) is widely observed in several industrial fields, such as seawater desalination, fog collection, energy utilization, chemical industry, energy saving, water saving, nuclear industry, etc. 1,2 The mechanism of condensation involving NCG is obviously different from that observed in pure water vapor condensation. 3 When the dew point evaporation technology is used for desalination, air is used as the carrier gas, and pre-heated seawater or brackish water is used to humidify and dehumidify, yielding fresh water: the NCG in the condensation process has a mass content of 50–90%. 1 In the process of flue gas latent heat recovery in gas boilers, NCG accounts for about 80% of the total volume, and the volume of water vapor accounts for about 20%. 2 These industrial processes inevitably face the problem of water vapor condensation in the presence of NCG, whose presence complicates the condensation heat transfer process, increases power consumption, and deteriorates performance. Therefore, it is of significant importance to determine methods to enhance the heat transfer performance of vapor condensation in the presence of NCG. In 1929, Othmer 4 firstly conducted experimental research of steam condensation in the presence of NCG. His experimental results showed that the surface heat transfer coefficient of a copper tube could decrease to 50% when the air volume fraction in the boiler rose from 0 to 0.5%. Thereafter, several experimental studies 5–7 have been conducted to investigate the effect of NCG during condensation on the horizontal or vertical surfaces and a variety of NCG types have been employed, namely, air, nitrogen, argon, neon, hydrogen, etc. The results of these studies have shown that every type of NCG can inhibit condensation heat transfer. It has been reported that vapor condensation includes filmwise condensation (FWC) and dropwise condensation (DWC). DWC is the preferred condensation mode due to its better thermal performance. However, there have been a large number of studies on FWC with NCG involving experimental 8,9 and numerical simulation methods; 10,11 however, studies related to DWC with NCG are limited. In the 1960s, Tanner et al. 12 firstly conducted an experimental study to compare pure-steam DWC with NCG DWC on a vertical wall under low-pressure conditions. The results revealed that the heat transfer coefficient was independent of NCG at lower concentrations, but the inhibition of condensation increased with the NCG concentrations. Since then, studies on DWC with NCG have not been conducted for a long time until the recent successful fabrication of several functioned surfaces with specific wettability, which has highlighted this topic once again. 13,14 It is widely observed and believed that DWC forms on a hydrophobic surface, while FWC occurs on a hydrophilic surface. According to this idea, various hydrophobic surfaces (HPOs) or superhydrophobic surfaces (SHPOs) have been fabricated based on the lotus effect. 15 Here, water droplets are suspended on a composite solid–air interface caused by the air trapped in the micro/nanostructures and high surface free energy, because of which the droplets can roll away easily. The improved mobility of droplets is demonstrated to successfully enhance the heat transfer characteristics of DWC, particularly when coalescence-induced droplet jumping occurs. 16 However, it is found that a SHPO could impede the nucleation process because according to the nucleation theory, the presence of a large contact angle (CA) elevates the formation energy barrier of a liquid cluster. 17 This problem is particularly serious when the condensation suffers from small subcooling degree and the existence of NCG. Moreover, the coalescence-induced jumping effect on SHPOs does not only requires extremely small surface adhesion and weak wettability, but it is also prone to losing the jumping ability under a large sub-cooling degree (common occurrence in condensation with NCGs). 18 Water collection ability of living things has enabled various inspirational ways of promoting condensation. For example, a superhydrophilic surface with a honeycomb structure on the skin of several lizards facilitates water absorption from the air and the condensed water is transported into their mouth by a capillary network on the skin. 19 Spider silk is composed of humidity-sensitive hydrophilic flabelliform proteins and can form periodic spindle-shaped knots after wetting. 20 Directional water collection can begin owing to the capillary propulsion stimulated by the conical geometry of the spindle knots. Certain cacti possess conical spines and trichomes with a hydrophilic surface and hierarchical grooves such that water droplets can be efficiently transported by the Laplace pressure and wettability gradient. 21,22 All the abovementioned examples have one thing in common: a hydrophilic surface. It has been reported that a hydrophilic surface ensures a higher nucleation process than the hydrophobic surface. These living prototypes demonstrate that the nucleation ability is also a critical factor contributing toward efficient water harvesting apart from the high droplet mobility, particularly for the cases with NCG in which the nucleation process is impeded by the air layer. Motivated by these living prototypes, various biomimetic artificial surfaces have been developed that can combine hydrophilicity with high droplet mobility. The existing methods to combine these two factors are summarized in three ways according to different driving forces. (i) Fabricate a hydrophilic surface and arrange conical geometries inspired by spider silk. Droplets can condensate easily on the hydrophilic surface and spontaneously move due to the Laplace pressure gradient. 23,24 (ii) Fabricate a surface with a wettability gradient starting from the hydrophilic patterns. Without external forces, droplets can move directionally from the hydrophobic region to the hydrophilic region due to the wettability gradient. 25–28 The first two ways can be combined by fabricating a conical fiber surface with a wettability gradient, 29 mimicking a cactus spine 21 with a gradient in both curvature and wettability. (iii) Fabricate a surface with mixed wettability patterns of hydrophilicity and hydrophobicity, mimicking a desert beetle's back. For the first and second ways, in spite of the effective water collection on the roughness-gradient conical fiber, it is difficult to apply conical geometry in the case of large-scale flat surfaces extensively existing in industries. It seems to be acceptable that the hydrophobicity for efficient water transport essentially contradicts the hydrophilicity for condensation enhancement. Therefore, the third way provides an attractive solution to this contradiction based on a spatial combination of hydrophobicity and hydrophilicity. Various methods have been reported in the literature to fabricate hydrophilic–hydrophobic surfaces, and the nucleation density and condensate removal efficiency are successfully improved. 30–33 However, an undesired pinning effect can be observed on the border of different wettability under certain conditions. 34,35 Although hydrophobicity is essentially incompatible with hydrophilicity, there is no evidence that high nucleation ability and efficient droplet shedding are naturally contradictory. An alternative inspiration to improve droplet mobility, different from the lotus effect, is observed in the Nepenthes pitcher plant, 36,37 where an intermediary liquid is locked within the surface microtextures, resulting in a liquid film on its surface. Inspired by this, a slippery liquid-infused porous surface or an oil-infused hydrophobic surface 38 have been recently proposed, recommending a lubricating liquid coating on the microstructured surface. This surface exhibits excellent omniphobic, anti-icing, antifrosting, and antifouling characteristics. 39–43 The high droplet mobility of an oil-infused hydrophobic surface could be attributed to the low friction of the composite water–liquid/solid interface instead of the traditional water–solid interface. It is promising to fabricate a hydrophilic surface with a small sliding angle (SA) to combine high nucleation ability and efficient droplet shedding. As mentioned above, a condensing surface is required to have both high nucleation ability and efficient droplet mobility to enhance the heat transfer performance of vapor condensation. The nucleation ability is an important, but often neglected, factor, which is highlighted in the cases with NCG where the nucleation process is impeded by the air layer. Superhydrophobicity—favorable for high droplet mobility—is incompatible with superhydrophilicity that favors efficient nucleation. In this work, we proposed a copper-based hydrophilic-slippery surface (HPI-SLIPS) motivated by the Nepenthes pitcher plant, having both small CA and small SA. This design provides a promising approach for enhancing both water nucleation and droplet departure, which successfully reconciles the two seemingly conflicting requirements and exhibits an enhanced heat and mass transfer performance.",
"discussion": "4. Results and discussion 4.1 Effect of lubricant thickness on surface wettability It is well known that surface wettability is sensitive to the thickness of the lubricant oil. To determine what lubricant thickness needs to be used in the experiments, firstly, we investigated the effect of lubricant thickness on surface wettability. Considering that the volume of the lubricant droplet is difficult to control and a large measurement error is involved in such small volumes, we take advantage of lubricant evaporation to make the lubricant thinner. There were a total of 60 samples placed in a large clean box under ambient conditions. Every sample was deposited with 12 lubricant droplets with around 3 μL of each droplet on the copper surface (3 cm × 3 cm) with microstructures. We took 5 samples in an electronic balance (AL204, METTLER TOLEDO) and recorded their masses every 12 h. At the same time, three samples were taken out to measure their CA and SA. The averaged values were recorded and plotted ( Fig. 4 ). Corresponding schematic diagrams were also plotted. The lubricant thickness was calculated as follows: 5 th = m O /( ρ O s ), where th refers to the lubricant thickness; m O refers to the lubricant mass, which equals to the difference between the sample mass measured on that very day and the mass measured before lubricant deposition; and s refers to the sample area. It should be noted that we did not consider the microstructures on the sample surface for simplicity. The actual lubricant thickness could be marginally higher if the lubricant volume was maintained to be the same. Fig. 4 Effect of lubricant thickness on surface wettability. Four wetting states are observed and the corresponding illustrative diagrams are shown. (A) Hydrophilic slippery state; (B) slippery Wenzel state; (C) sticky Wenzel state; (D) superhydrophobic Cassie state. From Fig. 4 , it is evident that four different wetting states were found as the lubricant got thinner. The lubricant decreased quickly in the first three days, but the CA and SA were still very low. The lubricant was able to completely cover the microtextures, so the droplet was still in a hydrophilic slippery state, marked as A. As the lubricant thickness continued to decrease, the CA and SA increased abruptly on the third day, reaching 55° and 10°, respectively. Thereafter, the CA was nearly unchanged and the SA increased slowly as the lubricant thickness decreased. This can be attributed to the fact that the microtextures were exposed and wetted by the droplet, forming a Wenzel state. However, the droplet still had high mobility on the surface due to sufficient lubricant retained in the microtextures, which can be attributed to the so-called slippery Wenzel state, 44 marked as B. With further evaporation, the decreased lubricant occupied only parts of the gaps between the microtextures. The remaining lubricant was not sufficient to maintain the slippery Wenzel state, while the exposed spaces were not able to form air pockets, so the droplet completely wets the textures and pins onto the surface, forming a traditional sticky Wenzel state, marked as C. On the 13 th day, there was almost no lubricant left on the surface and the air pockets between the microtextures were able to form induced by the microtextures and low free energy self-assembled monolayer, resulting in the superhydrophobic Cassie state with a large CA and small SA, marked as D. Therefore, to obtain the hydrophilic slippery state, we deposited 12 droplets with around 3 μL of each droplet when fabricating the HPI-SLIPS. 4.2 Surface wettability The measured CA and SA of the water droplet on the five surfaces are shown in Fig. 5 with corresponding schematic diagrams for each surface. The untreated copper surface is hydrophilic (CA = 72 ± 1°), which is favorable for nucleation, but the droplet strongly pins on the surface (SA = ∞), which is difficult to move. The hydrophobicity on HPO (CA = 128 ± 1°; SA = 41 ± 2°) is induced by the low free energy self-assembled monolayer without any microtextures. The water droplet on the SHPO could form the Cassie–Baxter state due to the air trapped within the microtextures and the low free energy self-assembled monolayer, resulting in large CA (153 ± 2°) and small SA (41 ± 2°). Both the SAs on HPO-SLIPS and HPI-SLIPS are extremely low (2 ± 1°), which is attributed to the composite water–liquid/solid interface, whose friction is much lower when compared with the traditional water–solid interface. HPO-SLIPS exhibits hydrophobicity (CA = 118 ± 3°), while HPI-SLIPS (CA = 32 ± 3°) exhibits hydrophilicity, which is caused by different interface tensions between the water and lubricant oil. Fig. 5 Surface wettability and corresponding illustrative diagrams of (a) untreated copper surface, (b) HPO, (c) SHPO, (d) HPO-SLIPS, and (e) HPI-SLIPS. Not to scale. 4.3 Droplets behavior on condensing surface To investigate the dynamics of the condensed droplets, the condensation processes with and without nitrogen on the five tested surfaces were observed and captured with a high-speed camera (Phantom Miro M110, USA) and a microscopic lens (QM100, QUESTAR). Example transient images of the condensed droplets on the five surfaces under stable condensation are shown in Fig. 6 . From this figure, it is evident that condensates on the untreated copper surface flooded and formed a film. In contrast, condensates formed discrete droplets on SHPO, HPO, and HPO-SLIPS with large CAs independent of the presence of nitrogen. It should be noted that a sustainable DWC was also observed on HPI-SLIPS, whose CA is only 32°, which is even smaller than the untreated copper surface. Fig. 6 Example transient images of condensed droplets on the five surfaces under (a) pure vapor condition and (b) with NCG when the condensation becomes steady. Scale bar: 5 mm. According to the classical nucleation theory, 45 the minimal formation work for a liquid cluster with a diameter of r is equal to the change in the Gibbs free energy, which is known as the Gibbs free energy of the formed cluster. With increasing r , the Gibbs free energy initially increases to the maximum Δ G and then decreases. Δ G is called the Gibbs free energy barrier, which describes the formation resistance of a liquid nucleus on a flat surface and strongly depends on the CA θ LV of the surface: 6 Δ G = π γ LV r * 2 (2 − 3 cos θ LV + cos 3 θ LV )/3, where γ LV is the liquid–vapor surface energy and r * is the critical radius, which can be obtained by Kelvin's classical equation. The nucleation rate, J , is strongly influenced by the CA θ LV of the surface as follows: 7 J = J o exp(−Δ G / kT ), where k is the Boltzmann constant and J o is a kinetic constant. The energy barrier and nucleation rate are normalized as Δ G * = Δ G /Δ G (180°) and J * = J / J (180°), respectively, whose dependence on the CA is plotted in Fig. 7 according to eqn (6) and (7) , with subcooling degree Δ T = 3.05 K and vapor temperature T = 373 K. For comparison, the CAs of the five test surfaces are also plotted in the figure. From this figure, it is evident that the nucleation energy barrier continuously increases with the CA, indicating that the hydrophobic surfaces have higher Δ G * as compared to hydrophilic surfaces under identical conditions. Consequently, the nucleation rate on the HPI-SLIPS is significantly higher than that on the other surfaces, particularly compared to the surfaces with hydrophobic and superhydrophobic properties. Fig. 7 Dependence of normalized nucleation energy barrier Δ G * (dashed line) and normalized nucleation rate J * (solid line) on the CA with Δ T = 3.05 K and T = 373 K. The corresponding CAs of the four tested surfaces are also marked by dashed lines. To enable a detailed comparison of the four surfaces with DWC, randomly timed images with higher magnification when the condensation became stable were captured, as shown in Fig. 8 (under pure vapor conditions) and Fig. 9 (under nitrogen). The moment when the capture process is started was defined as 0 s. Droplets observed moving are marked with dashed circles. Moreover, to formulate a quantitative analysis of the condensates, an automated image post-processing script was developed in MATLAB to recognize and characterize the size of the droplets on the surface. The surface coverage, droplet density, average droplet radius, and critical departure radius on the four surfaces were calculated every 30 s and averaged for 30 min. The results are shown in Fig. 10 . Fig. 8 Example transient images of condensed droplets on the surfaces. φ v = 0%; Scale bar: 500 μm. Fig. 9 Example transient images of condensed droplets on the surfaces. φ v = 9.2%; Scale bar: 500 μm. Fig. 10 Quantitative condensate analysis of the steady DWC in 30 min. The droplet dynamics were fairly the same with and without NCG. In the early stages, all the condensates were uniformly formed on the surfaces. Droplets began to merge with their neighbors after increasing in size, and the areas were then refreshed with new small condensates beginning to nucleate and grow. When reaching the critical departure radius, condensates began falling and sweeping droplets on their way. The surface was then refreshed, allowing more sites for heat transfer and nucleation. The falling droplets increased in size by falling and coalescing with other droplets on their way, whose velocity increased due to gravity and the surface-to-kinetic energy transfer during droplet coalescences. From Fig. 8 , it is evident that under pure vapor condensation, one droplet was observed to be moving in the field of view on SHPO and HPO until 16.34 s and 7.62 s, respectively. In contrast, both HPO-SLIPS and HPI-SLIPS exhibit efficient nucleation, taking less than 0.44 s for droplets to nucleate and increase up to sufficient size to be visible. In the case with NCG, the droplet growth rate on SHPO was higher than HPO, but all of them were slower than SLIPS. It took 3.46 s to observe the droplet to move on HPO-SLIPS. However, it is surprisingly quick for droplets to move on HPI-SLIPS, which was as short as 0.44 s—the same as that in the case of pure vapor condensation, exhibiting efficient nucleation even with NCG. There were obviously more droplets observed moving on the HPI-SLIPS than the other three surfaces under pure vapor conditions. When NCG was present, a nitrogen layer was formed beneath the condensing surface. The nucleation was impeded and droplets found moving were, therefore, decreased. HPI-SLIPS still had the largest number of moving droplets observed despite being remarkably less than that under pure vapor condition. The critical departure radii on SHPO and HPO were larger than the field of view, so there were no moving droplets marked in Fig. 9(a) and (b) . The calculated critical departure radius on the HPI-SLIPS was about 220 μm ( Fig. 10(a) ), which was the smallest among these four surfaces. Moreover, the number of condensed droplets on the HPI-SLIPS was far more than those on the other three surfaces with smaller average droplet radii ( Fig. 10(b) and (c) ). Hence, the condensates were able to actively merge with each other from the earliest stage and were removed with small departure radii under gravity during condensation. It should be noted that the critical departure radius of the HPI-SLIPS was smaller than that of HPO-SLIPS despite the same small SA ( Fig. 10(d) ). According to ref. 46 , the critical departure radius should be the same if the SA is maintained to the same value. However, in our cases, condensates on HPI-SLIPS cannot only be driven by gravity, but also be driven by droplet coalescence. Considering that there were many more condensed droplets on HPI-SLIPS, droplets coalesced more frequently, thereby generating more kinetic energy and resulting in a smaller critical departure radius and larger droplet velocity than those in HPO-SLIPS. It should be noted that the droplet falling velocities on SHPO and HPO were higher, which should be attributed to the much larger droplet weight and dominating effect of gravity. However, the large critical departure radius made the high falling velocity meaningless. With regard to condensation with nitrogen, the critical departure radius was increased, droplet density and average droplet radius decreased, and droplets moved slower on the condensing surface, which was caused by the nitrogen layer beneath the condensing surface. 4.4 Heat transfer performance on HPI-SLIPS with and without NCG The heat transfer coefficients of pure vapor condensation on the untreated copper surface, SHPO, HPO, HPO-SLIPS, and HPI-SLIPS under different subcooling degrees were measured and shown in Fig. 11(a) . The untreated copper surface had the lowest heat transfer coefficient because of the large thermal resistance caused by the flooded liquid film. All the other four surfaces maintained steady DWC mode but differed in heat transfer performance. The heat transfer coefficients of SHPO, HPO, HPO-SLIPS, and HPI-SLIPS increased sequentially, regardless of the subcooling degree. The condensation heat transfer coefficient on HPI-SLIPS was enhanced by 38.7–55.4% and 2.2–18.9% when compared to the typical SHPO and HPO-SLIPS within the subcooling degree ranging from 24 K to 2 K, respectively. This improvement was higher when the subcooling degree was lower. Fig. 11 (a) Overall heat transfer performance of pure vapor condensation against subcooling degree. (b) Overall heat transfer performance of condensation with NCG against subcooling degree. (c) Overall heat transfer performance of condensation against NCG content. The effects of the NCG on the heat transfer performance along with the subcooling degree and nitrogen content are shown in Fig. 11(b) and (c) . The untreated copper surface still had the worst heat transfer performance, but the SHPO performed better than the HPO. This is because SHPO had smaller CA hysteresis when there was nitrogen beneath the surface. Air pockets were able to form in the microtextures and Cassie–Baxter state condensates were able to form, leading to higher droplet mobility than that observed in the HPO. However, in pure vapor condensation, droplets condensed within the microstructures and formed the undesired sticky Wenzel state. The pinning contact line made the droplet hard to move. It should be noted that the HPI-SLIPS still had the highest heat transfer coefficient with nitrogen. The condensation heat transfer coefficient on HPI-SLIPS was enhanced by 113.8–120.3% and 27.4–44.8% as compared to those of HPO and HPO-SLIPS within the subcooling degree from 33.6 K to 10 K when the nitrogen volumetric content was 9.2% (see Fig. 11(b) ). Moreover, this enhancement improved when the nitrogen content increased (see Fig. 11(c) ). The measured droplet density on the HPI-SLIPS reached 2.71 × 10 9 /m 2 ( φ v = 0%) and 2.56 × 10 9 /m 2 ( φ v = 9.2%), which was almost twice as the other surfaces. Combined with a smaller average droplet radius and a smaller droplet departure radius, there were an increased number of smaller droplets existing on HPI-SLIPS. Droplets with small radii and small CAs were more efficient for heat transfer than large droplets. On the other hand, a large NCG amount leads to an increase in the thickness of the NCG layer above the condensates and further results in larger thermal resistance. Therefore, the actual temperature of the saturated vapor above the solid surface is reduced, which means that the effective subcooling degree is reduced. A decreased subcooling degree leads to an increase in the energy barrier of the condensation formation. Therefore, the enhancement of the HPI-SLIPS on the nucleation ability is highlighted, particularly when the NCG content is very high when compared with the HPO-SLIPS surface. Further, these frequently sliding droplets cause disturbance to the NCG layer, which also reduces the thermal resistance and facilitates heat transfer. In addition, considering that the evaporation rate of the lubricant decreases as the circumstance temperature decreases, the HPI-SLIPS will have higher sustainability with an increase in NCG content. The promising potential of HPI-SLIPS, with efficient nucleation and high droplet mobility, to enhance condensation heat transfer was shown, particularly for cases with NCG. However, there are still certain issues that need to be resolved for engineering applications. Lubricant drainage faced by the lubricant-impregnated surface is an inevitable problem. For the application of condensation, the temperature of the condensing surface could be higher than the ambient conditions, which might expedite the lubricant evaporation process. Methods to retain the lubricant on the surface and achieve longer sustainability should be explored in future works."
} | 7,176 |
37091376 | PMC10116617 | pmc | 845 | {
"abstract": "Water-based superamphiphobic coatings are environment-friendly,\nwhich have attracted tremendous attention recently, but the performances\nare severely limited by the dispersibility of hydrophobic particles.\nTo solve the poor dispersibility of modified silica powder with hydrophobicity,\nsilica dispersion was blended with polytetrafluoroethylene (PTFE)\nemulsion and modified aluminum tripolyphosphate (ATP) dispersion to\nsuccessfully prepare water-based coatings. Multifunctional coatings\nwere prepared by one-step spraying. It possessed good adhesion (grade\n1), excellent antifouling, impact resistance, chemical stability (acid\nand alkali resistance for 96 h of immersion), and corrosion resistance\n(3.5 wt % NaCl solutions for 20 days). More importantly, the superamphiphobic\ncoatings had high contact angles (CAs) and low slide angles (SAs)\nfor ethylene glycol (CAs = 154 ± 0.8°; SAs = 13 ± 0.7°)\nand water (CAs = 158 ± 0.7°; SAs = 4 ± 0.3°).\nFurthermore, the composite coating was still hydrophobic after 35\ncycles of wear with high roughness sandpaper (120 mesh) under three\ndifferent loads, which maintained superamphiphobicity at 425 °C.\nThis work is expected to provide a facile idea and method for the\npreparation of waterborne superamphiphobic coatings.",
"conclusion": "4 Conclusions In summary, modified silica\ndispersion was blended with PTFE emulsion\nand modified ATP dispersion to prepare an aqueous coating without\nintroducing an external surfactant. The composite coating possessed\ngood adhesion (grade 1), thermal stability (<425 °C), excellent\nantifouling, and impact resistance. More importantly, the superamphiphobic\nsurfaces had obtained strong repulsion to ethylene glycol (CAs = 154\n± 0.8°; SAs = 13 ± 0.7°) and water (CAs = 158\n± 0.7°; SAs = 4 ± 0.3°). Under different loads\n(50, 100, and 200 g), the coating still possessed hydrophobicity after\n35 cycles of wear of high-roughness sandpaper (120 mesh). Moreover,\nthe composite coating had long-term chemical stability to HCl solution\n(pH 1) and KOH (pH 14) solution for 96 h. The polarization curve and\nelectrochemical impedance analysis showed that the PTFE-ATP-SiO 2 composite coating exhibited long-term corrosion resistance\n(20 days), which was mainly due to the superhydrophobicity of the\ncomposite coating and physical barrier effect. It is believed that\nthe PTFE-ATP-SiO 2 composite coating has promising potential\nin the corrosion protection.",
"introduction": "1 Introduction The study of superamphiphobic\nsurfaces originated from bionics,\nand oil-repellent organisms on the surface were found in nature, 1 such as bacterial biofilm ( Bacillus\nsubtilis ), 2 leafhopper\nskin, 3 fish skin, and springworm skin. 4 − 6 Inspired by nature, researchers have attempted to develop superamphiphobic\ncoatings to solve industrial problems. 7 Due to excellent water and oil repellency, superamphiphobic coatings\nare widely used in self-cleaning, 8 antifouling, 9 anti-icing, 10 drag\nreduction, 11 anticorrosion, 12 chemical shielding, 13 and crude oil transportation. 14 The micro/nanostructures\nand low surface energy chemistry are inseparable factors for obtaining\nsuperamphiphobic coatings. 15 − 18 To obtain superamphiphobic coatings with excellent\nperformances, volatile organic compounds (VOCs) are generally used\nto a disperser to enhance the dispersion of coatings, such as toluene, 19 , 20 acetone, 21 , 22 ethyl acetate, 23 and dichloropentafluoropropane. 24 They are harmful to the environment and human health. In recent\nyears, the preparation of environment-friendly waterborne superamphiphobic\ncoatings attracted researchers’ interest. However, it seemed\ncontradictory to disperse hydrophobic substances in water, which tended\nto float on the water surface. 25 Poor dispersion\nwill lead to the low performance of superhydrophobic coatings, and\nit is not easy to realize oil repellency on the surface. Therefore,\nimproving the dispersion of the waterborne coating will be conducive\nto replacing solvent superhydrophobic and oil repellent coatings. At present, a lot of work has been done to solve dispersibility,\nand many researchers have made attempts. Some studies have reported\nsimple methods for obtaining dispersions of aqueous superhydrophobic\ncoating by dispersing silica particles in silane/siloxane/organic\npolymer waterborne emulsion and coating on different substrates to\nobtain superhydrophobic surfaces. 26 − 30 Although these waterborne superhydrophobic coatings\nwere prepared by various methods to enhance the dispersion of coatings,\nthere were few reports on the exploration of dispersion stability.\nIn the preparation methods of a rough surface, some reports had chemically\netched the surface of aluminum alloy with HNO 3 /HCl/HF, 31 , 32 CuCl 2 , 33 and NaOH solution. 34 It had been treated with low surface energy\nmaterials to fabricate the superamphiphobic surfaces. Chemical etching\navoided harmful VOCs, but acid and alkali solutions endangered human\nhealth and affected the mechanical properties of the alloy to a certain\nextent. Zhang et al. prepared the dispersion of tetraethyl orthosilicate\n(TEOS), 1 H ,1 H ,2 H ,2 H -perfluorodecyltriethoxysilane (PFDTES), and\nnanosilica by improving the Stöber method and then sprayed\nit on an AA5083 alloy substrate to obtain superamphiphobic coatings. 35 Although the dispersion of superamphiphobic\ncoatings with excellent dispersibility was fabricated by one pot,\nthe wear resistance of the surface has not been discussed. In addition,\nthe stability was not explored. Li et al. used a four-step spraying\nmethod to successively form a coating system of polyurethane, silica\nfluoride, methyl silicone, and silica fluoride, which obtained superamphiphobic\nsurfaces with excellent performance. 36 However,\nthe preparation steps are too cumbersome. In this study, dodecyltrimethoxysilane\nwas used to modify silica,\nwhich reduced the hydrophilic groups of nanosilica and enhanced the\nhydrophobicity. Compared with the zeta potential of silica dispersion\n(0.1 g/L, 17.45 ± 0.95 mV), the zeta potential of modified silica\ndispersion (0.1 g/L, 32.8 ± 0.9 mV) was larger due to the steric\nhindrance of dodecyltrimethoxysilane on the silica surface. 37 − 39 Therefore, the modified silica dispersion was more stable. To solve\nthe dispersibility of hydrophobic particles, silica dispersion was\nblended with polytetrafluoroethylene (PTFE) emulsion and modified\naluminum tripolyphosphate (ATP) dispersion to successfully prepare\nwater-based dispersion. The water-based dispersion was sprayed and\ncured at 320 °C to fabricate superamphiphobic surfaces. It obtained\nexcellent moisture resistance for ethylene glycol (CAs = 154 ±\n0.8°; SAs = 13 ± 0.7°) and water (CAs = 158 ±\n0.7°; SAs = 4 ± 0.3°). Furthermore, the polytetrafluoroethylene-modified\naluminum tripolyphosphate-modified silica (PTFE-ATP-SiO 2 ) composite coating possessed good adhesion (grade 1), thermal stability\n(<425 °C), excellent antifouling, and impact resistance. More\nimportantly, the modified ATP improved the density of the PTFE-ATP-SiO 2 composite coating and was regarded as an anticorrosive material\nto enhance corrosion resistance. 40 , 41 After 20 days\nof immersion in 3.5% NaCl solution, the PTFE-ATP-SiO 2 composite\ncoating still had 89.1% protection efficiency. This work is expected\nto provide a facile idea and method for the preparation of waterborne\nsuperamphiphobic coatings with high temperature resistance and anticorrosion\nby one-step spraying.",
"discussion": "3 Results and Discussion 3.1 Analysis of Modified Silica Nanoparticles The prepared modified silica nanoparticles are shown in Figure 2 a. The appearance\nwas uniformly spherical, and the CAs of particles were 153 ±\n0.5° for deionized water. The average particle size was 462.25\n± 4.65 nm ( Figure 2 b). Compared with silica, the infrared spectra of modified silica\nshowed absorption peaks at 2855 cm –1 (symmetric\nstretching vibration peak) and 2924 cm –1 (antisymmetric\nstretching vibration peak), which belonged to methylene (CH 2 ) and methyl (CH 3 ), respectively. In addition, the hydroxyl\nabsorption peaks at 1638 and 3420 cm –1 were weakened\nand hydrophilic groups were reduced compared with silica ( Figure 2 c). 43 − 45 It indicated that silica was successfully modified by dodecyltrimethoxysilane.\nThe mass percentage of silica-branched dodecyltrimethoxysilane was\nmeasured by a synchronous thermal analyzer. For modified silica, the\nadditional weight loss rate was about 3.91% compared with unmodified\nsilica ( Figure 2 d). Figure 2 Characterization\nof modified silica nanoparticles. (a) SEM and\nCAs of modified silica nanoparticles. (b) Particle size distribution.\n(c) FT-IR spectra. (d) Thermogravimetry of modified silica nanoparticles. 3.2 Analysis of the Composite Coating Dispersion For water-based superamphiphobic coatings, dispersibility decides\nthe performance. However, the hydrophobic substances often float on\nthe water surface, resulting in poor dispersibility ( Figure 3 a,b). Therefore, we solved\nthe dispersibility by preparing modified silica dispersion instead\nof modified silica powder and obtained water-based superamphiphobic\ndispersion ( Figure 3 e,f). The prepared modified silica dispersion was milky white ( Figure 3 c). When the dispersion\nof modified nanosilica was stored for 210 days, it was clearly found\nthat the dispersion was not layered ( Figure 3 d). As shown in Figure 3 g, it was layered when the coating dispersion\nwas placed for 120 min, but the coating dispersion returned to the\noriginal state after passing through the high-speed stirrer ( Figure 3 h). In addition,\nthe coating dispersion after redispersion does not affect the coating\nspraying and performance ( Figure S1 ). Figure 3 Dispersion\nstability. (a) Nanosilica powder suspended in water.\n(b) Nanosilica powder suspended in the coating dispersion. (c) Dispersion\nof modified silica nanoparticles. (d) Dispersion of modified silica\nnanoparticles after 210 days. (e, f) PTFE-ATP-SiO 2 composite\ncoating dispersion. (g) PTFE-ATP-SiO 2 composite coating\ndispersion after 120 min. (h) PTFE-ATP-SiO 2 composite coating\ndispersion after high-speed dispersion again. Silica dispersion has a large phase interface,\nso the colloidal\nparticles are in an unstable state to be destroyed and agglomerated. 46 , 47 Consequently, the one-step synthesis and hydrophobic modification\nhave been adopted to improve stability for fabricating the modified\nsilica dispersion with excellent dispersibility. During the process\nof one-step synthesis of the modified silica dispersion, dodecyltrimethoxysilane\nwas used to modify the surface of dispersed phase particles (silica),\nreduce the surface energy, and play a role in steric hindrance on\nthe surface of silica. 37 − 39 Zeta potential (absolute value) is usually an important\nindex to evaluate the stability of the colloidal dispersion system. 48 , 49 The silica dispersion with 437.2 ± 13.1 nm average particle\nsize and the modified silica dispersion with 462.25 ± 4.65 nm\naverage particle size were synthesized, it was found that the modified\nsilica dispersion had a large zeta potential (0.1 g/L, 32.8 ±\n0.9 mV), and its stability was better than that of the silica dispersion\nby comparing with the zeta potential of the silica dispersion (0.1\ng/L, 17.45 ± 0.95 mV). To evaluate the stability of the\ncoating dispersion, the zeta potential\nof coating dispersions with different concentrations of silica (437.2\n± 13.1 nm) and modified silica (462.25 ± 4.65 nm) was measured\n( Figure 4 a). It was\nfound that with the increase in concentration, the zeta potential\nof the coating dispersion increased, indicating that the stability\nincreased. As shown in Figure 5 a, with the increase in particle concentration, the average\nparticle size of the coating dispersion prepared with modified silica\nincreased relatively slowly, while the average particle size of the\ncoating dispersion prepared with silica increased rapidly. The maximum\ndifference in average particle size between the two was 70.45 nm.\nTherefore, compared with the coating dispersion prepared with modified\nsilica, the coating dispersion prepared with silica was easier to\naccumulate. Furthermore, the zeta potential of modified silica was\nhigher than that of silica, which indicated that the stability of\nthe coating dispersion prepared with modified silica was good. Figure 4 Zeta potential\nof coating dispersion. (a) Effect of concentration\non zeta potential. (b) Effect of pH on zeta potential. Figure 5 Particle size distribution. (a) Effect of concentration\non the\naverage particle size. (b) Particle size distribution of dispersion. The zeta potential of coating dispersions prepared\nwith silica\n(437.2 ± 13.1 nm) and modified silica (462.25 ± 4.65 nm)\nwith different pH levels (2, 4, 6, 8, 10, and 12) was measured so\nas to explore the influence of pH on the stability of the coating\ndispersion ( Figure 4 b). The results showed that the zeta potential of the coating dispersion\nprepared with modified silica and silica under the alkaline condition\nwas higher than that under the acidic condition, and alkaline condition\nwas conducive to the stability of the coating dispersion. The composite\ncoating dispersion included PTFE emulsion (pH 9–10), modified\nsilica dispersion (neutral), and modified ATP dispersion (pH 6–7),\nof which PTFE emulsion was the main substance ( Figure 8 ). Therefore, the stability of the composite\ncoating dispersion was mainly determined by the stability of PTFE\nemulsion. For the stability of PTFE emulsion, it was determined by\nthe synthesis process and stable environment (acid–base properties).\nIn this paper, PTFE emulsion (TE3893) was an alkaline dispersion,\nso alkaline condition was conducive to improving the stability of\nthe composite coating dispersion. In addition, the effect of\nmodified silica particle size on the\nstability of the coating dispersion was investigated. It was found\nthat there was little difference in the zeta potential of the coating\ndispersion prepared with 278.1 ± 2.2 nm (0.1 g/L, 22.75 ±\n0.15 mV) and 462.25 ± 4.65 nm (0.1 g/L, 23.25 ± 0.15 mV)\nmodified silica. As shown in Figure 5 b, compared with the coating dispersion with 462.25\n± 4.65 nm modified silica, the coating dispersion with 278.1\n± 2.2 nm modified silica had a larger particle size (352.95 ±\n5.95 nm). Consequently, small particle size was not conducive to the\nstability of dispersion and led to easier accumulation. To sum up,\nthe modification, concentration, particle size, and pH have a great\nimpact on the stability of the coating dispersion. 3.3 Analysis of Wettability of the Composite Coating Since the coating dispersion was composed of PTFE, modified silica\nnanoparticles, and modified ATP, weight percentages of modified silica\nnanoparticles and modified ATP were changed to explore the influence\non the wettability of coating. Under the condition of 50 g of PTFE\nemulsion (60 wt %) and 50 g of modified ATP (20 wt %), the amount\nof modified silica was changed (five kinds of formulations). It was\nobvious that the waterproof ability of the coating can be improved\nby adding modified silica nanoparticles ( Figure 6 a,b). When the weight percentage of modified\nsilica nanoparticles was 36 wt % (25 g), the maximum CA of water reached\n158.7°, and the CAs of ethylene glycol were 151 ± 0.6°.\nWhen the weight percentage of modified silica nanoparticles was 40.3\nwt % (30 g), the composite coating had the maximum CA (154.8°)\nfor ethylene glycol, the CAs of water were 155 ± 1° ( Figure 6 a). When the weight\npercentage of modified silica nanoparticles changed from 18.37 wt\n% (10 g) to 47.37 wt % (40 g), the SAs of water were transformed from\n9 ± 0.2° to 3 ± 0.1°, and the SAs were changed\nfrom 17.9 ± 0.6° to 13 ± 0.5° for ethylene glycol.\nWith the amount of filled modified silica increasing, the composite\ncoating surface easily formed an inhomogeneous rough structure and\nhad high apparent contact angle. It conformed to the Cassie–Baxter\nmodel. 50 In the Cassie–Baxter model,\nit was assumed that the droplet only contacts with the solid tip (inhomogeneous\nsurface) and intercepts air in an inhomogeneous rough structure under\nthe droplet so that the inhomogeneous surface is not wetted. The apparent\ncontact angle increased with the decrease in the contact between the\ndroplet and the solid tip, which was basically consistent with our\nexperimental results ( Figure S2 ). Therefore,\nunder the quantitative condition of PTFE (50 g) and modified ATP (50\ng), the composite coating with 36 wt % (25 g) modified silica nanoparticles\nobtained high CAs. Then, the amount of modified ATP was optimized,\nwhich was changed under the dosage of 50 g of PTFE emulsion (60 wt\n%) and 25 g of modified silica (90 wt %). Five kinds of formulations\nwere evaluated ( Figure 6 c,d). When the weight percentage of modified ATP was changed from\n10.26 wt % (30 g) to 21.05 wt % (70 g), the CAs of water altered from\n151 ± 0.9° to 156 ± 0.7°, and the SAs of water\ndecreased to less than 4°. However, when the weight percentage\nof modified ATP exceeded 14.63 wt % (45 g), the SAs of ethylene glycol\nwere increased, and the CAs of ethylene glycol were decreased. Because\nwhen the weight percentage of modified ATP exceeded 14.63 wt % (45\ng) in the coating system, the hydrophilicity and lipophilicity of\nmodified ATP gradually appeared. Ten kinds of formulations were comprehensively\nconsidered, and the composite coating with 25 g of modified silica\n(90 wt %), 50 g of modified ATP (20 wt %), and 50 g of PTFE (60 wt\n%) obtained the best superamphiphobicity. Therefore, we chose the\ncomposite coating as the follow-up study and named it the PTFE-ATP-SiO 2 composite coating. When the modified ATP was 50 g and PTFE\nwas 50 g, the coating was named the PTFE-ATP composite coating. Figure 6 Wettability\nof the composite coating. (a, b) Effect of modified\nsilica content on the wettability of the composite coating for water\nand ethylene glycol. (c, d) Effect of modified ATP content on the\nwettability of the composite coating for water and ethylene glycol.\n(e, f) CAs and SAs of the PTFE-ATP-SiO 2 composite coating\nto different pH solutions. The PTFE-ATP-SiO 2 composite coating\nhad high CAs and\nlow SAs from pH 1 to 14 solution, proving that the PTFE-ATP-SiO 2 composite coating possessed good repulsion for corrosive\ndroplets and prevented corrosive liquid from wetting the composite\ncoating surface ( Figure 6 e,f). As a result, the PTFE-ATP-SiO 2 composite coating\nhas a promising application in a corrosive environment. 3.4 Analysis of Morphology and Composition As shown in Figure 7 a, the surface of coating was prepared by spraying pure PTFE emulsion.\nDue to many voids, it was not conducive to anticorrosion of coating.\nPure PTFE emulsion was mixed with modified ATP dispersion (20 wt %)\nto prepare the coating, which improved the compactness of coating\nand formed a continuous surface ( Figure 7 b). SEM images of the prepared PTFE-ATP-SiO 2 composite coating are shown in Figure 7 c,d. It was obvious that the surface of the\nPTFE-ATP-SiO 2 composite coating owned many hierarchical\nmicro/nanostructures similar to the surface of lotus leaves. This\nstructure intercepts air to form a stable gas film and enables the\ncomposite coating to show excellent water repellency. 51 According to the surface scanning energy spectrum of the\ncomposite coating, it was composed of carbon, oxygen, fluorine, zinc,\naluminum, silicon, and phosphorus ( Figure 8 a). The content of\nfluorine was the highest to reach 57.7 wt %, which provided low surface\nenergy for the composite coating. The micro/nanostructures were also\ncomposed of these seven elements, and the content was basically the\nsame as that of the composite coating surface ( Figure 8 c). From the element distribution ( Figure 8 b,d), it was clear\nthat the elements were distributed uniformly in the composite coating\nsurface, and the dispersibility of the prepared composite coating\nwas excellent. As shown in Figure 9 c, the PTFE-ATP-SiO 2 composite coating with\n20 μm thickness was obtained after high-temperature curing (320\n°C). The thickness of the film was obtained by measuring the\nheight of the step at the junction of the substrate leak and the film\n( Figure 9 b). The 3D\nsurface morphology of the composite coating is shown in Figure 9 a, and the arithmetic mean\ndeviation of the profile (Ra) was 1.648 μm. Figure 7 SEM of the composite\ncoating. (a) SEM of pure PTFE coating. (b)\nSEM of the PTFE-ATP composite coating. (c, d) SEM of the PTFE-ATP-SiO 2 composite coating. Figure 8 Element distribution of the composite coating. (a, b)\nElement content\nand element distribution of the PTFE-ATP-SiO 2 composite\ncoating. (c, d) Element content and element distribution of emulsion\nprotrusion. Figure 9 Morphology of the PTFE-ATP-SiO 2 composite coating.\n(a)\n3D surface morphology of the composite coating. (b, c) Profile of\nthe composite coating. Compared with the infrared spectra and XRD phase\nanalysis for modified\nATP at normal atmospheric temperature, it was found that some changes\nhave taken place in the infrared spectra and XRD phase when modified\nATP was kept at 320 °C ( Figures S3 and S4 ). The corresponding diffraction peaks of AlPO 4 were found\nafter high-temperature curing, but the corresponding diffraction peaks\nof H 2 AlP 3 O 10 ·H 2 O\nand Zn 3 (PO 4 ) 2 ·4H 2 O were not found ( Figures S5 and S6 ).\nIt was also consistent with the infrared spectrum result; the peak\nat 424 cm –1 (Zn–O) disappears ( Figure S3 and Table S1 ). Moreover, the infrared\nspectra of PTFE were changed, but the XRD spectra were not changed\nat a high temperature of 320 °C for 20 min ( Figures S3 and S4 ). These results affect the infrared spectra\nand XRD phase analysis of the PTFE-ATP-SiO 2 composite coating\nafter high-temperature curing (Figures S3, S4, and S7). 3.5 Analysis of Self-Cleaning and Antipollution A 10 min continuous rolling test (water and glycerol) and slurry\ndumping (45 wt %) test were executed to evaluate the antifouling performance\nof the PTFE-ATP-SiO 2 composite coating. The test results\nshowed that the PTFE-ATP-SiO 2 composite coating had an\nexcellent antipollution ability ( Figure 10 a–c). During the test (10 min), deionized\nwater and glycerol were nearly spherical and quickly dripped ( Movies S1 and S2 ).\nIt still possessed high CAs for water after the 10 min continuous\nrolling test. It was worth mentioning that the slurry (45 wt %) was\ntested continuously for 10 min without any sediment invading the composite\ncoating ( Figure 10 d and Movie S3 ). The CAs were 145 ±\n0.5° for slurry, which showed that the stability of the PTFE-ATP-SiO 2 composite coating was good. The self-cleaning performance\nof the PTFE-ATP-SiO 2 composite coating needs to be observed.\nSilica and sand were placed on the coating surface, and then water\ndrops fell on the polluted area, which took away pollutants and formed\na clean surface ( Figure 10 e–h). The PTFE-ATP-SiO 2 composite coating\nshowed good self-cleaning performance. Therefore, the composite coating\nhas good stability and self-cleaning performance, which are conducive\nto practical application. Figure 10 Self-cleaning and antipollution test. (a) PTFE-ATP-SiO 2 composite coating surface. (b) Deionized water test. (c)\nGlycerol\ntest. (d) 45 wt % slurry test. (e, f) Self-cleaning test of silica.\n(g, h) Self-cleaning test of silt. 3.6 Analysis of Mechanical Properties The PTFE-ATP-SiO 2 composite coating was tested with a\ncross-cut tape test according to GB/T 9286 ( Figure S8 ). The PTFE-ATP-SiO 2 composite coating was cut\nto a 1 mm × 1 mm grid by exposing the surface with a wallpaper\nknife and then pressed, and the 3M test tape was pulled until it was\nremoved from the scored surface within 5 min, which was repeated three\ntimes. The peeling of scratch intersection after removal by the 3M\ntest tape was neglected, and it was recognized as grade 1 adhesion\n( Figure 11 a,b). After\nthe cross-cut tape test, the composite coating still had high repellency\nfor glycerol, ethylene glycol, and water ( Figure 11 d,e). Moreover, the composite coating surface\nstill had an anti-adhesion ability ( Figure 11 c). The impact resistance tester is shown\nin Figure 11 g, and\nit evaluated the impact resistance of the PTFE-ATP-SiO 2 composite coating when a heavy hammer (1 kg) was dropped at different\nheights. The impact resistance of the composite coating was evaluated\naccording to GB/T 1732-1993. The impact resistance was judged by observing\nthe crack and peeling of the delamination ring on the surface of the\ncomposite coating. There was no crack on the surface of the PTFE-ATP-SiO 2 composite coating ( Figure 11 f) because of the high content of PTFE in the film-forming\nmaterial and certain toughness after film formation. PTFE played a\nleading role in the PTFE-ATP-SiO 2 composite coating showing\nexcellent impact resistance and good adhesion. Figure 11 Mechanical property\ntest. (a, b, e) Surface of the PTFE-ATP-SiO 2 composite\ncoating before and after the cross-cut tape test.\n(c) Deionized water pull test. (d) Ethylene glycol rolling test. (f)\nThe surface of the PTFE-ATP-SiO 2 composite coating was\nimpacted by a heavy hammer at different heights. (g) The heavy hammer\nis a 1 kg coating impact instrument. High-roughness sandpaper (120 mesh) was adhered\non weights of different\nmasses with double-sided adhesive, which was placed on the composite\ncoating surface and pulled at constant speed. Then, it was pulled\nto the other side and returned to the original position again to record\nit once ( Figure 12 a). The CAs and SAs of the PTFE-ATP-SiO 2 composite coating\nwere recorded every five times. The surface morphology of the PTFE-ATP-SiO 2 composite coating under 50 g loads is shown in Figure 13 a,b; it was slightly\ndamaged, the micro/nanostructures were also worn, and the CAs of deionized\nwater were less than 150° after 35 cycles of wear. After 35 times\nof abrasion of the PTFE-ATP-SiO 2 composite coating under\n100 g loads, it was found that the micro/nanostructures on the coating\nsurface were ground flat, the modified silica in the composite coating\nwas exposed ( Figure 13 c,d), and the CAs were reduced to 145 ± 1.1° for water.\nIt was observed that the high-roughness surface of the PTFE-ATP-SiO 2 composite coating was completely polished after 35 times\nof wear under 200 g loads ( Figure 13 e), there were no micro/nanostructures ( Figure 13 f), and the CAs of water were\nreduced to 136 ± 0.7°. After 35 cycles of wear, compared\nwith 50 g loads, the wear of the composite coating was intensified\nunder 200 g loads. The micro/nanostructures disappeared and the rough\nsurface was lost, which affected the surface wettability of the PTFE-ATP-SiO 2 composite coating. As shown in Figure 12 b, the CAs of the PTFE-ATP-SiO 2 composite coating were decreased under three different loads, which\nintuitively proved the influence of micro/nanostructures and rough\nsurface on hydrophobicity. 52 , 53 Furthermore, the hardness\nof the PTFE-ATP-SiO 2 composite coating was grade 2B ( Figure S9 ). Figure 12 Abrasion resistance test. (a) Process\nof PTFE-ATP-SiO 2 composite coating wear resistance test.\n(b) Effect of different\nloads on the CAs. Figure 13 SEM of the PTFE-ATP-SiO 2 composite coating\nafter the\nwear resistance test. (a, b) SEM of the composite coating after 35\ncycles under 50 g loads. (c, d) SEM of the composite coating after\n35 cycles under 100 g loads. (e, f) SEM of the composite coating after\n35 cycles under 200 g loads. 3.7 Analysis of Thermal Stability For\nthe sake of testing the thermal stability, coating annealing tests\nwere carried out. The wettability of the PTFE-ATP-SiO 2 composite\ncoating after annealing for 1.5 h at different temperatures is shown\nin Figure 14 . The\nCAs of the PTFE-ATP-SiO 2 composite coating decreased slightly\nwith the increase in heat treatment temperature (within 425 °C)\nfor water (CAs = 152 ± 0.8°; SAs = 7 ± 0.6°) and\nethylene glycol (CAs = 150 ± 0.2°; SAs = 26 ± 1°).\nHowever, when the temperature exceeded 435 °C, the PTFE-ATP-SiO 2 composite coating lost superamphiphobicity. The results showed\nthat the composite coating can maintain excellent thermal stability\nin a wide high-temperature range. Figure 14 High temperature resistance test of the\nPTFE-ATP-SiO 2 composite coating. (a) Effect of heat treatment\non CAs and SAs of\nwater. (b) Effect of heat treatment on CAs and SAs of ethylene glycol. 3.8 Analysis of Corrosion Resistance The PTFE-ATP-SiO 2 composite coating was soaked in acid–base\nsolution to evaluate the hydrophobicity and chemical stability. As\nshown in Figure 15 , the change trend of the CAs and SAs was shown after soaking for\ndifferent times in the environment of pH 1 and pH 14, respectively.\nThe chemical stability of the composite coating was excellent for\nthe hydrochloric acid solution with pH 1 (within 96 h). There was\nno significant change in the CAs and SAs of the composite coating\nbefore and after soaking, and the composite coating surface was not\ndamaged ( Figure 16 a,b). Compared with hydrochloric acid solution with pH 1, the repellency\nof the composite coating was worse in KOH solution with pH 14. After\nsoaking for 96 h, the CAs dropped to 145 ± 0.9°, and SAs\nincreased to 10 ± 0.7°. It may be that with the increase\nin immersion time, the PTFE-ATP-SiO 2 composite coating\ncontained acid-modified ATP that causes a chemical reaction with alkaline\nKOH, leading to the weakening of surface hydrophobicity. However,\nthe morphology of the PTFE-ATP-SiO 2 composite coating had\nnot changed significantly, and the micro/nanostructures had not been\ndamaged ( Figure 16 c,d). Therefore, the composite coating possessed excellent chemical\nstability. Figure 15 Acid–base immersion test. (a) CAs of the composite\ncoating\nsoaked in acid–base solution for different times. (b) SAs of\nthe composite coating soaked in acid–base solution for different\ntimes. Figure 16 SEM of the PTFE-ATP-SiO 2 composite coating\nafter the\nacid–base immersion test. (a, b) SEM of the composite coating\nafter soaking in pH 1 solution for 96 h. (c, d) SEM of the composite\ncoating after soaking in pH 14 solution for 96 h. To explore the corrosion resistance of the PTFE-ATP-SiO 2 composite coating, it was soaked in 3.5 wt % NaCl solution\nto obtain\nTafel plots and electrochemical impedance spectra ( Figure 17 ). The relevant electrochemical\nparameters were obtained according to the polarization curve ( Table 1 ). As shown in Figure 17 a, compared with\npure tinplate and PTFE coating, the PTFE-ATP coating had an E corr positive shift, lower corrosion rate (−0.603\nV), and higher protection efficiency (96.367%). It was attributed\nto the anticorrosive substance (modified ATP) and reduction of coating\nsurface defects ( Figure 7 b). The corrosion resistance of the PTFE-ATP-SiO 2 composite\ncoating was better than that of the PTFE-ATP coating due to the physical\nbarrier in the composite coating (modified silica). Figure 17 Corrosion resistance\ntest. (a) Polarization curve test. (b) Polarization\ncurve of the PTFE-ATP-SiO 2 composite coating for different\ntimes. (c) EIS of the PTFE-ATP-SiO 2 composite coating for\ndifferent times. Table 1 Polarization Values for Uncoated and\nCoated Tinplate Substrates sample immersion time E corr (V) I corr (A/cm 2 ) CR (μm/year) PE (%) tinplate 6 h –0.767 3.457 × 10 –4 4021 PTFE 6 h –0.621 2.620 × 10 –5 305 92.421 PTFE-ATP 6 h –0.603 1.256 × 10 –5 146 96.367 PTFE-ATP-SiO 2 6 h –0.142 1.850 × 10 –6 21 99.465 PTFE-ATP-SiO 2 5 days –0.181 3.484 × 10 –6 40 98.992 PTFE-ATP-SiO 2 10 days –0.423 1.071 × 10 –5 125 96.902 PTFE-ATP-SiO 2 20 days –0.605 3.768 × 10 –5 438 89.100 The corrosion rate (CR) and protection efficiency\n(PE) of the coating\nwere obtained by corrosion current ( I corr ) according to formulas 1 and 2 . 54 , 55 1 2 where 3270 = 0.01\n× [1 year (in s)/96,497.8] and 96,497.8 = 1 F in Coulombs. Furthermore, V , M , and d represent\nthe valence (+2), atomic mass (55.845), and density (7.85 g/cm 3 ) of the tinplate substrate, respectively. I corr 0 and I corr c belong to the corrosion current of the bare tinplate substrate and\nthe corrosion current of the coated tinplate substrate (determined\nby the intersection of the linear portions of the anodic and cathodic\ncurves), respectively. For the sake of exploring the long-term\nprotection efficiency of\nthe PTFE-ATP-SiO 2 composite coating, the polarization curves\nand electrochemical impedance spectra of the PTFE-ATP-SiO 2 composite coating at different times were obtained ( Figure 17 b). As shown in Table 1 , the PTFE-ATP-SiO 2 composite coating had the largest\npositive displacement, low corrosion rate, and high protection efficiency\nafter soaking for 6 h. However, after soaking for 5 days, the polarization\ncurve of the PTFE-ATP-SiO 2 composite coating had a slight\nnegative shift compared with that of the coating soaking for 6 h.\nWith the increase in immersion time, the degree of negative shift\nof the PTFE-ATP-SiO 2 composite coating polarization curve\nincreased. After 20 days of immersion, the degree of negative shift\n(−0.605 V) was higher than that of the PTFE-ATP-SiO 2 composite coating soaked for 6 h, and the protection efficiency\ndecreased to 89.1%. Electrochemical impedance spectroscopy was in\ngood agreement with the polarization curve, which also proved that\nthe corrosion resistance of the PTFE-ATP-SiO 2 composite\ncoating was reduced ( Figure 17 c). With the increase in immersion time, the semicircle diameter\nof EIS decreased, and the protection efficiency decreased. The PTFE-ATP-SiO 2 composite coating had a long-term protection efficiency of\n20 days, because the modified silica in the composite coating prolonged\nthe time to reach the tinplate surface for corrosive substances. In\naddition, the air was intercepted by the rough structure of the composite\ncoating to improve the corrosion resistance. Therefore, the PTFE-ATP-SiO 2 composite coating can be used in a corrosive environment."
} | 8,524 |
23468605 | PMC3585391 | pmc | 847 | {
"abstract": "The spontaneous emergence of pattern formation is ubiquitous in nature, often arising as a collective phenomenon from interactions among a large number of individual constituents or sub-systems. Understanding, and controlling, collective behavior is dependent on determining the low-level dynamical principles from which spatial and temporal patterns emerge; a key question is whether different group-level patterns result from all components of a system responding to the same external factor, individual components changing behavior but in a distributed self-organized way, or whether multiple collective states co-exist for the same individual behaviors. Using schooling fish (golden shiners, in groups of 30 to 300 fish) as a model system, we demonstrate that collective motion can be effectively mapped onto a set of order parameters describing the macroscopic group structure, revealing the existence of at least three dynamically-stable collective states; swarm, milling and polarized groups. Swarms are characterized by slow individual motion and a relatively dense, disordered structure. Increasing swim speed is associated with a transition to one of two locally-ordered states, milling or highly-mobile polarized groups. The stability of the discrete collective behaviors exhibited by a group depends on the number of group members. Transitions between states are influenced by both external (boundary-driven) and internal (changing motion of group members) factors. Whereas transitions between locally-disordered and locally-ordered group states are speed dependent, analysis of local and global properties of groups suggests that, congruent with theory, milling and polarized states co-exist in a bistable regime with transitions largely driven by perturbations. Our study allows us to relate theoretical and empirical understanding of animal group behavior and emphasizes dynamic changes in the structure of such groups.",
"conclusion": "Conclusions Despite the multitude of local interactions that result in coordinated group motion we demonstrate that schooling golden shiners predominantly exist in three ‘fundamental’ dynamically-stable states of the underlying dynamics: swarm, milling and polarized motion. We establish that group states, and transitional behavior, can be represented in low-dimensional space, a projection that allows us to see the path taken by groups between the three dynamically-stable states as well as to relate the collective states exhibited to properties such as group size, individual speed and perturbations to the group. We note that it is possible that further collective states may be found within the classified dynamically stable regimes described here, but the present states are highly consistent with the theoretical predictions of three regimes A key question in the study of collective behavior is whether different group-level patterns result from all individuals responding to the same external factor [27] , or individuals changing behavior [1] , or whether multiple dynamically-stable collective states co-exist for the same individual behaviors [3] . Our results provide evidence for the importance of the two latter processes in the behavior of schooling fish: transitions from the swarm, to the milling or parallel group states (and vice versa) involve a social feedback whereby individuals adjust behavior—in this case their speed—in response to prevailing local conditions. Low average speeds among group members correspond to them occupying the dense, disordered swarm regime. Higher speeds correspond to higher local order (alignment among group members) and groups existing in either the milling or polarized state. Transitions between these states occur with negligible, or no, change in local density, order or speed; instead perturbations such as collisions with the boundary, or (seemingly stochastic) fluctuations in motion at the group edge (in the case of milling to polarized state transitions) or front (in the case of polarized to milling transitions) result in the group leaving one dynamically-stable state, and either then returning to that state, or transitioning to the alternative locally-ordered regime. Thus the milling and polarized states appear to be bistable; the state exhibited by the group effectively being dependent on starting conditions and/or the nature of perturbations, as well as the group size. Theoretically [3] and experimentally (analysis of shoals of 30 fish in small tank), milling states are seen to be less stable for small groups, when controlling for boundary condition effects. It is likely that the relationship we found between speed and local order is a generic feature of mobile groups with local interactions. Furthermore, qualitatively similar features have been observed in small groups (4 and 8 fish) of the giant danios [24] . A key challenge for animal behavior in this, and future, decades is to understand how the microscopic mechanisms of interactions among molecules, physiological systems and neural circuits result in behavior at higher levels of organization. Whereas we focused on collective behavior resulting from interactions among individual organisms, the general approach adopted shares commonalities with approaches that have successfully been used to characterize the dynamical properties of gene interaction networks [28] , neuronal circuits [29] , how locomotion is coordinated among limbs, each of which has many degrees of freedom [30] , and how the behavior of individual organisms (such as Caenorhabditis elegans ), despite apparent complexity, can be deconstructed into a discrete number of low-dimensional behavioral dynamically-stable states (see Stephens et al. [31] ). We suggest that development, and adoption of, such techniques in the behavioral sciences could facilitate the advent of increasingly integrative and quantitative insights. Our work demonstrates that such an approach to data collection and analysis can reveal underlying simplicity in the dynamical properties of collective behavior in groups. Collective behavioral states appear to result from both behavioral feedback processes whereby individuals both adopt, and influence, the behavior of near neighbors and also as multi-stable regimes in which individual behavior does not change, but rather perturbations induce relatively abrupt transitions between alternate and co-existing dynamically-stable behavioral states. Prey groups have been observed to switch states upon detecting a predator [32] and risk can be dependent on these states [9] , [33] . Whether the mechanisms for switching between states as identified here are somehow themselves adaptive would be an interesting question to address in future work.",
"introduction": "Introduction Many animal groups display coordinated motion in which individuals exhibit attraction towards others and also a tendency to align their direction of travel with near-neighbors [1] . The functional complexity of such aggregates are thought to result from relatively local, self-organizing interactions among individuals that endow many groups, such as flocking birds and schooling fish, with the capacity to move, respond to threats and make decisions collectively [2] . Thus the way in which collective dynamics emerge from inter-individual social interactions likely has profound consequences on the selection pressures experienced by group living organisms. Theoretical considerations of the self-structuring properties of groups have suggested that certain features of interaction among individuals may give rise to a relatively small number of specific collective states [3] – [7] . For example, models representing local repulsion, directional alignment and longer-range attraction among individuals [3] , [5] predict that groups in which individuals follow these behavioral rules exhibit only three collective states; a swarm state in which individuals aggregate but are locally and globally disordered, a milling state in which individuals have a high degree of alignment with local neighbors but overall the group exhibits a rotating milling formation (or torus, in three-dimensional space) and a polarized state in which individuals tend to be aligned with each other over a long range and consequently the group experiences net movement. While these patterns have all been observed in nature among different organisms [8] and can evolve in simulations of simple predator and prey behaviour [9] , experimental support for the existence of these states, and for any one system transitioning between them, has not previously been determined. In addition to displaying commonly observed group structures these models also emphasize an important unifying feature—that collectively moving animal groups can be considered as a dynamical system in which multiple group states, or dynamically-stable states, exist, and that collective properties, such as the spatio-temporal configurations exhibited, may be robust to exactly how behavioral tendencies such as repulsion, alignment and attraction are mediated. Furthermore, under this scenario, animal groups are also predicted to exhibit ‘multistability’ whereby more than one collective state coexist for identical individual behavior, with groups transitioning relatively quickly between the alternate structural configurations. In the model of Couzin et al. [3] , for example, the milling and polarized state co-exist over a region of parameter space (see Paley et al. [10] for a formal analysis of this bistable regime). The question of which collective behavior is adopted therefore depends on the initial configuration of the system, in this case the positions and orientations of individuals: if groups start in a relatively disordered state they tend to form a milling formation; but tend to remain in the polarized configuration if they begin sufficiently aligned with one another. Perturbations, such as disruption induced by predator attacks [1] , or more generally, sources of intrinsic [11] , [12] or extrinsic noise [13] , can cause the system to leave its existing dynamically-stable state and enter an unstable transitional regime. Depending on the degree and type of perturbation the group may find itself either drawn back towards the previous state, or if perturbed sufficiently far, to be drawn into the alternative dynamically-stable state. Thus even though the group behavior results from a large number of relatively local interactions, the group-level dynamics can be described using relatively few and simple lower-dimensional ‘order parameters’ that portray the collective dynamics, such as global polarization and the degree of collective rotation. This approach is familiar to us in physical systems where a multitude of different inter-molecular interactions result in only four fundamental states of matter; solid, liquid, gas and plasma. When properties such as density and energy are altered, a physical system can undergo ‘phase transitions’ between these states. Similarly, at a certain level of description, we can view animal groups as having the potential to exhibit abrupt changes in spatial or temporal patterns, and thus phase transition-like behavior (see Buhl et al. [14] for an experimental example of density-driven transitions in locust swarms). However, biological systems are not in equilibrium. There are no conserved thermodynamic quantities, such as momentum and energy, and individual motion typically results not from thermal fluctuations or external forcing, but from individual self-propulsion and decision-making. Nevertheless, the concept that local interactions reduce to common non-equilibrium collective states is a key insight provided by computational modeling of collective animal behavior [3] , [4] , [15] which relates more generally to phase transition theory in non-equilibrium systems [16] . Additionally, despite individual motion likely being governed by a complex stochastic decision-making process based on the positions, movement and size of neighbors (and even hidden features like individual's state), two recent studies of schooling fish—Katz et al. [17] and Herbert-Read et al. [18] —demonstrate that interactions can be effectively reduced to local tendencies to be repelled from, or attracted towards, neighbors. Similar evidence exists for aggregating ducks [19] and swarming locusts [20] . Despite these advances, to date, no experimental study has quantified the dynamical states of collective motion exhibited by any species of group-living animal, nor determined whether the general predictions of existing models of collective behavior hold—notably, that groups will exhibit, and transition among, relatively few (dynamically-stable) states. Here we investigate the emergence of macroscopic collective states under highly controlled laboratory conditions using schooling fish (golden shiner, Notemigonus crysoleucas ). This is a convenient model system for investigating collective behavior since individuals are relatively small (average length approximately 5 cm in our study), and naturally form highly cohesive schools in very shallow and still water [21] . Digital tracking of fish in a range of group sizes (from 30 to 300 fish) allows us to obtain detailed data regarding the individual positions and velocities of schooling fish over long periods of time. We use these data to analyze how group size and perturbations (driven both by inevitable contact with the boundary of the tank, but also by changes in motion by individuals within the group in the absence of boundary influence) affect group behavior and function to transition the group between alternative dynamical states.",
"discussion": "Results/Discussion Seven replicate experiments were conducted for group sizes 30, 70 and 150 fish, and three replicates were conducted for 300 fish (due to limitations in our capacity to house very large numbers of fish). Each replicate consisted of filming fish swimming in a large shallow tank (2.1 m×1.2 m, water depth 5 cm) for 56 minutes (at 30 frames per second). A similar approach has been taken previously [17] , [18] , [22] , [23] . Individual fish were tracked following the methodology of Katz et al. [17] to obtain time series of the positions and velocities. These time series constitute the raw data from which we base our analyses. To describe the collective structure of the fish shoals we use two order parameters, identical to previous categorization in simulation models [3] , [11] . First, the polarization order parameter O p , which provides a measure of how aligned the individuals in a group are. It is defined as the absolute value of the mean individual heading, where is the unit direction of fish number i . O p takes values between 0 (no alignment on average) and 1 (all fish are aligned). Second, the rotation order parameter O r , which describes a group's degree of rotation about its center of mass. To define this measure we introduce the unit vector pointing from the shoal's center of mass towards fish i . The rotation order parameter O r is then defined by the mean (normalized) angular momentum which, by construction also takes values between 0 (no rotation) and 1 (strong rotation). Time series of these two order parameters give us valuable information about the global structure of a group and how the structure changes during an experiment. However, important pieces of information are not captured, like the group density, the average individual speed, and how close a group swims to the tank boundary. For a fuller picture of the collective dynamics, we will also use these order parameters. Other group properties could also be measured. Collective states exhibited and the role of group size Throughout our entire period of filming the fish were cohesive, and for all group sizes three dynamically-stable collective behaviors were observed [3] ; the swarm (S), polarized (P) and milling (M) group state. Snapshots of these distinct patterns are shown in Fig. 1A for a group of 150 fish ( Videos S1 , S2 , S3 , S4 contain video extracts of all group sizes). As in [3] , these separate modes of motion can be categorized by only two structural properties (order parameters) of the group—its polarization O p and its degree of collective rotation O r . Groups repeatedly transitioned between these collective states, as is evident in representative time series of the order parameters shown in Fig. 1B . 10.1371/journal.pcbi.1002915.g001 Figure 1 Dynamical states of schooling fish. ( A ) Snapshots of a group of 150 golden shiners swimming in a shallow tank. The different images (thresholded for clarity) demonstrate the typical configurations displayed by the fish school: swarm state (S), polarized state (P) and milling state (M). ( B ) Extracts of time series of order parameters for groups of 30, 70, 150, and 300 golden shiners. Polarization O p (in blue) measures how aligned the fish are, while rotation O r (in red) measures the degree of rotation around the center of mass of the fish shoal. To demonstrate more clearly these dynamically-stable states, in Fig. 2 we consider the proportion of time groups spend in different regions of the two-dimensional phase space spanned by the order parameters O p and O r (red representing more time spent in a given region and blue the least time; black areas signify regions of the phase space not visited by groups in our experiments). While all three states of motion were manifest in all groups, there are also visible differences relating to increasing group size. For the smallest group size of 30 fish we see that the polarized group state predominates (high O p and low O r ). Only rarely did groups of this size exhibit swarm behavior (low O p and low O r ), and even less frequently did they adopt the rotating group state (low O p and high O r ). The fluctuations in the order parameters are also most frequent for this group size ( Fig. 1B ). For a group size of 70 fish the frequency of transitions decreases and the collective states corresponding to the three dynamically-stable states become clearly distinguishable as ‘hotspots’: the polarized state is no longer dominant, with milling and swarm behavior also being common. As group size is increased further, to 150 and then 300 fish, groups spend most of their time milling, displaying fewer transitions into (and among) the polar state and swarm state. For all group sizes the milling state has an equal probability of rotating clockwise and counter-clockwise, i.e. groups did not exhibit a handedness (see Fig. S1 ). 10.1371/journal.pcbi.1002915.g002 Figure 2 Density plots of polarization vs. rotation from experiments. The data shown are averaged over all replicates for each of the groups of 30, 70, 150, and 300 golden shiners. The order parameter space is divided into four regions—swarm (S), polarized (P), milling (M), and transition (T)—each being characterized by the dominant dynamical state of the fish school in that particular region. Different values of p min and p max were used for each group size to emphasize the density patterns and regions with no data are colored black. The insert in the 30 fish plot shows the density plot from an experiment with 30 fish and the tank area reduced to one tenth of the original. A natural consequence of increasing the number of fish in the tank is that the mean density within the experimental arena becomes higher, and hence the effects of the tank boundary become more pronounced. To reveal whether the higher density of fish per tank area in larger groups could cause the increased stability of the milling state we performed an experiment (4 replicates) with 30 fish in a smaller tank (0.66×0.38 m), which corresponds to the mean density of 300 fish in the larger tank. The density plot of the order parameters is shown as an inset in the 30 fish density plot in Fig. 2 , and reveals that confinement by the boundaries and higher mean density do not lead to increased time spent milling. However, the time spent in the polarized state was reduced; contact with the smaller tank caused this group size to exhibit more frequent transitions to the swarm state than when it was in the larger tank. At least for the 30 fish, and possibly as a general result, the presence of the boundary does not increase the stability of the milling state per se. Rather, the stability of milling is largely determined by the size of the group. Although the functional reason for milling is not yet known, it does, however, allow individuals to be locally polarized, which could be important for information transfer, while allowing the group to remain in a specific area. Swarm behavior allows the group to remain in an area but is locally disordered and this may be more susceptible to predation [1] . To gain further understanding of the relationship between group size and the stability of the different group structures we employed the canonical model of grouping of Couzin et al. [3] , in which there are no boundary interactions. Exploring the collective behavior of simulated individuals (see Methods for simulation set up and details) we find that the model produces qualitatively similar results across the range of group size in our experiments (30, 70, 150 and 300 agents), with the polarized states being dominant for the smallest group size, and an increasing proportion of the group's time is spent in the milling state as group size increases (see Fig. 3 ). 10.1371/journal.pcbi.1002915.g003 Figure 3 Density plots of polarization vs. rotation from simulations. The data shown are from simulations with 30, 70, 150, and 300 agents, employing a constant-speed agent based simulation model of collective behavior where no boundary is present (See Methods for simulation details). Regions with no data are colored black. As for the experimental data, the milling state grows in stability with group size. Another aspect of group size is the self-regulation of density. Theoretically, when more members are added to a group of self-propelled particles, the density can either remain approximately constant, in which case the system is H-stable, or the density can increase, and the system is catastrophic [4] . In our experiments, the individuals within the group regulate their spacing such that density tends to remain stable regardless of group size. The mean area occupied by the fish grows approximately linearly with group size and the packing fraction (and density) remains nearly constant (See Fig. S2 ). This regulatory behavior places the fish in the category of H-stable systems. Transitions between collective states To quantify the relation between group size and collective state we need to explicitly define the different states. Given the relatively clear demarcation of states revealed by our data in Fig. 2 we employed a simple approach in which we discretize the phase space in terms of the order parameters. The specific range of values were motivated by the high-density regions observed in Fig. 2 . We thus define that the school is in: the polar state (P) when O p >0.65 and O r <0.35; the milling state (M) when O p <0.35 and O r >0.65; and the swarm state (S) when O p <0.35 and O r <0.35. Outside these ranges we define the system to be in a transitional regime (T). On average, therefore, each region is characterized by the dominant dynamical state of the fish school within that particular region. The regions defining the dynamical states are overlaid the density plots in Fig. 2 (the qualitative nature of our results does not depend on the precise nature of how these regions are defined, see Fig. S3 ). As shown in Fig. 1B , groups frequently transitioned between the three collective states. A transition is considered completed if the group moves from one of the three states to another. By this definition, a school can move from one of the dynamical states, into the transition region, and then back to its previous state, without having undergone a transition. We quantify, statistically, the transitions between states and investigate how the transition patterns depend on the group size. As we previously saw in Fig. 2 , the proportion of time spent in the polarized state decreases strongly with group size from 0.57 to 0.26, 0.18 and 0.05 (30 to 300 fish shoals, respectively, light blue columns in Fig. 4A ). Likewise, the proportion of time in the milling state increases with group size from 0.03 to 0.18, 0.36 and 0.45 (yellow columns Fig. 4A ). The group size has however little effect on the proportion of time spent in the swarm state or in the transition region, which varies between 0.09 and 0.11 (dark blue columns Fig. 4A ), and 0.31 and 0.44 (brown columns in Fig. 4A ), respectively. The fraction of transitions between states, as illustrated by Fig. 4B , exhibit little variation between the group sizes. The only visible trend is a small increase in the number of transitions between the milling state and the swarm state (see also Fig. S4 for alternative graphics). 10.1371/journal.pcbi.1002915.g004 Figure 4 Statistics of state transitions. ( A ) Fraction of time spent in the different dynamical states shown for each group size. The error bars are showing the standard deviation measured across replicates. The group of 30 fish is predominantly in the polarized state, but less time is spent in this state with increasing group size (GLM: F1,22 = 36.21, P = 4.6e-06). As the group size increases, the groups gradually spend more time in the milling state (GLM: F1,22 = 19.31, P = 0.00023). The amount of time spent in the transition regime is high, but constant, for all group sizes (GLM: F1,22 = 0.67, P = 0.42). Across all group sizes (GLM: F1,22 = 0.053, P = 0.82) little time is spent in the swarm state. ( B ) Fraction of transitions from one state to another for the different groups. The error bars are showing the standard deviation measured across replicates. For all group sizes, the polar state predominantly transitions into the swarm state compared to the milling state (GLMM: F1,23 = 58.77, P<0.0001). The swarm state is dominated by transitions into the polarized state (GLMM: F1,23 = 55.69, P<0.0001). Although these transitions are consistent across group sizes, there is a significant interaction between group size and the frequencies of transitioning from the milling state to the swarm and polarized states (GLMM: F1,22 = 13.30, P = 0.0014). While the milling state tended to transition into the polar state, for 300 fish there was a roughly equal probability of transitioning to the polar or swarm states. ( C ) Rank plots showing the probability of being in a state longer than time T s before moving into a different regime. There was no significant difference between group sizes in the persistence of the polar state (GLMM: F1,22 = 0.58, P = 0.45), although group size increased the persistence of the transition (F1,22 = 24.52, P = 1e-04) and milling states (F1,22 = 17.54, P = 4.0e-04), and to a lesser degree, the swarm state (F1,22 = 5.31, P = 0.031). To complement this picture it is important to note two (inter-related) features that do change with group size: Firstly the rate at which groups exhibit transitions decreases as a function of group size (from 2.0 transitions/min for 30 fish, to 1.4, 1.2, and 0.8 transitions/min for 70, 150, and 300 fish respectively); Secondly, the stability of the milling state increases as a function of group size (the longest time a group of 30 spent milling was 17 s, this increased to 110 s, 708 s, and 1245 s for group sizes 70, 150 and 300, respectively. See Fig. 4C for rank plots of time spent in a state before transitioning). This means that the large proportion of time the 30 fish spent in the polarized state is an accumulated effect of many visits into the state, while the proportion of time the 300 fish spent in the milling state is greatly affected by the milling state being more stable for larger groups. A further feature of the transitional behavior of groups is that transitions from one dynamical state to another only accounted for 47% (n = 1943) of the total number of visits into the transition zone (n = 4119), counting only visits lasting longer than 1 s. This demonstrates that the schools experience frequent perturbations, of which some result in transitions to another state ( Fig. 4B ), and the others back to the preceding state. From observing the schooling behavior it appears that perturbations to the group act as triggers to transitions between collective states (see Videos S1 , S2 , S3 , S4 ) and we can identify two main sources for fluctuations that result in transitions; interactions with the tank wall (boundary effects) and fluctuations due to the intrinsically noisy nature of individual motion. We note that these processes are not mutually exclusive. Boundary effects and state transitions In order to reveal more clearly the role of boundary effects on state transitions we use our extensive time series to characterize the typical nature of transitions and relate these to whether the group tends to be relatively close to, or far from, the boundary. We present data for 150 fish in Fig. 5 (for other group sizes see Fig. S5 ). In Fig. 5A the arrows represent the average trajectories that groups take through O p and O r space when transitions occur and, unlike Fig. 2 , the density plot now depicts the distance d b from the center of mass of the group to the closest point at the tank boundary; red colors represent a relatively large distance and blue colors relative proximity to the boundary. A more detailed view of the transition dynamics is presented in Figs. 5B–D . Here, for each of the transitions from polar to milling, polar to swarm and milling to swarm state, the average trajectories are plotted as a velocity field in the O p and O r phase space overlaid on the density plot showing the distribution of trajectories (for the reverse transitions and other group sizes see Figs. S6 and S7 ). 10.1371/journal.pcbi.1002915.g005 Figure 5 Transition patterns for 150 fish. ( A ) Density plot of the smallest distance from the center of mass of the fish shoal to the tank boundary as a function of rotation and polarization (d min = 26 and d max = 52 cm). The overlaid arrows are the averaged trajectories of all transitions in the rotation-polarization phase space. ( B–D ) Density plots of transitions from polarized to milling state ( B ), from polarized to swarm state (C) and from milling to swarm state (D). Overlaid the density plots are the corresponding velocity fields of the transition data (in the rotation-polarization phase space). Plots of the reverse transitions and group sizes 30, 70 and 300 are provided in SI. The data shown in Fig. 5 verify that transitions happen both close to the wall of the tank, and in the center of the tank. For large parts of the transition region schools are relatively close to the boundary, such as from the polarized state to the swarm state, where most transitions take place ( Fig. 5C ). Although not as clear, the transitions between the milling state and the swarm state are also characterized by being, on average, closer to the wall ( Fig. 5D ). The exception is for transitions that occur with high values of the order parameters, that is, between the polarized state and the milling state ( Fig. 5B ), and vice versa. These tend to occur both close to and also away from the wall. This indicates, as evidenced by the video footage ( Videos S1 , S2 , S3 , S4 ), that both boundary and other triggering mechanisms are important for inducing transitions between collective states. When in the milling state, interactions with the boundary can result in a local increase in density near the wall, due to the inherently constrained nature of motion when abutting the boundary. This can cause the mill to transition into a polarized state as shown in Fig. 6A . Another way in which the mill can break down is due to the action of individuals at the group edge; if fish turn or move away from the edge of the group this can seed the unraveling of the milling formation into a polarized state. Conversely, when in a polarized state individuals at the front of the group can turn towards the main mass resulting in a perturbation that prompts the group to turn, potentially initiating the mill formation. This last example is evidenced in Fig. 6B and demonstrates that the milling formation can emerge as a group effect from the individual interactions—without direct interaction with the tank boundary. 10.1371/journal.pcbi.1002915.g006 Figure 6 Time-lapse examples demonstrating transition mechanisms. ( A ) Transition initiated by interaction with the tank wall. In the first picture, the fish is in a milling state, indicated by the blue arrow, and the lower part of the group is close to the tank wall (grey line). In the second picture, the interaction with the wall has caused a local increase in density, marked by the blue region, and a few individuals have started to turn opposite the milling direction. This cascades, and in the third picture the flow of the mill is interrupted as a large proportion is breaking away from the milling direction. The result is total unraveling of the milling state and transition into the polar state, seen in the last picture. ( B ) Transition from the polar state to the milling state, initiated by individuals in the shoal. In the first picture, the group is in a polar state, signified by the red arrow. A few individuals in the front, visible in the red region, have started to turn downwards. This leads the group as a whole into a sharp right turn, and as picture two demonstrates, the group is forced into a shape with larger curvature. Now, when the individuals in front of the group can spot the back of the group, they continue the turning and start following the back, as seen in the third picture. In the final picture, as the front individuals catches up with the tail, the loop closes and the transition into the milling state is complete. In the case of polarized groups the boundary also has inevitable consequences on transitions; a polarized group may swim directly towards a wall, or corner, of the tank. Individuals reaching the wall slow down and tend to become disordered (unaligned) and the group can transition into the swarm state. That this mechanism of transition is dominating is demonstrated both by the average transition path from polarized to swarm in Fig. 5A , which crosses a region where d b is small, as well as the short average time groups spend in the polarized state before transitioning, as shown in Fig. 4C . The relationship between speed, density and group state For tractability many previous models of animal grouping (including that of Couzin et al. [3] ) have assumed that individuals move at constant speed and social response is represented by adjusting direction of travel in response to the positions and/or orientations of near neighbors. Recently, however, two experimental studies on schooling fish, Katz et al. [17] , involving golden shiners ( Notemigonus crysoleucas —the species used here), and Herbert-read et al. [18] , involving mosquitofish ( Gambusia holbrooki ), have highlighted the importance of speed regulation to collective behaviors. In the former study it was found that individual social interactions can be approximated qualitatively by pairwise interactions that are functions of the position and speed of each individual. While the spatial nature of these interactions was found to be relatively independent of individual speed, the magnitude, or strength, of response to neighbors decreased greatly as individual speed decreased. Also Viscido et al. [24] found a positive correlation between average group speed and polarity for shoals of 4 and 8 giant danios ( Devario aequipinnatus ). This suggests that there is an important relationship between individual speed and the degree to which individuals coordinate their motion with neighbors, a relationship that is not captured in many models of collective motion [3] , [15] , [25] , [26] . Examining the relationship between the mean speed of individuals in the group and the ‘packing fraction’ (a measure of the density of individuals within the group) and the order parameters O p and O r , we observe that low speed is associated with the group being relatively dense and both locally- and globally-disordered (the swarm state). The two locally-ordered (milling and polarized) states are characterized by higher mean speed and a decreased packing fraction (see Fig. 7A for group size of 150 fish; this relationship is common among all group sizes as shown in Fig. S8 ). Consequently the relationship between density and order is the opposite of that predicted by the most studied models of grouping behavior, notably the Vicsek model [15] ; although we note that such simple models have been extremely useful in developing understanding of group dynamics for other animal aggregates, such as locusts [14] , and other species of schooling fish [23] . 10.1371/journal.pcbi.1002915.g007 Figure 7 Structural properties. ( A ) Density plots of packing fraction and average individual speed (averaged per frame) as functions of rotation O p and polarization O r for 150 fish. ( B ) The plot illustrates the correlation between individual speed and local polarization estimated in two ways from the underlying density maps (the example shown in the background is for 150 fish). The stapled curves are produced by averaging across individual speeds for each value of the order parameter; the solid curves from averaging across the order parameter values for each individual speed. The local polarization of an individual fish is defined as the polarization Op restricted to the area inside a circle with radius 15.6 cm (approximately 3 body lengths) centered at the individual fish. ( C ) Average individual speed at different radial positions in the milling state. ( D ) Average rotational order parameter at the same positions. The radial division of the milling state in ( C ) and ( D ) is constructed by centering six shells outside each other, where the outermost shell has a radius defined by the distance from the group's center of mass to the median distance of the five most peripheral fish (see illustration in Fig. S11 ). The width of each shell is the radius of the outer shell divided by six. The averages are calculated for each shell, where the outer shell even includes peripheral fish. From our data we cannot distinguish between two, not mutually exclusive, hypotheses regarding the causal relationship between speed and order; does decreasing speed induce local disorder through weakened social interactions, or does perception of local disorder reduce an individual's speed? Since golden shiners do not appear to explicitly respond to the body orientation of neighbors, and rather respond more-or-less exclusively to individuals' positions in space [17] , increasing speed likely increases local order. However, a dense, slow moving and disordered group is also likely to further reduce individual speed (not least through increased risk of collision) - thus both causal relationships likely co-exist. To demonstrate the plausibility (and indeed, generality) of speed-induced transitions, we return to the model employed above, from [3] . We verify that changing individual speed does result in qualitatively the same transitional behavior seen here; swarm behavior for relatively low speed and bistable milling and parallel group motion as individual speed increases. This result holds up to 150 agents. For groups of 300 agents the milling state is dominant and no instances of the polar state are found. (see Fig. 8 for results from simulations with 150 agents and Fig. S9 for remaining group sizes. Simulation details are found in Methods ). 10.1371/journal.pcbi.1002915.g008 Figure 8 Relationship between agent speed and order in constant-speed agent based simulation model with 150 agents. ( A ) Density plot of agent speed as function of rotation O r and polarization O p , revealing a bistable regime between the milling and the polar states for high speeds. ( B ) Normalized probability plot of polarization O p as function of agent speed. ( C ) Normalized probability plot of rotation O r as function of agent speed. The last two plots illustrate the bifurcation that occurs as the speed is increased, where the system transitions from a swarm state and is found either in a highly polarized state or in a milling state. (See Methods for simulation details). Local social interactions are similar across group size and dynamical state In order to deepen our understanding of the local dynamics we also quantified the relationship between the speed of an individual and the degree of order in its immediate vicinity (within a radius distance of 15.5 cm). In Fig. 7B we show the resulting relationship as a contour plot for groups of 150 fish (see Fig. S10 for corresponding plots of 30, 70 and 300 fish, and for different radial proximity distances). A strong association is evident between individual speed and local order. Assuming an unknown causal direction in the relation between the speed and the local order, as discussed above, there are two ways we can average the contour plots; either over all values of the speed for each value of the local order, or vice versa, for each value of the speed we average over the local order. The results of both procedures, for all group sizes, are overlaid on the contour plot in Fig. 7B . While the two approaches yield disparate curves, they both demonstrate a similar relationship between individual speed and local order. Although the fish have to align at higher speeds to maintain group cohesion, it is unclear why they should become disordered at low speeds. There is also a much greater variance in local order at low speeds, demonstrating a wide degree of flexibility when individual speeds are low. Interestingly, across the group sizes the two sets of curves are close to identical. This suggests that, from the perspective of a focal individual, it may simply adopt the same local rules regardless of group size (consistent with the findings of Katz et al. [17] for groups of 10 and 30 fish). Contrary to the polarization order parameter, the rotation order parameter has little meaning on a purely local scale. Rather we compute the rotation order parameter separately for a series of shells placed around the center of mass of the group, as illustrated in Fig. S11 . This allows us to obtain a well-defined measure that provides insight into the structural organization of the milling state. As can be seen in Figs. 7C and 7D this state is characterized by a center with low speed and low degree of structure that contributes little to the milling state, while as we move towards the edge of the group the speed and impact from each shell on the milling state increases. In Fig. 7D the curves are almost identical for all group sizes suggesting that, again, scaling the size of the group has little effect on the local structural signature of this collective state. These results demonstrate that the ordered polarized and milling states are locally near identical from the perspective of a focal individual, regardless of group size. These data also support the prediction of a multi-stable locally-ordered regime in which the group can transition back and forth between the polarized and milling state through stochastic and boundary-induced effects. Conclusions Despite the multitude of local interactions that result in coordinated group motion we demonstrate that schooling golden shiners predominantly exist in three ‘fundamental’ dynamically-stable states of the underlying dynamics: swarm, milling and polarized motion. We establish that group states, and transitional behavior, can be represented in low-dimensional space, a projection that allows us to see the path taken by groups between the three dynamically-stable states as well as to relate the collective states exhibited to properties such as group size, individual speed and perturbations to the group. We note that it is possible that further collective states may be found within the classified dynamically stable regimes described here, but the present states are highly consistent with the theoretical predictions of three regimes A key question in the study of collective behavior is whether different group-level patterns result from all individuals responding to the same external factor [27] , or individuals changing behavior [1] , or whether multiple dynamically-stable collective states co-exist for the same individual behaviors [3] . Our results provide evidence for the importance of the two latter processes in the behavior of schooling fish: transitions from the swarm, to the milling or parallel group states (and vice versa) involve a social feedback whereby individuals adjust behavior—in this case their speed—in response to prevailing local conditions. Low average speeds among group members correspond to them occupying the dense, disordered swarm regime. Higher speeds correspond to higher local order (alignment among group members) and groups existing in either the milling or polarized state. Transitions between these states occur with negligible, or no, change in local density, order or speed; instead perturbations such as collisions with the boundary, or (seemingly stochastic) fluctuations in motion at the group edge (in the case of milling to polarized state transitions) or front (in the case of polarized to milling transitions) result in the group leaving one dynamically-stable state, and either then returning to that state, or transitioning to the alternative locally-ordered regime. Thus the milling and polarized states appear to be bistable; the state exhibited by the group effectively being dependent on starting conditions and/or the nature of perturbations, as well as the group size. Theoretically [3] and experimentally (analysis of shoals of 30 fish in small tank), milling states are seen to be less stable for small groups, when controlling for boundary condition effects. It is likely that the relationship we found between speed and local order is a generic feature of mobile groups with local interactions. Furthermore, qualitatively similar features have been observed in small groups (4 and 8 fish) of the giant danios [24] . A key challenge for animal behavior in this, and future, decades is to understand how the microscopic mechanisms of interactions among molecules, physiological systems and neural circuits result in behavior at higher levels of organization. Whereas we focused on collective behavior resulting from interactions among individual organisms, the general approach adopted shares commonalities with approaches that have successfully been used to characterize the dynamical properties of gene interaction networks [28] , neuronal circuits [29] , how locomotion is coordinated among limbs, each of which has many degrees of freedom [30] , and how the behavior of individual organisms (such as Caenorhabditis elegans ), despite apparent complexity, can be deconstructed into a discrete number of low-dimensional behavioral dynamically-stable states (see Stephens et al. [31] ). We suggest that development, and adoption of, such techniques in the behavioral sciences could facilitate the advent of increasingly integrative and quantitative insights. Our work demonstrates that such an approach to data collection and analysis can reveal underlying simplicity in the dynamical properties of collective behavior in groups. Collective behavioral states appear to result from both behavioral feedback processes whereby individuals both adopt, and influence, the behavior of near neighbors and also as multi-stable regimes in which individual behavior does not change, but rather perturbations induce relatively abrupt transitions between alternate and co-existing dynamically-stable behavioral states. Prey groups have been observed to switch states upon detecting a predator [32] and risk can be dependent on these states [9] , [33] . Whether the mechanisms for switching between states as identified here are somehow themselves adaptive would be an interesting question to address in future work."
} | 12,139 |
40128811 | PMC11934532 | pmc | 848 | {
"abstract": "Background Biofuels produced from algae have enormous advantages in replacing fossil fuels, and Microcystis aeruginosa has a great potential for biofuel production, due to growing fast to form large amounts of biomass. Light is essential for algal growth, and the optimum light quality can promote the biomass and lipid accumulation for increasing feedstock for biofuel production. Results We investigated the biomass accumulation, photosynthetic ability, carbohydrate, and lipid yield as well as related gene expression in M. aeruginosa under red, blue, purple, and white light to promote biofuel production using this alga under the optimal light quality. Compared with white light, purple light promoted the cell growth during the 5 days, while blue light showed inhibitory effect. Red light had no effect on the cell growth, but improved the biomass content to the highest level. Red light improved the photosynthetic ability by raising chlorophyll level, and up-regulating expression of the genes in chlorophyll biosynthesis, photosynthetic electron transfer, and CO 2 fixation. Among these light qualities, red light showed the maximum effect on soluble, insoluble, and total carbohydrate accumulation by up-regulating the genes in polysaccharide and starch formation, and down-regulating the genes in glycolysis and tricarboxylic acid cycle. Red light also exhibited the maximum effect on lipid accumulation, which might be caused by up-regulating five genes in fatty acid biosynthesis. Conclusion Red light can promote M. aeruginosa accumulating carbohydrates and lipids by regulating related gene expression, which should be the optimal light quality for improving feedstock yield for biofuel production. Supplementary Information The online version contains supplementary material available at 10.1186/s13068-025-02615-8.",
"conclusion": "Conclusion In contrast to white light, the photosynthetic ability in M. aeruginosa was improved under red light by increasing chlorophyll content and promoting related gene expression. Among red, blue, and purple light, red light showed the maximum effect on the biomass accumulation as well as carbohydrate and lipid yield, which may result from the high photosynthetic ability, up-regulation of primary metabolite synthesis-related genes, and down-regulation of sugar degradation-related genes. Therefore, red light should be the optimal light quality for M. aeruginosa accumulating carbohydrates and lipids, which is beneficial for supplying feedstock for biofuel production.",
"discussion": "Results and discussion Biomass yield under different light qualities When M. aeruginosa cells were cultured under white, red, purple, and blue light, their density gradually increased with prolonging the culturing time. Compared with white light, purple light significantly ( P < 0.05) promoted the cell growth, but blue light lowered the cell growth. During the 5 days, red light weakly promoted the cell growth in contrast to white light, with no significant difference between them (Fig. 1 A). These findings were similar with the previous studies. Compared with white light, blue light inhibited the growth of Nannochloropsis sp. strain MUR266 and C. reinhardtii strain CC-125 [ 22 , 23 ], while red light had no impact on the growth of Nannochloropsis sp. strain MUR266 [ 22 ]. In contrast to blue light, red light markedly promoted the growth of A. fertilissima strain PUPCCC410.5 and Nannochloropsis sp. strain MUR266 [ 22 , 24 ]. Fig. 1 Effects of different light qualities on the cell growth ( A ) and biomass content ( B ) on M. aeruginosa. White: White light, the control; purple: purple light; blue: blue light; red: red light. On the same day, the different lowercase letters indicate the significant difference at P < 0.05. Means ± SE ( n = 4) In the treatments with the four light qualities, the biomass content in M. aeruginosa cultures gradually increased with prolonging the culturing time. Under red and purple light, the biomass content was significantly ( P < 0.05) higher than that under white light, and red light showed the maximum promoting effect on the biomass accumulation since the 2nd day. At the 5th day, the biomass content under red light increased by 41.1% ( P < 0.05), 15.0% ( P < 0.05), and 99.2% ( P < 0.05) compared to that under white, purple, and blue light, respectively. In contrast to white light, blue light significantly ( P < 0.05) lowered the biomass content (Fig. 1 B). Similarly, red light also promoted biomass accumulation in S. platensis cells, and blue light showed inhibitory effect [ 26 ]. In contrast to blue light, red light improved the biomass content in A. fertilissima , Dunaliella salina , D. tertiolecta , Nodularia sphaerocarpa, and H. pluvialis [ 24 , 33 – 35 ], and red–orange light improved the biomass content in C. reinhardtii [ 23 ]. Among the treatments with the four light qualities, M. aeruginosa cell length and width not significantly changed (Supplementary Fig. 1), indicating that the variations of the algal biomass under the three monochromatic light qualities are caused by the cell contents but not the cell size. Photosynthetic abilities under different light qualities Chlorophylls are essential photosynthetic pigments, which serve crucial functions in absorbing solar energy and converting it into electric energy. Under red light, an increase was found in the chlorophyll content in N. sphaeroides and S. platensis cells [ 36 , 37 ], while a decrease was found in S. platensis , N. sphaeroides , and N. sphaerocarpa under blue light [ 33 , 36 , 37 ]. Under purple light, the chlorophyll a and b content in strawberry was not changed in contrast to white light [ 38 ]. Compared with white light, the increase was found in chlorophyll content in M. aeruginosa cultured under red light since the 1st day, whereas a decrease was detected under blue light. For purple light, no obvious variation was detected during the 5-day treatment (Fig. 2 A). These results demonstrate that red light has positive effect on chlorophyll formation, which is beneficial for light energy absorption and photosynthetic electron generation. Fig. 2 Effects of different light qualities on the chlorophyll content ( A ), ETo/RC ( B ), and O 2 evolution rate ( C ) in M. aeruginosa. White: white light, the control; purple: purple light; blue: blue light; red: red light. On the same day, the different lowercase letters indicate the significant difference at P < 0.05. Means ± SE ( n = 4) Among white, red, blue, yellow, and green light, Porphyra leucosticta cells exhibited the highest photosynthetic electron transport rate under red light, while the lowest under blue light [ 39 ]. When N. flagelliforme was cultured under white, red, blue, purple, yellow, and green light, the highest photosynthetic rate was found under red light, while the lowest under green and blue light [ 27 ]. In addition, red light promoted photosynthetic electron production and transfer in strawberry, but blue light exhibited adverse effect [ 38 ]. For 3 cyanobacteria, UV-B raised their photosynthetic abilities by promoting photosynthetic electron production and transfer [ 26 ]. The ETo/RC in M. aeruginosa raised under red and purple light compared with white light, with the increase of 33.6% ( P < 0.05) and 31.2% ( P < 0.05) after 5 days, respectively (Fig. 2 B). This indicates that the two light qualities are beneficial for photosynthetic electron production and transfer, which can supply abundant assimilatory power (ATP and NADPH) for CO 2 fixation. Similarly, red and purple light also improved O 2 evolution rate (Fig. 2 C), demonstrating that the two light qualities promoted the increase in photosynthetic rate [ 40 ]. For photosynthetic products, they generated from the high photosynthesis under purple light could be distributed to cell growth, leading to the highest cell density, while they might be converted into carbohydrates and lipids under red light, resulting in the biomass accumulating to the highest level (Fig. 1 ). Under blue light, the decrease was detected in ETo/RC and O 2 evolution rate, which may result in the low photosynthetic ability, cell growth, and biomass accumulation (Figs. 1 , 2 ). Carbohydrate accumulation under different light qualities Carbohydrates are one of the main types of primary metabolites, which mainly include sugar monomers and polymers as well as sugar derivatives (amino sugars and uronic acids) [ 15 ]. Sugar polymers exhibit widely varied molecular weights for the different polymerization degree. They function as storage (starch) or structural (cellulose) compounds with accounting for large proportion in biomass sources, and are used to produce bioethanol through hydrolysis [ 41 ]. In contrast to blue light, red light promoted the accumulation of carbohydrates in Chlorella sp. [ 29 ]. For C. reinhardtii , an increase was found in carbohydrate content under red–orange light [ 23 ]. Among 6 light qualities, red and purple light improved the extracellular polysaccharide yield from N. flagelliforme in contrast to white light, with the maximum promoting effect under red light, while blue light did not impact the polysaccharide yield [ 27 ]. Under blue light, a decrease was found in carbohydrate content in Nannochloropsis sp . , due to the increase in respiration consumption [ 22 ]. Compared with white, purple, and blue light, red light remarkably promoted extracellular and intracellular polysaccharide formation for raising carbon flow into the synthetic process [ 42 ]. In M. aeruginosa , the soluble carbohydrates primarily contain monosaccharides and some oligosaccharides, while the insoluble carbohydrates primarily contain stored and structural polysaccharides [ 15 ]. In this study, M. aeruginosa cells significantly ( P < 0.05) increased the soluble carbohydrate content since the 1st day under red and purple light, but declined the content under blue light. Among the treatments with the four light qualities, the highest soluble carbohydrate content was always detected under red light during the 5 days, and it was increased by 60.4% ( P < 0.05), 20.3% ( P < 0.05), and 93.0% ( P < 0.05), respectively, compared to that under white, purple, and blue light after 5 days (Fig. 3 A). Compared with white light, red and purple light significantly ( P < 0.05) improved the soluble carbohydrate concentration during the 5 days, with the maximum effect under red light, whereas blue light significantly ( P < 0.05) lowered the soluble carbohydrate concentration, with the decrease of 36.9% after 5 days (Fig. 3 B). Among the four light qualities, the highest soluble carbohydrate productivity was found under red light, which was increased by 2.4 folds ( P < 0.05), 43.3% ( P < 0.05) and 7.1 folds ( P < 0.05) compared to that under white, purple, and blue light, respectively (Fig. 3 C). Fig. 3 Effects of different light qualities on the accumulation of soluble ( A – C ), insoluble ( D – F ), and total ( G – I ) carbohydrate in M. aeruginosa. White: white light, the control; purple: purple light; blue: blue light; red: red light. On the same day, the different lowercase letters indicate the significant difference at P < 0.05. Means ± SE ( n = 4) Insoluble carbohydrates (Fig. 3 D–F), total carbohydrates (Fig. 3 G–I), and polysaccharides (mainly intracellular polysaccharides) (Supplementary Fig. 2) also showed the similar variations under white, red, purple, and blue light. Red light exhibited the maximum effect on the accumulation of these carbohydrates, which may be caused by the conversion of massive photosynthetic products from high photosynthetic ability. Blue light declined the carbohydrate accumulation, indicating that this light was not beneficial for carbohydrate production. Lipid accumulation under different light qualities Lipids are another major type of primary metabolites, and used as feedstock to produce biodiesel. It has been reported that the biodiesel produced from M. aeruginosa lipid transesterification conforms to the standard of American ASTMD6751 [ 31 ]. For Nannochloropsis sp., light quality played an important role in the lipid accumulation, and the highest lipid content was found under blue light among five light qualities [ 22 ]. Similarly, blue light also increased the eicosapentaenoic acid content in N. oceanica [ 28 ] and fatty acid (C16:2 and C18:2) content in Chlorella sp. cells [ 29 ]. For H. pluvialis , white–red light was beneficial for the formation of saturated fatty acids [ 43 ]. For Cyanobium sp., supplementation of red light can induce the accumulation of saturated and unsaturated fatty acids as well as lipids [ 44 ]. Compared with blue light, red–orange light was more effective for lipid accumulation in C. reinhardtii [ 23 ]. In addition, different ratios of red and blue light also affected fatty acid and lipid biosynthesis in Phaeodactylum tricornutum , with the highest saturated fatty acid content, lipid content, and lipid production under red–blue (5:2) light [ 45 ]. These results suggest that the optimal light quality for lipid formation varied with algal species. Lipid accumulation in M. aeruginosa was also markedly influenced by light quality. With prolonging the culturing time, the lipid content and concentration in the cells gradually increased under purple, blue, and red light. Among these light qualities, blue light always showed the maximum promoting effect on the increase in lipid content, but had minor effect on lipid concentration and productivity (Fig. 4 ). This demonstrated that blue light was beneficial for lipid formation, but the low cell density and biomass content limited the lipid yield. Fig. 4 Effects of different light qualities on the lipid content ( A ), lipid concentration ( B ) and lipid productivity ( C ) in M. aeruginosa. White: white light, the control; Purple: purple light; blue: blue light; red: red light. On the same day, the different lowercase letters indicate the significant difference at P < 0.05. Means ± SE ( n = 4) For red light, it improved the lipid content to the highest level at the 5th day, with no significant difference from that under blue light (Fig. 4 A). However, the increase in lipid content under the two light qualities showed different patterns, with a gradual increase under red light and a fast increase to the plateau under blue light. Blue light may have stimulated the immediately formed and stored photosynthetic products to form lipids quickly, with a fast increase at the 1st day. However, the low photosynthetic ability under blue light might limit the supply of photosynthetic products, resulting in a slow increase trend after 3 days. The lipid conversation from photosynthetic products was slowly initiated under red light, but the high photosynthetic ability might supply abundant photosynthetic products, leading to a gradually increasing trend in lipid content. Under red light, M. aeruginosa raised the lipid concentration to the highest level since the 3rd day (Fig. 4 B). For lipid productivity, the highest value was detected under red light, with the increase of 2.6-fold ( P < 0.05), 74.0% ( P < 0.05), and 1.7-fold ( P < 0.05), respectively, compared to that under white, purple, and blue light (Fig. 4 C). These results suggest that red light is suitable for culturing M. aeruginosa to harvest lipids for biodiesel production. Related gene expression in photosynthesis under red light For algae, light quality can act as a signal to adjust gene expression in energy and matter metabolism, resulting in the variation of primary and secondary metabolites [ 40 ]. In chlorophyll biosynthesis, glutamate is the precursor, and 15 enzymes take part in the catalysis [ 46 ]. In N. oceanica , red light improved the chlorophyll content by up-regulating 11 genes in the biosynthetic process [ 47 ]. When M. aeruginosa was kept under red light for 5 days, the up-regulation was found in ten genes associated with chlorophyll biosynthesis in contrast to white light, including gltX , hemB , hemE , hemF , hemJ , bchI , bchM , acsF , por , and chlG (Fig. 5 A, the gene functions and expression levels in Supplementary Table 1). For chlG , it encodes chlorophyll synthase, which catalyzes the formation of chlorophyll a and b. Under red light, its expression level in M. aeruginosa was increased by 2.6-fold ( P < 0.05) (Fig. 6 ). Then, the up-regulation of these genes under red light should promote the formation of chlorophylls and improve their levels in M. aeruginosa (Fig. 2 A). Fig. 5 Expression of the genes in chlorophyll biosynthesis ( A ) and photosynthesis ( B ) in M. aeruginosa under red light . White: White light, the control; red: red light. fpkm: Fragments per kilobase per million mapped reads. Means ( n = 3) are shown Fig. 6 Expression of three genes ( chlG , psb29, and PRK ) related with photosynthesis in M. aeruginosa . White: White light, the control; Red: Red light. Different lowercase letters indicate the significant difference at P < 0.05. Means ± SE ( n = 4) In Synechococcus sp., the up-regulation was found in three genes related with photosynthetic electron transfer under red light, which led to the improvement of the photosynthetic ability [ 48 ]. Under red light, Eustigmatos cf. polyphem improved the expression of 11 genes in PSII assembly, 6 genes in cytochrome (Cyt) b 6 /f, 2 genes in PSI assembly, and 2 gene in ATP synthase, while blue light lowered their expression [ 49 ]. Similarly, red light also raised expression of the genes associated with light-harvesting complex, PSII, Cyt b 6 /f, PSI, photosynthetic electron transfer, and ATP synthase in grape seedlings [ 50 ]. In this study, red light up-regulated the expression of 4 genes ( psba , psbn , psb27 , and psb29 ) in PSII assembly, 1 gene ( psaM ) in PSI assembly, and 3 genes ( xfp , rpiA , and PRK ) in CO 2 fixation compared with white light (Fig. 5 B, Supplementary Table 1). psb29 and PRK encoded PSII biogenesis protein Psb29 and phosphoribulokinase, respectively, whose expression levels were increased by 11.6 ( P < 0.05) and 8.1 folds ( P < 0.05) under red light, respectively (Fig. 6 ). The up-regulation of these genes was beneficial for M. aeruginosa adsorbing light, transporting photosynthetic electron and fixing CO 2 , resulting in the high photosynthetic ability under red light (Fig. 2 B). In the previous study, purple light also raised the expression of 12 photosynthesis-related genes in M. aeruginosa [ 48 ], which should contribute to the improvement of the photosynthetic ability under this light (Fig. 2 B). Related gene expression in primary metabolism under red light In algal cells, glycolysis and tricarboxylic acid cycle (TCA) are two main pathways for sugar degradation that can be adjusted by light quality. Under red light, the expression of several genes in glycolysis and TCA cycle was down-regulated according to the transcriptome analysis [ 47 , 50 ], and the activities of several key enzymes, such as hexokinase, pyruvate kinase, and succinic dehydrogenase, in the two pathways were declined [ 51 , 52 ]. These were not beneficial for sugar degradation under red light, leading to the reduction of intermediate product levels, such as malic acid, citrate, succinic acid, 2-hydroxyglutaric acid, and fumarate [ 27 , 53 ]. In this study, the down-regulation was found in the expression of eight genes ( OA58_RS22055 , OA58_RS13225 , OA58_RS12055 , fbaA , OA58_RS00740 , pgk , eno , and OA58_RS17895 ) in glycolysis and 5 genes ( acoB , OA58_RS02030 , acnB , OA58_RS16450 , and mdh ) in TCA cycle under red light (Fig. 7 A, Supplementary Table 2). This was consistent with the previous findings in N. oceanica and grape seedlings under red light [ 47 , 50 ]. fbaA codes for fructose-diphosphate aldolase that cleaves fructose-1,6-bisphosphate to generate glyceraldehyde-3-phosphate and dihydroxyacetone phosphate. Under red light, its relative expression level was declined by 41.9% ( P < 0.05) compared to that under white light (Fig. 8 ). The down-regulation of these genes may reduce the consumption of photosynthetic products in M. aeruginosa under red light, resulting in the biomass accumulation (Fig. 1 B). Fig. 7 Expression of the genes in sugar degradation ( A ), fucose, and starch biosynthesis ( B ) as well as fatty acid biosynthesis ( C ) in M. aeruginosa under red light . White: White light, the control; Red: Red light. fpkm: Fragments per kilobase per million mapped reads. Means ( n = 3) are shown Fig. 8 Expression of four genes ( fbaA , gmd , glgB, and accD ) related with primary metabolism in M. aeruginosa . White: white light, the control; red: red light. Different lowercase letters indicate the significant difference at P < 0.05. Means ± SE ( n = 4) OA58_RS06690 , rfbM , and gmd code for mannose-6-phosphate isomerase, mannose-1-phosphate guanylate transferase, and GDP- d -mannose dehydratase in fucose biosynthesis, respectively. Compared with white light, their expression was significantly raised under red light (Fig. 7 B, Supplementary Table 2), and the expression level of gmd was increased by 7.1 folds ( P < 0.05) (Fig. 8 ), which should lead to the synthesis and accumulation of polysaccharides. In exposure to red light, an increase was detected in the content of extracellular polysaccharides in N. flagelliforme by raising expression of mannose-6-phosphate isomerase and improving its activity [ 29 , 42 ]. In addition, red light also improved monosaccharide composition by raising UDP-glucose pyrophosphorylase activity for providing various sugar nucleotides as the substrates [ 42 ]. In starch biosynthesis, phosphoglucomutase is encoded by OA58_RS05120 , and catalyzes glucose-6-phosphate to form glucose-1-phosphate. When phosphoglucomutase activity was inhibited, a decrease was detected in the starch content in Gracilariopsis lemaneiformis [ 54 ]. For phosphoglucomutase-deficient mutant of Arabidopsis thaliana , no starch was formed in the guard cells [ 55 ], whereas phosphoglucomutase overexpression promoted the formation of glucose-1-phosphate, β-1,3-glucan, chrysolaminarin, and starch in Phaeodactylum tricornutum [ 56 ]. glgB encodes 1,4-α-glucan branching enzyme, which plays a key role in catalyzing amylose to form starch in plants and glycogen in cyanobacteria. Streptomyces aureofaciens and S. elongatus were two species in cyanobacteria, and the former glgB- disrupted strains and the latter glgB mutant lowered glycogen content by blocking its formation [ 57 , 58 ]. Under red light, the up-regulation was found in OA58_RS05120 and glgB expression (Fig. 7 B, Supplementary Table 2), with the increase of 3.6-fold ( P < 0.05) in glgB expression level (Fig. 8 ). This was consistent with the up-regulation of the two genes and other genes in starch biosynthesis in grape seedlings under red light [ 50 ]. The up-regulation of the two genes may facilitate glycogen biosynthesis in M. aeruginosa under red light. Compared with white light, the expressions of five genes ( accD , fabD , fabF , fabG , and fabZ ) in fatty acid biosynthesis in M. aeruginosa were remarkably up-regulated under red light (Fig. 7 C). accD codes for β-carboxyl transferase subunit of acetyl-CoA carboxylase, which is a rate-limiting enzyme in fatty acid formation [ 49 ]. Its expression level was increased by 9.0 folds ( P < 0.05) under red light (Fig. 8 ). Similarly, the up-regulation of these genes was also detected in E. cf. polyphem under red light. Meanwhile, several genes related with lipid formation were also up-regulated [ 49 ]. For C. reinhardtii , red light improved three gene expression in fatty acid biosynthesis [ 59 ]. The up-regulation of these genes may promote the generation and accumulation of fatty acids and lipids in algal cells (Fig. 4 ) [ 43 , 49 , 59 ]. In contrast to white light, red light may decline the sugar degradation in M. aeruginosa by lowering the gene expression in glycolysis and TCA cycle. Then, the abundant photosynthetic products that were generated from high photosynthetic abilities might be converted into sugars and glycogen by raising the related gene expression (Figs. 7 A, B, 8 ), leading to the accumulation of soluble and insoluble carbohydrates (Fig. 3 ). Meanwhile, the abundant photosynthetic products can also be used to form fatty acids by raising the related gene expression (Figs. 7 C, 8 ), resulting in the accumulation of lipids (Fig. 4 ). This is beneficial for the production of bioethanol and biodiesel by supplying feedstock. Similar with red light, purple light also declined the expression of the genes in glycolysis and TCA cycle ( OA58_RS13225 , OA58_RS00740 , eno , and acoB ), and increased expression of the genes in biosynthesis of polysaccharide ( rfbm and gmd ), glycogen ( glgB ), and fatty acids ( accD ) (Supplementary Fig. 3). However, the less altered gene numbers might lead to the lower accumulation of carbohydrates and lipids under purple light in contrast to red light (Figs. 3 , 4 )."
} | 6,320 |
21347422 | PMC3036657 | pmc | 850 | {
"abstract": "Microbial consortia constitute a majority of the earth's biomass, but little is known about how these cooperating communities persist despite competition among community members. Theory suggests that non-random spatial structures contribute to the persistence of mixed communities; when particular structures form, they may provide associated community members with a growth advantage over unassociated members. If true, this has implications for the rise and persistence of multi-cellular organisms. However, this theory is difficult to study because we rarely observe initial instances of non-random physical structure in natural populations. Using two engineered strains of Escherichia coli that constitute a synthetic symbiotic microbial consortium, we fortuitously observed such spatial self-organization. This consortium forms a biofilm and, after several days, adopts a defined layered structure that is associated with two unexpected, measurable growth advantages. First, the consortium cannot successfully colonize a new, downstream environment until it self-organizes in the initial environment; in other words, the structure enhances the ability of the consortium to survive environmental disruptions. Second, when the layered structure forms in downstream environments the consortium accumulates significantly more biomass than it did in the initial environment; in other words, the structure enhances the global productivity of the consortium. We also observed that the layered structure only assembles in downstream environments that are colonized by aggregates from a previous, structured community. These results demonstrate roles for self-organization and aggregation in persistence of multi-cellular communities, and also illustrate a role for the techniques of synthetic biology in elucidating fundamental biological principles.",
"introduction": "Introduction The vast majority of living biomass consists of single-celled organisms, but the existence of higher organisms demonstrates that interacting networks of cell populations can thrive despite competition between them [1] , [2] . How nascent communities gain a growth advantage over unassociated individuals is an open question [2] – [9] , but cell–cell interactions [10] – [14] and the formation of specific multi-cellular structures [15] – [18] are thought to contribute. Evaluating the role of physical structure in the initiation and persistence of natural consortia poses a causality dilemma [2] , and de novo design of synthetic consortia that self-organize into specific structures is difficult. Thus, experimental studies of the formation and benefits of specific physical structures in mixed microbial communities are few. We describe a synthetic symbiotic microbial consortium that allows us to address some of these questions. An advantage of using synthetic, or engineered, consortia for studies of this nature is that complex communal behaviors such as symbiosis can be implemented under defined and tunable experimental control [9] , [19] – [21] . Although very simple relative to naturally-occurring microbial consortia, engineered ecosystems can nonetheless exhibit behaviors that mimic those found in nature and, because the interactions of engineered consortia can be controlled and more fully characterized, can provide insight into the development and persistence of natural communities. It is useful to study the relationship between microbial community structure and persistence in biofilm communities for three primary reasons. First, biofilm spatial structure and productivity (total biomass accumulation) can be observed and quantified as a function of time using confocal laser scanning microscopy (CLSM) [22] . Second, stable micro-communities with very different properties and behaviors can form and persist within biofilms [23] – [26] . Finally, cells and sub-communities that detach from a biofilm subjected to fluid flow will flow downstream and may colonize downstream environments, where composition, spatial structure, and productivity can be observed. Thus, the effects of composition and spatial structure on the productivity of a consortium can be easily quantified when it grows as a biofilm. A structure that is beneficial should increase productivity in the local environment; it might also improve colonization or productivity when the consortium moves to downstream environments. The synthetic microbial consortium was engineered to rely on biofilm formation so that these effects could be measured.",
"discussion": "Discussion Biofilms can propagate when single cells dissociate from the outer regions of mature biofilms and adhere downstream, and this is considered the primary mode by which biofilms spread [31] , [32] . Our results suggest this as one mechanism of proliferation. The regions of the symbiotic biofilm that are exposed to flow are predominantly populated by yellow cells, and more than 99% of cells in the single-cell fraction are yellow (Supporting Information S4 ). This fraction can form biofilms downstream, indicating that the yellow population adapts to adhere better, but these biofilms are weak and monomorphic ( Fig. 2F ). However, when effluent is never treated or sorted, the whole consortium colonizes downstream environments through multiple transfers and these biofilms accumulate more biomass than the initial biofilm. Therefore, aggregates that dissociate from upstream biofilms and colonize downstream environments enhance overall growth and proliferation of this consortium. These results indicate that a critical mechanism by which microbial communities propagate is the movement of aggregated members into downstream environments. What are these aggregates, and how do they work? Aggregates preserve the physical relationship between the blue and yellow populations, and enhance yellow adhesion in downstream biofilms. Aggregates are distinct but conjoined clusters of blue and yellow biomass (Supporting Information S4 , information regarding aggregate composition can be found in Supporting Information S5 ). When aggregates are transferred, average downstream biofilms contain five times more yellow biomass, after initial adhesion, than when aggregates are disrupted, even though both inocula contain the same numbers of blue and yellow cells (Supporting Information S3 , S4 , S5 ). Aggregates appear to be pre-organized pieces of the layered structure that quickly grow to recapitulate it in new environments. Additionally, it is possible that the proximity of blue and yellow cells in the aggregates enhances collaboration and therefore productivity (this is difficult to assess and remains untested). We have used an engineered symbiotic microbial consortium to explore spatial self-organization and its benefits to a microbial community. Here, two symbiotic populations of E. coli grow to form a defined, layered structure which provides a growth advantage to both. This engineered consortium allowed us to observe the critical roles of self-organization, layering, and aggregation in the growth, movement, and ability of a microbial consortium to colonize new environments. The persistence assured by aggregates allows evolution and adaptation of interacting microbial communities despite environmental disruptions. It may eventually be possible to use engineered consortia like this one to determine how relationships between interacting, co-evolving populations are enhanced and preserved by particular physical structures."
} | 1,869 |
20557974 | null | s2 | 851 | {
"abstract": "Predominant frameworks for understanding plant ecology have an aboveground bias that neglects soil micro-organisms. This is inconsistent with recent work illustrating the importance of soil microbes in terrestrial ecology. Microbial effects have been incorporated into plant community dynamics using ideas of niche modification and plant-soil community feedbacks. Here, we expand and integrate qualitative conceptual models of plant niche and feedback to explore implications of microbial interactions for understanding plant community ecology. At the same time we review the empirical evidence for these processes. We also consider common mycorrhizal networks, and propose that these are best interpreted within the feedback framework. Finally, we apply our integrated model of niche and feedback to understanding plant coexistence, monodominance and invasion ecology."
} | 217 |
34354559 | PMC8329666 | pmc | 852 | {
"abstract": "This paper presents a novel spiking neural network (SNN) classifier architecture for enabling always-on artificial intelligent (AI) functions, such as keyword spotting (KWS) and visual wake-up, in ultra-low-power internet-of-things (IoT) devices. Such always-on hardware tends to dominate the power efficiency of an IoT device and therefore it is paramount to minimize its power dissipation. A key observation is that the input signal to always-on hardware is typically sparse in time. This is a great opportunity that a SNN classifier can leverage because the switching activity and the power consumption of SNN hardware can scale with spike rate. To leverage this scalability, the proposed SNN classifier architecture employs event-driven architecture, especially fine-grained clock generation and gating and fine-grained power gating, to obtain very low static power dissipation. The prototype is fabricated in 65 nm CMOS and occupies an area of 1.99 mm 2 . At 0.52 V supply voltage, it consumes 75 nW at no input activity and less than 300 nW at 100% input activity. It still maintains competitive inference accuracy for KWS and other always-on classification workloads. The prototype achieved a power consumption reduction of over three orders of magnitude compared to the state-of-the-art for SNN hardware and of about 2.3X compared to the state-of-the-art KWS hardware.",
"introduction": "Introduction An spiking neural network (SNN) classifier is an attractive option for ultra-low-power intelligent internet-of-things (IoT) devices. It is promising especially for always-on functions due to their spike-based operation for computation and communication, allowing their switching activity and power to scale smoothly with the input activity rate. An SNN, therefore, is suitable for applications like keyword spotting (KWS) or face recognition in surveillance, thanks to its event-driven operation. Spiking neural network based hardware work so far, however, focused on either the acceleration of neural simulations or the improvement of both performance and energy efficiency. In other words, they are not designed for always-on function. For example, Neurogrid ( Benjamin et al., 2014 ) targets large-scale neural simulations. It employs analog neurons and address event representation (AER) for communication, the latter using a multi-bit bus. SpiNNaker ( Painkras et al., 2013 ) also targets neural simulation and employs an array of embedded digital processors communicating asynchronously. Yang et al. presented multiple works that targeted large scale neural simulations. In CerebelluMorphic ( Yang et al., 2021b ) they simulated portions of the cerebellum related to motor learning using 6 field programmable gate array (FPGA) chips that communicate using a multicast router. In BiCoSS ( Yang et al., 2021c ) they presented a platform with 35 FPGA chips connected to realize real-time computation of biological activities in multiple brain areas. In another work ( Yang et al., 2021d ), they presented an event-based processing algorithm that used piecewise linear approximation and binarization for efficient implementation of credit assignment to neurons in neuromorphic hardware. On the other hand, TrueNorth ( Akopyan et al., 2015 ) was designed to be a scalable low power neurosynaptic inference engine for SNNs. The architecture was event-driven and employed synchronous circuits for computation blocks and asynchronous circuits for communication. Also, Tianjic chip was designed to support inference only with both neuromorphic and deep-learning models ( Pei et al., 2019 ). Some works proposed architectures for both the training and inference of SNNs. Koo et al. (2020) introduced the implementation of a stochastic bit and used it in the realization of a neuron and synapse. They support on-chip training and inference with the synapse being stochastic in training and neuron being stochastic in both training and inference. Chen et al. (2018) presented an SNN accelerator with on-chip spike-timing-dependent plasticity (STDP) based learning. This chip has 64 cores that communicate using a network-on-chip (NoC) with each core supporting 64 leaky integrate and fire (LIF) neurons. Also, Loihi ( Davies et al., 2018 ) was designed to support a variation of the current based dynamics LIF neuron model and a wide range of synaptic learning rules for both supervised and unsupervised learning. It is built for performance. It has 128 cores, three x86 cores, off-chip interfaces and an asynchronous NoC for communication between cores. Also, Seo et al. (2011) , implemented a scalable architecture with a set of 256 neurons and transposable memory for synapses in near-threshold voltage (NTV) circuits. It mapped an auto-associative memory model. Some other works implemented different learning rules for on-chip training. Knag et al. (2015) , implemented a feature extractor based on a sparse coding algorithm using LIF neurons. Park et al. (2019) , developed a new neuromorphic training algorithm and hardware which supports low overhead on-chip learning. Some of these chips e.g., ( Akopyan et al., 2015 ; Davies et al., 2018 ) employ asynchronous logic such as quasi-delay-insensitive (QDI) dual-rail dynamic logic or bundled data communication. Asynchronous logic circuits are, however, generally bulkier and power-hungrier than the single-rail static counterpart and also not very voltage-scalable ( Chen et al., 2013 ; Liu et al., 2013 ) and bundled data communication incurs significant overhead because of the handshake. Some other chips employ power-efficient static logic ( Chen et al., 2018 ; Davies et al., 2018 ; Park et al., 2019 ; Pei et al., 2019 ), but they target high throughput, not always-on function. As a result, they exhibit a power consumption of more than tens of mW, which makes it difficult to use them for always-on functions. In this work, we focus on ultra-low-power always-on inference hardware and propose an SNN classifier consuming less than 300 nW. Our architecture uses fully spike-based event-driven operation and only static logic operating at a NTV to achieve such low power. Specifically, our design is centered around the neurosynaptic core. It is implemented using static gates and spike-driven (i) spatiotemporally fine-grained clock generation, (ii) clock-gating, and (iii) power-gating. Also, the communication between neurosynaptic cores is free from information loss due to the collision of spikes, despite using only wires to connect the cores. The architecture exhibits active power consumption that is proportional to the input rate due to its event-driven nature. We also employ the technique in Cao et al. (2014) to train a neural network with binary weights and use the weights for the SNN we intend to deploy. The use of binary weights is a recent development in deep learning for making inference efficient ( Courbariaux et al., 2015 ). They are of special interest because of their reduced memory footprint and simple computations. They are well suited for low power hardware and attain close to state of the art accuracy on datasets like MNIST. On the other hand, we keep the activations as spike-rate-coded multi-bit values, which improves the model’s inference accuracy. We prototyped an SNN classifier in 65-nm LP CMOS technology. It has 5-layers and a total of 650 neurons and 67,000 synapses. It consumes 2.3–6.8X lower power at state-of-the-art accuracies on two well-known KWS benchmarks, i.e., Google Speech Command Dataset (GSCD) for multi-keyword recognition ( Warden, 2018 ) and HeySnips for single-keyword spotting ( Coucke et al., 2019 ). In the remaining portion of this manuscript, we will present our SNN hardware architecture and the experimental results. In section “Materials and Methods,” we discuss the high-level SNN classifier architecture, elaborate on each of the components of the neurosynaptic core and introduce the experiment setup. In section “Results,” we present the results and finally conclude in section “Discussion.”",
"discussion": "Discussion Prior works on SNN hardware have focused on non-always-on application ( Akopyan et al., 2015 ; Chen et al., 2018 ; Davies et al., 2018 ), support for on-chip training ( Chen et al., 2018 ; Davies et al., 2018 ; Park et al., 2019 ) and support for both deep learning and neuromorphic workloads ( Pei et al., 2019 ). The absence of any prior work on SNNs for targeting always-on hardware motivated us to explore a new architecture for SNNs. We presented a fully spike-event-driven SNN classifier for an always-on intelligent function. We employed a fine-grained clock and power-gating to take advantage of the input signal sparsity, low leakage SRAM and a fixed priority arbiter to achieve a very low standby power of 75 nW. We trained the SNN for multiple always-on functions, notably multi- and single-keyword spotting benchmarks, achieving competitive accuracies. The average power consumption of the SNN chip scales with the input activity rate. It ranges from 75 nW with no input activity and 220 nW with the maximum input activity for the KWS benchmarks. Table 1 summarizes the comparison of our work with other recent KWS accelerators. Our design achieves 2.3–6.8X power savings compared to Shan et al. (2020) among KWS accelerators. If we scale the area of our design to 28 nm it would be 0.37 mm 2 which is still slightly higher than Shan et al. (2020) . The higher area usage of our work is possibly because it does not adopt time-sharing in neuron hardware. TABLE 1 Comparisons with recent KWS hardware. This work Shan et al. (2020) Guo et al. (2019) Giraldo and Marian (2018) Technology (nm) 65 28 65 65 Algorithm SNN DSCNN RNN LSTM Area (mm 2 ) 1.99 0.23 6.2 1.035 VDD (V) 0.52–1 0.41 0.9–1.1 0.575 Clock frequency 70 kHz @ 0.52 V 40 kHz 75 MHz 250 kHz Benchmark 1 GSCD (4 Keywords) GSCD (2 Keywords) GSCD (10 Keywords) TIMIT (4 Keywords) Accuracy (%) 91.8 94.6 90.2 92.0 Benchmark 2 HeySnips (1 Keyword) GSCD (1 Keyword) HeySnips (1 Keyword) N/A Accuracy (%) 95.8 98.0 91.9 N/A Power 75–220 nW* 510 nW** 134 μW 5 μW *Power consumption scales with input rate; **feature extraction circuits included. Our work does not have feature extraction circuits. They would increase the area and power when included. We can consider two feature extraction circuits ( Yang et al., 2019 , 2021a ), as candidates for the analog front end for our chip. Yang et al. (2021a) is the improved version of Yang et al. (2019) . We used the software model of the analog front end presented in Yang et al. (2019) . The power consumed by the analog front end and the feature extraction circuits is 50 nW in the improved version and 380 nW in the older version. The use of multiple supplies (VDD = 0.52 V and VDDH = 0.8 V) in our work can add some hardware and power overhead. There would be a significant increase in power consumption if we use only 0.8 V as the power supply for our chip. For example, if we assume that power consumption increases quadratically with VDD, then the power increases by 2.4X. We can consider two scenarios that can provide two different supplies and avoid a large increase in power. In one scenario, we assume an external DC-DC converter provides VDD while we can generate VDDH using a capacitor-based charge pump circuits ( Kim et al., 2021 ). The current load of the VDDH is not high since it is used in only a small part of SRAM. Therefore, even if the charge pump efficiency is not high, the overall impact is small. In the other scenario, we assume an external DC-DC converter provides VDDH and then we can generate VDD using an on-chip digital LDO. This LDO would have a power efficiency of 65% (VDD/VDDH), which increases total chip power dissipation by 53.8%. Table 2 summarizes the comparison of our design with other SNN hardware work (TrueNorth’s power is estimated from Cheng et al., 2017 ). Our design achieves over 30,000X power savings compared to Chen et al. (2018) in Table 2 . Our design is optimized for ultra-low power always-on functions while others are optimized for a balance between higher throughput and energy efficiency. High-performance SNN accelerators generally assume that input will be presented at a much higher rate, therefore, the time interval between spike events would be much smaller, limiting the benefit of fine-grained clock gating. Our design achieves competitive accuracies among both KWS and SNN hardware works and contributes to a growing body of literature that supports SNNs as an attractive low-power alternative to deep learning based hardware architectures. TABLE 2 Comparisons with recent SNN hardware. This work Koo et al. (2020) Park et al. (2019) Chen et al. (2018) TrueNorth Technology (nm) 65 90 65 10 28 Neuron count 650 1 410* 4096 1M Synapse count 67k 1 N/A 1M 256M Area (mm 2 ) 1.99 0.15 10.08 1.72 430 Clock frequency 70 kHz @ 0.52 V 37.5 MHz 20 MHz 105 MHz @ 0.5 V N/A MNIST classification Power 305 nW 282.8 mW † 23.6 mW 9.42 mW** 63 mW Accuracy (%) 97.6 92.3 97.8 97.9 97.6*** Throughput (inf/s) 2 N/A 100K N/A N/A Energy per inference (nJ) 195 N/A 236 1700 N/A Energy per SOP (pJ) 1.5 8.4 pJ/1.84 pJ †† N/A 3.8 26 *Input layer not included; **estimated from neuron’s power dissipation; ***estimated from Hsin-Pai Cheng et al., IEEE DATE 2017; † power reported in Koo et al. (2020) based on network size and power for one neuron and synapse; †† energy with sequencing circuits / Energy without sequencing circuits."
} | 3,364 |
36073844 | PMC9685438 | pmc | 853 | {
"abstract": "Abstract Resistive random‐access memories are promising candidates for novel computer architectures such as in‐memory computing, multilevel data storage, and neuromorphics. Their working principle is based on electrically stimulated materials changes that allow access to two (digital), multiple (multilevel), or quasi‐continuous (analog) resistive states. However, the stochastic nature of forming and switching the conductive pathway involves complex atomistic defect configurations resulting in considerable variability. This paper reveals that the intricate interplay of 0D and 2D defects can be engineered to achieve reproducible and controlled low‐voltage formation of conducting filaments. The author find that the orientation of grain boundaries in polycrystalline HfO \n x \n is directly related to the required forming voltage of the conducting filaments, unravelling a neglected origin of variability. Based on the realistic atomic structure of grain boundaries obtained from ultra‐high resolution imaging combined with first‐principles calculations including local strain, this paper shows how oxygen vacancy segregation energies and the associated electronic states in the vicinity of the Fermi level govern the formation of conductive pathways in memristive devices. These findings are applicable to non‐amorphous valence change filamentary type memristive device. The results demonstrate that a fundamental atomistic understanding of defect chemistry is pivotal to design memristors as key element of future electronics.",
"conclusion": "3 Conclusion Our combined experimental and ab initio study reveals that an intricate interplay of defects with different dimensionality plays the key role in predefining the formation path of the conducting filament. The controlled induction of a specific grain boundary type by texture transfer is a promising way to overcome present limitations set by the variability of forming voltages and is likely to be favorable for improved device endurance and cycle stability. For the sake of this study we have used MBE growth parameters outside the CMOS temperature budget. However, the favorable combination of a (111) oriented TiN electrode with a (11‐1) oriented HfO 2 functional layer can also be achieved with industrial relevant methods such as atomic layer deposition. The here suggested method of grain boundary and defect engineering is scalable well below 10 nm as the grain size is comparable to the thickness of the dielectric layer.",
"introduction": "1 Introduction A memristor is based on the controlled (digital or analog) [ \n \n 1 \n , \n 2 \n , \n 3 \n , \n 4 \n \n ] change of the resistance of a conductive pathway. In complementary metal oxide semiconductor (CMOS) relevant materials such as Hf—O or Ta—O, the conductive pathways consist of a local enrichment of oxygen vacancies. [ \n \n 5 \n , \n 6 \n , \n 7 \n , \n 8 \n \n ] Pre‐existing oxygen vacancies allow for low‐voltage and thickness independent electroforming as only a reshuffle of vacancies is required. [ \n \n 9 \n , \n 10 \n , \n 11 \n , \n 12 \n , \n 13 \n , \n 14 \n \n ] However, the oxygen vacancy distribution itself is still a random process associated with forming and operating voltage variability. [ \n \n 15 \n , \n 16 \n , \n 17 \n \n ] Defect engineering to reduce the device variability has been investigated in several works using dislocations and nanocomposites. [ \n \n 18 \n , \n 19 \n , \n 20 \n \n ] The introduction of threading grain boundaries (GBs) via GB engineering results in highly reproducible low‐voltage electroforming. [ \n \n 17 \n \n ] Here, we show that the complete materials picture is disclosed only when taking into account the intricate interplay between point‐defects (oxygen vacancies) and 2D defect planes (grain boundaries). The specific atomic configurations of the grain boundaries including strain effects result in enhanced or suppressed attraction between both types of defects. As a consequence, the selection of the proper GB allows us to create a predefined region of increased oxygen vacancy concentration which is associated with electronic defect states close to the Fermi level. These defect states, in turn, are the nuclei for the soft dielectric breakdown via formation of a defined conducting filament. This insight into the materials defect chemistry suggests new experimental methods of controlling the conducting filament, and thus also serves as a valuable guideline for future memristor designs. [ \n \n 21 \n , \n 22 \n , \n 23 \n , \n 24 \n , \n 25 \n , \n 26 \n , \n 27 \n , \n 28 \n , \n 29 \n , \n 30 \n , \n 31 \n , \n 32 \n , \n 33 \n , \n 34 \n \n ]"
} | 1,141 |
26426118 | PMC4591267 | pmc | 854 | {
"abstract": "Bleaching episodes caused by increasing seawater temperatures may induce mass coral mortality and are regarded as one of the biggest threats to coral reef ecosystems worldwide. The current consensus is that this phenomenon results from enhanced production of harmful reactive oxygen species (ROS) that disrupt the symbiosis between corals and their endosymbiotic dinoflagellates, Symbiodinium . Here, the responses of two important antioxidant defence components, the host coenzyme Q (CoQ) and symbiont plastoquinone (PQ) pools, are investigated for the first time in colonies of the scleractinian coral, Acropora millepora , during experimentally-induced bleaching under ecologically relevant conditions. Liquid chromatography-mass spectrometry (LC-MS) was used to quantify the states of these two pools, together with physiological parameters assessing the general state of the symbiosis (including photosystem II photochemical efficiency, chlorophyll concentration and Symbiodinium cell densities). The results show that the responses of the two antioxidant systems occur on different timescales: ( i ) the redox state of the Symbiodinium PQ pool remained stable until twelve days into the experiment, after which there was an abrupt oxidative shift; ( ii ) by contrast, an oxidative shift of approximately 10% had occurred in the host CoQ pool after 6 days of thermal stress, prior to significant changes in any other physiological parameter measured. Host CoQ pool oxidation is thus an early biomarker of thermal stress in corals, and this antioxidant pool is likely to play a key role in quenching thermally-induced ROS in the coral-algal symbiosis. This study adds to a growing body of work that indicates host cellular responses may precede the bleaching process and symbiont dysfunction.",
"conclusion": "Conclusions This study demonstrated that hyperthermal stress in A . millepora was associated with oxidation of the coral host CoQ pool redox state. This oxidation occurred prior to any measurable loss of Symbiodinium cells from the host and major decline in PSII photochemical efficiency. Thus the oxidation of CoQ pool redox state is among the earliest known impacts of hyperthermal stress on the cellular chemistry of the coral host and adds to a growing body of work that indicates host cellular responses may precede the bleaching process and symbiont dysfunction. Furthermore, the Symbiodinium PQ pool redox state remained unaffected by hyperthermal stress until PSII photochemical efficiency was severely impaired. At this stage, the PQ pool exhibited a five-fold increase in concentration and a distinct oxidative shift.",
"introduction": "Introduction Elevated seawater temperatures in conjunction with high solar irradiance disrupt the relationship between reef-building corals (Cnidaria: Scleractinia) and their dinoflagellate symbionts ( Symbiodinium sp.) [ 1 ] and have been implicated in causing mass coral bleaching events [ 2 – 4 ]. Although the molecular events underlying the loss of Symbiodinium cells via exocytosis [ 5 ] and apoptosis [ 6 ] remain unclear, it is broadly accepted that coral bleaching is preceded by oxidative stress: the excessive formation of reactive oxygen species (ROS) which eventually overwhelm the antioxidant defence capacity of the symbiosis [ 7 – 9 ]. Initial impairment of photosynthesis is thought to increase ROS formation in the symbionts, leading to oxidative damage in the host, which then initiates bleaching [ 7 ]. Several potential primary damage sites have been identified in Symbiodinium during thermal stress, including photosystem II (PSII) reaction centres [ 10 – 12 ], antenna pigments [ 13 ], the Calvin cycle [ 14 ], and the thylakoid membranes [ 15 ]. Other evidence suggests that the primary site of thermal damage in Symbiodinium varies among coral species and symbiont types [ 16 ] which may explain some of the apparent contradictory results to date. In addition, there is increasing evidence suggesting that the cnidarian host plays a more significant role in the bleaching cascade than previously thought because thermal stress can compromise host cells prior to damaging the symbiont [ 17 – 20 ] and because bleaching can occur in darkness, independent of photosynthetically produced ROS [ 21 ]. Nonetheless, it is clear that the coral host has substantial antioxidant potential, indicating ROS scavenging during exposure to thermal and irradiance stress is essential in both symbiotic partners in order to prevent bleaching [ 22 – 26 ]. Hence, oxidative stress is likely to reflect an imbalance between the antioxidant capacity of both partners and the performance of the electron transport chains (ETC) of coral mitochondria and Symbiodinium chloroplasts [ 7 ]. As components of both antioxidant defence systems and the electron transport chains that generate ROS, the prenylquinones coenzyme Q (CoQ; ubiquinone) and plastoquinone (PQ) and their respective reduced (antioxidant) forms ubiquinol (CoQH 2 ) and plastoquinol (PQH 2 ) may play key roles in the bleaching response. These redox carriers play an integral role in electron transport (CoQ/CoQH 2 in the mitochondrial ETC and PQ/PQH 2 in the photosynthetic ETC) but also have important antioxidant functions within mitochondrial [ 27 ], cellular [ 28 ] and thylakoid [ 29 ] membranes. The reduced forms of these prenylquinones are highly effective lipid peroxidation chain breakers, and are involved in the regeneration of other antioxidants such as ascorbate and α-tocopherol [ 28 , 30 – 32 ]. In addition, PQH 2 is an effective singlet oxygen ( 1 O 2 ) quencher in chloroplasts [ 33 , 34 ]. Consequently, shifts in the proportion of reduced to oxidised prenylquinones (%CoQH 2 ; %PQH 2 ) have been used to infer oxidative stress and ROS scavenging activity in plant models [ 30 , 33 , 35 , 36 ]. Little is known about how the coral CoQ and symbiont PQ pools respond to hyperthermal stress. In a proof of concept study, Lutz et al. [ 37 ] demonstrated that the Acropora millepora CoQ and the Symbiodinium PQ pool redox states are maintained predominantly in their reduced forms (a prerequisite for antioxidant action), and acute heat-stress causes increased oxidation of the coral CoQ pool consistent with evidence that oxidative stress occurs in both host and symbiont [ 23 , 24 , 38 ]. However, due to the acute nature of the stress applied, it is unclear whether the observed oxidative shift was a consequence of metabolic failure, or whether the CoQ pool is sensitive to prolonged elevated temperature stress under more ecologically relevant conditions. Here, quantitative liquid chromatography-mass spectrometry (LC-MS) was used to estimate the redox states of host CoQ and Symbiodinium PQ pools in colonies of the scleractinian coral A . millepora during experimentally-induced bleaching under ecologically relevant temperature conditions. The data on CoQ and PQ pool redox status, in combination with PSII photochemical efficiency, chlorophyll concentration and Symbiodinium density estimates were used to follow the effects of thermal stress on the state of the symbiosis over time.",
"discussion": "Discussion Coenzyme Q pool redox state The results presented here demonstrate that the A . millepora host CoQ redox state is sensitive to hyperthermal stress, exhibiting an overall 13% decline of CoQH 2 in response to the stressor ( Fig 2H ). This oxidative shift was not caused by an increase of de novo synthesised CoQ, as the total CoQ concentration (CoQ + CoQH 2 ) did not increase during the heat stress. Oxidation of the CoQ pool occurred early in the thermal stress treatment, upon reaching 32°C (after seven days with a daily temperature increase of approximately 0.7°C per day) and prior to any measurable loss of Symbiodinium cells from the host tissue. Oxidation of the CoQ pool occurred before a major decline in PSII photochemical efficiency was observed, i.e., while the effects of the hyperthermal stress on the Symbiodinium photosynthesis apparatus were still limited (F V /F M > 0.65). Moderate irradiance levels (350 μmol photons m −2 s −1 ) were used in order to avoid major light stress concomitantly with the applied hyperthermal stress [ 2 , 54 , 55 ]. Maximum daily irradiance at 1–3 m depths regularly exceeds 1000 μmol photons m −2 s −1 for nearby (< 25 km), equally turbid Great Palm Island waters [ 56 ]. The results therefore indicate that the A . millepora CoQ pool is oxidised significantly in response to hyperthermal stress in the absence of strong light exacerbating ROS leakage from the symbiont [ 57 , 58 ]. Although a contribution of photosynthetically derived ROS to the observed CoQ pool oxidation cannot be discounted, the results presented here add to a growing body of work that indicates host cellular responses may precede the bleaching process and symbiont dysfunction [ 5 , 17 – 20 , 59 ]. In addition, other thermal stress-related responses such as transcriptional and physiological changes that were not measured here are expected to occur in both coral symbiosis partners prior to host CoQ pool oxidation. For example, other reported early changes include a reduction in epithelial tissue, signs of increased apoptosis in the gastrodermis, and changes to the transcriptome, which have been associated with an upregulation of chaperone and antioxidant defence genes alongside transcriptional changes that, by analogy to vertebrate models, are assumed to be linked to apoptosis [ 18 , 25 , 26 , 60 – 63 ]. It should also be noted that the heat/light sensitivity of the photosynthetic apparatus varies among different symbiont types and that this affects host sensitivity to bleaching [ 64 – 66 ]; however, bleaching susceptibility differs widely among different coral genera despite often hosting the same Symbiodinium types [ 67 – 69 ]. Considering this, the results presented here require confirmation in other symbiont-host associations prior to postulation of a generalized physiological response during the bleaching cascade. Nonetheless, the oxidative shift in the CoQ redox state observed here is among the earliest known metabolic changes in the coral partner in response to a realistic temperature level. Coenzyme Q pool redox state regulation CoQ/CoQH 2 is present (in varying quantities) in all intracellular membranes of every animal with the highest concentrations found in the mitochondrial membranes at the primary site of ROS production [ 32 , 70 ]. In eukaryotes, CoQ redox processes are relatively complex ( Fig 4 ; for relevant enzymes identified in Acropora sp. see S1 Table ). Within mitochondrial membranes, CoQH 2 is continuously regenerated by the respiratory chain (complex I, II and alternative NAD(P)H dehydrogenases) [ 71 ] and other mitochondrial enzymes (glycerol-3-phosphate dehydrogenase, electron-transferring flavoprotein dehydrogenase, dihydroorotate dehydrogenase; [ 72 ]). In other membranes, several enzymes catalyse CoQ reduction including a NADH-cytochrome b 5 reductase [ 73 ] and a distinct, unresolved NADPH-CoQ reductase [ 74 ]. Interestingly, a cytosolic NAD(P)H:quinone reductase (NQO1; formerly DT-diaphorase) [ 75 ]–the most studied CoQ reducing enzyme–appears to be absent in cnidarians along with other NQO genes [ 76 ]. 10.1371/journal.pone.0139290.g004 Fig 4 Schematic diagram of electron transfer reactions using the coenzyme Q (CoQ) pool in the coral mitochondrial and plasma membrane electron transport. Respiratory “linear” electron flows (black arrows) proceed from NADH in the mitochondrial matrix to H 2 O via the CoQ pool and the enzyme complexes I, II, III, and IV, forming ubiquinol (CoQH 2 ) as an intermediary product. The electron flows via complexes I, III and IV occur (mostly) via tunnelling or micro-diffusion of CoQ/CoQH 2 in I-II-IV supercomplexes rather than via the larger mobile CoQ pool [ 72 ]. “Non-linear” electron flows (dark blue arrows) proceed from electron donors (e.g. NAD(P)H) via several quinone dehydrogenases to the CoQ pool, and to H 2 O from CoQH 2 via AOX. Plasma membrane electron transport occurs from NAD(P)H to H 2 O via one or more type of NAD(P)H-CoQ reductases, the plasma membrane CoQ pool and Ecto-NOX. CoQH 2 ROS scavenging occurs continuously in O 2 metabolism primarily via chain breaking of lipid peroxidation (LPO) caused by O 2 \n •− and H 2 O 2 . Abbreviations: AOX, alternative oxidase; cyt-c, cytochrome c; DHAP, dihydroxyacetone phosphate; DHO, dihydroorotate; DHODH, dihydroorotate dehydrogenase; Ecto-NOX, external quinone oxidase; ETF red/ox , reduced/oxidised electron-transferring-flavoprotein; ETFDH, electron-transferring-flavoprotein dehydrogenase reduced/oxidised; Ecto-NOX, external quinone oxidase; GPDH, glycerol-3-phosphate dehydrogenase; G-3-P, glycerol-3-phosphate; H 2 O 2 , hydrogen peroxide ; LPO, lipid peroxidation; pmNDH/mNDH, plasma membrane/mitochondrial NAD(P)H dehydrogenases; OA, orotate; O 2 \n •− , superoxide. In contrast to the reduction of the CoQ pool, CoQH 2 is oxidised by direct interaction with ROS, in particular lipid peroxyl radicals and the lipid peroxidation initiating perferryl radicals (Fe 3 O 2 \n •− ) found in all membranes [ 28 ], by complex III and alternative oxidases (AOX) of the mitochondrial ETC [ 77 ], and by the external oxidases of the plasma membrane electron transport (Ecto-NOX; [ 78 ]). Considering the known CoQ pool redox mechanisms of other animals, the oxidative shift in the CoQ redox state in A . millepora can therefore be attributed to: (1) an increase in ROS scavenging by CoQH 2 ; (2) a decline in net CoQ reduction by the mitochondrial ETC; (3) extra-mitochondrial pathways; or 4) any combination of these processes. Coenzyme Q pool reactive oxygen species scavenging in corals Attributing the thermal stress-induced oxidative shift in CoQ redox state of A . millepora to a specific physiological mechanism is difficult, primarily because current understanding of CoQ functions in the coral- Symbiodinium symbiosis is very limited and existing methods cannot distinguish between functionally and spatially different CoQ pools present in different organelles [ 79 ]. Theoretically, a net decline in CoQ reduction caused by the mitochondrial ETC or extra-mitochondrial pathways are conceivable by postulating a decline in CoQ reducing or an increase in CoQH 2 oxidising enzyme activities; however, no such direct impact of thermal stress on the CoQ pool has been demonstrated so far. In particular, the emerging consensus that the complexes I-III-IV occur mostly as supercomplexes further complicates attributing shifts in the CoQ redox state to a specific location in the mitochondrial ETC as electron transfer in these supercomplexes appears to occur via tunnelling or microdiffusion of CoQ/CoQH 2 rather than via a mobile CoQ pool in mitochondrial membranes [ 72 ]. Short term heat stress in the bleaching model Aiptasia has been reported to cause the degradation of host mitochondria prior to symbiont impairment and to lead to the downregulation of genes associated with ATP production and electron transport at the site of, and downstream from, cytochrome c [ 19 ]. However, the report did not include any genes upstream of complex III ( Fig 4 ), thus there is no indication that the CoQ pool reducing side of the mitochondrial ETC was affected. A recent transcriptional analysis provided further evidence of the thermal stability of complex III gene expression in A . millepora [ 80 ] but analyses at the enzyme activity level in cnidarians remain outstanding. The visible, pre-bleaching mitochondrial damage in Aiptasia [ 19 ] would be expected to incapacitate the mitochondrial ETC and potentially lead to CoQ oxidation due to a decline in electron flux to the pool; however, there is currently no data available to lend support to such a model. On the other hand, an inefficient mitochondrial ETC is likely leading to the increased formation of ROS [ 8 , 59 , 81 ], in particular during daytime hyperoxia [ 24 , 82 ]. Thus, even though it is possible that any components of the CoQ pool redox mechanisms are affected by thermal stress, attributing the oxidative shift in the CoQ redox state to increased CoQH 2 ROS scavenging in response to hyperthermal stress currently provides the most parsimonious explanation. In mammals, oxidative shifts in the CoQ pool redox state have been observed in a variety of pathological conditions that are associated with oxidative stress [ 35 , 36 , 83 ]. These oxidative shifts are understood to result from an increasingly challenged antioxidant defence [ 28 ]. In cnidarian- Symbiodinium symbioses, it has been repeatedly demonstrated that the cnidarian host reacts to thermal stress and high light by increasing its antioxidant activities, which indicates an increased requirement to detoxify ROS in the host tissues [ 7 , 38 , 54 , 84 – 86 ]. ROS formation also occurs in aposymbiotic cnidarians upon exposure to light and elevated temperatures, although in symbiosis, the hyperoxia caused by algal photosynthesis aggravates the coral’s innate ROS formation because it increases relative to oxygen concentration [ 24 , 59 , 87 – 91 ]. Bleaching in symbiotic cnidarians can also be triggered in the absence of photosynthetically produced ROS by thermal stress in darkness [ 21 ]. It is not yet understood what role non-photosynthetically produced ROS play in this dark-bleaching; however, mitochondria would appear to be the most likely origin for these ROS. Mitochondria (the primary source of ROS in animals) are the location of the highest cellular CoQ/CoQH 2 concentration in eukaryotes [ 28 , 78 ]. Here, superoxide (O 2 \n •− ) and other ROS are generated by enzymes involved in the ETC, particularly the NADH dehydrogenase of complex I, and the interface between the CoQ pool and complex III [ 91 , 92 ]. The co-localisation of the ROS producing respiratory ETC and the CoQ pool within mitochondrial membranes is likely an important factor in the high antioxidant effectiveness of CoQH 2 [ 28 , 93 ]. In addition to ROS generation by the coral, ROS leakage from Symbiodinium probably exceeds the host’s innate ROS generation [ 24 , 57 , 58 ]. Moreover, impaired or damaged photosynthetic ETC may further increase ROS formation and, ultimately, ROS leaking into the host [ 7 , 8 , 94 ]. The expulsion of Symbiodinium cells by the coral host has therefore been regarded as a protective mechanism: the coral prevents further ROS leakage from Symbiodinium into host cells by removing the primary source of ROS production and also by reducing tissue hyperoxia during daylight exposure [ 95 , 96 ]. The CoQ pool likely provides an early line of antioxidant defence because ROS leaking from Symbiodinium cells would need to cross the host-derived symbiosomal membrane, which like all animal membranes is expected to contain CoQ/CoQH 2 [ 32 , 70 ]. Nonetheless, it would be expected that an increase in Symbiodinium cellular ROS concentrations to a point where leakage into the host occurs would manifest as a distinct decline in PSII photochemical efficiency. However, a major decline in F V /F M was only observed here after significant CoQ pool oxidation had already occurred ( Fig 2F and 2H ). This suggests that ROS leakage is unlikely to be a major contributing factor to the initial oxidation of the CoQ pool, although this cannot be ruled out in the later stages where a major decline in F V /F M was observed. Plastoquinone pool redox state In contrast to the host CoQ pool redox state, the Symbiodinium PQ redox state remained stable until the point at which PSII photochemical efficiency was severely impaired and coral nubbins were distinctly bleached ( Fig 2C and 2D ). The observed initial stability of the PQ redox state, despite hyperthermal stress, is consistent with short-term acute heat stress results [ 37 ]. By analogy with high light stress [ 29 , 97 ], the oxidative shift due to hyperthermal stress could be caused by increased ROS scavenging of PQH 2 within Symbiodinium chloroplasts or changes in photosynthetic ETC such as increased plastid terminal oxidase activity. PQH 2 is a highly effective quencher of 1 O 2 [ 33 , 34 ] and, like CoQ, acts as a lipid peroxidation chain breaker either directly or via the regeneration of α-tocopherol [ 30 , 31 ]. Even though irradiance was maintained at a moderate level during the experiment described here, the applied temperature stress caused chronic photoinhibition of PSII, which is commonly reported in coral bleaching experiments (e.g. [ 14 , 15 , 98 ]) and a known indicator of ROS formation within the photosynthetic ETC [ 99 ]. A five-fold increase in PQ pool concentrations was recorded concomitantly with the observed PQ pool oxidation ( Fig 2I ). Newly synthesized PQ is predominantly in the reduced form (PQH 2 , not PQ) [ 97 ], thus the observed oxidative shift in the PQ pool at this stage should be the result of increased non-enzymatic formation of PQ from PQH 2 after its interaction with ROS, which are increasingly generated by a thermally damaged photosynthetic ETC [ 7 , 14 , 81 , 94 ]. In plants and algae, a considerable proportion of the PQ pool is associated with the chloroplast plastoglobuli which are thought to act as PQH 2 reservoirs [ 100 , 101 ]. Consequently, the size of the PQ pool increases when plants and algae are exposed to conditions that induce the formation of 1 O 2 , such as high light exposure [ 29 , 33 , 97 ]. Accordingly, the increase in the total PQ pool observed here could be seen as a cellular protective mechanism against the oxidative stress caused by the increasingly impaired photosynthetic ETC. The lack of compromised cell walls in the TEM images ( Fig 3 ) implies that this increase is unlikely to be the result of a normalisation artefact: i.e. loss of structurally compromised cells during the extraction procedure, thus underestimating cell counts. Nonetheless, the concomitant loss of internal cellular structure and the highly compromised state of thylakoid membranes at this time point indicate a need for further experimental work before a definitive protective mechanism can be attributed to de novo synthesised (reduced) PQ during bleaching."
} | 5,576 |
36248639 | PMC9560128 | pmc | 855 | {
"abstract": "Recently, spiking neural networks (SNNs) have been widely studied by researchers due to their biological interpretability and potential application of low power consumption. However, the traditional clock-driven simulators have the problem that the accuracy is limited by the time-step and the lateral inhibition failure. To address this issue, we introduce EvtSNN (Event SNN), a faster SNN event-driven simulator inspired by EDHA (Event-Driven High Accuracy). Two innovations are proposed to accelerate the calculation of event-driven neurons. Firstly, the intermediate results can be reused in population computing without repeated calculations. Secondly, unnecessary peak calculations will be skipped according to a condition. In the MNIST classification task, EvtSNN took 56 s to complete one epoch of unsupervised training and achieved 89.56% accuracy, while EDHA takes 642 s. In the benchmark experiments, the simulation speed of EvtSNN is 2.9–14.0 times that of EDHA under different network scales.",
"conclusion": "6. Conclusion and future work Based on the SNN event-driven framework EDHA, a simulator named EvtSNN is introduced. In this paper, two innovations are proposed to speed up the simulation without any accuracy loss. Firstly, repeated calculations are avoided according to the hidden information of the population. Secondly, the unnecessary calculation is filtered by the conditions derived from differential inequality. In the benchmark experiment, without the learning rule, the EvtSNN was the fastest in small network scale simulation (hundreds of neurons). EvtSNN always kept the lead when using the STDP learning rule. In the unsupervised training task of MNIST, EvtSNN only took 56 s to complete one epoch and reached 89.59% accuracy, which is 11.4 times faster than EDHA. Our work can be further improved. Firstly, large-scale network simulation can be optimized in combination with the clock-driven method. Secondly, multithreading acceleration and parallel computing can be used with the single-layer parallel structure of Inception (Szegedy et al., 2015 ; Meng et al., 2021 ) for population-level concurrent acceleration.",
"introduction": "1. Introduction Spiking neural networks (SNNs) (Maass, 1997 ) have attracted increasing attention because of their characteristics, including preferable biological interpretability and low-power processing potential (Akopyan et al., 2015 ; Shen et al., 2016 ; Davies et al., 2018 ; Moradi et al., 2018 ; Pei et al., 2019 ; Li et al., 2021 ; Pham et al., 2021 ). Compared to traditional artificial neural networks (ANNs), SNNs increase the time dimension so that they naturally support information processing in the temporal domain. To introduce the extra time dimension into the calculation, two methods are usually adopted: clock-driven and event-driven. The idea of clock-driven is to discretize the time and update the state of all neurons in each timestamp. The clock-driven method is widely used in the existing SNN frameworks (simulators) (Goodman and Brette, 2008 ; Hazan et al., 2018 ; Stimberg et al., 2019 ) because it is simulated by the iterative method which can be compatible with the differential equations of most neuron models. However, this method has two problems that cannot be ignored. Firstly, there is a conflict between simulation accuracy and calculation speed. The smaller the time step, the higher the simulation accuracy and the larger the calculation amount. Secondly, lateral inhibition cannot be effective on other neurons that fire lately in the same time slice. In the event-driven method, the state of neurons is updated when spikes are received, which means that the sparsity of spikes can be fully utilized to reduce computations. The realization of event-driven simulation on hardware (Davies et al., 2018 ; Li et al., 2021 ) has the ability of parallel computing and the potential of low-power processing, but it is costly and less flexible than software. Our team previously proposed an event-driven software simulation framework EDHA (Event-Driven High Accuracy), whose core task is to maintain the pulse priority queue (Mo et al., 2021 ). During the simulation, the earliest spike is popped from the queue, and then postsynaptic neurons are updated independently. However, the high complexity of its single update limits the overall simulation speed. In this paper, an event-driven software simulator named EvtSNN (Event SNN) is introduced, which includes two contributions. To begin with, neurons are clustered into populations, which means that intermediate results can be reused. In addition, pre-filtering is adopted to avoid unnecessary calculations according to the condition. After rewriting the framework code with the C++ programming language and combining these two innovations, the simulation speed of EvtSNN has been greatly improved. In the ablation experiment task, the processing capacity of EvtSNN(C++) reached 117.8 M spikes × fan-outs/s, which was 13 times that of EDHA(java). In the unsupervised training task of MNIST, the network (784–400) took 56 s to train one epoch with an accuracy of 89.59%, which is 11.4 times faster than EDHA. In Section 2, we describe the related work, including unsupervised learning and supervised learning, as well as clock-driven and event-driven simulation. In Section 3, the principle of EDHA is reviewed, and two innovations are proposed to accelerate the event-driven simulation. Section 4 contains several comparison experiments and results. And the discussion is in Section 5. Finally, Section 6 summarizes the current work.",
"discussion": "5. Discussion 5.1. Simulation accuracy In the experiment of Section 4.1, the membrane potential of neurons during the simulation of Brian2 and EvtSNN were recorded and plotted, as shown in Figure 9 . It can be seen that the voltage variation trend in the two simulations is basically the same, but sometimes there are small errors that affect spike delivery. Interestingly, even if the pulses do not match, the voltage will tend to be consistent after a while. This may be due to the decay of membrane potential. As time passes, the subsequent state is mainly affected by the input pattern rather than the initial state. Figure 9 Voltage curves of neurons in Brian2 and EvtSNN simulations. The clock-driven framework Brian2 has higher simulation accuracy when using smaller time steps (e.g., dt = 0.01 ms), and its voltage is close to the result of the event-driven framework EvtSNN. When using a larger time step (e.g., dt = 1 ms), the simulation accuracy of Brian2 decreases and leads to changes in the firing pattern. The simulation accuracy of the clock-driven method is limited by the time step, which may cause a small number of spike mismatches. However, in most cases, there are similar voltage curves in clock-driven (Brian2) and event-driven (EvtSNN) simulation, and the overall spike pattern is not much different, which means that the error of clock-driven simulation could be ignored many times. To sum up, there are some simulation errors in the clock-driven method, which can be ignored in most cases; the precision of the event-driven method can be very high, and it can use network sparsity to reduce the amount of calculation, which has a higher potential. 5.2. Quantitative analysis of sub-steps acceleration In this section, a dynamic code analysis tool (Clion Profiler) was employed to count the time-consuming of each part. With the 10 kHz sampling frequency and 100 times repeated tasks (same as Section 4.1), the simulation time-consuming composition is shown in Figure 10 . Population computing reduces the calculation time of sub-step (a) from 77.75 to 34.34 ms, while pre-filtering reduces the time consumption of sub-step (c) from 76.3 to 0.86 ms. It can be seen that the accelerating effect of pre-filtering is commendable so that the time consumption of sub-step (c) can be neglected. Even poor filtering in extreme cases does not slow down the overall simulation because its computational overhead is negligible. Figure 10 Time-consuming components in simulation. Using Pre-filtering (+filter) avoids unnecessary computation and can significantly reduce the time-consuming of sub-step (c). Enabling population computing (+popu) avoids repeated calculations and speeds up sub-step (a). (A) EvtSNN (base), (B) EvtSNN (+filter), (C) EvtSNN (+filter+popo). 5.3. Acceleration ability in multi-scale network To measure the contribution of each innovation in different network scales, we combined the benchmark and ablation experiment, and the results are shown in Figure 11 . Firstly, the acceleration effect of population calculation and pre-filtering under different network scales is relatively stable, reducing the time consumption by about 25 and 35%, respectively compared with EvtSNN (base). Secondly, when the average input frequency ( Fr ) of the neuron group is higher than the output frequency, the delayed update term can have some acceleration effect, otherwise, it will have a negative effect. In addition, after code optimization and rewriting, the speed of EvtSNN (base) is 2.6–4.5 times that of EDHA under the same calculation flow. Finally, EvtSNN using all optimization items is 2.9–14.0 times faster than EDHA. Figure 11 Time-consuming comparison chart. (A–C) EvtSNN time-consuming reduction after enabling popu, lazy, and filter, respectively. (D,E) Time-consuming comparison of EvtSNN and EDHA with none/all optimizations. 5.4. Limitations Of course, there are some limitations to our framework. First, as an event-driven framework, EvtSNN has poor model compatibility, requiring the derivation of time-domain equations and the solution of spike firing time. Secondly, the pre-filtering formulation is only used for the neuronal model used. However, pre-filtering formulas for other neuronal models can draw on the derivation process in this paper (3.5)."
} | 2,470 |
30708986 | PMC6387213 | pmc | 856 | {
"abstract": "The creation of functional materials from renewable resources has attracted much interest. We previously reported on the genetic code expansion of the domesticated silkworm Bombyx mori to functionalize silk fiber with synthetic amino acids such as 4-azido-L-phenylalanine (AzPhe). The azido groups act as selective handles for biorthogonal chemical reactions. Here we report the characterization and scaled-up production of azido-functionalized silk fiber for textile, healthcare, and medical applications. To increase the productivity of azido-functionalized silk fiber, the original transgenic line was hybridized with a high silk-producing strain. The F 1 hybrid produced circa 1.5 times more silk fibroin than the original transgenic line. The incorporation efficiency of AzPhe into silk fibroin was retained after hybridization. The tensile properties of the azido-functionalized silk fiber were equal to those of normal silk fiber. Scaled-up production of the azido-functionalized silk fiber was demonstrated by rearing circa 1000 transgenic silkworms. Differently-colored fluorescent silk fibers were successfully prepared by click chemistry reactions, demonstrating the utility of the azido-functionalized silk fiber for developing silk-based materials with desired functions.",
"conclusion": "4. Conclusions An azido-bearing synthetic amino acid, AzPhe, can be incorporated into silk fiber during protein synthesis by expressing a mutant of BmPheRS in the PSGs of transgenic silkworms. The azido-functionalized silk fiber can be easily modified by click chemistry reactions. The highest efficiency of AzPhe incorporation was previously observed with the H06 transgenic line, which expresses the F432V mutant of BmPheRS. For industrial applications, it was necessary to improve the productivity of the azido-functionalized silk fiber. We generated an F 1 hybrid of the H06 line with a high silk-producing strain and investigated its fibroin production and AzPhe incorporation into silk fiber. We demonstrated that fibroin production increased approximately 1.5 times by hybridization, whereas AzPhe incorporation was not affected. The mechanical properties of the azido-functionalized silk fiber exhibited no significant differences from normal fiber. Scaled-up production was performed using approximately 1000 transgenic silkworms, which produced circa 160 g of the azido-functionalized silk fiber. As a demonstration of click modification, we prepared differently-colored fluorescent silk fiber by SPAAC with fluorescent molecules. This study realized a rare opportunity to apply genetic code expansion methodology to the production of protein-based functional materials on an industrial scale.",
"introduction": "1. Introduction Materials development based on naturally derived renewable resources is of great interest because such sources are independent of fossil resources and have lower impact on the environment [ 1 , 2 , 3 ]. Silk fiber produced by the domesticated silkworm, Bombyx mori , is one such natural resource and has the advantageous properties of mechanical toughness, elasticity, biocompatibility, and biodegradability [ 4 , 5 , 6 ]. The development of novel technologies to utilize B. mori silk could thus lead to the creation of high-performance bio-based materials for textile, healthcare, and medical fields. We have been developing a novel technology to incorporate synthetic amino acids with unnatural functional groups into silk fibroin [ 7 , 8 , 9 , 10 ], which is the major protein component of B. mori silk and is regarded as a heterodimer of fibroin heavy chain (FibH; ~390 kDa) and fibroin light chain (FibL; ~26 kDa). The technology for incorporating synthetic amino acids into proteins is referred to as genetic code expansion, and it could enhance the utility of proteins as renewable materials. In the case of silk fibroin, incorporation of synthetic amino acids bearing azido groups endows the fibers with novel functionalities. Azido groups can act as selective chemical handles for further modifications with desired functional molecules for specific applications. To achieve the in vivo incorporation of an azido-bearing synthetic amino acid, 4-azido-L-phenaylalanine (AzPhe) ( Figure 1 A), into silk fibroin, several mutants of B. mori phenylalanyl-tRNA synthetase (BmPheRS) which efficiently accommodate AzPhe as a substrate were selected in E. coli by a growth inhibition assay [ 10 ]. The selected mutants were expressed in the posterior silk glands (PSGs) of transgenic B. mori larvae. The PSGs are the organs where silk fibroin is synthesized. The transgenic line with the highest AzPhe incorporation efficiency was designated as the H06 line, which expresses the F432V mutant of BmPheRS [ 10 ]. When the H06 larvae were fed a diet containing AzPhe (0.05 wt% in dry diet) from the third day in their final (fifth) instar until the start of spinning, 6.6% of 35 phenylalanine (Phe) residues in silk fibroin, corresponding to an average 2.3 residues per molecule, were replaced to AzPhe. The azido groups of AzPhe in silk fibroin were confirmed to be available for bioorthogonal modifications by the click chemistry reaction, azido-alkyne [3+2] cycloaddition [ 11 , 12 ]. The reaction proceeded in three different material forms: fiber, film, and porous sponge [ 9 ]. The azido groups can be converted to alkyne groups by reaction with a bifunctional compound, DBCO-PEG4-DBCO [ 13 ]. The H06 line was derived from a small-sized strain suited for transgenesis experiments and thus produced smaller cocoons and thinner fibers than strains used for practical silk production. In this study, the H06 line was hybridized with a high silk-producing strain to achieve scaled-up production of the azido-functionalized silk fiber using a conventional automatic reeling machine. The F 1 hybrid produced circa 1.5 times larger amounts of silk fibroin, and the incorporation efficiency of AzPhe was maintained after hybridization. The mechanical properties were not affected by AzPhe incorporation. We demonstrated the preparation of differently-colored fluorescent silk fibers by click chemistry reactions. This study realized a rare opportunity to apply genetic code expansion methodology to the production of proteins containing synthetic amino acids on an industrial scale.",
"discussion": "2. Results and Discussion 2.1. Fibroin Production and AzPhe Incorporation The H06 line was generated based on the MCS601 strain, which is a relatively small-sized strain appropriate for genetic transformation experiments. The cocoon shells of MCS601 are thinner than those of the high silk-producing strains generally used for silk production on an industrial scale. To improve the productivity of the azido-functionalized silk fiber, the H06 line was hybridized with a high silk-producing wildtype strain, Nichi509 × Nichi510 [ 14 ]. When a transgenic line bearing homozygous transgene(s) is hybridized with a wildtype strain, the copy number of transgenes decreases by half. In the case of the H06 line, it has one homozygous BmPheRS mutant (F432V) gene; that is, it has two copies of the transgene in its genome. Its F 1 hybrid with a wildtype strain is heterozygous and has one copy of the transgene. A decrease in the copy number would lead to decreased expression of the mutant enzyme, which might lead to lowered incorporation of AzPhe into silk fibroin. Male larvae of the H06 line and its F 1 hybrid with Nichi509×Nichi510 were fed a diet containing AzPhe (0 or 0.05 wt% in dry diet) from their third day of fifth instar until the start of spinning. No adverse effects on their larval growth were observed ( Figure S1 ); the F 1 hybrid grew heavier than the H06 line as expected. Larger amounts of diet were required for maturation of the F 1 hybrid than the H06 line ( Figure S2 ). Figure 1 B shows the fibroin production and AzPhe incorporation rate in the H06 and its F 1 hybrid with Nichi509×Nichi510. As reported previously [ 10 ], the fibroin production decreased upon AzPhe administration. The F 1 hybrid produced larger amounts of fibroin than the H06 line, whereas the AzPhe incorporation rate, estimated from the ratio of peak intensities in matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF-MS) spectra ( Figure S3 ), was not significantly different between the two lines. It is noteworthy that the AzPhe incorporation rate was not affected by hybridization despite the decrease in the copy number of the transgene encoding the BmPheRS mutant (F432V). We speculated that the concentration of the mutant enzyme in PSG cells was high enough for sufficient enzymatic activity to aminoacylate tRNA Phe with AzPhe even when the copy number of the transgene decreased by half. In the above comparison, only male larvae were used for the experiments to avoid data variation due to sex differences. However, in industrial production of silk fiber, both male and female larvae were reared without discrepancy. Varied amounts of AzPhe (0, 0.02, 0.05, or 0.1 wt%) were then administered separately to male and female larvae of the F 1 hybrid, H06 × (Nichi509 × Nichi510), to investigate the sex difference as well as to verify the optimal AzPhe concentration in diet. Fibroin production and the AzPhe incorporation rate exhibited no statistically significant differences between male and female larvae ( Figure 1 C), demonstrating that both male and female larvae of the F 1 hybrid were available for industrial silk production. We verified that 0.05 wt% AzPhe in dry diet was optimal to balance fibroin production and AzPhe incorporation ( Figure 1 C). This result was consistent with the previous result obtained with the H06 line [ 10 ]. 2.2. Cocoon Reeling and Mechanical Property Fifth instar larvae of the F 1 hybrid were reared on a normal or an AzPhe-containing diet and their cocoons were harvested. Raw silk fiber was reeled out from the cocoons using a conventional automatic reeling machine. Characterization of the raw silk fiber revealed no significant differences in the fiber length, fiber size, or degumming ratio ( Table 1 ). On the other hand, the raw silk ratio became significantly lower when AzPhe was incorporated into silk fibroin. AzPhe administration decreased the fibroin production, as shown in Figure 1 B, which would be one reason for the lower raw silk ratio. Another reason might be that AzPhe administration influenced the weight of the pupae. However, the relationship between AzPhe administration and the weight of pupae has not been investigated. Mechanical properties of the raw silk fiber (Young’s modulus, maximum stress, and maximum strain) were compared ( Table 2 ). None of the properties were significantly changed by AzPhe incorporation. In our system, AzPhe replaces a portion of the 35 Phe residues found in silk fibroin. Phe are not observed in the crystalline regions, which are mainly made of Gly-Ala repeats [ 15 ]. From the data in Figure 1 C, the replacement ratio of Phe to AzPhe was calculated to be circa 6.8% in the F 1 hybrid (including both male and female larvae) when 0.05 wt% AzPhe was added to the dry diet. The 6.8% replacement roughly corresponds to 2.4 residues of AzPhe in one silk fibroin molecule comprised of circa 5500 amino acids. We speculated that such a small number of AzPhe residues in the non-crystalline regions probably had a negligible impact on the mechanical properties of raw silk fiber. The azido-functionalized silk fiber produced by the F 1 hybrid can thus be directly used for industrial applications via the same processing methods as normal raw silk fiber. As a trial to scale-up the production of the azido-functionalized silk fiber, we reared circa 1000 larvae of the F 1 hybrid on an AzPhe-containing diet (0.05 wt% in dry diet) and obtained circa 160 g of raw silk fiber ( Figure 2 ). 2.3. Click Modification The sericin coating of the raw silk fiber was removed by a degumming process to obtain pure fibroin fiber. To verify the reactivity of azido groups in fibroin, the degummed fiber was treated with fluorescent reagents bearing dibenzocyclooctyne (DBCO) groups. DBCO groups selectively react with azido groups by a click chemistry reaction, strain-promoted azido-alkyne cycloaddition (SPAAC) [ 11 , 12 ] ( Figure 3 ). The azido-functionalized degummed silk fiber was soaked in a solvent containing green or red fluorescent reagent and reacted overnight at room temperature. The fiber was then thoroughly washed and dried. Distinct fluorescence was observed on the fiber, whereas the control fiber not containing azido groups only exhibited weak background fluorescence due to the intrinsic blue autofluorescence and nonspecific binding of the reagents ( Figure 4 A). The azido-functionalized degummed silk fiber was reacted with excess amounts of DBCO-PEG4-DBCO to convert azido groups into DBCO groups. The DBCO-functionalized fiber was then reacted with a blue fluorescent reagent. The distinct blue fluorescence was observed on silk fiber ( Figure 4 B). The above results demonstrated that the azido-functionalized silk can be readily modified with fluorophores in its native fibrous form by SPAAC, in this case yielding differently-colored fluorescent silk fiber. By changing fluorophores to other functional molecules, such as pigments, drugs, polymers, peptides, and proteins, silk fiber with any desired functions can be easily produced. Silk can also be processed into various forms other than fibers, including aqueous solutions, hydrogels, transparent films, porous sponges, nanofibers, and nanoparticles [ 16 , 17 , 18 ]. We previously prepared a transparent film and porous sponge from the azido-functionalized silk and demonstrated that these silk materials were readily modified by SPAAC [ 9 ]. Since SPAAC requires no toxic catalysts or ligands and exhibits high chemical selectivity, variously-modified silk fiber or silk-based materials would be useful for a broad range of applications, including applications in the textile, healthcare, and medical fields."
} | 3,507 |
35377486 | PMC9325543 | pmc | 857 | {
"abstract": "Summary \n Renewed interests in the development of bioenergy, biochemicals, and biomaterials have elicited new strategies for engineering the lignin of biomass feedstock plants. This study shows, for the first time, that 3,4‐dihydroxybenzoate (DHB) is compatible with the radical coupling reactions that assemble polymeric lignin in plants. We introduced a bacterial 3‐dehydroshikimate dehydratase into hybrid poplar ( Populus alba × grandidentata ) to divert carbon flux away from the shikimate pathway, which lies upstream of lignin biosynthesis. Transgenic poplar wood had up to 33% less lignin with p ‐hydroxyphenyl units comprising as much as 10% of the lignin. Mild alkaline hydrolysis of transgenic wood released fewer ester‐linked p ‐hydroxybenzoate groups than control trees, and revealed the novel incorporation of cell‐wall‐bound DHB, as well as glycosides of 3,4‐dihydroxybenzoic acid (DHBA). Two‐dimensional nuclear magnetic resonance (2D‐NMR) analysis uncovered DHBA‐derived benzodioxane structures suggesting that DHB moieties were integrated into the lignin polymer backbone. In addition, up to 40% more glucose was released from transgenic wood following ionic liquid pretreatment and enzymatic hydrolysis. This work highlights the potential of diverting carbon flux from the shikimate pathway for lignin engineering and describes a new type of ‘zip‐lignin’ derived from the incorporation of DHB into poplar lignin.",
"introduction": "Introduction Lignin is a phenolic polymer found predominately in the secondary cell walls of xylem vessels, tracheids, and fibres where it plays crucial roles facilitating water transport and providing structural support to plants. Lignin is assembled primarily from three 4‐hydroxycinnamyl alcohols known as monolignols – p ‐coumaryl, coniferyl, and sinapyl alcohols – which form the p ‐hydroxyphenyl (H), guaiacyl (G), and syringyl (S) subunits of lignin, respectively (Boerjan et al ., 2003 ). Following biosynthesis in the cytosol via the general phenylpropanoid pathway, lignin monomers are then exported to the cell wall and, once oxidized by laccase and/or peroxidase enzymes, undergo radical coupling reactions to form polymeric lignin (Freudenberg & Neish, 1968 ; Ralph et al ., 2004). The molecular processes underpinning lignin formation are highly flexible and it has been shown that various noncanonical monomers are compatible with lignification. Among these are the acylated monolignol conjugates that result in ester‐linked acetate, benzoate, p ‐hydroxybenzoate, p ‐coumarate, and ferulate (Nakamura & Higuchi, 1976 ; Ralph & Lu, 1998 ; Lu & Ralph, 2002 ; Wilkerson et al ., 2014 ; Karlen et al ., 2016 ; Kim et al ., 2020 ; Goacher et al ., 2021 ); pathway intermediates such as hydroxycinnamaldehydes, caffeyl alcohol, and 5‐hydroxyconiferyl alcohol, (Kim et al ., 2000 ; Ralph et al ., 2001 ; Kim et al ., 2003 ) and even phenolic compounds arising from outside the monolignol biosynthetic pathway such as flavonoids, hydroxystilbenes, and hydroxycinnamamides (del Río et al ., 2012 , 2018 , 2020 ; Lan et al ., 2015 ; Mahon et al ., 2021 ). Woody feedstocks represent an abundant and fast‐growing source of lignocellulosic biomass for use in the production of pulp and paper, biofuels, biomaterials, and biochemicals (de Vries et al ., 2021 ). Genetic engineering of economically important feedstock species is an attractive and proven strategy for improving the efficiency of biomass utilization (Mahon & Mansfield, 2019 ). Moreover, by harnessing the plasticity of lignification, noncanonical monomers can become incorporated into lignin leading to polymers with improved digestibility and/or high‐value constituents (Sederoff et al ., 1999 ; Ralph et al ., 2004 , 2008a ; Vanholme et al ., 2012 ; Mottiar et al ., 2016 ; del Río et al ., 2020 ). Lignin engineering efforts have been further buoyed by advancements in lignin valorization, microbial strain engineering, and the utilization of lignin‐derived phenols via metabolic upgrading to create high‐value biochemicals (Beckham et al ., 2016 ). In previous work, expression of a bacterial 3‐dehydroshikimate dehydratase gene ( QsuB ) from Corynebacterium glutamicum in Arabidopsis (Eudes et al ., 2015 ) led to the conversion of 3‐dehydroshikimic acid, an intermediate of the shikimate pathway, into 3,4‐dihydroxybenzoic acid (hereafter denoted as DHBA, with DHB being used to denote the dihydroxybenzoate ester and the phenolic moiety present in lignin, and also known as protocatechuic acid). This intervention diverted carbon flux away from the production of phenylalanine and resulted in reduced lignin content and improved saccharification potential. Herein, we investigated the effects of heterologous expression of QsuB in the plastids of hybrid poplar ( Populus alba × grandidentata ). Not only did this strategy result in reduced lignin content and improved saccharification, but it also led to the participation of monolignol–DHB conjugates in lignification, resulting in pendent DHB moieties on the lignin as well as backbone‐integrated DHB units, ultimately producing a novel type of ‘zip‐lignin’.",
"discussion": "Discussion Lignin engineering with QsuB \n Fast‐growing and highly productive hardwood species such as eucalypts, poplars, and willows are abundant sources of lignocellulosic biomass, and have increasingly become targets for breeding and/or genetic modifications aimed at altering wood properties to improve the efficiency of industrial biomass utilization (Mahon & Mansfield, 2019 ; Murphy et al ., 2021 ). Gain‐of‐function approaches that not only reduce cell wall recalcitrance but which also impart additional value to biomass represent a particularly promising strategy to enable more cost‐effective processing of woody biomass. In this study, we transformed hybrid poplar with a bacterial 3‐dehydroshikimate dehydratase gene ( QsuB ) that was previously shown to reduce lignin deposition and improve saccharification yields in Arabidopsis and switchgrass (Eudes et al ., 2015 ; Hao et al ., 2021 ). Expression of QsuB in hybrid poplar led to the hyperaccumulation of soluble DHB in the xylem, and effectively diverted carbon flux away from the phenylpropanoid pathway resulting in decreased lignin content, altered lignin monomer composition, and improved saccharification potential compared to WT trees. Cell wall analysis of the QsuB ‐poplars showed that conjugated forms of DHBA were actually polymerized into the lignin leaving some DHB as ester‐linked pendent groups. Moreover, the discovery of DHB‐derived benzodioxane structures in the 2D‐NMR spectra suggests that the DHB moieties of these conjugates can also participate in radical coupling reactions themselves leading to backbone‐integrated DHB and the possibility of a new type of ‘zip‐lignin’ resulting from the ester bonds derived from lignification using monolignol–DHB conjugates. Hyperaccumulation of DHB in QsuB ‐poplar Transgenic QsuB ‐poplar accumulated significant amounts of soluble DHB/DHBA, presumably as glycosylated forms (Fig. 1 ). DHBA produced in planta could be extracted as soluble compounds that can, for example, be further converted into cis , cis ‐muconic acid ( cc MA), a precursor for chemicals such as adipic acid, terephthalic acid, and trimellitic acid, all of which are used in the production of various plastics and nylons (Xie et al ., 2014 ). Recent studies have demonstrated that DHBA produced by genetically engineered tobacco expressing QsuB can be converted into cc MA using an engineered Escherichia coli strain (Wu et al ., 2017 ). Employing the same lignin‐specific promoter, the QsuB ‐poplar lines in this study produced up to 11.5 mg g −1 of extractable DHBA in xylem tissue (Fig. 1 ) compared to just 1.45, 0.133 and 0.375 mg g −1 produced by QsuB ‐expressing tobacco, Arabidopsis, and switchgrass, respectively (Eudes et al ., 2015 ; Wu et al ., 2017 ; Hao et al ., 2021 ). Given the fast growth rates of hybrid poplar and the proven capacity to generate substantial amounts of DHB, these trees could prove to be useful as value‐added woody feedstocks. Incorporation of DHB conjugates into lignin Lignin is synthesized via dehydrogenative polymerization driven by the radical coupling of, primarily, the three canonical monolignols p ‐coumaryl, coniferyl, and sinapyl alcohol. However, lignification is a highly flexible process and a growing number of noncanonical subunits have been reported in diverse plant taxa (del Río et al ., 2020 ). In this study, we have shown that QsuB ‐poplar trees not only accumulate DHB in the soluble fraction, but DHB conjugates were also incorporated into the lignin resulting in ester‐linked DHB pendent groups, as well as lignin backbone‐integrated units (Figs 2 , 4 ). Unlike in most other dicots, the lignin of poplar is abundantly decorated with p HB pendent groups, which are conjugated to monolignols in the cytosol prior to export to the cell wall space where the monolignol moieties of these conjugates radically couple into lignin, giving rise to free‐phenolic pendent groups of p HB (Ralph, 2010 ). The biosynthesis of p HBA proceeds from p ‐coumarate, an intermediate of the general phenylpropanoid biosynthetic pathway (Loscher & Heide, 1994 ; Sircar & Mitra, 2008 ). By diverting carbon flux in the shikimate pathway away from the production of phenylpropanoids including p ‐coumarate, expression of QsuB in poplar evidently reduced the availability of p HBA for lignin acylation. \n p ‐Hydroxybenzoylation of lignin occurs via the incorporation of acylated monolignols rather than by post‐polymerization acylation (Lu et al ., 2015 ). And, the gene encoding this specific p ‐hydroxybenzoyl‐coenzyme A (CoA) monolignol transferase has recently been identified and functionally tested in poplar in two independent studies (Zhao et al ., 2021 ; de Vries et al ., 2022 ). In QsuB ‐poplar, we observed significant reductions in p HBA released from the cell wall following alkaline hydrolysis. By contrast, several conjugates of DHBA were detected (Fig. 2 ), with several of these being glycosylated forms. It is plausible that these novel cell‐wall‐bound DHB groups could have arisen from the same monolignol transferase(s) that are normally responsible for the production of monolignol– p HB conjugates (Fig. S9 ). If so, our findings imply that these transferases possess significant promiscuity in substrate tolerance and that monolignol–DHB conjugates can be effectively exported to the cell wall and incorporated into lignin. Competition for the transferase between endogenous p HBA and excessive DHBA could also partly explain the observed decreases in cell‐wall‐bound p HB groups. Surprisingly, the glycosylated forms may also be accepted by the transferase as these too occurred on the lignin of QsuB ‐poplar. In addition to the transferase, 4CL or 4CL‐like enzymes must also then act on DHBA to produce the CoA thioester form (Fig. S9 ). Alternatively, given that DHB esters can evidently participate in radical coupling reactions, it is possible that DHB esters other than those conjugated with a monolignol could be exported and incorporated into lignin without the involvement of any monolignol acyltransferases. As support for this hypothesis, cell wall analyses of the leaves of Vitis vinifera and the roots of Ginkgo biloba (both species that are not known to possess highly acylated lignin) showed that DHB can occur naturally in a cell‐wall‐bound form at low levels (Codignola et al ., 1989 ; Weber et al ., 1995 ). Whether or not DHB‐acylation of lignin in QsuB ‐poplar is facilitated by monolignol acyltransferases is a question that will be best answered in due course using enzyme activity assays of the recently characterized p ‐hydroxybenzoyl‐CoA monolignol transferase. NMR analysis of QsuB ‐poplar lignin revealed signatures of DHB‐derived benzodioxane structures (Fig. 4 ). As was observed in transgenic poplars expressing an exotic feruloyl‐CoA monolignol transferase in which increased levels of readily cleavable ester bonds were incorporated into the lignin backbone (Wilkerson et al ., 2014 ), DHB evidently participates in radical coupling reactions and thereby incorporates into the backbone structure of lignin. Unlike p HB esters that prefer radical transfer over radical coupling reactions and thus occur as pendent groups in poplar, the second hydroxyl group of DHB appears to render it more compatible with the radical coupling reactions of lignification, leading to DHB units that integrally incorporate into the polymer. Alkaline hydrolysis also revealed the presence of cell‐wall‐bound glycosides of 4‐hydroxy‐3‐methoxybenzoate and 3‐hydroxy‐4‐methoxybenzoate in QsuB ‐expressing poplar. As these compounds were absent from the alkaline hydrolysates of WT poplar, we postulate that they may have been produced directly from DHBA. This would involve one or more O ‐methyltransferase enzymes that methylate the para or meta hydroxyl groups of DHBA. Although none of the 26 O ‐methyltransferases in poplar has been specifically reported to have such activity, such enzymes are known to accept diverse phenolic substrates (Lam et al ., 2007 ; Barakat et al ., 2011 ). Indeed, nonspecific O ‐methyltransferase activity towards DHBA has been described previously in poplar and in various other plant taxa as well (Finkle & Masri, 1964 ; Kuroda, 1983 ). Although we have not conclusively demonstrated their association with lignin, it is apparent that phenol‐glycosylated monolignol–DHB esters can be incorporated into the cell wall. Phenols involved in radical coupling following one‐electron oxidation (‘radicalization’) need to be in their free‐phenolic form, but monomers or monomer conjugates in which other noncrucial phenolic groups are glycosylated can still enter lignification via the radical generated from the molecule's free phenolic group. As a precedent, hydroxystilbene glucosides were recently identified in the bark of Norway spruce (Rencoret et al ., 2019 ). We anticipate that future studies will uncover further examples of such lignin‐bound sugars, particularly in plant tissues in which phenolic glycosides accumulate to high levels. Effects on lignin and biomass recalcitrance Targeted expression of QsuB in the plastids of lignifying xylem tissue diverts carbon flux away from the shikimate pathway and towards the production of DHBA. In transgenic Arabidopsis lines expressing QsuB there was no reduction in tryptophan, phenylalanine, tyrosine or salicylate, all downstream products of the shikimate pathway; even so, there was an accumulation of the precursors for the H lignin units ( p ‐coumarate, p‐ coumaraldehyde, and p‐ coumaryl alcohol) (Eudes et al ., 2015 ). In poplar, we found that heterologous expression of QsuB results in reduced lignin content and a significant accumulation of H‐lignin units along with a higher S : G lignin monomer ratio compared to WT trees. Reduction of carbon flux into or within the phenylpropanoid pathway often results in reduced overall lignin, combined with an increase in the S : G ratio, as metabolic flux into G‐lignin is more severely impacted by a reduction in substrate availability (Wang et al ., 2014 , 2018 ). \n QsuB acts to divert carbon flux upstream of shikimate and, in so doing, could lead to a reduction in the pool of shikimate available during lignin biosynthesis. As p ‐hydroxycinnamoyl‐CoA shikimate p ‐hydroxycinnamoyl transferase (HCT) requires shikimate as a co‐substrate, a reduction in the available shikimate would clearly hinder the progression of monolignol biosynthesis beyond HCT leading to coniferyl and sinapyl alcohols (Bonawitz & Chapple, 2010 ). In this way, proportionately more p ‐coumaryl alcohol would be produced resulting in an increase in H‐lignin units, as we saw with QsuB ‐poplar. In previous work, it has been shown that at least one poplar HCT can accept DHBA as a co‐substrate instead of shikimate (Eudes et al ., 2016 ). While it remains to be seen whether the resulting p ‐coumaroyl–DHB is tolerated by coumarate 3‐hydroxylase (C3′H, p ‐coumaroyl shikimate 3′‐hydroxylase), and whether the ester moiety can be subsequently released by HCT or caffeoyl shikimate esterase (CSE), this could provide another possible explanation for the altered flux through the metabolic grid that is monolignol biosynthesis. In addition to compositional changes to lignin, we observed a reduced degree of polymerization (DP) in the lignin fraction of transgenic trees compared to WT controls. Reduced DP has been observed in lignin exhibiting higher amount of H units and has been proposed as an alternative strategy to improve saccharification efficiency (Ziebell et al ., 2010 ; Eudes et al ., 2012 ; Mottiar et al ., 2016 ). Altogether, the reductions in lignin content and DP likely played significant roles in improving saccharification yields by up to 40% for glucose and up to 33% for xylose. Perspectives on lignin engineering Perturbations of the phenylpropanoid pathway in poplar in previous studies have often resulted in a reduction of total lignin accompanied by compensatory increases in cell wall polysaccharides, typically cellulose (Li et al ., 2003 ; Leplé et al ., 2007 ; Coleman et al ., 2008b ; Bjurhager et al ., 2010 ). Elevated levels of cellulose in low‐lignin plants are generally assumed to be the result of increased carbon availability for polysaccharide biosynthesis. However, in QsuB ‐poplar we observed no increase in glucose (i.e. cellulose), despite achieving comparable reductions in lignin content. QsuB ‐expressing lines displayed more xylose (i.e. xylan) in cell walls compared to WT. Increases in hemicellulose content and changes in hemicellulose composition have been reported in other low‐lignin poplar trees (Coleman et al ., 2008a ; Van Acker et al ., 2014 ), as well as in aspen mutants with reduced cellulose (Joshi et al ., 2011 ), all of which displayed an irregular xylem phenotype similar to QsuB ‐poplar. Cell‐wall‐bound DHB evidently occurs in the lignin of QsuB ‐poplar in two forms: as ester‐linked pendent groups, and as backbone‐integrated units (Fig. S9 ). As mentioned earlier, the latter form introduces ester linkages directly into the backbone of lignin polymers to form zip‐lignin structures. These alkali‐labile linkages readily translate into improved biomass processability by rendering the lignin more amenable to chemical deconstruction. However, the pendent DHB groups also offer industrial advantages as these ester‐linked moieties can be easily cleaved during processing. Once separated, these clip‐off phenolics could be used as platform compounds in chemical or microbial upgrading to produce an array of high‐value biochemicals. In this way, lignin engineering strategies such as that exemplified by QsuB ‐poplar can provide both reduced recalcitrance and value‐added lignin. This study adds to a growing body of evidence which shows that any compounds that are compatible with radical coupling and that are present in the cell wall during lignin deposition can become incorporated into lignin (Ralph et al ., 2008b ). In the case of the QsuB ‐poplar, the plants are clearly capable of producing and transporting DHB esters (and the various glycosylated conjugates as well) to the cell wall for polymerization. These new monolignol–DHB conjugates offer another example of how engineering bioenergy crops can not only improve the efficiency of industrial biomass processing but also potentially increase the value of lignin as DHBA itself could be a valuable coproduct in future biorefineries."
} | 4,943 |
30979882 | PMC6461649 | pmc | 858 | {
"abstract": "On coral reefs, microorganisms are essential for recycling nutrients to primary producers through the remineralization of benthic-derived organic matter. Diel investigations of reef processes are required to holistically understand the functional roles of microbial players in these ecosystems. Here we report a metagenomic analysis characterizing microbial communities in the water column overlying 16 remote forereef sites over a diel cycle. Our results show that microbial community composition is more dissimilar between day and night samples collected from the same site than between day or night samples collected across geographically distant reefs. Diel community differentiation is largely driven by the flux of Psychrobacter sp., which is two-orders of magnitude more abundant during the day. Nighttime communities are enriched with species of Roseobacter , Halomonas , and Alteromonas encoding a greater variety of pathways for carbohydrate catabolism, further illustrating temporal patterns of energetic provisioning between different marine microbes. Dynamic diel fluctuations of microbial populations could also support the efficient trophic transfer of energy posited in coral reef food webs.",
"introduction": "Introduction Microorganisms comprise the majority of biomass in the oceans and their role in the decomposition of organic substrates is critical to nutrient cycling, as well as for channeling nutrients and energy to higher trophic levels 1 – 3 . The water column overlying tropical coral reefs comprises a complex mixture of oligotrophic offshore waters and reef water enriched with the organic carbon and nitrogen substrates exuded by the benthic community 4 , 5 , which establishes the base of a robust microbial food web 6 , 7 . Coral reef benthic communities can directly consume bacterioplankton from the overlying water column via suspension feeding 8 – 11 thereby reducing energy loss that would otherwise be respired by microbes. These top down consumption processes reduce microbial production in the water column and simultaneously enhance transfer of microbial biomass to metazoan consumers in the benthos. The mechanisms sustaining these retention processes, yet poorly understood, promote the tight recycling of materials and production of high consumer biomass observed in coral reef ecosystems 12 . A key question in reef ecology is identifying how benthic communities influence reef microbial community structure and function. Microbial biomass and community structure on reefs are linked to local conditions, such as the composition of benthic assemblages 13 , 14 and allochthonous inputs 15 , 16 . As reefs shift toward algal-dominated states microbial production increases and becomes a greater energetic sink compared to that observed in a coral dominated system 17 , 18 . Diel investigations on reefs are required to better understand the heterotrophic metabolisms that dominate at night, including fundamental processes influenced by microbial communities, such as reef dissolution and boundary layer anoxia 19 – 21 . Furthermore, the influence of rhythmic growth patterns versus predation on reef microbial community structure remains virtually unknown. The majority of studies on coral reefs have been conducted during the day when diurnal processes predominately associated with corals and benthic algae exhibit highest rates of primary production and calcification. The hazards of apex predator feeding behaviors and of navigating boats through the reef during the night complicate the logistics of acquiring field diel measurements and nocturnal sample collections outside of controlled but artificial environments such as aquaria. Despite these challenges, few studies have shown that the dark reef (i.e., the matrix of crevices and caves) is a hotspot for microorganisms 22 and meiofauna (e.g., amphipods and other tiny invertebrates) that are active during the night 23 . Autochthonous copepods have been observed moving into the water column at night to feast on the planktonic communities 24 , 25 . Remote cameras and hydrophones have also recorded nocturnal migrations of invertebrates from the dark reef to the benthic surfaces 23 . To further distinguish trophic linkages between macroorganisms and microorganisms, and provide a more holistic understanding of the structure and function of coral reef communities, measurements of microbial dynamics in natural reefs over a complete diel cycle are needed. The study described here used a novel apparatus to collect diel biochemical and metagenomic samples from remote coral reefs to characterize microbial community dynamics in the water column overlying 16 forereef plots over a 24 h period. Metagenomic characterizations of environmental microbes can reveal linkages between spatial and temporal community dynamics, biogeochemical fluxes, and ecological niche partitioning (e.g. refs. 26 – 28 ). Our results illustrate a dramatic and consistent shift in day versus night microbial communities, which is reflected both in the taxonomic structure and the metabolic capacity encoded by the populations. This study highlights ecosystem functions on reefs that support dynamic fluctuations of diel microbial populations, capturing a key aspect of microbial ecology implicated in promoting trophic transfer of energetic resources through the microbial food web in tropical reef ecosystems.",
"discussion": "Discussion Marine microbes generally demonstrate similar community profiles during the day and night. Previous studies from the North Pacific Subtropical Gyre 36 , the English Channel 37 , and Monterey Bay 38 all report marginal changes in community structure and gene content between day and night communities. While transcriptional activity of metabolic functions in marine bacterioplankton communities are strongly influenced by diel patterns 36 , 38 , 39 , significant shifts in composition are more commonly linked to seasonal changes 37 , 40 , 41 , spatial patterns, such as distance to land 42 and oceanographic processes, such as upwelling 43 and mesoscale eddies 44 . In contrast to open ocean bacterioplankton, our results indicate that the microbial populations inhabiting coral reef waters are subjected to strong diel shifts in relative abundance. For instance, one species of Gammaproteobacteria, most closely related to the Genus Psychrobacter , dominated the day community on all reef sites (up to 70% on Vostok), but was depleted to an average of 0.5% at night (Fig. 2c, d ). The high prevalence of Psychrobacter spp. (Moraxellaceae Family) on these reefs was striking but not unexpected. For comparison, reef microbes collected between 2009 and 2016 from 22 Pacific islands were investigated for the presence of Psychrobacter . The abundances of Psychrobacter spp. were higher on the reef compared to surface and offshore waters (Supplementary Fig. 4a ) and varied widely across reefs ranging from <1% to 40% (Supplementary Fig. 4b ). The genus Psychrobacter has been characterized as aerobic, non-motile, cold-adapted, and mesophilic 45 , 46 . Designating the most abundant taxon inhabiting coral reef waters as psychrophilic was surprising. Hence, this discovery that a closely related population of Psychrobacter spp. (Supplementary Fig. 3 ) can dominate tropical marine microbial communities warrants further investigation to better understand the ecology of this clade. There were also a number of microbial taxa that were common during the night across all islands including one phylotype from the family Rhodobacteraceae (Alphaproteobacteria) that on average represented 17% of the night population and up to 33% on Starbuck. Gammaproteobacteria from the Genus Halomonas (Oceanospirillales) and Alteromonas also represented a significant proportion of the night community (up to 40% at sites on Starbuck and Malden and >50% at sites on Vostok and Malden, respectively). The difference in diel patterns between open ocean communities and those associated with coral reef habitats reflects the potential for a benthic–pelagic coupling of the microbial food web. The metabolic profile during the day was strongly influenced by the high prevalence of Psychrobacter spp. The metabolic profile of genes encoded during the day were predominately anabolic pathways including the biosynthesis of cofactors, vitamins, cell walls, and membranes (lipids), as well as DNA replication and repair. Thus, day community metabolism reflected a strategy for cellular growth, but a limited capacity to catabolize a variety of substrates. Nighttime communities showed a greater diversity of genes encoding for catabolic functions including a higher abundance of fermentative pathways for energy acquisition. Key pathways in central carbohydrate metabolism also differed between the day and night communities. The more efficient Embden–Meyerhof–Parnas (EMP pathway) was enriched in day communities whereas genes encoding the Entner–Doudoroff (ED pathway) were significantly more abundant at night, suggesting alternative strategies for catabolism at night 18 . Enriched carbohydrate pathways encoded by the night community also reflect greater availability of sugar alcohols and cycling of C1–C3 compounds that could be metabolized more favorably in low oxygen conditions 47 . Prior research has suggested that temporal cascades of transcriptional activity by different oceanic heterotrophs reflect a mechanism for partitioning energetic resources between members of the microbial community 39 , 48 , 49 . These organisms potentially respond to rhythmic metabolic fluctuations through molecular regulation that result in oscillatory patterns or circadian clocks 50 . Coral reef habitats create strong temporal fluxes of biochemicals including oxygen, pH, labile-dissolved organic matter, and inorganic nutrients (e.g. Fig. 1c, d ) that could provide essential cues for local biota to evolve oscillatory mechanisms to maximize utilization of resources over a diel cycle. Coral reef communities appear to maximize efficient processes related to nutrient cycling; dynamic growth and removal processes foster high production in these otherwise resource-limited environments 7 , 42 , 51 – 53 . Temporal synchrony of certain members of the microbial community to exhibit high metabolic rates during the day that correlate with primary productivity may represent another example of such mechanisms. These fast-growing diurnal microbes are diminished at night by (1) reduced growth rates, (2) protist predation that could channel energy into the benthic food web, or (3) viral lysis coordinated to remove a large proportion of the bacterial biomass, providing both reduced activity and available energetic substrates to different members of the community. The influence of predation versus rhythmic growth patterns on community structure are fundamental questions in microbial ecology that remain largely unanswered. The use of commercial autonomous samplers could provide the capacity to sample different size fractions (i.e., particle associated versus free-living cells), at finer temporal resolution (e.g., hourly) and over consecutive diel cycles. This enhanced time resolution would also provide a means to further characterize dynamics of microbial populations and help resolve the roles of predators, both phage and protists, in community shifts on diel time scales. While it is less logistically feasible to deploy larger sampling equipment on cruises, land-based field studies both on populated islands and remote regions (e.g., Palmyra; the northwest Hawaiian Islands) would provide greater geographic context to these observed diel patterns and allow for comparison across intact and degraded coral reef ecosystems. We hypothesize that the ecological functions driving these microbial fluxes will be diminished as habitats become influenced by anthropogenic perturbations. Further studies are required to better describe the retention of microbes from the water column into the benthic habitat, how this capacity serves to influence production and respiration processes in coral reef ecosystems, and to what degree this functionality is lost as habitats degrade."
} | 3,044 |
37177655 | PMC10181665 | pmc | 859 | {
"abstract": "Conventional processing of sensory input often relies on uniform sampling leading to redundant information and unnecessary resource consumption throughout the entire processing pipeline. Neuromorphic computing challenges these conventions by mimicking biology and employing distributed event-based hardware. Based on the task of lateral auditory sound source localization (SSL), we propose a generic approach to map biologically inspired neural networks to neuromorphic hardware. First, we model the neural mechanisms of SSL based on the interaural level difference (ILD). Afterward, we identify generic computational motifs within the model and transform them into spike-based components. A hardware-specific step then implements them on neuromorphic hardware. We exemplify our approach by mapping the neural SSL model onto two platforms, namely the IBM TrueNorth Neurosynaptic System and SpiNNaker. Both implementations have been tested on synthetic and real-world data in terms of neural tunings and readout characteristics. For synthetic stimuli, both implementations provide a perfect readout ( 100 % accuracy). Preliminary real-world experiments yield accuracies of 78 % (TrueNorth) and 13 % (SpiNNaker), RMSEs of 41 ∘ and 39 ∘ , and MAEs of 18 ∘ and 29 ∘ , respectively. Overall, the proposed mapping approach allows for the successful implementation of the same SSL model on two different neuromorphic architectures paving the way toward more hardware-independent neural SSL.",
"conclusion": "5. Conclusions In the present study, we outlined a generic approach for mapping rate-based neural architectures to spike-based neuromorphic hardware applied to the case of auditory sound source localization (SSL). We investigated the functional equivalence of the mapping for two neuromorphic architectures, namely TrueNorth [ 5 ] and SpiNNaker [ 29 ]. The neural SSL model consisted of neurons smoothing gammatone-filtered input across frequency bands and time per hemisphere (S-T units), neurons differentially weighting input from S-T units of both hemispheres (W-S units) to compute ILD values, and a linear readout population with weights to W-S units that have been fit via regression to perform azimuthal SSL. Neural models of SSL offer a way to study the complete range of functionality (adaptation to input statistics, top-down context-adaptivity, and learning) required to harness neuromorphic SSL computation for everyday real-world applications and usage in neuromorphic robotics. Thus, being able to bring one and the same SSL model to different neuromorphic platforms will help to pursue these goals further. To this end, we presented the first on-chip simulation results for the SSL model on both neuromorphic platforms. To demonstrate the feasibility of our approach, we selected two neuromorphic architectures, TrueNorth and SpiNNaker. Not only do further architectures, such as Loihi [ 43 ] or BrainScaleS-2 [ 44 ] exist, the development in the field of neuromorphic computing is fast-paced so that new architectures, such as Loihi 2 [ 45 ] or SpiNNaker 2 [ 46 ], are expected to find broader research usage soon and further architectures will keep being developed. By having investigated implementations on TrueNorth and SpiNNaker, we expect that our procedure would also work considerably well for bringing SSL to other architectures as the frameworks and capabilities offered by TrueNorth and SpiNNaker are diametrically different—TrueNorth provides a specific neuron type and specific connectivity capabilities being optimized for high energy efficiency, while SpiNNaker offers broad flexibility for implementational decisions but offers fewer neurons and energy efficiency. Fittingly, the two have been termed “[...] good examples of the extremes one can take with digital hardware implementations”. [ 8 ], which makes them likewise good examples for the exemplification of bringing neural SSL to neuromorphic hardware by means of a generic mapping approach.",
"introduction": "1. Introduction 1.1. Motivation Audition plays an important role in survival as it helps us to sense our environment omnidirectionally and localize sound sources precisely. Such localization is not trivial since perceived sound signals do not explicitly convey localization cues. Instead, these cues need to be computed from differences in the incoming sounds of the left and right ear. In the case of lateral localization, two binaural cues are computed: the interaural time difference (ITD) and the interaural level difference (ILD) [ 1 ]. As the name implies, the ITD is based on the difference in the arrival time of sound signals on one ear compared to the other ear, and its encoding in mammals is based on a sophisticated inhibition circuit [ 2 ]. The interaural level difference (ILD) is the other cue for lateral sound source localization (SSL). ILD cues are primarily used for high-frequency sounds (>1500 Hz [ 3 ]) since high frequencies are attenuated by the head and, therefore, create a difference in the intensity levels between ipsi- and contralateral side ( Figure 1 A). The encoding of these cues takes place in the lateral superior olive (LSO) [ 4 ]. By a weighted combination of excitatory ipsilateral and inhibitory contralateral inputs single LSO neurons exhibit tuning curves that relate their sensory stream inputs monotonously to ILD values (see Figure 1 B for a typical tuning curve). Beyond providing an understanding of brain function, models of ITD and ILD computation can be used for real-world applications of biologically inspired SSL. For such applications, biological inspiration can not only be taken on an algorithmic level. Moreover, the employed hardware can be inspired by principles from biology in terms of neuromorphic computing paradigms. Neuromorphic hardware platforms provide specialized computer architectures that mirror the structure and function of neurons and their interaction in networks [ 5 , 6 , 7 ]. Such hardware comes with benefits in terms of real-time capabilities and energy efficiency. These are achieved by adhering to event-based processing principles instead of sampling and processing sensory information temporally uniformly. So far, there is no generally agreed-upon neuromorphic hardware architecture, but many different kinds exist, each with its own strengths and weaknesses, and use cases [ 6 , 8 , 9 ]. Likewise, improving the energy efficiency of neuromorphic systems, and hence developing new hardware, is still an ongoing effort [ 10 , 11 , 12 ]. As an implication, also software frameworks and available components to implement neuron models vary broadly. Developing even one neural network for one architecture can, thus, already be an endeavor. While some approaches of SSL have been brought to neuromorphic platforms already, efforts to allow for utilizing the same SSL model on multiple different such platforms are still lacking. Here, we propose a mapping procedure to allow for using the same underlying biologically inspired SSL model across different neuromorphic platforms. Following the mapping, procedure we implement the same SSL model on two different neuromorphic platforms and compare the results between both implementations. Figure 1 Overview of relevant concepts for neuromorphic sound source localization. ( A ) Principle of interaural level difference (ILD) computation. High-frequency sounds coming from the ipsilateral side are attenuated by the head and, therefore, have a smaller amplitude at the contralateral ear. This difference in amplitude, or sound level, between the ipsi- and contralateral ear varies systematically with the azimuth angle α and is encoded in the interaural level difference in lateral superior olive (LSO) neurons. ( B ) Typical LSO neuron response over different ipsi- and contralateral sound levels. Such a response curve varies for different LSO neurons, i.e., it shifts along the x-axis and the steepness varies. ( C ) Comparison of hardware specifications for the neuromorphic chips used in our approach (comparison based on [ 13 ]). 1.2. Related Works Many proposals of biologically inspired SSL exist already, yet only a few of them have been brought to neuromorphic hardware ( Table 1 ). While we focus on these neuromorphic biologically inspired SSL approaches here, broader overviews concerning SSL in general or with applications to robotics can be found in Desai and Mehendale [ 14 ] and Rascon and Meza [ 15 ], respectively. Studying biologically inspired SSL in neuromorphic hardware, so far, is mostly concerned with processing ITD cues and implementations tailored toward specific hardware. Moreover, testing of the systems varies widely in terms of input data. A very early implementation of Lazzaro and Mead [ 16 ] built an analog integrated circuit (IC) and tested the linear relationship between neuron responses and input ITD with simple click stimuli fed to the board. Glackin et al. [ 17 ] developed their spiking neural network based on a model of the medial superior olive (MSO) at first on a conventional computer and then brought it to field programmable gate array (FPGA) based hardware to accelerate the system’s performance. The model has been trained using spike-timing-dependent plasticity (STDP) based on ear canal recordings of a domestic cat acquired in a sound-dampened chamber. Similarly, Xu et al. [ 18 ] utilized a correlational approach based on ITD-related onset-timing features extracted from an auditory filter device mimicking cochlear processing. Using sound samples from speakers in a reverberant environment the extracted features have been used in either a regression or extreme learning machine (ELM) approach for SSL. Escudero et al. [ 19 ] developed an FPGA-based sound tracking system, which turned a robotic platform towards the sound source. Notably, the system relied on ILD cues. It was tested with pure tones recorded in a classroom and ILD cues have been used in a control loop after an initial calibration phase per scenario. A recent investigation by Schoepe et al. [ 20 ] combined an FPGA and a SpiNNaker board to establish a closed-loop robotic control platform. The system capacity was tested in a real-world setting with pure tones and human speech played back from a loudspeaker. Additional components of time-to-rate, ring attractor, and center detector networks transform ITD-based encodings from a time-difference encoder (TDE) to motor commands. The TDE computation itself was not conducted on SpiNNaker, but on the FPGA board to meet the required timing constraints of the ILD-based approach. Oess et al. [ 21 ] first reported a model of ILD-based SSL on the TrueNorth NeuroSynaptic System, which we subsume and extend in our current communication. The system was tested on synthetic data and recordings of natural sounds played back from a speaker in a sound-dampened chamber. Despite these approaches on either generic (ICs, FPGAs) or specific neuromorphic (SpiNNaker, TrueNorth) hardware, lately, also novel biologically inspired resistive random access memory (RRAM) approaches have been proposed. So far, these provided proof-of-concept results based on ITD cues and controlled laboratory setups [ 22 , 23 ]. Taken together, biologically inspired SSL approaches on neuromorphic hardware, so far, focused on either specific hardware platforms or rather freely definable FPGA implementations. Notably, most of the approaches followed the route of ITD-based SSL and tests were mostly restricted to laboratory or indoor settings and synthetic or played-back sounds. 1.3. Summary Here, we propose a generic approach for how to map biologically inspired neural networks to different neuromorphic platforms and, subsequently, apply it to a binaural SSL model, which encodes ILD values for the horizontal plane. The employed neural architecture yields a mechanistic explanation of the LSO and, hence, gives rise to a biologically inspired algorithm of SSL. It is based on single-compartment conductance-based neuron models, which serve as computational units for signal integration and generation of output responses. The activations of single units in the model represent the average membrane potential of a group of neurons calculated as the sum of excitatory, inhibitory, and leak conductances. Their dynamics are described by gradual activation dynamics of a neuronal population [ 24 ]. These dynamics can be simulated by numerically integrating systems of ordinary differential equations [ 25 ]. However, such an approach is neither very cost- nor time-efficient to compute. To circumvent these computational issues and to account for the biological principles of neural information processing in real neurons, we adopt an event-based, i.e., spiking, representation of responses. Responses are generated asynchronously based on an integrate-and-fire principle [ 26 ]. To make full use of the event-based model paradigm, we implement the model on neuromorphic platforms. Importantly, since no generally agreed-upon platform exists yet, we here propose a generic approach for mapping such models to different platforms coming from an original rate-based model description. As a first step of the mapping, we identify generic computational motifs within the original rate-based model. During a second step, a transformation of this generic component onto patterns of spiking response characteristics in neuromorphic hardware is performed. While the first two steps are platform independent, we suggest a consecutive step that operationalizes the mapping onto specific brain-inspired hardware platforms. We exemplify this mapping based on two neuromorphic platforms, namely the TrueNorth Neurosynaptic System [ 5 , 27 ] and SpiNNaker [ 28 ] ( Figure 1 C). Both systems are designed to utilize digital computing technology relying on a CMOS hardware process. TrueNorth has been developed by IBM as part of DARPA’s SyNAPSE project. The architecture is functionally organized as a network of neurosynaptic cores, each one defining a canonical cortical microcircuit [ 5 ]. TrueNorth’s computation is extremely energy efficient, consuming 70 mW of power in operation. This is achieved by asynchronous event-driven neuron activation, hardware implementation of specific neuron models, and an on-chip network realization that interconnects all neurosynaptic cores avoiding an off-chip memory component. A single chip contains one million neurons with 256 million synapses. The system is delivered as 1-, 4-, and 16-chip board architecture [ 13 , 27 ]. The SpiNNaker network architecture utilizes a general-purpose parallel computing system that can run neurons of different specifications in software. It has been developed at Manchester University within the EU Flagship Human Brain Project (HBP). SpiNNaker is a many-core architecture utilizing small integer cores and incorporating a communication framework that is optimized to send large numbers of very small data packages (conveying neural spikes) to many destinations following a multi-cast principle [ 6 , 29 ]. Delivering spike activations to arbitrary receiver neurons is based upon a packet-switched Address Event Representation (AER). A SpiNNaker node contains 18 ARM processor cores and consumes 1 W power for a fully loaded 18-core package. The system is delivered in two circuit board configurations, namely a 4-node (72-core) and a 48-node (864-core) board. We present on-chip model simulations for the defined neural network mechanisms for ILD-based auditory SSL computations of synthetic and naturalistic sounds. The model has been successfully implemented on the different hardware designs of TrueNorth and the SpiNNaker neuromorphic architectures, respectively, based on the proposed generic mapping approach. We demonstrate similar performance in the response characteristics of the resulting neuromorphic algorithms implemented on the different platforms. These model implementations realize model circuits towards a real-time and energy-efficient SSL for real-world applications.",
"discussion": "4. Discussion In this paper, we propose a generic mapping approach to implement neural mechanisms for sound source localization (SSL) in the horizontal plane on multiple neuromorphic platforms. The SSL model utilizes the differences in sound level of the source signal at the two ears, or microphones, corresponding to the interaural level difference (ILD) principle. The main aim of this communication article is to propose a principled approach to map the underlying neural mechanism onto a target neuromorphic platform for the neuromorphic implementation of the core functionality. We exemplify the mapping procedure from two different platforms, namely IBM’s TrueNorth neurosynaptic chip (developed in the DARPA Synapse program; [ 5 , 27 ]) and the SpiNNaker platform (developed in the EU Flagship Human Brain Project, HBP; [ 28 ]). We emphasized that the first step in this mapping cascade is intrinsically related to the neural model, while the last part is specifically designed to adopt the properties of the neuromorphic target architecture. During the first mapping step, connectivity motifs are identified and neurons of the original model are characterized in terms of a generic neuron model [ 30 ]. Next, the connectivity motifs and neurons are mapped onto neuromorphic-compatible simplified connection patterns and spike-based neurons [ 35 ]. After these generic steps, a final step specific to the target platform takes place, during which connections and neurons are formalized within the target neuromorphic framework and parameters are fine-tuned. Using the mapping procedure to implement the neural SSL model on both platforms, TrueNorth and SpiNNaker, respectively, we have conducted two main experiments to demonstrate the functionality as a proof of concept. On synthetically generated input, we validated that the neuromorphic implementations of the SSL model for both architectures are well capable to encode ILD values, and hence azimuth location, across the complete range of inputs. The input–output behavior of the model was investigated in this experiment for a single frequency band. Computing the accuracy of the readout population resulted in a perfect score of 100 % for both implementations indicating that ILD information can be decoded linearly from the model’s W-S units. This validates the feasibility of the proposed mapping procedure and, in line with previous findings [ 21 ], the functionality of the SSL model. During a second experiment, we tested the real-world capabilities of the mapped models. Both models showed similar qualitative behavior under real-world input in providing azimuth estimates that follow the overall course of calculated ground truth values. Quantitatively, the performances in the real-world setting between the TrueNorth and SpiNNaker implementation differ for accuracy ( 78 % vs. 13 % ) and MAE ( 18 ∘ vs. 29 ∘ ), while being more similar in terms of RMSE ( 41 ∘ vs. 39 ∘ ). The discrepancy might be attributable to multiple sources of influence. While in the synthetic setting performances between both models were well aligned, differences only became apparent in the real-world experiment. For these input types, the SpiNNaker experimental protocol deviated slightly from the TrueNorth one, providing one possible reason. For the second architecture, i.e., SpiNNaker, the results of the second experiment are to be interpreted as a first proof-of-concept, as we restricted the mapping approach to the SSL model’s core ILD mechanism. As an implication, we deliberately chose to not perform most of the proposed additional pre- and postprocessing steps (cf. [ 21 ]). On a post-processing side, this means that we did not exclude any poorly performing frequency bands from forming the azimuth estimate. On a pre-processing side, this means, that no time-bin-wise cross-normalization of left and right input was performed. Thus, the quantitative results can be seen as lower bound performance estimates for the given experiments, which nevertheless capture the essentials of biologically inspired SSL. While quantitative performance measurements are better for the TrueNorth data in terms of accuracy, a visual comparison of mean azimuth estimates between the architectures and the ground truth estimates reveals similar qualitative traits. Both implementations follow the general trend and desired behavior of the ground truth’s ILD response curve. We are confident, that applying such pre- and postprocessing techniques likewise to the SpiNNaker model will lower or even close the quantitative gap. As the basic underlying SSL mechanism is well captured in both architectures, these quantitative discrepancies are rather suggestive of an avenue towards future development directions, than revealing shortcomings of the proposed mapping. Namely, to overcome the current dependency of the employed SSL model on pre- and post-processing of the data i.e., cross-normalization of input ranges, and selective admission of frequency bands into the ILD estimate formation). These processing steps strongly reduce confounders to the ILD estimate stemming from differences in overall loudness (cross-normalization step) and variability of frequency bands dependent on sound source type (frequency admission step). A remedy to these situation-dependent effects lies in making the model adaptive to context. Such context-adaptivity lends itself perfectly for an all-neuronal implementation within the already utilized canonical neuron model. For example, eliminating absolute level information while retaining contrasting one between computational elements (as required for the cross-normalization step) has been shown in the context of biologically plausible visual information processing of oriented contrast information [ 38 ]. Likewise, context-dependent selection of information (as required for the frequency admission step) is a prime example of top-down attention in the context of visual object detection, where the visual system needs to selectively filter its input for features matching those of a higher-level aggregate representation of the object [ 39 ]. Indeed, both functionalities have been studied previously using the canonical neuron model that comprises as well the mapped SSL model [ 30 , 40 ]. In such context-adaptability lies the advantage of using neuronal models over conventional approaches. By way of the dynamic computation of the neurons employed, they can be naturally extended to display such adaptive behavior (cf. [ 31 ] for the case of history-dependent adaptation of ILD selectivity). Currently, the neuromorphic model implementations occupy only a fraction of the overall available compute resources, for example, 16 out of 864 cores for SpiNNaker). Thus, the implementation would offer enough space for further model components and adaptations, even if the additional resource requirements for input handling and monitoring are taken into account. As part of this communications article, we aimed to exemplify the proposed mapping procedure and compared the SSL model implementations on TrueNorth and SpiNNaker. To make for a proper comparison, the same experiments were performed with the new SpiNNaker implementation as previously reported for TrueNorth [ 21 ]. This set of experiments served our purpose well by providing a common ground to compare both implementations. However, it has several limitations when it comes to capturing the variability of the real world. The synthetic data were restricted to a single frequency band. On the one hand, this allowed for precise control and investigation of the neural tuning properties. On the other hand, it lacks the variability and bandwidth of real-world sounds. The real-world experiments were based on a dataset first reported by Oess et al. [ 21 ]. The data were acquired in a sound-dampened chamber and provided a single stationary sound source per recording from a limited set of sound types. Thus, to draw stronger conclusions about the performance of the neural SSL model itself, further experiments will be needed, on either platform, under more demanding conditions. For instance, synthetic data could be generated comprised of more than one frequency band and defined temporal variations. Likewise, real-world recordings under realistic conditions would further test the model. These could consist of multiple sound sources from different angles at once recorded in a reverberating environment. The model on which we exemplified the mapping adheres to principles of biological auditory processing of ILD. Thus, it will underlie uncertainties and ambiguities, such as the cone of confusion [ 41 ]. Incorporating further acoustic features, such as ITD or spectral cues, would be interesting extensions to the model. This way, model performance might be increased and extended to estimates along the elevation angle as well. Likewise, replacing the readout procedure (Equation ( 3 )) with a deep neural network could harness the model for technical applications and result in more precise location estimates [ 42 ]."
} | 6,274 |
34622046 | PMC8482434 | pmc | 863 | {
"abstract": "Vehicle routing problem is a widely researched combinatorial optimization problem. We developed a hybrid of three strategies—a modified ant system, a sweep algorithm, and a path relinking—for solving a capacitated vehicle routing optimization problem, a vehicle routing problem with a capacity constraint. A sweep algorithm was used to generate initial solutions and assign customers to vehicles, followed by a modified ant system to generate new generations of good solutions. Path relinking was used for building a better solution (candidate) from a pair of guiding and initial solutions. Finally, a local search method—swap, insert, reverse and greedy search operations—was used to prevent solutions from getting trapped in a local minimum. Performance of the proposed algorithm was evaluated on three datasets from CVRPLIB. Our proposed method was at least competitive to state-of-the-art algorithms in terms of the total route lengths. It even surpassed the best-known solution in the P-n55-k8 instance. Our findings can lower the transportation cost by reducing the travelling distance and efficiently utilizing the vehicle capacity.",
"conclusion": "5 Conclusions We developed and evaluated a hybrid MAS-SA-PR algorithm for solving capacitated vehicle routing problems. A sweep algorithm (SA) was used to generate initial solutions and assign customers to vehicles. A modified ant system (MAS) was used to construct offspring from parents of the current generation. Path relinking (PR) was used to further improve a solution by a guiding from the elite solution. The performance of our algorithm was compared with that of the individual MAS and three state-of-the-art algorithms. Our algorithm was very effective with small-sized and medium-sized datasets—the deviations from the best-known solutions were all less than 1.71%. Even though it did not perform any better or worse with large-sized datasets than the other state-of-the-art algorithms, for one benchmark instance, it generated a better solution than the best-known solution. However, the complexity of our new method caused its computation time to be quite long. We will attempt to reduce this complexity in our future work, so that it will be more efficient.",
"introduction": "1 Introduction Currently, online purchasing is a popular choice for many people around the world because it offers many kinds of products without having to spend time traveling to real stores. During the Covid-19 pandemic, online purchasing has been even more popular. Optimal vehicle routing is an important logistic factor that helps make online purchasing successful and efficient, leading to acceptance of online purchasing by former and new customers. In addition, optimal vehicle routing helps reduce oil (a limited energy resource) consumption and air pollution from vehicle exhausts. The Vehicle Routing Problem (VRP) is a combinatorial optimization problem; it aims to find an optimal route with the lowest cost for delivering goods to specified customers in various locations. Since a vehicle has a limited load capacity, a VRP solution needs to consider the load capacity as a constraint. A Capacitated Vehicle Routing Problem (CVRP), a similar problem, finds sets of routes to be traversed by a fleet of vehicles. The vehicles start from a depot, deliver goods to customers, and then return to the depot. A customer is served by only one vehicle. Each vehicle can serve several customers when their demands do not exceed the vehicle capacity [ 1 ]. Optimization algorithms can be classified into algorithms that give an exact or an approximate solution. An exact algorithm is guaranteed to find the best solution. However, generally it requires a large amount of computational time to get the solution. Since CVRP is an NP-hard problem, in which finding the optimal solution is very hard and time-consuming, algorithms that provide an approximate solution are more practical and commonly used to solve large CVRP problems [ 2 ]. Algorithms that provide an approximate solution are of 2 groups: approximation and meta-heuristic algorithms. Approximation algorithms have theoretical performance guarantees for the solution and the computational time [ 3 ]. Haimovich and Rinnooy Kan [ 4 ] introduced the first Polynomial-Time Approximation Scheme (PTAS) for the two-dimensional Euclidean CVRP with time complexity of O(log log n) . Then Asano et al. [ 5 ] improved the result to q = O(log n/log log n) . More recently, Das and Mathieu [ 6 ] introduced the first Quasi-Polynomial-Time Approximation Scheme (QPTAS) for the two-dimensional Euclidean CVRP. Their scheme gave a (1 + ε)-approximation for the two-dimensional Euclidean CVRP in time n ( log n ) O ( 1 / ε ) . However, their approximation scheme is no longer a QPTAS in general metric space of a fixed doubling dimension. Khachay et al. [ 7 ] extended the QPTAS proposed by Das and Mathieu [ 6 ] to the case of metric spaces of a fixed doubling dimension. They replaced the exhaustive search in the original QPTAS with the novel internal dynamic program to ensure that the resulting approximation scheme became QPTAS for an arbitrary fixed doubling dimension d > 1. Meta-heuristic algorithms commonly used to solve CVRP can be further divided into an individual search algorithm and a population-based algorithm. The individual search algorithms generate one solution at a time and improves it until it becomes the best solution; the population-based algorithms generate many solutions (a swarm of solutions) at a time, then improve them, and select the best solution among the improved solutions. Examples of algorithms of the former type are Simulated Annealing, Tabu Search (TS) and Greedy Randomized Adaptive Search Procedure (GRASP), while algorithms of the latter type are Ant Colony Optimization (ACO), Firefly Algorithm (FA), Genetic Algorithm (GA), and Particle Swarm Optimization (PSO). Ezugwu and Adewumi [ 8 ] presented a Discrete Symbiotic Organism Search (DSOS) algorithm for solving the Travelling Salesman Problem (TSP). The original Symbiotic Organism Search (SOS) algorithm was inspired by symbiotic relationships among organisms in an ecosystem: mutualism, commensalism, and parasitism. It has been very effective in solving numerical optimization problems. DSOS is an extended version of SOS for solving combinatorial optimization problems. Three mutations were used in DSOS: a swap mutation operator, that randomly picked two placeholders and swapped the two cities at those placeholders; an inversion mutation operator, that randomly picked two placeholders and reversed the order of the cities to be visited in those two places; and an insertion mutation operator, that randomly picked two placeholders, i and j , and inserted the city in placeholder i into placeholder j . Zhong et al. [ 9 ] described a hybrid method for solving the TSP based on an artificial bee colony (ABC) algorithm and a threshold acceptance criterion. This method converted a continuous ABC to a discrete ABC, in part, by redesigning the solution updating equation, that depended on the information from other bees and the problem at hand. The threshold acceptance criterion was used in place of a greedy acceptance strategy to decide whether to accept a new solution. Liao and Liu [ 10 ] described a hierarchical algorithm that decomposed a Traveling Salesman Problem (TSP) into subproblems by Density Peaks Clustering algorithm, which was used for clustering a node and defining the center of each cluster. Ant Colony Optimization Algorithm was used to find the optimal route for each subproblem, and then connect all center nodes of all subproblems together. Finally, the global route was optimized by k-opt algorithm. Osaba et al. [ 11 ] presented a Discrete Water Cycle Algorithm (DWCA), inspired by the water cycle process, i.e. how streams flow into rivers and the sea. DWCA is a discrete version of the Water Cycle Algorithm (WCA). Authors used DWCA to solve the Symmetric Traveling Salesman Problem (STSP) and Asymmetric Traveling Salesman Problem (ATSP). In DWCA, Hamming distance was used to determine the distance between the streams and their corresponding rivers or sea. The distance influenced the movement of streams. Streams had two movement methods: insertion for slow moving and 2-opt for fast moving. The evaporation condition and raining process were used to prevent overly fast convergence and becoming stuck in a local optimum. Marinaki and Marinakis [ 12 ] showed a hybrid algorithm for solving vehicle routing problems with stochastic demands; it merged Combinatorial Neighborhood Topology Glowworm Swarm Optimization (CNTGSO) with Variable Neighborhood Search (VNS) and Path Relinking (PR). Two types of vectors were used in this algorithm: x i , with continuous values for calculation of movement of glowworms by GSO, and y i , with discrete values for representation of routes. A local search technique based on VNS was applied to each glowworm to enhance the exploitation capabilities. Goel and Maini [ 13 ] presented a hybrid of Ant Colony System (ACS) and Firefly Algorithm (FA) for solving Capacitated Vehicle Routing Problem (CVRP). In this hybrid algorithm, a discrete version of the ACS served as the basic framework, while the FA was used to explore the search space for new solutions. Since ACS has a drawback of premature convergence, pheromone shaking process was used to prevent the algorithm from getting trapped at local optima. Pop et al. [ 14 ] presented a novel two-level optimization approach for solving a clustered vehicle routing problem (CluVRP), in which customers were grouped into a given number of clusters. The two-level optimization decomposed CluVRP into two sub-problems: upper-level and lower-level ones. The upper-level sub-problem created a global route, using a genetic algorithm to connect all clusters. The lower-level sub-problem used the Concorde TSP solver, after transforming the global route into a TSP, to determine the visiting order within a cluster. Yousefikhoshbakht and Sedighpour [ 15 ] proposed a hybrid of the sweep algorithm and the elite ant colony optimization for solving the multiple traveling salesman problem (MTSP). First, initial routes were created using the sweep algorithm. Then the elite ant colony optimization and 3-opt local search were used to further improve the routes. Chen et al. [ 16 ] used a hybrid of two-stage sweep algorithm and greedy search, for solving CVRP. Sweep algorithm was used to cluster all customers into groups. Then, it re-clustered the customers, adding customers from adjacent clusters, based on vehicle capacity constraints. These steps helped avoiding local optima. Lastly, a greedy search was used to determine the shortest route for each vehicle. Osaba et al. [ 17 ] designed a Discrete Firefly Algorithm (DFA) for solving a newspaper distribution system problem, with a recycling policy, i.e., a Rich Vehicle Routing Problem (R-VRP). Their R-VRP took many constraints into account, for instance, asymmetry, simultaneous pickup and delivery, variable traveling times, and forbidden paths. Since the original FA was for continuous data, their DFA was designed to deal with discrete data. The Hamming distance was used to determine the distance between two different fireflies, and an insertion function was used to move fireflies from one place to another. Hannan et al. [ 18 ] presented a modified particle swarm optimization (PSO) algorithm for finding an optimized route for waste collection. The idea was that finding an optimized route could be assisted by real-time data, from smart bin sensors, to determine which bin was full, and the waste in it had to be collected. Sub-routes were constructed from neighboring locations of smart bins according to vehicle capacity and threshold waste level. Four local search algorithms—2-Opt∗, Or-Opt-1, 2-Opt, and Or-Opt—were used to improve routes. Altabeeb et al. [ 19 ] presented a hybrid firefly algorithm for a capacitated vehicle routing problem (CVRP-FA) that integrated 2 local search algorithms and genetic operators to FA. In CVRP-FA, initial fireflies were randomly generated under the capacity constraint. The Hamming distance was used to calculate the distance between two fireflies. Partially matched crossover was used to generate two new offspring. The better one was chosen as a new firefly. CVRP-FA applied two local search algorithms—improved 2-opt and 2-h-opt—to enhance the solution. Two types of mutation—swapping two customers within the same sub-route and swapping two customers in two different sub-routes—were used to prevent the premature convergence. Although many researchers have studied VRP, there remain many questions in this field that need to be addressed. Since VRP is an NP-hard problem, researchers are still attempting to find more efficient methods to solve it. This paper describes a hybrid algorithm that combines sweep algorithm for generating good initial solutions, Modified Ant System (MAS) for subsequent building of new generations of good solutions, and path relinking for finding a better solution from a pair of guiding and initial solutions. Finally, four local search methods were used to further improve the solutions. The paper is organized as follows: Section 2 describes several background concepts; Sections 3 and 4 discuss our methods and experimental results. Section 5 concludes the paper."
} | 3,357 |
29588660 | PMC5863372 | pmc | 865 | {
"abstract": "Background Lignin is a heterogeneous polymer representing a renewable source of aromatic and phenolic bio-derived products for the chemical industry. However, the inherent structural complexity and recalcitrance of lignin makes its conversion into valuable chemicals a challenge. Natural microbial communities produce biocatalysts derived from a large number of microorganisms, including those considered unculturable, which operate synergistically to perform a variety of bioconversion processes. Thus, metagenomic approaches are a powerful tool to reveal novel optimized metabolic pathways for lignin conversion and valorization. Results The lignin-degrading consortium (LigMet) was obtained from a sugarcane plantation soil sample. The LigMet taxonomical analyses (based on 16S rRNA) indicated prevalence of Proteobacteria , Actinobacteria and Firmicutes members, including the Alcaligenaceae and Micrococcaceae families, which were enriched in the LigMet compared to sugarcane soil. Analysis of global DNA sequencing revealed around 240,000 gene models, and 65 draft bacterial genomes were predicted. Along with depicting several peroxidases, dye-decolorizing peroxidases, laccases, carbohydrate esterases, and lignocellulosic auxiliary (redox) activities, the major pathways related to aromatic degradation were identified, including benzoate (or methylbenzoate) degradation to catechol (or methylcatechol), catechol ortho-cleavage, catechol meta-cleavage, and phthalate degradation. A novel Paenarthrobacter strain harboring eight gene clusters related to aromatic degradation was isolated from LigMet and was able to grow on lignin as major carbon source. Furthermore, a recombinant pathway for vanillin production was designed based on novel gene sequences coding for a feruloyl-CoA synthetase and an enoyl-CoA hydratase/aldolase retrieved from the metagenomic data set. Conclusion The enrichment protocol described in the present study was successful for a microbial consortium establishment towards the lignin and aromatic metabolism, providing pathways and enzyme sets for synthetic biology engineering approaches. This work represents a pioneering study on lignin conversion and valorization strategies based on metagenomics, revealing several novel lignin conversion enzymes, aromatic-degrading bacterial genomes, and a novel bacterial strain of potential biotechnological interest. The validation of a biosynthetic route for vanillin synthesis confirmed the applicability of the targeted metagenome discovery approach for lignin valorization strategies. Electronic supplementary material The online version of this article (10.1186/s13068-018-1073-4) contains supplementary material, which is available to authorized users.",
"conclusion": "Conclusions Different approaches were applied to validate the establishment of a lignin-degrading consortium and applicability of the metagenomic strategy. From the analysis of massive DNA sequencing data, several microorganisms and enzymes linked to lignin-degrading metabolic pathways were identified in LigMet. Furthermore, gene clusters involved in the non-beta-oxidative pathway for vanillin production were depicted, followed by production of recombinant FerA and FerB in functional form for successful production of the phenolic aldehyde. The genome sequencing analysis of a bacterium revealed eight gene clusters encoding proteins related to aromatic degradation, suggesting potential for biotechnological application. The LigMet metagenome represents a vast reservoir of genes coding for enzymes involved in lignin depolymerization and assimilation. The targeted metagenomic discovery platform described in the present study is of potential interest to reveal optimized gene sets and microorganisms for initiatives based on synthetic biology principles to produce high value chemicals from lignocellulose. The straightforward metagenomic strategy described could also be applied to other fields such as the development of antibiotic producing microorganisms or the recycling of plastic polymers.",
"discussion": "Results and discussion Establishment of lignin-degrading consortium In Brazil, sugarcane straw is usually returned to the field after harvest. In the present study, the hypothesis under investigation was whether the sugarcane field soil covered by straws, after the plants were harvested, could be a potential source of novel lignolytic enzymes, because it is expected that the microbial population involved in degradation of plant biomass polymers is enriched in this environment (Fig. 1 ). According to the previous studies, the establishment of microbial consortia is a powerful strategy to develop an enriched community with a particular metabolic ability, e.g., consortia for lignocellulosic biomass, phenanthrene, and bitumen degradation have been reported [ 21 – 23 ]. Thus, sugarcane field soil was used as inoculum in media containing minimal nutrients together with lignin fragments as the major carbon source. To establish a consortium able to degrade and modify lignin and/or lignin aromatic compounds, a sugarcane soil-derived consortium was adapted and enrichment on lignin medium for up to 50 consecutive weeks. As illustrated in Fig. 2 a, the liquid waste stream used for LigMet cultivation contains lignin fragments ranging from 300 to 1200 Da, including phenolic monomers such as ferulic and cinnamic acid (Additional file 1 : Table S1), along with minimal sugar concentrations (450 µM l −1 ). According to Fig. 2 b, it is possible to observe that compounds ranging from ~ 700 to ~ 300 Da decreased during the first 34 h of growth, indicating that the consortium was able to use the soluble lignin fragments as a carbon source. Moreover, during the first 40 h, an increase in OD 600 and consumption of sugars available in the medium was observed, indicating the growth of microbial community (Additional file 1 : Figure S1). Fig. 2 GPC chromatograms showing the molecular weight distribution of a lignin-waste stream used as the sole carbon source for establishment of the lignin-degrading consortium (LigMet) and b high- and low-molecular-weight fragments detected during 14 days of LigMet cultivation. Above the peaks are the molecular weight distributions in Dalton (Da), which are inversely correlated with the elution time. Different phenolic compounds and polymers were used as internal standards \n The LigMet structure evaluated through amplicon sequencing Alpha diversity is used to describe microbial community composition of a single sample. The calculated index, which includes species richness and diversity, allows to detect changes in its composition and to compare with other samples. Therefore, the microbial richness and diversity of the sugarcane soil and LigMet were quantified through amplicons sequencing. Statistical data summarizing the sequencing are shown in Additional file 1 : Table S2. Clustering of 16S rRNA gene amplicon resulted in 1558 and 355 of OTUs for the sugarcane soil and LigMet, respectively. ACE and Chao1 (richness estimators) and rarefaction analysis suggested that the bacterial species richness in soil and LigMet were entire covered (Additional file 1 : Table S3 and Figure S2A, S2B, S3). Similarly, Good’s coverage index of 0.99 for both LigMet and soil indicated that the sequencing was enough to cover the whole bacterial species. Moreover, Shannon and Simpson’s diversity, which calculated species richness and evenness based on different taxa and their relative abundance, were higher for soil (6.3 and 0.99, respectively) in comparison to LigMet (3.4 and 0.92) (Additional file 1 : Table S3). The lower microbial diversity, as well as richness and evenness, in enriched consortia is an expected selection response based on selective media [ 21 , 22 , 24 , 27 ] and the microorganisms best adapted become dominant [ 24 ]. In our study, four OTUs represented 48% of total sequences (Fig. 3 b), consequently, decreasing the diversity in LigMet compared to its microbial source (soil). Fig. 3 Taxonomic composition of microbial communities from LigMet and the sugarcane soil samples. a Bacterial composition at phylum level. b Bacterial families of highest abundance in the LigMet and in the sugarcane soil sample. c and d Fungal communities at phylum and species level, respectively, from LigMet sample based on ITS2 sequencing \n To compare the taxonomic composition of LigMet and sugarcane soil, the OTU representative sequences were assigned from the phylum to genus level. Figure 3 shows the sugarcane soil and LigMet taxonomic profile. A total of 17 phyla were identified in soil amplicon data and the most abundant phyla were Proteobacteria (29% of the total reads), Acidobacteria (22%), and Actinobacteria (15%) (Fig. 3 a). At the class rank, Alphaproteobacteria (15%), Actinobacteria (14%), and Acidobacteria (14%) were predominant in LigMet (Additional file 1 : Figure S4). The overall taxonomic profile of the sugarcane soil corroborated with the previous sequencing studies addressing soil microbial communities [ 64 – 66 ]. The LigMet exhibited a significant difference in the relative abundance of the phylogenetic groups compared to the sugarcane soil. The LigMet 16S sequences of Proteobacteria origin reached up to 58% of the total sequences classified, which were assigned to the Betaproteobacteria (25%), Alphaproteobacteria (24%), and Gammaproteobacteria (9%) classes (Fig. 3 a). Members of Actinobacteria and Firmicutes were also enriched in LigMet, representing the second and the third predominant phyla with 21% and 13% of the classified sequences, respectively. On the other hand, sequences assigned to Acidobacteria class decreased to less than 0.1% in the LigMet, while, in the soil, this group represented around 22% of the total sequences (Fig. 3 a). The most abundant bacterial families included Alcaligenaceae (24%) , Micrococcaceae (11%), Phyllobacteriaceae (9%), and Paenibacillaceaea (5%), totalizing 48% of the total sequences assigned. Interestingly, the predominant family in LigMet ( Alcaligenaceae ) was virtually non-existent in the soil (less than 0.5%). At the genera level, almost half of total the sequences in the LigMet were assigned to Achromobacter (belongs to Betaproteobacteria ), Paenarthrobacter (Actinobacteria) , Pseudaminobacter ( Alphaproteobacteria ), and Paenibacillus (Bacilli). The result suggests that these microorganisms were key players in LigMet. Betaproteobacteria , Alphaproteobacteria , Actinobacteria, and Bacilli classes were frequently reported as degraders lignin and/or lignin-derived aromatic compounds [ 14 , 17 , 67 , 68 ]. For example, Achromobacter and Paenarthrobacter have been characterized for their ability to degrade biphenyl and polycyclic aromatic hydrocarbons, respectively [ 69 , 70 ]. Ochrobactrum , Rhizobiales , Sphingobium, and Sphingomonas are examples of the genus belong to Alphaproteobacteria described as lignin-degrading bacteria [ 14 , 18 , 69 ]. Nonetheless, Pseudomaniobacter genus is being reported for the first time involved in aromatic compounds degradation. The presence of fungal community was also analyzed in LigMet. The richness estimators indicated only 12 OTUs, considering an evolutionary distance of 0.03, while Shannon and Simpson diversity were 1.44 and 0.71, respectively (Additional file 1 : Table S3). OTUs were assigned to Basidiomycota and Ascomycota members (Fig. 3 c). On average, approximately 94% of all the analyzed sequences could be assigned at the genus level (Fig. 3 d). Among them, Candida sp. (39%), followed by Rhodosporidium sp. (26%), Trichosporon sp. (20%), and Cyberlindnera sp. (9%), were predominant. Candida tropicalis was described as being capable of degrading phenolic compounds [ 71 , 72 ]. Trichosporon and Cyberlindnera were reported as dye-decolorizing yeasts [ 73 ]. The community had very few sequences assigned to the Archaea domain, such as OTUs assigned to the phylum Thaumarchaeota , which was identified in sugarcane soil and LigMet with 0.038% and 0.002% of total relative OTU abundance, respectively (Additional file 1 : Figure S5). The phylum Euryarchaeota was identified only in the LigMet (0.01% relative abundance). Moreover, sequences affiliated with the protozoan class Litostomatea were detected in LigMet based on 18S rDNA analyses (Additional file 1 : Table S4). According to Simek et al. [ 74 ] and Jürgens et al. [ 75 ], the taxonomic structure of microbial communities can be shaped by protozoa due to its preferential predation of particular bacterial taxa. Overall, the statistical analysis showed the enrichment of a few phylogenetic groups in response to adaption of the carbon source added to the medium. As mentioned previously, the taxonomic analysis revealed several microorganisms previously reported as lignin and/or phenolic compounds degraders, thus encouraging further analyses on the genetic content of LigMet. Genome assembly and overall functional annotation of LigMet A total of 97.5 million high-quality paired-end reads (~ 18 gigabase of sequences) were generated from the LigMet sample. Assembled reads resulted in approximately 260 megabases (Mb) of contiguous sequences ≥ 1 Kilobase (Kb). The N50 statistic from the assembled data was 35.5 Kb and the longest contig assembled reached up to 1.3 Mbp. Eukaryotic sequences were not detected in the assembled metagenome, and therefore, downstream analyses focused only on prokaryotic genes and genomes. Considering the complete assembly, 282,237 gene models were predicted (prokaryotic protein-coding regions), where 243,896 of these were larger than 300 bp. The functional annotation was determined by different methods, where all predicted proteins obtained from the metagenome data set were annotated using EggNOG, KEGG, Pfam, and dbCAN databases. According to the results, 242,299 unigenes (85.8% of the predicted proteins) were assigned based on the EggNOG database, 129,140 (45.7%) based on KEGG, and 208,248 (73.78%) on Pfam, and 8800 (3.1%) had at least one protein domain annotated in dbCAN. The functional annotation obtained based on EggNOG and KEGG was quite similar (Fig. 4 ); the majority of proteins were classified as belonging to the metabolism category, which harbors a variety of pathways involved in degradation of aromatic compounds. According to the KEGG and Pfam annotation, several aromatic compound degradation pathways were found complete in the metagenome data set, including benzoate degradation (or methyl benzoate) to catechol (or methylcatechol), catechol ortho-cleavage, catechol meta-cleavage, and phthalate degradation (Additional file 1 : Figure S6 and Fig. 5 ). The pathways for metabolism of lignin components and gene designations were described and demonstrated in the previous studies [ 13 , 76 – 78 ]. Fig. 4 Summary of the annotated predicted proteins from the LigMet Metagenome according to KEGG ( a ) and eggNOG databases ( b ) \n Fig. 5 Pfam domains related to aromatic compound degradation pathways. Pfam domains were identified in bacterial draft genomes recovered from LigMet, Paenarthrobacter sp. HW13 genome, and in prokaryotic predicted proteins from LigMet \n Reconstruction of draft genomes from the metagenomic data set The assembly of metagenomic data resulted in the reconstruction of 65 bins (Additional file 1 : Table S5), of which 20 were considered draft bacterial genomes, since they presented completeness greater than 75% and contamination less than 10% (Table 1 ). The taxonomic assignments show that most of the genome bins were assigned to Alphaproteobacteria (9 bins), followed by Actinobacteria (6 bins) and Betaproteobacteria (2 bins). The remaining genome bins were assigned to Gammaproteobacteria , Bacilli , and Sphingobacteriia . These results are in accordance with the previous studies that reported Rhodococcus , Sphingobacterium , Ochrobactrum , Bacillus , Sphingobium, and Arthrobacter as microorganisms that harbor lignolytic enzymes and metabolize aromatic compounds [ 14 , 16 , 79 ]. The 20 draft genomes were further analyzed to depict their metabolic potential on lignin and aromatic compound degradation (Fig. 5 ). Table 1 Assembly statistics from draft genomes binned from CLig metagenome (20 best results) Bin Id Taxonomic affiliation Completeness (%) Contamination (%) Heterogeneity (%) Total length (bp) # contigs # Predicted genes GC (%) 3 \n Sphingomonadaceae \n 97.9 2.1 0 2,600,224 14 2487 63.8 4 \n Alcaligenaceae \n 96.6 3.2 8.3 4,071,016 139 3831 63.9 6 \n Microbacteriaceae \n 93.0 2.3 25.0 2,775,244 23 2744 70.2 7 Mycobacterium sp. 98.6 4.7 7.4 8,113,983 152 7824 67.0 14 Agrobacterium sp. 97.0 1.9 0 3,723,529 25 3637 62.4 17 Rhodococcus sp. 96.4 1.4 12.5 5,940,093 80 5417 67.9 19 Rhodobacter sp. 97.5 3.5 0 4,829,944 102 4636 66.4 20 \n Comamonadaceae \n 76.1 1.8 0 5,317,169 797 5668 69.5 24 Sphingobacterium sp. 92.2 2.2 0 5,342,826 905 5155 40.9 28 Ochrobactrum sp. 99.1 10.5 0 8,098,959 173 7885 56.5 30 Bacillus sp. 92.6 3.7 0 4,962,376 118 4977 35.4 35 \n Acetobacteraceae \n 96.7 2.0 0 5,217,664 48 5050 70.2 37 Sphingobium sp. 98.1 9.3 5.7 6,313,527 428 6265 59.7 47 Pseudoxanthomonas sp. 97.3 1.5 0 3,192,941 22 2886 71.6 48 Paenarthrobacter sp. 97.0 4.1 0 4,259,515 57 3930 68.2 55 Paenarthrobacter sp. 97.3 1.6 0 4,537,449 80 4185 63.5 56 \n Acetobacteraceae \n 92.7 1.7 25.0 5,069,773 252 4671 70.4 60 Hyphomicrobium sp. 97.9 0.9 0 3,563,628 12 3278 64.7 64 Methylobacterium sp. 94.8 1.6 25.0 3,854,043 176 3682 70.1 65 \n Rhodococcus \n 96.8 1.8 0 5,770,250 74 5292 70.8 \n Biocatalysts and pathways involved in lignin degradation and aromatic compound conversion Pfam-based analysis identified several conserved domains related to lignin and aromatic compound degradation in the LigMet data set and genome bins (Fig. 5 ). Among these, peroxidases (PF00141), dye-decolorizing peroxidases (PF04261), and laccases (PF00394, PF02578, PF07731, and PF07732) domains, and enzymes that can generate radicals for cleavage of several lignin linkages, were found predominantly in genome bins belonging to the Actinobacteria class (bins 48,7,17, and 65) (Fig. 5 ). These findings corroborated the previous studies, reporting that Actinobacteria is able to produce extracellular lignolytic enzymes [ 14 , 17 , 80 ]. Glutathione-dependent beta-etherases of Sphingobium sp. were recently described as able to cleave beta-aryl ether bonds of lignin from softwoods and hardwoods [ 81 ]. Corresponding domains for the beta-etherases LigE, LigP, LigG, and LigF (considering the PF13417, PF00043, and PF02798) were predominantly found in bins 4, 20, 28, 35, and 37, which were assigned with Alpha and Betaproteobacteria origin (Fig. 5 ). Conversion of the beta-aryl ether lignin dimer involves LigEFG and LigD (PF00106) and results in vanillic acid [ 16 ], which can also be generated from vanillin by the action of LigV (PF00171). In this pathway, vanillic acid is converted into protocatechuic acid by LigM (PF01571 and PF08669). All these protein domains (LigEFG, LigD, LigV, and LigM) were identified in several bins, indicating their potential to metabolize vanillin and vanillic acid. Nonetheless, they were found predominantly in bins assigned to Mycobacterium (bin 7). The biphenyl linkage is also a common component of lignin structure. Its cleavage is carried out by the cascade pathway LigX (PF0035), LigZ (PF02900), and LigW, LigW2, and LigY (PF04909) [ 16 ], also resulting in the central intermediate vanillic acid. With the exception of bins 6 and 28, several Actinobacteria genome bins displayed the complete set of proteins needed for biphenyl and vanillic acid conversion (Fig. 5 ). However, it is possible to notice that this cleavage cascade pathway was identified in only a few bins of Proteobacteria origin (20, 37, and 3) (Fig. 5 ). The depolymerization of lignin releases a mixture of aromatic monomers that can be used as carbon and energy sources by several microorganisms [ 82 , 83 ]. Phenolic aromatic monomers can be converted into metabolic intermediates, via catechol and protocatechuate, by the action of dioxygenases, which are classified according to the relative positions of hydroxyl groups (ortho- and meta-cleavage) [ 13 ]. The routes can be divided in three blocks: (i) the branch of catechol intermediate (ortho-cleavage), which involves the following enzymes: CatA (PF0775 and PF04444), CatB (PF02746 and PF13378), and CatC (PF02426); (ii) the branch of protocatechuate (meta-cleavage), which involves PcaG (PF00775), PcaH (PF00775 and PF12391), PcaB (PF10397 and PF00206), and PcaC (PF02627); and finally, (iii) the reactions common for both branches, catalyzed by PcaD (PF00561), PcaI (PF01144), PcaJ (PF01144), and PcaF (PF02803 and PF00108). Three Actinomycetes genome bins (17, 48, and 65) and two of Proteobacteria origin (4 and 20) presented all protein domains related to routes i, ii, and iii. Regarding route (ii), all Actinobacteria genome bins and two of Proteobacteria origin (bins 20 and 35) were clearly enriched with coding genes involved in protocatechuate degradation. With exception of the Sphingobacterium genome bin (24) and the Bacillus bin (30), the protein domains corresponding to enzymes related to route iii were mapped in all bins. The present work disclosed several novel aromatic-degrading enzymes of a high and low degree of homology to previously identified variants. For instance, the BLASTp comparison with beta-etherase (LigE) from S. paucimobilis SYK-6 identified 18 orthologous in LigMet, with sequence identity varying from 31 to 64% (E-value from 4e−30 to 0.0). Accordingly, LigMet orthologous to beta-etherase LigF (26 hits), Glutathione S-transferase homolog LigG (3 hits) and beta-etherase LigP (18 hits) from S. paucimobilis were recovered as well, with sequence identity varying from 26 to 93% (4e−178 to 9e−11), 29 to 84% (4e−164 to 3e−18), and 31 to 88% (4e−30 to 0.0), respectively. Moreover, to obtain deeper insight on potential of the microbial community to carry out redox and hydrolytic mechanisms related to lignocellulose degradation, the auxiliary activity (AA) and carbohydrate esterase (CE) families, which act with CAZymes, were predicted based on dbCAN analysis (Additional file 1 : Figure S7). The carbohydrate esterases from family 1 (CE1), including feruloyl esterases and p -coumaroyl esterases, can break down ester cross links of lignin and hemicelluloses [ 84 , 85 ]; and glucuronyl esterases (CE15) were reported to catalyze ester linkage hydrolysis between glucuronoxylan and lignin [ 86 ]. Furthermore, acetyl esterases (described in families CE01, 02, 03, 04, 05, 06, 07, 12, and 16) play a role in degradation of xylose units esterified with acetic acid [ 85 ]. Nonetheless, only the families CE1 and CE4 were predominant in the LigMet metagenome data set and widespread in all genome bins (Additional file 1 : Figure S7 and S8). Similar to CE families, AA were abundant in the LigMet metagenome, notably the AA3 family members, which may be related to peroxidase activity, thus corroborating with the previous PFAM-based functional analysis. Paenarthrobacter sp. HW13 genome To complement the metagenomic discovery approach, the isolation of a novel lignin-degrading microorganism from LigMet was performed using HW or LW as the sole carbon source in media. Among the obtained colonies, only a single strain cultivated on HW exhibited a decolorization halo after being covered with an agar solution supplemented with Azure-B dye. Based on the 16S rRNA gene comparison, the analysis revealed that the strain HW13 shares 96.8% 16S rRNA gene sequence similarity with P. ureafaciens DSM 20126 T as the closest related species with a valid nomenclature (Additional file 1 : Figure S9). Therefore, the strain may represent a new species within the Paenarthrobacter genus. The strain was denominated Paenarthrobacter sp. HW13. The genome sequence analysis of the strain HW13 depicted its potential to metabolize lignin fragments and phenolic compounds (Fig. 6 ). Two libraries were sequenced on an Illumina/MiSeq system, resulting in 3,746,638 pair-end reads and 1,671,986 mate-pair reads, accounting for 1083 Mb. After data processing, the assembly resulted in three contigs consisting of 4,091,031 bases and featuring a CG content of 63%. Additional details of the genome features can be found in Additional file 1 : Table S6. To determine whether strain HW13 correspond to bin 48 or 55 (both assigned to the Paenarthrobacter genera), the microbial species identified (MiSi) method [ 87 ] was conducted, as implemented in IMG/ER. The P probability of bins 48 and 55 being assigned to the same species as HW13 was 0.0 and 0.99, respectively. Moreover, strain HW13 showed 100% 16S rRNA gene sequence identity to the 16S rRNA gene found in genome bin 55. The genomic analysis confirmed the prediction of several previously identified genetic determinants involved in aromatic compound metabolism, including the protocatechuate, catechol, phenylacetate, gentisate, and phenylpropionate degradation pathways (Fig. 6 ). Fig. 6 Schematic representation of gene clusters involved in aromatic degradation pathways identified in the Paenarthrobacter sp. HW13 genome. Homologous genes are shown in the same color and the predicted gene products are indicated \n Biosynthesis of vanillin from ferulic acid The occurrence of aromatic conversion pathways in LigMet to produce high value chemicals was validated through biochemical vanillin production. Ferulic acid is found in lignocellulosic biomass, and it is a precursor for biovanillin production [ 88 ]. The conversion of ferulic acid into vanillin has been reported in several microorganisms via coenzyme A-dependent, a non-beta-oxidative pathway [ 58 , 89 – 91 ], including feruloyl-CoA synthetase (FerB) and enoyl-CoA hydratase/aldolase, (FerB) [ 92 ], which are regulated by ferC , an MarR-type transcriptional regulator [ 93 ]. In this study, the protein domains related to FerA (PF13380) and FerB (PF00378) were found in several bins described in Fig. 5 . The clusters of genes related to ferulic acid conversion into vanillin were manually identified in the LigMet, two presenting similar genetic structures previously described in P. fluorescens BF13 [ 93 ] and S. paucimobilis SYK-6 [ 78 ] (Fig. 7 ). In addition, another gene cluster was found in LigMet of similar organization to the gluconate operon of B. subtilis [ 94 ], which is regulated by a GntR family protein (Fig. 7 ). Fig. 7 Structure of the F erA and FerB genes clusters identified in the LigMet metagenome. The structure of FerA and FerB genes clusters described in a \n Sphingobium sp. and b \n Bacillus subtilis , in which the regulator gene belongs to the GntR family, and c \n Pseudomonas sp., showing different arrangements of the regulator ( FerC ). The three gene cluster configurations were identified in the LigMet [ 17 – 19 ], as well as on draft genomes recovered from LigMet, as indicated. Homologous genes are shown in the same color and the predicted gene products are indicated \n Candidates’ genes for ferA (derived from bin 3) and ferB (derived from bin 11) were selected for further analyses. The FerA_B3 (bin 3) when compared to other bacterial feruloyl-CoA synthetases displays 60% amino acid identity to S. paucimobilis (A0A031JK39), followed by Sphingobium sp. (60%; BAK67177), Delftia acidovorans (55%; CAC83622), and Burkholderia glumae (52%; ACR31088). The FerB_B11 (bin 11) compared to other bacterial enoyl-CoA hydratases shows 63% amino acid identity to Pseudomonas nitroreducens (C3VA24), followed by Amycolatopsis methanolica (59%; A0A076MUC3), Streptomyces sp. (55%; S5LPF1), and D. acidovorans (55%; Q8VNW7). Amino acid sequences from NCBI database and Uniprot with similarity higher than 20% to FerA_B3 and FerB_B11 were considered to construct phylogenetic trees (Additional file 1 : Figure S10). It evidenced the relationship FerA_B3 and FerB_B11 with other bacterial feruloyl-CoA synthetases and enoyl-CoA hydratases, respectively, but the low bootstrap value at most of the clades denoted the high divergence of FerA_B3 and FerB_B11. For the bioproduction of vanillin, candidates genes for ferA and ferB were then synthesized and successfully cloned for expression in E. coli, followed by purification by liquid chromatography. The conversion of ferulic acid into vanillin, after combining the two purified enzymes, was confirmed by GC–MS. The vanillin production from ferulic acid was detected after 6 and 24 h incubation using the purified recombinant proteins (Fig. 8 ). The biochemical and structural characterization of the two proteins, FerA_B3 and FerB_B11, will be the focus for future studies. Fig. 8 GC–MS chromatograms of products from the reaction of ferulic acid into vanillin by the enzymes FerA and FerB. The Y -axis represents relative abundance (ion count) and the X -axis represents retention time (in seconds). Vanillin and ferulic acid peaks are depicted in chromatograms A and B by the letters ( v ) and ( f ), respectively. a 6 h reaction; b 24 h reaction"
} | 7,337 |
28000139 | PMC5174012 | pmc | 866 | {
"abstract": "The metabolic state of microflora (mixed microbial cultures) in microbial fuel cells (MFCs) is currently unclear. Metabolomic analyses were conducted of microflora growing on the anodic electrodes of MFCs operated at pH 7.0, 5.5, or 4.0 and utilizing starch as the major carbon substrate. A much higher current was produced at pH 7.0 than at pH 5.5 and 4.0, correlating with an increased population ratio of Geobacter species to the total bacteria growing on the electrode. Most intracellular metabolites related to the tricarboxylic acid (TCA) cycle were present at a higher level at pH 7.0 than at pH 5.5 and 4.0, and the levels of metabolites correlated well with the obtained current densities. A high intracellular adenosine triphosphate (ATP)/adenosine diphosphate (ADP) ratio at pH 7.0, compared to at pH 5.5 and 4.0, likewise supported current production. Overall, the metabolomic analyses demonstrated that activation of the TCA cycle and increased ATP generation are critical parameters for electricity generation by microflora. Electronic supplementary material The online version of this article (doi:10.1186/s13568-016-0299-4) contains supplementary material, which is available to authorized users.",
"introduction": "Introduction Microbial fuel cells (MFCs) are being investigated for the simultaneous treatment of organic and inorganic materials such as wastewater and for generating electricity by using mixed cultures of microflora as the catalyst (Logan et al. 2006 ; Lovley 2008 ). Electrons generated by oxidation of the reduced substrate at the anode flow to the cathode through the circuit, resulting in the reduction of oxygen (Logan 2009 ). In most cases, MFCs employ microflora rather than a pure culture because mixed cultures can utilize complex organic materials as substrates and enhance anodophilic electron transfer due to interspecies connections (Logan et al. 2006 ). The effects of system architecture and operational parameters on electricity generation in MFCs have been intensively studied (Logan et al. 2006 ). Air-cathode MFCs, in which the cathode is in direct contact with the air, are useful because there is no need for aeration or chemical catholytes, there is high electrical output, and they are suitable for scale-up (Shimoyama et al. 2008 ). Electricity generation can be modulated by pH, which is one important parameter (He et al. 2008 ); however, although the pH range for adequate current generation is relatively broad (between pH 7 and 10), pH conditions close to neutral are usually used in MFCs to support the growth of anodic bacteria in air-cathode systems (He et al. 2008 ). To date, the conditions supporting efficient bacterial growth and current generation have been investigated, but less effort has been devoted to investigating the intracellular metabolic states of anode-respiring microflora in air-cathode MFCs. \n Geobacter sulfurreducens is the primary microorganism used for current production and its electron transfer mechanism has been well studied (Lovley 2012 ). Recent advances in metabolomic analysis allow clarification of the intracellular metabolic states of microorganisms (Toya and Shimizu 2013 ) and this approach has shown that pure cultures of G. sulfurreducens activate the tricarboxylic acid (TCA) cycle and down-regulate gluconeogenesis under conditions compatible with electricity generation (Song et al. 2016 ). Metabolomic analysis can also be applied to microflora to understand the metabolic state of a bacterial community as a whole. In microflora used for methane fermentation, the Embden-Meyerhof (EM) pathway is inhibited and simultaneously the reductive branch of the TCA cycle is stimulated when the pH is decreased from 7.5 to 5.0, and methane fermentation is inhibited (Sasaki et al. 2014 ). This finding suggests that metabolomic analysis would be useful for evaluating the conditions for optimizing the performance of microflora in MFCs. The aim of this study was to clarify the metabolic state of microflora on the surface of the anodic electrode in an air-cathode MFC operated at pH 7.0 to generate current. The metabolic states of microflora in comparable MFCs operated at pH 5.5 or 4.0 and producing a relatively low current were used as controls. Starch was the major carbon source in the substrate, simulating wastewater containing carbohydrate.",
"discussion": "Discussion The observed good current generation at neutral pH and poor current generation below pH 5.0 were in accordance with the previous study of Patil et al. ( 2011 ). In the present study, microorganisms related to Geobacter species would play a major role in current generation at the anode electrode (Lovley 2008 ). Geobacter cells oxidize acetate via the TCA cycle to generate electricity (Mahadevan et al. 2006 ). Good correlation between the levels of metabolites relating to the TCA cycle and current generation is corroborated by our previous study showing that the intracellular metabolites in the TCA cycle are present at high concentration and activation of the TCA cycle supports higher electricity generation by a pure Geobacter culture (Song et al. 2016 ). The observed decrease in Geobacter cells at lower pH values and current densities indicates that the metabolic state of the TCA cycle of microflora directly reflects the state of the Geobacter cells, even though the biofilm on the surface of the anode included microorganisms other than Geobacter . Thus, the present results indicate that acetate produced from starch-derived glucose through acetyl-CoA is consumed via the TCA cycle in microflora, comprised mainly of Geobacter cells, for electricity generation at pH 7.0 (Fig. 3 ). In addition, a higher intracellular concentration of glutamate was observed at pH 7.0 (Fig. 3 ). This correlates to higher cell numbers on the electrode at pH 7.0 compared to at pH 5.5 and 4.0 (data not shown) and supports the central importance of glutamate at the intersection between carbon and nitrogen metabolism required for bacterial growth (Commichau et al. 2008 ). The profile of intercellular metabolites related to glycolysis (EM and PP pathways) at pH 7.0, 5.5, and pH 4.0 showed no relationship with current density (Additional file 1 : Figure S1). Geobacter cells reportedly down-regulate gluconeogenesis during electricity generation (Meng et al. 2013 ; Song et al. 2016 ). Because G. sulfurreducens cells cannot directly utilize glucose (Mahadevan et al. 2006 ), we can separate the phases of current production by microflora into an acetate-producing phase from starch via glycolysis, and a current-producing phase from acetate via the TCA cycle. Thus, intracellular metabolites related to glycolysis would be affected by carbon flux from starch-derived glucose in microorganisms other than Geobacter cells. In contrast, the intracellular concentrations of acetyl-CoA were higher at pH 5.5 and 4.0 than at pH 7.0, implying that the accumulation of organic acids at pH 5.5 and 4.0 inhibited carbon flux via acetyl-CoA (Fig. 3 ). The same profiles were observed in microflora used for methane fermentation from glucose (Sasaki et al. 2014 ). ATP is produced in the respiratory electron transfer chain and oxidative consumption of NADH occurs when electricity generation is high in Geobacter species (Meng et al. 2013 ). ATP is also produced in the EM pathway by substrate level phosphorylation. Thus, increased ATP/ADP would be due to activation of the EM pathway (by consumption of starch-derived glucose) and TCA cycle (by electricity generation) at pH 7.0 (Fig. 4 ). In addition, the observed decrease in the NADH/NAD + ratio of the microflora at pH 7.0 was as expected, whereas the reason for the fluctuating NADH/NAD + ratio at pH 5.5 remains unclear. In contrast, the NADPH/NADP + ratio in the microflora was the same at pH 7.0, 5.5, and 4.0, although oxidative consumption of NADPH is faster in electricity-generating Geobacter cells (Song et al. 2016 ). During electricity generation, NADPH is produced in the TCA cycle (isocitrate to α-ketoglutarate) and consumed during glutamate production and in the respiratory chain. In this study, NADPH was additionally supplied from the PP pathway, resulting in an increased NADPH/NADP + ratio at pH 7.0 and 5.5. In this study, metabolomic analyses of the microflora on the anodic electrode of a MFC showed for the first time that activation of the TCA cycle and increased ATP generation are necessary for efficient current generation from starch-like waste, as observed in single kind of microorganism, Geobacter cells."
} | 2,146 |
25336453 | PMC4212834 | pmc | 867 | {
"abstract": "ABSTRACT Microbial conversion of carbon dioxide to organic commodities via syngas metabolism or microbial electrosynthesis is an attractive option for production of renewable biocommodities. The recent development of an initial genetic toolbox for the acetogen Clostridium ljungdahlii has suggested that C. ljungdahlii may be an effective chassis for such conversions. This possibility was evaluated by engineering a strain to produce butyrate, a valuable commodity that is not a natural product of C. ljungdahlii metabolism. Heterologous genes required for butyrate production from acetyl-coenzyme A (CoA) were identified and introduced initially on plasmids and in subsequent strain designs integrated into the C. ljungdahlii chromosome. Iterative strain designs involved increasing translation of a key enzyme by modifying a ribosome binding site, inactivating the gene encoding the first step in the conversion of acetyl-CoA to acetate, disrupting the gene which encodes the primary bifunctional aldehyde/alcohol dehydrogenase for ethanol production, and interrupting the gene for a CoA transferase that potentially represented an alternative route for the production of acetate. These modifications yielded a strain in which ca. 50 or 70% of the carbon and electron flow was diverted to the production of butyrate with H 2 or CO as the electron donor, respectively. These results demonstrate the possibility of producing high-value commodities from carbon dioxide with C. ljungdahlii as the catalyst.",
"introduction": "INTRODUCTION Acetogenic microorganisms are attractive catalysts for the conversion of syngas or carbon monoxide to organic commodities ( 1 – 7 ) and also have the ability to convert carbon dioxide to organic products with electrons derived from an electrode in a process known as microbial electrosynthesis ( 5 , 8 , 9 ). An attractive feature favoring the use of acetogens in these processes is that the Wood-Ljungdahl pathway, which is naturally present in acetogens, provides the most effective known pathway for converting carbon dioxide to organic compounds that are then excreted from the cell ( 5 , 10 ). Acetogens typically produce acetate as their primary end product, but other natural products include ethanol, 2,3-butanediol, butyrate, and butanol ( 1 – 3 , 6 , 7 , 11 – 13 ). Furthermore, under the appropriate conditions, products other than acetate may predominate. For example, modifying growth conditions permitted Clostridium ljungdahlii and Clostridium autoethanogenum (C. D. Mihalcea, J. M. Y. Fung, B. Al-Sinawi, and L. P. Tran, U.S. patent application publication no. US/0230894 A1) to produce ethanol as the dominant product from syngas metabolism and allowed Butyribacterium methylotrophicum adapted for growth on carbon monoxide to produce butyrate at molar levels comparable with those of acetate ( 14 ). Genetic engineering is another approach to enhance the production of commodities other than acetate ( 1 ). Transient production of butanol was achieved by expressing genes required for butanol production on a plasmid introduced into C. ljungdahlii ( 15 ). Expression of heterologous genes for acetone in C. ljungdahlii diverted carbon and electron flow to this product ( 16 ). Moorella thermoacetica was genetically manipulated to produce lactate ( 17 ). However, in none of these studies was production of the target commodity optimized with additional genetic modifications, such as deletion of genes for competing pathways. Much more impressive yields of multicarbon products from Clostridium strains were reported in studies by researchers at Syngas Biofuels Energy, Inc., but the validity of these results is doubtful ( 1 ). An improved array of tools for genetic manipulation of C. ljungdahlii ( 16 , 18 ) and a steadily improving understanding of its physiology ( 11 , 15 , 19 – 23 ) suggested that C. ljungdahlii might be an effective acetogen chassis for the construction of strains that produce organic products with more than two carbons from carbon dioxide with high yields. We chose butyrate for proof-of-concept studies. Butyrate is a food supplement that benefits colon function ( 24 , 25 ) and is also used in food flavorings ( 26 , 27 ). Butyrate is a feedstock for production of cellulose acetate butyrate ( 28 ), which is a thermoplastic that has a wide range of applications, such as paints ( 29 ) and polymers, such as poly(3-hydroxybutyrate)/cellulose acetate butyrate ( 30 ). Poly(3-hydroxybutyrate-co-3-hydroxyhexanoate) was synthesized from butyrate using an engineered strain of Ralstonia eutropha ( 31 ). Furthermore, butyrate is a precursor to butanol, which has value as a fuel ( 32 , 33 ). Current industrial production of butyrate relies on chemical synthesis from petroleum. Another potential biological route is fermentation of sugars ( 34 – 38 ). Several strains naturally produce high titers of butyrate from sugars, and butyrate production has been enhanced with genetic engineering or adaptation approaches ( 37 – 45 ). However, producing butyrate from carbon dioxide with waste gases ( 1 ) or renewable electricity ( 5 ) as the energy source may be more environmentally sustainable and does not consume high-quality biomass feedstocks that may be more appropriately consumed as food. Therefore, we evaluated the possibility of engineering C. ljungdahlii to produce butyrate from carbon dioxide.",
"discussion": "RESULTS AND DISCUSSION Initial engineering of C. ljungdahlii for butyrate production: strain B1. As previously reported ( 46 , 47 ), wild-type cells grew with either H 2 , CO, or fructose as the electron donor, with acetate as the primary end product ( Fig. 1 and 2 ). Growth rates and yields were highest with fructose, intermediate with CO, and lowest with H 2 . FIG 1 Engineering a pathway for butyrate synthesis. The C. ljungdahlii native pathways for acetate and ethanol formation are indicated in grey. The added butyrate pathway is in black. Genes encoding enzymes are in parentheses. The thick black “x” indicates pathways that were disrupted in this study. FIG 2 Growth profiles of the C. ljungdahlii wild-type strain. Data are representative of duplicate cultures. In order to redirect carbon and electron flow to the production of butyrate, the following genes from Clostridium acetobutylicum were introduced into C. ljungdahlii : thl , which encodes thiolase for the synthesis of acetoacetyl-coenzyme A (CoA) from acetyl-CoA; hbd , which encodes 3-hydroxybutyryl-CoA dehydrogenase for the synthesis of 3-hydroxybutyryl-CoA from acetoacetyl-CoA; crt , which encodes crotonase for the synthesis of crotonyl-CoA from 3-hydroxybutyryl-CoA; bcd , which encodes butyryl-CoA dehydrogenase for the synthesis of butyryl-CoA from crotonyl-CoA; etfA and etfB , which encode electron transfer flavoproteins that form an enzyme complex with butyryl-CoA dehydrogenase for the synthesis of butyryl-CoA; ptb , which encodes phosphotransbutyrylase for the synthesis of butyryl phosphate from butyryl-CoA; and buk , which encodes butyrate kinase for synthesis of butyrate from butyryl phosphate ( Fig. 1 ). The genes thl , crt , bcd , etfA , etfB , and hbd were on one plasmid, and the genes ptb and buk were on a second plasmid (see Fig. S1 and S2 in the supplemental material). The expression of all eight genes was under the control of the putative promoter of the pta gene from C. ljungdahlii , which is considered to be highly expressed under a diversity of conditions. This strain was designated B1. Both acetate synthesis via the Pta/Ack pathway and butyrate synthesis via the Ptb/Buk pathway result in ATP production ( Fig. 1 ). The conversion of acetyl-CoA to acetate is expected to yield 1 ATP via substrate-level phosphorylation, whereas only 0.5 ATP is expected to be derived from each acetyl-CoA converted to butyrate ( Fig. 1 ). However, the conversion of crotonyl-CoA to butyryl-CoA catalyzed by the complex of Bcd and EtfAB is considered to be coupled with energy generation via the Rnf complex and ATP synthase ( Fig. 1 ) to yield ca. 0.5 ATP ( 48 , 49 ), but the stoichiometry of the proton gradient and ATP generation has not been clarified for C. ljungdahlii . Butyrate synthesis from acetyl-CoA requires additional reducing equivalents in the form of NADH for Hbd to reduce acetoacetyl-CoA to 3-hydroxybutyryl-CoA ( 50 ) and for Bcd to reduce crotonyl-CoA to butyryl-CoA ( 51 ). When grown with H 2 as the electron donor, strain B1 produced butyrate as well as acetate and ethanol over time ( Fig. 3 and 4 ). Acetate remained the primary product, with minor production of ethanol, but the butyrate produced (8.5 mM) accounted for 13% of the carbon and 16% of the electrons appearing in acetate. More butyrate was produced with CO as the electron donor than with H 2 ( Fig. 3 and 4 ), and the proportion of carbon and electrons in butyrate compared to that in acetate, 25% and 29%, respectively, was also greater. Butyrate production was lowest during growth on fructose, with only slightly more butyrate than ethanol produced ( Fig. 3 and 4 ). When grown with H 2 or CO as the electron donor, strain B1 had longer lag periods and lower growth rates than the wild-type strain, but the final cell yields were comparable to those of the wild-type strain with all three substrates. FIG 3 Growth profiles of strain B1. Data are representative of duplicate cultures. FIG 4 Summary of yields of biomass (OD 600 ), acetate (mM), ethanol (mM), butyrate (mM), and electrons appearing in butyrate versus acetate (%). Data are means from duplicate cultures. Enhancing Crt expression through ribosome binding site modification: strain B2. Proteomic analysis demonstrated that with the exception of crotonase (Crt), all of the heterologous enzymes were expressed at levels comparable to those for highly expressed native proteins, such as phosphotransacetylase (Pta) and acetate kinase (Ack) ( Fig. 5A ). However, the transcript abundance of crt was comparable to that of other heterologous genes ( Fig. 5B ). These results suggested that the low abundance of the Crt protein resulted from inefficient translation of crt transcripts. The distance between the putative ribosome binding site (RBS) and translation initiation codon of crt was shorter than those for the other heterologous genes ( Fig. 6A ). Therefore, a crt sequence in which the length between the putative RBS and translation initiation codon was increased and substituted for the original crt sequence in constructing strain B2. Western blot analysis revealed that strain B2 produced substantially more Crt than strain B1 ( Fig. 6B ). Strain B2 produced more butyrate with all three electron donors ( Fig. 4 and 7 ), suggesting that the low level of Crt was a bottleneck in butyrate production in strain B1. FIG 5 Proteomic and transcriptomic analyses of strain B1. (A) Proteomic analysis. Numbers of identified peptides are shown. The wild type (WT) and strain B1 were analyzed. (B) Transcriptomic analysis. Transcript abundance in strain B1 is presented as transcript copy numbers (×10 7 ). FIG 6 Modification of the ribosomal binding site (RBS) of the crt gene. (A) Comparison of putative RBSs. Putative RBSs are indicated in bold. Putative translation initiation codons are underlined. Lowercase indicates sequence from restriction sites used for cloning. “ crt -M” represents the modified sequence used in strain B2. (B) Western blot analysis for the Crt protein. Cell extracts prepared from the wild-type strain (WT), strain B1, and strain B2 were analyzed using an antibody against Crt. “S” represents protein standards, whose sizes are shown in kDa. FIG 7 Growth profiles of strain B2. Data are representative of duplicate cultures. Integration of butyrate pathway genes into chromosome: strain B3. Chromosomal integration of the genes for butyrate synthesis may be preferable to maintaining the genes on plasmids, to avoid the need to add antibiotics to retain plasmids and to reduce the cellular metabolic burden ( 52 , 53 ). A fragment containing the genes ptb and buk was integrated at the adhE1 locus on the chromosome with double-crossover homologous recombination, but a longer fragment containing thl - crt - bcd - etfB - etfA - hbd could not be integrated in this manner (data not shown). However, the DNA fragment of thl - crt - bcd - etfB - etfA - hbd - ptb - buk ( crt with the modified RBS), under the control of the putative pta promoter region, was successfully integrated into the putative pta promoter region via single-crossover homologous recombination (see Fig. S3 in the supplemental material). Compared with strain B2, which had the same genes expressed on plasmids, strain B3 grew faster but with a lower final biomass yield when H 2 was the electron donor, ( Fig. 8 ). A final butyrate yield by strain B3 was slightly lower than that by strain B2, but butyrate production compared to acetate production was improved ( Fig. 4 ). In contrast, strain B3 grew faster than strain B2 with a similar final biomass yield on CO, and butyrate production was lower than that of strain B2. With fructose as the substrate, strain B3 grew faster than strain B2, but yields of biomass, acetate, and butyrate were comparable. In addition to acetate and butyrate, strain B3 produced ethanol at lower levels under all growth conditions ( Fig. 8 ). FIG 8 Growth profiles of strain B3. Data are representative of duplicate cultures. Inactivating Pta-dependent acetate synthesis: strain B4. In an attempt to further enhance the diversion of carbon and electron flux from acetate production to butyrate synthesis, Pta, which is thought to catalyze the first step in the conversion of acetyl-CoA to acetate ( Fig. 1 ), was disrupted with a single-crossover homologous recombination, which simultaneously integrated the butyrate pathway genes (see Fig. S4 in the supplemental material). The Cre-lox system, which allows reuse of an antibiotic resistance gene ( 54 ), was applied to C. ljungdahlii strain engineering (see Fig. S5 , S6 , and S7 ). This strain was designated B4. Strain B4 grew slower than strain B3 with H 2 as the electron donor but with a comparable final cell yield ( Fig. 9 ). Surprisingly, strain B4 produced acetate at amounts comparable to those for strain B3 ( Fig. 4 ), despite the fact that the expected primary route for acetate synthesis was disrupted. Strain B4 produced slightly more butyrate and less acetate than strain B3 during growth with CO or fructose ( Fig. 4 and 9 ), but acetate production remained substantial ( Fig. 4 and 9 ), suggesting that there are one or more unknown pathways for acetate synthesis in C. ljungdahlii . FIG 9 Growth profiles of strain B4. Data are representative of duplicate cultures. Inactivating Pta-dependent acetate synthesis and AdhE1-dependent ethanol synthesis: strain B5. In an attempt to divert the NAD(P)H being consumed for ethanol production toward the production of butyrate, the ethanol pathway in strain B4 was inactivated by disrupting the adhE1 gene, which was shown ( 18 ) to encode the major bifunctional aldehyde/alcohol dehydrogenase for ethanol production, via single-crossover homologous recombination ( Fig. 1 ; see also Fig. S8 in the supplemental material), creating strain B5. Strain B5 produced significantly less ethanol than strain B4, slightly improving butyrate production during growth on CO or fructose but not H 2 ( Fig. 4 and 10 ). FIG 10 Growth profiles of strain B5. Data are representative of duplicate cultures. Inactivating Pta-dependent acetate synthesis and CoA transferase: strain B6. It was considered that a CoA transferase might convert butyrate back to butyryl-CoA with simultaneous production of acetate or other fatty acids ( Fig. 1 ). In an attempt to avoid this possibility, the gene Clju_c39430, which encodes a homolog of CoA transferases, was disrupted via single-crossover homologous recombination (see Fig. S8 in the supplemental material) in strain B4, yielding strain B6. Acetate continued to be produced in strain B6, but this strain was the best butyrate-producing strain in terms of butyrate yield under all three growth conditions tested ( Fig. 4 and 11 ). Carbon and electron yields in butyrate were 42% and 48% (H 2 ), 68% and 73% (CO), and 71% and 75% (fructose), respectively. FIG 11 Growth profiles of strain B6. Data are representative of duplicate cultures. Implications. The results demonstrate that it was possible through genetic manipulation to redirect carbon and electron flow to the production of butyrate from carbon dioxide in C. ljungdahlii . However, further genetic manipulation will be required before the goal of producing butyrate as the sole product of metabolism will be achieved. The continued production of acetate when pta is disrupted was previously observed in C. acetobutylicum , which has a fermentative metabolism ( 55 ). A potential alternative route for acetate production in C. ljungdahlii , even during autotrophic growth, is conversion of acetyl-CoA to acetate via an acetaldehyde intermediate ( 15 ). This could yield energy to support cell growth and maintenance through the Rnf complex and ATP synthase. Conversion of acetyl-CoA to acetaldehyde and acetaldehyde to acetate is thought to proceed with oxidation of NADH to NAD + and reduction of ferredoxin, respectively ( Fig. 1 ) ( 15 ). Proton gradients generated with the reactions coupled with the reduction of NAD + and the oxidation of ferredoxin via the Rnf complex result in ATP synthesis by ATP synthase ( 48 , 49 ). Therefore, net ATP synthesis may be possible via this route of acetate production. ATP generated in this manner would yield ca. 0.5 ATP, which would be comparable to ATP generation via butyrate production ( Fig. 1 ). A previous study demonstrated that deleting adhE1 , which encodes a bifunctional aldehyde/alcohol dehydrogenase, prevented ethanol formation from acetyl-CoA, presumably by eliminating acetaldehyde formation ( 18 ). However, strain B5, in which both the pta gene and the adhE1 gene were inactivated, still produced acetate. There are other homologs for aldehyde dehydrogenase in the C. ljungdahlii genome (Clju_c11960, Clju_c39730, and Clju_c39840). Therefore, deletion of one or more of these genes may be required in order to completely eliminate acetate production. Deletion of many genes in the same strain is currently not possible because there are only three antibiotic resistance markers that have been identified, severely limiting the number of genes that can be deleted if they are not colocalized on the chromosome. We adopted the Cre-lox system in order to permit simultaneous disruption of three genes, but this method leaves the recombination site on the chromosome, and thus they can be used only once in a strain. Negative- or counterselection markers, such as pyrF ( 56 ), galK ( 57 ), and mazF ( 58 ), which can be used repeatedly and have been applied to other Clostridium species, would be useful, but attempts to apply them to C. ljungdahlii have not yet been successful (unpublished data). Another possibility for the unexpected acetate production is that the enzyme phosphotransbutyrylase (Ptb) introduced in the synthetic butyrate pathway may also act on acetyl-CoA, as previously described for the Ptb purified from C. acetobutylicum ( 59 ). The specificity for acetyl-CoA was only 1.6% of that for butyryl-CoA, but even low reactivity with acetyl-CoA, which is likely to be a more abundant metabolite, could be significant. Acetate production via this pathway could be favored because it would yield more ATP and have lower reducing equivalent demands than the desired production of butyrate. When the appropriate tools are available, deleting ack may eliminate this possibility because acetate kinase is essential for this potential pathway. Substituting bukII , which encodes another butyrate kinase that is more specific for butyrate synthesis ( 37 ), for buk in the ptb - buk operon might also improve butyrate yields. These and previous ( 16 , 18 ) findings that carbon and electron flow associated with the Wood-Ljungdahl pathway in C. ljungdahlii can be genetically redirected suggest that C. ljungdahlii may be an effective chassis for the production of organic commodities from carbon monoxide-containing waste gases ( 11 , 15 ) or from carbon dioxide via electrosynthesis ( 9 ). Further development of the genetic tools needed to accomplish this goal is under way."
} | 5,149 |
39597724 | PMC11596381 | pmc | 868 | {
"abstract": "Rhizosphere microorganisms are crucial for enhancing plant stress resistance. Current studies have shown that Arbuscular mycorrhizal fungi (AMF) can facilitate vegetation recovery in heavy metal-contaminated soils through interactions with rhizosphere microbiota. However, the mechanisms by which AMF influences rhizosphere microbiota and plant growth under cadmium (Cd) stress remain unclear. In this study, Lolium perenne L. was inoculated with AMF ( Rhizophagus irregularis ) and grown in soils supplemented with Cd (0 mg kg −1 , Cd0; 100 mg kg −1 , Cd100). Plant biomass, antioxidant enzyme activities, peroxide content, Cd uptake, and rhizosphere bacterial community composition were evaluated. AMF inoculation reduced Cd influx in aboveground tissues, enhanced nutrient availability in the rhizosphere, and mitigated Cd biotoxicity. Additionally, AMF inoculation improved the scavenging efficiency of reactive oxygen species and alleviated oxidative stress in L. perenne , thereby mitigating biomass reduction. Moreover, AMF treatment increased leaf and root biomass by 342.94% and 41.31%, respectively. Furthermore, under the same Cd concentration, AMF inoculation increased bacterial diversity (as measured by the Shannon index) and reduced bacterial enrichment (as indicated by the ACE index). AMF promoted the enrichment of certain bacterial genera (e.g., Proteobacteria and Actinobacteria ) in the Cd100 group. These findings suggest that AMF regulated the composition of the rhizosphere bacterial community and promoted the growth of potentially beneficial microorganisms, thereby enhancing the resistance of L. perenne to Cd stress. Cd contamination in soil severely limits plant growth and threatens ecosystem stability, highlighting the need to understand how AMF and rhizosphere microbes can enhance Cd tolerance in L. perenne . Therefore, inoculating plants with AMF is a promising strategy for enhancing their adaptability to Cd-contaminated soils.",
"conclusion": "5. Conclusions This study indicated that AMF inoculation significantly enhanced the growth and stress resistance of L. perenne in Cd-contaminated soils. AMF inoculation improved plant biomass, promoted antioxidant enzyme activities, and regulated the composition of the rhizosphere bacterial community. Additionally, AMF treatment increased nutrient availability in the rhizosphere soil and reduced Cd toxicity in the aboveground parts of the plant, thereby promoting Cd accumulation in the roots. Moreover, AMF facilitated the enrichment of beneficial microorganisms, further enhancing plant resistance to Cd stress. These results suggest that AMF inoculation is an effective method for improving plant adaptability to Cd pollution. Further studies are needed to identify rhizobacterial species that are more responsive to host plants for plant stability. Systematic studies are also required to understand host–AMF compatibility and the mechanisms by which AMF contributes to enhanced plant stability.",
"introduction": "1. Introduction Heavy metal contamination, particularly from cadmium (Cd), poses significant global environmental hazards [ 1 ]. Among heavy metals and metalloids, Cd exhibits the highest percentage of soil samples (7.0%), exceeding the limit specified by the Ministry of Environmental Protection [ 2 ]. Notably, Cd exhibits strong mobility in the biosphere, posing significant risks to both soil and plants [ 3 , 4 ]. Cd, mainly existing in its divalent state, is released into the environment from various sources, including human activities, industrial emissions, volcanic eruptions, and other natural processes [ 5 ]. The high mobility of Cd in soil facilitates easy absorption by plant roots, leading to rapid translocation throughout the plant and significant accumulation in the aboveground parts of vegetation [ 6 ]. This accumulation can reduce plant biomass and lead to Cd buildup in plants [ 7 ]. Therefore, addressing Cd pollution is crucial. Arbuscular mycorrhizal fungi (AMF) can enhance plant adaptability to challenging environmental conditions through both direct and indirect mechanisms. AMF directly forms an extensive mycelial network around the roots of the host plant, creating a reciprocal exchange system in which plant-derived carbon is traded for essential nutrients such as nitrogen and phosphorus. This symbiosis enables AMF to utilize 4% to 20% of the total photosynthetic output of plants to facilitate their metabolic processes, thereby promoting plant growth and development in a positive feedback loop [ 8 ]. To indirectly mitigate heavy metal toxicity, AMF release specific effector proteins and peptides that reduce the mobility and bioavailability of heavy metals in the soil. This process modulates the uptake and distribution of heavy metal ions in both plants and the surrounding environment [ 9 , 10 ]. AMF sequester significant amounts of Cd within their spores and vesicles, thereby effectively reducing environmental heavy metal concentrations, limiting Cd translocation from roots to aboveground tissues, and alleviating heavy metal stress on plants [ 11 ]. Additionally, AMF enhance rhizosphere community structure and diversity, strengthen interspecies interactions, and attract beneficial microbiota to optimize nutrient acquisition from the soil, further promoting plant growth [ 12 ]. Lolium perenne L. is a vital pasture and forage species commonly found in temperate regions and native to Europe and was introduced into China as a lawn grass in the 1930s [ 13 ]. The L. perenne used in this study is a high-quality cool-season turfgrass variety that is widely cultivated in urban areas of temperate and cool climate regions in China, including the northeast, southwest, and eastern regions [ 14 ]. L. perenne , a perennial monocot in the Poaceae family, is highly valued for its ease of cultivation, high biomass yield, excellent regenerative capacity, and strong resistance to pests and diseases [ 15 ]. The economic and ecological significance of L. perenne is well recognized, particularly for its role as a vital turfgrass and green manure crop [ 16 ]. Notably, ryegrass exhibits a high tolerance to heavy metals and can accumulate these contaminants, making it an ideal candidate for phytoremediation, particularly in turf management on golf courses and in heavy-metal polluted areas [ 17 , 18 ]. This study investigated the effectiveness of AMF, specifically Rhizophagus irregularis , in mitigating plant growth inhibition in heavy metal-contaminated soils [ 19 ]. Previous studies have shown that Rhizophagus irregularis reduces the toxicity of Cd in the soil, thereby altering the rhizosphere soil microbial community, alleviating Cd-induced damage to rice, and enhancing plant tolerance to heavy metal stress. In this study, pot experiments were conducted to artificially inoculate Rhizophagus irregularis to evaluate its potential impact on the rhizosphere microbial community and its ability to reduce Cd toxicity in L. perenne . Plant growth and ectopic Cd accumulation were measured to assess the responses of the rhizosphere microbial community.",
"discussion": "4. Discussion Cd is a highly toxic heavy metal with high mobility in soil, which poses a significant risk of soil pollution, particularly in northeast China [ 41 ]. Cd accumulation can significantly reduce soil nutrient content, thereby inhibiting plant growth. This study indicates that the contents of key soil nutrients, including NN, AN, P, and K, significantly decreased under Cd contamination. This reduction may be attributed to Cd binding with these nutrients, forming stable chemical compounds that hinder plant absorption [ 42 ]. Additionally, Cd can disrupt the protein structure of enzymes, further reducing the availability of vital nutrients in the soil and leading to a decline in soil quality [ 43 ]. Conversely, under Cd contamination, the soil organic carbon content significantly increased owing to the secretion of organic metabolites by plant roots, which promotes the growth and reproduction of soil microorganisms [ 44 ]. These microorganisms enhanced the nutritional status of the microenvironment, promoting the formation of complex microbial communities [ 44 ]. Yuan, et al. [ 45 ] found that Cd-contaminated soils harbored Cd-resistant microbial communities. These microorganisms utilized mechanisms such as biosorption and bioconcentration to effectively remove large amounts of Cd from the soil. Moreover, the microorganisms can transform Cd into non-toxic or less toxic forms [ 46 ]. Furthermore, the microorganisms released organic matter during their metabolic processes, contributing to soil organic matter accumulation and enhancing the soil microenvironment [ 47 ]. The toxic effects of heavy metals can disrupt the metabolic activities and growth regulation of plants, which affects their growth rate and seedling development [ 48 ]. Research has shown that high Cd concentrations in soil can hinder plant growth and reduce physiological functions [ 49 ]. This study found that AMF inoculation significantly reduced Cd content in the aerial parts of L. perennial but increased Cd accumulation in the roots and rhizosphere soil, thereby promoting plant growth. Cd exposure significantly increased MDA and H 2 O 2 levels in plants, leading to oxidative stress. The Cd-treated control group (NMCd100) exhibited significantly higher MDA levels. However, the AMCd100 group had lower MDA levels, indicating that AMF inoculation mitigated Cd-induced lipid peroxidation. Similarly, the Cd-treated control group exhibited significantly higher H 2 O 2 levels, but AMF-treated groups had significantly lower H 2 O 2 levels [ 50 ]. The extensive network of intra- and extraradical hyphae formed by AMF directly sequestered Cd, which significantly limited its upward translocation in plants [ 51 ]. Additionally, AMF produced glomalin-related soil protein, a glycoprotein that binds and stabilizes metals, which facilitated Cd immobilization within the rhizosphere soil and root tissues, thereby reducing its transfer to the shoots [ 52 ]. The significant increase in BCF and decrease in TF further supported this mechanism. AMF inoculation reduced Cd concentration in the aerial parts, increased biomass, and alleviated oxidative stress, thereby effectively mitigating Cd toxicity in plants. Under Cd contamination, the rhizosphere soil microbial community at the phylum level exhibited significant changes after inoculation with AMF ( Figure 4 ). Notably, Actinobacteriota emerged as the dominant phylum across all treatment groups, particularly in the AMF-inoculated groups (AMCd0 asnd AMCd100), with a significant increase in their relative abundance [ 8 ]. A previous study has revealed that a combination of actinomycetes and mucor can clean up contaminated soil, particularly zinc, lead, and manganese compounds [ 52 ]. At the genus level, Phenylobacterium , Pullulanibacillus , Jatrophihabitans , and Marmoricola were significantly enriched in the AMCd100 group. These genera may promote growth mechanisms in plants, such as nutrient absorption and water balance, which can enhance plant tolerance and promote root system development, potentially mitigating the effects of Cd contamination [ 53 ]. However, microbial interaction can be complex, with the effects on plants varying based on environmental factors and competitive or symbiotic relationships [ 54 ]. Therefore, in applying beneficial microorganisms such as AMF to improve plant adaptability to Cd contamination, it is crucial to consider various factors such as soil conditions and plant varieties to achieve optimal outcomes [ 55 ]. For example, studies have shown that different soil conditions can influence the effectiveness of AMF-plant interactions in Cd-contaminated environments. In acidic soils, AMF can significantly enhance the Cd tolerance of plants and reduce Cd accumulation in plant tissues, but this effect may be weaker in alkaline soils [ 56 ]. Moreover, plant species exhibit varying levels of Cd tolerance after AMF inoculation. For example, maize varieties differ in their Cd uptake and utilization after treatment with AMF [ 57 ]. This highlights the importance of selecting suitable plant species to enhance the effectiveness of AMF applications. Symbiotic network analysis revealed that AMF inoculation (AMCd0 and NMCd100) reshaped the rhizosphere microbial community. The results indicated that AMF inoculation promoted an increase in plant-associated growth-promoting bacteria, which provided vital organic compounds to AMF. This interaction created a more favorable environment for both the fungi and the plant. Consequently, the synergistic relationship between AMF and rhizosphere microorganisms stimulated root growth in L. perenne , enhanced nutrient uptake from the soil, and contributed to an overall improvement in plant growth rate and quality [ 54 , 58 ]. Mental tests and SEM analysis revealed that plant physiological and soil nutritional indicators significantly influenced the abundance and distribution of microbial species. SEM results indicated the complex relationships between AMF infestation, soil nutrients, bacterial communities, plant growth, and Cd uptake and transport. AMF inoculation significantly promoted bacterial diversity and improved soil quality, thereby enhancing Cd uptake by plants. However, AMF directly reduced Cd translocation through mechanisms such as chelation or sequestration. Improved soil quality positively affected plant growth, while bacterial communities had a minimal negative impact on plant growth owing to resource competition [ 59 ]. Overall, AMF played a critical role in regulating both microbial interactions and Cd dynamics, which aided in mitigating Cd contamination and improving nutrient uptake. The synergistic effects of AMF and rhizosphere microorganisms can significantly enhance nutrient uptake, promote growth under adverse conditions, and improve the tolerance of L. perenne to pollution."
} | 3,497 |
38970122 | PMC11225162 | pmc | 869 | {
"abstract": "Background Despite rapid advances in genomic-resolved metagenomics and remarkable explosion of metagenome-assembled genomes (MAGs), the function of uncultivated anaerobic lineages and their interactions in carbon mineralization remain largely uncertain, which has profound implications in biotechnology and biogeochemistry. Results In this study, we combined long-read sequencing and metatranscriptomics-guided metabolic reconstruction to provide a genome-wide perspective of carbon mineralization flow from polymers to methane in an anaerobic bioreactor. Our results showed that incorporating long reads resulted in a substantial improvement in the quality of metagenomic assemblies, enabling the effective recovery of 132 high-quality genomes meeting stringent criteria of minimum information about a metagenome-assembled genome (MIMAG). In addition, hybrid assembly obtained 51% more prokaryotic genes in comparison to the short-read-only assembly. Metatranscriptomics-guided metabolic reconstruction unveiled the remarkable metabolic flexibility of several novel Bacteroidales -affiliated bacteria and populations from Mesotoga sp. in scavenging amino acids and sugars. In addition to recovering two circular genomes of previously known but fragmented syntrophic bacteria, two newly identified bacteria within Syntrophales were found to be highly engaged in fatty acid oxidation through syntrophic relationships with dominant methanogens Methanoregulaceae bin.74 and Methanothrix sp. bin.206. The activity of bin.206 preferring acetate as substrate exceeded that of bin.74 with increasing loading, reinforcing the substrate determinantal role. Conclusion Overall, our study uncovered some key active anaerobic lineages and their metabolic functions in this complex anaerobic ecosystem, offering a framework for understanding carbon transformations in anaerobic digestion. These findings advance the understanding of metabolic activities and trophic interactions between anaerobic guilds, providing foundational insights into carbon flux within both engineered and natural ecosystems. \n Video Abstract Supplementary Information The online version contains supplementary material available at 10.1186/s40168-024-01830-z.",
"introduction": "Introduction Anaerobic digestion is regarded as a representative engineered biotechnology that contributes to creating a circular economy and combatting climate change by organic waste resource recovery and renewable methane production. The global operation of approximately 132,000 digesters (≥ 100 m 3 tank size) and the growth of biogas-based electricity generation from 64,854 GWh in 2012 to 96,565 GWh in 2020, as reported by the World Gas Association and The International Renewable Energy Agency, highlight the significant global interest of anaerobic digestion biotechnology [ 1 , 2 ]. Within methanogenic environments, multiple microbial trophic guilds collaborate to convert organic matters into methane and carbon dioxide, playing a pivotal role in anaerobic carbon flux across both natural ecosystems and engineered bioreactors [ 3 ]. Despite the widespread application of anaerobic digestion and well-known theories about a cascade of anaerobic degradation steps, knowledge about the uncultured anaerobes, including their genetic diversity and ecological functions, is still relatively scarce not only due to their vast phylogenetic and metabolic diversity but also technical limitations, e.g., challenges in cultivating slow-growing anaerobes [ 4 , 5 ]. An improved understanding of the metabolic capability of uncultured anaerobes and their adaptations to environmental shifts will contribute to optimizing operational strategies of anaerobic waste treatment and deciphering the black box of anaerobic carbon transformation and the global carbon flux [ 6 ]. In recent years, genome-centric metagenomics has extensively been used to explore intricate microbial communities, offering crucial insights into the potential ecological functions of microbial populations within anaerobic environments [ 3 , 7 ]. However, despite advancements in sequencing technologies and bioinformatic workflows, short-read-based genome-resolved metagenomics is still subject to substantive limitations partly owing to the challenges in reconstructing complete/high-quality genomes and retrieving comprehensive genetic information [ 8 ]. Long-read sequencing approaches provide unique solutions for metagenomic assembly challenges by recovering rRNA operons and spanning the long repetitive regions [ 9 , 10 ], enabling the retrieval of complete and previously unexplored metagenome-assembled genomes (MAGs) in diverse environments, e.g., activated sludge ecosystem [ 11 ] and rumen [ 12 ]. Several investigations have reported large collections of biogas microbiomes by short-read-based metagenomics. Ma et al. assembled 2426 draft MAGs from 56 full-scale biogas plants in China [ 13 ]. And Campanaro et al. retrieved 1401 archaeal and bacterial genomes derived from 134 public metagenomes from various biogas reactors [ 14 ], which was further complemented to 4568 non-redundant anaerobic species by integrating 192 additional datasets [ 15 ]. However, only 108 MAGs were shared in the collections between Ma et al. and Campanaro et al. [ 13 ], highlighting high portion of undiscovered anaerobes inhabiting various methanogenic environments and necessity of recovering the microbial wealth from anaerobic communities by long-read metagenomics. Given the complicated nature of engineered and natural methanogenic environments, the majority of anaerobic digestion studies focused on a specific facet of the microbial communities, e.g., the methanogenic stage, using simplified model systems [ 16 , 17 ] or long-term artificial enrichment of substrate specification experiments, such as syntrophic consortia enrichment [ 6 , 18 , 19 ]. Such pioneering studies provided important knowledge about the metabolic function of different flora and their metabolic trophic relationships. However, it is crucial to recognize that both natural and engineered ecosystems consistently harbor a complex mixture of different substrates, rather than simple ones. These substrates collectively determine the ecological niches of microbial populations and significantly influence their metabolic interactions. Therefore, to capture functionally important species and uncover their trophic interactions in complex methanogenic habitats, it is vital to incorporate individual-level functional assignment and community-level carbon mineralization routes. Up to date, few attempts have been made to unravel a genome-wide understanding of carbon flow from sugars and amino acids (AAs) to central carbon metabolism, from syntrophic oxidation of long-chain fatty acids (LCFA) to short-chain fatty acids (SCFA), and finally to methanogenesis within the complex anaerobic food web [ 20 – 22 ]. Furthermore, prior attempts in exploring the metabolic potential of keystone species were mainly based on the presence of targeted pathways or biomarker genes [ 22 – 24 ], with fewer investigations harnessing multi-omics approaches to provide expression-based evidence regarding the in situ metabolic activity of highly active microbes and their biogeochemical functions in intricate ecosystems. In this study, we investigated in situ metabolic activities and trophic interactions of uncultivated anaerobes by utilizing a lab-scale anaerobic bioreactor as a methanogenic ecosystem. Using metatranscriptomics-guided metabolic reconstruction, we unraveled a community-level carbon flow based on high-quality genomes reconstructed by the hybrid assembly using short and long reads. Through this work, we connected microbial community structures to the functional potentials of individual populations, identified novel keystone lineages with previously undescribed functions, as well as pinpointed specific genomic characteristics of active anaerobic lineages helping them stand out from substrate competition and niche dominance. Our study improves the understanding of metabolic underpinnings and trophic interactions between uncultivated key anaerobic guilds, providing a foundational framework to connect the metabolic attributes of uncultured anaerobic lineages with their ecological function within the intricate anaerobic ecosystems.",
"discussion": "Discussion Studying the metabolic activity and trophic interactions among microbial communities during various stages of carbon mineralization offers insights into their contributions to the global carbon cycle, spanning implications from biotechnological management to efforts in addressing climate change. Through metatranscriptomics-guided genome-scale metabolic reconstruction, this study offered a genome-wide insight into carbon mineralization flow within the methanogenic community in details, revealing metabolic functions from the whole anaerobic community to the specific active populations, including some novel uncultured syntrophic fatty acid-oxidizing bacteria (i.e., new Smithellaceae and Syntrophales lineages) and bacteria with previously undescribed functions (e.g., sugar- and AAs-scavengers within family VadinHA17 and Bacteroidales ). By harnessing the long reads advantage, we recovered eight circular genomes, including two previously recognized but highly incomplete SPOB, yet-uncharacterized fermenters with versatile metabolism ( Petrimonas sp.bin.267, etc.) and a genome from yet-to-be underexplored phylum ( Caldisericota bin.236), advancing the holistic comprehension of anaerobic keystone species from the whole genomic perspective. In this study, we employed a 100% pathway completeness threshold (except 90% for MMC pathway) to mitigate inaccurate functional assignments. Our approach involves assessing the complete pathway transcriptional activity to pinpoint pivotal contributors in the targeted functional guild. The utilization of complete pathway transcriptional activity could address the deficiency of the mcr -biomarker comparison method in differentiating between hydrogenotrophic and aceticlastic methanogenesis. Furthermore, our finding emphasized the significance of metatranscriptomics in identifying the ecological functions and in situ metabolism of uncultured species. Representative example is Syntrophobacteraceae bin.487, it encodes the reversed Wood–Ljungdahl pathway, MMC pathway and beta-oxidation for acetate, propionate, and butyrate degradation, yet our study indicates significant transcriptional activity primarily in the MMC pathway within bin.487, consistent with the prior research classifying it as a SPOB based on isolated pure culture experiments [ 60 ]. Overall, our study establishes a framework for elucidating carbon transformations in anaerobic microbiota, linking genomic pathways of uncultured microbes with their ecological functions, and evaluating the significance of certain microbes within functional guilds via comparisons of targeted pathway activities. This framework could undoubtedly be applied to uncover novel microbes and trophic interactions in other engineered and natural ecosystems, offering significant insights into biotechnologies and global biogeochemistry. In the phylogenetically and functionally diverse methanogenic community, the metabolic repertoire demonstrates a high level of functional redundancy across various phyla [ 65 ]. For example, Bacteroidota -related bacteria (bin.267, bin.202, and bin.334), along with Thermotogota -affiliated species (bin.190 and bin.512) and members from Synergistota (e.g., Thermovirgaceae bin.148) showed metabolic flexibility in scavenging varied AAs and sugars to generate metabolites (such as glucose, β -D-fructose-6P, and pyruvate) for downstream central carbon metabolism. Such a high degree of functional redundancy contributes to greater system stability, particularly in fluctuating environments with high microbial density and significant turnover of dead biomass, e.g., in anaerobic digesters with rapidly increased organic loadings. Despite having comparable metabolic capacities, different taxonomic lineages utilized functional adaptation strategies, i.e., distinct transport mechanisms and substrate preference, to mitigate fierce competition. Thermotogota member bin.190 transcriptionally expressed genes encoding ABC-type transporters to catalyze simple (K02056-K02058) and multiple sugar transport (K02025-K02027) as well as AA uptake, differentiating them from Bacteroidota -affiliated degrader bin.267 that utilizes the unique TonB/SusC system to facilitate sugars and AAs across the outer membrane (Supplementary Data 10). And substrate-preference-driven interactions were also observed among the fermenters from different families within Bacteroidota phylum, as revealed by expression comparisons of overall pathway activities. That is, Dysgonomonadaceae bin.267 and VadinHA17 bin.202 favored glucose cleavage by the complete EMP pathway, whereas Bacteroidales bin.334 (f__4484-276) exhibited a preference for downstream pyruvate fermentation. These findings advanced the understanding how functionally diverse anaerobic populations evade substrate competition and create niche specialization, thereby forming a cohesive metabolic network within the intricate methanogenic communities. In addition, our findings highlight that the active anaerobic lineages inhabiting nutrient-rich bioreactors exhibit remarkable metabolic flexibility and plasticity, representing an additional metabolic trait that contributes to the stability of the methanogenic community. Strong evidence of such adaptability was observed in predominant syntrophic bacterial genomes, Smithellaceae bin.332 and Syntrophales bin.292, both functioning in LCFA and propionate/butyrate metabolism. This metabolic flexibility allowed them to switch their metabolic processes responding to changes in fatty acids with different lengths and/or saturation levels in substrate. Similar to how functional redundancy across diverse phyla enhances the community-level resilience to environmental disturbances, the metabolic flexibility of anaerobic populations could provide individual lineage an adaptive survival strategy to cope with variable environmental conditions, which are also witnessed in other ecosystems, like the seafloor microbiomes [ 65 , 66 ] and the gut microbiota [ 67 , 68 ]. A crucial question revolves around the genomic features within each redundant functional guild that contribute to the prominence of dominant species. Integrating genomic feature analysis and transcriptional evidence of key enzymes, we conducted comparative analyses on active lineages within the same functional guilds, revealing the potential genomic determinants driving their substrate competition advantage and niche dominance. For the sugar- and AA-scavengers bin.267 and bin.202 within phylum Bacteroidota , their distinction from competitors could be attributed to rapid cross-membrane transport, supported by the presence and high transcriptional expression of more susC/susD homolog genes for protein and carbohydrate hydrolysate uptake and transport (Supplementary Fig. 8). And Smithellaceae bin.332 could stand out from LCFA-degrading function, primarily benefiting from exceptional bacterial motility (pilus and flagellar assembly), chemotaxis, rapid substrate-activating mechanism, and adaptive cellular degradation (Supplementary Fig. 9). The breakdown of LCFA is assumed to be adsorbed and attached by functional taxa [ 54 , 69 ]. In comparison to other active LCFA-degraders, Smithellaceae bin.332 exhibited higher transcriptional expression of chemotaxis genes ( wspABCDEF ), type IV pilus assembly ( pilABCMNOPQTVWX ), and flagellar assembly ( flhG and flgP ), enhancing its mobility and attachment to LCFA molecules. Bin.332 also harbors diverse homologous for each step of beta-oxidation pathway, e.g., 15 genes for long-chain acyl-CoA synthetase ( ACSL ), allowing this syntrophic bacterium adaptive to varied types of fatty acid and/or fluctuating fatty acid concentrations in bioreactors with increased organic loadings due to differing kinetics and/or affinities of the homologous genes [ 6 , 70 ]. Regarding methanogen, Methanomicrobiales bin.74 possessed 24 genes encoding V/A-type ATPase and four sets of membrane-bound hydrogenases ( Ech , Eha , Ehb , and Mbh ), a notably higher count compared to those in the bin.206 genome (9 ATPase and one hydrogenase) (Supplementary Data 12). The three clusters of alternative ATPase may confer adaptive advantages for varying growth rates or distinct concentrations of Na + /H + in eutrophic habitats, e.g., bioreactors [ 70 ]. These specific energy-related features of bin.74 contribute to its dominance at archaeal community under low organic loading to some extent, but the extreme amount of acetate in fresh leachate substrate at high organic loading leads to a rapid increase in the activity of bin.206 that exhibits a preference for acetate utilization. Incorporating the above discoveries, the specific genetic patterns and environmental determinants foster niche specialization and ecological dominance among anaerobic community, enhancing our understanding of environmental adaptation of bacterial and archaeal lineages."
} | 4,324 |
39888291 | PMC12005752 | pmc | 870 | {
"abstract": "Abstract In this work, the regulation of liquid self‐transport is achieved through architectural and thermal coupling, transitioning from free surfaces to open channels. Hierarchical structures inspired by the skin of a Texas horned lizard are designed, with the primary structure of wedged grooves and the secondary structure of capillary crura. This design enables advantages including long‐distance self‐transport, liquid storage and active reflux management on free surfaces, directional transportation, synthesis and detection of reagents in confined spaces, as well as controllable motion and enhanced heat dissipation in open channels. The regulation capacity can be precisely controlled by adjusting the secondary capillary crura and external thermal gradients. The regulation mechanism is further elucidated through microscopic flow observation and a deduced theoretical model. The proposed structures are expected to introduce a new concept for designing lubrication systems, microfluidic chips, methods for chemical synthesis, and heat transfer in the future.",
"conclusion": "3 Conclusion This study introduces an innovative concept for liquid self‐transport, leveraging architectural‐thermal coupling to transition from free surfaces to open channels. We designed and optimized hierarchical structures composed of primary wedged grooves and secondary capillary crura. These hierarchical structures effectively facilitate the movement of lubricants with varying viscosities and surface tensions, enabling several key functions. Long‐distance self‐transport: the design allows for efficient transport of liquids over considerable distances, addressing challenges where traditional methods may fall short. Liquid storage and active reflux management: the structures enable effective liquid storage and management of reflux on free surfaces, enhancing overall system efficiency. Directional transportation: the design supports the directional flow of liquids, which is crucial for applications such as microfluidics and chemical synthesis. Reagent synthesis and detection: the capability to control liquid flow in confined spaces provides opportunities for reagent synthesis and precise detection. Controllable motion in open channels: with the secondary structures in the form of capillary crura, we can regulate liquid movement within open channels, allowing the system to adapt to various operational conditions. Enhanced heat dissipation: compared to a normal channel, the optimized surface pattern in the open channel can enhance cooling capacity. The regulatory mechanism governing these functions was further elucidated through microscopic flow observations and theoretical modeling, which deepened our understanding of the underlying principles. Experimental validation underscores the practicality of these hierarchical structures for applications in lubrication under starved conditions, chemical reagent synthesis, efficient transportation, and enhanced heat transfer within open channel systems.",
"introduction": "1 Introduction In nature, liquid transport facilitates the transfer of mass and energy between nonliving and living matter. [ \n \n 1 \n , \n 2 \n , \n 3 \n , \n 4 \n , \n 5 \n , \n 6 \n , \n 7 \n , \n 8 \n , \n 9 \n \n ] In recent decades, controllable liquid transport has rapidly advanced due to its significant potential in scientific and engineering applications such as microfluidics, [ \n \n 10 \n \n ] chemical synthesis, [ \n \n 11 \n \n ] energy recovery, [ \n \n 12 \n , \n 13 \n \n ] enhanced heat transfer, [ \n \n 14 \n , \n 15 \n \n ] and lubrication systems. [ \n \n 16 \n \n ] The Marangoni effect, [ \n \n 17 \n \n ] where an unbalanced surface tension gradient in a liquid causes motion from low to high‐tension regions, is key to this. Researchers have developed two main approaches for regulating liquid self‐transport: applying external electric, [ \n \n 18 \n , \n 19 \n , \n 20 \n , \n 21 \n , \n 22 \n \n ] thermal, [ \n \n 23 \n , \n 24 \n , \n 25 \n \n ] magnetic, [ \n \n 26 \n , \n 27 \n \n ] or optical [ \n \n 28 \n , \n 29 \n , \n 30 \n \n ] fields to create a surface tension gradient, and constructing micro‐nano structures to form a wettability gradient on solid surfaces. [ \n \n 31 \n , \n 32 \n , \n 33 \n , \n 34 \n , \n 35 \n , \n 36 \n , \n 37 \n \n ] The latter is often preferred as it doesn't require additional devices. Nature offers numerous inspirations for designing micro‐nano structures. Through the processes of natural selection and evolution, many organisms have developed specialized surfaces that exhibit remarkable liquid transport capabilities to adapt to environmental changes. [ \n \n 38 \n , \n 39 \n , \n 40 \n \n ] For instance, the conical tip of a cactus captures droplets from mist and transports them from narrow to wide end. [ \n \n 41 \n \n ] The Janus structure on the tip of pine needles can merge small droplets and direct them to the needle's root. [ \n \n 42 \n \n ] The skin of the Texas horned lizard, a honeycomb‐shaped wedge array, can provide directional, passive liquid transport as a model for a biomimetic liquid diode [ \n \n 43 \n \n ] \n The driving force behind the self‐transport capabilities of these biological surfaces is the asymmetric Laplace pressure. The wedged or patterned structures create a curvature difference across opposing sides of a liquid, resulting in an unbalanced Laplace pressure that causes spontaneous flow. [ \n \n 44 \n , \n 45 \n , \n 46 \n \n ] Inspired by these natural systems, Alheshibri et al. [ \n \n 47 \n \n ] fabricated a wedge‐shaped hydrophilic region against a hydrophobic background on a solid surface and experimentally observed the liquid self‐transport process. Dai et al. [ \n \n 48 \n \n ] designed superoleophobic surfaces with wedge‐shaped superoleophilic grooves to control the thermocapillary migration. Additionally, researchers have developed theoretical models using Lattice‐Boltzmann [ \n \n 49 \n \n ] and molecular dynamics [ \n \n 50 \n \n ] to elucidate the transport mechanism. The wettability difference and the divergent angle of a wedged structure are key parameters influencing its self‐transport capability. [ \n \n 51 \n \n ] In such structures, the wedged area is typically hydrophilic, while the surrounding area is hydrophobic. However, there is a limitation to the wettability difference since the maximum difference in contact angles is less than 180°. Although increasing the divergent angle can improve self‐transport efficiency, it also enlarges the hydrophilic area, leading to significant liquid loss and a reduction in the overall transport effectiveness. These limitations of a single wedge groove become apparent when dealing with liquid transport on complex surfaces, confined spaces, or in open channels. Song et al. [ \n \n 52 \n \n ] demonstrated long‐distance transport of underwater bubbles using a continuous gradient surface by connecting multiple wedged structures. Hou et al. [ \n \n 53 \n \n ] created patterned superhydrophilic wedged grooves on a superhydrophobic surface, enabling water condensation collection against gravity. Liu et al. [ \n \n 54 \n \n ] optimized the design of connected wedged structures and developed a series of cycloidal wedge grooves to enhance water transport. Patterned wedge‐shaped structures indeed offer significant advantages for driving liquid flow. However, most research has focused on enhancing the transport of water droplets, with less attention given to self‐transport regulation for other liquids like mineral or lubricating oils, which differ in viscosity and surface tension. Accurately controlling the flow of these varied liquids remains a major challenge. Currently, the directional transport of liquids in open channels is a burgeoning area in microfluidics. [ \n \n 55 \n , \n 56 \n , \n 57 \n , \n 58 \n \n ] The flow state of the liquid in the open flow channel is different from that on the free surface. There are two more side walls in the open channel, and the liquid flow is related to the spontaneous capillary flow and the depth‐to‐width ratio of the section. [ \n \n 59 \n , \n 60 \n , \n 61 \n , \n 62 \n \n ] It remains uncertain whether patterned structures can effectively regulate liquid flow in open channels. Successfully addressing this could enable long‐distance transportation of large volumes of liquid within such channels. In this study, hierarchical structures inspired by the skin of a Texas horned lizard were proposed to regulate liquid self‐transport coupling thermal effect. The hierarchical structures can effectively drive the movement of typical lubricants. By designing an appropriate chamfer radius at the connection points of the wedged units, we were able to significantly enhance self‐transport performance. The regulation capacity could be precisely controlled by adjusting the secondary capillary crura and thermal gradient. Experimental confirmation showed their potential applications in lubrication under starved conditions, reagent synthesis and detection, long‐distance transportation, and enhanced heat dissipation in open channels. The regulation mechanism was further elucidated through microscopic flow observations. The proposed hierarchical structures are expected to introduce a new concept for designing lubrication systems, microfluidic chips, chemical synthesis, and enhanced heat transfer methods in the future.",
"discussion": "2 Results and Discussion 2.1 Long‐Distance Self‐Transport Capability of Designed Surface Three types of wedged groove patterns were fabricated, each with increasing chamfer radii (R) of 0, 0.8, and 2.5 mm at the connecting region (Design details on the parameters were provided in Figure S1a , Supporting Information). For comparison, a single wedged groove was also tested. The length (L) of the designed structure was 62.58 mm, with a divergent angle (α) of 7°. The wedged groove area was oleophilic, while the surrounding area was oleophobic, with contact angles of ≈1° and 158°, respectively. Figure \n \n 1 a illustrates the self‐transport performance using 8 µL paraffin oil droplets. The patterned wedged grooves exhibited superior self‐transport capacity compared to the single one. Chamfer radius is a positive correlation with the transport performance, thus, a chamfer radius of 2.5 mm was selected in the subsequent sections. Note that the self‐transport capacity is related to the wettability and wedged angle, [ \n \n 51 \n , \n 52 \n , \n 53 \n , \n 63 \n \n ] increasing the size of the openings at both ends of the wedge shape would not affect the self‐transport capacity, but it can transport more volume of liquid. The architectural‐thermal coupling of liquid self‐transport capacity was further validated by different lubricants. As shown in Figure 1b1 , the designed surface demonstrated controllable self‐transport for Hexadecane, 10# aviation hydraulic oil, Poly‐alfa‐olefins 4 (PAO4), PAO25, water, and silicon oil. PAO25 exhibited a shorter transport distance due to its higher viscosity (Movie S1 , Supporting Information). When a thermal gradient (ΔT = 1.7 °C mm −1 ) was applied (Figure 1b2 ), the self‐transport capacity improved significantly for all tested liquids. For instance, the transport distance of PAO25 increased from 27.17 to 46.18 mm within 35 s (Movie S2 , Supporting Information). Figure 1 a) Relationship between distance and time for on surfaces with wedged groove patterns with increasing chamfer radii (R). Performance of six different lubricants on surfaces with a wedged groove pattern (R = 2.5 mm), exhibiting a thermal gradient of (b1) ΔT = 0 °C mm −1 , and (b2) ΔT = 1.7 °C mm −1 . c) Comparison of the maximum transport distance and average velocity between this work and opening literature for water‐to‐water and oil‐to‐oil systems. Figure 1c presents a comparison of the maximum transport distance and average velocity between this work and the opening literature for water‐to‐water and oil‐to‐oil systems, specifically regarding the wedged‐like structures. It is confirmed that the proposed structures can realize a long‐distance self‐transport for different lubricants. 2.2 Capacity for Liquid Storage and Active Reflux Management The performance of liquid storage and active reflux on the designed surfaces was evaluated. As illustrated in Figure \n \n 2 a , 10 µL droplets of paraffin oil were applied to the surface every 15 s, totaling 50 µL. Without a thermal gradient, the majority of the liquid remained in the wedged grooves and did not fully flow into the pool after 240 s. However, when a thermal gradient of ΔT = 2 °C mm −1 was introduced in the divergent direction of the wedged groove, the liquid flowed rapidly and reached the pool area in just 46 s, significantly faster than under simple architectural conditions. Figure 2b shows the active reflux performance of the liquid on the designed surfaces (the reflux distance equals 62.58 mm minus the movement distance of the black arrow, as shown in Figure 2b(ii,iii) . Two thermal gradients were applied: 2 °C mm −1 in the divergent direction and −2 °C mm −1 in the convergent direction. The process can be divided into three stages. Stage I: For ΔT = ±2 °C mm −1 , the liquid reached the pools on the cold side within 41 s. Stage II: As the transport continued, liquid flowed from the wedged grooves into the pools between 41 and 70 s. Stage III: After the thermal gradients were removed at the 70s, the liquid exhibited different behaviors based on the direction of the gradient. For ΔT = 2 °C mm −1 (divergent direction), the liquid flowed back slightly. In contrast, for ΔT = −2 °C mm −1 (convergent direction), the main portion of the liquid quickly flowed back to its initial position from right to left. Figure 2 Liquid storage performance and active reflux management of the designed surfaces: a) The transport distance and detailed transportation process for liquid storage under a thermal gradient of ΔT = 0 °C mm −1 , and ΔT = 1.7 °C mm −1 . b) Active reflux management performance under thermal gradients of ±2 °C mm −1 . Based on the experiments conducted, it is confirmed that architectural‐thermal coupling offers a controllable self‐transport capacity. The designed surface effectively returns lubricants to the initial region, even when faced with a thermal gradient. This capability could be utilized to transport or collect lubricants to the designated area within advanced lubrication systems that encounter varying thermal gradients. 2.3 Directional Self‐Transport Capability in Confined Spaces Compared to straight wedged groove patterns, incorporating a deflection angle ( β ) between each wedged groove allows for directional self‐transport capacity in confined spaces. As illustrated in Figure \n \n 3 a,b , wedged groove patterns with varying deflection angles ( β ) of 180°, 135°, and 105° were fabricated. The geometric dimensions of each unit (wedged groove) remained unchanged, with the only modification being the introduction of an inner chamfer radius (R = 0.5 mm) at the junction region to facilitate the bent pattern (more details were provided in Figure S1b , Supporting Information). The results of the directional self‐transport are shown in Figure 3c . Velocity is a vector, since its direction is changing due to varying deflection angles, here, instantaneous rate (just magnitude) is used for comparison. The transport distance is inversely correlated with the deflection angle ( β ). Specifically, when comparing β = 180° to β = 105°, the time required to reach the maximum distance is reduced by 36 s. Additionally, the instantaneous rate indicates that there is no significant difference among all designed surfaces before reaching the first junction region. Figure 3 Directional self‐transport capability in confined spaces: a) Detailed transportation process on wedged groove patterns with varying deflection angles ( β ) of 180°, 135°, and 105°. b) Design schematic of a chamfered series of wedged grooves featuring a deflection angle. c) Results showing transport distance and instantaneous rate. It is interesting to see that once the liquid flowed over the deflection angle at ≈2 s, a significant difference was observed, revealing an interesting phenomenon: the liquid exhibited an acceleration process around the junction region. To understand the acceleration mechanism, the microscopic transport process of the liquid flowing through the connection region was analyzed. \n Figure \n \n 4 a illustrates the microscopic self‐transport process at the junction region. The second connection point was designated as the reference, and the moment when the liquid precursor reached the tangent of the chamfer radius was defined as 0 s. A 3D coordinate system was established to provide a perspective on the different sections of paraffin oil flowing through the junction region. Figure 4b presents a side view (X‐Z coordinate) of the detailed self‐transport process. When β equals 180°, the convex meniscus becomes increasingly flatter. In contrast, at β equal to 105°, the precursor rapidly elongates, and the level height gradually increases after passing through the junction region. The extracted contour curve of the liquid demonstrates the differences in the surface profile of the convex meniscus. From a bird's eye view (X‐Y coordinate) in Figure 4c , it is evident that the liquid precursors on these two surfaces exhibited convex menisci and were obstructed at the junction area at 0 s. For β = 180°, the continuous accumulation of liquid in the junction region hindered the transport process, resulting in only a thin film flowing through. In contrast, for β = 105°, the transport behavior near the inner chamfer was more prominent than that near the outer chamfer. This led to the elongation of the precursor, preventing liquid accumulation at the junction region. Consequently, the convex meniscus became more pronounced during the turning process, initiating an acceleration phase. Figure 4 Acceleration mechanism at the junction region: Microscopic self‐transport processes of lubricant as it flows through the junction region illustrated in a) a coordinate diagram, b) a side view, and c) a bird's eye view. d) The forces present at the junction region and e) a simplified theoretical model. f) The experimental results and theoretical predictions of the instantaneous rate. Based on the microscopic process described above, we can gain insight into the acceleration mechanism. As illustrated in Figure 4d‐i , the Laplace pressure (resulting from the convex meniscus) serves as the driving force for self‐transportation. For β = 180°, the inner and outer chamfer radii at the junction region are equal ( R 1 \n = R 2 \n ), causing the liquid to encounter equivalent pinning resistance ( F pin1 \n = F pin2 \n ) as it flows over the connection point. In contrast, for β = 105° (Figure 4d‐ii ), the inner and outer chamfer radii at the structural connection are unequal ( R 1 \n < R 2 \n ), leading to a disparity in pinning resistance. Meanwhile, the contact line of the liquid precursor becomes elongated and bent, generating additional Laplace pressure in that region. This asymmetry in pinning resistance, combined with the extra Laplace pressure, accelerates the self‐transport process. The external force ( F ) acting on the liquid can be expressed as follows:\n \n (1) \n F = F L a p + F C a p − F P i n − F f \n where F Lap \n represents the Laplace driving force (mainly attributed by the micro wedged groove, that is, the curvatures at the right and left sides), F Cap \n signifies the structure‐induced capillary driving force (mainly attributed by the micro‐nano roughness), F Pin \n indicates the pinning resistance at the triple‐phase contact line, and F T \n denotes the frictional resistance. Referring to the inserts in Figure 4e , the Laplace pressure (Δ P ) can be determined via the Young‐Laplace equation:\n \n (2) \n Δ P = γ S L 1 R a + 1 R r \n where γ SL \n is the surface tension of a liquid ( γ SL \n can be simplified to γ ), R a \n and R r \n represent the radii of curvature of the liquid. As shown in Figure 4e , the relationship between R ( x ) and θ( x ) can be written as follows:\n \n (3) \n cos θ − π 2 = l x 2 R x R x = l x 2 sin θ x \n where l ( x ) is the length of the contact line. Substituting Equation 3 into Equation 2 , the Laplace pressure (Δ P ) can be written as:\n \n (4) \n F L a p = Δ P S x = 2 γ sin θ a l a + sin θ r l r S x \n where S(x) represents the cross‐sectional area of the liquid in the direction of the y ‐axis. When considering the solid‐liquid capillary forces involved in the self‐transport process, the capillary driving force F Cap \n along the depth of the wedged groove can be determined as: [ \n \n 64 \n \n ] \n \n (5) \n F C a p = γ l a cos α 2 h \n where α represents the divergent angle, and h denotes the depth of the wedged structure. The pinning resistance at the triple‐phase contact line, denoted as F Pin \n , can be calculated using the Furmidge equation. [ \n \n 65 \n \n ] :\n \n (6) \n F P i n = γ l a cos θ a − cos θ r \n where θ \n a \n and θ \n r \n represents the advancing and receding contact angle, respectively. The frictional resistance ( F f \n ) can be written as:\n \n (7) \n F f = 3 μ L l a h 3 cos α 2 \n where μ represents the viscosity, L represents the length of the hydrophilic region. Substituting Equations (4) , (5) , (6) , (7) into Equation 1 , the external force ( F ) acting on the liquid can be obtained:\n \n (8) \n F = 2 γ l r sin θ a + l a sin θ r l a l r S x + γ l a cos α 2 h − cos θ a + cos θ r − 3 μ L l a h 3 cos α 2 \n \n Given that the flow rate is low (indicating a small Reynolds number) and the liquid has low viscosity, the self‐transport process can be considered laminar flow. Therefore, the inertia term in the Navier‐Stokes equation can be neglected. Focusing on the directional movement along the x ‐axis, the simplified 2D Navier‐Stokes equation can be expressed as follows:\n \n (9) \n ∂ 2 u ∂ y 2 = 1 μ ∂ P ∂ x \n \n Combing Equations ( 8 ) and ( 9 ), the theoretical instantaneous rate ( u ) can be obtained by integrating Equation 9 twice with the boundary conditions ( u = 0 at x 1 \n = 0, y 1 \n = 0 and x 2 \n = 62.58 mm, y 2 \n = 2.5 mm):\n \n (10) \n u = 1 2 μ y 2 x F S x + C e x p y \n where C exp \n is the coefficient related to the initial volume of liquid, y represents the length of the contact line. By substituting all variables into Equation 10 , the instantaneous rate ( u ) can be calculated. As illustrated in Figure 4f , although the amplitude of the experimental results is slightly higher than that of the theoretical predictions, the general trends of the experimental and theoretical instantaneous rates ( u ) are remarkably similar. By defining the time at which the liquid reaches the second junction region as 0 s, the bent structure induces an acceleration effect that is observable within the time interval of 0–2 s. 2.4 Synthesis and Detection of Reagents in a Confined Space \n Figure \n \n 5 a illustrates the potential application for amino acid chromogenic reaction in a limited space with the designed “double halberd” structures. 80 µL ninhydrin aqueous solution (4 wt.%, glacial acetic acid: water = 100:3) was used as the detection solution (Movie S3 , Supporting Information). The areas marked ① and ② were randomly dropped 30 µL liquids of proline (1 wt.%) and glycine (1 wt.%) aqueous solutions. As shown in Figure 5a1 , the detection solution was rapidly transported to the detection areas within 6 s, but there was no color change until 30 s. When the surface was heated to 75 °C (Figure 5a2 ), obvious color changes appeared at the detection areas at 10 s, and the colors were stable after 30 s. It can be confirmed that purple area ① was glycine solution and yellow area ② was proline solution, respectively. Figure 5 Synthesis and detection of reagents in a confined space: amino acid chromogenic reaction at room temperature a1) and heated to 75 °C a2) on surfaces with “double halberd” structures. b‐i) Design diagram of the bent structure with regions ①‐④ marked for detection. b‐ii–v) Flow charts depicting the transportation of the blue indigo disulfonate sodium‐glucose solution to the four designated detection areas. Figure 5b illustrates another demonstrative experiment in a limited space for reagent synthesis and detection. Notably, the sodium indigo disulfonate‐glucose aqueous solution will exhibit different colors when mixed with alkaline liquids of varying pH levels, serving as a demonstrative experiment for acid‐base identification. In Figure 5b‐i , the diagram of the designed structure was presented, with the areas marked ①, ②, ③, and ④ representing four different alkaline liquids to be tested. Initially, ultrapure water (colorless), a NaOH aqueous solution (colorless, pH = 12.4), and another NaOH aqueous solution (colorless, pH = 14) were randomly added to three of the areas (①, ②, and ③) designated for detection. As a control, the sodium indigo disulfonate‐glucose aqueous solution (blue, at pH < 11) was added to area ④, as shown in Figure 5b‐ii . Subsequently, 60 µL of the blue sodium indigo disulfonate‐glucose aqueous solution was applied at the entrance. After ≈10 s, the detection solution self‐transported to the four designated areas, as illustrated in Figure 5b‐iii . Following the reaction process, the results were as follows: Region ① displayed a light blue color, indicating ultrapure water (pH < 11); Region ② turned light green, representing a NaOH aqueous solution with a pH of 12.4 (11.4 < pH < 13); and Region ③ became yellow, corresponding to a NaOH aqueous solution with a pH of 14 (pH > 13). These results are shown in Figure 5b(iv,v) . This indicates that the proposed structures can effectively facilitate reagent synthesis and detection within a limited space, especially when the reaction requires to be heated. These break through the limitations of limited space and the inability to heat the reaction. 2.5 Controllable Self‐Transport Capacity of Hierarchical Structures Patterns The structures mentioned above demonstrated excellent self‐transport capacity, this capacity could not be precisely controlled as needed. To achieve this goal, wedged grooves were regarded as the primary structure, and secondary structures in the form of capillary crura were fabricated around the wedged grooves. As illustrated in Figure \n \n 6 a,b , capillary crura patterns with different orientation angles ( φ ) of 150°, 90°, and 30° were prepared, while keeping all other parameters constant. The self‐transport capacity of 8 µL silicone oil on the hierarchical structures was evaluated with and without external thermal gradients. Single wedged grooves exhibited the fastest self‐transport distance when ΔT = 0 °C mm −1 (Figure 6b ). The inclusion of capillary crura patterns allowed for controllable self‐transport performance, with controllability ranking in descending order for φ = 150°, 90°, and 30°. When subjected to a thermal gradient (ΔT = 1.8 °C mm −1 ), all corresponding self‐transport distances increased; however, the overall trends and controllability of the designed hierarchical structures remained unchanged, as shown in Figure 6c (Movie S4 , Supporting Information). Figure 6 Controllable self‐transport capacity of hierarchical structure patterns: a) Design schematic of wedged grooves featuring capillary crura. Detailed self‐transport distance and transportation process on surfaces with capillary crura with orientation angles ( φ ) of 150°, 90°, and 30°, under two different thermal gradients: b) ΔT = 0 °C mm −1 , and c) ΔT = 1.8 °C mm −1 . The experimental results demonstrated that designing capillary crura provides an effective method for controllable self‐transport capability. The overall self‐transport distances ranked in descending order as follows: structures without capillary crura, followed by those with capillary crura at φ = 150°, 90°, and 30°. 2.6 Long‐Distance Directional Self‐Transport in Open Channels The hierarchical structure patterns can be further utilized for directional self‐transport in open channels. Since silicone oil has the lowest surface tension, it was chosen in the following experiments to better reflect the influence of the designed structures. As illustrated in Figure \n \n 7 a,b , single wedged groove, and single wedged groove decorated with capillary crura ( φ = 30°) were prepared, along with micronano structures (after a boiling process) and PFDTS modification (after low surface energy treatment) as control groups. The maximum transport distance ( L target \n ) achieved was 34 mm, utilizing 35 µL of silicone oil. In the open channel with a single wedged groove, silicone oil was able to transport directionally and rapidly forward to 34 mm within 12 s (blue line, Figure 7b‐ii ). Decorating the groove with capillary crura (φ = 30°) resulted in a reduced flow rate (black line, Figure 7b‐iv ). The silicone oil spread in both directions within the channel containing micronano structures (red line, Figure 7b‐i ), while it remained stationary in the PFDTS‐modified channel (yellow line, Figure 7b‐iii ) (Movie S5 , Supporting Information). Figure 7 The transport distance and processes observed on various surfaces: a,b) Four distinct surfaces are analyzed, including micronano structures, PFDTS modification, single wedged groove, and single wedged groove featuring capillary crura ( φ = 30°). c,d) Two surfaces are examined, highlighting the wedged groove pattern and wedged groove pattern with capillary crura ( φ = 90°).,e) Self‐transport capacity in open channels evaluated via a dimensionless number ( C ). Note that single wedged grooves have structural limitations; decreasing the divergent angle can extend their length but at the expense of weakening their self‐transport capacity. Therefore, hierarchical structure patterns were employed for long‐distance transportation in open channels. As shown in Figure 7c,d (Movie S6 , Supporting Information), both wedged groove patterns and wedged groove patterns with capillary crura ( φ = 90°) were prepared, achieving a maximum transport distance ( L target \n ) of 62.58 mm. The results confirm that the silicone oil can reach the target transport distance of 62.58 mm within 24 s (purple line, Figure 7d‐i ), and the addition of capillary crura ( φ = 90°) provides the capability for flow rate control (red line, Figure 7d‐ii ). The long‐distance directional self‐transport capacity in open channels is evaluated by involving a dimensionless number, C , which is defined as C = | L m a x | − | L m i n | L t a r g e t (For a specific liquid, L max \n and L min \n refer to the maximum and minimum values in the y ‐axis in Figure 7a,c ). As illustrated in Figure 7e , the wedged groove pattern demonstrates the best long‐distance directional self‐transport capacity and controllable flow rates can be achieved by incorporating capillary crura. 2.7 Enhanced Heat Dissipation in Open Channels The open channel with an optimized surface pattern can enhance the heat dissipation effect of the liquid in the channel. As shown in Figure \n \n 8 \n , a constant power heating plate is used to heat the bottom 3/4 of the open channel, maintaining its temperature at 62 °C. The aluminum plate is tilted at 2° to increase the flow rate of the liquid, and a J‐type thermocouple is used to measure the temperature on its surface. A micro constant flow pump and three syringes are used to inject cooling water at a constant rate of 15° into each channel, with the water flow rate set to 28.75 µL s −1 , continuous injection for 120 s. The detailed data from the heat dissipation experiment is shown in Figure 8b . The open channel with a micronano rough surface reduced the temperature of the aluminum plate from 62 to 54 °C in the 60s, after that the aluminum plate temperature remained constant, while the stable surface temperature of the open channel with the optimized pattern was just 50 °C. Figure 8 a) Diagram of enhanced heat dissipation experimental device. b) Temperature‐cooling time relationship in open channels with micronano structure and wedged pattern. The results indicate that the optimized pattern accelerated the flow of cooling water, allowing the liquid that had exchanged heat to promptly carry away the heat from the channel surface. It is believed that structure design in open channels can enhance the heat dissipation effect, which has potential applications in finned heat exchangers, electronic components, and liquid cooling fields. Note that fabricating the proposed structure on a large area surface with connected open channels will no longer need tilting the system."
} | 8,162 |
38251164 | PMC11154518 | pmc | 871 | {
"abstract": "A synaptic device with a multilayer structure is proposed to reduce the operating power of neuromorphic computing systems while maintaining a high-density integration. A simple metal–insulator–metal (MIM)-structured multilayer synaptic device is developed using an 8-inch wafer-based and complementary metal–oxide–semiconductor (CMOS) fabrication process. The three types of MIM-structured synaptic devices are compared to assess their effects on reducing the operating power. The obtained results exhibited low-power operation owing to the inserted layers acting as an internal resistor. The modulated operational conductance level and simple MIM structure demonstrate the feasibility of implementing both low-power operation and high-density integration in multilayer synaptic devices.",
"conclusion": "4. Conclusions In this study, the synaptic device with multilayer MIM-structured synaptic devices suitable for high-density integration and low-power operation were developed using 8 inch wafer-based CMOS fabrication processes. Compared to the double layer, the triple layer demonstrated a low-power operation as the power consumption was reduced by approximately 31%. The synaptic device for neuromorphic systems achieved a low-power consumption due to the reduced or modulated conductance level, because the AlO \n X \n layer inserted in the triple layer not only acts as a barrier layer but also acts as an internal resistor. In addition, the triple layer does not degrade the synaptic characteristics even when the AlO \n X \n layer is added, so the recognition rate shows the undegraded performance of 84.11%. Therefore, the obtained results demonstrate the feasibility of achieving both a low-power operation and high-density integration in multilayer synaptic devices.",
"introduction": "1. Introduction The recent exponential growth in unstructured data has led to a significant increase in the amount of data required for efficient processing [ 1 , 2 ]. However, conventional von Neumann computing systems have limitations that result in slow data processing owing to the bottleneck effect caused by the sequential transfer of data between the central processing unit and memory [ 3 , 4 , 5 ]. To address this issue, researchers have explored neuromorphic computing systems that use parallel data processing, which enables faster and more energy-efficient processing of large amounts of data [ 6 , 7 , 8 ]. To implement this neuromorphic computing system in the hardware, current-based vector–matrix multiplication (VMM) is commonly used via a synaptic device array [ 9 , 10 , 11 ]. Because a larger synaptic device array can process more data in parallel, the high-density integration of the synaptic device is necessary. For this purpose, in this research, a simple two-terminal (2T)-based metal–insulator–metal (MIM)-structured memristor which has been studied for memory application is utilized as the synaptic device [ 12 , 13 , 14 , 15 , 16 , 17 , 18 ]. The 2T-based memristor devices have been investigated, including resistive random-access memory (ReRAM), phase-change memory (PCM) [ 19 ], ferroelectric random-access memory (FeRAM) [ 20 , 21 ], and Magnetic random-access memory (MRAM) [ 22 ]. Among these memristor devices, ReRAM is the most attractive candidate owing to its simple structure, high-density integration, fast switching speed, and excellent scalability [ 23 , 24 , 25 , 26 , 27 , 28 ]. Although the memristor-based synaptic device array can lead to faster parallel data processing using VMM, further research is required to minimize its power consumption. However, ReRAM has been studied for memory application [ 16 , 29 ], research on the device operation mechanism [ 29 , 30 , 31 ], and research on ReRAMs composed of materials that are not CMOS-compatible [ 32 , 33 ]. Thus, in this study, a memristor-based 2T synaptic device with a multilayer structure was proposed to reduce the operating power while maintaining high-density integration. Moreover, 8 inch wafer-based CMOS fabrication processes and an oxide-based W/TaO \n X \n /AlO \n X \n /WO \n X \n /TiN stack were used to assess the feasibility of mass production. The obtained result showed that the AlO \n X \n layer acted as an internal resistor (and barrier layer) without degradation of the synaptic characteristics and exhibited a low-power operation.",
"discussion": "3. Results and Discussion As mentioned above, the synaptic devices of single, double, and triple layers were fabricated. To confirm the synaptic characteristics of each device, each weight-update curve was measured ( Figure 2 a–c). The inset of Figure 2 a–c show the pulse conditions for potentiation (conductance increase) and depression (conductance decrease). In the single layer, it exhibited resistive switching, which refers to resistance changes from a high-resistance state to a low-resistance state in the negative bias region, and vice versa. When a positive bias is applied to the TE, the oxygen ions of the WO \n X \n layer are migrated to the TE. This migration results in the formation of an induced oxide layer at the interface between the WO \n X \n layer and TE, resulting in decreased conductance. The thickness of the induced oxide layer increased as a continuous positive pulse bias was applied, and thus the conductance was modulated ( Figure 2 a,d) [ 36 , 37 ]. In contrast, the weight update curve occurs at the opposite polarity for the double and triple layer ( Figure 2 b,c). The inset of Figure 2 b shows the current-voltage (I–V) curve characteristic of the double layer. Gradual resistive switching of the double layer was observed under optimized conditions. The set process in the positive bias and the reset process in the negative bias are observed. Switching behavior occurred in the WO \n X \n layer depending on the mobile oxygen ions between WO \n X \n and TaO \n X \n layers [ 35 , 38 ]. When the positive bias was applied to the TE, the oxygen ions in the WO \n X \n layer moved to the TaO \n X \n layer. Thus, the amount of oxygen vacancies in WO \n X \n increased, resulting in the potentiation process. Conversely, when the negative bias was applied, the oxygen ions that had moved to the TaO \n X \n layer during the potentiation process moved back to the WO \n X \n layer, resulting in the depression process. To achieve synaptic characteristics based on this operating mechanism, the fabrication conditions (such as the Ar: O \n 2 \n ratio of the WO \n X \n layer and the working pressure of the TaO \n X \n ) were optimized, as shown in Figure 3 . A higher initial resistance was observed during the deposition of the WO \n X \n when the Ar: O \n 2 \n ratio was increased ( Figure 3 a). However, resistive switching was only obtained when the ratio of Ar to O \n 2 \n was 20:5. This result can be explained in terms of the oxygen vacancy density in the WO \n X \n layer [ 39 ] ( Figure 3 b). When the Ar:O \n 2 \n ratio changed to 20:1, more oxygen vacancies were present in the WO \n X \n layer, resulting in an electrically short state. In contrast, when the Ar:O \n 2 \n ratio was 20:10, sufficient oxygen ions were supplied during the deposition of the WO \n X \n . Consequently, an insulating WO \n X \n layer was formed, leading to an electrically insulating behavior. Based on the optimized WO \n X \n oxygen partial pressure condition, the working pressure of the TaO \n X \n layer was also varied to achieve synaptic characteristics, as shown in Figure 3 c,d. When the working pressure was changed from 20 to 10 and 5 mTorr, resistive switching was observed only at 10 mTorr. Considering that a higher working pressure can result in a porous film, deposition at 20 mTorr forms a more porous TaO \n X \n layer [ 37 , 40 ]. Similarly, a denser TaO \n X \n layer was deposited at 5 mTorr. Because the effective area of the interface between the TaO \n X \n and WO \n X \n layers can be increased by higher porosity, more oxygen absorption, resulting in an electrically short state, can occur at 20 mTorr. Additionally, at 5 mTorr, the reduced effective interfacial area and formation of a denser TaO \n X \n layer prevented oxygen absorption. Based on these results, conditions such as an Ar:O \n 2 \n ratio of 20:5 and a working pressure of 10 mTorr were selected as the optimal fabrication conditions for the WO \n X \n and TaO \n X \n layers. The double layer exhibited a lower conductance level than the single layer; however, it was still unacceptably high for the low-power operation of synaptic devices. To further reduce the operating power of the synaptic device, an AlO \n X \n layer was inserted into the interface between the TaO \n X \n and WO \n X \n layers (triple layer). The AlO \n X \n layer was added between the TaO \n X \n and WO \n X \n layers, rather than elsewhere, to obtain the synaptic characteristic. When the AlO \n X \n layer was added to the interface between the WO \n X \n and TiN layers (W/TaO \n X \n /WO \n X \n /AlO \n X \n /TiN), no switching characteristic was observed. The triple layer has an operating mechanism similar to the double layer. The switching occurs in the WO \n X \n layer according to the mobility of oxygen ions between the WO \n X \n layer and TaO \n X \n layer, as shown in Figure 2 f. When the positive bias is applied to the top electrode, oxygen ions in the WO \n X \n layer migrate through the AlO \n X \n layer to the TaO \n X \n layer, causing switching in the WO \n X \n layer. Thus, the potentiation process occurs in which the conductance increases under a positive bias. The conductance level of potentiation and depression decreased with the insertion of the AlO \n X \n layer. The thickness of the AlO \n X \n layer was varied from 22.5 to 30 nm for optimization. The initial resistance increased with increasing AlO \n X \n layer thickness. Owing to the increased initial resistance, the conductance levels of potentiation and depression decreased. The conductance levels of potentiation and depression were compared in three types of synapse devices. The conductance levels of potentiation and depression decreased with increasing number of layers ( Figure 4 a). The synaptic characteristics of the devices were verified by normalizing and comparing the potentiation and depression behaviors of the single, double, and triple layers using Equation ( 1 ), where G \n m a x \n and G \n m i n \n are the maximum conductance state and minimum conductance state, respectively. The normalized synaptic potentiation and depression behaviors of each device were similar, indicating that the multilayer structure can reduce the operating power without significantly degrading the synaptic characteristics ( Figure 4 b).\n \n (1) \n G n o r m a l = ( G − G m i n ) ( G m a x − G m i n ) \n To investigate the role of the inserted AlO \n X \n layer, three cases, namely a double layer, a double layer with an external commercial resistor (200 k \n Ω \n ), and a triple layer, were compared in Figure 5 . Figure 5 a compares the double and triple layers, revealing an obvious decrease in the conductance level of the triple layer. As shown in Figure 5 b, the conductance of double layer with an external commercial resistor was measured by connecting a 200 k \n Ω \n commercial resistor in series through the wiring outside of the double layer device. When the external resistor was connected to the double layer, the conductance level decreased. Compared with the double layer, as shown in Figure 5 c, both the triple and double layers with an external resistor exhibited significantly decreased conductance levels. Furthermore, the triple layer exhibited the same operating conductance level as the double layer connected to the external resistor. This result implies that the inserted AlO \n X \n layer can serve as an internal 200 k \n Ω \n resistor to efficiently reduce the conductance level. In addition, the composition ratio of the WO \n X \n layer, which is a switching layer, was changed compared to the double layer because the AlO \n X \n layer was inserted between the WO \n X \n layer and the TaO \n X \n layer in the triple layer. When the AlO \n X \n layer, which acts as a barrier layer (or shielding layer) [ 41 ], was deposited on the WO \n X \n layer, the amount of oxygen ions absorbed from the WO \n X \n layer was reduced. Accordingly, compared with the double layer, the oxygen vacancy density of the WO \n X \n layer of the triple layer decreases. These results were quantitatively analyzed by X-ray photoelectron spectroscopy (XPS) measurements in Figure 6 . Figure 6 a,b show the XPS analysis spectra of O 1s in the WO \n X \n layer of the double layer and the triple layer, respectively. The XPS spectrum showed a broad peak, which can be deconvoluted into three individual peaks: the W-O bond peak, oxygen vacancy density, and chemisorbed oxygen species. The green peak of the double layer (530.75 eV) and the triple layer (531.09 eV) can be assigned to the oxygen atoms (O \n 2 − \n ) which form W-O bonds. In addition, the violet peaks represent chemisorbed oxygen species (H \n 2 \n O). Finally, the pink peaks can be assigned to species adsorbed on the surface (OH \n − \n , O \n − \n , or oxygen vacancies); the OH \n − \n groups bond with the metal cations to maintain a charge balance. This implies that the intensity of the OH \n − \n peak indicates oxygen vacancy density [ 42 , 43 ]. The oxygen vacancy density of the double and triple layer are 30.34% and 27.42%, respectively. Therefore, the triple layer has a lower oxygen vacancy density than the double layer. The stoichiometric ratio between tungsten and oxygen can be determined from the composition ratio. In double layer, the tungsten atomic ratio is 30.76% and the oxygen atomic ratio is 69.24%. Thus, the ratio of the tungsten to the oxygen is about 1:2.25 (WO \n 2.25 \n ). In the same way, the atomic ratio of tungsten in the triple layer is 29.08% and the atomic ratio of oxygen is 70.92%, so the ratio is 1:2.44 (WO \n 2.44 \n ) ( Table 1 ). This indicates that the WO \n X \n of the triple layer contains a smaller number of oxygen vacancies compared to the WO \n X \n of the double layer. As a result, the AlO \n X \n layer plays the role of 200 k \n Ω \n because the defect in the switching layer (WO \n X \n layer) decreases the resistance of the AlO \n X \n layer itself. Therefore, the conductance level of the triple layer decreases. Owing to the decreased or modulated conductance level, the synaptic device for the neuromorphic system can achieve low power consumption. The power consumption of the single, double, and triple layer was numerically calculated as shown in Figure 7 a. When comparing the single and double layer, the power consumption of the double layer was slightly decreased, from 28.24 \n μ \n J to 25.03 \n μ \n J. This is because a voltage drop occurred by inserting a TaO \n X \n layer. Thus, a larger pulse amplitude is required for the double layer, and the power consumption was only slightly decreased. However, the power consumption of the triple layer was reduced by 31.2% compared to the double layer (from 25.03 \n μ \n J to 17.22 \n μ \n J), with the same pulse width and amplitude. Considering the huge size of the synaptic array in the neuromorphic system, a significant reduction in power consumption can be expected. Additionally, to verify the influence at the system level, an image recognition simulation consisting of four-layer neural networks was conducted, as shown in Figure 7 b–e. The IBM Analog Hardware Acceleration Kit (AIHWK it )), which can simulate devices in real-world applications, is used to simulate training and inference [ 44 ]. This provides several device models. We used a “LinearStepDevice” among them. Each parameter required for the simulation was extracted from the measured potentiation/depression weight update curve of the single, double, and triple layer. The neural network was constructed with an input layer of 784 neuron nodes, hidden layer 1 of 256 neuron nodes, hidden layer 2 of 128 neuron nodes, and an output layer of 10 neuron nodes ( Figure 7 b). A synapse device model was used to connect each neuron node. For the implementation of the deep neural network of Figure 7 b at the device level, a synaptic device acting as a weight value can be constructed by a cross-point array [ 45 ]. To perform the Multiply and Accumulation operation, the input voltage bias is applied to all row lines, and the output is obtained as a summed current by multiplying the conductance stored at the synaptic devices ( Figure 7 c). We utilized the Modified National Institute of Standards and Technology (MNIST) dataset (28 × 28) as an input image. Figure 7 d,e show the recognition rate according to the training epoch. The image recognition rates are 85.10%, 71.51%, and 84.11% for a single, double, and triple layer when the numerically ideal case is 92.57%. Even though the triple layer has the lowest power consumption, it exhibited a similar recognition rate to others. This is because the linearity of the weight update curve was not degraded with the addition of the layer compared to the single layer. The image recognition rate reaches about 85%, which is respectable but could be even higher with a wider dynamic range."
} | 4,326 |
37072642 | PMC10238223 | pmc | 872 | {
"abstract": "Abstract In the era of big data and artificial intelligence (AI), advanced data storage and processing technologies are in urgent demand. The innovative neuromorphic algorithm and hardware based on memristor devices hold a promise to break the von Neumann bottleneck. In recent years, carbon nanodots (CDs) have emerged as a new class of nano‐carbon materials, which have attracted widespread attention in the applications of chemical sensors, bioimaging, and memristors. The focus of this review is to summarize the main advances of CDs‐based memristors, and their state‐of‐the‐art applications in artificial synapses, neuromorphic computing, and human sensory perception systems. The first step is to systematically introduce the synthetic methods of CDs and their derivatives, providing instructive guidance to prepare high‐quality CDs with desired properties. Then, the structure–property relationship and resistive switching mechanism of CDs‐based memristors are discussed in depth. The current challenges and prospects of memristor‐based artificial synapses and neuromorphic computing are also presented. Moreover, this review outlines some promising application scenarios of CDs‐based memristors, including neuromorphic sensors and vision, low‐energy quantum computation, and human–machine collaboration.",
"conclusion": "9 Conclusions CDs and their derivative composites have been successfully explored in memristor, artificial synapse, and neuromorphic computing applications, due to their excellent properties of biocompatibility, low toxicity, environmental friendliness, and simple preparation. As compared with traditional organic carbon materials, CDs‐based electronic devices are more resistant to temperature, but more sensitive to heat and light. This review summarizes recent advances in the synthetic route of diversified CDs and their derivatives, switching mechanisms, smart memristors, artificial synapses, and neuromorphic computing applications. Despite these achievements in recent years, there remain a number of challenges and limitations: 1) The introduction of CDs into self‐assembled films helps to improve device stability and reproducibility. However, it should be noted that CDs have a lot of active groups (such as amides, hydroxyls, or carboxyls) anchored onto their surface, which may not be conducive to long‐term stability under ambient conditions. Thus, developing a surface passivation strategy to prepare desired CDs candidates could assure solid memristive performances. 2) Normally, most CDs exhibit low quantum yields below 10%, making them unsuitable for multifunctional photosensitive memristors, artificial synapses and neuromorphic applications. How to prepare CDs materials with a high quantum yield is one of the popular research topics. 3) Given that nanoscale CDs often serve as memristive materials, the integration of nano‐memory cells with nanostructured assembly to fabricate high‐density nanoarrays is fascinating but challenging. 4) The investigation of CDs‐based artificial synapse is still in its infancy, despite the excellent structural stability, nano‐size, and quantum confinement effects of CDs. Novel bionic synaptic devices with exceptional transparency, mechanical flexibility, and transferability need to be pursued. 5) An integrated device with the integration of sensory storage and computation functions should be constructed for future human sensory perception systems. Furthermore, advanced algorithms and logic operations are urgently needed for competitive neuromorphic platforms. In general, we anticipate that this review will lead to further in‐depth research on CDs‐based smart memristors, low‐power artificial synapses, and neuromorphic computing applications.",
"introduction": "1 Introduction Since the information industry has grown rapidly over the past years, developing high‐performance data storage and computing systems is becoming urgent. [ \n \n 1 \n \n ] The traditional solution is to scale down electronic components and increase the density of device cells per unit area. Nevertheless, this strategy is limited by Von Neumann bottleneck and technological challenges in design, manufacturing, and energy dissipation. To address these issues, researchers are devoting substantial efforts to exploring a new generation of ultrahigh‐density non‐volatile memory (NVM) technology and algorithm. [ \n \n 2 \n \n ] Among the current burgeoning techniques, two‐terminal memristors with a metal/insulator/metal (MIM) structure have emerged as promising information storage and processing technology, due to their excellent characteristics of low power consumption, high‐frequency and high‐density data programming, and reliable compatibility with complementary metal‐oxide‐semiconductor (CMOS) circuits for multi‐integration applications. [ \n \n 3 \n \n ] \n Recent studies have demonstrated that memristors with MIM structures can be used to construct efficient in‐memory computing systems with enhanced computation capability. [ \n \n 4 \n \n ] Unlike conventional computers with separated memory and processing units, the two‐terminal memristor with integrated storage and computing architectures is capable of overcoming the von Neumann bottleneck. Further cutting‐edge research is to develop brain‐like bionic electronic devices that adopt memristors as equivalent electronic components. Considering that the human brain can perform the tasks of memory, learning, computation, cognition, and emotions, a biomimetic neuromorphic network comprising trillions of neurons connected by more than 10 15 synapses is anticipated to be established via memristor technology. In the long run, exploiting a memristor device array to simulate synaptic connections in a biomimetic neuromorphic brain or human sensory perception system will become a revolutionary technology. [ \n \n 1c \n \n ] \n During the past decade, a large number of studies have been conducted on the preparation methods of memristive materials, tunable performance, and device manufacturing. To date, a wide range of inorganic and organic memristive materials have been intensively studied, including inorganic oxides, [ \n \n 5 \n \n ] phase‐change materials, [ \n \n 6 \n \n ] organic small molecules, [ \n \n 4 \n , \n 7 \n \n ] polymers, [ \n \n 8 \n \n ] organic nanocrystals, [ \n \n 9 \n \n ] semiconductor quantum dots (SQDs), [ \n \n 10 \n \n ] van der Waals and 2D materials, [ \n \n 11 \n \n ] novel carbon materials, [ \n \n 12 \n \n ] biomaterials, [ \n \n 13 \n \n ] MXenes, [ \n \n 14 \n \n ] ferroelectric materials, [ \n \n 15 \n \n ] perovskites, [ \n \n 16 \n \n ] as well as their hybrid composites. [ \n \n 17 \n \n ] For organic memristive materials, one of the challenges is their insufficient stability and low tolerance to ambient conditions, whereas inorganic memristive materials are limited by their poor flexibility and tunability. By contrast, the emerging solution‐processable carbon nanodots (CDs) exhibit unique physicochemical and photochemical properties, such as excellent charge storage capability, tunable energy level, and photo‐/electro‐luminescence, offering an appealing path to achieve high‐performance photosensitive logic sensors and electronic devices. [ \n \n 18 \n \n ] More importantly, 0D CDs possess attractive low toxicity, environmental friendliness, biocompatibility, and simple synthesis, which have shown wide applications in the field of biosensors, chemical sensors, nanomedicine, bioimaging, solid‐state batteries, photocatalysis, electrocatalysis, and so on. [ \n \n 19 \n \n ] Particularly, CDs‐based memristors have emerged as promising candidates to implement artificial synapses and human sensory perception systems, due to their outstanding structural stability and quantum confinement effects. In this contribution, we focus on the recent progress of CDs‐based memristors for artificial synapses and neuromorphic computing applications. This review begins with a brief introduction of CDs materials, involving general chemical makeup and basic physical properties. The various synthetic approaches for CDs are summarized. Then, we present the strategies to construct CDs‐related composite materials. Finally, the memristive, synaptic, and neuromorphic computing performances of CDs‐based memristors, as well as their intrinsic mechanisms, are well discussed. It is believed that this article will offer a comprehensive overview on current achievements and challenges of CDs‐based memristors ( Figure \n \n 1 \n ), and open new perspectives toward the future memristor development in the post‐Moore era. Figure 1 Overview diagram of CDs, involving synthesis method, material design, the memristive, synaptic, neuromorphic computing applications, and possible application scenario (memristor+). Image at the bottom: Reproduced with permission. [ \n \n 55 \n \n ] Copyright 2016, Royal Society of Chemistry. Image at the top left: Reproduced with permission. [ \n \n 2b \n \n ] Copyright 2022, Wiley."
} | 2,225 |
33990331 | PMC8121431 | pmc | 873 | {
"abstract": "High dimensionality and fading memory for in-sensor reservoir computing are achieved via two-dimensional memristors.",
"introduction": "INTRODUCTION The massive amount of data produced and transmitted by the Internet of Things (IoT) requires new insight into real-time information processing by edge computing devices based on novel materials and architectures ( 1 ). Proactively interpreting and learning with temporal and sequential information represent key tasks for edge computing devices ( 2 ). However, current edge computing systems mostly rely on physically separated sensors and digital processing units, leading to high rate of energy consumption and long-time latencies when sequentially digitizing the analog pixel signals ( 3 – 9 ). The direct processing of time-varying optical data accounting for more than 80% of the collected information ( 10 ) or the perceptions of humans in a bio-plausible fashion would provide a breakthrough by reducing the communication and computation loads of edge computing devices that operate on the IoT. While recurrent neural networks (RNNs) are highly capable of processing time-series data and various dynamic sequential events ( 11 , 12 ), the high training complexity in RNNs limits their practical use with regard to edge computing on the IoT ( 2 ). Among various frameworks of RNNs, reservoir computing (RC) has demonstrated substantial reductions of the computational cost of learning, offering a promising solution by which to develop edge devices for temporal pattern classification, prediction, and generation ( 2 ). In RC, a dynamic reservoir is used to map complex inputs nonlinearly (e.g., spatiotemporal patterns) into high-dimensional states, which can be further augmented by virtual nodes, allowing a simple and fast training of readout weights at a low computation cost. Thus far, RC systems have been physically implemented on conventional digital platforms and previously unidentified hardware dynamic systems powered by different physical mechanisms (e.g., electronic, photonic, spintronic, mechanical, and biological RC implementations) ( 6 , 9 , 13 – 18 ). For example, RC systems have been built on memristors with a simple structure, high density, good energy efficiency, and three-dimensional (3D) stackability ( 9 ). Memristors have demonstrated the ability to mimic the leaky integrate-and-fire operations of neurons in a manner that is more energy and area efficient than digital alternatives, as reported by Du et al. , Moon et al. , and Midya et al. ( 6 , 8 , 9 ). Conventional memristors mostly rely on redox reactions and the migration of ions and cannot directly respond to optical stimuli. Creating RC systems with high computing efficiency and direct responses to optical inputs without extra sensors/processors remains a challenge. Accordingly, memristors based on novel nanomaterials or hybrid materials ( 19 – 31 ) that offer different degrees of plasticity by means of either electrical or optical signals have been reported. However, the realization of an RC system with high computing efficiency enabled by in-memristor computing and direct responses to optical inputs without extra sensors/processors with a simple device structure remains a challenge. In this study, we demonstrate optoelectronic RC for language learning with dynamic memristors built on a two-terminal tin sulfide (SnS) device structure. The synergy of the charge trap/detrap dynamics and the photogating effects in the atomically thin material used in this study enables high-performance capabilities and versatile memristive behaviors with dual-mode operation (i.e., driven by electrical and optical stimuli), as well as good nonlinearity and fading memory. These memristors constitute an optoelectronic reservoir by responding to sequential electrical and broadband optical stimuli. Mapping complex temporal optoelectronic inputs into high-dimensional reservoir states, the optoelectronic RC described here demonstrates an accuracy of 91% in classifying practical Korean sentences with small natural errors. The low training cost and real-time processing of spatiotemporal signals for optoelectrical stimuli pave the way for efficient edge machine learning.",
"discussion": "DISCUSSION The unique electrical and optical responses of 2D SnS, originating from its rich defect states with both Sn and S atomic vacancies, have been successfully adopted for the novel in-sensor RC for language learning with dual-mode operation. The unique and opposite trends of the electrical and optical responses of integrated SnS-based memristors effectively lead to the high-dimensional states of optoelectronic RC with complicated spatiotemporal input signals. As a proof of concept, we achieved an accuracy of 91% during the classification of five practical Korean sentences each containing intentional noise. Our finding overcomes the hardware bottleneck associated with physically separated sensors and processors, where analog-to-digital data conversion incurs large energy consumption and latency penalties in conventional sensing-computing systems. In addition, RC in this case greatly reduces the learning complexity. This energy-efficient method enables efficient machine learning applications with temporal inputs at the edge, an advance in strong demand in the era of the IoT. Compared with biological neurons of different/specialized sensing modes, our dual-mode processing could be considered as a breakthrough for efficient machine learning and neuromorphic computing."
} | 1,374 |
25247576 | PMC4200787 | pmc | 874 | {
"abstract": "Microbial fuel cells (MFCs) represent a novel platform for treating wastewater and at the same time generating electricity. Using Pseudomonas \n putida ( BCRC 1059 ), a wild-type bacterium, we demonstrated that the refinery wastewater could be treated and also generate electric current in an air-cathode chamber over four-batch cycles for 63 cumulative days. Our study indicated that the oil refinery wastewater containing 2213 mg/L (ppm) chemical oxygen demand (COD) could be used as a substrate for electricity generation in the reactor of the MFC. A maximum voltage of 355 mV was obtained with the highest power density of 0.005 mW/cm 2 in the third cycle with a maximum current density of 0.015 mA/cm 2 in regard to the external resistor of 1000 Ω. A maximum coulombic efficiency of 6 × 10 −2 % was obtained in the fourth cycle. The removal efficiency of the COD reached 30% as a function of time. Electron transfer mechanism was studied using cyclic voltammetry, which indicated the presence of a soluble electron shuttle in the reactor. Our study demonstrated that oil refinery wastewater could be used as a substrate for electricity generation.",
"conclusion": "4. Conclusions Our study demonstrated that treatment of oil refinery wastewater could simultaneously generate electricity in an air-cathode MFC. The maximum voltage output of 355 mV was reached with the highest power density of 0.005 mW/cm 2 in the third cycle. A maximum current density of 0.015 mA/cm 2 was reached in the third cycle and the maximum coulombic efficiency (6 × 10 −2 ) was observed in the fourth cycle, mainly caused by other electron acceptors in refinery wastewater and oxygen diffusion during the long operation period. A COD removal rate up to 30% was achieved. The combination with some well-known strains in MFC such as Shewanella and Acinetobacter would be the future prospects to achieve maximum power density and maximum COD removal in refinery waste water. Bacterial population dynamics can be detected and monitored in this design using the molecular biotechnology technique of RT-PCR to understand the interactions of enriched bacterial communities.",
"introduction": "1. Introduction Energy requirements have been increasing exponentially worldwide. At present, global energy requirements are mostly dependent on fossil fuels, which will eventually lead to an exhaustion of limited fossil energy sources. Combustion of fossil fuels also has serious negative effects on the environment due to CO 2 emissions, which could be the main reason for climate change. Increased global demand for finite oil and natural gas reserves, and energy security concerns have intensified the search for alternatives to fossil fuels [ 1 ]. Bioelectrochemical systems such as microbial fuel cells (MFCs) are devices that exploit the ability of exo-electrogenic microbes to respire through the transfer of electrons outside the cell [ 2 ]. MFCs have been shown to convert the energy in organic matter present in wastewaters into electrical current [ 3 , 4 , 5 , 6 ]. In comparison to conventional fuel cells the key advantages of biological fuel cells are the mild operating conditions such as ambient temperature and near neutral pH. It could allow essentially infinite applications of potential fuel. However, there is a scarcity of suitable electrocatalysts for oxidation [ 7 ]. The principle of MFCs is based on the fact that generation of electricity is one of the basic properties of microorganisms, as they transfer electrons from an oxidized electron donor to an electron acceptor at a higher electrochemical potential [ 7 ]. Exo-electrogenic bacteria are mostly employed in MFC because exo-electrogenic bacteria transfer electrons to the anode of a MFC either through direct contact via highly conductive nanowires or membrane-associated proteins [ 2 , 7 , 8 ], or by using soluble electron shuttles [ 9 ]. Bioelectrogenesis was first demonstrated in 1911 by Potter who used Saccharomyces cerevisiae and some other species of bacteria with a Pt electrode immersed in sterile medium in a battery-like setup, and the chemical action of their vital process was utilized to develop electrical energy. Electrical energy was generated due to disintegration of organic compounds by microorganisms; Potter reported that he had obtained a voltage ranging from 0.3 to 0.5 volts [ 10 ]. Over the last few years, MFCs have been the focus of increasing interest due to their sustainable approach towards wastewater treatment along with use as an alternative source for power generation [ 11 , 12 ]. Domestic wastewater was used for electricity generation in several MFC configurations [ 13 , 14 , 15 , 16 ] and a maximum power density of 204 mW/m 2 was demonstrated [ 1 ]. Huang and Logan reported the effectiveness of electricity production with paper recycling plant wastewater with maximum power density reaching 672 mW/m 2 [ 17 ]. Beer brewery wastewater treatment using an air-cathode MFC was investigated by Feng and Wang [ 18 , 19 ] and a maximum power density of 528 mW/m 2 was achieved [ 18 ]. In 2006, Krishnan et al . reported electrolytic treatment of beer brewery wastewater on the basis of in-situ hypochlorous acid generation, and a maximum current density of 74.5 mA/cm 2 was achieved [ 20 ]. Starch processing wastewater was reported as being used for power generation using an air-cathode MFC with a maximum power density of 239.4 mW/m 2 and a current density of 893.3 mA/m 2 [ 21 ]. Swine wastewater treatment using a single chamber air-cathode MFC was studied and maximum power density of 261 mW/m 2 was achieved [ 22 ]. Removal of odor from swine wastewater was investigated by Kim et al ., and the maximum power density achieved was 228 mW/m 2 [ 23 ]. Chocolate industry wastewater treatment using a double chamber MFC was studied by Patil and colleagues and the maximum current achieved was 3.02 and 2.3 A/m 2 using a membrane and salt bridge respectively [ 6 ]. A single chamber MFC with an air-cathode was successfully created using a glucose-penicillin mixture or only penicillin as the fuel and a maximum current density of 10.73 A/m 2 was achieved for penicillin and a maximum power density of 101.2 W/m 3 was achieved for glucose-penicillin mixtures [ 24 ]. A MFC was employed to dispose of Cr 6+ in real electroplating wastewater and the maximum power density of 1.6 W/m 2 was generated at a coulombic efficiency of 12% [ 25 ]. So far, very few studies have been performed using oil refinery wastewater in MFC. Oil refinery wastewater is one of the major environmental pollutants if the treatment is not proper. It causes much environmental damage such as the abolition of microflora and fauna in water bodies that could spoil water quality. Most importantly it causes abnormal changes in the water eco-system. The major problem with oil refinery wastewater is the vast chemical oxygen demand (COD). We are reporting here the result of reduced COD using Pseudomonas \n putida (BCRC 1059) for the treatment of oil refinery wastewater in a single chamber air-cathode MFC.",
"discussion": "2. Results and Discussion 2.1. Current Generation from Refinery Wastewater The power generation shown in Figure 1 was observed over a period of four batch cycles with a fixed external resistance (1000 Ω). During the startup stage, there is a lag period of two days followed by inoculation and the course lasted for a total of 63 days from the beginning of the first cycle. An initial peak current of 0.08 mA was achieved in the first cycle. The bacterial population was restored by new substrate inoculation at the beginning of the second cycle and there was an immediate power generation of 81.26 mV. This result could be due to the difference in potential between the two electrodes based on both chemical and biological factors. Thereafter a sharp increase in current was observed to reach 0.31 mA, which might indicate the bio-electrochemical activity of the microorganisms that then gradually started to decrease after 15 days. The third and the fourth cycles were performed as per the second cycle. As shown in Figure 1 , greater current output was observed in the later feed-batch cycles, i.e. , third and fourth feed batches as compared with the first batch. Each cycle can be divided into three phases—ascending, stationary and declining. From Figure 1 it was observed that the stationary phase was longest in the second and third cycles. It is probably due to the formation of a biofilm seen in the reactor by electrochemically-active bacteria and the successful degradation of organic matter. The ascending phase was longer in the first cycle than in any other cycles owing to functioning of the biofilm. In the fourth cycle, however there was a sharp decrease from the stationary phase to the declining phase, which could be the inhibition of electron transfer from the bacteria to the anode surface by the matured bacterial biofilm. Figure 1 Current generation in air-cathode MFC for four successive batch cycles. Those arrows indicate the substrate inoculum addition as an end of each cycle. 2.2. Characterization of Microbial Fuel Cells (MFC) One of the most important parameters of the MFC is the polarization curve, which is used to assess performance on the basis of current generation. A polarization curve represents voltage as a function of current. In a single batch cycle, the MFC was stabilized at the maximum steady voltage, and the power density and polarization curves were measured at several points by changing the external resistance from 0.27 to 10 kΩ. Figure 2 shows the curves of voltage, current and power density, which followed a similar pattern for all four cycles. Allocations of open circuit voltage (OCV), maximum power density, and maximum current output are presented in Table 1 . According to Table 1 , the OCVs of the four cycles were consecutively at 312 mV for the first cycle and 402 mV for the second cycle, 409 mV for third cycle and 401 mV for fourth cycle. In the status of open circuit, no current is circuited through the circuit and hence power production is null. There are three phases in the polarization curve of Figure 2 : Activation losses, ohmic losses and mass transport limitation [ 1 , 2 ]. At the beginning of power production there is a drop of voltage, which could be attributed to activation loss of substrate diffusion. This finding is in agreement with results in the literature [ 1 , 2 , 26 ]. Similarly, a decrement of the current has been observed as the resistance increased, which is in agreement with the literature [ 1 , 2 , 12 , 26 ]. During the second phase of ohmic losses the current rose, and a linear relationship of voltage and current was exhibited. This is because of the resistance of electron and ion movement. This region formed an overshoot except during the first cycle, and hence reached their maximum power densities. The highest power density achieved was 0.0011 and 0.0015 mW/cm 2 respectively for the third and fourth cycle, and the associated currents were 0.25 and 0.28 mA ( Table 1 ). The third cycle and fourth cycle produced better power curves than the two initial ones. There is a steep drop of power density near to the maximum cell voltage, which could be attributed to mass transport losses respective of all cycles ( Figure 2 ). According to the polarization curve, the optimum resistance for our study was 1000 Ω, which applies to all cycles. ijms-15-16772-t001_Table 1 Table 1 Different polarization and current output values for four fed batch cycles during operation with different external resistance (0.27–10 kΩ). Cycle Number Open Circuit Potential (mV) Maximum Current (mA) Maximum Power Density (mW/cm 2 ) Cycle 1 312 0.079612121 0.000092 Cycle 2 402 0.257575758 0.000973 Cycle 3 409 0.284848485 0.00119 Cycle 4 401 0.321212121 0.0015132 Figure 2 Polarization curve for four cycles. 2.3. Cyclic Voltammetry Cyclic voltammetry (CV) is one of the most familiar and versatile techniques used to reflect electrochemical reactions, which allow probing of the mechanics of redox and transport properties of a system in solution. It also enables measurement of redox activities between the components involved in a biochemical system and components bound to the bacteria. The voltammetry profiles ( Figure 3 ) for the four batch cycles revealed there were noticeable variations in the electron discharge and energy generation patterns in the individual cycles. During the highest output of current, a voltammogram was recorded in situ with a scan rate of 0.1 Vs −1 . From the CV profile a significant peak was found in the second and third cycles but not in the fourth cycle, in both the forward and reverse scans ( Figure 3 b–d). A small peak in the first cycle was also found ( Figure 3 a). In the 1st cycle, the oxidation peak of 0.12 mA was found at −100 mV ( vs . Ag/AgCl) and a corresponding reduction peak was found near 0.1 mA ( Figure 3 a). For the second cycle an oxidation peak of 0.2 mA was found at −200 mV ( vs . Ag/AgCl) and the reduction peak was shown at the same voltage ( Figure 3 b). For the third cycle, an oxidation peak of 0.21 mA was detected at −110 mV ( vs . Ag/AgCl) and a corresponding reduction peak of −0.13 mA was detected at almost 30 mV ( Figure 3 c). However, for the fourth cycle the oxidative peak of 0.05 mA was very low at nearly −320 mV ( vs . Ag/AgCl), and a reduction peak of 0.15 mA was detected ( Figure 3 d). It is well known that the mechanism of anodic bacterial electron transfer is governed by three different mechanisms. One is the direct electron transfer between the electrode surface and bacterial membrane. Second is the mediated electron transfer which uses a redox active compound for the shuttle of the electron between the electrode and bacteria. The third one is wire electron transfer, which uses facilitated nanowire by bacteria for the transfer of electron to electrode [ 7 ]. In the case of the low current in first cycle, the biofilm was immature and was hence considered not to contribute much to the electron transfer ( Figure 1 and Figure 3 a). A strong oxidative peak was detected in both the second and third cycles, especially in the third cycle which indicated the biofilm could develop and mature after a long incubation time ( Figure 1 and Figure 3 b,c). The strong oxidation/reduction peaks in the voltammogram have been found in the second and third cycles ( Figure 3 b,c), which implied that the studied bacteria became more electrochemically active. The active bacteria could produce redox active compounds that facilitate electron transfer. The production of current would rely on the availability of such compounds. Those profound concentrations of redox mediators could improve the electron transfer and hence generate more current. However, there was a fall in current in the fourth cycle, which reveals that the biofilm might become fully developed and aged. It does not play a key role in electron transfer but acts as an inhibitor for the block of the electrochemical property ( Figure 3 d). It could be concluded that the observed peaks in voltammogram were due to redox active complexes produced by bacteria, which could control the electron transfer process for power generation. These redox active compounds might be attributed to the degradation of the substrate from refinery wastewater. The proficiency of the MFC reactor is dependent on the availability of the degraded derivatives. 2.4. Coulombic Efficiency as a Function of Power Density Using Refinery Wastewater Coulombic efficiency (CE) and power density are the two major parameters used to describe and understand the phenomena that take place in the reactor of MFC. The data of Figure 4 showed that CE has been increasing with the increase of cycles. It ranged from 1 × 10 −3 % to 6 × 10 −3 % with a maximum of 6 × 10 −3 % for the fourth cycle. Maximum power densities were proportional to the changes in CE. This result is consistent with the result of Lu et al . [ 21 ]. However, maximum power density in the fourth cycle was found to decrease compared with the third cycle. This variation agreed with the result of CV and could be due to biological interference of biofilm aging. The low CE is a general issue in MFCs using real wastewater. Several probable factors including reactor configuration that could contribute to lower CE in MFCs were reported [ 19 ]. The other is the intake of electrons for other bacterial metabolic activities such as methanogenesis and fermentation [ 19 ]. Figure 3 Cyclic voltammogram was recorded at a scan rate of 0.1 V/s for four batch cycles. Cycles are shown here as a clockwise direction ( a : first cycle; b : second cycle; c : third cycle; and d : fourth cycle). In contradiction to the literature, our study did not reach the same maximum CE. The probable cause is the existence of additional existing electron acceptors such as nitrate and sulfate in oil refinery wastewater, which consumes electrons and thus lowers the CE [ 21 ]. This causes the additional flow of electrons from substrates to different acceptors. The dissipation of electrons hampered the gain of the highest CE. Oxygen diffusion through the cathode also accounts for loss of carbon compounds to aerobic respiration and degradation, resulting in a low CE [ 8 ]. Higher recovery of electrons represents more effective organic oxidization and lower loss of cells contributing to production of electricity. The highest CE achieved in this study, was not as high as compared with that achieved using other wastewater. This is probably because of the presence of other electron acceptors in the wastewater and oxygen diffusion during the batch process. Reactor configuration also could be improved to achieve high CE. 2.5. Chemical Oxygen Demand (COD) Removal and Wastewater Treatment Efficiency One of the major goals of the MFC system is the treatment of wastewater by removing the load of COD. COD removal efficiency and coulombic efficiency was computed for each cycle as shown in Figure 5 . The data showed that there is a proportional relationship between COD removal and CE. Initially, the COD removal efficiency was very low, as was the CE of the system, but a significant increase in both parameters was found as the cycle proceeded. COD removal was increased by higher CE. The CE could be increased in future studies by adding two layers of cloth to the cathode surface as shown by Fan et al. [ 27 ]. Figure 4 Coulombic efficiency as a function of power density using refinery wastewater for fourth batch cycles. Figure 5 Refinery wastewater chemical oxygen demand (COD) removal efficiency and coulombic efficiency (CE) was obtained for four batch cycles. The function of the MFC is to produce electricity as a result of the degradation of organic matter in wastewater. In our study, a MFC was developed that was able to perform electricity generation and wastewater treatment, simultaneously. The effluent COD concentrations and removal efficiencies for all of the four cycles are shown in Figure 6 . The data showed that the COD removal efficiency increased as a function of time, ranging from 10.0% to 30.0%. Degradation patterns for petrochemical fractions for four batch cycles have been provided in Figure 7 . The high COD removal was attributed to the long operation period in this study, which also extended the time for oxygen diffusion into the system, as a consequence of low CE. A COD removal efficiency of 30.0% and the low corresponding CE of 6 × 10 −3 % indicated that the major consumption of soluble and insoluble organic matter in refinery wastewater was not associated with power generation. Oxygen diffusion is not the only reason for this result and other factors such as other electron acceptors, biomass production and fermentation also need to be included. The wastewater treatment efficiency achieved is not sufficient as compared with other studies. We propose that the longer operation time is needed because the efficiency was found to increase as a function of time. Figure 6 COD and removal efficiency was computed for four batch cycles of this study. Substrate degradation is one of the main factors that lead to high current. Substrate degradation in this study did not achieve high capacity according to the efficiency of COD removal. The enhancement of substrate degradation by other methods is needed. It is also necessary to improve the electron transfer procedure between substrates and electrodes. The respective electrodes need to be rearranged. The improvement of reactor architecture needs to be taken into consideration. In addition, the use of a proper separator for the air cathode reactor as per Sevda et \n al. , led to increased COD removal [ 28 ]. In that study 77.56% of COD was removed using a Zirfon ® membrane compared with the 51.92% of removal using a Fumasep ® one. The degradation of the petrochemical fraction at the beginning and at the end of each cycle was performed. The Figure 7 shows the degradation pattern for four cycles and the Table 2 shows the list of all poly-aromatic contaminants. The data shows the presence of aromatic fractions which has been found more often than aliphatic fractions in refinery waste-water after degradation. The electroactive bacteria are solely responsible for the degradation of organic pollutants with long incubation times. The spatial barrier of biofilm for electron transfer could be the reason for the reduction of current production in the fourth cycle. Figure 7 Degradation pattern of organic pollutants in refinery waste water. ijms-15-16772-t002_Table 2 Table 2 Details of the poly-aromatic hydrocarbons. Abbreviation Full Name Formula Molecular Weight Nap Naphthalene C 10 H 8 128 Acpy Acenaphthylene C 12 H 8 152 Acp Acenaphthene C 12 H 10 154 Flu Fluorene C 13 H 10 166 Phen Phenanthrene C 14 H 10 178 Anthr Anthracene C 14 H 10 178 Fl Fluoranthene C 16 H 10 202 Pyr Pyrene C 16 H 12 202 Baa Benzo[a]anthracene C 18 H 12 228 Chry Chrysene C 18 H 12 228 Bbf Benzo[b]fluoranthene C 20 H 12 252 Bkf Benzo[k]fluoranthene C 20 H 12 252 Bap Benzo[a]pyrene C 20 H 12 252 Pery Perylene C 20 H 12 252 IP Ideno[1,2,3-c,d]pyrene C 22 H 12 276 Dbah Dibenzo[a,h]anthracene C 22 H 14 278 Bghip Benzo[g,h,i] perylene C 22 H 12 276"
} | 5,575 |
28083531 | PMC5183647 | pmc | 875 | {
"abstract": "Ion-conducting memristors comprised of the layered materials Ge 2 Se 3 /SnSe/Ag are promising candidates for neuromorphic computing applications. Here, the spike-timing dependent plasticity (STDP) application is demonstrated for the first time with a single memristor type operating as a synapse over a timescale of 10 orders of magnitude, from nanoseconds through seconds. This large dynamic range allows the memristors to be useful in applications that require slow biological times, as well as fast times such as needed in neuromorphic computing, thus allowing multiple functions in one design for one memristor type—a “one size fits all” approach. This work also investigated the effects of varying the spike pulse shapes on the STDP response of the memristors. These results showed that small changes in the pre- and postsynaptic pulse shape can have a significant impact on the STDP. These results may provide circuit designers with insights into how pulse shape affects the actual memristor STDP response and aid them in the design of neuromorphic circuits and systems that can take advantage of certain features in the memristor STDP response that are programmable via the pre- and postsynaptic pulse shapes. In addition, the energy requirement per memristor is approximated based on the pulse shape and timing responses. The energy requirement estimated per memristor operating on slower biological timescales (milliseconds to seconds) is larger (nanojoules range), as expected, than the faster (nanoseconds) operating times (~0.1 pJ in some cases). Lastly, the memristors responded in a similar manner under normal STDP conditions (pre- and post-spikes applied to opposite memristor terminals) as they did to the case where a waveform corresponding to the difference between pre- and post-spikes was applied to only one electrode, with the other electrode held at ground potential. By applying the difference signal to only one terminal, testing of the memristor in various applications can be achieved with a simplified test set-up, and thus be easier to accomplish in most laboratories.",
"conclusion": "Conclusion Ge 2 Se 3 /SnSe/Ag-based ion-conducting memristive devices perform over the seconds to nanoseconds timescale as synapses in an STDP experiment. STDP tests were performed with four different spike wave shapes in order to demonstrate the influence of the resultant waveforms on the STDP response. This wave shape analysis can be used to help choose the pulse shape to be used in future circuit designs. Furthermore, the ability of the memristor to operate on a large dynamic range timescale, allows for one memristor type to be used for applications requiring short or long timescales. This provides an opportunity for a single integrated circuit to include both long and short timescale applications and to have only one type of memristor that needs to be integrated with the circuit during fabrication. Without this possibility, it would be unlikely that multiple memristor types which catered to different timescales could be fabricated together on one integrated circuit. The data retention measurements, or lifetime of a particular resistance state after the application of a spike pair, are currently in progress.",
"introduction": "Introduction Bioinspired neuromorphic computing has the potential of becoming realizable through the application of memristors (Chua, 2015 ) as artificial synapses (Jo et al., 2010 ; Chang et al., 2011 ; Erokhin and Fontana, 2011 ; Rose et al., 2011a , b ; Gaba et al., 2013 ; Serrano-Gotarredona et al., 2013a ; Subramaniam et al., 2013 ; Thomas, 2013 ; Mahalanabis et al., 2016 ). The spike-timing dependent plasticity (STDP) synaptic learning rule, inspired from the behavior of the biological neural system (Dayan and Abbott, 2001 ) and dominant in the brain, has been proposed and experimentally demonstrated with memristors acting as synapses by several groups over the past few years in many material systems, such as oxides (Yu et al., 2011 ; Wang et al., 2012a , b , 2016 ; Wu et al., 2012 ; Pickett et al., 2013 ; Mandal et al., 2014 ; Kim et al., 2015 ), chalcogenides (Li et al., 2013b ; Mahalanabis et al., 2014a , b , 2016 ; La Barbera et al., 2015 ), silicon (Jo et al., 2010 ; Subramaniam et al., 2013 ), organic materials (Alibart et al., 2012 ; Li et al., 2013a ; Cabaret et al., 2014 ; Luo et al., 2015 ), and even magnetic tunnel junctions (Krzysteczko et al., 2012 ). Illustrations of memristor effectiveness have also been shown in simulation and with transistor and/or complementary metal oxide semiconductor (CMOS)-based memristors (Rachmuth et al., 2011 ; Rose et al., 2011a , b ; Cruz-Albrecht et al., 2012 ; Noack et al., 2015 ) and graphics processing units (Snider et al., 2011 ). The exploration of new memristor materials systems is driven by the advantage of analog, memristor-based learning implementations compared to the digital-based learning, where the analog, memristor-based learning was shown to provide an improvement of at least a factor of 10 for power and density (Rajendran et al., 2013 ) over digital-based learning. The larger area and power requirement for CMOS-based memristors have driven the research into novel material-based memristor STDP to find a lower power/area alternative for neuromorphic computing. Some of the issues with previous experimental implementations of memristors in the synaptic role in the STDP application (Chang et al., 2011 ; Rose et al., 2011a , b ; Li et al., 2013a ; Subramaniam et al., 2013 ; Luo et al., 2015 ; Mahalanabis et al., 2016 ) include the lack of analog programmability of the memristor, high power requirements, and requirement of very specific programing spike shapes in order to effectively program the synaptic weights. Recent work, using a TaO x memristor as a synapse has demonstrated incremental switching in memristors, through the use of repetitive pulses and a pulse train with increasingly higher amplitudes. However, the pulses used in this study were limited to 100 ns pulse width (Wang et al., 2016 ). A similar TaO 5− x memristor was also used (Kim et al., 2015 ) to demonstrate the effects of incremental pulses on the memristor resistance tuning, as well as use of pulses in the range of 100 ns to 10 µs to demonstrate the STDP response. It should be noted that the incremental resistance programing response was also demonstrated in a chalcogenide-based memristor based on a phase-change mechanism, using 30 ns pulses in a five pulse train with increasing pulse amplitude from 1 to 1.8 V to increase resistance and −0.6 to −0.8 V to decrease resistance (Li et al., 2013b ). The memristor used in this work is based on the ion-conducting self-directed-channel (SDC) memristor, which has demonstrated lifetime endurance greater than one billion cycles, operation at temperatures of 150°C without degradation, and analog programmability (Campbell, 2017 ). This device is comprised of chalcogenide material layers (Figure 1 ) (Campbell, 2008a , b , 2017 ). It uses a Ge 2 Se 3 chalcogenide layer, which is activated for analog resistance tuning operation by Sn ions that migrate from an adjacent SnSe layer during the initial forming process (Campbell and Anderson, 2007 ; Devasia et al., 2010 , 2012 ). A layer of ternary GeSeAg is the ion source during operation. In contrast to other Ag-based GeSe or GeS ion-conducting device types (Mitkova and Kozicki, 2002 ; Kozicki and Mitkova, 2006 ; Kamalanathan et al., 2009 ; Waser et al., 2009 ; Wang et al., 2011 ; Mahalanabis et al., 2014a , b ; Rajabi et al., 2015 ; Ielmini and Waser, 2016 ), no photodoping or thermal annealing steps are required, simplifying the fabrication steps, and producing more consistent device operation. These differences also enable the device used in this work to withstand higher fabrication (at least 300°C) and operating temperatures (operation at 150°C is routinely performed). Additionally, this device can be integrated into a back-end-of-line (BEOL) CMOS process (Regner et al., 2009 ) making it compatible with CMOS architectures (Serrano-Gotarredona et al., 2013b ). Figure 1 Memristor device structure showing (from bottom to top) the bottom tungsten (W) electrode, the insulating nitride layer which defines the active layer contact area to the bottom electrode, the active layer (Ge 2 Se 3 ) followed by the SnSe, GeSeAg, and top tungsten electrode layers . The function of each layer is described in Campbell ( 2017 ). The advantage of the SDC memristor used in this work, over all of the memristors described to date in the STDP application is that it is the only memristor that has been shown to be simultaneously capable of (1) operation over nanoseconds to seconds timescale in STDP; (2) analog programmability over at least four orders of magnitude of resistance; (3) operation at high temperature (150°C); (4) cycling in excess of one billion times; and (5) demonstrated incorporation into a BEOL CMOS process. The scalability is predicted to be easily below 20 nm due to the one dimensional aspect of device operation (based on success at 27 nm node of CuTe-based 16 Gb memory; Fackenthal et al., 2014 ). The analog resistance programing capabilities of the SDC memristor used as a synapse are demonstrated in this work through the memristor’s synaptic weight change induced during the STDP experiment over the nanoseconds to seconds timescale, and the response to four different synaptic pulse shapes (Figure 2 ). These pulse shapes were used to explore the effects of the spike shape on the STDP response. As has been previously noted (Zhu et al., 2014 ; Qu et al., 2016 ), the ion-conducting memristive devices are logical candidates for this purpose since they have functional similarities to biological synapses in that both synapse types have a dependence on ion species to alter the synaptic strength. Figure 2 Spike shapes 1–4 used during spike-timing dependent plasticity testing . In addition to the STDP pulse shape tests, the memristor response to only the resultant waveform applied to one electrode while the other electrode was held at ground potential was also measured.",
"discussion": "Discussion Spike-timing dependent plasticity, as developed from Hebbian learning, requires that if the presynaptic spike arrives before the postsynaptic spike that the synapse exhibits long-term potentiation or increased synaptic weight and that if the presynaptic spike arrives after the postsynaptic spike that the synapse exhibits long-term depression or decreased synaptic weight. This effect has been shown for all of the pulse shapes and all the time scales presented in this paper. These change in weight plots do not perfectly match the plots experimentally measured for biologic systems (Bi and Poo, 2001 ), but since they contain regions of both potentiation and depression, they will be suitable for neuromorphic learning. Through the experiments in this paper, the nature of the potentiation and depression response of this memristor relative to different pulse shapes can be seen. Neither the null region observed for pulse shapes 2 and 3 nor the spiking region for pulse shape 4, near Δ T = 0 was initially expected since the differences between the pulse shapes are subtle. However, the differences are significant enough to impact the STDP results. This “tunability” of the STDP outcome, based on a slight change in the spike pulse, can be both beneficial and detrimental. Since electronics can be prone to noise, stray capacitance and to device mismatch during fabrication, among other possible interferences, it is possible that the spike pulse generated by the circuit could alter the STDP outcome in an undesirable way. Alternatively, one may be able to use this feature to, for example, suppress a circuit output by adding a slight slope or flat peak to the spike. Either way, this result demonstrates the significant influence the pulse shape can have on the STDP outcome and could prove to be a useful feature for circuit designers. Because there are essentially an infinite number of possible pulse shapes, and the pre- and postsynaptic pulses are not required to be identical in shape, this study provides guidance to allow the design of pulse shape pairs that will have the desired response. For some applications continuing learning (change in synapse resistance) when there is a large Δ T is desirable, and the range of Δ T to which this continues will depend on the rest of the design components. This was clearly shown in the “trailing edge cancelation” test. In this test, it was shown that a larger Δ T is possible by modification of the resultant waveform so that the trailing edge was completely suppressed. In the event that a circuit designer wants to maintain a higher Δ w % in a longer Δ T , suppression of the trailing edge would provide that opportunity. Additionally, partial suppression of the trailing edge pulse could achieve incremental weight changes if desired. Compared to a pure CMOS implementation of the STDP learning rule, memristor synapses require significantly less area on a chip than an equivalent CMOS-based synapse and are suitable for use in a cross bar array architecture. For example, a switched-capacitor realization of synapses in 28 nm CMOS was developed (Noack et al., 2015 ), which minimizes the leakage current problems present when CMOS-based architectures are scaled down in size. However, this system requires 0.36 mm 2 area and a power consumption of 1.9 mW for only 128 presynapses and 8,192 “stop-learning” synapses which corresponds to roughly 2.27 × 10 4 synapses/mm 2 and an energy requirement of 0.23 nJ to 0.23 mJ per synapse. By contrast, 16 Gb ReRAM ion-conducting, chalcogenide-based memory chips have been fabricated at the 27 nm node (Fackenthal et al., 2014 ) which have a total area of 168 mm 2 , including all periphery circuits required of a memory chip, as well as the memory elements. The memory elements on this chip, CuTe-based ion-conducting devices, are actually memristors, thus providing a good analogy to a high density memristor array. Using this entire chip area, a worst case approximation for synapse density would be at least 9.52 × 10 7 synapses/mm 2 , or a factor of 1,000 more synapses per mm 2 than the CMOS-based architecture, even taking into account all of the CMOS circuitry incorporated into the memory chip (periphery circuits and access transistors). Without knowing the power requirement of the 16 Gb ReRAM chip, it is not possible to know the energy requirement per memristor directly for this chip. However, an estimate of the energy requirements for the memristors used in this work, which are also chalcogenide ion-conductors, is provided in Table 1 for each experiment conducted and is a reasonable approximation to the energy requirements of the memristors in the 16 Gb ReRAM chip given that they are both chalcogenide-based ion-conducting device types. For pulse timing ranging from seconds to nanoseconds and considering the pulse shapes used in this work, the memristor displays lower energy requirements than the CMOS counterpart. In fact, as the memristor is driven with faster pulses, the overall energy requirement decreases even more (nanojoules to picojoules per memristor). From this, it can easily be seen that two major advantages of the memristor in the STDP application are: (1) denser area achievable and (2) lower energy requirements. In fact, from the energy requirements, it is clear that one of the advantages of operating a memristor on the faster timescale (nanoseconds), even though it is not a biological timescale, is that artificial synapses are not limited to slow, biological speeds and can therefore take advantage of the greatly reduced energy requirements of the memristors at fast speeds. Even though to date there has been no report of large scale integration of memristors as synapses, it is promising that there has been demonstrated a large scale memory chip which uses memristors (Fackenthal et al., 2014 ). While this report focused on the ion-conducting CuTe device as a binary memory, this device is a memristor and thus the memory chip does demonstrate the large scale feasibility of memristor incorporation into a large scale CMOS-based integrated circuit. Given this example, the major challenges of integration of new memristor device technologies into large scale integrated circuits can be overcome. These challenges include how to repeatably achieve novel material deposition for device-to-device and lot-to-lot consistency, as well as fundamental issues with trying to access a variable resistance device with transistors which, in the traditional architecture with the memristor in the source of the transistor, do not allow a constant switching voltage across a device due to a voltage divider between the access transistor ON resistance and the memristor."
} | 4,232 |
21791580 | PMC3143844 | pmc | 876 | {
"abstract": "ABSTRACT Carbonate chimneys at the Lost City hydrothermal field are coated in biofilms dominated by a single phylotype of archaea known as Lost City Methanosarcinales . In this study, we have detected surprising physiological complexity in single-species biofilms, which is typically indicative of multispecies biofilm communities. Multiple cell morphologies were visible within the biofilms by transmission electron microscopy, and some cells contained intracellular membranes that may facilitate methane oxidation. Both methane production and oxidation were detected at 70 to 80°C and pH 9 to 10 in samples containing the single-species biofilms. Both processes were stimulated by the presence of hydrogen (H 2 ), indicating that methane production and oxidation are part of a syntrophic interaction. Metagenomic data included a sequence encoding AMP-forming acetyl coenzyme A synthetase, indicating that acetate may play a role in the methane-cycling syntrophy. A wide range of nitrogen fixation genes were also identified, many of which were likely acquired via lateral gene transfer (LGT). Our results indicate that cells within these single-species biofilms may have differentiated into multiple physiological roles to form multicellular communities linked by metabolic interactions and LGT. Communities similar to these Lost City biofilms are likely to have existed early in the evolution of life, and we discuss how the multicellular characteristics of ancient hydrogen-fueled biofilm communities could have stimulated ecological diversification, as well as unity of biochemistry, during the earliest stages of cellular evolution.",
"conclusion": "Conclusions It is unclear whether the functional types identified in this study correlate with the genotypic microdiversity reported previously ( 25 ). It is possible that the methanogenic and methanotrophic cell types within the LCMS phylotype are not phylogenetically distinct; i.e., they may not represent ecotypes, as defined by Cohan ( 66 ). Differentiation into multiple cell types seems to imply a kind of niche partitioning, similar to that reported for vibrioplankton populations ( 16 , 17 ), but it may be that genetically identical LCMS cells are capable of filling any of multiple niches, as necessitated by local conditions within the biofilm. The same LCMS phylotype dominates chimneys differing in age by at least 100 years ( 9 ), so the LCMS community pangenome may encode the necessary adaptations to fill the many niches that arise during chimney development. The available data are still limited, but if further work corroborates this model, then the LCMS community could be considered one evolutionary unit comprising multiple ecological units, making it a “poster child” for a pluralistic view of microbial species ( 67 , 68 ). It has long been known that clonal bacterial populations can form multiple cell types ( 57 ); the results of this study implicate syntrophic metabolic interactions and LGT as potentially important aspects of cellular differentiation within biofilms. The Lost City biofilms are also excellent models for studying the diversification of life from a common ancestor, and they inhabit environmental conditions that were probably widespread on the ancient Earth ( 4 , 5 ). Frequent LGT during the early stages of evolution ( Fig. 4 ) ( 41 ) would have favored the utilization of identical biochemical building blocks, leading to a unity of biochemistry. Syntrophic metabolic interactions among closely related organisms ( Fig. 1 and 3 ) would have increased the overall efficiency and fitness of the community while promoting diversification into multiple ecological niches. In short, the ecological dynamics operating in Lost City biofilms today can help explain how the tremendous physiological diversity of life could have evolved from a community of organisms sharing a unity of biochemistry.",
"introduction": "Introduction The set of geochemical reactions known as serpentinization presents unique opportunities and challenges for biological communities, but little is known about the adaptations required of organisms that inhabit serpentinization-dominated habitats. Serpentinization is exothermic anywhere water reacts with the mineral olivine, and it results in warm, alkaline fluids and abiotic synthesis of hydrogen (H 2 ), methane, and larger hydrocarbons ( 1 , 2 ). It typically occurs in ultramafic oceanic crust and is also expected to occur on other planetary bodies where liquid water and olivine are present ( 3 ). Alkaline hydrothermal systems driven by serpentinization were probably widespread on the seafloor of the ancient Archean ocean, and H 2 and methane generated by serpentinization may have supported the earliest biological communities ( 4 – 7 ). A modern example of such an ecosystem is the Lost City hydrothermal field, where microbial biofilm communities attached to the carbonate chimneys are fueled by ≤90°C, pH 9 to 11, serpentinization-derived fluids rich in H 2 and methane ( 1 ). Therefore, the Lost City ecosystem provides an opportunity to directly investigate geochemical and biological processes thought to be important in the early evolution of life. The densely populated (up to 10 9 cells/g carbonate) mucilaginous biofilms within Lost City carbonate chimneys appear to be dominated by a single “species” of archaea. In the hottest and highest-pH zones of actively venting chimneys, nearly 100% of the archaeal sequences in 16S rRNA gene clone libraries and tag pyrosequencing data sets belong to a single phylotype referred to as Lost City Methanosarcinales (LCMS) ( 8 , 9 ). Furthermore, a FISH (fluorescent in situ hybridization) probe specific to the LCMS phylotype comprises up to 32.5% of the cells in the interior of actively venting chimneys. A FISH probe universal to bacteria detected only 4.2% of the cells in the same sample, so the LCMS phylotype represents ~80% of the cells detectable by FISH ( 8 ). The bacteria in these samples are dominated by taxa expected to be mesophilic and aerobic or microaerophilic and therefore would not occupy the same hot, anoxic zones as the members of the order Methanosarcinales ( 9 , 10 ). Such extreme dominance by a single species is remarkable, especially considering the longevity of the Lost City hydrothermal field. Radiometric dating of Lost City chimneys indicates that hydrothermal activity has been ongoing for at least 100,000 years and possibly up to 1 million years ( 11 , 12 ). Although we cannot know whether the LCMS phylotype has been present throughout that history, it has been found to dominate actively venting chimneys at least ~100 years old ( 9 ). In general, ecosystem stability is thought to require high biological diversity ( 13 ). If this model applies to the Lost City ecosystem, then two possible explanations for the extremely low diversity of the Lost City archaeal biofilms are apparent. One possibility is that the low-diversity biofilms are inherently unstable but do not experience environmental perturbations or do not experience competition. A potential example of this scenario is a deep mine ecosystem apparently consisting of a single bacterial species ( 14 ). The extreme energy limitation of this habitat, along with its isolation, may contribute to ecological simplicity/stability by reducing the number of available competitors. This explanation is consistent with the long-lived nature (compared to other hydrothermal systems) and physiological constraints of Lost City chimneys, but significant changes in chimney fluid chemistry, temperature, and mineralogy do occur over time ( 15 ). Therefore, the chimney biofilms are not completely protected from environmental perturbations, but the unusual stability and extreme conditions of their habitat may contribute to their low diversity. A second explanation (that is not mutually exclusive with respect to the first) is that the archaeal biofilms harbor more ecologically relevant diversity than is apparent in molecular surveys of 16S rRNA genes. It is now commonly accepted that organisms with nearly identical 16S rRNA genes may differ greatly in genomic content and physiological characteristics. For example, Vibrio splendidus isolates with >99% 16S rRNA gene sequence similarity can have widely different genome sizes and contain extensive heterogeneity in protein-coding genes ( 16 ). Hunt et al. ( 17 ) have suggested that these isolates represent recent or ongoing sympatric speciation events resulting from fine partitioning of resources. Niche partitioning among closely related species is particularly well documented for cyanobacteria. Several studies have delineated multiple cyanobacterial “ecotypes” that inhabit distinct regions within oceanic water columns ( 18 , 19 ) or within hot spring microbial mats ( 20 ). Although the 16S rRNA genes of cyanobacterial ecotypes may be highly similar, the genomic contents can vary widely ( 21 – 23 ). In another study, metagenomic sequencing revealed two coexisting strains of “ Cenarchaeum symbiosum ” (an archaeal symbiont of marine sponges) with >99% 16S rRNA gene sequence similarity ( 24 ). The authors conclude that the two strains may occupy different niches within the host sponge or else participate in a metabolic interdependence that maintains their coexistence. Recent evidence suggests that the LCMS phylotype also comprises multiple closely related, coexisting strains. Tag pyrosequencing of the V6 hypervariable region of the 16S rRNA gene of natural LCMS biofilm populations revealed hundreds of very rare sequences that differed by 1 to 2 bp from the most common sequence, possibly representing a dormant pool of variants ( 9 ). Clone library sequencing of the intergenic transcribed spacer (ITS) region between the 16S and 23S rRNA genes of LCMS biofilms revealed multiple ITS genotypes associated with the same V6 genotype ( 25 ). The relative abundance of its genotypes differed among samples, suggesting a possible ecological signature. In this paper, we present additional evidence for microdiversity within the LCMS biofilm population in the form of syntrophic physiological activity, cell morphological differentiation, and functional gene diversity. In particular, we focus on the ability of LCMS biofilms to generate or oxidize methane. LCMS has resisted laboratory cultivation because of the difficulty in replicating its in situ environmental conditions and/or because its growth rate is too low to be detected by typical laboratory experiments. Consequently, the physiology of LCMS is unknown. All known members of the order Methanosarcinales either generate or oxidize methane; therefore, LCMS cells are presumed to utilize either H 2 or methane, both of which are present at high concentrations (1 to 14 mmol/kg and 1 to 2 mmol/kg, respectively) in Lost City hydrothermal fluids. Carbon dioxide (CO 2 ), however, is virtually absent from Lost City fluids ( 2 ), so if methanogenesis occurs in active chimneys, it is either severely carbon limited ( 26 ) or else driven by a substrate other than CO 2 . In large part because of the unusual combination of high H 2 , high methane, and low CO 2 concentrations, previous studies have provided apparently conflicting data regarding whether LCMS cells are generating or oxidizing methane. Isotopic analyses of methane from Lost City fluids indicate a negligible biogenic contribution ( 2 , 27 ), suggesting that LCMS biofilms are unlikely to be methanogenic. However, isotopic ratios of archaeal lipids extracted from Lost City carbonate chimneys are inconsistent with LCMS being methanotrophic ( 26 ). Therefore, whether the LCMS phylotype represents methanogenic or methanotrophic organisms is unresolved. We present evidence for the ability of LCMS biofilms to mediate both methanogenesis and methanotrophy and describe observations that are consistent with cellular differentiation within the biofilm.",
"discussion": "RESULTS AND DISCUSSION Metabolic diversity: methane production and oxidation. To test the capability of LCMS to mediate methane production and oxidation, we incubated at high temperature ~3-g subsamples of carbonate (aragonite and calcite) chimneys in anaerobic 50-ml serum bottles (containing high-pH artificial seawater medium including sulfate) with either 13 C-labeled sodium bicarbonate (NaH 13 CO 3 , which exchanges with CO 2 ) or 13 C-labeled methane ( 13 CH 4 ). Incubations were stopped after 8 days, and the bottles were analyzed for the 13 C contents of CO 2 and methane by gas chromatography-mass spectrometry. The results ( Fig. 1 ) show that in the absence of oxygen and at 70 to 80°C and pH 9 to 10, both methanogenesis (NaH 13 CO 3 conversion to 13 CH 4 ) and methanotrophy ( 13 CH 4 conversion to 13 CO 2 ) proceeded at similar rates (5 to 150 nmol g −1 day −1 ). These rates are in the same range as those reported for anaerobic oxidation of methane (AOM) in flowthrough bioreactors (9 to 138 nmol g −1 day −1 ) ( 28 ) at 5°C but much lower than rates measured for bottle incubations at 9°C of methanotrophic microbial mats from the Black Sea (7,000 to 20,000 nmol g −1 day −1 ) ( 29 ). In situ rates at Lost City chimneys, however, could be much greater than those measured in this experiment due to the increased solubility of H 2 and methane at water depths of 700 to 900 m and because of the high fluid flux from the carbonate chimneys. Although analyses of biomarkers ( 30 ) and laboratory incubations ( 31 ) of marine sediments have indicated that AOM is possible at high temperatures, our results represent the first experimental evidence that AOM occurs in high-temperature hydrothermal chimneys. FIG 1 The rates (in nmol·C·day −1 g −1 of carbonate chimney material) of methane (CH 4 ) production (A) and oxidation (B) are both stimulated by the addition of hydrogen (H 2 ) gas to the incubation vessels. The H 2 stimulation effect was observed in two different carbonate samples under two sets of temperature and pH conditions simulating natural mixing of hydrothermal fluid and seawater. The biochemical pathway of AOM remains unknown ( 32 ), but genomic and biochemical analyses suggest that AOM utilizes many of the same enzymes required for methanogenesis but in the opposite direction ( 33 ). Accordingly, we hypothesized that high concentrations of H 2 would increase the rate of methanogenesis (4H 2 + CO 2 → CH 4 + 2H 2 O) and inhibit AOM in Lost City chimney biofilm samples. Intriguingly, our results ( Fig. 1 ) show that both methanogenesis and AOM were stimulated by the addition of H 2 , indicating that the two processes are not in competition. It has been shown that methane is reoxidized at trace levels during methanogenesis ( 34 ) such that the rate of methane oxidation is correlated with the rate of methane production ( 35 ). In such studies, however, methane oxidation rates are >200-fold lower than methane production rates, in contrast to our experiment, in which production and oxidation were roughly equal ( Fig. 1 ). Furthermore, Fig. 1 shows that methane oxidation does not require the addition of H 2 and is therefore not dependent on H 2 -fueled methanogenesis from CO 2 occurring in the same bottle. This result is consistent with the chemical composition of Lost City fluids, which are nearly devoid of CO 2 but rich in methane, and suggests that methane is the root carbon source fueling the biofilm community. Alternatively, it is possible that the methane oxidation in our experiment is a small fraction of the total methane production from a substrate other than CO 2 , such as formate or acetate. The stimulation of methane oxidation by H 2 as measured in our experiment, however, is not easily explained by this scenario because no known pathway involves the production of methane via oxidation of H 2 by formate or acetate. Regardless of the methanogenic substrate, it is clear that both methanogenesis and AOM occurred during the experiment and both reactions were stimulated by H 2 . Thus, it appears that Lost City chimney biofilms are capable of consuming both H 2 and methane, the two most abundant reactive species in Lost City fluids. Our previous work has shown that the LCMS phylotype comprises >80% of the cells detectable by FISH ( 8 ) and nearly 100% of the archaeal 16S rRNA gene sequences ( 9 ) in Lost City chimneys. It is possible that representatives of the ANME-1 group (anaerobic methane oxidizer group 1) of anaerobic methanotrophic archaea were present in our experiment, but ANME-1 sequences are infinitesimally rare (0.0001% of the 16S rRNA gene pyrotags) ( 9 ) in hot, actively venting chimneys such as the one used in this experiment. Furthermore, there is no evidence for the presence or activity of ANME-1 organisms in any high-temperature environment. Bacteria are also unlikely to have played a role in this experiment. Sequences corresponding to potential sulfate-reducing bacteria related to the genus Desulfotomaculum have been detected at low levels in Lost City chimneys, but there is no evidence linking these organisms to AOM ( 10 ). The most abundant bacterial sequences at this chimney derive from the Thiomicrospira group of microaerophilic, sulfur-oxidizing bacteria ( 9 , 10 , 36 ). These organisms are not expected to be active under the conditions of this experiment, nor are they likely to be involved in methane cycling. Therefore, both methanogenesis and AOM were most likely mediated by a single organism or a syntrophic assemblage of organisms belonging to the LCMS phylotype. Other studies have reported production of methane associated with AOM in marine sediments, suggesting that the associated archaea (all of which are phylogenetically affiliated with or closely related to Methanosarcinales ) are capable of both methane production and oxidation (reference 37 and references therein). This interpretation is consistent with micrometer-scale diffusion-reaction models of these habitats, which indicate that archaea which are expected to be methanotrophs may be net producers of methane when H 2 and methane are both available ( 38 , 39 ). Our experiments provide the first evidence for the stimulation of both methane production and oxidation by H 2 in a biofilm community dominated by a single phylotype. Metabolic diversity: formate and acetate. CO 2 is scarce (0.1 to 26 µmol/kg) in Lost City fluids ( 2 ), and the low methanogenesis rates shown in Fig. 1 could reflect substrate limitation at high pH. Indeed, isotopic measurements of lipids extracted from Lost City chimneys could indicate in situ carbon limitation if the chimney biofilms are dependent on CO 2 utilization ( 26 ). Recently it has been reported that formate and acetate are present in relatively high concentrations (36 to 158 µmol/kg and 1 to 35 µmol/kg, respectively) in Lost City fluids ( 40 ); therefore, these species may be more suitable as substrates for methanogenesis than is CO 2 . During a preliminary analysis of metagenomic data ( 41 ) from the same Lost City carbonate chimney sample with which the methane production and oxidation experiments were conducted, we identified an open reading frame encoding AMP-forming acetyl coenzyme A (CoA) synthetase (ACS). The Lost City ACS sequence has high overall sequence similarity to that from the Methanosaeta genus of aceticlastic methanogens ( Fig. 2 ) and shares several conserved residues diagnostic of a strong preference for acetate as a substrate ( 42 ) (see Fig. S1 in the supplemental material). Therefore, it is possible that at least some of the cells within the LCMS biofilm are capable of utilizing acetate as a carbon source and/or a methanogenic substrate. Further work is necessary to determine whether the consumption of acetate is also stimulated by the addition of H 2 and whether acetate may serve as a shuttle between methanogenic and methanotrophic cells ( 43 ). In addition to acetate, formate may also be utilized as a methanogenic substrate. Although no members of the order Methanosarcinales have been shown to utilize formate, the genome of Methanosarcina barkeri contains a complete formate dehydrogenase operon ( 44 ). We attempted, but failed, to amplify with degenerate primers the gene encoding the alpha subunit of formate dehydrogenase from Lost City chimney samples. Rapid cycling of methane with formate and/or acetate within the LCMS biofilm may account for the apparent discrepancies between isotopic ratios of methane and organic lipids described above, but further work is required to fully characterize the putative processes. FIG 2 Phylogenetic tree illustrating the relationship between AMP-forming ACS of aceticlastic methanogens and the metagenomic sequence from a Lost City carbonate chimney. The Lost City partial sequence includes 276 amino acids 65% identical and 79% similar to Methanosaeta thermophila Mthe_1413. Quartet puzzling support values are shown at nodes. Note that other aceticlastic methanogens (e.g., Methanosarcina acetivorans ) utilize a different pathway featuring the enzyme acetate kinase rather than AMP-forming ACS ( 75 ). Morphological diversity. Additional evidence for the ability of the LCMS biofilm to oxidize methane is shown by transmission electron micrographs (TEMs) of a carbonate chimney in thin section ( Fig. 3 ). Many of the cells in the TEMs contain stacks of what appear to be intracellular membranes ( Fig. 3B ). These structures bear some resemblance to the putative intracellular membranes observed in cylinder-shaped ANME-1 cells by Reitner et al. ( 45 ), who noted the similarity to membrane stacks utilized by aerobic methanotrophic bacteria. The cells in our TEMs that contain putative intracellular membranes are abundant and exhibit the characteristic sarcinal morphology of the Methanosarcinales ( 46 , 47 ), contrasting sharply with the expected morphology of either bacterial or ANME-1 cells. Furthermore, previous work has shown that 80% of the cells visible by microscopy via FISH belong to the LCMS phylotype ( 8 ). Therefore, it seems plausible that the LCMS phylotype includes both morphologies and that some LCMS cells oxidize methane with the aid of intracellular membranes. Future studies should test this hypothesis with single-cell imaging and sorting techniques ( 48 , 49 ). FIG 3 TEMs of carbonate chimney thin sections. (A and B) A cell type with sarcinal morphology is prevalent, but multiple cell types are present and closely associated with each other. Furthermore, a viscous matrix appears to surround each cell cluster and may aid in the attachment of cells to the carbonate minerals (bright white areas). (C) At least three lightly stained cells surround a densely stained cell near the bottom of the panel. The outlined area corresponds to the area shown in panel D, which shows that the sarcinal cells contain intracellular membrane stacks putatively involved in methane oxidation. Other remarkable features of the TEMs include (i) the close association between most of the cells and the carbonate minerals, (ii) the lightly stained matrix that encloses cell clusters, and (iii) the range of cell morphologies in a sample dominated by the LCMS phylotype. Densely stained cells that are enclosed by the lightly stained matrix exhibit the characteristic sarcinal morphology, appear to have recently divided, and commonly contain stacks of intracellular membranes. The more lightly stained, slightly smaller cells, in contrast, are enclosed by a grainier, darker matrix and are rarely in sarcinal packets. Some of the latter cells contain a central densely stained body, while others show more diffuse staining. We have previously reported low but significant levels of diversity within the ITS region of LCMS sequences ( 25 ). Diversity in the ITS region has been linked to important ecological and physiological variations within populations previously considered to be one species ( 17 – 20 ), so it would be very interesting to know whether the morphological diversity observed in our TEMs is correlated with genetic diversity. Alternatively, cell differentiation may arise within genetically homogenous populations; indeed, the diversity of cell morphologies in Fig. 3 is similar to that of multicellular aggregates seen in pure cultures of Methanosarcina mazei . Such aggregates commonly contain multiple cell types, some of which contain cytoplasmic granules similar to the densely stained bodies in Fig. 3 , and some have groups of intracellular tubules that may form membranes ( 47 ). A closely related species, Methanosarcina acetivorans , also forms cell clusters containing dense cytoplasmic granules and intracellular vesicles ( 46 ). This species can mediate trace levels of AOM during methanogenesis ( 34 ), so it is possible that the diversity of cell morphologies in Methanosarcina spp. and in the LCMS biofilm is related to the ability to mediate simultaneous methanogenesis and AOM. Functional gene diversity. We have presented evidence that the LCMS biofilm has adapted to utilize the most abundant electron donor (H 2 ) and two of the most abundant sources of carbon (methane and acetate) in its environment. Our evidence suggests that this biofilm community has also adapted to utilize the most abundant source of nitrogen in Lost City fluids—dissolved nitrogen gas (N 2 ). Although N 2 has not yet been measured in Lost City fluids, the N 2 concentrations at the Rainbow and Logatchev hydrothermal fields (1.8 mM and 3.0 mM, respectively), which are also hosted on serpentinized peridotites on the Mid-Atlantic Ridge, are 3- to 5-fold greater than those in ambient seawater ( 50 ). Fixed nitrogen, in contrast, is relatively scarce in end-member Lost City fluids (nitrate and ammonium concentrations are <4 µM [D. Butterfield, personal communication]). Although micromolar quantities of fixed nitrogen may be enough to sustain the densely populated LCMS biofilms, N 2 fixation is likely to occur because biofilms are typically nitrogen limited due to their ability to quickly remove fixed nitrogen from their surroundings ( 51 , 52 ). Furthermore, the highly reducing conditions within Lost City chimneys may be much more favorable for N 2 fixation than those in surface environments, where N 2 fixation is often inhibited by oxygen. These observations led us to investigate LCMS biofilms for the presence of genes encoding the nitrogenase enzyme complex, which is required for biological N 2 fixation. We identified a partial nitrogenase operon containing four open reading frames in preliminary metagenomic data from the same carbonate chimney sample utilized for the incubation experiments described above ( Fig. 4A ). The 5′ end consists of a partial open reading frame with 75% amino acid sequence identity to the 3′ end of the product of nifH from Methanothermobacter thermautotrophicus , a thermophilic, N 2 -fixing methanogen ( 53 ). The other open reading frames contain nifI 1 , nifI 2 , and nifD ( Fig. 4A ) in the same order as in M. thermautotrophicus , and the amino acid sequences they encode are 54 to 60% identitical to those encoded by the corresponding genes in that genome. Although nifH sequences have been detected in organisms unable to fix N 2 , our identification of an apparent nitrogenase operon containing at least four genes, including nifH , with high sequence similarity to a known thermophilic N 2 fixer strongly supports the existence of N 2 fixers in high-temperature Lost City chimneys. FIG 4 (A) A partial nitrogenase operon recovered from metagenomic sequencing of a carbonate chimney including the 3′ end of nifH (258 bp), nifI 1 (106 bp), nifI 2 (111 bp), and the 5′ end of nifD (323 bp). The sequence assembly contig (GenBank accession no. ACQI01004781) is included in the metagenomic data set described in references 36 and 41 . (B) Diversity of NifH sequences in clone libraries constructed from actively venting carbonate chimneys. NifH sequences from this study are underlined and followed by the number of additional NifH sequences with ≥97% amino acid identity in parentheses. Although the partial metagenomic nifH sequence in panel A does not overlap the region included in the nifH clone libraries, it most likely corresponds to clone 1408.9 in panel B. Quartet puzzling support values are shown at nodes. The tree is outgroup rooted with Plectonema boryanum FrxC, a dinitrogenase reductase-like protein involved in the light-independent reduction of protochlorophyllide. We further explored the phylogenetic diversity of nitrogenase genes in actively venting Lost City chimneys by targeted amplification and sequencing of nifH . A total of 181 nifH clones were sequenced from the four samples shown in Table 1 . These 181 NifH sequences fell into 31 clusters based on >97% amino acid identity. Eighteen clusters (representing 121 sequences and forming at least 5 distinct clades) are most closely related to methanogenic NifH and are shown in Fig. 4B . This represents a surprisingly wide range of NifH diversity, given that these samples were dominated by a single archaeal 16S rRNA gene phylotype and that no other methanogens have been detected at Lost City. The Lost City NifH sequences represented by clone 1408.30 ( Fig. 4B ) are identical to that of NifH expressed at 92°C by the N 2 -fixing methanogen isolated from deep-sea hydrothermal fluid on the Juan de Fuca Ridge in the northeastern Pacific ( 54 ). These sequences were detected only in the hottest Lost City carbonate chimney sampled, which vented fluids reaching 81°C at the time of sampling. Interestingly, this sequence is phylogenetically distinct from the NifH sequence encoded by the partial nitrogenase operon ( Fig. 4A ), indicating that there are at least two different nifH genes in Lost City chimneys with high sequence similarity to known thermophilic, methanogenic N 2 fixers. The phylogenetic affiliations of the many other NifH sequences in Fig. 4B are less clear, and further work is necessary to determine whether they encode functional nitrogenases. TABLE 1 Lost City carbonate chimney samples that were used to create nifH clone libraries Sample Marker Description Temp (°C) No. of unique NifH sequences/total no. sequenced Dominant archaeal 16S phylotype a LC1149 2 Active flange covered in biofilm 55 20/40 LCMS 3864-1537 2 Active flange covered in biofilm 53.5 28/50 LCMS LC1022 3 Active chimney covered in biofilm 75 26/52 LCMS 3881-1408 3 Active chimney covered in biofilm 81 21/39 LCMS a From Schrenk et al. ( 8 ) and Brazelton et al. ( 9 ). Regardless of their expression and function, it is clear that Lost City chimneys contain a remarkable diversity of nifH sequences, given that the most likely host for these genes is the LCMS phylotype. One possible explanation is that LCMS cells have variable genomes, with each cell containing different nifH genes. Members of the order Methanosarcinales are known to have acquired a significant fraction of their genomes, including nitrogenase genes, via lateral gene transfer (LGT) ( 55 ), so it may be that many of these nifH genes were acquired via lateral transfer from other organisms. We have previously reported an extremely high abundance and diversity of transposase sequences in Lost City metagenomic data ( 41 ), which is consistent with LGT influencing the genomic content of LCMS biofilms. Nitrogen fixation and LGT may be commonly associated with AOM, as two recent studies have also recovered a wide diversity of nifH genes from N 2 -fixing anaerobic methanotrophic archaea in marine sediments ( 48 , 49 ). Differentiation and syntrophy within a single-species biofilm. All of the evidence that has been presented in this paper and published previously ( 1 , 9 , 10 , 25 ) is consistent with the model illustrated in Fig. 5 , in which the LCMS phylotype includes genetically heterogenous populations of cells that are differentiated into multiple cell types engaging in metabolic cooperation and LGT. Metabolic and genetic interactions among organisms within biofilm communities are well documented. For thorough reviews, see references 56 to 58 . Because individual cells benefit in numerous ways from their membership in a diverse community, a successful biofilm requires at least some level of cooperation or, at the very least, noncompetition among the members of the community. Although selection still acts on competing lineages, the ability to participate in the community is also a selectable trait ( 58 ). Therefore, biofilm communities can be considered to be inherently social and multicellular. The observations reported in this paper highlight the potential for multicellular characteristics in biofilms even when they are dominated by a single 16S rRNA gene phylotype (e.g., LCMS). Our results are consistent with the provocative study by Boles et al. ( 59 ), who showed that during biofilm growth, a clonal population of Pseudomonas aeruginosa rapidly generates specialized subpopulations that have increased fitness. FIG 5 Hypothetical syntrophic reactions between methanogenic and methanotrophic cell types that are consistent with experimental data. The observation that both the production and oxidation of methane (CH 4 ) are stimulated by H 2 is explained by the dependence of methane oxidation on the uptake of its waste product by a nearby H 2 -utilizing cell. CO 2 , formate, and acetate are potential transfer molecules. The terminal electron acceptor of the overall reaction is unknown, but sulfate (SO 4 2− ) is a likely candidate. In the case of the LCMS biofilms, differentiation and syntrophy likely evolved as adaptations to maximize the metabolic potential of Lost City fluids, which are among the most H 2 - and methane-rich fluids ever measured. To our knowledge, no organisms capable of utilizing both H 2 and methane have been characterized; indeed, the simultaneous consumption of H 2 and methane does not seem thermodynamically sensible for a single cell. This process is feasible, however, for a population that is differentiated into H 2 -utilizing and methane-utilizing cell types that consume each other’s waste products ( Fig. 5 ). Any other explanation of the data would involve novel and seemingly implausible metabolic pathways. This putative differentiation might involve only a few specialized cells, akin to heterocyst formation in cyanobacteria, or all cells may differentiate into one of the possible types according to environmental stimuli. The exact reactions and metabolites involved in our proposed syntrophic assemblage have yet to be determined, but initial metagenomic data indicate that acetate may play an important role as a carbon source, methanogenic substrate, or metabolite shuttle between cell types ( Fig. 2 ). The terminal electron acceptor is also unknown, but the most obvious candidate is sulfate, which is surprisingly abundant in end-member Lost City fluids (1 to 4 mM) and increases with seawater mixing ( 1 ). LGT, as evidenced in the LCMS biofilms by the high diversity of nitrogenase ( Fig. 4 ) and transposase ( 41 ) genes, may be directly related to the multicellular nature of the LCMS biofilm community. In addition to increasing the genetic diversity of the community by introducing novel DNA from unrelated organisms, a moderate rate of LGT within a biofilm community is also predicted to stabilize the coexistence of multiple phenotypes ( 60 ). Thus, LGT provides another mechanism, in addition to metabolic interactions, whereby genetic variants can evolve into codependent populations. Biofilms and the unity of biochemistry. There is no reason to suppose that the multicellular nature of biofilms should be considered a novelty or a recently derived trait of a few lineages. Although traditional microbiological research has focused on clonal cultures in nutrient-rich broth, the predominant mode of growth for most microorganisms in natural environments is as surface-attached, multispecies biofilms ( 61 ). It is therefore plausible that the earliest biological communities utilized the biofilm mode of growth ( 56 ) and that evolutionary dynamics on the ancient Earth were strongly shaped by physiological interactions and LGT events within biofilms. In particular, we propose that these processes acting within ancient biofilm communities would have stimulated the diversification of organisms that shared biochemical pathways. One of the most fundamental observations of biology is the unity of biochemistry ( 62 ); i.e., all forms of life on Earth use the same biochemicals and relatively minor variations of the same biochemical pathways. The unity of biochemistry strongly argues that all modern organisms have a common ancestor. The nature of the most recent common ancestor is widely debated, but it would be naive to assume that it consisted of only a single cell living in isolation. On the contrary, it seems plausible that it inhabited a biofilm-like community in which cells interacted with each other by exchanging nutrients, metabolites, and genes. Differentiation of certain populations to fill niches within the community would have led to the diversification of life from a common ancestral pool. It has been previously proposed that promiscuous LGT within an ancestral “multiphenotypical population” ( 63 ) of “precells” ( 64 ) or “progenotes” ( 65 ) could explain the vast diversity observed in extant organisms within the unity of biochemistry. We suggest that the LCMS biofilms within Lost City carbonate chimneys, where a single ancestral population appears to have differentiated into metabolically linked cells undergoing frequent LGT, are reminiscent of this hypothetical ancestral community and provide an intriguing model for the study of potentially ancient evolutionary processes. Conclusions It is unclear whether the functional types identified in this study correlate with the genotypic microdiversity reported previously ( 25 ). It is possible that the methanogenic and methanotrophic cell types within the LCMS phylotype are not phylogenetically distinct; i.e., they may not represent ecotypes, as defined by Cohan ( 66 ). Differentiation into multiple cell types seems to imply a kind of niche partitioning, similar to that reported for vibrioplankton populations ( 16 , 17 ), but it may be that genetically identical LCMS cells are capable of filling any of multiple niches, as necessitated by local conditions within the biofilm. The same LCMS phylotype dominates chimneys differing in age by at least 100 years ( 9 ), so the LCMS community pangenome may encode the necessary adaptations to fill the many niches that arise during chimney development. The available data are still limited, but if further work corroborates this model, then the LCMS community could be considered one evolutionary unit comprising multiple ecological units, making it a “poster child” for a pluralistic view of microbial species ( 67 , 68 ). It has long been known that clonal bacterial populations can form multiple cell types ( 57 ); the results of this study implicate syntrophic metabolic interactions and LGT as potentially important aspects of cellular differentiation within biofilms. The Lost City biofilms are also excellent models for studying the diversification of life from a common ancestor, and they inhabit environmental conditions that were probably widespread on the ancient Earth ( 4 , 5 ). Frequent LGT during the early stages of evolution ( Fig. 4 ) ( 41 ) would have favored the utilization of identical biochemical building blocks, leading to a unity of biochemistry. Syntrophic metabolic interactions among closely related organisms ( Fig. 1 and 3 ) would have increased the overall efficiency and fitness of the community while promoting diversification into multiple ecological niches. In short, the ecological dynamics operating in Lost City biofilms today can help explain how the tremendous physiological diversity of life could have evolved from a community of organisms sharing a unity of biochemistry."
} | 10,023 |
39549075 | PMC11568993 | pmc | 877 | {
"abstract": "Abstract Biochemical methane potential (BMP) test is an important tool to evaluate the methane production biodegradability and toxicity of different wastes or wastewaters. This is a key parameter for assessing design and feasibility issues in the full-scale implementation of anaerobic digestion processes. A standardized and storable inoculum is the key to obtain reproducible results. In Uruguay, a local enterprise dedicated to design and install anaerobic digesters operated a lab-scale bioreactor as a source of biomass for BMP tests, using a protocol previously described. This reactor was controlled and fed with a mixture of varied organic compounds (lipids, cellulolytic wastes, proteins). Biomass was reintroduced into the reactor after BMP assays to maintain a constant volume and biomass concentration. The aim of this work was to evaluate how the microbial community evolved during this operation and the effect of storing biomass in the refrigerator. The composition of the microbial communities was analyzed by 16S rRNA amplicon sequencing using primers for Bacteria and Archaea. The methanogenic activity was determined, and the methanogens were quantified by mcr A qPCR. One sample was stored for a 5-month period in the refrigerator (4 °C); the activity and the microbial community composition were analyzed before and after storage. Results showed that applying the reported methodology, a reliable methanogenic sludge with an acceptable SMA was obtained even though the reactor suffered biomass alterations along the evaluated period. Refrigerating the acclimatized biomass for 5 months did not affect its activity nor its microbial composition according to the 16S rRNA gene sequence analysis, even though changes in the mcr A abundance were observed. Key points • The applied methodology was successful to obtain biomass suitable to perform BMP assays. • The microbial community was resilient to external biomass addition. • Biomass storage at 4 °C for 5 months did not alter the methanogenic activity. Graphical Abstract \n Supplementary Information The online version contains supplementary material available at 10.1007/s00253-024-13305-0.",
"introduction": "Introduction Anaerobic digestion (AD) is a consolidated technology applied worldwide to obtain methane from wastes and wastewaters (Ampese et al. 2022 ; Grando et al. 2017 ). High organic content wastes and wastewaters can be converted into energy and fertilizers. The application of this technology is essential to move towards a circular economy. In Latin American countries whose economy is based on food production, the production of energy from the large amount of waste generated is essential to obtain more sustainable production (Moreda 2016 ). The AD process includes four major microbial steps, i.e., hydrolysis, fermentation, acetogenesis, and methanogenesis, which are accomplished jointly by different types of microorganisms in tandem with their syntrophic interactions (Amin et al. 2020 ). In the acetogenic step, the acetogenic bacteria convert the organic acid products into acetate and hydrogen which become available for methanogenic archaea consumption (Schink 1997 ). Methanogenic archaea are critical for efficient degradation and play an important role in the syntrophic process. On the other hand, microbial composition of the inoculum source deterministically contributes in shaping the community structure and specific ecosystem function (De Vrieze et al. 2014 ; T. Liu et al. 2017 ; Perrotta et al. 2017 ). Thus, selecting an appropriate inoculum is important to ensure degradation of a wide variety of substrates to methane. Taking this process to full scale needs high knowledge of these complex process variables (Carballa et al. 2015 ). This knowledge allows us to monitor its proper functioning and therefore to obtain a predictable and stable product (Werner et al. 2011 ). To apply this technology for the treatment of wastes and wastewaters, it is necessary to determine the potential to obtain methane using these wastes. BMP tests are simple and inexpensive procedures employed to evaluate the biodegradability and toxicity of different organic matter feed source to anaerobic treatment process and one of the key parameters for scaling up the process to further assessing design, economic, and managing issues for the full-scale implementation of anaerobic digestion processes (Angelidaki et al. 2009 ). In these tests, variations in the microbial community composition, different source, and origin inocula could lead to different biodegradability results (Raposo et al. 2011a , b ). Thus, guaranteeing a proper microbial community acclimated to the test conditions or adapted to the substrate is necessary to obtain reliable and reproducible results for BMP tests. Insufficient activity or quality of the inoculum can lead to wrong results (Steinmetz et al. 2016 ). The specific methanogenic activity (SMA) is another parameter of great importance which can be described as a measure of the specific chemical oxygen demand (COD) digestion or biotransformation rate. The degradation rate is higher when the inoculum is adapted to a substrate (Koch et al. 2017 ). Also, higher SMA results reflect higher capacity of the inoculum to degrade the organic matter into methane (Moreno-Andrade and Buitrón 2004 ). Regular determination of the sludge SMA provides an idea of its methanogenic performance and evolution. This value is estimated from the methane production rate and the amount of inoculum added and it varies with substrate type, operating temperature, etc. (Hussain and Dubey 2014 ). It is reported that an appropriate inoculum should have a minimum SMA value (with acetate as substrate) between 0.10 and 0.15 g CH 4 -COD/g VSS·day (Angelidaki et al. 2009 ; Hussain and Dubey 2014 ). Additional information can be gained by studying the methanogenic community abundance inside the acclimatized and maintained reactor by qPCR (Oka et al. 2011 ). For this purpose, the mcr A gene was used as a key functional marker. This gene encodes part of the terminal enzyme complex in the methane production pathway and is thought to be unique and ubiquitous in methanogens which makes it perfect to target their detection (Cisek et al. 2023 ; Franchi et al. 2018 ; Wäge-Recchioni et al. 2020 ). Obtaining an adapted inoculum to perform BMP tests is a common problem in laboratories and enterprises focused on testing the production of methane using different wastes, especially in Latin American countries where biogas plant availability is limited. For that, Steinmetz et al. ( 2016 ) proposed a protocol to enrich and maintain an active hydrolytic and methanogenic community to perform BMP tests. The authors propose to operate a lab-scale bioreactor inoculated with a mixture of different methanogenic sludge and fed with complex organic biomass easy to obtain for laboratories. As the objective was to obtain a microbial community with the capacity to degrade complex wastes, the feeding was carried out using organic matter balanced in its composition of lipids, proteins, and cellulose. For that, the authors use a mixture of dried grass (30% of VS), milk powder (25% of VS), swine feed (maize basis) (30% of VS), and vegetable oil (15% of VS). An adapted biomass with good methanogenic activity was obtained after 25 days of operation. BMP tests performed using the adapted inoculum and microcrystalline cellulose, maize silage, and dried distilled grain gave values similar to those reported in the bibliography indicating that the strategy was successful. Once the sludge is acclimatized, the possibility of preserving it without altering its characteristics (microbial composition thus its metabolic performance) would represent an opportunity to have availability of a standardized inoculum for reproducible BMP results. This is the reason why many research groups have been studying different temperature long-term preservation methods (Astals et al. 2020 ; Bhattad et al. 2017 ; Castro et al. 2002 ; Heerenklage et al. 2019 ; Soldano et al. 2019 ). Although these methodologies are frequently applied in laboratories, it is not yet known how the microbial community adapts during this acclimatization process and how much long-term storage at 4 °C affects the microbial community. In the current study, therefore, we envisioned the opportunity to monitor a microbial community adapted within a reactor acclimatized with the main goal of producing biomass by the company NETUM S.R.L. We aim to give light on how the microbial community is adapted during the acclimatization and storage procedures, a routine methodology performed in anaerobic digestion studies. For that purpose, we asked ourselves the following questions: (1) Does the sludge’s activity sustain in time? (2) How does the microbial community evolve during acclimatization? (3) Does long-term storage at 4 °C affect the methanogenic activity and the microbial community composition?",
"discussion": "Discussion Biomass with good methanogenic activity was obtained and sustained Both the SMA and the qPCR results showed that the built-up community was enriched in methanogens and that they were maintained during the reactor operation. The acetoclastic methanogenic activity reached values of 0.241 g CH 4 -COD/gVSS·day which is considered an appropriate methanogenic activity for a sludge (Angelidaki et al. 2009 ; Callejas et al. 2022 , 2019 ; Hussain and Dubey 2014 ). The proportion of methanogens determined by qPCR also showed values similar to those obtained in previous studies for functional reactors. Callejas et al. ( 2019 ) reported similar values for a full-scale UASB reactor designed for dairy wastewater treatment where its microbiome was monitored during the startup and operation processes applying the same qPCR technique. We observed an inconsistent relationship between the SMA and the mcr A copy number/ng DNA in some of the samples (Morris et al. 2014 ). While the SMA increases in time, a one order decrease in magnitude was observed in the mcr A abundance in samples taken at days 186, 218, and 239. This drop in mcr A abundance could be due to a decrease in the proportion of hydrogenotrophic methanogens (detected by mcr A quantification but not detected by the acetoclastic methanogenic activity), or to the difficulty of the PCR methods to detect all the methanogens. The lack of consistency between both techniques was previously reported (Dellagnezze et al. 2023 ) and it is a problem caused by the bias of the different methods. The hydrolytic and fermentative capacity of the biomass was tested in a Latin American interlaboratory assay using four compounds with different chemical composition: microcrystalline cellulose, ground coffee beans, cornmeal, and gelatin (Steinmetz et al. 2020 ). The results obtained in our laboratory were satisfactory according to the statistical analysis evaluation (Supplementary Material: Fig. S5 ). These results on this variety of complex substrates tested the entire community performance indicating a well-functioning hydrolytic and methanogenic community. A microbial community with hydrolytic and methanogenic microorganisms was enriched and maintained during reactor operation Both the bacterial and archaeal microbial community analysis showed a community with a taxonomic composition of a well-established methanogenic community. The Firmicutes, Synergistota, Actinobacteria, Chloroflexota, and Proteobacteria phyla dominated the bacterial community. These phyla are key for degrading complex organic matter as carbohydrate, lipids, protein, etc., to simpler compounds more accessible to methanogens (Li et al. 2015 ; Liu et al. 2009 ). Their presence could be an indicator of a well-adapted to a wide variety of substrates sludge. At the same time, microorganisms syntrophic to hydrogen-consuming methanogen, such as those belonging to Synergistota phylum, were also detected in these samples (Ito et al. 2011 ; Militon et al. 2015 ). These results confirm that the chosen strategy to obtain an adapted biomass to perform BMP tests was successful. When the analysis was performed at Family level (Fig. 4 A), we observed a microbial community composed mainly of hydrolytic and fermentative microorganisms ( Synergistaceae , Anaerolineaceae , Christensenellaceae , Peptostreptococcaceae , Clostridiaceae ), along with syntrophic oxidation microorganisms ( Syntrophomonadaceae , Synrophorhabdaceae , Smithellaceae ) (Rosenberg et al. 2013 ; McInerney et al. 2008 ). The family Desulfotomaculaceae , composed of sulfate-reducing bacteria, was detected only in 3 out of 14 samples analyzed showing a very high relative abundance in sample B131 (25.8%) and B61 (8%) (Barton and Hamilton 2007 ), indicating that sulfate may be present in the complex digested substrates. Whenever sulfate is present, it competes with methanogenesis in an anaerobic ecosystem, but, on the other hand, in its absence, species from this family are also known to perform syntrophic oxidation of organic acids, an important role for methanogenesis. In the present case study, sulfate content in the reactor was not measured; then, this genus role cannot be predicted from the 16S rRNA gene sequence alone. De Francisci et al. ( 2015 ) detected an increase in Desulfotomaculum in a reactor overfed with proteins, in our case the increase in this family occurred after the addition of BMP tests digestate using as substrates: gelatin, cornmeal, coffee, microcrystalline cellulose (at day 127). Within the archaeal community, the genus Methanosaeta , acetoclastic methanogens, predominated showing high abundance in some samples (from 32.7 to 87.6%). This genus was reclassified as Methanothrix and it is a known acetoclastic methanogen (Carr et al. 2018 ; Chen and He 2015 ). Several previous studies have noted the dominance of Methanosaeta suggesting that this occurrence might be widespread in the anaerobic digestion process (Moertelmaier et al. 2014 ; van Haandel et al. 2014 ; Yilmaz et al. 2014 ). Another microorganism, represented by ASV 597, exhibited a high level of dominance. This ASV could not be classified using the QIIME analysis; however, using the BLAST tool, we found a high-sequence homology with the sequence corresponding to Methanothrix soehngenii , which is known as an acetoclastic methanogen (Ferry 2020 ). According to Fig. 4 , the following genera from hydrogenotrophic archaea were observed: Methanoregula , Methanospirillum , Methanobacterium , Methanobrevibacter , Methanocorpusculum , along with several unknown genera within the Methanomicrobiales order (Garcia et al. 2022 ). The relative abundance of each of these genera was less than 10%. Therefore, hydrogenotrophic methanogenesis can occur but in lower dominance. These results showed that acetoclastic methanogenesis predominates in all samples, which is important for a stable process performance as according to literature; this population accounts for most of the methane production in anaerobic digestion (Chen and He 2015 ). Disturbances occurred during the external biomass addition As the reactor was operated to generate biomass stock to be used in BMP tests, the reactor was periodically disturbed by biomass extraction. Consequently, to stabilize the biomass level, we reintroduced the biomass back into the reactor after BMP tests were conducted and whenever it was available in the lab. Different substrates were used in each test, so it was reasonable to expect that the new biomass adapted to the different substrate digestion. To evaluate the effect of these perturbations in the microbial communities, we also analyzed two of the added to the reactor biomass samples (EB1 and EB2). These two samples showed different microbial composition according to the 16S rRNA gene analysis (Fig. 3 A, B). As these biomass samples were taken from BMP tests performed with different substrates, we expected a different selection of microorganisms during each batch test. Sample EB1 was taken from a test performed using poultry manure and breading bed wastes, and presented high abundance of Firmicutes, while sample EB2 was taken from a BMP test performed with chicken feathers and enzymatically hydrolyzed chicken feathers. The microbial composition in sample EB2 was similar to the biomass composition within the reactor. Samples B61, B111, B131, B321, and B347 were taken shortly after the addition of external biomass. Samples B61 and B131 presented a higher dominance of Firmicutes compared to other samples, while in sample B321 low abundance phyla increased their proportion (Fig. 3 A). It has been postulated that substrate type determines differences in the reactor microbial phylogenetic structure (Zhang et al. 2014 ). Therefore, the relationship between the external biomass sludge characteristics and the affected samples from the present case study was analyzed. In the case of the initial seed sludge mixture prepared on day 1 (Table 1 ), it was composed globally of 25% sludge from poultry wastes origin and 5% of cattle wastes origin. The next external biomass, added on day 47 and 79 to the reactor, was basically cattle manure and anaerobic manure digester sludge. Firmicutes and Bacteroidota were reported as dominant in most manure digesters as expected coming from cattle ruminal content (Liu et al. 2009 ; Narihiro et al. 2015 ; Pitta et al. 2010 ). Consequently, this could explain Firmicutes predominance in samples B1, B61, and B131. On the other hand, Bacteroidota phylum was practically absent in all samples. The latest biomass addition to the reactor was poultry wastes origin samples (poultry manure, poultry litter, feathers). According to Zhang et al. ( 2018 ), chicken manure contains diverse gut microbes, mainly species derived from Proteobacteria. On the other hand, poultry litter, which is primarily a mixture of bedding materials and bird excreta, are reported to contain, depending on the number of reuse cycles, members of the Proteobacteria phylum (Cressman et al. 2010 ). Moreover, feathers are high content protein structures (dry feathers contain 91% protein) (Mézes and Tamás 2015 ), and substrates with these characteristics could be contributing with microorganisms related to protein degradation (Javůrková et al. 2019 ; Shawkey et al. 2005 ). Firmicutes was the predominant phylum in samples 321 and 347 (Fig. 3 A) even though Proteobacteria is known to be one of the dominant phyla in anaerobic digesters (Li et al. 2015 ; Bovio-Winkler et al. 2021 ) and the protein-rich characteristics of the poultry biomass reinoculated into the reactor on days 284 and 339 (Table 3 ). In summary, a highly diverse bacterial community dominated by fermenting and hydrolyzing bacteria was observed throughout the studied period. Despite noticing variations in relative abundance during the evaluated period, the sludge’s methanogenic activity did not decline (Fig. 2 ), which indicates that the acetoclastic methanogenic population remained active. The results obtained demonstrated resilience and robustness of the microbial community to the disturbance produced by the biomass addition episodes. Disturbance and community stability are necessarily related, as stability is defined as the community’s response to disturbance (Shade et al. 2012 ). Probable functional redundancy within the community may explain sludge’s activity maintenance throughout the whole evaluated period (Allison and Martiny 2009 ). Effect of the storage at 4 °C Previous studies have been reported for long-term preservation methods at room temperature, refrigerating, freezing, and lyophilization of inoculum samples prior to its use in anaerobic batch tests (Astals et al. 2020 ; Bhattad et al. 2017 ; Castro et al. 2002 ; Heerenklage et al. 2019 ; Soldano et al. 2019 ). Results from these experiments demonstrated high biogas or methane production recovery for room temperature and refrigerated (4 °C) samples while biogas production and the lag phase were drastically affected in the case of frozen or lyophilized samples. Contrary to this, another report suggested a drop in methane yields and slower production kinetics after 1-month storage for both refrigerated and frozen samples (4 °C and − 20 °C respectively) (Hagen et al. 2015 ). From the abovementioned reports, only a few employed SMA tests to evaluate storage impact in the inoculum activity (Astals et al. 2020 ; Bhattad et al. 2017 ; Castro et al. 2002 ). Instead, most of them evaluated inoculum’s quality based on BMP tests results which can be less sensitive to changes in the methanogenic activity than metabolic assays (e.g., SMA) as they comprise the entire degradation process while metabolic assays directly target a specific biological step (e.g., methanogenesis) (Astals et al. 2020 ). Castro et al. ( 2002 ) analyzed the entire anaerobic population during an SMA test using glucose as the substrate and specifically targeted acetoclastic methanogens by employing acetate as the substrate. After a 5-month storage experiment, they observed lower SMA results for the experiments tested with glucose than for those with acetate. From this, it could be concluded a higher resistance from the acetoclastic methanogens to different storage methods. The lower SMA results obtained with glucose might be attributed to the lower sensitivity previously mentioned to changes in the inoculum activity than metabolic assays. This group argued that the selection of the preservation method and the storage characteristics are inoculum specific. According to our results, the microbial community was not affected, but, as this analysis was based on DNA, we cannot infer the activity was not affected. The results from the present work were consistent with those reported by several groups. Acetoclastic activity remained stable after a 5-month storage at 4 °C (Zitomer et al. 2007 ). To our knowledge, there are no previous studies monitoring an acclimatization to produce methane microbial community sludge at this storage temperature for a 5-month period combining SMA, qPCR, and massive sequencing tools. The increase in the methanogens proportion after storage could be explained by the higher sensitivity to low temperature of the hydrolytic and fermentative microorganisms. If their degradation rate is higher, their DNA will not persist in the sample; consequently, mcr A copy number versus total DNA proportion will increase. Despite the increase shown in the methanogens quantification after storage by qPCR, the SMA maintained. A possible explanation for this fact is that, as the SMA is determined by VSS gram even though DNA was degraded, the amount of biomass remains, or the change is undetectable by the applied method. Further work is needed to confirm this hypothesis. The abovementioned results contradict those reported by Hagen et al. who argue a decay in the sludge activity after a 1-month 4 °C storage (Hagen et al. 2015 ). In the cited article, the most abundant phyla were hydrogenotrophic Methanomicrobiales (genus; Methanoculleus ) and “Miscellaneous Crenarchaeota Group” (MCG) (clone GrfC26) the first 2 months of storage, but longer storage led to a shift in dominance to acetoclastic methanogens, Methanosarcinales (genus; Methanosarcina ). In the present case study, acetoclastic Methanosaeta (now reclassified as Methanothrix ) was the predominant archaeal genus before and after storage (Fig. 4 B). From these observations, acetoclastic community resistance to refrigeration could be argued explaining why no archaeal community differences were observed. On the other hand, as we did not test the BMP of complex substrates after their storage, we cannot affirm other steps in the anaerobic chain were not affected; this should be further investigated. In this work, the effect of storage at 4 °C was evaluated as it is a common practice in laboratories which perform BMP tests. It would be interesting to evaluate other sludge storage conditions such as preservation at room temperature, freeze-drying, among others, therefore, selecting the best storage condition which does not affect the microbial community composition in time. Deeper work is necessary to evaluate how the different storage methods affect the microbial communities. In summary, applying the reported methodology to produce an appropriate inoculum guarantees a microbial community able to perform BMP tests on a wide variety of substrates. The microbial community composition was affected by external biomass addition episodes; despite this, the sludge’s activity as well as its methanogens abundance maintained over the evaluated period. The microbial profile changed predominating different microorganisms according to the external sludge origin, returning to its initial structure rapidly. Storing an acclimatized inoculum in the refrigerator for 5 months did not affect its methanogenic activity nor its microbial composition although changes in mcr A abundance were observed through qPCR, likely attributable to the degradation of DNA from other bacteria."
} | 6,290 |
30374283 | PMC6196279 | pmc | 878 | {
"abstract": "In computational neuroscience, synaptic plasticity learning rules are typically studied using the full 64-bit floating point precision computers provide. However, for dedicated hardware implementations, the precision used not only penalizes directly the required memory resources, but also the computing, communication, and energy resources. When it comes to hardware engineering, a key question is always to find the minimum number of necessary bits to keep the neurocomputational system working satisfactorily. Here we present some techniques and results obtained when limiting synaptic weights to 1-bit precision, applied to a Spike-Timing-Dependent-Plasticity (STDP) learning rule in Spiking Neural Networks (SNN). We first illustrate the 1-bit synapses STDP operation by replicating a classical biological experiment on visual orientation tuning, using a simple four neuron setup. After this, we apply 1-bit STDP learning to the hidden feature extraction layer of a 2-layer system, where for the second (and output) layer we use already reported SNN classifiers. The systems are tested on two spiking datasets: a Dynamic Vision Sensor (DVS) recorded poker card symbols dataset and a Poisson-distributed spike representation MNIST dataset version. Tests are performed using the in-house MegaSim event-driven behavioral simulator and by implementing the systems on FPGA (Field Programmable Gate Array) hardware.",
"introduction": "1. Introduction One goal of neuromorphic engineering is to map efficiently neurocomputational algorithms onto compact, low power, and fast hardware, while preserving satisfactorily the functionalities of the theoretical algorithms. The main first question (digital) hardware neuromorphic engineers ask themselves is about the minimum required bits to represent parameters and states. Theoretical neurocomputists, which use as tools conventional computers, use typically the full 64-bit floating point precision available to develop and study their algorithms. However, using 64-bit floating point precision for neuromorphic hardware, where typically massive parallelism is physically implemented, imposes a severe resources penalty not only for memory usage, but also for all computing and communication resources. Because of this, for example, the SpiNNaker spiking neuromorphic platform, although it uses 32-bit precision hardware, it restricts itself to integer arithmetic (instead of floating point), thus introducing some hardware simplifications (Furber et al., 2014 ). In another example, the TrueNorth platform (Cassidy et al., 2013 ) is built upon multiple 256 × 256 1-bit synaptic weight crossbars, although it includes extra circuitry to allow assigning up to four possible 8-bit values to the synapses (with some restrictions). In the world of non-spiking Deep Neural Networks (DNN), where there is now a strong quest for providing dedicated efficient hardware (Chen et al., 2016 ; Sim et al., 2016 ; Bong et al., 2017 ; Whatmough et al., 2017 ; Biswas and Chandrakasan, 2018 ; Gonugondla et al., 2018 ; Khwa et al., 2018 ), some theorists are studying ways to reduce bit precision of the weights down to 1-bit (Courbariaux et al., 2015 ; Rastegari et al., 2016 ) to help simplifying hardware. Here we focus on spiking neural network (SNN) hardware capable of on-line unsupervised learning through Spike-Time-Dependent-Plasticity (STDP). In a previous work (Yousefzadeh et al., 2017 ) we implemented deterministic STDP hardware with 9-bit resolution for the weights. Our goal here is to obtain a working learning (stochastic) STDP layer by restricting the weights to 1-bit precision (“0” or “1”) for both learning and inference phases. For this, we consider a feed-forward neural system made of two layers. The first layer, restricted to 1-bit weights, uses stochastic STDP unsupervised learning for learning representative features. The second layer uses some trainable SNN classifier with supervised learning for pattern classification (Stromatias et al., 2017 ; Yousefzadeh et al., 2017 ). Interestingly, the conventional full-precision additive STDP learning rule will result in a bimodal weight distribution after learning (Barbour et al., 2007 ; Galluppi et al., 2015 ). This means that, even in the case of graded synaptic weights, the final weight values tend to saturate to the minimum (disconnected) or maximum (fully connected) value. STDP was originally proposed by Gerstner et al. ( 1993 ), evolving later on to successfully learn hidden spiking patterns (Masquelier et al., 2008 ), to perform competitive spike pattern learning (Masquelier et al., 2009 ), to achieve reward modulated (pseudo-supervised) learning (Mozafari et al., 2017 ), or to be successfully applied to deep spiking neural networks (Kheradpisheh et al., 2018 ). Surprisingly, experimental evidence of STDP in biological synapses was reported by neuroscientists shortly after proposing the computational algorithms for the first time (Markram et al., 1997 ; Bi and Poo, 1998 , 2001 ; Zhang et al., 1998 ; Feldman, 2000 ; Mu and Poo, 2006 ; Cassenaer and Laurent, 2006 ; Jacob et al., 2007 ). However, STDP works typically by performing very small weight changes, which implies high resolution for the weights, consequently imposing high hardware resources demands for memory, computing, and communication circuits. Most of the neuromorphic hardware designs use on-chip memory to reduce power consumption (Cassidy et al., 2013 ; Davies et al., 2018 ) 1 and therefore memory is a limiting factor for the number of neurons and synaptic connections. By using 1-bit synaptic weights, not only the memory, but all the processing elements inside the chip will be much simpler. Reducing weight resolution from 9-bit to 1-bit, should in principle allow for about one order of magnitude reduction in memory, computing, communication resources and power consumption. When restricting to 1-bit weights, one option is to consider some type of stochastic weight update. This is, instead of applying a given weight change from the STDP rule, one changes the weight from either “0” to “1” or from “1” to “0” with a probability given by the STDP rule. This idea has already been used before. Suri et al. ( 2013 ) applied it to cluster vehicle trajectories recorded with spiking retinas (Dynamic Vision Sensors -DVS-) (Lichtsteiner et al., 2008 ; Posch et al., 2011 ; Serrano-Gotarredona and Linares-Barranco, 2013 ; Son et al., 2017 ; Guo et al., 2017 ) into highway lane segments. Seo and Seok ( 2015 ) applied it to simple classification problems, but found out that they could not learn to separate more than five patterns. Here we want to learn to classify either DVS recorded poker card symbols (Soto, 2017 ) or a spiking representation of the MNIST dataset (LeCun et al., 1998 ). At the beginning, we were not able to obtain any reasonable learning by simply applying an STDP binary update with stochasticity. It was not until we started applying some additional “tricks” that we started to observe the formation of characteristic features together with overall reasonable accuracy results. These “tricks” were weight normalization, individual neuron threshold adjustment, or using separate thresholds for learning and inference. Since there is no other work in the literature reporting classification accuracy results for 1-bit weights STDP feature extraction (FE) layers, we compare with purely random 1-bit weights FE layers. It is well-known that it is possible to build excellent performance pattern learning and classification systems by using a sufficiently large hidden FE layer with random weights, followed by a trainable high performance classification layer (Huang et al., 2006 ). Based on this, we compare our STDP systems with a “parallel” one using the same classifier output layer and a hidden layer with the same number of neurons but with 1-bit random weights. As shown later in the section 3, there is always a consistent improvement when using STDP with respect to using pure random weights, although this improvement reduces as the number of hidden layer neurons increases. Throughout the paper we will use always a very simple neuron model, namely, the linear-leak (or piece-wise linear-leak) integrate-and-fire (LIF) model, together with instantaneous synapses (Camuñas-Mesa et al., 2010 , 2011 , 2012 ; Pérez-Carrasco et al., 2013 ; Serrano-Gotarredona et al., 2015 ), restricted to positive neural state values. The paper is structured as follows. In the section 2 we will first include a quick review on different STDP rules, signaling the difference between more conventional time-based STDP rules vs. less conventional order-based STDP rules, which we will use here. Then we will briefly explain how we “engineer” STDP and, in particular, stochastic STDP with 1-bit weights by adding some “tricks.” Then we present quickly two previously reported spiking classifiers we have used, followed by a quick description of the software and hardware experimentation platforms used. In the section 3 we provide software and hardware results for three experiments. A first experiment is a very simple 4-neuron system that replicates a biological experiment of visual orientation tuning (Bienenstock et al., 1982 ; Moore and Freeman, 2012 ; Jeyabalaratnam et al., 2013 ). It is a very simple starting point used as reference in other computational studies (Galluppi et al., 2015 ). The second experiment learns to classify poker card symbols (Soto, 2017 ) recorded with a spiking retina Dynamic Vision Sensor (DVS) (Serrano-Gotarredona and Linares-Barranco, 2013 ). The third experiment learns to classify the MNIST dataset. More specifically, a spiking version of it, obtained by generating Poisson distributed spike trains from the original pixel gray levels. This is a common technique for converting static images to synthetic spike-trains and has been used in a number of previous studies (O'Connor et al., 2013 ; Querlioz et al., 2013 ; Diehl et al., 2015 ; Diehl and Cook, 2015 ; Galluppi et al., 2015 ; Stromatias et al., 2017 ). Finally, the paper finishes by presenting some Discussions and Conclusions.",
"discussion": "4. Discussion There is a growing interest in exploiting SNNs for practical hardware applications, not only because they approximate better the inherent operations of the brain, but also because only meaningful information, represented by spikes/events, consume computing resources and energy. Originally, there was almost no work on training SNNs directly in the spiking domain, and many research efforts were inverted in efficiently mapping conventional (frame-driven) ANNs (Artificial Neural Networks), conveniently trained in the frame-domain, to their SNN counterpart (O'Connor et al., 2013 ; Pérez-Carrasco et al., 2013 ; Diehl et al., 2015 ). However, recently, there has been a growing success in training networks directly in the spiking domain, either by developing some type of backpropagation technique for the spiking domain (Lee et al., 2016 ; Neftci et al., 2017 ; Mostafa, 2018 ), by exploiting STDP at some level of the network (Querlioz et al., 2013 ; Diehl and Cook, 2015 ; Neftci et al., 2015 , 2016 ; Kheradpisheh et al., 2016 ), or by some ad-hoc technique (Lagorce et al., 2016 ; Negri, 2018 ). For the simple 4-class DVS-recorded poker card recognition datasets (Serrano-Gotarredona and Linares-Barranco, 2015 ; Soto, 2017 ), reported results are summarized in Table 3 . There are two poker card dataset versions. The fast one (Serrano-Gotarredona and Linares-Barranco, 2015 ) of visual field resolution 32 × 32 pixels, where symbols cross the screen in about 20–30 ms. And the slow one (Soto, 2017 ) of visual field resolution 128 × 128 pixels, where symbols move slowly and recordings are cut into sequences of about between 0.75 (10 kesl) and 3.75 (50 kesl) seconds. The fast set has been used by several researchers. In the original paper using the fast dataset (Pérez-Carrasco et al., 2013 ), 91% of CA was achieved with around the first 500 events, which corresponds to about the first 1–3 ms of a recording. This was achieved by training a 3-layer ConvNet in the frame-domain using backpropagation and then mapping it to an SNN. Orchard (Orchard et al., 2015 ) used a 1-layer ConvNet with 18 Gabor filters with a heuristic classification method, achieving 97.5% CA after processing the full recordings, each with about 5 k events. The HOTS technique (Lagorce et al., 2016 ) based on computing with space-time surfaces, achieved 100% CA by processing also the full recordings. Recently, another mapping method from the frame-domain to SNN was reported (Kaiser et al., 2017 ) based on applying Contrastive Divergence (CD) on a hidden layer of a generative model with 10 convolutional feature maps, achieving 91% CA after processing the full recordings. Negri developed a method based on histrogramming, which is capable of recognizing extremely fast (0.6 ms or less than 50 events), with a reasonable accuracy of 96%. Stromatias et al. (Stromatias et al., 2017 ) used a 1-layer ConvNet with 18 Gabor filters followed by an SNN classifier. They achieved 99.77% CA on the fast set and 100% on the slow set. For all the above methods, synaptic weights used the full resolution of 64-bit floating point precision, except for (Stromatias et al., 2017 ) who used integer precision of 24-bit for the classifier layer and 32-bit for the convolution layer. In this work we used just 1-bit precision for the first layer, trained with STDP directly in the spiking domain, with the same 24-bit integer resolution classifier layer than (Stromatias et al., 2017 ), achieving 100% CA on the slow set. Unfortunately, we could not use the fast set because we needed more training samples for STDP. Table 3 Comparison of reported results on DVS-recorded Poker datasets (“Fast” and “Slow”). Table 4 shows SNN results reported for spiking versions of the MNIST dataset. As indicated in the last column, the original frame-based data set is converted artificially into spikes by either mapping a pixel value into an average frequency of a spike train with Poisson distribution, by mapping the pixel value to a delay (resulting in one spike per pixel), or by flashing the MNIST digit on a monitor while recording spikes with a DVS spiking retina sensor (O'Connor et al., 2013 ) 14 . Table 4 is divided in two parts by a double horizontal line. The top part uses synaptic weights with high resolution, while for the bottom part synapse weight resolution has been reduced to 8-bit or less. The entries shown in red correspond to networks that have used STDP for some of their weights training. The second last column indicates whether the models are generative or not 15 . Column “Structure” indicates the number of layers of each system and the number of neurons (or convolutional feature maps) per layer. Column “Type” indicates whether the architecture is a pure MLP (multi-layer preceptron), includes convolutional layers (CONV), or is a generative model using a DBN (deep belief network) with RBMs (Restricted Boltzmann Machine). Columns “FE Training” and “Classifier Training” specify how the hidden and output classifier layers have been trained. The columns on“Resolution” indicate the resolution used for the weights of the synapses in the hidden and the classifier layers. Column “Classification Accuracy” (CA) compares the recognition performance obtained for the different architectures and methods. We can see that for the systems using some type of synaptic precision reduction (below 8-bit) combined with STDP, the technique presented in this paper shows the best CA for the specific cases analyzed. The only system with weight precision reduction (8-bit) but without STDP that improves our CA is the one by Neftci (Neftci et al., 2017 ) based on event-driven random backpropagation. For the cases with full precision but using STDP, only (Kheradpisheh et al., 2018 ) improves our CA, but at the expense of using three convolution layers. Table 4 also includes (under “This Work”) the results we obtained with fixed-random weights, which are very similar to the ones obtained by STDP. Table 4 Comparison of reported results on spiking versions of the static MNIST dataset. Experimental results with STDP learning rule are marked in red color . The above Tables 3 , 4 show that using 1-bit weights for the FE layer (whether fixed-random or trained by STDP) results, from the computational point of view, in overall CAs which are comparable to related state of the art results on limited precision weights and STDP learning systems. However, the most interesting benefits of the presented 1-bit weight technique is its efficiency for hardware implementations. There are not many STDP hardware systems reported, implemented on FPGAs. Table 5 compares similar hardware systems implementing on-line STDP on spiking neural networks. Spartan-3 FPGAs use (according to our experience) about 10% more resources than Spartan-6 for the same system. As we can see, our technique results in about two orders of magnitude in resources consumption (slices per neuron) efficiency with respect to (Rice et al., 2009 ), and about one order of magnitude with respect to (Cassidy et al., 2007 ) (and without using Block-RAM). This is achieved thanks to the fact of using 1-bit weights, resulting in high reductions in memory resources, computing resources, and communication resources. Table 5 Comparison of reported results on STDP MLP FPGA Hardware. There are some more recent works reporting on STDP based synapses implemented on FPGAs (Pedroni et al., 2016 ; Jokar and Soleimani, 2017 ; Nouri et al., 2018 ; Lammie et al., 2018 ). Unfortunately, hardware resources are reported only for a single STDP synapse/unit, making it difficult to compare with a full STDP system implementation, since it is not clear how reported synaptic resources can be shared or time-multiplexed at system level. Also, the resources for the rest of computing/memory/communication requirements are not specified."
} | 4,529 |
35149672 | PMC8837800 | pmc | 879 | {
"abstract": "Advances in synthetic biology permit the genetic encoding of synthetic chemistries at monomeric precision, enabling the synthesis of programmable proteins with tunable properties. Bacterial pili serve as an attractive biomaterial for the development of engineered protein materials due to their ability to self-assemble into mechanically robust filaments. However, most biomaterials lack electronic functionality and atomic structures of putative conductive proteins are not known. Here, we engineer high electronic conductivity in pili produced by a genomically-recoded E. coli strain. Incorporation of tryptophan into pili increased conductivity of individual filaments >80-fold. Computationally-guided ordering of the pili into nanostructures increased conductivity 5-fold compared to unordered pili networks. Site-specific conjugation of pili with gold nanoparticles, facilitated by incorporating the nonstandard amino acid propargyloxy-phenylalanine, increased filament conductivity ~170-fold. This work demonstrates the sequence-defined production of highly-conductive protein nanowires and hybrid organic-inorganic biomaterials with genetically-programmable electronic functionalities not accessible in nature or through chemical-based synthesis.",
"introduction": "Introduction Materials produced from synthetic chemical processes provide access to a broad range of chemical structures yet are constrained by the lack of sequence-defined polymerization methods. In contrast, biological systems employ sequence-controlled processes to synthesize biomolecules, in which the molecular information encoded by nucleic acids is converted into sequence-controlled protein polymers 1 . Protein polymers have evolved to assume specialized functions in nature, among which is the formation of dynamic protein-based materials (e.g., collagen, silk, and elastin). These multifunctional materials possess versatile functions spanning a range of strength, elasticity, and stability, but lack electronic or optical functionality. Engineered living materials with programmable functionalities and environmental resilience are attractive biomaterials due to their ability to regenerate, sense, and adapt to environmental cues 2 . However, nature is constrained to a small set of organic monomeric building blocks, the 20 canonical amino acids, thereby limiting the chemical diversity of polymeric biomaterials. Expanding the chemical palette of genetically encoded chemistries could yield new classes of enzymes, materials, and therapeutics produced in a sequence-defined manner with diverse chemistries. Many bacteria produce filamentous protein appendages on their surface called pili, which are critical to bacterial infections due to their roles in host colonization and surface sensing 3 , bacterial motility 4 , and natural competence 5 . In addition to their biomedical importance, pili filaments are attractive biomaterials due to their capacity to self-assemble through natural polymerization while retaining extraordinary mechanical stability and robustness, being able to withstand a wide range of temperatures, pH, and protein-denaturing agents such as SDS and urea 6 – 9 . However, there are several major challenges in the use of pili as multifunctional biomaterials. First, like most proteins, pili lack electronic or optical functionality, which are critical for the development of next-generation bioelectronics. It was previously thought that some soil bacteria such as Geobacter sulfurreducens produce conductive type IV pili. However, structural studies revealed that conductive filaments on the bacterial surface are polymerized cytochromes whereas pili remain inside the cell and are involved in the secretion of filamentous cytochromes 10 – 14 . G. sulfurreducens pili are bipartite comprised of the PilA-N and PilA-C proteins, show overall structure similar to the type IV pili 15 , but extend past the bacterial surface only when artificially overexpressed, and show very low conductivity 10 . A previous study has claimed conductivity in synthetic pili 16 , however the conductivity of the individual synthetic filaments has not been demonstrated along their length, only across their diameter, providing no evidence that the pili could be conductive down their length like a nanowire. Furthermore, their exact biochemical composition is unknown as discussed at length in a previous study 10 . Second, structures of these putative conductive “pili are not available, hindering the elucidation and prediction of structure-function correlations. The final obstacle in using pili as multifunctional biomaterials is that, although in vitro assembly of conductive proteins is feasible, they tend to aggregate, which is not suitable for mass production 16 . Therefore, novel methods for in vivo biomaterial production are required for large-scale production of pili biomaterials endowed with tunable electronic and mechanical functionalities. In this study we address these challenges by pursuing three strategies that demonstrate large-scale production of conductive pili proteins with tunable electronic properties. First, we strategically encode standard aromatic amino acid mutations into E. coli type 1 pili (Fig. 1a, b ) using a cryo-EM structure of E. coli type 1 pili as a guide (PDB: 6c53) 9 , 17 , and demonstrate mutation-dependent 10- to 84-fold increases in the conductivity of the filaments. Importantly, our work utilizes methods to measure the conductivity of individual filaments rather than filament films, which is distinct to prior conductivity studies of protein networks where high contact resistance between filaments and measurement electrodes has masked the intrinsic electronic properties of the individual filaments 18 – 20 . Our approach also eliminates network artifacts where conductivity is dominated by percolation behavior, and reduces the impact of inter-filament contact resistance on the conductivity measurements 21 . Second, we engineer long-range conductivity over the micrometer scale by generating networks of hierarchical assemblies of conductive pili using molecular recognition self-assembly 22 (Fig. 1c ). Finally, we use a genomically recoded strain of E. coli 23 to genetically encode the nonstandard amino acid (nsAA) propargyloxy-phenylalanine (PrOF) to generate pili that form a click-chemistry-functional scaffold for the precise, site-specific conjugation of gold nanoparticles (AuNPs), leading to the biosynthesis of sequence-controlled organic-inorganic hybrid biomaterials endowed with conductivity enhanced by 170-fold (Fig. 1d ). Fig. 1 Strategy to engineer electronic conductivity into E. coli pili nanofilaments. a Representative TEM image of an E. coli cell (left) expressing pili and purified pili (right). Scale bars, 200 nm (left) and 100 nm (right). b Cryo-EM structure of 8 FimA monomers forming mature pilus. Positions 80 and 109 in each monomer are highlighted with cyan and red spheres respectively. side view (left), front view (right). a , b experiments replicated independently greater than 15 times with similar results. c Strategy to develop hierarchical ordered structures with enhanced conductivity. d Schematic of creating organic-inorganic hybrid pili using gold nanoparticles clicked on through azido-alkyne click chemistry functionality encoded with nsAAs.",
"discussion": "Discussion Our work combines the precise, site-specific engineering of bacterial nanowires enabled using synthetic biology methods with highly accurate conductivity measurements on individual filaments and filament nanostructures to demonstrate the production of multi-functional, highly conductive pili biomaterials. Guided by cryo-EM structures 9 , 17 , we used MAGE to generate a targeted combinatorial library of genomic type 1 pili mutants to screen for aromatic amino acid mutations that retain pili self-assembly and exhibit increased conductivity. Notably, pili mutants exhibited a wide range of electronic conductivities based on the insertion site and aromaticity of the amino acids inserted into the pili. The insertion of aromatic amino acids into type 1 pili increased pili up to 84-fold with the double tryptophan mutant. Our use of single-filament conductivity measurements allowed us to directly ascribe an increase in conductivity of a filament to the incorporation of aromatic amino acids, providing a higher level of engineered pili characterization compared to commonly used film measurements. The use of molecular dynamics simulations demonstrated more frequent association between pilus proteins in the presence of HMD, which informed the construction of hierarchical multidimensional nanostructures that demonstrated the efficient transport of charges over micrometer distances under ordinary thermal conditions. Bundling of pili increased conductivity 5-fold over the micrometer scale, implying that bundling may further facilitate electron transport down the filaments. Finally, using the genomically recoded organism (GRO) and a highly efficient orthogonal translation system (OTS) able to incorporate nsAAs into proteins with >95% efficiency 25 , we inserted 2NaA, pAzF, and PrOF at genetically encoded locations within the FimA monomer and thus across the entire surface of the pilus. We then conjugated azide-covered AuNPs to the terminal alkyne groups on PrOF residues in pili using the Cu-catalyzed cycloaddition “click” chemistry reaction. Using PrOF to directly label pili proteins with AuNP at single-residue precision is a strategy presented herein. The use of GROs and an efficient OTS to incorporate “click”-functional nsAAs into a bacterial nanowire allowed us to produce a class of hybrid organic-inorganic biomaterials which would otherwise be difficult to produce at scale without recoded bacteria capable of efficiently incorporating nsAAs. The efficient and site-specific incorporation of PrOF into individual FimA monomers allowed pili, large filaments made of polymerized FimA protein, to be uniformly covered in AuNPs—this advance demonstrates the ability to site-specifically functionalize large, micrometer-scale protein structures by re-purposing open codons (e.g., UAG) to encode nsAAs at high efficiency in a recoded bacterial host. This functionalization allowed the pili to be used as a chassis for a hybrid organic-inorganic material engineered with conductivity ~170-fold higher than wild-type. Looking forward, our studies demonstrate the potential to use recoded E. coli strains for the production of genetically encoded hybrid biomaterials comprising organic and inorganic components. We have demonstrated the capability to endow functionality into pili using multisite incorporation of nsAAs with diverse chemical modalities. Building on our work, a large and diverse array of chemical groups can be “clicked” onto pili-based materials. For example, taking inspiration from the stacked hemes in conductive G. sulfurreducens filaments 11 , 12 , hemes modified with click-able moieties can be uniformly attached down the length of the pili chassis as another method of increasing of filament conductivity. As our approach to generate hierarchical structures can lead to multidimensional assembly of pili, future studies can incorporate photoreactive, cross-linkable nsAAs such as pAzF 44 into pili to generate light-activated arrays of conductive protein biomaterials. Furthermore, since our engineered pili are almost identical to those used previously to create pili nanostructures 22 , other programmable assemblies, with varying effects on conductivity, are possible using our conductive pili. Applied work could examine the use of these electronically active pili in device applications such as sensors, transistors, and capacitors that span biological and electrical systems. More broadly, the capability of using GROs to produce multifunctional, structurally complex materials can be expanded toward the development of a programmable class of genetically encoded biomaterials with diverse chemistries and functions."
} | 3,009 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.